Re: mpii zfs xen unstable on -current

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

On Fri, Jul 01, 2022 at 09:05:17PM +0200, oskar%fessel.org@localhost wrote:
> 
> 
> > Am 25.06.2022 um 01:26 schrieb Robert Elz <kre%munnari.OZ.AU@localhost>:
> > 
> >    Date:        Sat, 25 Jun 2022 01:04:04 +0200
> >    From:        "oskar%fessel.org@localhost" <oskar%fessel.org@localhost>
> >    Message-ID:  <056EBA16-27DC-42B3-80A4-7CEF3F10162D%fessel.org@localhost>
> > 
> > 
> >  | [ 236025.4127414] panic: kernel diagnostic assertion "xs->resid =3D=3D =
> >  | xs->datalen" failed: file "/hurz/src/sys/dev/pci/mpii.c", line 3207=20
> >  | [ 236025.4127414] cpu0: Begin traceback...
> >  | [ 236025.4227374] vpanic() at netbsd:vpanic+0x177
> >  | [ 236025.4227374] kern_assert() at netbsd:kern_assert+0x4b
> >  | [ 236025.4227374] mpii_scsi_cmd_done() at =
> >  | netbsd:mpii_scsi_cmd_done+0x30b
> >  | [ 236025.4227374] mpii_intr() at netbsd:mpii_intr+0x21e
> > 
> >  | The machine i want to fix is running NetBSD 9.99.97 (XEN3_DOM0) #4: Thu =
> >  | Jun 16 13:02:43 CEST 2022
> > 
> > Try updating to a very new -current, there have been fixes to the mpii
> > driver (I have observed happening, I don't have that hardware) in the
> > very recent past (June 16 is ages ago!).   There might be other issues
> > (there always is that possibility with very new code) but if so, they
> > might not affect you as much as this one does.
> 
> Have done that, even though the hope was more on Xen related changes to solve problems.
> > 
> > That you don't see issues with a non-Xen system and the same hardware
> > might be because of the way interrupts are delivered (a bit different
> > for Xen clients than it is for bare metal systems).
> > 
> > If this is the problem, then that might be an easy fix, and all of the
> > rest of your questions (if still material) can be considered without
> > imagining that they may alter this issue.
> > 
> 
> So, after a week away from home with the system running for 4 hours until every filesystem access hung infinitely?
> After that i just restarted everything and booted from new kernels with a coherent userland just to  achieve the following the day after:
> 
> [ 77118.8430199] panic: kernel diagnostic assertion "xs->resid == xs->datalen" failed: file "/hurz/src/sys/dev/pci/mpii.c", line 3207 
> [ 77118.8430199] cpu0: Begin traceback...
> [ 77118.8430199] vpanic() at netbsd:vpanic+0x177
> [ 77118.8430199] kern_assert() at netbsd:kern_assert+0x4b
> [ 77118.8530960] mpii_scsi_cmd_done() at netbsd:mpii_scsi_cmd_done+0x30b
> [ 77118.8530960] mpii_intr() at netbsd:mpii_intr+0x21e
> [ 77118.8530960] evtchn_do_event() at netbsd:evtchn_do_event+0x114
> [ 77118.8530960] do_hypervisor_callback() at netbsd:do_hypervisor_callback+0x167
> [ 77118.8530960] Xhandle_hypervisor_callback() at netbsd:Xhandle_hypervisor_callback+0x19
> [ 77118.8530960] --- interrupt ---
> [ 77118.8630225] hypercall_page() at netbsd:hypercall_page+0x3aa
> [ 77118.8630225] idle_loop() at netbsd:idle_loop+0x146
> [ 77118.8630225] cpu0: End traceback...
> 
> [ 77118.8630225] dumping to dev 168,9 (offset=33482590, size=0): not possible
> [ 77118.8630225] rebooting...
> (XEN) Hardware Dom0 shutdown: rebooting machine
> 
> So, this is somewhere different context but nonethelss not so good. Any pointers on how to fix this greatly appreciated.

I see no reason to think this is Xen-specific at this point

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--