Oops in 2.2.18 with pppd dial in server

Oops in 2.2.18 with pppd dial in server

[root]

[-- Attachment #1: Type: text/plain, Size: 194 bytes --]


Alan,

I was able to reproduce this oops with a somewhat more reliable ksymoops (I was ready for this nasty bug this time).  Looks like the problem is in the sockets
code.

See attached.

Jeff

[-- Attachment #2: oops.txt --]
[-- Type: text/plain, Size: 602 bytes --]

ksymoops 0.7c on i686 2.2.18pre21.  Options used
EIP:  0010: [<c013365d>]
Using defaults from ksymoops -t elf32-i386 -a i386
Call Trace: [<c014c43d>] [<c014cb15>] [<c012d79a>] [<c012dda8>] [<c014cb82>] [<c014d7cb>] [<c010a038>]
Warning (Oops_read): Code line not seen, dumping what data is available

>>EIP; c013365d <get_empty_inode+15/98>   <=====
Trace; c014c43d <get_fd+91/9c>
Trace; c014cb15 <sock_create+85/d8>
Trace; c012d79a <permission+1a/2c>
Trace; c012dda8 <open_namei+1dc/34c>
Trace; c014cb82 <sys_socket+1a/7c>
Trace; c014d7cb <sys_socketcall+5f/1a8>
Trace; c010a038 <system_call+34/38>

Re: Oops in 2.2.18 with pppd dial in server

[Alan Cox]

> I was able to reproduce this oops with a somewhat more reliable ksymoops (I was ready for this nasty bug this time).  Looks like the problem is in the sockets
> code.

The traces so far all match one description , this one included. Its the
'something scribbled a while ago and I just walked the list and found it'

Is your ppp module getting loaded/unloaded a lot. Im wondering if we have
a module related race. That would explain why folks running large ppp dialin
servers simply dont see any problems

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

Re: Oops in 2.2.18 with pppd dial in server

[Jeff V. Merkey]

On Fri, Dec 01, 2000 at 12:46:18PM +0000, Alan Cox wrote:
> > I was able to reproduce this oops with a somewhat more reliable ksymoops (I was ready for this nasty bug this time).  Looks like the problem is in the sockets
> > code.
> 
> The traces so far all match one description , this one included. Its the
> 'something scribbled a while ago and I just walked the list and found it'
> 
> Is your ppp module getting loaded/unloaded a lot. Im wondering if we have

Humm.  The kernel ppp module is not getting unloaded.  I am using both the
pppd 2.3.11 and pppd-mppe 2.3.10 versions (pppd-mppe has CHAP-80 and MPPE
patched into the code).  The user space module does load and unload a lot 
due to all the testing we are doing.  

I took the CWorthy ncurses interface I wrote for the NWFS tools, and am 
using it to write a pppconfig utility that handles modem setup, inittab, 
mgetty, options.ttyX, etc. so setting up a dial-in server will be as 
easy with IPX and IP as it is in NT.  

I have also noticed that pppd is not "kicking" modprobe -v the way it 
probably should, and the ppp kernel modules have to be loaded and 
configured manually.

Jeff

> a module related race. That would explain why folks running large ppp dialin
> servers simply dont see any problems
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> Please read the FAQ at http://www.tux.org/lkml/
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/