Bug 125770

Summary: [CSS Regions][CSS Shapes] ASSERTION FAILED: m_segmentRanges.size() < m_segments.size()
Product: WebKit Reporter: Zoltan Horvath <zoltan>
Component: CSSAssignee: Zoltan Horvath <zoltan>
Status: RESOLVED FIXED    
Severity: Normal CC: commit-queue, esprehn+autocc, glenn, kondapallykalyan, WebkitBugTracker
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: Unspecified   
OS: Unspecified   
Bug Depends on:    
Bug Blocks: 89256    
Attachments:
Description Flags
Patch
none
Patch - address Darin's comment
bjonesbe: review-, bjonesbe: commit-queue-
Patch none

Zoltan Horvath
Reported 2013-12-16 01:35:40 PST
If we have an e.g. up-side-down triangle, then at the bottom part of the triangle the content doesn't fit, then it needs to be adjusted. The adjustment to the next region happens correctly, but the current region/shape doesn't get updated, which leads to the the shape mismatch, which leads to layout error / assertion.
Attachments
Patch (5.88 KB, patch)
2013-12-16 01:53 PST, Zoltan Horvath 		no flags
DetailsFormatted DiffDiff
Patch - address Darin's comment (5.92 KB, patch)
2013-12-16 13:06 PST, Zoltan Horvath 		bjonesbe: review-
bjonesbe: commit-queue-
DetailsFormatted DiffDiff
Patch (6.47 KB, patch)
2014-01-06 15:34 PST, Zoltan Horvath 		no flags
DetailsFormatted DiffDiff
Show Obsolete (2) View All Add attachment proposed patch, testcase, etc.
Zoltan Horvath
Comment 1 2013-12-16 01:53:38 PST
Created attachment 219304 [details] Patch
Darin Adler
Comment 2 2013-12-16 10:04:52 PST
Comment on attachment 219304 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=219304&action=review > Source/WebCore/rendering/RenderBlockLineLayout.cpp:1231 > + RenderRegion* regionAtNewLogicalHeight = regionAtBlockOffset(logicalHeight()); > + if (!currentRegion->isLastRegion() && currentRegion != regionAtNewLogicalHeight) { > + currentRegion = regionAtNewLogicalHeight; > + shapeInsideInfo = currentRegion->shapeInsideInfo(); > + } What if regionAtBlockOffset returns nullptr?
Zoltan Horvath
Comment 3 2013-12-16 13:06:12 PST
Created attachment 219345 [details] Patch - address Darin's comment > What if regionAtBlockOffset returns nullptr? It should never happen at this point. I added an ASSERT for checking it.
Bem Jones-Bey
Comment 4 2014-01-06 12:00:06 PST
Comment on attachment 219345 [details] Patch - address Darin's comment View in context: https://bugs.webkit.org/attachment.cgi?id=219345&action=review > Source/WebCore/rendering/RenderBlockLineLayout.cpp:1226 > + // flows into the next region with a shape, we need to update the current shape and region This code seems like it belongs in the "overflowsToNextRegion" case above, as it is very strange to have two different places in the same function that move to the next region, and it also makes the code hard to follow. If it doesn't make the patch too big, I would prefer if this is refactored to make that possible. If not, then a FIXME comment here is ok, with a reference to a bug to do the refactor.
Zoltan Horvath
Comment 5 2014-01-06 15:34:29 PST
Created attachment 220463 [details] Patch
Bem Jones-Bey
Comment 6 2014-01-06 15:50:28 PST
Comment on attachment 220463 [details] Patch That looks a lot better!
WebKit Commit Bot
Comment 7 2014-01-06 16:38:50 PST
Comment on attachment 220463 [details] Patch Clearing flags on attachment: 220463 Committed r161384: <http://trac.webkit.org/changeset/161384>
WebKit Commit Bot
Comment 8 2014-01-06 16:38:52 PST
All reviewed patches have been landed. Closing bug.
Note You need to log in before you can comment on or make changes to this bug.