diff options
| author | pliard@chromium.org <pliard@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2013-08-01 13:45:46 +0000 |
|---|---|---|
| committer | pliard@chromium.org <pliard@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2013-08-01 13:45:46 +0000 |
| commit | bf1efbfe52864430d3446f5e8b7a5afe0d22d72f (patch) | |
| tree | 47c9fa9d915729f3d4e752ab5b95e5ba30b7f0ee /content/child/fileapi | |
| parent | 728bc5ce79e7275dfd4b7107be16a63131a2ea54 (diff) | |
| download | chromium_src-bf1efbfe52864430d3446f5e8b7a5afe0d22d72f.zip chromium_src-bf1efbfe52864430d3446f5e8b7a5afe0d22d72f.tar.gz chromium_src-bf1efbfe52864430d3446f5e8b7a5afe0d22d72f.tar.bz2 | |
Reland r212020: Move Python setup/tear down logic into Forwarder ...
The initial attempt caused flakiness (random deadlocks). It turns out that
testserver.py was inheriting undesirable file descriptors when it was spawned
(including file lock file descriptors). Closing the unnecessary file
descriptors between fork() and exec() solves the problem.
Forwarder used to be a pain to setup/tear down across all the various
harnesses.
This CL should hopefully solve these issues by hiding these
implementation details. The host daemon is now killed once the first time that
the Forwarder class is used and the daemon running on the devices is also
killed the first time a port is forwarded for a specific device.
TBR=bulach@chromium.org
BUG=242846
Review URL: https://codereview.chromium.org/20824008
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@215023 0039d316-1c4b-4281-b951-d872f2087c98
Diffstat (limited to 'content/child/fileapi')
0 files changed, 0 insertions, 0 deletions