This started to happen in 1107 build. Also in 2000110808Mtrunk. Browser fails to load a lot of items like bug 50357, and starts to crash. Setting cache size to zero prevents this crash. Stack strace is pretty short: Calling chain using A6/R1 links Back chain ISA Caller Return addresses on the stack Stack Addr Frame Addr ISA Caller 0B83D6D6 68K 05FE100E 0B83D558 PPC 1A655660 nsScriptSecurityManager::CanGetProperty(JSContext*, const nsID&, nsISupports*, nsIInterfaceInfo*, unsigned short, long)+00030 0B83D538 PPC 1A8ED79C xpcPerThreadData::GetData()+000E8 0B83D4EA 68K 00000A3C 0B83D4DC 68K 09A96186 0B83D422 68K 00000A38 0B83D404 PPC 1A85A7D4 nsHTTPPipelinedRequest::AdvanceToNextRequest()+000E8 0B83D3EC PPC 1A85A800 nsHTTPPipelinedRequest::AdvanceToNextRequest()+00114 0B83D352 68K 0000000A 0B83D2F4 PPC 6806E228

I suspect the fix for bug 53843 that was checked in 11-06 has something to do with this.

It crashes far less often when Cache is zero, but it is still crashing. Changing summary and component toreflect this point. Stack trace of a crash with zero cache: Calling chain using A6/R1 links Back chain ISA Caller Return addresses on the stack Stack Addr Frame Addr ISA Caller 0B83CF60 PPC 6806E228 0B83CF50 68K 0C0CADB0 HGETVOL+FD370 0B83CF38 68K 00408A64 0B83CF24 68K FFC1736A _UnimplTrap+0108C 0B83CEFC 68K 0083E31C 0B83CEF8 68K FFC10C24 _CursorDeviceDispatch+01684 0B83CED8 68K 0064ECCE 0B83CE30 PPC 1954A660 nsScriptSecurityManager::CanGetProperty(JSContext*, const nsID&, nsISupports*, nsIInterfaceInfo*, unsigned short, long)+00030 0B83CE20 68K 08B0AD56 0B83CE10 PPC 1A93579C xpcPerThreadData::GetData()+000E8 0B83CCDC PPC 1A8AC7D4 nsHTTPPipelinedRequest::AdvanceToNextRequest()+000E 8 0B83CCC4 PPC 1A8AC800 nsHTTPPipelinedRequest::AdvanceToNextRequest()+0011 4 0B83CC72 68K 0000000A 0B83CC64 PPC 1BAE7A68 OTOpenEndpointPriv+00034 0B83CC5C PPC FFD01508 LowLevelExceptionHandler+002A0 0B83CC2A 68K 0000000A 0B83CBCC PPC 6806E228

Again, a typo. The bug 53849. Sorry, I am getting used to my new key board.

>I suspect the fix for bug 53843 [that is, 53849] that was checked in 11-06 has something to do >with this. Two things: 1. Could you provide a simple, reproducable test case. 2. What makes you think that 53849 has anything to do with this?

1. I can not. because it is occuring randomly. It just keeps failing to load and crash. Going back and forth between unfinished pages may increase the likelyhood. Furthermore, I do not have enough knowledge to figure out from these stack traces exactly what is happening. 2. In 53849, there is a description of another bug fix that is not in Bugzilla. When I queried for similar bug in Bugzilla, 53849 was the only one that seemed to have something to do with nsScriptSecurityManager, and in it was the description of a check-in for the unpublicized fix done on 11-06.

A description of what you did with the product leading up to the problem is the next best thing to a reproducible test case.

The behavior of the back and forward buttons was also changed for framesets. I think that was bug 46828, but I might be off since that's from memory. If you were using back and forward to recreate the problem, it may be related.

I poked around, testing new profiles. My conclusion is the prefs.js file has something in it that causes the trouble. I can easily transfer crash by copying prefs.js file from the old profile to a freshly created profile folder, even though the preferences dialog shows nothing suspect.

The HTTP pipelining in Advanced item in Preferences was turned on. Turning it off stabilized the browser. Changing to Dupe of bug 55714. I will test further a bit to see if there is anything else. *** This bug has been marked as a duplicate of 55714 ***

verified dup