Heh. I don't think this is related to naving's startup crash, but rather this would be a side-effect of bienvenu's checkin last night. With naving's patch backed out, I crashed whilst downloading POP3 mail. This time, folderInfo is null when we try to get it's charsetptr: rv = folderInfo->GetCharPtrCharacterSet(getter_Copies(charset)); nsMsgDBFolder::GetCharset(nsMsgDBFolder * const 0x02f15a9c, unsigned short * * 0x03df9eec) line 233 + 38 bytes nsMsgFilter::MatchHdr(nsMsgFilter * const 0x03dcc4a0, nsIMsgDBHdr * 0x03df9af0, nsIMsgFolder * 0x02f15a9c, nsIMsgDatabase * 0x03ce2d30, const char * 0x010a51fc, unsigned int 1058, int * 0x006ef648) line 362 nsMsgFilterList::ApplyFiltersToHdr(nsMsgFilterList * const 0x03dcc640, int 1, nsIMsgDBHdr * 0x03df9af0, nsIMsgFolder * 0x02f15a9c, nsIMsgDatabase * 0x03ce2d30, const char * 0x010a51fc, unsigned int 1058, nsIMsgFilterHitNotify * 0x03dccec8, nsIMsgWindow * 0x03c5a1c0) line 165 + 47 bytes nsParseNewMailState::ApplyFilters(int * 0x006ef6f4, nsIMsgWindow * 0x03c5a1c0) line 1644 + 104 bytes nsParseNewMailState::PublishMsgHeader(nsIMsgWindow * 0x03c5a1c0) line 1555 nsPop3Sink::IncorporateComplete(nsPop3Sink * const 0x03d0ab80, nsIMsgWindow * 0x03c5a1c0) line 470 nsPop3Protocol::HandleLine(char * 0x010e8488, unsigned int 3) line 2430 + 38 bytes nsMsgLineBuffer::ConvertAndSendBuffer() line 270 + 58 bytes nsMsgLineBuffer::BufferInput(const char * 0x03380aa4, int 2) line 206 + 8 bytes nsPop3Protocol::RetrResponse(nsIInputStream * 0x03d0cf40, unsigned int 1262) line 2249 + 19 bytes nsPop3Protocol::ProcessProtocolState(nsIURI * 0x03d0aa44, nsIInputStream * 0x03d0cf40, unsigned int 345, unsigned int 1262) line 2872 + 16 bytes nsMsgProtocol::OnDataAvailable(nsMsgProtocol * const 0x03d0a4c0, nsIRequest * 0x03d0cd30, nsISupports * 0x03d0aa44, nsIInputStream * 0x03d0cf40, unsigned int 345, unsigned int 1262) line 261 + 32 bytes nsOnDataAvailableEvent::HandleEvent() line 193 + 70 bytes nsARequestObserverEvent::HandlePLEvent(PLEvent * 0x03dfac64) line 80 PL_HandleEvent(PLEvent * 0x03dfac64) line 590 + 10 bytes PL_ProcessPendingEvents(PLEventQueue * 0x00adddf0) line 520 + 9 bytes _md_EventReceiverProc(HWND__ * 0x00000ebc, unsigned int 55030, unsigned int 0, long 11394544) line 1071 + 9 bytes KERNEL32! bff7363b() KERNEL32! bff94407() 006e8a0a()
My emails was lost when they were downloading and the crash occured. :(
I don't see either of these problems, so there must be something about your configuration (or your build). I set up my pop3 account to get new mail on startup and download the messages on checking for new mail, and the pop3 account is not my default account. Is that different from your setup?
POP3 -> Sheela.
The pop3 account is my default account, but this crash seems to happen in db code so are you sure it is pop3 specific? Anyway, this is a regular windows build from the tip, and I don't see anything strange in general with it.
I'm not sure how much opportunity you have to see other strangeness if you're crashing on startup of mail. I don't see this crash (or the other one you're seeing). I'd suggest a clobber build of mailnews.
Sure, I'll back naving out, clobber, and get back to you.
Sorry for the delay. This crash is still happening in my debug build, and is triggered or possibly revealed by naving's patch from bug 103683. It's a blocker for me, since I can't debug.
you can debug this. Try stepping through the code above your crash where we get the db folder info and db, and see why the dbfolder is null, but the return value is NS_OK.
Not to spam you David, but this prevented me from verifying the POP3 password leak last night (and earlier today). I just deleted my tree and repulled, so I'll see when that's done. I was crashing, but couldn't get a stack (in fact, MSVC6 became hung). This was with an opt build with MOZ_PROFILE=1, running under Purify (when I ran the build by itself, I never crashed, and yes, I deleted Purify's cache).
Just adding a datapoint. There are currently no talkback crashes on this.
D'oh! Didn't I close this? I meant to close this some days ago when I pulled a fresh tree. It is not occuring any longer.
verifying worksforme based on the comments. Reopen if you start seeing the crash again.