Segmentation Fault in Linux v0.9.9 client

[gbusler]

Has anybody else been getting segmentation faults with the v0.9.9 linux client? It's happened twice on me so far.

All intermediate blocks transfer OK but when the last intermediate block gets transferred and the client need to get another complete block to work on it seg faults.

[Sun Nov 10 21:47:23 2002] n.high = 2173220 . 1 blocks left in test
[Sun Nov 10 21:52:35 2002] iteration: 2180000/2224092 (98.02%) k = 27653 n = 2224077
[Sun Nov 10 22:00:15 2002] iteration: 2190000/2224092 (98.47%) k = 27653 n = 2224077
[Sun Nov 10 22:07:54 2002] iteration: 2200000/2224092 (98.92%) k = 27653 n = 2224077
[Sun Nov 10 22:16:10 2002] iteration: 2210000/2224092 (99.37%) k = 27653 n = 2224077
[Sun Nov 10 22:24:31 2002] iteration: 2220000/2224092 (99.82%) k = 27653 n = 2224077
[Sun Nov 10 22:27:42 2002] logging into server
[Sun Nov 10 22:27:43 2002] login successful
[Sun Nov 10 22:27:44 2002] n.high = 2223760 . 0 blocks left in test
[Sun Nov 10 22:28:01 2002] residue: 33e5caf256f8f194
[Sun Nov 10 22:28:01 2002] completed proth test(k=27653, n=2224077): result 3
[Sun Nov 10 22:28:01 2002] connecting to server
[Sun Nov 10 22:28:02 2002] logging into server
Segmentation fault

Grant

[Alien88]

as posted in the 0.9.9 thread:

And.. if anyone happens to see the linux client core (0.9.8 was coring on one of my machines), can you please run:

gdb sb core
bt

and paste the output here.. thanks

-Alien88

[CrazySaint]

Ok, I've finally got a core dump - here's the output:

------------------------------------------
[wingnut@localhost SOB]$ ./sb sclient.conf
[Mon Nov 11 01:41:51 2002] client process [v0.9.9] invoked
[Mon Nov 11 01:41:51 2002] priority set to low
[Mon Nov 11 01:41:51 2002] got k and n from cache
[Mon Nov 11 01:41:51 2002] restarting proth test from cache (k=27653, n=2376537) [52.8%]
Segmentation fault (core dumped)
[wingnut@localhost SOB]$ dir
cache core.2382 sb sclient.conf sclient.conf~ sclient.log ÿ¿\030õÿ¿ z2376537
[wingnut@localhost SOB]$ gdb sb core.2382
GNU gdb 5.2.1-2mdk (Mandrake Linux)
Copyright 2002 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB. Type "show warranty" for details.
This GDB was configured as "i586-mandrake-linux-gnu"...
Core was generated by `./sb sclient.conf'.
Program terminated with signal 11, Segmentation fault.
#0 0x081fd97c in ??00F4 ()
(gdb) bt
#0 0x081fd97c in ??00F4 ()
#1 0x40115000 in ?? ()
#2 0x08056900 in block_loop ()
#3 0x08056892 in main ()
#4 0x0805a1b2 in __libc_start_main (main=0x80566f4 <main>, argc=2, ubp_av=0xbffff8c4, init=0x80480b4 <_init>,
fini=0x80a9a00 <_fini>, rtld_fini=0, stack_end=0xbffff8bc) at ../sysdeps/generic/libc-start.c:129
(gdb)
------------------------------------------

[Firebirth]

I got a similar message, but I guess, it was because I copied the 0.9.9 version over the 0.9.8 version, while the 0.9.8 version was running; it stopped at the same point as you, don't know if that is the case here though!

[jjjjL]

I think I've tracked the bug down.

I'm running 18 linux clients on random computers to test it out. So far so good. It's appears to be a bad call to free(). Hopefully this is it so I can work on compiling a BeOS client :)

-Louie

[huraxprax]

Another segfault appears when ending the client with a sigterm or control-c.
(client 0.99)
caught signal 15. dumping unfreed memory:
unfreed ptr 0x8248e70, size 36, in file client.c:350 at time 1037350473
and similar.

[res0r9lm]

I got the same thing trying to close client but signal 2 when shutting down