[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: e2fsprogs & endian questions



On Mon, Jul 30, 2001 at 01:43:47PM -0700, Tom Rini wrote:
> On Mon, Jul 30, 2001 at 12:56:46PM -0700, Tom Rini wrote:
> > On Tue, Jul 31, 2001 at 02:00:20AM +1000, Andrew Morton wrote:
> > 
> > > > I'm assuming it's just a generally good idea.  Second, has anyone tested
> > > > this on a BE machine?  I'd like to use this on my PPC.  I don't mind testing
> > > > things out, but it's always nice to know what's coming up.  Thanks.
> > > 
> > > I've seen positive reports for ia64 and Alpha (big-endian). It
> > > should be OK.  You don't hear from people when things work correctly :)
> > 
> > Okay, on my G4/466 running Linux-2.4.8-pre2 and ext3-0.9.5 everything seems
> > to be going great.  Bonnie++ ran fine in all 3 with each of the data= options,
> > and I'm going to pound on it a bit with dbench just to check.
> 
> Okay, I spoke too soon.  Running dbench 8 seems to have oopsed with:
> kernel BUG at buffer.c:1530!
> Oops: Exception in kernel mode, sig: 4
> NIP: C003C990 XER: 00000000 LR: C003C990 SP: C231DDF0 REGS: c231dd40 TRAP: 0700
> Using defaults from ksymoops -t elf32-powerpc -a powerpc:common
> MSR: 00089032 EE: 1 PR: 0 FP: 0 ME: 1 IR/DR: 11
> TASK = c231c000[9724] 'dbench' Last syscall: 4 
> last math cdfc2000 last altivec 00000000
> GPR00: C003C990 C231DDF0 C231C000 0000001D 00001032 00000001 FFFFFFFF C01BC17C 
> GPR08: 00000000 00000000 00000000 C231DD30 44000828 1001A7F0 00000000 CD54E460 
> GPR16: 0000B001 CE4654A0 D21CEE80 C231DE18 00001000 C906F000 C4B10500 CE465440 
> GPR24: 0000001B 00000001 00000000 00001000 C0591ECC C0110000 00001000 CBC156C0 
> Call backtrace: 
> C003C990 C003CD14 C003D5A0 D21CF3DC C002B2DC D21CCE00 C00396B0 
> C0003F9C 00000006 10002018 100019A8 10001B54 6FED6478 00000000 
> Warning (Oops_read): Code line not seen, dumping what data is available
> 
> >>NIP; c003c990 <unmap_underlying_metadata+e4/15c>   <=====
> Trace; c003c990 <unmap_underlying_metadata+e4/15c>
> Trace; c003cd14 <__block_prepare_write+fc/254>
> Trace; c003d5a0 <block_prepare_write+34/68>
> Trace; d21cf3dc <END_OF_CODE+11ffcd8c/????>
> Trace; c002b2dc <generic_file_write+418/624>
> Trace; d21cce00 <END_OF_CODE+11ffa7b0/????>
> Trace; c00396b0 <sys_write+c8/114>
> Trace; c0003f9c <ret_from_syscall_1+0/b4>
> Trace; 00000006 Before first symbol
> Trace; 10002018 Before first symbol
> Trace; 100019a8 Before first symbol
> Trace; 10001b54 Before first symbol
> Trace; 6fed6478 Before first symbol
> Trace; 00000000 Before first symbol

This is reproducable, but only w/ jbd-debug = 0, anything else the load
on the machine becomes very high and the tests go slowly and all seems
well.  Here's one last thing I got from the logs:
Bill-The-Cat kernel: Assertion failure in jbd_preclean_buffer_check() at jbd-kernel.c:74: "(jh->b_jlist == 0 || jh->b_jlist == 7 || jh->b_jlist == 5 || jh->b_jlist == 4 || buffer_jbd_data(bh))"

-- 
Tom Rini (TR1265)
http://gate.crashing.org/~trini/





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]