illumos gate

To test this, running a BE with the change, I created zpools with the following configurations:

single disk (used the existing rpool in the VM)
2 unmirrored disks (zpool create yolopool disk1 disk2)
2 mirrored disks (zpool create testpool disk1 disk2)
3 disk raidz (zpool create testpool raidz c3t0d0 c4t0d0 c5t0d0)
3 disk raidz3 (zpool create testpool raidz2 c3t0d0 c4t0d0 c5t0d0)

For each pool, I created a dump zvol (zfs create -V 8G .../dump), then set the dump device to that zvol (dumpadm -d /dev/zvol/dsk/..../dump). Then I used the age-old method of echo rootdir/Z 0 | mdb -kw to force a panic and dump.

During each panic, I made sure the dump was successful (as reported on the console), then after the system was running (post-panic), I uncompressed the resulting core file, loaded it in mdb, and ran '::status' to ensure mdb no errors were reported while processing/reading the resulting dump.

git commit ac04831df9010f533275633d3bb9829b86c171d3

commit  ac04831df9010f533275633d3bb9829b86c171d3
Author: Mike Gerdts <mike.gerdts@joyent.com>
Date:   2020-01-11T00:01:41.000Z

    12085 zvol_dumpio would be cleaner with vdev_op_dumpio
    Reviewed by: Dan McDonald <danmcd@joyent.com>
    Reviewed by: Toomas Soome <tsoome@me.com>
    Approved by: Robert Mustacchi <rm@fingolfin.org>