|Summary:||SVG DOM manipulation crash|
|Product:||WebKit||Reporter:||Cris Neckar <cdn>|
|Component:||SVG||Assignee:||Stephen Chenney <schenney>|
|Severity:||Normal||CC:||commit-queue, fmalita, inferno, krit, pdr, schenney, webkit.review.bot, zimmermann|
|Version:||528+ (Nightly build)|
Description Cris Neckar 2013-02-01 15:27:01 PST
Created attachment 186162 [details] repro Originally filed by firstname.lastname@example.org at https://crbug.com/171363 Chrome Version: 25.0.1364.36 Operating System: Ubuntu 10.04 URL (if applicable) where crash occurred: http://shinobi.dempsky.org/~matthew/misc/chrome-svg-crash.html Can you reproduce this crash? Yes, 100% reliable. What steps will reproduce this crash? (or if it's not reproducible, what were you doing just before the crash)? 1. Navigate to http://shinobi.dempsky.org/~matthew/misc/chrome-svg-crash.html 2. Click the left orange square. 3. Crash. The bug exists when there are multiple event handlers on a node in an SVG use tree and the first handler causes the tree to be rebuilt. The target of the event remains the now removed node from the original use tree. I am not certain that this is a security issue, but I suspect it is due to a heap-use-after-free scenario. Maybe the ref-counting on the target is enough to avoid the heap-use-after-free. It is NOT a security bug in the example because it is a Chrome mouse event handler that is second to be invoked, and Chrome cannot access a WebNode for the deleted node (it's null).
Comment 2 WebKit Review Bot 2013-02-11 16:38:23 PST
Comment on attachment 187703 [details] Patch Clearing flags on attachment: 187703 Committed r142548: <http://trac.webkit.org/changeset/142548>
Comment 3 WebKit Review Bot 2013-02-11 16:38:26 PST
All reviewed patches have been landed. Closing bug.