null pointer dereferencing / segfault (reproducible)

[hp-mythtv.org@…]

current cvs mythtv segfaults when beginning to record.

a) manually scheduling through mythweb: backend segfaults at scheduledrecording.cpp:84 (type is 0x0)

b) manually scheduling through frontend: frontend segfaults with one-liner complaining about the empty type

No errors when compiling. Please instruct me for further information gathering, as I just learned how to "backtrace" with gdb (hence the line 84 above :) )

[Isaac Richards]

This will be marked as invalid unless a backtrace as specified in the docs is attached.

http://www.mythtv.org/docs/mythtv-HOWTO-21.html#ss21.2

[hp-mythtv.org@…]

I did the following:

a) start mythbackend b) get on mythweb to schedule_manually.php c) select an immediate, two minute recording on any FTA-DVB-Channel d) backend segfaults

I will add: a) output of gdb b) backtrace according to instructions

ATTENTION: backtrace might be "corrupted", because after about 80'000 zero lines, gdb crashes.

[hp-mythtv.org@…]

backtrace according to instructions

[hp-mythtv.org@…]

gdb console output up until crash

[Isaac Richards]

Since your gdb appears broken, you need to switch threads manually and get a backtrace for each one. 'info threads' and 'thread X' to switch.

[hp-mythtv.org@…]

bt full per thread

[hp-mythtv.org@…]

I added a backtrace of all 14 threads. I found that those five threads being stopped in pthread_cond_wait have an indefinitely long backtrace and several others end in "Previous frame inner to this frame (corrupt stack)?".

I'm using stock Mandriva Linux 2005 on Athlon64. I really hope, the development tools are ok.

[Isaac Richards]

(In [6802]) Add the sr_* stuff to the backend-only build (gui items can still be used on the backend, since they're accessors for the data as well as the ui item). Fixes #88. Use pointers, not references in sr_* stuff.