Skip to contentSkip to conversation

Add kernel seccomp support for m68k

PATCH Superseded Michael Schmitz 4 participants 10 messages Jan 27, 2022 – May 9, 2022

Subsystems (4)

v10 of 13 Superseded Latest is v13 →

lore
Share & tools
lore
#0 lore raw

v10 cover Add kernel seccomp support for m68k

Patch 3 adds the test for TIF_SECCOMP thread info flag to call into syscall_trace_enter() when seccomp syscall filtering is active, which was missing from earlier versions (classic MMU m68k only - need some help with the test for 68000 and coldfire).

The previous version required the TIF_SYSCALL_TRACE flag to be set for syscall filtering to work

The rest of the series remains unchanged from v9.

Tested on ARAnyM, 63 out of 89 seccomp_bpf tests now pass (33 with the old version).

I hope you didn't waste too much effort on testing v8/v9, Adrian. Should have a suitably large brown paper bag somewhere yet :-(

Cheers,

Michael

#1 lore raw

PATCH v10 1/3 m68k/kernel - wire up syscall_trace_enter/leave for m68k

m68k (other than Coldfire) uses syscall_trace for both trace entry and trace exit. Seccomp support requires separate entry points for trace entry and exit which are already provided for Coldfire.

Replace syscall_trace by syscall_trace_enter and syscall_trace_leave in preparation for seccomp support.

No regression seen in testing with strace on ARAnyM.

--
Changes from v8:

- adapt ptrace.c patch to recent changes by Eric Biederman

Changes from v7:

Al Viro:
- split first patch to separate the switch to syscall_trace_enter/leave() from return code checks

Changes from v6:

Geert Uytterhoeven:
- add syscall_trace_enter() return code check for 68000 and coldfire

Changes from v5:

- add comment to explain optimization

Changes from v4:

Andreas Schwab:
- optimize return code test (addql #1,%d0 for cmpil #-1,%d0)
- spelling fix in commit message

Changes from v3:

- change syscall_trace_enter return code test from !=0 to ==-1

revert syscall trace return code checks

---
 arch/m68k/kernel/entry.S  | 4 ++--
 arch/m68k/kernel/ptrace.c | 7 -------
 2 files changed, 2 insertions(+), 9 deletions(-)
arch/m68k/kernel/entry.S +2 -2
index 9434fca68de5..18f278bdbd21 100644
@@ -181,7 +181,7 @@ do_trace_entry:
 	movel	#-ENOSYS,%sp@(PT_OFF_D0)| needed for strace
 	subql	#4,%sp
 	SAVE_SWITCH_STACK
-	jbsr	syscall_trace
+	jbsr	syscall_trace_enter
 	RESTORE_SWITCH_STACK
 	addql	#4,%sp
 	movel	%sp@(PT_OFF_ORIG_D0),%d0
@@ -194,7 +194,7 @@ badsys:
 do_trace_exit:
 	subql	#4,%sp
 	SAVE_SWITCH_STACK
-	jbsr	syscall_trace
+	jbsr	syscall_trace_leave
 	RESTORE_SWITCH_STACK
 	addql	#4,%sp
 	jra	.Lret_from_exception
arch/m68k/kernel/ptrace.c +0 -7
index aa3a0b8d07e9..74d58a82a135 100644
@@ -271,12 +271,6 @@ long arch_ptrace(struct task_struct *child, long request,
 	return -EIO;
 }
 
-asmlinkage void syscall_trace(void)
-{
-	ptrace_report_syscall(0);
-}
-
-#if defined(CONFIG_COLDFIRE) || !defined(CONFIG_MMU)
 asmlinkage int syscall_trace_enter(void)
 {
 	int ret = 0;
@@ -291,4 +285,3 @@ asmlinkage void syscall_trace_leave(void)
 	if (test_thread_flag(TIF_SYSCALL_TRACE))
 		tracehook_report_syscall_exit(task_pt_regs(current), 0);
 }
-#endif /* CONFIG_COLDFIRE */
Signature · 1 line
2.17.1
#3 lore raw

PATCH v10 2/3 m68k/kernel - check syscall_trace_enter() return code on m68k

In reply to Michael Schmitz

Check return code of syscall_trace_enter(), and skip syscall if -1. Return code will be left at what had been set by ptrace or seccomp (in regs->d0).

