Started here https://build.chromium.org/p/chromium.lkgr/builders/Win%20ASan%20Release/builds/3961

Nothing obvious, jochen's cdb change probably looks most related if you want to try and speculatively revert something.

I am OOO. Oliver, can you try the revert and see if it fixes.

Real error seems to be:

[10803/35819] ACTION //v8:run_mkpeephole(//build/toolchain/win:clang_x86)
FAILED: gen/v8/bytecode-peephole-table.cc 
C:/b/depot_tools/python276_bin/python.exe ../../v8/tools/run.py ./mkpeephole gen/v8/bytecode-peephole-table.cc
=================================================================
==7368==ERROR: AddressSanitizer: attempting free on address which was not malloc()-ed: 0x0018fe58 in thread T0
    #0 0xcd381a in free e:\b\build\slave\win_upload_clang\build\src\third_party\llvm\projects\compiler-rt\lib\asan\asan_malloc_win.cc:45
    #1 0x6a4b0ba1 in ltoa+0x1a1 (C:\b\c\b\Win_ASan_Release\src\out\Release\ucrtbase.DLL+0x10030ba1)
    #2 0x77668ea4 in RtlIsCurrentThreadAttachExempt+0x5e (C:\Windows\SysWOW64\ntdll.dll+0x7dea8ea4)
    #3 0x77689d91 in LdrShutdownProcess+0x96 (C:\Windows\SysWOW64\ntdll.dll+0x7dec9d91)
    #4 0x77689cdd in RtlExitUserProcess+0x73 (C:\Windows\SysWOW64\ntdll.dll+0x7dec9cdd)
    #5 0x754d79dc in ExitProcess+0x14 (C:\Windows\syswow64\kernel32.dll+0x7dd779dc)
    #6 0xcf9470 in exit_or_terminate_process d:\rs1\minkernel\crts\ucrt\src\appcrt\startup\exit.cpp:129
    #7 0xcf942e in common_exit d:\rs1\minkernel\crts\ucrt\src\appcrt\startup\exit.cpp:265
    #8 0xcf9567 in exit d:\rs1\minkernel\crts\ucrt\src\appcrt\startup\exit.cpp:278
    #9 0xcecea7 in __scrt_common_main_seh f:\dd\vctools\crt\vcstartup\src\startup\exe_common.inl:260
    #10 0x754d3379 in BaseThreadInitThunk+0x11 (C:\Windows\syswow64\kernel32.dll+0x7dd73379)
    #11 0x776692e1 in RtlInitializeExceptionChain+0x62 (C:\Windows\SysWOW64\ntdll.dll+0x7dea92e1)
    #12 0x776692b4 in RtlInitializeExceptionChain+0x35 (C:\Windows\SysWOW64\ntdll.dll+0x7dea92b4)

Address 0x0018fe58 is a wild pointer.
SUMMARY: AddressSanitizer:

Reverting jochen's CL doesn't fix this. Might be an issue with the V8 roll?

Looks identical to the issue Etienne was fixing here: https://reviews.llvm.org/D25946

Which also showed up on the TOT bots here: https://bugs.chromium.org/p/chromium/issues/detail?id=693718

I'm going to merge that issue into this one.

Did someone re-image the VM this bot uses? Basically, this bug depends on which version of dbghelp.dll is installed on the machine running ASan.

As Etienne mentioned in the other bug, if mkpeephole was recently switched from being 32-bit to 64-bit that could have also caused this.

I think this will be fixed in the next Clang roll.

 Issue 693718  has been merged into this issue.

I observed on 64-bit that recent CRT are doing allocations before asan is able to hook.

But, I believe this may also happen on 32-bit architecture.

So, as Reid mentioned, did someone update CRT, compiler, etc..?


> I think this will be fixed in the next Clang roll.

No, if this issue is on a 32-bit executable. The fix that landed in LLVM 
only apply to windows-64.

Roll landed and helped.

That means it was a bug caused by that fixed:
  https://reviews.llvm.org/D25946

The executable was probably built for x64.