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

Issue 756832 link

Starred by 2 users

Issue metadata

Status: Fixed
Owner:
Closed: Aug 2017
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug



Sign in to add a comment

V8 side GPU bots fail with dcheck PromotedSpaceSizeOfObjects() >= survived_watermark

Project Member Reported by machenb...@chromium.org, Aug 18 2017

Issue description

Cc: hablich@chromium.org
@hablich: This might not have rolled into chromium yet, and we should hold off the roll to not screw up the dev channel release. I close rolling until this is resolved.
Please reopen https://v8-roll.appspot.com/ with google.com account when this is resolved...
Cc: petermarshall@chromium.org
Owner: u...@chromium.org
Ulan is cooking up a fix...
Abandoned current roll: https://chromium-review.googlesource.com/c/620531
Project Member

Comment 6 by bugdroid1@chromium.org, Aug 18 2017

The following revision refers to this bug:
  https://chromium.googlesource.com/v8/v8.git/+/3294594895271be600a0100b9b4265cb9f151814

commit 3294594895271be600a0100b9b4265cb9f151814
Author: Ulan Degenbaev <ulan@chromium.org>
Date: Fri Aug 18 15:14:04 2017

[heap] Fix code that assumes monotonicity of PromotedSpaceSizeOfObjects.

After dfc6b4d the space size can decrease if the sweeper discovers
new fillers added after marking (e.g. by array trimming).

Bug:  chromium:756832 
Change-Id: Ibf420593bd12a4fe13a1e47f862302025b52ad58
Reviewed-on: https://chromium-review.googlesource.com/620734
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#47431}
[modify] https://crrev.com/3294594895271be600a0100b9b4265cb9f151814/src/heap/heap.cc
[modify] https://crrev.com/3294594895271be600a0100b9b4265cb9f151814/src/heap/heap.h

Comment 7 by u...@chromium.org, Aug 18 2017

Reopened the roll
Cc: kbr@chromium.org ynovikov@chromium.org hpayer@chromium.org jmad...@chromium.org enne@chromium.org machenb...@chromium.org
 Issue 756890  has been merged into this issue.

Comment 9 by kbr@chromium.org, Aug 18 2017

There's a compile error on the V8 side with msvc causing a couple of Windows trybot failures here:

https://chromium-review.googlesource.com/621191

Comment 10 by kbr@chromium.org, Aug 18 2017

FAILED: obj/v8/v8_base/wasm-module.obj 
ninja -t msvc -e environment.x86 -- E:\b\c\goma_client/gomacc.exe "e:\b\c\win_toolchain\vs_files\f53e4598951162bad6330f7a167486c7ae5db1e5\vc\bin\amd64_x86/cl.exe" /nologo /showIncludes  @obj/v8/v8_base/wasm-module.obj.rsp /c ../../v8/src/wasm/wasm-module.cc /Foobj/v8/v8_base/wasm-module.obj /Fd"obj/v8/v8_base_cc.pdb"
../../v8/src/wasm/wasm-module.cc(247): error C2440: 'reinterpret_cast': cannot convert from 'uint32_t' to 'uint32_t'
../../v8/src/wasm/wasm-module.cc(247): note: Conversion is a valid standard conversion, which can be performed implicitly or by use of static_cast, C-style cast or function-style cast
../../v8/src/wasm/wasm-module.cc(250): error C3852: 'v8::internal::compiler::ModuleEnv::globals_start' having type 'const uintptr_t': aggregate initialization could not initialize this member
../../v8/src/wasm/wasm-module.cc(250): note: const members cannot be default initialized unless their type has a user defined default constructor
[680/17152] CXX obj/v8/v8_base/wasm-compiler.obj
FAILED: obj/v8/v8_base/wasm-compiler.obj 
ninja -t msvc -e environment.x86 -- E:\b\c\goma_client/gomacc.exe "e:\b\c\win_toolchain\vs_files\f53e4598951162bad6330f7a167486c7ae5db1e5\vc\bin\amd64_x86/cl.exe" /nologo /showIncludes  @obj/v8/v8_base/wasm-compiler.obj.rsp /c ../../v8/src/compiler/wasm-compiler.cc /Foobj/v8/v8_base/wasm-compiler.obj /Fd"obj/v8/v8_base_cc.pdb"
../../v8/src/compiler/wasm-compiler.cc(2955): error C2440: 'reinterpret_cast': cannot convert from 'const uintptr_t' to 'uintptr_t'
../../v8/src/compiler/wasm-compiler.cc(2955): note: Conversion is a valid standard conversion, which can be performed implicitly or by use of static_cast, C-style cast or function-style cast

Comment 11 by kbr@chromium.org, Aug 18 2017

Stopped rolls. Rolling back the most recent roll.

Cc: mtrofin@chromium.org titzer@chromium.org
Thanks for the revert. The current roll seems blocked by:
https://chromium-review.googlesource.com/c/613880

The revert doesn't apply cleanly. Could somebody make that work or add a fix?
https://chromium-review.googlesource.com/c/622547

Guess I managed. Sorry for reverting also three other CLs. I'll make the merges to get the roll in.
Labels: -Pri-1 Pri-2
Here's a roll with ulan's fix and some patching for the other compile errors of the previous attempt:
https://chromium-review.googlesource.com/c/621555

Comment 16 by u...@chromium.org, Aug 19 2017

Ouch.

Thank you for reverting, Ken.

Thank you for fixing the roll, Michael.

Comment 17 by u...@chromium.org, Aug 21 2017

Status: Fixed (was: Assigned)

Sign in to add a comment