Skip to contentSkip to conversation

siginfo_t ABI break on sparc64 from si_addr_lsb move 3y ago

RFC PATCH Superseded Marco Elver 9 participants 118 messages Apr 29, 2021 – Jul 16, 2021 Page 1 of 2

Subsystems (12)
Arm (32-bit) x86 VFS Kselftest +8 more subsystems
lore
Share & tools
lore
#0 lore raw

Hello, Eric,

By inspecting the logs I've seen that about 3 years ago there had been a number of siginfo_t cleanups. This included moving si_addr_lsb:

b68a68d3dcc1 ("signal: Move addr_lsb into the _sigfault union for clarity") 859d880cf544 ("signal: Correct the offset of si_pkey in struct siginfo") 8420f71943ae ("signal: Correct the offset of si_pkey and si_lower in struct siginfo on m68k")

In an ideal world, we could just have si_addr + the union in _sigfault, but it seems there are more corner cases. :-/

The reason I've stumbled upon this is that I wanted to add the just merged si_perf [1] field to glibc. But what I noticed is that glibc's definition and ours are vastly different around si_addr_lsb, si_lower, si_upper, and si_pkey.

[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=42dec9a936e7696bea1f27d3c5a0068cd9aa95fd

In our current definition of siginfo_t, si_addr_lsb is placed into the same union as si_lower, si_upper, and si_pkey (and now si_perf). From the logs I see that si_lower, si_upper, and si_pkey are padded because si_addr_lsb used to be outside the union, which goes back to "signal: Move addr_lsb into the _sigfault union for clarity".

Since then, si_addr_lsb must also be pointer-aligned, because the union containing it must be pointer-aligned (because si_upper, si_lower). On all architectures where si_addr_lsb is right after si_addr, this is perfectly fine, because si_addr itself is a pointer...

... except for the anomaly that are 64-bit architectures that define __ARCH_SI_TRAPNO and want that 'int si_trapno'. Like, for example sparc64, which means siginfo_t's ABI has been subtly broken on sparc64 since v4.16.

The following static asserts illustrate this:

arch/sparc/kernel/signal_64.c +28 -0
@@ -556,3 +556,37 @@ void do_notify_resume(struct pt_regs *regs, unsigned long orig_i0, unsigned long
 	user_enter();
 }
 
+static_assert(offsetof(siginfo_t, si_signo)	== 0);
+static_assert(offsetof(siginfo_t, si_errno)	== 4);
+static_assert(offsetof(siginfo_t, si_code)	== 8);
+static_assert(offsetof(siginfo_t, si_pid)	== 16);
+static_assert(offsetof(siginfo_t, si_uid)	== 20);
+static_assert(offsetof(siginfo_t, si_tid)	== 16);
+static_assert(offsetof(siginfo_t, si_overrun)	== 20);
+static_assert(offsetof(siginfo_t, si_status)	== 24);
+static_assert(offsetof(siginfo_t, si_utime)	== 32);
+static_assert(offsetof(siginfo_t, si_stime)	== 40);
+static_assert(offsetof(siginfo_t, si_value)	== 24);
+static_assert(offsetof(siginfo_t, si_int)	== 24);
+static_assert(offsetof(siginfo_t, si_ptr)	== 24);
+static_assert(offsetof(siginfo_t, si_addr)	== 16);
+static_assert(offsetof(siginfo_t, si_trapno)	== 24);
+#if 1 /* Correct offsets, obtained from v4.14 */
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 28);
+static_assert(offsetof(siginfo_t, si_lower)	== 32);
+static_assert(offsetof(siginfo_t, si_upper)	== 40);
+static_assert(offsetof(siginfo_t, si_pkey)	== 32);
+#else /* Current offsets, as of v4.16 */
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 32);
+static_assert(offsetof(siginfo_t, si_lower)	== 40);
+static_assert(offsetof(siginfo_t, si_upper)	== 48);
+static_assert(offsetof(siginfo_t, si_pkey)	== 40);
+#endif
+static_assert(offsetof(siginfo_t, si_band)	== 16);
+static_assert(offsetof(siginfo_t, si_fd)	== 20);

---

Granted, nobody seems to have noticed because I don't even know if these fields have use on sparc64. But I don't yet see this as justification to leave things as-is...

The collateral damage of this, and the acute problem that I'm having is defining si_perf in a sort-of readable and portable way in siginfo_t definitions that live outside the kernel, where sparc64 does not yet have broken si_addr_lsb. And the same difficulty applies to the kernel if we want to unbreak sparc64, while not wanting to move si_perf for other architectures.

There are 2 options I see to solve this:

1. Make things simple again. We could just revert the change moving si_addr_lsb into the union, and sadly accept we'll have to live with that legacy "design" mistake. (si_perf stays in the union, but will unfortunately change its offset for all architectures... this one-off move might be ok because it's new.)

2. Add special cases to retain si_addr_lsb in the union on architectures that do not have __ARCH_SI_TRAPNO (the majority). I have added a draft patch that would do this below (with some refactoring so that it remains sort-of readable), as an experiment to see how complicated this gets.

Option (1) means we'll forever be wasting the space where si_addr_lsb lives (including the padding). It'd also mean we'd move si_perf for _all_ architectures -- this might be acceptable, given there is no stable release with it yet -- the fix just needs to be merged before the release of v5.13! It is the simpler option though -- and I don't know if we need all this complexity.

Option (2) perhaps results in better space utilization. Maybe that's better long-term if we worry about space in some rather distant future -- where we need those 8 bytes on 64-bit architectures to not exceed 128 bytes. This option, however, doesn't just make us carry this complexity forever, but also forces it onto everyone else, like glibc and other libcs (including those in other languages with C FFIs) which have their own definition of siginfo_t.

Which option do you prefer? Are there better options?

Many thanks,
-- Marco

------ >8 ------

Option #2 draft:

arch/sparc/kernel/signal_64.c +34 -0
index a0eec62c825d..150ee27b1423 100644
@@ -556,3 +556,37 @@ void do_notify_resume(struct pt_regs *regs, unsigned long orig_i0, unsigned long
 	user_enter();
 }
 
+/*
+ * Compile-time assertions for siginfo_t offsets. Unlike other architectures,
+ * sparc64 is special, because it requires si_trapno (int), and the following
+ * si_addr_lsb (short) need not be word aligned. Accidental changes around the
+ * offset of si_addr_lsb and the following fields would only be caught here.
+ */
+static_assert(offsetof(siginfo_t, si_signo)	== 0);
+static_assert(offsetof(siginfo_t, si_errno)	== 4);
+static_assert(offsetof(siginfo_t, si_code)	== 8);
+static_assert(offsetof(siginfo_t, si_pid)	== 16);
+static_assert(offsetof(siginfo_t, si_uid)	== 20);
+static_assert(offsetof(siginfo_t, si_tid)	== 16);
+static_assert(offsetof(siginfo_t, si_overrun)	== 20);
+static_assert(offsetof(siginfo_t, si_status)	== 24);
+static_assert(offsetof(siginfo_t, si_utime)	== 32);
+static_assert(offsetof(siginfo_t, si_stime)	== 40);
+static_assert(offsetof(siginfo_t, si_value)	== 24);
+static_assert(offsetof(siginfo_t, si_int)	== 24);
+static_assert(offsetof(siginfo_t, si_ptr)	== 24);
+static_assert(offsetof(siginfo_t, si_addr)	== 16);
+static_assert(offsetof(siginfo_t, si_trapno)	== 24);
+#if 1 /* Correct offsets, obtained from v4.14 */
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 28);
+static_assert(offsetof(siginfo_t, si_lower)	== 32);
+static_assert(offsetof(siginfo_t, si_upper)	== 40);
+static_assert(offsetof(siginfo_t, si_pkey)	== 32);
+#else /* Current offsets, as of v4.16 */
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 32);
+static_assert(offsetof(siginfo_t, si_lower)	== 40);
+static_assert(offsetof(siginfo_t, si_upper)	== 48);
+static_assert(offsetof(siginfo_t, si_pkey)	== 40);
+#endif
+static_assert(offsetof(siginfo_t, si_band)	== 16);
+static_assert(offsetof(siginfo_t, si_fd)	== 20);
include/linux/compat.h +29 -12

61 lines not on this page · raw message

include/uapi/asm-generic/siginfo.h +45 -15

92 lines not on this page · raw message

#1 lore raw
Quoted 93 lines · Marco Elver

Hello, Eric,

By inspecting the logs I've seen that about 3 years ago there had been a number of siginfo_t cleanups. This included moving si_addr_lsb:

b68a68d3dcc1 ("signal: Move addr_lsb into the _sigfault union for clarity") 859d880cf544 ("signal: Correct the offset of si_pkey in struct siginfo") 8420f71943ae ("signal: Correct the offset of si_pkey and si_lower in struct siginfo on m68k")

In an ideal world, we could just have si_addr + the union in _sigfault, but it seems there are more corner cases. :-/

The reason I've stumbled upon this is that I wanted to add the just merged si_perf [1] field to glibc. But what I noticed is that glibc's definition and ours are vastly different around si_addr_lsb, si_lower, si_upper, and si_pkey.

[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=42dec9a936e7696bea1f27d3c5a0068cd9aa95fd

In our current definition of siginfo_t, si_addr_lsb is placed into the same union as si_lower, si_upper, and si_pkey (and now si_perf). From the logs I see that si_lower, si_upper, and si_pkey are padded because si_addr_lsb used to be outside the union, which goes back to "signal: Move addr_lsb into the _sigfault union for clarity".

Since then, si_addr_lsb must also be pointer-aligned, because the union containing it must be pointer-aligned (because si_upper, si_lower). On all architectures where si_addr_lsb is right after si_addr, this is perfectly fine, because si_addr itself is a pointer...

... except for the anomaly that are 64-bit architectures that define __ARCH_SI_TRAPNO and want that 'int si_trapno'. Like, for example sparc64, which means siginfo_t's ABI has been subtly broken on sparc64 since v4.16.

The following static asserts illustrate this:

arch/sparc/kernel/signal_64.c +28 -1
@@ -556,3 +556,37 @@ void do_notify_resume(struct pt_regs *regs, unsigned long orig_i0, unsigned long
 	user_enter();
 }
 
+static_assert(offsetof(siginfo_t, si_signo)	== 0);
+static_assert(offsetof(siginfo_t, si_errno)	== 4);
+static_assert(offsetof(siginfo_t, si_code)	== 8);
+static_assert(offsetof(siginfo_t, si_pid)	== 16);
+static_assert(offsetof(siginfo_t, si_uid)	== 20);
+static_assert(offsetof(siginfo_t, si_tid)	== 16);
+static_assert(offsetof(siginfo_t, si_overrun)	== 20);
+static_assert(offsetof(siginfo_t, si_status)	== 24);
+static_assert(offsetof(siginfo_t, si_utime)	== 32);
+static_assert(offsetof(siginfo_t, si_stime)	== 40);
+static_assert(offsetof(siginfo_t, si_value)	== 24);
+static_assert(offsetof(siginfo_t, si_int)	== 24);
+static_assert(offsetof(siginfo_t, si_ptr)	== 24);
+static_assert(offsetof(siginfo_t, si_addr)	== 16);
+static_assert(offsetof(siginfo_t, si_trapno)	== 24);
+#if 1 /* Correct offsets, obtained from v4.14 */
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 28);
+static_assert(offsetof(siginfo_t, si_lower)	== 32);
+static_assert(offsetof(siginfo_t, si_upper)	== 40);
+static_assert(offsetof(siginfo_t, si_pkey)	== 32);
+#else /* Current offsets, as of v4.16 */
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 32);
+static_assert(offsetof(siginfo_t, si_lower)	== 40);
+static_assert(offsetof(siginfo_t, si_upper)	== 48);
+static_assert(offsetof(siginfo_t, si_pkey)	== 40);
+#endif
+static_assert(offsetof(siginfo_t, si_band)	== 16);
+static_assert(offsetof(siginfo_t, si_fd)	== 20);
 
---

Granted, nobody seems to have noticed because I don't even know if these fields have use on sparc64. But I don't yet see this as justification to leave things as-is...

The collateral damage of this, and the acute problem that I'm having is defining si_perf in a sort-of readable and portable way in siginfo_t definitions that live outside the kernel, where sparc64 does not yet have broken si_addr_lsb. And the same difficulty applies to the kernel if we want to unbreak sparc64, while not wanting to move si_perf for other architectures.

There are 2 options I see to solve this:

1. Make things simple again. We could just revert the change moving si_addr_lsb into the union, and sadly accept we'll have to live with that legacy "design" mistake. (si_perf stays in the union, but will unfortunately change its offset for all architectures... this one-off move might be ok because it's new.)

Quoted 7 lines

2. Add special cases to retain si_addr_lsb in the union on architectures that do not have __ARCH_SI_TRAPNO (the majority). I have added a draft patch that would do this below (with some refactoring so that it remains sort-of readable), as an experiment to see how complicated this gets.

Which option do you prefer? Are there better options?

Personally the most important thing to have is a single definition shared by all architectures so that we consolidate testing.

A little piece of me cries a little whenever I see how badly we implemented the POSIX design. As specified by POSIX the fields can be place in siginfo such that 32bit and 64bit share a common definition. Unfortunately we did not addpadding after si_addr on 32bit to accommodate a 64bit si_addr.

I find it unfortunate that we are adding yet another definition that requires translation between 32bit and 64bit, but I am glad that at least the translation is not architecture specific. That common definition is what has allowed this potential issue to be caught and that makes me very happy to see.

Let's go with Option 3.

Confirm BUS_MCEERR_AR, BUS_MCEERR_AO, SEGV_BNDERR, SEGV_PKUERR are not in use on any architecture that defines __ARCH_SI_TRAPNO, and then fixup the userspace definitions of these fields.

To the kernel I would add some BUILD_BUG_ON's to whatever the best maintained architecture (sparc64?) that implements __ARCH_SI_TRAPNO just to confirm we don't create future regressions by accident.

I did a quick search and the architectures that define __ARCH_SI_TRAPNO are sparc, mips, and alpha. All have 64bit implementations. A further quick search shows that none of those architectures have faults that use BUS_MCEERR_AR, BUS_MCEERR_AO, SEGV_BNDERR, SEGV_PKUERR, nor do they appear to use mm/memory-failure.c

So it doesn't look like we have an ABI regression to fix.

Eric

#2 lore raw

On Thu, 29 Apr 2021 at 19:24, Eric W. Biederman <[email protected]> wrote:
[...]

Quoted 27 lines
Quoted 19 lines

Granted, nobody seems to have noticed because I don't even know if these fields have use on sparc64. But I don't yet see this as justification to leave things as-is...

The collateral damage of this, and the acute problem that I'm having is defining si_perf in a sort-of readable and portable way in siginfo_t definitions that live outside the kernel, where sparc64 does not yet have broken si_addr_lsb. And the same difficulty applies to the kernel if we want to unbreak sparc64, while not wanting to move si_perf for other architectures.

There are 2 options I see to solve this:

1. Make things simple again. We could just revert the change moving si_addr_lsb into the union, and sadly accept we'll have to live with that legacy "design" mistake. (si_perf stays in the union, but will unfortunately change its offset for all architectures... this one-off move might be ok because it's new.)

Quoted 7 lines

2. Add special cases to retain si_addr_lsb in the union on architectures that do not have __ARCH_SI_TRAPNO (the majority). I have added a draft patch that would do this below (with some refactoring so that it remains sort-of readable), as an experiment to see how complicated this gets.

Which option do you prefer? Are there better options?

Quoted 8 lines

Personally the most important thing to have is a single definition shared by all architectures so that we consolidate testing.

A little piece of me cries a little whenever I see how badly we implemented the POSIX design. As specified by POSIX the fields can be place in siginfo such that 32bit and 64bit share a common definition. Unfortunately we did not addpadding after si_addr on 32bit to accommodate a 64bit si_addr.

I think it's even worse than that, see the fun I had with siginfo last
week: https://lkml.kernel.org/r/[email protected]
... because of the 3 initial ints and no padding after them, we can't portably add __u64 fields to siginfo, and are forever forced to have subtly different behaviour between 32-bit and 64-bit architectures. :-/

Quoted 16 lines

I find it unfortunate that we are adding yet another definition that requires translation between 32bit and 64bit, but I am glad that at least the translation is not architecture specific. That common definition is what has allowed this potential issue to be caught and that makes me very happy to see.

Let's go with Option 3.

Confirm BUS_MCEERR_AR, BUS_MCEERR_AO, SEGV_BNDERR, SEGV_PKUERR are not in use on any architecture that defines __ARCH_SI_TRAPNO, and then fixup the userspace definitions of these fields.

To the kernel I would add some BUILD_BUG_ON's to whatever the best maintained architecture (sparc64?) that implements __ARCH_SI_TRAPNO just to confirm we don't create future regressions by accident.

Quoted 7 lines

I did a quick search and the architectures that define __ARCH_SI_TRAPNO are sparc, mips, and alpha. All have 64bit implementations. A further quick search shows that none of those architectures have faults that use BUS_MCEERR_AR, BUS_MCEERR_AO, SEGV_BNDERR, SEGV_PKUERR, nor do they appear to use mm/memory-failure.c

So it doesn't look like we have an ABI regression to fix.

That sounds fine to me -- my guess was that they're not used on these architectures, but I just couldn't make that call.

I have patches adding compile-time asserts for sparc64, arm, arm64 ready to go. I'll send them after some more testing.

Thanks,
-- Marco

maintainer
#3 lore raw
In reply to Eric W. Biederman
Quoted 23 lines · Eric W. Biederman
Quoted 1 line

Which option do you prefer? Are there better options?

Personally the most important thing to have is a single definition shared by all architectures so that we consolidate testing.

A little piece of me cries a little whenever I see how badly we implemented the POSIX design. As specified by POSIX the fields can be place in siginfo such that 32bit and 64bit share a common definition. Unfortunately we did not addpadding after si_addr on 32bit to accommodate a 64bit si_addr.

I find it unfortunate that we are adding yet another definition that requires translation between 32bit and 64bit, but I am glad that at least the translation is not architecture specific. That common definition is what has allowed this potential issue to be caught and that makes me very happy to see.

Let's go with Option 3.

Confirm BUS_MCEERR_AR, BUS_MCEERR_AO, SEGV_BNDERR, SEGV_PKUERR are not in use on any architecture that defines __ARCH_SI_TRAPNO, and then fixup the userspace definitions of these fields.

Quoted 6 lines

To the kernel I would add some BUILD_BUG_ON's to whatever the best maintained architecture (sparc64?) that implements __ARCH_SI_TRAPNO just to confirm we don't create future regressions by accident.

I did a quick search and the architectures that define __ARCH_SI_TRAPNO are sparc, mips, and alpha. All have 64bit implementations.

I think you (slightly) misread: mips has "#undef __ARCH_SI_TRAPNO", not "#define __ARCH_SI_TRAPNO". This means it's only sparc and alpha.

I can see that the alpha instance was added to the kernel during linux-2.5, but never made it into the glibc or uclibc copy of the struct definition, and musl doesn't support alpha or sparc. Debian codesearch only turns up sparc (and BSD) references to si_trapno.

Quoted 7 lines

I did a quick search and the architectures that define __ARCH_SI_TRAPNO are sparc, mips, and alpha. All have 64bit implementations. A further quick search shows that none of those architectures have faults that use BUS_MCEERR_AR, BUS_MCEERR_AO, SEGV_BNDERR, SEGV_PKUERR, nor do they appear to use mm/memory-failure.c

So it doesn't look like we have an ABI regression to fix.

Even better!

So if sparc is the only user of _trapno and it uses none of the later fields in _sigfault, I wonder if we could take even more liberty at trying to have a slightly saner definition. Can you think of anything that might break if we put _trapno inside of the union along with _perf and _addr_lsb?

I suppose in theory sparc64 or alpha might start using the other fields in the future, and an application might be compiled against mismatched headers, but that is unlikely and is already broken with the current headers.

Arnd

#4 lore raw
Quoted 51 lines · Arnd Bergmann
Quoted 23 lines · Eric W. Biederman
Quoted 1 line

Which option do you prefer? Are there better options?

Personally the most important thing to have is a single definition shared by all architectures so that we consolidate testing.

A little piece of me cries a little whenever I see how badly we implemented the POSIX design. As specified by POSIX the fields can be place in siginfo such that 32bit and 64bit share a common definition. Unfortunately we did not addpadding after si_addr on 32bit to accommodate a 64bit si_addr.

I find it unfortunate that we are adding yet another definition that requires translation between 32bit and 64bit, but I am glad that at least the translation is not architecture specific. That common definition is what has allowed this potential issue to be caught and that makes me very happy to see.

Let's go with Option 3.

Confirm BUS_MCEERR_AR, BUS_MCEERR_AO, SEGV_BNDERR, SEGV_PKUERR are not in use on any architecture that defines __ARCH_SI_TRAPNO, and then fixup the userspace definitions of these fields.

Quoted 6 lines

To the kernel I would add some BUILD_BUG_ON's to whatever the best maintained architecture (sparc64?) that implements __ARCH_SI_TRAPNO just to confirm we don't create future regressions by accident.

I did a quick search and the architectures that define __ARCH_SI_TRAPNO are sparc, mips, and alpha. All have 64bit implementations.

I think you (slightly) misread: mips has "#undef __ARCH_SI_TRAPNO", not "#define __ARCH_SI_TRAPNO". This means it's only sparc and alpha.

I can see that the alpha instance was added to the kernel during linux-2.5, but never made it into the glibc or uclibc copy of the struct definition, and musl doesn't support alpha or sparc. Debian codesearch only turns up sparc (and BSD) references to si_trapno.

Quoted 7 lines

I did a quick search and the architectures that define __ARCH_SI_TRAPNO are sparc, mips, and alpha. All have 64bit implementations. A further quick search shows that none of those architectures have faults that use BUS_MCEERR_AR, BUS_MCEERR_AO, SEGV_BNDERR, SEGV_PKUERR, nor do they appear to use mm/memory-failure.c

So it doesn't look like we have an ABI regression to fix.

Quoted 7 lines

Even better!

So if sparc is the only user of _trapno and it uses none of the later fields in _sigfault, I wonder if we could take even more liberty at trying to have a slightly saner definition. Can you think of anything that might break if we put _trapno inside of the union along with _perf and _addr_lsb?

On sparc si_trapno is only set when SIGILL ILL_TRP is set. So we can limit si_trapno to that combination, and it should not be a problem for a new signal/si_code pair to use that storage. Precisely because it is new.

Similarly on alpha si_trapno is only set for:

SIGFPE {FPE_INTOVF, FPE_INTDIV, FPE_FLTOVF, FPE_FLTDIV, FPE_FLTUND, FPE_FLTINV, FPE_FLTRES, FPE_FLTUNK} and SIGTRAP {TRAP_UNK}.

Placing si_trapno into the union would also make the problem that the union is pointer aligned a non-problem as then the union immediate follows a pointer.

I hadn't had a chance to look before but we must deal with this. The definition of perf_sigtrap in 42dec9a936e7696bea1f27d3c5a0068cd9aa95fd is broken on sparc, alpha, and ia64 as it bypasses the code in kernel/signal.c that ensures the si_trapno or the ia64 special fields are set.

Not to mention that perf_sigtrap appears to abuse si_errno.

The code is only safe if the analysis that says we can move si_trapno and perhaps the ia64 fields into the union is correct. It looks like ia64 much more actively uses it's signal extension fields including for SIGTRAP, so I am not at all certain the generic definition of perf_sigtrap is safe on ia64.

Quoted 4 lines

I suppose in theory sparc64 or alpha might start using the other fields in the future, and an application might be compiled against mismatched headers, but that is unlikely and is already broken with the current headers.

If we localize the use of si_trapno to just a few special cases on alpha and sparc I think we don't even need to worry about breaking userspace on any architecture. It will complicate siginfo_layout, but it is a complication that reflects reality.

I don't have a clue how any of this affects ia64. Does perf work on ia64? Does perf work on sparc, and alpha?

If perf works on ia64 we need to take a hard look at what is going on there as well.

Eric

#5 lore raw
Quoted 1 line · Eric W. Biederman

Arnd Bergmann <[email protected]> writes:

[...]

Quoted 30 lines
Quoted 8 lines
Quoted 7 lines

I did a quick search and the architectures that define __ARCH_SI_TRAPNO are sparc, mips, and alpha. All have 64bit implementations. A further quick search shows that none of those architectures have faults that use BUS_MCEERR_AR, BUS_MCEERR_AO, SEGV_BNDERR, SEGV_PKUERR, nor do they appear to use mm/memory-failure.c

So it doesn't look like we have an ABI regression to fix.

Quoted 7 lines

Even better!

So if sparc is the only user of _trapno and it uses none of the later fields in _sigfault, I wonder if we could take even more liberty at trying to have a slightly saner definition. Can you think of anything that might break if we put _trapno inside of the union along with _perf and _addr_lsb?

On sparc si_trapno is only set when SIGILL ILL_TRP is set. So we can limit si_trapno to that combination, and it should not be a problem for a new signal/si_code pair to use that storage. Precisely because it is new.

Similarly on alpha si_trapno is only set for:

SIGFPE {FPE_INTOVF, FPE_INTDIV, FPE_FLTOVF, FPE_FLTDIV, FPE_FLTUND, FPE_FLTINV, FPE_FLTRES, FPE_FLTUNK} and SIGTRAP {TRAP_UNK}.

Placing si_trapno into the union would also make the problem that the union is pointer aligned a non-problem as then the union immediate follows a pointer.

Quoted 7 lines

I hadn't had a chance to look before but we must deal with this. The definition of perf_sigtrap in 42dec9a936e7696bea1f27d3c5a0068cd9aa95fd is broken on sparc, alpha, and ia64 as it bypasses the code in kernel/signal.c that ensures the si_trapno or the ia64 special fields are set.

Not to mention that perf_sigtrap appears to abuse si_errno.

There are a few other places in the kernel that repurpose si_errno similarly, e.g. arch/arm64/kernel/ptrace.c, kernel/seccomp.c -- it was either that or introduce another field or not have it. It is likely we could do without, but if there are different event types the user would have to sacrifice a few bits of si_perf to encode the event type, and I'd rather keep those bits for something else. Thus the decision fell to use si_errno.

Given it'd be wasted space otherwise, and we define the semantics of whatever is stored in siginfo on the new signal, it'd be good to keep.

Quoted 5 lines

The code is only safe if the analysis that says we can move si_trapno and perhaps the ia64 fields into the union is correct. It looks like ia64 much more actively uses it's signal extension fields including for SIGTRAP, so I am not at all certain the generic definition of perf_sigtrap is safe on ia64.

Trying to understand the requirements of si_trapno myself: safe here would mean that si_trapno is not required if we fire our SIGTRAP / TRAP_PERF.

As far as I can tell that is the case -- see below.

Quoted 10 lines
Quoted 4 lines

I suppose in theory sparc64 or alpha might start using the other fields in the future, and an application might be compiled against mismatched headers, but that is unlikely and is already broken with the current headers.

If we localize the use of si_trapno to just a few special cases on alpha and sparc I think we don't even need to worry about breaking userspace on any architecture. It will complicate siginfo_layout, but it is a complication that reflects reality.

Quoted 5 lines

I don't have a clue how any of this affects ia64. Does perf work on ia64? Does perf work on sparc, and alpha?

If perf works on ia64 we need to take a hard look at what is going on there as well.

No perf on ia64, but it seems alpha and sparc have perf:

	$ git grep 'select.*HAVE_PERF_EVENTS$' -- arch/
	arch/alpha/Kconfig:	select HAVE_PERF_EVENTS    <--
	arch/arc/Kconfig:	select HAVE_PERF_EVENTS
	arch/arm/Kconfig:	select HAVE_PERF_EVENTS
	arch/arm64/Kconfig:	select HAVE_PERF_EVENTS
	arch/csky/Kconfig:	select HAVE_PERF_EVENTS
	arch/hexagon/Kconfig:	select HAVE_PERF_EVENTS
	arch/mips/Kconfig:	select HAVE_PERF_EVENTS
	arch/nds32/Kconfig:	select HAVE_PERF_EVENTS
	arch/parisc/Kconfig:	select HAVE_PERF_EVENTS
	arch/powerpc/Kconfig:	select HAVE_PERF_EVENTS
	arch/riscv/Kconfig:	select HAVE_PERF_EVENTS
	arch/s390/Kconfig:	select HAVE_PERF_EVENTS
	arch/sh/Kconfig:	select HAVE_PERF_EVENTS
	arch/sparc/Kconfig:	select HAVE_PERF_EVENTS    <--
	arch/x86/Kconfig:	select HAVE_PERF_EVENTS
	arch/xtensa/Kconfig:	select HAVE_PERF_EVENTS

Now, given ia64 is not an issue, I wanted to understand the semantics of si_trapno. Per https://man7.org/linux/man-pages/man2/sigaction.2.html, I see:

int si_trapno; /* Trap number that caused
hardware-generated signal (unused on most architectures) */

... its intended semantics seem to suggest it would only be used by some architecture-specific signal like you identified above. So if the semantics is some code of a hardware trap/fault, then we're fine and do not need to set it.

Also bearing in mind we define the semantics any new signal, and given most architectures do not have si_trapno, definitions of new generic signals should probably not include odd architecture specific details related to old architectures.

From all this, my understanding now is that we can move si_trapno into the union, correct? What else did you have in mind?

Thanks,
-- Marco

#6 lore raw
Quoted 41 lines · Marco Elver
Quoted 1 line · Eric W. Biederman

Arnd Bergmann <[email protected]> writes:

[...]

Quoted 30 lines
Quoted 8 lines
Quoted 7 lines

I did a quick search and the architectures that define __ARCH_SI_TRAPNO are sparc, mips, and alpha. All have 64bit implementations. A further quick search shows that none of those architectures have faults that use BUS_MCEERR_AR, BUS_MCEERR_AO, SEGV_BNDERR, SEGV_PKUERR, nor do they appear to use mm/memory-failure.c

So it doesn't look like we have an ABI regression to fix.

Quoted 7 lines

Even better!

So if sparc is the only user of _trapno and it uses none of the later fields in _sigfault, I wonder if we could take even more liberty at trying to have a slightly saner definition. Can you think of anything that might break if we put _trapno inside of the union along with _perf and _addr_lsb?

On sparc si_trapno is only set when SIGILL ILL_TRP is set. So we can limit si_trapno to that combination, and it should not be a problem for a new signal/si_code pair to use that storage. Precisely because it is new.

Similarly on alpha si_trapno is only set for:

SIGFPE {FPE_INTOVF, FPE_INTDIV, FPE_FLTOVF, FPE_FLTDIV, FPE_FLTUND, FPE_FLTINV, FPE_FLTRES, FPE_FLTUNK} and SIGTRAP {TRAP_UNK}.

Placing si_trapno into the union would also make the problem that the union is pointer aligned a non-problem as then the union immediate follows a pointer.

Quoted 7 lines

I hadn't had a chance to look before but we must deal with this. The definition of perf_sigtrap in 42dec9a936e7696bea1f27d3c5a0068cd9aa95fd is broken on sparc, alpha, and ia64 as it bypasses the code in kernel/signal.c that ensures the si_trapno or the ia64 special fields are set.

Not to mention that perf_sigtrap appears to abuse si_errno.

Quoted 7 lines

There are a few other places in the kernel that repurpose si_errno similarly, e.g. arch/arm64/kernel/ptrace.c, kernel/seccomp.c -- it was either that or introduce another field or not have it. It is likely we could do without, but if there are different event types the user would have to sacrifice a few bits of si_perf to encode the event type, and I'd rather keep those bits for something else. Thus the decision fell to use si_errno.

arm64 only abuses si_errno in compat code for bug compatibility with arm32.

Quoted 2 lines

Given it'd be wasted space otherwise, and we define the semantics of whatever is stored in siginfo on the new signal, it'd be good to keep.

Except you don't completely. You are not defining a new signal. You are extending the definition of SIGTRAP. Anything generic that responds to all SIGTRAPs can reasonably be looking at si_errno.

Further you are already adding a field with si_perf you can just as easily add a second field with well defined semantics for that data.

Quoted 61 lines
Quoted 5 lines

The code is only safe if the analysis that says we can move si_trapno and perhaps the ia64 fields into the union is correct. It looks like ia64 much more actively uses it's signal extension fields including for SIGTRAP, so I am not at all certain the generic definition of perf_sigtrap is safe on ia64.

Trying to understand the requirements of si_trapno myself: safe here would mean that si_trapno is not required if we fire our SIGTRAP / TRAP_PERF.

As far as I can tell that is the case -- see below.

Quoted 10 lines
Quoted 4 lines

I suppose in theory sparc64 or alpha might start using the other fields in the future, and an application might be compiled against mismatched headers, but that is unlikely and is already broken with the current headers.

If we localize the use of si_trapno to just a few special cases on alpha and sparc I think we don't even need to worry about breaking userspace on any architecture. It will complicate siginfo_layout, but it is a complication that reflects reality.

Quoted 5 lines

I don't have a clue how any of this affects ia64. Does perf work on ia64? Does perf work on sparc, and alpha?

If perf works on ia64 we need to take a hard look at what is going on there as well.

No perf on ia64, but it seems alpha and sparc have perf:

	$ git grep 'select.*HAVE_PERF_EVENTS$' -- arch/
	arch/alpha/Kconfig:	select HAVE_PERF_EVENTS    <--
	arch/arc/Kconfig:	select HAVE_PERF_EVENTS
	arch/arm/Kconfig:	select HAVE_PERF_EVENTS
	arch/arm64/Kconfig:	select HAVE_PERF_EVENTS
	arch/csky/Kconfig:	select HAVE_PERF_EVENTS
	arch/hexagon/Kconfig:	select HAVE_PERF_EVENTS
	arch/mips/Kconfig:	select HAVE_PERF_EVENTS
	arch/nds32/Kconfig:	select HAVE_PERF_EVENTS
	arch/parisc/Kconfig:	select HAVE_PERF_EVENTS
	arch/powerpc/Kconfig:	select HAVE_PERF_EVENTS
	arch/riscv/Kconfig:	select HAVE_PERF_EVENTS
	arch/s390/Kconfig:	select HAVE_PERF_EVENTS
	arch/sh/Kconfig:	select HAVE_PERF_EVENTS
	arch/sparc/Kconfig:	select HAVE_PERF_EVENTS    <--
	arch/x86/Kconfig:	select HAVE_PERF_EVENTS
	arch/xtensa/Kconfig:	select HAVE_PERF_EVENTS

Now, given ia64 is not an issue, I wanted to understand the semantics of si_trapno. Per https://man7.org/linux/man-pages/man2/sigaction.2.html, I see:

int si_trapno; /* Trap number that caused
hardware-generated signal (unused on most architectures) */

... its intended semantics seem to suggest it would only be used by some architecture-specific signal like you identified above. So if the semantics is some code of a hardware trap/fault, then we're fine and do not need to set it.

Quoted 7 lines

Also bearing in mind we define the semantics any new signal, and given most architectures do not have si_trapno, definitions of new generic signals should probably not include odd architecture specific details related to old architectures.

From all this, my understanding now is that we can move si_trapno into the union, correct? What else did you have in mind?

Yes. Let's move si_trapno into the union.

That implies a few things like siginfo_layout needs to change.

The helpers in kernel/signal.c can change to not imply that if you define __ARCH_SI_TRAPNO you must always define and pass in si_trapno. A force_sig_trapno could be defined instead to handle the cases that alpha and sparc use si_trapno.

It would be nice if a force_sig_perf_trap could be factored out of perf_trap and placed in kernel/signal.c.

My experience (especially this round) is that it becomes much easier to audit the users of siginfo if there is a dedicated function in kernel/signal.c that is simply passed the parameters that need to be placed in siginfo.

So I would very much like to see if I can make force_sig_info static.

Eric

#7 lore raw

On Fri, 30 Apr 2021 at 22:15, Eric W. Biederman <[email protected]> wrote:
[...]

Quoted 9 lines

arm64 only abuses si_errno in compat code for bug compatibility with arm32.

Quoted 2 lines

Given it'd be wasted space otherwise, and we define the semantics of whatever is stored in siginfo on the new signal, it'd be good to keep.

Except you don't completely. You are not defining a new signal. You are extending the definition of SIGTRAP. Anything generic that responds to all SIGTRAPs can reasonably be looking at si_errno.

I see where you're coming from, and agree with this if si_errno already had some semantics for some subset of SIGTRAPs. I've tried to analyze the situation a bit further, since siginfo seems to be a giant minefield and semantics is underspecified at best. :-)

Do any of the existing SIGTRAPs define si_errno to be set? As far as I can tell, they don't.

If this is true, I think there are benefits and downsides to repurposing si_errno (similar to what SIGSYS did). The obvious downside is as you suggest, it's not always a real errno. The benefit is that we avoid introducing more and more fields -- i.e. if we permit si_errno to be repurposed for SIGTRAP and its value depends on the precise si_code, too, we simplify siginfo's overall definition (also given every new field needs more code in kernel/signal.c, too).

Given SIGTRAPs are in response to some user-selected event in the user's code (breakpoints, ptrace, etc. ... now perf events), the user must already check the si_code to select the right action because SIGTRAPs are not alike (unlike, e.g. SIGSEGV). Because of this, I think that repurposing si_errno in an si_code-dependent way for SIGTRAPs is safe.

If you think it is simply untenable to repurpose si_errno for SIGTRAPs, please confirm -- I'll just send a minimal patch to fix (I'd probably just remove setting it... everything else is too intrusive as a "fix".. sigh).

The cleanups as you outline below seem orthogonal and not urgent for 5.13 (all changes and cleanups for __ARCH_SI_TRAPNO seem too intrusive without -next exposure).

Thanks,
-- Marco

Quoted 93 lines

Further you are already adding a field with si_perf you can just as easily add a second field with well defined semantics for that data.

Quoted 61 lines
Quoted 5 lines

The code is only safe if the analysis that says we can move si_trapno and perhaps the ia64 fields into the union is correct. It looks like ia64 much more actively uses it's signal extension fields including for SIGTRAP, so I am not at all certain the generic definition of perf_sigtrap is safe on ia64.

Trying to understand the requirements of si_trapno myself: safe here would mean that si_trapno is not required if we fire our SIGTRAP / TRAP_PERF.

As far as I can tell that is the case -- see below.

Quoted 10 lines
Quoted 4 lines

I suppose in theory sparc64 or alpha might start using the other fields in the future, and an application might be compiled against mismatched headers, but that is unlikely and is already broken with the current headers.

If we localize the use of si_trapno to just a few special cases on alpha and sparc I think we don't even need to worry about breaking userspace on any architecture. It will complicate siginfo_layout, but it is a complication that reflects reality.

Quoted 5 lines

I don't have a clue how any of this affects ia64. Does perf work on ia64? Does perf work on sparc, and alpha?

If perf works on ia64 we need to take a hard look at what is going on there as well.

No perf on ia64, but it seems alpha and sparc have perf:

      $ git grep 'select.*HAVE_PERF_EVENTS$' -- arch/
      arch/alpha/Kconfig:     select HAVE_PERF_EVENTS    <--
      arch/arc/Kconfig:       select HAVE_PERF_EVENTS
      arch/arm/Kconfig:       select HAVE_PERF_EVENTS
      arch/arm64/Kconfig:     select HAVE_PERF_EVENTS
      arch/csky/Kconfig:      select HAVE_PERF_EVENTS
      arch/hexagon/Kconfig:   select HAVE_PERF_EVENTS
      arch/mips/Kconfig:      select HAVE_PERF_EVENTS
      arch/nds32/Kconfig:     select HAVE_PERF_EVENTS
      arch/parisc/Kconfig:    select HAVE_PERF_EVENTS
      arch/powerpc/Kconfig:   select HAVE_PERF_EVENTS
      arch/riscv/Kconfig:     select HAVE_PERF_EVENTS
      arch/s390/Kconfig:      select HAVE_PERF_EVENTS
      arch/sh/Kconfig:        select HAVE_PERF_EVENTS
      arch/sparc/Kconfig:     select HAVE_PERF_EVENTS    <--
      arch/x86/Kconfig:       select HAVE_PERF_EVENTS
      arch/xtensa/Kconfig:    select HAVE_PERF_EVENTS

Now, given ia64 is not an issue, I wanted to understand the semantics of si_trapno. Per https://man7.org/linux/man-pages/man2/sigaction.2.html, I see:

int si_trapno; /* Trap number that caused
hardware-generated signal (unused on most architectures) */

... its intended semantics seem to suggest it would only be used by some architecture-specific signal like you identified above. So if the semantics is some code of a hardware trap/fault, then we're fine and do not need to set it.

Quoted 7 lines

Also bearing in mind we define the semantics any new signal, and given most architectures do not have si_trapno, definitions of new generic signals should probably not include odd architecture specific details related to old architectures.

From all this, my understanding now is that we can move si_trapno into the union, correct? What else did you have in mind?

Yes. Let's move si_trapno into the union.

That implies a few things like siginfo_layout needs to change.

The helpers in kernel/signal.c can change to not imply that if you define __ARCH_SI_TRAPNO you must always define and pass in si_trapno. A force_sig_trapno could be defined instead to handle the cases that alpha and sparc use si_trapno.

It would be nice if a force_sig_perf_trap could be factored out of perf_trap and placed in kernel/signal.c.

My experience (especially this round) is that it becomes much easier to audit the users of siginfo if there is a dedicated function in kernel/signal.c that is simply passed the parameters that need to be placed in siginfo.

So I would very much like to see if I can make force_sig_info static.

Eric

#8 lore raw

cover signal: Move si_trapno into the _si_fault union

In reply to Marco Elver

Eric W. Biederman (3):
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Use dedicated helpers to send signals with si_trapno set

 arch/alpha/kernel/osf_sys.c        |  2 +-
 arch/alpha/kernel/signal.c         |  4 +-
 arch/alpha/kernel/traps.c          | 24 ++++++------
 arch/alpha/mm/fault.c              |  4 +-
 arch/sparc/kernel/process_64.c     |  2 +-
 arch/sparc/kernel/sys_sparc_32.c   |  2 +-
 arch/sparc/kernel/sys_sparc_64.c   |  2 +-
 arch/sparc/kernel/traps_32.c       | 22 +++++------
 arch/sparc/kernel/traps_64.c       | 44 ++++++++++------------
 arch/sparc/kernel/unaligned_32.c   |  2 +-
 arch/sparc/mm/fault_32.c           |  2 +-
 arch/sparc/mm/fault_64.c           |  2 +-
 fs/signalfd.c                      |  7 +---
 include/linux/compat.h             |  4 +-
 include/linux/sched/signal.h       | 12 ++----
 include/linux/signal.h             |  1 +
 include/uapi/asm-generic/siginfo.h |  6 +--
 kernel/signal.c                    | 77 ++++++++++++++++++++++----------------
 18 files changed, 107 insertions(+), 112 deletions(-)
#9 lore raw

PATCH 1/3 siginfo: Move si_trapno inside the union inside _si_fault

It turns out that linux uses si_trapno very sparingly, and as such it can be considered extra information for a very narrow selection of signals, rather than information that is present with every fault reported in siginfo.

As such move si_trapno inside the union inside of _si_fault. This results in no change in placement, and makes it eaiser to extend _si_fault in the future as this reduces the number of special cases. In particular with si_trapno included in the union it is no longer a concern that the union must be pointer alligned on most architectures because the union followes immediately after si_addr which is a pointer.

---
 include/linux/compat.h             | 4 +---
 include/uapi/asm-generic/siginfo.h | 6 +-----
 2 files changed, 2 insertions(+), 8 deletions(-)
include/linux/compat.h +1 -3
index f0d2dd35d408..24462ed63af4 100644
@@ -214,12 +214,10 @@ typedef struct compat_siginfo {
 		/* SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT */
 		struct {
 			compat_uptr_t _addr;	/* faulting insn/memory ref. */
-#ifdef __ARCH_SI_TRAPNO
-			int _trapno;	/* TRAP # which caused the signal */
-#endif
 #define __COMPAT_ADDR_BND_PKEY_PAD  (__alignof__(compat_uptr_t) < sizeof(short) ? \
 				     sizeof(short) : __alignof__(compat_uptr_t))
 			union {
+				int _trapno;	/* TRAP # which caused the signal */
 				/*
 				 * used when si_code=BUS_MCEERR_AR or
 				 * used when si_code=BUS_MCEERR_AO
include/uapi/asm-generic/siginfo.h +1 -5
index 03d6f6d2c1fe..2abdf1d19aad 100644
@@ -63,9 +63,6 @@ union __sifields {
 	/* SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT */
 	struct {
 		void __user *_addr; /* faulting insn/memory ref. */
-#ifdef __ARCH_SI_TRAPNO
-		int _trapno;	/* TRAP # which caused the signal */
-#endif
 #ifdef __ia64__
 		int _imm;		/* immediate value for "break" */
 		unsigned int _flags;	/* see ia64 si_flags */
@@ -75,6 +72,7 @@ union __sifields {
 #define __ADDR_BND_PKEY_PAD  (__alignof__(void *) < sizeof(short) ? \
 			      sizeof(short) : __alignof__(void *))
 		union {
+			int _trapno;	/* TRAP # which caused the signal */
 			/*
 			 * used when si_code=BUS_MCEERR_AR or
 			 * used when si_code=BUS_MCEERR_AO
@@ -150,9 +148,7 @@ typedef struct siginfo {
 #define si_int		_sifields._rt._sigval.sival_int
 #define si_ptr		_sifields._rt._sigval.sival_ptr
 #define si_addr		_sifields._sigfault._addr
-#ifdef __ARCH_SI_TRAPNO
 #define si_trapno	_sifields._sigfault._trapno
-#endif
 #define si_addr_lsb	_sifields._sigfault._addr_lsb
 #define si_lower	_sifields._sigfault._addr_bnd._lower
 #define si_upper	_sifields._sigfault._addr_bnd._upper
Signature · 1 line
2.30.1
author Acked-by on v1 1/3
#10 lore raw

Re: [PATCH 1/3] siginfo: Move si_trapno inside the union inside _si_fault

Quoted 6 lines · Eric W. Biederman

It turns out that linux uses si_trapno very sparingly, and as such it can be considered extra information for a very narrow selection of signals, rather than information that is present with every fault reported in siginfo.

Quoted 8 lines

As such move si_trapno inside the union inside of _si_fault. This results in no change in placement, and makes it eaiser to extend _si_fault in the future as this reduces the number of special cases. In particular with si_trapno included in the union it is no longer a concern that the union must be pointer alligned on most architectures because the union followes immediately after si_addr which is a pointer.

Maybe add "Link:
https://lkml.kernel.org/r/CAK8P3a0+uKYwL1NhY6Hvtieghba2hKYGD6hcKx5n8=4Gtt+pHA@mail.gmail.com"

Quoted 1 line

By no longer guarding it with __ARCH_SI_TRAPNO we run the risk that it will be used by something else at some point. Is that intentional?

Thanks,
-- Marco

Quoted 57 lines
---
 include/linux/compat.h             | 4 +---
 include/uapi/asm-generic/siginfo.h | 6 +-----
 2 files changed, 2 insertions(+), 8 deletions(-)
include/linux/compat.h +1 -3
index f0d2dd35d408..24462ed63af4 100644
@@ -214,12 +214,10 @@ typedef struct compat_siginfo {
                /* SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT */
                struct {
                        compat_uptr_t _addr;    /* faulting insn/memory ref. */
-#ifdef __ARCH_SI_TRAPNO
-                       int _trapno;    /* TRAP # which caused the signal */
-#endif
 #define __COMPAT_ADDR_BND_PKEY_PAD  (__alignof__(compat_uptr_t) < sizeof(short) ? \
                                     sizeof(short) : __alignof__(compat_uptr_t))
                        union {
+                               int _trapno;    /* TRAP # which caused the signal */
                                /*
                                 * used when si_code=BUS_MCEERR_AR or
                                 * used when si_code=BUS_MCEERR_AO
include/uapi/asm-generic/siginfo.h +1 -6
index 03d6f6d2c1fe..2abdf1d19aad 100644
@@ -63,9 +63,6 @@ union __sifields {
        /* SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT */
        struct {
                void __user *_addr; /* faulting insn/memory ref. */
-#ifdef __ARCH_SI_TRAPNO
-               int _trapno;    /* TRAP # which caused the signal */
-#endif
 #ifdef __ia64__
                int _imm;               /* immediate value for "break" */
                unsigned int _flags;    /* see ia64 si_flags */
@@ -75,6 +72,7 @@ union __sifields {
 #define __ADDR_BND_PKEY_PAD  (__alignof__(void *) < sizeof(short) ? \
                              sizeof(short) : __alignof__(void *))
                union {
+                       int _trapno;    /* TRAP # which caused the signal */
                        /*
                         * used when si_code=BUS_MCEERR_AR or
                         * used when si_code=BUS_MCEERR_AO
@@ -150,9 +148,7 @@ typedef struct siginfo {
 #define si_int         _sifields._rt._sigval.sival_int
 #define si_ptr         _sifields._rt._sigval.sival_ptr
 #define si_addr                _sifields._sigfault._addr
-#ifdef __ARCH_SI_TRAPNO
 #define si_trapno      _sifields._sigfault._trapno
-#endif
 #define si_addr_lsb    _sifields._sigfault._addr_lsb
 #define si_lower       _sifields._sigfault._addr_bnd._lower
 #define si_upper       _sifields._sigfault._addr_bnd._upper
--

2.30.1

#11 lore raw

Re: [PATCH 1/3] siginfo: Move si_trapno inside the union inside _si_fault

Quoted 19 lines · Marco Elver
Quoted 6 lines · Eric W. Biederman

It turns out that linux uses si_trapno very sparingly, and as such it can be considered extra information for a very narrow selection of signals, rather than information that is present with every fault reported in siginfo.

Quoted 8 lines

As such move si_trapno inside the union inside of _si_fault. This results in no change in placement, and makes it eaiser to extend _si_fault in the future as this reduces the number of special cases. In particular with si_trapno included in the union it is no longer a concern that the union must be pointer alligned on most architectures because the union followes immediately after si_addr which is a pointer.

Maybe add "Link:
https://lkml.kernel.org/r/CAK8P3a0+uKYwL1NhY6Hvtieghba2hKYGD6hcKx5n8=4Gtt+pHA@mail.gmail.com"

Quoted 6 lines
Quoted 1 line

By no longer guarding it with __ARCH_SI_TRAPNO we run the risk that it will be used by something else at some point. Is that intentional?

The motivation was letting the code be tested on other architectures.

But once si_trapno falls inside the union instead of being present for every signal reporting a fault it doesn't really matter.

I think it would be poor taste but harmless to use si_trapno, mostly because defining a new entry in the union could be more specific and well defined.

Eric

#12 lore raw

PATCH 2/3 signal: Implement SIL_FAULT_TRAPNO

In reply to Eric W. Biederman

Now that si_trapno is part of the union in _si_fault and available on all architectures, add SIL_FAULT_TRAPNO and update siginfo_layout to return SIL_FAULT_TRAPNO when si_trapno is actually used.

Update the code that uses siginfo_layout to deal with SIL_FAULT_TRAPNO and have the same code ignore si_trapno in in all other cases.

---
 fs/signalfd.c          |  7 ++-----
 include/linux/signal.h |  1 +
 kernel/signal.c        | 36 ++++++++++++++----------------------
 3 files changed, 17 insertions(+), 27 deletions(-)
fs/signalfd.c +2 -5
index 040a1142915f..126c681a30e7 100644
@@ -123,15 +123,12 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		 */
 	case SIL_FAULT:
 		new.ssi_addr = (long) kinfo->si_addr;
-#ifdef __ARCH_SI_TRAPNO
+	case SIL_FAULT_TRAPNO:
+		new.ssi_addr = (long) kinfo->si_addr;
 		new.ssi_trapno = kinfo->si_trapno;
-#endif
 		break;
 	case SIL_FAULT_MCEERR:
 		new.ssi_addr = (long) kinfo->si_addr;
-#ifdef __ARCH_SI_TRAPNO
-		new.ssi_trapno = kinfo->si_trapno;
-#endif
 		new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
 		break;
 	case SIL_PERF_EVENT:
include/linux/signal.h +1 -0
index 1e98548d7cf6..5160fd45e5ca 100644
@@ -40,6 +40,7 @@ enum siginfo_layout {
 	SIL_TIMER,
 	SIL_POLL,
 	SIL_FAULT,
+	SIL_FAULT_TRAPNO,
 	SIL_FAULT_MCEERR,
 	SIL_FAULT_BNDERR,
 	SIL_FAULT_PKUERR,
kernel/signal.c +14 -22
index c3017aa8024a..7b2d61cb7411 100644
@@ -1194,6 +1194,7 @@ static inline bool has_si_pid_and_uid(struct kernel_siginfo *info)
 	case SIL_TIMER:
 	case SIL_POLL:
 	case SIL_FAULT:
+	case SIL_FAULT_TRAPNO:
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
@@ -2527,6 +2528,7 @@ static void hide_si_addr_tag_bits(struct ksignal *ksig)
 {
 	switch (siginfo_layout(ksig->sig, ksig->info.si_code)) {
 	case SIL_FAULT:
+	case SIL_FAULT_TRAPNO:
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
@@ -3206,6 +3208,12 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
 			if ((sig == SIGBUS) &&
 			    (si_code >= BUS_MCEERR_AR) && (si_code <= BUS_MCEERR_AO))
 				layout = SIL_FAULT_MCEERR;
+			else if (IS_ENABLED(ALPHA) &&
+				 ((sig == SIGFPE) ||
+				  ((sig == SIGTRAP) && (si_code == TRAP_UNK))))
+				layout = SIL_FAULT_TRAPNO;
+			else if (IS_ENABLED(SPARC) && (sig == SIGILL) && (si_code == ILL_ILLTRP))
+				layout = SIL_FAULT_TRAPNO;
 			else if ((sig == SIGSEGV) && (si_code == SEGV_BNDERR))
 				layout = SIL_FAULT_BNDERR;
 #ifdef SEGV_PKUERR
@@ -3317,30 +3325,22 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 		break;
 	case SIL_FAULT:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+		break;
+	case SIL_FAULT_TRAPNO:
+		to->si_addr = ptr_to_compat(from->si_addr);
 		to->si_trapno = from->si_trapno;
-#endif
 		break;
 	case SIL_FAULT_MCEERR:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_addr_lsb = from->si_addr_lsb;
 		break;
 	case SIL_FAULT_BNDERR:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_lower = ptr_to_compat(from->si_lower);
 		to->si_upper = ptr_to_compat(from->si_upper);
 		break;
 	case SIL_FAULT_PKUERR:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_pkey = from->si_pkey;
 		break;
 	case SIL_PERF_EVENT:
@@ -3401,30 +3401,22 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 		break;
 	case SIL_FAULT:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+		break;
+	case SIL_FAULT_TRAPNO:
+		to->si_addr = compat_ptr(from->si_addr);
 		to->si_trapno = from->si_trapno;
-#endif
 		break;
 	case SIL_FAULT_MCEERR:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_addr_lsb = from->si_addr_lsb;
 		break;
 	case SIL_FAULT_BNDERR:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_lower = compat_ptr(from->si_lower);
 		to->si_upper = compat_ptr(from->si_upper);
 		break;
 	case SIL_FAULT_PKUERR:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_pkey = from->si_pkey;
 		break;
 	case SIL_PERF_EVENT:
Signature · 1 line
2.30.1
#13 lore raw

Re: [PATCH 2/3] signal: Implement SIL_FAULT_TRAPNO

Quoted 6 lines · [email protected] (Eric W. Biederman) writes:

Now that si_trapno is part of the union in _si_fault and available on all architectures, add SIL_FAULT_TRAPNO and update siginfo_layout to return SIL_FAULT_TRAPNO when si_trapno is actually used.

Update the code that uses siginfo_layout to deal with SIL_FAULT_TRAPNO and have the same code ignore si_trapno in in all other cases.

This change is missing a break in signalfd.

Eric

Quoted 12 lines
---
 fs/signalfd.c          |  7 ++-----
 include/linux/signal.h |  1 +
 kernel/signal.c        | 36 ++++++++++++++----------------------
 3 files changed, 17 insertions(+), 27 deletions(-)
fs/signalfd.c +0 -0
index 040a1142915f..126c681a30e7 100644

fs/signalfd.c static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo, · line 125

Quoted 3 lines
fs/signalfd.c +0 -0
@@ -123,15 +123,12 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		 */
 	case SIL_FAULT:
 		new.ssi_addr = (long) kinfo->si_addr;

+ break;

Quoted 127 lines
fs/signalfd.c +2 -5
-#ifdef __ARCH_SI_TRAPNO
+	case SIL_FAULT_TRAPNO:
+		new.ssi_addr = (long) kinfo->si_addr;
 		new.ssi_trapno = kinfo->si_trapno;
-#endif
 		break;
 	case SIL_FAULT_MCEERR:
 		new.ssi_addr = (long) kinfo->si_addr;
-#ifdef __ARCH_SI_TRAPNO
-		new.ssi_trapno = kinfo->si_trapno;
-#endif
 		new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
 		break;
 	case SIL_PERF_EVENT:
include/linux/signal.h +1 -0
index 1e98548d7cf6..5160fd45e5ca 100644
@@ -40,6 +40,7 @@ enum siginfo_layout {
 	SIL_TIMER,
 	SIL_POLL,
 	SIL_FAULT,
+	SIL_FAULT_TRAPNO,
 	SIL_FAULT_MCEERR,
 	SIL_FAULT_BNDERR,
 	SIL_FAULT_PKUERR,
kernel/signal.c +14 -22
index c3017aa8024a..7b2d61cb7411 100644
@@ -1194,6 +1194,7 @@ static inline bool has_si_pid_and_uid(struct kernel_siginfo *info)
 	case SIL_TIMER:
 	case SIL_POLL:
 	case SIL_FAULT:
+	case SIL_FAULT_TRAPNO:
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
@@ -2527,6 +2528,7 @@ static void hide_si_addr_tag_bits(struct ksignal *ksig)
 {
 	switch (siginfo_layout(ksig->sig, ksig->info.si_code)) {
 	case SIL_FAULT:
+	case SIL_FAULT_TRAPNO:
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
@@ -3206,6 +3208,12 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
 			if ((sig == SIGBUS) &&
 			    (si_code >= BUS_MCEERR_AR) && (si_code <= BUS_MCEERR_AO))
 				layout = SIL_FAULT_MCEERR;
+			else if (IS_ENABLED(ALPHA) &&
+				 ((sig == SIGFPE) ||
+				  ((sig == SIGTRAP) && (si_code == TRAP_UNK))))
+				layout = SIL_FAULT_TRAPNO;
+			else if (IS_ENABLED(SPARC) && (sig == SIGILL) && (si_code == ILL_ILLTRP))
+				layout = SIL_FAULT_TRAPNO;
 			else if ((sig == SIGSEGV) && (si_code == SEGV_BNDERR))
 				layout = SIL_FAULT_BNDERR;
 #ifdef SEGV_PKUERR
@@ -3317,30 +3325,22 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 		break;
 	case SIL_FAULT:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+		break;
+	case SIL_FAULT_TRAPNO:
+		to->si_addr = ptr_to_compat(from->si_addr);
 		to->si_trapno = from->si_trapno;
-#endif
 		break;
 	case SIL_FAULT_MCEERR:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_addr_lsb = from->si_addr_lsb;
 		break;
 	case SIL_FAULT_BNDERR:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_lower = ptr_to_compat(from->si_lower);
 		to->si_upper = ptr_to_compat(from->si_upper);
 		break;
 	case SIL_FAULT_PKUERR:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_pkey = from->si_pkey;
 		break;
 	case SIL_PERF_EVENT:
@@ -3401,30 +3401,22 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 		break;
 	case SIL_FAULT:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+		break;
+	case SIL_FAULT_TRAPNO:
+		to->si_addr = compat_ptr(from->si_addr);
 		to->si_trapno = from->si_trapno;
-#endif
 		break;
 	case SIL_FAULT_MCEERR:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_addr_lsb = from->si_addr_lsb;
 		break;
 	case SIL_FAULT_BNDERR:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_lower = compat_ptr(from->si_lower);
 		to->si_upper = compat_ptr(from->si_upper);
 		break;
 	case SIL_FAULT_PKUERR:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_pkey = from->si_pkey;
 		break;
 	case SIL_PERF_EVENT:
#14 lore raw

Re: [PATCH 2/3] signal: Implement SIL_FAULT_TRAPNO

In reply to Eric W. Biederman
Quoted 70 lines · Eric W. Biederman

Now that si_trapno is part of the union in _si_fault and available on all architectures, add SIL_FAULT_TRAPNO and update siginfo_layout to return SIL_FAULT_TRAPNO when si_trapno is actually used.

Update the code that uses siginfo_layout to deal with SIL_FAULT_TRAPNO and have the same code ignore si_trapno in in all other cases.

---
 fs/signalfd.c          |  7 ++-----
 include/linux/signal.h |  1 +
 kernel/signal.c        | 36 ++++++++++++++----------------------
 3 files changed, 17 insertions(+), 27 deletions(-)
fs/signalfd.c +2 -5
index 040a1142915f..126c681a30e7 100644
@@ -123,15 +123,12 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
                 */
        case SIL_FAULT:
                new.ssi_addr = (long) kinfo->si_addr;
-#ifdef __ARCH_SI_TRAPNO
+       case SIL_FAULT_TRAPNO:
+               new.ssi_addr = (long) kinfo->si_addr;
                new.ssi_trapno = kinfo->si_trapno;
-#endif
                break;
        case SIL_FAULT_MCEERR:
                new.ssi_addr = (long) kinfo->si_addr;
-#ifdef __ARCH_SI_TRAPNO
-               new.ssi_trapno = kinfo->si_trapno;
-#endif
                new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
                break;
        case SIL_PERF_EVENT:
include/linux/signal.h +1 -0
index 1e98548d7cf6..5160fd45e5ca 100644
@@ -40,6 +40,7 @@ enum siginfo_layout {
        SIL_TIMER,
        SIL_POLL,
        SIL_FAULT,
+       SIL_FAULT_TRAPNO,
        SIL_FAULT_MCEERR,
        SIL_FAULT_BNDERR,
        SIL_FAULT_PKUERR,
kernel/signal.c +2 -0
index c3017aa8024a..7b2d61cb7411 100644
@@ -1194,6 +1194,7 @@ static inline bool has_si_pid_and_uid(struct kernel_siginfo *info)
        case SIL_TIMER:
        case SIL_POLL:
        case SIL_FAULT:
+       case SIL_FAULT_TRAPNO:
        case SIL_FAULT_MCEERR:
        case SIL_FAULT_BNDERR:
        case SIL_FAULT_PKUERR:
@@ -2527,6 +2528,7 @@ static void hide_si_addr_tag_bits(struct ksignal *ksig)
 {
        switch (siginfo_layout(ksig->sig, ksig->info.si_code)) {
        case SIL_FAULT:
+       case SIL_FAULT_TRAPNO:
        case SIL_FAULT_MCEERR:
        case SIL_FAULT_BNDERR:
        case SIL_FAULT_PKUERR:

kernel/signal.c enum siginfo_layout siginfo_layout(unsigned sig, int si_code) · line 3216

Quoted 9 lines
kernel/signal.c +6 -0
@@ -3206,6 +3208,12 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
                        if ((sig == SIGBUS) &&
                            (si_code >= BUS_MCEERR_AR) && (si_code <= BUS_MCEERR_AO))
                                layout = SIL_FAULT_MCEERR;
+                       else if (IS_ENABLED(ALPHA) &&
+                                ((sig == SIGFPE) ||
+                                 ((sig == SIGTRAP) && (si_code == TRAP_UNK))))
+                               layout = SIL_FAULT_TRAPNO;
+                       else if (IS_ENABLED(SPARC) && (sig == SIGILL) && (si_code == ILL_ILLTRP))
+                               layout = SIL_FAULT_TRAPNO;

The breakage isn't apparent here, but in later patches. These need to become CONFIG_SPARC and CONFIG_ALPHA.

Quoted 73 lines
kernel/signal.c +6 -23
                        else if ((sig == SIGSEGV) && (si_code == SEGV_BNDERR))
                                layout = SIL_FAULT_BNDERR;
 #ifdef SEGV_PKUERR
@@ -3317,30 +3325,22 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
                break;
        case SIL_FAULT:
                to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+               break;
+       case SIL_FAULT_TRAPNO:
+               to->si_addr = ptr_to_compat(from->si_addr);
                to->si_trapno = from->si_trapno;
-#endif
                break;
        case SIL_FAULT_MCEERR:
                to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_addr_lsb = from->si_addr_lsb;
                break;
        case SIL_FAULT_BNDERR:
                to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_lower = ptr_to_compat(from->si_lower);
                to->si_upper = ptr_to_compat(from->si_upper);
                break;
        case SIL_FAULT_PKUERR:
                to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_pkey = from->si_pkey;
                break;
        case SIL_PERF_EVENT:
@@ -3401,30 +3401,22 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
                break;
        case SIL_FAULT:
                to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+               break;
+       case SIL_FAULT_TRAPNO:
+               to->si_addr = compat_ptr(from->si_addr);
                to->si_trapno = from->si_trapno;
-#endif
                break;
        case SIL_FAULT_MCEERR:
                to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_addr_lsb = from->si_addr_lsb;
                break;
        case SIL_FAULT_BNDERR:
                to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_lower = compat_ptr(from->si_lower);
                to->si_upper = compat_ptr(from->si_upper);
                break;
        case SIL_FAULT_PKUERR:
                to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_pkey = from->si_pkey;
                break;
        case SIL_PERF_EVENT:
--

2.30.1

#15 lore raw

Re: [PATCH 2/3] signal: Implement SIL_FAULT_TRAPNO

Quoted 75 lines · Marco Elver
Quoted 74 lines · Eric W. Biederman

Now that si_trapno is part of the union in _si_fault and available on all architectures, add SIL_FAULT_TRAPNO and update siginfo_layout to return SIL_FAULT_TRAPNO when si_trapno is actually used.

