[llvm-commits] [llvm-gcc-4.2] r52263 - in /llvm-gcc-4.2/trunk/gcc: c-cppbuiltin.c cfgexpand.c common.opt function.c toplev.c

Dale Johannesen dalej at apple.com
Fri Jun 13 17:49:23 PDT 2008


On Jun 13, 2008, at 5:37 PM, Chris Lattner wrote:

>
> On Jun 13, 2008, at 4:57 PM, Dale Johannesen wrote:
>
>>
>> On Jun 13, 2008, at 4:48 PM, Devang Patel wrote:
>>>
>>> Indeed. Right now, I do not have the setup handy. Would it be
>>> possible
>>> to try (and apply if it works)  this ?
>>> Thanks,
>>> -
>>> Devang
>>
>> OK.  It takes a while on rs6000 as you know.
>
> Won't this require updating every target?  This seems like the wrong
> fix.
>
> -Chris

Every target that uses flag_stack_protector.  But Devang is right that  
there's no way to conditionalize the .opt file.

fwiw, devang's patch does make the build succeed on rs6000.

>>> Index: config/rs6000/rs6000.h
>>> ===================================================================
>>> --- config/rs6000/rs6000.h	(revision 52264)
>>> +++ config/rs6000/rs6000.h	(working copy)
>>> @@ -1212,8 +1212,11 @@
>>>
>>>   On the RS/6000, we grow upwards, from the area after the outgoing
>>>   arguments.  */
>>> -#define FRAME_GROWS_DOWNWARD (flag_stack_protect != 0)
>>> +/* LLVM LOCAL begin stack protector */
>>> +#define FRAME_GROWS_DOWNWARD 0
>>> +/* LLVM LOCAL end stack protector */
>>>
>>> +
>>> /* Size of the outgoing register save area */
>>> #define RS6000_REG_SAVE ((DEFAULT_ABI == ABI_AIX			\
>>> 			  || DEFAULT_ABI == ABI_DARWIN)			\
>>>
>>>
>>> _______________________________________________
>>> llvm-commits mailing list
>>> llvm-commits at cs.uiuc.edu
>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits




More information about the llvm-commits mailing list