New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 630742 link

Starred by 3 users

Issue metadata

Status: Fixed
Owner: ----
Closed: Aug 2016
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 1
Type: Bug

Blocking:
issue 629966



Sign in to add a comment

Clang ToT: ASAN in broken on Win32

Project Member Reported by krasin@chromium.org, Jul 22 2016

Issue description

A number of Clang ToT bots are broken due to some ASAN reports like below:
https://build.chromium.org/p/chromium.fyi/builders/CrWinAsanCov/builds/2652/steps/compile/logs/stdio

319/27494] ACTION Generating yasm tokens for lc3b
FAILED: obj/third_party/yasm/yasm.gen/third_party/yasm/lc3bid.c 
C:\b\depot_tools\python276_bin\python.exe gyp-win-tool action-wrapper environment.x86 yasm_target_generate_lc3b_token_2995c673986e86381842b3a605763f4d..rsp ..\..\third_party\yasm
==3852==ERROR: AddressSanitizer failed to allocate 0x20000000 (536870912) bytes of CovInit (error code: 1455)
==3852==Dumping process modules:
	0x07730000-0x7772c000 C:\Windows\syswow64\msvcrt.dll
	0x5bbd0000-0x6bbd3000 C:\Windows\system32\api-ms-win-core-synch-l1-2-0.DLL
	0x65500000-0x7550c000 C:\Windows\syswow64\CRYPTBASE.dll
	0x74b40000-0x77679000 C:\Windows\SysWOW64\sechost.dll
	0xf7aa0000-0x756e0000 C:\Windows\syswow64\RPCRT4.dll
	0xf7c70000-0x75570000 C:\Windows\syswow64\SspiCli.dll
	0xf7db0000-0x75c20000 C:\Windows\syswow64\kernel32.dll
	0xf84c0000-0x75d57000 C:\Windows\syswow64\KERNELBASE.dll
	0xf9d20000-0x77d10000 C:\Windows\SysWOW64\ntdll.dll
	0xfea20000-0x76720000 C:\Windows\syswow64\ADVAPI32.dll
	0xffc80000-0x00589000 C:\b\c\b\CrWinAsanCov\src\out\Release\re2c.exe
==3852==AddressSanitizer CHECK failed: C:\b\c\b\CrWinAsanCov\src\third_party\llvm\projects\compiler-rt\lib\sanitizer_common\sanitizer_common.cc:120 "((0 && "unable to mmap")) != (0)" (0x0, 0x0)
    #0 0xa8527 in __sanitizer_update_counter_bitset_and_clear_counters+0x97 (C:\b\c\b\CrWinAsanCov\src\out\Release\re2c.exe+0x428527)
    #1 0xac9f3 in __sanitizer::CheckFailed+0x43 (C:\b\c\b\CrWinAsanCov\src\out\Release\re2c.exe+0x42c9f3)
    #2 0xb3efa in __sanitizer::ReportMmapFailureAndDie+0x5a (C:\b\c\b\CrWinAsanCov\src\out\Release\re2c.exe+0x433efa)
    #3 0xb5063 in __sanitizer::MmapNoReserveOrDie+0x33 (C:\b\c\b\CrWinAsanCov\src\out\Release\re2c.exe+0x435063)
    #4 0xa6e0a in __sanitizer::CoverageData::Enable+0x1a (C:\b\c\b\CrWinAsanCov\src\out\Release\re2c.exe+0x426e0a)
    #5 0xa833b in __sanitizer_cov_module_init+0x1b (C:\b\c\b\CrWinAsanCov\src\out\Release\re2c.exe+0x42833b)
    #6 0x81e32 in __local_stdio_scanf_options+0x62 (C:\b\c\b\CrWinAsanCov\src\out\Release\re2c.exe+0x401e32)
    #7 0xc1455 in __scrt_common_main_seh f:\dd\vctools\crt\vcstartup\src\startup\exe_common.inl:221
    #8 0x75b23379 in BaseThreadInitThunk+0x11 (C:\Windows\syswow64\kernel32.dll+0x7dd73379)
    #9 0x77bc92b1 in RtlInitializeExceptionChain+0x62 (C:\Windows\SysWOW64\ntdll.dll+0x7dea92b1)
    #10 0x77bc9284 in RtlInitializeExceptionChain+0x35 (C:\Windows\SysWOW64\ntdll.dll+0x7dea9284)

The immediate issue is inability to allocate 512MB of memory, and the error code 1455 stands for:

ERROR_COMMITMENT_LIMIT
1455 (0x5AF)
The paging file is too small for this operation to complete.

So, we could either increase the page file on the bots, or stop requesting 512MB (sorry, I don't know specifics, why do we do that), or make sure there's such a contiguous chunk in the memory.

This is currently blocking the Clang roll.
 

Comment 1 by r...@chromium.org, Jul 27 2016

Cc: etienneb@chromium.org thakis@chromium.org
 Issue 632029  has been merged into this issue.

Comment 2 by r...@chromium.org, Jul 27 2016

The same failure is present on CrWinAsan (no coverage).

Long term, we might want to consider doing what we do on win64 asan, which is to implement on-demand paging in user space with an exception handler, so we don't have to worry about allocating real swap for most shadow memory.

Comment 3 by r...@chromium.org, Aug 3 2016

Status: Fixed (was: Untriaged)
We fixed this by rebooting those slaves.

Sign in to add a comment