[PATCH] fix bad syscall on ARM architecture

[Hans-Christian Egtvedt]

On Fri, 20 Feb 2009 23:20:40 +0100
Denys Vlasenko <dvlasenk at redhat.com> wrote:

> On Fri, 2009-02-20 at 13:23 +0100, Hans-Christian Egtvedt wrote:

<snipp>

> --- a/syscall.c
> +++ b/syscall.c
> @@ -1060,6 +1060,14 @@ get_scno(struct tcb *tcp)
>                 /*
>                  * Note: we only deal with only 32-bit CPUs here.
>                  */
> +
> +               if (!(tcp->flags & TCB_INSYSCALL) &&
> +                   (tcp->flags & TCB_WAITEXECVE)) {
> +                       /* caught a fake syscall from the execve's
> exit */
> +                       tcp->flags &= ~TCB_WAITEXECVE;
> +                       return 0;
> +               }
> +
>                 if (regs.ARM_cpsr & 0x20) {
> 
> But this is already checked a bit earlier. Here is a bigger fragment.
> Look as the second "if":
> 
>         /*
>          * We only need to grab the syscall number on syscall entry.
>          */
>         if (regs.ARM_ip == 0) {
>                 if (!(tcp->flags & TCB_INSYSCALL)) {
>                         /* Check if we return from execve. */
>                         if (tcp->flags & TCB_WAITEXECVE) {

Aha (-:

Then the patch can probably be nuked all together. My 0.02 € is that it
was needed for a previous version of strace, but was not nuked when
strace version was bumped in Buildroot.

<snipp>

-- 
Best regards,
Hans-Christian Egtvedt