Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(474)

Issue 27541003: add initial support ofr multi-threaded execution (Closed)

Created:
7 years, 2 months ago by djsollen
Modified:
7 years, 1 month ago
Reviewers:
mtklein
CC:
skia-review_googlegroups.com
Visibility:
Public.

Description

add initial support ofr multi-threaded execution Committed: https://code.google.com/p/skia/source/detail?r=11823

Patch Set 1 #

Patch Set 2 : #

Total comments: 2
Unified diffs Side-by-side diffs Delta from patch set Stats (+56 lines, -9 lines) Patch
M tools/skpdiff/SkDiffContext.h View 3 chunks +4 lines, -0 lines 0 comments Download
M tools/skpdiff/SkDiffContext.cpp View 8 chunks +47 lines, -9 lines 2 comments Download
M tools/skpdiff/skpdiff_main.cpp View 2 chunks +5 lines, -0 lines 0 comments Download

Messages

Total messages: 2 (0 generated)
djsollen
Committed patchset #2 manually as r11823 (presubmit successful).
7 years, 2 months ago (2013-10-16 18:36:58 UTC) #1
mtklein
7 years, 1 month ago (2013-10-23 02:47:12 UTC) #2
Message was sent while issue was closed.
https://codereview.chromium.org/27541003/diff/3001/tools/skpdiff/SkDiffContex...
File tools/skpdiff/SkDiffContext.cpp (right):

https://codereview.chromium.org/27541003/diff/3001/tools/skpdiff/SkDiffContex...
tools/skpdiff/SkDiffContext.cpp:114:
fDiffContext->addDiff(fBaselinePath.c_str(), fTestPath.c_str());
I'm crashing all over the place when I run skpdiff with more than one thread.

Are you certain that addDiff is threadsafe?  It doesn't look like any of the
code it calls in skpdiff/ is itself threadsafe.  We'd need to at least guard
fRecords and we definitely can't share fDiffers like we're doing here (maybe we
can just put them on the stack?  thread-local at worst?).

I'd strongly suggest setting the default --threads to 0 until we can get this
cleaned up.

https://codereview.chromium.org/27541003/diff/3001/tools/skpdiff/SkDiffContex...
tools/skpdiff/SkDiffContext.cpp:152: threadPool.wait();
These wait() calls are harmless, but you can cut them out if you want.  The
threadpool destructor will wait too.

Powered by Google App Engine
This is Rietveld 408576698