[PATCH] PR16490: Fix a crash in ARM inline asm

[Tinti]

Hi,

Sorry guys. The patch does not fix the problem. False alarm.

Can someone help me to check the following asm's? I have made clang print
the filename when it reaches some parts of the patch.

arch/arm/vfp/vfpdouble.c
arch/arm/vfp/vfpsingle.c
fs/eventpoll.c
fs/readdir.c
fs/signalfd.c

Best regards,
Tinti


On Fri, Jul 12, 2013 at 1:45 PM, Tinti <viniciustinti at gmail.com> wrote:

> Hi Renato,
>
> By "it works again" I mean it does compile and also does boot in VEXPRESS,
> I have not tested in Android.
>
> Also seems that this patch fixed the problem:
> http://permalink.gmane.org/gmane.comp.compilers.llvm.cvs/151589 if some
> one could try it would be great.
>
> Best regards,
> Tinti
>
>
> On Fri, Jul 12, 2013 at 1:14 PM, Renato Golin <renato.golin at linaro.org>wrote:
>
>> Hi Vinicius,
>>
>> Is that without any extra patch? Or do you apply the old Weimin's patch
>> that you were before?
>>
>> Does "it works again" mean it boots? Is this just the kernel or Android
>> user
>>
>> cheers,
>> --renato
>> Hi all,
>>
>> Looks like if you revert the following commits it works again.
>>
>> 8b97c23fe95db0bba23855205e2318188ceea947 (HEAD, test) Revert "Bug 13662:
>> Enable GPRPair for all i64 operands of inline asm on ARM"
>> a189501caeea75082f70646afd449700bdd67ad3 Revert "Remove unused variables."
>> 20f544be14e5096ee59c5c383731fb425191d5e7 Revert "PR16490: fix a crash in
>> ARMDAGToDAGISel::SelectInlineAsm."
>> 5762e77973991da82bc0fd3947df1a5d590de12e Revert "Add a comment to this
>> change, requested by Eric Christopher."
>>
>> On top of these:
>> 12fa02841ca1f2115d38aadd3e5d7afe08c65337 (origin/master, origin/HEAD,
>> master) Merge these tests.
>> aad52dfef9b64ed00aed3626513f3d2eea0fb8e0 Use a more unique name to avoid
>> conflicting with directory.ll tests when running in parallel.
>>
>> Best regards,
>> Tinti
>>
>>
>> On Tue, Jul 9, 2013 at 6:06 PM, Jan-Simon Möller <dl9pf at gmx.de> wrote:
>>
>>> Hi all,
>>>
>>> On Monday 08 July 2013 14:09:07 Joey Gouly wrote:
>>> > So, I think my patch is actually orthogonal to r185169. It’s just that
>>> > r185169 exposed a latent bug, that I fixed with my commit.
>>> >
>>> > I can't think of a good way to find the problem here, apart from
>>> diffing the
>>> > kernels before r185169 and at r185169 (with my patch to let it
>>> compile).
>>> > Seeing where the differences come from, and finding the inline assembly
>>> > that produced that.
>>>
>>> @Joe - the output is here:
>>>
>>> http://88.198.106.157/tmp_vmlinux1_old.xz   (pre r185169)
>>> http://88.198.106.157/tmp_vmlinux1_new.xz   (r185169+your commit)
>>>
>>> What are the next steps to compare them ? What do you propose ?
>>>
>>> -- JS
>>>
>>>
>>
>>
>> --
>> Simplicity is the ultimate sophistication
>>
>
>
>
> --
> Simplicity is the ultimate sophistication
>



-- 
Simplicity is the ultimate sophistication
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130712/560d2f52/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: asm.tar.gz
Type: application/x-gzip
Size: 264491 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130712/560d2f52/attachment.bin>