[next-20210827][ppc][multipathd] INFO: task hung in dm_table_add_target

Started by Abdul Haleem on Sep 1, 2021 · 4 messages from 2 authors · Last reply Sep 29, 2021

open the first message on its own page

Message details and related content
Greeting's

multiple task hung while adding the vfc disk back to the multipath on my powerpc box running linux-next kernel

Test:
$ multipathd -k"remove path sde"
$ multipathd -k"add path sde"

After above command multipathd task hung with continuous call traces on console, it requires reboot to stop call traces

systemd-udevd[180359]: Process '/sbin/mdadm -I /dev/dm-8' failed with exit code 1.
multipathd[180272]: mpathf: load table [0 62914560 multipath 1 queue_if_no_path 1 alua 2 1 round-robin 0 2 1 8:64 1 65:0 1 round-robin 0 2 1 8:144 1 65:80 1]
systemd[1]: Started Device-Mapper Multipath Device Controller.
multipathd[180272]: mpatha: sdl - tur checker reports path is up
multipathd[180272]: 8:176: reinstated
multipathd[180272]: mpatha: remaining active paths: 2
multipathd[180272]: sde: remove path (operator)
multipathd[180272]: mpathf: load table [0 62914560 multipath 1 queue_if_no_path 1 alua 2 1 round-robin 0 1 1 65:0 1 round-robin 0 2 1 8:144 1 65:80 1]
multipathd[180272]: sde [8:64]: path removed from map mpathf
multipathd[180272]: sdq: remove path (operator)
multipathd[180272]: mpathf: load table [0 62914560 multipath 1 queue_if_no_path 1 alua 1 1 round-robin 0 2 1 8:144 1 65:80 1]
multipathd[180272]: sdq [65:0]: path removed from map mpathf
multipathd[180272]: sdj: remove path (operator)
multipathd[180272]: mpathf: load table [0 62914560 multipath 1 queue_if_no_path 1 alua 1 1 round-robin 0 1 1 65:80 1]
multipathd[180272]: sdj [8:144]: path removed from map mpathf
multipathd[180272]: sdv: remove path (operator)
multipathd[180272]: mpathf: map in use
multipathd[180272]: mpathf: can't flush
multipathd[180272]: mpathf: load table [0 62914560 multipath 1 queue_if_no_path 0 0 0]
multipathd[180272]: sdv [65:80]: path removed from map mpathf
systemd[1]: Starting system activity accounting tool...
systemd[1]: sysstat-collect.service: Succeeded.
systemd[1]: Started system activity accounting tool.
systemd-udevd[1156]: seq 5678 '/devices/virtual/block/dm-10' is taking a long time
systemd-udevd[1156]: seq 5678 '/devices/virtual/block/dm-10' killed
multipathd[180272]: sde: add path (operator)
systemd[1]: Stopping Device-Mapper Multipath Device Controller...
systemd[1]: multipathd.service: State 'stop-sigterm' timed out. Killing.
systemd[1]: multipathd.service: Killing process 180272 (multipathd) with signal SIGKILL.
systemd[1]: multipathd.service: Processes still around after SIGKILL. Ignoring.

68 more lines

Message details and related content
On Wed, Sep 01, 2021 at 04:47:26PM +0530, Abdul Haleem wrote:
Greeting's

multiple task hung while adding the vfc disk back to the multipath on my 
powerpc box running linux-next kernel
Can you retry to reproduce this with lockdep enabled to see if there
is anything interesting holding this lock?
Message details and related content
On 9/1/21 7:06 PM, Christoph Hellwig wrote:
On Wed, Sep 01, 2021 at 04:47:26PM +0530, Abdul Haleem wrote:
quoted
Greeting's

multiple task hung while adding the vfc disk back to the multipath on my
powerpc box running linux-next kernel
Can you retry to reproduce this with lockdep enabled to see if there
is anything interesting holding this lock?
LOCKDEP was earlier enabled by default

# cat .config | grep LOCKDEP
CONFIG_LOCKDEP_SUPPORT=y

BTW, Recreated again on 5.15.0-rc2 mainline kernel and attaching the logs

-- 
Regard's

Abdul Haleem
IBM Linux Technology Center
Message details and related content
On Tue, Sep 28, 2021 at 03:53:47PM +0530, Abdul Haleem wrote:
On 9/1/21 7:06 PM, Christoph Hellwig wrote:
quoted
On Wed, Sep 01, 2021 at 04:47:26PM +0530, Abdul Haleem wrote:
quoted
Greeting's

multiple task hung while adding the vfc disk back to the multipath on my
powerpc box running linux-next kernel
Can you retry to reproduce this with lockdep enabled to see if there
is anything interesting holding this lock?
LOCKDEP was earlier enabled by default

# cat .config | grep LOCKDEP
CONFIG_LOCKDEP_SUPPORT=y

BTW, Recreated again on 5.15.0-rc2 mainline kernel and attaching the logs
It seems the reinstate is blocking on the close which is blocking on
flushing dirty data.  In other words it looks like the link blocking
looks like the symptom and not the cause.
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help