Bug 141402

Summary: Check for self-assignment in Length::operator=(const Length&)
Product: WebKit Reporter: Chris Dumez <cdumez>
Component: CSSAssignee: Chris Dumez <cdumez>
Status: RESOLVED FIXED    
Severity: Normal CC: andersca, commit-queue, ddkilzer, kling, koivisto, simon.fraser
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: Unspecified   
OS: Unspecified   
Attachments:
Description Flags
Patch
none
Patch none

Chris Dumez
Reported 2015-02-09 14:31:55 PST
Check for self-assignment in Length::operator=(const Length&) as calling memset() with the same source and destination addresses has undefined behavior.
Attachments
Patch (1.33 KB, patch)
2015-02-09 14:34 PST, Chris Dumez 		no flags
DetailsFormatted DiffDiff
Patch (1.33 KB, patch)
2015-02-09 14:40 PST, Chris Dumez 		no flags
DetailsFormatted DiffDiff
Show Obsolete (1) View All Add attachment proposed patch, testcase, etc.
Chris Dumez
Comment 1 2015-02-09 14:34:17 PST
Created attachment 246292 [details] Patch
Chris Dumez
Comment 2 2015-02-09 14:40:37 PST
Created attachment 246293 [details] Patch
WebKit Commit Bot
Comment 3 2015-02-09 17:17:48 PST
Comment on attachment 246293 [details] Patch Clearing flags on attachment: 246293 Committed r179860: <http://trac.webkit.org/changeset/179860>
WebKit Commit Bot
Comment 4 2015-02-09 17:17:52 PST
All reviewed patches have been landed. Closing bug.
David Kilzer (:ddkilzer)
Comment 5 2015-02-10 09:18:06 PST
(In reply to comment #0) > Check for self-assignment in Length::operator=(const Length&) as calling > memset() with the same source and destination addresses has undefined > behavior. On Darwin platforms (iOS, OS X), the undefined behavior has been well-defined as being identical to memmove(). (However, that's not true on other platforms, so it's still a good idea to fix this.)
Chris Dumez
Comment 6 2015-02-10 09:49:00 PST
(In reply to comment #5) > (In reply to comment #0) > > Check for self-assignment in Length::operator=(const Length&) as calling > > memset() with the same source and destination addresses has undefined > > behavior. > > On Darwin platforms (iOS, OS X), the undefined behavior has been > well-defined as being identical to memmove(). (However, that's not true on > other platforms, so it's still a good idea to fix this.) Oh, I did not know that. If I do a "man memcpy" on my Mac, it says "If dst and src overlap, behavior is undefined".
Note You need to log in before you can comment on or make changes to this bug.