Re: [Qemu-devel] [PATCH 4/5] cpu-exec: Move TB execution stuff out of cpu_exec()

[Sergey Fedorov]

On 15/07/16 09:45, Stefan Weil wrote:
> Hi,
>
> Am 11.05.2016 um 12:21 schrieb Sergey Fedorov:
> [...]
>>   int cpu_exec(CPUState *cpu)
>> @@ -516,8 +576,6 @@ int cpu_exec(CPUState *cpu)
>>      CPUArchState *env = &x86_cpu->env;
>>  #endif
>>      int ret;
>> -    TranslationBlock *tb, *last_tb;
>> -    int tb_exit = 0;
> Here tb_exit was only once set to 0, ...
>
>>      SyncClocks sc;
>>  
>>      /* replay_interrupt may need current_cpu */
>> @@ -544,6 +602,9 @@ int cpu_exec(CPUState *cpu)
>>      init_delay_params(&sc, cpu);
>>  
>>      for(;;) {
>> +        TranslationBlock *tb, *last_tb;
>> +        int tb_exit = 0;
> ... while now it is zeroed in each iteration of the for loop.
> I'm not sure whether the new code is still correct.

That is okay because 'tb_exit' only makes sense when "last_tb != NULL".
But we always reset 'last_tb' in this loop:

    last_tb = NULL; /* forget the last executed TB after exception */

>
> If it is, ...
>
>> +
>>          /* prepare setjmp context for exception handling */
>>          if (sigsetjmp(cpu->jmp_env, 0) == 0) {
> ... the declaration of tb_exit could also be done here, after the sigsetjmp.
> That would fix a compiler warning which I get when compiling with
> -Wclobbered:
>
>
>     cpu-exec.c:603:13: warning: variable ‘tb_exit’ might be clobbered by
> ‘longjmp’ or ‘vfork’ [-Wclobbered]

I've sent the patch to fix this:

    Message-Id: <address@hidden>

Thanks,
Sergey