[PATCH] ARM:stacktrace: make stacktrace skip always right.

From: Wang Kai
Date: Thu Sep 11 2014 - 05:14:58 EST


This patch is for commit 3683f44c42e991d313dc301504ee0fca1aeb8580,
that makes stacktrace NOT begin from stacktrace self function,
which are: __save_stack_trace and save_stack_trace.

But save_stack_trace will NOT have stacktrace information when build
with '-O2', assemble code looks like below:
00000154 <save_stack_trace>:
154: e1a0200d mov r2, sp
158: e1a01000 mov r1, r0
15c: e3c23d7f bic r3, r2, #8128 ; 0x1fc0
160: e3a02000 mov r2, #0
164: e3c3303f bic r3, r3, #63 ; 0x3f
168: e593000c ldr r0, [r3, #12]
16c: eaffffd0 b b4 <__save_stack_trace>

In this situation, the "data.skip +=2" operation will skip the last
user call function, make user stacktrace strange/incorrect.

To fix this, we have to make sure save_stack_trace has stacktrace info,
then skip 2 function will be OK.

Signed-off-by: Wang Kai <morgan.wang@xxxxxxxxxx>
---
arch/arm/kernel/stacktrace.c | 1 +
1 file changed, 1 insertion(+)

diff --git a/arch/arm/kernel/stacktrace.c b/arch/arm/kernel/stacktrace.c
index 6582c4a..1eb1b5ae 100644
--- a/arch/arm/kernel/stacktrace.c
+++ b/arch/arm/kernel/stacktrace.c
@@ -134,6 +134,7 @@ void save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace)
void save_stack_trace(struct stack_trace *trace)
{
__save_stack_trace(current, trace, 0);
+ asm volatile("" : : : "memory");
}
EXPORT_SYMBOL_GPL(save_stack_trace);
#endif
--
1.8.5.5

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/