Closed
Bug 1241364
Opened 8 years ago
Closed 8 years ago
Weird rendering problem occurs when opening a new window in Nightly build (20160120030239)
Categories
(Core :: Graphics, defect)
Tracking
()
RESOLVED
DUPLICATE
of bug 1239861
Tracking | Status | |
---|---|---|
firefox45 | --- | unaffected |
firefox46 | + | fixed |
firefox47 | + | fixed |
People
(Reporter: magicp.jp, Assigned: bas.schouten)
References
Details
(Keywords: regression, reproducible, Whiteboard: [gfx-noted])
Attachments
(4 files)
User Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:46.0) Gecko/20100101 Firefox/46.0 Build ID: 20160120030239 Steps to reproduce: Please refer to attached video. 1. Start Nightly 46.0a1 BuildID: 20160120030239 with new profile 2. Install the Classic Theme Restorer v1.4.6.1 and Themes Menu v1.24 https://addons.mozilla.org/addon/classicthemerestorer https://addons.mozilla.org/addon/themes-menu 3. In Customize, move Themes Menu icon to nav-bar 4. Open a new window Actual results: Weird rendering problem occurs when opening a new window. Please refer to attached image. It occurs from Build 20160120030239. No problem in other versions and builds. Expected results: Weird rendering problem should not be occurred.
Has STR: --- → yes
status-firefox45:
--- → unaffected
status-firefox46:
--- → affected
Component: Untriaged → Graphics
OS: Unspecified → Windows 10
Product: Firefox → Core
Hardware: Unspecified → x86_64
Please use mozregression to narrow this down: http://mozilla.github.io/mozregression/install.html
Keywords: regression,
regressionwindow-wanted
Whiteboard: [gfx-noted]
(In reply to Anthony Hughes, QA Mentor (:ashughes) from comment #2) > Please use mozregression to narrow this down: > http://mozilla.github.io/mozregression/install.html pushlog_url: https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=b67316254602a63bf4e568198a5c7d3288a9db27&tochange=2e50b83954e62d52d2ef294e850c4380d457d96a
Updated•8 years ago
|
Severity: normal → major
OS: Windows 10 → Windows
Hardware: x86_64 → All
Comment 4•8 years ago
|
||
[Tracking Requested - why for this release]: Regression
tracking-firefox46:
--- → ?
Updated•8 years ago
|
Version: Trunk → 46 Branch
Updated•8 years ago
|
status-firefox47:
--- → affected
tracking-firefox47:
--- → ?
Is this still an issue with the latest nightlies? I'm not quite sure the regression range from comment 3 can be trusted (bustage sometimes messes things up.)
Flags: needinfo?(magicp.jp)
Assignee | ||
Comment 7•8 years ago
|
||
I can't reproduce this as far as I can tell. But also looking at the video, this is a brief, transient, non-invasive artifact that doesn't seem like it's something we should actually be tracking, and doesn't seem of severity 'major' to me.
(In reply to Milan Sreckovic [:milan] from comment #6) > Is this still an issue with the latest nightlies? I'm not quite sure the > regression range from comment 3 can be trusted (bustage sometimes messes > things up.) Yes, this is still an issue with the latest nightlies. I confirmed the mozregression again, and then I got different regression range. pushlog_url: https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=d2b4d101d00691f835c4f109524a3506d5f51626&tochange=4f897d0f494b281d9f52c6fc1f4848ab414a85a8
Flags: needinfo?(magicp.jp)
That shouldn't have done it, but - Bas, it should be easy enough to throw in gfxCriticalNote into the modified code and show us what the max was before, and what the max is now, after the change in bug 1239743, and have :magicp run the custom build and report the values captured in about:support? I agree that the issues are brief and transient, but it'd be good to understand a bit more. After we understand it a bit more, we can probably stop tracking, based on the effect, if it turns out this is all.
Flags: needinfo?(bas)
So, perhaps no need for that custom build, but when I did the same, I'm seeing values like 1073741824 get adjusted to 16384. Those numbers make sense? The second one certainly looks like a reasonable "16k" limit, but did we really have a 1 << 30 limit before? Are we running into zoom situations? Perhaps a good custom build would be one where we (temporarily) introduce a preference where we can set the value we cap with, instead of the max texture size, and see at which point these artifacts go away.
