Re: [SSI-devel] CVS:OPENSS-RH kernel oops CFS/NFSD interaction
Brought to you by:
brucewalker,
rogertsang
From: Roger T. <pe...@ho...> - 2004-12-17 21:16:15
|
Hi John, As you know what platform I'm running, I'm running that SSI kernel from CVS as of 2pm EST yesterday and includes your nfsd.patch Good news: your patch fixes my nfsd kernel oops. Bad news: The new changes to the SSI kernel from Laura's recent check-ins to OPENSSI-RH branch or from this nfsd.patch causes kernel crash at prune_dcache.c or the following kernel oops. This can be reproduced while running /etc/cron.daily/slocate.cron. My root fs is CFS over ext3 with dir_index on local ATA drive. fsck checks out fine on reboot and prior to inducing crash. Dec 17 04:31:23 node1 kernel: kernel BUG in header file at line 325 Dec 17 04:31:23 node1 kernel: ------------[ cut here ]------------ Dec 17 04:31:23 node1 kernel: kernel BUG at panic.c:297! Dec 17 04:31:23 node1 kernel: invalid operand: 0000 Dec 17 04:31:23 node1 kernel: nfsd ipt_REJECT ipt_multiport ipt_state ip_conntrack iptable_filter ip_tables floppy ide-cd sr_mod cdrom dm-mod keybdev mousedev hid input usb-uhci ehci-hcd u Dec 17 04:31:23 node1 kernel: CPU: 0 Dec 17 04:31:23 node1 kernel: EIP: 0060:[<c0128a39>] Not tainted Dec 17 04:31:23 node1 kernel: EFLAGS: 00010286 Dec 17 04:31:23 node1 kernel: Dec 17 04:31:23 node1 kernel: EIP is at __out_of_line_bug [kernel] 0x19 (2.4.22-1.2199.nptl_ssi_5develsmp) Dec 17 04:31:23 node1 kernel: eax: 00000026 ebx: dcf97980 ecx: 00000000 edx: c066f980 Dec 17 04:31:23 node1 kernel: esi: dcf97bbc edi: dcf979ef ebp: e4a17c90 esp: e4a17c88 Dec 17 04:31:23 node1 kernel: ds: 0068 es: 0068 ss: 0068 Dec 17 04:31:23 node1 kernel: Process updatedb (pid: 68374, stackpage=e4a17000) Dec 17 04:31:23 node1 kernel: Stack: xxx Dec 17 04:31:23 node1 kernel: Call Trace: Dec 17 04:31:23 node1 kernel: [<c0173b39>] d_alloc [kernel] 0x199 (0xe4a17c94) Dec 17 04:31:23 node1 kernel: [<c01684b8>] lookup_hash_it [kernel] 0x88 (0xe4a17cb4) Dec 17 04:31:23 node1 kernel: [<c016858e>] lookup_hash [kernel] 0x1e (0xe4a17cd8) Dec 17 04:31:23 node1 kernel: [<c02537f4>] cfsd_lookup [kernel] 0x54 (0xe4a17cec) Dec 17 04:31:23 node1 kernel: [<c024f023>] cfs_proc_lookup [kernel] 0x1b3 (0xe4a17d24) Dec 17 04:31:23 node1 kernel: [<c0174dbb>] __mark_inode_dirty [kernel] 0xbb (0xe4a17d34) Dec 17 04:31:23 node1 kernel: [<c01771ab>] update_atime [kernel] 0x6b (0xe4a17d4c) Dec 17 04:31:23 node1 kernel: [<c0192ba4>] ext3_readdir [kernel] 0x304 (0xe4a17d60) Dec 17 04:31:23 node1 kernel: [<c023fee7>] tok_hold [kernel] 0x87 (0xe4a17dd0) Dec 17 04:31:23 node1 kernel: [<c025a917>] cfstok_req [kernel] 0x127 (0xe4a17df0) Dec 17 04:31:23 node1 kernel: [<c024d701>] cfs_lookup [kernel] 0xe1 (0xe4a17e68) Dec 17 04:31:23 node1 kernel: [<c016717a>] real_lookup [kernel] 0x19a (0xe4a17ea4) Dec 17 04:31:23 node1 kernel: [<c0166f61>] cached_lookup [kernel] 0x21 (0xe4a17eb0) Dec 17 04:31:23 node1 kernel: [<c0167a8c>] link_path_walk_it [kernel] 0x5bc (0xe4a17ecc) Dec 17 04:31:23 node1 kernel: [<c016803e>] path_walk_it [kernel] 0x2e (0xe4a17f0c) Dec 17 04:31:23 node1 kernel: [<c016871f>] __user_walk_it [kernel] 0x6f (0xe4a17f20) Dec 17 04:31:23 node1 kernel: [<c01632b7>] sys_lstat64 [kernel] 0x37 (0xe4a17f44) Dec 17 04:31:23 node1 kernel: [<c016d6bd>] vfs_readdir [kernel] 0xad (0xe4a17f4c) Dec 17 04:31:23 node1 kernel: [<c0157e5e>] sys_fchdir [kernel] 0x4e (0xe4a17f90) Dec 17 04:31:23 node1 kernel: [<c010be37>] system_call [kernel] 0x33 (0xe4a17fc0) Dec 17 04:31:23 node1 kernel: Dec 17 04:31:23 node1 kernel: Code: 0f 0b 29 01 a1 2b 3e c0 eb 0d 90 90 90 90 90 90 90 90 90 90 > >Please apply the attached patch to you kernel and see if it fixes your >NFS problems. The patch fixes the problem Dave reported, but unless your >panics all involved umounting the filesystem, I have no reason to >believe it will help you. (But I can hope.) > >John > |