No regression seen in testing with strace on ARAnyM.

--
Changes from v7:

Al Viro:
- split return code checks from switch to syscall_trace_enter()

Changes from v6:

Geert Uytterhoeven:
- add syscall_trace_enter() return code check for 68000 and coldfire

Changes from v5:

- add comment to explain optimization

Changes from v4:

Andreas Schwab:
- optimize return code test (addql #1,%d0 for cmpil #-1,%d0)
- spelling fix in commit message

Changes from v3:

- change syscall_trace_enter return code test from !=0 to ==-1

---
 arch/m68k/68000/entry.S    | 2 ++
 arch/m68k/coldfire/entry.S | 2 ++
 arch/m68k/kernel/entry.S   | 2 ++
 3 files changed, 6 insertions(+)
arch/m68k/68000/entry.S +2 -0
index 997b54933015..7d63e2f1555a 100644
@@ -45,6 +45,8 @@ do_trace:
 	jbsr	syscall_trace_enter
 	RESTORE_SWITCH_STACK
 	addql	#4,%sp
+	addql	#1,%d0
+	jeq	ret_from_exception
 	movel	%sp@(PT_OFF_ORIG_D0),%d1
 	movel	#-ENOSYS,%d0
 	cmpl	#NR_syscalls,%d1
arch/m68k/coldfire/entry.S +2 -0
index 9f337c70243a..35104c5417ff 100644
@@ -90,6 +90,8 @@ ENTRY(system_call)
 	jbsr	syscall_trace_enter
 	RESTORE_SWITCH_STACK
 	addql	#4,%sp
+	addql	#1,%d0
+	jeq	ret_from_exception
 	movel	%d3,%a0
 	jbsr	%a0@
 	movel	%d0,%sp@(PT_OFF_D0)		/* save the return value */
arch/m68k/kernel/entry.S +2 -0
index 18f278bdbd21..0d5b7085d76f 100644
@@ -184,6 +184,8 @@ do_trace_entry:
 	jbsr	syscall_trace_enter
 	RESTORE_SWITCH_STACK
 	addql	#4,%sp
+	addql	#1,%d0			| optimization for cmpil #-1,%d0
+	jeq	ret_from_syscall
 	movel	%sp@(PT_OFF_ORIG_D0),%d0
 	cmpl	#NR_syscalls,%d0
 	jcs	syscall
Signature · 1 line
2.17.1
#4 lore raw

PATCH v10 3/3 m68k: add kernel seccomp support

In reply to Michael Schmitz

Add secure_computing() call to syscall_trace_enter to actually filter system calls.

Add necessary arch Kconfig options, define TIF_SECCOMP trace flag and provide basic seccomp filter support in asm/syscall.h

syscall_get_nr currently uses the syscall nr stored in orig_d0 because we change d0 to a default return code before starting a syscall trace. This may be inconsistent with syscall_rollback copying orig_d0 to d0 (which we never check upon return from trace). We use d0 for the return code from syscall_trace_enter in entry.S currently, and could perhaps expand that to store a new syscall number returned by the seccomp filter before executing the syscall. This clearly needs some discussion.

seccomp_bpf self test on ARAnyM passes 63 out of 89 tests.

--
Changes from v9:

- add test for TIF_SECCOMP bit to call syscall_trace_enter() if seccomp syscall filtering is active

Changes from v6:

Geert Uytterhoeven:
- add syscall_get_error(), syscall_get_return_value(), and syscall_set_arguments() (not needed to compile!)

