arm64: expose FAR_EL1 tag bits in siginfo

Most architectures with the exception of alpha, mips, parisc and sparc use the same values for these flags. Move their definitions into asm-generic/signal-defs.h and allow the architectures with non-standard values to override them. Also, document the non-standard flag values in order to make it easier to add new generic flags in the future.

Signed-off-by Peter Collingbourne <[email protected]>

 arch/m68k/include/uapi/asm/signal.h    | 24 -------------

include/uapi/asm-generic/signal-defs.h +0 -0

@@ -4,6 +4,53 @@
 
 #include <linux/compiler.h>

include/uapi/asm-generic/signal-defs.h +15 -0

+ */
+#ifndef SA_NOCLDSTOP
+#define SA_NOCLDSTOP   0x00000001
+#endif
+#ifndef SA_NOCLDWAIT
+#define SA_NOCLDWAIT   0x00000002
+#endif
+#ifndef SA_SIGINFO
+#define SA_SIGINFO     0x00000004
+#endif
+#ifndef SA_ONSTACK
+#define SA_ONSTACK     0x08000000
+#endif
+#ifndef SA_RESTART
+#define SA_RESTART     0x10000000

Hi Peter,

Quoted 7 lines · Peter Collingbourne

Most architectures with the exception of alpha, mips, parisc and sparc use the same values for these flags. Move their definitions into asm-generic/signal-defs.h and allow the architectures with non-standard values to override them. Also, document the non-standard flag values in order to make it easier to add new generic flags in the future.

Signed-off-by Peter Collingbourne <[email protected]>

Thanks for your patch!

Quoted 1 line

 arch/m68k/include/uapi/asm/signal.h    | 24 -------------

Acked-by Geert Uytterhoeven <[email protected]>

Quoted 6 lines

include/uapi/asm-generic/signal-defs.h +0 -0

@@ -4,6 +4,53 @@
 
 #include <linux/compiler.h>
 

include/uapi/asm-generic/signal-defs.h line 15

Quoted 9 lines

include/uapi/asm-generic/signal-defs.h +9 -0

+/*
+ * SA_FLAGS values:
+ *
+ * SA_ONSTACK indicates that a registered stack_t will be used.
+ * SA_RESTART flag to get restarting signals (which were the default long ago)
+ * SA_NOCLDSTOP flag to turn off SIGCHLD when children stop.
+ * SA_RESETHAND clears the handler when the signal is delivered.
+ * SA_NOCLDWAIT flag on SIGCHLD to inhibit zombies.
+ * SA_NODEFER prevents the current signal from being masked in the handler.

Perhaps the order should match the order of the definitions below? Then it becomes obvious SA_SIGINFO is not documented (it was on arm).

include/uapi/asm-generic/signal-defs.h line 18

Quoted 3 lines

include/uapi/asm-generic/signal-defs.h +3 -0

+ *
+ * SA_ONESHOT and SA_NOMASK are the historical Linux names for the Single
+ * Unix names RESETHAND and NODEFER respectively.

I think moving this last comment below, just above the definitions of SA_NOMASK and SA_ONESHOT, would make this easier to read.

Quoted 19 lines

include/uapi/asm-generic/signal-defs.h +19 -0

+ */
+#ifndef SA_NOCLDSTOP
+#define SA_NOCLDSTOP   0x00000001
+#endif
+#ifndef SA_NOCLDWAIT
+#define SA_NOCLDWAIT   0x00000002
+#endif
+#ifndef SA_SIGINFO
+#define SA_SIGINFO     0x00000004
+#endif
+#ifndef SA_ONSTACK
+#define SA_ONSTACK     0x08000000
+#endif
+#ifndef SA_RESTART
+#define SA_RESTART     0x10000000
+#endif
+#ifndef SA_NODEFER
+#define SA_NODEFER     0x40000000
+#endif

include/uapi/asm-generic/signal-defs.h line 43

Quoted 6 lines

include/uapi/asm-generic/signal-defs.h +6 -0

+#ifndef SA_RESETHAND
+#define SA_RESETHAND   0x80000000
+#endif
+/*
+ * The following bits are used in architecture-specific SA_* definitions and
+ * should be avoided for new generic flags: 3, 4, 5, 6, 7, 8, 9, 16, 24, 25, 26.

And I think this belongs to the previous comment block.

Most architectures with the exception of alpha, mips, parisc and sparc use the same values for these flags. Move their definitions into asm-generic/signal-defs.h and allow the architectures with non-standard values to override them. Also, document the non-standard flag values in order to make it easier to add new generic flags in the future.

Signed-off-by Peter Collingbourne <[email protected]>

---
View this change in Gerrit: https://linux-review.googlesource.com/q/Ia3849f18b8009bf41faca374e701cdca36974528

 arch/alpha/include/uapi/asm/signal.h   | 14 --------
 arch/arm/include/uapi/asm/signal.h     | 28 ++-------------
 arch/h8300/include/uapi/asm/signal.h   | 24 -------------
 arch/ia64/include/uapi/asm/signal.h    | 24 -------------
 arch/m68k/include/uapi/asm/signal.h    | 24 -------------
 arch/mips/include/uapi/asm/signal.h    | 12 -------
 arch/parisc/include/uapi/asm/signal.h  | 13 -------
 arch/powerpc/include/uapi/asm/signal.h | 24 -------------
 arch/s390/include/uapi/asm/signal.h    | 24 -------------
 arch/sparc/include/uapi/asm/signal.h   |  4 +--
 arch/x86/include/uapi/asm/signal.h     | 24 -------------
 arch/xtensa/include/uapi/asm/signal.h  | 24 -------------
 include/uapi/asm-generic/signal-defs.h | 47 ++++++++++++++++++++++++++
 include/uapi/asm-generic/signal.h      | 29 ----------------
 14 files changed, 51 insertions(+), 264 deletions(-)

arch/alpha/include/uapi/asm/signal.h +0 -14

index 74c750bf1c1a..a69dd8d080a8 100644
@@ -60,20 +60,6 @@ typedef unsigned long sigset_t;
 #define SIGRTMIN	32
 #define SIGRTMAX	_NSIG
 
-/*
- * SA_FLAGS values:
- *
- * SA_ONSTACK indicates that a registered stack_t will be used.
- * SA_RESTART flag to get restarting signals (which were the default long ago)
- * SA_NOCLDSTOP flag to turn off SIGCHLD when children stop.
- * SA_RESETHAND clears the handler when the signal is delivered.
- * SA_NOCLDWAIT flag on SIGCHLD to inhibit zombies.
- * SA_NODEFER prevents the current signal from being masked in the handler.
- *
- * SA_ONESHOT and SA_NOMASK are the historical Linux names for the Single
- * Unix names RESETHAND and NODEFER respectively.
- */
-
 #define SA_ONSTACK	0x00000001
 #define SA_RESTART	0x00000002
 #define SA_NOCLDSTOP	0x00000004

arch/arm/include/uapi/asm/signal.h +0 -12

index 9b4185ba4f8a..7727f0984d26 100644
@@ -60,33 +60,11 @@ typedef unsigned long sigset_t;
 #define SIGSWI		32
 
 /*
- * SA_FLAGS values:
- *
- * SA_NOCLDSTOP		flag to turn off SIGCHLD when children stop.
- * SA_NOCLDWAIT		flag on SIGCHLD to inhibit zombies.
- * SA_SIGINFO		deliver the signal with SIGINFO structs
- * SA_THIRTYTWO		delivers the signal in 32-bit mode, even if the task 
- *			is running in 26-bit.
- * SA_ONSTACK		allows alternate signal stacks (see sigaltstack(2)).
- * SA_RESTART		flag to get restarting signals (which were the default long ago)
- * SA_NODEFER		prevents the current signal from being masked in the handler.
- * SA_RESETHAND		clears the handler when the signal is delivered.
- *

Quoted 10 lines · Peter Collingbourne

Most architectures with the exception of alpha, mips, parisc and sparc use the same values for these flags. Move their definitions into asm-generic/signal-defs.h and allow the architectures with non-standard values to override them. Also, document the non-standard flag values in order to make it easier to add new generic flags in the future.

Signed-off-by Peter Collingbourne <[email protected]>

---
View this change in Gerrit: https://linux-review.googlesource.com/q/Ia3849f18b8009bf41faca374e701cdca36974528

Quoted 15 lines

 arch/alpha/include/uapi/asm/signal.h   | 14 --------
 arch/arm/include/uapi/asm/signal.h     | 28 ++-------------
 arch/h8300/include/uapi/asm/signal.h   | 24 -------------
 arch/ia64/include/uapi/asm/signal.h    | 24 -------------
 arch/m68k/include/uapi/asm/signal.h    | 24 -------------
 arch/mips/include/uapi/asm/signal.h    | 12 -------
 arch/parisc/include/uapi/asm/signal.h  | 13 -------
 arch/powerpc/include/uapi/asm/signal.h | 24 -------------
 arch/s390/include/uapi/asm/signal.h    | 24 -------------
 arch/sparc/include/uapi/asm/signal.h   |  4 +--
 arch/x86/include/uapi/asm/signal.h     | 24 -------------
 arch/xtensa/include/uapi/asm/signal.h  | 24 -------------
 include/uapi/asm-generic/signal-defs.h | 47 ++++++++++++++++++++++++++
 include/uapi/asm-generic/signal.h      | 29 ----------------
 14 files changed, 51 insertions(+), 264 deletions(-)

Nice diffstat!

Quoted 51 lines

arch/alpha/include/uapi/asm/signal.h +0 -14

index 74c750bf1c1a..a69dd8d080a8 100644
@@ -60,20 +60,6 @@ typedef unsigned long sigset_t;
 #define SIGRTMIN     32
 #define SIGRTMAX     _NSIG
 
-/*
- * SA_FLAGS values:
- *
- * SA_ONSTACK indicates that a registered stack_t will be used.
- * SA_RESTART flag to get restarting signals (which were the default long ago)
- * SA_NOCLDSTOP flag to turn off SIGCHLD when children stop.
- * SA_RESETHAND clears the handler when the signal is delivered.
- * SA_NOCLDWAIT flag on SIGCHLD to inhibit zombies.
- * SA_NODEFER prevents the current signal from being masked in the handler.
- *
- * SA_ONESHOT and SA_NOMASK are the historical Linux names for the Single
- * Unix names RESETHAND and NODEFER respectively.
- */
-
 #define SA_ONSTACK   0x00000001
 #define SA_RESTART   0x00000002
 #define SA_NOCLDSTOP 0x00000004

arch/arm/include/uapi/asm/signal.h +3 -14

index 9b4185ba4f8a..7727f0984d26 100644
@@ -60,33 +60,11 @@ typedef unsigned long sigset_t;
 #define SIGSWI               32
 
 /*
- * SA_FLAGS values:
- *
- * SA_NOCLDSTOP              flag to turn off SIGCHLD when children stop.
- * SA_NOCLDWAIT              flag on SIGCHLD to inhibit zombies.
- * SA_SIGINFO                deliver the signal with SIGINFO structs
- * SA_THIRTYTWO              delivers the signal in 32-bit mode, even if the task
- *                   is running in 26-bit.
- * SA_ONSTACK                allows alternate signal stacks (see sigaltstack(2)).
- * SA_RESTART                flag to get restarting signals (which were the default long ago)
- * SA_NODEFER                prevents the current signal from being masked in the handler.
- * SA_RESETHAND              clears the handler when the signal is delivered.
- *
- * SA_ONESHOT and SA_NOMASK are the historical Linux names for the Single
- * Unix names RESETHAND and NODEFER respectively.
+ * SA_THIRTYTWO historically meant deliver the signal in 32-bit mode, even if
+ * the task is running in 26-bit. But since the kernel no longer supports
+ * 26-bit mode, the flag has no effect.

arch/arm/include/uapi/asm/signal.h typedef unsigned long sigset_t; · line 67

Quoted 5 lines

arch/arm/include/uapi/asm/signal.h +0 -3

  */
-#define SA_NOCLDSTOP 0x00000001
-#define SA_NOCLDWAIT 0x00000002
-#define SA_SIGINFO   0x00000004
 #define SA_THIRTYTWO 0x02000000

Can we add a placeholder for this in the common header? We don't want people accidentally defining a new generic flag that clashes with this.

This allows userspace to detect missing support for flag bits and allows the kernel to use non-uapi bits internally, as we are already doing in arch/x86 for two flag bits. Now that this change is in place, we no longer need the code in arch/x86 that was hiding these bits from userspace, so remove it.

