[compiler-rt] r243686 - [Sanitizers] [Bug 24151] Generalize type of offset in internal_mmap

[Hans Wennborg]

On Fri, Jul 31, 2015 at 2:22 PM, Daniel Sanders
<Daniel.Sanders at imgtec.com> wrote:
>> On build 5935, which is the first one that includes this commit, the
>> first error is actually "==22192==ERROR: AddressSanitizer failed to
>> allocate 0x10000 (65536) bytes at address 617000f40000 (errno: 12)"
>>
>> That's what made me suspect it could be related.
>
> Reading the logs again, it looks like I was reading 5936's log which first shows the compilation failure. The four patches on the blamelist for that build look innocent at first glance, at least, there's nothing obviously related to subtarget features.
>
>> It's entirely possible that this is all unrelated, and the bot is just
>> in a bad state, but it was green before, and turned red when this
>> patch landed. I'd feel much better merging this patch if the bot were
>> green.
>
> I agree. I'm testing rc2 with this patch applied for the moment but if it breaks other targets then we can't merge it.
> Fortunately, it's not a regression since asan is new for LLVM 3.7.0. We could potentially accept the failing test and release anyway.

The bot has turned green again, unclear why.

I've merged r243686 in r244114.

Thanks,
Hans



> ________________________________________
> From: hwennborg at google.com [hwennborg at google.com] on behalf of Hans Wennborg [hans at chromium.org]
> Sent: 31 July 2015 21:25
> To: Daniel Sanders; kcc at google.com
> Cc: Mohit Bhakkad; llvm-commits
> Subject: Re: [compiler-rt] r243686 - [Sanitizers] [Bug 24151] Generalize type of offset in internal_mmap
>
> Kostya: earthdok is listed as the owner of
> http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-fast/. Do you
> know who might be able to figure out why it's red?
>
>
> On Fri, Jul 31, 2015 at 2:01 AM, Daniel Sanders
> <Daniel.Sanders at imgtec.com> wrote:
>> I don't think it's related but it's not a false positive either. The annotate step is failing with:
>>         /mnt/b/sanitizer-buildbot3/sanitizer-x86_64-linux-fast/build/llvm/lib/Target/X86/AsmParser/X86AsmInstrumentation.cpp:264:38: error: no member named 'Mode64Bit' in namespace 'llvm::X86'
>>             return STI.getFeatureBits()[X86::Mode64Bit];
>>                                         ~~~~~^
>> and other compilation errors in areas this patch didn't touch.
>
> On build 5935, which is the first one that includes this commit, the
> first error is actually "==22192==ERROR: AddressSanitizer failed to
> allocate 0x10000 (65536) bytes at address 617000f40000 (errno: 12)"
>
> That's what made me suspect it could be related.
>
> http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-fast/builds/5954/steps/annotate/logs/stdio
> is the build where this patch got reverted, but it seems the bot did
> not rebuild everything that was built when the patch landed.
>
> It's entirely possible that this is all unrelated, and the bot is just
> in a bad state, but it was green before, and turned red when this
> patch landed. I'd feel much better merging this patch if the bot were
> green.
>
>
>>> -----Original Message-----
>>> From: hwennborg at google.com [mailto:hwennborg at google.com] On Behalf
>>> Of Hans Wennborg
>>> Sent: 31 July 2015 00:14
>>> To: Daniel Sanders
>>> Cc: llvm-commits
>>> Subject: Re: [compiler-rt] r243686 - [Sanitizers] [Bug 24151] Generalize type
>>> of offset in internal_mmap
>>>
>>> On Thu, Jul 30, 2015 at 2:00 PM, Daniel Sanders
>>> <daniel.sanders at imgtec.com> wrote:
>>> > Author: dsanders
>>> > Date: Thu Jul 30 16:00:48 2015
>>> > New Revision: 243686
>>> >
>>> > URL: http://llvm.org/viewvc/llvm-project?rev=243686&view=rev
>>> > Log:
>>> > [Sanitizers] [Bug 24151] Generalize type of offset in internal_mmap
>>> >
>>> > Summary:
>>> > Using u64 as type for offset changes its value, changing starting address for
>>> map in file.
>>> > This patch solves Bug 24151, which raises issue while mapping file in mips32.
>>> >
>>> > Patch by Mohit Bhakkad
>>> >
>>> > Reviewers: dsanders, kcc
>>> >
>>> > Subscribers: hans, llvm-commits, samsonov, nitesh.jain, sagar, bhushan,
>>> jaydeep
>>> >
>>> > Differential Revision: http://reviews.llvm.org/D11588
>>>
>>> It might not be related (I'm not entirely sure what's going on), but
>>> it seems this change made a bot go red:
>>> http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-fast/builds/5935
>>>
>>> Given that, I'm not going to try and squeeze this into 3.7 right before RC2.