[OpenAFS] Compatibility w/ Linux 2.2.19?

Steven N. Hirsch shirsch@adelphia.net
Mon, 9 Apr 2001 18:48:33 -0400 (EDT)


On Sat, 7 Apr 2001, Steven N. Hirsch wrote:

> Previous attempts at building the kernel modules against 2.2.19preX failed
> rather miserably.  Is it worth trying again with release 2.2.19, or is
> this known to be broken?

Thanks to all the respondants, I tried this again with slightly better
results.  I'm running:

kernel 2.2.19 w/
 autofs-v4 patch
 ext3 filesystem patch
 hedrick's unified IDE driver
 Justin Gibb's "official" Adaptec SCSI driver

on a dual-PIII Xeon IBM Intellistation.

This system had been stable for months using the Transarc AFS-3.6 client
software and module (libafs-2.2.16.o, IIRC) with a 2.2.17 kernel (had only
the autofs-v4 patch in that one).  After rebuilding OpenAFS 1.0.3 against
the new kernel, I tried bringing up AFS in 2.2.19, leaving the OEM afsd
binary in place.

At first, everything seemed to work properly except for syslog complaints
about:

kernel: osi_linux_free: failed to remove chunk from hashtable

at regular intervals.  Eventually, afsd CPU usage climbed up to 99% and
things became unresponsive.

Remembering that I had not cleared out the cache files, I rebooted into
single-user mode, erased 8,000-odd V* files and brought it up
again.  At this point I was unable to access anything under /afs.  The
afsd process died (klogd was nice enough to decode it - not run through
ksymoops):

Apr  9 16:59:11 bucky kernel: rxi_AllocPacket error<1>Unable to handle kernel paging request at virtual address ffffffff 
Apr  9 16:59:11 bucky kernel: current->tss.cr3 = 37f24000, %cr3 = 37f24000 
Apr  9 16:59:11 bucky kernel: *pde = 002b8067 
Apr  9 16:59:11 bucky kernel: *pte = 00000000 
Apr  9 16:59:11 bucky kernel: Oops: 0002 
Apr  9 16:59:11 bucky kernel: CPU:    1 
Apr  9 16:59:11 bucky kernel: EIP:    0010:[<fc8b0852>] 
Apr  9 16:59:11 bucky kernel: EFLAGS: 00010286 
Apr  9 16:59:11 bucky kernel: eax: 00000015   ebx: fb45eb40   ecx: 0000004a   edx: 00000069 
Apr  9 16:59:11 bucky kernel: esi: fa2ab440   edi: fb45eb40   ebp: 00000000   esp: f7f23be0 
Apr  9 16:59:11 bucky kernel: ds: 0018   es: 0018   ss: 0018 
Apr  9 16:59:11 bucky kernel: Process kdm (pid: 997, process nr: 67, stackpage=f7f23000) 
Apr  9 16:59:11 bucky kernel: Stack: fc8b192a 00000002 fc8b18b0 fc8c7270 fb45eb40 fc8b192a 00000002 fc8abf8b  
Apr  9 16:59:11 bucky kernel:        00000002 fb45eb40 fa2ab440 00000003 fa2ab440 00000004 00000001 c0293c20  
Apr  9 16:59:11 bucky kernel:        f7f23c38 00000286 fb45eea0 fb45eeb0 f7f22000 fc8aeb02 fb45eb40 00000000  
Apr  9 16:59:11 bucky kernel: Call Trace: [<fc8b192a>] [<fc8b18b0>] [<fc8c7270>] [<fc8b192a>] [<fc8abf8b>] [<fc8aeb02>] [<fc8aed44>]  
Apr  9 16:59:11 bucky kernel:        [<fc8850f0>] [<fc8850d5>] [<fc9507f4>] [<fc8d5540>] [<fc884ff4>] [<fc950744>] [<fc8868e5>] [<fca83158>]  
Apr  9 16:59:11 bucky kernel:        [<fc950744>] [<fc892c22>] [<fc894840>] [<fc8d1fdc>] [<fc950744>] [<fc950744>] [<fc950744>] [<fc911774>]  
Apr  9 16:59:11 bucky kernel:        [<fc884ff4>] [<fc9507f0>] [<fca831ac>] [<fc9505d0>] [<fca83158>] [<fc9505d0>] [<fc8d1878>] [<fc950744>]  
Apr  9 16:59:11 bucky kernel:        [<fc89c4c0>] [<fc950744>] [<fc950744>] [<fc9507f0>] [<fc950744>] [<fc950744>] [<fc9507f0>] [<fc950744>]  
Apr  9 16:59:11 bucky kernel:        [<fc895cfe>] [<fc895dec>] [<fc8b9adc>] [<fc950744>] [<fc895901>] [<fc950744>] [<fc950744>] [<fc895b5a>]  
Apr  9 16:59:11 bucky kernel:        [<fc950744>] [<fc8ba952>] [<fc950744>] [<fc950744>] [<fc950790>] [real_lookup+80/160] [<fc950744>] [lookup_dentry+296/488]  
Apr  9 16:59:11 bucky kernel:        [open_namei+102/848] [filp_open+68/240] [sys_open+82/172] [system_call+52/56]  
Apr  9 16:59:11 bucky kernel: Code: c6 05 ff ff ff ff 2a 83 c4 10 c3 8d 76 00 83 ec 04 55 57 56  

