Issue metadata
Sign in to add a comment
|
Wrong background-color for <html> and <body> with display:table styling
Reported by
fris...@jeka.info,
Jan 23 2018
|
||||||||||||||||||||
Issue descriptionUserAgent: Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:57.0) Gecko/20100101 Firefox/57.0 Steps to reproduce the problem: 1. See attached testcase or visit https://nagelpilz.de.cool/ What is the expected behavior? Background should be solid red What went wrong? there is a correct colored area in the middle, but a wrong pale red (orange) area around Did this work before? Yes - same error in chrome 63, but I'm pretty sure it worked some releases ago Does this work in other browsers? Yes Chrome version: Version 66.0.3329.1 Channel: n/a OS Version: 6.1 (Windows 7, Windows Server 2008 R2) Flash Version:
,
Jan 24 2018
,
Jan 25 2018
Able to reproduce the issue on reported chrome version 66.0.3329.1 using Ubuntu 14.04 hence providing Bisect Info Note: Unable to reproduce the issue on chrome version 66.0.3329.1 using Windows-7 and Mac 10.13.1, also same issue is not observerd on other Ubuntu 14.04 machines. Bisect Info: ================ Good build: 61.0.3143.0 Bad build: 61.0.3144.0 You are probably looking for a change made after 482903 (known good), but no later than 482904 (first known bad). https://chromium.googlesource.com/chromium/src/+log/8b4643f67dc1c6b142a152aa2b76c771287e3a22..1894d423068be735560e0171922c833c4091b5d1 Reviewed-on: https://chromium-review.googlesource.com/551092 @Christopher Cameron: Please confirm the issue and help in re-assigning if it is not related to your change. Thanks!
,
Jan 25 2018
,
Jan 30 2018
Please attach a screenshot. I suspect that there is a ~1 pixel value difference (which is the consequence of rounding error).
,
Jan 30 2018
Additional info: Resizing window changes the affected areas (red gets larger or smaller and/or unsymmetrical)
,
Jan 31 2018
,
Jan 31 2018
,
Mar 2 2018
GPU triage: ping?
,
Mar 5 2018
Hm, this reproducable with a much simpler testcase: <!DOCTYPE html> <title>Test Issue 804864</title> <body style=background:red> Background should be evenly red, no red rectangle on orange area
,
Mar 5 2018
And issue 805893 clearly isn't a duplicate.
,
Mar 5 2018
btw, this is not Linux only, my OS is Windows 7
,
Mar 8 2018
Please attach your ICC profile, and also go to about:gpu, print the result, and attach it as a PDF. Solid color tiles are represented as sRGB through the whole pipeline, while non-solid-color tiles are drawn in the output color space. A consequence can be that rounding error in the color conversion pipeline can result in the colors not matching exactly. With your ICC profile, I'll be able to do the math myself, to see what went wrong and where.
,
Mar 9 2018
Thanks ... I've installed your profile and compared the about:gpu results on my test machine, and it gives the same values as you have:
{primaries:[[0.3530,0.4527,0.1585,],[0.1882,0.7526,0.0592,],[0.0106,0.0357,0.7786,],], transfer:1.0000*x + 0.0000 if x < 0.0000 else (1.0000*x + 0.0000)**2.2004 + 0.0000, matrix:RGB, range:FULL}
But the bug still doesn't reproduce for me on Linux or Windows 10 (I don't have a Windows 7 setup, but I don't see how it would be different).
,
Mar 9 2018
Btw, this is very similar to issue 766175 , which was somewhat-mysteriously fixed.
,
Mar 9 2018
Also of note is that - the "correct" color is the one that we get for the tiles with content, and is 255,0,14 - the "incorrect" color is the one that we get for the solid color tiles, and is 255,43,14 So this isn't rounding error -- it's just somehow completely wrong. I wonder if this could be because of one of the driver bug workarounds. If you go to about:flags and enable "Override software rendering list", does the issue go away?
,
Mar 9 2018
enabling "Override software rendering list" does not fix it but changes behaviour. In both cases https://nagelpilz.de.cool/i/ and testcase from comment 11 the red rectangle is wider (covers the scren horizontally)
,
Mar 9 2018
The NextAction date has arrived: 2018-03-09
,
Mar 9 2018
The size of the rectangle is an artifact of our layer tiling scheme (you can see the boundaries in devtools). Hmm. Without being able to reproduce this to debug it, I'm pretty stumped.
,
Mar 9 2018
One more thought is that this may be related to issue 820233, where we don't support A2B profiles.
,
Mar 9 2018
Per comment 3 this is a regression in Chrome 61 Good build: 61.0.3143.0 Bad build: 61.0.3144.0 Is it not easy to identify what's going on?
,
Nov 16
ccameron: Is there any action we can take here, or should we drop priority?
,
Nov 29
I don't see any action to take -- I haven't been able to reproduce this |
|||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||
Comment 1 by fris...@jeka.info
, Jan 23 2018312 bytes
312 bytes View Download