Signed-off-by Peter Collingbourne <[email protected]>

---
View this change in Gerrit: https://linux-review.googlesource.com/q/I35aab6f5be932505d90f3b3450c083b4db1eca86

 arch/arm/include/asm/signal.h    |  4 ++++
 arch/parisc/include/asm/signal.h |  4 ++++
 arch/x86/kernel/signal_compat.c  |  7 -------
 include/linux/signal_types.h     | 12 ++++++++++++
 kernel/signal.c                  | 10 ++++++++++
 5 files changed, 30 insertions(+), 7 deletions(-)

arch/arm/include/asm/signal.h +0 -0

index 65530a042009..d1070a783993 100644

arch/arm/include/asm/signal.h +0 -0

 #define __ARCH_HAS_SA_RESTORER
 
 #include <asm/sigcontext.h>

arch/parisc/include/asm/signal.h +4 -0

index 715c96ba2ec8..ad06e14f6e8a 100644
@@ -21,6 +21,10 @@ typedef struct {
 	unsigned long sig[_NSIG_WORDS];
 } sigset_t;
 
+#define SA_UAPI_FLAGS                                                          \
+	(SA_ONSTACK | SA_RESETHAND | SA_NOCLDSTOP | SA_SIGINFO | SA_NODEFER |  \
+	 SA_RESTART | SA_NOCLDWAIT | _SA_SIGGFAULT)
+
 #include <asm/sigcontext.h>
 
 #endif /* !__ASSEMBLY */

arch/x86/kernel/signal_compat.c +0 -7

index 9ccbf0576cd0..c599013ae8cb 100644
@@ -165,16 +165,9 @@ void sigaction_compat_abi(struct k_sigaction *act, struct k_sigaction *oact)
 {
 	signal_compat_build_tests();
 
-	/* Don't leak in-kernel non-uapi flags to user-space */
-	if (oact)
-		oact->sa.sa_flags &= ~(SA_IA32_ABI | SA_X32_ABI);
-
 	if (!act)
 		return;
 
-	/* Don't let flags to be set from userspace */
-	act->sa.sa_flags &= ~(SA_IA32_ABI | SA_X32_ABI);
-
 	if (in_ia32_syscall())
 		act->sa.sa_flags |= SA_IA32_ABI;
 	if (in_x32_syscall())

include/linux/signal_types.h +12 -0

index f8a90ae9c6ec..e792f29b5846 100644
@@ -68,4 +68,16 @@ struct ksignal {
 	int sig;
 };
 
+#ifndef SA_UAPI_FLAGS
+#ifdef SA_RESTORER
+#define SA_UAPI_FLAGS                                                          \
+	(SA_NOCLDSTOP | SA_NOCLDWAIT | SA_SIGINFO | SA_ONSTACK | SA_RESTART |  \
+	 SA_NODEFER | SA_RESETHAND | SA_RESTORER)
+#else
+#define SA_UAPI_FLAGS                                                          \
+	(SA_NOCLDSTOP | SA_NOCLDWAIT | SA_SIGINFO | SA_ONSTACK | SA_RESTART |  \
+	 SA_NODEFER | SA_RESETHAND)
+#endif
+#endif
+
 #endif /* _LINUX_SIGNAL_TYPES_H */

kernel/signal.c +0 -0

index 42b67d2cea37..348b7981f1ff 100644
@@ -3984,6 +3984,16 @@ int do_sigaction(int sig, struct k_sigaction *act, struct k_sigaction *oact)
 	if (oact)
 		*oact = *k;
 

kernel/signal.c +0 -0

 	sigaction_compat_abi(act, oact);
 
 	if (act) {

Signature · 7 lines

2.28.0.220.ged08abb693-goog


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

On Mon, Aug 17, 2020 at 08:33:48PM -0700, Peter Collingbourne wrote:

Nit: please say what the patch does. Subject line should summarise what is done, but should not add new information that is not present in the description proper.

(Same for all the other patches.)

Quoted 22 lines

This allows userspace to detect missing support for flag bits and allows the kernel to use non-uapi bits internally, as we are already doing in arch/x86 for two flag bits. Now that this change is in place, we no longer need the code in arch/x86 that was hiding these bits from userspace, so remove it.

Signed-off-by Peter Collingbourne <[email protected]>

---
View this change in Gerrit: https://linux-review.googlesource.com/q/I35aab6f5be932505d90f3b3450c083b4db1eca86

 arch/arm/include/asm/signal.h    |  4 ++++
 arch/parisc/include/asm/signal.h |  4 ++++
 arch/x86/kernel/signal_compat.c  |  7 -------
 include/linux/signal_types.h     | 12 ++++++++++++
 kernel/signal.c                  | 10 ++++++++++
 5 files changed, 30 insertions(+), 7 deletions(-)

arch/arm/include/asm/signal.h +0 -0

index 65530a042009..d1070a783993 100644

arch/arm/include/asm/signal.h typedef struct { · line 23

Quoted 7 lines

arch/arm/include/asm/signal.h +4 -0

@@ -17,6 +17,10 @@ typedef struct {
      unsigned long sig[_NSIG_WORDS];
 } sigset_t;
 
+#define SA_UAPI_FLAGS                                                          \
+     (SA_NOCLDSTOP | SA_NOCLDWAIT | SA_SIGINFO | SA_THIRTYTWO |             \
+      SA_RESTORER | SA_ONSTACK | SA_RESTART | SA_NODEFER | SA_RESETHAND)
+

I wonder whether all these per-arch definitions will tend to bitrot when people add new common flags.

Can we have a common definition for the common bits, and just add the extra arch-specific ones here?

Also, I wonder whether we should avoid the "SA_" prefix here. Maybe UAPI_SA_FLAGS?

Quoted 69 lines

arch/arm/include/asm/signal.h +0 -0

 #define __ARCH_HAS_SA_RESTORER
 
 #include <asm/sigcontext.h>

arch/parisc/include/asm/signal.h +4 -0

index 715c96ba2ec8..ad06e14f6e8a 100644
@@ -21,6 +21,10 @@ typedef struct {
      unsigned long sig[_NSIG_WORDS];
 } sigset_t;
 
+#define SA_UAPI_FLAGS                                                          \
+     (SA_ONSTACK | SA_RESETHAND | SA_NOCLDSTOP | SA_SIGINFO | SA_NODEFER |  \
+      SA_RESTART | SA_NOCLDWAIT | _SA_SIGGFAULT)
+
 #include <asm/sigcontext.h>
 
 #endif /* !__ASSEMBLY */

arch/x86/kernel/signal_compat.c +0 -7

index 9ccbf0576cd0..c599013ae8cb 100644
@@ -165,16 +165,9 @@ void sigaction_compat_abi(struct k_sigaction *act, struct k_sigaction *oact)
 {
      signal_compat_build_tests();
 
-     /* Don't leak in-kernel non-uapi flags to user-space */
-     if (oact)
-             oact->sa.sa_flags &= ~(SA_IA32_ABI | SA_X32_ABI);
-
      if (!act)
              return;
 
-     /* Don't let flags to be set from userspace */
-     act->sa.sa_flags &= ~(SA_IA32_ABI | SA_X32_ABI);
-
      if (in_ia32_syscall())
              act->sa.sa_flags |= SA_IA32_ABI;
      if (in_x32_syscall())

include/linux/signal_types.h +12 -0

index f8a90ae9c6ec..e792f29b5846 100644
@@ -68,4 +68,16 @@ struct ksignal {
      int sig;
 };
 
+#ifndef SA_UAPI_FLAGS
+#ifdef SA_RESTORER
+#define SA_UAPI_FLAGS                                                          \
+     (SA_NOCLDSTOP | SA_NOCLDWAIT | SA_SIGINFO | SA_ONSTACK | SA_RESTART |  \
+      SA_NODEFER | SA_RESETHAND | SA_RESTORER)
+#else
+#define SA_UAPI_FLAGS                                                          \
+     (SA_NOCLDSTOP | SA_NOCLDWAIT | SA_SIGINFO | SA_ONSTACK | SA_RESTART |  \
+      SA_NODEFER | SA_RESETHAND)
+#endif
+#endif
+
 #endif /* _LINUX_SIGNAL_TYPES_H */

kernel/signal.c +0 -0

index 42b67d2cea37..348b7981f1ff 100644
@@ -3984,6 +3984,16 @@ int do_sigaction(int sig, struct k_sigaction *act, struct k_sigaction *oact)
      if (oact)
              *oact = *k;
 

kernel/signal.c int do_sigaction(int sig, struct k_sigaction *act, struct k_sigaction *oact) · line 3996

Quoted 10 lines

kernel/signal.c +10 -0

+     /*
+      * Clear unknown flag bits in order to allow userspace to detect missing
+      * support for flag bits and to allow the kernel to use non-uapi bits
+      * internally.
+      */
+     if (act)
+             act->sa.sa_flags &= SA_UAPI_FLAGS;
+     if (oact)
+             oact->sa.sa_flags &= SA_UAPI_FLAGS;
+

Seems reasonable.

Quoted 8 lines · Dave Martin

On Mon, Aug 17, 2020 at 08:33:48PM -0700, Peter Collingbourne wrote:

Nit: please say what the patch does. Subject line should summarise what is done, but should not add new information that is not present in the description proper.

(Same for all the other patches.)

Will do.

Quoted 16 lines

arch/arm/include/asm/signal.h typedef struct { · line 23

Quoted 12 lines

arch/arm/include/asm/signal.h +4 -0

index 65530a042009..d1070a783993 100644
@@ -17,6 +17,10 @@ typedef struct {
      unsigned long sig[_NSIG_WORDS];
 } sigset_t;
 
+#define SA_UAPI_FLAGS                                                          \
+     (SA_NOCLDSTOP | SA_NOCLDWAIT | SA_SIGINFO | SA_THIRTYTWO |             \
+      SA_RESTORER | SA_ONSTACK | SA_RESTART | SA_NODEFER | SA_RESETHAND)
+

I wonder whether all these per-arch definitions will tend to bitrot when people add new common flags.

Quoted 2 lines

Can we have a common definition for the common bits, and just add the extra arch-specific ones here?

I think so. We could have something like:

#define ARCH_UAPI_SA_FLAGS SA_THIRTYTWO

here. Then in signal_types.h we can do:

#ifndef ARCH_UAPI_SA_FLAGS
#define ARCH_UAPI_SA_FLAGS 0
#endif

#define UAPI_SA_FLAGS (... | ARCH_UAPI_SA_FLAGS)

I'll do that in v10.

Quoted 2 lines

Also, I wonder whether we should avoid the "SA_" prefix here. Maybe UAPI_SA_FLAGS?

Sounds good to me.

Quoted 8 lines

Quoted 8 lines

kernel/signal.c +0 -0

index 42b67d2cea37..348b7981f1ff 100644
@@ -3984,6 +3984,16 @@ int do_sigaction(int sig, struct k_sigaction *act, struct k_sigaction *oact)
      if (oact)
              *oact = *k;
 

Quoted 12 lines

kernel/signal.c int do_sigaction(int sig, struct k_sigaction *act, struct k_sigaction *oact) · line 3996

Quoted 10 lines

kernel/signal.c +10 -0

+     /*
+      * Clear unknown flag bits in order to allow userspace to detect missing
+      * support for flag bits and to allow the kernel to use non-uapi bits
+      * internally.
+      */
+     if (act)
+             act->sa.sa_flags &= SA_UAPI_FLAGS;
+     if (oact)
+             oact->sa.sa_flags &= SA_UAPI_FLAGS;
+

Seems reasonable.

Thanks. I also decided to check how other operating systems handle unknown flag bits in sigaction.sa_flags. It looks like OpenBSD and illumos also accept unknown bits but (implicitly, as a result of using a different internal representation) end up clearing them in oldact:

https://github.com/openbsd/src/blob/f634a6a4b5bf832e9c1de77f7894ae2625e74484/sys/kern/kern_sig.c#L278
https://github.com/illumos/illumos-gate/blob/76f19f5fdc974fe5be5c82a556e43a4df93f1de1/usr/src/uts/common/syscall/sigaction.c#L86

and FreeBSD and NetBSD fail the syscall if unknown bits are set:

https://github.com/freebsd/freebsd/blob/eded70c37057857c6e23fae51f86b8f8f43cd2d0/sys/kern/kern_sig.c#L699
https://github.com/NetBSD/src/blob/3365779becdcedfca206091a645a0e8e22b2946e/sys/kern/sys_sig.c#L473

So there is some precedent for doing what we're planning to do here, which makes it yet more likely that we'll be okay doing this.

Quoted 11 lines · Peter Collingbourne

Quoted 8 lines · Dave Martin

On Mon, Aug 17, 2020 at 08:33:48PM -0700, Peter Collingbourne wrote:

Nit: please say what the patch does. Subject line should summarise what is done, but should not add new information that is not present in the description proper.

(Same for all the other patches.)

Will do.

Thanks

[...]

Quoted 25 lines

Quoted 16 lines

arch/arm/include/asm/signal.h typedef struct { · line 23

Quoted 12 lines

arch/arm/include/asm/signal.h +4 -0

index 65530a042009..d1070a783993 100644
@@ -17,6 +17,10 @@ typedef struct {
      unsigned long sig[_NSIG_WORDS];
 } sigset_t;
 
+#define SA_UAPI_FLAGS                                                          \
+     (SA_NOCLDSTOP | SA_NOCLDWAIT | SA_SIGINFO | SA_THIRTYTWO |             \
+      SA_RESTORER | SA_ONSTACK | SA_RESTART | SA_NODEFER | SA_RESETHAND)
+

I wonder whether all these per-arch definitions will tend to bitrot when people add new common flags.

Quoted 2 lines

Can we have a common definition for the common bits, and just add the extra arch-specific ones here?

I think so. We could have something like:

#define ARCH_UAPI_SA_FLAGS SA_THIRTYTWO

here. Then in signal_types.h we can do:

Quoted 7 lines

#ifndef ARCH_UAPI_SA_FLAGS
#define ARCH_UAPI_SA_FLAGS 0
#endif

#define UAPI_SA_FLAGS (... | ARCH_UAPI_SA_FLAGS)

I'll do that in v10.

Yes, something like that would be fine, I should think.

Quoted 4 lines

Quoted 2 lines

Also, I wonder whether we should avoid the "SA_" prefix here. Maybe UAPI_SA_FLAGS?

Sounds good to me.

OK, great.

[...]

Quoted 28 lines

Quoted 8 lines

Quoted 8 lines

kernel/signal.c +0 -0

index 42b67d2cea37..348b7981f1ff 100644
@@ -3984,6 +3984,16 @@ int do_sigaction(int sig, struct k_sigaction *act, struct k_sigaction *oact)
      if (oact)
              *oact = *k;
 

Quoted 12 lines

kernel/signal.c int do_sigaction(int sig, struct k_sigaction *act, struct k_sigaction *oact) · line 3996

Quoted 10 lines

kernel/signal.c +10 -0

+     /*
+      * Clear unknown flag bits in order to allow userspace to detect missing
+      * support for flag bits and to allow the kernel to use non-uapi bits
+      * internally.
+      */
+     if (act)
+             act->sa.sa_flags &= SA_UAPI_FLAGS;
+     if (oact)
+             oact->sa.sa_flags &= SA_UAPI_FLAGS;
+

Seems reasonable.