Update the code that uses siginfo_layout to deal with SIL_FAULT_TRAPNO and have the same code ignore si_trapno in in all other cases.

---
 fs/signalfd.c          |  7 ++-----
 include/linux/signal.h |  1 +
 kernel/signal.c        | 36 ++++++++++++++----------------------
 3 files changed, 17 insertions(+), 27 deletions(-)
fs/signalfd.c +2 -5
index 040a1142915f..126c681a30e7 100644
@@ -123,15 +123,12 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
                 */
        case SIL_FAULT:
                new.ssi_addr = (long) kinfo->si_addr;
-#ifdef __ARCH_SI_TRAPNO
+       case SIL_FAULT_TRAPNO:
+               new.ssi_addr = (long) kinfo->si_addr;
                new.ssi_trapno = kinfo->si_trapno;
-#endif
                break;
        case SIL_FAULT_MCEERR:
                new.ssi_addr = (long) kinfo->si_addr;
-#ifdef __ARCH_SI_TRAPNO
-               new.ssi_trapno = kinfo->si_trapno;
-#endif
                new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
                break;
        case SIL_PERF_EVENT:
include/linux/signal.h +1 -0
index 1e98548d7cf6..5160fd45e5ca 100644
@@ -40,6 +40,7 @@ enum siginfo_layout {
        SIL_TIMER,
        SIL_POLL,
        SIL_FAULT,
+       SIL_FAULT_TRAPNO,
        SIL_FAULT_MCEERR,
        SIL_FAULT_BNDERR,
        SIL_FAULT_PKUERR,
kernel/signal.c +3 -0
index c3017aa8024a..7b2d61cb7411 100644
@@ -1194,6 +1194,7 @@ static inline bool has_si_pid_and_uid(struct kernel_siginfo *info)
        case SIL_TIMER:
        case SIL_POLL:
        case SIL_FAULT:
+       case SIL_FAULT_TRAPNO:
        case SIL_FAULT_MCEERR:
        case SIL_FAULT_BNDERR:
        case SIL_FAULT_PKUERR:
@@ -2527,6 +2528,7 @@ static void hide_si_addr_tag_bits(struct ksignal *ksig)
 {
        switch (siginfo_layout(ksig->sig, ksig->info.si_code)) {
        case SIL_FAULT:
+       case SIL_FAULT_TRAPNO:
        case SIL_FAULT_MCEERR:
        case SIL_FAULT_BNDERR:
        case SIL_FAULT_PKUERR:
@@ -3206,6 +3208,12 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
                        if ((sig == SIGBUS) &&
                            (si_code >= BUS_MCEERR_AR) && (si_code <= BUS_MCEERR_AO))
                                layout = SIL_FAULT_MCEERR;
+                       else if (IS_ENABLED(ALPHA) &&
Quoted 8 lines

kernel/signal.c enum siginfo_layout siginfo_layout(unsigned sig, int si_code) · line 3216

Quoted 5 lines
kernel/signal.c +5 -0
+                                ((sig == SIGFPE) ||
+                                 ((sig == SIGTRAP) && (si_code == TRAP_UNK))))
+                               layout = SIL_FAULT_TRAPNO;
+                       else if (IS_ENABLED(SPARC) && (sig == SIGILL) && (si_code == ILL_ILLTRP))
+                               layout = SIL_FAULT_TRAPNO;

The breakage isn't apparent here, but in later patches. These need to become CONFIG_SPARC and CONFIG_ALPHA.

Good catch. For some reason I thought IS_ENABLED added the CONFIG_ prefix but I looked and it doesn't.

Quoted 73 lines
Quoted 73 lines
kernel/signal.c +6 -23
                        else if ((sig == SIGSEGV) && (si_code == SEGV_BNDERR))
                                layout = SIL_FAULT_BNDERR;
 #ifdef SEGV_PKUERR
@@ -3317,30 +3325,22 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
                break;
        case SIL_FAULT:
                to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+               break;
+       case SIL_FAULT_TRAPNO:
+               to->si_addr = ptr_to_compat(from->si_addr);
                to->si_trapno = from->si_trapno;
-#endif
                break;
        case SIL_FAULT_MCEERR:
                to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_addr_lsb = from->si_addr_lsb;
                break;
        case SIL_FAULT_BNDERR:
                to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_lower = ptr_to_compat(from->si_lower);
                to->si_upper = ptr_to_compat(from->si_upper);
                break;
        case SIL_FAULT_PKUERR:
                to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_pkey = from->si_pkey;
                break;
        case SIL_PERF_EVENT:
@@ -3401,30 +3401,22 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
                break;
        case SIL_FAULT:
                to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+               break;
+       case SIL_FAULT_TRAPNO:
+               to->si_addr = compat_ptr(from->si_addr);
                to->si_trapno = from->si_trapno;
-#endif
                break;
        case SIL_FAULT_MCEERR:
                to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_addr_lsb = from->si_addr_lsb;
                break;
        case SIL_FAULT_BNDERR:
                to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_lower = compat_ptr(from->si_lower);
                to->si_upper = compat_ptr(from->si_upper);
                break;
        case SIL_FAULT_PKUERR:
                to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_pkey = from->si_pkey;
                break;
        case SIL_PERF_EVENT:
--

2.30.1

#16 lore raw

PATCH 3/3 signal: Use dedicated helpers to send signals with si_trapno set

In reply to Eric W. Biederman

Now that si_trapno is no longer expected to be present for every fault reported using siginfo on alpha and sparc remove the trapno parameter from force_sig_fault, force_sig_fault_to_task and send_sig_fault.

Add two new helpers force_sig_fault_trapno and send_sig_fault_trapno for those signals where trapno is expected to be set.

---
 arch/alpha/kernel/osf_sys.c      |  2 +-
 arch/alpha/kernel/signal.c       |  4 +--
 arch/alpha/kernel/traps.c        | 24 ++++++++---------
 arch/alpha/mm/fault.c            |  4 +--
 arch/sparc/kernel/process_64.c   |  2 +-
 arch/sparc/kernel/sys_sparc_32.c |  2 +-
 arch/sparc/kernel/sys_sparc_64.c |  2 +-
 arch/sparc/kernel/traps_32.c     | 22 ++++++++--------
 arch/sparc/kernel/traps_64.c     | 44 ++++++++++++++------------------
 arch/sparc/kernel/unaligned_32.c |  2 +-
 arch/sparc/mm/fault_32.c         |  2 +-
 arch/sparc/mm/fault_64.c         |  2 +-
 include/linux/sched/signal.h     | 12 +++------
 kernel/signal.c                  | 41 +++++++++++++++++++++--------
 14 files changed, 88 insertions(+), 77 deletions(-)
arch/alpha/kernel/osf_sys.c +1 -1
index d5367a1c6300..80c5d7fbe66a 100644
@@ -878,7 +878,7 @@ SYSCALL_DEFINE5(osf_setsysinfo, unsigned long, op, void __user *, buffer,
 
 			send_sig_fault(SIGFPE, si_code,
 				       (void __user *)NULL,  /* FIXME */
-				       0, current);
+				       current);
  		}
 		return 0;
 	}
arch/alpha/kernel/signal.c +2 -2
index 948b89789da8..bc077babafab 100644
@@ -219,7 +219,7 @@ do_sigreturn(struct sigcontext __user *sc)
 
 	/* Send SIGTRAP if we're single-stepping: */
 	if (ptrace_cancel_bpt (current)) {
-		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc, 0,
+		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc,
 			       current);
 	}
 	return;
@@ -247,7 +247,7 @@ do_rt_sigreturn(struct rt_sigframe __user *frame)
 
 	/* Send SIGTRAP if we're single-stepping: */
 	if (ptrace_cancel_bpt (current)) {
-		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc, 0,
+		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc,
 			       current);
 	}
 	return;
arch/alpha/kernel/traps.c +12 -12
index 921d4b6e4d95..0dddf9ecc1f4 100644
@@ -227,7 +227,7 @@ do_entArith(unsigned long summary, unsigned long write_mask,
 	}
 	die_if_kernel("Arithmetic fault", regs, 0, NULL);
 
-	send_sig_fault(SIGFPE, si_code, (void __user *) regs->pc, 0, current);
+	send_sig_fault_trapno(SIGFPE, si_code, (void __user *) regs->pc, 0, current);
 }
 
 asmlinkage void
@@ -268,12 +268,12 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 			regs->pc -= 4;	/* make pc point to former bpt */
 		}
 
-		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc, 0,
+		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc,
 			       current);
 		return;
 
 	      case 1: /* bugcheck */
-		send_sig_fault(SIGTRAP, TRAP_UNK, (void __user *) regs->pc, 0,
+		send_sig_fault(SIGTRAP, TRAP_UNK, (void __user *) regs->pc,
 			       current);
 		return;
 		
@@ -335,8 +335,8 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 			break;
 		}
 
-		send_sig_fault(signo, code, (void __user *) regs->pc, regs->r16,
-			       current);
+		send_sig_fault_trapno(signo, code, (void __user *) regs->pc,
+				      regs->r16, current);
 		return;
 
 	      case 4: /* opDEC */
@@ -360,9 +360,9 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 			if (si_code == 0)
 				return;
 			if (si_code > 0) {
-				send_sig_fault(SIGFPE, si_code,
-					       (void __user *) regs->pc, 0,
-					       current);
+				send_sig_fault_trapno(SIGFPE, si_code,
+						      (void __user *) regs->pc,
+						      0, current);
 				return;
 			}
 		}
@@ -387,7 +387,7 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 		      ;
 	}
 
-	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, 0, current);
+	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, current);
 }
 
 /* There is an ifdef in the PALcode in MILO that enables a 
@@ -402,7 +402,7 @@ do_entDbg(struct pt_regs *regs)
 {
 	die_if_kernel("Instruction fault", regs, 0, NULL);
 
-	force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, 0);
+	force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc);
 }
 
 
@@ -964,12 +964,12 @@ do_entUnaUser(void __user * va, unsigned long opcode,
 			si_code = SEGV_MAPERR;
 		mmap_read_unlock(mm);
 	}
-	send_sig_fault(SIGSEGV, si_code, va, 0, current);
+	send_sig_fault(SIGSEGV, si_code, va, current);
 	return;
 
 give_sigbus:
 	regs->pc -= 4;
-	send_sig_fault(SIGBUS, BUS_ADRALN, va, 0, current);
+	send_sig_fault(SIGBUS, BUS_ADRALN, va, current);
 	return;
 }
 
arch/alpha/mm/fault.c +2 -2
index 09172f017efc..eee5102c3d88 100644
@@ -219,13 +219,13 @@ do_page_fault(unsigned long address, unsigned long mmcsr,
 	mmap_read_unlock(mm);
 	/* Send a sigbus, regardless of whether we were in kernel
 	   or user mode.  */
-	force_sig_fault(SIGBUS, BUS_ADRERR, (void __user *) address, 0);
+	force_sig_fault(SIGBUS, BUS_ADRERR, (void __user *) address);
 	if (!user_mode(regs))
 		goto no_context;
 	return;
 
  do_sigsegv:
-	force_sig_fault(SIGSEGV, si_code, (void __user *) address, 0);
+	force_sig_fault(SIGSEGV, si_code, (void __user *) address);
 	return;
 
 #ifdef CONFIG_ALPHA_LARGE_VMALLOC
arch/sparc/kernel/process_64.c +1 -1
index 7afd0a859a78..29e67854d5a4 100644
@@ -518,7 +518,7 @@ void synchronize_user_stack(void)
 
 static void stack_unaligned(unsigned long sp)
 {
-	force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) sp, 0);
+	force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) sp);
 }
 
 static const char uwfault32[] = KERN_INFO \
arch/sparc/kernel/sys_sparc_32.c +1 -1
index be77538bc038..082a551897ed 100644
@@ -151,7 +151,7 @@ sparc_breakpoint (struct pt_regs *regs)
 #ifdef DEBUG_SPARC_BREAKPOINT
         printk ("TRAP: Entering kernel PC=%x, nPC=%x\n", regs->pc, regs->npc);
 #endif
-	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc, 0);
+	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc);
 
 #ifdef DEBUG_SPARC_BREAKPOINT
 	printk ("TRAP: Returning to space: PC=%x nPC=%x\n", regs->pc, regs->npc);
arch/sparc/kernel/sys_sparc_64.c +1 -1
index 6b92fadb6ec7..1e9a9e016237 100644
@@ -514,7 +514,7 @@ asmlinkage void sparc_breakpoint(struct pt_regs *regs)
 #ifdef DEBUG_SPARC_BREAKPOINT
         printk ("TRAP: Entering kernel PC=%lx, nPC=%lx\n", regs->tpc, regs->tnpc);
 #endif
-	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->tpc, 0);
+	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->tpc);
 #ifdef DEBUG_SPARC_BREAKPOINT
 	printk ("TRAP: Returning to space: PC=%lx nPC=%lx\n", regs->tpc, regs->tnpc);
 #endif
arch/sparc/kernel/traps_32.c +11 -11
index 247a0d9683b2..5630e5a395e0 100644
@@ -102,8 +102,8 @@ void do_hw_interrupt(struct pt_regs *regs, unsigned long type)
 	if(regs->psr & PSR_PS)
 		die_if_kernel("Kernel bad trap", regs);
 
-	force_sig_fault(SIGILL, ILL_ILLTRP,
-			(void __user *)regs->pc, type - 0x80);
+	force_sig_fault_trapno(SIGILL, ILL_ILLTRP,
+			       (void __user *)regs->pc, type - 0x80);
 }
 
 void do_illegal_instruction(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -116,7 +116,7 @@ void do_illegal_instruction(struct pt_regs *regs, unsigned long pc, unsigned lon
 	       regs->pc, *(unsigned long *)regs->pc);
 #endif
 
-	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc, current);
 }
 
 void do_priv_instruction(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -124,7 +124,7 @@ void do_priv_instruction(struct pt_regs *regs, unsigned long pc, unsigned long n
 {
 	if(psr & PSR_PS)
 		die_if_kernel("Penguin instruction from Penguin mode??!?!", regs);
-	send_sig_fault(SIGILL, ILL_PRVOPC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_PRVOPC, (void __user *)pc, current);
 }
 
 /* XXX User may want to be allowed to do this. XXX */
@@ -145,7 +145,7 @@ void do_memaccess_unaligned(struct pt_regs *regs, unsigned long pc, unsigned lon
 #endif
 	send_sig_fault(SIGBUS, BUS_ADRALN,
 		       /* FIXME: Should dig out mna address */ (void *)0,
-		       0, current);
+		       current);
 }
 
 static unsigned long init_fsr = 0x0UL;
@@ -291,7 +291,7 @@ void do_fpe_trap(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 		else if (fsr & 0x01)
 			code = FPE_FLTRES;
 	}
-	send_sig_fault(SIGFPE, code, (void __user *)pc, 0, fpt);
+	send_sig_fault(SIGFPE, code, (void __user *)pc, fpt);
 #ifndef CONFIG_SMP
 	last_task_used_math = NULL;
 #endif
@@ -305,7 +305,7 @@ void handle_tag_overflow(struct pt_regs *regs, unsigned long pc, unsigned long n
 {
 	if(psr & PSR_PS)
 		die_if_kernel("Penguin overflow trap from kernel mode", regs);
-	send_sig_fault(SIGEMT, EMT_TAGOVF, (void __user *)pc, 0, current);
+	send_sig_fault(SIGEMT, EMT_TAGOVF, (void __user *)pc, current);
 }
 
 void handle_watchpoint(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -327,13 +327,13 @@ void handle_reg_access(struct pt_regs *regs, unsigned long pc, unsigned long npc
 	printk("Register Access Exception at PC %08lx NPC %08lx PSR %08lx\n",
 	       pc, npc, psr);
 #endif
-	force_sig_fault(SIGBUS, BUS_OBJERR, (void __user *)pc, 0);
+	force_sig_fault(SIGBUS, BUS_OBJERR, (void __user *)pc);
 }
 
 void handle_cp_disabled(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 			unsigned long psr)
 {
-	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, current);
 }
 
 void handle_cp_exception(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -343,13 +343,13 @@ void handle_cp_exception(struct pt_regs *regs, unsigned long pc, unsigned long n
 	printk("Co-Processor Exception at PC %08lx NPC %08lx PSR %08lx\n",
 	       pc, npc, psr);
 #endif
-	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, current);
 }
 
 void handle_hw_divzero(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 		       unsigned long psr)
 {
-	send_sig_fault(SIGFPE, FPE_INTDIV, (void __user *)pc, 0, current);
+	send_sig_fault(SIGFPE, FPE_INTDIV, (void __user *)pc, current);
 }
 
 #ifdef CONFIG_DEBUG_BUGVERBOSE
arch/sparc/kernel/traps_64.c +19 -25

144 lines not on this page · raw message

arch/sparc/kernel/unaligned_32.c +1 -1
index ef5c5207c9ff..455f0258c745 100644
@@ -278,5 +278,5 @@ asmlinkage void user_unaligned_trap(struct pt_regs *regs, unsigned int insn)
 {
 	send_sig_fault(SIGBUS, BUS_ADRALN,
 		       (void __user *)safe_compute_effective_address(regs, insn),
-		       0, current);
+		       current);
 }
arch/sparc/mm/fault_32.c +1 -1
index de2031c2b2d7..fa858626b85b 100644
@@ -83,7 +83,7 @@ static void __do_fault_siginfo(int code, int sig, struct pt_regs *regs,
 		show_signal_msg(regs, sig, code,
 				addr, current);
 
-	force_sig_fault(sig, code, (void __user *) addr, 0);
+	force_sig_fault(sig, code, (void __user *) addr);
 }
 
 static unsigned long compute_si_addr(struct pt_regs *regs, int text_fault)
arch/sparc/mm/fault_64.c +1 -1
index 0a6bcc85fba7..9a9652a15fed 100644
@@ -176,7 +176,7 @@ static void do_fault_siginfo(int code, int sig, struct pt_regs *regs,
 	if (unlikely(show_unhandled_signals))
 		show_signal_msg(regs, sig, code, addr, current);
 
-	force_sig_fault(sig, code, (void __user *) addr, 0);
+	force_sig_fault(sig, code, (void __user *) addr);
 }
 
 static unsigned int get_fault_insn(struct pt_regs *regs, unsigned int insn)
include/linux/sched/signal.h +4 -8
index 3f6a0fcaa10c..7daa425f3055 100644
@@ -298,11 +298,6 @@ static inline void kernel_signal_stop(void)
 
 	schedule();
 }
-#ifdef __ARCH_SI_TRAPNO
-# define ___ARCH_SI_TRAPNO(_a1) , _a1
-#else
-# define ___ARCH_SI_TRAPNO(_a1)
-#endif
 #ifdef __ia64__
 # define ___ARCH_SI_IA64(_a1, _a2, _a3) , _a1, _a2, _a3
 #else
@@ -310,14 +305,11 @@ static inline void kernel_signal_stop(void)
 #endif
 
 int force_sig_fault_to_task(int sig, int code, void __user *addr
-	___ARCH_SI_TRAPNO(int trapno)
 	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
 	, struct task_struct *t);
 int force_sig_fault(int sig, int code, void __user *addr
-	___ARCH_SI_TRAPNO(int trapno)
 	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr));
 int send_sig_fault(int sig, int code, void __user *addr
-	___ARCH_SI_TRAPNO(int trapno)
 	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
 	, struct task_struct *t);
 
@@ -327,6 +319,10 @@ int send_sig_mceerr(int code, void __user *, short, struct task_struct *);
 int force_sig_bnderr(void __user *addr, void __user *lower, void __user *upper);
 int force_sig_pkuerr(void __user *addr, u32 pkey);
 
+int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno);
+int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
+			  struct task_struct *task);
+
 int force_sig_ptrace_errno_trap(int errno, void __user *addr);
 
 extern int send_sig_info(int, struct kernel_siginfo *, struct task_struct *);
kernel/signal.c +31 -10

78 lines not on this page · raw message

Signature · 1 line
2.30.1
#17 lore raw

Re: [PATCH 3/3] signal: Use dedicated helpers to send signals with si_trapno set

Quoted 9 lines · Eric W. Biederman

Now that si_trapno is no longer expected to be present for every fault reported using siginfo on alpha and sparc remove the trapno parameter from force_sig_fault, force_sig_fault_to_task and send_sig_fault.

Add two new helpers force_sig_fault_trapno and send_sig_fault_trapno for those signals where trapno is expected to be set.

Quoted 16 lines
---
 arch/alpha/kernel/osf_sys.c      |  2 +-
 arch/alpha/kernel/signal.c       |  4 +--
 arch/alpha/kernel/traps.c        | 24 ++++++++---------
 arch/alpha/mm/fault.c            |  4 +--
 arch/sparc/kernel/process_64.c   |  2 +-
 arch/sparc/kernel/sys_sparc_32.c |  2 +-
 arch/sparc/kernel/sys_sparc_64.c |  2 +-
 arch/sparc/kernel/traps_32.c     | 22 ++++++++--------
 arch/sparc/kernel/traps_64.c     | 44 ++++++++++++++------------------
 arch/sparc/kernel/unaligned_32.c |  2 +-
 arch/sparc/mm/fault_32.c         |  2 +-
 arch/sparc/mm/fault_64.c         |  2 +-
 include/linux/sched/signal.h     | 12 +++------
 kernel/signal.c                  | 41 +++++++++++++++++++++--------
 14 files changed, 88 insertions(+), 77 deletions(-)

This still breaks sparc64:

Quoted 2 lines

sparc64-linux-gnu-ld: arch/sparc/kernel/traps_64.o: in function `bad_trap':
(.text+0x2a4): undefined reference to `force_sig_fault_trapno'

[...]

Quoted 1 line

+#if IS_ENABLED(SPARC)

This should be 'IS_ENABLED(CONFIG_SPARC)'.

Quoted 5 lines
kernel/signal.c +5 -0
+int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno)
+{
+       struct kernel_siginfo info;
+
+       clear_siginfo(&info);

kernel/signal.c int force_sig_pkuerr(void __user *addr, u32 pkey) · line 1771

Quoted 10 lines
kernel/signal.c +10 -0
+       info.si_signo = sig;
+       info.si_errno = 0;
+       info.si_code  = code;
+       info.si_addr  = addr;
+       info.si_trapno = trapno;
+       return force_sig_info(&info);
+}
+#endif
+
+#if IS_ENABLED(ALPHA)

CONFIG_ALPHA

Quoted 21 lines
kernel/signal.c +15 -1
+int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
+                         struct task_struct *t)
+{
+       struct kernel_siginfo info;
+
+       clear_siginfo(&info);
+       info.si_signo = sig;
+       info.si_errno = 0;
+       info.si_code  = code;
+       info.si_addr  = addr;
+       info.si_trapno = trapno;
+       return send_sig_info(info.si_signo, &info, t);
+}
+#endif
+
 /* For the crazy architectures that include trap information in
  * the errno field, instead of an actual errno value.
  */
--

2.30.1

#18 lore raw

[PATCH 4/3] signal: Remove __ARCH_SI_TRAPNO

In reply to Eric W. Biederman

Now that this define is no longer used remove it from the kernel.

---
 arch/alpha/include/uapi/asm/siginfo.h | 2 --
 arch/mips/include/uapi/asm/siginfo.h  | 2 --
 arch/sparc/include/uapi/asm/siginfo.h | 3 ---
 3 files changed, 7 deletions(-)
arch/alpha/include/uapi/asm/siginfo.h +0 -2
index 6e1a2af2f962..e08eae88182b 100644
@@ -2,8 +2,6 @@
 #ifndef _ALPHA_SIGINFO_H
 #define _ALPHA_SIGINFO_H
 
-#define __ARCH_SI_TRAPNO
-
 #include <asm-generic/siginfo.h>
 
 #endif
arch/mips/include/uapi/asm/siginfo.h +0 -2
index c34c7eef0a1c..8cb8bd061a68 100644
@@ -10,9 +10,7 @@
 #ifndef _UAPI_ASM_SIGINFO_H
 #define _UAPI_ASM_SIGINFO_H
 
-
 #define __ARCH_SIGEV_PREAMBLE_SIZE (sizeof(long) + 2*sizeof(int))
-#undef __ARCH_SI_TRAPNO /* exception code needs to fill this ...  */
 
 #define __ARCH_HAS_SWAPPED_SIGINFO
 
arch/sparc/include/uapi/asm/siginfo.h +0 -3
index 68bdde4c2a2e..0e7c27522aed 100644
@@ -8,9 +8,6 @@
 
 #endif /* defined(__sparc__) && defined(__arch64__) */
 
-
-#define __ARCH_SI_TRAPNO
-
 #include <asm-generic/siginfo.h>
 
 
Signature · 1 line
2.30.1
#19 lore raw

Is perf_sigtrap synchronous?

In reply to Eric W. Biederman

I am looking at perf_sigtrap and I am confused by the code.

	/*
	 * We'd expect this to only occur if the irq_work is delayed and either
	 * ctx->task or current has changed in the meantime. This can be the
	 * case on architectures that do not implement arch_irq_work_raise().
	 */
	if (WARN_ON_ONCE(event->ctx->task != current))
		return;

	/*
	 * perf_pending_event() can race with the task exiting.
	 */
	if (current->flags & PF_EXITING)
		return;

It performs tests that absolutely can never fail if we are talking about a synchronous exception. The code force_sig family of functions only make sense to use with and are only safe to use with synchronous exceptions.

Are the tests in perf_sigtrap necessary or is perf_sigtrap not reporting a synchronous event?

Eric

#20 lore raw

Re: Is perf_sigtrap synchronous?

Quoted 19 lines · Eric W. Biederman

I am looking at perf_sigtrap and I am confused by the code.

        /*
         * We'd expect this to only occur if the irq_work is delayed and either
         * ctx->task or current has changed in the meantime. This can be the
         * case on architectures that do not implement arch_irq_work_raise().
         */
        if (WARN_ON_ONCE(event->ctx->task != current))
                return;

        /*
         * perf_pending_event() can race with the task exiting.
         */
        if (current->flags & PF_EXITING)
                return;
Quoted 7 lines

It performs tests that absolutely can never fail if we are talking about a synchronous exception. The code force_sig family of functions only make sense to use with and are only safe to use with synchronous exceptions.

Are the tests in perf_sigtrap necessary or is perf_sigtrap not reporting a synchronous event?

Yes it's synchronous, insofar that the user will receive the signal right when the event happens (I've tested this extensively, also see tools/testing/selftests/perf_events). Of course, there's some effort involved from the point where the event triggered to actually safely delivering the signal. In particular, for HW events, these arrive in NMI, and we can't do much in NMI, and therefore will queue an irq_work.

On architectures that properly implement irq_work, it will do a self-IPI, so that once it is safe to do so, another interrupt is delivered where we process the event and do the force_sig_info(). The task where the event occurred never got a chance to run -- except for bad architectures with broken irq_work, and the first WARN_ON() is there so we don't crash the kernel if somebody botched their irq_work.

Since we're talking about various HW events, these can still trigger while the task is exiting, before perf_event_exit_task() being called during do_exit(). That's why we have the 2nd check.

Thanks,
-- Marco

#21 lore raw

[PATCH 5/3] signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency

In reply to Eric W. Biederman

It helps to know which part of the siginfo structure the siginfo_layout value is talking about.

---
 fs/signalfd.c          |  2 +-
 include/linux/signal.h |  2 +-
 kernel/signal.c        | 10 +++++-----
 3 files changed, 7 insertions(+), 7 deletions(-)
fs/signalfd.c +1 -1
index e87e59581653..83130244f653 100644
@@ -132,7 +132,7 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		new.ssi_addr = (long) kinfo->si_addr;
 		new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
 		break;
-	case SIL_PERF_EVENT:
+	case SIL_FAULT_PERF_EVENT:
 		new.ssi_addr = (long) kinfo->si_addr;
 		new.ssi_perf = kinfo->si_perf;
 		break;
include/linux/signal.h +1 -1
index 5160fd45e5ca..ed896d790e46 100644
@@ -44,7 +44,7 @@ enum siginfo_layout {
 	SIL_FAULT_MCEERR,
 	SIL_FAULT_BNDERR,
 	SIL_FAULT_PKUERR,
-	SIL_PERF_EVENT,
+	SIL_FAULT_PERF_EVENT,
 	SIL_CHLD,
 	SIL_RT,
 	SIL_SYS,
kernel/signal.c +5 -5
index 0517ff950d38..690921960d8b 100644
@@ -1198,7 +1198,7 @@ static inline bool has_si_pid_and_uid(struct kernel_siginfo *info)
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
-	case SIL_PERF_EVENT:
+	case SIL_FAULT_PERF_EVENT:
 	case SIL_SYS:
 		ret = false;
 		break;
@@ -2553,7 +2553,7 @@ static void hide_si_addr_tag_bits(struct ksignal *ksig)
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
-	case SIL_PERF_EVENT:
+	case SIL_FAULT_PERF_EVENT:
 		ksig->info.si_addr = arch_untagged_si_addr(
 			ksig->info.si_addr, ksig->sig, ksig->info.si_code);
 		break;
@@ -3242,7 +3242,7 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
 				layout = SIL_FAULT_PKUERR;
 #endif
 			else if ((sig == SIGTRAP) && (si_code == TRAP_PERF))
-				layout = SIL_PERF_EVENT;
+				layout = SIL_FAULT_PERF_EVENT;
 		}
 		else if (si_code <= NSIGPOLL)
 			layout = SIL_POLL;
@@ -3364,7 +3364,7 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 		to->si_addr = ptr_to_compat(from->si_addr);
 		to->si_pkey = from->si_pkey;
 		break;
-	case SIL_PERF_EVENT:
+	case SIL_FAULT_PERF_EVENT:
 		to->si_addr = ptr_to_compat(from->si_addr);
 		to->si_perf = from->si_perf;
 		break;
@@ -3440,7 +3440,7 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 		to->si_addr = compat_ptr(from->si_addr);
 		to->si_pkey = from->si_pkey;
 		break;
-	case SIL_PERF_EVENT:
+	case SIL_FAULT_PERF_EVENT:
 		to->si_addr = compat_ptr(from->si_addr);
 		to->si_perf = from->si_perf;
 		break;
Signature · 1 line
2.30.1
author Acked-by
#22 lore raw

Re: [PATCH 5/3] signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency

Quoted 3 lines · Eric W. Biederman

It helps to know which part of the siginfo structure the siginfo_layout value is talking about.

Your Signed-off-by seems to be missing.

Otherwise,

Quoted 84 lines
---
 fs/signalfd.c          |  2 +-
 include/linux/signal.h |  2 +-
 kernel/signal.c        | 10 +++++-----
 3 files changed, 7 insertions(+), 7 deletions(-)
fs/signalfd.c +1 -1
index e87e59581653..83130244f653 100644
@@ -132,7 +132,7 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
                new.ssi_addr = (long) kinfo->si_addr;
                new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
                break;
-       case SIL_PERF_EVENT:
+       case SIL_FAULT_PERF_EVENT:
                new.ssi_addr = (long) kinfo->si_addr;
                new.ssi_perf = kinfo->si_perf;
                break;
include/linux/signal.h +1 -1
index 5160fd45e5ca..ed896d790e46 100644
@@ -44,7 +44,7 @@ enum siginfo_layout {
        SIL_FAULT_MCEERR,
        SIL_FAULT_BNDERR,
        SIL_FAULT_PKUERR,
-       SIL_PERF_EVENT,
+       SIL_FAULT_PERF_EVENT,
        SIL_CHLD,
        SIL_RT,
        SIL_SYS,
kernel/signal.c +5 -6
index 0517ff950d38..690921960d8b 100644
@@ -1198,7 +1198,7 @@ static inline bool has_si_pid_and_uid(struct kernel_siginfo *info)
        case SIL_FAULT_MCEERR:
        case SIL_FAULT_BNDERR:
        case SIL_FAULT_PKUERR:
-       case SIL_PERF_EVENT:
+       case SIL_FAULT_PERF_EVENT:
        case SIL_SYS:
                ret = false;
                break;
@@ -2553,7 +2553,7 @@ static void hide_si_addr_tag_bits(struct ksignal *ksig)
        case SIL_FAULT_MCEERR:
        case SIL_FAULT_BNDERR:
        case SIL_FAULT_PKUERR:
-       case SIL_PERF_EVENT:
+       case SIL_FAULT_PERF_EVENT:
                ksig->info.si_addr = arch_untagged_si_addr(
                        ksig->info.si_addr, ksig->sig, ksig->info.si_code);
                break;
@@ -3242,7 +3242,7 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
                                layout = SIL_FAULT_PKUERR;
 #endif
                        else if ((sig == SIGTRAP) && (si_code == TRAP_PERF))
-                               layout = SIL_PERF_EVENT;
+                               layout = SIL_FAULT_PERF_EVENT;
                }
                else if (si_code <= NSIGPOLL)
                        layout = SIL_POLL;
@@ -3364,7 +3364,7 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
                to->si_addr = ptr_to_compat(from->si_addr);
                to->si_pkey = from->si_pkey;
                break;
-       case SIL_PERF_EVENT:
+       case SIL_FAULT_PERF_EVENT:
                to->si_addr = ptr_to_compat(from->si_addr);
                to->si_perf = from->si_perf;
                break;
@@ -3440,7 +3440,7 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
                to->si_addr = compat_ptr(from->si_addr);
                to->si_pkey = from->si_pkey;
                break;
-       case SIL_PERF_EVENT:
+       case SIL_FAULT_PERF_EVENT:
                to->si_addr = compat_ptr(from->si_addr);
                to->si_perf = from->si_perf;
                break;
--

2.30.1

#23 lore raw

[PATCH 6/3] signal: Factor force_sig_perf out of perf_sigtrap

In reply to Eric W. Biederman

Separate generating the signal from deciding it needs to be sent.

---
 include/linux/sched/signal.h |  1 +
 kernel/events/core.c         | 11 ++---------
 kernel/signal.c              | 13 +++++++++++++
 3 files changed, 16 insertions(+), 9 deletions(-)
include/linux/sched/signal.h +1 -0
index 7daa425f3055..1e2f61a1a512 100644
@@ -318,6 +318,7 @@ int send_sig_mceerr(int code, void __user *, short, struct task_struct *);
 
 int force_sig_bnderr(void __user *addr, void __user *lower, void __user *upper);
 int force_sig_pkuerr(void __user *addr, u32 pkey);
+int force_sig_perf(void __user *addr, u32 type, u64 sig_data);
 
 int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno);
 int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
kernel/events/core.c +2 -9
index 928b166d888e..48ea8863183b 100644
@@ -6394,8 +6394,6 @@ void perf_event_wakeup(struct perf_event *event)
 
 static void perf_sigtrap(struct perf_event *event)
 {
-	struct kernel_siginfo info;
-
 	/*
 	 * We'd expect this to only occur if the irq_work is delayed and either
 	 * ctx->task or current has changed in the meantime. This can be the
@@ -6410,13 +6408,8 @@ static void perf_sigtrap(struct perf_event *event)
 	if (current->flags & PF_EXITING)
 		return;
 
-	clear_siginfo(&info);
-	info.si_signo = SIGTRAP;
-	info.si_code = TRAP_PERF;
-	info.si_errno = event->attr.type;
-	info.si_perf = event->attr.sig_data;
-	info.si_addr = (void __user *)event->pending_addr;
-	force_sig_info(&info);
+	force_sig_perf((void __user *)event->pending_addr,
+		       event->attr.type, event->attr.sig_data);
 }
 
 static void perf_pending_event_disable(struct perf_event *event)
kernel/signal.c +13 -0
index 690921960d8b..5b1ad7f080ab 100644
@@ -1753,6 +1753,19 @@ int force_sig_pkuerr(void __user *addr, u32 pkey)
 }
 #endif
 
+int force_sig_perf(void __user *pending_addr, u32 type, u64 sig_data)
+{
+	struct kernel_siginfo info;
+
+	clear_siginfo(&info);
+	info.si_signo = SIGTRAP;
+	info.si_errno = type;
+	info.si_code  = TRAP_PERF;
+	info.si_addr  = pending_addr;
+	info.si_perf  = sig_data;
+	return force_sig_info(&info);
+}
+
 #if IS_ENABLED(SPARC)
 int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno)
 {
Signature · 1 line
2.30.1
#24 lore raw

Re: [PATCH 6/3] signal: Factor force_sig_perf out of perf_sigtrap

Quoted 56 lines · Eric W. Biederman

Separate generating the signal from deciding it needs to be sent.

---
 include/linux/sched/signal.h |  1 +
 kernel/events/core.c         | 11 ++---------
 kernel/signal.c              | 13 +++++++++++++
 3 files changed, 16 insertions(+), 9 deletions(-)
include/linux/sched/signal.h +1 -0
index 7daa425f3055..1e2f61a1a512 100644
@@ -318,6 +318,7 @@ int send_sig_mceerr(int code, void __user *, short, struct task_struct *);
 
 int force_sig_bnderr(void __user *addr, void __user *lower, void __user *upper);
 int force_sig_pkuerr(void __user *addr, u32 pkey);
+int force_sig_perf(void __user *addr, u32 type, u64 sig_data);
 
 int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno);
 int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
kernel/events/core.c +2 -9
index 928b166d888e..48ea8863183b 100644
@@ -6394,8 +6394,6 @@ void perf_event_wakeup(struct perf_event *event)
 
 static void perf_sigtrap(struct perf_event *event)
 {
-       struct kernel_siginfo info;
-
        /*
         * We'd expect this to only occur if the irq_work is delayed and either
         * ctx->task or current has changed in the meantime. This can be the
@@ -6410,13 +6408,8 @@ static void perf_sigtrap(struct perf_event *event)
        if (current->flags & PF_EXITING)
                return;
 
-       clear_siginfo(&info);
-       info.si_signo = SIGTRAP;
-       info.si_code = TRAP_PERF;
-       info.si_errno = event->attr.type;
-       info.si_perf = event->attr.sig_data;
-       info.si_addr = (void __user *)event->pending_addr;
-       force_sig_info(&info);
+       force_sig_perf((void __user *)event->pending_addr,
+                      event->attr.type, event->attr.sig_data);
 }
 
 static void perf_pending_event_disable(struct perf_event *event)
kernel/signal.c +0 -0
index 690921960d8b..5b1ad7f080ab 100644

kernel/signal.c int force_sig_pkuerr(void __user *addr, u32 pkey) · line 1756

Quoted 4 lines
kernel/signal.c +1 -0
@@ -1753,6 +1753,19 @@ int force_sig_pkuerr(void __user *addr, u32 pkey)
 }
 #endif
 
+int force_sig_perf(void __user *pending_addr, u32 type, u64 sig_data)

s/pending_addr/addr/

to match force_sig_perf() declaration.

Quoted 17 lines
kernel/signal.c +12 -1
+{
+       struct kernel_siginfo info;
+
+       clear_siginfo(&info);
+       info.si_signo = SIGTRAP;
+       info.si_errno = type;
+       info.si_code  = TRAP_PERF;
+       info.si_addr  = pending_addr;
+       info.si_perf  = sig_data;
+       return force_sig_info(&info);
+}
+
 #if IS_ENABLED(SPARC)
 int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno)
 {
--

2.30.1

#25 lore raw

[PATCH 7/3] signal: Deliver all of the perf_data in si_perf

In reply to Eric W. Biederman

Don't abuse si_errno and deliver all of the perf data in si_perf.

---
 fs/signalfd.c                      |  3 ++-
 include/linux/compat.h             |  5 ++++-
 include/uapi/asm-generic/siginfo.h |  5 ++++-
 include/uapi/linux/signalfd.h      |  4 ++--
 kernel/signal.c                    | 18 +++++++++++-------
 5 files changed, 23 insertions(+), 12 deletions(-)
fs/signalfd.c +2 -1
index 83130244f653..9686af56f073 100644
@@ -134,7 +134,8 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		break;
 	case SIL_FAULT_PERF_EVENT:
 		new.ssi_addr = (long) kinfo->si_addr;
-		new.ssi_perf = kinfo->si_perf;
+		new.ssi_perf_type = kinfo->si_perf.type;
+		new.ssi_perf_data = kinfo->si_perf.data;
 		break;
 	case SIL_CHLD:
 		new.ssi_pid    = kinfo->si_pid;
include/linux/compat.h +4 -1
index 24462ed63af4..0726f9b3a57c 100644
@@ -235,7 +235,10 @@ typedef struct compat_siginfo {
 					u32 _pkey;
 				} _addr_pkey;
 				/* used when si_code=TRAP_PERF */
-				compat_ulong_t _perf;
+				struct {
+					compat_ulong_t data;
+					u32 type;
+				} _perf;
 			};
 		} _sigfault;
 
include/uapi/asm-generic/siginfo.h +4 -1
index 2abdf1d19aad..19b6310021a3 100644
@@ -90,7 +90,10 @@ union __sifields {
 				__u32 _pkey;
 			} _addr_pkey;
 			/* used when si_code=TRAP_PERF */
-			unsigned long _perf;
+			struct {
+				unsigned long data;
+				u32 type;
+			} _perf;
 		};
 	} _sigfault;
 
include/uapi/linux/signalfd.h +2 -2
index 7e333042c7e3..e78dddf433fc 100644
@@ -39,8 +39,8 @@ struct signalfd_siginfo {
 	__s32 ssi_syscall;
 	__u64 ssi_call_addr;
 	__u32 ssi_arch;
-	__u32 __pad3;
-	__u64 ssi_perf;
+	__u32 ssi_perf_type;
+	__u64 ssi_perf_data;
 
 	/*
 	 * Pad strcture to 128 bytes. Remember to update the
kernel/signal.c +11 -7
index 5b1ad7f080ab..cb3574b7319c 100644
@@ -1758,11 +1758,13 @@ int force_sig_perf(void __user *pending_addr, u32 type, u64 sig_data)
 	struct kernel_siginfo info;
 
 	clear_siginfo(&info);
-	info.si_signo = SIGTRAP;
-	info.si_errno = type;
-	info.si_code  = TRAP_PERF;
-	info.si_addr  = pending_addr;
-	info.si_perf  = sig_data;
+	info.si_signo     = SIGTRAP;
+	info.si_errno     = 0;
+	info.si_code      = TRAP_PERF;
+	info.si_addr      = pending_addr;
+	info.si_perf.data = sig_data;
+	info.si_perf.type = type;
+
 	return force_sig_info(&info);
 }
 
@@ -3379,7 +3381,8 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 		break;
 	case SIL_FAULT_PERF_EVENT:
 		to->si_addr = ptr_to_compat(from->si_addr);
-		to->si_perf = from->si_perf;
+		to->si_perf.data = from->si_perf.data;
+		to->si_perf.type = from->si_perf.type;
 		break;
 	case SIL_CHLD:
 		to->si_pid = from->si_pid;
@@ -3455,7 +3458,8 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 		break;
 	case SIL_FAULT_PERF_EVENT:
 		to->si_addr = compat_ptr(from->si_addr);
-		to->si_perf = from->si_perf;
+		to->si_perf.data = from->si_perf.data;
+		to->si_perf.type = from->si_perf.type;
 		break;
 	case SIL_CHLD:
 		to->si_pid    = from->si_pid;
Signature · 1 line
2.30.1
#26 lore raw

Re: [PATCH 7/3] signal: Deliver all of the perf_data in si_perf

Quoted 5 lines · Eric W. Biederman

Don't abuse si_errno and deliver all of the perf data in si_perf.

---

Thank you for the fix, this looks cleaner.

Just note that this patch needs to include updates to tools/testing/selftests/perf_events. This should do it:

Quoted 1 line

sed -i 's/si_perf/si_perf.data/g; s/si_errno/si_perf.type/g' tools/testing/selftests/perf_events/*.c

Subject: s/perf_data/perf data/ ?

For uapi, need to switch to __u32, see below.

Quoted 42 lines
 fs/signalfd.c                      |  3 ++-
 include/linux/compat.h             |  5 ++++-
 include/uapi/asm-generic/siginfo.h |  5 ++++-
 include/uapi/linux/signalfd.h      |  4 ++--
 kernel/signal.c                    | 18 +++++++++++-------
 5 files changed, 23 insertions(+), 12 deletions(-)
fs/signalfd.c +2 -1
index 83130244f653..9686af56f073 100644
@@ -134,7 +134,8 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
                break;
        case SIL_FAULT_PERF_EVENT:
                new.ssi_addr = (long) kinfo->si_addr;
-               new.ssi_perf = kinfo->si_perf;
+               new.ssi_perf_type = kinfo->si_perf.type;
+               new.ssi_perf_data = kinfo->si_perf.data;
                break;
        case SIL_CHLD:
                new.ssi_pid    = kinfo->si_pid;
include/linux/compat.h +4 -1
index 24462ed63af4..0726f9b3a57c 100644
@@ -235,7 +235,10 @@ typedef struct compat_siginfo {
                                        u32 _pkey;
                                } _addr_pkey;
                                /* used when si_code=TRAP_PERF */
-                               compat_ulong_t _perf;
+                               struct {
+                                       compat_ulong_t data;
+                                       u32 type;
+                               } _perf;
                        };
                } _sigfault;
 
include/uapi/asm-generic/siginfo.h +0 -0
index 2abdf1d19aad..19b6310021a3 100644

include/uapi/asm-generic/siginfo.h union __sifields { · line 95

Quoted 7 lines
include/uapi/asm-generic/siginfo.h +3 -1
@@ -90,7 +90,10 @@ union __sifields {
                                __u32 _pkey;
                        } _addr_pkey;
                        /* used when si_code=TRAP_PERF */
-                       unsigned long _perf;
+                       struct {
+                               unsigned long data;
+                               u32 type;

This needs to be __u32.

Quoted 64 lines
include/uapi/asm-generic/siginfo.h +1 -0
+                       } _perf;
                };
        } _sigfault;
 
include/uapi/linux/signalfd.h +2 -2
index 7e333042c7e3..e78dddf433fc 100644
@@ -39,8 +39,8 @@ struct signalfd_siginfo {
        __s32 ssi_syscall;
        __u64 ssi_call_addr;
        __u32 ssi_arch;
-       __u32 __pad3;
-       __u64 ssi_perf;
+       __u32 ssi_perf_type;
+       __u64 ssi_perf_data;
 
        /*
         * Pad strcture to 128 bytes. Remember to update the
kernel/signal.c +11 -8
index 5b1ad7f080ab..cb3574b7319c 100644
@@ -1758,11 +1758,13 @@ int force_sig_perf(void __user *pending_addr, u32 type, u64 sig_data)
        struct kernel_siginfo info;
 
        clear_siginfo(&info);
-       info.si_signo = SIGTRAP;
-       info.si_errno = type;
-       info.si_code  = TRAP_PERF;
-       info.si_addr  = pending_addr;
-       info.si_perf  = sig_data;
+       info.si_signo     = SIGTRAP;
+       info.si_errno     = 0;
+       info.si_code      = TRAP_PERF;
+       info.si_addr      = pending_addr;
+       info.si_perf.data = sig_data;
+       info.si_perf.type = type;
+
        return force_sig_info(&info);
 }
 
@@ -3379,7 +3381,8 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
                break;
        case SIL_FAULT_PERF_EVENT:
                to->si_addr = ptr_to_compat(from->si_addr);
-               to->si_perf = from->si_perf;
+               to->si_perf.data = from->si_perf.data;
+               to->si_perf.type = from->si_perf.type;
                break;
        case SIL_CHLD:
                to->si_pid = from->si_pid;
@@ -3455,7 +3458,8 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
                break;
        case SIL_FAULT_PERF_EVENT:
                to->si_addr = compat_ptr(from->si_addr);
-               to->si_perf = from->si_perf;
+               to->si_perf.data = from->si_perf.data;
+               to->si_perf.type = from->si_perf.type;
                break;
        case SIL_CHLD:
                to->si_pid    = from->si_pid;
--

2.30.1

#27 lore raw

Re: [PATCH 7/3] signal: Deliver all of the perf_data in si_perf

Quoted 9 lines · Marco Elver
Quoted 5 lines · Eric W. Biederman

Don't abuse si_errno and deliver all of the perf data in si_perf.

---

Thank you for the fix, this looks cleaner.

Quoted 7 lines

Just note that this patch needs to include updates to tools/testing/selftests/perf_events. This should do it:

Quoted 1 line

sed -i 's/si_perf/si_perf.data/g; s/si_errno/si_perf.type/g' tools/testing/selftests/perf_events/*.c

Subject: s/perf_data/perf data/ ?

For uapi, need to switch to __u32, see below.

Good point.

The one thing that this doesn't do is give you a 64bit field on 32bit architectures.

On 32bit builds the layout is:

	int si_signo;
	int si_errno;
	int si_code;
	void __user *_addr;

So I believe if the first 3 fields were moved into the _sifields union si_perf could define a 64bit field as it's first member and it would not break anything else.

Given that the data field is 64bit that seems desirable.

Eric

Quoted 117 lines
Quoted 42 lines
 fs/signalfd.c                      |  3 ++-
 include/linux/compat.h             |  5 ++++-
 include/uapi/asm-generic/siginfo.h |  5 ++++-
 include/uapi/linux/signalfd.h      |  4 ++--
 kernel/signal.c                    | 18 +++++++++++-------
 5 files changed, 23 insertions(+), 12 deletions(-)
fs/signalfd.c +2 -1
index 83130244f653..9686af56f073 100644
@@ -134,7 +134,8 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
                break;
        case SIL_FAULT_PERF_EVENT:
                new.ssi_addr = (long) kinfo->si_addr;
-               new.ssi_perf = kinfo->si_perf;
+               new.ssi_perf_type = kinfo->si_perf.type;
+               new.ssi_perf_data = kinfo->si_perf.data;
                break;
        case SIL_CHLD:
                new.ssi_pid    = kinfo->si_pid;
include/linux/compat.h +4 -1
index 24462ed63af4..0726f9b3a57c 100644
@@ -235,7 +235,10 @@ typedef struct compat_siginfo {
                                        u32 _pkey;
                                } _addr_pkey;
                                /* used when si_code=TRAP_PERF */
-                               compat_ulong_t _perf;
+                               struct {
+                                       compat_ulong_t data;
+                                       u32 type;
+                               } _perf;
                        };
                } _sigfault;
 
include/uapi/asm-generic/siginfo.h +0 -0
index 2abdf1d19aad..19b6310021a3 100644

include/uapi/asm-generic/siginfo.h union __sifields { · line 95

Quoted 7 lines
include/uapi/asm-generic/siginfo.h +3 -1
@@ -90,7 +90,10 @@ union __sifields {
                                __u32 _pkey;
                        } _addr_pkey;
                        /* used when si_code=TRAP_PERF */
-                       unsigned long _perf;
+                       struct {
+                               unsigned long data;
+                               u32 type;

This needs to be __u32.

Quoted 64 lines
include/uapi/asm-generic/siginfo.h +1 -0
+                       } _perf;
                };
        } _sigfault;
 
include/uapi/linux/signalfd.h +2 -2
index 7e333042c7e3..e78dddf433fc 100644
@@ -39,8 +39,8 @@ struct signalfd_siginfo {
        __s32 ssi_syscall;
        __u64 ssi_call_addr;
        __u32 ssi_arch;
-       __u32 __pad3;
-       __u64 ssi_perf;
+       __u32 ssi_perf_type;
+       __u64 ssi_perf_data;
 
        /*
         * Pad strcture to 128 bytes. Remember to update the
kernel/signal.c +11 -8
index 5b1ad7f080ab..cb3574b7319c 100644
@@ -1758,11 +1758,13 @@ int force_sig_perf(void __user *pending_addr, u32 type, u64 sig_data)
        struct kernel_siginfo info;
 
        clear_siginfo(&info);
-       info.si_signo = SIGTRAP;
-       info.si_errno = type;
-       info.si_code  = TRAP_PERF;
-       info.si_addr  = pending_addr;
-       info.si_perf  = sig_data;
+       info.si_signo     = SIGTRAP;
+       info.si_errno     = 0;
+       info.si_code      = TRAP_PERF;
+       info.si_addr      = pending_addr;
+       info.si_perf.data = sig_data;
+       info.si_perf.type = type;
+
        return force_sig_info(&info);
 }
 
@@ -3379,7 +3381,8 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
                break;
        case SIL_FAULT_PERF_EVENT:
                to->si_addr = ptr_to_compat(from->si_addr);
-               to->si_perf = from->si_perf;
+               to->si_perf.data = from->si_perf.data;
+               to->si_perf.type = from->si_perf.type;
                break;
        case SIL_CHLD:
                to->si_pid = from->si_pid;
@@ -3455,7 +3458,8 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
                break;
        case SIL_FAULT_PERF_EVENT:
                to->si_addr = compat_ptr(from->si_addr);
-               to->si_perf = from->si_perf;
+               to->si_perf.data = from->si_perf.data;
+               to->si_perf.type = from->si_perf.type;
                break;
        case SIL_CHLD:
                to->si_pid    = from->si_pid;
--

2.30.1

#28 lore raw

Re: [PATCH 7/3] signal: Deliver all of the perf_data in si_perf

Quoted 32 lines · Eric W. Biederman
Quoted 9 lines · Marco Elver
Quoted 5 lines · Eric W. Biederman

Don't abuse si_errno and deliver all of the perf data in si_perf.

---

Thank you for the fix, this looks cleaner.

Quoted 7 lines

Just note that this patch needs to include updates to tools/testing/selftests/perf_events. This should do it:

Quoted 1 line

sed -i 's/si_perf/si_perf.data/g; s/si_errno/si_perf.type/g' tools/testing/selftests/perf_events/*.c

Subject: s/perf_data/perf data/ ?

For uapi, need to switch to __u32, see below.

Good point.

The one thing that this doesn't do is give you a 64bit field on 32bit architectures.

On 32bit builds the layout is:

        int si_signo;
        int si_errno;
        int si_code;
        void __user *_addr;
Quoted 5 lines

So I believe if the first 3 fields were moved into the _sifields union si_perf could define a 64bit field as it's first member and it would not break anything else.

Given that the data field is 64bit that seems desirable.

Yes, it's quite unfortunate -- it was __u64 at first, but then we noticed it broke 32-bit architectures like arm:
https://lore.kernel.org/linux-arch/[email protected]/

Thanks,
-- Marco

maintainer
#29 lore raw

Re: [PATCH 7/3] signal: Deliver all of the perf_data in si_perf

In reply to Eric W. Biederman
Quoted 10 lines · Eric W. Biederman

The one thing that this doesn't do is give you a 64bit field on 32bit architectures.

On 32bit builds the layout is:

	int si_signo;
	int si_errno;
	int si_code;
	void __user *_addr;
Quoted 5 lines

So I believe if the first 3 fields were moved into the _sifields union si_perf could define a 64bit field as it's first member and it would not break anything else.

Given that the data field is 64bit that seems desirable.

The data field is fundamentally an address, it is internally a u64 because the perf ring buffer has u64 alignment and it saves on compat crap etc.

So for the 32bit/compat case the high bits will always be 0 and truncating into an unsigned long is fine.

#30 lore raw

Re: [PATCH 7/3] signal: Deliver all of the perf_data in si_perf

Quoted 16 lines · Peter Zijlstra
Quoted 14 lines · Eric W. Biederman

The one thing that this doesn't do is give you a 64bit field on 32bit architectures.

On 32bit builds the layout is:

	int si_signo;
	int si_errno;
	int si_code;
	void __user *_addr;

So I believe if the first 3 fields were moved into the _sifields union si_perf could define a 64bit field as it's first member and it would not break anything else.

Quoted 8 lines
Quoted 1 line

Given that the data field is 64bit that seems desirable.

The data field is fundamentally an address, it is internally a u64 because the perf ring buffer has u64 alignment and it saves on compat crap etc.

So for the 32bit/compat case the high bits will always be 0 and truncating into an unsigned long is fine.

I see why it is fine to truncate the data field into an unsigned long.

Other than technical difficulties in extending siginfo_t is there any reason not to define data as a __u64?

Eric

#31 lore raw

Re: [PATCH 7/3] signal: Deliver all of the perf_data in si_perf

Quoted 24 lines · Eric W. Biederman
Quoted 22 lines · Peter Zijlstra
Quoted 10 lines · Eric W. Biederman

The one thing that this doesn't do is give you a 64bit field on 32bit architectures.

On 32bit builds the layout is:

     int si_signo;
     int si_errno;
     int si_code;
     void __user *_addr;
Quoted 5 lines

So I believe if the first 3 fields were moved into the _sifields union si_perf could define a 64bit field as it's first member and it would not break anything else.

Given that the data field is 64bit that seems desirable.

The data field is fundamentally an address, it is internally a u64 because the perf ring buffer has u64 alignment and it saves on compat crap etc.

Quoted 7 lines
Quoted 2 lines

So for the 32bit/compat case the high bits will always be 0 and truncating into an unsigned long is fine.

I see why it is fine to truncate the data field into an unsigned long.

Other than technical difficulties in extending siginfo_t is there any reason not to define data as a __u64?

No -- like I pointed at earlier, si_perf used to be __u64, but we can't because of the siginfo_t limitation. What we have now is fine, and not worth dwelling over given siginfo limitations.

Thanks,
-- Marco

#32 lore raw

Re: [RFC][PATCH 0/3] signal: Move si_trapno into the _si_fault union

In reply to Eric W. Biederman

Well with 7 patches instead of 3 that was a little more than I thought I was going to send.

However that does demonstrate what I am thinking, and I think most of the changes are reasonable at this point.

I am very curious how synchronous this all is, because if this code is truly synchronous updating signalfd to handle this class of signal doesn't really make sense.

If the code is not synchronous using force_sig is questionable.

Eric W. Biederman (7):
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Use dedicated helpers to send signals with si_trapno set
signal: Remove __ARCH_SI_TRAPNO
signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency
signal: Factor force_sig_perf out of perf_sigtrap
signal: Deliver all of the perf_data in si_perf

 arch/alpha/include/uapi/asm/siginfo.h |   2 -
 arch/alpha/kernel/osf_sys.c           |   2 +-
 arch/alpha/kernel/signal.c            |   4 +-
 arch/alpha/kernel/traps.c             |  24 ++++----
 arch/alpha/mm/fault.c                 |   4 +-
 arch/mips/include/uapi/asm/siginfo.h  |   2 -
 arch/sparc/include/uapi/asm/siginfo.h |   3 -
 arch/sparc/kernel/process_64.c        |   2 +-
 arch/sparc/kernel/sys_sparc_32.c      |   2 +-
 arch/sparc/kernel/sys_sparc_64.c      |   2 +-
 arch/sparc/kernel/traps_32.c          |  22 +++----
 arch/sparc/kernel/traps_64.c          |  44 ++++++--------
 arch/sparc/kernel/unaligned_32.c      |   2 +-
 arch/sparc/mm/fault_32.c              |   2 +-
 arch/sparc/mm/fault_64.c              |   2 +-
 fs/signalfd.c                         |  13 ++--
 include/linux/compat.h                |   9 +--
 include/linux/sched/signal.h          |  13 ++--
 include/linux/signal.h                |   3 +-
 include/uapi/asm-generic/siginfo.h    |  11 ++--
 include/uapi/linux/signalfd.h         |   4 +-
 kernel/events/core.c                  |  11 +---
 kernel/signal.c                       | 108 ++++++++++++++++++++++------------
 23 files changed, 149 insertions(+), 142 deletions(-)
#33 lore raw

Re: [RFC][PATCH 0/3] signal: Move si_trapno into the _si_fault union

Quoted 13 lines · Eric W. Biederman

Well with 7 patches instead of 3 that was a little more than I thought I was going to send.

However that does demonstrate what I am thinking, and I think most of the changes are reasonable at this point.

I am very curious how synchronous this all is, because if this code is truly synchronous updating signalfd to handle this class of signal doesn't really make sense.

If the code is not synchronous using force_sig is questionable.

Quoted 8 lines

Eric W. Biederman (7):
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Use dedicated helpers to send signals with si_trapno set
signal: Remove __ARCH_SI_TRAPNO
signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency
signal: Factor force_sig_perf out of perf_sigtrap
signal: Deliver all of the perf_data in si_perf

Thank you for doing this so quickly -- it looks much cleaner. I'll have a more detailed look next week and also run some tests myself.

At a first glance, you've broken our tests in
tools/testing/selftests/perf_events/ -- needs a
s/si_perf/si_perf.data/, s/si_errno/si_perf.type/

Thanks!

-- Marco

#34 lore raw

Re: [RFC][PATCH 0/3] signal: Move si_trapno into the _si_fault union

Quoted 23 lines · Marco Elver
Quoted 13 lines · Eric W. Biederman

Well with 7 patches instead of 3 that was a little more than I thought I was going to send.

However that does demonstrate what I am thinking, and I think most of the changes are reasonable at this point.

I am very curious how synchronous this all is, because if this code is truly synchronous updating signalfd to handle this class of signal doesn't really make sense.

If the code is not synchronous using force_sig is questionable.

Quoted 8 lines

Eric W. Biederman (7):
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Use dedicated helpers to send signals with si_trapno set
signal: Remove __ARCH_SI_TRAPNO
signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency
signal: Factor force_sig_perf out of perf_sigtrap
signal: Deliver all of the perf_data in si_perf

Quoted 6 lines

Thank you for doing this so quickly -- it looks much cleaner. I'll have a more detailed look next week and also run some tests myself.

At a first glance, you've broken our tests in
tools/testing/selftests/perf_events/ -- needs a
s/si_perf/si_perf.data/, s/si_errno/si_perf.type/

Yeah. I figured I did, but I couldn't figure out where the tests were and I didn't have a lot of time. I just wanted to get this out so we can do as much as reasonable before the ABI starts being actively used by userspace and we can't change it.

Eric

#35 lore raw

Re: [RFC][PATCH 0/3] signal: Move si_trapno into the _si_fault union

Quoted 29 lines · Eric W. Biederman
Quoted 26 lines · Marco Elver
Quoted 13 lines · Eric W. Biederman

Well with 7 patches instead of 3 that was a little more than I thought I was going to send.

However that does demonstrate what I am thinking, and I think most of the changes are reasonable at this point.

I am very curious how synchronous this all is, because if this code is truly synchronous updating signalfd to handle this class of signal doesn't really make sense.

If the code is not synchronous using force_sig is questionable.

Quoted 8 lines

Eric W. Biederman (7):
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Use dedicated helpers to send signals with si_trapno set
signal: Remove __ARCH_SI_TRAPNO
signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency
signal: Factor force_sig_perf out of perf_sigtrap
signal: Deliver all of the perf_data in si_perf

Thank you for doing this so quickly -- it looks much cleaner. I'll have a more detailed look next week and also run some tests myself.

Quoted 8 lines
Quoted 3 lines

At a first glance, you've broken our tests in
tools/testing/selftests/perf_events/ -- needs a
s/si_perf/si_perf.data/, s/si_errno/si_perf.type/

Yeah. I figured I did, but I couldn't figure out where the tests were and I didn't have a lot of time. I just wanted to get this out so we can do as much as reasonable before the ABI starts being actively used by userspace and we can't change it.

No worries, and agreed. I've run tools/testing/selftests/perf_events tests on x86-64 (native + 32-bit compat), and compile-tested x86-32, arm64, arm (with my static asserts), m68k, and sparc64. Some trivial breakages, note comments in other patches.

With the trivial fixes this looks good to me. I'll happily retest v2 when you send it.

Thanks,
-- Marco

#36 lore raw

cover signal: sort out si_trapno and si_perf

This is my attempt to sort out the ABI issues with SIGTRAP TRAP_PERF before any userspace code starts using the new ABI.

The big ideas are:
- Placing the asserts first to prevent unexpected ABI changes
- si_trapno can become an ordinary fault subfield.
- Reworking siginfo so that si_perf_data can be a 64bit field.
- struct signalfd_siginfo is almost full

Marco I have incorporated your static_assert changes and built on them to prevent having unexpected ABI changes.

The field si_trapno is changed to become an ordinary extension of the _sigfault member of siginfo.

The code is refactored a bit and then si_perf_data is made a 64bit, and si_perf_type is made distinct from si_errno.

