Summary: | VeryHigh priority loads are actually loading at VeryLow priority | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Ben Nham <nham> | ||||||||||
Component: | WebKit2 | Assignee: | Ben Nham <nham> | ||||||||||
Status: | RESOLVED FIXED | ||||||||||||
Severity: | Normal | CC: | achristensen, cdumez, commit-queue, koivisto, nham, webkit-bug-importer | ||||||||||
Priority: | P2 | Keywords: | InRadar | ||||||||||
Version: | WebKit Nightly Build | ||||||||||||
Hardware: | Unspecified | ||||||||||||
OS: | Unspecified | ||||||||||||
Bug Depends on: | 203594 | ||||||||||||
Bug Blocks: | |||||||||||||
Attachments: |
|
Description
Ben Nham
2019-10-25 10:02:50 PDT
Lol Nice find! Created attachment 382243 [details]
Patch
Created attachment 382249 [details]
Patch
The attached patch shows a consistent 20-25% improvement for cold loading of nytimes-article in PLT5 on the perf bots, which is exactly what I expected since that is the main resource load most affected by connection congestion in the test. Because we load 13 websites in PLT5, this shows up as a 1.5-2.25% cold PLT5 improvement. This shows up as a neutral change in overall PLT5 because: 1. We weight warm/cold loads in PLT5 80:20. 2. Some of the resource load timings changed enough in the warm loads that we ended up capturing more JS before deciding on first paint, which slightly regressed warm PLT5 (by ~1%) in certain configurations. However, I don't think we should hold off on making this change since I'm planning to do a pass on first paint timings soon. Comment on attachment 382249 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=382249&action=review > Source/WebCore/platform/network/cf/ResourceRequestCFNet.h:43 > + // TODO(nham): switch VeryLow back to 0 priority when CFNetwork fixes <rdar://problem/56621205> We usually just say FIXME: without a name. Anyone is free to fix problems and if the FIXME originator is important it can be found with blame. Created attachment 383480 [details]
Patch
Rebased and replaced TODO with FIXME. Comment on attachment 383480 [details] Patch Rejecting attachment 383480 [details] from commit-queue. Failed to run "['/Volumes/Data/EWS/WebKit/Tools/Scripts/webkit-patch', '--status-host=webkit-queues.webkit.org', '--bot-id=webkit-cq-01', 'validate-changelog', '--check-oops', '--non-interactive', 383480, '--port=mac']" exit_code: 1 cwd: /Volumes/Data/EWS/WebKit /Volumes/Data/EWS/WebKit/Source/WebKit/ChangeLog neither lists a valid reviewer nor contains the string "Unreviewed" or "Rubber stamp" (case insensitive). Full output: https://webkit-queues.webkit.org/results/13248130 Created attachment 383484 [details]
Patch
Comment on attachment 383484 [details]
Patch
Add reviewed-by line to WebKit ChangeLog.
Comment on attachment 383484 [details] Patch Clearing flags on attachment: 383484 Committed r252431: <https://trac.webkit.org/changeset/252431> All reviewed patches have been landed. Closing bug. |