Thanks. I also decided to check how other operating systems handle unknown flag bits in sigaction.sa_flags. It looks like OpenBSD and illumos also accept unknown bits but (implicitly, as a result of using a different internal representation) end up clearing them in oldact:

https://github.com/openbsd/src/blob/f634a6a4b5bf832e9c1de77f7894ae2625e74484/sys/kern/kern_sig.c#L278
https://github.com/illumos/illumos-gate/blob/76f19f5fdc974fe5be5c82a556e43a4df93f1de1/usr/src/uts/common/syscall/sigaction.c#L86

Quoted 6 lines

and FreeBSD and NetBSD fail the syscall if unknown bits are set:

https://github.com/freebsd/freebsd/blob/eded70c37057857c6e23fae51f86b8f8f43cd2d0/sys/kern/kern_sig.c#L699
https://github.com/NetBSD/src/blob/3365779becdcedfca206091a645a0e8e22b2946e/sys/kern/sys_sig.c#L473

Quoted 2 lines

So there is some precedent for doing what we're planning to do here, which makes it yet more likely that we'll be okay doing this.

Ack, it's good to have that extra evidence to support this approach.

This also means that other OSes could adopt the new Linux flag(s) with comatible semantics, if they wanted to. Or have I misunderstood something there?

Quoted 5 lines · Dave Martin

Quoted 3 lines · Peter Collingbourne

Quoted 2 lines · Dave Martin

On Mon, Aug 17, 2020 at 08:33:48PM -0700, Peter Collingbourne wrote:

Quoted 21 lines

Quoted 21 lines

Quoted 20 lines

Quoted 8 lines

kernel/signal.c +0 -0

index 42b67d2cea37..348b7981f1ff 100644
@@ -3984,6 +3984,16 @@ int do_sigaction(int sig, struct k_sigaction *act, struct k_sigaction *oact)
      if (oact)
              *oact = *k;
 

kernel/signal.c int do_sigaction(int sig, struct k_sigaction *act, struct k_sigaction *oact) · line 3996

Quoted 10 lines

kernel/signal.c +10 -0

+     /*
+      * Clear unknown flag bits in order to allow userspace to detect missing
+      * support for flag bits and to allow the kernel to use non-uapi bits
+      * internally.
+      */
+     if (act)
+             act->sa.sa_flags &= SA_UAPI_FLAGS;
+     if (oact)
+             oact->sa.sa_flags &= SA_UAPI_FLAGS;
+

Seems reasonable.

Quoted 7 lines

Quoted 7 lines

Thanks. I also decided to check how other operating systems handle unknown flag bits in sigaction.sa_flags. It looks like OpenBSD and illumos also accept unknown bits but (implicitly, as a result of using a different internal representation) end up clearing them in oldact:

https://github.com/openbsd/src/blob/f634a6a4b5bf832e9c1de77f7894ae2625e74484/sys/kern/kern_sig.c#L278
https://github.com/illumos/illumos-gate/blob/76f19f5fdc974fe5be5c82a556e43a4df93f1de1/usr/src/uts/common/syscall/sigaction.c#L86

XNU does the same:

https://github.com/apple/darwin-xnu/blob/a449c6a3b8014d9406c2ddbdc81795da24aa7443/bsd/kern/kern_sig.c#L480

Quoted 6 lines

Quoted 6 lines

and FreeBSD and NetBSD fail the syscall if unknown bits are set:

https://github.com/freebsd/freebsd/blob/eded70c37057857c6e23fae51f86b8f8f43cd2d0/sys/kern/kern_sig.c#L699
https://github.com/NetBSD/src/blob/3365779becdcedfca206091a645a0e8e22b2946e/sys/kern/sys_sig.c#L473

Quoted 8 lines

Quoted 2 lines

So there is some precedent for doing what we're planning to do here, which makes it yet more likely that we'll be okay doing this.

Ack, it's good to have that extra evidence to support this approach.

This also means that other OSes could adopt the new Linux flag(s) with comatible semantics, if they wanted to. Or have I misunderstood something there?

The other OSs could adopt SA_XFLAGS, but they would probably have no need for SA_UNSUPPORTED because they already have a protocol for detecting missing flag support in the kernel (Linux is really the odd one out here in not supporting such a protocol from the start). Userspace programs running on OpenBSD, illumos and XNU could use the Linux protocol without the SA_UNSUPPORTED part, while programs running on FreeBSD and NetBSD could do something like:

static bool has_xflags = true;
[...]
struct sigaction act;
act.sa_flags = SA_SIGINFO | SA_XFLAGS;
if (sigaction(SIGSEGV, &act, 0) != 0) {
has_xflags = false;
act.sa_flags = SA_SIGINFO;
if (sigaction(SIGSEGV, &act, 0) != 0)

    perror("sigaction");
}

It would probably be possible to write a unified function that would support all three protocols.

This bit will never be supported in the uapi. The purpose of this flag bit is to allow userspace to distinguish an old kernel that does not clear unknown sa_flags bits from a kernel that supports every flag bit.

In other words, if userspace finds that this bit remains set in oldact.sa_flags, it means that the kernel cannot be trusted to have cleared unknown flag bits from sa_flags, so no assumptions about flag bit support can be made.

Signed-off-by Peter Collingbourne <[email protected]>

---
View this change in Gerrit: https://linux-review.googlesource.com/q/Ic2501ad150a3a79c1cf27fb8c99be342e9dffbcb

 include/uapi/asm-generic/signal-defs.h | 7 +++++++
 kernel/signal.c                        | 6 ++++++
 2 files changed, 13 insertions(+)

include/uapi/asm-generic/signal-defs.h +6 -0

index 91000b6b97e0..c30a9c1a77b2 100644
@@ -13,6 +13,12 @@
  * SA_RESETHAND clears the handler when the signal is delivered.
  * SA_NOCLDWAIT flag on SIGCHLD to inhibit zombies.
  * SA_NODEFER prevents the current signal from being masked in the handler.
+ * SA_UNSUPPORTED is a flag bit that will never be supported. Kernels from
+ * before the introduction of SA_UNSUPPORTED did not clear unknown bits from
+ * sa_flags when read using the oldact argument to sigaction and rt_sigaction,
+ * so this bit allows flag bit support to be detected from userspace while
+ * allowing an old kernel to be distinguished from a kernel that supports every
+ * flag bit.
  *
  * SA_ONESHOT and SA_NOMASK are the historical Linux names for the Single
  * Unix names RESETHAND and NODEFER respectively.

Quoted 34 lines · Peter Collingbourne

This bit will never be supported in the uapi. The purpose of this flag bit is to allow userspace to distinguish an old kernel that does not clear unknown sa_flags bits from a kernel that supports every flag bit.

In other words, if userspace finds that this bit remains set in oldact.sa_flags, it means that the kernel cannot be trusted to have cleared unknown flag bits from sa_flags, so no assumptions about flag bit support can be made.

Signed-off-by Peter Collingbourne <[email protected]>

---
View this change in Gerrit: https://linux-review.googlesource.com/q/Ic2501ad150a3a79c1cf27fb8c99be342e9dffbcb

 include/uapi/asm-generic/signal-defs.h | 7 +++++++
 kernel/signal.c                        | 6 ++++++
 2 files changed, 13 insertions(+)

include/uapi/asm-generic/signal-defs.h +6 -0

index 91000b6b97e0..c30a9c1a77b2 100644
@@ -13,6 +13,12 @@
  * SA_RESETHAND clears the handler when the signal is delivered.
  * SA_NOCLDWAIT flag on SIGCHLD to inhibit zombies.
  * SA_NODEFER prevents the current signal from being masked in the handler.
+ * SA_UNSUPPORTED is a flag bit that will never be supported. Kernels from
+ * before the introduction of SA_UNSUPPORTED did not clear unknown bits from
+ * sa_flags when read using the oldact argument to sigaction and rt_sigaction,
+ * so this bit allows flag bit support to be detected from userspace while
+ * allowing an old kernel to be distinguished from a kernel that supports every
+ * flag bit.
  *
  * SA_ONESHOT and SA_NOMASK are the historical Linux names for the Single
  * Unix names RESETHAND and NODEFER respectively.

include/uapi/asm-generic/signal-defs.h line 51

Quoted 5 lines

include/uapi/asm-generic/signal-defs.h +1 -0

@@ -42,6 +48,7 @@
  * The following bits are used in architecture-specific SA_* definitions and
  * should be avoided for new generic flags: 3, 4, 5, 6, 7, 8, 9, 16, 24, 25, 26.
  */
+#define SA_UNSUPPORTED       0x00000400

This concept confused me a bit initially, since in a sense this flag is supported, just with a rather peculiar meaning.

Since the main (only) purpose of this bit will be to check whether

SA_XFLAGS is actually supported, I wonder whether it makes sense to weld the two together, say:

#define SA_REQUEST_XFLAGS       0x00000c00
#define SA_XFLAGS_MASK          0x00000c00
#define SA_HAVE_XFLAGS          0x00000800

This is a departure from the current style of definitions though.

        sa.sa_flags |= SA_REQUEST_XFLAGS;
        sigaction(..., &sa, &sa);
        if ((sa.sa_flags & SA_XFLAGS_MASK) == SA_HAVE_XFLAGS)
                /* xflags available */

This would require some juggling of the way SA_UAPI_FLAGS works though. Maybe not worth it, so long as the semantics get clearly documented.

Quoted 36 lines · Dave Martin

Quoted 34 lines · Peter Collingbourne

This bit will never be supported in the uapi. The purpose of this flag bit is to allow userspace to distinguish an old kernel that does not clear unknown sa_flags bits from a kernel that supports every flag bit.

In other words, if userspace finds that this bit remains set in oldact.sa_flags, it means that the kernel cannot be trusted to have cleared unknown flag bits from sa_flags, so no assumptions about flag bit support can be made.

Signed-off-by Peter Collingbourne <[email protected]>

---
View this change in Gerrit: https://linux-review.googlesource.com/q/Ic2501ad150a3a79c1cf27fb8c99be342e9dffbcb

 include/uapi/asm-generic/signal-defs.h | 7 +++++++
 kernel/signal.c                        | 6 ++++++
 2 files changed, 13 insertions(+)

include/uapi/asm-generic/signal-defs.h +6 -0

index 91000b6b97e0..c30a9c1a77b2 100644
@@ -13,6 +13,12 @@
  * SA_RESETHAND clears the handler when the signal is delivered.
  * SA_NOCLDWAIT flag on SIGCHLD to inhibit zombies.
  * SA_NODEFER prevents the current signal from being masked in the handler.
+ * SA_UNSUPPORTED is a flag bit that will never be supported. Kernels from
+ * before the introduction of SA_UNSUPPORTED did not clear unknown bits from
+ * sa_flags when read using the oldact argument to sigaction and rt_sigaction,
+ * so this bit allows flag bit support to be detected from userspace while
+ * allowing an old kernel to be distinguished from a kernel that supports every
+ * flag bit.
  *
  * SA_ONESHOT and SA_NOMASK are the historical Linux names for the Single
  * Unix names RESETHAND and NODEFER respectively.

Quoted 8 lines

include/uapi/asm-generic/signal-defs.h line 51

Quoted 5 lines

include/uapi/asm-generic/signal-defs.h +1 -0

@@ -42,6 +48,7 @@
  * The following bits are used in architecture-specific SA_* definitions and
  * should be avoided for new generic flags: 3, 4, 5, 6, 7, 8, 9, 16, 24, 25, 26.
  */
+#define SA_UNSUPPORTED       0x00000400

This concept confused me a bit initially, since in a sense this flag is supported, just with a rather peculiar meaning.

Hmm. Maybe it should be named "SA_UNKNOWN" to mean that the bit will always be "unknown" to the kernel in the sense that it shall be treated in the same way as any other "unknown" bit. Then we can define the kernel's behavior in terms of what happens if a bit is "known". I don't know if this is just shuffling terms around though. At any rate, this seems like a problem to be solved with documentation.

Quoted 1 line

Since the main (only) purpose of this bit will be to check whether

I wouldn't necessarily say that it is the only purpose. If another new sa_flags bit were to be introduced in the future, SA_UN(whatever) could be used to detect kernel support for that bit in the same way as SA_XFLAGS.

Quoted 15 lines

SA_XFLAGS is actually supported, I wonder whether it makes sense to weld the two together, say:

#define SA_REQUEST_XFLAGS       0x00000c00
#define SA_XFLAGS_MASK          0x00000c00
#define SA_HAVE_XFLAGS          0x00000800

This is a departure from the current style of definitions though.

        sa.sa_flags |= SA_REQUEST_XFLAGS;
        sigaction(..., &sa, &sa);
        if ((sa.sa_flags & SA_XFLAGS_MASK) == SA_HAVE_XFLAGS)
                /* xflags available */

Quoted 2 lines

This would require some juggling of the way SA_UAPI_FLAGS works though. Maybe not worth it, so long as the semantics get clearly documented.

I'm not sure about this. I personally think that it would be clearer to keep the flags orthogonal.

This field will contain flags that may be used by signal handlers to determine whether other fields in the _sigfault portion of siginfo are valid. An example use case is the following patch, which introduces the si_addr_ignored_bits{,_mask} fields.

A new sigcontext flag, SA_XFLAGS, is introduced in order to allow a signal handler to require the kernel to set the field (but note that the field will be set anyway if the kernel supports the flag, regardless of its value). In combination with the previous patches, this allows a userspace program to determine whether the kernel will set the field.

Ideally this field could have just been named si_flags, but that name was already taken by ia64, so a different name was chosen.

Alternatively, we may consider making ia64's si_flags a generic field and having it appear at the end of _sigfault (in the same place as this patch has si_xflags) on non-ia64, keeping it in the same place on ia64. ia64's si_flags is a 32-bit field with only one flag bit allocated, so we would have 31 bits to use if we do this.