Finally the signalfd_siginfo fields are removed as they appear to be filling up the structure without userspace actually being able to use them.

v1: https://lkml.kernel.org/r/[email protected]

Eric W. Biederman (9):
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Use dedicated helpers to send signals with si_trapno set
signal: Remove __ARCH_SI_TRAPNO
signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency
signal: Factor force_sig_perf out of perf_sigtrap
signal: Redefine signinfo so 64bit fields are possible
signal: Deliver all of the siginfo perf data in _perf
signalfd: Remove SIL_FAULT_PERF_EVENT fields from signalfd_siginfo

Marco Elver (3):
sparc64: Add compile-time asserts for siginfo_t offsets
arm: Add compile-time asserts for siginfo_t offsets
arm64: Add compile-time asserts for siginfo_t offsets

 arch/alpha/include/uapi/asm/siginfo.h              |   2 -
 arch/alpha/kernel/osf_sys.c                        |   2 +-
 arch/alpha/kernel/signal.c                         |   4 +-
 arch/alpha/kernel/traps.c                          |  24 ++---
 arch/alpha/mm/fault.c                              |   4 +-
 arch/arm/kernel/signal.c                           |  37 +++++++
 arch/arm64/kernel/signal.c                         |  37 +++++++
 arch/arm64/kernel/signal32.c                       |  37 +++++++
 arch/mips/include/uapi/asm/siginfo.h               |   2 -
 arch/sparc/include/uapi/asm/siginfo.h              |   3 -
 arch/sparc/kernel/process_64.c                     |   2 +-
 arch/sparc/kernel/signal32.c                       |  35 +++++++
 arch/sparc/kernel/signal_64.c                      |  34 +++++++
 arch/sparc/kernel/sys_sparc_32.c                   |   2 +-
 arch/sparc/kernel/sys_sparc_64.c                   |   2 +-
 arch/sparc/kernel/traps_32.c                       |  22 ++--
 arch/sparc/kernel/traps_64.c                       |  44 ++++----
 arch/sparc/kernel/unaligned_32.c                   |   2 +-
 arch/sparc/mm/fault_32.c                           |   2 +-
 arch/sparc/mm/fault_64.c                           |   2 +-
 arch/x86/kernel/signal_compat.c                    |  20 ++--
 fs/signalfd.c                                      |  23 ++---
 include/linux/compat.h                             |  38 ++++---
 include/linux/sched/signal.h                       |  13 +--
 include/linux/signal.h                             |   3 +-
 include/uapi/asm-generic/siginfo.h                 |  57 +++++++----
 include/uapi/linux/signalfd.h                      |   4 +-
 kernel/events/core.c                               |  11 +-
 kernel/signal.c                                    | 113 +++++++++++++--------
 .../selftests/perf_events/sigtrap_threads.c        |  12 +--
 30 files changed, 402 insertions(+), 191 deletions(-)
#37 lore raw

PATCH 1/12 sparc64: Add compile-time asserts for siginfo_t offsets

From: Marco Elver <[email protected]>

To help catch ABI breaks at compile-time, add compile-time assertions to verify the siginfo_t layout. Unlike other architectures, sparc64 is special, because it is one of few architectures requiring si_trapno. ABI breaks around that field would only be caught here.

---
 arch/sparc/kernel/signal32.c  | 34 ++++++++++++++++++++++++++++++++++
 arch/sparc/kernel/signal_64.c | 33 +++++++++++++++++++++++++++++++++
 2 files changed, 67 insertions(+)
arch/sparc/kernel/signal32.c +34 -0
index e9695a06492f..778ed5c26d4a 100644
@@ -745,3 +745,37 @@ asmlinkage int do_sys32_sigstack(u32 u_ssptr, u32 u_ossptr, unsigned long sp)
 out:
 	return ret;
 }
+
+/*
+ * Compile-time assertions for siginfo_t offsets. Check NSIG* as well, as
+ * changes likely come with new fields that should be added below.
+ */
+static_assert(NSIGILL	== 11);
+static_assert(NSIGFPE	== 15);
+static_assert(NSIGSEGV	== 9);
+static_assert(NSIGBUS	== 5);
+static_assert(NSIGTRAP	== 6);
+static_assert(NSIGCHLD	== 6);
+static_assert(NSIGSYS	== 2);
+static_assert(offsetof(compat_siginfo_t, si_signo)	== 0x00);
+static_assert(offsetof(compat_siginfo_t, si_errno)	== 0x04);
+static_assert(offsetof(compat_siginfo_t, si_code)	== 0x08);
+static_assert(offsetof(compat_siginfo_t, si_pid)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_uid)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_tid)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_overrun)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_status)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_utime)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_stime)	== 0x1c);
+static_assert(offsetof(compat_siginfo_t, si_value)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_int)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_ptr)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_addr)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_trapno)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_addr_lsb)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_lower)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_upper)	== 0x1c);
+static_assert(offsetof(compat_siginfo_t, si_pkey)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_perf)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_band)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_fd)		== 0x10);
arch/sparc/kernel/signal_64.c +33 -0
index a0eec62c825d..c9bbf5f29078 100644
@@ -556,3 +556,36 @@ void do_notify_resume(struct pt_regs *regs, unsigned long orig_i0, unsigned long
 	user_enter();
 }
 
+/*
+ * Compile-time assertions for siginfo_t offsets. Check NSIG* as well, as
+ * changes likely come with new fields that should be added below.
+ */
+static_assert(NSIGILL	== 11);
+static_assert(NSIGFPE	== 15);
+static_assert(NSIGSEGV	== 9);
+static_assert(NSIGBUS	== 5);
+static_assert(NSIGTRAP	== 6);
+static_assert(NSIGCHLD	== 6);
+static_assert(NSIGSYS	== 2);
+static_assert(offsetof(siginfo_t, si_signo)	== 0x00);
+static_assert(offsetof(siginfo_t, si_errno)	== 0x04);
+static_assert(offsetof(siginfo_t, si_code)	== 0x08);
+static_assert(offsetof(siginfo_t, si_pid)	== 0x10);
+static_assert(offsetof(siginfo_t, si_uid)	== 0x14);
+static_assert(offsetof(siginfo_t, si_tid)	== 0x10);
+static_assert(offsetof(siginfo_t, si_overrun)	== 0x14);
+static_assert(offsetof(siginfo_t, si_status)	== 0x18);
+static_assert(offsetof(siginfo_t, si_utime)	== 0x20);
+static_assert(offsetof(siginfo_t, si_stime)	== 0x28);
+static_assert(offsetof(siginfo_t, si_value)	== 0x18);
+static_assert(offsetof(siginfo_t, si_int)	== 0x18);
+static_assert(offsetof(siginfo_t, si_ptr)	== 0x18);
+static_assert(offsetof(siginfo_t, si_addr)	== 0x10);
+static_assert(offsetof(siginfo_t, si_trapno)	== 0x18);
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x20);
+static_assert(offsetof(siginfo_t, si_lower)	== 0x28);
+static_assert(offsetof(siginfo_t, si_upper)	== 0x30);
+static_assert(offsetof(siginfo_t, si_pkey)	== 0x28);
+static_assert(offsetof(siginfo_t, si_perf)	== 0x20);
+static_assert(offsetof(siginfo_t, si_band)	== 0x10);
+static_assert(offsetof(siginfo_t, si_fd)	== 0x14);
Signature · 1 line
2.30.1
#38 lore raw

PATCH 2/12 arm: Add compile-time asserts for siginfo_t offsets

From: Marco Elver <[email protected]>

To help catch ABI breaks at compile-time, add compile-time assertions to verify the siginfo_t layout.

This could have caught that we cannot portably add 64-bit integers to siginfo_t on 32-bit architectures like Arm before reaching -next:
https://lkml.kernel.org/r/[email protected]

---
 arch/arm/kernel/signal.c | 36 ++++++++++++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)
arch/arm/kernel/signal.c +36 -0
index a3a38d0a4c85..2dac5d2c5cf6 100644
@@ -725,3 +725,39 @@ asmlinkage void do_rseq_syscall(struct pt_regs *regs)
 	rseq_syscall(regs);
 }
 #endif
+
+/*
+ * Compile-time assertions for siginfo_t offsets. Check NSIG* as well, as
+ * changes likely come with new fields that should be added below.
+ */
+static_assert(NSIGILL	== 11);
+static_assert(NSIGFPE	== 15);
+static_assert(NSIGSEGV	== 9);
+static_assert(NSIGBUS	== 5);
+static_assert(NSIGTRAP	== 6);
+static_assert(NSIGCHLD	== 6);
+static_assert(NSIGSYS	== 2);
+static_assert(offsetof(siginfo_t, si_signo)	== 0x00);
+static_assert(offsetof(siginfo_t, si_errno)	== 0x04);
+static_assert(offsetof(siginfo_t, si_code)	== 0x08);
+static_assert(offsetof(siginfo_t, si_pid)	== 0x0c);
+static_assert(offsetof(siginfo_t, si_uid)	== 0x10);
+static_assert(offsetof(siginfo_t, si_tid)	== 0x0c);
+static_assert(offsetof(siginfo_t, si_overrun)	== 0x10);
+static_assert(offsetof(siginfo_t, si_status)	== 0x14);
+static_assert(offsetof(siginfo_t, si_utime)	== 0x18);
+static_assert(offsetof(siginfo_t, si_stime)	== 0x1c);
+static_assert(offsetof(siginfo_t, si_value)	== 0x14);
+static_assert(offsetof(siginfo_t, si_int)	== 0x14);
+static_assert(offsetof(siginfo_t, si_ptr)	== 0x14);
+static_assert(offsetof(siginfo_t, si_addr)	== 0x0c);
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x10);
+static_assert(offsetof(siginfo_t, si_lower)	== 0x14);
+static_assert(offsetof(siginfo_t, si_upper)	== 0x18);
+static_assert(offsetof(siginfo_t, si_pkey)	== 0x14);
+static_assert(offsetof(siginfo_t, si_perf)	== 0x10);
+static_assert(offsetof(siginfo_t, si_band)	== 0x0c);
+static_assert(offsetof(siginfo_t, si_fd)	== 0x10);
+static_assert(offsetof(siginfo_t, si_call_addr)	== 0x0c);
+static_assert(offsetof(siginfo_t, si_syscall)	== 0x10);
+static_assert(offsetof(siginfo_t, si_arch)	== 0x14);
Signature · 1 line
2.30.1
#39 lore raw

PATCH 3/12 arm64: Add compile-time asserts for siginfo_t offsets

In reply to Eric W. Beiderman

From: Marco Elver <[email protected]>

To help catch ABI breaks at compile-time, add compile-time assertions to verify the siginfo_t layout.

---
 arch/arm64/kernel/signal.c   | 36 ++++++++++++++++++++++++++++++++++++
 arch/arm64/kernel/signal32.c | 36 ++++++++++++++++++++++++++++++++++++
 2 files changed, 72 insertions(+)
arch/arm64/kernel/signal.c +36 -0
index 6237486ff6bb..af8bd2af1298 100644
@@ -973,3 +973,39 @@ void __init minsigstksz_setup(void)
 		round_up(sizeof(struct frame_record), 16) +
 		16; /* max alignment padding */
 }
+
+/*
+ * Compile-time assertions for siginfo_t offsets. Check NSIG* as well, as
+ * changes likely come with new fields that should be added below.
+ */
+static_assert(NSIGILL	== 11);
+static_assert(NSIGFPE	== 15);
+static_assert(NSIGSEGV	== 9);
+static_assert(NSIGBUS	== 5);
+static_assert(NSIGTRAP	== 6);
+static_assert(NSIGCHLD	== 6);
+static_assert(NSIGSYS	== 2);
+static_assert(offsetof(siginfo_t, si_signo)	== 0x00);
+static_assert(offsetof(siginfo_t, si_errno)	== 0x04);
+static_assert(offsetof(siginfo_t, si_code)	== 0x08);
+static_assert(offsetof(siginfo_t, si_pid)	== 0x10);
+static_assert(offsetof(siginfo_t, si_uid)	== 0x14);
+static_assert(offsetof(siginfo_t, si_tid)	== 0x10);
+static_assert(offsetof(siginfo_t, si_overrun)	== 0x14);
+static_assert(offsetof(siginfo_t, si_status)	== 0x18);
+static_assert(offsetof(siginfo_t, si_utime)	== 0x20);
+static_assert(offsetof(siginfo_t, si_stime)	== 0x28);
+static_assert(offsetof(siginfo_t, si_value)	== 0x18);
+static_assert(offsetof(siginfo_t, si_int)	== 0x18);
+static_assert(offsetof(siginfo_t, si_ptr)	== 0x18);
+static_assert(offsetof(siginfo_t, si_addr)	== 0x10);
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x18);
+static_assert(offsetof(siginfo_t, si_lower)	== 0x20);
+static_assert(offsetof(siginfo_t, si_upper)	== 0x28);
+static_assert(offsetof(siginfo_t, si_pkey)	== 0x20);
+static_assert(offsetof(siginfo_t, si_perf)	== 0x18);
+static_assert(offsetof(siginfo_t, si_band)	== 0x10);
+static_assert(offsetof(siginfo_t, si_fd)	== 0x18);
+static_assert(offsetof(siginfo_t, si_call_addr)	== 0x10);
+static_assert(offsetof(siginfo_t, si_syscall)	== 0x18);
+static_assert(offsetof(siginfo_t, si_arch)	== 0x1c);
arch/arm64/kernel/signal32.c +36 -0
index 2f507f565c48..b6afb646515f 100644
@@ -457,3 +457,39 @@ void compat_setup_restart_syscall(struct pt_regs *regs)
 {
        regs->regs[7] = __NR_compat_restart_syscall;
 }
+
+/*
+ * Compile-time assertions for siginfo_t offsets. Check NSIG* as well, as
+ * changes likely come with new fields that should be added below.
+ */
+static_assert(NSIGILL	== 11);
+static_assert(NSIGFPE	== 15);
+static_assert(NSIGSEGV	== 9);
+static_assert(NSIGBUS	== 5);
+static_assert(NSIGTRAP	== 6);
+static_assert(NSIGCHLD	== 6);
+static_assert(NSIGSYS	== 2);
+static_assert(offsetof(compat_siginfo_t, si_signo)	== 0x00);
+static_assert(offsetof(compat_siginfo_t, si_errno)	== 0x04);
+static_assert(offsetof(compat_siginfo_t, si_code)	== 0x08);
+static_assert(offsetof(compat_siginfo_t, si_pid)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_uid)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_tid)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_overrun)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_status)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_utime)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_stime)	== 0x1c);
+static_assert(offsetof(compat_siginfo_t, si_value)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_int)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_ptr)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_addr)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_addr_lsb)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_lower)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_upper)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_pkey)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_perf)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_band)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_fd)		== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_call_addr)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_syscall)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_arch)	== 0x14);
Signature · 1 line
2.30.1
#40 lore raw

PATCH 4/12 siginfo: Move si_trapno inside the union inside _si_fault

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

It turns out that linux uses si_trapno very sparingly, and as such it can be considered extra information for a very narrow selection of signals, rather than information that is present with every fault reported in siginfo.

As such move si_trapno inside the union inside of _si_fault. This results in no change in placement, and makes it eaiser
to extend _si_fault in the future as this reduces the number of special cases. In particular with si_trapno included in the union it is no longer a concern that the union must be pointer alligned on most architectures because the union followes immediately after si_addr which is a pointer.

This change results in a difference in siginfo field placement on sparc and alpha for the fields si_addr_lsb, si_lower, si_upper, si_pkey, and si_perf. These architectures do not implement the signals that would use si_addr_lsb, si_lower, si_upper, si_pkey, and si_perf. Further these architecture have not yet implemented the userspace that would use si_perf.

The point of this change is in fact to correct these placement issues before sparc or alpha grow userspace that cares. This change was discussed[1] and the agreement is that this change is currently safe.

[1]: https://lkml.kernel.org/r/CAK8P3a0+uKYwL1NhY6Hvtieghba2hKYGD6hcKx5n8=4Gtt+pHA@mail.gmail.com

v1: https://lkml.kernel.org/r/[email protected]

---
 arch/sparc/kernel/signal32.c       | 10 +++++-----
 arch/sparc/kernel/signal_64.c      | 10 +++++-----
 arch/x86/kernel/signal_compat.c    |  3 +++
 include/linux/compat.h             |  5 ++---
 include/uapi/asm-generic/siginfo.h |  7 ++-----
 kernel/signal.c                    |  1 +
 6 files changed, 18 insertions(+), 18 deletions(-)
arch/sparc/kernel/signal32.c +5 -5
index 778ed5c26d4a..73fd8700df3e 100644
@@ -772,10 +772,10 @@ static_assert(offsetof(compat_siginfo_t, si_int)	== 0x14);
 static_assert(offsetof(compat_siginfo_t, si_ptr)	== 0x14);
 static_assert(offsetof(compat_siginfo_t, si_addr)	== 0x0c);
 static_assert(offsetof(compat_siginfo_t, si_trapno)	== 0x10);
-static_assert(offsetof(compat_siginfo_t, si_addr_lsb)	== 0x14);
-static_assert(offsetof(compat_siginfo_t, si_lower)	== 0x18);
-static_assert(offsetof(compat_siginfo_t, si_upper)	== 0x1c);
-static_assert(offsetof(compat_siginfo_t, si_pkey)	== 0x18);
-static_assert(offsetof(compat_siginfo_t, si_perf)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_addr_lsb)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_lower)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_upper)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_pkey)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_perf)	== 0x10);
 static_assert(offsetof(compat_siginfo_t, si_band)	== 0x0c);
 static_assert(offsetof(compat_siginfo_t, si_fd)		== 0x10);
arch/sparc/kernel/signal_64.c +5 -5
index c9bbf5f29078..17913daa66c6 100644
@@ -582,10 +582,10 @@ static_assert(offsetof(siginfo_t, si_int)	== 0x18);
 static_assert(offsetof(siginfo_t, si_ptr)	== 0x18);
 static_assert(offsetof(siginfo_t, si_addr)	== 0x10);
 static_assert(offsetof(siginfo_t, si_trapno)	== 0x18);
-static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x20);
-static_assert(offsetof(siginfo_t, si_lower)	== 0x28);
-static_assert(offsetof(siginfo_t, si_upper)	== 0x30);
-static_assert(offsetof(siginfo_t, si_pkey)	== 0x28);
-static_assert(offsetof(siginfo_t, si_perf)	== 0x20);
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x18);
+static_assert(offsetof(siginfo_t, si_lower)	== 0x20);
+static_assert(offsetof(siginfo_t, si_upper)	== 0x28);
+static_assert(offsetof(siginfo_t, si_pkey)	== 0x20);
+static_assert(offsetof(siginfo_t, si_perf)	== 0x18);
 static_assert(offsetof(siginfo_t, si_band)	== 0x10);
 static_assert(offsetof(siginfo_t, si_fd)	== 0x14);
arch/x86/kernel/signal_compat.c +3 -0
index 0e5d0a7e203b..a9fcabd8a5e5 100644
@@ -127,6 +127,9 @@ static inline void signal_compat_build_tests(void)
 	BUILD_BUG_ON(offsetof(siginfo_t, si_addr) != 0x10);
 	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_addr) != 0x0C);
 
+	BUILD_BUG_ON(offsetof(siginfo_t, si_trapno) != 0x18);
+	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_trapno) != 0x10);
+
 	BUILD_BUG_ON(offsetof(siginfo_t, si_addr_lsb) != 0x18);
 	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_addr_lsb) != 0x10);
 
include/linux/compat.h +2 -3
index f0d2dd35d408..6af7bef15e94 100644
@@ -214,12 +214,11 @@ typedef struct compat_siginfo {
 		/* SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT */
 		struct {
 			compat_uptr_t _addr;	/* faulting insn/memory ref. */
-#ifdef __ARCH_SI_TRAPNO
-			int _trapno;	/* TRAP # which caused the signal */
-#endif
 #define __COMPAT_ADDR_BND_PKEY_PAD  (__alignof__(compat_uptr_t) < sizeof(short) ? \
 				     sizeof(short) : __alignof__(compat_uptr_t))
 			union {
+				/* used on alpha and sparc */
+				int _trapno;	/* TRAP # which caused the signal */
 				/*
 				 * used when si_code=BUS_MCEERR_AR or
 				 * used when si_code=BUS_MCEERR_AO
include/uapi/asm-generic/siginfo.h +2 -5
index 03d6f6d2c1fe..e663bf117b46 100644
@@ -63,9 +63,6 @@ union __sifields {
 	/* SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT */
 	struct {
 		void __user *_addr; /* faulting insn/memory ref. */
-#ifdef __ARCH_SI_TRAPNO
-		int _trapno;	/* TRAP # which caused the signal */
-#endif
 #ifdef __ia64__
 		int _imm;		/* immediate value for "break" */
 		unsigned int _flags;	/* see ia64 si_flags */
@@ -75,6 +72,8 @@ union __sifields {
 #define __ADDR_BND_PKEY_PAD  (__alignof__(void *) < sizeof(short) ? \
 			      sizeof(short) : __alignof__(void *))
 		union {
+			/* used on alpha and sparc */
+			int _trapno;	/* TRAP # which caused the signal */
 			/*
 			 * used when si_code=BUS_MCEERR_AR or
 			 * used when si_code=BUS_MCEERR_AO
@@ -150,9 +149,7 @@ typedef struct siginfo {
 #define si_int		_sifields._rt._sigval.sival_int
 #define si_ptr		_sifields._rt._sigval.sival_ptr
 #define si_addr		_sifields._sigfault._addr
-#ifdef __ARCH_SI_TRAPNO
 #define si_trapno	_sifields._sigfault._trapno
-#endif
 #define si_addr_lsb	_sifields._sigfault._addr_lsb
 #define si_lower	_sifields._sigfault._addr_bnd._lower
 #define si_upper	_sifields._sigfault._addr_bnd._upper
kernel/signal.c +1 -0
index c3017aa8024a..65888aec65a0 100644
@@ -4607,6 +4607,7 @@ static inline void siginfo_buildtime_checks(void)
 
 	/* sigfault */
 	CHECK_OFFSET(si_addr);
+	CHECK_OFFSET(si_trapno);
 	CHECK_OFFSET(si_addr_lsb);
 	CHECK_OFFSET(si_lower);
 	CHECK_OFFSET(si_upper);
Signature · 1 line
2.30.1
#41 lore raw

PATCH 5/12 signal: Implement SIL_FAULT_TRAPNO

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Now that si_trapno is part of the union in _si_fault and available on all architectures, add SIL_FAULT_TRAPNO and update siginfo_layout to return SIL_FAULT_TRAPNO when si_trapno is actually used.

Update the code that uses siginfo_layout to deal with SIL_FAULT_TRAPNO and have the same code ignore si_trapno in in all other cases.

v1: https://lkml.kernel.org/r/[email protected]

---
 fs/signalfd.c          |  8 +++-----
 include/linux/signal.h |  1 +
 kernel/signal.c        | 37 +++++++++++++++----------------------
 3 files changed, 19 insertions(+), 27 deletions(-)
fs/signalfd.c +3 -5
index 040a1142915f..e87e59581653 100644
@@ -123,15 +123,13 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		 */
 	case SIL_FAULT:
 		new.ssi_addr = (long) kinfo->si_addr;
-#ifdef __ARCH_SI_TRAPNO
+		break;
+	case SIL_FAULT_TRAPNO:
+		new.ssi_addr = (long) kinfo->si_addr;
 		new.ssi_trapno = kinfo->si_trapno;
-#endif
 		break;
 	case SIL_FAULT_MCEERR:
 		new.ssi_addr = (long) kinfo->si_addr;
-#ifdef __ARCH_SI_TRAPNO
-		new.ssi_trapno = kinfo->si_trapno;
-#endif
 		new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
 		break;
 	case SIL_PERF_EVENT:
include/linux/signal.h +1 -0
index 1e98548d7cf6..5160fd45e5ca 100644
@@ -40,6 +40,7 @@ enum siginfo_layout {
 	SIL_TIMER,
 	SIL_POLL,
 	SIL_FAULT,
+	SIL_FAULT_TRAPNO,
 	SIL_FAULT_MCEERR,
 	SIL_FAULT_BNDERR,
 	SIL_FAULT_PKUERR,
kernel/signal.c +15 -22
index 65888aec65a0..3d3ba7949788 100644
@@ -1194,6 +1194,7 @@ static inline bool has_si_pid_and_uid(struct kernel_siginfo *info)
 	case SIL_TIMER:
 	case SIL_POLL:
 	case SIL_FAULT:
+	case SIL_FAULT_TRAPNO:
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
@@ -2527,6 +2528,7 @@ static void hide_si_addr_tag_bits(struct ksignal *ksig)
 {
 	switch (siginfo_layout(ksig->sig, ksig->info.si_code)) {
 	case SIL_FAULT:
+	case SIL_FAULT_TRAPNO:
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
@@ -3206,6 +3208,13 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
 			if ((sig == SIGBUS) &&
 			    (si_code >= BUS_MCEERR_AR) && (si_code <= BUS_MCEERR_AO))
 				layout = SIL_FAULT_MCEERR;
+			else if (IS_ENABLED(CONFIG_ALPHA) &&
+				 ((sig == SIGFPE) ||
+				  ((sig == SIGTRAP) && (si_code == TRAP_UNK))))
+				layout = SIL_FAULT_TRAPNO;
+			else if (IS_ENABLED(CONFIG_SPARC) &&
+				 (sig == SIGILL) && (si_code == ILL_ILLTRP))
+				layout = SIL_FAULT_TRAPNO;
 			else if ((sig == SIGSEGV) && (si_code == SEGV_BNDERR))
 				layout = SIL_FAULT_BNDERR;
 #ifdef SEGV_PKUERR
@@ -3317,30 +3326,22 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 		break;
 	case SIL_FAULT:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+		break;
+	case SIL_FAULT_TRAPNO:
+		to->si_addr = ptr_to_compat(from->si_addr);
 		to->si_trapno = from->si_trapno;
-#endif
 		break;
 	case SIL_FAULT_MCEERR:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_addr_lsb = from->si_addr_lsb;
 		break;
 	case SIL_FAULT_BNDERR:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_lower = ptr_to_compat(from->si_lower);
 		to->si_upper = ptr_to_compat(from->si_upper);
 		break;
 	case SIL_FAULT_PKUERR:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_pkey = from->si_pkey;
 		break;
 	case SIL_PERF_EVENT:
@@ -3401,30 +3402,22 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 		break;
 	case SIL_FAULT:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+		break;
+	case SIL_FAULT_TRAPNO:
+		to->si_addr = compat_ptr(from->si_addr);
 		to->si_trapno = from->si_trapno;
-#endif
 		break;
 	case SIL_FAULT_MCEERR:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_addr_lsb = from->si_addr_lsb;
 		break;
 	case SIL_FAULT_BNDERR:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_lower = compat_ptr(from->si_lower);
 		to->si_upper = compat_ptr(from->si_upper);
 		break;
 	case SIL_FAULT_PKUERR:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_pkey = from->si_pkey;
 		break;
 	case SIL_PERF_EVENT:
Signature · 1 line
2.30.1
#42 lore raw

PATCH 6/12 signal: Use dedicated helpers to send signals with si_trapno set

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Now that si_trapno is no longer expected to be present for every fault reported using siginfo on alpha and sparc remove the trapno parameter from force_sig_fault, force_sig_fault_to_task and send_sig_fault.

Add two new helpers force_sig_fault_trapno and send_sig_fautl_trapno for those signals where trapno is expected to be set.

v1: https://lkml.kernel.org/r/[email protected]

---
 arch/alpha/kernel/osf_sys.c      |  2 +-
 arch/alpha/kernel/signal.c       |  4 +--
 arch/alpha/kernel/traps.c        | 24 ++++++++---------
 arch/alpha/mm/fault.c            |  4 +--
 arch/sparc/kernel/process_64.c   |  2 +-
 arch/sparc/kernel/sys_sparc_32.c |  2 +-
 arch/sparc/kernel/sys_sparc_64.c |  2 +-
 arch/sparc/kernel/traps_32.c     | 22 ++++++++--------
 arch/sparc/kernel/traps_64.c     | 44 ++++++++++++++------------------
 arch/sparc/kernel/unaligned_32.c |  2 +-
 arch/sparc/mm/fault_32.c         |  2 +-
 arch/sparc/mm/fault_64.c         |  2 +-
 include/linux/sched/signal.h     | 12 +++------
 kernel/signal.c                  | 41 +++++++++++++++++++++--------
 14 files changed, 88 insertions(+), 77 deletions(-)
arch/alpha/kernel/osf_sys.c +1 -1
index d5367a1c6300..80c5d7fbe66a 100644
@@ -878,7 +878,7 @@ SYSCALL_DEFINE5(osf_setsysinfo, unsigned long, op, void __user *, buffer,
 
 			send_sig_fault(SIGFPE, si_code,
 				       (void __user *)NULL,  /* FIXME */
-				       0, current);
+				       current);
  		}
 		return 0;
 	}
arch/alpha/kernel/signal.c +2 -2
index 948b89789da8..bc077babafab 100644
@@ -219,7 +219,7 @@ do_sigreturn(struct sigcontext __user *sc)
 
 	/* Send SIGTRAP if we're single-stepping: */
 	if (ptrace_cancel_bpt (current)) {
-		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc, 0,
+		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc,
 			       current);
 	}
 	return;
@@ -247,7 +247,7 @@ do_rt_sigreturn(struct rt_sigframe __user *frame)
 
 	/* Send SIGTRAP if we're single-stepping: */
 	if (ptrace_cancel_bpt (current)) {
-		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc, 0,
+		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc,
 			       current);
 	}
 	return;
arch/alpha/kernel/traps.c +12 -12
index 921d4b6e4d95..0dddf9ecc1f4 100644
@@ -227,7 +227,7 @@ do_entArith(unsigned long summary, unsigned long write_mask,
 	}
 	die_if_kernel("Arithmetic fault", regs, 0, NULL);
 
-	send_sig_fault(SIGFPE, si_code, (void __user *) regs->pc, 0, current);
+	send_sig_fault_trapno(SIGFPE, si_code, (void __user *) regs->pc, 0, current);
 }
 
 asmlinkage void
@@ -268,12 +268,12 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 			regs->pc -= 4;	/* make pc point to former bpt */
 		}
 
-		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc, 0,
+		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc,
 			       current);
 		return;
 
 	      case 1: /* bugcheck */
-		send_sig_fault(SIGTRAP, TRAP_UNK, (void __user *) regs->pc, 0,
+		send_sig_fault(SIGTRAP, TRAP_UNK, (void __user *) regs->pc,
 			       current);
 		return;
 		
@@ -335,8 +335,8 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 			break;
 		}
 
-		send_sig_fault(signo, code, (void __user *) regs->pc, regs->r16,
-			       current);
+		send_sig_fault_trapno(signo, code, (void __user *) regs->pc,
+				      regs->r16, current);
 		return;
 
 	      case 4: /* opDEC */
@@ -360,9 +360,9 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 			if (si_code == 0)
 				return;
 			if (si_code > 0) {
-				send_sig_fault(SIGFPE, si_code,
-					       (void __user *) regs->pc, 0,
-					       current);
+				send_sig_fault_trapno(SIGFPE, si_code,
+						      (void __user *) regs->pc,
+						      0, current);
 				return;
 			}
 		}
@@ -387,7 +387,7 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 		      ;
 	}
 
-	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, 0, current);
+	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, current);
 }
 
 /* There is an ifdef in the PALcode in MILO that enables a 
@@ -402,7 +402,7 @@ do_entDbg(struct pt_regs *regs)
 {
 	die_if_kernel("Instruction fault", regs, 0, NULL);
 
-	force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, 0);
+	force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc);
 }
 
 
@@ -964,12 +964,12 @@ do_entUnaUser(void __user * va, unsigned long opcode,
 			si_code = SEGV_MAPERR;
 		mmap_read_unlock(mm);
 	}
-	send_sig_fault(SIGSEGV, si_code, va, 0, current);
+	send_sig_fault(SIGSEGV, si_code, va, current);
 	return;
 
 give_sigbus:
 	regs->pc -= 4;
-	send_sig_fault(SIGBUS, BUS_ADRALN, va, 0, current);
+	send_sig_fault(SIGBUS, BUS_ADRALN, va, current);
 	return;
 }
 
arch/alpha/mm/fault.c +2 -2
index 09172f017efc..eee5102c3d88 100644
@@ -219,13 +219,13 @@ do_page_fault(unsigned long address, unsigned long mmcsr,
 	mmap_read_unlock(mm);
 	/* Send a sigbus, regardless of whether we were in kernel
 	   or user mode.  */
-	force_sig_fault(SIGBUS, BUS_ADRERR, (void __user *) address, 0);
+	force_sig_fault(SIGBUS, BUS_ADRERR, (void __user *) address);
 	if (!user_mode(regs))
 		goto no_context;
 	return;
 
  do_sigsegv:
-	force_sig_fault(SIGSEGV, si_code, (void __user *) address, 0);
+	force_sig_fault(SIGSEGV, si_code, (void __user *) address);
 	return;
 
 #ifdef CONFIG_ALPHA_LARGE_VMALLOC
arch/sparc/kernel/process_64.c +1 -1
index 7afd0a859a78..29e67854d5a4 100644
@@ -518,7 +518,7 @@ void synchronize_user_stack(void)
 
 static void stack_unaligned(unsigned long sp)
 {
-	force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) sp, 0);
+	force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) sp);
 }
 
 static const char uwfault32[] = KERN_INFO \
arch/sparc/kernel/sys_sparc_32.c +1 -1
index be77538bc038..082a551897ed 100644
@@ -151,7 +151,7 @@ sparc_breakpoint (struct pt_regs *regs)
 #ifdef DEBUG_SPARC_BREAKPOINT
         printk ("TRAP: Entering kernel PC=%x, nPC=%x\n", regs->pc, regs->npc);
 #endif
-	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc, 0);
+	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc);
 
 #ifdef DEBUG_SPARC_BREAKPOINT
 	printk ("TRAP: Returning to space: PC=%x nPC=%x\n", regs->pc, regs->npc);
arch/sparc/kernel/sys_sparc_64.c +1 -1
index 6b92fadb6ec7..1e9a9e016237 100644
@@ -514,7 +514,7 @@ asmlinkage void sparc_breakpoint(struct pt_regs *regs)
 #ifdef DEBUG_SPARC_BREAKPOINT
         printk ("TRAP: Entering kernel PC=%lx, nPC=%lx\n", regs->tpc, regs->tnpc);
 #endif
-	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->tpc, 0);
+	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->tpc);
 #ifdef DEBUG_SPARC_BREAKPOINT
 	printk ("TRAP: Returning to space: PC=%lx nPC=%lx\n", regs->tpc, regs->tnpc);
 #endif
arch/sparc/kernel/traps_32.c +11 -11
index 247a0d9683b2..5630e5a395e0 100644
@@ -102,8 +102,8 @@ void do_hw_interrupt(struct pt_regs *regs, unsigned long type)
 	if(regs->psr & PSR_PS)
 		die_if_kernel("Kernel bad trap", regs);
 
-	force_sig_fault(SIGILL, ILL_ILLTRP,
-			(void __user *)regs->pc, type - 0x80);
+	force_sig_fault_trapno(SIGILL, ILL_ILLTRP,
+			       (void __user *)regs->pc, type - 0x80);
 }
 
 void do_illegal_instruction(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -116,7 +116,7 @@ void do_illegal_instruction(struct pt_regs *regs, unsigned long pc, unsigned lon
 	       regs->pc, *(unsigned long *)regs->pc);
 #endif
 
-	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc, current);
 }
 
 void do_priv_instruction(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -124,7 +124,7 @@ void do_priv_instruction(struct pt_regs *regs, unsigned long pc, unsigned long n
 {
 	if(psr & PSR_PS)
 		die_if_kernel("Penguin instruction from Penguin mode??!?!", regs);
-	send_sig_fault(SIGILL, ILL_PRVOPC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_PRVOPC, (void __user *)pc, current);
 }
 
 /* XXX User may want to be allowed to do this. XXX */
@@ -145,7 +145,7 @@ void do_memaccess_unaligned(struct pt_regs *regs, unsigned long pc, unsigned lon
 #endif
 	send_sig_fault(SIGBUS, BUS_ADRALN,
 		       /* FIXME: Should dig out mna address */ (void *)0,
-		       0, current);
+		       current);
 }
 
 static unsigned long init_fsr = 0x0UL;
@@ -291,7 +291,7 @@ void do_fpe_trap(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 		else if (fsr & 0x01)
 			code = FPE_FLTRES;
 	}
-	send_sig_fault(SIGFPE, code, (void __user *)pc, 0, fpt);
+	send_sig_fault(SIGFPE, code, (void __user *)pc, fpt);
 #ifndef CONFIG_SMP
 	last_task_used_math = NULL;
 #endif
@@ -305,7 +305,7 @@ void handle_tag_overflow(struct pt_regs *regs, unsigned long pc, unsigned long n
 {
 	if(psr & PSR_PS)
 		die_if_kernel("Penguin overflow trap from kernel mode", regs);
-	send_sig_fault(SIGEMT, EMT_TAGOVF, (void __user *)pc, 0, current);
+	send_sig_fault(SIGEMT, EMT_TAGOVF, (void __user *)pc, current);
 }
 
 void handle_watchpoint(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -327,13 +327,13 @@ void handle_reg_access(struct pt_regs *regs, unsigned long pc, unsigned long npc
 	printk("Register Access Exception at PC %08lx NPC %08lx PSR %08lx\n",
 	       pc, npc, psr);
 #endif
-	force_sig_fault(SIGBUS, BUS_OBJERR, (void __user *)pc, 0);
+	force_sig_fault(SIGBUS, BUS_OBJERR, (void __user *)pc);
 }
 
 void handle_cp_disabled(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 			unsigned long psr)
 {
-	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, current);
 }
 
 void handle_cp_exception(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -343,13 +343,13 @@ void handle_cp_exception(struct pt_regs *regs, unsigned long pc, unsigned long n
 	printk("Co-Processor Exception at PC %08lx NPC %08lx PSR %08lx\n",
 	       pc, npc, psr);
 #endif
-	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, current);
 }
 
 void handle_hw_divzero(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 		       unsigned long psr)
 {
-	send_sig_fault(SIGFPE, FPE_INTDIV, (void __user *)pc, 0, current);
+	send_sig_fault(SIGFPE, FPE_INTDIV, (void __user *)pc, current);
 }
 
 #ifdef CONFIG_DEBUG_BUGVERBOSE
arch/sparc/kernel/traps_64.c +19 -25

144 lines not on this page · raw message

arch/sparc/kernel/unaligned_32.c +1 -1
index ef5c5207c9ff..455f0258c745 100644
@@ -278,5 +278,5 @@ asmlinkage void user_unaligned_trap(struct pt_regs *regs, unsigned int insn)
 {
 	send_sig_fault(SIGBUS, BUS_ADRALN,
 		       (void __user *)safe_compute_effective_address(regs, insn),
-		       0, current);
+		       current);
 }
arch/sparc/mm/fault_32.c +1 -1
index de2031c2b2d7..fa858626b85b 100644
@@ -83,7 +83,7 @@ static void __do_fault_siginfo(int code, int sig, struct pt_regs *regs,
 		show_signal_msg(regs, sig, code,
 				addr, current);
 
-	force_sig_fault(sig, code, (void __user *) addr, 0);
+	force_sig_fault(sig, code, (void __user *) addr);
 }
 
 static unsigned long compute_si_addr(struct pt_regs *regs, int text_fault)
arch/sparc/mm/fault_64.c +1 -1
index 0a6bcc85fba7..9a9652a15fed 100644
@@ -176,7 +176,7 @@ static void do_fault_siginfo(int code, int sig, struct pt_regs *regs,
 	if (unlikely(show_unhandled_signals))
 		show_signal_msg(regs, sig, code, addr, current);
 
-	force_sig_fault(sig, code, (void __user *) addr, 0);
+	force_sig_fault(sig, code, (void __user *) addr);
 }
 
 static unsigned int get_fault_insn(struct pt_regs *regs, unsigned int insn)
include/linux/sched/signal.h +4 -8
index 3f6a0fcaa10c..7daa425f3055 100644
@@ -298,11 +298,6 @@ static inline void kernel_signal_stop(void)
 
 	schedule();
 }
-#ifdef __ARCH_SI_TRAPNO
-# define ___ARCH_SI_TRAPNO(_a1) , _a1
-#else
-# define ___ARCH_SI_TRAPNO(_a1)
-#endif
 #ifdef __ia64__
 # define ___ARCH_SI_IA64(_a1, _a2, _a3) , _a1, _a2, _a3
 #else
@@ -310,14 +305,11 @@ static inline void kernel_signal_stop(void)
 #endif
 
 int force_sig_fault_to_task(int sig, int code, void __user *addr
-	___ARCH_SI_TRAPNO(int trapno)
 	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
 	, struct task_struct *t);
 int force_sig_fault(int sig, int code, void __user *addr
-	___ARCH_SI_TRAPNO(int trapno)
 	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr));
 int send_sig_fault(int sig, int code, void __user *addr
-	___ARCH_SI_TRAPNO(int trapno)
 	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
 	, struct task_struct *t);
 
@@ -327,6 +319,10 @@ int send_sig_mceerr(int code, void __user *, short, struct task_struct *);
 int force_sig_bnderr(void __user *addr, void __user *lower, void __user *upper);
 int force_sig_pkuerr(void __user *addr, u32 pkey);
 
+int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno);
+int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
+			  struct task_struct *task);
+
 int force_sig_ptrace_errno_trap(int errno, void __user *addr);
 
 extern int send_sig_info(int, struct kernel_siginfo *, struct task_struct *);
kernel/signal.c +31 -10

78 lines not on this page · raw message

Signature · 1 line
2.30.1
#43 lore raw

PATCH 7/12 signal: Remove __ARCH_SI_TRAPNO

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Now that this define is no longer used remove it from the kernel.

v1: https://lkml.kernel.org/r/[email protected]

---
 arch/alpha/include/uapi/asm/siginfo.h | 2 --
 arch/mips/include/uapi/asm/siginfo.h  | 2 --
 arch/sparc/include/uapi/asm/siginfo.h | 3 ---
 3 files changed, 7 deletions(-)
arch/alpha/include/uapi/asm/siginfo.h +0 -2
index 6e1a2af2f962..e08eae88182b 100644
@@ -2,8 +2,6 @@
 #ifndef _ALPHA_SIGINFO_H
 #define _ALPHA_SIGINFO_H
 
-#define __ARCH_SI_TRAPNO
-
 #include <asm-generic/siginfo.h>
 
 #endif
arch/mips/include/uapi/asm/siginfo.h +0 -2
index c34c7eef0a1c..8cb8bd061a68 100644
@@ -10,9 +10,7 @@
 #ifndef _UAPI_ASM_SIGINFO_H
 #define _UAPI_ASM_SIGINFO_H
 
-
 #define __ARCH_SIGEV_PREAMBLE_SIZE (sizeof(long) + 2*sizeof(int))
-#undef __ARCH_SI_TRAPNO /* exception code needs to fill this ...  */
 
 #define __ARCH_HAS_SWAPPED_SIGINFO
 
arch/sparc/include/uapi/asm/siginfo.h +0 -3
index 68bdde4c2a2e..0e7c27522aed 100644
@@ -8,9 +8,6 @@
 
 #endif /* defined(__sparc__) && defined(__arch64__) */
 
-
-#define __ARCH_SI_TRAPNO
-
 #include <asm-generic/siginfo.h>
 
 
Signature · 1 line
2.30.1
#44 lore raw

PATCH 8/12 signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

It helps to know which part of the siginfo structure the siginfo_layout value is talking about.

v1: https://lkml.kernel.org/r/[email protected]

---
 fs/signalfd.c          |  2 +-
 include/linux/signal.h |  2 +-
 kernel/signal.c        | 10 +++++-----
 3 files changed, 7 insertions(+), 7 deletions(-)
fs/signalfd.c +1 -1
index e87e59581653..83130244f653 100644
@@ -132,7 +132,7 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		new.ssi_addr = (long) kinfo->si_addr;
 		new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
 		break;
-	case SIL_PERF_EVENT:
+	case SIL_FAULT_PERF_EVENT:
 		new.ssi_addr = (long) kinfo->si_addr;
 		new.ssi_perf = kinfo->si_perf;
 		break;
include/linux/signal.h +1 -1
index 5160fd45e5ca..ed896d790e46 100644
@@ -44,7 +44,7 @@ enum siginfo_layout {
 	SIL_FAULT_MCEERR,
 	SIL_FAULT_BNDERR,
 	SIL_FAULT_PKUERR,
-	SIL_PERF_EVENT,
+	SIL_FAULT_PERF_EVENT,
 	SIL_CHLD,
 	SIL_RT,
 	SIL_SYS,
kernel/signal.c +5 -5
index 7eaa8d84db4c..697c5fe58db8 100644
@@ -1198,7 +1198,7 @@ static inline bool has_si_pid_and_uid(struct kernel_siginfo *info)
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
-	case SIL_PERF_EVENT:
+	case SIL_FAULT_PERF_EVENT:
 	case SIL_SYS:
 		ret = false;
 		break;
@@ -2553,7 +2553,7 @@ static void hide_si_addr_tag_bits(struct ksignal *ksig)
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
-	case SIL_PERF_EVENT:
+	case SIL_FAULT_PERF_EVENT:
 		ksig->info.si_addr = arch_untagged_si_addr(
 			ksig->info.si_addr, ksig->sig, ksig->info.si_code);
 		break;
@@ -3243,7 +3243,7 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
 				layout = SIL_FAULT_PKUERR;
 #endif
 			else if ((sig == SIGTRAP) && (si_code == TRAP_PERF))
-				layout = SIL_PERF_EVENT;
+				layout = SIL_FAULT_PERF_EVENT;
 		}
 		else if (si_code <= NSIGPOLL)
 			layout = SIL_POLL;
@@ -3365,7 +3365,7 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 		to->si_addr = ptr_to_compat(from->si_addr);
 		to->si_pkey = from->si_pkey;
 		break;
-	case SIL_PERF_EVENT:
+	case SIL_FAULT_PERF_EVENT:
 		to->si_addr = ptr_to_compat(from->si_addr);
 		to->si_perf = from->si_perf;
 		break;
@@ -3441,7 +3441,7 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 		to->si_addr = compat_ptr(from->si_addr);
 		to->si_pkey = from->si_pkey;
 		break;
-	case SIL_PERF_EVENT:
+	case SIL_FAULT_PERF_EVENT:
 		to->si_addr = compat_ptr(from->si_addr);
 		to->si_perf = from->si_perf;
 		break;
Signature · 1 line
2.30.1
#45 lore raw

PATCH 9/12 signal: Factor force_sig_perf out of perf_sigtrap

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Separate generating the signal from deciding it needs to be sent.

v1: https://lkml.kernel.org/r/[email protected]

---
 include/linux/sched/signal.h |  1 +
 kernel/events/core.c         | 11 ++---------
 kernel/signal.c              | 13 +++++++++++++
 3 files changed, 16 insertions(+), 9 deletions(-)
include/linux/sched/signal.h +1 -0
index 7daa425f3055..1e2f61a1a512 100644
@@ -318,6 +318,7 @@ int send_sig_mceerr(int code, void __user *, short, struct task_struct *);
 
 int force_sig_bnderr(void __user *addr, void __user *lower, void __user *upper);
 int force_sig_pkuerr(void __user *addr, u32 pkey);
+int force_sig_perf(void __user *addr, u32 type, u64 sig_data);
 
 int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno);
 int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
kernel/events/core.c +2 -9
index 928b166d888e..48ea8863183b 100644
@@ -6394,8 +6394,6 @@ void perf_event_wakeup(struct perf_event *event)
 
 static void perf_sigtrap(struct perf_event *event)
 {
-	struct kernel_siginfo info;
-
 	/*
 	 * We'd expect this to only occur if the irq_work is delayed and either
 	 * ctx->task or current has changed in the meantime. This can be the
@@ -6410,13 +6408,8 @@ static void perf_sigtrap(struct perf_event *event)
 	if (current->flags & PF_EXITING)
 		return;
 
-	clear_siginfo(&info);
-	info.si_signo = SIGTRAP;
-	info.si_code = TRAP_PERF;
-	info.si_errno = event->attr.type;
-	info.si_perf = event->attr.sig_data;
-	info.si_addr = (void __user *)event->pending_addr;
-	force_sig_info(&info);
+	force_sig_perf((void __user *)event->pending_addr,
+		       event->attr.type, event->attr.sig_data);
 }
 
 static void perf_pending_event_disable(struct perf_event *event)
kernel/signal.c +13 -0
index 697c5fe58db8..49560ceac048 100644
@@ -1753,6 +1753,19 @@ int force_sig_pkuerr(void __user *addr, u32 pkey)
 }
 #endif
 
+int force_sig_perf(void __user *addr, u32 type, u64 sig_data)
+{
+	struct kernel_siginfo info;
+
+	clear_siginfo(&info);
+	info.si_signo = SIGTRAP;
+	info.si_errno = type;
+	info.si_code  = TRAP_PERF;
+	info.si_addr  = addr;
+	info.si_perf  = sig_data;
+	return force_sig_info(&info);
+}
+
 #if IS_ENABLED(CONFIG_SPARC)
 int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno)
 {
Signature · 1 line
2.30.1
#46 lore raw

PATCH 10/12 signal: Redefine signinfo so 64bit fields are possible

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

The si_perf code really wants to add a u64 field. This change enables that by reorganizing the definition of siginfo_t, so that a 64bit field can be added without increasing the alignment of other fields.

---
 arch/x86/kernel/signal_compat.c    |  9 +++----
 include/linux/compat.h             | 28 +++++++++++++-------
 include/uapi/asm-generic/siginfo.h | 42 ++++++++++++++++++++----------
 3 files changed, 49 insertions(+), 30 deletions(-)
arch/x86/kernel/signal_compat.c +3 -6
index a9fcabd8a5e5..a5cd01c52dfb 100644
@@ -17,8 +17,6 @@
  */
 static inline void signal_compat_build_tests(void)
 {
-	int _sifields_offset = offsetof(compat_siginfo_t, _sifields);
-
 	/*
 	 * If adding a new si_code, there is probably new data in
 	 * the siginfo.  Make sure folks bumping the si_code
@@ -40,8 +38,7 @@ static inline void signal_compat_build_tests(void)
 	 * in the ABI, of course.  Make sure none of them ever
 	 * move and are always at the beginning:
 	 */
-	BUILD_BUG_ON(offsetof(compat_siginfo_t, _sifields) != 3 * sizeof(int));
-#define CHECK_CSI_OFFSET(name)	  BUILD_BUG_ON(_sifields_offset != offsetof(compat_siginfo_t, _sifields.name))
+#define CHECK_CSI_OFFSET(name)	  BUILD_BUG_ON(0 != offsetof(compat_siginfo_t, _sifields.name))
 
 	BUILD_BUG_ON(offsetof(siginfo_t, si_signo) != 0);
 	BUILD_BUG_ON(offsetof(siginfo_t, si_errno) != 4);
@@ -63,8 +60,8 @@ static inline void signal_compat_build_tests(void)
 	 * structure stays within the padding size (checked
 	 * above).
 	 */
-#define CHECK_CSI_SIZE(name, size) BUILD_BUG_ON(size != sizeof(((compat_siginfo_t *)0)->_sifields.name))
-#define CHECK_SI_SIZE(name, size) BUILD_BUG_ON(size != sizeof(((siginfo_t *)0)->_sifields.name))
+#define CHECK_CSI_SIZE(name, size) BUILD_BUG_ON(((3*sizeof(int))+(size)) != sizeof(((compat_siginfo_t *)0)->_sifields.name))
+#define CHECK_SI_SIZE(name, size) BUILD_BUG_ON(((4*sizeof(int))+(size)) != sizeof(((siginfo_t *)0)->_sifields.name))
 
 	CHECK_CSI_OFFSET(_kill);
 	CHECK_CSI_SIZE  (_kill, 2*sizeof(int));
include/linux/compat.h +18 -10
index 6af7bef15e94..d81493248bf3 100644
@@ -158,27 +158,28 @@ typedef union compat_sigval {
 	compat_uptr_t	sival_ptr;
 } compat_sigval_t;
 
-typedef struct compat_siginfo {
-	int si_signo;
-#ifndef __ARCH_HAS_SWAPPED_SIGINFO
-	int si_errno;
-	int si_code;
-#else
-	int si_code;
-	int si_errno;
-#endif
+#define __COMPAT_SIGINFO_COMMON	\
+	___SIGINFO_COMMON;	\
+	int	_common_pad[__alignof__(compat_uptr_t) != __alignof__(int)]
 
+typedef struct compat_siginfo {
+	union {
+		struct {
+			__COMPAT_SIGINFO_COMMON;
+		};
 	union {
-		int _pad[128/sizeof(int) - 3];
+		int _pad[128/sizeof(int)];
 
 		/* kill() */
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_pid_t _pid;	/* sender's pid */
 			__compat_uid32_t _uid;	/* sender's uid */
 		} _kill;
 
 		/* POSIX.1b timers */
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_timer_t _tid;	/* timer id */
 			int _overrun;		/* overrun count */
 			compat_sigval_t _sigval;	/* same as below */
@@ -186,6 +187,7 @@ typedef struct compat_siginfo {
 
 		/* POSIX.1b signals */
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_pid_t _pid;	/* sender's pid */
 			__compat_uid32_t _uid;	/* sender's uid */
 			compat_sigval_t _sigval;
@@ -193,6 +195,7 @@ typedef struct compat_siginfo {
 
 		/* SIGCHLD */
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_pid_t _pid;	/* which child */
 			__compat_uid32_t _uid;	/* sender's uid */
 			int _status;		/* exit code */
@@ -203,6 +206,7 @@ typedef struct compat_siginfo {
 #ifdef CONFIG_X86_X32_ABI
 		/* SIGCHLD (x32 version) */
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_pid_t _pid;	/* which child */
 			__compat_uid32_t _uid;	/* sender's uid */
 			int _status;		/* exit code */
@@ -213,6 +217,7 @@ typedef struct compat_siginfo {
 
 		/* SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT */
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_uptr_t _addr;	/* faulting insn/memory ref. */
 #define __COMPAT_ADDR_BND_PKEY_PAD  (__alignof__(compat_uptr_t) < sizeof(short) ? \
 				     sizeof(short) : __alignof__(compat_uptr_t))
@@ -242,16 +247,19 @@ typedef struct compat_siginfo {
 
 		/* SIGPOLL */
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_long_t _band;	/* POLL_IN, POLL_OUT, POLL_MSG */
 			int _fd;
 		} _sigpoll;
 
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_uptr_t _call_addr; /* calling user insn */
 			int _syscall;	/* triggering system call number */
 			unsigned int _arch;	/* AUDIT_ARCH_* of syscall */
 		} _sigsys;
 	} _sifields;
+	};
 } compat_siginfo_t;
 
 struct compat_rlimit {
include/uapi/asm-generic/siginfo.h +28 -14

96 lines not on this page · raw message

Signature · 1 line
2.30.1
#47 lore raw

Re: [PATCH 10/12] signal: Redefine signinfo so 64bit fields are possible

Quoted 5 lines · Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

The si_perf code really wants to add a u64 field. This change enables that by reorganizing the definition of siginfo_t, so that a 64bit field can be added without increasing the alignment of other fields.

I decided to include this change because it is not that complicated, and it allows si_perf_data to have the definition that was originally desired.

If this looks difficult to make in the userspace definitions, or is otherwise a problem I don't mind dropping this change. I just figured since it was not too difficult and we are changing things anyway I should try for everything.

Eric

Quoted 241 lines
---
 arch/x86/kernel/signal_compat.c    |  9 +++----
 include/linux/compat.h             | 28 +++++++++++++-------
 include/uapi/asm-generic/siginfo.h | 42 ++++++++++++++++++++----------
 3 files changed, 49 insertions(+), 30 deletions(-)
arch/x86/kernel/signal_compat.c +3 -6
index a9fcabd8a5e5..a5cd01c52dfb 100644
@@ -17,8 +17,6 @@
  */
 static inline void signal_compat_build_tests(void)
 {
-	int _sifields_offset = offsetof(compat_siginfo_t, _sifields);
-
 	/*
 	 * If adding a new si_code, there is probably new data in
 	 * the siginfo.  Make sure folks bumping the si_code
@@ -40,8 +38,7 @@ static inline void signal_compat_build_tests(void)
 	 * in the ABI, of course.  Make sure none of them ever
 	 * move and are always at the beginning:
 	 */
-	BUILD_BUG_ON(offsetof(compat_siginfo_t, _sifields) != 3 * sizeof(int));
-#define CHECK_CSI_OFFSET(name)	  BUILD_BUG_ON(_sifields_offset != offsetof(compat_siginfo_t, _sifields.name))
+#define CHECK_CSI_OFFSET(name)	  BUILD_BUG_ON(0 != offsetof(compat_siginfo_t, _sifields.name))
 
 	BUILD_BUG_ON(offsetof(siginfo_t, si_signo) != 0);
 	BUILD_BUG_ON(offsetof(siginfo_t, si_errno) != 4);
@@ -63,8 +60,8 @@ static inline void signal_compat_build_tests(void)
 	 * structure stays within the padding size (checked
 	 * above).
 	 */
-#define CHECK_CSI_SIZE(name, size) BUILD_BUG_ON(size != sizeof(((compat_siginfo_t *)0)->_sifields.name))
-#define CHECK_SI_SIZE(name, size) BUILD_BUG_ON(size != sizeof(((siginfo_t *)0)->_sifields.name))
+#define CHECK_CSI_SIZE(name, size) BUILD_BUG_ON(((3*sizeof(int))+(size)) != sizeof(((compat_siginfo_t *)0)->_sifields.name))
+#define CHECK_SI_SIZE(name, size) BUILD_BUG_ON(((4*sizeof(int))+(size)) != sizeof(((siginfo_t *)0)->_sifields.name))
 
 	CHECK_CSI_OFFSET(_kill);
 	CHECK_CSI_SIZE  (_kill, 2*sizeof(int));
include/linux/compat.h +18 -10
index 6af7bef15e94..d81493248bf3 100644
@@ -158,27 +158,28 @@ typedef union compat_sigval {
 	compat_uptr_t	sival_ptr;
 } compat_sigval_t;
 
-typedef struct compat_siginfo {
-	int si_signo;
-#ifndef __ARCH_HAS_SWAPPED_SIGINFO
-	int si_errno;
-	int si_code;
-#else
-	int si_code;
-	int si_errno;
-#endif
+#define __COMPAT_SIGINFO_COMMON	\
+	___SIGINFO_COMMON;	\
+	int	_common_pad[__alignof__(compat_uptr_t) != __alignof__(int)]
 
+typedef struct compat_siginfo {
+	union {
+		struct {
+			__COMPAT_SIGINFO_COMMON;
+		};
 	union {
-		int _pad[128/sizeof(int) - 3];
+		int _pad[128/sizeof(int)];
 
 		/* kill() */
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_pid_t _pid;	/* sender's pid */
 			__compat_uid32_t _uid;	/* sender's uid */
 		} _kill;
 
 		/* POSIX.1b timers */
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_timer_t _tid;	/* timer id */
 			int _overrun;		/* overrun count */
 			compat_sigval_t _sigval;	/* same as below */
@@ -186,6 +187,7 @@ typedef struct compat_siginfo {
 
 		/* POSIX.1b signals */
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_pid_t _pid;	/* sender's pid */
 			__compat_uid32_t _uid;	/* sender's uid */
 			compat_sigval_t _sigval;
@@ -193,6 +195,7 @@ typedef struct compat_siginfo {
 
 		/* SIGCHLD */
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_pid_t _pid;	/* which child */
 			__compat_uid32_t _uid;	/* sender's uid */
 			int _status;		/* exit code */
@@ -203,6 +206,7 @@ typedef struct compat_siginfo {
 #ifdef CONFIG_X86_X32_ABI
 		/* SIGCHLD (x32 version) */
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_pid_t _pid;	/* which child */
 			__compat_uid32_t _uid;	/* sender's uid */
 			int _status;		/* exit code */
@@ -213,6 +217,7 @@ typedef struct compat_siginfo {
 
 		/* SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT */
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_uptr_t _addr;	/* faulting insn/memory ref. */
 #define __COMPAT_ADDR_BND_PKEY_PAD  (__alignof__(compat_uptr_t) < sizeof(short) ? \
 				     sizeof(short) : __alignof__(compat_uptr_t))
@@ -242,16 +247,19 @@ typedef struct compat_siginfo {
 
 		/* SIGPOLL */
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_long_t _band;	/* POLL_IN, POLL_OUT, POLL_MSG */
 			int _fd;
 		} _sigpoll;
 
 		struct {
+			__COMPAT_SIGINFO_COMMON;
 			compat_uptr_t _call_addr; /* calling user insn */
 			int _syscall;	/* triggering system call number */
 			unsigned int _arch;	/* AUDIT_ARCH_* of syscall */
 		} _sigsys;
 	} _sifields;
+	};
 } compat_siginfo_t;
 
 struct compat_rlimit {
include/uapi/asm-generic/siginfo.h +28 -14
index e663bf117b46..1fcede623a73 100644
@@ -29,15 +29,33 @@ typedef union sigval {
 #define __ARCH_SI_ATTRIBUTES
 #endif
 
+#ifndef __ARCH_HAS_SWAPPED_SIGINFO
+#define ___SIGINFO_COMMON	\
+	int	si_signo;	\
+	int	si_errno;	\
+	int	si_code
+#else
+#define ___SIGINFO_COMMON	\
+	int	si_signo;	\
+	int	si_code;	\
+	int	si_errno
+#endif /* __ARCH_HAS_SWAPPED_SIGINFO */
+
+#define __SIGINFO_COMMON	\
+	___SIGINFO_COMMON;	\
+	int	_common_pad[__alignof__(void *) != __alignof(int)]
+
 union __sifields {
 	/* kill() */
 	struct {
+		__SIGINFO_COMMON;
 		__kernel_pid_t _pid;	/* sender's pid */
 		__kernel_uid32_t _uid;	/* sender's uid */
 	} _kill;
 
 	/* POSIX.1b timers */
 	struct {
+		__SIGINFO_COMMON;
 		__kernel_timer_t _tid;	/* timer id */
 		int _overrun;		/* overrun count */
 		sigval_t _sigval;	/* same as below */
@@ -46,6 +64,7 @@ union __sifields {
 
 	/* POSIX.1b signals */
 	struct {
+		__SIGINFO_COMMON;
 		__kernel_pid_t _pid;	/* sender's pid */
 		__kernel_uid32_t _uid;	/* sender's uid */
 		sigval_t _sigval;
@@ -53,6 +72,7 @@ union __sifields {
 
 	/* SIGCHLD */
 	struct {
+		__SIGINFO_COMMON;
 		__kernel_pid_t _pid;	/* which child */
 		__kernel_uid32_t _uid;	/* sender's uid */
 		int _status;		/* exit code */
@@ -62,6 +82,7 @@ union __sifields {
 
 	/* SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT */
 	struct {
+		__SIGINFO_COMMON;
 		void __user *_addr; /* faulting insn/memory ref. */
 #ifdef __ia64__
 		int _imm;		/* immediate value for "break" */
@@ -97,35 +118,28 @@ union __sifields {
 
 	/* SIGPOLL */
 	struct {
+		__SIGINFO_COMMON;
 		__ARCH_SI_BAND_T _band;	/* POLL_IN, POLL_OUT, POLL_MSG */
 		int _fd;
 	} _sigpoll;
 
 	/* SIGSYS */
 	struct {
+		__SIGINFO_COMMON;
 		void __user *_call_addr; /* calling user insn */
 		int _syscall;	/* triggering system call number */
 		unsigned int _arch;	/* AUDIT_ARCH_* of syscall */
 	} _sigsys;
 };
 
-#ifndef __ARCH_HAS_SWAPPED_SIGINFO
-#define __SIGINFO 			\
-struct {				\
-	int si_signo;			\
-	int si_errno;			\
-	int si_code;			\
-	union __sifields _sifields;	\
-}
-#else
+
 #define __SIGINFO 			\
-struct {				\
-	int si_signo;			\
-	int si_code;			\
-	int si_errno;			\
+union {					\
+	struct {			\
+		__SIGINFO_COMMON;	\
+	};				\
 	union __sifields _sifields;	\
 }
-#endif /* __ARCH_HAS_SWAPPED_SIGINFO */
 
 typedef struct siginfo {
 	union {
#48 lore raw

Re: [PATCH 10/12] signal: Redefine signinfo so 64bit fields are possible

Quoted 6 lines · Eric W. Biederman
Quoted 5 lines · Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

The si_perf code really wants to add a u64 field. This change enables that by reorganizing the definition of siginfo_t, so that a 64bit field can be added without increasing the alignment of other fields.

If you can, it'd be good to have an explanation for this, because it's not at all obvious -- some future archeologist will wonder how we ever came up with this definition of siginfo...

(I see the trick here is that before the union would have changed alignment, introducing padding after the 3 ints -- but now because the 3 ints are inside the union the union's padding no longer adds padding for these ints. Perhaps you can explain it better than I can. Also see below.)

Quoted 3 lines

I decided to include this change because it is not that complicated, and it allows si_perf_data to have the definition that was originally desired.

Neat, and long-term I think this seems to be worth having. Sooner or later something else might want __u64, too.

But right now, due to the slight increase in complexity, we need to take extra care ensuring nothing broke -- at a high-level I see why this seems to be safe.

Quoted 4 lines

If this looks difficult to make in the userspace definitions, or is otherwise a problem I don't mind dropping this change. I just figured since it was not too difficult and we are changing things anyway I should try for everything.

Languages that support inheritance might end up with the simpler definition here (and depending on which fields they want to access, they'd have to cast the base siginfo into the one they want). What will become annoying is trying to describe siginfo_t.