Changes from v5:

Geert Uytterhoeven:
- correct wrong offset for d1-d5 register copy
- update Documentation/features/seccomp/seccomp-filter/arch-support.txt

add syscall_get_error(), syscall_get_return_value(), syscall_set_arguments()

---
 .../seccomp/seccomp-filter/arch-support.txt   |  2 +-
 arch/m68k/Kconfig                             |  2 +
 arch/m68k/include/asm/seccomp.h               | 11 ++++
 arch/m68k/include/asm/syscall.h               | 56 +++++++++++++++++++
 arch/m68k/include/asm/thread_info.h           |  2 +
 arch/m68k/kernel/entry.S                      |  3 +
 arch/m68k/kernel/ptrace.c                     |  5 ++
 7 files changed, 80 insertions(+), 1 deletion(-)
 create mode 100644 arch/m68k/include/asm/seccomp.h
Documentation/features/seccomp/seccomp-filter/arch-support.txt +1 -1
index 26eec58ab819..be71f2066981 100644
@@ -14,7 +14,7 @@
     |       h8300: | TODO |
     |     hexagon: | TODO |
     |        ia64: | TODO |
-    |        m68k: | TODO |
+    |        m68k: |  ok  |
     |  microblaze: | TODO |
     |        mips: |  ok  |
     |       nds32: | TODO |
arch/m68k/Kconfig +2 -0
index 936e1803c7c7..c0ef187fb3ed 100644
@@ -18,6 +18,8 @@ config M68K
 	select GENERIC_IOMAP
 	select GENERIC_IRQ_SHOW
 	select HAVE_AOUT if MMU
+	select HAVE_ARCH_SECCOMP
+	select HAVE_ARCH_SECCOMP_FILTER
 	select HAVE_ASM_MODVERSIONS
 	select HAVE_DEBUG_BUGVERBOSE
 	select HAVE_EFFICIENT_UNALIGNED_ACCESS if !CPU_HAS_NO_UNALIGNED
arch/m68k/include/asm/seccomp.h new file +11 -0
new file mode 100644
index 000000000000..de8a94e1fb3f
@@ -0,0 +1,11 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+#ifndef _ASM_SECCOMP_H
+#define _ASM_SECCOMP_H
+
+#include <asm-generic/seccomp.h>
+
+#define SECCOMP_ARCH_NATIVE		AUDIT_ARCH_M68K
+#define SECCOMP_ARCH_NATIVE_NR		NR_syscalls
+#define SECCOMP_ARCH_NATIVE_NAME	"m68k"
+
+#endif /* _ASM_SECCOMP_H */
arch/m68k/include/asm/syscall.h +56 -0

62 lines not on this page · raw message

arch/m68k/include/asm/thread_info.h +2 -0
index c952658ba792..31be2ad999ca 100644
@@ -61,6 +61,7 @@ static inline struct thread_info *current_thread_info(void)
 #define TIF_NOTIFY_RESUME	5	/* callback before returning to user */
 #define TIF_SIGPENDING		6	/* signal pending */
 #define TIF_NEED_RESCHED	7	/* rescheduling necessary */
+#define TIF_SECCOMP		13	/* seccomp syscall filtering active */
 #define TIF_DELAYED_TRACE	14	/* single step a syscall */
 #define TIF_SYSCALL_TRACE	15	/* syscall trace active */
 #define TIF_MEMDIE		16	/* is terminating due to OOM killer */
@@ -69,6 +70,7 @@ static inline struct thread_info *current_thread_info(void)
 #define _TIF_NOTIFY_RESUME	(1 << TIF_NOTIFY_RESUME)
 #define _TIF_SIGPENDING		(1 << TIF_SIGPENDING)
 #define _TIF_NEED_RESCHED	(1 << TIF_NEED_RESCHED)
+#define _TIF_SECCOMP		(1 << TIF_SECCOMP)
 #define _TIF_DELAYED_TRACE	(1 << TIF_DELAYED_TRACE)
 #define _TIF_SYSCALL_TRACE	(1 << TIF_SYSCALL_TRACE)
 #define _TIF_MEMDIE		(1 << TIF_MEMDIE)
arch/m68k/kernel/entry.S +3 -0
index 0d5b7085d76f..ee0a905b5f74 100644
@@ -213,6 +213,9 @@ ENTRY(system_call)
 	| syscall trace?
 	tstb	%a1@(TINFO_FLAGS+2)
 	jmi	do_trace_entry
+	| seccomp filter active?
+	btst	#5,%a1@(TINFO_FLAGS+2)
+	bnes	do_trace_entry
 	cmpl	#NR_syscalls,%d0
 	jcc	badsys
 syscall:
arch/m68k/kernel/ptrace.c +5 -0
index 74d58a82a135..bc2490c3fb52 100644
@@ -19,6 +19,7 @@
 #include <linux/ptrace.h>
 #include <linux/user.h>
 #include <linux/signal.h>
+#include <linux/seccomp.h>
 #include <linux/tracehook.h>
 
