Re: All processes go tstile

["J. Hannken-Illjes" <hannken%eis.cs.tu-bs.de@localhost>]

> On 16. Aug 2018, at 03:16, Emmanuel Dreyfus <manu%netbsd.org@localhost> wrote:
> 
> Hello
> 
> Another dump/snapshot problem that seems to have surfaced with
> NetBSD 8.0: I run a dump with snapshot, and after a while, 
> all processes get stuck in tstile.
> 
> This happens with GENERIC kernel, though I added LOCKDEBUG to 
> investigate.
> 
> Processes backtrace show namei/VOP_LOCK/rw_enter and  following
> the locks owner, it seems everyone is  ultimately waiting for ioflush, i
> which itself is stuck with this backtrace:
> 
> sleepq_block
> cv_wait
> fstrans_start
> VOP_BWRITE
> getnewbuf
> getblk
> bio_doread
> bread
> ffs_update.part.3
> ffs_sync
> VFS_SYNC
> sched_sync
> 
> If I understand correctly, that means a n I/O that never completed, right?

Looks like a deadlock where we sync a file system, need a new buffer
and try to free a buffer on a currently suspending file system.

VFS_SYNC and VOP_BWRITE should be on different mounts here.

Do you have a crash dump?

--
J. Hannken-Illjes - hannken%eis.cs.tu-bs.de@localhost - TU Braunschweig (Germany)