I will run some tests in the morning.

Thanks,
-- Marco

[...]

Quoted 8 lines
Quoted 8 lines
include/uapi/asm-generic/siginfo.h +0 -0
index e663bf117b46..1fcede623a73 100644
@@ -29,15 +29,33 @@ typedef union sigval {
 #define __ARCH_SI_ATTRIBUTES
 #endif
 
Quoted 15 lines
Quoted 5 lines
include/uapi/asm-generic/siginfo.h +5 -0
+#ifndef __ARCH_HAS_SWAPPED_SIGINFO
+#define ___SIGINFO_COMMON    \
+     int     si_signo;       \
+     int     si_errno;       \
+     int     si_code

include/uapi/asm-generic/siginfo.h typedef union sigval { · line 46

Quoted 10 lines
include/uapi/asm-generic/siginfo.h +10 -0
+#else
+#define ___SIGINFO_COMMON    \
+     int     si_signo;       \
+     int     si_code;        \
+     int     si_errno
+#endif /* __ARCH_HAS_SWAPPED_SIGINFO */
+
+#define __SIGINFO_COMMON     \
+     ___SIGINFO_COMMON;      \
+     int     _common_pad[__alignof__(void *) != __alignof(int)]

Just to verify my understanding of _common_pad: this is again a legacy problem, right? I.e. if siginfo was designed from the start like this, we wouldn't need the _common_pad.

While this looks quite daunting, this is effectively turning siginfo from a struct with a giant union, into lots of smaller structs, each of which share a common header a'la inheritance -- until now the distinction didn't matter, but it starts to matter as soon as alignment of one child-struct would affect the offsets of another child-struct (i.e. the old version). Right?

I was wondering if it would make things look cleaner if the above was encapsulated in a struct, say __sicommon? Then the outermost union would use 'struct __sicommon _sicommon' and we need #defines for si_signo, si_code, and si_errno.

Or would it break alignment somewhere?

I leave it to your judgement -- just an idea.

Quoted 81 lines
Quoted 81 lines
include/uapi/asm-generic/siginfo.h +12 -14
 union __sifields {
      /* kill() */
      struct {
+             __SIGINFO_COMMON;
              __kernel_pid_t _pid;    /* sender's pid */
              __kernel_uid32_t _uid;  /* sender's uid */
      } _kill;
 
      /* POSIX.1b timers */
      struct {
+             __SIGINFO_COMMON;
              __kernel_timer_t _tid;  /* timer id */
              int _overrun;           /* overrun count */
              sigval_t _sigval;       /* same as below */
@@ -46,6 +64,7 @@ union __sifields {
 
      /* POSIX.1b signals */
      struct {
+             __SIGINFO_COMMON;
              __kernel_pid_t _pid;    /* sender's pid */
              __kernel_uid32_t _uid;  /* sender's uid */
              sigval_t _sigval;
@@ -53,6 +72,7 @@ union __sifields {
 
      /* SIGCHLD */
      struct {
+             __SIGINFO_COMMON;
              __kernel_pid_t _pid;    /* which child */
              __kernel_uid32_t _uid;  /* sender's uid */
              int _status;            /* exit code */
@@ -62,6 +82,7 @@ union __sifields {
 
      /* SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT */
      struct {
+             __SIGINFO_COMMON;
              void __user *_addr; /* faulting insn/memory ref. */
 #ifdef __ia64__
              int _imm;               /* immediate value for "break" */
@@ -97,35 +118,28 @@ union __sifields {
 
      /* SIGPOLL */
      struct {
+             __SIGINFO_COMMON;
              __ARCH_SI_BAND_T _band; /* POLL_IN, POLL_OUT, POLL_MSG */
              int _fd;
      } _sigpoll;
 
      /* SIGSYS */
      struct {
+             __SIGINFO_COMMON;
              void __user *_call_addr; /* calling user insn */
              int _syscall;   /* triggering system call number */
              unsigned int _arch;     /* AUDIT_ARCH_* of syscall */
      } _sigsys;
 };
 
-#ifndef __ARCH_HAS_SWAPPED_SIGINFO
-#define __SIGINFO                    \
-struct {                             \
-     int si_signo;                   \
-     int si_errno;                   \
-     int si_code;                    \
-     union __sifields _sifields;     \
-}
-#else
+
 #define __SIGINFO                    \
-struct {                             \
-     int si_signo;                   \
-     int si_code;                    \
-     int si_errno;                   \
+union {                                      \
+     struct {                        \
+             __SIGINFO_COMMON;       \
+     };                              \
      union __sifields _sifields;     \
 }
-#endif /* __ARCH_HAS_SWAPPED_SIGINFO */
 
 typedef struct siginfo {
      union {
#49 lore raw

Re: [PATCH 10/12] signal: Redefine signinfo so 64bit fields are possible

Quoted 12 lines · Marco Elver
Quoted 6 lines · Eric W. Biederman
Quoted 5 lines · Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

The si_perf code really wants to add a u64 field. This change enables that by reorganizing the definition of siginfo_t, so that a 64bit field can be added without increasing the alignment of other fields.

If you can, it'd be good to have an explanation for this, because it's not at all obvious -- some future archeologist will wonder how we ever came up with this definition of siginfo...

Quoted 5 lines

(I see the trick here is that before the union would have changed alignment, introducing padding after the 3 ints -- but now because the 3 ints are inside the union the union's padding no longer adds padding for these ints. Perhaps you can explain it better than I can. Also see below.)

Yes. The big idea is adding a 64bit field into the second union in the _sigfault case will increase the alignment of that second union to 64bit.

In the 64bit case the alignment is already 64bit so it is not an issue.

In the 32bit case there are 3 ints followed by a pointer. When the 64bit member is added the alignment of _segfault becomes 64bit. That 64bit alignment after 3 ints changes the location of the 32bit pointer.

By moving the 3 preceding ints into _segfault that does not happen.

There remains one very subtle issue that I think isn't a problem but I would appreciate someone else double checking me.

The old definition of siginfo_t on 32bit almost certainly had 32bit alignment. With the addition of a 64bit member siginfo_t gains 64bit alignment. This difference only matters if the 64bit field is accessed. Accessing a 64bit field with 32bit alignment will cause unaligned access exceptions on some (most?) architectures.

For the 64bit field to be accessed the code needs to be recompiled with the new headers. Which implies that when everything is recompiled siginfo_t will become 64bit aligned.

So the change should be safe unless someone is casting something with 32bit alignment into siginfo_t.

Quoted 11 lines
Quoted 4 lines

I decided to include this change because it is not that complicated, and it allows si_perf_data to have the definition that was originally desired.

Neat, and long-term I think this seems to be worth having. Sooner or later something else might want __u64, too.

But right now, due to the slight increase in complexity, we need to take extra care ensuring nothing broke -- at a high-level I see why this seems to be safe.

Absolutely.

Quoted 35 lines
Quoted 4 lines

If this looks difficult to make in the userspace definitions, or is otherwise a problem I don't mind dropping this change. I just figured since it was not too difficult and we are changing things anyway I should try for everything.

Languages that support inheritance might end up with the simpler definition here (and depending on which fields they want to access, they'd have to cast the base siginfo into the one they want). What will become annoying is trying to describe siginfo_t.

I will run some tests in the morning.

Thanks,
-- Marco

[...]

Quoted 19 lines
Quoted 19 lines
include/uapi/asm-generic/siginfo.h +11 -0
index e663bf117b46..1fcede623a73 100644
@@ -29,15 +29,33 @@ typedef union sigval {
 #define __ARCH_SI_ATTRIBUTES
 #endif
 
+#ifndef __ARCH_HAS_SWAPPED_SIGINFO
+#define ___SIGINFO_COMMON    \
+     int     si_signo;       \
+     int     si_errno;       \
+     int     si_code
+#else
+#define ___SIGINFO_COMMON    \
+     int     si_signo;       \
+     int     si_code;        \
+     int     si_errno
+#endif /* __ARCH_HAS_SWAPPED_SIGINFO */
Quoted 8 lines
Quoted 4 lines

include/uapi/asm-generic/siginfo.h typedef union sigval { · line 46

Quoted 4 lines
include/uapi/asm-generic/siginfo.h +4 -0
+
+#define __SIGINFO_COMMON     \
+     ___SIGINFO_COMMON;      \
+     int     _common_pad[__alignof__(void *) != __alignof(int)]

Just to verify my understanding of _common_pad: this is again a legacy problem, right? I.e. if siginfo was designed from the start like this, we wouldn't need the _common_pad.

Correct.

If we are designing things carefully from the start the code could have had a 64bit or a 128bit header that was common to all siginfo instances. Then the trick of moving the header into each union member would not have been necessary.

Unfortunately no one noticed until it was much too late to do anything about it.

Quoted 6 lines

While this looks quite daunting, this is effectively turning siginfo from a struct with a giant union, into lots of smaller structs, each of which share a common header a'la inheritance -- until now the distinction didn't matter, but it starts to matter as soon as alignment of one child-struct would affect the offsets of another child-struct (i.e. the old version). Right?

Correct.

Quoted 8 lines

I was wondering if it would make things look cleaner if the above was encapsulated in a struct, say __sicommon? Then the outermost union would use 'struct __sicommon _sicommon' and we need #defines for si_signo, si_code, and si_errno.

Or would it break alignment somewhere?

I leave it to your judgement -- just an idea.

I tried it and kernel/signal.c would not compile because a few functions have a parameter named si_code. The kernel I could fix but there is a chance some user space application has similar issues.

So to be safe si_signo, si_code, si_errno have to be available without having a define.

Eric

#50 lore raw

Re: [PATCH 10/12] signal: Redefine signinfo so 64bit fields are possible

Quoted 46 lines · Eric W. Biederman
Quoted 12 lines · Marco Elver
Quoted 6 lines · Eric W. Biederman
Quoted 5 lines · Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

The si_perf code really wants to add a u64 field. This change enables that by reorganizing the definition of siginfo_t, so that a 64bit field can be added without increasing the alignment of other fields.

If you can, it'd be good to have an explanation for this, because it's not at all obvious -- some future archeologist will wonder how we ever came up with this definition of siginfo...

Quoted 5 lines

(I see the trick here is that before the union would have changed alignment, introducing padding after the 3 ints -- but now because the 3 ints are inside the union the union's padding no longer adds padding for these ints. Perhaps you can explain it better than I can. Also see below.)

Yes. The big idea is adding a 64bit field into the second union in the _sigfault case will increase the alignment of that second union to 64bit.

In the 64bit case the alignment is already 64bit so it is not an issue.

In the 32bit case there are 3 ints followed by a pointer. When the 64bit member is added the alignment of _segfault becomes 64bit. That 64bit alignment after 3 ints changes the location of the 32bit pointer.

By moving the 3 preceding ints into _segfault that does not happen.

There remains one very subtle issue that I think isn't a problem but I would appreciate someone else double checking me.

The old definition of siginfo_t on 32bit almost certainly had 32bit alignment. With the addition of a 64bit member siginfo_t gains 64bit alignment. This difference only matters if the 64bit field is accessed. Accessing a 64bit field with 32bit alignment will cause unaligned access exceptions on some (most?) architectures.

Quoted 7 lines

For the 64bit field to be accessed the code needs to be recompiled with the new headers. Which implies that when everything is recompiled siginfo_t will become 64bit aligned.

So the change should be safe unless someone is casting something with 32bit alignment into siginfo_t.

How about if someone has a field of type siginfo_t as an element of a struct? For example:

struct foo {
int x;
siginfo_t y;
};

With this change wouldn't the y field move from offset 4 to offset 8?

Peter

#51 lore raw

Re: [PATCH 10/12] signal: Redefine signinfo so 64bit fields are possible

Quoted 56 lines · Peter Collingbourne
Quoted 44 lines · Eric W. Biederman
Quoted 12 lines · Marco Elver
Quoted 6 lines · Eric W. Biederman
Quoted 5 lines · Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

The si_perf code really wants to add a u64 field. This change enables that by reorganizing the definition of siginfo_t, so that a 64bit field can be added without increasing the alignment of other fields.

If you can, it'd be good to have an explanation for this, because it's not at all obvious -- some future archeologist will wonder how we ever came up with this definition of siginfo...

Quoted 5 lines

(I see the trick here is that before the union would have changed alignment, introducing padding after the 3 ints -- but now because the 3 ints are inside the union the union's padding no longer adds padding for these ints. Perhaps you can explain it better than I can. Also see below.)

Yes. The big idea is adding a 64bit field into the second union in the _sigfault case will increase the alignment of that second union to 64bit.

In the 64bit case the alignment is already 64bit so it is not an issue.

In the 32bit case there are 3 ints followed by a pointer. When the 64bit member is added the alignment of _segfault becomes 64bit. That 64bit alignment after 3 ints changes the location of the 32bit pointer.

By moving the 3 preceding ints into _segfault that does not happen.

There remains one very subtle issue that I think isn't a problem but I would appreciate someone else double checking me.

The old definition of siginfo_t on 32bit almost certainly had 32bit alignment. With the addition of a 64bit member siginfo_t gains 64bit alignment. This difference only matters if the 64bit field is accessed. Accessing a 64bit field with 32bit alignment will cause unaligned access exceptions on some (most?) architectures.

Quoted 7 lines

For the 64bit field to be accessed the code needs to be recompiled with the new headers. Which implies that when everything is recompiled siginfo_t will become 64bit aligned.

So the change should be safe unless someone is casting something with 32bit alignment into siginfo_t.

How about if someone has a field of type siginfo_t as an element of a struct? For example:

Quoted 6 lines

struct foo {
int x;
siginfo_t y;
};

With this change wouldn't the y field move from offset 4 to offset 8?

This is a problem if such a struct is part of the ABI -- in the kernel I found these that might be problematic:

| arch/csky/kernel/signal.c:struct rt_sigframe {
| arch/csky/kernel/signal.c-	/*
| arch/csky/kernel/signal.c-	 * pad[3] is compatible with the same struct defined in
| arch/csky/kernel/signal.c-	 * gcc/libgcc/config/csky/linux-unwind.h
| arch/csky/kernel/signal.c-	 */
| arch/csky/kernel/signal.c-	int pad[3];
| arch/csky/kernel/signal.c-	struct siginfo info;
| arch/csky/kernel/signal.c-	struct ucontext uc;
| arch/csky/kernel/signal.c-};
| [...]
| arch/parisc/include/asm/rt_sigframe.h-#define SIGRETURN_TRAMP 4
| arch/parisc/include/asm/rt_sigframe.h-#define SIGRESTARTBLOCK_TRAMP 5 
| arch/parisc/include/asm/rt_sigframe.h-#define TRAMP_SIZE (SIGRETURN_TRAMP + SIGRESTARTBLOCK_TRAMP)
| arch/parisc/include/asm/rt_sigframe.h-
| arch/parisc/include/asm/rt_sigframe.h:struct rt_sigframe {
| arch/parisc/include/asm/rt_sigframe.h-	/* XXX: Must match trampoline size in arch/parisc/kernel/signal.c 
| arch/parisc/include/asm/rt_sigframe.h-	        Secondary to that it must protect the ERESTART_RESTARTBLOCK
| arch/parisc/include/asm/rt_sigframe.h-		trampoline we left on the stack (we were bad and didn't 
| arch/parisc/include/asm/rt_sigframe.h-		change sp so we could run really fast.) */
| arch/parisc/include/asm/rt_sigframe.h-	unsigned int tramp[TRAMP_SIZE];
| arch/parisc/include/asm/rt_sigframe.h-	struct siginfo info;
| [..]
| arch/parisc/kernel/signal32.h-#define COMPAT_SIGRETURN_TRAMP 4
| arch/parisc/kernel/signal32.h-#define COMPAT_SIGRESTARTBLOCK_TRAMP 5
| arch/parisc/kernel/signal32.h-#define COMPAT_TRAMP_SIZE (COMPAT_SIGRETURN_TRAMP + \
| arch/parisc/kernel/signal32.h-				COMPAT_SIGRESTARTBLOCK_TRAMP)
| arch/parisc/kernel/signal32.h-
| arch/parisc/kernel/signal32.h:struct compat_rt_sigframe {
| arch/parisc/kernel/signal32.h-        /* XXX: Must match trampoline size in arch/parisc/kernel/signal.c
| arch/parisc/kernel/signal32.h-                Secondary to that it must protect the ERESTART_RESTARTBLOCK
| arch/parisc/kernel/signal32.h-                trampoline we left on the stack (we were bad and didn't
| arch/parisc/kernel/signal32.h-                change sp so we could run really fast.) */
| arch/parisc/kernel/signal32.h-        compat_uint_t tramp[COMPAT_TRAMP_SIZE];
| arch/parisc/kernel/signal32.h-        compat_siginfo_t info;

Adding these static asserts to parisc shows the problem:

| diff --git a/arch/parisc/kernel/signal.c b/arch/parisc/kernel/signal.c
| index fb1e94a3982b..0be582fb81be 100644
| --- a/arch/parisc/kernel/signal.c
| +++ b/arch/parisc/kernel/signal.c
| @@ -610,3 +610,6 @@ void do_notify_resume(struct pt_regs *regs, long in_syscall)
|  	if (test_thread_flag(TIF_NOTIFY_RESUME))
|  		tracehook_notify_resume(regs);
|  }
| +
| +static_assert(sizeof(unsigned long) == 4); // 32 bit build
| +static_assert(offsetof(struct rt_sigframe, info) == 9 * 4);

This passes without the siginfo rework in this patch. With it:

| ./include/linux/build_bug.h:78:41: error: static assertion failed: "offsetof(struct rt_sigframe, info) == 9 * 4"

As sad as it is, I don't think we can have our cake and eat it, too. :-(

Unless you see why this is fine, I think we need to drop this patch and go back to the simpler version you had.

Thanks,
-- Marco

#52 lore raw

Re: [PATCH 10/12] signal: Redefine signinfo so 64bit fields are possible

Quoted 116 lines · Marco Elver
Quoted 56 lines · Peter Collingbourne
Quoted 44 lines · Eric W. Biederman
Quoted 12 lines · Marco Elver
Quoted 6 lines · Eric W. Biederman
Quoted 5 lines · Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

The si_perf code really wants to add a u64 field. This change enables that by reorganizing the definition of siginfo_t, so that a 64bit field can be added without increasing the alignment of other fields.

If you can, it'd be good to have an explanation for this, because it's not at all obvious -- some future archeologist will wonder how we ever came up with this definition of siginfo...

Quoted 5 lines

(I see the trick here is that before the union would have changed alignment, introducing padding after the 3 ints -- but now because the 3 ints are inside the union the union's padding no longer adds padding for these ints. Perhaps you can explain it better than I can. Also see below.)

Yes. The big idea is adding a 64bit field into the second union in the _sigfault case will increase the alignment of that second union to 64bit.

In the 64bit case the alignment is already 64bit so it is not an issue.

In the 32bit case there are 3 ints followed by a pointer. When the 64bit member is added the alignment of _segfault becomes 64bit. That 64bit alignment after 3 ints changes the location of the 32bit pointer.

By moving the 3 preceding ints into _segfault that does not happen.

There remains one very subtle issue that I think isn't a problem but I would appreciate someone else double checking me.

The old definition of siginfo_t on 32bit almost certainly had 32bit alignment. With the addition of a 64bit member siginfo_t gains 64bit alignment. This difference only matters if the 64bit field is accessed. Accessing a 64bit field with 32bit alignment will cause unaligned access exceptions on some (most?) architectures.

Quoted 7 lines

For the 64bit field to be accessed the code needs to be recompiled with the new headers. Which implies that when everything is recompiled siginfo_t will become 64bit aligned.

So the change should be safe unless someone is casting something with 32bit alignment into siginfo_t.

How about if someone has a field of type siginfo_t as an element of a struct? For example:

Quoted 6 lines

struct foo {
int x;
siginfo_t y;
};

With this change wouldn't the y field move from offset 4 to offset 8?

This is a problem if such a struct is part of the ABI -- in the kernel I found these that might be problematic:

| arch/csky/kernel/signal.c:struct rt_sigframe {
| arch/csky/kernel/signal.c-	/*
| arch/csky/kernel/signal.c-	 * pad[3] is compatible with the same struct defined in
| arch/csky/kernel/signal.c-	 * gcc/libgcc/config/csky/linux-unwind.h
| arch/csky/kernel/signal.c-	 */
| arch/csky/kernel/signal.c-	int pad[3];
| arch/csky/kernel/signal.c-	struct siginfo info;
| arch/csky/kernel/signal.c-	struct ucontext uc;
| arch/csky/kernel/signal.c-};
| [...]
| arch/parisc/include/asm/rt_sigframe.h-#define SIGRETURN_TRAMP 4
| arch/parisc/include/asm/rt_sigframe.h-#define SIGRESTARTBLOCK_TRAMP 5 
| arch/parisc/include/asm/rt_sigframe.h-#define TRAMP_SIZE (SIGRETURN_TRAMP + SIGRESTARTBLOCK_TRAMP)
| arch/parisc/include/asm/rt_sigframe.h-
| arch/parisc/include/asm/rt_sigframe.h:struct rt_sigframe {
| arch/parisc/include/asm/rt_sigframe.h-	/* XXX: Must match trampoline size in arch/parisc/kernel/signal.c 
| arch/parisc/include/asm/rt_sigframe.h-	        Secondary to that it must protect the ERESTART_RESTARTBLOCK
| arch/parisc/include/asm/rt_sigframe.h-		trampoline we left on the stack (we were bad and didn't 
| arch/parisc/include/asm/rt_sigframe.h-		change sp so we could run really fast.) */
| arch/parisc/include/asm/rt_sigframe.h-	unsigned int tramp[TRAMP_SIZE];
| arch/parisc/include/asm/rt_sigframe.h-	struct siginfo info;
| [..]
| arch/parisc/kernel/signal32.h-#define COMPAT_SIGRETURN_TRAMP 4
| arch/parisc/kernel/signal32.h-#define COMPAT_SIGRESTARTBLOCK_TRAMP 5
| arch/parisc/kernel/signal32.h-#define COMPAT_TRAMP_SIZE (COMPAT_SIGRETURN_TRAMP + \
| arch/parisc/kernel/signal32.h-				COMPAT_SIGRESTARTBLOCK_TRAMP)
| arch/parisc/kernel/signal32.h-
| arch/parisc/kernel/signal32.h:struct compat_rt_sigframe {
| arch/parisc/kernel/signal32.h-        /* XXX: Must match trampoline size in arch/parisc/kernel/signal.c
| arch/parisc/kernel/signal32.h-                Secondary to that it must protect the ERESTART_RESTARTBLOCK
| arch/parisc/kernel/signal32.h-                trampoline we left on the stack (we were bad and didn't
| arch/parisc/kernel/signal32.h-                change sp so we could run really fast.) */
| arch/parisc/kernel/signal32.h-        compat_uint_t tramp[COMPAT_TRAMP_SIZE];
| arch/parisc/kernel/signal32.h-        compat_siginfo_t info;

Adding these static asserts to parisc shows the problem:

| diff --git a/arch/parisc/kernel/signal.c b/arch/parisc/kernel/signal.c
| index fb1e94a3982b..0be582fb81be 100644
| --- a/arch/parisc/kernel/signal.c
| +++ b/arch/parisc/kernel/signal.c
| @@ -610,3 +610,6 @@ void do_notify_resume(struct pt_regs *regs, long in_syscall)
|  	if (test_thread_flag(TIF_NOTIFY_RESUME))
|  		tracehook_notify_resume(regs);
|  }
| +
| +static_assert(sizeof(unsigned long) == 4); // 32 bit build
| +static_assert(offsetof(struct rt_sigframe, info) == 9 * 4);
Quoted 8 lines

This passes without the siginfo rework in this patch. With it:

| ./include/linux/build_bug.h:78:41: error: static assertion failed: "offsetof(struct rt_sigframe, info) == 9 * 4"

As sad as it is, I don't think we can have our cake and eat it, too. :-(

Unless you see why this is fine, I think we need to drop this patch and go back to the simpler version you had.

No. I really can't. I think we are stuck with 32bit alignment on 32bit architectures at this point. Which precludes 32bit architectures from including a 64bit field.

The variant of this that concerns me the most is siginfo_t embedded in a structure in a library combined with code that is compiled with new headers. The offset of the embedded siginfo_t could very easily change and break things.

That makes the alignment an ABI property we can't mess with. Shame.

I will figure out some static asserts to verify this property remains on 32bit and respin this series.

Eric

#53 lore raw

PATCH 11/12 signal: Deliver all of the siginfo perf data in _perf

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Don't abuse si_errno and deliver all of the perf data in _perf member of siginfo_t.

Because it is possible increase the size of perf_data to 64bits.

v1: https://lkml.kernel.org/r/[email protected]

---
 arch/arm/kernel/signal.c                      |  3 ++-
 arch/arm64/kernel/signal.c                    |  3 ++-
 arch/arm64/kernel/signal32.c                  |  3 ++-
 arch/sparc/kernel/signal32.c                  |  3 ++-
 arch/sparc/kernel/signal_64.c                 |  3 ++-
 arch/x86/kernel/signal_compat.c               |  8 ++++---
 fs/signalfd.c                                 |  3 ++-
 include/linux/compat.h                        |  5 ++++-
 include/uapi/asm-generic/siginfo.h            |  8 +++++--
 include/uapi/linux/signalfd.h                 |  4 ++--
 kernel/signal.c                               | 21 ++++++++++++-------
 .../selftests/perf_events/sigtrap_threads.c   | 12 +++++------
 12 files changed, 48 insertions(+), 28 deletions(-)
arch/arm/kernel/signal.c +2 -1
index 2dac5d2c5cf6..ffd3cb8dadea 100644
@@ -755,7 +755,8 @@ static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x10);
 static_assert(offsetof(siginfo_t, si_lower)	== 0x14);
 static_assert(offsetof(siginfo_t, si_upper)	== 0x18);
 static_assert(offsetof(siginfo_t, si_pkey)	== 0x14);
-static_assert(offsetof(siginfo_t, si_perf)	== 0x10);
+static_assert(offsetof(siginfo_t, si_perf_data)	== 0x10);
+static_assert(offsetof(siginfo_t, si_perf_type)	== 0x18);
 static_assert(offsetof(siginfo_t, si_band)	== 0x0c);
 static_assert(offsetof(siginfo_t, si_fd)	== 0x10);
 static_assert(offsetof(siginfo_t, si_call_addr)	== 0x0c);
arch/arm64/kernel/signal.c +2 -1
index af8bd2af1298..f6cbbc54861e 100644
@@ -1003,7 +1003,8 @@ static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x18);
 static_assert(offsetof(siginfo_t, si_lower)	== 0x20);
 static_assert(offsetof(siginfo_t, si_upper)	== 0x28);
 static_assert(offsetof(siginfo_t, si_pkey)	== 0x20);
-static_assert(offsetof(siginfo_t, si_perf)	== 0x18);
+static_assert(offsetof(siginfo_t, si_perf_data)	== 0x18);
+static_assert(offsetof(siginfo_t, si_perf_type)	== 0x20);
 static_assert(offsetof(siginfo_t, si_band)	== 0x10);
 static_assert(offsetof(siginfo_t, si_fd)	== 0x18);
 static_assert(offsetof(siginfo_t, si_call_addr)	== 0x10);
arch/arm64/kernel/signal32.c +2 -1
index b6afb646515f..8241dcc43d7f 100644
@@ -487,7 +487,8 @@ static_assert(offsetof(compat_siginfo_t, si_addr_lsb)	== 0x10);
 static_assert(offsetof(compat_siginfo_t, si_lower)	== 0x14);
 static_assert(offsetof(compat_siginfo_t, si_upper)	== 0x18);
 static_assert(offsetof(compat_siginfo_t, si_pkey)	== 0x14);
-static_assert(offsetof(compat_siginfo_t, si_perf)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_perf_data)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_perf_type)	== 0x18);
 static_assert(offsetof(compat_siginfo_t, si_band)	== 0x0c);
 static_assert(offsetof(compat_siginfo_t, si_fd)		== 0x10);
 static_assert(offsetof(compat_siginfo_t, si_call_addr)	== 0x0c);
arch/sparc/kernel/signal32.c +2 -1
index 73fd8700df3e..1c8b60dd27d8 100644
@@ -776,6 +776,7 @@ static_assert(offsetof(compat_siginfo_t, si_addr_lsb)	== 0x10);
 static_assert(offsetof(compat_siginfo_t, si_lower)	== 0x14);
 static_assert(offsetof(compat_siginfo_t, si_upper)	== 0x18);
 static_assert(offsetof(compat_siginfo_t, si_pkey)	== 0x14);
-static_assert(offsetof(compat_siginfo_t, si_perf)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_perf_data)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_perf_type)	== 0x18);
 static_assert(offsetof(compat_siginfo_t, si_band)	== 0x0c);
 static_assert(offsetof(compat_siginfo_t, si_fd)		== 0x10);
arch/sparc/kernel/signal_64.c +2 -1
index 17913daa66c6..a58e0cc45d24 100644
@@ -586,6 +586,7 @@ static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x18);
 static_assert(offsetof(siginfo_t, si_lower)	== 0x20);
 static_assert(offsetof(siginfo_t, si_upper)	== 0x28);
 static_assert(offsetof(siginfo_t, si_pkey)	== 0x20);
-static_assert(offsetof(siginfo_t, si_perf)	== 0x18);
+static_assert(offsetof(siginfo_t, si_perf_data)	== 0x18);
+static_assert(offsetof(siginfo_t, si_perf_type)	== 0x20);
 static_assert(offsetof(siginfo_t, si_band)	== 0x10);
 static_assert(offsetof(siginfo_t, si_fd)	== 0x14);
arch/x86/kernel/signal_compat.c +5 -3
index a5cd01c52dfb..a81310fe9e8a 100644
@@ -118,7 +118,7 @@ static inline void signal_compat_build_tests(void)
 #endif
 
 	CHECK_CSI_OFFSET(_sigfault);
-	CHECK_CSI_SIZE  (_sigfault, 4*sizeof(int));
+	CHECK_CSI_SIZE  (_sigfault, 5*sizeof(int));
 	CHECK_SI_SIZE   (_sigfault, 8*sizeof(int));
 
 	BUILD_BUG_ON(offsetof(siginfo_t, si_addr) != 0x10);
@@ -138,8 +138,10 @@ static inline void signal_compat_build_tests(void)
 	BUILD_BUG_ON(offsetof(siginfo_t, si_pkey) != 0x20);
 	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_pkey) != 0x14);
 
-	BUILD_BUG_ON(offsetof(siginfo_t, si_perf) != 0x18);
-	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_perf) != 0x10);
+	BUILD_BUG_ON(offsetof(siginfo_t, si_perf_data) != 0x18);
+	BUILD_BUG_ON(offsetof(siginfo_t, si_perf_type) != 0x20);
+	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_perf_data) != 0x10);
+	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_perf_type) != 0x18);
 
 	CHECK_CSI_OFFSET(_sigpoll);
 	CHECK_CSI_SIZE  (_sigpoll, 2*sizeof(int));
fs/signalfd.c +2 -1
index 83130244f653..335ad39f3900 100644
@@ -134,7 +134,8 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		break;
 	case SIL_FAULT_PERF_EVENT:
 		new.ssi_addr = (long) kinfo->si_addr;
-		new.ssi_perf = kinfo->si_perf;
+		new.ssi_perf_type = kinfo->si_perf_type;
+		new.ssi_perf_data = kinfo->si_perf_data;
 		break;
 	case SIL_CHLD:
 		new.ssi_pid    = kinfo->si_pid;
include/linux/compat.h +4 -1
index d81493248bf3..9190a1c2aac7 100644
@@ -241,7 +241,10 @@ typedef struct compat_siginfo {
 					u32 _pkey;
 				} _addr_pkey;
 				/* used when si_code=TRAP_PERF */
-				compat_ulong_t _perf;
+				struct {
+					u64 _data;
+					u32 _type;
+				} _perf;
 			};
 		} _sigfault;
 
include/uapi/asm-generic/siginfo.h +6 -2
index 1fcede623a73..dbe3dd90876f 100644
@@ -112,7 +112,10 @@ union __sifields {
 				__u32 _pkey;
 			} _addr_pkey;
 			/* used when si_code=TRAP_PERF */
-			unsigned long _perf;
+			struct {
+				__u64 _data;
+				__u32 _type;
+			} _perf;
 		};
 	} _sigfault;
 
@@ -168,7 +171,8 @@ typedef struct siginfo {
 #define si_lower	_sifields._sigfault._addr_bnd._lower
 #define si_upper	_sifields._sigfault._addr_bnd._upper
 #define si_pkey		_sifields._sigfault._addr_pkey._pkey
-#define si_perf		_sifields._sigfault._perf
+#define si_perf_data	_sifields._sigfault._perf._data
+#define si_perf_type	_sifields._sigfault._perf._type
 #define si_band		_sifields._sigpoll._band
 #define si_fd		_sifields._sigpoll._fd
 #define si_call_addr	_sifields._sigsys._call_addr
include/uapi/linux/signalfd.h +2 -2
index 7e333042c7e3..e78dddf433fc 100644
@@ -39,8 +39,8 @@ struct signalfd_siginfo {
 	__s32 ssi_syscall;
 	__u64 ssi_call_addr;
 	__u32 ssi_arch;
-	__u32 __pad3;
-	__u64 ssi_perf;
+	__u32 ssi_perf_type;
+	__u64 ssi_perf_data;
 
 	/*
 	 * Pad strcture to 128 bytes. Remember to update the
kernel/signal.c +13 -8
index 49560ceac048..7fec9d1c5b11 100644
@@ -1758,11 +1758,13 @@ int force_sig_perf(void __user *addr, u32 type, u64 sig_data)
 	struct kernel_siginfo info;
 
 	clear_siginfo(&info);
-	info.si_signo = SIGTRAP;
-	info.si_errno = type;
-	info.si_code  = TRAP_PERF;
-	info.si_addr  = addr;
-	info.si_perf  = sig_data;
+	info.si_signo     = SIGTRAP;
+	info.si_errno     = 0;
+	info.si_code      = TRAP_PERF;
+	info.si_addr      = addr;
+	info.si_perf_data = sig_data;
+	info.si_perf_type = type;
+
 	return force_sig_info(&info);
 }
 
@@ -3380,7 +3382,8 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 		break;
 	case SIL_FAULT_PERF_EVENT:
 		to->si_addr = ptr_to_compat(from->si_addr);
-		to->si_perf = from->si_perf;
+		to->si_perf_data = from->si_perf_data;
+		to->si_perf_type = from->si_perf_type;
 		break;
 	case SIL_CHLD:
 		to->si_pid = from->si_pid;
@@ -3456,7 +3459,8 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 		break;
 	case SIL_FAULT_PERF_EVENT:
 		to->si_addr = compat_ptr(from->si_addr);
-		to->si_perf = from->si_perf;
+		to->si_perf_data = from->si_perf_data;
+		to->si_perf_type = from->si_perf_type;
 		break;
 	case SIL_CHLD:
 		to->si_pid    = from->si_pid;
@@ -4639,7 +4643,8 @@ static inline void siginfo_buildtime_checks(void)
 	CHECK_OFFSET(si_lower);
 	CHECK_OFFSET(si_upper);
 	CHECK_OFFSET(si_pkey);
-	CHECK_OFFSET(si_perf);
+	CHECK_OFFSET(si_perf_data);
+	CHECK_OFFSET(si_perf_type);
 
 	/* sigpoll */
 	CHECK_OFFSET(si_band);
tools/testing/selftests/perf_events/sigtrap_threads.c +6 -6
index 78ddf5e11625..fde123066a8c 100644
@@ -164,8 +164,8 @@ TEST_F(sigtrap_threads, enable_event)
 	EXPECT_EQ(ctx.signal_count, NUM_THREADS);
 	EXPECT_EQ(ctx.tids_want_signal, 0);
 	EXPECT_EQ(ctx.first_siginfo.si_addr, &ctx.iterate_on);
-	EXPECT_EQ(ctx.first_siginfo.si_errno, PERF_TYPE_BREAKPOINT);
-	EXPECT_EQ(ctx.first_siginfo.si_perf, TEST_SIG_DATA(&ctx.iterate_on));
+	EXPECT_EQ(ctx.first_siginfo.si_perf_type, PERF_TYPE_BREAKPOINT);
+	EXPECT_EQ(ctx.first_siginfo.si_perf_data, TEST_SIG_DATA(&ctx.iterate_on));
 
 	/* Check enabled for parent. */
 	ctx.iterate_on = 0;
@@ -183,8 +183,8 @@ TEST_F(sigtrap_threads, modify_and_enable_event)
 	EXPECT_EQ(ctx.signal_count, NUM_THREADS);
 	EXPECT_EQ(ctx.tids_want_signal, 0);
 	EXPECT_EQ(ctx.first_siginfo.si_addr, &ctx.iterate_on);
-	EXPECT_EQ(ctx.first_siginfo.si_errno, PERF_TYPE_BREAKPOINT);
-	EXPECT_EQ(ctx.first_siginfo.si_perf, TEST_SIG_DATA(&ctx.iterate_on));
+	EXPECT_EQ(ctx.first_siginfo.si_perf_type, PERF_TYPE_BREAKPOINT);
+	EXPECT_EQ(ctx.first_siginfo.si_perf_data, TEST_SIG_DATA(&ctx.iterate_on));
 
 	/* Check enabled for parent. */
 	ctx.iterate_on = 0;
@@ -203,8 +203,8 @@ TEST_F(sigtrap_threads, signal_stress)
 	EXPECT_EQ(ctx.signal_count, NUM_THREADS * ctx.iterate_on);
 	EXPECT_EQ(ctx.tids_want_signal, 0);
 	EXPECT_EQ(ctx.first_siginfo.si_addr, &ctx.iterate_on);
-	EXPECT_EQ(ctx.first_siginfo.si_errno, PERF_TYPE_BREAKPOINT);
-	EXPECT_EQ(ctx.first_siginfo.si_perf, TEST_SIG_DATA(&ctx.iterate_on));
+	EXPECT_EQ(ctx.first_siginfo.si_perf_type, PERF_TYPE_BREAKPOINT);
+	EXPECT_EQ(ctx.first_siginfo.si_perf_data, TEST_SIG_DATA(&ctx.iterate_on));
 }
 
 TEST_HARNESS_MAIN
Signature · 1 line
2.30.1
#54 lore raw

PATCH 12/12 signalfd: Remove SIL_FAULT_PERF_EVENT fields from signalfd_siginfo

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

With the addition of ssi_perf_data and ssi_perf_type struct signalfd_siginfo is dangerously close to running out of space. All that remains is just enough space for two additional 64bit fields. A practice of adding all possible siginfo_t fields into struct singalfd_siginfo can not be supported as adding the missing fields ssi_lower, ssi_upper, and ssi_pkey would require two 64bit fields and one 32bit fields. In practice the fields ssi_perf_data and ssi_perf_type can never be used by signalfd as the signal that generates them always delivers them synchronously to the thread that triggers them.

Therefore until someone actually needs the fields ssi_perf_data and ssi_perf_type in signalfd_siginfo remove them. This leaves a bit more room for future expansion.

---
 fs/signalfd.c                 | 16 ++++++----------
 include/uapi/linux/signalfd.h |  4 +---
 2 files changed, 7 insertions(+), 13 deletions(-)
fs/signalfd.c +6 -10
index 335ad39f3900..040e1cf90528 100644
@@ -114,12 +114,13 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		break;
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
+	case SIL_FAULT_PERF_EVENT:
 		/*
-		 * Fall through to the SIL_FAULT case.  Both SIL_FAULT_BNDERR
-		 * and SIL_FAULT_PKUERR are only generated by faults that
-		 * deliver them synchronously to userspace.  In case someone
-		 * injects one of these signals and signalfd catches it treat
-		 * it as SIL_FAULT.
+		 * Fall through to the SIL_FAULT case.  SIL_FAULT_BNDERR,
+		 * SIL_FAULT_PKUERR, and SIL_FAULT_PERF_EVENT are only
+		 * generated by faults that deliver them synchronously to
+		 * userspace.  In case someone injects one of these signals
+		 * and signalfd catches it treat it as SIL_FAULT.
 		 */
 	case SIL_FAULT:
 		new.ssi_addr = (long) kinfo->si_addr;
@@ -132,11 +133,6 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		new.ssi_addr = (long) kinfo->si_addr;
 		new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
 		break;
-	case SIL_FAULT_PERF_EVENT:
-		new.ssi_addr = (long) kinfo->si_addr;
-		new.ssi_perf_type = kinfo->si_perf_type;
-		new.ssi_perf_data = kinfo->si_perf_data;
-		break;
 	case SIL_CHLD:
 		new.ssi_pid    = kinfo->si_pid;
 		new.ssi_uid    = kinfo->si_uid;
include/uapi/linux/signalfd.h +1 -3
index e78dddf433fc..83429a05b698 100644
@@ -39,8 +39,6 @@ struct signalfd_siginfo {
 	__s32 ssi_syscall;
 	__u64 ssi_call_addr;
 	__u32 ssi_arch;
-	__u32 ssi_perf_type;
-	__u64 ssi_perf_data;
 
 	/*
 	 * Pad strcture to 128 bytes. Remember to update the
@@ -51,7 +49,7 @@ struct signalfd_siginfo {
 	 * comes out of a read(2) and we really don't want to have
 	 * a compat on read(2).
 	 */
-	__u8 __pad[16];
+	__u8 __pad[28];
 };
 
 
Signature · 1 line
2.30.1
#55 lore raw

v3 cover signal: sort out si_trapno and si_perf

In reply to Eric W. Biederman

This set of changes sorts out the ABI issues with SIGTRAP TRAP_PERF, and hopefully will can get merged before any userspace code starts using the new ABI.

The big ideas are:
- Placing the asserts first to prevent unexpected ABI changes
- si_trapno becomming ordinary fault subfield.
- struct signalfd_siginfo is almost full

This set of changes starts out with Marco's static_assert changes and additional one of my own that enforces the fact that the alignment of siginfo_t is also part of the ABI. Together these build time checks verify there are no unexpected ABI changes in the changes that follow.

The field si_trapno is changed to become an ordinary extension of the _sigfault member of siginfo.

The code is refactored a bit and then si_perf_type is added along side si_perf_data in the _perf subfield of _sigfault of siginfo_t.

Finally the signalfd_siginfo fields are removed as they appear to be filling up the structure without userspace actually being able to use them.

v2: https://lkml.kernel.org/r/[email protected]
v1: https://lkml.kernel.org/r/[email protected]

Eric W. Biederman (9):
signal: Verify the alignment and size of siginfo_t
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Use dedicated helpers to send signals with si_trapno set
signal: Remove __ARCH_SI_TRAPNO
signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency
signal: Factor force_sig_perf out of perf_sigtrap
signal: Deliver all of the siginfo perf data in _perf
signalfd: Remove SIL_FAULT_PERF_EVENT fields from signalfd_siginfo

Marco Elver (3):
sparc64: Add compile-time asserts for siginfo_t offsets
arm: Add compile-time asserts for siginfo_t offsets
arm64: Add compile-time asserts for siginfo_t offsets

 arch/alpha/include/uapi/asm/siginfo.h              |   2 -
 arch/alpha/kernel/osf_sys.c                        |   2 +-
 arch/alpha/kernel/signal.c                         |   4 +-
 arch/alpha/kernel/traps.c                          |  24 ++---
 arch/alpha/mm/fault.c                              |   4 +-
 arch/arm/kernel/signal.c                           |  39 +++++++
 arch/arm64/kernel/signal.c                         |  39 +++++++
 arch/arm64/kernel/signal32.c                       |  39 +++++++
 arch/mips/include/uapi/asm/siginfo.h               |   2 -
 arch/sparc/include/uapi/asm/siginfo.h              |   3 -
 arch/sparc/kernel/process_64.c                     |   2 +-
 arch/sparc/kernel/signal32.c                       |  37 +++++++
 arch/sparc/kernel/signal_64.c                      |  36 +++++++
 arch/sparc/kernel/sys_sparc_32.c                   |   2 +-
 arch/sparc/kernel/sys_sparc_64.c                   |   2 +-
 arch/sparc/kernel/traps_32.c                       |  22 ++--
 arch/sparc/kernel/traps_64.c                       |  44 ++++----
 arch/sparc/kernel/unaligned_32.c                   |   2 +-
 arch/sparc/mm/fault_32.c                           |   2 +-
 arch/sparc/mm/fault_64.c                           |   2 +-
 arch/x86/kernel/signal_compat.c                    |  15 ++-
 fs/signalfd.c                                      |  23 ++---
 include/linux/compat.h                             |  10 +-
 include/linux/sched/signal.h                       |  13 +--
 include/linux/signal.h                             |   3 +-
 include/uapi/asm-generic/siginfo.h                 |  20 ++--
 include/uapi/linux/signalfd.h                      |   4 +-
 kernel/events/core.c                               |  11 +-
 kernel/signal.c                                    | 113 +++++++++++++--------
 .../selftests/perf_events/sigtrap_threads.c        |  12 +--
 30 files changed, 373 insertions(+), 160 deletions(-)
#56 lore raw

Re: [PATCH v3 00/12] signal: sort out si_trapno and si_perf

Quoted 40 lines · Eric W. Biederman

This set of changes sorts out the ABI issues with SIGTRAP TRAP_PERF, and hopefully will can get merged before any userspace code starts using the new ABI.

The big ideas are:
- Placing the asserts first to prevent unexpected ABI changes
- si_trapno becomming ordinary fault subfield.
- struct signalfd_siginfo is almost full

This set of changes starts out with Marco's static_assert changes and additional one of my own that enforces the fact that the alignment of siginfo_t is also part of the ABI. Together these build time checks verify there are no unexpected ABI changes in the changes that follow.

The field si_trapno is changed to become an ordinary extension of the _sigfault member of siginfo.

The code is refactored a bit and then si_perf_type is added along side si_perf_data in the _perf subfield of _sigfault of siginfo_t.

Finally the signalfd_siginfo fields are removed as they appear to be filling up the structure without userspace actually being able to use them.

v2: https://lkml.kernel.org/r/[email protected]
v1: https://lkml.kernel.org/r/[email protected]

Eric W. Biederman (9):
signal: Verify the alignment and size of siginfo_t
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Use dedicated helpers to send signals with si_trapno set
signal: Remove __ARCH_SI_TRAPNO
signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency
signal: Factor force_sig_perf out of perf_sigtrap
signal: Deliver all of the siginfo perf data in _perf
signalfd: Remove SIL_FAULT_PERF_EVENT fields from signalfd_siginfo

Quoted 4 lines

Marco Elver (3):
sparc64: Add compile-time asserts for siginfo_t offsets
arm: Add compile-time asserts for siginfo_t offsets
arm64: Add compile-time asserts for siginfo_t offsets

I can't seem to see the rest of them in my inbox. LKML also is missing
them: https://lore.kernel.org/linux-api/[email protected]/

Something must have swallowed them. Could you resend? I'll then test in the morning.

Thanks,
-- Marco

#57 lore raw

Re: [PATCH v3 00/12] signal: sort out si_trapno and si_perf

Quoted 46 lines · Marco Elver
Quoted 40 lines · Eric W. Biederman

This set of changes sorts out the ABI issues with SIGTRAP TRAP_PERF, and hopefully will can get merged before any userspace code starts using the new ABI.

The big ideas are:
- Placing the asserts first to prevent unexpected ABI changes
- si_trapno becomming ordinary fault subfield.
- struct signalfd_siginfo is almost full

This set of changes starts out with Marco's static_assert changes and additional one of my own that enforces the fact that the alignment of siginfo_t is also part of the ABI. Together these build time checks verify there are no unexpected ABI changes in the changes that follow.

The field si_trapno is changed to become an ordinary extension of the _sigfault member of siginfo.

The code is refactored a bit and then si_perf_type is added along side si_perf_data in the _perf subfield of _sigfault of siginfo_t.

Finally the signalfd_siginfo fields are removed as they appear to be filling up the structure without userspace actually being able to use them.

v2: https://lkml.kernel.org/r/[email protected]
v1: https://lkml.kernel.org/r/[email protected]

Eric W. Biederman (9):
signal: Verify the alignment and size of siginfo_t
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Use dedicated helpers to send signals with si_trapno set
signal: Remove __ARCH_SI_TRAPNO
signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency
signal: Factor force_sig_perf out of perf_sigtrap
signal: Deliver all of the siginfo perf data in _perf
signalfd: Remove SIL_FAULT_PERF_EVENT fields from signalfd_siginfo

Quoted 4 lines

Marco Elver (3):
sparc64: Add compile-time asserts for siginfo_t offsets
arm: Add compile-time asserts for siginfo_t offsets
arm64: Add compile-time asserts for siginfo_t offsets

Quoted 5 lines

I can't seem to see the rest of them in my inbox. LKML also is missing
them: https://lore.kernel.org/linux-api/[email protected]/

Something must have swallowed them. Could you resend? I'll then test in the morning.

They got stuck going out you should see them any time now. Sorry about that.

Eric

#58 lore raw

PATCH v3 1/12 sparc64: Add compile-time asserts for siginfo_t offsets

In reply to Eric W. Biederman

From: Marco Elver <[email protected]>

To help catch ABI breaks at compile-time, add compile-time assertions to verify the siginfo_t layout. Unlike other architectures, sparc64 is special, because it is one of few architectures requiring si_trapno. ABI breaks around that field would only be caught here.

---
 arch/sparc/kernel/signal32.c  | 34 ++++++++++++++++++++++++++++++++++
 arch/sparc/kernel/signal_64.c | 33 +++++++++++++++++++++++++++++++++
 2 files changed, 67 insertions(+)
arch/sparc/kernel/signal32.c +34 -0
index e9695a06492f..778ed5c26d4a 100644
@@ -745,3 +745,37 @@ asmlinkage int do_sys32_sigstack(u32 u_ssptr, u32 u_ossptr, unsigned long sp)
 out:
 	return ret;
 }
+
+/*
+ * Compile-time assertions for siginfo_t offsets. Check NSIG* as well, as
+ * changes likely come with new fields that should be added below.
+ */
+static_assert(NSIGILL	== 11);
+static_assert(NSIGFPE	== 15);
+static_assert(NSIGSEGV	== 9);
+static_assert(NSIGBUS	== 5);
+static_assert(NSIGTRAP	== 6);
+static_assert(NSIGCHLD	== 6);
+static_assert(NSIGSYS	== 2);
+static_assert(offsetof(compat_siginfo_t, si_signo)	== 0x00);
+static_assert(offsetof(compat_siginfo_t, si_errno)	== 0x04);
+static_assert(offsetof(compat_siginfo_t, si_code)	== 0x08);
+static_assert(offsetof(compat_siginfo_t, si_pid)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_uid)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_tid)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_overrun)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_status)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_utime)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_stime)	== 0x1c);
+static_assert(offsetof(compat_siginfo_t, si_value)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_int)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_ptr)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_addr)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_trapno)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_addr_lsb)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_lower)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_upper)	== 0x1c);
+static_assert(offsetof(compat_siginfo_t, si_pkey)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_perf)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_band)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_fd)		== 0x10);
arch/sparc/kernel/signal_64.c +33 -0
index a0eec62c825d..c9bbf5f29078 100644
@@ -556,3 +556,36 @@ void do_notify_resume(struct pt_regs *regs, unsigned long orig_i0, unsigned long
 	user_enter();
 }
 
+/*
+ * Compile-time assertions for siginfo_t offsets. Check NSIG* as well, as
+ * changes likely come with new fields that should be added below.
+ */
+static_assert(NSIGILL	== 11);
+static_assert(NSIGFPE	== 15);
+static_assert(NSIGSEGV	== 9);
+static_assert(NSIGBUS	== 5);
+static_assert(NSIGTRAP	== 6);
+static_assert(NSIGCHLD	== 6);
+static_assert(NSIGSYS	== 2);
+static_assert(offsetof(siginfo_t, si_signo)	== 0x00);
+static_assert(offsetof(siginfo_t, si_errno)	== 0x04);
+static_assert(offsetof(siginfo_t, si_code)	== 0x08);
+static_assert(offsetof(siginfo_t, si_pid)	== 0x10);
+static_assert(offsetof(siginfo_t, si_uid)	== 0x14);
+static_assert(offsetof(siginfo_t, si_tid)	== 0x10);
+static_assert(offsetof(siginfo_t, si_overrun)	== 0x14);
+static_assert(offsetof(siginfo_t, si_status)	== 0x18);
+static_assert(offsetof(siginfo_t, si_utime)	== 0x20);
+static_assert(offsetof(siginfo_t, si_stime)	== 0x28);
+static_assert(offsetof(siginfo_t, si_value)	== 0x18);
+static_assert(offsetof(siginfo_t, si_int)	== 0x18);
+static_assert(offsetof(siginfo_t, si_ptr)	== 0x18);
+static_assert(offsetof(siginfo_t, si_addr)	== 0x10);
+static_assert(offsetof(siginfo_t, si_trapno)	== 0x18);
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x20);
+static_assert(offsetof(siginfo_t, si_lower)	== 0x28);
+static_assert(offsetof(siginfo_t, si_upper)	== 0x30);
+static_assert(offsetof(siginfo_t, si_pkey)	== 0x28);
+static_assert(offsetof(siginfo_t, si_perf)	== 0x20);
+static_assert(offsetof(siginfo_t, si_band)	== 0x10);
+static_assert(offsetof(siginfo_t, si_fd)	== 0x14);
Signature · 1 line
2.30.1
#59 lore raw

PATCH v3 2/12 arm: Add compile-time asserts for siginfo_t offsets

From: Marco Elver <[email protected]>

To help catch ABI breaks at compile-time, add compile-time assertions to verify the siginfo_t layout.

This could have caught that we cannot portably add 64-bit integers to siginfo_t on 32-bit architectures like Arm before reaching -next:
https://lkml.kernel.org/r/[email protected]

---
 arch/arm/kernel/signal.c | 36 ++++++++++++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)
arch/arm/kernel/signal.c +36 -0
index a3a38d0a4c85..2dac5d2c5cf6 100644
@@ -725,3 +725,39 @@ asmlinkage void do_rseq_syscall(struct pt_regs *regs)
 	rseq_syscall(regs);
 }
 #endif
+
+/*
+ * Compile-time assertions for siginfo_t offsets. Check NSIG* as well, as
+ * changes likely come with new fields that should be added below.
+ */
+static_assert(NSIGILL	== 11);
+static_assert(NSIGFPE	== 15);
+static_assert(NSIGSEGV	== 9);
+static_assert(NSIGBUS	== 5);
+static_assert(NSIGTRAP	== 6);
+static_assert(NSIGCHLD	== 6);
+static_assert(NSIGSYS	== 2);
+static_assert(offsetof(siginfo_t, si_signo)	== 0x00);
+static_assert(offsetof(siginfo_t, si_errno)	== 0x04);
+static_assert(offsetof(siginfo_t, si_code)	== 0x08);
+static_assert(offsetof(siginfo_t, si_pid)	== 0x0c);
+static_assert(offsetof(siginfo_t, si_uid)	== 0x10);
+static_assert(offsetof(siginfo_t, si_tid)	== 0x0c);
+static_assert(offsetof(siginfo_t, si_overrun)	== 0x10);
+static_assert(offsetof(siginfo_t, si_status)	== 0x14);
+static_assert(offsetof(siginfo_t, si_utime)	== 0x18);
+static_assert(offsetof(siginfo_t, si_stime)	== 0x1c);
+static_assert(offsetof(siginfo_t, si_value)	== 0x14);
+static_assert(offsetof(siginfo_t, si_int)	== 0x14);
+static_assert(offsetof(siginfo_t, si_ptr)	== 0x14);
+static_assert(offsetof(siginfo_t, si_addr)	== 0x0c);
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x10);
+static_assert(offsetof(siginfo_t, si_lower)	== 0x14);
+static_assert(offsetof(siginfo_t, si_upper)	== 0x18);
+static_assert(offsetof(siginfo_t, si_pkey)	== 0x14);
+static_assert(offsetof(siginfo_t, si_perf)	== 0x10);
+static_assert(offsetof(siginfo_t, si_band)	== 0x0c);
+static_assert(offsetof(siginfo_t, si_fd)	== 0x10);
+static_assert(offsetof(siginfo_t, si_call_addr)	== 0x0c);
+static_assert(offsetof(siginfo_t, si_syscall)	== 0x10);
+static_assert(offsetof(siginfo_t, si_arch)	== 0x14);
Signature · 1 line
2.30.1
#60 lore raw

PATCH v3 3/12 arm64: Add compile-time asserts for siginfo_t offsets

In reply to Eric W. Beiderman

From: Marco Elver <[email protected]>

To help catch ABI breaks at compile-time, add compile-time assertions to verify the siginfo_t layout.

---
 arch/arm64/kernel/signal.c   | 36 ++++++++++++++++++++++++++++++++++++
 arch/arm64/kernel/signal32.c | 36 ++++++++++++++++++++++++++++++++++++
 2 files changed, 72 insertions(+)
arch/arm64/kernel/signal.c +36 -0
index 6237486ff6bb..af8bd2af1298 100644
@@ -973,3 +973,39 @@ void __init minsigstksz_setup(void)
 		round_up(sizeof(struct frame_record), 16) +
 		16; /* max alignment padding */
 }
+
+/*
+ * Compile-time assertions for siginfo_t offsets. Check NSIG* as well, as
+ * changes likely come with new fields that should be added below.
+ */
+static_assert(NSIGILL	== 11);
+static_assert(NSIGFPE	== 15);
+static_assert(NSIGSEGV	== 9);
+static_assert(NSIGBUS	== 5);
+static_assert(NSIGTRAP	== 6);
+static_assert(NSIGCHLD	== 6);
+static_assert(NSIGSYS	== 2);
+static_assert(offsetof(siginfo_t, si_signo)	== 0x00);
+static_assert(offsetof(siginfo_t, si_errno)	== 0x04);
+static_assert(offsetof(siginfo_t, si_code)	== 0x08);
+static_assert(offsetof(siginfo_t, si_pid)	== 0x10);
+static_assert(offsetof(siginfo_t, si_uid)	== 0x14);
+static_assert(offsetof(siginfo_t, si_tid)	== 0x10);
+static_assert(offsetof(siginfo_t, si_overrun)	== 0x14);
+static_assert(offsetof(siginfo_t, si_status)	== 0x18);
+static_assert(offsetof(siginfo_t, si_utime)	== 0x20);
+static_assert(offsetof(siginfo_t, si_stime)	== 0x28);
+static_assert(offsetof(siginfo_t, si_value)	== 0x18);
+static_assert(offsetof(siginfo_t, si_int)	== 0x18);
+static_assert(offsetof(siginfo_t, si_ptr)	== 0x18);
+static_assert(offsetof(siginfo_t, si_addr)	== 0x10);
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x18);
+static_assert(offsetof(siginfo_t, si_lower)	== 0x20);
+static_assert(offsetof(siginfo_t, si_upper)	== 0x28);
+static_assert(offsetof(siginfo_t, si_pkey)	== 0x20);
+static_assert(offsetof(siginfo_t, si_perf)	== 0x18);
+static_assert(offsetof(siginfo_t, si_band)	== 0x10);
+static_assert(offsetof(siginfo_t, si_fd)	== 0x18);
+static_assert(offsetof(siginfo_t, si_call_addr)	== 0x10);
+static_assert(offsetof(siginfo_t, si_syscall)	== 0x18);
+static_assert(offsetof(siginfo_t, si_arch)	== 0x1c);
arch/arm64/kernel/signal32.c +36 -0
index 2f507f565c48..b6afb646515f 100644
@@ -457,3 +457,39 @@ void compat_setup_restart_syscall(struct pt_regs *regs)
 {
        regs->regs[7] = __NR_compat_restart_syscall;
 }
+
+/*
+ * Compile-time assertions for siginfo_t offsets. Check NSIG* as well, as
+ * changes likely come with new fields that should be added below.
+ */
+static_assert(NSIGILL	== 11);
+static_assert(NSIGFPE	== 15);
+static_assert(NSIGSEGV	== 9);
+static_assert(NSIGBUS	== 5);
+static_assert(NSIGTRAP	== 6);
+static_assert(NSIGCHLD	== 6);
+static_assert(NSIGSYS	== 2);
+static_assert(offsetof(compat_siginfo_t, si_signo)	== 0x00);
+static_assert(offsetof(compat_siginfo_t, si_errno)	== 0x04);
+static_assert(offsetof(compat_siginfo_t, si_code)	== 0x08);
+static_assert(offsetof(compat_siginfo_t, si_pid)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_uid)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_tid)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_overrun)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_status)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_utime)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_stime)	== 0x1c);
+static_assert(offsetof(compat_siginfo_t, si_value)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_int)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_ptr)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_addr)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_addr_lsb)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_lower)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_upper)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_pkey)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_perf)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_band)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_fd)		== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_call_addr)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_syscall)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_arch)	== 0x14);
Signature · 1 line
2.30.1
#61 lore raw

PATCH v3 4/12 signal: Verify the alignment and size of siginfo_t

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Update the static assertions about siginfo_t to also describe it's alignment and size.

While investigating if it was possible to add a 64bit field into siginfo_t[1] it became apparent that the alignment of siginfo_t is as much a part of the ABI as the size of the structure.

If the alignment changes siginfo_t when embedded in another structure can move to a different offset. Which is not acceptable from an ABI structure.

So document that fact and add static assertions to notify developers if they change change the alignment by accident.

[1] https://lkml.kernel.org/r/[email protected]

---
 arch/arm/kernel/signal.c           | 2 ++
 arch/arm64/kernel/signal.c         | 2 ++
 arch/arm64/kernel/signal32.c       | 2 ++
 arch/sparc/kernel/signal32.c       | 2 ++
 arch/sparc/kernel/signal_64.c      | 2 ++
 arch/x86/kernel/signal_compat.c    | 6 ++++++
 include/uapi/asm-generic/siginfo.h | 5 +++++
 7 files changed, 21 insertions(+)
arch/arm/kernel/signal.c +2 -0
index 2dac5d2c5cf6..643bcb0f091b 100644
@@ -737,6 +737,8 @@ static_assert(NSIGBUS	== 5);
 static_assert(NSIGTRAP	== 6);
 static_assert(NSIGCHLD	== 6);
 static_assert(NSIGSYS	== 2);
+static_assert(sizeof(siginfo_t) == 128);
+static_assert(__alignof__(siginfo_t) == 4);
 static_assert(offsetof(siginfo_t, si_signo)	== 0x00);
 static_assert(offsetof(siginfo_t, si_errno)	== 0x04);
 static_assert(offsetof(siginfo_t, si_code)	== 0x08);
arch/arm64/kernel/signal.c +2 -0
index af8bd2af1298..ad4bd27fc044 100644
@@ -985,6 +985,8 @@ static_assert(NSIGBUS	== 5);
 static_assert(NSIGTRAP	== 6);
 static_assert(NSIGCHLD	== 6);
 static_assert(NSIGSYS	== 2);
+static_assert(sizeof(siginfo_t) == 128);
+static_assert(__alignof__(siginfo_t) == 8);
 static_assert(offsetof(siginfo_t, si_signo)	== 0x00);
 static_assert(offsetof(siginfo_t, si_errno)	== 0x04);
 static_assert(offsetof(siginfo_t, si_code)	== 0x08);
arch/arm64/kernel/signal32.c +2 -0
index b6afb646515f..ee6c7484e130 100644
@@ -469,6 +469,8 @@ static_assert(NSIGBUS	== 5);
 static_assert(NSIGTRAP	== 6);
 static_assert(NSIGCHLD	== 6);
 static_assert(NSIGSYS	== 2);
+static_assert(sizeof(compat_siginfo_t) == 128);
+static_assert(__alignof__(compat_siginfo_t) == 4);
 static_assert(offsetof(compat_siginfo_t, si_signo)	== 0x00);
 static_assert(offsetof(compat_siginfo_t, si_errno)	== 0x04);
 static_assert(offsetof(compat_siginfo_t, si_code)	== 0x08);
arch/sparc/kernel/signal32.c +2 -0
index 778ed5c26d4a..32b977f253e3 100644
@@ -757,6 +757,8 @@ static_assert(NSIGBUS	== 5);
 static_assert(NSIGTRAP	== 6);
 static_assert(NSIGCHLD	== 6);
 static_assert(NSIGSYS	== 2);
+static_assert(sizeof(compat_siginfo_t) == 128);
+static_assert(__alignof__(compat_siginfo_t) == 4);
 static_assert(offsetof(compat_siginfo_t, si_signo)	== 0x00);
 static_assert(offsetof(compat_siginfo_t, si_errno)	== 0x04);
 static_assert(offsetof(compat_siginfo_t, si_code)	== 0x08);
arch/sparc/kernel/signal_64.c +2 -0
index c9bbf5f29078..e9dda9db156c 100644
@@ -567,6 +567,8 @@ static_assert(NSIGBUS	== 5);
 static_assert(NSIGTRAP	== 6);
 static_assert(NSIGCHLD	== 6);
 static_assert(NSIGSYS	== 2);
+static_assert(sizeof(siginfo_t) == 128);
+static_assert(__alignof__(siginfo_t) == 8);
 static_assert(offsetof(siginfo_t, si_signo)	== 0x00);
 static_assert(offsetof(siginfo_t, si_errno)	== 0x04);
 static_assert(offsetof(siginfo_t, si_code)	== 0x08);
arch/x86/kernel/signal_compat.c +6 -0
index 0e5d0a7e203b..e735bc129331 100644
@@ -34,7 +34,13 @@ static inline void signal_compat_build_tests(void)
 	BUILD_BUG_ON(NSIGSYS  != 2);
 
 	/* This is part of the ABI and can never change in size: */
+	BUILD_BUG_ON(sizeof(siginfo_t) != 128);
 	BUILD_BUG_ON(sizeof(compat_siginfo_t) != 128);
+
+	/* This is a part of the ABI and can never change in alignment */
+	BUILD_BUG_ON(__alignof__(siginfo_t) != 8);
+	BUILD_BUG_ON(__alignof__(compat_siginfo_t) != 4);
+
 	/*
 	 * The offsets of all the (unioned) si_fields are fixed
 	 * in the ABI, of course.  Make sure none of them ever
include/uapi/asm-generic/siginfo.h +5 -0
index 03d6f6d2c1fe..91c80d0c10c5 100644
@@ -29,6 +29,11 @@ typedef union sigval {
 #define __ARCH_SI_ATTRIBUTES
 #endif
 
+/*
+ * Be careful when extending this union.  On 32bit siginfo_t is 32bit
+ * aligned.  Which means that a 64bit field or any other field that
+ * would increase the alignment of siginfo_t will break the ABI.
+ */
 union __sifields {
 	/* kill() */
 	struct {
Signature · 1 line
2.30.1
author Acked-by on v3 4/12
#62 lore raw

Re: [PATCH v3 04/12] signal: Verify the alignment and size of siginfo_t

Quoted 13 lines · Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Update the static assertions about siginfo_t to also describe it's alignment and size.

