Issue metadata
Sign in to add a comment
|
SVG Paths render with an offset when hardware acceleration is activated (probably only when in an iFrame)
Reported by
tombi...@wix.com,
Mar 7 2018
|
||||||||||||||||||||||
Issue descriptionUserAgent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_13_3) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/64.0.3282.186 Safari/537.36 Example URL: https://codepen.io/tombigel/full/paMKGN/ Steps to reproduce the problem: 1. Add an SVG to an html file 2. Create a new html with the SVG html embedded as an iFrame 3. Add a hardware accelerated CSS transform on the <svg> element What is the expected behavior? The SVG paths should render the same way with and without hardware acceleration What went wrong? On MacOS, some or all the paths re-render with a slight visual offset. The paths BBox values does not change. If you don't see the described behaviour, try to change the window size, sometimes this does not happen on small windows. We tested this on multiple Macs with multiple browsers, could only recreate on Chrome. Attached an animated GIF taken from the Codepen Does it occur on multiple sites: Yes Is it a problem with a plugin? No Did this work before? Yes Does this work in other browsers? Yes Chrome version: 64.0.3282.186 Channel: stable OS Version: OS X 10.13.3 Flash Version: The attached Codepen recreates the bug only when in edit/full mode where the SVG is in an iframe. When i opened the Codepen in debug mode (https://s.codepen.io/tombigel/debug/paMKGN) where the SVG is in the top frame, I couldn't recreate
,
Mar 7 2018
Is this the issue with OOPIF and SVG that we have fixed?
,
Mar 7 2018
Doesn't sound (or look) like it would be OOPIF related. I'd suspect some GPU raster issue.
,
Mar 8 2018
Able to reproduce the issue on mac 10.13.3 using chrome reported version #64.0.3282.186, latest stable #65.0.3325.146 and latest canary #67.0.3364.0. Bisect Information: ===================== Good build: 63.0.3239.0 Bad Build : 64.0.3240.0 Change Log URL: https://chromium.googlesource.com/chromium/src/+log/b902d921a5dbdfb7fde34c5ddfb442fab19f8dfa..1a24eadedcaf0ccdaac8648c812dffad37b9fce5 Skia Change Log URL: https://skia.googlesource.com/skia.git/+log/e60ca7eae432..f543a60ef06a From the above change log suspecting below change Change-Id: I8b872af26150d93cf39cf8eeba23c91e1decace3 Reviewed-on: https://skia-review.googlesource.com/58863 fmalita@ - Could you please check whether this is caused with respect to your change, if not please help us in assigning it to the right owner. Thanks...!!
,
Mar 8 2018
https://skia-review.googlesource.com/c/skia/+/57942 (Skia commit 5698c8a8a0dcce3bdabab3551f068bda61d741a3) seems more likely.
,
Mar 8 2018
Wondering whether crbug.com/819966 is a duplicate. That's SVG paths as a background image in a sticky container where the path gets sometime rendered at an offset, sometimes not, and not all the paths.
,
Mar 8 2018
I can't seem to duplicate this even in full-screen mode -- does it only happen with Retina displays? fmalita@ was able to duplicate on his laptop, but not with ToT build.
,
Mar 8 2018
crbug.com/819966 might be better place to start, on a higher profile site. It's not yet reduced though. I can look at doing that this afternoon if my other work goes smoothly.
,
Mar 9 2018
For those who can reproduce this, this patch (applied to third_party/skia) will probably fix it: https://skia-review.googlesource.com/c/skia/+/113428. If you could verify that would be appreciated.
,
Mar 9 2018
The following revision refers to this bug: https://skia.googlesource.com/skia/+/a64a5850d946a713a38eeafb68e57a91e00a18cc commit a64a5850d946a713a38eeafb68e57a91e00a18cc Author: Jim Van Verth <jvanverth@google.com> Date: Fri Mar 09 19:50:20 2018 Fix issue with negative translations of small paths Truncating negative values will produce a negative fractional part when you subtract out the integer value, which causes havoc with rendering the path and searching for it in the cache. What we want is the floor. Bug: 819966 , 819522 Change-Id: I2feb451e28bfb863111479f70fd99fa1299abac0 Reviewed-on: https://skia-review.googlesource.com/113428 Commit-Queue: Jim Van Verth <jvanverth@google.com> Reviewed-by: Brian Salomon <bsalomon@google.com> [modify] https://crrev.com/a64a5850d946a713a38eeafb68e57a91e00a18cc/src/gpu/ops/GrSmallPathRenderer.cpp
,
Mar 9 2018
I'll verify on Monday. Thanks for the rapid fix.
,
Mar 12 2018
The NextAction date has arrived: 2018-03-12
,
Mar 12 2018
,
Mar 15 2018
The following revision refers to this bug: https://skia.googlesource.com/skia/+/46dfb183c8d3ea386e03fb2c4f74d58203dac7dc commit 46dfb183c8d3ea386e03fb2c4f74d58203dac7dc Author: Jim Van Verth <jvanverth@google.com> Date: Thu Mar 15 14:22:27 2018 Fix issue with negative translations of small paths Truncating negative values will produce a negative fractional part when you subtract out the integer value, which causes havoc with rendering the path and searching for it in the cache. What we want is the floor. No-Tree-Checks: true No-Try: true No-Presubmit: true Bug: 819966 , 819522 Change-Id: I2feb451e28bfb863111479f70fd99fa1299abac0 Reviewed-On: https://skia-review.googlesource.com/113428 Commit-Queue: Jim Van Verth <jvanverth@google.com> Reviewed-By: Brian Salomon <bsalomon@google.com> Reviewed-on: https://skia-review.googlesource.com/114496 Reviewed-by: Jim Van Verth <jvanverth@google.com> [modify] https://crrev.com/46dfb183c8d3ea386e03fb2c4f74d58203dac7dc/src/gpu/ops/GrSmallPathRenderer.h [modify] https://crrev.com/46dfb183c8d3ea386e03fb2c4f74d58203dac7dc/src/gpu/ops/GrSmallPathRenderer.cpp
,
Mar 22 2018
Issue 823614 has been merged into this issue. |
|||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||
Comment 1 by tkent@chromium.org
, Mar 7 2018Labels: -Type-Bug Needs-Bisect Type-Bug-Regression