Reporter | ||
Comment 11•8 years ago
|
||
Assignee | ||
Comment 12•8 years ago
|
||
(In reply to Milan Sreckovic [:milan] from comment #10) > So, perhaps no need for that custom build, but when I did the same, I'm > seeing values like 1073741824 get adjusted to 16384. Those numbers make > sense? The second one certainly looks like a reasonable "16k" limit, but > did we really have a 1 << 30 limit before? Are we running into zoom > situations? > > Perhaps a good custom build would be one where we (temporarily) introduce a > preference where we can set the value we cap with, instead of the max > texture size, and see at which point these artifacts go away. I'm guessing it's a bug in the extension requesting those values. I'm totally willing to believe values could have gotten that's large before. Zoom shouldn't really apply here as these are actual 'widget' sizes. I can make that build if we really want to but I'm not entirely sure what we could do with the information that comes out of it :-). We can't draw windows bigger than the maximum texture size. I think I might know what's going on though. This problem probably has to do with before us failing to draw completely and sort of choking up, while now we might be creating a backbuffer at the max texture size, probably not drawing anything sensible into it, and that gets sized down briefly by windows as it resizes the window to the 'proper' size. We could verify it's something like this by creating a build that clear's the back buffer right after creating it if we really want to investigate this in detail.
Flags: needinfo?(bas)
That would be good. I think I noticed different behaviour in the debug build, when things are slow, when the first window is created. I'm not sure I can describe it, but it feels like there is an extra "flash", though for me it's just white.
Assignee: nobody → bas
Reporter | ||
Comment 14•8 years ago
|
||
Current situation: Nightly 46.0a1 (2016-02-04) with my existing profile. weird rendering is more clearly.
Reporter | ||
Comment 15•8 years ago
|
||
(In reply to magicp from comment #14) > Current situation: Nightly 46.0a1 (2016-02-04) with my existing profile. ^ 47.0a1
Comment 16•8 years ago
|
||
I want to add that if happens also with a new clean fresh profile without any addons (extensions & plugins). Mostly it occurs on vertical menu list. Just go to bookmark button and look on some folders with many bookmarks to see flashing.
Reporter | ||
Comment 17•8 years ago
|
||
(In reply to Virtual_ManPL [:Virtual] - (ni? me) from comment #16) > I want to add that if happens also with a new clean fresh profile without > any addons (extensions & plugins). Mostly it occurs on vertical menu list. > Just go to bookmark button and look on some folders with many bookmarks to > see flashing. Sure! I can reproduce this bug with a new profile. This flashing is more clearly with any addons. Especially, CTR button color on titlebar is easy to find when flashing.
Reporter | ||
Comment 18•8 years ago
|
||
In 47.0a1 (2016-02-06), rendering works fine. something changed? 46.0a2 still reproduce.
Reporter | ||
Comment 19•8 years ago
|
||
(In reply to magicp from comment #18) > In 47.0a1 (2016-02-06), rendering works fine. something changed? 46.0a2 > still reproduce. pushlog_url: https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=5ca719a45f5d7fe36296efe7e638b866523000db&tochange=59ae52ad58801ef311710dd91fc94f73201cc91f
Comment 20•8 years ago
|
||
(In reply to magicp from comment #18) > In 47.0a1 (2016-02-06), rendering works fine. something changed? 46.0a2 > still reproduce. Are you saying your problem is now fixed after bug 1239861? Thanks!
Flags: needinfo?(magicp.jp)
Reporter | ||
Comment 21•8 years ago
|
||
(In reply to Mason Chang [:mchang] from comment #20) > Are you saying your problem is now fixed after bug 1239861? Thanks! Yes, it is fixed after bug 1239861 without aurora.
Flags: needinfo?(magicp.jp)
Updated•8 years ago
|
Status: NEW → RESOLVED
Closed: 8 years ago
Resolution: --- → DUPLICATE
Comment 23•8 years ago
|
||
It's only fixed on Firefox 47. Firefox 46 is still affected. What are we going to do with it?
Has Regression Range: --- → yes
Flags: needinfo?(mchang)
Keywords: regressionwindow-wanted → reproducible
Comment 25•8 years ago
|
||
Fixed in 46 now according to bug 1239861.
You need to log in
before you can comment on or make changes to this bug.
Description
•