While investigating if it was possible to add a 64bit field into siginfo_t[1] it became apparent that the alignment of siginfo_t is as much a part of the ABI as the size of the structure.

If the alignment changes siginfo_t when embedded in another structure can move to a different offset. Which is not acceptable from an ABI structure.

Quoted 5 lines

So document that fact and add static assertions to notify developers if they change change the alignment by accident.

[1] https://lkml.kernel.org/r/[email protected]

Quoted 28 lines
---
 arch/arm/kernel/signal.c           | 2 ++
 arch/arm64/kernel/signal.c         | 2 ++
 arch/arm64/kernel/signal32.c       | 2 ++
 arch/sparc/kernel/signal32.c       | 2 ++
 arch/sparc/kernel/signal_64.c      | 2 ++
 arch/x86/kernel/signal_compat.c    | 6 ++++++
 include/uapi/asm-generic/siginfo.h | 5 +++++
 7 files changed, 21 insertions(+)
arch/arm/kernel/signal.c +2 -0
index 2dac5d2c5cf6..643bcb0f091b 100644
@@ -737,6 +737,8 @@ static_assert(NSIGBUS       == 5);
 static_assert(NSIGTRAP == 6);
 static_assert(NSIGCHLD == 6);
 static_assert(NSIGSYS  == 2);
+static_assert(sizeof(siginfo_t) == 128);
+static_assert(__alignof__(siginfo_t) == 4);
 static_assert(offsetof(siginfo_t, si_signo)    == 0x00);
 static_assert(offsetof(siginfo_t, si_errno)    == 0x04);
 static_assert(offsetof(siginfo_t, si_code)     == 0x08);
arch/arm64/kernel/signal.c +0 -0
index af8bd2af1298..ad4bd27fc044 100644

arch/arm64/kernel/signal.c static_assert(NSIGBUS == 5); · line 988

Quoted 4 lines
arch/arm64/kernel/signal.c +1 -0
@@ -985,6 +985,8 @@ static_assert(NSIGBUS       == 5);
 static_assert(NSIGTRAP == 6);
 static_assert(NSIGCHLD == 6);
 static_assert(NSIGSYS  == 2);
+static_assert(sizeof(siginfo_t) == 128);

Would using SI_MAX_SIZE be appropriate? Perhaps not.. in case somebody changes it, given these static asserts are meant to double-check.

I leave it to you to decide what makes more sense.

Quoted 80 lines
arch/arm64/kernel/signal.c +1 -0
+static_assert(__alignof__(siginfo_t) == 8);
 static_assert(offsetof(siginfo_t, si_signo)    == 0x00);
 static_assert(offsetof(siginfo_t, si_errno)    == 0x04);
 static_assert(offsetof(siginfo_t, si_code)     == 0x08);
arch/arm64/kernel/signal32.c +2 -0
index b6afb646515f..ee6c7484e130 100644
@@ -469,6 +469,8 @@ static_assert(NSIGBUS       == 5);
 static_assert(NSIGTRAP == 6);
 static_assert(NSIGCHLD == 6);
 static_assert(NSIGSYS  == 2);
+static_assert(sizeof(compat_siginfo_t) == 128);
+static_assert(__alignof__(compat_siginfo_t) == 4);
 static_assert(offsetof(compat_siginfo_t, si_signo)     == 0x00);
 static_assert(offsetof(compat_siginfo_t, si_errno)     == 0x04);
 static_assert(offsetof(compat_siginfo_t, si_code)      == 0x08);
arch/sparc/kernel/signal32.c +2 -0
index 778ed5c26d4a..32b977f253e3 100644
@@ -757,6 +757,8 @@ static_assert(NSIGBUS       == 5);
 static_assert(NSIGTRAP == 6);
 static_assert(NSIGCHLD == 6);
 static_assert(NSIGSYS  == 2);
+static_assert(sizeof(compat_siginfo_t) == 128);
+static_assert(__alignof__(compat_siginfo_t) == 4);
 static_assert(offsetof(compat_siginfo_t, si_signo)     == 0x00);
 static_assert(offsetof(compat_siginfo_t, si_errno)     == 0x04);
 static_assert(offsetof(compat_siginfo_t, si_code)      == 0x08);
arch/sparc/kernel/signal_64.c +2 -0
index c9bbf5f29078..e9dda9db156c 100644
@@ -567,6 +567,8 @@ static_assert(NSIGBUS       == 5);
 static_assert(NSIGTRAP == 6);
 static_assert(NSIGCHLD == 6);
 static_assert(NSIGSYS  == 2);
+static_assert(sizeof(siginfo_t) == 128);
+static_assert(__alignof__(siginfo_t) == 8);
 static_assert(offsetof(siginfo_t, si_signo)    == 0x00);
 static_assert(offsetof(siginfo_t, si_errno)    == 0x04);
 static_assert(offsetof(siginfo_t, si_code)     == 0x08);
arch/x86/kernel/signal_compat.c +6 -0
index 0e5d0a7e203b..e735bc129331 100644
@@ -34,7 +34,13 @@ static inline void signal_compat_build_tests(void)
        BUILD_BUG_ON(NSIGSYS  != 2);
 
        /* This is part of the ABI and can never change in size: */
+       BUILD_BUG_ON(sizeof(siginfo_t) != 128);
        BUILD_BUG_ON(sizeof(compat_siginfo_t) != 128);
+
+       /* This is a part of the ABI and can never change in alignment */
+       BUILD_BUG_ON(__alignof__(siginfo_t) != 8);
+       BUILD_BUG_ON(__alignof__(compat_siginfo_t) != 4);
+
        /*
         * The offsets of all the (unioned) si_fields are fixed
         * in the ABI, of course.  Make sure none of them ever
include/uapi/asm-generic/siginfo.h +5 -1
index 03d6f6d2c1fe..91c80d0c10c5 100644
@@ -29,6 +29,11 @@ typedef union sigval {
 #define __ARCH_SI_ATTRIBUTES
 #endif
 
+/*
+ * Be careful when extending this union.  On 32bit siginfo_t is 32bit
+ * aligned.  Which means that a 64bit field or any other field that
+ * would increase the alignment of siginfo_t will break the ABI.
+ */
 union __sifields {
        /* kill() */
        struct {
--

2.30.1

#63 lore raw

PATCH v3 5/12 siginfo: Move si_trapno inside the union inside _si_fault

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

It turns out that linux uses si_trapno very sparingly, and as such it can be considered extra information for a very narrow selection of signals, rather than information that is present with every fault reported in siginfo.

As such move si_trapno inside the union inside of _si_fault. This results in no change in placement, and makes it eaiser
to extend _si_fault in the future as this reduces the number of special cases. In particular with si_trapno included in the union it is no longer a concern that the union must be pointer alligned on most architectures because the union followes immediately after si_addr which is a pointer.

This change results in a difference in siginfo field placement on sparc and alpha for the fields si_addr_lsb, si_lower, si_upper, si_pkey, and si_perf. These architectures do not implement the signals that would use si_addr_lsb, si_lower, si_upper, si_pkey, and si_perf. Further these architecture have not yet implemented the userspace that would use si_perf.

The point of this change is in fact to correct these placement issues before sparc or alpha grow userspace that cares. This change was discussed[1] and the agreement is that this change is currently safe.

[1]: https://lkml.kernel.org/r/CAK8P3a0+uKYwL1NhY6Hvtieghba2hKYGD6hcKx5n8=4Gtt+pHA@mail.gmail.com

v1: https://lkml.kernel.org/r/[email protected]

---
 arch/sparc/kernel/signal32.c       | 10 +++++-----
 arch/sparc/kernel/signal_64.c      | 10 +++++-----
 arch/x86/kernel/signal_compat.c    |  3 +++
 include/linux/compat.h             |  5 ++---
 include/uapi/asm-generic/siginfo.h |  7 ++-----
 kernel/signal.c                    |  1 +
 6 files changed, 18 insertions(+), 18 deletions(-)
arch/sparc/kernel/signal32.c +5 -5
index 32b977f253e3..5573722e34ad 100644
@@ -774,10 +774,10 @@ static_assert(offsetof(compat_siginfo_t, si_int)	== 0x14);
 static_assert(offsetof(compat_siginfo_t, si_ptr)	== 0x14);
 static_assert(offsetof(compat_siginfo_t, si_addr)	== 0x0c);
 static_assert(offsetof(compat_siginfo_t, si_trapno)	== 0x10);
-static_assert(offsetof(compat_siginfo_t, si_addr_lsb)	== 0x14);
-static_assert(offsetof(compat_siginfo_t, si_lower)	== 0x18);
-static_assert(offsetof(compat_siginfo_t, si_upper)	== 0x1c);
-static_assert(offsetof(compat_siginfo_t, si_pkey)	== 0x18);
-static_assert(offsetof(compat_siginfo_t, si_perf)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_addr_lsb)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_lower)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_upper)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_pkey)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_perf)	== 0x10);
 static_assert(offsetof(compat_siginfo_t, si_band)	== 0x0c);
 static_assert(offsetof(compat_siginfo_t, si_fd)		== 0x10);
arch/sparc/kernel/signal_64.c +5 -5
index e9dda9db156c..a69a78984c36 100644
@@ -584,10 +584,10 @@ static_assert(offsetof(siginfo_t, si_int)	== 0x18);
 static_assert(offsetof(siginfo_t, si_ptr)	== 0x18);
 static_assert(offsetof(siginfo_t, si_addr)	== 0x10);
 static_assert(offsetof(siginfo_t, si_trapno)	== 0x18);
-static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x20);
-static_assert(offsetof(siginfo_t, si_lower)	== 0x28);
-static_assert(offsetof(siginfo_t, si_upper)	== 0x30);
-static_assert(offsetof(siginfo_t, si_pkey)	== 0x28);
-static_assert(offsetof(siginfo_t, si_perf)	== 0x20);
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x18);
+static_assert(offsetof(siginfo_t, si_lower)	== 0x20);
+static_assert(offsetof(siginfo_t, si_upper)	== 0x28);
+static_assert(offsetof(siginfo_t, si_pkey)	== 0x20);
+static_assert(offsetof(siginfo_t, si_perf)	== 0x18);
 static_assert(offsetof(siginfo_t, si_band)	== 0x10);
 static_assert(offsetof(siginfo_t, si_fd)	== 0x14);
arch/x86/kernel/signal_compat.c +3 -0
index e735bc129331..c9601f092a1e 100644
@@ -133,6 +133,9 @@ static inline void signal_compat_build_tests(void)
 	BUILD_BUG_ON(offsetof(siginfo_t, si_addr) != 0x10);
 	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_addr) != 0x0C);
 
+	BUILD_BUG_ON(offsetof(siginfo_t, si_trapno) != 0x18);
+	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_trapno) != 0x10);
+
 	BUILD_BUG_ON(offsetof(siginfo_t, si_addr_lsb) != 0x18);
 	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_addr_lsb) != 0x10);
 
include/linux/compat.h +2 -3
index f0d2dd35d408..6af7bef15e94 100644
@@ -214,12 +214,11 @@ typedef struct compat_siginfo {
 		/* SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT */
 		struct {
 			compat_uptr_t _addr;	/* faulting insn/memory ref. */
-#ifdef __ARCH_SI_TRAPNO
-			int _trapno;	/* TRAP # which caused the signal */
-#endif
 #define __COMPAT_ADDR_BND_PKEY_PAD  (__alignof__(compat_uptr_t) < sizeof(short) ? \
 				     sizeof(short) : __alignof__(compat_uptr_t))
 			union {
+				/* used on alpha and sparc */
+				int _trapno;	/* TRAP # which caused the signal */
 				/*
 				 * used when si_code=BUS_MCEERR_AR or
 				 * used when si_code=BUS_MCEERR_AO
include/uapi/asm-generic/siginfo.h +2 -5
index 91c80d0c10c5..3503282021aa 100644
@@ -68,9 +68,6 @@ union __sifields {
 	/* SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT */
 	struct {
 		void __user *_addr; /* faulting insn/memory ref. */
-#ifdef __ARCH_SI_TRAPNO
-		int _trapno;	/* TRAP # which caused the signal */
-#endif
 #ifdef __ia64__
 		int _imm;		/* immediate value for "break" */
 		unsigned int _flags;	/* see ia64 si_flags */
@@ -80,6 +77,8 @@ union __sifields {
 #define __ADDR_BND_PKEY_PAD  (__alignof__(void *) < sizeof(short) ? \
 			      sizeof(short) : __alignof__(void *))
 		union {
+			/* used on alpha and sparc */
+			int _trapno;	/* TRAP # which caused the signal */
 			/*
 			 * used when si_code=BUS_MCEERR_AR or
 			 * used when si_code=BUS_MCEERR_AO
@@ -155,9 +154,7 @@ typedef struct siginfo {
 #define si_int		_sifields._rt._sigval.sival_int
 #define si_ptr		_sifields._rt._sigval.sival_ptr
 #define si_addr		_sifields._sigfault._addr
-#ifdef __ARCH_SI_TRAPNO
 #define si_trapno	_sifields._sigfault._trapno
-#endif
 #define si_addr_lsb	_sifields._sigfault._addr_lsb
 #define si_lower	_sifields._sigfault._addr_bnd._lower
 #define si_upper	_sifields._sigfault._addr_bnd._upper
kernel/signal.c +1 -0
index c3017aa8024a..65888aec65a0 100644
@@ -4607,6 +4607,7 @@ static inline void siginfo_buildtime_checks(void)
 
 	/* sigfault */
 	CHECK_OFFSET(si_addr);
+	CHECK_OFFSET(si_trapno);
 	CHECK_OFFSET(si_addr_lsb);
 	CHECK_OFFSET(si_lower);
 	CHECK_OFFSET(si_upper);
Signature · 1 line
2.30.1
#64 lore raw

PATCH v3 6/12 signal: Implement SIL_FAULT_TRAPNO

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Now that si_trapno is part of the union in _si_fault and available on all architectures, add SIL_FAULT_TRAPNO and update siginfo_layout to return SIL_FAULT_TRAPNO when si_trapno is actually used.

Update the code that uses siginfo_layout to deal with SIL_FAULT_TRAPNO and have the same code ignore si_trapno in in all other cases.

v1: https://lkml.kernel.org/r/[email protected]

---
 fs/signalfd.c          |  8 +++-----
 include/linux/signal.h |  1 +
 kernel/signal.c        | 37 +++++++++++++++----------------------
 3 files changed, 19 insertions(+), 27 deletions(-)
fs/signalfd.c +3 -5
index 040a1142915f..e87e59581653 100644
@@ -123,15 +123,13 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		 */
 	case SIL_FAULT:
 		new.ssi_addr = (long) kinfo->si_addr;
-#ifdef __ARCH_SI_TRAPNO
+		break;
+	case SIL_FAULT_TRAPNO:
+		new.ssi_addr = (long) kinfo->si_addr;
 		new.ssi_trapno = kinfo->si_trapno;
-#endif
 		break;
 	case SIL_FAULT_MCEERR:
 		new.ssi_addr = (long) kinfo->si_addr;
-#ifdef __ARCH_SI_TRAPNO
-		new.ssi_trapno = kinfo->si_trapno;
-#endif
 		new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
 		break;
 	case SIL_PERF_EVENT:
include/linux/signal.h +1 -0
index 1e98548d7cf6..5160fd45e5ca 100644
@@ -40,6 +40,7 @@ enum siginfo_layout {
 	SIL_TIMER,
 	SIL_POLL,
 	SIL_FAULT,
+	SIL_FAULT_TRAPNO,
 	SIL_FAULT_MCEERR,
 	SIL_FAULT_BNDERR,
 	SIL_FAULT_PKUERR,
kernel/signal.c +15 -22
index 65888aec65a0..3d3ba7949788 100644
@@ -1194,6 +1194,7 @@ static inline bool has_si_pid_and_uid(struct kernel_siginfo *info)
 	case SIL_TIMER:
 	case SIL_POLL:
 	case SIL_FAULT:
+	case SIL_FAULT_TRAPNO:
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
@@ -2527,6 +2528,7 @@ static void hide_si_addr_tag_bits(struct ksignal *ksig)
 {
 	switch (siginfo_layout(ksig->sig, ksig->info.si_code)) {
 	case SIL_FAULT:
+	case SIL_FAULT_TRAPNO:
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
@@ -3206,6 +3208,13 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
 			if ((sig == SIGBUS) &&
 			    (si_code >= BUS_MCEERR_AR) && (si_code <= BUS_MCEERR_AO))
 				layout = SIL_FAULT_MCEERR;
+			else if (IS_ENABLED(CONFIG_ALPHA) &&
+				 ((sig == SIGFPE) ||
+				  ((sig == SIGTRAP) && (si_code == TRAP_UNK))))
+				layout = SIL_FAULT_TRAPNO;
+			else if (IS_ENABLED(CONFIG_SPARC) &&
+				 (sig == SIGILL) && (si_code == ILL_ILLTRP))
+				layout = SIL_FAULT_TRAPNO;
 			else if ((sig == SIGSEGV) && (si_code == SEGV_BNDERR))
 				layout = SIL_FAULT_BNDERR;
 #ifdef SEGV_PKUERR
@@ -3317,30 +3326,22 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 		break;
 	case SIL_FAULT:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+		break;
+	case SIL_FAULT_TRAPNO:
+		to->si_addr = ptr_to_compat(from->si_addr);
 		to->si_trapno = from->si_trapno;
-#endif
 		break;
 	case SIL_FAULT_MCEERR:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_addr_lsb = from->si_addr_lsb;
 		break;
 	case SIL_FAULT_BNDERR:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_lower = ptr_to_compat(from->si_lower);
 		to->si_upper = ptr_to_compat(from->si_upper);
 		break;
 	case SIL_FAULT_PKUERR:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_pkey = from->si_pkey;
 		break;
 	case SIL_PERF_EVENT:
@@ -3401,30 +3402,22 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 		break;
 	case SIL_FAULT:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+		break;
+	case SIL_FAULT_TRAPNO:
+		to->si_addr = compat_ptr(from->si_addr);
 		to->si_trapno = from->si_trapno;
-#endif
 		break;
 	case SIL_FAULT_MCEERR:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_addr_lsb = from->si_addr_lsb;
 		break;
 	case SIL_FAULT_BNDERR:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_lower = compat_ptr(from->si_lower);
 		to->si_upper = compat_ptr(from->si_upper);
 		break;
 	case SIL_FAULT_PKUERR:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_pkey = from->si_pkey;
 		break;
 	case SIL_PERF_EVENT:
Signature · 1 line
2.30.1
author Reviewed-by on v3 6/12
#65 lore raw

Re: [PATCH v3 06/12] signal: Implement SIL_FAULT_TRAPNO

Quoted 11 lines · Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Now that si_trapno is part of the union in _si_fault and available on all architectures, add SIL_FAULT_TRAPNO and update siginfo_layout to return SIL_FAULT_TRAPNO when si_trapno is actually used.

Update the code that uses siginfo_layout to deal with SIL_FAULT_TRAPNO and have the same code ignore si_trapno in in all other cases.

v1: https://lkml.kernel.org/r/[email protected]

Quoted 145 lines
---
 fs/signalfd.c          |  8 +++-----
 include/linux/signal.h |  1 +
 kernel/signal.c        | 37 +++++++++++++++----------------------
 3 files changed, 19 insertions(+), 27 deletions(-)
fs/signalfd.c +3 -5
index 040a1142915f..e87e59581653 100644
@@ -123,15 +123,13 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
                 */
        case SIL_FAULT:
                new.ssi_addr = (long) kinfo->si_addr;
-#ifdef __ARCH_SI_TRAPNO
+               break;
+       case SIL_FAULT_TRAPNO:
+               new.ssi_addr = (long) kinfo->si_addr;
                new.ssi_trapno = kinfo->si_trapno;
-#endif
                break;
        case SIL_FAULT_MCEERR:
                new.ssi_addr = (long) kinfo->si_addr;
-#ifdef __ARCH_SI_TRAPNO
-               new.ssi_trapno = kinfo->si_trapno;
-#endif
                new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
                break;
        case SIL_PERF_EVENT:
include/linux/signal.h +1 -0
index 1e98548d7cf6..5160fd45e5ca 100644
@@ -40,6 +40,7 @@ enum siginfo_layout {
        SIL_TIMER,
        SIL_POLL,
        SIL_FAULT,
+       SIL_FAULT_TRAPNO,
        SIL_FAULT_MCEERR,
        SIL_FAULT_BNDERR,
        SIL_FAULT_PKUERR,
kernel/signal.c +15 -23
index 65888aec65a0..3d3ba7949788 100644
@@ -1194,6 +1194,7 @@ static inline bool has_si_pid_and_uid(struct kernel_siginfo *info)
        case SIL_TIMER:
        case SIL_POLL:
        case SIL_FAULT:
+       case SIL_FAULT_TRAPNO:
        case SIL_FAULT_MCEERR:
        case SIL_FAULT_BNDERR:
        case SIL_FAULT_PKUERR:
@@ -2527,6 +2528,7 @@ static void hide_si_addr_tag_bits(struct ksignal *ksig)
 {
        switch (siginfo_layout(ksig->sig, ksig->info.si_code)) {
        case SIL_FAULT:
+       case SIL_FAULT_TRAPNO:
        case SIL_FAULT_MCEERR:
        case SIL_FAULT_BNDERR:
        case SIL_FAULT_PKUERR:
@@ -3206,6 +3208,13 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
                        if ((sig == SIGBUS) &&
                            (si_code >= BUS_MCEERR_AR) && (si_code <= BUS_MCEERR_AO))
                                layout = SIL_FAULT_MCEERR;
+                       else if (IS_ENABLED(CONFIG_ALPHA) &&
+                                ((sig == SIGFPE) ||
+                                 ((sig == SIGTRAP) && (si_code == TRAP_UNK))))
+                               layout = SIL_FAULT_TRAPNO;
+                       else if (IS_ENABLED(CONFIG_SPARC) &&
+                                (sig == SIGILL) && (si_code == ILL_ILLTRP))
+                               layout = SIL_FAULT_TRAPNO;
                        else if ((sig == SIGSEGV) && (si_code == SEGV_BNDERR))
                                layout = SIL_FAULT_BNDERR;
 #ifdef SEGV_PKUERR
@@ -3317,30 +3326,22 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
                break;
        case SIL_FAULT:
                to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+               break;
+       case SIL_FAULT_TRAPNO:
+               to->si_addr = ptr_to_compat(from->si_addr);
                to->si_trapno = from->si_trapno;
-#endif
                break;
        case SIL_FAULT_MCEERR:
                to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_addr_lsb = from->si_addr_lsb;
                break;
        case SIL_FAULT_BNDERR:
                to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_lower = ptr_to_compat(from->si_lower);
                to->si_upper = ptr_to_compat(from->si_upper);
                break;
        case SIL_FAULT_PKUERR:
                to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_pkey = from->si_pkey;
                break;
        case SIL_PERF_EVENT:
@@ -3401,30 +3402,22 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
                break;
        case SIL_FAULT:
                to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+               break;
+       case SIL_FAULT_TRAPNO:
+               to->si_addr = compat_ptr(from->si_addr);
                to->si_trapno = from->si_trapno;
-#endif
                break;
        case SIL_FAULT_MCEERR:
                to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_addr_lsb = from->si_addr_lsb;
                break;
        case SIL_FAULT_BNDERR:
                to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_lower = compat_ptr(from->si_lower);
                to->si_upper = compat_ptr(from->si_upper);
                break;
        case SIL_FAULT_PKUERR:
                to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-               to->si_trapno = from->si_trapno;
-#endif
                to->si_pkey = from->si_pkey;
                break;
        case SIL_PERF_EVENT:
--

2.30.1

#66 lore raw

PATCH v3 7/12 signal: Use dedicated helpers to send signals with si_trapno set

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Now that si_trapno is no longer expected to be present for every fault reported using siginfo on alpha and sparc remove the trapno parameter from force_sig_fault, force_sig_fault_to_task and send_sig_fault.

Add two new helpers force_sig_fault_trapno and send_sig_fautl_trapno for those signals where trapno is expected to be set.

v1: https://lkml.kernel.org/r/[email protected]

---
 arch/alpha/kernel/osf_sys.c      |  2 +-
 arch/alpha/kernel/signal.c       |  4 +--
 arch/alpha/kernel/traps.c        | 24 ++++++++---------
 arch/alpha/mm/fault.c            |  4 +--
 arch/sparc/kernel/process_64.c   |  2 +-
 arch/sparc/kernel/sys_sparc_32.c |  2 +-
 arch/sparc/kernel/sys_sparc_64.c |  2 +-
 arch/sparc/kernel/traps_32.c     | 22 ++++++++--------
 arch/sparc/kernel/traps_64.c     | 44 ++++++++++++++------------------
 arch/sparc/kernel/unaligned_32.c |  2 +-
 arch/sparc/mm/fault_32.c         |  2 +-
 arch/sparc/mm/fault_64.c         |  2 +-
 include/linux/sched/signal.h     | 12 +++------
 kernel/signal.c                  | 41 +++++++++++++++++++++--------
 14 files changed, 88 insertions(+), 77 deletions(-)
arch/alpha/kernel/osf_sys.c +1 -1
index d5367a1c6300..80c5d7fbe66a 100644
@@ -878,7 +878,7 @@ SYSCALL_DEFINE5(osf_setsysinfo, unsigned long, op, void __user *, buffer,
 
 			send_sig_fault(SIGFPE, si_code,
 				       (void __user *)NULL,  /* FIXME */
-				       0, current);
+				       current);
  		}
 		return 0;
 	}
arch/alpha/kernel/signal.c +2 -2
index 948b89789da8..bc077babafab 100644
@@ -219,7 +219,7 @@ do_sigreturn(struct sigcontext __user *sc)
 
 	/* Send SIGTRAP if we're single-stepping: */
 	if (ptrace_cancel_bpt (current)) {
-		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc, 0,
+		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc,
 			       current);
 	}
 	return;
@@ -247,7 +247,7 @@ do_rt_sigreturn(struct rt_sigframe __user *frame)
 
 	/* Send SIGTRAP if we're single-stepping: */
 	if (ptrace_cancel_bpt (current)) {
-		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc, 0,
+		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc,
 			       current);
 	}
 	return;
arch/alpha/kernel/traps.c +12 -12
index 921d4b6e4d95..0dddf9ecc1f4 100644
@@ -227,7 +227,7 @@ do_entArith(unsigned long summary, unsigned long write_mask,
 	}
 	die_if_kernel("Arithmetic fault", regs, 0, NULL);
 
-	send_sig_fault(SIGFPE, si_code, (void __user *) regs->pc, 0, current);
+	send_sig_fault_trapno(SIGFPE, si_code, (void __user *) regs->pc, 0, current);
 }
 
 asmlinkage void
@@ -268,12 +268,12 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 			regs->pc -= 4;	/* make pc point to former bpt */
 		}
 
-		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc, 0,
+		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc,
 			       current);
 		return;
 
 	      case 1: /* bugcheck */
-		send_sig_fault(SIGTRAP, TRAP_UNK, (void __user *) regs->pc, 0,
+		send_sig_fault(SIGTRAP, TRAP_UNK, (void __user *) regs->pc,
 			       current);
 		return;
 		
@@ -335,8 +335,8 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 			break;
 		}
 
-		send_sig_fault(signo, code, (void __user *) regs->pc, regs->r16,
-			       current);
+		send_sig_fault_trapno(signo, code, (void __user *) regs->pc,
+				      regs->r16, current);
 		return;
 
 	      case 4: /* opDEC */
@@ -360,9 +360,9 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 			if (si_code == 0)
 				return;
 			if (si_code > 0) {
-				send_sig_fault(SIGFPE, si_code,
-					       (void __user *) regs->pc, 0,
-					       current);
+				send_sig_fault_trapno(SIGFPE, si_code,
+						      (void __user *) regs->pc,
+						      0, current);
 				return;
 			}
 		}
@@ -387,7 +387,7 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 		      ;
 	}
 
-	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, 0, current);
+	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, current);
 }
 
 /* There is an ifdef in the PALcode in MILO that enables a 
@@ -402,7 +402,7 @@ do_entDbg(struct pt_regs *regs)
 {
 	die_if_kernel("Instruction fault", regs, 0, NULL);
 
-	force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, 0);
+	force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc);
 }
 
 
@@ -964,12 +964,12 @@ do_entUnaUser(void __user * va, unsigned long opcode,
 			si_code = SEGV_MAPERR;
 		mmap_read_unlock(mm);
 	}
-	send_sig_fault(SIGSEGV, si_code, va, 0, current);
+	send_sig_fault(SIGSEGV, si_code, va, current);
 	return;
 
 give_sigbus:
 	regs->pc -= 4;
-	send_sig_fault(SIGBUS, BUS_ADRALN, va, 0, current);
+	send_sig_fault(SIGBUS, BUS_ADRALN, va, current);
 	return;
 }
 
arch/alpha/mm/fault.c +2 -2
index 09172f017efc..eee5102c3d88 100644
@@ -219,13 +219,13 @@ do_page_fault(unsigned long address, unsigned long mmcsr,
 	mmap_read_unlock(mm);
 	/* Send a sigbus, regardless of whether we were in kernel
 	   or user mode.  */
-	force_sig_fault(SIGBUS, BUS_ADRERR, (void __user *) address, 0);
+	force_sig_fault(SIGBUS, BUS_ADRERR, (void __user *) address);
 	if (!user_mode(regs))
 		goto no_context;
 	return;
 
  do_sigsegv:
-	force_sig_fault(SIGSEGV, si_code, (void __user *) address, 0);
+	force_sig_fault(SIGSEGV, si_code, (void __user *) address);
 	return;
 
 #ifdef CONFIG_ALPHA_LARGE_VMALLOC
arch/sparc/kernel/process_64.c +1 -1
index 7afd0a859a78..29e67854d5a4 100644
@@ -518,7 +518,7 @@ void synchronize_user_stack(void)
 
 static void stack_unaligned(unsigned long sp)
 {
-	force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) sp, 0);
+	force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) sp);
 }
 
 static const char uwfault32[] = KERN_INFO \
arch/sparc/kernel/sys_sparc_32.c +1 -1
index be77538bc038..082a551897ed 100644
@@ -151,7 +151,7 @@ sparc_breakpoint (struct pt_regs *regs)
 #ifdef DEBUG_SPARC_BREAKPOINT
         printk ("TRAP: Entering kernel PC=%x, nPC=%x\n", regs->pc, regs->npc);
 #endif
-	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc, 0);
+	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc);
 
 #ifdef DEBUG_SPARC_BREAKPOINT
 	printk ("TRAP: Returning to space: PC=%x nPC=%x\n", regs->pc, regs->npc);
arch/sparc/kernel/sys_sparc_64.c +1 -1
index 6b92fadb6ec7..1e9a9e016237 100644
@@ -514,7 +514,7 @@ asmlinkage void sparc_breakpoint(struct pt_regs *regs)
 #ifdef DEBUG_SPARC_BREAKPOINT
         printk ("TRAP: Entering kernel PC=%lx, nPC=%lx\n", regs->tpc, regs->tnpc);
 #endif
-	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->tpc, 0);
+	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->tpc);
 #ifdef DEBUG_SPARC_BREAKPOINT
 	printk ("TRAP: Returning to space: PC=%lx nPC=%lx\n", regs->tpc, regs->tnpc);
 #endif
arch/sparc/kernel/traps_32.c +11 -11
index 247a0d9683b2..5630e5a395e0 100644
@@ -102,8 +102,8 @@ void do_hw_interrupt(struct pt_regs *regs, unsigned long type)
 	if(regs->psr & PSR_PS)
 		die_if_kernel("Kernel bad trap", regs);
 
-	force_sig_fault(SIGILL, ILL_ILLTRP,
-			(void __user *)regs->pc, type - 0x80);
+	force_sig_fault_trapno(SIGILL, ILL_ILLTRP,
+			       (void __user *)regs->pc, type - 0x80);
 }
 
 void do_illegal_instruction(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -116,7 +116,7 @@ void do_illegal_instruction(struct pt_regs *regs, unsigned long pc, unsigned lon
 	       regs->pc, *(unsigned long *)regs->pc);
 #endif
 
-	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc, current);
 }
 
 void do_priv_instruction(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -124,7 +124,7 @@ void do_priv_instruction(struct pt_regs *regs, unsigned long pc, unsigned long n
 {
 	if(psr & PSR_PS)
 		die_if_kernel("Penguin instruction from Penguin mode??!?!", regs);
-	send_sig_fault(SIGILL, ILL_PRVOPC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_PRVOPC, (void __user *)pc, current);
 }
 
 /* XXX User may want to be allowed to do this. XXX */
@@ -145,7 +145,7 @@ void do_memaccess_unaligned(struct pt_regs *regs, unsigned long pc, unsigned lon
 #endif
 	send_sig_fault(SIGBUS, BUS_ADRALN,
 		       /* FIXME: Should dig out mna address */ (void *)0,
-		       0, current);
+		       current);
 }
 
 static unsigned long init_fsr = 0x0UL;
@@ -291,7 +291,7 @@ void do_fpe_trap(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 		else if (fsr & 0x01)
 			code = FPE_FLTRES;
 	}
-	send_sig_fault(SIGFPE, code, (void __user *)pc, 0, fpt);
+	send_sig_fault(SIGFPE, code, (void __user *)pc, fpt);
 #ifndef CONFIG_SMP
 	last_task_used_math = NULL;
 #endif
@@ -305,7 +305,7 @@ void handle_tag_overflow(struct pt_regs *regs, unsigned long pc, unsigned long n
 {
 	if(psr & PSR_PS)
 		die_if_kernel("Penguin overflow trap from kernel mode", regs);
-	send_sig_fault(SIGEMT, EMT_TAGOVF, (void __user *)pc, 0, current);
+	send_sig_fault(SIGEMT, EMT_TAGOVF, (void __user *)pc, current);
 }
 
 void handle_watchpoint(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -327,13 +327,13 @@ void handle_reg_access(struct pt_regs *regs, unsigned long pc, unsigned long npc
 	printk("Register Access Exception at PC %08lx NPC %08lx PSR %08lx\n",
 	       pc, npc, psr);
 #endif
-	force_sig_fault(SIGBUS, BUS_OBJERR, (void __user *)pc, 0);
+	force_sig_fault(SIGBUS, BUS_OBJERR, (void __user *)pc);
 }
 
 void handle_cp_disabled(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 			unsigned long psr)
 {
-	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, current);
 }
 
 void handle_cp_exception(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -343,13 +343,13 @@ void handle_cp_exception(struct pt_regs *regs, unsigned long pc, unsigned long n
 	printk("Co-Processor Exception at PC %08lx NPC %08lx PSR %08lx\n",
 	       pc, npc, psr);
 #endif
-	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, current);
 }
 
 void handle_hw_divzero(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 		       unsigned long psr)
 {
-	send_sig_fault(SIGFPE, FPE_INTDIV, (void __user *)pc, 0, current);
+	send_sig_fault(SIGFPE, FPE_INTDIV, (void __user *)pc, current);
 }
 
 #ifdef CONFIG_DEBUG_BUGVERBOSE
arch/sparc/kernel/traps_64.c +19 -25

144 lines not on this page · raw message

arch/sparc/kernel/unaligned_32.c +1 -1
index ef5c5207c9ff..455f0258c745 100644
@@ -278,5 +278,5 @@ asmlinkage void user_unaligned_trap(struct pt_regs *regs, unsigned int insn)
 {
 	send_sig_fault(SIGBUS, BUS_ADRALN,
 		       (void __user *)safe_compute_effective_address(regs, insn),
-		       0, current);
+		       current);
 }
arch/sparc/mm/fault_32.c +1 -1
index de2031c2b2d7..fa858626b85b 100644
@@ -83,7 +83,7 @@ static void __do_fault_siginfo(int code, int sig, struct pt_regs *regs,
 		show_signal_msg(regs, sig, code,
 				addr, current);
 
-	force_sig_fault(sig, code, (void __user *) addr, 0);
+	force_sig_fault(sig, code, (void __user *) addr);
 }
 
 static unsigned long compute_si_addr(struct pt_regs *regs, int text_fault)
arch/sparc/mm/fault_64.c +1 -1
index 0a6bcc85fba7..9a9652a15fed 100644
@@ -176,7 +176,7 @@ static void do_fault_siginfo(int code, int sig, struct pt_regs *regs,
 	if (unlikely(show_unhandled_signals))
 		show_signal_msg(regs, sig, code, addr, current);
 
-	force_sig_fault(sig, code, (void __user *) addr, 0);
+	force_sig_fault(sig, code, (void __user *) addr);
 }
 
 static unsigned int get_fault_insn(struct pt_regs *regs, unsigned int insn)
include/linux/sched/signal.h +4 -8
index 3f6a0fcaa10c..7daa425f3055 100644
@@ -298,11 +298,6 @@ static inline void kernel_signal_stop(void)
 
 	schedule();
 }
-#ifdef __ARCH_SI_TRAPNO
-# define ___ARCH_SI_TRAPNO(_a1) , _a1
-#else
-# define ___ARCH_SI_TRAPNO(_a1)
-#endif
 #ifdef __ia64__
 # define ___ARCH_SI_IA64(_a1, _a2, _a3) , _a1, _a2, _a3
 #else
@@ -310,14 +305,11 @@ static inline void kernel_signal_stop(void)
 #endif
 
 int force_sig_fault_to_task(int sig, int code, void __user *addr
-	___ARCH_SI_TRAPNO(int trapno)
 	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
 	, struct task_struct *t);
 int force_sig_fault(int sig, int code, void __user *addr
-	___ARCH_SI_TRAPNO(int trapno)
 	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr));
 int send_sig_fault(int sig, int code, void __user *addr
-	___ARCH_SI_TRAPNO(int trapno)
 	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
 	, struct task_struct *t);
 
@@ -327,6 +319,10 @@ int send_sig_mceerr(int code, void __user *, short, struct task_struct *);
 int force_sig_bnderr(void __user *addr, void __user *lower, void __user *upper);
 int force_sig_pkuerr(void __user *addr, u32 pkey);
 
+int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno);
+int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
+			  struct task_struct *task);
+
 int force_sig_ptrace_errno_trap(int errno, void __user *addr);
 
 extern int send_sig_info(int, struct kernel_siginfo *, struct task_struct *);
kernel/signal.c +31 -10

78 lines not on this page · raw message

Signature · 1 line
2.30.1
author Reviewed-by on v3 7/12
#67 lore raw

Re: [PATCH v3 07/12] signal: Use dedicated helpers to send signals with si_trapno set

Quoted 11 lines · Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Now that si_trapno is no longer expected to be present for every fault reported using siginfo on alpha and sparc remove the trapno parameter from force_sig_fault, force_sig_fault_to_task and send_sig_fault.

Add two new helpers force_sig_fault_trapno and send_sig_fautl_trapno for those signals where trapno is expected to be set.

v1: https://lkml.kernel.org/r/[email protected]

Quoted 620 lines
---
 arch/alpha/kernel/osf_sys.c      |  2 +-
 arch/alpha/kernel/signal.c       |  4 +--
 arch/alpha/kernel/traps.c        | 24 ++++++++---------
 arch/alpha/mm/fault.c            |  4 +--
 arch/sparc/kernel/process_64.c   |  2 +-
 arch/sparc/kernel/sys_sparc_32.c |  2 +-
 arch/sparc/kernel/sys_sparc_64.c |  2 +-
 arch/sparc/kernel/traps_32.c     | 22 ++++++++--------
 arch/sparc/kernel/traps_64.c     | 44 ++++++++++++++------------------
 arch/sparc/kernel/unaligned_32.c |  2 +-
 arch/sparc/mm/fault_32.c         |  2 +-
 arch/sparc/mm/fault_64.c         |  2 +-
 include/linux/sched/signal.h     | 12 +++------
 kernel/signal.c                  | 41 +++++++++++++++++++++--------
 14 files changed, 88 insertions(+), 77 deletions(-)
arch/alpha/kernel/osf_sys.c +1 -1
index d5367a1c6300..80c5d7fbe66a 100644
@@ -878,7 +878,7 @@ SYSCALL_DEFINE5(osf_setsysinfo, unsigned long, op, void __user *, buffer,
 
                        send_sig_fault(SIGFPE, si_code,
                                       (void __user *)NULL,  /* FIXME */
-                                      0, current);
+                                      current);
                }
                return 0;
        }
arch/alpha/kernel/signal.c +2 -2
index 948b89789da8..bc077babafab 100644
@@ -219,7 +219,7 @@ do_sigreturn(struct sigcontext __user *sc)
 
        /* Send SIGTRAP if we're single-stepping: */
        if (ptrace_cancel_bpt (current)) {
-               send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc, 0,
+               send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc,
                               current);
        }
        return;
@@ -247,7 +247,7 @@ do_rt_sigreturn(struct rt_sigframe __user *frame)
 
        /* Send SIGTRAP if we're single-stepping: */
        if (ptrace_cancel_bpt (current)) {
-               send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc, 0,
+               send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc,
                               current);
        }
        return;
arch/alpha/kernel/traps.c +12 -12
index 921d4b6e4d95..0dddf9ecc1f4 100644
@@ -227,7 +227,7 @@ do_entArith(unsigned long summary, unsigned long write_mask,
        }
        die_if_kernel("Arithmetic fault", regs, 0, NULL);
 
-       send_sig_fault(SIGFPE, si_code, (void __user *) regs->pc, 0, current);
+       send_sig_fault_trapno(SIGFPE, si_code, (void __user *) regs->pc, 0, current);
 }
 
 asmlinkage void
@@ -268,12 +268,12 @@ do_entIF(unsigned long type, struct pt_regs *regs)
                        regs->pc -= 4;  /* make pc point to former bpt */
                }
 
-               send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc, 0,
+               send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc,
                               current);
                return;
 
              case 1: /* bugcheck */
-               send_sig_fault(SIGTRAP, TRAP_UNK, (void __user *) regs->pc, 0,
+               send_sig_fault(SIGTRAP, TRAP_UNK, (void __user *) regs->pc,
                               current);
                return;
 
@@ -335,8 +335,8 @@ do_entIF(unsigned long type, struct pt_regs *regs)
                        break;
                }
 
-               send_sig_fault(signo, code, (void __user *) regs->pc, regs->r16,
-                              current);
+               send_sig_fault_trapno(signo, code, (void __user *) regs->pc,
+                                     regs->r16, current);
                return;
 
              case 4: /* opDEC */
@@ -360,9 +360,9 @@ do_entIF(unsigned long type, struct pt_regs *regs)
                        if (si_code == 0)
                                return;
                        if (si_code > 0) {
-                               send_sig_fault(SIGFPE, si_code,
-                                              (void __user *) regs->pc, 0,
-                                              current);
+                               send_sig_fault_trapno(SIGFPE, si_code,
+                                                     (void __user *) regs->pc,
+                                                     0, current);
                                return;
                        }
                }
@@ -387,7 +387,7 @@ do_entIF(unsigned long type, struct pt_regs *regs)
                      ;
        }
 
-       send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, 0, current);
+       send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, current);
 }
 
 /* There is an ifdef in the PALcode in MILO that enables a
@@ -402,7 +402,7 @@ do_entDbg(struct pt_regs *regs)
 {
        die_if_kernel("Instruction fault", regs, 0, NULL);
 
-       force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, 0);
+       force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc);
 }
 
 
@@ -964,12 +964,12 @@ do_entUnaUser(void __user * va, unsigned long opcode,
                        si_code = SEGV_MAPERR;
                mmap_read_unlock(mm);
        }
-       send_sig_fault(SIGSEGV, si_code, va, 0, current);
+       send_sig_fault(SIGSEGV, si_code, va, current);
        return;
 
 give_sigbus:
        regs->pc -= 4;
-       send_sig_fault(SIGBUS, BUS_ADRALN, va, 0, current);
+       send_sig_fault(SIGBUS, BUS_ADRALN, va, current);
        return;
 }
 
arch/alpha/mm/fault.c +2 -2
index 09172f017efc..eee5102c3d88 100644
@@ -219,13 +219,13 @@ do_page_fault(unsigned long address, unsigned long mmcsr,
        mmap_read_unlock(mm);
        /* Send a sigbus, regardless of whether we were in kernel
           or user mode.  */
-       force_sig_fault(SIGBUS, BUS_ADRERR, (void __user *) address, 0);
+       force_sig_fault(SIGBUS, BUS_ADRERR, (void __user *) address);
        if (!user_mode(regs))
                goto no_context;
        return;
 
  do_sigsegv:
-       force_sig_fault(SIGSEGV, si_code, (void __user *) address, 0);
+       force_sig_fault(SIGSEGV, si_code, (void __user *) address);
        return;
 
 #ifdef CONFIG_ALPHA_LARGE_VMALLOC
arch/sparc/kernel/process_64.c +1 -1
index 7afd0a859a78..29e67854d5a4 100644
@@ -518,7 +518,7 @@ void synchronize_user_stack(void)
 
 static void stack_unaligned(unsigned long sp)
 {
-       force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) sp, 0);
+       force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) sp);
 }
 
 static const char uwfault32[] = KERN_INFO \
arch/sparc/kernel/sys_sparc_32.c +1 -1
index be77538bc038..082a551897ed 100644
@@ -151,7 +151,7 @@ sparc_breakpoint (struct pt_regs *regs)
 #ifdef DEBUG_SPARC_BREAKPOINT
         printk ("TRAP: Entering kernel PC=%x, nPC=%x\n", regs->pc, regs->npc);
 #endif
-       force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc, 0);
+       force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc);
 
 #ifdef DEBUG_SPARC_BREAKPOINT
        printk ("TRAP: Returning to space: PC=%x nPC=%x\n", regs->pc, regs->npc);
arch/sparc/kernel/sys_sparc_64.c +1 -1
index 6b92fadb6ec7..1e9a9e016237 100644
@@ -514,7 +514,7 @@ asmlinkage void sparc_breakpoint(struct pt_regs *regs)
 #ifdef DEBUG_SPARC_BREAKPOINT
         printk ("TRAP: Entering kernel PC=%lx, nPC=%lx\n", regs->tpc, regs->tnpc);
 #endif
-       force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->tpc, 0);
+       force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->tpc);
 #ifdef DEBUG_SPARC_BREAKPOINT
        printk ("TRAP: Returning to space: PC=%lx nPC=%lx\n", regs->tpc, regs->tnpc);
 #endif
arch/sparc/kernel/traps_32.c +11 -11
index 247a0d9683b2..5630e5a395e0 100644
@@ -102,8 +102,8 @@ void do_hw_interrupt(struct pt_regs *regs, unsigned long type)
        if(regs->psr & PSR_PS)
                die_if_kernel("Kernel bad trap", regs);
 
-       force_sig_fault(SIGILL, ILL_ILLTRP,
-                       (void __user *)regs->pc, type - 0x80);
+       force_sig_fault_trapno(SIGILL, ILL_ILLTRP,
+                              (void __user *)regs->pc, type - 0x80);
 }
 
 void do_illegal_instruction(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -116,7 +116,7 @@ void do_illegal_instruction(struct pt_regs *regs, unsigned long pc, unsigned lon
               regs->pc, *(unsigned long *)regs->pc);
 #endif
 
-       send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc, 0, current);
+       send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc, current);
 }
 
 void do_priv_instruction(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -124,7 +124,7 @@ void do_priv_instruction(struct pt_regs *regs, unsigned long pc, unsigned long n
 {
        if(psr & PSR_PS)
                die_if_kernel("Penguin instruction from Penguin mode??!?!", regs);
-       send_sig_fault(SIGILL, ILL_PRVOPC, (void __user *)pc, 0, current);
+       send_sig_fault(SIGILL, ILL_PRVOPC, (void __user *)pc, current);
 }
 
 /* XXX User may want to be allowed to do this. XXX */
@@ -145,7 +145,7 @@ void do_memaccess_unaligned(struct pt_regs *regs, unsigned long pc, unsigned lon
 #endif
        send_sig_fault(SIGBUS, BUS_ADRALN,
                       /* FIXME: Should dig out mna address */ (void *)0,
-                      0, current);
+                      current);
 }
 
 static unsigned long init_fsr = 0x0UL;
@@ -291,7 +291,7 @@ void do_fpe_trap(struct pt_regs *regs, unsigned long pc, unsigned long npc,
                else if (fsr & 0x01)
                        code = FPE_FLTRES;
        }
-       send_sig_fault(SIGFPE, code, (void __user *)pc, 0, fpt);
+       send_sig_fault(SIGFPE, code, (void __user *)pc, fpt);
 #ifndef CONFIG_SMP
        last_task_used_math = NULL;
 #endif
@@ -305,7 +305,7 @@ void handle_tag_overflow(struct pt_regs *regs, unsigned long pc, unsigned long n
 {
        if(psr & PSR_PS)
                die_if_kernel("Penguin overflow trap from kernel mode", regs);
-       send_sig_fault(SIGEMT, EMT_TAGOVF, (void __user *)pc, 0, current);
+       send_sig_fault(SIGEMT, EMT_TAGOVF, (void __user *)pc, current);
 }
 
 void handle_watchpoint(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -327,13 +327,13 @@ void handle_reg_access(struct pt_regs *regs, unsigned long pc, unsigned long npc
        printk("Register Access Exception at PC %08lx NPC %08lx PSR %08lx\n",
               pc, npc, psr);
 #endif
-       force_sig_fault(SIGBUS, BUS_OBJERR, (void __user *)pc, 0);
+       force_sig_fault(SIGBUS, BUS_OBJERR, (void __user *)pc);
 }
 
 void handle_cp_disabled(struct pt_regs *regs, unsigned long pc, unsigned long npc,
                        unsigned long psr)
 {
-       send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, 0, current);
+       send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, current);
 }
 
 void handle_cp_exception(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -343,13 +343,13 @@ void handle_cp_exception(struct pt_regs *regs, unsigned long pc, unsigned long n
        printk("Co-Processor Exception at PC %08lx NPC %08lx PSR %08lx\n",
               pc, npc, psr);
 #endif
-       send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, 0, current);
+       send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, current);
 }
 
 void handle_hw_divzero(struct pt_regs *regs, unsigned long pc, unsigned long npc,
                       unsigned long psr)
 {
-       send_sig_fault(SIGFPE, FPE_INTDIV, (void __user *)pc, 0, current);
+       send_sig_fault(SIGFPE, FPE_INTDIV, (void __user *)pc, current);
 }
 
 #ifdef CONFIG_DEBUG_BUGVERBOSE
arch/sparc/kernel/traps_64.c +19 -25
index a850dccd78ea..6863025ed56d 100644
@@ -107,8 +107,8 @@ void bad_trap(struct pt_regs *regs, long lvl)
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGILL, ILL_ILLTRP,
-                       (void __user *)regs->tpc, lvl);
+       force_sig_fault_trapno(SIGILL, ILL_ILLTRP,
+                              (void __user *)regs->tpc, lvl);
 }
 
 void bad_trap_tl1(struct pt_regs *regs, long lvl)
@@ -201,8 +201,7 @@ void spitfire_insn_access_exception(struct pt_regs *regs, unsigned long sfsr, un
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGSEGV, SEGV_MAPERR,
-                       (void __user *)regs->tpc, 0);
+       force_sig_fault(SIGSEGV, SEGV_MAPERR, (void __user *)regs->tpc);
 out:
        exception_exit(prev_state);
 }
@@ -237,7 +236,7 @@ void sun4v_insn_access_exception(struct pt_regs *regs, unsigned long addr, unsig
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGSEGV, SEGV_MAPERR, (void __user *) addr, 0);
+       force_sig_fault(SIGSEGV, SEGV_MAPERR, (void __user *) addr);
 }
 
 void sun4v_insn_access_exception_tl1(struct pt_regs *regs, unsigned long addr, unsigned long type_ctx)
@@ -321,7 +320,7 @@ void spitfire_data_access_exception(struct pt_regs *regs, unsigned long sfsr, un
        if (is_no_fault_exception(regs))
                return;
 
-       force_sig_fault(SIGSEGV, SEGV_MAPERR, (void __user *)sfar, 0);
+       force_sig_fault(SIGSEGV, SEGV_MAPERR, (void __user *)sfar);
 out:
        exception_exit(prev_state);
 }
@@ -385,13 +384,13 @@ void sun4v_data_access_exception(struct pt_regs *regs, unsigned long addr, unsig
         */
        switch (type) {
        case HV_FAULT_TYPE_INV_ASI:
-               force_sig_fault(SIGILL, ILL_ILLADR, (void __user *)addr, 0);
+               force_sig_fault(SIGILL, ILL_ILLADR, (void __user *)addr);
                break;
        case HV_FAULT_TYPE_MCD_DIS:
-               force_sig_fault(SIGSEGV, SEGV_ACCADI, (void __user *)addr, 0);
+               force_sig_fault(SIGSEGV, SEGV_ACCADI, (void __user *)addr);
                break;
        default:
-               force_sig_fault(SIGSEGV, SEGV_MAPERR, (void __user *)addr, 0);
+               force_sig_fault(SIGSEGV, SEGV_MAPERR, (void __user *)addr);
                break;
        }
 }
@@ -568,7 +567,7 @@ static void spitfire_ue_log(unsigned long afsr, unsigned long afar, unsigned lon
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGBUS, BUS_OBJERR, (void *)0, 0);
+       force_sig_fault(SIGBUS, BUS_OBJERR, (void *)0);
 }
 
 void spitfire_access_error(struct pt_regs *regs, unsigned long status_encoded, unsigned long afar)
@@ -2069,8 +2068,7 @@ void do_mcd_err(struct pt_regs *regs, struct sun4v_error_entry ent)
        /* Send SIGSEGV to the userspace process with the right signal
         * code
         */
-       force_sig_fault(SIGSEGV, SEGV_ADIDERR, (void __user *)ent.err_raddr,
-                       0);
+       force_sig_fault(SIGSEGV, SEGV_ADIDERR, (void __user *)ent.err_raddr);
 }
 
 /* We run with %pil set to PIL_NORMAL_MAX and PSTATE_IE enabled in %pstate.
@@ -2184,7 +2182,7 @@ bool sun4v_nonresum_error_user_handled(struct pt_regs *regs,
        }
        if (attrs & SUN4V_ERR_ATTRS_PIO) {
                force_sig_fault(SIGBUS, BUS_ADRERR,
-                               (void __user *)sun4v_get_vaddr(regs), 0);
+                               (void __user *)sun4v_get_vaddr(regs));
                return true;
        }
 
@@ -2340,8 +2338,7 @@ static void do_fpe_common(struct pt_regs *regs)
                        else if (fsr & 0x01)
                                code = FPE_FLTRES;
                }
-               force_sig_fault(SIGFPE, code,
-                               (void __user *)regs->tpc, 0);
+               force_sig_fault(SIGFPE, code, (void __user *)regs->tpc);
        }
 }
 
@@ -2395,8 +2392,7 @@ void do_tof(struct pt_regs *regs)
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGEMT, EMT_TAGOVF,
-                       (void __user *)regs->tpc, 0);
+       force_sig_fault(SIGEMT, EMT_TAGOVF, (void __user *)regs->tpc);
 out:
        exception_exit(prev_state);
 }
@@ -2415,8 +2411,7 @@ void do_div0(struct pt_regs *regs)
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGFPE, FPE_INTDIV,
-                       (void __user *)regs->tpc, 0);
+       force_sig_fault(SIGFPE, FPE_INTDIV, (void __user *)regs->tpc);
 out:
        exception_exit(prev_state);
 }
@@ -2612,7 +2607,7 @@ void do_illegal_instruction(struct pt_regs *regs)
                        }
                }
        }
-       force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc, 0);
+       force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc);
 out:
        exception_exit(prev_state);
 }
@@ -2632,7 +2627,7 @@ void mem_address_unaligned(struct pt_regs *regs, unsigned long sfar, unsigned lo
        if (is_no_fault_exception(regs))
                return;
 
-       force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *)sfar, 0);
+       force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *)sfar);
 out:
        exception_exit(prev_state);
 }
@@ -2650,7 +2645,7 @@ void sun4v_do_mna(struct pt_regs *regs, unsigned long addr, unsigned long type_c
        if (is_no_fault_exception(regs))
                return;
 
-       force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) addr, 0);
+       force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) addr);
 }
 
 /* sun4v_mem_corrupt_detect_precise() - Handle precise exception on an ADI
@@ -2697,7 +2692,7 @@ void sun4v_mem_corrupt_detect_precise(struct pt_regs *regs, unsigned long addr,
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGSEGV, SEGV_ADIPERR, (void __user *)addr, 0);
+       force_sig_fault(SIGSEGV, SEGV_ADIPERR, (void __user *)addr);
 }
 
 void do_privop(struct pt_regs *regs)
@@ -2712,8 +2707,7 @@ void do_privop(struct pt_regs *regs)
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGILL, ILL_PRVOPC,
-                       (void __user *)regs->tpc, 0);
+       force_sig_fault(SIGILL, ILL_PRVOPC, (void __user *)regs->tpc);
 out:
        exception_exit(prev_state);
 }
arch/sparc/kernel/unaligned_32.c +1 -1
index ef5c5207c9ff..455f0258c745 100644
@@ -278,5 +278,5 @@ asmlinkage void user_unaligned_trap(struct pt_regs *regs, unsigned int insn)
 {
        send_sig_fault(SIGBUS, BUS_ADRALN,
                       (void __user *)safe_compute_effective_address(regs, insn),
-                      0, current);
+                      current);
 }
arch/sparc/mm/fault_32.c +1 -1
index de2031c2b2d7..fa858626b85b 100644
@@ -83,7 +83,7 @@ static void __do_fault_siginfo(int code, int sig, struct pt_regs *regs,
                show_signal_msg(regs, sig, code,
                                addr, current);
 
-       force_sig_fault(sig, code, (void __user *) addr, 0);
+       force_sig_fault(sig, code, (void __user *) addr);
 }
 
 static unsigned long compute_si_addr(struct pt_regs *regs, int text_fault)
arch/sparc/mm/fault_64.c +1 -1
index 0a6bcc85fba7..9a9652a15fed 100644
@@ -176,7 +176,7 @@ static void do_fault_siginfo(int code, int sig, struct pt_regs *regs,
        if (unlikely(show_unhandled_signals))
                show_signal_msg(regs, sig, code, addr, current);
 
-       force_sig_fault(sig, code, (void __user *) addr, 0);
+       force_sig_fault(sig, code, (void __user *) addr);
 }
 
 static unsigned int get_fault_insn(struct pt_regs *regs, unsigned int insn)
include/linux/sched/signal.h +4 -8
index 3f6a0fcaa10c..7daa425f3055 100644
@@ -298,11 +298,6 @@ static inline void kernel_signal_stop(void)
 
        schedule();
 }
-#ifdef __ARCH_SI_TRAPNO
-# define ___ARCH_SI_TRAPNO(_a1) , _a1
-#else
-# define ___ARCH_SI_TRAPNO(_a1)
-#endif
 #ifdef __ia64__
 # define ___ARCH_SI_IA64(_a1, _a2, _a3) , _a1, _a2, _a3
 #else
@@ -310,14 +305,11 @@ static inline void kernel_signal_stop(void)
 #endif
 
 int force_sig_fault_to_task(int sig, int code, void __user *addr
-       ___ARCH_SI_TRAPNO(int trapno)
        ___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
        , struct task_struct *t);
 int force_sig_fault(int sig, int code, void __user *addr
-       ___ARCH_SI_TRAPNO(int trapno)
        ___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr));
 int send_sig_fault(int sig, int code, void __user *addr
-       ___ARCH_SI_TRAPNO(int trapno)
        ___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
        , struct task_struct *t);
 
@@ -327,6 +319,10 @@ int send_sig_mceerr(int code, void __user *, short, struct task_struct *);
 int force_sig_bnderr(void __user *addr, void __user *lower, void __user *upper);
 int force_sig_pkuerr(void __user *addr, u32 pkey);
 
+int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno);
+int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
+                         struct task_struct *task);
+
 int force_sig_ptrace_errno_trap(int errno, void __user *addr);
 
 extern int send_sig_info(int, struct kernel_siginfo *, struct task_struct *);
kernel/signal.c +31 -11
index 3d3ba7949788..7eaa8d84db4c 100644
@@ -1651,7 +1651,6 @@ void force_sigsegv(int sig)
 }
 
 int force_sig_fault_to_task(int sig, int code, void __user *addr
-       ___ARCH_SI_TRAPNO(int trapno)
        ___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
        , struct task_struct *t)
 {
@@ -1662,9 +1661,6 @@ int force_sig_fault_to_task(int sig, int code, void __user *addr
        info.si_errno = 0;
        info.si_code  = code;
        info.si_addr  = addr;
-#ifdef __ARCH_SI_TRAPNO
-       info.si_trapno = trapno;
-#endif
 #ifdef __ia64__
        info.si_imm = imm;
        info.si_flags = flags;
@@ -1674,16 +1670,13 @@ int force_sig_fault_to_task(int sig, int code, void __user *addr
 }
 
 int force_sig_fault(int sig, int code, void __user *addr
-       ___ARCH_SI_TRAPNO(int trapno)
        ___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr))
 {
        return force_sig_fault_to_task(sig, code, addr
-                                      ___ARCH_SI_TRAPNO(trapno)
                                       ___ARCH_SI_IA64(imm, flags, isr), current);
 }
 
 int send_sig_fault(int sig, int code, void __user *addr
-       ___ARCH_SI_TRAPNO(int trapno)
        ___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
        , struct task_struct *t)
 {
@@ -1694,9 +1687,6 @@ int send_sig_fault(int sig, int code, void __user *addr
        info.si_errno = 0;
        info.si_code  = code;
        info.si_addr  = addr;
-#ifdef __ARCH_SI_TRAPNO
-       info.si_trapno = trapno;
-#endif
 #ifdef __ia64__
        info.si_imm = imm;
        info.si_flags = flags;
@@ -1763,6 +1753,37 @@ int force_sig_pkuerr(void __user *addr, u32 pkey)
 }
 #endif
 
+#if IS_ENABLED(CONFIG_SPARC)
+int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno)
+{
+       struct kernel_siginfo info;
+
+       clear_siginfo(&info);
+       info.si_signo = sig;
+       info.si_errno = 0;
+       info.si_code  = code;
+       info.si_addr  = addr;
+       info.si_trapno = trapno;
+       return force_sig_info(&info);
+}
+#endif
+
+#if IS_ENABLED(CONFIG_ALPHA)
+int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
+                         struct task_struct *t)
+{
+       struct kernel_siginfo info;
+
+       clear_siginfo(&info);
+       info.si_signo = sig;
+       info.si_errno = 0;
+       info.si_code  = code;
+       info.si_addr  = addr;
+       info.si_trapno = trapno;
+       return send_sig_info(info.si_signo, &info, t);
+}
+#endif
+
 /* For the crazy architectures that include trap information in
  * the errno field, instead of an actual errno value.
  */
--

2.30.1

#68 lore raw

PATCH v3 8/12 signal: Remove __ARCH_SI_TRAPNO

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Now that this define is no longer used remove it from the kernel.

v1: https://lkml.kernel.org/r/[email protected]

---
 arch/alpha/include/uapi/asm/siginfo.h | 2 --
 arch/mips/include/uapi/asm/siginfo.h  | 2 --
 arch/sparc/include/uapi/asm/siginfo.h | 3 ---
 3 files changed, 7 deletions(-)
arch/alpha/include/uapi/asm/siginfo.h +0 -2
index 6e1a2af2f962..e08eae88182b 100644
@@ -2,8 +2,6 @@
 #ifndef _ALPHA_SIGINFO_H
 #define _ALPHA_SIGINFO_H
 
-#define __ARCH_SI_TRAPNO
-
 #include <asm-generic/siginfo.h>
 
 #endif
arch/mips/include/uapi/asm/siginfo.h +0 -2
index c34c7eef0a1c..8cb8bd061a68 100644
@@ -10,9 +10,7 @@
 #ifndef _UAPI_ASM_SIGINFO_H
 #define _UAPI_ASM_SIGINFO_H
 
-
 #define __ARCH_SIGEV_PREAMBLE_SIZE (sizeof(long) + 2*sizeof(int))
-#undef __ARCH_SI_TRAPNO /* exception code needs to fill this ...  */
 
 #define __ARCH_HAS_SWAPPED_SIGINFO
 
arch/sparc/include/uapi/asm/siginfo.h +0 -3
index 68bdde4c2a2e..0e7c27522aed 100644
@@ -8,9 +8,6 @@
 
 #endif /* defined(__sparc__) && defined(__arch64__) */
 
-
-#define __ARCH_SI_TRAPNO
-
 #include <asm-generic/siginfo.h>
 
 
Signature · 1 line
2.30.1
author Reviewed-by on v3 8/12
#69 lore raw

Re: [PATCH v3 08/12] signal: Remove __ARCH_SI_TRAPNO

Quoted 6 lines · Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Now that this define is no longer used remove it from the kernel.

v1: https://lkml.kernel.org/r/[email protected]

Quoted 49 lines
---
 arch/alpha/include/uapi/asm/siginfo.h | 2 --
 arch/mips/include/uapi/asm/siginfo.h  | 2 --
 arch/sparc/include/uapi/asm/siginfo.h | 3 ---
 3 files changed, 7 deletions(-)
arch/alpha/include/uapi/asm/siginfo.h +0 -2
index 6e1a2af2f962..e08eae88182b 100644
@@ -2,8 +2,6 @@
 #ifndef _ALPHA_SIGINFO_H
 #define _ALPHA_SIGINFO_H
 
-#define __ARCH_SI_TRAPNO
-
 #include <asm-generic/siginfo.h>
 
 #endif
arch/mips/include/uapi/asm/siginfo.h +0 -2
index c34c7eef0a1c..8cb8bd061a68 100644
@@ -10,9 +10,7 @@
 #ifndef _UAPI_ASM_SIGINFO_H
 #define _UAPI_ASM_SIGINFO_H
 
-
 #define __ARCH_SIGEV_PREAMBLE_SIZE (sizeof(long) + 2*sizeof(int))
-#undef __ARCH_SI_TRAPNO /* exception code needs to fill this ...  */
 
 #define __ARCH_HAS_SWAPPED_SIGINFO
 
arch/sparc/include/uapi/asm/siginfo.h +0 -4
index 68bdde4c2a2e..0e7c27522aed 100644
@@ -8,9 +8,6 @@
 
 #endif /* defined(__sparc__) && defined(__arch64__) */
 
-
-#define __ARCH_SI_TRAPNO
-
 #include <asm-generic/siginfo.h>
 
 
--

2.30.1

#70 lore raw

PATCH v3 9/12 signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

It helps to know which part of the siginfo structure the siginfo_layout value is talking about.

v1: https://lkml.kernel.org/r/[email protected]

---
 fs/signalfd.c          |  2 +-
 include/linux/signal.h |  2 +-
 kernel/signal.c        | 10 +++++-----
 3 files changed, 7 insertions(+), 7 deletions(-)
fs/signalfd.c +1 -1
index e87e59581653..83130244f653 100644
@@ -132,7 +132,7 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		new.ssi_addr = (long) kinfo->si_addr;
 		new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
 		break;
-	case SIL_PERF_EVENT:
+	case SIL_FAULT_PERF_EVENT:
 		new.ssi_addr = (long) kinfo->si_addr;
 		new.ssi_perf = kinfo->si_perf;
 		break;
include/linux/signal.h +1 -1
index 5160fd45e5ca..ed896d790e46 100644
@@ -44,7 +44,7 @@ enum siginfo_layout {
 	SIL_FAULT_MCEERR,
 	SIL_FAULT_BNDERR,
 	SIL_FAULT_PKUERR,
-	SIL_PERF_EVENT,
+	SIL_FAULT_PERF_EVENT,
 	SIL_CHLD,
 	SIL_RT,
 	SIL_SYS,
kernel/signal.c +5 -5
index 7eaa8d84db4c..697c5fe58db8 100644
@@ -1198,7 +1198,7 @@ static inline bool has_si_pid_and_uid(struct kernel_siginfo *info)
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
-	case SIL_PERF_EVENT:
+	case SIL_FAULT_PERF_EVENT:
 	case SIL_SYS:
 		ret = false;
 		break;
@@ -2553,7 +2553,7 @@ static void hide_si_addr_tag_bits(struct ksignal *ksig)
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
-	case SIL_PERF_EVENT:
+	case SIL_FAULT_PERF_EVENT:
 		ksig->info.si_addr = arch_untagged_si_addr(
 			ksig->info.si_addr, ksig->sig, ksig->info.si_code);
 		break;
@@ -3243,7 +3243,7 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
 				layout = SIL_FAULT_PKUERR;
 #endif
 			else if ((sig == SIGTRAP) && (si_code == TRAP_PERF))
-				layout = SIL_PERF_EVENT;
+				layout = SIL_FAULT_PERF_EVENT;
 		}
 		else if (si_code <= NSIGPOLL)
 			layout = SIL_POLL;
@@ -3365,7 +3365,7 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 		to->si_addr = ptr_to_compat(from->si_addr);
 		to->si_pkey = from->si_pkey;
 		break;
-	case SIL_PERF_EVENT:
+	case SIL_FAULT_PERF_EVENT:
 		to->si_addr = ptr_to_compat(from->si_addr);
 		to->si_perf = from->si_perf;
 		break;
@@ -3441,7 +3441,7 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 		to->si_addr = compat_ptr(from->si_addr);
 		to->si_pkey = from->si_pkey;
 		break;
-	case SIL_PERF_EVENT:
+	case SIL_FAULT_PERF_EVENT:
 		to->si_addr = compat_ptr(from->si_addr);
 		to->si_perf = from->si_perf;
 		break;
Signature · 1 line
2.30.1
author Reviewed-by on v3 9/12
#71 lore raw

Re: [PATCH v3 09/12] signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency

Quoted 8 lines · Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

It helps to know which part of the siginfo structure the siginfo_layout value is talking about.

v1: https://lkml.kernel.org/r/[email protected]

Quoted 88 lines
---
 fs/signalfd.c          |  2 +-
 include/linux/signal.h |  2 +-
 kernel/signal.c        | 10 +++++-----
 3 files changed, 7 insertions(+), 7 deletions(-)
fs/signalfd.c +1 -1
index e87e59581653..83130244f653 100644
@@ -132,7 +132,7 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
                new.ssi_addr = (long) kinfo->si_addr;
                new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
                break;
-       case SIL_PERF_EVENT:
+       case SIL_FAULT_PERF_EVENT:
                new.ssi_addr = (long) kinfo->si_addr;
                new.ssi_perf = kinfo->si_perf;
                break;
include/linux/signal.h +1 -1
index 5160fd45e5ca..ed896d790e46 100644
@@ -44,7 +44,7 @@ enum siginfo_layout {
        SIL_FAULT_MCEERR,
        SIL_FAULT_BNDERR,
        SIL_FAULT_PKUERR,
-       SIL_PERF_EVENT,
+       SIL_FAULT_PERF_EVENT,
        SIL_CHLD,
        SIL_RT,
        SIL_SYS,
kernel/signal.c +5 -6
index 7eaa8d84db4c..697c5fe58db8 100644
@@ -1198,7 +1198,7 @@ static inline bool has_si_pid_and_uid(struct kernel_siginfo *info)
        case SIL_FAULT_MCEERR:
        case SIL_FAULT_BNDERR:
        case SIL_FAULT_PKUERR:
-       case SIL_PERF_EVENT:
+       case SIL_FAULT_PERF_EVENT:
        case SIL_SYS:
                ret = false;
                break;
@@ -2553,7 +2553,7 @@ static void hide_si_addr_tag_bits(struct ksignal *ksig)
        case SIL_FAULT_MCEERR:
        case SIL_FAULT_BNDERR:
        case SIL_FAULT_PKUERR:
-       case SIL_PERF_EVENT:
+       case SIL_FAULT_PERF_EVENT:
                ksig->info.si_addr = arch_untagged_si_addr(
                        ksig->info.si_addr, ksig->sig, ksig->info.si_code);
                break;
@@ -3243,7 +3243,7 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
                                layout = SIL_FAULT_PKUERR;
 #endif
                        else if ((sig == SIGTRAP) && (si_code == TRAP_PERF))
-                               layout = SIL_PERF_EVENT;
+                               layout = SIL_FAULT_PERF_EVENT;
                }
                else if (si_code <= NSIGPOLL)
                        layout = SIL_POLL;
@@ -3365,7 +3365,7 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
                to->si_addr = ptr_to_compat(from->si_addr);
                to->si_pkey = from->si_pkey;
                break;
-       case SIL_PERF_EVENT:
+       case SIL_FAULT_PERF_EVENT:
                to->si_addr = ptr_to_compat(from->si_addr);
                to->si_perf = from->si_perf;
                break;
@@ -3441,7 +3441,7 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
                to->si_addr = compat_ptr(from->si_addr);
                to->si_pkey = from->si_pkey;
                break;
-       case SIL_PERF_EVENT:
+       case SIL_FAULT_PERF_EVENT:
                to->si_addr = compat_ptr(from->si_addr);
                to->si_perf = from->si_perf;
                break;
--

2.30.1

--
You received this message because you are subscribed to the Google Groups "kasan-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/kasan-dev/20210505141101.11519-9-ebiederm%40xmission.com.

