malloc bugreport on Mozilla exit

[fyysik]

When starting SeaMonkey from terminal and then exit app at some time, complaint from libroot appears in terminal: "seamonkey-bin: src/system/libroot/posix/malloc/superblock.h:194:void BPrivate::superblock::putBlock(BPrivate::block *): getNumAvailable() < getNumBlocks() ./run-mozilla.sh: line 424: 192 Kill Thread "$prog" ${1+"$@"}"

I wish to know what happens from libroot/malloc POV, this problem may happen in BeOS aswell, as Mozilla for BeOS exit code (and bootstrap code btw) is neither standard nor trivial.

[bonefish]

That's an assert() in the memory allocator, probably caused by a double free(), free()ing not allocated memory, or some memory corruption. Actually the failed assert() should call the debugger(), where you could at least see the back trace. Was anything printed in the serial debug output (respectively syslog)?

[axeld]

The allocator likely has its own assert implementation that we might want to replace :)

[fyysik]

Replying to bonefish:

where you could at least see the back trace. Was anything printed in the serial debug output (respectively syslog)?

Unfortunately my laptop lacks serial port (inspite there is "debug port" listed in BIOS - with resources which remind normal COM! very closely. That unfortunate Thinkpad R40e - very simplified model:(

[bonefish]

Replying to axeld:

The allocator likely has its own assert implementation that we might want to replace :)

Unless I overlooked something, it doesn't. arch-specific.h includes <assert.h>.

Replying to fyysik:

Replying to bonefish:

where you could at least see the back trace. Was anything printed in the serial debug output (respectively syslog)?

Unfortunately my laptop lacks serial port (inspite there is "debug port" listed in BIOS - with resources which remind normal COM! very closely. That unfortunate Thinkpad R40e - very simplified model:(

As long as Haiku itself doesn't crash, you should have a syslog (/var/log/syslog). It is enabled by default.