Signed-off-by Peter Collingbourne <[email protected]>

---
View this change in Gerrit: https://linux-review.googlesource.com/q/Ide155ce29366c3eab2a944ae4c51205982e5b8b2

 arch/arm/include/asm/signal.h              |  3 ++-
 arch/parisc/include/asm/signal.h           |  2 +-
 arch/powerpc/platforms/powernv/vas-fault.c |  1 +
 include/linux/compat.h                     |  2 ++
 include/linux/signal_types.h               |  4 ++--
 include/uapi/asm-generic/siginfo.h         |  3 +++
 include/uapi/asm-generic/signal-defs.h     |  4 ++++
 kernel/signal.c                            | 15 +++++++++++++++
 8 files changed, 30 insertions(+), 4 deletions(-)

arch/arm/include/asm/signal.h +2 -1

index d1070a783993..6b2630dfe1df 100644
@@ -19,7 +19,8 @@ typedef struct {
 
 #define SA_UAPI_FLAGS                                                          \
 	(SA_NOCLDSTOP | SA_NOCLDWAIT | SA_SIGINFO | SA_THIRTYTWO |             \
-	 SA_RESTORER | SA_ONSTACK | SA_RESTART | SA_NODEFER | SA_RESETHAND)
+	 SA_RESTORER | SA_ONSTACK | SA_RESTART | SA_NODEFER | SA_RESETHAND |   \
+	 SA_XFLAGS)
 
 #define __ARCH_HAS_SA_RESTORER
 

arch/parisc/include/asm/signal.h +1 -1

index ad06e14f6e8a..3582bce44520 100644
@@ -23,7 +23,7 @@ typedef struct {
 
 #define SA_UAPI_FLAGS                                                          \
 	(SA_ONSTACK | SA_RESETHAND | SA_NOCLDSTOP | SA_SIGINFO | SA_NODEFER |  \
-	 SA_RESTART | SA_NOCLDWAIT | _SA_SIGGFAULT)
+	 SA_RESTART | SA_NOCLDWAIT | _SA_SIGGFAULT | SA_XFLAGS)
 
 #include <asm/sigcontext.h>
 

arch/powerpc/platforms/powernv/vas-fault.c +1 -0

index 3d21fce254b7..3bbb335561f5 100644
@@ -154,6 +154,7 @@ static void update_csb(struct vas_window *window,
 	info.si_errno = EFAULT;
 	info.si_code = SEGV_MAPERR;
 	info.si_addr = csb_addr;
+	info.si_xflags = 0;
 
 	/*
 	 * process will be polling on csb.flags after request is sent to

include/linux/compat.h +2 -0

index d38c4d7e83bd..55d4228dfd88 100644
@@ -231,7 +231,9 @@ typedef struct compat_siginfo {
 					char _dummy_pkey[__COMPAT_ADDR_BND_PKEY_PAD];
 					u32 _pkey;
 				} _addr_pkey;
+				compat_uptr_t _pad[6];
 			};
+			compat_uptr_t _xflags;
 		} _sigfault;
 
 		/* SIGPOLL */

include/linux/signal_types.h +2 -2

index e792f29b5846..cd3d08dde47a 100644
@@ -72,11 +72,11 @@ struct ksignal {
 #ifdef SA_RESTORER
 #define SA_UAPI_FLAGS                                                          \
 	(SA_NOCLDSTOP | SA_NOCLDWAIT | SA_SIGINFO | SA_ONSTACK | SA_RESTART |  \
-	 SA_NODEFER | SA_RESETHAND | SA_RESTORER)
+	 SA_NODEFER | SA_RESETHAND | SA_RESTORER | SA_XFLAGS)
 #else
 #define SA_UAPI_FLAGS                                                          \
 	(SA_NOCLDSTOP | SA_NOCLDWAIT | SA_SIGINFO | SA_ONSTACK | SA_RESTART |  \
-	 SA_NODEFER | SA_RESETHAND)
+	 SA_NODEFER | SA_RESETHAND | SA_XFLAGS)
 #endif
 #endif
 

include/uapi/asm-generic/siginfo.h +3 -0

index cb3d6c267181..413d804623c0 100644
@@ -91,7 +91,9 @@ union __sifields {
 				char _dummy_pkey[__ADDR_BND_PKEY_PAD];
 				__u32 _pkey;
 			} _addr_pkey;
+			void *_pad[6];
 		};
+		unsigned long _xflags;
 	} _sigfault;
 
 	/* SIGPOLL */

@@ -152,6 +154,7 @@ typedef struct siginfo {
 #define si_trapno	_sifields._sigfault._trapno
 #endif
 #define si_addr_lsb	_sifields._sigfault._addr_lsb
+#define si_xflags	_sifields._sigfault._xflags
 #define si_lower	_sifields._sigfault._addr_bnd._lower
 #define si_upper	_sifields._sigfault._addr_bnd._upper
 #define si_pkey		_sifields._sigfault._addr_pkey._pkey

include/uapi/asm-generic/signal-defs.h +4 -0

index c30a9c1a77b2..aeee6bb0763b 100644
@@ -19,6 +19,9 @@
  * so this bit allows flag bit support to be detected from userspace while
  * allowing an old kernel to be distinguished from a kernel that supports every
  * flag bit.
+ * SA_XFLAGS indicates that the signal handler requires the siginfo.si_xflags
+ * field to be valid. Note that if the kernel supports SA_XFLAGS, the field will
+ * be valid regardless of the value of this flag.
  *
  * SA_ONESHOT and SA_NOMASK are the historical Linux names for the Single
  * Unix names RESETHAND and NODEFER respectively.

@@ -49,6 +52,7 @@
  * should be avoided for new generic flags: 3, 4, 5, 6, 7, 8, 9, 16, 24, 25, 26.
  */
 #define SA_UNSUPPORTED	0x00000400
+#define SA_XFLAGS	0x00000800
 
 #define SA_NOMASK	SA_NODEFER
 #define SA_ONESHOT	SA_RESETHAND

kernel/signal.c +14 -0

index 664a6c31137e..72182eed1b8d 100644
@@ -1669,6 +1669,7 @@ int force_sig_fault_to_task(int sig, int code, void __user *addr
 	info.si_flags = flags;
 	info.si_isr = isr;
 #endif
+	info.si_xflags = 0;
 	return force_sig_info_to_task(&info, t);
 }
 

@@ -1701,6 +1702,7 @@ int send_sig_fault(int sig, int code, void __user *addr
 	info.si_flags = flags;
 	info.si_isr = isr;
 #endif
+	info.si_xflags = 0;
 	return send_sig_info(info.si_signo, &info, t);
 }
 

@@ -1715,6 +1717,7 @@ int force_sig_mceerr(int code, void __user *addr, short lsb)
 	info.si_code = code;
 	info.si_addr = addr;
 	info.si_addr_lsb = lsb;
+	info.si_xflags = 0;
 	return force_sig_info(&info);
 }
 

@@ -1729,6 +1732,7 @@ int send_sig_mceerr(int code, void __user *addr, short lsb, struct task_struct *
 	info.si_code = code;
 	info.si_addr = addr;
 	info.si_addr_lsb = lsb;
+	info.si_xflags = 0;
 	return send_sig_info(info.si_signo, &info, t);
 }
 EXPORT_SYMBOL(send_sig_mceerr);

@@ -1744,6 +1748,7 @@ int force_sig_bnderr(void __user *addr, void __user *lower, void __user *upper)
 	info.si_addr  = addr;
 	info.si_lower = lower;
 	info.si_upper = upper;
+	info.si_xflags = 0;
 	return force_sig_info(&info);
 }
 

@@ -1758,6 +1763,7 @@ int force_sig_pkuerr(void __user *addr, u32 pkey)
 	info.si_code  = SEGV_PKUERR;
 	info.si_addr  = addr;
 	info.si_pkey  = pkey;
+	info.si_xflags = 0;
 	return force_sig_info(&info);
 }
 #endif

@@ -1774,6 +1780,7 @@ int force_sig_ptrace_errno_trap(int errno, void __user *addr)
 	info.si_errno = errno;
 	info.si_code  = TRAP_HWBKPT;
 	info.si_addr  = addr;
+	info.si_xflags = 0;
 	return force_sig_info(&info);
 }
 

@@ -3290,6 +3297,7 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 #ifdef __ARCH_SI_TRAPNO
 		to->si_trapno = from->si_trapno;
 #endif
+		to->si_xflags = from->si_xflags;
 		break;
 	case SIL_FAULT_MCEERR:
 		to->si_addr = ptr_to_compat(from->si_addr);

@@ -3297,6 +3305,7 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 		to->si_trapno = from->si_trapno;
 #endif
 		to->si_addr_lsb = from->si_addr_lsb;
+		to->si_xflags = from->si_xflags;
 		break;
 	case SIL_FAULT_BNDERR:
 		to->si_addr = ptr_to_compat(from->si_addr);

@@ -3305,6 +3314,7 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 #endif
 		to->si_lower = ptr_to_compat(from->si_lower);
 		to->si_upper = ptr_to_compat(from->si_upper);
+		to->si_xflags = from->si_xflags;
 		break;
 	case SIL_FAULT_PKUERR:
 		to->si_addr = ptr_to_compat(from->si_addr);

@@ -3312,6 +3322,7 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 		to->si_trapno = from->si_trapno;
 #endif
 		to->si_pkey = from->si_pkey;
+		to->si_xflags = from->si_xflags;
 		break;
 	case SIL_CHLD:
 		to->si_pid = from->si_pid;

@@ -3370,6 +3381,7 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 #ifdef __ARCH_SI_TRAPNO
 		to->si_trapno = from->si_trapno;
 #endif
+		to->si_xflags = from->si_xflags;
 		break;
 	case SIL_FAULT_MCEERR:
 		to->si_addr = compat_ptr(from->si_addr);

@@ -3377,6 +3389,7 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 		to->si_trapno = from->si_trapno;
 #endif
 		to->si_addr_lsb = from->si_addr_lsb;
+		to->si_xflags = from->si_xflags;
 		break;
 	case SIL_FAULT_BNDERR:
 		to->si_addr = compat_ptr(from->si_addr);

@@ -3385,6 +3398,7 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 #endif
 		to->si_lower = compat_ptr(from->si_lower);
 		to->si_upper = compat_ptr(from->si_upper);
+		to->si_xflags = from->si_xflags;
 		break;
 	case SIL_FAULT_PKUERR:
 		to->si_addr = compat_ptr(from->si_addr);

Quoted 15 lines · Peter Collingbourne

This field will contain flags that may be used by signal handlers to determine whether other fields in the _sigfault portion of siginfo are valid. An example use case is the following patch, which introduces the si_addr_ignored_bits{,_mask} fields.

A new sigcontext flag, SA_XFLAGS, is introduced in order to allow a signal handler to require the kernel to set the field (but note that the field will be set anyway if the kernel supports the flag, regardless of its value). In combination with the previous patches, this allows a userspace program to determine whether the kernel will set the field.

Ideally this field could have just been named si_flags, but that name was already taken by ia64, so a different name was chosen.

Quoted 5 lines

Alternatively, we may consider making ia64's si_flags a generic field and having it appear at the end of _sigfault (in the same place as this patch has si_xflags) on non-ia64, keeping it in the same place on ia64. ia64's si_flags is a 32-bit field with only one flag bit allocated, so we would have 31 bits to use if we do this.

For clarity, is the new si_xflags field supposed to be valid for all signal types, or just certain signals and si_codes?

What happens for things like a rt_sigqueueinfo() from userspace?

Quoted 246 lines

Signed-off-by Peter Collingbourne <[email protected]>

---
View this change in Gerrit: https://linux-review.googlesource.com/q/Ide155ce29366c3eab2a944ae4c51205982e5b8b2

 arch/arm/include/asm/signal.h              |  3 ++-
 arch/parisc/include/asm/signal.h           |  2 +-
 arch/powerpc/platforms/powernv/vas-fault.c |  1 +
 include/linux/compat.h                     |  2 ++
 include/linux/signal_types.h               |  4 ++--
 include/uapi/asm-generic/siginfo.h         |  3 +++
 include/uapi/asm-generic/signal-defs.h     |  4 ++++
 kernel/signal.c                            | 15 +++++++++++++++
 8 files changed, 30 insertions(+), 4 deletions(-)

arch/arm/include/asm/signal.h +2 -1

index d1070a783993..6b2630dfe1df 100644
@@ -19,7 +19,8 @@ typedef struct {
 
 #define SA_UAPI_FLAGS                                                          \
      (SA_NOCLDSTOP | SA_NOCLDWAIT | SA_SIGINFO | SA_THIRTYTWO |             \
-      SA_RESTORER | SA_ONSTACK | SA_RESTART | SA_NODEFER | SA_RESETHAND)
+      SA_RESTORER | SA_ONSTACK | SA_RESTART | SA_NODEFER | SA_RESETHAND |   \
+      SA_XFLAGS)
 
 #define __ARCH_HAS_SA_RESTORER
 

arch/parisc/include/asm/signal.h +1 -1

index ad06e14f6e8a..3582bce44520 100644
@@ -23,7 +23,7 @@ typedef struct {
 
 #define SA_UAPI_FLAGS                                                          \
      (SA_ONSTACK | SA_RESETHAND | SA_NOCLDSTOP | SA_SIGINFO | SA_NODEFER |  \
-      SA_RESTART | SA_NOCLDWAIT | _SA_SIGGFAULT)
+      SA_RESTART | SA_NOCLDWAIT | _SA_SIGGFAULT | SA_XFLAGS)
 
 #include <asm/sigcontext.h>
 

arch/powerpc/platforms/powernv/vas-fault.c +1 -0

index 3d21fce254b7..3bbb335561f5 100644
@@ -154,6 +154,7 @@ static void update_csb(struct vas_window *window,
      info.si_errno = EFAULT;
      info.si_code = SEGV_MAPERR;
      info.si_addr = csb_addr;
+     info.si_xflags = 0;
 
      /*
       * process will be polling on csb.flags after request is sent to

include/linux/compat.h +2 -0

index d38c4d7e83bd..55d4228dfd88 100644
@@ -231,7 +231,9 @@ typedef struct compat_siginfo {
                                      char _dummy_pkey[__COMPAT_ADDR_BND_PKEY_PAD];
                                      u32 _pkey;
                              } _addr_pkey;
+                             compat_uptr_t _pad[6];
                      };
+                     compat_uptr_t _xflags;
              } _sigfault;
 
