arm syscall fast path can miss a ptrace syscall-exit

[Russell King - ARM Linux]

On Thu, May 14, 2015 at 12:13:40PM -0700, Josh Stone wrote:
> I've discovered a case where both arm and arm64 will miss a ptrace
> syscall-exit that they should report.  If the syscall is entered without
> TIF_SYSCALL_TRACE set, then it goes on the fast path.  It's then
> possible to have TIF_SYSCALL_TRACE added in the middle of the syscall,
> but ret_fast_syscall doesn't check this flag again.

Yes, we assume that if TIF_SYSCALL_TRACE was not set before the call, it
isn't set after.  That appears to be an invalid assumption.

Here's a patch for ARM - untested atm.

There's still a possible hole - if we exit the syscall, then do "work"
before returning (such as reschedling to another process), and _then_
have syscall tracing enabled, we won't trace the exit.  I think that's
acceptable as I see no difference between that and having restored
state for userspace, and then immediately processing an interrupt and
scheduling on the IRQ exit path.

 arch/arm/kernel/entry-common.S | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/arch/arm/kernel/entry-common.S b/arch/arm/kernel/entry-common.S
index f8ccc21fa032..4e7f40c577e6 100644
--- a/arch/arm/kernel/entry-common.S
+++ b/arch/arm/kernel/entry-common.S
@@ -33,7 +33,9 @@ ret_fast_syscall:
  UNWIND(.fnstart	)
  UNWIND(.cantunwind	)
 	disable_irq				@ disable interrupts
-	ldr	r1, [tsk, #TI_FLAGS]
+	ldr	r1, [tsk, #TI_FLAGS]		@ re-check for syscall tracing
+	tst	r1, #_TIF_SYSCALL_WORK
+	bne	__sys_trace_return
 	tst	r1, #_TIF_WORK_MASK
 	bne	fast_work_pending
 	asm_trace_hardirqs_on


-- 
FTTC broadband for 0.8mile line: currently at 10.5Mbps down 400kbps up
according to speedtest.net.