diff options
| author | eroman@chromium.org <eroman@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2009-10-29 22:21:00 +0000 |
|---|---|---|
| committer | eroman@chromium.org <eroman@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2009-10-29 22:21:00 +0000 |
| commit | 27f99c872ffb123323e03bfcaa8bd89a1b0579d8 (patch) | |
| tree | 12985ccdc3454e5adc9935cb62d8b4192fc1e6d8 /chrome/common | |
| parent | 617176577ba1075f139b7627950876b017510d18 (diff) | |
| download | chromium_src-27f99c872ffb123323e03bfcaa8bd89a1b0579d8.zip chromium_src-27f99c872ffb123323e03bfcaa8bd89a1b0579d8.tar.gz chromium_src-27f99c872ffb123323e03bfcaa8bd89a1b0579d8.tar.bz2 | |
Make LoadLog refcounted-thread safe.
This also backs out r26610 (these are the changes to host_resolver_impl.cc), since that code is no longer necessary if LoadLog is refcounted thread safe.
LoadLog itself is not thread safe -- however making its refcounting threadsafe simplifies passing it across thread boundaries.
Initially I had avoided doing this, and instead was trying to be very careful in when callers would AddRef/Release to avoid races.
This approach however turned out to be fragile, and it was difficult to chase down all of the dependencies (they just kept popping up).
So now I am changing that decision, and opting for the simpler approach which will be much more robust.
The performance impact should be negligable.
BUG=22272
Review URL: http://codereview.chromium.org/347021
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@30525 0039d316-1c4b-4281-b951-d872f2087c98
Diffstat (limited to 'chrome/common')
0 files changed, 0 insertions, 0 deletions