 #include <linux/uaccess.h>
@@ -277,6 +278,10 @@ asmlinkage int syscall_trace_enter(void)
 
 	if (test_thread_flag(TIF_SYSCALL_TRACE))
 		ret = tracehook_report_syscall_entry(task_pt_regs(current));
+
+	if (secure_computing() == -1)
+		return -1;
+
 	return ret;
 }
 
Signature · 1 line
2.17.1
#5 lore raw
In reply to Michael Schmitz

Hi Michael!

Quoted 9 lines · Michael Schmitz

Patch 3 adds the test for TIF_SECCOMP thread info flag to call into syscall_trace_enter() when seccomp syscall filtering is active, which was missing from earlier versions (classic MMU m68k only - need some help with the test for 68000 and coldfire).

The previous version required the TIF_SYSCALL_TRACE flag to be set for syscall filtering to work

Quoted 7 lines

The rest of the series remains unchanged from v9.

Tested on ARAnyM, 63 out of 89 seccomp_bpf tests now pass (33 with the old version).

I hope you didn't waste too much effort on testing v8/v9, Adrian. Should have a suitably large brown paper bag somewhere yet :-(

Sorry, I haven't found the time for the tests yet, I'm really busy at the moment. If you want to test libsecomp, you can check it out from my Github, then build it and run the testsuite.

See: https://github.com/glaubitz/libseccomp/tree/m68k

Adrian

Signature · 4 lines
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - [email protected]
`. `'   Freie Universitaet Berlin - [email protected]
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913
#6 lore raw

Hi Adrian,

not to worry - there is no rush on this. I may have to debug this a little more at the kernel level anyway (things like syscall redirection does not work yet, and neither does setting a particular errno when skipping a syscall). Plain syscall filtering does appear to work however, and AFAIR that would be enough for your purpose.

None of my test systems are recent enough to allow compiling current user space code, unfortunately.

Outdated user space might conntribute to some of the selftests failing, so I could use a little help with that perhaps. I know Geert tried the selftests independently some time ago, so I'll post my changes to the selftest code and we'll see whether that improves the score.

Cheers,

Michael

Am 27.01.2022 um 22:09 schrieb John Paul Adrian Glaubitz:

Quoted 28 lines

Hi Michael!

Quoted 9 lines · Michael Schmitz

Patch 3 adds the test for TIF_SECCOMP thread info flag to call into syscall_trace_enter() when seccomp syscall filtering is active, which was missing from earlier versions (classic MMU m68k only - need some help with the test for 68000 and coldfire).

The previous version required the TIF_SYSCALL_TRACE flag to be set for syscall filtering to work

Quoted 7 lines

The rest of the series remains unchanged from v9.

Tested on ARAnyM, 63 out of 89 seccomp_bpf tests now pass (33 with the old version).

I hope you didn't waste too much effort on testing v8/v9, Adrian. Should have a suitably large brown paper bag somewhere yet :-(

Sorry, I haven't found the time for the tests yet, I'm really busy at the moment. If you want to test libsecomp, you can check it out from my Github, then build it and run the testsuite.

See: https://github.com/glaubitz/libseccomp/tree/m68k

Adrian

#7 lore raw
In reply to Michael Schmitz
Quoted 9 lines · Michael Schmitz

Patch 3 adds the test for TIF_SECCOMP thread info flag to call into syscall_trace_enter() when seccomp syscall filtering is active, which was missing from earlier versions (classic MMU m68k only - need some help with the test for 68000 and coldfire).

