Linux server crashes immediately with 2.6.5 kernel

[SirReal 0]

Hi, I'm running gentoo with a 2.6.5 SMP kernel, OFP 1.96.

I've two problems. One with the 1.96 shar update, and one with the server. The shar first.

Don't, as in DO NOT execute that tolower without verifying that you're in the right directory! If someone happens to run that in their root, they're scr*wed royally!

On the server... well, starting the server results in an almost immediate segmentation fault:

1:02:36 Dedicated server created, memory used: 226480 KB

Segmentation fault

Running an strace, the trace ends in:

read(35, "\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0\3\0\1\0\0\0\325\\\1"..., 1024) = 1024

fstat64(35, {st_mode=S_IFREG|0755, st_size=1272540, ...}) = 0

old_mmap(NULL, 1210724, PROT_READ|PROT_EXEC, MAP_PRIVATE, 35, 0) = 0x56bb2000

mprotect(0x56cd4000, 22884, PROT_NONE) = 0

old_mmap(0x56cd4000, 12288, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED, 35, 0x121000) = 0x56cd4000

old_mmap(0x56cd7000, 10596, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) = 0x56

cd7000

close(35) = 0

open("/lib/ld-linux.so.2", O_RDONLY) = 35

read(35, "\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0\3\0\1\0\0\0\360\t\0"..., 1024) = 1024

fstat64(35, {st_mode=S_IFREG|0755, st_size=82432, ...}) = 0

old_mmap(NULL, 79600, PROT_READ|PROT_EXEC, MAP_PRIVATE, 35, 0) = 0x56cda000

mprotect(0x56ced000, 1776, PROT_NONE) = 0

old_mmap(0x56ced000, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED, 35, 0x13000) = 0x56ced000

close(35) = 0

--- SIGSEGV (Segmentation fault) @ 0 (0) ---

+++ killed by SIGSEGV +++

Although without symbols, you probably can't do much with it, here's the backtrace from gdb:

#0 0x08648b25 in ?? ()

#1 0x086c40e0 in ?? ()

#2 0x086c40e0 in ?? ()

#3 0x091115b0 in ?? ()

#4 0xbfffd19c in ?? ()

#5 0x086c44f0 in ?? ()

#6 0x091115b0 in ?? ()

#7 0xbfffd158 in ?? ()

#8 0xbfffd340 in ?? ()

#9 0x00000002 in ?? ()

The machine is a 2.8GHz P4 with 1GB of RAM. The motherboard has an Intel 865 chipset.

/SirReal

[SirReal 0]

As per benu's suggestion:

Quote[/b] ]Search for segfault and libs in this forum. You need older versions of ld-linux.so.2, libc.so.6, libnss_dns.so.2, libnss_files.so.2 and libresolv.so.2 and set LD_LIBRARY_PATH accordingly.

I tried that and it seems to work... so far

/SirReal

[hell_fire 0]

Check that nscd is running, I missed this because Slackware doesn't run NSCD by default, and my crash was at exactly the same place... I'm running a Dual P3-500 box with Slackware 9.1

Edit: Opps, didn't see the thread was in April... :P I was just searchin

Another Edit: Why is there still a reliance on nscd?

[RN Malboeuf 12]

I'm running a Dual P3-500 box

OMG, the worst server i ever run was a 1.4 AMD Athlon

500 mhz lol

[hell_fire 0]

hey! i got it for free, a total of 1GHz between the CPUs, and it doesn't have to handle much, just a few mates at the occasional LAN... see my other thread about the race conditions on SMP boxes though

[shinRaiden 0]

NSCD is a Redhat<tm>, er "Pinktie", or Fedora issue.

Sorry, a bit OT here, but since there was also a tolower.c included in the server.shar iirc, could somebody with a bit of code capability (not me) add a little if...else loop to look for the presence of a flashpoint.cfg and a specific server file to help prevent nuking your qwayke dir or root? Thanks.

[RN Malboeuf 12]

hey! i got it for free, a total of 1GHz between the CPUs, and it doesn't have to handle much, just a few mates at the occasional LAN... see my other thread about the race conditions on SMP boxes though

ofp is single thread, which means it will only use 1 cpu, some will argure it will help but it wont

it will do good as a 10 player server, dont bother hosting CTIs since 2.1 have problems as it is

[hell_fire 0]

Done, posting it in its own thread...

Edit: yeah, i know it won't help, but it does spread the load as the server can have one CPU all by itself :P