Re: panic: vm_domainset_iter_first: Unknown policy 15168

From: Mark Johnston <markj_at_freebsd.org>
Date: Tue, 20 Jul 2021 18:36:39 UTC
On Tue, Jul 20, 2021 at 10:13:20AM -0700, Steve Kargl wrote:
> On Tue, Jul 20, 2021 at 12:15:58PM -0400, Mark Johnston wrote:
> > On Tue, Jul 20, 2021 at 09:07:04AM -0700, Steve Kargl wrote:
> > > On Mon, Jul 19, 2021 at 07:05:03PM -0700, Steve Kargl wrote:
> > > > On Mon, Jul 19, 2021 at 07:55:07PM -0400, Mark Johnston wrote:
> > > > > On Mon, Jul 19, 2021 at 03:02:19PM -0700, Steve Kargl wrote:
> > > > > > 
> > > > > > (kgdb) #0  __curthread () at /usr/src/sys/amd64/include/pcpu_aux.h:55
> > > > > > #1  doadump (textdump=textdump@entry=1)
> > > > > >     at /usr/src/sys/kern/kern_shutdown.c:399
> > > > > > #2  0xffffffff805fe263 in kern_reboot (howto=260)
> > > > > >     at /usr/src/sys/kern/kern_shutdown.c:486
> > > > > > #3  0xffffffff805fe6b0 in vpanic (fmt=<optimized out>, ap=<optimized out>)
> > > > > >     at /usr/src/sys/kern/kern_shutdown.c:919
> > > > > > #4  0xffffffff805fe4b3 in panic (fmt=<unavailable>)
> > > > > >     at /usr/src/sys/kern/kern_shutdown.c:843
> > > > > > #5  0xffffffff8085dcbb in vm_domainset_iter_first (di=<optimized out>, 
> > > > > >     domain=<optimized out>) at /usr/src/sys/vm/vm_domainset.c:189
> > > > > > #6  0xffffffff8085dbd2 in vm_domainset_iter_page_init (
> > > > > >     di=di@entry=0xfffffe012ae5e2a0, obj=obj@entry=0xfffff8003c21f420, 
> > > > > >     pindex=<optimized out>, pindex@entry=16931, 
> > > > > >     domain=domain@entry=0xfffffe012ae5e2f4, req=<unavailable>, 
> > > > > >     req@entry=0xfffffe012ae5e2f0) at /usr/src/sys/vm/vm_domainset.c:217
> > > > > 
> > > > > Could you please show output from:
> > > > > 
> > > > > (kgdb) frame 6
> > > > > (kgdb) p *dr
> > > > > (kgdb) p obj->domain
> > > > > 
> > > > 
> > > > The system is at work.  I'll do this tomorrow morning.
> > > > Thanks for asking for additional info.
> > > >
> > > 
> > > Hi Mark, I poked around and tried to supply the request info
> > > along with content of other structs. 
> > 
> > Is the problem reproducible?
> 
> Don't know if it matters, but the DVD was prepared on a MS
> Windows 10 system.  I don't know what software was used.
> 
> I got 4 different panics while trying to copy data from
> the UDF disk.
> 
> % mount_udf /dev/cd0 /cdrom
> 
> This mounts the drive and I can ls and cd into it.  If I do 
> 
> % cd ${HOME}/data      <-- UFS2, softupdate
> % cp -R /cdrom/data .
> 
> where /cdrom/data is a directory with 4000-ish files, I
> eventually get a panic.  Two of the other panics are of
> the form (which may be a result of UDF handing FFS bad
> data).

I see, it looks like there is some generic memory corruption triggered
by the UDF implementation.  Our implementation is fairly stale,
unfortunately, but it shouldn't trigger a panic.

If you don't already have INVARIANTS configured, that would probably
help narrow this down.  If you're able to share the image that can
trigger this, I could try to debug from that.