Bad pty handling 2.1.1 (was: Oops in kNFS code (vanilla 2.1.1) (?))

Philippe Troin (phil@fifi.org)
01 Feb 1999 12:58:42 -0800


Seems like there's a problem with ptys.
I'm trying to run the egcs testsuite in two egcs trees (a patched one
and an original one) in parallel in a two-CPU SMP box.
Everytime it ended up crashing the machine.

When the test suites are running I get bunches of "Warning: dev
(03:07) tty->count(0) != #fd's(2) in do_tty_hangup" (with varying
figures for device and numbers).

Everytime (3 times so far) it ended up crashing in [schedule+210/864]
like the enclosed oops. Once when called from xprt_receive(enclosed
Oops), once from interruptible_sleep_on (oops available) and once from
sys_select (oops available too).

The kernel I use is NOT compiled with Unix98 PTY support.

I'm trying to have a quicker and easier way to reproduce it (probably
with several processes racing to grab PTY's).

I'm using vanilla 2.1.1 (config available upon request)

Phil.

I wrote:

> Got this ooops while compiling egcs, and the machine died later on...
>
> ceramic kernel: Oops: 0000
> ceramic kernel: CPU: 0
> ceramic kernel: EIP: 0010:[<20302037>]
> ceramic kernel: EFLAGS: 00010202
> ceramic kernel: eax: 20302037 ebx: c0236000 ecx: 00000000 edx: 00000001
> ceramic kernel: esi: c1995000 edi: c7ac3de8 ebp: 00000000 esp: c7ac3d94
> ceramic kernel: ds: 0018 es: 0018 ss: 0018
> ceramic kernel: Process collect2 (pid: 24829, process nr: 108, stackpage=c7ac3000)
> ceramic kernel: Stack: 00000000 c7ac2000 c7ac3de8 c7ac3dd4 c1995000 00000000 c011172e c1995000
> ceramic kernel: c7ac2000 00000086 c7ac3de8 c0188b20 c6e31058 c7ac3e4c 00000000 c0242000
> ceramic kernel: c7ac3df0 c0111e5c c7ac2000 c7ac3e4c 00000001 c7ac2000 c7ac3ea8 00000000
> ceramic kernel: Call Trace: [schedule+210/864] [xprt_receive+116/164] [sleep_on+92/176] [__rpc_execute+247/436] [rpc_execute+83/96] [rpc_do_call+214/272] [xprt_receive_status+0/44]
> ceramic kernel: [call_status+0/232] [xprt_timer+0/100] [nfs_proc_readlink+94/164] [nfs_follow_link+88/248] [do_follow_link+71/132] [lookup_dentry+324/440] [__namei+41/92] [sys_access+148/252]
> ceramic kernel: > 7[<c0108co00>] [<c0100d02b>] <4>Code: _<1>Uunable to hande kernl paing request at )virtual address 2030207h
>
> NFS client & server were running knfsd+kmountd+kstatd and had volumes
> cross-mounted on each other... Running vanilla 2.1.1 on both hosts.
>
> Phil.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/