[i386] : Unbreak bootstrap for x64 SEH enabled target

2011/10/12 Richard Henderson <rth@redhat.com>:
> On 10/12/2011 12:07 AM, Kai Tietz wrote:
>> Hello,
>>
>> by recent changes gcc begun to move code into the prologue region.
>> This is for x64 SEH an issue, as here the table-information for
>> prologue is limited to 255 bytes size.  So we need to avoid moving
>> additional code into prologue.  To achieve this we mark all standard
>> and xmm registers as prologue-used at the end of prologue.   Also we
>> need to emit a memory blockage.
>>
>> ChangeLog
>>
>> 2011-10-12  Kai Tietz  <ktietz@redhat.com>
>>
>>         * config/i386/i386.c (ix86_expand_prologue): Mark
>>         for TARGET_SEH all sse/integer registers as prologue-used.
>>
>> Tested for x86_64-w64-mingw32.  Ok for apply?
>>
>> Regards,
>> Kai
>>
>> Index: i386.c
>> ===================================================================
>> --- i386.c      (revision 179824)
>> +++ i386.c      (working copy)
>> @@ -10356,7 +10356,24 @@
>>       Further, prevent alloca modifications to the stack pointer from being
>>       combined with prologue modifications.  */
>>    if (TARGET_SEH)
>> -    emit_insn (gen_prologue_use (stack_pointer_rtx));
>> +    {
>> +      int i;
>> +
>> +      /* Due limited size of prologue-code size of 255 bytes,
>> +         we need to prevent scheduler to sink instructions into
>> +         prologue code.  Therefore we mark all standard, sse, fpu,
>> +         and the pc registers as prologue-used to prevent this.
>> +         Also an memory-blockage is necessary.  */
>> +      emit_insn (gen_memory_blockage ());
>> +
>> +      for (i = 0; i <= 7; i++)
>> +        {
>> +         emit_insn (gen_prologue_use (gen_rtx_REG (Pmode, AX_REG + i)));
>> +         emit_insn (gen_prologue_use (gen_rtx_REG (Pmode, R8_REG + i)));
>> +         emit_insn (gen_prologue_use (gen_rtx_REG (TImode, XMM0_REG + i)));
>> +         emit_insn (gen_prologue_use (gen_rtx_REG (TImode, XMM8_REG + i)));
>> +       }
>> +    }
>
> This is overkill.  We simply need to disable shrink-wrapping for SEH.
> The easiest way to do that is to add !TARGET_SEH (and a comment) to
> the simple_return pattern predicate.
>
>
> r~

Thanks,  this is indeed more simple.  I wasn't aware that enamed
"return_simple" expand also enables shrink-wrapping into prologue.

Patch tested for x86_64-w64-mingw32.  Ok?

Regards,
Kai

On 10/12/2011 09:54 AM, Kai Tietz wrote:
> +;; We need to disable this for TARGET_SEH, as otherwise
> +;; shrink-wrapped prologue gets enabled too.  This might exceed
> +;; the maximum size of prologue in unwind information.
> +
>  (define_expand "simple_return"
>    [(simple_return)]
> -  ""
> +  "!TARGET_SEH"

Ok.


r~