Opened 14 years ago
Closed 14 years ago
#8547 closed defect (fixed)
Mythcommflag broken in trunk
Reported by: | Owned by: | danielk | |
---|---|---|---|
Priority: | minor | Milestone: | 0.24 |
Component: | MythTV - Mythcommflag | Version: | Master Head |
Severity: | low | Keywords: | |
Cc: | Ticket locked: | no |
Description
As late as today (revision 25063) mythcommflag has serious problems.
The trivial one is from Changeset 24694 for trunk/mythtv/programs/mythcommflag/main.cpp.
@@ -868,5 +872,5 @@
QString outputfilename = QString::null;
- QString chanid;
+ uint chanid;
QString starttime; QString allStart = "19700101000000";
which makes chanid an uninitialized uint, which causes the test to fail in
--- @@ -1285,6 +1290,6 @@
}
- (!chanid.isEmpty() && starttime.isEmpty()))
+ (chanid && starttime.isEmpty()))
{
VERBOSE(VB_IMPORTANT, "You must specify both the Channel ID "
---
As a result mythcommflag with no arguments will always fail with the message above and terminate.
The more serious problem is a non-deterministic one. Even when mythcommflag is invoked with a file argument, it will often segfault in the middle of processing the file. If run repeatedly on the same file it segfaults at different frame numbers in each run. I'm attaching gdb output of successive runs.
Attachments (1)
Change History (5)
comment:1 Changed 14 years ago by
Milestone: | unknown → 0.24 |
---|---|
Owner: | changed from cpinkham to danielk |
Priority: | major → minor |
Severity: | medium → low |
Status: | new → accepted |
The chanid bug is mine. The random crashes are unrelated and at least a few months old.
comment:2 Changed 14 years ago by
I didn't see these segfaults on a build from the trunk as recently as May 30. What ever segfaults I've seen in the past have been on corrupted recordings and were deterministic. These occur on good recordings and get caught in different frames in repeated trials on the same recording. Sometimes it even gets all the way through.
comment:3 Changed 14 years ago by
re. the mythcommflag crashes, can you please attached the output of a crash instance with '-v playback' logging.
It would also help if you could confirm whether or not you see this behaviour if you roll back to r24971 (i.e. before the libmythui-osd branch was merged into trunk).
gdb output showing segfaults at different frames