              /* SIGPOLL */

include/linux/signal_types.h +2 -2

index e792f29b5846..cd3d08dde47a 100644
@@ -72,11 +72,11 @@ struct ksignal {
 #ifdef SA_RESTORER
 #define SA_UAPI_FLAGS                                                          \
      (SA_NOCLDSTOP | SA_NOCLDWAIT | SA_SIGINFO | SA_ONSTACK | SA_RESTART |  \
-      SA_NODEFER | SA_RESETHAND | SA_RESTORER)
+      SA_NODEFER | SA_RESETHAND | SA_RESTORER | SA_XFLAGS)
 #else
 #define SA_UAPI_FLAGS                                                          \
      (SA_NOCLDSTOP | SA_NOCLDWAIT | SA_SIGINFO | SA_ONSTACK | SA_RESTART |  \
-      SA_NODEFER | SA_RESETHAND)
+      SA_NODEFER | SA_RESETHAND | SA_XFLAGS)
 #endif
 #endif
 

include/uapi/asm-generic/siginfo.h +3 -0

index cb3d6c267181..413d804623c0 100644
@@ -91,7 +91,9 @@ union __sifields {
                              char _dummy_pkey[__ADDR_BND_PKEY_PAD];
                              __u32 _pkey;
                      } _addr_pkey;
+                     void *_pad[6];
              };
+             unsigned long _xflags;
      } _sigfault;
 
      /* SIGPOLL */

@@ -152,6 +154,7 @@ typedef struct siginfo {
 #define si_trapno    _sifields._sigfault._trapno
 #endif
 #define si_addr_lsb  _sifields._sigfault._addr_lsb
+#define si_xflags    _sifields._sigfault._xflags
 #define si_lower     _sifields._sigfault._addr_bnd._lower
 #define si_upper     _sifields._sigfault._addr_bnd._upper
 #define si_pkey              _sifields._sigfault._addr_pkey._pkey

include/uapi/asm-generic/signal-defs.h +4 -0

index c30a9c1a77b2..aeee6bb0763b 100644
@@ -19,6 +19,9 @@
  * so this bit allows flag bit support to be detected from userspace while
  * allowing an old kernel to be distinguished from a kernel that supports every
  * flag bit.
+ * SA_XFLAGS indicates that the signal handler requires the siginfo.si_xflags
+ * field to be valid. Note that if the kernel supports SA_XFLAGS, the field will
+ * be valid regardless of the value of this flag.
  *
  * SA_ONESHOT and SA_NOMASK are the historical Linux names for the Single
  * Unix names RESETHAND and NODEFER respectively.

@@ -49,6 +52,7 @@
  * should be avoided for new generic flags: 3, 4, 5, 6, 7, 8, 9, 16, 24, 25, 26.
  */
 #define SA_UNSUPPORTED       0x00000400
+#define SA_XFLAGS    0x00000800
 
 #define SA_NOMASK    SA_NODEFER
 #define SA_ONESHOT   SA_RESETHAND

kernel/signal.c +14 -0

index 664a6c31137e..72182eed1b8d 100644
@@ -1669,6 +1669,7 @@ int force_sig_fault_to_task(int sig, int code, void __user *addr
      info.si_flags = flags;
      info.si_isr = isr;
 #endif
+     info.si_xflags = 0;
      return force_sig_info_to_task(&info, t);
 }
 

@@ -1701,6 +1702,7 @@ int send_sig_fault(int sig, int code, void __user *addr
      info.si_flags = flags;
      info.si_isr = isr;
 #endif
+     info.si_xflags = 0;
      return send_sig_info(info.si_signo, &info, t);
 }
 

@@ -1715,6 +1717,7 @@ int force_sig_mceerr(int code, void __user *addr, short lsb)
      info.si_code = code;
      info.si_addr = addr;
      info.si_addr_lsb = lsb;
+     info.si_xflags = 0;
      return force_sig_info(&info);
 }
 

@@ -1729,6 +1732,7 @@ int send_sig_mceerr(int code, void __user *addr, short lsb, struct task_struct *
      info.si_code = code;
      info.si_addr = addr;
      info.si_addr_lsb = lsb;
+     info.si_xflags = 0;
      return send_sig_info(info.si_signo, &info, t);
 }
 EXPORT_SYMBOL(send_sig_mceerr);

@@ -1744,6 +1748,7 @@ int force_sig_bnderr(void __user *addr, void __user *lower, void __user *upper)
      info.si_addr  = addr;
      info.si_lower = lower;
      info.si_upper = upper;
+     info.si_xflags = 0;
      return force_sig_info(&info);
 }
 

@@ -1758,6 +1763,7 @@ int force_sig_pkuerr(void __user *addr, u32 pkey)
      info.si_code  = SEGV_PKUERR;
      info.si_addr  = addr;
      info.si_pkey  = pkey;
+     info.si_xflags = 0;
      return force_sig_info(&info);
 }
 #endif

@@ -1774,6 +1780,7 @@ int force_sig_ptrace_errno_trap(int errno, void __user *addr)
      info.si_errno = errno;
      info.si_code  = TRAP_HWBKPT;
      info.si_addr  = addr;
+     info.si_xflags = 0;
      return force_sig_info(&info);
 }
 

@@ -3290,6 +3297,7 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 #ifdef __ARCH_SI_TRAPNO
              to->si_trapno = from->si_trapno;
 #endif
+             to->si_xflags = from->si_xflags;
              break;
      case SIL_FAULT_MCEERR:
              to->si_addr = ptr_to_compat(from->si_addr);

@@ -3297,6 +3305,7 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
              to->si_trapno = from->si_trapno;
 #endif
              to->si_addr_lsb = from->si_addr_lsb;
+             to->si_xflags = from->si_xflags;
              break;
      case SIL_FAULT_BNDERR:
              to->si_addr = ptr_to_compat(from->si_addr);

@@ -3305,6 +3314,7 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
 #endif
              to->si_lower = ptr_to_compat(from->si_lower);
              to->si_upper = ptr_to_compat(from->si_upper);
+             to->si_xflags = from->si_xflags;
              break;
      case SIL_FAULT_PKUERR:
              to->si_addr = ptr_to_compat(from->si_addr);

@@ -3312,6 +3322,7 @@ void copy_siginfo_to_external32(struct compat_siginfo *to,
              to->si_trapno = from->si_trapno;
 #endif
              to->si_pkey = from->si_pkey;
+             to->si_xflags = from->si_xflags;
              break;
      case SIL_CHLD:
              to->si_pid = from->si_pid;

@@ -3370,6 +3381,7 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 #ifdef __ARCH_SI_TRAPNO
              to->si_trapno = from->si_trapno;
 #endif
+             to->si_xflags = from->si_xflags;
              break;
      case SIL_FAULT_MCEERR:
              to->si_addr = compat_ptr(from->si_addr);

@@ -3377,6 +3389,7 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
              to->si_trapno = from->si_trapno;
 #endif
              to->si_addr_lsb = from->si_addr_lsb;
+             to->si_xflags = from->si_xflags;
              break;
      case SIL_FAULT_BNDERR:
              to->si_addr = compat_ptr(from->si_addr);

@@ -3385,6 +3398,7 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 #endif
              to->si_lower = compat_ptr(from->si_lower);
              to->si_upper = compat_ptr(from->si_upper);
+             to->si_xflags = from->si_xflags;
              break;
      case SIL_FAULT_PKUERR:
              to->si_addr = compat_ptr(from->si_addr);

kernel/signal.c static int post_copy_siginfo_from_user32(kernel_siginfo_t *to, · line 3409

Quoted 5 lines

kernel/signal.c +1 -0

@@ -3392,6 +3406,7 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
              to->si_trapno = from->si_trapno;
 #endif
              to->si_pkey = from->si_pkey;
+             to->si_xflags = from->si_xflags;

How did you figure out the list of places to make these changes? I'm not sure how to confirm that it's exhaustive.

It's a shame if we can't simply apply the change in one place. Would the refactoring be too invasive to accomplish that?

Quoted 23 lines · Dave Martin

Quoted 20 lines · Peter Collingbourne

This field will contain flags that may be used by signal handlers to determine whether other fields in the _sigfault portion of siginfo are valid. An example use case is the following patch, which introduces the si_addr_ignored_bits{,_mask} fields.

A new sigcontext flag, SA_XFLAGS, is introduced in order to allow a signal handler to require the kernel to set the field (but note that the field will be set anyway if the kernel supports the flag, regardless of its value). In combination with the previous patches, this allows a userspace program to determine whether the kernel will set the field.

Ideally this field could have just been named si_flags, but that name was already taken by ia64, so a different name was chosen.

Alternatively, we may consider making ia64's si_flags a generic field and having it appear at the end of _sigfault (in the same place as this patch has si_xflags) on non-ia64, keeping it in the same place on ia64. ia64's si_flags is a 32-bit field with only one flag bit allocated, so we would have 31 bits to use if we do this.

Quoted 2 lines

For clarity, is the new si_xflags field supposed to be valid for all signal types, or just certain signals and si_codes?

It is intended to be valid for all signal types that use the _sigfault union member of siginfo. As listed in siginfo.h these are: SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT.

Quoted 1 line

What happens for things like a rt_sigqueueinfo() from userspace?

Hmm. Let's enumerate each of these things, which I believe are all of the call sites of the function copy_siginfo_from_user and related functions (correct me if I'm wrong):

- ptrace(PTRACE_SETSIGINFO)
- pidfd_send_signal
- rt_sigqueueinfo
- rt_tgsigqueueinfo

We can handle the last three by observing that the kernel forbids sending a signal with these syscalls if si_code >= 0, so we can say that the value of si_xflags is only valid if si_code >= 0.

As for the first one, it's more tricky. Arguably something like a debugger should be able to send arbitrary signals to a debuggee, and there's no reason why it shouldn't be able to set si_xflags in siginfo, but on the other hand who knows how existing debuggers end up setting this field today. Maybe all that we can do is have the kernel clear si_xflags if it detects that the signal uses _sigfault, and let si_xflags aware debuggers opt out of this behavior, perhaps by introducing a PTRACE_SETSIGINFO2 or something.

Quoted 4 lines

Quoted 4 lines

kernel/signal.c +0 -0

index 664a6c31137e..72182eed1b8d 100644

Quoted 16 lines

Quoted 16 lines

kernel/signal.c +2 -0

@@ -3377,6 +3389,7 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
              to->si_trapno = from->si_trapno;
 #endif
              to->si_addr_lsb = from->si_addr_lsb;
+             to->si_xflags = from->si_xflags;
              break;
      case SIL_FAULT_BNDERR:
              to->si_addr = compat_ptr(from->si_addr);

@@ -3385,6 +3398,7 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
 #endif
              to->si_lower = compat_ptr(from->si_lower);
              to->si_upper = compat_ptr(from->si_upper);
+             to->si_xflags = from->si_xflags;
              break;
      case SIL_FAULT_PKUERR:
              to->si_addr = compat_ptr(from->si_addr);

Quoted 8 lines

kernel/signal.c static int post_copy_siginfo_from_user32(kernel_siginfo_t *to, · line 3409

Quoted 5 lines

kernel/signal.c +1 -0

@@ -3392,6 +3406,7 @@ static int post_copy_siginfo_from_user32(kernel_siginfo_t *to,
              to->si_trapno = from->si_trapno;
 #endif
              to->si_pkey = from->si_pkey;
+             to->si_xflags = from->si_xflags;

How did you figure out the list of places to make these changes? I'm not sure how to confirm that it's exhaustive.

I looked for all places in the kernel where the si_addr field was being assigned to (basically: git grep '\bsi_addr\s*=') and added my code there.

Quoted 2 lines

It's a shame if we can't simply apply the change in one place. Would the refactoring be too invasive to accomplish that?

I believe that Eric mentioned that kernel/signal.c is supposed to be "the" file that knows about all of the fields of siginfo (I'm not sure what's going on in arch/powerpc, I think that's a recent regression). Maybe it's possible to eliminate the duplication between the various SIL_FAULT* cases, but aside from that it looks like there wouldn't be much duplication in the rest of the code.

Quoted 24 lines · Peter Collingbourne

Quoted 23 lines · Dave Martin

Quoted 20 lines · Peter Collingbourne

This field will contain flags that may be used by signal handlers to determine whether other fields in the _sigfault portion of siginfo are valid. An example use case is the following patch, which introduces the si_addr_ignored_bits{,_mask} fields.

A new sigcontext flag, SA_XFLAGS, is introduced in order to allow a signal handler to require the kernel to set the field (but note that the field will be set anyway if the kernel supports the flag, regardless of its value). In combination with the previous patches, this allows a userspace program to determine whether the kernel will set the field.

Ideally this field could have just been named si_flags, but that name was already taken by ia64, so a different name was chosen.

Alternatively, we may consider making ia64's si_flags a generic field and having it appear at the end of _sigfault (in the same place as this patch has si_xflags) on non-ia64, keeping it in the same place on ia64. ia64's si_flags is a 32-bit field with only one flag bit allocated, so we would have 31 bits to use if we do this.

Quoted 6 lines

Quoted 2 lines

For clarity, is the new si_xflags field supposed to be valid for all signal types, or just certain signals and si_codes?

It is intended to be valid for all signal types that use the _sigfault union member of siginfo. As listed in siginfo.h these are: SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT.

SIGSYS is similar to SIGILL, is that included also?

Quoted 6 lines

Quoted 1 line

What happens for things like a rt_sigqueueinfo() from userspace?

Hmm. Let's enumerate each of these things, which I believe are all of the call sites of the function copy_siginfo_from_user and related functions (correct me if I'm wrong):

Quoted 8 lines

- ptrace(PTRACE_SETSIGINFO)
- pidfd_send_signal
- rt_sigqueueinfo
- rt_tgsigqueueinfo

We can handle the last three by observing that the kernel forbids sending a signal with these syscalls if si_code >= 0, so we can say that the value of si_xflags is only valid if si_code >= 0.

