wapbl_flush: current transaction too big to flush (was: WAPBL panic)

[Emmanuel Dreyfus <manu%netbsd.org@localhost>]

On Thu, Nov 19, 2015 at 02:21:35PM +0000, Emmanuel Dreyfus wrote:
> On NetBSD 6.1.5/amd64, I get 
> 
> panic: wapbl_flush: current transaction too big to flush

I got it again with VOP_WRITE this time, and with a few odd
messages on the console before the panic:
free inode /mail/122024449 had 4294967264 blocks
free inode /mail/122024452 had 4294967264 blocks
panic: wapbl_flush: current transaction too big to flush
 
cpu0: Begin traceback...
printf_nolog() at netbsd:printf_nolog
wapbl_begin() at netbsd:wapbl_begin
wapbl_begin() at netbsd:wapbl_begin+0x45
ffs_write() at netbsd:ffs_write+0x8b9
VOP_WRITE() at netbsd:VOP_WRITE+0x37
vn_write() at netbsd:vn_write+0xf9
dofilewrite() at netbsd:dofilewrite+0x7d
sys_pwrite() at netbsd:sys_pwrite+0x80
syscall() at netbsd:syscall+0xc4
cpu0: End traceback...

4294967264 in 0xffffffe0 with suggests some int32 underrun. 
Is it possible to crash is not caused by a big transaction,
but by a transaction whose block count went wrong?
  


-- 
Emmanuel Dreyfus
manu%netbsd.org@localhost