Re: WAPBL panic at mount time

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Tue, May 12, 2009 at 03:57:54PM +0200, Manuel Bouyer wrote:
> Hi,
> trying to mount a 3.5TB filesystem with -o log on 5.0_STABLE, I get:
> raid2: WARNING: raid2: total sector size in disklabel (3050553088) != the 
> size of raid (7345520384)                                                     
>         
> panic: kernel diagnostic assertion "bp->b_freelistindex == -1" failed: file 
> "/dsk/l1/misc/bouyer/netbsd-5/src/sys/kern/vfs_bio.c", line 327               
>       

As this is a large partition, I'm using the raw device (raid2d) and
there's no partition table. I wonder what ffs_statvfs() will get in such a
case. 
> fatal breakpoint trap in supervisor mode
> trap type 1 code 0 rip ffffffff803d2935 cs 8 rflags 246 cr2  7f7ffd96a740 cpl 
> 0 rsp ffff80005f89f5e0                                                        
>     
> Stopped in pid 183.1 (mount_ffs) at     netbsd:breakpoint+0x5:  leave
> db{1}> tr
> breakpoint() at netbsd:breakpoint+0x5
> panic() at netbsd:panic+0x24d
> __kernassert() at netbsd:__kernassert+0x2d
> brelsel() at netbsd:brelsel+0x524
> wapbl_discard() at netbsd:wapbl_discard+0xfe
> wapbl_stop() at netbsd:wapbl_stop+0x2d
> ffs_wapbl_stop() at netbsd:ffs_wapbl_stop+0x135
> ffs_wapbl_start() at netbsd:ffs_wapbl_start+0x3ca

If I trust gdb, ffs_wapbl_start+0x3ca is at ffs_wapbl.c:376
373                                     UFS_WAPBL_END(mp);
374                                     error = wapbl_flush(mp->mnt_wapbl, 1);
375                                     if (error) {
376                                             ffs_wapbl_stop(mp, MNT_FORCE);
377                                             return error;
378                                     }

so wapbl_flush() returned an error.
ffs_wapbl_stop+0x135 is the wapbl_stop() just after forceout:
wapbl_stop+0x2d is vfs_wapbl.c:621


-- 
Manuel Bouyer, LIP6, Universite Paris VI.           
Manuel.Bouyer%lip6.fr@localhost
     NetBSD: 26 ans d'experience feront toujours la difference
--