Electrolysis/Meetings/2016-04-07

From MozillaWiki
Jump to: navigation, search


« previous week | index | next week »

kanru

  • bug 1258312 (p1, crash in mozalloc_abort | NS_DebugBreak | mozilla::ipc::FatalError | mozilla::dom::PBrowserChild::FatalError | mozilla::dom::PBrowserChild::OnMessageReceived) - landed diagnosis patches
  • some reviews

cervantes

  • bug 1257486 (Add additional memory annotations to content process crash reports) - landed
  • bug 1258663 (Annotate system call failure regarding IPC transports) - requested uplifting to aurora
  • bug 1253575 (p2, crash in mozalloc_abort | NS_DebugBreak | mozilla::ipc::GeckoChildProcessHost::OpenPrivilegedHandle) - tentative patch, awaiting feedback

ting

  • bug 1257869 (p1, Window Resize of large page takes seconds to change viewport) - partially fixed by 1255968, still investigating
  • bug 1260736 (3.71 - 5.47% damp / tp5o (windows7-32) regression on push 4de1094b41b2) - regression caused by 1255968, investigating but not blocking

haik

  • bug 1258925 ([e10s] Browser window is resized when click) - Testing a patch, feedback requested, not sure if approach is right
  • bug 1248683 ([e10s] Find bar sometimes gets keypresses in the wrong order when system is under heavy load)

bsmedberg

I may not be attending today: I'm traveling.

Neil

  • bug 1256952 - caret not visible after dragging in some cases - fixed
  • bug 1253486 - hide dropdown when <select> or tab is removed - patch ready
  • bug 1101975 - menu is overriding in-page access keys - working on

placeholder

chutten

Beta46 APZ "experiment"

  • e10s_experiment.ipynb has been run against the new numbers and is up for review.[1]
  • e10s_plugin_lag.ipynb now has BLOCKED_ON_PLUGIN_MODULE_INIT_MS numbers, and is up for review.[2]

mconley

  • Perf P1 - bug 1226564 - [e10s] CACHE V2 telemetry probe regressions
  • Perf P2 bug 1174770 - tpaint regressions (11%-30%) in e10s mode compared to non-e10s mode
    • bug 1251032 - Have ContentParent send RenderFrameInfo down when responding to the CreateWindow sync message
    • bug 1258465 - Only nsIWebBrowserFocus->activate a TabChild once it has composited
      • This is likely not the right way forward, as this breaks the assumption that follow-up messages regarding focus will affect the right things. Closed it out.
    • bug 1262570 - [IME] Defer IME messages until after the content of a new window has been painted and composited
      • Also unlikely to work out, according to masayuki - though he posted a patch to try that combines two sync IME messages in bug 1251063, which will help.
    • Next trick up my sleeve is to try to wait until content has painted before focusing the initial browser of a new window. This should take out the majority of the remaining regression.
    • bug 1261842 - Make initial browser remote sooner if we're defaulting to using remote tabs
      • We probably want to do this anyway - it will, however, require a bit of a deep-dive in nsWindowWatcher to make sure we're doing the right things with DocShells (or lack thereof).
      • Can be investigated if we need more tpaint win. Initial testing suggests this will cause e10s to beat non-e10s for tpaint.
  • Perf P1 - bug 1186585 - 19.66% - 44.84% tps talos regression with e10s enabled and APZ disabled

bobowen

  • bug 1256992 - crash in mozilla::SandboxBroker::SandboxBroker - caused the top crash on Aurora when this was uplifted, fortunately backed out from Beta before it built (thanks Jim). New patch landed, which calls the initialisation earlier. Also with null checks to prevent the crashes and Telemetry to see if the Init is failing.


felipe

gabor

  • bug 1247420 - (m9 1255138 - Resizing windows from JS is not supported in e10s mode) - It took my entire week to fix all the various APIs that can move and resize windows from JS. Every one of them were broken in its own way or non-existing... One thing that is left to do is to figure out a way to tell the tab count from content process, because we have to block these operations if that count is greater than one. GetTargetableShellCount were used for it before but now the treeowner in the content process has no idea about it any longer, nor the tree owner on the parent side since it is only counting the docshells and we would need to count the targetable tab parents instead... Not sure how to do it yet.

mrbkap

  • Not too much to report. I'm going back and forth on a few of my patches trying to get review. Should be done with my m9 and a P1 by next week.

gw280

  • bug 1114647 - content process rename landed and stuck. there's been quite a bit of breakage from it though, mainly due to various pieces of software hardcoding plugin-container.exe in order to do weird things. Spent pretty much all of my week dealing with that.
  • bug 1237769 - e10s disabled for XP when acceleration is on. landed.
  • bug 1251939 - RecvUpdateNoSwap. Spoke to the nical and we have a plan to move the TextureClientPool so that we don't trigger badness when doing texture uploads.
  • Final thing - I'm now on the graphics team rather than e10s but I'm still the team liaison for gfx/e10s issues.

gsvelto

  • bug 1258857 - Enable test_copypaste.html with e10s - Landed a fix in mozilla-central for the issue in the widget code that handled copying HTML elements, evaluating a simpler approach to remove the HTMLFormatConverter which seems unused outside of this case.
  • bug 1258895 - Make test_bug426082.html and test_bug656379-1.html work with e10s - Investigating a fix for this.
  • bug 1222128 - Fix test_bug1011748.html so it works with e10s - Wrote an alternative browser mochitest for this, waiting for feedback if this is the right approach.
  • bug 1257785 - Fix test_offlineNotification.html to work on e10s - Fixed test ready but we might not land it as the entire feature might get removed.

aklotz

  • Continuing work on bug 1261107 - want to get the naive version of this up ASAP so we can quantify the perf situation
  • Have some ideas on reducing chattiness of remoted COM objects if necessary.
Retrieved from "https://wiki.mozilla.org/index.php?title=Electrolysis/Meetings/2016-04-07&oldid=1126530"