Hmmm, that's what the code says (actually >= 0 or SI_TKILL), but it's illogical. Those are user signals, so there's no obvious reason why userspace shouldn't be allowed to generate their siginfo. It would probably be better for the kernel to police si_pid etc. in the SI_USER and SI_TKILL cases rather than flatly refusing, but I guess that's a discussion for another day.

I guess the combination of SI_FROMKERNEL() and the signal number being a known fault signal if probably sufficient for now.

It might be helpful to have a helper to identify fault signals, but we don't have this today, and it's unlikely that a new kind of fault signal will crop up any time soon.

Handlers that handle specific signal types won't care, but debuggers and generic backtracer code would have to be hand-hacked to add new kinds of fault signal today -- not a huge priority though, and orthogonal to this series.

Quoted 8 lines

As for the first one, it's more tricky. Arguably something like a debugger should be able to send arbitrary signals to a debuggee, and there's no reason why it shouldn't be able to set si_xflags in siginfo, but on the other hand who knows how existing debuggers end up setting this field today. Maybe all that we can do is have the kernel clear si_xflags if it detects that the signal uses _sigfault, and let si_xflags aware debuggers opt out of this behavior, perhaps by introducing a PTRACE_SETSIGINFO2 or something.

Most likely a debugger usually amends an existing siginfo from a trapped signal than generating a new one from scratch.

Given the other things that ptrace can do to the target process I don't think we need to police here, but your suggestion about a PTRACE_SETSIGINFO2 or similar, and zeroing this field by default with PTRACE_SETSIGINFO, does make sense.

Quoted 26 lines · Dave Martin

Quoted 24 lines · Peter Collingbourne

Quoted 23 lines · Dave Martin

Quoted 20 lines · Peter Collingbourne

This field will contain flags that may be used by signal handlers to determine whether other fields in the _sigfault portion of siginfo are valid. An example use case is the following patch, which introduces the si_addr_ignored_bits{,_mask} fields.

A new sigcontext flag, SA_XFLAGS, is introduced in order to allow a signal handler to require the kernel to set the field (but note that the field will be set anyway if the kernel supports the flag, regardless of its value). In combination with the previous patches, this allows a userspace program to determine whether the kernel will set the field.

Ideally this field could have just been named si_flags, but that name was already taken by ia64, so a different name was chosen.

Alternatively, we may consider making ia64's si_flags a generic field and having it appear at the end of _sigfault (in the same place as this patch has si_xflags) on non-ia64, keeping it in the same place on ia64. ia64's si_flags is a 32-bit field with only one flag bit allocated, so we would have 31 bits to use if we do this.

Quoted 8 lines

Quoted 6 lines

Quoted 2 lines

For clarity, is the new si_xflags field supposed to be valid for all signal types, or just certain signals and si_codes?

It is intended to be valid for all signal types that use the _sigfault union member of siginfo. As listed in siginfo.h these are: SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT.

SIGSYS is similar to SIGILL, is that included also?

I think that SIGSYS is covered by a separate _sigsys union member.

Quoted 22 lines

Quoted 6 lines

Quoted 1 line

What happens for things like a rt_sigqueueinfo() from userspace?

Hmm. Let's enumerate each of these things, which I believe are all of the call sites of the function copy_siginfo_from_user and related functions (correct me if I'm wrong):

Quoted 8 lines

- ptrace(PTRACE_SETSIGINFO)
- pidfd_send_signal
- rt_sigqueueinfo
- rt_tgsigqueueinfo

We can handle the last three by observing that the kernel forbids sending a signal with these syscalls if si_code >= 0, so we can say that the value of si_xflags is only valid if si_code >= 0.

Hmmm, that's what the code says (actually >= 0 or SI_TKILL), but it's illogical. Those are user signals, so there's no obvious reason why userspace shouldn't be allowed to generate their siginfo. It would probably be better for the kernel to police si_pid etc. in the SI_USER and SI_TKILL cases rather than flatly refusing, but I guess that's a discussion for another day.

Quoted 2 lines

I guess the combination of SI_FROMKERNEL() and the signal number being a known fault signal if probably sufficient for now.

In v10 I ended up adding a comment saying that si_xflags is only valid if 0 <= si_code < SI_KERNEL (the SI_KERNEL part was due to my discovery of kernel code that was calling force_sig(SIGSEGV) where force_sig uses the _kill union member). Your comment about SI_USER

made me realize that is not exactly true (since kill and pidfd_send_signal can send a fault signal with si_code == SI_USER). I was not aware of the SI_FROMKERNEL() macro. In v11 I will update the comment to say that SI_FROMKERNEL(si) && si->si_code != SI_KERNEL must be true in order for si_xflags to be valid.

Quoted 19 lines

It might be helpful to have a helper to identify fault signals, but we don't have this today, and it's unlikely that a new kind of fault signal will crop up any time soon.

Handlers that handle specific signal types won't care, but debuggers and generic backtracer code would have to be hand-hacked to add new kinds of fault signal today -- not a huge priority though, and orthogonal to this series.

Quoted 8 lines

As for the first one, it's more tricky. Arguably something like a debugger should be able to send arbitrary signals to a debuggee, and there's no reason why it shouldn't be able to set si_xflags in siginfo, but on the other hand who knows how existing debuggers end up setting this field today. Maybe all that we can do is have the kernel clear si_xflags if it detects that the signal uses _sigfault, and let si_xflags aware debuggers opt out of this behavior, perhaps by introducing a PTRACE_SETSIGINFO2 or something.

Quoted 2 lines

Most likely a debugger usually amends an existing siginfo from a trapped signal than generating a new one from scratch.

Right, but it could have copied the fields by hand from a kernel-supplied siginfo before amending it.

Quoted 4 lines

Given the other things that ptrace can do to the target process I don't think we need to police here, but your suggestion about a PTRACE_SETSIGINFO2 or similar, and zeroing this field by default with PTRACE_SETSIGINFO, does make sense.

Ack. I've implemented that in v10.

Quoted 60 lines · Peter Collingbourne

Quoted 26 lines · Dave Martin

Quoted 24 lines · Peter Collingbourne

Quoted 23 lines · Dave Martin

Quoted 20 lines · Peter Collingbourne

This field will contain flags that may be used by signal handlers to determine whether other fields in the _sigfault portion of siginfo are valid. An example use case is the following patch, which introduces the si_addr_ignored_bits{,_mask} fields.

A new sigcontext flag, SA_XFLAGS, is introduced in order to allow a signal handler to require the kernel to set the field (but note that the field will be set anyway if the kernel supports the flag, regardless of its value). In combination with the previous patches, this allows a userspace program to determine whether the kernel will set the field.

Ideally this field could have just been named si_flags, but that name was already taken by ia64, so a different name was chosen.

Alternatively, we may consider making ia64's si_flags a generic field and having it appear at the end of _sigfault (in the same place as this patch has si_xflags) on non-ia64, keeping it in the same place on ia64. ia64's si_flags is a 32-bit field with only one flag bit allocated, so we would have 31 bits to use if we do this.

Quoted 8 lines

Quoted 6 lines

Quoted 2 lines

For clarity, is the new si_xflags field supposed to be valid for all signal types, or just certain signals and si_codes?

It is intended to be valid for all signal types that use the _sigfault union member of siginfo. As listed in siginfo.h these are: SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT.

SIGSYS is similar to SIGILL, is that included also?

I think that SIGSYS is covered by a separate _sigsys union member.

Quoted 22 lines

Quoted 6 lines

Quoted 1 line

What happens for things like a rt_sigqueueinfo() from userspace?

Hmm. Let's enumerate each of these things, which I believe are all of the call sites of the function copy_siginfo_from_user and related functions (correct me if I'm wrong):

Quoted 8 lines

- ptrace(PTRACE_SETSIGINFO)
- pidfd_send_signal
- rt_sigqueueinfo
- rt_tgsigqueueinfo

We can handle the last three by observing that the kernel forbids sending a signal with these syscalls if si_code >= 0, so we can say that the value of si_xflags is only valid if si_code >= 0.

Hmmm, that's what the code says (actually >= 0 or SI_TKILL), but it's illogical. Those are user signals, so there's no obvious reason why userspace shouldn't be allowed to generate their siginfo. It would probably be better for the kernel to police si_pid etc. in the SI_USER and SI_TKILL cases rather than flatly refusing, but I guess that's a discussion for another day.

Quoted 7 lines

Quoted 2 lines

I guess the combination of SI_FROMKERNEL() and the signal number being a known fault signal if probably sufficient for now.

In v10 I ended up adding a comment saying that si_xflags is only valid if 0 <= si_code < SI_KERNEL (the SI_KERNEL part was due to my discovery of kernel code that was calling force_sig(SIGSEGV) where force_sig uses the _kill union member). Your comment about SI_USER

Although it's been there a long time, is this a bug?

sigaction(2) says that SI_KERNEL can be reported for any signal, but doesn't say how/why. It also says that si_addr is [unconditionally] valid for [kernel-generated] SIGSEGV. ([] are my insertions).

While it may be reasonable to expect userspace code to filter out user signals before assuming that siginfo fields are value, requiring user code to check for specific si_codes is a bit nastier.

I rather suspect that little or no code out there is explicity checking for SI_KERNEL before assuming that si_addr etc. are valid today.

Quoted 5 lines

made me realize that is not exactly true (since kill and pidfd_send_signal can send a fault signal with si_code == SI_USER). I was not aware of the SI_FROMKERNEL() macro. In v11 I will update the comment to say that SI_FROMKERNEL(si) && si->si_code != SI_KERNEL must be true in order for si_xflags to be valid.

Given the above, maybe it would be better to say nothing explicit about SI_KERNEL, but make sure that the additional siginfo fields are sanely zeroed anyway.

For kernel-generated signals we can guarantee this, so I think that requiring userspace to check explicitly for SI_KERNEL is too unrealistic (i.e., 90% of the time, people will forget ... and 99% of the time they will get away with it).

I'm not sure that SI_FROMKERNEL() is standard btw, but I observed that it is present in the Linux UAPI header.

The "si_code < 0 means userspace" convention seems well established, but I haven't found anywhere that this is clearly specified.

Quoted 22 lines

Quoted 21 lines

It might be helpful to have a helper to identify fault signals, but we don't have this today, and it's unlikely that a new kind of fault signal will crop up any time soon.

Handlers that handle specific signal types won't care, but debuggers and generic backtracer code would have to be hand-hacked to add new kinds of fault signal today -- not a huge priority though, and orthogonal to this series.

Quoted 8 lines

As for the first one, it's more tricky. Arguably something like a debugger should be able to send arbitrary signals to a debuggee, and there's no reason why it shouldn't be able to set si_xflags in siginfo, but on the other hand who knows how existing debuggers end up setting this field today. Maybe all that we can do is have the kernel clear si_xflags if it detects that the signal uses _sigfault, and let si_xflags aware debuggers opt out of this behavior, perhaps by introducing a PTRACE_SETSIGINFO2 or something.

Most likely a debugger usually amends an existing siginfo from a trapped signal than generating a new one from scratch.

Quoted 2 lines

Right, but it could have copied the fields by hand from a kernel-supplied siginfo before amending it.

It could have, but we can't cover every eventuality. With a PTRACE_SETSIGINFO2, we at least get a promise from userspace that it processed the fields correctly. (If userspace is wrong, or lying, that's its own funeral. The ptracer is as good as root from the point of view of the traced process anyway.)

Quoted 76 lines · Dave Martin

Quoted 60 lines · Peter Collingbourne

Quoted 26 lines · Dave Martin

Quoted 24 lines · Peter Collingbourne

Quoted 23 lines · Dave Martin

Quoted 20 lines · Peter Collingbourne

This field will contain flags that may be used by signal handlers to determine whether other fields in the _sigfault portion of siginfo are valid. An example use case is the following patch, which introduces the si_addr_ignored_bits{,_mask} fields.

A new sigcontext flag, SA_XFLAGS, is introduced in order to allow a signal handler to require the kernel to set the field (but note that the field will be set anyway if the kernel supports the flag, regardless of its value). In combination with the previous patches, this allows a userspace program to determine whether the kernel will set the field.

Ideally this field could have just been named si_flags, but that name was already taken by ia64, so a different name was chosen.

Alternatively, we may consider making ia64's si_flags a generic field and having it appear at the end of _sigfault (in the same place as this patch has si_xflags) on non-ia64, keeping it in the same place on ia64. ia64's si_flags is a 32-bit field with only one flag bit allocated, so we would have 31 bits to use if we do this.

Quoted 8 lines

Quoted 6 lines

Quoted 2 lines

For clarity, is the new si_xflags field supposed to be valid for all signal types, or just certain signals and si_codes?

It is intended to be valid for all signal types that use the _sigfault union member of siginfo. As listed in siginfo.h these are: SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT.

SIGSYS is similar to SIGILL, is that included also?

I think that SIGSYS is covered by a separate _sigsys union member.

Quoted 22 lines

Quoted 6 lines

Quoted 1 line

What happens for things like a rt_sigqueueinfo() from userspace?

Hmm. Let's enumerate each of these things, which I believe are all of the call sites of the function copy_siginfo_from_user and related functions (correct me if I'm wrong):

Quoted 8 lines

- ptrace(PTRACE_SETSIGINFO)
- pidfd_send_signal
- rt_sigqueueinfo
- rt_tgsigqueueinfo

We can handle the last three by observing that the kernel forbids sending a signal with these syscalls if si_code >= 0, so we can say that the value of si_xflags is only valid if si_code >= 0.

Hmmm, that's what the code says (actually >= 0 or SI_TKILL), but it's illogical. Those are user signals, so there's no obvious reason why userspace shouldn't be allowed to generate their siginfo. It would probably be better for the kernel to police si_pid etc. in the SI_USER and SI_TKILL cases rather than flatly refusing, but I guess that's a discussion for another day.

Quoted 7 lines

Quoted 2 lines

I guess the combination of SI_FROMKERNEL() and the signal number being a known fault signal if probably sufficient for now.

