Closed Bug 1398402 Opened 3 years ago Closed 3 years ago
.0b3 crashing repeatedly when open 1st unread mail .
78.40 KB, image/jpeg
User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/60.0.3112.113 Safari/537.36 Steps to reproduce: 56.0b3 running in Win10 64 bit. Open 1st unread mail. View message for 1-2 secs, then crash. 8 crashes today. No add ons. Not in Safe Mode. Crash reporter shows crashes, says submitted. 1k data. Running POP client. Actual results: Crash and restart Expected results: Not crash, read mail
Can you please post the crash IDs here, you'll find them in "Help > Troubleshooting Information". Wayne, this one is for you ;-)
bp-3fb37ba8-1103-4074-99d8-c140b0170909,bp-5ea810fc-ccb7-411e-8ed4-3a4370170908, bp-1117d2dd-f9b3-4fb0-8bde-3ab070170908, bp-61f53490-6a3c-44e8-831a-f6d040170908, bp-732121f1-9b6a-45e5-84ab-9da250170908, bp-c4edd9f8-bf82-4a53-bc88-7b3fd0170908, bp-c755a3d9-dddf-49ad-83dd-4c4470170908, bp-82312a18-37f9-4460-bd0e-a13550170908, bp-1f17a89e-3bb4-4886-a425-45ae80170908, bp-a03d98bb-c99b-4e04-972f-8d6f40170907
Thanks the crash is at morkRow.cpp:516 which makes it a duplicate of bug 1368786. We've been tearing our hair out to find a reproducible case. Maybe you would consider making your e-mail data (or a subset that will reproduce the crash) available to us. Can you identify a crashing message or a crashing folder?
Status: UNCONFIRMED → RESOLVED
Closed: 3 years ago
Resolution: --- → DUPLICATE
Duplicate of bug: 1368786
Always in Inbox, reading 1st unread message. Does not happen every time. Open unread, message window appears on screen (not tab), 1-2 seconds, crash. Message is marked read when TBird restarts. Then read same message, go on to other mail, no problem - usually. (Has happened twice in succession if I pick the new oldest unread to open instead of the last read message.) I just tried marking a number of messages as unread and then opening them one at a time. No problem. I shut down TBird and restarted in safe mode and tried again. No problem. So, I can't reproduce it. Currently have 17 pieces of read mail in InBox, which has been about typical. Do messages read and then marked unread have different attributes than initially unread? What final process takes place as the message is displayed in a window. Something that would crash not just the window, but TBird? Some kind of linking or garbage collection or prep for whatever comes next or final display steps in the window (e.g., menu items, updating the unread count in the main window)? Got a new piece of mail, tried opening it. No problem. As far as I know, it has never been the same piece of mail or sender for two crashes. Always in Inbox (main identity). Let me know if you can think of some scenario to try.
Reporter, would you be willing to test a Daily version? If so, do you think that the crash is happening frequently enough for you that if you can't get that Daily version to crash, we can assume that the problem is indeed solved?
Thought I replied? Yes, just tell me how to download a daily. I can't make it happen. It crashed 8 times yesterday and 3 times the day before but has not crashed today. Agree that it feels like uninitalized data or bad pointer. Doesn't feel like race condition or other timing issue.
Yes, you replied in comment #5, but I had more questions in comment #6. We know that the problem can't be produced at will, but I analysed it further and you can read my analysis in bug 1368786 comment #17. I'll compile you a special Daily version. In two hours you can download it from here: https://email@example.com/ Hmm, just noticing that the compile process hasn't started yet, so there might be a technical problem, so please don't hold your breath. I'll check it tomorrow.
OK, I finally managed to compile that version. Download from here: https://firstname.lastname@example.org/try-comm-central-win32/thunderbird-57.0a1.en-US.win32.installer.exe You can install this in parallel to your existing beta version. Please let me know if this fixes the crashes. Note, this is a Daily version and sadly due to bug 1398579 is doesn't allow you to drag messages between folders. That bug is almost fixed, so if that's an impediment, I can get you a totally working version.
Ran the first one for a few hours. Now running the 2nd. Will report any crashes here.
Thanks, first and second one behave the same re. the crashes, only that the second one lets you drag messages again.
It would be great if you could let me know by the end of Tuesday, tomorrow, whether you think that this version solves the problem.
Have read about 30 new messages from InBox plus 30-40 more in a filter-managed folder. No crashes so far. BTW, I've had problems for years on two different machines and several versions of Windows and TBird with unread mail windows disappearing after I opened them, but the mail marked read and no crash. I reported/joined s a bug that should be somewhere in my list. My conjecture is that it was a previous instance of this bug, but the new code "fixed" it and then caused the crash.
Let's wait until Tuesday for a verdict, but thanks for the update, it's looking promising. As for the bug you mentioned, you're talking about bug 711204? Here's your list: https://bugzilla.mozilla.org/buglist.cgi?query_format=advanced&emailtype1=exact&emailreporter1=1&email1=dmccammishjr%40gmail.com&list_id=13772035
Yes big #711204 is the old problem.
So far, 2 possible issues. (1) One or two occurrences of the old problem 711204 new message window opens and then closes immediately. (2) (minor) the new mail icon does not appear consistently when there is new mail. (The icon appearance has been inconsistent for a long time, but in the other direction - 2 icons or the icon does not go away when all mail is read). I'm going to watch and test #1 today as much as I can. Even so, these are better than crashes.
I didn't promise to fix all problems you have with TB, only the crashes ;-) Thanks for your reports so far, maybe a final report tonight since we'll assess the patch you've been testing tomorrow.
Still no crashes on this daily. Definite regression on old problem 711204, however. New message window disappears (crashes?), message marked read.
You need to log in before you can comment on or make changes to this bug.