Closed Bug 10797 Opened 20 years ago Closed 20 years ago
[PP][blocker] Necko: Messenger hangs trying to send or recieve mail
Composing a new message and sending it (clicking the Send button) or trying to recieve new mail (clicking Get Msg button) freezes up messenger (and apprunner). (This occurs in today's NECKO build, it was not a problem with yesterdays M9 w/o NECKO build) Step: Self explanatory - try sending a message and after you freeze and restart messenger, compose a new message in 4.7 and try to recieve it with messenger and freeze again. Machine and Build: MacOS 8.5.1 with build 1999072908 Note: Not a problem with Win32 or Linux builds
cc: pmock and mscott. Scott- would this belong to you?
Component: Necko → Networking-Mail
Summary: [PP] Messenger hangs trying to send or recieve mail → [PP] Necko: Messenger hangs trying to send or recieve mail
just changing component. Per mscott suggestion, we don't need a Necko component for mail/news.
I think this should be assigned to me and not gagan. taking it from him...setting milestone to m9 and accepting.
adding hangas to the cc line. Paul, can you try sending me a message with your build?
Summary: [PP] Necko: Messenger hangs trying to send or recieve mail → [PP][blocker] Necko: Messenger hangs trying to send or recieve mail
We need to get this fixed soon. Peter sees this too.
This should be P1, marking so, hope that's ok
Hey Paul, can we sit at your machine and step through the debugger this morning?
Whiteboard: Looks like a necko problem with blocking writes on the mac
This appears to be a necko specific Mac bug with blocking writes and the socket transport. It looks like the socket transport on the Mac hangs whenever we make a blocking write on the socket interface. Calling write on the output stream for the socket never returns and we freeze the Mac. I haven't stepped into the write to see where we freeze yet. This is why we freeze both when sending messages and getting new mail. These are the two netwerk operations we perform on the Mac which involve blocking writes to the socket transport. Adding rick potts and warren to the cc list as this is looking like a core necko problem and not a mailnews specifc one.
Assignee: mscott → gordon
Status: ASSIGNED → NEW
Component: Networking-Mail → Necko
Re-assigning to gordon as this is a necko bug (changing the component too =)). Gordon, you should be able to test this using the necko socket test instead of messenger. I'm performining a blocking write on the nsSocketTransport.
Hey Gordon, do you have any updates on this bug? This is really hurting the mac mail client as you can't use it for dogfood at all right now. Please let me know if I can help.
I'm working on bringing up the socket test on Mac. I hope to have more information later today.
we see this when we first get pop mail. here's a stack trace: 00000000 PPC 0EB4AAD4 03309D80 PPC 0EB49EE8 main+00060 03309D30 PPC 0EB49738 main1(int, char**)+00E2C 03309B90 PPC 0EA504D8 nsAppShellService::Run()+00020 03309B50 PPC 0E91C46C nsAppShell::Run()+00050 03309AD0 PPC 0E91D210 nsMacMessagePump::DoMessagePump()+00044 03309A80 PPC 0E91D518 nsMacMessagePump::DispatchEvent(int, EventRecord*)+00188 03309A30 PPC 0EAA9E9C Repeater::DoRepeaters(const EventRecord&)+00034 033099F0 PPC 0E8FEEF0 nsToolkit::RepeatAction(const EventRecord&)+00044 033099B0 PPC 0E13921C nsEventQueueServiceImpl::ProcessEvents()+0002C 03309970 PPC 0E1347A4 nsHashtable::Enumerate(int (*)(nsHashKey*, void*, void*), void*) +0003C 03309930 PPC 0E1BC63C PL_HashTableEnumerateEntries+00060 033098C0 PPC 0E133FE4 _hashEnumerate(PLHashEntry*, int, void*)+00030 03309880 PPC 0E13917C EventDispatchingFunc(nsHashKey*, void*, void*)+0003C 03309840 PPC 0E150228 nsEventQueueImpl::ProcessPendingEvents()+00018 03309800 PPC 0E1BB508 PL_ProcessPendingEvents+00078 033097B0 PPC 0E1BB63C PL_HandleEvent+0004C 03309770 PPC 0EB2A224 nsStreamListenerEvent::HandlePLEvent(PLEvent*)+0002C 03309720 PPC 0EB2B104 nsOnDataAvailableEvent::HandleEvent()+00044 033096D0 PPC 0DABFC14 nsMsgProtocol::OnDataAvailable(nsIChannel*, nsISupports*, nsIInp utStream*, unsigned int, unsigned int)+00078 03309680 PPC 0DAA943C nsPop3Protocol::ProcessProtocolState(nsIURI*, nsIInputStream*, u nsigned int, unsigned int)+00374 03309570 PPC 0DAA5C10 nsPop3Protocol::SendUsername()+00104 033094E0 PPC 0DAA5584 nsPop3Protocol::SendData(nsIURI*, const char*)+0002C 033094A0 PPC 0DABFB40 nsMsgProtocol::SendData(nsIURI*, const char*)+00088 03309440 PPC 0E1864CC nsBufferOutputStream::Write(const char*, unsigned int, unsigned int*)+000D4
*** Bug 10971 has been marked as a duplicate of this bug. ***
This bug may also be causing http://bugzilla.mozilla.org/show_bug.cgi?id=11749 so that News on the Mac is not able to be used.
*** Bug 11749 has been marked as a duplicate of this bug. ***
So I've got the TestSocketTransport up and running on the Mac, and it seems to work fine for both the async and sync cases. I'm trying to reproduce the problem with my build of AppRunner, but I'm currently blocked by strange menu behavior. Therefore I am unable to set the user name and server for mail. Patrick Beard's build (pulled at the same time) doesn't have the same problem, so I'm going to ask some of the XPFE folks how my setup could be causing this.
*** Bug 11760 has been marked as a duplicate of this bug. ***
*** Bug 11752 has been marked as a duplicate of this bug. ***
Status: ASSIGNED → RESOLVED
Closed: 20 years ago
Resolution: --- → FIXED
I checked in 2 fixes that I think will fix this problem. Please try it again. Thanks to Gordon and Patrick for tracking it down.
I'm pulling and building the tip now on my mac. I'll report back with the results.
I'm able to get new news articles and post a news article from the mac. I'm also able to send email and get IMAP mail. nice work, gordon, patrick and warren.
Great work guys. Using today Mac commercial seamonkey build 1999-08-13-m9 build, I can successfully send and receive on POP, IMAP, and news. Marking bug as verified. Note: I find that I have to wait for the status bar says done before I can perform the next mail operation.
Bulk move of all Necko (to be deleted component) bugs to new Networking component.
You need to log in before you can comment on or make changes to this bug.