458227 - JIT crashes jsMSX demo.

Status: RESOLVED DUPLICATE of bug 455547

(Core :: JavaScript Engine, defect, P2)

Description

[nemo]

Clicking on Start, with JIT enabled, crashes.
Disable JIT, works fine.

Was testing to see if JIT improved performance of this demo (interestingly, Webkit was flatlining at 32 ips so I was curious if that was the natural speed of the demo for that ROM, or Webkit).

Since it has been crashing for a week or two, figured I'd better make a note of it.

http://crash-stats.mozilla.com/report/index/30c9825c-8c16-11dd-bd57-001a4bd43e5c
http://crash-stats.mozilla.com/report/index/2735e1fd-8c16-11dd-b43a-001cc45a2c28
A couple of the crashes.

Comment 1

[Matthias Versen [:Matti]]

Can you plaese retest with a build form today (and please add the used build in the bug report)

0  	js3250.dll  	nanojit::LIns::isop  	 js/src/nanojit/LIR.h:365
1 	js3250.dll 	nanojit::LIns::isTramp 	js/src/nanojit/LIR.h:383
2 	js3250.dll 	nanojit::LIns::deref 	js/src/nanojit/LIR.cpp:244
3 	js3250.dll 	nanojit::LIns::oprnd2 	js/src/nanojit/LIR.cpp:627
4 	js3250.dll 	TraceRecorder::set 	js/src/jstracer.cpp:1498
5 	js3250.dll 	TraceRecorder::record_JSOP_GETELEM 	js/src/jstracer.cpp:4650
6 	js3250.dll 	js3250.dll@0x66f15 	
7 	js3250.dll 	js_Invoke 	js/src/jsinterp.cpp:1324
8 	js3250.dll 	js_fun_apply 	js/src/jsfun.cpp:1731
9 	js3250.dll 	js_Interpret 	js/src/jsinterp.cpp:4977
10 	js3250.dll 	js_Execute 	js/src/jsinterp.cpp:1550
11 	js3250.dll 	JS_EvaluateUCScriptForPrincipals 	js/src/jsapi.cpp:5078
12 	xul.dll 	nsJSContext::EvaluateString 	dom/src/base/nsJSEnvironment.cpp:1574
13 	xul.dll 	xul.dll@0x2f2be0

Comment 2

[Robert Sayre]

Still crashes:
http://crash-stats.mozilla.com/report/index/62eafcee-90a7-11dd-9de9-001a4bd43e5c?p=1

Comment 3

[Andreas Gal :gal]

David is working on a crash that looks related.

Comment 4

[nemo]

Right. Sorry.  Should have been more explicit that I was testing with that day's trunk build.

Re: Comment #3 - even if it doesn't warrant duping  yet, mind referencing the bug?  I of course would be interested to know when the patches are committed and such.

Comment 5

[nemo]

http://crash-stats.mozilla.com/report/index/5f9761d7-9557-11dd-9f7e-001cc45a2c28?p=1

The trace appears to have changed slightly, but the interesting thing I wanted to note was while retesting with latest nightly, the first time I tried firefox froze up with 100% of CPU and I killed it after a few minutes.

This trace is from the 2nd attempt, which crashes as usual.

Comment 6

[Alfred Kayser]

Just to confirm, with build:
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.1b2pre) Gecko/20081020 Minefield/3.1b2pre
it still crashes...

Comment 7

[David Anderson [:dvander] - inactive, e-mail if emergency]

Looks the same as the SCUMM bug, the fix works for me on both as well.