home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.linux
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!sdd.hp.com!cs.utexas.edu!hermes.chpc.utexas.edu!news.utdallas.edu!ramesh
- From: ramesh@utdallas.edu (R. Ramesh)
- Subject: 0.99 kernel crash with Named etc.
- Message-ID: <BzB82F.F91@utdallas.edu>
- Sender: usenet@utdallas.edu
- Nntp-Posting-Host: ramesh.utdallas.edu
- Organization: Univ. of Texas at Dallas
- Date: Tue, 15 Dec 1992 16:49:27 GMT
- Lines: 46
-
- Hi:
-
- I would like to report kernel crash and related info to help to hunt bugs,
- if any, in the kernel. First a small prolog.
-
- I got hold of the named, rshd and inetd binaries from tsx-11. I have only gcc
- 222d7 but jump4.2 as realeased by hlu (Nov 30, I think). I got the rsh source
- and compiled my self as the binary is out of date (lloks for libc.so.2.2)
- I have 0.99 kernel compiled
- by myself without SCSI, sound, special serial card support. Yes, the kernel
- is standard 0.99 with the inode.c and fs.h patches reported by Linus. I do not
- have ethernet card of any kind. This test is only for loopback stuff.
-
- I set up all the necessary files for named in /usr/etc/inet (named.boot,
- named.ca, named.hosts, named.local). config loopback to address 129.110.21.2
- and 127.0.0.1. started inetd and named.
-
- Now the bug report:
-
- I did a "rsh lata ls" (lata is my host name). I get Permission denied. But
- that is not the real problem. Now the kernel compalins "OOPS: 0000 ... etc
- at address EIP: 0008:0001EAE1 and kernel unable to handle paging request
- at C000000C". The EIP looks like it is around kfree_skb routine (System.map ->
- kfree_skb = 0x1eab..). I ignore it and do another "rsh lata ls". This time
- I bring the system down to grinding halt. This is the error report:
-
-
- Unable to handle paging request at C0000018
- OOPS: 0000
- EIP: 0008:0001F0055 (destroy_sock = 0x1efac and destroy sock calls kfree_skb)
- ....
- ....
- PID : 0 process:0
- task[0] swapper killed: unable to recover
- Kernel panic: Trying to free swapper memory space
- In swapper task not syncing
- <grinding halt>
-
- BTW, the named also complains (just once) "No root name server for class 1"
- when rsh is invoked the first time. Further, I am able to repeat this kernel
- crashing doing exactly what I said above (i.e., config loopback, start named,
- inetd, rsh first to get OOPS and second time to OOPS and crash)
-
- Hope I did not waste bandwidth reporting silly thing.
-
- Ramesh
-