Bug 18264 - Adding 2,000,000 spans containing different fonts is slow
Summary: Adding 2,000,000 spans containing different fonts is slow
Status: NEW
Alias: None
Product: WebKit
Classification: Unclassified
Component: WebCore Misc. (show other bugs)
Version: 528+ (Nightly build)
Hardware: Mac OS X 10.5
: P2 Normal
Assignee: Nobody
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2008-04-01 16:57 PDT by Eric Seidel (no email)
Modified: 2011-08-09 11:54 PDT (History)
2 users (show)

See Also:

Attachments
test case (crashes Safari) (1.15 KB, text/html)
2008-04-01 16:57 PDT, Eric Seidel (no email) 		no flags Details
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Eric Seidel (no email) 2008-04-01 16:57:20 PDT 
Crash in array instance

I ran into this while trying to make Safari run out of GDI objects. :)  See test case.

Process:         Safari [18170]
Path:            /Applications/Safari.app/Contents/MacOS/Safari
Identifier:      com.apple.Safari
Version:         3.1 (5525.13)
Build Info:      WebBrowser-55251300~1
Code Type:       X86 (Native)
Parent Process:  launchd [341]

Date/Time:       2008-04-01 16:54:35.691 -0700
OS Version:      Mac OS X 10.5.2 (9C31)
Report Version:  6

Exception Type:  EXC_BAD_ACCESS (SIGSEGV)
Exception Codes: KERN_PROTECTION_FAILURE at 0x00000000bf7ffffc
Crashed Thread:  0

Thread 0 Crashed:
0   com.apple.JavaScriptCore      	0x00340cb2 KJS::ArrayInstance::mark() + 114
1   com.apple.JavaScriptCore      	0x0033fd6d KJS::PropertyMap::mark() const + 221
2   com.apple.JavaScriptCore      	0x0033fd6d KJS::PropertyMap::mark() const + 221
3   com.apple.JavaScriptCore      	0x0033fd6d KJS::PropertyMap::mark() const + 221
4   com.apple.JavaScriptCore      	0x0033fd6d KJS::PropertyMap::mark() const + 221
5   com.apple.JavaScriptCore      	0x0033fd6d KJS::PropertyMap::mark() const + 221
6   com.apple.JavaScriptCore      	0x0033fd6d KJS::PropertyMap::mark() const + 221
.
.
.
Comment 1 Eric Seidel (no email) 2008-04-01 16:57:59 PDT 
Created attachment 20277 [details]
test case (crashes Safari)
Comment 2 Gavin Barraclough 2011-08-08 16:53:30 PDT 
I don't think this is a JSC bug any more - I don't see a crash in ToT.  Looks like this test takes an unreasonably long time to run, so might still be worth looking at the performance at some point.
Comment 3 Alexey Proskuryakov 2011-08-09 10:50:29 PDT 
Is that worth tracking as a bug? Two million spans is a large amount.
Comment 4 Gavin Barraclough 2011-08-09 11:54:39 PDT 
Hey Alexey,

Sam suggested that this test may be exposing some n^2 behaviour in adding sibling spans, so could be worth profiling.  It may be worth nothing that this also thrashes on font rendering, since the contents of no two spans are the same.

I'll bow out of the conversation to leave the decision to someone more expert in the field - if you don't think this is a useful test then please do close.

G.