The previous version required the TIF_SYSCALL_TRACE flag to be set for syscall filtering to work

Quoted 7 lines

The rest of the series remains unchanged from v9.

Tested on ARAnyM, 63 out of 89 seccomp_bpf tests now pass (33 with the old version).

I hope you didn't waste too much effort on testing v8/v9, Adrian. Should have a suitably large brown paper bag somewhere yet :-(

Has any progress been made on testing or merging this patch series?

I was just thinking it might be nice if the only callers of ptrace_report_syscall were ptrace_report_syscall_entry and ptrace_report_syscall_exit.

Eric

#8 lore raw

Hi Eric,

Am 04.05.2022 um 09:53 schrieb Eric W. Biederman:

Quoted 13 lines
Quoted 11 lines · Michael Schmitz

Patch 3 adds the test for TIF_SECCOMP thread info flag to call into syscall_trace_enter() when seccomp syscall filtering is active, which was missing from earlier versions (classic MMU m68k only - need some help with the test for 68000 and coldfire).

The previous version required the TIF_SYSCALL_TRACE flag to be set for syscall filtering to work

The rest of the series remains unchanged from v9.

Quoted 7 lines
Quoted 5 lines

Tested on ARAnyM, 63 out of 89 seccomp_bpf tests now pass (33 with the old version).

I hope you didn't waste too much effort on testing v8/v9, Adrian. Should have a suitably large brown paper bag somewhere yet :-(

Has any progress been made on testing or merging this patch series?

This series totally dropped off my radar, sorry. I had run out of options to test, and have been too busy to do much about that.

Quoted 3 lines

I was just thinking it might be nice if the only callers of ptrace_report_syscall were ptrace_report_syscall_entry and ptrace_report_syscall_exit.

Just merging patch 1 would be enough there AFAICS. Any reason not to do that, Geert?

Cheers,

Michael

Quoted 4 lines

Eric

maintainer Reviewed-by on v10 1/3
#2 lore raw

Re: [PATCH v10 1/3] m68k/kernel - wire up syscall_trace_enter/leave for m68k

In reply to Michael Schmitz
Quoted 11 lines · Michael Schmitz

m68k (other than Coldfire) uses syscall_trace for both trace entry and trace exit. Seccomp support requires separate entry points for trace entry and exit which are already provided for Coldfire.

Replace syscall_trace by syscall_trace_enter and syscall_trace_leave in preparation for seccomp support.

No regression seen in testing with strace on ARAnyM.

i.e. will queue in the m68k for-v5.19 branch.

Gr{oetje,eeting}s,

Geert

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

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

#9 lore raw
In reply to Michael Schmitz

Hi Michael,

Quoted 24 lines · Michael Schmitz

Am 04.05.2022 um 09:53 schrieb Eric W. Biederman:

Quoted 12 lines
Quoted 11 lines · Michael Schmitz

Patch 3 adds the test for TIF_SECCOMP thread info flag to call into syscall_trace_enter() when seccomp syscall filtering is active, which was missing from earlier versions (classic MMU m68k only - need some help with the test for 68000 and coldfire).

The previous version required the TIF_SYSCALL_TRACE flag to be set for syscall filtering to work

The rest of the series remains unchanged from v9.

Quoted 7 lines
Quoted 5 lines

Tested on ARAnyM, 63 out of 89 seccomp_bpf tests now pass (33 with the old version).

I hope you didn't waste too much effort on testing v8/v9, Adrian. Should have a suitably large brown paper bag somewhere yet :-(

Has any progress been made on testing or merging this patch series?

This series totally dropped off my radar, sorry. I had run out of options to test, and have been too busy to do much about that.

Quoted 6 lines
Quoted 3 lines

I was just thinking it might be nice if the only callers of ptrace_report_syscall were ptrace_report_syscall_entry and ptrace_report_syscall_exit.

Just merging patch 1 would be enough there AFAICS. Any reason not to do that, Geert?

Indeed no reason. Applied and queued for v5.19.

Gr{oetje,eeting}s,

Geert

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

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