[llvm] r238362 - Reapply "AsmPrinter: Change DIEValue to be stored by value"
Duncan P. N. Exon Smith
dexonsmith at apple.com
Thu May 28 11:02:09 PDT 2015
> On 2015-May-28, at 08:33, Rafael EspĂndola <rafael.espindola at gmail.com> wrote:
>
> On 28 May 2015 at 10:58, Duncan P. N. Exon Smith <dexonsmith at apple.com> wrote:
>>
>>> On 2015 May 28, at 07:28, Aaron Ballman <aaron at aaronballman.com> wrote:
>>>
>>> On Thu, May 28, 2015 at 10:23 AM, Duncan Exon Smith
>>> <dexonsmith at apple.com> wrote:
>>>> Thanks for pointing it out. My build is clean, as are the bots I looked at that build with -Werror.
>>>>
>>>> What compiler are you using? Looks like some version of GCC maybe?
>>>>
>>>> Chandler, you helped me yesterday... any ideas for me here? Any other experts on AlignedCharArrayUnion?
>>>
>>> This is coming from the bot used to compile our attribute
>>> documentation server-side. I don't know the exact details of the
>>> configuration, but Tanya would know if we need to find out. I am happy
>>> to forward the bot email in its entirety if you think it would help.
>>>
>>> That being said, there's at least one bot showing signs of this as
>>> well: http://bb.pgr.jp/builders/clang-3stage-x86_64-linux/builds/8597/steps/stage1_build/logs/warnings%20%28240%29
>>>
>>> ~Aaron
>>
>> Looks like chapuni's bot is using GCC 4.7 for stage 1 (and the warnings
>> are happening in stage 1). Do you have a link to the other bot?
>>
>> If this is happening locally for you (or someone else), maybe try out
>> the attached patch. I won't have time to run tests and commit it
>> myself until I get to work (another couple of hours), but if someone
>> is able to confirm it fixes the problem please feel free to commit on
>> my behalf.
>>
>> I'm not terribly confident it will help, since it just moves the
>> dereference to a different location, but perhaps it's good enough to
>> confuse GCC?
>
> It fixes the warning for me using gcc 4.9.2.
Finally sitting down at a computer (crazy morning with some personal
stuff). Thanks for checking. I'll commit once `ninja check` finishes.
More information about the llvm-commits
mailing list