mozilla :: #gfx

20 Mar 2017
06:44pulsebotCheck-in: https://hg.mozilla.org/projects/graphics/rev/93b5f8f13084 - Ethan Lin - Bug 1343450 - Convert nsDisplayCanvasBackgroundColor to WebRenderDisplayItemLayer. r=mattwoodrow
06:53pulsebotCheck-in: https://hg.mozilla.org/projects/graphics/rev/0d2742faf779 - Ethan Lin - Bug 1343450 - Keep gfxPrefs sorted alphabetically. r=me
08:30pulsebotCheck-in: https://hg.mozilla.org/projects/graphics/rev/af630ec864b5 - Ethan Lin - Bug 1347871 - Do not create webrender commands if the image renderer doesn't have image container. r=mattwoodrow
13:03mrobinsonmstange: Now that WebRender has support the things you suggested with respect to clipping, do you have some time to talk about how to take advantage of this in Gecko?
13:43jwattis there a doc anywhere describing the various defines, prefs, etc for debugging graphics issues
13:43jwattwith tips on how to use them effectively
14:35mstangejwatt: https://developer.mozilla.org/en-US/docs/Mozilla/Performance/Debugging_a_Performance_Problem has some
14:38jwattmstange: thanks
14:44pulsebotCheck-in: https://hg.mozilla.org/projects/graphics/pushloghtml?startID=399&endID=400 - 175 changesets - Merge m-c to graphics
15:28nicalkats: do you know why PWebRenderBridge::AddImage and other messages are sync ?
15:35katsnical: probably leftover from when they were originally added
15:35katsnical: at that point everything was sync
15:35nicalok thanks
15:39mstangemrobinson: how about tomorrow after gfx standup?
15:44mrobinsonmstange: Sure. Works for me.
15:44mstangecool
16:02pulsebotCheck-in: https://hg.mozilla.org/projects/graphics/rev/c637fe322357 - Mason Chang - Bug 1348408 - Only create inset box shadow WR layers for cases WR supports. r=rhunt
16:15Basjrmuizel: Do you know if we still support Win7 pre-platform update?
16:16BasOr more specifically, if we support D3D11 on systems with win7 pre-platform update.
17:41pulsebotCheck-in: https://hg.mozilla.org/projects/graphics/rev/5974b444127a - Sotaro Ikeda - Bug 1325911 - Update Gleam to v0.4. r=kats
17:41pulsebotCheck-in: https://hg.mozilla.org/projects/graphics/rev/bedec3e899fe - Kartikaya Gupta - Bug 1346856 - Update webrender to dab674d09a51ff5253da0797d129bf29b25896e5. r=jrmuizel
19:50jrmuizelbotond: ping
19:50botondjrmuizel: pong
19:50jrmuizelbotond: can you help me with C++?
20:32dvanderashughes: you mentioned there were still driver crashes in the browser process, when the gpu process was running. do we have an example of one of those?
20:32ashughesto be clear there are driver crashes that don't show up as process_type=gpu in Socorro
20:33ashughesI don't know if GPU Process was running or not
20:33dvanderashughes: oh, that'd be an important distinction
20:33dvanderif it's running, the statement that it eliminates that crash type should be true
20:36ashughesdvander: as an example, igd10iumd32.dll | CContext::EmptyOutAllDDIBindPoints has 137 reports in gpu, 14 reports in content, and 277 reports total for 53.0b in the last week
20:38dvanderashughes: it looks like those are not using the gpu process
20:39ashughesand I can't supersearch on TelemetryEnvironment which would tell me if gpu process was available or not
20:39dvanderi opened a bunch at random and I see windows 7, D2D1.1-
20:39dvanderwhich would indicate no platform update, so no gpu process support
20:39dvanderso i think it's safe to say that it does eliminate that as a source of crashes
20:40dvanderoh, right, i forgot to add an easy supersearch key :) i'll do that today. though it's possible to infer it in that other way
20:41ashughesD2D1.1+: 88, D2D1.1-: 203, D2D1.1?: 291
20:45dvanderashughes: hrm. maybe nota s good as a proxy as I'd hoped. I opened a few d2d1.1+ ones and gpu process has a status of "unavailable"
20:45dvandernot sure how they get d2d1.1 without the platform update
20:47dvanderanyway, it's safe to make the claim that having the gpu process means you won't get driver crashes in the UI
21 Mar 2017
No messages
   
Last message: 151 days and 20 hours ago