Apr  9 16:59:11 bucky kernel: rxi_AllocPacket error<1>Unable to handle kernel paging request at virtual address ffffffff 
Apr  9 16:59:11 bucky kernel: current->tss.cr3 = 3ac5a000, %cr3 = 3ac5a000 
Apr  9 16:59:11 bucky kernel: *pde = 002b8067 
Apr  9 16:59:11 bucky kernel: *pte = 00000000 
Apr  9 16:59:11 bucky kernel: Oops: 0002 
Apr  9 16:59:11 bucky kernel: CPU:    0 
Apr  9 16:59:11 bucky kernel: EIP:    0010:[<fc8b0852>] 
Apr  9 16:59:11 bucky kernel: EFLAGS: 00010296 
Apr  9 16:59:11 bucky kernel: eax: 00000015   ebx: fa2a8da0   ecx: c026a128   edx: 00000021 
Apr  9 16:59:11 bucky kernel: esi: fb45eb40   edi: fb45eb40   ebp: 00000000   esp: fa491bac 
Apr  9 16:59:11 bucky kernel: ds: 0018   es: 0018   ss: 0018 
Apr  9 16:59:11 bucky kernel: Process afsd (pid: 656, process nr: 34, stackpage=fa491000) 
Apr  9 16:59:11 bucky kernel: Stack: fc8b192a 00000002 fc8b18b0 fc8c7270 fa2a8da0 fc8b192a 00000002 fc8abf8b  
Apr  9 16:59:11 bucky kernel:        00000002 fa2a8da0 fb45eb40 fb45eb50 00000000 0000058c fb45eb50 fb45eb50  
Apr  9 16:59:11 bucky kernel:        00000002 0400058c fa491c68 fb45eb40 fb45eb50 fc8aaa91 fb45eb40 00000000  
Apr  9 16:59:11 bucky kernel: Call Trace: [<fc8b192a>] [<fc8b18b0>] [<fc8c7270>] [<fc8b192a>] [<fc8abf8b>] [<fc8aaa91>] [<fc8b953f>]  
Apr  9 16:59:11 bucky kernel:        [<fc8aa3be>] [<fc8b10ec>] [<fc8bb915>] [<fc8c8d0c>] [<fc8c8ce8>] [<fc8d1fdd>] [<fc8c8ce0>] [<fc8d43a8>]  
Apr  9 16:59:11 bucky kernel:        [<fc8d1f1a>] [do_wp_page+432/512] [<fc8bc3a6>] [do_page_fault+290/960] [sys_setpriority+120/252] [system_call+52/56]  
Apr  9 16:59:11 bucky kernel: Code: c6 05 ff ff ff ff 2a 83 c4 10 c3 8d 76 00 83 ec 04 55 57 56  


So, before running around in circles can anyone advise me as whether the
Transarc 3.6 afsd binary should be expected to work with the OpenAFS
kernel module loaded?  Or, is this asking for trouble?

The system is being run at a site with a LARGE AFS installation and is
used for production, so I'd rather not do too much blind hack-and-reboot
if I can help it.

Thanks for any advice and input!

Steve