summaryrefslogtreecommitdiffstats
path: root/extensions/renderer
diff options
context:
space:
mode:
authordalecurtis <dalecurtis@chromium.org>2015-11-10 10:25:58 -0800
committerCommit bot <commit-bot@chromium.org>2015-11-10 18:26:52 +0000
commitdd1b4ad61b745e8d75195cd0b4938086d3b6062d (patch)
tree2c9c7b807f5e10ced053650cdd95bb66a08b60b9 /extensions/renderer
parent2641bbb33465cf9d2442886162e4911cd91c011c (diff)
downloadchromium_src-dd1b4ad61b745e8d75195cd0b4938086d3b6062d.zip
chromium_src-dd1b4ad61b745e8d75195cd0b4938086d3b6062d.tar.gz
chromium_src-dd1b4ad61b745e8d75195cd0b4938086d3b6062d.tar.bz2
Revert of Stall audio thread after device changes to avoid 3rd party deadlock. (patchset #2 id:20001 of https://codereview.chromium.org/1428413003/ )
Reason for revert: Did not reduce the number of crash reports, so reverting. :( Original issue's description: > Stall audio thread after device changes to avoid 3rd party deadlock. > > Speculative change to see if stalling callbacks into the Windows > audio subsystem after a device change is detected will reduce the > frequency of deadlock in 3rd party WDM drivers. > > This change reroutes both input and output device changes through > the AudioManagerWin, where a simple sleep on the audio thread > stalls any outstanding activity. > > The stall should be fine as the audio thread is used primarily as > an asychronous control channel. I've chosen a stall over a post > delayed task to ensure that any outstanding tasks are prevented. > > If this CL is successful we can experiment with loosening to a > PostDelayedTask will still bear fruit. > > BUG=422522 > TEST=device change works as normal. > > Committed: https://crrev.com/6ffe80ed4dea8b85925ba6dfa2a0cb8f472fb951 > Cr-Commit-Position: refs/heads/master@{#358707} TBR=watk@chromium.org NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG=422522 Review URL: https://codereview.chromium.org/1430423002 Cr-Commit-Position: refs/heads/master@{#358854}
Diffstat (limited to 'extensions/renderer')
0 files changed, 0 insertions, 0 deletions