Re: Regression of GCC >= 4.8.5 for 32-bit powerpc

On 2016/12/14 13:11, Matt Thomas wrote:
>
>> On Dec 13, 2016, at 8:06 PM, Rin Okuyama <[email protected]> wrote:
>>
>> Thank you for your correction. I slightly modified your version
>> in order to avoid dangling ".machine". With this patch, I obtained
>> working kernel and userland for IBM 405 (evbppc). Also, build.sh
>> successfully completes for evbppc-powerpc64. Is this OK with you?
>>
>> Rin
>
> Look fine.
>
>> +	p = "ppc64";
>> +      else if (!global_options_set.x_rs6000_cpu_index)
>> +	p = "ppc";
>> +      if (p != NULL)
>> +	fprintf (asm_out_file, "\t.machine %s\n", p);
>>     }
>> #endif
>
> However.  You can drop the "if (p != NULL)" if you do
>
> 	else
> 	  return;
> 	gcc_assert(p != NULL);
> 	fprintf (asm_out_file, "\t.machine %s\n", p);
>
> Pendantic I know.
>

Thank you for your reply. However, we cannot return here; it is not
the end of this function.

Rin