Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <baef87c3-5dad-3b47-44c1-6914bfc90108@cybernetics.com>
Date:   Wed, 25 May 2022 18:03:34 -0400
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.9.1
Content-Language: en-US
To:     Saurav Kashyap <skashyap@marvell.com>,
        Nilesh Javali <njavali@marvell.com>,
        GR-QLogic-Storage-Upstream@marvell.com
Cc:     linux-scsi@vger.kernel.org, linux-kernel@vger.kernel.org,
        regressions@lists.linux.dev
From:   Tony Battersby <tonyb@cybernetics.com>
Subject: [REGRESSION] qla2xxx: tape drive not removed after unplug FC cable
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Precedence: bulk

#regzbot introduced: 44c57f205876

I have several different QLogic FC HBAs (8, 16, 32 Gbps) and several
different FC LTO tape drives (IBM Ultrium 8 & 9).  When I plug in the FC
cable, the tape drive shows up as a SCSI device as expected.  With older
kernels, when I unplug the FC cable, the tape drive SCSI device would
disappear after about 30 seconds.  But with newer kernels (including
5.18), when I unplug the FC cable, the tape drive SCSI device never
disappears.  I have bisected the change in behavior to the following
commit in kernel 5.15:

44c57f205876 ("scsi: qla2xxx: Changes to support FCP2 Target")

This commit has been backported to various -stable kernels, so they are
also affected.

When testing with two different tape drives:
1) Plug FC cable into tape drive A.  Tape drive A shows up as a SCSI device.
2) Unplug FC cable; wait 60 seconds.  Tape drive A does not disappear.
3) Plug FC cable into tape drive B.  Tape drive A disappears 30 seconds
later, but tape drive B does not show up.
4) Unplug FC cable and plug it back into tape drive B.  Tape drive B
shows up as a SCSI device.

So I can actually make a tape drive disappear by plugging the cable into
a different tape drive, but then I have to reseat the cable again to
make the new tape drive show up.

lspci -n
83:00.0 0c04: 1077:2031 (rev 02)
83:00.1 0c04: 1077:2031 (rev 02)

When plugging in cable:
qla2xxx [0000:83:00.1]-500a:7: LOOP UP detected (8 Gbps).

When unplugging cable with old kernel:
qla2xxx [0000:83:00.1]-500b:7: LOOP DOWN detected (2 7 0 0).
rport-7:0-2: blocked FC remote port time out: removing target and saving binding

When unplugging cable with new kernel:
qla2xxx [0000:83:00.1]-500b:7: LOOP DOWN detected (2 7 0 0).

/sys/class/fc_remote_ports/rport-*/
dev_loss_tmo: 30
supported_classes: Class 3
port_state: Online
(port_state remains Online even when FC cable unplugged)

/proc/scsi/scsi
Host: scsi7 Channel: 00 Id: 01 Lun: 00
  Vendor: IBM      Model: ULTRIUM-HH8      Rev: K4K1
  Type:   Sequential-Access                ANSI  SCSI revision: 06

Tony Battersby
Cybernetics