#72 lore raw

PATCH v3 10/12 signal: Factor force_sig_perf out of perf_sigtrap

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Separate generating the signal from deciding it needs to be sent.

v1: https://lkml.kernel.org/r/[email protected]

---
 include/linux/sched/signal.h |  1 +
 kernel/events/core.c         | 11 ++---------
 kernel/signal.c              | 13 +++++++++++++
 3 files changed, 16 insertions(+), 9 deletions(-)
include/linux/sched/signal.h +1 -0
index 7daa425f3055..1e2f61a1a512 100644
@@ -318,6 +318,7 @@ int send_sig_mceerr(int code, void __user *, short, struct task_struct *);
 
 int force_sig_bnderr(void __user *addr, void __user *lower, void __user *upper);
 int force_sig_pkuerr(void __user *addr, u32 pkey);
+int force_sig_perf(void __user *addr, u32 type, u64 sig_data);
 
 int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno);
 int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
kernel/events/core.c +2 -9
index 928b166d888e..48ea8863183b 100644
@@ -6394,8 +6394,6 @@ void perf_event_wakeup(struct perf_event *event)
 
 static void perf_sigtrap(struct perf_event *event)
 {
-	struct kernel_siginfo info;
-
 	/*
 	 * We'd expect this to only occur if the irq_work is delayed and either
 	 * ctx->task or current has changed in the meantime. This can be the
@@ -6410,13 +6408,8 @@ static void perf_sigtrap(struct perf_event *event)
 	if (current->flags & PF_EXITING)
 		return;
 
-	clear_siginfo(&info);
-	info.si_signo = SIGTRAP;
-	info.si_code = TRAP_PERF;
-	info.si_errno = event->attr.type;
-	info.si_perf = event->attr.sig_data;
-	info.si_addr = (void __user *)event->pending_addr;
-	force_sig_info(&info);
+	force_sig_perf((void __user *)event->pending_addr,
+		       event->attr.type, event->attr.sig_data);
 }
 
 static void perf_pending_event_disable(struct perf_event *event)
kernel/signal.c +13 -0
index 697c5fe58db8..49560ceac048 100644
@@ -1753,6 +1753,19 @@ int force_sig_pkuerr(void __user *addr, u32 pkey)
 }
 #endif
 
+int force_sig_perf(void __user *addr, u32 type, u64 sig_data)
+{
+	struct kernel_siginfo info;
+
+	clear_siginfo(&info);
+	info.si_signo = SIGTRAP;
+	info.si_errno = type;
+	info.si_code  = TRAP_PERF;
+	info.si_addr  = addr;
+	info.si_perf  = sig_data;
+	return force_sig_info(&info);
+}
+
 #if IS_ENABLED(CONFIG_SPARC)
 int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno)
 {
Signature · 1 line
2.30.1
author Reviewed-by on v3 10/12
#73 lore raw

Re: [PATCH v3 10/12] signal: Factor force_sig_perf out of perf_sigtrap

Quoted 6 lines · Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Separate generating the signal from deciding it needs to be sent.

v1: https://lkml.kernel.org/r/[email protected]

Quoted 74 lines
---
 include/linux/sched/signal.h |  1 +
 kernel/events/core.c         | 11 ++---------
 kernel/signal.c              | 13 +++++++++++++
 3 files changed, 16 insertions(+), 9 deletions(-)
include/linux/sched/signal.h +1 -0
index 7daa425f3055..1e2f61a1a512 100644
@@ -318,6 +318,7 @@ int send_sig_mceerr(int code, void __user *, short, struct task_struct *);
 
 int force_sig_bnderr(void __user *addr, void __user *lower, void __user *upper);
 int force_sig_pkuerr(void __user *addr, u32 pkey);
+int force_sig_perf(void __user *addr, u32 type, u64 sig_data);
 
 int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno);
 int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
kernel/events/core.c +2 -9
index 928b166d888e..48ea8863183b 100644
@@ -6394,8 +6394,6 @@ void perf_event_wakeup(struct perf_event *event)
 
 static void perf_sigtrap(struct perf_event *event)
 {
-       struct kernel_siginfo info;
-
        /*
         * We'd expect this to only occur if the irq_work is delayed and either
         * ctx->task or current has changed in the meantime. This can be the
@@ -6410,13 +6408,8 @@ static void perf_sigtrap(struct perf_event *event)
        if (current->flags & PF_EXITING)
                return;
 
-       clear_siginfo(&info);
-       info.si_signo = SIGTRAP;
-       info.si_code = TRAP_PERF;
-       info.si_errno = event->attr.type;
-       info.si_perf = event->attr.sig_data;
-       info.si_addr = (void __user *)event->pending_addr;
-       force_sig_info(&info);
+       force_sig_perf((void __user *)event->pending_addr,
+                      event->attr.type, event->attr.sig_data);
 }
 
 static void perf_pending_event_disable(struct perf_event *event)
kernel/signal.c +13 -1
index 697c5fe58db8..49560ceac048 100644
@@ -1753,6 +1753,19 @@ int force_sig_pkuerr(void __user *addr, u32 pkey)
 }
 #endif
 
+int force_sig_perf(void __user *addr, u32 type, u64 sig_data)
+{
+       struct kernel_siginfo info;
+
+       clear_siginfo(&info);
+       info.si_signo = SIGTRAP;
+       info.si_errno = type;
+       info.si_code  = TRAP_PERF;
+       info.si_addr  = addr;
+       info.si_perf  = sig_data;
+       return force_sig_info(&info);
+}
+
 #if IS_ENABLED(CONFIG_SPARC)
 int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno)
 {
--

2.30.1

maintainer Acked-by on v3 10/12
#74 lore raw

Re: [PATCH v3 10/12] signal: Factor force_sig_perf out of perf_sigtrap

In reply to Eric W. Beiderman
Quoted 6 lines · Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Separate generating the signal from deciding it needs to be sent.

v1: https://lkml.kernel.org/r/[email protected]

Thanks for cleaning all that up.

#75 lore raw

PATCH v3 11/12 signal: Deliver all of the siginfo perf data in _perf

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Don't abuse si_errno and deliver all of the perf data in _perf member of siginfo_t.

The data field in the perf data structures in a u64 to allow a pointer to be encoded without needed to implement a 32bit and 64bit version of the same structure. There already exists a 32bit and 64bit versions siginfo_t, and the 32bit version can not include a 64bit member as it only has 32bit alignment. So unsigned long is used in siginfo_t instead of a u64 as unsigned long can encode a pointer on all architectures linux supports.

v1: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]

---
 arch/arm/kernel/signal.c                      |  3 ++-
 arch/arm64/kernel/signal.c                    |  3 ++-
 arch/arm64/kernel/signal32.c                  |  3 ++-
 arch/sparc/kernel/signal32.c                  |  3 ++-
 arch/sparc/kernel/signal_64.c                 |  3 ++-
 arch/x86/kernel/signal_compat.c               |  6 ++++--
 fs/signalfd.c                                 |  3 ++-
 include/linux/compat.h                        |  5 ++++-
 include/uapi/asm-generic/siginfo.h            |  8 +++++--
 include/uapi/linux/signalfd.h                 |  4 ++--
 kernel/signal.c                               | 21 ++++++++++++-------
 .../selftests/perf_events/sigtrap_threads.c   | 12 +++++------
 12 files changed, 47 insertions(+), 27 deletions(-)
arch/arm/kernel/signal.c +2 -1
index 643bcb0f091b..f3800c0f428b 100644
@@ -757,7 +757,8 @@ static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x10);
 static_assert(offsetof(siginfo_t, si_lower)	== 0x14);
 static_assert(offsetof(siginfo_t, si_upper)	== 0x18);
 static_assert(offsetof(siginfo_t, si_pkey)	== 0x14);
-static_assert(offsetof(siginfo_t, si_perf)	== 0x10);
+static_assert(offsetof(siginfo_t, si_perf_data)	== 0x10);
+static_assert(offsetof(siginfo_t, si_perf_type)	== 0x14);
 static_assert(offsetof(siginfo_t, si_band)	== 0x0c);
 static_assert(offsetof(siginfo_t, si_fd)	== 0x10);
 static_assert(offsetof(siginfo_t, si_call_addr)	== 0x0c);
arch/arm64/kernel/signal.c +2 -1
index ad4bd27fc044..b3978b468bd4 100644
@@ -1005,7 +1005,8 @@ static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x18);
 static_assert(offsetof(siginfo_t, si_lower)	== 0x20);
 static_assert(offsetof(siginfo_t, si_upper)	== 0x28);
 static_assert(offsetof(siginfo_t, si_pkey)	== 0x20);
-static_assert(offsetof(siginfo_t, si_perf)	== 0x18);
+static_assert(offsetof(siginfo_t, si_perf_data)	== 0x18);
+static_assert(offsetof(siginfo_t, si_perf_type)	== 0x20);
 static_assert(offsetof(siginfo_t, si_band)	== 0x10);
 static_assert(offsetof(siginfo_t, si_fd)	== 0x18);
 static_assert(offsetof(siginfo_t, si_call_addr)	== 0x10);
arch/arm64/kernel/signal32.c +2 -1
index ee6c7484e130..d3be01c46bec 100644
@@ -489,7 +489,8 @@ static_assert(offsetof(compat_siginfo_t, si_addr_lsb)	== 0x10);
 static_assert(offsetof(compat_siginfo_t, si_lower)	== 0x14);
 static_assert(offsetof(compat_siginfo_t, si_upper)	== 0x18);
 static_assert(offsetof(compat_siginfo_t, si_pkey)	== 0x14);
-static_assert(offsetof(compat_siginfo_t, si_perf)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_perf_data)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_perf_type)	== 0x14);
 static_assert(offsetof(compat_siginfo_t, si_band)	== 0x0c);
 static_assert(offsetof(compat_siginfo_t, si_fd)		== 0x10);
 static_assert(offsetof(compat_siginfo_t, si_call_addr)	== 0x0c);
arch/sparc/kernel/signal32.c +2 -1
index 5573722e34ad..4276b9e003ca 100644
@@ -778,6 +778,7 @@ static_assert(offsetof(compat_siginfo_t, si_addr_lsb)	== 0x10);
 static_assert(offsetof(compat_siginfo_t, si_lower)	== 0x14);
 static_assert(offsetof(compat_siginfo_t, si_upper)	== 0x18);
 static_assert(offsetof(compat_siginfo_t, si_pkey)	== 0x14);
-static_assert(offsetof(compat_siginfo_t, si_perf)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_perf_data)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_perf_type)	== 0x14);
 static_assert(offsetof(compat_siginfo_t, si_band)	== 0x0c);
 static_assert(offsetof(compat_siginfo_t, si_fd)		== 0x10);
arch/sparc/kernel/signal_64.c +2 -1
index a69a78984c36..cea23cf95600 100644
@@ -588,6 +588,7 @@ static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x18);
 static_assert(offsetof(siginfo_t, si_lower)	== 0x20);
 static_assert(offsetof(siginfo_t, si_upper)	== 0x28);
 static_assert(offsetof(siginfo_t, si_pkey)	== 0x20);
-static_assert(offsetof(siginfo_t, si_perf)	== 0x18);
+static_assert(offsetof(siginfo_t, si_perf_data)	== 0x18);
+static_assert(offsetof(siginfo_t, si_perf_type)	== 0x20);
 static_assert(offsetof(siginfo_t, si_band)	== 0x10);
 static_assert(offsetof(siginfo_t, si_fd)	== 0x14);
arch/x86/kernel/signal_compat.c +4 -2
index c9601f092a1e..b52407c56000 100644
@@ -147,8 +147,10 @@ static inline void signal_compat_build_tests(void)
 	BUILD_BUG_ON(offsetof(siginfo_t, si_pkey) != 0x20);
 	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_pkey) != 0x14);
 
-	BUILD_BUG_ON(offsetof(siginfo_t, si_perf) != 0x18);
-	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_perf) != 0x10);
+	BUILD_BUG_ON(offsetof(siginfo_t, si_perf_data) != 0x18);
+	BUILD_BUG_ON(offsetof(siginfo_t, si_perf_type) != 0x20);
+	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_perf_data) != 0x10);
+	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_perf_type) != 0x14);
 
 	CHECK_CSI_OFFSET(_sigpoll);
 	CHECK_CSI_SIZE  (_sigpoll, 2*sizeof(int));
fs/signalfd.c +2 -1
index 83130244f653..335ad39f3900 100644
@@ -134,7 +134,8 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		break;
 	case SIL_FAULT_PERF_EVENT:
 		new.ssi_addr = (long) kinfo->si_addr;
-		new.ssi_perf = kinfo->si_perf;
+		new.ssi_perf_type = kinfo->si_perf_type;
+		new.ssi_perf_data = kinfo->si_perf_data;
 		break;
 	case SIL_CHLD:
 		new.ssi_pid    = kinfo->si_pid;
include/linux/compat.h +4 -1
index 6af7bef15e94..a27fffaae121 100644
@@ -236,7 +236,10 @@ typedef struct compat_siginfo {
 					u32 _pkey;
 				} _addr_pkey;
 				/* used when si_code=TRAP_PERF */
-				compat_ulong_t _perf;
+				struct {
+					compat_ulong_t _data;
+					u32 _type;
+				} _perf;
 			};
 		} _sigfault;
 
include/uapi/asm-generic/siginfo.h +6 -2
index 3503282021aa..3ba180f550d7 100644
@@ -96,7 +96,10 @@ union __sifields {
 				__u32 _pkey;
 			} _addr_pkey;
 			/* used when si_code=TRAP_PERF */
-			unsigned long _perf;
+			struct {
+				unsigned long _data;
+				__u32 _type;
+			} _perf;
 		};
 	} _sigfault;
 
@@ -159,7 +162,8 @@ typedef struct siginfo {
 #define si_lower	_sifields._sigfault._addr_bnd._lower
 #define si_upper	_sifields._sigfault._addr_bnd._upper
 #define si_pkey		_sifields._sigfault._addr_pkey._pkey
-#define si_perf		_sifields._sigfault._perf
+#define si_perf_data	_sifields._sigfault._perf._data
+#define si_perf_type	_sifields._sigfault._perf._type
 #define si_band		_sifields._sigpoll._band
 #define si_fd		_sifields._sigpoll._fd
 #define si_call_addr	_sifields._sigsys._call_addr
include/uapi/linux/signalfd.h +2 -2
index 7e333042c7e3..e78dddf433fc 100644
@@ -39,8 +39,8 @@ struct signalfd_siginfo {
 	__s32 ssi_syscall;
 	__u64 ssi_call_addr;
 	__u32 ssi_arch;
-	__u32 __pad3;
-	__u64 ssi_perf;
+	__u32 ssi_perf_type;
+	__u64 ssi_perf_data;
 
 	/*
 	 * Pad strcture to 128 bytes. Remember to update the
kernel/signal.c +13 -8
index 49560ceac048..7fec9d1c5b11 100644
@@ -1758,11 +1758,13 @@ int force_sig_perf(void __user *addr, u32 type, u64 sig_data)
 	struct kernel_siginfo info;
 
 	clear_siginfo(&info);
-	info.si_signo = SIGTRAP;
-	info.si_errno = type;
-	info.si_code  = TRAP_PERF;
-	info.si_addr  = addr;
-	info.si_perf  = sig_data;
+	info.si_signo     = SIGTRAP;
+	info.si_errno     = 0;
+	info.si_code      = TRAP_PERF;
+	info.si_addr      = addr;
+	info.si_perf_data = sig_data;
+	info.si_perf_type = type;
+
 	return force_sig_info(&info);
 }
 
@@ -3380,7 +3382,8 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 		break;
 	case SIL_FAULT_PERF_EVENT:
 		to->si_addr = ptr_to_compat(from->si_addr);
-		to->si_perf = from->si_perf;
+		to->si_perf_data = from->si_perf_data;
+		to->si_perf_type = from->si_perf_type;
 		break;
 	case SIL_CHLD:
 		to->si_pid = from->si_pid;
@@ -3456,7 +3459,8 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 		break;
 	case SIL_FAULT_PERF_EVENT:
 		to->si_addr = compat_ptr(from->si_addr);
-		to->si_perf = from->si_perf;
+		to->si_perf_data = from->si_perf_data;
+		to->si_perf_type = from->si_perf_type;
 		break;
 	case SIL_CHLD:
 		to->si_pid    = from->si_pid;
@@ -4639,7 +4643,8 @@ static inline void siginfo_buildtime_checks(void)
 	CHECK_OFFSET(si_lower);
 	CHECK_OFFSET(si_upper);
 	CHECK_OFFSET(si_pkey);
-	CHECK_OFFSET(si_perf);
+	CHECK_OFFSET(si_perf_data);
+	CHECK_OFFSET(si_perf_type);
 
 	/* sigpoll */
 	CHECK_OFFSET(si_band);
tools/testing/selftests/perf_events/sigtrap_threads.c +6 -6
index 78ddf5e11625..fde123066a8c 100644
@@ -164,8 +164,8 @@ TEST_F(sigtrap_threads, enable_event)
 	EXPECT_EQ(ctx.signal_count, NUM_THREADS);
 	EXPECT_EQ(ctx.tids_want_signal, 0);
 	EXPECT_EQ(ctx.first_siginfo.si_addr, &ctx.iterate_on);
-	EXPECT_EQ(ctx.first_siginfo.si_errno, PERF_TYPE_BREAKPOINT);
-	EXPECT_EQ(ctx.first_siginfo.si_perf, TEST_SIG_DATA(&ctx.iterate_on));
+	EXPECT_EQ(ctx.first_siginfo.si_perf_type, PERF_TYPE_BREAKPOINT);
+	EXPECT_EQ(ctx.first_siginfo.si_perf_data, TEST_SIG_DATA(&ctx.iterate_on));
 
 	/* Check enabled for parent. */
 	ctx.iterate_on = 0;
@@ -183,8 +183,8 @@ TEST_F(sigtrap_threads, modify_and_enable_event)
 	EXPECT_EQ(ctx.signal_count, NUM_THREADS);
 	EXPECT_EQ(ctx.tids_want_signal, 0);
 	EXPECT_EQ(ctx.first_siginfo.si_addr, &ctx.iterate_on);
-	EXPECT_EQ(ctx.first_siginfo.si_errno, PERF_TYPE_BREAKPOINT);
-	EXPECT_EQ(ctx.first_siginfo.si_perf, TEST_SIG_DATA(&ctx.iterate_on));
+	EXPECT_EQ(ctx.first_siginfo.si_perf_type, PERF_TYPE_BREAKPOINT);
+	EXPECT_EQ(ctx.first_siginfo.si_perf_data, TEST_SIG_DATA(&ctx.iterate_on));
 
 	/* Check enabled for parent. */
 	ctx.iterate_on = 0;
@@ -203,8 +203,8 @@ TEST_F(sigtrap_threads, signal_stress)
 	EXPECT_EQ(ctx.signal_count, NUM_THREADS * ctx.iterate_on);
 	EXPECT_EQ(ctx.tids_want_signal, 0);
 	EXPECT_EQ(ctx.first_siginfo.si_addr, &ctx.iterate_on);
-	EXPECT_EQ(ctx.first_siginfo.si_errno, PERF_TYPE_BREAKPOINT);
-	EXPECT_EQ(ctx.first_siginfo.si_perf, TEST_SIG_DATA(&ctx.iterate_on));
+	EXPECT_EQ(ctx.first_siginfo.si_perf_type, PERF_TYPE_BREAKPOINT);
+	EXPECT_EQ(ctx.first_siginfo.si_perf_data, TEST_SIG_DATA(&ctx.iterate_on));
 }
 
 TEST_HARNESS_MAIN
Signature · 1 line
2.30.1
author Reviewed-by on v3 11/12
#76 lore raw

Re: [PATCH v3 11/12] signal: Deliver all of the siginfo perf data in _perf

Quoted 5 lines · Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Don't abuse si_errno and deliver all of the perf data in _perf member of siginfo_t.

Quoted 7 lines

The data field in the perf data structures in a u64 to allow a pointer to be encoded without needed to implement a 32bit and 64bit version of the same structure. There already exists a 32bit and 64bit versions siginfo_t, and the 32bit version can not include a 64bit member as it only has 32bit alignment. So unsigned long is used in siginfo_t instead of a u64 as unsigned long can encode a pointer on all architectures linux supports.

Since there is no functional change (we already had the unsigned long here), the explanation made me think there was a change in that area (but there wasn't). (Can leave as-is, but a "Note: " or similar if you deem it appropriate.)

Quoted 3 lines

v1: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]

Thank you!

Quoted 264 lines
---
 arch/arm/kernel/signal.c                      |  3 ++-
 arch/arm64/kernel/signal.c                    |  3 ++-
 arch/arm64/kernel/signal32.c                  |  3 ++-
 arch/sparc/kernel/signal32.c                  |  3 ++-
 arch/sparc/kernel/signal_64.c                 |  3 ++-
 arch/x86/kernel/signal_compat.c               |  6 ++++--
 fs/signalfd.c                                 |  3 ++-
 include/linux/compat.h                        |  5 ++++-
 include/uapi/asm-generic/siginfo.h            |  8 +++++--
 include/uapi/linux/signalfd.h                 |  4 ++--
 kernel/signal.c                               | 21 ++++++++++++-------
 .../selftests/perf_events/sigtrap_threads.c   | 12 +++++------
 12 files changed, 47 insertions(+), 27 deletions(-)
arch/arm/kernel/signal.c +2 -1
index 643bcb0f091b..f3800c0f428b 100644
@@ -757,7 +757,8 @@ static_assert(offsetof(siginfo_t, si_addr_lsb)      == 0x10);
 static_assert(offsetof(siginfo_t, si_lower)    == 0x14);
 static_assert(offsetof(siginfo_t, si_upper)    == 0x18);
 static_assert(offsetof(siginfo_t, si_pkey)     == 0x14);
-static_assert(offsetof(siginfo_t, si_perf)     == 0x10);
+static_assert(offsetof(siginfo_t, si_perf_data)        == 0x10);
+static_assert(offsetof(siginfo_t, si_perf_type)        == 0x14);
 static_assert(offsetof(siginfo_t, si_band)     == 0x0c);
 static_assert(offsetof(siginfo_t, si_fd)       == 0x10);
 static_assert(offsetof(siginfo_t, si_call_addr)        == 0x0c);
arch/arm64/kernel/signal.c +2 -1
index ad4bd27fc044..b3978b468bd4 100644
@@ -1005,7 +1005,8 @@ static_assert(offsetof(siginfo_t, si_addr_lsb)    == 0x18);
 static_assert(offsetof(siginfo_t, si_lower)    == 0x20);
 static_assert(offsetof(siginfo_t, si_upper)    == 0x28);
 static_assert(offsetof(siginfo_t, si_pkey)     == 0x20);
-static_assert(offsetof(siginfo_t, si_perf)     == 0x18);
+static_assert(offsetof(siginfo_t, si_perf_data)        == 0x18);
+static_assert(offsetof(siginfo_t, si_perf_type)        == 0x20);
 static_assert(offsetof(siginfo_t, si_band)     == 0x10);
 static_assert(offsetof(siginfo_t, si_fd)       == 0x18);
 static_assert(offsetof(siginfo_t, si_call_addr)        == 0x10);
arch/arm64/kernel/signal32.c +2 -1
index ee6c7484e130..d3be01c46bec 100644
@@ -489,7 +489,8 @@ static_assert(offsetof(compat_siginfo_t, si_addr_lsb)       == 0x10);
 static_assert(offsetof(compat_siginfo_t, si_lower)     == 0x14);
 static_assert(offsetof(compat_siginfo_t, si_upper)     == 0x18);
 static_assert(offsetof(compat_siginfo_t, si_pkey)      == 0x14);
-static_assert(offsetof(compat_siginfo_t, si_perf)      == 0x10);
+static_assert(offsetof(compat_siginfo_t, si_perf_data) == 0x10);
+static_assert(offsetof(compat_siginfo_t, si_perf_type) == 0x14);
 static_assert(offsetof(compat_siginfo_t, si_band)      == 0x0c);
 static_assert(offsetof(compat_siginfo_t, si_fd)                == 0x10);
 static_assert(offsetof(compat_siginfo_t, si_call_addr) == 0x0c);
arch/sparc/kernel/signal32.c +2 -1
index 5573722e34ad..4276b9e003ca 100644
@@ -778,6 +778,7 @@ static_assert(offsetof(compat_siginfo_t, si_addr_lsb)       == 0x10);
 static_assert(offsetof(compat_siginfo_t, si_lower)     == 0x14);
 static_assert(offsetof(compat_siginfo_t, si_upper)     == 0x18);
 static_assert(offsetof(compat_siginfo_t, si_pkey)      == 0x14);
-static_assert(offsetof(compat_siginfo_t, si_perf)      == 0x10);
+static_assert(offsetof(compat_siginfo_t, si_perf_data) == 0x10);
+static_assert(offsetof(compat_siginfo_t, si_perf_type) == 0x14);
 static_assert(offsetof(compat_siginfo_t, si_band)      == 0x0c);
 static_assert(offsetof(compat_siginfo_t, si_fd)                == 0x10);
arch/sparc/kernel/signal_64.c +2 -1
index a69a78984c36..cea23cf95600 100644
@@ -588,6 +588,7 @@ static_assert(offsetof(siginfo_t, si_addr_lsb)      == 0x18);
 static_assert(offsetof(siginfo_t, si_lower)    == 0x20);
 static_assert(offsetof(siginfo_t, si_upper)    == 0x28);
 static_assert(offsetof(siginfo_t, si_pkey)     == 0x20);
-static_assert(offsetof(siginfo_t, si_perf)     == 0x18);
+static_assert(offsetof(siginfo_t, si_perf_data)        == 0x18);
+static_assert(offsetof(siginfo_t, si_perf_type)        == 0x20);
 static_assert(offsetof(siginfo_t, si_band)     == 0x10);
 static_assert(offsetof(siginfo_t, si_fd)       == 0x14);
arch/x86/kernel/signal_compat.c +4 -2
index c9601f092a1e..b52407c56000 100644
@@ -147,8 +147,10 @@ static inline void signal_compat_build_tests(void)
        BUILD_BUG_ON(offsetof(siginfo_t, si_pkey) != 0x20);
        BUILD_BUG_ON(offsetof(compat_siginfo_t, si_pkey) != 0x14);
 
-       BUILD_BUG_ON(offsetof(siginfo_t, si_perf) != 0x18);
-       BUILD_BUG_ON(offsetof(compat_siginfo_t, si_perf) != 0x10);
+       BUILD_BUG_ON(offsetof(siginfo_t, si_perf_data) != 0x18);
+       BUILD_BUG_ON(offsetof(siginfo_t, si_perf_type) != 0x20);
+       BUILD_BUG_ON(offsetof(compat_siginfo_t, si_perf_data) != 0x10);
+       BUILD_BUG_ON(offsetof(compat_siginfo_t, si_perf_type) != 0x14);
 
        CHECK_CSI_OFFSET(_sigpoll);
        CHECK_CSI_SIZE  (_sigpoll, 2*sizeof(int));
fs/signalfd.c +2 -1
index 83130244f653..335ad39f3900 100644
@@ -134,7 +134,8 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
                break;
        case SIL_FAULT_PERF_EVENT:
                new.ssi_addr = (long) kinfo->si_addr;
-               new.ssi_perf = kinfo->si_perf;
+               new.ssi_perf_type = kinfo->si_perf_type;
+               new.ssi_perf_data = kinfo->si_perf_data;
                break;
        case SIL_CHLD:
                new.ssi_pid    = kinfo->si_pid;
include/linux/compat.h +4 -1
index 6af7bef15e94..a27fffaae121 100644
@@ -236,7 +236,10 @@ typedef struct compat_siginfo {
                                        u32 _pkey;
                                } _addr_pkey;
                                /* used when si_code=TRAP_PERF */
-                               compat_ulong_t _perf;
+                               struct {
+                                       compat_ulong_t _data;
+                                       u32 _type;
+                               } _perf;
                        };
                } _sigfault;
 
include/uapi/asm-generic/siginfo.h +6 -2
index 3503282021aa..3ba180f550d7 100644
@@ -96,7 +96,10 @@ union __sifields {
                                __u32 _pkey;
                        } _addr_pkey;
                        /* used when si_code=TRAP_PERF */
-                       unsigned long _perf;
+                       struct {
+                               unsigned long _data;
+                               __u32 _type;
+                       } _perf;
                };
        } _sigfault;
 
@@ -159,7 +162,8 @@ typedef struct siginfo {
 #define si_lower       _sifields._sigfault._addr_bnd._lower
 #define si_upper       _sifields._sigfault._addr_bnd._upper
 #define si_pkey                _sifields._sigfault._addr_pkey._pkey
-#define si_perf                _sifields._sigfault._perf
+#define si_perf_data   _sifields._sigfault._perf._data
+#define si_perf_type   _sifields._sigfault._perf._type
 #define si_band                _sifields._sigpoll._band
 #define si_fd          _sifields._sigpoll._fd
 #define si_call_addr   _sifields._sigsys._call_addr
include/uapi/linux/signalfd.h +2 -2
index 7e333042c7e3..e78dddf433fc 100644
@@ -39,8 +39,8 @@ struct signalfd_siginfo {
        __s32 ssi_syscall;
        __u64 ssi_call_addr;
        __u32 ssi_arch;
-       __u32 __pad3;
-       __u64 ssi_perf;
+       __u32 ssi_perf_type;
+       __u64 ssi_perf_data;
 
        /*
         * Pad strcture to 128 bytes. Remember to update the
kernel/signal.c +13 -8
index 49560ceac048..7fec9d1c5b11 100644
@@ -1758,11 +1758,13 @@ int force_sig_perf(void __user *addr, u32 type, u64 sig_data)
        struct kernel_siginfo info;
 
        clear_siginfo(&info);
-       info.si_signo = SIGTRAP;
-       info.si_errno = type;
-       info.si_code  = TRAP_PERF;
-       info.si_addr  = addr;
-       info.si_perf  = sig_data;
+       info.si_signo     = SIGTRAP;
+       info.si_errno     = 0;
+       info.si_code      = TRAP_PERF;
+       info.si_addr      = addr;
+       info.si_perf_data = sig_data;
+       info.si_perf_type = type;
+
        return force_sig_info(&info);
 }
 
@@ -3380,7 +3382,8 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
                break;
        case SIL_FAULT_PERF_EVENT:
                to->si_addr = ptr_to_compat(from->si_addr);
-               to->si_perf = from->si_perf;
+               to->si_perf_data = from->si_perf_data;
+               to->si_perf_type = from->si_perf_type;
                break;
        case SIL_CHLD:
                to->si_pid = from->si_pid;
@@ -3456,7 +3459,8 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
                break;
        case SIL_FAULT_PERF_EVENT:
                to->si_addr = compat_ptr(from->si_addr);
-               to->si_perf = from->si_perf;
+               to->si_perf_data = from->si_perf_data;
+               to->si_perf_type = from->si_perf_type;
                break;
        case SIL_CHLD:
                to->si_pid    = from->si_pid;
@@ -4639,7 +4643,8 @@ static inline void siginfo_buildtime_checks(void)
        CHECK_OFFSET(si_lower);
        CHECK_OFFSET(si_upper);
        CHECK_OFFSET(si_pkey);
-       CHECK_OFFSET(si_perf);
+       CHECK_OFFSET(si_perf_data);
+       CHECK_OFFSET(si_perf_type);
 
        /* sigpoll */
        CHECK_OFFSET(si_band);
tools/testing/selftests/perf_events/sigtrap_threads.c +6 -7
index 78ddf5e11625..fde123066a8c 100644
@@ -164,8 +164,8 @@ TEST_F(sigtrap_threads, enable_event)
        EXPECT_EQ(ctx.signal_count, NUM_THREADS);
        EXPECT_EQ(ctx.tids_want_signal, 0);
        EXPECT_EQ(ctx.first_siginfo.si_addr, &ctx.iterate_on);
-       EXPECT_EQ(ctx.first_siginfo.si_errno, PERF_TYPE_BREAKPOINT);
-       EXPECT_EQ(ctx.first_siginfo.si_perf, TEST_SIG_DATA(&ctx.iterate_on));
+       EXPECT_EQ(ctx.first_siginfo.si_perf_type, PERF_TYPE_BREAKPOINT);
+       EXPECT_EQ(ctx.first_siginfo.si_perf_data, TEST_SIG_DATA(&ctx.iterate_on));
 
        /* Check enabled for parent. */
        ctx.iterate_on = 0;
@@ -183,8 +183,8 @@ TEST_F(sigtrap_threads, modify_and_enable_event)
        EXPECT_EQ(ctx.signal_count, NUM_THREADS);
        EXPECT_EQ(ctx.tids_want_signal, 0);
        EXPECT_EQ(ctx.first_siginfo.si_addr, &ctx.iterate_on);
-       EXPECT_EQ(ctx.first_siginfo.si_errno, PERF_TYPE_BREAKPOINT);
-       EXPECT_EQ(ctx.first_siginfo.si_perf, TEST_SIG_DATA(&ctx.iterate_on));
+       EXPECT_EQ(ctx.first_siginfo.si_perf_type, PERF_TYPE_BREAKPOINT);
+       EXPECT_EQ(ctx.first_siginfo.si_perf_data, TEST_SIG_DATA(&ctx.iterate_on));
 
        /* Check enabled for parent. */
        ctx.iterate_on = 0;
@@ -203,8 +203,8 @@ TEST_F(sigtrap_threads, signal_stress)
        EXPECT_EQ(ctx.signal_count, NUM_THREADS * ctx.iterate_on);
        EXPECT_EQ(ctx.tids_want_signal, 0);
        EXPECT_EQ(ctx.first_siginfo.si_addr, &ctx.iterate_on);
-       EXPECT_EQ(ctx.first_siginfo.si_errno, PERF_TYPE_BREAKPOINT);
-       EXPECT_EQ(ctx.first_siginfo.si_perf, TEST_SIG_DATA(&ctx.iterate_on));
+       EXPECT_EQ(ctx.first_siginfo.si_perf_type, PERF_TYPE_BREAKPOINT);
+       EXPECT_EQ(ctx.first_siginfo.si_perf_data, TEST_SIG_DATA(&ctx.iterate_on));
 }
 
 TEST_HARNESS_MAIN
--

2.30.1

#77 lore raw

PATCH v3 12/12 signalfd: Remove SIL_FAULT_PERF_EVENT fields from signalfd_siginfo

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

With the addition of ssi_perf_data and ssi_perf_type struct signalfd_siginfo is dangerously close to running out of space. All that remains is just enough space for two additional 64bit fields. A practice of adding all possible siginfo_t fields into struct singalfd_siginfo can not be supported as adding the missing fields ssi_lower, ssi_upper, and ssi_pkey would require two 64bit fields and one 32bit fields. In practice the fields ssi_perf_data and ssi_perf_type can never be used by signalfd as the signal that generates them always delivers them synchronously to the thread that triggers them.

Therefore until someone actually needs the fields ssi_perf_data and ssi_perf_type in signalfd_siginfo remove them. This leaves a bit more room for future expansion.

v1: https://lkml.kernel.org/r/[email protected]

---
 fs/signalfd.c                 | 16 ++++++----------
 include/uapi/linux/signalfd.h |  4 +---
 2 files changed, 7 insertions(+), 13 deletions(-)
fs/signalfd.c +6 -10
index 335ad39f3900..040e1cf90528 100644
@@ -114,12 +114,13 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		break;
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
+	case SIL_FAULT_PERF_EVENT:
 		/*
-		 * Fall through to the SIL_FAULT case.  Both SIL_FAULT_BNDERR
-		 * and SIL_FAULT_PKUERR are only generated by faults that
-		 * deliver them synchronously to userspace.  In case someone
-		 * injects one of these signals and signalfd catches it treat
-		 * it as SIL_FAULT.
+		 * Fall through to the SIL_FAULT case.  SIL_FAULT_BNDERR,
+		 * SIL_FAULT_PKUERR, and SIL_FAULT_PERF_EVENT are only
+		 * generated by faults that deliver them synchronously to
+		 * userspace.  In case someone injects one of these signals
+		 * and signalfd catches it treat it as SIL_FAULT.
 		 */
 	case SIL_FAULT:
 		new.ssi_addr = (long) kinfo->si_addr;
@@ -132,11 +133,6 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		new.ssi_addr = (long) kinfo->si_addr;
 		new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
 		break;
-	case SIL_FAULT_PERF_EVENT:
-		new.ssi_addr = (long) kinfo->si_addr;
-		new.ssi_perf_type = kinfo->si_perf_type;
-		new.ssi_perf_data = kinfo->si_perf_data;
-		break;
 	case SIL_CHLD:
 		new.ssi_pid    = kinfo->si_pid;
 		new.ssi_uid    = kinfo->si_uid;
include/uapi/linux/signalfd.h +1 -3
index e78dddf433fc..83429a05b698 100644
@@ -39,8 +39,6 @@ struct signalfd_siginfo {
 	__s32 ssi_syscall;
 	__u64 ssi_call_addr;
 	__u32 ssi_arch;
-	__u32 ssi_perf_type;
-	__u64 ssi_perf_data;
 
 	/*
 	 * Pad strcture to 128 bytes. Remember to update the
@@ -51,7 +49,7 @@ struct signalfd_siginfo {
 	 * comes out of a read(2) and we really don't want to have
 	 * a compat on read(2).
 	 */
-	__u8 __pad[16];
+	__u8 __pad[28];
 };
 
 
Signature · 1 line
2.30.1
author Reviewed-by on v3 12/12
#78 lore raw

Re: [PATCH v3 12/12] signalfd: Remove SIL_FAULT_PERF_EVENT fields from signalfd_siginfo

Quoted 12 lines · Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

With the addition of ssi_perf_data and ssi_perf_type struct signalfd_siginfo is dangerously close to running out of space. All that remains is just enough space for two additional 64bit fields. A practice of adding all possible siginfo_t fields into struct singalfd_siginfo can not be supported as adding the missing fields ssi_lower, ssi_upper, and ssi_pkey would require two 64bit fields and one 32bit fields. In practice the fields ssi_perf_data and ssi_perf_type can never be used by signalfd as the signal that generates them always delivers them synchronously to the thread that triggers them.

Quoted 6 lines

Therefore until someone actually needs the fields ssi_perf_data and ssi_perf_type in signalfd_siginfo remove them. This leaves a bit more room for future expansion.

v1: https://lkml.kernel.org/r/[email protected]

Quoted 64 lines
---
 fs/signalfd.c                 | 16 ++++++----------
 include/uapi/linux/signalfd.h |  4 +---
 2 files changed, 7 insertions(+), 13 deletions(-)
fs/signalfd.c +6 -10
index 335ad39f3900..040e1cf90528 100644
@@ -114,12 +114,13 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
                break;
        case SIL_FAULT_BNDERR:
        case SIL_FAULT_PKUERR:
+       case SIL_FAULT_PERF_EVENT:
                /*
-                * Fall through to the SIL_FAULT case.  Both SIL_FAULT_BNDERR
-                * and SIL_FAULT_PKUERR are only generated by faults that
-                * deliver them synchronously to userspace.  In case someone
-                * injects one of these signals and signalfd catches it treat
-                * it as SIL_FAULT.
+                * Fall through to the SIL_FAULT case.  SIL_FAULT_BNDERR,
+                * SIL_FAULT_PKUERR, and SIL_FAULT_PERF_EVENT are only
+                * generated by faults that deliver them synchronously to
+                * userspace.  In case someone injects one of these signals
+                * and signalfd catches it treat it as SIL_FAULT.
                 */
        case SIL_FAULT:
                new.ssi_addr = (long) kinfo->si_addr;
@@ -132,11 +133,6 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
                new.ssi_addr = (long) kinfo->si_addr;
                new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
                break;
-       case SIL_FAULT_PERF_EVENT:
-               new.ssi_addr = (long) kinfo->si_addr;
-               new.ssi_perf_type = kinfo->si_perf_type;
-               new.ssi_perf_data = kinfo->si_perf_data;
-               break;
        case SIL_CHLD:
                new.ssi_pid    = kinfo->si_pid;
                new.ssi_uid    = kinfo->si_uid;
include/uapi/linux/signalfd.h +1 -4
index e78dddf433fc..83429a05b698 100644
@@ -39,8 +39,6 @@ struct signalfd_siginfo {
        __s32 ssi_syscall;
        __u64 ssi_call_addr;
        __u32 ssi_arch;
-       __u32 ssi_perf_type;
-       __u64 ssi_perf_data;
 
        /*
         * Pad strcture to 128 bytes. Remember to update the
@@ -51,7 +49,7 @@ struct signalfd_siginfo {
         * comes out of a read(2) and we really don't want to have
         * a compat on read(2).
         */
-       __u8 __pad[16];
+       __u8 __pad[28];
 };
 
 
--

2.30.1

#79 lore raw

Re: [PATCH v3 00/12] signal: sort out si_trapno and si_perf

In reply to Eric W. Biederman
Quoted 67 lines · Eric W. Biederman

This set of changes sorts out the ABI issues with SIGTRAP TRAP_PERF, and hopefully will can get merged before any userspace code starts using the new ABI.

The big ideas are:
- Placing the asserts first to prevent unexpected ABI changes
- si_trapno becomming ordinary fault subfield.
- struct signalfd_siginfo is almost full

This set of changes starts out with Marco's static_assert changes and additional one of my own that enforces the fact that the alignment of siginfo_t is also part of the ABI. Together these build time checks verify there are no unexpected ABI changes in the changes that follow.

The field si_trapno is changed to become an ordinary extension of the _sigfault member of siginfo.

The code is refactored a bit and then si_perf_type is added along side si_perf_data in the _perf subfield of _sigfault of siginfo_t.

Finally the signalfd_siginfo fields are removed as they appear to be filling up the structure without userspace actually being able to use them.

v2: https://lkml.kernel.org/r/[email protected]
v1: https://lkml.kernel.org/r/[email protected]

Eric W. Biederman (9):
signal: Verify the alignment and size of siginfo_t
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Use dedicated helpers to send signals with si_trapno set
signal: Remove __ARCH_SI_TRAPNO
signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency
signal: Factor force_sig_perf out of perf_sigtrap
signal: Deliver all of the siginfo perf data in _perf
signalfd: Remove SIL_FAULT_PERF_EVENT fields from signalfd_siginfo

Marco Elver (3):
sparc64: Add compile-time asserts for siginfo_t offsets
arm: Add compile-time asserts for siginfo_t offsets
arm64: Add compile-time asserts for siginfo_t offsets

 arch/alpha/include/uapi/asm/siginfo.h              |   2 -
 arch/alpha/kernel/osf_sys.c                        |   2 +-
 arch/alpha/kernel/signal.c                         |   4 +-
 arch/alpha/kernel/traps.c                          |  24 ++---
 arch/alpha/mm/fault.c                              |   4 +-
 arch/arm/kernel/signal.c                           |  39 +++++++
 arch/arm64/kernel/signal.c                         |  39 +++++++
 arch/arm64/kernel/signal32.c                       |  39 +++++++
 arch/mips/include/uapi/asm/siginfo.h               |   2 -
 arch/sparc/include/uapi/asm/siginfo.h              |   3 -
 arch/sparc/kernel/process_64.c                     |   2 +-
 arch/sparc/kernel/signal32.c                       |  37 +++++++
 arch/sparc/kernel/signal_64.c                      |  36 +++++++
 arch/sparc/kernel/sys_sparc_32.c                   |   2 +-
 arch/sparc/kernel/sys_sparc_64.c                   |   2 +-
 arch/sparc/kernel/traps_32.c                       |  22 ++--
 arch/sparc/kernel/traps_64.c                       |  44 ++++----
 arch/sparc/kernel/unaligned_32.c                   |   2 +-
 arch/sparc/mm/fault_32.c                           |   2 +-
 arch/sparc/mm/fault_64.c                           |   2 +-
 arch/x86/kernel/signal_compat.c                    |  15 ++-
 fs/signalfd.c                                      |  23 ++---
 include/linux/compat.h                             |  10 +-
Quoted 8 lines
 include/linux/sched/signal.h                       |  13 +--
 include/linux/signal.h                             |   3 +-
 include/uapi/asm-generic/siginfo.h                 |  20 ++--
 include/uapi/linux/signalfd.h                      |   4 +-
 kernel/events/core.c                               |  11 +-
 kernel/signal.c                                    | 113 +++++++++++++--------
 .../selftests/perf_events/sigtrap_threads.c        |  12 +--
 30 files changed, 373 insertions(+), 160 deletions(-)

Looks good, thanks a lot! I ran selftests/perf_events on x86-64, and build-tested x86-32, arm, arm64, sparc, alpha.

I added my Reviewed/Acked-by to the various patches.

Thanks,
-- Marco

#80 lore raw

Re: [PATCH v3 00/12] signal: sort out si_trapno and si_perf

In reply to Eric W. Biederman

Hi Eric,

Quoted 18 lines · Eric W. Biederman

This set of changes sorts out the ABI issues with SIGTRAP TRAP_PERF, and hopefully will can get merged before any userspace code starts using the new ABI.

The big ideas are:
- Placing the asserts first to prevent unexpected ABI changes
- si_trapno becomming ordinary fault subfield.
- struct signalfd_siginfo is almost full

This set of changes starts out with Marco's static_assert changes and additional one of my own that enforces the fact that the alignment of siginfo_t is also part of the ABI. Together these build time checks verify there are no unexpected ABI changes in the changes that follow.

The field si_trapno is changed to become an ordinary extension of the _sigfault member of siginfo.

Quoted 6 lines

The code is refactored a bit and then si_perf_type is added along side si_perf_data in the _perf subfield of _sigfault of siginfo_t.

Finally the signalfd_siginfo fields are removed as they appear to be filling up the structure without userspace actually being able to use them.

Thanks for your series, which is now in next-20210506.

Quoted 13 lines
 arch/alpha/include/uapi/asm/siginfo.h              |   2 -
 arch/alpha/kernel/osf_sys.c                        |   2 +-
 arch/alpha/kernel/signal.c                         |   4 +-
 arch/alpha/kernel/traps.c                          |  24 ++---
 arch/alpha/mm/fault.c                              |   4 +-
 arch/arm/kernel/signal.c                           |  39 +++++++
 arch/arm64/kernel/signal.c                         |  39 +++++++
 arch/arm64/kernel/signal32.c                       |  39 +++++++
 arch/mips/include/uapi/asm/siginfo.h               |   2 -
 arch/sparc/include/uapi/asm/siginfo.h              |   3 -
 arch/sparc/kernel/process_64.c                     |   2 +-
 arch/sparc/kernel/signal32.c                       |  37 +++++++
 arch/sparc/kernel/signal_64.c                      |  36 +++++++
Quoted 8 lines
 arch/sparc/kernel/sys_sparc_32.c                   |   2 +-
 arch/sparc/kernel/sys_sparc_64.c                   |   2 +-
 arch/sparc/kernel/traps_32.c                       |  22 ++--
 arch/sparc/kernel/traps_64.c                       |  44 ++++----
 arch/sparc/kernel/unaligned_32.c                   |   2 +-
 arch/sparc/mm/fault_32.c                           |   2 +-
 arch/sparc/mm/fault_64.c                           |   2 +-
 arch/x86/kernel/signal_compat.c                    |  15 ++-

No changes needed for other architectures?
All m68k configs are broken with

arch/m68k/kernel/signal.c:626:35: error: 'siginfo_t' {aka 'struct siginfo'} has no member named 'si_perf'; did you mean 'si_errno'?

See e.g. http://kisskb.ellerman.id.au/kisskb/buildresult/14537820/

There are still a few more references left to si_perf:

$ git grep -n -w si_perf
Next/merge.log:2902:Merging userns/for-next (4cf4e48fff05 signal: sort
out si_trapno and si_perf)
arch/m68k/kernel/signal.c:626:  BUILD_BUG_ON(offsetof(siginfo_t,
si_perf) != 0x10);
include/uapi/linux/perf_event.h:467:     * siginfo_t::si_perf, e.g. to
permit user to identify the event.
tools/testing/selftests/perf_events/sigtrap_threads.c:46:/* Unique
value to check si_perf is correctly set from
perf_event_attr::sig_data. */

Thanks!

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#81 lore raw

Re: [PATCH v3 00/12] signal: sort out si_trapno and si_perf

On Thu, May 06, 2021 at 09:00AM +0200, Geert Uytterhoeven wrote:
[...]

Quoted 10 lines

No changes needed for other architectures?
All m68k configs are broken with

arch/m68k/kernel/signal.c:626:35: error: 'siginfo_t' {aka 'struct siginfo'} has no member named 'si_perf'; did you mean 'si_errno'?

See e.g. http://kisskb.ellerman.id.au/kisskb/buildresult/14537820/

There are still a few more references left to si_perf:

Quoted 10 lines
$ git grep -n -w si_perf
Next/merge.log:2902:Merging userns/for-next (4cf4e48fff05 signal: sort
out si_trapno and si_perf)
arch/m68k/kernel/signal.c:626:  BUILD_BUG_ON(offsetof(siginfo_t,
si_perf) != 0x10);
include/uapi/linux/perf_event.h:467:     * siginfo_t::si_perf, e.g. to
permit user to identify the event.
tools/testing/selftests/perf_events/sigtrap_threads.c:46:/* Unique
value to check si_perf is correctly set from
perf_event_attr::sig_data. */

I think we're missing the below in "signal: Deliver all of the siginfo perf data in _perf".

Thanks,
-- Marco

------ >8 ------

arch/m68k/kernel/signal.c +2 -1
index a4b7ee1df211..8f215e79e70e 100644
@@ -623,7 +623,8 @@ static inline void siginfo_build_tests(void)
 	BUILD_BUG_ON(offsetof(siginfo_t, si_pkey) != 0x12);
 
 	/* _sigfault._perf */
-	BUILD_BUG_ON(offsetof(siginfo_t, si_perf) != 0x10);
+	BUILD_BUG_ON(offsetof(siginfo_t, si_perf_data) != 0x10);
+	BUILD_BUG_ON(offsetof(siginfo_t, si_perf_type) != 0x14);
 
 	/* _sigpoll */
 	BUILD_BUG_ON(offsetof(siginfo_t, si_band)   != 0x0c);
include/uapi/linux/perf_event.h +1 -1
index bf8143505c49..f92880a15645 100644
@@ -464,7 +464,7 @@ struct perf_event_attr {
 
 	/*
 	 * User provided data if sigtrap=1, passed back to user via
-	 * siginfo_t::si_perf, e.g. to permit user to identify the event.
+	 * siginfo_t::si_perf_data, e.g. to permit user to identify the event.
 	 */
 	__u64	sig_data;
 };
tools/testing/selftests/perf_events/sigtrap_threads.c +1 -1
index fde123066a8c..8e83cf91513a 100644
@@ -43,7 +43,7 @@ static struct {
 	siginfo_t first_siginfo;	/* First observed siginfo_t. */
 } ctx;
 
-/* Unique value to check si_perf is correctly set from perf_event_attr::sig_data. */
+/* Unique value to check si_perf_data is correctly set from perf_event_attr::sig_data. */
 #define TEST_SIG_DATA(addr) (~(unsigned long)(addr))
 
 static struct perf_event_attr make_event_attr(bool enabled, volatile void *addr)
#82 lore raw

Re: [PATCH v3 00/12] signal: sort out si_trapno and si_perf

For the moment I am adding this to my for-next branch. I plan to respin and fold this in but I am not certain what my schedule looks like today. So I figure making certain I have a fix out (so I stop breaking m68k) is more important than having a perfect patch.

Eric

From: "Eric W. Biederman" <[email protected]>
Date: Thu, 6 May 2021 10:17:10 -0500
Subject: [PATCH] signal: Remove the last few si_perf references

I accidentially overlooked a few references to si_perf when sorting out the ABI update those references now.

Fixes f6a2c711f1e3 ("signal: Deliver all of the siginfo perf data in _perf")
---
 arch/m68k/kernel/signal.c                             | 3 ++-
 include/uapi/linux/perf_event.h                       | 2 +-
 tools/testing/selftests/perf_events/sigtrap_threads.c | 2 +-
 3 files changed, 4 insertions(+), 3 deletions(-)
arch/m68k/kernel/signal.c +2 -1
index a4b7ee1df211..8f215e79e70e 100644
@@ -623,7 +623,8 @@ static inline void siginfo_build_tests(void)
 	BUILD_BUG_ON(offsetof(siginfo_t, si_pkey) != 0x12);
 
 	/* _sigfault._perf */
-	BUILD_BUG_ON(offsetof(siginfo_t, si_perf) != 0x10);
+	BUILD_BUG_ON(offsetof(siginfo_t, si_perf_data) != 0x10);
+	BUILD_BUG_ON(offsetof(siginfo_t, si_perf_type) != 0x14);
 
 	/* _sigpoll */
 	BUILD_BUG_ON(offsetof(siginfo_t, si_band)   != 0x0c);
include/uapi/linux/perf_event.h +1 -1
index e54e639248c8..7b14753b3d38 100644
@@ -464,7 +464,7 @@ struct perf_event_attr {
 
 	/*
 	 * User provided data if sigtrap=1, passed back to user via
-	 * siginfo_t::si_perf, e.g. to permit user to identify the event.
+	 * siginfo_t::si_perf_data, e.g. to permit user to identify the event.
 	 */
 	__u64	sig_data;
 };
tools/testing/selftests/perf_events/sigtrap_threads.c +1 -1
index fde123066a8c..8e83cf91513a 100644
@@ -43,7 +43,7 @@ static struct {
 	siginfo_t first_siginfo;	/* First observed siginfo_t. */
 } ctx;
 
-/* Unique value to check si_perf is correctly set from perf_event_attr::sig_data. */
+/* Unique value to check si_perf_data is correctly set from perf_event_attr::sig_data. */
 #define TEST_SIG_DATA(addr) (~(unsigned long)(addr))
 
 static struct perf_event_attr make_event_attr(bool enabled, volatile void *addr)
Signature · 1 line
2.30.1
#83 lore raw

Re: [PATCH v3 00/12] signal: sort out si_trapno and si_perf

In reply to Geert Uytterhoeven
Quoted 52 lines · Geert Uytterhoeven

Hi Eric,

Quoted 18 lines · Eric W. Biederman

This set of changes sorts out the ABI issues with SIGTRAP TRAP_PERF, and hopefully will can get merged before any userspace code starts using the new ABI.

The big ideas are:
- Placing the asserts first to prevent unexpected ABI changes
- si_trapno becomming ordinary fault subfield.
- struct signalfd_siginfo is almost full

This set of changes starts out with Marco's static_assert changes and additional one of my own that enforces the fact that the alignment of siginfo_t is also part of the ABI. Together these build time checks verify there are no unexpected ABI changes in the changes that follow.

The field si_trapno is changed to become an ordinary extension of the _sigfault member of siginfo.

Quoted 6 lines

The code is refactored a bit and then si_perf_type is added along side si_perf_data in the _perf subfield of _sigfault of siginfo_t.

Finally the signalfd_siginfo fields are removed as they appear to be filling up the structure without userspace actually being able to use them.

Thanks for your series, which is now in next-20210506.

