From: David Teigland <teigland@redhat.com>
To: target-devel@vger.kernel.org
Subject: [BUG] LIO iSCSI target deadlock on PR PREEMPT AND ABORT with concurrent I/O
Date: Tue, 23 Jun 2026 13:14:42 -0500 [thread overview]
Message-ID: <ajrNEj-B63SDtZ9W@redhat.com> (raw)
Hi, I've been hitting the deadlocked threads copied below while running
some new lvm tests that are doing preempt-abort between VMs sharing an
iscsi target exported by a third VM. The problem occurs when sg_persist
register triggers udev probing/reads, followed quickly by sg_persist
preempt-abort (preempt without abort doesn't deadlock.) The combination
of probing reads, and preempt-abort often hits the deadlock. (I'm leaving
out the more extensive AI analysis and patch which I can't personally
vouch for, but can provide if it's useful.)
An effective work-around for now is calling udevadm settle after register.
Seen on Fedora kernel 6.19.10-300.fc44.x86_64 and a recent 7.1.0-rc.
Thanks,
Dave
task:iscsi_trx state:D stack:0 pid:1899 tgid:1899 ppid:2 task_flags:0x208140 flags:0x00080006
Call Trace:
<TASK>
__schedule+0x2b2/0x5d0
schedule+0x27/0x80
schedule_timeout+0xdc/0x120
wait_for_completion+0x9e/0x170
target_put_cmd_and_wait+0x6e/0xc0 [target_core_mod]
core_tmr_drain_state_list+0x2eb/0x3f0 [target_core_mod]
core_tmr_lun_reset+0xa0/0x280 [target_core_mod]
core_scsi3_pro_preempt+0x7a8/0x8e0 [target_core_mod]
target_scsi3_emulate_pr_out+0x298/0x3c0 [target_core_mod]
__target_execute_cmd+0x36/0xd0 [target_core_mod]
iscsit_execute_cmd+0x279/0x2b0 [iscsi_target_mod]
iscsit_sequence_cmd+0xf3/0x170 [iscsi_target_mod]
iscsit_get_immediate_data+0x71/0x190 [iscsi_target_mod]
iscsit_get_rx_pdu+0xe1/0x300 [iscsi_target_mod]
iscsi_target_rx_thread+0x8e/0x120 [iscsi_target_mod]
kthread+0xfa/0x240
ret_from_fork+0x130/0x1a0
ret_from_fork_asm+0x1a/0x30
</TASK>
task:iscsi_ttx state:D stack:0 pid:1898 tgid:1898 ppid:2 task_flags:0x208040 flags:0x00080002
Call Trace:
<TASK>
__schedule+0x2b2/0x5d0
schedule+0x27/0x80
schedule_timeout+0xdc/0x120
wait_for_completion+0x9e/0x170
kthread_stop+0x6e/0x190
iscsit_close_connection+0x49c/0x7a0 [iscsi_target_mod]
iscsit_take_action_for_connection_exit+0x80/0x140 [iscsi_target_mod]
iscsi_target_tx_thread+0x162/0x240 [iscsi_target_mod]
kthread+0xfa/0x240
ret_from_fork+0x130/0x1a0
ret_from_fork_asm+0x1a/0x30
</TASK>
task:iscsi_np state:D stack:0 pid:1885 tgid:1885 ppid:2 task_flags:0x208040 flags:0x00080002
Call Trace:
<TASK>
__schedule+0x2b2/0x5d0
schedule+0x27/0x80
schedule_timeout+0xdc/0x120
wait_for_completion+0x9e/0x170
iscsi_check_for_session_reinstatement+0x1f5/0x2d0 [iscsi_target_mod]
iscsi_target_handle_csg_one+0x114/0x130 [iscsi_target_mod]
iscsi_target_do_login+0x2d/0xd0 [iscsi_target_mod]
iscsi_target_start_negotiation+0x67/0x170 [iscsi_target_mod]
__iscsi_target_login_thread+0x37f/0x7d0 [iscsi_target_mod]
iscsi_target_login_thread+0x24/0x60 [iscsi_target_mod]
kthread+0xfa/0x240
ret_from_fork+0x130/0x1a0
ret_from_fork_asm+0x1a/0x30
</TASK>
reply other threads:[~2026-06-23 18:14 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=ajrNEj-B63SDtZ9W@redhat.com \
--to=teigland@redhat.com \
--cc=target-devel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.