#355 SIGILL, Illegal instruction in libFLAC crashes audacious

[Alister Hood]

Running puppy linux 4.1.1, with an old (maybe 2000/2001) AMD cpu.
I was having trouble with audacious crashes on startup, caused by libFLAC, and also aqualung crashes (I can't remember the details of that - I think it may have been when I tried to play a FLAC in aqualung).
I recompiled libFLAC myself, which stopped the audacious crashes on startup (I'm not sure about the aqualung crashes). But if I run `gdb audacious` libFLAC still crashes it:

(gdb) run
Starting program: /usr/bin/audacious

Program received signal SIGILL, Illegal instruction.
0xb3b46fd8 in FLAC__cpu_info () from /usr/lib/libFLAC.so.8
(gdb) backtrace
#0 0xb3b46fd8 in FLAC__cpu_info () from /usr/lib/libFLAC.so.8
#1 0xb3b69395 in ?? () from /usr/lib/libFLAC.so.8
#2 0x0828d5a4 in ?? ()
#3 0xb3b84ac3 in ?? () from /mnt/sdb3/audacious/lib/audacious/Input/flacng.so
#4 0xb3b84000 in ?? ()
#5 0xb3b692de in ?? () from /usr/lib/libFLAC.so.8
#6 0xb3b348f0 in ?? () from /usr/lib/libFLAC.so.8
#7 0xb3b89720 in ?? () from /mnt/sdb3/audacious/lib/audacious/Input/flacng.so
#8 0xb3b898f4 in main_info ()
from /mnt/sdb3/audacious/lib/audacious/Input/flacng.so
#9 0xb3b898fc in test_info ()
from /mnt/sdb3/audacious/lib/audacious/Input/flacng.so
#10 0xb3b87600 in ?? () at seekable_stream_callbacks.c:125
from /mnt/sdb3/audacious/lib/audacious/Input/flacng.so
#11 0xb3b87680 in ?? () at seekable_stream_callbacks.c:101
from /mnt/sdb3/audacious/lib/audacious/Input/flacng.so
#12 0x08283b18 in ?? ()
#13 0xb3b89720 in ?? () from /mnt/sdb3/audacious/lib/audacious/Input/flacng.so
#14 0xb3b898f4 in main_info ()
from /mnt/sdb3/audacious/lib/audacious/Input/flacng.so
#15 0xb3b898fc in test_info ()
from /mnt/sdb3/audacious/lib/audacious/Input/flacng.so
#16 0xbfa449e8 in ?? ()
---Type <return> to continue, or q <return> to quit---
#17 0xb3b867c8 in flac_init () at plugin.c:117
Backtrace stopped: frame did not save the PC

The audacious people sent me here, and also told me to make sure my libflac "isn't compiled to use features only present on newer CPUs". I'm not sure how to do that, but since I didn't specify any options when I configured it (except --prefix), if it is then I would think that was itself a bug.

[lvqcl]

After a quick googling I think that's normal:

http://www.raspberrypi.org/forums/viewtopic.php?p=155085
http://stackoverflow.com/questions/15071625/gdb-ddd-program-received-signal-sigill