Quoted 13 lines
 arch/alpha/include/uapi/asm/siginfo.h              |   2 -
 arch/alpha/kernel/osf_sys.c                        |   2 +-
 arch/alpha/kernel/signal.c                         |   4 +-
 arch/alpha/kernel/traps.c                          |  24 ++---
 arch/alpha/mm/fault.c                              |   4 +-
 arch/arm/kernel/signal.c                           |  39 +++++++
 arch/arm64/kernel/signal.c                         |  39 +++++++
 arch/arm64/kernel/signal32.c                       |  39 +++++++
 arch/mips/include/uapi/asm/siginfo.h               |   2 -
 arch/sparc/include/uapi/asm/siginfo.h              |   3 -
 arch/sparc/kernel/process_64.c                     |   2 +-
 arch/sparc/kernel/signal32.c                       |  37 +++++++
 arch/sparc/kernel/signal_64.c                      |  36 +++++++
Quoted 8 lines
 arch/sparc/kernel/sys_sparc_32.c                   |   2 +-
 arch/sparc/kernel/sys_sparc_64.c                   |   2 +-
 arch/sparc/kernel/traps_32.c                       |  22 ++--
 arch/sparc/kernel/traps_64.c                       |  44 ++++----
 arch/sparc/kernel/unaligned_32.c                   |   2 +-
 arch/sparc/mm/fault_32.c                           |   2 +-
 arch/sparc/mm/fault_64.c                           |   2 +-
 arch/x86/kernel/signal_compat.c                    |  15 ++-
Quoted 2 lines

No changes needed for other architectures?
All m68k configs are broken with

Thanks. I hadn't realized that si_perf asserts existed on m68k. Thankfully linux-next caught this these.

Looking a little more deeply, it is strange that this is tested on m68k. The architecture does not implement HAVE_PERF_EVENTS so it is impossible for this signal to be generated.

On the off chance this these new signals will appear on m68k someday I will update the assertion.

Quoted 7 lines

arch/m68k/kernel/signal.c:626:35: error: 'siginfo_t' {aka 'struct siginfo'} has no member named 'si_perf'; did you mean 'si_errno'?

See e.g. http://kisskb.ellerman.id.au/kisskb/buildresult/14537820/

There are still a few more references left to si_perf:

Quoted 10 lines
$ git grep -n -w si_perf
Next/merge.log:2902:Merging userns/for-next (4cf4e48fff05 signal: sort
out si_trapno and si_perf)
arch/m68k/kernel/signal.c:626:  BUILD_BUG_ON(offsetof(siginfo_t,
si_perf) != 0x10);
include/uapi/linux/perf_event.h:467:     * siginfo_t::si_perf, e.g. to
permit user to identify the event.
tools/testing/selftests/perf_events/sigtrap_threads.c:46:/* Unique
value to check si_perf is correctly set from
perf_event_attr::sig_data. */

I will sweep them up as well.

Eric

#84 lore raw

[GIT PULL] siginfo: ABI fixes for v5.13-rc2

In reply to Eric W. Biederman

Linus,

Please pull the for-v5.13-rc2 branch from the git tree:

git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace.git for-v5.13-rc2

HEAD: addd6821190ebf1e9fece0b7848db667fd280e2e signalfd: Remove SIL_FAULT_PERF_EVENT fields from signalfd_siginfo

During the merge window an issue with si_perf and the siginfo ABI came up. The alpha and sparc siginfo structure layout had changed with the addition of SIGTRAP TRAP_PERF and the new field si_perf.

The reason only alpha and sparc were affected is that they are the only architectures that use si_trapno.

Looking deeper it was discovered that si_trapno is used for only a few select signals on alpha and sparc, and that none of the other _sigfault fields past si_addr are used at all. Which means technically no regression on alpha and sparc.

While the alignment concerns might be dismissed the abuse of si_errno by SIGTRAP TRAP_PERF does have the potential to cause regressions in existing userspace.

While we still have time before userspace starts using and depending on the new definition siginfo for SIGTRAP TRAP_PERF this set of changes cleans up siginfo_t.

- The si_trapno field is demoted from magic alpha and sparc status and made an ordinary union member of the _sigfault member of siginfo_t. Without moving it of course.

- si_perf is replaced with si_perf_data and si_perf_type ending the abuse of si_errno.

- BUILD_BUG_ONs are added and various helpers are modified to accommodate this change.

- Unnecessary additions to signalfd_siginfo are removed.

v3: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]
v1: https://lkml.kernel.org/r/[email protected]

You might notice a recent rebase. This code has been sitting in linux-next as ef566ba2d7d9 ("signal: Remove the last few si_perf references"). Which results in the exact same code as the branch I am sending you but the commits differ to keep git bisect working.

The difference is that I squashed a fix for a mips BUILD_BUG_ON about si_perf into the commit that replaces si_perf with si_perf_data and si_perf_type. This keeps the kernel building on all architectures for all commits keeping git-bisect working for everyone.

Eric W. Biederman (9):
signal: Verify the alignment and size of siginfo_t
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Use dedicated helpers to send signals with si_trapno set
signal: Remove __ARCH_SI_TRAPNO
signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency
signal: Factor force_sig_perf out of perf_sigtrap
signal: Deliver all of the siginfo perf data in _perf
signalfd: Remove SIL_FAULT_PERF_EVENT fields from signalfd_siginfo

Marco Elver (3):
sparc64: Add compile-time asserts for siginfo_t offsets
arm: Add compile-time asserts for siginfo_t offsets
arm64: Add compile-time asserts for siginfo_t offsets

 arch/alpha/include/uapi/asm/siginfo.h              |   2 -
 arch/alpha/kernel/osf_sys.c                        |   2 +-
 arch/alpha/kernel/signal.c                         |   4 +-
 arch/alpha/kernel/traps.c                          |  24 ++---
 arch/alpha/mm/fault.c                              |   4 +-
 arch/arm/kernel/signal.c                           |  39 +++++++
 arch/arm64/kernel/signal.c                         |  39 +++++++
 arch/arm64/kernel/signal32.c                       |  39 +++++++
 arch/m68k/kernel/signal.c                          |   3 +-
 arch/mips/include/uapi/asm/siginfo.h               |   2 -
 arch/sparc/include/uapi/asm/siginfo.h              |   3 -
 arch/sparc/kernel/process_64.c                     |   2 +-
 arch/sparc/kernel/signal32.c                       |  37 +++++++
 arch/sparc/kernel/signal_64.c                      |  36 +++++++
 arch/sparc/kernel/sys_sparc_32.c                   |   2 +-
 arch/sparc/kernel/sys_sparc_64.c                   |   2 +-
 arch/sparc/kernel/traps_32.c                       |  22 ++--
 arch/sparc/kernel/traps_64.c                       |  44 ++++----
 arch/sparc/kernel/unaligned_32.c                   |   2 +-
 arch/sparc/mm/fault_32.c                           |   2 +-
 arch/sparc/mm/fault_64.c                           |   2 +-
 arch/x86/kernel/signal_compat.c                    |  15 ++-
 fs/signalfd.c                                      |  23 ++---
 include/linux/compat.h                             |  10 +-
 include/linux/sched/signal.h                       |  13 +--
 include/linux/signal.h                             |   3 +-
 include/uapi/asm-generic/siginfo.h                 |  20 ++--
 include/uapi/linux/perf_event.h                    |   2 +-
 include/uapi/linux/signalfd.h                      |   4 +-
 kernel/events/core.c                               |  11 +-
 kernel/signal.c                                    | 113 +++++++++++++--------
 .../selftests/perf_events/sigtrap_threads.c        |  14 +--
 32 files changed, 377 insertions(+), 163 deletions(-)

Eric

#85 lore raw

Re: [GIT PULL] siginfo: ABI fixes for v5.13-rc2

Quoted 2 lines · Eric W. Biederman

Please pull the for-v5.13-rc2 branch from the git tree:

I really don't like this tree.

The immediate cause for "no" is the silly

#if IS_ENABLED(CONFIG_SPARC)

and

#if IS_ENABLED(CONFIG_ALPHA)

code in kernel/signal.c. It has absolutely zero business being there, when those architectures have a perfectly fine arch/*/kernel/signal.c file where that code would make much more sense *WITHOUT* any odd preprocessor games.

But there are other oddities too, like the new

    send_sig_fault_trapno(SIGFPE, si_code, (void __user *) regs->pc,

0, current);

in the alpha code, which fundamentally seems bogus: using send_sig_fault_trapno() with a '0' for trapno seems entirely incorrect, since the *ONLY* point of that function is to set si_trapno to something non-zero.

So it would seem that a plain send_sig_fault() without that 0 would be the right thing to do.

This also mixes in a lot of other stuff than just the fixes. Which would have been ok during the merge window, but I'm definitely not happy about it now.

Linus

#86 lore raw

Re: [GIT PULL] siginfo: ABI fixes for v5.13-rc2

Quoted 10 lines · Linus Torvalds
Quoted 2 lines · Eric W. Biederman

Please pull the for-v5.13-rc2 branch from the git tree:

I really don't like this tree.

The immediate cause for "no" is the silly

#if IS_ENABLED(CONFIG_SPARC)

Quoted 8 lines

and

#if IS_ENABLED(CONFIG_ALPHA)

code in kernel/signal.c. It has absolutely zero business being there, when those architectures have a perfectly fine arch/*/kernel/signal.c file where that code would make much more sense *WITHOUT* any odd preprocessor games.

The code is generic it just happens those functions are only used on sparc and alpha. Further I really want to make filling out siginfo_t happen in dedicated functions as much as possible in kernel/signal.c. The probably of getting it wrong without a helper functions is very strong. As the code I am fixing demonstrates.

The IS_ENABLED(arch) is mostly there so we can delete the code if/when the architectures are retired in another decade or so.

Quoted 5 lines

But there are other oddities too, like the new

    send_sig_fault_trapno(SIGFPE, si_code, (void __user *) regs->pc,

0, current);

Quoted 7 lines

in the alpha code, which fundamentally seems bogus: using send_sig_fault_trapno() with a '0' for trapno seems entirely incorrect, since the *ONLY* point of that function is to set si_trapno to something non-zero.

So it would seem that a plain send_sig_fault() without that 0 would be the right thing to do.

As it happens the floating point emulation code on alpha is inconsistent with the non floating point emulation code. When using real floating point hardware SIGFPE on alpha always set si_trapno. The floating point emulation code does not look like it has ever set si_trapno.

I continued to used send_sig_fault_trapno to point out that inconsistency.

If alpha floating point emulation was in active use I expect we would care enough to put something other than 0 in there.

Quoted 3 lines

This also mixes in a lot of other stuff than just the fixes. Which would have been ok during the merge window, but I'm definitely not happy about it now.

If the breakage that came with SIGTRAP TRAP_PERF had not been discovered during the merge window I would not be sending this now. It took a little time to dig to the bottom, then the code needed just a little extra time to sit, so there were not surprises.

As for mixing things, I am not quite certain what you are referring to. All of the changes relate to keeping people from shooting themselves in the foot with when using siginfo.

The most noise comes from send_sig_fault vs send_sig_fault_trapno, and force_sig_fault vs force_sig_fault_trapno. That is fundamental to the siginfo fix as it is there to ensure that is safe to treat si_trapno as an ordinary _sigfault union member. Which in turns makes alpha and sparc no longer special with respect to _sigfault, just a little eccentric.

I will concede that renaming SIL_PERF_EVENT to SIL_FAULT_PERF_EVENT is unnecessary, but it certainly makes it clear that we are dealing with _sigfault and not some other part of siginfo_t.

Eric

#87 lore raw

Re: [GIT PULL] siginfo: ABI fixes for v5.13-rc2

Quoted 21 lines · [email protected] (Eric W. Biederman) writes:
Quoted 10 lines · Linus Torvalds
Quoted 2 lines · Eric W. Biederman

Please pull the for-v5.13-rc2 branch from the git tree:

I really don't like this tree.

The immediate cause for "no" is the silly

#if IS_ENABLED(CONFIG_SPARC)

Quoted 8 lines

and

#if IS_ENABLED(CONFIG_ALPHA)

code in kernel/signal.c. It has absolutely zero business being there, when those architectures have a perfectly fine arch/*/kernel/signal.c file where that code would make much more sense *WITHOUT* any odd preprocessor games.

Quoted 8 lines

The code is generic it just happens those functions are only used on sparc and alpha. Further I really want to make filling out siginfo_t happen in dedicated functions as much as possible in kernel/signal.c. The probably of getting it wrong without a helper functions is very strong. As the code I am fixing demonstrates.

The IS_ENABLED(arch) is mostly there so we can delete the code if/when the architectures are retired in another decade or so.

There is also the question of why alpha allows userspace to block SIGFPE.

If it turns out that alpha is just silly by allowing synchronous exceptions to be blocked, then the code really becomes generic and shared shared between sparc and alpha.

Which is really why the code does not make sense in some architecture specific version of signal.c. That and the fact the two functions are almost identical.

If you want I can remove the #ifdefs and we can take up slightly more space until someone implements -ffunction-sections.

Do you know if alpha will be stuck triggering the same floating point error if the SIGFPE is blocked or can alpha somehow continue past it?

If alpha using send_sig instead of force_sig is historical and does not reflect the reality of the hardware alpha can be converted and several of the send_sig variants can be removed. Otherwise alpha remains the odd man out, and the code can remain until all of the alpha hardware dies. (I don't think anyone is manufacturing alpha hardware anymore).

I would look it up but I have lost access to whatever alpha documentation I had.

Eric

#88 lore raw

Re: [GIT PULL] siginfo: ABI fixes for v5.13-rc2

In reply to Eric W. Biederman
Quoted 4 lines · Eric W. Biederman

Looking deeper it was discovered that si_trapno is used for only a few select signals on alpha and sparc, and that none of the other _sigfault fields past si_addr are used at all. Which means technically no regression on alpha and sparc.

If there's no functional regression on any platform, could much of this wait until v5.14, or do we want some of these cleanups right now?

The fixes seem to be for long-existing bugs, not fresh regressions, AFAICS. The asserts & cleanups are useful, but not regression fixes.

I.e. this is a bit scary:

Quoted 1 line
 32 files changed, 377 insertions(+), 163 deletions(-)

at -rc2 time.

Thanks,

Ingo

#89 lore raw

Re: [GIT PULL] siginfo: ABI fixes for v5.13-rc2

Quoted 7 lines · Ingo Molnar
Quoted 4 lines · Eric W. Biederman

Looking deeper it was discovered that si_trapno is used for only a few select signals on alpha and sparc, and that none of the other _sigfault fields past si_addr are used at all. Which means technically no regression on alpha and sparc.

Quoted 7 lines

If there's no functional regression on any platform, could much of this wait until v5.14, or do we want some of these cleanups right now?

The fixes seem to be for long-existing bugs, not fresh regressions, AFAICS. The asserts & cleanups are useful, but not regression fixes.

I.e. this is a bit scary:

The new ABI for SIGTRAP TRAP perf that came in the merge window is broken and wrong. We need to revert/disable the new SIGTRAP TRAP_PERF or have a fix before v5.13.

The issue is old crap getting in the way of a new addition. I think I might see a smaller code change on how to get to something compatible with this.

Quoted 3 lines
Quoted 1 line
 32 files changed, 377 insertions(+), 163 deletions(-)

at -rc2 time.

The additions are all tests to make certain everything is fine. The actual code change without the assertions (tests) is essentially a wash.

Eric

Tested-by Marco Elver
#90 lore raw

[GIT PULL] siginfo: ABI fixes for v5.13-rc3

In reply to Eric W. Biederman

Linus,

Please pull the for-v5.13-rc3 branch from the git tree:

git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace.git for-v5.13-rc3

HEAD: 922e3013046b79b444c87eda5baf43afae1326a8 signalfd: Remove SIL_FAULT_PERF_EVENT fields from signalfd_siginfo

During the merge window an issue with si_perf and the siginfo ABI came up. The alpha and sparc siginfo structure layout had changed with the addition of SIGTRAP TRAP_PERF and the new field si_perf.

The reason only alpha and sparc were affected is that they are the only architectures that use si_trapno.

Looking deeper it was discovered that si_trapno is used for only a few select signals on alpha and sparc, and that none of the other _sigfault fields past si_addr are used at all. Which means technically no regression on alpha and sparc.

While the alignment concerns might be dismissed the abuse of si_errno by SIGTRAP TRAP_PERF does have the potential to cause regressions in existing userspace.

While we still have time before userspace starts using and depending on the new definition siginfo for SIGTRAP TRAP_PERF this set of changes cleans up siginfo_t.

- The si_trapno field is demoted from magic alpha and sparc status and made an ordinary union member of the _sigfault member of siginfo_t. Without moving it of course.

- si_perf is replaced with si_perf_data and si_perf_type ending the abuse of si_errno.

- Unnecessary additions to signalfd_siginfo are removed.

v4: https://lkml.kernel.org/r/[email protected]
v3: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]
v1: https://lkml.kernel.org/r/[email protected]

This version drops the tests and fine grained handling of si_trapno on alpha and sparc (replaced assuming si_trapno is valid for all but the faults that defined different data).

Hopefully this is enough to not be scary as a fix for the ABI issues.

Eric W. Biederman (5):
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Factor force_sig_perf out of perf_sigtrap
signal: Deliver all of the siginfo perf data in _perf
signalfd: Remove SIL_PERF_EVENT fields from signalfd_siginfo

 arch/m68k/kernel/signal.c                          |  3 +-
 arch/x86/kernel/signal_compat.c                    |  9 +++-
 fs/signalfd.c                                      | 23 ++++-----
 include/linux/compat.h                             | 10 ++--
 include/linux/sched/signal.h                       |  1 +
 include/linux/signal.h                             |  1 +
 include/uapi/asm-generic/siginfo.h                 | 15 +++---
 include/uapi/linux/perf_event.h                    |  2 +-
 include/uapi/linux/signalfd.h                      |  4 +-
 kernel/events/core.c                               | 11 +---
 kernel/signal.c                                    | 59 +++++++++++++---------
 .../selftests/perf_events/sigtrap_threads.c        | 14 ++---
 12 files changed, 79 insertions(+), 73 deletions(-)

Eric

merge bot Merged
#91 lore raw

Re: [GIT PULL] siginfo: ABI fixes for v5.13-rc3

The pull request you sent on Fri, 21 May 2021 09:59:53 -0500:

The pull request you sent on Fri, 21 May 2021 09:59:53 -0500:

Quoted 1 line

git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace.git for-v5.13-rc3

has been merged into torvalds/linux.git:
https://git.kernel.org/torvalds/c/a0e31f3a38e77612ed8967aaad28db6d3ee674b5

Thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/prtracker.html
#92 lore raw

v4 cover siginfo: ABI fixes for TRAP_PERF

In reply to Eric W. Biederman

During the merge window an issue with si_perf and the siginfo ABI came up. The alpha and sparc siginfo structure layout had changed with the addition of SIGTRAP TRAP_PERF and the new field si_perf.

The reason only alpha and sparc were affected is that they are the only architectures that use si_trapno.

Looking deeper it was discovered that si_trapno is used for only a few select signals on alpha and sparc, and that none of the other _sigfault fields past si_addr are used at all. Which means technically no regression on alpha and sparc.

While the alignment concerns might be dismissed the abuse of si_errno by SIGTRAP TRAP_PERF does have the potential to cause regressions in existing userspace.

While we still have time before userspace starts using and depending on the new definition siginfo for SIGTRAP TRAP_PERF this set of changes cleans up siginfo_t.

- The si_trapno field is demoted from magic alpha and sparc status and made an ordinary union member of the _sigfault member of siginfo_t. Without moving it of course.

- si_perf is replaced with si_perf_data and si_perf_type ending the abuse of si_errno.

- Unnecessary additions to signalfd_siginfo are removed.

v3: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]
v1: https://lkml.kernel.org/r/[email protected]

This version drops the tests and fine grained handling of si_trapno on alpha and sparc (replaced assuming si_trapno is valid for all but the faults that defined different data).

Eric W. Biederman (5):
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Factor force_sig_perf out of perf_sigtrap
signal: Deliver all of the siginfo perf data in _perf
signalfd: Remove SIL_PERF_EVENT fields from signalfd_siginfo

 arch/m68k/kernel/signal.c                          |  3 +-
 arch/x86/kernel/signal_compat.c                    |  9 +++-
 fs/signalfd.c                                      | 23 ++++-----
 include/linux/compat.h                             | 10 ++--
 include/linux/sched/signal.h                       |  1 +
 include/linux/signal.h                             |  1 +
 include/uapi/asm-generic/siginfo.h                 | 15 +++---
 include/uapi/linux/perf_event.h                    |  2 +-
 include/uapi/linux/signalfd.h                      |  4 +-
 kernel/events/core.c                               | 11 +---
 kernel/signal.c                                    | 59 +++++++++++++---------
 .../selftests/perf_events/sigtrap_threads.c        | 14 ++---
 12 files changed, 79 insertions(+), 73 deletions(-)
#93 lore raw

PATCH v4 1/5 siginfo: Move si_trapno inside the union inside _si_fault

From: "Eric W. Biederman" <[email protected]>

It turns out that linux uses si_trapno very sparingly, and as such it can be considered extra information for a very narrow selection of signals, rather than information that is present with every fault reported in siginfo.

As such move si_trapno inside the union inside of _si_fault. This results in no change in placement, and makes it eaiser
to extend _si_fault in the future as this reduces the number of special cases. In particular with si_trapno included in the union it is no longer a concern that the union must be pointer aligned on most architectures because the union follows immediately after si_addr which is a pointer.

This change results in a difference in siginfo field placement on sparc and alpha for the fields si_addr_lsb, si_lower, si_upper, si_pkey, and si_perf. These architectures do not implement the signals that would use si_addr_lsb, si_lower, si_upper, si_pkey, and si_perf. Further these architecture have not yet implemented the userspace that would use si_perf.

The point of this change is in fact to correct these placement issues before sparc or alpha grow userspace that cares. This change was discussed[1] and the agreement is that this change is currently safe.

[1]: https://lkml.kernel.org/r/CAK8P3a0+uKYwL1NhY6Hvtieghba2hKYGD6hcKx5n8=4Gtt+pHA@mail.gmail.com

v1: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]

---
 arch/x86/kernel/signal_compat.c    | 3 +++
 include/linux/compat.h             | 5 ++---
 include/uapi/asm-generic/siginfo.h | 7 ++-----
 kernel/signal.c                    | 1 +
 4 files changed, 8 insertions(+), 8 deletions(-)
arch/x86/kernel/signal_compat.c +3 -0
index 0e5d0a7e203b..a9fcabd8a5e5 100644
@@ -127,6 +127,9 @@ static inline void signal_compat_build_tests(void)
 	BUILD_BUG_ON(offsetof(siginfo_t, si_addr) != 0x10);
 	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_addr) != 0x0C);
 
+	BUILD_BUG_ON(offsetof(siginfo_t, si_trapno) != 0x18);
+	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_trapno) != 0x10);
+
 	BUILD_BUG_ON(offsetof(siginfo_t, si_addr_lsb) != 0x18);
 	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_addr_lsb) != 0x10);
 
include/linux/compat.h +2 -3
index f0d2dd35d408..6af7bef15e94 100644
@@ -214,12 +214,11 @@ typedef struct compat_siginfo {
 		/* SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT */
 		struct {
 			compat_uptr_t _addr;	/* faulting insn/memory ref. */
-#ifdef __ARCH_SI_TRAPNO
-			int _trapno;	/* TRAP # which caused the signal */
-#endif
 #define __COMPAT_ADDR_BND_PKEY_PAD  (__alignof__(compat_uptr_t) < sizeof(short) ? \
 				     sizeof(short) : __alignof__(compat_uptr_t))
 			union {
+				/* used on alpha and sparc */
+				int _trapno;	/* TRAP # which caused the signal */
 				/*
 				 * used when si_code=BUS_MCEERR_AR or
 				 * used when si_code=BUS_MCEERR_AO
include/uapi/asm-generic/siginfo.h +2 -5
index 03d6f6d2c1fe..e663bf117b46 100644
@@ -63,9 +63,6 @@ union __sifields {
 	/* SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT */
 	struct {
 		void __user *_addr; /* faulting insn/memory ref. */
-#ifdef __ARCH_SI_TRAPNO
-		int _trapno;	/* TRAP # which caused the signal */
-#endif
 #ifdef __ia64__
 		int _imm;		/* immediate value for "break" */
 		unsigned int _flags;	/* see ia64 si_flags */
@@ -75,6 +72,8 @@ union __sifields {
 #define __ADDR_BND_PKEY_PAD  (__alignof__(void *) < sizeof(short) ? \
 			      sizeof(short) : __alignof__(void *))
 		union {
+			/* used on alpha and sparc */
+			int _trapno;	/* TRAP # which caused the signal */
 			/*
 			 * used when si_code=BUS_MCEERR_AR or
 			 * used when si_code=BUS_MCEERR_AO
@@ -150,9 +149,7 @@ typedef struct siginfo {
 #define si_int		_sifields._rt._sigval.sival_int
 #define si_ptr		_sifields._rt._sigval.sival_ptr
 #define si_addr		_sifields._sigfault._addr
-#ifdef __ARCH_SI_TRAPNO
 #define si_trapno	_sifields._sigfault._trapno
-#endif
 #define si_addr_lsb	_sifields._sigfault._addr_lsb
 #define si_lower	_sifields._sigfault._addr_bnd._lower
 #define si_upper	_sifields._sigfault._addr_bnd._upper
kernel/signal.c +1 -0
index c3017aa8024a..65888aec65a0 100644
@@ -4607,6 +4607,7 @@ static inline void siginfo_buildtime_checks(void)
 
 	/* sigfault */
 	CHECK_OFFSET(si_addr);
+	CHECK_OFFSET(si_trapno);
 	CHECK_OFFSET(si_addr_lsb);
 	CHECK_OFFSET(si_lower);
 	CHECK_OFFSET(si_upper);
Signature · 1 line
2.30.1
#94 lore raw

PATCH v4 2/5 signal: Implement SIL_FAULT_TRAPNO

From: "Eric W. Biederman" <[email protected]>

Now that si_trapno is part of the union in _si_fault and available on all architectures, add SIL_FAULT_TRAPNO and update siginfo_layout to return SIL_FAULT_TRAPNO when the code assumes si_trapno is valid.

There is room for future changes to reduce when si_trapno is valid but this is all that is needed to make si_trapno and the other members of the the union in _sigfault mutually exclusive.

Update the code that uses siginfo_layout to deal with SIL_FAULT_TRAPNO and have the same code ignore si_trapno in in all other cases.

v1: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]

---
 fs/signalfd.c          |  8 +++-----
 include/linux/signal.h |  1 +
 kernel/signal.c        | 34 ++++++++++++----------------------
 3 files changed, 16 insertions(+), 27 deletions(-)
fs/signalfd.c +3 -5
index 040a1142915f..e87e59581653 100644
@@ -123,15 +123,13 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		 */
 	case SIL_FAULT:
 		new.ssi_addr = (long) kinfo->si_addr;
-#ifdef __ARCH_SI_TRAPNO
+		break;
+	case SIL_FAULT_TRAPNO:
+		new.ssi_addr = (long) kinfo->si_addr;
 		new.ssi_trapno = kinfo->si_trapno;
-#endif
 		break;
 	case SIL_FAULT_MCEERR:
 		new.ssi_addr = (long) kinfo->si_addr;
-#ifdef __ARCH_SI_TRAPNO
-		new.ssi_trapno = kinfo->si_trapno;
-#endif
 		new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
 		break;
 	case SIL_PERF_EVENT:
include/linux/signal.h +1 -0
index 1e98548d7cf6..5160fd45e5ca 100644
@@ -40,6 +40,7 @@ enum siginfo_layout {
 	SIL_TIMER,
 	SIL_POLL,
 	SIL_FAULT,
+	SIL_FAULT_TRAPNO,
 	SIL_FAULT_MCEERR,
 	SIL_FAULT_BNDERR,
 	SIL_FAULT_PKUERR,
kernel/signal.c +12 -22
index 65888aec65a0..597594ee72de 100644
@@ -1194,6 +1194,7 @@ static inline bool has_si_pid_and_uid(struct kernel_siginfo *info)
 	case SIL_TIMER:
 	case SIL_POLL:
 	case SIL_FAULT:
+	case SIL_FAULT_TRAPNO:
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
@@ -2527,6 +2528,7 @@ static void hide_si_addr_tag_bits(struct ksignal *ksig)
 {
 	switch (siginfo_layout(ksig->sig, ksig->info.si_code)) {
 	case SIL_FAULT:
+	case SIL_FAULT_TRAPNO:
 	case SIL_FAULT_MCEERR:
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
@@ -3214,6 +3216,10 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
 #endif
 			else if ((sig == SIGTRAP) && (si_code == TRAP_PERF))
 				layout = SIL_PERF_EVENT;
+#ifdef __ARCH_SI_TRAPNO
+			else if (layout == SIL_FAULT)
+				layout = SIL_FAULT_TRAPNO;
+#endif
 		}
 		else if (si_code <= NSIGPOLL)
 			layout = SIL_POLL;
@@ -3317,30 +3323,22 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 		break;
 	case SIL_FAULT:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+		break;
+	case SIL_FAULT_TRAPNO:
+		to->si_addr = ptr_to_compat(from->si_addr);
 		to->si_trapno = from->si_trapno;
-#endif
 		break;
 	case SIL_FAULT_MCEERR:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_addr_lsb = from->si_addr_lsb;
 		break;
 	case SIL_FAULT_BNDERR:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_lower = ptr_to_compat(from->si_lower);
 		to->si_upper = ptr_to_compat(from->si_upper);
 		break;
 	case SIL_FAULT_PKUERR:
 		to->si_addr = ptr_to_compat(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_pkey = from->si_pkey;
 		break;
 	case SIL_PERF_EVENT:
@@ -3401,30 +3399,22 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 		break;
 	case SIL_FAULT:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
+		break;
+	case SIL_FAULT_TRAPNO:
+		to->si_addr = compat_ptr(from->si_addr);
 		to->si_trapno = from->si_trapno;
-#endif
 		break;
 	case SIL_FAULT_MCEERR:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_addr_lsb = from->si_addr_lsb;
 		break;
 	case SIL_FAULT_BNDERR:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_lower = compat_ptr(from->si_lower);
 		to->si_upper = compat_ptr(from->si_upper);
 		break;
 	case SIL_FAULT_PKUERR:
 		to->si_addr = compat_ptr(from->si_addr);
-#ifdef __ARCH_SI_TRAPNO
-		to->si_trapno = from->si_trapno;
-#endif
 		to->si_pkey = from->si_pkey;
 		break;
 	case SIL_PERF_EVENT:
Signature · 1 line
2.30.1
#95 lore raw

PATCH v4 3/5 signal: Factor force_sig_perf out of perf_sigtrap

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Separate filling in siginfo for TRAP_PERF from deciding that siginal needs to be sent.

There are enough little details that need to be correct when properly filling in siginfo_t that it is easy to make mistakes if filling in the siginfo_t is in the same function with other logic. So factor out force_sig_perf to reduce the cognative load of on reviewers, maintainers and implementors.

v1: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]

---
 include/linux/sched/signal.h |  1 +
 kernel/events/core.c         | 11 ++---------
 kernel/signal.c              | 13 +++++++++++++
 3 files changed, 16 insertions(+), 9 deletions(-)
include/linux/sched/signal.h +1 -0
index 3f6a0fcaa10c..7f4278fa21fe 100644
@@ -326,6 +326,7 @@ int send_sig_mceerr(int code, void __user *, short, struct task_struct *);
 
 int force_sig_bnderr(void __user *addr, void __user *lower, void __user *upper);
 int force_sig_pkuerr(void __user *addr, u32 pkey);
+int force_sig_perf(void __user *addr, u32 type, u64 sig_data);
 
 int force_sig_ptrace_errno_trap(int errno, void __user *addr);
 
kernel/events/core.c +2 -9
index 928b166d888e..48ea8863183b 100644
@@ -6394,8 +6394,6 @@ void perf_event_wakeup(struct perf_event *event)
 
 static void perf_sigtrap(struct perf_event *event)
 {
-	struct kernel_siginfo info;
-
 	/*
 	 * We'd expect this to only occur if the irq_work is delayed and either
 	 * ctx->task or current has changed in the meantime. This can be the
@@ -6410,13 +6408,8 @@ static void perf_sigtrap(struct perf_event *event)
 	if (current->flags & PF_EXITING)
 		return;
 
-	clear_siginfo(&info);
-	info.si_signo = SIGTRAP;
-	info.si_code = TRAP_PERF;
-	info.si_errno = event->attr.type;
-	info.si_perf = event->attr.sig_data;
-	info.si_addr = (void __user *)event->pending_addr;
-	force_sig_info(&info);
+	force_sig_perf((void __user *)event->pending_addr,
+		       event->attr.type, event->attr.sig_data);
 }
 
 static void perf_pending_event_disable(struct perf_event *event)
kernel/signal.c +13 -0
index 597594ee72de..3a18d13c39b2 100644
@@ -1763,6 +1763,19 @@ int force_sig_pkuerr(void __user *addr, u32 pkey)
 }
 #endif
 
+int force_sig_perf(void __user *addr, u32 type, u64 sig_data)
+{
+	struct kernel_siginfo info;
+
+	clear_siginfo(&info);
+	info.si_signo = SIGTRAP;
+	info.si_errno = type;
+	info.si_code  = TRAP_PERF;
+	info.si_addr  = addr;
+	info.si_perf  = sig_data;
+	return force_sig_info(&info);
+}
+
 /* For the crazy architectures that include trap information in
  * the errno field, instead of an actual errno value.
  */
Signature · 1 line
2.30.1
#96 lore raw

PATCH v4 4/5 signal: Deliver all of the siginfo perf data in _perf

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

Don't abuse si_errno and deliver all of the perf data in _perf member of siginfo_t.

Note: The data field in the perf data structures in a u64 to allow a pointer to be encoded without needed to implement a 32bit and 64bit version of the same structure. There already exists a 32bit and 64bit versions siginfo_t, and the 32bit version can not include a 64bit member as it only has 32bit alignment. So unsigned long is used in siginfo_t instead of a u64 as unsigned long can encode a pointer on all architectures linux supports.

v1: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]
v3: https://lkml.kernel.org/r/[email protected]

---
 arch/m68k/kernel/signal.c                     |  3 ++-
 arch/x86/kernel/signal_compat.c               |  6 ++++--
 fs/signalfd.c                                 |  3 ++-
 include/linux/compat.h                        |  5 ++++-
 include/uapi/asm-generic/siginfo.h            |  8 +++++--
 include/uapi/linux/perf_event.h               |  2 +-
 include/uapi/linux/signalfd.h                 |  4 ++--
 kernel/signal.c                               | 21 ++++++++++++-------
 .../selftests/perf_events/sigtrap_threads.c   | 14 ++++++-------
 9 files changed, 41 insertions(+), 25 deletions(-)
arch/m68k/kernel/signal.c +2 -1
index a4b7ee1df211..8f215e79e70e 100644
@@ -623,7 +623,8 @@ static inline void siginfo_build_tests(void)
 	BUILD_BUG_ON(offsetof(siginfo_t, si_pkey) != 0x12);
 
 	/* _sigfault._perf */
-	BUILD_BUG_ON(offsetof(siginfo_t, si_perf) != 0x10);
+	BUILD_BUG_ON(offsetof(siginfo_t, si_perf_data) != 0x10);
+	BUILD_BUG_ON(offsetof(siginfo_t, si_perf_type) != 0x14);
 
 	/* _sigpoll */
 	BUILD_BUG_ON(offsetof(siginfo_t, si_band)   != 0x0c);
arch/x86/kernel/signal_compat.c +4 -2
index a9fcabd8a5e5..06743ec054d2 100644
@@ -141,8 +141,10 @@ static inline void signal_compat_build_tests(void)
 	BUILD_BUG_ON(offsetof(siginfo_t, si_pkey) != 0x20);
 	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_pkey) != 0x14);
 
-	BUILD_BUG_ON(offsetof(siginfo_t, si_perf) != 0x18);
-	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_perf) != 0x10);
+	BUILD_BUG_ON(offsetof(siginfo_t, si_perf_data) != 0x18);
+	BUILD_BUG_ON(offsetof(siginfo_t, si_perf_type) != 0x20);
+	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_perf_data) != 0x10);
+	BUILD_BUG_ON(offsetof(compat_siginfo_t, si_perf_type) != 0x14);
 
 	CHECK_CSI_OFFSET(_sigpoll);
 	CHECK_CSI_SIZE  (_sigpoll, 2*sizeof(int));
fs/signalfd.c +2 -1
index e87e59581653..373df2f12415 100644
@@ -134,7 +134,8 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		break;
 	case SIL_PERF_EVENT:
 		new.ssi_addr = (long) kinfo->si_addr;
-		new.ssi_perf = kinfo->si_perf;
+		new.ssi_perf_type = kinfo->si_perf_type;
+		new.ssi_perf_data = kinfo->si_perf_data;
 		break;
 	case SIL_CHLD:
 		new.ssi_pid    = kinfo->si_pid;
include/linux/compat.h +4 -1
index 6af7bef15e94..a27fffaae121 100644
@@ -236,7 +236,10 @@ typedef struct compat_siginfo {
 					u32 _pkey;
 				} _addr_pkey;
 				/* used when si_code=TRAP_PERF */
-				compat_ulong_t _perf;
+				struct {
+					compat_ulong_t _data;
+					u32 _type;
+				} _perf;
 			};
 		} _sigfault;
 
include/uapi/asm-generic/siginfo.h +6 -2
index e663bf117b46..5a3c221f4c9d 100644
@@ -91,7 +91,10 @@ union __sifields {
 				__u32 _pkey;
 			} _addr_pkey;
 			/* used when si_code=TRAP_PERF */
-			unsigned long _perf;
+			struct {
+				unsigned long _data;
+				__u32 _type;
+			} _perf;
 		};
 	} _sigfault;
 
@@ -154,7 +157,8 @@ typedef struct siginfo {
 #define si_lower	_sifields._sigfault._addr_bnd._lower
 #define si_upper	_sifields._sigfault._addr_bnd._upper
 #define si_pkey		_sifields._sigfault._addr_pkey._pkey
-#define si_perf		_sifields._sigfault._perf
+#define si_perf_data	_sifields._sigfault._perf._data
+#define si_perf_type	_sifields._sigfault._perf._type
 #define si_band		_sifields._sigpoll._band
 #define si_fd		_sifields._sigpoll._fd
 #define si_call_addr	_sifields._sigsys._call_addr
include/uapi/linux/perf_event.h +1 -1
index e54e639248c8..7b14753b3d38 100644
@@ -464,7 +464,7 @@ struct perf_event_attr {
 
 	/*
 	 * User provided data if sigtrap=1, passed back to user via
-	 * siginfo_t::si_perf, e.g. to permit user to identify the event.
+	 * siginfo_t::si_perf_data, e.g. to permit user to identify the event.
 	 */
 	__u64	sig_data;
 };
include/uapi/linux/signalfd.h +2 -2
index 7e333042c7e3..e78dddf433fc 100644
@@ -39,8 +39,8 @@ struct signalfd_siginfo {
 	__s32 ssi_syscall;
 	__u64 ssi_call_addr;
 	__u32 ssi_arch;
-	__u32 __pad3;
-	__u64 ssi_perf;
+	__u32 ssi_perf_type;
+	__u64 ssi_perf_data;
 
 	/*
 	 * Pad strcture to 128 bytes. Remember to update the
kernel/signal.c +13 -8
index 3a18d13c39b2..dca53515ae3f 100644
@@ -1768,11 +1768,13 @@ int force_sig_perf(void __user *addr, u32 type, u64 sig_data)
 	struct kernel_siginfo info;
 
 	clear_siginfo(&info);
-	info.si_signo = SIGTRAP;
-	info.si_errno = type;
-	info.si_code  = TRAP_PERF;
-	info.si_addr  = addr;
-	info.si_perf  = sig_data;
+	info.si_signo     = SIGTRAP;
+	info.si_errno     = 0;
+	info.si_code      = TRAP_PERF;
+	info.si_addr      = addr;
+	info.si_perf_data = sig_data;
+	info.si_perf_type = type;
+
 	return force_sig_info(&info);
 }
 
@@ -3356,7 +3358,8 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 		break;
 	case SIL_PERF_EVENT:
 		to->si_addr = ptr_to_compat(from->si_addr);
-		to->si_perf = from->si_perf;
+		to->si_perf_data = from->si_perf_data;
+		to->si_perf_type = from->si_perf_type;
 		break;
 	case SIL_CHLD:
 		to->si_pid = from->si_pid;
@@ -3432,7 +3435,8 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 		break;
 	case SIL_PERF_EVENT:
 		to->si_addr = compat_ptr(from->si_addr);
-		to->si_perf = from->si_perf;
+		to->si_perf_data = from->si_perf_data;
+		to->si_perf_type = from->si_perf_type;
 		break;
 	case SIL_CHLD:
 		to->si_pid    = from->si_pid;
@@ -4615,7 +4619,8 @@ static inline void siginfo_buildtime_checks(void)
 	CHECK_OFFSET(si_lower);
 	CHECK_OFFSET(si_upper);
 	CHECK_OFFSET(si_pkey);
-	CHECK_OFFSET(si_perf);
+	CHECK_OFFSET(si_perf_data);
+	CHECK_OFFSET(si_perf_type);
 
 	/* sigpoll */
 	CHECK_OFFSET(si_band);
tools/testing/selftests/perf_events/sigtrap_threads.c +7 -7
index 78ddf5e11625..8e83cf91513a 100644
@@ -43,7 +43,7 @@ static struct {
 	siginfo_t first_siginfo;	/* First observed siginfo_t. */
 } ctx;
 
-/* Unique value to check si_perf is correctly set from perf_event_attr::sig_data. */
+/* Unique value to check si_perf_data is correctly set from perf_event_attr::sig_data. */
 #define TEST_SIG_DATA(addr) (~(unsigned long)(addr))
 
 static struct perf_event_attr make_event_attr(bool enabled, volatile void *addr)
@@ -164,8 +164,8 @@ TEST_F(sigtrap_threads, enable_event)
 	EXPECT_EQ(ctx.signal_count, NUM_THREADS);
 	EXPECT_EQ(ctx.tids_want_signal, 0);
 	EXPECT_EQ(ctx.first_siginfo.si_addr, &ctx.iterate_on);
-	EXPECT_EQ(ctx.first_siginfo.si_errno, PERF_TYPE_BREAKPOINT);
-	EXPECT_EQ(ctx.first_siginfo.si_perf, TEST_SIG_DATA(&ctx.iterate_on));
+	EXPECT_EQ(ctx.first_siginfo.si_perf_type, PERF_TYPE_BREAKPOINT);
+	EXPECT_EQ(ctx.first_siginfo.si_perf_data, TEST_SIG_DATA(&ctx.iterate_on));
 
 	/* Check enabled for parent. */
 	ctx.iterate_on = 0;
@@ -183,8 +183,8 @@ TEST_F(sigtrap_threads, modify_and_enable_event)
 	EXPECT_EQ(ctx.signal_count, NUM_THREADS);
 	EXPECT_EQ(ctx.tids_want_signal, 0);
 	EXPECT_EQ(ctx.first_siginfo.si_addr, &ctx.iterate_on);
-	EXPECT_EQ(ctx.first_siginfo.si_errno, PERF_TYPE_BREAKPOINT);
-	EXPECT_EQ(ctx.first_siginfo.si_perf, TEST_SIG_DATA(&ctx.iterate_on));
+	EXPECT_EQ(ctx.first_siginfo.si_perf_type, PERF_TYPE_BREAKPOINT);
+	EXPECT_EQ(ctx.first_siginfo.si_perf_data, TEST_SIG_DATA(&ctx.iterate_on));
 
 	/* Check enabled for parent. */
 	ctx.iterate_on = 0;
@@ -203,8 +203,8 @@ TEST_F(sigtrap_threads, signal_stress)
 	EXPECT_EQ(ctx.signal_count, NUM_THREADS * ctx.iterate_on);
 	EXPECT_EQ(ctx.tids_want_signal, 0);
 	EXPECT_EQ(ctx.first_siginfo.si_addr, &ctx.iterate_on);
-	EXPECT_EQ(ctx.first_siginfo.si_errno, PERF_TYPE_BREAKPOINT);
-	EXPECT_EQ(ctx.first_siginfo.si_perf, TEST_SIG_DATA(&ctx.iterate_on));
+	EXPECT_EQ(ctx.first_siginfo.si_perf_type, PERF_TYPE_BREAKPOINT);
+	EXPECT_EQ(ctx.first_siginfo.si_perf_data, TEST_SIG_DATA(&ctx.iterate_on));
 }
 
 TEST_HARNESS_MAIN
Signature · 1 line
2.30.1
#97 lore raw

PATCH v4 5/5 signalfd: Remove SIL_PERF_EVENT fields from signalfd_siginfo

In reply to Eric W. Beiderman

From: "Eric W. Biederman" <[email protected]>

With the addition of ssi_perf_data and ssi_perf_type struct signalfd_siginfo is dangerously close to running out of space. All that remains is just enough space for two additional 64bit fields. A practice of adding all possible siginfo_t fields into struct singalfd_siginfo can not be supported as adding the missing fields ssi_lower, ssi_upper, and ssi_pkey would require two 64bit fields and one 32bit fields. In practice the fields ssi_perf_data and ssi_perf_type can never be used by signalfd as the signal that generates them always delivers them synchronously to the thread that triggers them.

Therefore until someone actually needs the fields ssi_perf_data and ssi_perf_type in signalfd_siginfo remove them. This leaves a bit more room for future expansion.

v1: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]

---
 fs/signalfd.c                 | 16 ++++++----------
 include/uapi/linux/signalfd.h |  4 +---
 2 files changed, 7 insertions(+), 13 deletions(-)
fs/signalfd.c +6 -10
index 373df2f12415..167b5889db4b 100644
@@ -114,12 +114,13 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		break;
 	case SIL_FAULT_BNDERR:
 	case SIL_FAULT_PKUERR:
+	case SIL_PERF_EVENT:
 		/*
-		 * Fall through to the SIL_FAULT case.  Both SIL_FAULT_BNDERR
-		 * and SIL_FAULT_PKUERR are only generated by faults that
-		 * deliver them synchronously to userspace.  In case someone
-		 * injects one of these signals and signalfd catches it treat
-		 * it as SIL_FAULT.
+		 * Fall through to the SIL_FAULT case.  SIL_FAULT_BNDERR,
+		 * SIL_FAULT_PKUERR, and SIL_PERF_EVENT are only
+		 * generated by faults that deliver them synchronously to
+		 * userspace.  In case someone injects one of these signals
+		 * and signalfd catches it treat it as SIL_FAULT.
 		 */
 	case SIL_FAULT:
 		new.ssi_addr = (long) kinfo->si_addr;
@@ -132,11 +133,6 @@ static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
 		new.ssi_addr = (long) kinfo->si_addr;
 		new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
 		break;
-	case SIL_PERF_EVENT:
-		new.ssi_addr = (long) kinfo->si_addr;
-		new.ssi_perf_type = kinfo->si_perf_type;
-		new.ssi_perf_data = kinfo->si_perf_data;
-		break;
 	case SIL_CHLD:
 		new.ssi_pid    = kinfo->si_pid;
 		new.ssi_uid    = kinfo->si_uid;
include/uapi/linux/signalfd.h +1 -3
index e78dddf433fc..83429a05b698 100644
@@ -39,8 +39,6 @@ struct signalfd_siginfo {
 	__s32 ssi_syscall;
 	__u64 ssi_call_addr;
 	__u32 ssi_arch;
-	__u32 ssi_perf_type;
-	__u64 ssi_perf_data;
 
 	/*
 	 * Pad strcture to 128 bytes. Remember to update the
@@ -51,7 +49,7 @@ struct signalfd_siginfo {
 	 * comes out of a read(2) and we really don't want to have
 	 * a compat on read(2).
 	 */
-	__u8 __pad[16];
+	__u8 __pad[28];
 };
 
 
Signature · 1 line
2.30.1
#98 lore raw

Re: [PATCH v4 0/5] siginfo: ABI fixes for TRAP_PERF

In reply to Eric W. Biederman
Quoted 30 lines · Eric W. Biederman

During the merge window an issue with si_perf and the siginfo ABI came up. The alpha and sparc siginfo structure layout had changed with the addition of SIGTRAP TRAP_PERF and the new field si_perf.

The reason only alpha and sparc were affected is that they are the only architectures that use si_trapno.

Looking deeper it was discovered that si_trapno is used for only a few select signals on alpha and sparc, and that none of the other _sigfault fields past si_addr are used at all. Which means technically no regression on alpha and sparc.

While the alignment concerns might be dismissed the abuse of si_errno by SIGTRAP TRAP_PERF does have the potential to cause regressions in existing userspace.

While we still have time before userspace starts using and depending on the new definition siginfo for SIGTRAP TRAP_PERF this set of changes cleans up siginfo_t.

- The si_trapno field is demoted from magic alpha and sparc status and made an ordinary union member of the _sigfault member of siginfo_t. Without moving it of course.

- si_perf is replaced with si_perf_data and si_perf_type ending the abuse of si_errno.

- Unnecessary additions to signalfd_siginfo are removed.

Quoted 7 lines

v3: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]
v1: https://lkml.kernel.org/r/[email protected]

This version drops the tests and fine grained handling of si_trapno on alpha and sparc (replaced assuming si_trapno is valid for all but the faults that defined different data).

And just to clarify, the rest of the series (including static-asserts) for the next merge-window will be sent once this series is all sorted, correct?

Quoted 6 lines

Eric W. Biederman (5):
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Factor force_sig_perf out of perf_sigtrap
signal: Deliver all of the siginfo perf data in _perf
signalfd: Remove SIL_PERF_EVENT fields from signalfd_siginfo

Looks good, thank you! I build-tested (defconfig -- x86_64, i386, arm, arm64, m68k, sparc, alpha) this series together with a local patch to pull in the static asserts from v3. Also re-ran perf_events kselftests on x86_64 (native and 32bit compat).

Thanks,
-- Marco

#99 lore raw

Re: [PATCH v4 0/5] siginfo: ABI fixes for TRAP_PERF

Quoted 35 lines · Marco Elver
Quoted 34 lines · Eric W. Biederman

During the merge window an issue with si_perf and the siginfo ABI came up. The alpha and sparc siginfo structure layout had changed with the addition of SIGTRAP TRAP_PERF and the new field si_perf.

The reason only alpha and sparc were affected is that they are the only architectures that use si_trapno.

Looking deeper it was discovered that si_trapno is used for only a few select signals on alpha and sparc, and that none of the other _sigfault fields past si_addr are used at all. Which means technically no regression on alpha and sparc.

While the alignment concerns might be dismissed the abuse of si_errno by SIGTRAP TRAP_PERF does have the potential to cause regressions in existing userspace.

While we still have time before userspace starts using and depending on the new definition siginfo for SIGTRAP TRAP_PERF this set of changes cleans up siginfo_t.

- The si_trapno field is demoted from magic alpha and sparc status and made an ordinary union member of the _sigfault member of siginfo_t. Without moving it of course.

- si_perf is replaced with si_perf_data and si_perf_type ending the abuse of si_errno.

- Unnecessary additions to signalfd_siginfo are removed.

v3: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]
v1: https://lkml.kernel.org/r/[email protected]

Quoted 7 lines
Quoted 3 lines

This version drops the tests and fine grained handling of si_trapno on alpha and sparc (replaced assuming si_trapno is valid for all but the faults that defined different data).

And just to clarify, the rest of the series (including static-asserts) for the next merge-window will be sent once this series is all sorted, correct?

That is the plan.

I really wonder about alphas use of si_trapno, and alphas use send_sig instead of force_sig. It could be worth looking into those as it has the potential to simplify the code.

Quoted 11 lines
Quoted 6 lines

Eric W. Biederman (5):
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Factor force_sig_perf out of perf_sigtrap
signal: Deliver all of the siginfo perf data in _perf
signalfd: Remove SIL_PERF_EVENT fields from signalfd_siginfo

Looks good, thank you! I build-tested (defconfig -- x86_64, i386, arm, arm64, m68k, sparc, alpha) this series together with a local patch to pull in the static asserts from v3. Also re-ran perf_events kselftests on x86_64 (native and 32bit compat).

Thanks,

Can I have your Tested-by?

Eric

author Tested-by
#100 lore raw

Re: [PATCH v4 0/5] siginfo: ABI fixes for TRAP_PERF

Quoted 59 lines · Eric W. Biederman
Quoted 35 lines · Marco Elver
Quoted 34 lines · Eric W. Biederman

During the merge window an issue with si_perf and the siginfo ABI came up. The alpha and sparc siginfo structure layout had changed with the addition of SIGTRAP TRAP_PERF and the new field si_perf.

The reason only alpha and sparc were affected is that they are the only architectures that use si_trapno.

Looking deeper it was discovered that si_trapno is used for only a few select signals on alpha and sparc, and that none of the other _sigfault fields past si_addr are used at all. Which means technically no regression on alpha and sparc.

While the alignment concerns might be dismissed the abuse of si_errno by SIGTRAP TRAP_PERF does have the potential to cause regressions in existing userspace.

While we still have time before userspace starts using and depending on the new definition siginfo for SIGTRAP TRAP_PERF this set of changes cleans up siginfo_t.

- The si_trapno field is demoted from magic alpha and sparc status and made an ordinary union member of the _sigfault member of siginfo_t. Without moving it of course.

- si_perf is replaced with si_perf_data and si_perf_type ending the abuse of si_errno.

- Unnecessary additions to signalfd_siginfo are removed.

v3: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]
v1: https://lkml.kernel.org/r/[email protected]

Quoted 7 lines
Quoted 3 lines

This version drops the tests and fine grained handling of si_trapno on alpha and sparc (replaced assuming si_trapno is valid for all but the faults that defined different data).

And just to clarify, the rest of the series (including static-asserts) for the next merge-window will be sent once this series is all sorted, correct?

That is the plan.

I really wonder about alphas use of si_trapno, and alphas use send_sig instead of force_sig. It could be worth looking into those as it has the potential to simplify the code.

Quoted 7 lines
Quoted 6 lines

Eric W. Biederman (5):
siginfo: Move si_trapno inside the union inside _si_fault
signal: Implement SIL_FAULT_TRAPNO
signal: Factor force_sig_perf out of perf_sigtrap
signal: Deliver all of the siginfo perf data in _perf
signalfd: Remove SIL_PERF_EVENT fields from signalfd_siginfo

Quoted 8 lines
Quoted 4 lines

Looks good, thank you! I build-tested (defconfig -- x86_64, i386, arm, arm64, m68k, sparc, alpha) this series together with a local patch to pull in the static asserts from v3. Also re-ran perf_events kselftests on x86_64 (native and 32bit compat).

Thanks,

Can I have your Tested-by?

Of course,

Thanks,
-- Marco

#101 lore raw

Re: [RFC][PATCH 0/3] signal: Move si_trapno into the _si_fault union

In reply to Marco Elver
Quoted 11 lines · Marco Elver
Quoted 10 lines · Eric W. Biederman

Well with 7 patches instead of 3 that was a little more than I thought I was going to send.

However that does demonstrate what I am thinking, and I think most of the changes are reasonable at this point.

I am very curious how synchronous this all is, because if this code is truly synchronous updating signalfd to handle this class of signal doesn't really make sense.

Just a note on this: the reason for adding signalfd support was based on the comment at SIL_FAULT_PKUERR:

Quoted 7 lines

/*
* Fall through to the SIL_FAULT case. Both SIL_FAULT_BNDERR
* and SIL_FAULT_PKUERR are only generated by faults that
* deliver them synchronously to userspace. In case someone
* injects one of these signals and signalfd catches it treat
* it as SIL_FAULT.
*/

The same would hold for SIL_FAULT_PERF_EVENT, where somebody injects (re-injects perhaps?) such an event. But otherwise, yes, non-synchronous handling of SIGTRAP/TRAP_PERF is pretty useless for almost all usecases I can think of.

Thanks,
-- Marco

#102 lore raw

cover Final si_trapno bits

In reply to Eric W. Biederman

As a part of a fix for the ABI of the newly added SIGTRAP TRAP_PERF a si_trapno was reduced to an ordinary extention of the _sigfault case of struct siginfo.

When Linus saw the complete set of changes come in as a fix he requested that the set of changes be trimmed down to just what was necessary to fix the SIGTRAP TRAP_PERF ABI.

I had intended to get the rest of the changes into the merge window for v5.14 but I dropped the ball.

I have made the changes to stop using __ARCH_SI_TRAPNO be per architecture so they are easier to review. In doing so I found one place on alpha where I used send_sig_fault instead of send_sig_fault_trapno(... si_trapno = 0). That would not have changed the userspace behavior but it did make the kernel code less clear.

My rule in these patches is everywhere that siginfo layout calls for SIL_FAULT_TRAPNO the code uses either force_sig_fault_trapno or send_sig_fault_trapno.

And of course I have rebased and compile tested Marco's compile time assert patches.

Eric

Eric W. Biederman (3):
signal/sparc: si_trapno is only used with SIGILL ILL_ILLTRP signal/alpha: si_trapno is only used with SIGFPE and SIGTRAP TRAP_UNK
signal: Remove the generic __ARCH_SI_TRAPNO support

Marco Elver (3):
sparc64: Add compile-time asserts for siginfo_t offsets
arm: Add compile-time asserts for siginfo_t offsets
arm64: Add compile-time asserts for siginfo_t offsets

 arch/alpha/include/uapi/asm/siginfo.h |  2 --
 arch/alpha/kernel/osf_sys.c           |  2 +-
 arch/alpha/kernel/signal.c            |  4 +--
 arch/alpha/kernel/traps.c             | 26 +++++++++---------
 arch/alpha/mm/fault.c                 |  4 +--
 arch/arm/kernel/signal.c              | 37 +++++++++++++++++++++++++
 arch/arm64/kernel/signal.c            | 37 +++++++++++++++++++++++++
 arch/arm64/kernel/signal32.c          | 37 +++++++++++++++++++++++++
 arch/mips/include/uapi/asm/siginfo.h  |  2 --
 arch/sparc/include/uapi/asm/siginfo.h |  3 --
 arch/sparc/kernel/process_64.c        |  2 +-
 arch/sparc/kernel/signal32.c          | 35 +++++++++++++++++++++++
 arch/sparc/kernel/signal_64.c         | 34 +++++++++++++++++++++++
 arch/sparc/kernel/sys_sparc_32.c      |  2 +-
 arch/sparc/kernel/sys_sparc_64.c      |  2 +-
 arch/sparc/kernel/traps_32.c          | 22 +++++++--------
 arch/sparc/kernel/traps_64.c          | 44 +++++++++++++----------------
 arch/sparc/kernel/unaligned_32.c      |  2 +-
 arch/sparc/mm/fault_32.c              |  2 +-
 arch/sparc/mm/fault_64.c              |  2 +-
 include/linux/sched/signal.h          | 11 ++------
 kernel/signal.c                       | 52 ++++++++++++++++++++++++++---------
 22 files changed, 276 insertions(+), 88 deletions(-)
#103 lore raw

PATCH 1/6 sparc64: Add compile-time asserts for siginfo_t offsets

From: Marco Elver <[email protected]>

To help catch ABI breaks at compile-time, add compile-time assertions to verify the siginfo_t layout. Unlike other architectures, sparc64 is special, because it is one of few architectures requiring si_trapno. ABI breaks around that field would only be caught here.

---
 arch/sparc/kernel/signal32.c  | 35 +++++++++++++++++++++++++++++++++++
 arch/sparc/kernel/signal_64.c | 34 ++++++++++++++++++++++++++++++++++
 2 files changed, 69 insertions(+)
arch/sparc/kernel/signal32.c +35 -0
index e9695a06492f..65fd26ae9d25 100644
@@ -745,3 +745,38 @@ asmlinkage int do_sys32_sigstack(u32 u_ssptr, u32 u_ossptr, unsigned long sp)
 out:
 	return ret;
 }
+
+/*
+ * Compile-time assertions for siginfo_t offsets. Check NSIG* as well, as
+ * changes likely come with new fields that should be added below.
+ */
+static_assert(NSIGILL	== 11);
+static_assert(NSIGFPE	== 15);
+static_assert(NSIGSEGV	== 9);
+static_assert(NSIGBUS	== 5);
+static_assert(NSIGTRAP	== 6);
+static_assert(NSIGCHLD	== 6);
+static_assert(NSIGSYS	== 2);
+static_assert(offsetof(compat_siginfo_t, si_signo)	== 0x00);
+static_assert(offsetof(compat_siginfo_t, si_errno)	== 0x04);
+static_assert(offsetof(compat_siginfo_t, si_code)	== 0x08);
+static_assert(offsetof(compat_siginfo_t, si_pid)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_uid)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_tid)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_overrun)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_status)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_utime)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_stime)	== 0x1c);
+static_assert(offsetof(compat_siginfo_t, si_value)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_int)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_ptr)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_addr)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_trapno)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_addr_lsb)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_lower)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_upper)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_pkey)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_perf_data)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_perf_type)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_band)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_fd)		== 0x10);
arch/sparc/kernel/signal_64.c +34 -0
index a0eec62c825d..a58e0cc45d24 100644
@@ -556,3 +556,37 @@ void do_notify_resume(struct pt_regs *regs, unsigned long orig_i0, unsigned long
 	user_enter();
 }
 
+/*
+ * Compile-time assertions for siginfo_t offsets. Check NSIG* as well, as
+ * changes likely come with new fields that should be added below.
+ */
+static_assert(NSIGILL	== 11);
+static_assert(NSIGFPE	== 15);
+static_assert(NSIGSEGV	== 9);
+static_assert(NSIGBUS	== 5);
+static_assert(NSIGTRAP	== 6);
+static_assert(NSIGCHLD	== 6);
+static_assert(NSIGSYS	== 2);
+static_assert(offsetof(siginfo_t, si_signo)	== 0x00);
+static_assert(offsetof(siginfo_t, si_errno)	== 0x04);
+static_assert(offsetof(siginfo_t, si_code)	== 0x08);
+static_assert(offsetof(siginfo_t, si_pid)	== 0x10);
+static_assert(offsetof(siginfo_t, si_uid)	== 0x14);
+static_assert(offsetof(siginfo_t, si_tid)	== 0x10);
+static_assert(offsetof(siginfo_t, si_overrun)	== 0x14);
+static_assert(offsetof(siginfo_t, si_status)	== 0x18);
+static_assert(offsetof(siginfo_t, si_utime)	== 0x20);
+static_assert(offsetof(siginfo_t, si_stime)	== 0x28);
+static_assert(offsetof(siginfo_t, si_value)	== 0x18);
+static_assert(offsetof(siginfo_t, si_int)	== 0x18);
+static_assert(offsetof(siginfo_t, si_ptr)	== 0x18);
+static_assert(offsetof(siginfo_t, si_addr)	== 0x10);
+static_assert(offsetof(siginfo_t, si_trapno)	== 0x18);
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x18);
+static_assert(offsetof(siginfo_t, si_lower)	== 0x20);
+static_assert(offsetof(siginfo_t, si_upper)	== 0x28);
+static_assert(offsetof(siginfo_t, si_pkey)	== 0x20);
+static_assert(offsetof(siginfo_t, si_perf_data)	== 0x18);
+static_assert(offsetof(siginfo_t, si_perf_type)	== 0x20);
+static_assert(offsetof(siginfo_t, si_band)	== 0x10);
+static_assert(offsetof(siginfo_t, si_fd)	== 0x14);
Signature · 1 line
2.20.1
#104 lore raw

PATCH 2/6 arm: Add compile-time asserts for siginfo_t offsets

In reply to Eric W. Biederman

From: Marco Elver <[email protected]>

To help catch ABI breaks at compile-time, add compile-time assertions to verify the siginfo_t layout.

This could have caught that we cannot portably add 64-bit integers to siginfo_t on 32-bit architectures like Arm before reaching -next:
https://lkml.kernel.org/r/[email protected]

---
 arch/arm/kernel/signal.c | 37 +++++++++++++++++++++++++++++++++++++
 1 file changed, 37 insertions(+)
arch/arm/kernel/signal.c +37 -0
index a3a38d0a4c85..7ef453e8a96f 100644
@@ -725,3 +725,40 @@ asmlinkage void do_rseq_syscall(struct pt_regs *regs)
 	rseq_syscall(regs);
 }
 #endif
+
+/*
+ * Compile-time assertions for siginfo_t offsets. Check NSIG* as well, as
+ * changes likely come with new fields that should be added below.
+ */
+static_assert(NSIGILL	== 11);
+static_assert(NSIGFPE	== 15);
+static_assert(NSIGSEGV	== 9);
+static_assert(NSIGBUS	== 5);
+static_assert(NSIGTRAP	== 6);
+static_assert(NSIGCHLD	== 6);
+static_assert(NSIGSYS	== 2);
+static_assert(offsetof(siginfo_t, si_signo)	== 0x00);
+static_assert(offsetof(siginfo_t, si_errno)	== 0x04);
+static_assert(offsetof(siginfo_t, si_code)	== 0x08);
+static_assert(offsetof(siginfo_t, si_pid)	== 0x0c);
+static_assert(offsetof(siginfo_t, si_uid)	== 0x10);
+static_assert(offsetof(siginfo_t, si_tid)	== 0x0c);
+static_assert(offsetof(siginfo_t, si_overrun)	== 0x10);
+static_assert(offsetof(siginfo_t, si_status)	== 0x14);
+static_assert(offsetof(siginfo_t, si_utime)	== 0x18);
+static_assert(offsetof(siginfo_t, si_stime)	== 0x1c);
+static_assert(offsetof(siginfo_t, si_value)	== 0x14);
+static_assert(offsetof(siginfo_t, si_int)	== 0x14);
+static_assert(offsetof(siginfo_t, si_ptr)	== 0x14);
+static_assert(offsetof(siginfo_t, si_addr)	== 0x0c);
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x10);
+static_assert(offsetof(siginfo_t, si_lower)	== 0x14);
+static_assert(offsetof(siginfo_t, si_upper)	== 0x18);
+static_assert(offsetof(siginfo_t, si_pkey)	== 0x14);
+static_assert(offsetof(siginfo_t, si_perf_data)	== 0x10);
+static_assert(offsetof(siginfo_t, si_perf_type)	== 0x14);
+static_assert(offsetof(siginfo_t, si_band)	== 0x0c);
+static_assert(offsetof(siginfo_t, si_fd)	== 0x10);
+static_assert(offsetof(siginfo_t, si_call_addr)	== 0x0c);
+static_assert(offsetof(siginfo_t, si_syscall)	== 0x10);
+static_assert(offsetof(siginfo_t, si_arch)	== 0x14);
Signature · 1 line
2.20.1
#105 lore raw

PATCH 3/6 arm64: Add compile-time asserts for siginfo_t offsets

In reply to Eric W. Biederman

From: Marco Elver <[email protected]>

To help catch ABI breaks at compile-time, add compile-time assertions to verify the siginfo_t layout.

