Re: [PATCH] tracing: properly align linker defined symbols

[Zeev Tarantov]

On Sat, Jul 10, 2010 at 16:48, Sam Ravnborg <sam@xxxxxxxxxxxx> wrote:
> On Sat, Jul 10, 2010 at 07:34:05AM -0400, Steven Rostedt wrote:
>> On Sat, 2010-07-10 at 13:18 +0300, Zeev Tarantov wrote:
>> > On Sat, Jul 10, 2010 at 09:35, Sam Ravnborg <sam@xxxxxxxxxxxx> wrote:
>>
>> > > +/*
>> > > + * Align to a 32 byte boundary equal to the
>> > > + * alignment gcc 4.5 uses for a struct
>> > > + */
>> > > +#define STRUCT_ALIGN() . = ALIGN(32)
>> > > +
>>
>> What I'm nervous about is when gcc 4.8 decides to up the alignment to
>> 64.
>>
>> Maybe we should have both patches, just to be safe.
>
> Another approach could be to just stop playing games with alignment
> and use the fact that __stop_syscalls_metadata point to next byte
> after last entry.
>
> Something like the below untested patch.
>
> But to fix the current regression I prefer the simpler
> patch that just fixup the aligment.
>
>        Sam
>
>
> diff --git a/kernel/trace/trace_syscalls.c b/kernel/trace/trace_syscalls.c
> index 34e3580..50e9606 100644
> --- a/kernel/trace/trace_syscalls.c
> +++ b/kernel/trace/trace_syscalls.c
> @@ -79,15 +79,19 @@ static struct syscall_metadata *find_syscall_meta(unsigned long syscall)
>        stop = (struct syscall_metadata *)__stop_syscalls_metadata;
>        kallsyms_lookup(syscall, NULL, NULL, NULL, str);
>
> -       for ( ; start < stop; start++) {
> +       /*
> +        * start may point a few bytes before first entry
> +        * stop points at first byte after last entry
> +        */
> +       for (stop--; stop >= start; stop--) {
>                /*
>                 * Only compare after the "sys" prefix. Archs that use
>                 * syscall wrappers may have syscalls symbols aliases prefixed
>                 * with "SyS" instead of "sys", leading to an unwanted
>                 * mismatch.
>                 */
> -               if (start->name && !strcmp(start->name + 3, str + 3))
> -                       return start;
> +               if (stop->name && !strcmp(stop->name + 3, str + 3))
> +                       return stop;
>        }
>        return NULL;
>  }
>

This also boots (source from tarball with only this patch, same config
& compiler).
Tested-by: Zeev Tarantov <zeev.tarantov@xxxxxxxxx>

-Zeev
--
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/