Opened 18 years ago
Closed 18 years ago
#3995 closed patch (invalid)
backend segfault (with bt) when deleting in-progress zero-byte recordings
| Reported by: | xris | Owned by: | danielk |
|---|---|---|---|
| Priority: | critical | Milestone: | 0.22 |
| Component: | mythtv | Version: | head |
| Severity: | medium | Keywords: | |
| Cc: | Ticket locked: | no |
Description
Backend keeps crashing during/after recordings and/or while messing with files via mythweb.
Attachments (4)
Change History (16)
comment:1 by , 18 years ago
| Priority: | major → critical |
|---|---|
| Summary: | backend crash (with bt), possibly storage-group related → backend segfault (with bt) when deleting in-progress zero-byte recordings |
comment:3 by , 18 years ago
| Milestone: | unknown → 0.21 |
|---|
comment:4 by , 18 years ago
| Owner: | changed from to |
|---|---|
| Status: | new → assigned |
comment:5 by , 18 years ago
| Status: | assigned → infoneeded |
|---|
We're going to need a backtrace with debugging symbols for this to be fixable...
by , 18 years ago
| Attachment: | gdb.20070927.txt added |
|---|
by , 18 years ago
| Attachment: | gdb.20070930.txt added |
|---|
comment:6 by , 18 years ago
| Status: | infoneeded → assigned |
|---|
Attached what I have. There is an arch problem with debug packages in fedora, so a bunch of symbols are missing.
This problem seems to have gotten better recently. Not sure if it's gone for good or not, but it's definitely more stable in the last week or two.
comment:8 by , 18 years ago
| Type: | defect → patch |
|---|
comment:9 by , 18 years ago
As you already know (but for the record here), the patch has an infinite loop in it that makes it useless.
comment:10 by , 18 years ago
| Status: | assigned → infoneeded |
|---|
xris, I'm going to need a backtrace with the patch applied to figure out what the problem is. (run it in gdb, wait for it to deadlock, then hit Ctrl-C and run "thread apply all bt")
comment:12 by , 18 years ago
| Resolution: | → invalid |
|---|---|
| Status: | infoneeded → closed |
info needed not provided in 6 weeks, closing.
Still happening to me with [14935]... The crashes are pretty easily replicated now. If I get an in-progress zero-byte recording (see [3872]), the backend segfaults immediately when I try to delete or delete/rerecord the recording (I have the "deleted recgroup" stuff turned on, if that makes a difference)