In v10 I ended up adding a comment saying that si_xflags is only valid if 0 <= si_code < SI_KERNEL (the SI_KERNEL part was due to my discovery of kernel code that was calling force_sig(SIGSEGV) where force_sig uses the _kill union member). Your comment about SI_USER

Although it's been there a long time, is this a bug?

sigaction(2) says that SI_KERNEL can be reported for any signal, but doesn't say how/why. It also says that si_addr is [unconditionally] valid for [kernel-generated] SIGSEGV. ([] are my insertions).

Quoted 6 lines

While it may be reasonable to expect userspace code to filter out user signals before assuming that siginfo fields are value, requiring user code to check for specific si_codes is a bit nastier.

I rather suspect that little or no code out there is explicity checking for SI_KERNEL before assuming that si_addr etc. are valid today.

Right, but maybe that can be attributed to poor documentation (in the man page), so maybe the right thing to do here is to make the documentation more explicit. The kernel code itself is fairly clear that SI_KERNEL does not use the _sigfault layout:
https://github.com/torvalds/linux/blob/6a9dc5fd6170d0a41c8a14eb19e63d94bea5705a/kernel/signal.c#L3173

And note that force_sig does not make the si_addr field valid either, it sets it to 0 (on most architectures, as a result of si_addr overlapping si_pid/si_uid which get set to 0 by that function), which is not necessarily the correct value. For example, on 64-bit x86, executing this code:

  volatile auto ptr = (char *)0xfedcba9876543210;
  *ptr = 42;

(i.e. accessing outside of the TASK_SIZE limit) will result in a call to force_sig(SIGSEGV) setting si_addr=0. But this is clearly not an accurate fault address. I don't know how x86 reports the fault address to the kernel in this case but maybe it simply isn't available for addresses larger than TASK_SIZE, so the right thing for the kernel to do would be to indicate that the address is unavailable (for example, by setting si_code=SI_KERNEL, as it is already doing). Then through documentation updates, userspace can know that si_code=SI_KERNEL means that the address is unavailable.

Quoted 9 lines

Quoted 5 lines

made me realize that is not exactly true (since kill and pidfd_send_signal can send a fault signal with si_code == SI_USER). I was not aware of the SI_FROMKERNEL() macro. In v11 I will update the comment to say that SI_FROMKERNEL(si) && si->si_code != SI_KERNEL must be true in order for si_xflags to be valid.

Given the above, maybe it would be better to say nothing explicit about SI_KERNEL, but make sure that the additional siginfo fields are sanely zeroed anyway.

I think that for si_addr this happens as a result of setting si_pid/si_uid, and for the other fields this happens as a result of zeroing the padding between fields. I know that we'd prefer not to rely on zeroing padding, but perhaps the zero padding can more be seen as a last resort for keeping things working in case userspace fails to check for SI_KERNEL.

Quoted 5 lines

For kernel-generated signals we can guarantee this, so I think that requiring userspace to check explicitly for SI_KERNEL is too unrealistic (i.e., 90% of the time, people will forget ... and 99% of the time they will get away with it).

It's unfortunate that the conditions for accessing these fields are so complex, but again this seems like part of the hand that we've been dealt with this API. Fortunately the requirement to check for SI_KERNEL should only really apply in practice to code accessing our new fields. We can make it retroactively apply to existing fields, but since that wouldn't be a change to the kernel code, just the documentation, existing code will continue to operate in the same way as it did before.

The kernel currently clears the tag bits (i.e. bits 56-63) in the fault address exposed via siginfo.si_addr and sigcontext.fault_address. However, the tag bits may be needed by tools in order to accurately diagnose memory errors, such as HWASan [1] or future tools based on the Memory Tagging Extension (MTE).

We should not stop clearing these bits in the existing fault address fields, because there may be existing userspace applications that are expecting the tag bits to be cleared. Instead, create a new pair of union fields in siginfo._sigfault, and store the tag bits of FAR_EL1 there, together with a mask specifying which bits are valid.

A flag is added to si_xflags to allow userspace to determine whether the values in the fields are valid.

[1] http://clang.llvm.org/docs/HardwareAssistedAddressSanitizerDesign.html

Signed-off-by Peter Collingbourne <[email protected]>

---
View this change in Gerrit: https://linux-review.googlesource.com/q/Ia8876bad8c798e0a32df7c2ce1256c4771c81446

v9:
- make the ignored bits fields generic
- add some new dependent patches that prepare us to store the field in such a way that userspace can detect their presence

v8:
- rebase onto 5.8rc2

v7:
- switch to a new siginfo field instead of using sigcontext
- merge the patch back into one since the other patches are now unnecessary

v6:
- move fault address and fault code into the kernel_siginfo data structure
- split the patch in three since it was getting large and now has generic and arch-specific parts

v5:
- add padding to fault_addr_top_byte_context in order to ensure the correct size and preserve sp alignment

v4:
- expose only the tag bits in the context instead of the entire FAR_EL1
- remove mention of the new context from the sigcontext.__reserved[] note

v3:
- add documentation to tagged-pointers.rst
- update comments in sigcontext.h

v2:
- revert changes to hw_breakpoint.c
- rename set_thread_esr to set_thread_far_esr

 Documentation/arm64/tagged-pointers.rst | 21 +++++++---
 arch/arm64/include/asm/exception.h      |  2 +-
 arch/arm64/include/asm/traps.h          |  7 +++-
 arch/arm64/kernel/debug-monitors.c      |  4 +-
 arch/arm64/kernel/entry-common.c        |  2 -
 arch/arm64/kernel/ptrace.c              |  2 +-
 arch/arm64/kernel/traps.c               | 15 ++++---
 arch/arm64/mm/fault.c                   | 54 ++++++++++++++-----------
 arch/mips/kernel/traps.c                |  2 +-
 arch/parisc/kernel/ptrace.c             |  2 +-
 arch/parisc/mm/fault.c                  |  2 +-
 arch/powerpc/mm/fault.c                 |  2 +-
 arch/x86/mm/fault.c                     |  3 +-
 include/linux/compat.h                  |  2 +
 include/linux/sched/signal.h            | 12 +++++-
 include/uapi/asm-generic/siginfo.h      | 10 +++++
 kernel/signal.c                         | 50 +++++++++++++++++++----
 mm/memory-failure.c                     |  2 +-
 18 files changed, 136 insertions(+), 58 deletions(-)

Documentation/arm64/tagged-pointers.rst +15 -6

index eab4323609b9..bd23039841a5 100644
@@ -53,12 +53,21 @@ visibility.
 Preserving tags
 ---------------
 
-Non-zero tags are not preserved when delivering signals. This means that
-signal handlers in applications making use of tags cannot rely on the
-tag information for user virtual addresses being maintained for fields
-inside siginfo_t. One exception to this rule is for signals raised in
-response to watchpoint debug exceptions, where the tag information will
-be preserved.
+Non-zero tags are not preserved in the fault address fields
+siginfo.si_addr or sigcontext.fault_address when delivering
+signals. This means that signal handlers in applications making use
+of tags cannot rely on the tag information for user virtual addresses
+being maintained in these fields. One exception to this rule is for
+signals raised in response to watchpoint debug exceptions, where the
+tag information will be preserved.
+
+The fault address tag is preserved in the si_addr_ignored_bits field
+of siginfo, which is set for signals raised in response to data aborts
+and instruction aborts. The si_addr_ignored_bits_mask field indicates
+which bits of the field are valid. The validity of these fields is
+indicated by the SI_XFLAG_IGNORED_BITS flag in siginfo.si_xflags,
+and the validity of si_xflags in turn is indicated by the kernel
+indicating support for the sigaction.sa_flags flag SA_XFLAGS.
 
 The architecture prevents the use of a tagged PC, so the upper byte will
 be set to a sign-extension of bit 55 on exception return.

arch/arm64/include/asm/exception.h +1 -1

index 7577a754d443..950d55dae948 100644
@@ -32,7 +32,7 @@ static inline u32 disr_to_esr(u64 disr)
 }
 
 asmlinkage void enter_from_user_mode(void);
-void do_mem_abort(unsigned long addr, unsigned int esr, struct pt_regs *regs);
+void do_mem_abort(unsigned long far, unsigned int esr, struct pt_regs *regs);
 void do_undefinstr(struct pt_regs *regs);
 void do_bti(struct pt_regs *regs);
 asmlinkage void bad_mode(struct pt_regs *regs, int reason, unsigned int esr);

arch/arm64/include/asm/traps.h +5 -2

index cee5928e1b7d..8e4f6c5b97af 100644
@@ -26,8 +26,11 @@ void register_undef_hook(struct undef_hook *hook);
 void unregister_undef_hook(struct undef_hook *hook);
 void force_signal_inject(int signal, int code, unsigned long address);
 void arm64_notify_segfault(unsigned long addr);
-void arm64_force_sig_fault(int signo, int code, void __user *addr, const char *str);
-void arm64_force_sig_mceerr(int code, void __user *addr, short lsb, const char *str);
+void arm64_force_sig_fault(int signo, int code, void __user *addr,
+			   unsigned long far, unsigned char far_tb_mask,
+			   const char *str);
+void arm64_force_sig_mceerr(int code, void __user *addr, short lsb,
+			    unsigned long far, const char *str);
 void arm64_force_sig_ptrace_errno_trap(int errno, void __user *addr, const char *str);
 
 /*

arch/arm64/kernel/debug-monitors.c +2 -2

index 7310a4f7f993..bc9e02fbb710 100644
@@ -235,8 +235,8 @@ static void send_user_sigtrap(int si_code)
 		local_irq_enable();
 
 	arm64_force_sig_fault(SIGTRAP, si_code,
-			     (void __user *)instruction_pointer(regs),
-			     "User debug trap");
+			      (void __user *)instruction_pointer(regs), 0, 0,
+			      "User debug trap");
 }
 
 static int single_step_handler(unsigned long unused, unsigned int esr,

arch/arm64/kernel/entry-common.c +0 -2

index d3be9dbf5490..65ed01606480 100644
@@ -22,7 +22,6 @@ static void notrace el1_abort(struct pt_regs *regs, unsigned long esr)
 	unsigned long far = read_sysreg(far_el1);
 
 	local_daif_inherit(regs);
-	far = untagged_addr(far);
 	do_mem_abort(far, esr, regs);
 }
 NOKPROBE_SYMBOL(el1_abort);

@@ -104,7 +103,6 @@ static void notrace el0_da(struct pt_regs *regs, unsigned long esr)
 
 	user_exit_irqoff();
 	local_daif_restore(DAIF_PROCCTX);
-	far = untagged_addr(far);
 	do_mem_abort(far, esr, regs);
 }
 NOKPROBE_SYMBOL(el0_da);

arch/arm64/kernel/ptrace.c +1 -1

index d8ebfd813e28..c73312064220 100644
@@ -197,7 +197,7 @@ static void ptrace_hbptriggered(struct perf_event *bp,
 	}
 #endif
 	arm64_force_sig_fault(SIGTRAP, TRAP_HWBKPT,
-			      (void __user *)(bkpt->trigger),
+			      (void __user *)(bkpt->trigger), 0, 0,
 			      desc);
 }
 

arch/arm64/kernel/traps.c +10 -5

index 13ebd5ca2070..c945c5cc1312 100644
@@ -235,20 +235,25 @@ static void arm64_show_signal(int signo, const char *str)
 }
 
 void arm64_force_sig_fault(int signo, int code, void __user *addr,
+			   unsigned long far, unsigned char far_tb_mask,
 			   const char *str)
 {
+	unsigned long ignored_bits_mask = ((unsigned long)far_tb_mask) << 56;
+
 	arm64_show_signal(signo, str);
 	if (signo == SIGKILL)
 		force_sig(SIGKILL);
 	else
-		force_sig_fault(signo, code, addr);
+		force_sig_fault_with_ignored_bits(signo, code, addr,
+						  far & ignored_bits_mask,
+						  ignored_bits_mask);
 }
 
 void arm64_force_sig_mceerr(int code, void __user *addr, short lsb,
-			    const char *str)
+			    unsigned long far, const char *str)
 {
 	arm64_show_signal(SIGBUS, str);
-	force_sig_mceerr(code, addr, lsb);
+	force_sig_mceerr(code, addr, lsb, far & (0xffUL << 56), 0xffUL << 56);
 }
 
 void arm64_force_sig_ptrace_errno_trap(int errno, void __user *addr,

@@ -267,7 +272,7 @@ void arm64_notify_die(const char *str, struct pt_regs *regs,
 		current->thread.fault_address = 0;
 		current->thread.fault_code = err;
 
-		arm64_force_sig_fault(signo, sicode, addr, str);
+		arm64_force_sig_fault(signo, sicode, addr, 0, 0, str);
 	} else {
 		die(str, regs, err);
 	}

@@ -829,7 +834,7 @@ void bad_el0_sync(struct pt_regs *regs, int reason, unsigned int esr)
 	current->thread.fault_address = 0;
 	current->thread.fault_code = esr;
 
-	arm64_force_sig_fault(SIGILL, ILL_ILLOPC, pc,
+	arm64_force_sig_fault(SIGILL, ILL_ILLOPC, pc, 0, 0,
 			      "Bad EL0 synchronous exception");
 }
 

arch/arm64/mm/fault.c +30 -24

index f07333e86c2f..82f05644417e 100644
@@ -40,7 +40,7 @@
 #include <asm/traps.h>
 
 struct fault_info {
-	int	(*fn)(unsigned long addr, unsigned int esr,
+	int	(*fn)(unsigned long far, unsigned int esr,
 		      struct pt_regs *regs);
 	int	sig;
 	int	code;

@@ -383,8 +383,11 @@ static void set_thread_esr(unsigned long address, unsigned int esr)
 	current->thread.fault_code = esr;
 }
 
-static void do_bad_area(unsigned long addr, unsigned int esr, struct pt_regs *regs)
+static void do_bad_area(unsigned long far, unsigned int esr,
+			struct pt_regs *regs)
 {
+	unsigned long addr = untagged_addr(far);
+
 	/*
 	 * If we are in kernel mode at this point, we have no context to
 	 * handle this fault with.