---
 arch/arm64/kernel/signal.c   | 37 ++++++++++++++++++++++++++++++++++++
 arch/arm64/kernel/signal32.c | 37 ++++++++++++++++++++++++++++++++++++
 2 files changed, 74 insertions(+)
arch/arm64/kernel/signal.c +37 -0
index f8192f4ae0b8..4413b6a4e32a 100644
@@ -999,3 +999,40 @@ void __init minsigstksz_setup(void)
 		round_up(sizeof(struct frame_record), 16) +
 		16; /* max alignment padding */
 }
+
+/*
+ * Compile-time assertions for siginfo_t offsets. Check NSIG* as well, as
+ * changes likely come with new fields that should be added below.
+ */
+static_assert(NSIGILL	== 11);
+static_assert(NSIGFPE	== 15);
+static_assert(NSIGSEGV	== 9);
+static_assert(NSIGBUS	== 5);
+static_assert(NSIGTRAP	== 6);
+static_assert(NSIGCHLD	== 6);
+static_assert(NSIGSYS	== 2);
+static_assert(offsetof(siginfo_t, si_signo)	== 0x00);
+static_assert(offsetof(siginfo_t, si_errno)	== 0x04);
+static_assert(offsetof(siginfo_t, si_code)	== 0x08);
+static_assert(offsetof(siginfo_t, si_pid)	== 0x10);
+static_assert(offsetof(siginfo_t, si_uid)	== 0x14);
+static_assert(offsetof(siginfo_t, si_tid)	== 0x10);
+static_assert(offsetof(siginfo_t, si_overrun)	== 0x14);
+static_assert(offsetof(siginfo_t, si_status)	== 0x18);
+static_assert(offsetof(siginfo_t, si_utime)	== 0x20);
+static_assert(offsetof(siginfo_t, si_stime)	== 0x28);
+static_assert(offsetof(siginfo_t, si_value)	== 0x18);
+static_assert(offsetof(siginfo_t, si_int)	== 0x18);
+static_assert(offsetof(siginfo_t, si_ptr)	== 0x18);
+static_assert(offsetof(siginfo_t, si_addr)	== 0x10);
+static_assert(offsetof(siginfo_t, si_addr_lsb)	== 0x18);
+static_assert(offsetof(siginfo_t, si_lower)	== 0x20);
+static_assert(offsetof(siginfo_t, si_upper)	== 0x28);
+static_assert(offsetof(siginfo_t, si_pkey)	== 0x20);
+static_assert(offsetof(siginfo_t, si_perf_data)	== 0x18);
+static_assert(offsetof(siginfo_t, si_perf_type)	== 0x20);
+static_assert(offsetof(siginfo_t, si_band)	== 0x10);
+static_assert(offsetof(siginfo_t, si_fd)	== 0x18);
+static_assert(offsetof(siginfo_t, si_call_addr)	== 0x10);
+static_assert(offsetof(siginfo_t, si_syscall)	== 0x18);
+static_assert(offsetof(siginfo_t, si_arch)	== 0x1c);
arch/arm64/kernel/signal32.c +37 -0
index 2f507f565c48..ab1775216712 100644
@@ -457,3 +457,40 @@ void compat_setup_restart_syscall(struct pt_regs *regs)
 {
        regs->regs[7] = __NR_compat_restart_syscall;
 }
+
+/*
+ * Compile-time assertions for siginfo_t offsets. Check NSIG* as well, as
+ * changes likely come with new fields that should be added below.
+ */
+static_assert(NSIGILL	== 11);
+static_assert(NSIGFPE	== 15);
+static_assert(NSIGSEGV	== 9);
+static_assert(NSIGBUS	== 5);
+static_assert(NSIGTRAP	== 6);
+static_assert(NSIGCHLD	== 6);
+static_assert(NSIGSYS	== 2);
+static_assert(offsetof(compat_siginfo_t, si_signo)	== 0x00);
+static_assert(offsetof(compat_siginfo_t, si_errno)	== 0x04);
+static_assert(offsetof(compat_siginfo_t, si_code)	== 0x08);
+static_assert(offsetof(compat_siginfo_t, si_pid)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_uid)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_tid)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_overrun)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_status)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_utime)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_stime)	== 0x1c);
+static_assert(offsetof(compat_siginfo_t, si_value)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_int)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_ptr)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_addr)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_addr_lsb)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_lower)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_upper)	== 0x18);
+static_assert(offsetof(compat_siginfo_t, si_pkey)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_perf_data)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_perf_type)	== 0x14);
+static_assert(offsetof(compat_siginfo_t, si_band)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_fd)		== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_call_addr)	== 0x0c);
+static_assert(offsetof(compat_siginfo_t, si_syscall)	== 0x10);
+static_assert(offsetof(compat_siginfo_t, si_arch)	== 0x14);
Signature · 1 line
2.20.1
#106 lore raw

PATCH 4/6 signal/sparc: si_trapno is only used with SIGILL ILL_ILLTRP

In reply to Eric W. Biederman

While reviewing the signal handlers on sparc it became clear that si_trapno is only set to a non-zero value when sending SIGILL with si_code ILL_ILLTRP.

Add force_sig_fault_trapno and send SIGILL ILL_ILLTRP with it.

Remove the define of __ARCH_SI_TRAPNO and remove the always zero si_trapno parameter from send_sig_fault and force_sig_fault.

v1: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]

---
 arch/sparc/include/uapi/asm/siginfo.h |  3 --
 arch/sparc/kernel/process_64.c        |  2 +-
 arch/sparc/kernel/sys_sparc_32.c      |  2 +-
 arch/sparc/kernel/sys_sparc_64.c      |  2 +-
 arch/sparc/kernel/traps_32.c          | 22 +++++++-------
 arch/sparc/kernel/traps_64.c          | 44 ++++++++++++---------------
 arch/sparc/kernel/unaligned_32.c      |  2 +-
 arch/sparc/mm/fault_32.c              |  2 +-
 arch/sparc/mm/fault_64.c              |  2 +-
 include/linux/sched/signal.h          |  1 +
 kernel/signal.c                       | 19 ++++++++++++
 11 files changed, 56 insertions(+), 45 deletions(-)
arch/sparc/include/uapi/asm/siginfo.h +0 -3
index 68bdde4c2a2e..0e7c27522aed 100644
@@ -8,9 +8,6 @@
 
 #endif /* defined(__sparc__) && defined(__arch64__) */
 
-
-#define __ARCH_SI_TRAPNO
-
 #include <asm-generic/siginfo.h>
 
 
arch/sparc/kernel/process_64.c +1 -1
index d33c58a58d4f..547b06b49ce3 100644
@@ -518,7 +518,7 @@ void synchronize_user_stack(void)
 
 static void stack_unaligned(unsigned long sp)
 {
-	force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) sp, 0);
+	force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) sp);
 }
 
 static const char uwfault32[] = KERN_INFO \
arch/sparc/kernel/sys_sparc_32.c +1 -1
index be77538bc038..082a551897ed 100644
@@ -151,7 +151,7 @@ sparc_breakpoint (struct pt_regs *regs)
 #ifdef DEBUG_SPARC_BREAKPOINT
         printk ("TRAP: Entering kernel PC=%x, nPC=%x\n", regs->pc, regs->npc);
 #endif
-	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc, 0);
+	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc);
 
 #ifdef DEBUG_SPARC_BREAKPOINT
 	printk ("TRAP: Returning to space: PC=%x nPC=%x\n", regs->pc, regs->npc);
arch/sparc/kernel/sys_sparc_64.c +1 -1
index 6b92fadb6ec7..1e9a9e016237 100644
@@ -514,7 +514,7 @@ asmlinkage void sparc_breakpoint(struct pt_regs *regs)
 #ifdef DEBUG_SPARC_BREAKPOINT
         printk ("TRAP: Entering kernel PC=%lx, nPC=%lx\n", regs->tpc, regs->tnpc);
 #endif
-	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->tpc, 0);
+	force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->tpc);
 #ifdef DEBUG_SPARC_BREAKPOINT
 	printk ("TRAP: Returning to space: PC=%lx nPC=%lx\n", regs->tpc, regs->tnpc);
 #endif
arch/sparc/kernel/traps_32.c +11 -11
index 247a0d9683b2..5630e5a395e0 100644
@@ -102,8 +102,8 @@ void do_hw_interrupt(struct pt_regs *regs, unsigned long type)
 	if(regs->psr & PSR_PS)
 		die_if_kernel("Kernel bad trap", regs);
 
-	force_sig_fault(SIGILL, ILL_ILLTRP,
-			(void __user *)regs->pc, type - 0x80);
+	force_sig_fault_trapno(SIGILL, ILL_ILLTRP,
+			       (void __user *)regs->pc, type - 0x80);
 }
 
 void do_illegal_instruction(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -116,7 +116,7 @@ void do_illegal_instruction(struct pt_regs *regs, unsigned long pc, unsigned lon
 	       regs->pc, *(unsigned long *)regs->pc);
 #endif
 
-	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc, current);
 }
 
 void do_priv_instruction(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -124,7 +124,7 @@ void do_priv_instruction(struct pt_regs *regs, unsigned long pc, unsigned long n
 {
 	if(psr & PSR_PS)
 		die_if_kernel("Penguin instruction from Penguin mode??!?!", regs);
-	send_sig_fault(SIGILL, ILL_PRVOPC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_PRVOPC, (void __user *)pc, current);
 }
 
 /* XXX User may want to be allowed to do this. XXX */
@@ -145,7 +145,7 @@ void do_memaccess_unaligned(struct pt_regs *regs, unsigned long pc, unsigned lon
 #endif
 	send_sig_fault(SIGBUS, BUS_ADRALN,
 		       /* FIXME: Should dig out mna address */ (void *)0,
-		       0, current);
+		       current);
 }
 
 static unsigned long init_fsr = 0x0UL;
@@ -291,7 +291,7 @@ void do_fpe_trap(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 		else if (fsr & 0x01)
 			code = FPE_FLTRES;
 	}
-	send_sig_fault(SIGFPE, code, (void __user *)pc, 0, fpt);
+	send_sig_fault(SIGFPE, code, (void __user *)pc, fpt);
 #ifndef CONFIG_SMP
 	last_task_used_math = NULL;
 #endif
@@ -305,7 +305,7 @@ void handle_tag_overflow(struct pt_regs *regs, unsigned long pc, unsigned long n
 {
 	if(psr & PSR_PS)
 		die_if_kernel("Penguin overflow trap from kernel mode", regs);
-	send_sig_fault(SIGEMT, EMT_TAGOVF, (void __user *)pc, 0, current);
+	send_sig_fault(SIGEMT, EMT_TAGOVF, (void __user *)pc, current);
 }
 
 void handle_watchpoint(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -327,13 +327,13 @@ void handle_reg_access(struct pt_regs *regs, unsigned long pc, unsigned long npc
 	printk("Register Access Exception at PC %08lx NPC %08lx PSR %08lx\n",
 	       pc, npc, psr);
 #endif
-	force_sig_fault(SIGBUS, BUS_OBJERR, (void __user *)pc, 0);
+	force_sig_fault(SIGBUS, BUS_OBJERR, (void __user *)pc);
 }
 
 void handle_cp_disabled(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 			unsigned long psr)
 {
-	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, current);
 }
 
 void handle_cp_exception(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -343,13 +343,13 @@ void handle_cp_exception(struct pt_regs *regs, unsigned long pc, unsigned long n
 	printk("Co-Processor Exception at PC %08lx NPC %08lx PSR %08lx\n",
 	       pc, npc, psr);
 #endif
-	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, 0, current);
+	send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, current);
 }
 
 void handle_hw_divzero(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 		       unsigned long psr)
 {
-	send_sig_fault(SIGFPE, FPE_INTDIV, (void __user *)pc, 0, current);
+	send_sig_fault(SIGFPE, FPE_INTDIV, (void __user *)pc, current);
 }
 
 #ifdef CONFIG_DEBUG_BUGVERBOSE
arch/sparc/kernel/traps_64.c +19 -25

144 lines not on this page · raw message

arch/sparc/kernel/unaligned_32.c +1 -1
index ef5c5207c9ff..455f0258c745 100644
@@ -278,5 +278,5 @@ asmlinkage void user_unaligned_trap(struct pt_regs *regs, unsigned int insn)
 {
 	send_sig_fault(SIGBUS, BUS_ADRALN,
 		       (void __user *)safe_compute_effective_address(regs, insn),
-		       0, current);
+		       current);
 }
arch/sparc/mm/fault_32.c +1 -1
index de2031c2b2d7..fa858626b85b 100644
@@ -83,7 +83,7 @@ static void __do_fault_siginfo(int code, int sig, struct pt_regs *regs,
 		show_signal_msg(regs, sig, code,
 				addr, current);
 
-	force_sig_fault(sig, code, (void __user *) addr, 0);
+	force_sig_fault(sig, code, (void __user *) addr);
 }
 
 static unsigned long compute_si_addr(struct pt_regs *regs, int text_fault)
arch/sparc/mm/fault_64.c +1 -1
index 0a6bcc85fba7..9a9652a15fed 100644
@@ -176,7 +176,7 @@ static void do_fault_siginfo(int code, int sig, struct pt_regs *regs,
 	if (unlikely(show_unhandled_signals))
 		show_signal_msg(regs, sig, code, addr, current);
 
-	force_sig_fault(sig, code, (void __user *) addr, 0);
+	force_sig_fault(sig, code, (void __user *) addr);
 }
 
 static unsigned int get_fault_insn(struct pt_regs *regs, unsigned int insn)
include/linux/sched/signal.h +1 -0
index b9126fe06c3f..99a9ab2b169a 100644
@@ -329,6 +329,7 @@ int force_sig_pkuerr(void __user *addr, u32 pkey);
 int force_sig_perf(void __user *addr, u32 type, u64 sig_data);
 
 int force_sig_ptrace_errno_trap(int errno, void __user *addr);
+int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno);
 
 extern int send_sig_info(int, struct kernel_siginfo *, struct task_struct *);
 extern void force_sigsegv(int sig);
kernel/signal.c +19 -0
index a3229add4455..87a374225277 100644
@@ -1808,6 +1808,22 @@ int force_sig_ptrace_errno_trap(int errno, void __user *addr)
 	return force_sig_info(&info);
 }
 
+/* For the rare architectures that include trap information using
+ * si_trapno.
+ */
+int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno)
+{
+	struct kernel_siginfo info;
+
+	clear_siginfo(&info);
+	info.si_signo = sig;
+	info.si_errno = 0;
+	info.si_code  = code;
+	info.si_addr  = addr;
+	info.si_trapno = trapno;
+	return force_sig_info(&info);
+}
+
 int kill_pgrp(struct pid *pid, int sig, int priv)
 {
 	int ret;
@@ -3243,6 +3259,9 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
 #endif
 			else if ((sig == SIGTRAP) && (si_code == TRAP_PERF))
 				layout = SIL_PERF_EVENT;
+			else if (IS_ENABLED(CONFIG_SPARC) &&
+				 (sig == SIGILL) && (si_code == ILL_ILLTRP))
+				layout = SIL_FAULT_TRAPNO;
 #ifdef __ARCH_SI_TRAPNO
 			else if (layout == SIL_FAULT)
 				layout = SIL_FAULT_TRAPNO;
Signature · 1 line
2.20.1
author Reviewed-by on v1 4/6
#107 lore raw

Re: [PATCH 4/6] signal/sparc: si_trapno is only used with SIGILL ILL_ILLTRP

Quoted 4 lines · Eric W. Biederman

While reviewing the signal handlers on sparc it became clear that si_trapno is only set to a non-zero value when sending SIGILL with si_code ILL_ILLTRP.

Quoted 8 lines

Add force_sig_fault_trapno and send SIGILL ILL_ILLTRP with it.

Remove the define of __ARCH_SI_TRAPNO and remove the always zero si_trapno parameter from send_sig_fault and force_sig_fault.

v1: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]

Quoted 416 lines
---
 arch/sparc/include/uapi/asm/siginfo.h |  3 --
 arch/sparc/kernel/process_64.c        |  2 +-
 arch/sparc/kernel/sys_sparc_32.c      |  2 +-
 arch/sparc/kernel/sys_sparc_64.c      |  2 +-
 arch/sparc/kernel/traps_32.c          | 22 +++++++-------
 arch/sparc/kernel/traps_64.c          | 44 ++++++++++++---------------
 arch/sparc/kernel/unaligned_32.c      |  2 +-
 arch/sparc/mm/fault_32.c              |  2 +-
 arch/sparc/mm/fault_64.c              |  2 +-
 include/linux/sched/signal.h          |  1 +
 kernel/signal.c                       | 19 ++++++++++++
 11 files changed, 56 insertions(+), 45 deletions(-)
arch/sparc/include/uapi/asm/siginfo.h +0 -3
index 68bdde4c2a2e..0e7c27522aed 100644
@@ -8,9 +8,6 @@
 
 #endif /* defined(__sparc__) && defined(__arch64__) */
 
-
-#define __ARCH_SI_TRAPNO
-
 #include <asm-generic/siginfo.h>
 
 
arch/sparc/kernel/process_64.c +1 -1
index d33c58a58d4f..547b06b49ce3 100644
@@ -518,7 +518,7 @@ void synchronize_user_stack(void)
 
 static void stack_unaligned(unsigned long sp)
 {
-       force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) sp, 0);
+       force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) sp);
 }
 
 static const char uwfault32[] = KERN_INFO \
arch/sparc/kernel/sys_sparc_32.c +1 -1
index be77538bc038..082a551897ed 100644
@@ -151,7 +151,7 @@ sparc_breakpoint (struct pt_regs *regs)
 #ifdef DEBUG_SPARC_BREAKPOINT
         printk ("TRAP: Entering kernel PC=%x, nPC=%x\n", regs->pc, regs->npc);
 #endif
-       force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc, 0);
+       force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc);
 
 #ifdef DEBUG_SPARC_BREAKPOINT
        printk ("TRAP: Returning to space: PC=%x nPC=%x\n", regs->pc, regs->npc);
arch/sparc/kernel/sys_sparc_64.c +1 -1
index 6b92fadb6ec7..1e9a9e016237 100644
@@ -514,7 +514,7 @@ asmlinkage void sparc_breakpoint(struct pt_regs *regs)
 #ifdef DEBUG_SPARC_BREAKPOINT
         printk ("TRAP: Entering kernel PC=%lx, nPC=%lx\n", regs->tpc, regs->tnpc);
 #endif
-       force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->tpc, 0);
+       force_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->tpc);
 #ifdef DEBUG_SPARC_BREAKPOINT
        printk ("TRAP: Returning to space: PC=%lx nPC=%lx\n", regs->tpc, regs->tnpc);
 #endif
arch/sparc/kernel/traps_32.c +11 -11
index 247a0d9683b2..5630e5a395e0 100644
@@ -102,8 +102,8 @@ void do_hw_interrupt(struct pt_regs *regs, unsigned long type)
        if(regs->psr & PSR_PS)
                die_if_kernel("Kernel bad trap", regs);
 
-       force_sig_fault(SIGILL, ILL_ILLTRP,
-                       (void __user *)regs->pc, type - 0x80);
+       force_sig_fault_trapno(SIGILL, ILL_ILLTRP,
+                              (void __user *)regs->pc, type - 0x80);
 }
 
 void do_illegal_instruction(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -116,7 +116,7 @@ void do_illegal_instruction(struct pt_regs *regs, unsigned long pc, unsigned lon
               regs->pc, *(unsigned long *)regs->pc);
 #endif
 
-       send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc, 0, current);
+       send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc, current);
 }
 
 void do_priv_instruction(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -124,7 +124,7 @@ void do_priv_instruction(struct pt_regs *regs, unsigned long pc, unsigned long n
 {
        if(psr & PSR_PS)
                die_if_kernel("Penguin instruction from Penguin mode??!?!", regs);
-       send_sig_fault(SIGILL, ILL_PRVOPC, (void __user *)pc, 0, current);
+       send_sig_fault(SIGILL, ILL_PRVOPC, (void __user *)pc, current);
 }
 
 /* XXX User may want to be allowed to do this. XXX */
@@ -145,7 +145,7 @@ void do_memaccess_unaligned(struct pt_regs *regs, unsigned long pc, unsigned lon
 #endif
        send_sig_fault(SIGBUS, BUS_ADRALN,
                       /* FIXME: Should dig out mna address */ (void *)0,
-                      0, current);
+                      current);
 }
 
 static unsigned long init_fsr = 0x0UL;
@@ -291,7 +291,7 @@ void do_fpe_trap(struct pt_regs *regs, unsigned long pc, unsigned long npc,
                else if (fsr & 0x01)
                        code = FPE_FLTRES;
        }
-       send_sig_fault(SIGFPE, code, (void __user *)pc, 0, fpt);
+       send_sig_fault(SIGFPE, code, (void __user *)pc, fpt);
 #ifndef CONFIG_SMP
        last_task_used_math = NULL;
 #endif
@@ -305,7 +305,7 @@ void handle_tag_overflow(struct pt_regs *regs, unsigned long pc, unsigned long n
 {
        if(psr & PSR_PS)
                die_if_kernel("Penguin overflow trap from kernel mode", regs);
-       send_sig_fault(SIGEMT, EMT_TAGOVF, (void __user *)pc, 0, current);
+       send_sig_fault(SIGEMT, EMT_TAGOVF, (void __user *)pc, current);
 }
 
 void handle_watchpoint(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -327,13 +327,13 @@ void handle_reg_access(struct pt_regs *regs, unsigned long pc, unsigned long npc
        printk("Register Access Exception at PC %08lx NPC %08lx PSR %08lx\n",
               pc, npc, psr);
 #endif
-       force_sig_fault(SIGBUS, BUS_OBJERR, (void __user *)pc, 0);
+       force_sig_fault(SIGBUS, BUS_OBJERR, (void __user *)pc);
 }
 
 void handle_cp_disabled(struct pt_regs *regs, unsigned long pc, unsigned long npc,
                        unsigned long psr)
 {
-       send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, 0, current);
+       send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, current);
 }
 
 void handle_cp_exception(struct pt_regs *regs, unsigned long pc, unsigned long npc,
@@ -343,13 +343,13 @@ void handle_cp_exception(struct pt_regs *regs, unsigned long pc, unsigned long n
        printk("Co-Processor Exception at PC %08lx NPC %08lx PSR %08lx\n",
               pc, npc, psr);
 #endif
-       send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, 0, current);
+       send_sig_fault(SIGILL, ILL_COPROC, (void __user *)pc, current);
 }
 
 void handle_hw_divzero(struct pt_regs *regs, unsigned long pc, unsigned long npc,
                       unsigned long psr)
 {
-       send_sig_fault(SIGFPE, FPE_INTDIV, (void __user *)pc, 0, current);
+       send_sig_fault(SIGFPE, FPE_INTDIV, (void __user *)pc, current);
 }
 
 #ifdef CONFIG_DEBUG_BUGVERBOSE
arch/sparc/kernel/traps_64.c +19 -25
index a850dccd78ea..6863025ed56d 100644
@@ -107,8 +107,8 @@ void bad_trap(struct pt_regs *regs, long lvl)
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGILL, ILL_ILLTRP,
-                       (void __user *)regs->tpc, lvl);
+       force_sig_fault_trapno(SIGILL, ILL_ILLTRP,
+                              (void __user *)regs->tpc, lvl);
 }
 
 void bad_trap_tl1(struct pt_regs *regs, long lvl)
@@ -201,8 +201,7 @@ void spitfire_insn_access_exception(struct pt_regs *regs, unsigned long sfsr, un
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGSEGV, SEGV_MAPERR,
-                       (void __user *)regs->tpc, 0);
+       force_sig_fault(SIGSEGV, SEGV_MAPERR, (void __user *)regs->tpc);
 out:
        exception_exit(prev_state);
 }
@@ -237,7 +236,7 @@ void sun4v_insn_access_exception(struct pt_regs *regs, unsigned long addr, unsig
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGSEGV, SEGV_MAPERR, (void __user *) addr, 0);
+       force_sig_fault(SIGSEGV, SEGV_MAPERR, (void __user *) addr);
 }
 
 void sun4v_insn_access_exception_tl1(struct pt_regs *regs, unsigned long addr, unsigned long type_ctx)
@@ -321,7 +320,7 @@ void spitfire_data_access_exception(struct pt_regs *regs, unsigned long sfsr, un
        if (is_no_fault_exception(regs))
                return;
 
-       force_sig_fault(SIGSEGV, SEGV_MAPERR, (void __user *)sfar, 0);
+       force_sig_fault(SIGSEGV, SEGV_MAPERR, (void __user *)sfar);
 out:
        exception_exit(prev_state);
 }
@@ -385,13 +384,13 @@ void sun4v_data_access_exception(struct pt_regs *regs, unsigned long addr, unsig
         */
        switch (type) {
        case HV_FAULT_TYPE_INV_ASI:
-               force_sig_fault(SIGILL, ILL_ILLADR, (void __user *)addr, 0);
+               force_sig_fault(SIGILL, ILL_ILLADR, (void __user *)addr);
                break;
        case HV_FAULT_TYPE_MCD_DIS:
-               force_sig_fault(SIGSEGV, SEGV_ACCADI, (void __user *)addr, 0);
+               force_sig_fault(SIGSEGV, SEGV_ACCADI, (void __user *)addr);
                break;
        default:
-               force_sig_fault(SIGSEGV, SEGV_MAPERR, (void __user *)addr, 0);
+               force_sig_fault(SIGSEGV, SEGV_MAPERR, (void __user *)addr);
                break;
        }
 }
@@ -568,7 +567,7 @@ static void spitfire_ue_log(unsigned long afsr, unsigned long afar, unsigned lon
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGBUS, BUS_OBJERR, (void *)0, 0);
+       force_sig_fault(SIGBUS, BUS_OBJERR, (void *)0);
 }
 
 void spitfire_access_error(struct pt_regs *regs, unsigned long status_encoded, unsigned long afar)
@@ -2069,8 +2068,7 @@ void do_mcd_err(struct pt_regs *regs, struct sun4v_error_entry ent)
        /* Send SIGSEGV to the userspace process with the right signal
         * code
         */
-       force_sig_fault(SIGSEGV, SEGV_ADIDERR, (void __user *)ent.err_raddr,
-                       0);
+       force_sig_fault(SIGSEGV, SEGV_ADIDERR, (void __user *)ent.err_raddr);
 }
 
 /* We run with %pil set to PIL_NORMAL_MAX and PSTATE_IE enabled in %pstate.
@@ -2184,7 +2182,7 @@ bool sun4v_nonresum_error_user_handled(struct pt_regs *regs,
        }
        if (attrs & SUN4V_ERR_ATTRS_PIO) {
                force_sig_fault(SIGBUS, BUS_ADRERR,
-                               (void __user *)sun4v_get_vaddr(regs), 0);
+                               (void __user *)sun4v_get_vaddr(regs));
                return true;
        }
 
@@ -2340,8 +2338,7 @@ static void do_fpe_common(struct pt_regs *regs)
                        else if (fsr & 0x01)
                                code = FPE_FLTRES;
                }
-               force_sig_fault(SIGFPE, code,
-                               (void __user *)regs->tpc, 0);
+               force_sig_fault(SIGFPE, code, (void __user *)regs->tpc);
        }
 }
 
@@ -2395,8 +2392,7 @@ void do_tof(struct pt_regs *regs)
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGEMT, EMT_TAGOVF,
-                       (void __user *)regs->tpc, 0);
+       force_sig_fault(SIGEMT, EMT_TAGOVF, (void __user *)regs->tpc);
 out:
        exception_exit(prev_state);
 }
@@ -2415,8 +2411,7 @@ void do_div0(struct pt_regs *regs)
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGFPE, FPE_INTDIV,
-                       (void __user *)regs->tpc, 0);
+       force_sig_fault(SIGFPE, FPE_INTDIV, (void __user *)regs->tpc);
 out:
        exception_exit(prev_state);
 }
@@ -2612,7 +2607,7 @@ void do_illegal_instruction(struct pt_regs *regs)
                        }
                }
        }
-       force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc, 0);
+       force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)pc);
 out:
        exception_exit(prev_state);
 }
@@ -2632,7 +2627,7 @@ void mem_address_unaligned(struct pt_regs *regs, unsigned long sfar, unsigned lo
        if (is_no_fault_exception(regs))
                return;
 
-       force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *)sfar, 0);
+       force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *)sfar);
 out:
        exception_exit(prev_state);
 }
@@ -2650,7 +2645,7 @@ void sun4v_do_mna(struct pt_regs *regs, unsigned long addr, unsigned long type_c
        if (is_no_fault_exception(regs))
                return;
 
-       force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) addr, 0);
+       force_sig_fault(SIGBUS, BUS_ADRALN, (void __user *) addr);
 }
 
 /* sun4v_mem_corrupt_detect_precise() - Handle precise exception on an ADI
@@ -2697,7 +2692,7 @@ void sun4v_mem_corrupt_detect_precise(struct pt_regs *regs, unsigned long addr,
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGSEGV, SEGV_ADIPERR, (void __user *)addr, 0);
+       force_sig_fault(SIGSEGV, SEGV_ADIPERR, (void __user *)addr);
 }
 
 void do_privop(struct pt_regs *regs)
@@ -2712,8 +2707,7 @@ void do_privop(struct pt_regs *regs)
                regs->tpc &= 0xffffffff;
                regs->tnpc &= 0xffffffff;
        }
-       force_sig_fault(SIGILL, ILL_PRVOPC,
-                       (void __user *)regs->tpc, 0);
+       force_sig_fault(SIGILL, ILL_PRVOPC, (void __user *)regs->tpc);
 out:
        exception_exit(prev_state);
 }
arch/sparc/kernel/unaligned_32.c +1 -1
index ef5c5207c9ff..455f0258c745 100644
@@ -278,5 +278,5 @@ asmlinkage void user_unaligned_trap(struct pt_regs *regs, unsigned int insn)
 {
        send_sig_fault(SIGBUS, BUS_ADRALN,
                       (void __user *)safe_compute_effective_address(regs, insn),
-                      0, current);
+                      current);
 }
arch/sparc/mm/fault_32.c +1 -1
index de2031c2b2d7..fa858626b85b 100644
@@ -83,7 +83,7 @@ static void __do_fault_siginfo(int code, int sig, struct pt_regs *regs,
                show_signal_msg(regs, sig, code,
                                addr, current);
 
-       force_sig_fault(sig, code, (void __user *) addr, 0);
+       force_sig_fault(sig, code, (void __user *) addr);
 }
 
 static unsigned long compute_si_addr(struct pt_regs *regs, int text_fault)
arch/sparc/mm/fault_64.c +1 -1
index 0a6bcc85fba7..9a9652a15fed 100644
@@ -176,7 +176,7 @@ static void do_fault_siginfo(int code, int sig, struct pt_regs *regs,
        if (unlikely(show_unhandled_signals))
                show_signal_msg(regs, sig, code, addr, current);
 
-       force_sig_fault(sig, code, (void __user *) addr, 0);
+       force_sig_fault(sig, code, (void __user *) addr);
 }
 
 static unsigned int get_fault_insn(struct pt_regs *regs, unsigned int insn)
include/linux/sched/signal.h +1 -0
index b9126fe06c3f..99a9ab2b169a 100644
@@ -329,6 +329,7 @@ int force_sig_pkuerr(void __user *addr, u32 pkey);
 int force_sig_perf(void __user *addr, u32 type, u64 sig_data);
 
 int force_sig_ptrace_errno_trap(int errno, void __user *addr);
+int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno);
 
 extern int send_sig_info(int, struct kernel_siginfo *, struct task_struct *);
 extern void force_sigsegv(int sig);
kernel/signal.c +19 -1
index a3229add4455..87a374225277 100644
@@ -1808,6 +1808,22 @@ int force_sig_ptrace_errno_trap(int errno, void __user *addr)
        return force_sig_info(&info);
 }
 
+/* For the rare architectures that include trap information using
+ * si_trapno.
+ */
+int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno)
+{
+       struct kernel_siginfo info;
+
+       clear_siginfo(&info);
+       info.si_signo = sig;
+       info.si_errno = 0;
+       info.si_code  = code;
+       info.si_addr  = addr;
+       info.si_trapno = trapno;
+       return force_sig_info(&info);
+}
+
 int kill_pgrp(struct pid *pid, int sig, int priv)
 {
        int ret;
@@ -3243,6 +3259,9 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
 #endif
                        else if ((sig == SIGTRAP) && (si_code == TRAP_PERF))
                                layout = SIL_PERF_EVENT;
+                       else if (IS_ENABLED(CONFIG_SPARC) &&
+                                (sig == SIGILL) && (si_code == ILL_ILLTRP))
+                               layout = SIL_FAULT_TRAPNO;
 #ifdef __ARCH_SI_TRAPNO
                        else if (layout == SIL_FAULT)
                                layout = SIL_FAULT_TRAPNO;
--

2.20.1

--
You received this message because you are subscribed to the Google Groups "kasan-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/kasan-dev/87mtqnxx89.fsf_-_%40disp2133.

#108 lore raw

PATCH 5/6 signal/alpha: si_trapno is only used with SIGFPE and SIGTRAP TRAP_UNK

In reply to Eric W. Biederman

While reviewing the signal handlers on alpha it became clear that si_trapno is only set to a non-zero value when sending SIGFPE and when sending SITGRAP with si_code TRAP_UNK.

Add send_sig_fault_trapno and send SIGTRAP TRAP_UNK, and SIGFPE with it.

Remove the define of __ARCH_SI_TRAPNO and remove the always zero si_trapno parameter from send_sig_fault and force_sig_fault.

v1: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]

---
 arch/alpha/include/uapi/asm/siginfo.h |  2 --
 arch/alpha/kernel/osf_sys.c           |  2 +-
 arch/alpha/kernel/signal.c            |  4 ++--
 arch/alpha/kernel/traps.c             | 26 +++++++++++++-------------
 arch/alpha/mm/fault.c                 |  4 ++--
 include/linux/sched/signal.h          |  2 ++
 kernel/signal.c                       | 21 +++++++++++++++++++++
 7 files changed, 41 insertions(+), 20 deletions(-)
arch/alpha/include/uapi/asm/siginfo.h +0 -2
index 6e1a2af2f962..e08eae88182b 100644
@@ -2,8 +2,6 @@
 #ifndef _ALPHA_SIGINFO_H
 #define _ALPHA_SIGINFO_H
 
-#define __ARCH_SI_TRAPNO
-
 #include <asm-generic/siginfo.h>
 
 #endif
arch/alpha/kernel/osf_sys.c +1 -1
index d5367a1c6300..bbdb1a9a5fd8 100644
@@ -876,7 +876,7 @@ SYSCALL_DEFINE5(osf_setsysinfo, unsigned long, op, void __user *, buffer,
 			if (fex & IEEE_TRAP_ENABLE_DZE) si_code = FPE_FLTDIV;
 			if (fex & IEEE_TRAP_ENABLE_INV) si_code = FPE_FLTINV;
 
-			send_sig_fault(SIGFPE, si_code,
+			send_sig_fault_trapno(SIGFPE, si_code,
 				       (void __user *)NULL,  /* FIXME */
 				       0, current);
  		}
arch/alpha/kernel/signal.c +2 -2
index 948b89789da8..bc077babafab 100644
@@ -219,7 +219,7 @@ do_sigreturn(struct sigcontext __user *sc)
 
 	/* Send SIGTRAP if we're single-stepping: */
 	if (ptrace_cancel_bpt (current)) {
-		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc, 0,
+		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc,
 			       current);
 	}
 	return;
@@ -247,7 +247,7 @@ do_rt_sigreturn(struct rt_sigframe __user *frame)
 
 	/* Send SIGTRAP if we're single-stepping: */
 	if (ptrace_cancel_bpt (current)) {
-		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc, 0,
+		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc,
 			       current);
 	}
 	return;
arch/alpha/kernel/traps.c +13 -13
index 921d4b6e4d95..e9e3de18793b 100644
@@ -227,7 +227,7 @@ do_entArith(unsigned long summary, unsigned long write_mask,
 	}
 	die_if_kernel("Arithmetic fault", regs, 0, NULL);
 
-	send_sig_fault(SIGFPE, si_code, (void __user *) regs->pc, 0, current);
+	send_sig_fault_trapno(SIGFPE, si_code, (void __user *) regs->pc, 0, current);
 }
 
 asmlinkage void
@@ -268,13 +268,13 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 			regs->pc -= 4;	/* make pc point to former bpt */
 		}
 
-		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc, 0,
+		send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc,
 			       current);
 		return;
 
 	      case 1: /* bugcheck */
-		send_sig_fault(SIGTRAP, TRAP_UNK, (void __user *) regs->pc, 0,
-			       current);
+		send_sig_fault_trapno(SIGTRAP, TRAP_UNK,
+				      (void __user *) regs->pc, 0, current);
 		return;
 		
 	      case 2: /* gentrap */
@@ -335,8 +335,8 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 			break;
 		}
 
-		send_sig_fault(signo, code, (void __user *) regs->pc, regs->r16,
-			       current);
+		send_sig_fault_trapno(signo, code, (void __user *) regs->pc,
+				      regs->r16, current);
 		return;
 
 	      case 4: /* opDEC */
@@ -360,9 +360,9 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 			if (si_code == 0)
 				return;
 			if (si_code > 0) {
-				send_sig_fault(SIGFPE, si_code,
-					       (void __user *) regs->pc, 0,
-					       current);
+				send_sig_fault_trapno(SIGFPE, si_code,
+						      (void __user *) regs->pc,
+						      0, current);
 				return;
 			}
 		}
@@ -387,7 +387,7 @@ do_entIF(unsigned long type, struct pt_regs *regs)
 		      ;
 	}
 
-	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, 0, current);
+	send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, current);
 }
 
 /* There is an ifdef in the PALcode in MILO that enables a 
@@ -402,7 +402,7 @@ do_entDbg(struct pt_regs *regs)
 {
 	die_if_kernel("Instruction fault", regs, 0, NULL);
 
-	force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, 0);
+	force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc);
 }
 
 
@@ -964,12 +964,12 @@ do_entUnaUser(void __user * va, unsigned long opcode,
 			si_code = SEGV_MAPERR;
 		mmap_read_unlock(mm);
 	}
-	send_sig_fault(SIGSEGV, si_code, va, 0, current);
+	send_sig_fault(SIGSEGV, si_code, va, current);
 	return;
 
 give_sigbus:
 	regs->pc -= 4;
-	send_sig_fault(SIGBUS, BUS_ADRALN, va, 0, current);
+	send_sig_fault(SIGBUS, BUS_ADRALN, va, current);
 	return;
 }
 
arch/alpha/mm/fault.c +2 -2
index 09172f017efc..eee5102c3d88 100644
@@ -219,13 +219,13 @@ do_page_fault(unsigned long address, unsigned long mmcsr,
 	mmap_read_unlock(mm);
 	/* Send a sigbus, regardless of whether we were in kernel
 	   or user mode.  */
-	force_sig_fault(SIGBUS, BUS_ADRERR, (void __user *) address, 0);
+	force_sig_fault(SIGBUS, BUS_ADRERR, (void __user *) address);
 	if (!user_mode(regs))
 		goto no_context;
 	return;
 
  do_sigsegv:
-	force_sig_fault(SIGSEGV, si_code, (void __user *) address, 0);
+	force_sig_fault(SIGSEGV, si_code, (void __user *) address);
 	return;
 
 #ifdef CONFIG_ALPHA_LARGE_VMALLOC
include/linux/sched/signal.h +2 -0
index 99a9ab2b169a..6657184cef07 100644
@@ -330,6 +330,8 @@ int force_sig_perf(void __user *addr, u32 type, u64 sig_data);
 
 int force_sig_ptrace_errno_trap(int errno, void __user *addr);
 int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno);
+int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
+			struct task_struct *t);
 
 extern int send_sig_info(int, struct kernel_siginfo *, struct task_struct *);
 extern void force_sigsegv(int sig);
kernel/signal.c +21 -0
index 87a374225277..ae06a424aa72 100644
@@ -1824,6 +1824,23 @@ int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno)
 	return force_sig_info(&info);
 }
 
+/* For the rare architectures that include trap information using
+ * si_trapno.
+ */
+int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
+			  struct task_struct *t)
+{
+	struct kernel_siginfo info;
+
+	clear_siginfo(&info);
+	info.si_signo = sig;
+	info.si_errno = 0;
+	info.si_code  = code;
+	info.si_addr  = addr;
+	info.si_trapno = trapno;
+	return send_sig_info(info.si_signo, &info, t);
+}
+
 int kill_pgrp(struct pid *pid, int sig, int priv)
 {
 	int ret;
@@ -3262,6 +3279,10 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
 			else if (IS_ENABLED(CONFIG_SPARC) &&
 				 (sig == SIGILL) && (si_code == ILL_ILLTRP))
 				layout = SIL_FAULT_TRAPNO;
+			else if (IS_ENABLED(CONFIG_ALPHA) &&
+				 ((sig == SIGFPE) ||
+				  ((sig == SIGTRAP) && (si_code == TRAP_UNK))))
+				layout = SIL_FAULT_TRAPNO;
 #ifdef __ARCH_SI_TRAPNO
 			else if (layout == SIL_FAULT)
 				layout = SIL_FAULT_TRAPNO;
Signature · 1 line
2.20.1
author Reviewed-by on v1 5/6
#109 lore raw

Re: [PATCH 5/6] signal/alpha: si_trapno is only used with SIGFPE and SIGTRAP TRAP_UNK

Quoted 4 lines · Eric W. Biederman

While reviewing the signal handlers on alpha it became clear that si_trapno is only set to a non-zero value when sending SIGFPE and when sending SITGRAP with si_code TRAP_UNK.

Quoted 8 lines

Add send_sig_fault_trapno and send SIGTRAP TRAP_UNK, and SIGFPE with it.

Remove the define of __ARCH_SI_TRAPNO and remove the always zero si_trapno parameter from send_sig_fault and force_sig_fault.

v1: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]

Quoted 220 lines
---
 arch/alpha/include/uapi/asm/siginfo.h |  2 --
 arch/alpha/kernel/osf_sys.c           |  2 +-
 arch/alpha/kernel/signal.c            |  4 ++--
 arch/alpha/kernel/traps.c             | 26 +++++++++++++-------------
 arch/alpha/mm/fault.c                 |  4 ++--
 include/linux/sched/signal.h          |  2 ++
 kernel/signal.c                       | 21 +++++++++++++++++++++
 7 files changed, 41 insertions(+), 20 deletions(-)
arch/alpha/include/uapi/asm/siginfo.h +0 -2
index 6e1a2af2f962..e08eae88182b 100644
@@ -2,8 +2,6 @@
 #ifndef _ALPHA_SIGINFO_H
 #define _ALPHA_SIGINFO_H
 
-#define __ARCH_SI_TRAPNO
-
 #include <asm-generic/siginfo.h>
 
 #endif
arch/alpha/kernel/osf_sys.c +1 -1
index d5367a1c6300..bbdb1a9a5fd8 100644
@@ -876,7 +876,7 @@ SYSCALL_DEFINE5(osf_setsysinfo, unsigned long, op, void __user *, buffer,
                        if (fex & IEEE_TRAP_ENABLE_DZE) si_code = FPE_FLTDIV;
                        if (fex & IEEE_TRAP_ENABLE_INV) si_code = FPE_FLTINV;
 
-                       send_sig_fault(SIGFPE, si_code,
+                       send_sig_fault_trapno(SIGFPE, si_code,
                                       (void __user *)NULL,  /* FIXME */
                                       0, current);
                }
arch/alpha/kernel/signal.c +2 -2
index 948b89789da8..bc077babafab 100644
@@ -219,7 +219,7 @@ do_sigreturn(struct sigcontext __user *sc)
 
        /* Send SIGTRAP if we're single-stepping: */
        if (ptrace_cancel_bpt (current)) {
-               send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc, 0,
+               send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc,
                               current);
        }
        return;
@@ -247,7 +247,7 @@ do_rt_sigreturn(struct rt_sigframe __user *frame)
 
        /* Send SIGTRAP if we're single-stepping: */
        if (ptrace_cancel_bpt (current)) {
-               send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc, 0,
+               send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *) regs->pc,
                               current);
        }
        return;
arch/alpha/kernel/traps.c +13 -13
index 921d4b6e4d95..e9e3de18793b 100644
@@ -227,7 +227,7 @@ do_entArith(unsigned long summary, unsigned long write_mask,
        }
        die_if_kernel("Arithmetic fault", regs, 0, NULL);
 
-       send_sig_fault(SIGFPE, si_code, (void __user *) regs->pc, 0, current);
+       send_sig_fault_trapno(SIGFPE, si_code, (void __user *) regs->pc, 0, current);
 }
 
 asmlinkage void
@@ -268,13 +268,13 @@ do_entIF(unsigned long type, struct pt_regs *regs)
                        regs->pc -= 4;  /* make pc point to former bpt */
                }
 
-               send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc, 0,
+               send_sig_fault(SIGTRAP, TRAP_BRKPT, (void __user *)regs->pc,
                               current);
                return;
 
              case 1: /* bugcheck */
-               send_sig_fault(SIGTRAP, TRAP_UNK, (void __user *) regs->pc, 0,
-                              current);
+               send_sig_fault_trapno(SIGTRAP, TRAP_UNK,
+                                     (void __user *) regs->pc, 0, current);
                return;
 
              case 2: /* gentrap */
@@ -335,8 +335,8 @@ do_entIF(unsigned long type, struct pt_regs *regs)
                        break;
                }
 
-               send_sig_fault(signo, code, (void __user *) regs->pc, regs->r16,
-                              current);
+               send_sig_fault_trapno(signo, code, (void __user *) regs->pc,
+                                     regs->r16, current);
                return;
 
              case 4: /* opDEC */
@@ -360,9 +360,9 @@ do_entIF(unsigned long type, struct pt_regs *regs)
                        if (si_code == 0)
                                return;
                        if (si_code > 0) {
-                               send_sig_fault(SIGFPE, si_code,
-                                              (void __user *) regs->pc, 0,
-                                              current);
+                               send_sig_fault_trapno(SIGFPE, si_code,
+                                                     (void __user *) regs->pc,
+                                                     0, current);
                                return;
                        }
                }
@@ -387,7 +387,7 @@ do_entIF(unsigned long type, struct pt_regs *regs)
                      ;
        }
 
-       send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, 0, current);
+       send_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, current);
 }
 
 /* There is an ifdef in the PALcode in MILO that enables a
@@ -402,7 +402,7 @@ do_entDbg(struct pt_regs *regs)
 {
        die_if_kernel("Instruction fault", regs, 0, NULL);
 
-       force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc, 0);
+       force_sig_fault(SIGILL, ILL_ILLOPC, (void __user *)regs->pc);
 }
 
 
@@ -964,12 +964,12 @@ do_entUnaUser(void __user * va, unsigned long opcode,
                        si_code = SEGV_MAPERR;
                mmap_read_unlock(mm);
        }
-       send_sig_fault(SIGSEGV, si_code, va, 0, current);
+       send_sig_fault(SIGSEGV, si_code, va, current);
        return;
 
 give_sigbus:
        regs->pc -= 4;
-       send_sig_fault(SIGBUS, BUS_ADRALN, va, 0, current);
+       send_sig_fault(SIGBUS, BUS_ADRALN, va, current);
        return;
 }
 
arch/alpha/mm/fault.c +2 -2
index 09172f017efc..eee5102c3d88 100644
@@ -219,13 +219,13 @@ do_page_fault(unsigned long address, unsigned long mmcsr,
        mmap_read_unlock(mm);
        /* Send a sigbus, regardless of whether we were in kernel
           or user mode.  */
-       force_sig_fault(SIGBUS, BUS_ADRERR, (void __user *) address, 0);
+       force_sig_fault(SIGBUS, BUS_ADRERR, (void __user *) address);
        if (!user_mode(regs))
                goto no_context;
        return;
 
  do_sigsegv:
-       force_sig_fault(SIGSEGV, si_code, (void __user *) address, 0);
+       force_sig_fault(SIGSEGV, si_code, (void __user *) address);
        return;
 
 #ifdef CONFIG_ALPHA_LARGE_VMALLOC
include/linux/sched/signal.h +2 -0
index 99a9ab2b169a..6657184cef07 100644
@@ -330,6 +330,8 @@ int force_sig_perf(void __user *addr, u32 type, u64 sig_data);
 
 int force_sig_ptrace_errno_trap(int errno, void __user *addr);
 int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno);
+int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
+                       struct task_struct *t);
 
 extern int send_sig_info(int, struct kernel_siginfo *, struct task_struct *);
 extern void force_sigsegv(int sig);
kernel/signal.c +21 -1
index 87a374225277..ae06a424aa72 100644
@@ -1824,6 +1824,23 @@ int force_sig_fault_trapno(int sig, int code, void __user *addr, int trapno)
        return force_sig_info(&info);
 }
 
+/* For the rare architectures that include trap information using
+ * si_trapno.
+ */
+int send_sig_fault_trapno(int sig, int code, void __user *addr, int trapno,
+                         struct task_struct *t)
+{
+       struct kernel_siginfo info;
+
+       clear_siginfo(&info);
+       info.si_signo = sig;
+       info.si_errno = 0;
+       info.si_code  = code;
+       info.si_addr  = addr;
+       info.si_trapno = trapno;
+       return send_sig_info(info.si_signo, &info, t);
+}
+
 int kill_pgrp(struct pid *pid, int sig, int priv)
 {
        int ret;
@@ -3262,6 +3279,10 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
                        else if (IS_ENABLED(CONFIG_SPARC) &&
                                 (sig == SIGILL) && (si_code == ILL_ILLTRP))
                                layout = SIL_FAULT_TRAPNO;
+                       else if (IS_ENABLED(CONFIG_ALPHA) &&
+                                ((sig == SIGFPE) ||
+                                 ((sig == SIGTRAP) && (si_code == TRAP_UNK))))
+                               layout = SIL_FAULT_TRAPNO;
 #ifdef __ARCH_SI_TRAPNO
                        else if (layout == SIL_FAULT)
                                layout = SIL_FAULT_TRAPNO;
--

2.20.1

#110 lore raw

PATCH 6/6 signal: Remove the generic __ARCH_SI_TRAPNO support

In reply to Eric W. Biederman

Now that __ARCH_SI_TRAPNO is no longer set by any architecture remove all of the code it enabled from the kernel.

On alpha and sparc a more explict approach of using send_sig_fault_trapno or force_sig_fault_trapno in the very limited circumstances where si_trapno was set to a non-zero value.

The generic support that is being removed always set si_trapno on all fault signals. With only SIGILL ILL_ILLTRAP on sparc and SIGFPE and SIGTRAP TRAP_UNK on alpla providing si_trapno values asking all senders of fault signals to provide an si_trapno value does not make sense.

Making si_trapno an ordinary extension of the fault siginfo layout has enabled the architecture generic implementation of SIGTRAP TRAP_PERF, and enables other faulting signals to grow architecture generic senders as well.

v1: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]

---
 arch/mips/include/uapi/asm/siginfo.h |  2 --
 include/linux/sched/signal.h         |  8 --------
 kernel/signal.c                      | 14 --------------
 3 files changed, 24 deletions(-)
arch/mips/include/uapi/asm/siginfo.h +0 -2
index c34c7eef0a1c..8cb8bd061a68 100644
@@ -10,9 +10,7 @@
 #ifndef _UAPI_ASM_SIGINFO_H
 #define _UAPI_ASM_SIGINFO_H
 
-
 #define __ARCH_SIGEV_PREAMBLE_SIZE (sizeof(long) + 2*sizeof(int))
-#undef __ARCH_SI_TRAPNO /* exception code needs to fill this ...  */
 
 #define __ARCH_HAS_SWAPPED_SIGINFO
 
include/linux/sched/signal.h +0 -8
index 6657184cef07..928e0025d358 100644
@@ -298,11 +298,6 @@ static inline void kernel_signal_stop(void)
 
 	schedule();
 }
-#ifdef __ARCH_SI_TRAPNO
-# define ___ARCH_SI_TRAPNO(_a1) , _a1
-#else
-# define ___ARCH_SI_TRAPNO(_a1)
-#endif
 #ifdef __ia64__
 # define ___ARCH_SI_IA64(_a1, _a2, _a3) , _a1, _a2, _a3
 #else
@@ -310,14 +305,11 @@ static inline void kernel_signal_stop(void)
 #endif
 
 int force_sig_fault_to_task(int sig, int code, void __user *addr
-	___ARCH_SI_TRAPNO(int trapno)
 	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
 	, struct task_struct *t);
 int force_sig_fault(int sig, int code, void __user *addr
-	___ARCH_SI_TRAPNO(int trapno)
 	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr));
 int send_sig_fault(int sig, int code, void __user *addr
-	___ARCH_SI_TRAPNO(int trapno)
 	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
 	, struct task_struct *t);
 
kernel/signal.c +0 -14
index ae06a424aa72..2181423e562a 100644
@@ -1666,7 +1666,6 @@ void force_sigsegv(int sig)
 }
 
 int force_sig_fault_to_task(int sig, int code, void __user *addr
-	___ARCH_SI_TRAPNO(int trapno)
 	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
 	, struct task_struct *t)
 {
@@ -1677,9 +1676,6 @@ int force_sig_fault_to_task(int sig, int code, void __user *addr
 	info.si_errno = 0;
 	info.si_code  = code;
 	info.si_addr  = addr;
-#ifdef __ARCH_SI_TRAPNO
-	info.si_trapno = trapno;
-#endif
 #ifdef __ia64__
 	info.si_imm = imm;
 	info.si_flags = flags;
@@ -1689,16 +1685,13 @@ int force_sig_fault_to_task(int sig, int code, void __user *addr
 }
 
 int force_sig_fault(int sig, int code, void __user *addr
-	___ARCH_SI_TRAPNO(int trapno)
 	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr))
 {
 	return force_sig_fault_to_task(sig, code, addr
-				       ___ARCH_SI_TRAPNO(trapno)
 				       ___ARCH_SI_IA64(imm, flags, isr), current);
 }
 
 int send_sig_fault(int sig, int code, void __user *addr
-	___ARCH_SI_TRAPNO(int trapno)
 	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
 	, struct task_struct *t)
 {
@@ -1709,9 +1702,6 @@ int send_sig_fault(int sig, int code, void __user *addr
 	info.si_errno = 0;
 	info.si_code  = code;
 	info.si_addr  = addr;
-#ifdef __ARCH_SI_TRAPNO
-	info.si_trapno = trapno;
-#endif
 #ifdef __ia64__
 	info.si_imm = imm;
 	info.si_flags = flags;
@@ -3283,10 +3273,6 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
 				 ((sig == SIGFPE) ||
 				  ((sig == SIGTRAP) && (si_code == TRAP_UNK))))
 				layout = SIL_FAULT_TRAPNO;
-#ifdef __ARCH_SI_TRAPNO
-			else if (layout == SIL_FAULT)
-				layout = SIL_FAULT_TRAPNO;
-#endif
 		}
 		else if (si_code <= NSIGPOLL)
 			layout = SIL_POLL;
Signature · 1 line
2.20.1
author Reviewed-by on v1 6/6
#111 lore raw

Re: [PATCH 6/6] signal: Remove the generic __ARCH_SI_TRAPNO support

Quoted 12 lines · Eric W. Biederman

Now that __ARCH_SI_TRAPNO is no longer set by any architecture remove all of the code it enabled from the kernel.

On alpha and sparc a more explict approach of using send_sig_fault_trapno or force_sig_fault_trapno in the very limited circumstances where si_trapno was set to a non-zero value.

The generic support that is being removed always set si_trapno on all fault signals. With only SIGILL ILL_ILLTRAP on sparc and SIGFPE and SIGTRAP TRAP_UNK on alpla providing si_trapno values asking all senders of fault signals to provide an si_trapno value does not make sense.

Quoted 8 lines

Making si_trapno an ordinary extension of the fault siginfo layout has enabled the architecture generic implementation of SIGTRAP TRAP_PERF, and enables other faulting signals to grow architecture generic senders as well.

v1: https://lkml.kernel.org/r/[email protected]
v2: https://lkml.kernel.org/r/[email protected]

Quoted 114 lines
---
 arch/mips/include/uapi/asm/siginfo.h |  2 --
 include/linux/sched/signal.h         |  8 --------
 kernel/signal.c                      | 14 --------------
 3 files changed, 24 deletions(-)
arch/mips/include/uapi/asm/siginfo.h +0 -2
index c34c7eef0a1c..8cb8bd061a68 100644
@@ -10,9 +10,7 @@
 #ifndef _UAPI_ASM_SIGINFO_H
 #define _UAPI_ASM_SIGINFO_H
 
-
 #define __ARCH_SIGEV_PREAMBLE_SIZE (sizeof(long) + 2*sizeof(int))
-#undef __ARCH_SI_TRAPNO /* exception code needs to fill this ...  */
 
 #define __ARCH_HAS_SWAPPED_SIGINFO
 
include/linux/sched/signal.h +0 -8
index 6657184cef07..928e0025d358 100644
@@ -298,11 +298,6 @@ static inline void kernel_signal_stop(void)
 
        schedule();
 }
-#ifdef __ARCH_SI_TRAPNO
-# define ___ARCH_SI_TRAPNO(_a1) , _a1
-#else
-# define ___ARCH_SI_TRAPNO(_a1)
-#endif
 #ifdef __ia64__
 # define ___ARCH_SI_IA64(_a1, _a2, _a3) , _a1, _a2, _a3
 #else
@@ -310,14 +305,11 @@ static inline void kernel_signal_stop(void)
 #endif
 
 int force_sig_fault_to_task(int sig, int code, void __user *addr
-       ___ARCH_SI_TRAPNO(int trapno)
        ___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
        , struct task_struct *t);
 int force_sig_fault(int sig, int code, void __user *addr
-       ___ARCH_SI_TRAPNO(int trapno)
        ___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr));
 int send_sig_fault(int sig, int code, void __user *addr
-       ___ARCH_SI_TRAPNO(int trapno)
        ___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
        , struct task_struct *t);
 
kernel/signal.c +0 -15
index ae06a424aa72..2181423e562a 100644
@@ -1666,7 +1666,6 @@ void force_sigsegv(int sig)
 }
 
 int force_sig_fault_to_task(int sig, int code, void __user *addr
-       ___ARCH_SI_TRAPNO(int trapno)
        ___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
        , struct task_struct *t)
 {
@@ -1677,9 +1676,6 @@ int force_sig_fault_to_task(int sig, int code, void __user *addr
        info.si_errno = 0;
        info.si_code  = code;
        info.si_addr  = addr;
-#ifdef __ARCH_SI_TRAPNO
-       info.si_trapno = trapno;
-#endif
 #ifdef __ia64__
        info.si_imm = imm;
        info.si_flags = flags;
@@ -1689,16 +1685,13 @@ int force_sig_fault_to_task(int sig, int code, void __user *addr
 }
 
 int force_sig_fault(int sig, int code, void __user *addr
-       ___ARCH_SI_TRAPNO(int trapno)
        ___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr))
 {
        return force_sig_fault_to_task(sig, code, addr
-                                      ___ARCH_SI_TRAPNO(trapno)
                                       ___ARCH_SI_IA64(imm, flags, isr), current);
 }
 
 int send_sig_fault(int sig, int code, void __user *addr
-       ___ARCH_SI_TRAPNO(int trapno)
        ___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
        , struct task_struct *t)
 {
@@ -1709,9 +1702,6 @@ int send_sig_fault(int sig, int code, void __user *addr
        info.si_errno = 0;
        info.si_code  = code;
        info.si_addr  = addr;
-#ifdef __ARCH_SI_TRAPNO
-       info.si_trapno = trapno;
-#endif
 #ifdef __ia64__
        info.si_imm = imm;
        info.si_flags = flags;
@@ -3283,10 +3273,6 @@ enum siginfo_layout siginfo_layout(unsigned sig, int si_code)
                                 ((sig == SIGFPE) ||
                                  ((sig == SIGTRAP) && (si_code == TRAP_UNK))))
                                layout = SIL_FAULT_TRAPNO;
-#ifdef __ARCH_SI_TRAPNO
-                       else if (layout == SIL_FAULT)
-                               layout = SIL_FAULT_TRAPNO;
-#endif
                }
                else if (si_code <= NSIGPOLL)
                        layout = SIL_POLL;
--

2.20.1

#112 lore raw

Re: [PATCH 0/6] Final si_trapno bits

In reply to Eric W. Biederman
Quoted 32 lines · Eric W. Biederman

As a part of a fix for the ABI of the newly added SIGTRAP TRAP_PERF a si_trapno was reduced to an ordinary extention of the _sigfault case of struct siginfo.

When Linus saw the complete set of changes come in as a fix he requested that the set of changes be trimmed down to just what was necessary to fix the SIGTRAP TRAP_PERF ABI.

I had intended to get the rest of the changes into the merge window for v5.14 but I dropped the ball.

I have made the changes to stop using __ARCH_SI_TRAPNO be per architecture so they are easier to review. In doing so I found one place on alpha where I used send_sig_fault instead of send_sig_fault_trapno(... si_trapno = 0). That would not have changed the userspace behavior but it did make the kernel code less clear.

My rule in these patches is everywhere that siginfo layout calls for SIL_FAULT_TRAPNO the code uses either force_sig_fault_trapno or send_sig_fault_trapno.

And of course I have rebased and compile tested Marco's compile time assert patches.

Eric

Eric W. Biederman (3):
signal/sparc: si_trapno is only used with SIGILL ILL_ILLTRP signal/alpha: si_trapno is only used with SIGFPE and SIGTRAP TRAP_UNK
signal: Remove the generic __ARCH_SI_TRAPNO support

Quoted 4 lines

Marco Elver (3):
sparc64: Add compile-time asserts for siginfo_t offsets
arm: Add compile-time asserts for siginfo_t offsets
arm64: Add compile-time asserts for siginfo_t offsets

Nice, thanks for the respin. If I diffed it right, I see this is almost (modulo what you mentioned above) equivalent to:
https://lore.kernel.org/linux-api/[email protected]/
+ what's already in mainline. It's only missing:

signal: Verify the alignment and size of siginfo_t
signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency

Would this be appropriate for this series, or rather separately, or dropped completely?

Thanks,
-- Marco

#113 lore raw

Re: [PATCH 0/6] Final si_trapno bits

Quoted 43 lines · Marco Elver
Quoted 32 lines · Eric W. Biederman

As a part of a fix for the ABI of the newly added SIGTRAP TRAP_PERF a si_trapno was reduced to an ordinary extention of the _sigfault case of struct siginfo.

When Linus saw the complete set of changes come in as a fix he requested that the set of changes be trimmed down to just what was necessary to fix the SIGTRAP TRAP_PERF ABI.

I had intended to get the rest of the changes into the merge window for v5.14 but I dropped the ball.

I have made the changes to stop using __ARCH_SI_TRAPNO be per architecture so they are easier to review. In doing so I found one place on alpha where I used send_sig_fault instead of send_sig_fault_trapno(... si_trapno = 0). That would not have changed the userspace behavior but it did make the kernel code less clear.

My rule in these patches is everywhere that siginfo layout calls for SIL_FAULT_TRAPNO the code uses either force_sig_fault_trapno or send_sig_fault_trapno.

And of course I have rebased and compile tested Marco's compile time assert patches.

Eric

Eric W. Biederman (3):
signal/sparc: si_trapno is only used with SIGILL ILL_ILLTRP signal/alpha: si_trapno is only used with SIGFPE and SIGTRAP TRAP_UNK
signal: Remove the generic __ARCH_SI_TRAPNO support

Quoted 4 lines

Marco Elver (3):
sparc64: Add compile-time asserts for siginfo_t offsets
arm: Add compile-time asserts for siginfo_t offsets
arm64: Add compile-time asserts for siginfo_t offsets

Nice, thanks for the respin. If I diffed it right, I see this is almost (modulo what you mentioned above) equivalent to:
https://lore.kernel.org/linux-api/[email protected]/
+ what's already in mainline. It's only missing:

Quoted 5 lines

signal: Verify the alignment and size of siginfo_t
signal: Rename SIL_PERF_EVENT SIL_FAULT_PERF_EVENT for consistency

Would this be appropriate for this series, or rather separately, or dropped completely?

Appropriate I just overlooked them.

Eric