Bug 206630 - [Hyper-V]FreeBSD 10.2 on Windows 10 & 2016 server may not boot due to multiple invalid disks issue
Summary: [Hyper-V]FreeBSD 10.2 on Windows 10 & 2016 server may not boot due to multipl...
Status: Closed FIXED
Alias: None
Product: Base System
Classification: Unclassified
Component: kern (show other bugs)
Version: 10.2-STABLE
Hardware: amd64 Any
: --- Affects Many People
Assignee: freebsd-emulation (Nobody)
URL:
Keywords: patch
Depends on:
Blocks:
 
Reported: 2016-01-26 02:10 UTC by Hongjiang
Modified: 2018-10-25 04:46 UTC (History)
8 users (show)

See Also:

Attachments
The patch to fix multiple disk issues on windows 10 & 2016 server (57.96 KB, patch)
2016-01-26 02:15 UTC, Hongjiang 		no flags Details | Diff
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Hongjiang 2016-01-26 02:10:10 UTC 
The problem can be reproduced for FreeBSD 10.2 installed on windows 10 and 2016 server.

The valid disk was randomly switched with invalid disk. For example, from da0 to da1 when restarting the VM. As a result, the boot process will hang because system wants to load ufs on /dev/da0p2, but in fact it locates in an invalid disk.

Both "camcontrol" and "dmesg" can be used to check whether this issue exist.

camcontrol command lists 16 disk, but only one is valid. See the output of 'camcontrol devlist' on windows 10.

root@honzhan-dev2:/usr/src # camcontrol devlist
 <Msft Virtual CD/ROM 1.0> at scbus0 target 0 lun 0 (cd0,pass0)
 <Msft Virtual Disk 1.0> at scbus1 target 0 lun 0 (da1,pass2)
 < > at scbus1 target 1 lun 1 (da0,pass1)
 < > at scbus2 target 0 lun 1 (da2,pass3)
 < > at scbus2 target 0 lun 2 (da4,pass5)
 < > at scbus2 target 0 lun 3 (da6,pass7)
 < > at scbus2 target 0 lun 4 (da8,pass9)
 < > at scbus2 target 0 lun 5 (da10,pass11)
 < > at scbus2 target 0 lun 6 (da12,pass13)
 < > at scbus2 target 0 lun 7 (da14,pass15)
 < > at scbus2 target 1 lun 1 (da3,pass4)
 < > at scbus2 target 1 lun 2 (da5,pass6)
 < > at scbus2 target 1 lun 3 (da7,pass8)
 < > at scbus2 target 1 lun 4 (da9,pass10)
 < > at scbus2 target 1 lun 5 (da11,pass12)
 < > at scbus2 target 1 lun 6 (da13,pass14)
 < > at scbus2 target 1 lun 7 (da15,pass16)

dmesg also shows us the similar information:

dmesg

da0 at blkvsc0 bus 0 scbus1 target 1 lun 1
 da0: < > Fixed Direct Access SCSI device
 da0: 300.000MB/s transfers
 da0: 0MB (0 512 byte sectors: 0H 0S/T 0C)
 da0: Delete methods: <NONE(*)>
 da1 at blkvsc0 bus 0 scbus1 target 0 lun 0
 da1: <Msft Virtual Disk 1.0> Fixed Direct Access SPC-3 SCSI device
 da1: 300.000MB/s transfers
 da1: 20480MB (41943040 512 byte sectors: 255H 63S/T 2610C)
 da2: < > Fixed Direct Access SCSI device
 da2: 300.000MB/s transfers
 da2: 0MB (0 512 byte sectors: 0H 0S/T 0C)
 da2: Delete methods: <NONE(*)>
 da3 at storvsc1 bus 0 scbus2 target 1 lun 1
 da3: < > Fixed Direct Access SCSI device
 da3: 300.000MB/s transfers
 da3: 0MB (0 512 byte sectors: 0H 0S/T 0C)
 da3: Delete methods: <NONE(*)>
 da5: < > Fixed Direct Access SCSI device
 da5: 300.000MB/s transfers
 da5: 0MB (0 512 byte sectors: 0H 0S/T 0C)
 da5: Delete methods: <NONE(*)>
 da4: < > Fixed Direct Access SCSI device
 da4: 300.000MB/s transfers
 da4: 0MB (0 512 byte sectors: 0H 0S/T 0C)
 da4: Delete methods: <NONE(*)>
Comment 1 Hongjiang 2016-01-26 02:13:58 UTC 
The fix of this issue has already been committed to head. The code review for it is https://reviews.freebsd.org/D4928.
Comment 2 Hongjiang 2016-01-26 02:15:13 UTC 
Created attachment 166126 [details]
The patch to fix multiple disk issues on windows 10 & 2016 server
Comment 3 Hongjiang 2016-02-18 05:50:53 UTC 
This issue was fixed on both 10 stable branch (r294890) and head (r294557).
Comment 4 khushi 2018-10-25 04:46:59 UTC 
MARKED AS SPAM