Opened 14 years ago
Closed 14 years ago
#8006 closed defect (fixed)
segfault in frontend after deleting recordings
Reported by: | Owned by: | danielk | |
---|---|---|---|
Priority: | blocker | Milestone: | 0.23 |
Component: | MythTV - General | Version: | head |
Severity: | medium | Keywords: | |
Cc: | Ticket locked: | no |
Description
After watching live TV for a few seconds and switching channels, I then go the the Watch Recordings screen and delete all of the recordings from the Live TV group. The last recording seems to hang around and I delete it multiple times, but it still stays there on the screen. I wait for the backend to log the following
2010-02-01 16:11:06.413 GetRecordBasename found no entry for 1004 @ 2010-02-01T16:10:32 2010-02-01 16:11:06.413 MainServer: HandleGenPreviewPixmap: Unable to find file locally, unable to make preview image. 2010-02-01 16:11:09.850 ERROR retrieving program info when trying to delete program for chanid 1004 recorded at Mon Feb 1 16:10:32 2010. Recording will NOT be deleted.
Then I exit the Watch Recordings screen on the frontend and I get the segfault.
GDB backtrace attached.
Attachments (3)
Change History (6)
comment:1 Changed 14 years ago by
Milestone: | unknown → 0.23 |
---|---|
Owner: | changed from Isaac Richards to danielk |
Status: | new → assigned |
comment:2 Changed 14 years ago by
Priority: | minor → blocker |
---|
Can you run with the attached patch and provide the mythfrontend log it produces up until the segfault?
Changed 14 years ago by
Attachment: | mythfrontend.log added |
---|
attached log with patch applied as requested
comment:3 Changed 14 years ago by
Resolution: | → fixed |
---|---|
Status: | assigned → closed |
(In [23599]) Fixes #8006. Refs #7195. Fixes some regressions in preview generation. Most importantly this addresses the segfault. But it also removes some of the cruft in the PreviewGenerator? constructor that had built up to deal with places that called the preview generator without setting pathname as per the PreviewGenerator? documentation. This means we aren't doing the very expensive StorageGroup? search quite as many times.
This also saves the pathname after doing a StorageGroup? search in the PlaybackBoxHelper?.
There are still a number of ways left to speed up the PreviewGenerator?: For instance, using fork instead of running an external program to get fault tolerance, which will also address #7136. We also do not need to open a new command socket if we convert the backend preview generation to use a non-blocking call and just send a MythEvent? when we're done. But I wanted to limit this commit to really only fix regressions since we're in feature freeze (the pathname saving is pbbh is a bit of a stretch, but we actually did do this before in a different way).
Backtrace from frontend while doing the actions in the description