PING: PATCH [mainline] use of save_world/rest_world calls in apple-ppc-darwin prologue/epilogue

Dale Johannesen dalej@apple.com
Fri Aug 6 00:39:00 GMT 2004


On Aug 5, 2004, at 5:35 PM, Fariborz Jahanian wrote:
>>
>> Could you find out?  It seems really very unlikely that this patch is 
>> ever invoked during a build of gcc, since gcc doesn't actually use 
>> altivec anywhere.
>
> There is no difference in cc1's code size which this patch might 
> possibly impact. So, change in
> size is due to other factors. But this patch will benefit APPLE's 
> internal apps. and is the main
> reason for submission.
> Specifically this patch benefits functions with unknown control flows 
> (setjmp, computed gotos, etc).
> Please let me know if it is OK for FSF?

If Geoff doesn't want it for FSF, I will approve it for 
apple-ppc-branch (after testing of course).
You've spent enough time on this IMO.

> Size comparisons:
>
> 1)Without this patch
>
> size ./cc1
> __TEXT  __DATA  __OBJC  others  dec     hex
> 5672960 1074290688      0       13078528        1093042176      
> 41268000
>
> 2) With this patch
>
> size ./cc1
> __TEXT  __DATA  __OBJC  others  dec     hex
> 5672960 1074290688      0       13008896        1092972544      
> 41257000
>
> - fariborz
>
>>
>> Other than that, the patch looks OK, but I don't want to have 
>> something like this applied if it doesn't actually produce a benefit.
>



More information about the Gcc-patches mailing list