@@ -394,7 +397,7 @@ static void do_bad_area(unsigned long addr, unsigned int esr, struct pt_regs *re
 
 		set_thread_esr(addr, esr);
 		arm64_force_sig_fault(inf->sig, inf->code, (void __user *)addr,
-				      inf->name);
+				      far, 0xff, inf->name);
 	} else {
 		__do_kernel_fault(addr, esr, regs);
 	}

@@ -446,7 +449,7 @@ static bool is_write_abort(unsigned int esr)
 	return (esr & ESR_ELx_WNR) && !(esr & ESR_ELx_CM);
 }
 
-static int __kprobes do_page_fault(unsigned long addr, unsigned int esr,
+static int __kprobes do_page_fault(unsigned long far, unsigned int esr,
 				   struct pt_regs *regs)
 {
 	const struct fault_info *inf;

@@ -454,6 +457,7 @@ static int __kprobes do_page_fault(unsigned long addr, unsigned int esr,
 	vm_fault_t fault;
 	unsigned long vm_flags = VM_ACCESS_FLAGS;
 	unsigned int mm_flags = FAULT_FLAG_DEFAULT;
+	unsigned long addr = untagged_addr(far);
 
 	if (kprobe_page_fault(regs, esr))
 		return 0;

@@ -566,7 +570,7 @@ static int __kprobes do_page_fault(unsigned long addr, unsigned int esr,
 		 * this page fault.
 		 */
 		arm64_force_sig_fault(SIGBUS, BUS_ADRERR, (void __user *)addr,
-				      inf->name);
+				      far, 0xff, inf->name);
 	} else if (fault & (VM_FAULT_HWPOISON_LARGE | VM_FAULT_HWPOISON)) {
 		unsigned int lsb;
 

@@ -575,7 +579,7 @@ static int __kprobes do_page_fault(unsigned long addr, unsigned int esr,
 			lsb = hstate_index_to_shift(VM_FAULT_GET_HINDEX(fault));
 
 		arm64_force_sig_mceerr(BUS_MCEERR_AR, (void __user *)addr, lsb,
-				       inf->name);
+				       far, inf->name);
 	} else {
 		/*
 		 * Something tried to access memory that isn't in our memory

@@ -583,8 +587,7 @@ static int __kprobes do_page_fault(unsigned long addr, unsigned int esr,
 		 */
 		arm64_force_sig_fault(SIGSEGV,
 				      fault == VM_FAULT_BADACCESS ? SEGV_ACCERR : SEGV_MAPERR,
-				      (void __user *)addr,
-				      inf->name);
+				      (void __user *)addr, far, 0xff, inf->name);
 	}
 
 	return 0;

@@ -594,30 +597,32 @@ static int __kprobes do_page_fault(unsigned long addr, unsigned int esr,
 	return 0;
 }
 
-static int __kprobes do_translation_fault(unsigned long addr,
+static int __kprobes do_translation_fault(unsigned long far,
 					  unsigned int esr,
 					  struct pt_regs *regs)
 {
+	unsigned long addr = untagged_addr(far);
+
 	if (is_ttbr0_addr(addr))
-		return do_page_fault(addr, esr, regs);
+		return do_page_fault(far, esr, regs);
 
-	do_bad_area(addr, esr, regs);
+	do_bad_area(far, esr, regs);
 	return 0;
 }
 
-static int do_alignment_fault(unsigned long addr, unsigned int esr,
+static int do_alignment_fault(unsigned long far, unsigned int esr,
 			      struct pt_regs *regs)
 {
-	do_bad_area(addr, esr, regs);
+	do_bad_area(far, esr, regs);
 	return 0;
 }
 
-static int do_bad(unsigned long addr, unsigned int esr, struct pt_regs *regs)
+static int do_bad(unsigned long far, unsigned int esr, struct pt_regs *regs)
 {
 	return 1; /* "fault" */
 }
 
-static int do_sea(unsigned long addr, unsigned int esr, struct pt_regs *regs)
+static int do_sea(unsigned long far, unsigned int esr, struct pt_regs *regs)
 {
 	const struct fault_info *inf;
 	void __user *siaddr;

@@ -635,7 +640,7 @@ static int do_sea(unsigned long addr, unsigned int esr, struct pt_regs *regs)
 	if (esr & ESR_ELx_FnV)
 		siaddr = NULL;
 	else
-		siaddr  = (void __user *)addr;
+		siaddr  = (void __user *)untagged_addr(far);
 	arm64_notify_die(inf->name, regs, inf->sig, inf->code, siaddr, esr);
 
 	return 0;

@@ -708,11 +713,12 @@ static const struct fault_info fault_info[] = {
 	{ do_bad,		SIGKILL, SI_KERNEL,	"unknown 63"			},
 };
 
-void do_mem_abort(unsigned long addr, unsigned int esr, struct pt_regs *regs)
+void do_mem_abort(unsigned long far, unsigned int esr, struct pt_regs *regs)
 {
 	const struct fault_info *inf = esr_to_fault_info(esr);
+	unsigned long addr = untagged_addr(far);
 
-	if (!inf->fn(addr, esr, regs))
+	if (!inf->fn(far, esr, regs))
 		return;
 
 	if (!user_mode(regs)) {

@@ -721,8 +727,8 @@ void do_mem_abort(unsigned long addr, unsigned int esr, struct pt_regs *regs)
 		show_pte(addr);
 	}
 
-	arm64_notify_die(inf->name, regs,
-			 inf->sig, inf->code, (void __user *)addr, esr);
+	arm64_notify_die(inf->name, regs, inf->sig, inf->code,
+			 (void __user *)addr, esr);
 }
 NOKPROBE_SYMBOL(do_mem_abort);
 

@@ -735,8 +741,8 @@ NOKPROBE_SYMBOL(do_el0_irq_bp_hardening);
 
 void do_sp_pc_abort(unsigned long addr, unsigned int esr, struct pt_regs *regs)
 {
-	arm64_notify_die("SP/PC alignment exception", regs,
-			 SIGBUS, BUS_ADRALN, (void __user *)addr, esr);
+	arm64_notify_die("SP/PC alignment exception", regs, SIGBUS, BUS_ADRALN,
+			 (void __user *)addr, esr);
 }
 NOKPROBE_SYMBOL(do_sp_pc_abort);
 

@@ -862,8 +868,8 @@ void do_debug_exception(unsigned long addr_if_watchpoint, unsigned int esr,
 		arm64_apply_bp_hardening();
 
 	if (inf->fn(addr_if_watchpoint, esr, regs)) {
-		arm64_notify_die(inf->name, regs,
-				 inf->sig, inf->code, (void __user *)pc, esr);
+		arm64_notify_die(inf->name, regs, inf->sig, inf->code,
+				 (void __user *)pc, esr);
 	}
 
 	debug_exception_exit(regs);

arch/mips/kernel/traps.c +1 -1

index 38aa07ccdbcc..73a181c904b0 100644
@@ -775,7 +775,7 @@ void force_fcr31_sig(unsigned long fcr31, void __user *fault_addr,
 	else if (fcr31 & FPU_CSR_INE_X)
 		si_code = FPE_FLTRES;
 
-	force_sig_fault_to_task(SIGFPE, si_code, fault_addr, tsk);
+	force_sig_fault_to_task(SIGFPE, si_code, fault_addr, 0, 0, tsk);
 }
 
 int process_fpemu_return(int sig, void __user *fault_addr, unsigned long fcr31)

arch/parisc/kernel/ptrace.c +1 -1

index 2127974982df..62c8bb0d7d31 100644
@@ -89,7 +89,7 @@ void user_enable_single_step(struct task_struct *task)
 		   parent know something happened. */
 		force_sig_fault_to_task(SIGTRAP, TRAP_TRACE,
 					(void __user *) (task_regs(task)->iaoq[0] & ~3),
-					task);
+					0, 0, task);
 		/* notify_parent(task, SIGCHLD); */
 		return;
 	}

arch/parisc/mm/fault.c +1 -1

index 4bfe2da9fbe3..2a6064ea2dfc 100644
@@ -400,7 +400,7 @@ void do_page_fault(struct pt_regs *regs, unsigned long code,
 				lsb = PAGE_SHIFT;
 
 			force_sig_mceerr(BUS_MCEERR_AR, (void __user *) address,
-					 lsb);
+					 lsb, 0, 0);
 			return;
 		}
 #endif

arch/powerpc/mm/fault.c +1 -1

index 0add963a849b..4ffe2fb8e31a 100644
@@ -152,7 +152,7 @@ static int do_sigbus(struct pt_regs *regs, unsigned long address,
 		if (fault & VM_FAULT_HWPOISON)
 			lsb = PAGE_SHIFT;
 
-		force_sig_mceerr(BUS_MCEERR_AR, (void __user *)address, lsb);
+		force_sig_mceerr(BUS_MCEERR_AR, (void __user *)address, lsb, 0, 0);
 		return 0;
 	}
 

arch/x86/mm/fault.c +2 -1

index 35f1498e9832..a443eec157f6 100644
@@ -898,7 +898,8 @@ do_sigbus(struct pt_regs *regs, unsigned long error_code, unsigned long address,
 			lsb = hstate_index_to_shift(VM_FAULT_GET_HINDEX(fault));
 		if (fault & VM_FAULT_HWPOISON)
 			lsb = PAGE_SHIFT;
-		force_sig_mceerr(BUS_MCEERR_AR, (void __user *)address, lsb);
+		force_sig_mceerr(BUS_MCEERR_AR, (void __user *)address, lsb, 0,
+				 0);
 		return;
 	}
 #endif

include/linux/compat.h +2 -0

index 55d4228dfd88..273146cf30fd 100644
@@ -234,6 +234,8 @@ typedef struct compat_siginfo {
 				compat_uptr_t _pad[6];
 			};
 			compat_uptr_t _xflags;
+			compat_uptr_t _addr_ignored_bits;
+			compat_uptr_t _addr_ignored_bits_mask;
 		} _sigfault;
 
 		/* SIGPOLL */

include/linux/sched/signal.h +10 -2

index 1bad18a1d8ba..f62a014a5a4d 100644
@@ -311,16 +311,24 @@ static inline void kernel_signal_stop(void)
 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);
+	, unsigned long addr_ignored_bits, unsigned long addr_ignored_bits_mask,
+	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 force_sig_fault_with_ignored_bits(int sig, int code, void __user *addr
+	___ARCH_SI_TRAPNO(int trapno)
+	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr)
+	, unsigned long addr_ignored_bits,
+	unsigned long addr_ignored_bits_mask);
 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);
 
-int force_sig_mceerr(int code, void __user *, short);
+int force_sig_mceerr(int code, void __user *addr, short addr_lsb,
+		     unsigned long addr_ignored_bits,
+		     unsigned long addr_ignored_bits_mask);
 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);

include/uapi/asm-generic/siginfo.h +10 -0

index 413d804623c0..1fc6482b0ad4 100644
@@ -94,6 +94,8 @@ union __sifields {
 			void *_pad[6];
 		};
 		unsigned long _xflags;
+		unsigned long _addr_ignored_bits;
+		unsigned long _addr_ignored_bits_mask;
 	} _sigfault;
 
 	/* SIGPOLL */

@@ -155,6 +157,8 @@ typedef struct siginfo {
 #endif
 #define si_addr_lsb	_sifields._sigfault._addr_lsb
 #define si_xflags	_sifields._sigfault._xflags
+#define si_addr_ignored_bits		_sifields._sigfault._addr_ignored_bits
+#define si_addr_ignored_bits_mask	_sifields._sigfault._addr_ignored_bits_mask
 #define si_lower	_sifields._sigfault._addr_bnd._lower
 #define si_upper	_sifields._sigfault._addr_bnd._upper
 #define si_pkey		_sifields._sigfault._addr_pkey._pkey

@@ -295,6 +299,12 @@ typedef struct siginfo {
 #define EMT_TAGOVF	1	/* tag overflow */
 #define NSIGEMT		1
 
+/*
+ * SIGILL, SIGFPE, SIGSEGV, SIGBUS, SIGTRAP, SIGEMT si_xflags
+ */
+#define SI_XFLAG_IGNORED_BITS	1
+/* si_addr_ignored_bits{,_mask} fields valid */
+
 /*
  * sigevent definitions
  * 

kernel/signal.c +21 -5

index 72182eed1b8d..1f1e42adc57d 100644
@@ -1652,7 +1652,8 @@ 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)
+	, unsigned long addr_ignored_bits, unsigned long addr_ignored_bits_mask,
+	struct task_struct *t)
 {
 	struct kernel_siginfo info;
 

@@ -1669,17 +1670,32 @@ int force_sig_fault_to_task(int sig, int code, void __user *addr
 	info.si_flags = flags;
 	info.si_isr = isr;
 #endif
-	info.si_xflags = 0;
+	info.si_xflags = SI_XFLAG_IGNORED_BITS;
+	info.si_addr_ignored_bits = addr_ignored_bits;
+	info.si_addr_ignored_bits_mask = addr_ignored_bits_mask;
 	return force_sig_info_to_task(&info, t);
 }
 
-int force_sig_fault(int sig, int code, void __user *addr
+int force_sig_fault_with_ignored_bits(int sig, int code, void __user *addr
 	___ARCH_SI_TRAPNO(int trapno)
-	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr))
+	___ARCH_SI_IA64(int imm, unsigned int flags, unsigned long isr),
+	unsigned long addr_ignored_bits, unsigned long addr_ignored_bits_mask)
 {
 	return force_sig_fault_to_task(sig, code, addr
 				       ___ARCH_SI_TRAPNO(trapno)
-				       ___ARCH_SI_IA64(imm, flags, isr), current);
+				       ___ARCH_SI_IA64(imm, flags, isr),
+				       addr_ignored_bits,
+				       addr_ignored_bits_mask, current);
+}
+
+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_with_ignored_bits(sig, code, addr
+						 ___ARCH_SI_TRAPNO(trapno)
+						 ___ARCH_SI_IA64(imm, flags,
+								 isr), 0, 0);
 }
 
 int send_sig_fault(int sig, int code, void __user *addr