#43 segfault in weed_plugin_info_init

[Andrew Pantyukhin]

GDB output attached

[Andrew Pantyukhin]

GDB

[Salsaman]

Logged In: YES
user_id=158916
Originator: NO

What OS/distro are you using ?

[Andrew Pantyukhin]

Logged In: YES
user_id=1477658
Originator: YES

FreeBSD 7.x-CURRENT on amd64
I also see this issue here (on IRIX): http://forums.nekochan.net/viewtopic.php?p=100106#100106
Deleting all the .wo files works for me.

[Salsaman]

Logged In: YES
user_id=158916
Originator: NO

Yes, this is a known issue in IRIX, which is why I asked. May I enquire, when you compiled for FreeBSD, did you get any compilation errors/warnings in lives-plugins/weed-plugins ? My guess for IRIX was that since it was not accepting the -rdynamic flag, it was something to do with this.

[Andrew Pantyukhin]

LiVES build log

[Andrew Pantyukhin]

Logged In: YES
user_id=1477658
Originator: YES

Yes, in fact there are warnings (long long int vs long int)

I'm attaching the full build log

BTW, lives is now in our ports collection: http://www.freshports.org/multimedia/lives/
File Added: lives-build.log

[Salsaman]

Logged In: YES
user_id=158916
Originator: NO

These warnings can be ignored, they come from 32bit code which is not used by AMD64.

However, it is interesting that you were getting the segfault even with -rdynamic. The implication is that a value cannot be assigned to static function pointer, even with -rdynamic set.

[Salsaman]

Logged In: YES
user_id=158916
Originator: NO

Hi,
it's been a while, but I think I finally came up with a fix for this. Please can you test with the current CVS version.

Thanks for your cooperation.

[Andrew Pantyukhin]

Logged In: YES
user_id=1477658
Originator: YES

I'm afraid it's quite the same with 0.9.8.10, both on i386 and amd64 (FreeBSD 8-current here):
Program received signal SIGSEGV, Segmentation fault.
0x00000008040154ea in weed_plugin_info_init ()
from /usr/local/share/lives/plugins/effects/realtime/weed/alien_overlay.wo
(gdb) bt

0 0x00000008040154ea in weed_plugin_info_init ()

from /usr/local/share/lives/plugins/effects/realtime/weed/alien_overlay.wo

1 0x0000000804016ad7 in weed_setup ()

from /usr/local/share/lives/plugins/effects/realtime/weed/alien_overlay.wo

2 0x0000000000434d0a in load_weed_plugin (

plugin_name=0x10b3e80 "alien_overlay.wo",
plugin_path=0x10ec580 "/usr/local/share/lives/plugins/effects/realtime/weed/alien_overlay.wo") at effects-weed.c:2215

3 0x00000000004359b2 in weed_load_all () at effects-weed.c:2371

4 0x0000000000420b25 in lives_init (ign_opts=Variable "ign_opts" is not available.

) at main.c:682

5 0x0000000000422485 in main (argc=1, argv=0x7fffffffe520) at main.c:1365

[Salsaman]

Logged In: YES
user_id=158916
Originator: NO

Strange. I could understand why it was crashing before - I was trying to convert between a function pointer and a void pointer, which is not such a good idea. I changed this by passing a pointer to a function pointer instead.

I do not understand why it is still crashing.

[Salsaman]

Hi,
if you are still monitoring this, would you mind testing with the latest release, 1.2.0 ?

I realised that the overloaded memory functions were not defined in the inital bootstrap process, so the bootstrap now uses standard memory functions only.

Hopefully this will resolve the problem.