> The following reply was made to PR kernel/5777; it has been noted by GNATS.
>
> From: RD Thrush <rd@thrush.com>
> To:
kurt@intricatesoftware.com
> Cc:
gnats@cvs.openbsd.org
> Subject: Re: kernel/5777: kernel hang
> Date: Wed, 09 Apr 2008 02:40:28 -0400
>
> RD Thrush wrote:
> > Kurt Miller wrote:
> >> I noticed you are using mfs on these systems. How big are they?
> >> It would be convenient if you provided a summary of the following
> >> for the systems experiencing the hang:
> >>
> >> System #, arch, real mem size, total swap size, total mfs size
> >> --------------------------------------------------------------------------------------------
> >>
> >>
> > 1 Athlon64 X2 amd64 1023MB 3,147,752K 503,854K
> > 2 Sempron amd64 479MB 1,048,824K 201,454K
> > 3 Mobile Athlon 64 amd64 1278MB 1,572,984K 403,054K
> > Mobile Athlon 64 i386 1278MB 1,572,984K 403,054K
> > 4 Soekris 5501 i386 511MB 1,052,257K 100,654K
> >
> >
> >
> >> Also please try some tests without mfs involved.
> >>
> >
> > Ok, I'll be able to work with the Soekris later today and will remove
> > mfs from fstab. I could boot a kernel with UVMHIST (that I have with
> > debug symbols) or a snapshot. Any preference?
>
> I installed an i386 snapshot, 4.3-current (GENERIC) #792: Sun Apr 6
> 22:49:37 MDT 2008, on the Soekris without mfs over 28 hours ago. It has
> continuously been running the test without hanging. I installed the
> same snapshot on the laptop (#3 above) without mfs and ran the test for
> 12 hours hanging. I then added mfs, rebooted the same snapshot and had
> the hang within 30 minutes. So, it seems you've definitely isolated the
> conditions that cause the hang.
>
> I typically have something like "/dev/wd0b /tmp mfs
> rw,-s=819200,nosuid,noexec,nodev 0 0" in /etc/fstab.
>
> Is mfs busted? Am I abusing mfs? Where do we go from here?
