All sysfs mgmt show functions print available parameters/attributes
without '.' at the end, except for scst_luns_mgmt_show(). So fix it.
This patch doesn't change any functionality.
The SCST device may receive several events almost simultaneously to
update its designators. Each such event calls scst_cm_update_dev(),
which frees all device designators in the global list and then submits
a inquiry that fills a new designators into the scst_cm_desig_list list.
This is racy because submiting the inquiry is asynchronous and can be
finished in another thread.
scst_cm_update_dev() 1 scst_cm_update_dev() 2
---------------------- ----------------------
[1] mutex_lock(&scst_mutex)
[2] scst_cm_send_init_inquiry(dev, lun, NULL)
[3] mutex_unlock(&scst_mutex)
[4] mutex_lock(&scst_mutex)
[5] scst_cm_dev_free_designators(dev)
[6] scst_cm_init_inq_finish()
[7] scst_cm_send_init_inquiry(dev, lun, NULL)
[8] mutex_unlock(&scst_mutex)
As a result we may get the scst_cm_desig_list list, which contains
SCST device designators from several inquiries.
Hence serialize scst_cm_desig_list list updates.
Introduce the helper function scst_cm_dev_free_designators(), which
purpose is to free copy manager designators for scst dev.
This patch doesn't change any functionality.
This patch should fix the following bug:
list_del corruption. next->prev should be ffff955cb1ea2540, but was ffff955c54a32440
------------[ cut here ]------------
kernel BUG at lib/list_debug.c:54!
invalid opcode: 0000 [#1] SMP PTI
Workqueue: events vdev_inq_changed_fn [scst_vdisk]
RIP: 0010:__list_del_entry_valid.cold+0x1d/0x47
Call Trace:
scst_cm_dev_unregister+0x66/0xd0 [scst]
scst_cm_update_dev+0x41/0xc0 [scst]
process_one_work+0x1ee/0x390
worker_thread+0x53/0x3e0
kthread+0x124/0x150
ret_from_fork+0x1f/0x30
scst_cm_desig_list is a global list for all SCST devices. It must be
protected with scst_cm_mutex because it can be modified by
scst_cm_init_inq_finish() from another thread when scst_cm_update_dev()
is called.
Fixes: https://github.com/SCST-project/scst/issues/99
Introduce the helper function scst_cm_get_free_lun(), which purpose is
to return the next free copy manager LUN.
This patch doesn't change any functionality.
Support for scst_tgt_template detect() method was declared obsolete
in 2015. Remove support for scst_tgt_template detect() method.
See also commit 4ac6d7a26d ("[PATCH] scst: Drop detect() method").
This reverts commit 1a15ee560c.
Commit 398d41e753 ("scst: Revert "Increase tgt_dev_cmd_count by one"
(r8465)") reverted commit[1].
But there was no revert for commit[2], so add it.
[1] Commit 5b0f9ef0b9 ("scst: Increase tgt_dev_cmd_count by one")
[2] Commit 1a15ee560c ("scst: Fix __scst_init_cmd()")
1. atomic_inc_return() avoids a race in between atomic_read() &
atomic_inc().
2. atomic_inc_return() is a few cycles faster than
atomic_inc()/atomic_read() sequence.
Commit b6faaaf796d7 ("scsi: qla2xxx: Serialize mailbox request") serialized
mailbox requests from userspace using the 'optrom' mutex. However, in the
case of DCBX TLV data, if the memory for it is already allocated, then the
mailbox request ends up not being serialized because it is done without
holding the 'optrom' mutex.
Link: https://lore.kernel.org/r/20220926230245.790508-1-rafaelmendsr@gmail.com
Fixes: b6faaaf796d7 ("scsi: qla2xxx: Serialize mailbox request")
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Rafael Mendonca <rafaelmendsr@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
[ commit 3ddeabd1536a upstream ]
qla2x00_get_fw_version_str() has been removed since commit abbd8870b9cb
("[SCSI] qla2xxx: Factor-out ISP specific functions to method-based call
tables.").
qla2x00_release_nvram_protection() has been removed since commit
459c537807bd ("[SCSI] qla2xxx: Add ISP24xx flash-manipulation routines.").
qla82xx_rdmem() and qla82xx_wrmem() have been removed since commit
3711333dfbee ("[SCSI] qla2xxx: Updates for ISP82xx.").
qla25xx_rd_req_reg(), qla24xx_rd_req_reg(), qla25xx_wrt_rsp_reg(),
qla24xx_wrt_rsp_reg(), qla25xx_wrt_req_reg() and qla24xx_wrt_req_reg() have
been removed since commit 08029990b25b ("[SCSI] qla2xxx: Refactor
request/response-queue register handling.").
qla2x00_async_login_done() has been removed since commit 726b85487067
("qla2xxx: Add framework for async fabric discovery").
qlt_24xx_process_response_error() has been removed since commit
c5419e2618b9 ("scsi: qla2xxx: Combine Active command arrays.").
Remove the declarations for them from header file.
Link: https://lore.kernel.org/r/20220913023722.547249-2-cuigaosheng1@huawei.com
Signed-off-by: Gaosheng Cui <cuigaosheng1@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
[ commit 1b80addaae09 upstream ]
Commit 7c15fc8f0d93 ("qla2x00t-32gbit: Enhance driver tracing with
separate tunable and more") introduced the use of the trace.h/trace_events.h
API.
Due to support for older kernel versions, limit the minimum kernel version to
use this enhance driver tracing to v5.5.
See also commit 288797871473 ("tracing: Adding new functions for kernel
access to Ftrace instances") # v5.5.
Older tracing of driver messages was to:
- log only debug messages to kernel main trace buffer; and
- log only if extended logging bits corresponding to this message is
off
This has been modified and extended as follows:
- Tracing is now controlled via ql2xextended_error_logging_ktrace
module parameter. Bit usages same as ql2xextended_error_logging.
- Tracing uses "qla2xxx" trace instance, unless instance creation have
issues.
- Tracing is enabled (compile time tunable).
- All driver messages, include debug and log messages are now traced in
kernel trace buffer.
Trace messages can be viewed by looking at the qla2xxx instance at:
/sys/kernel/tracing/instances/qla2xxx/trace
Trace tunable that takes the same bit mask as ql2xextended_error_logging
is:
ql2xextended_error_logging_ktrace (default=1)
Link: https://lore.kernel.org/r/20220826102559.17474-6-njavali@marvell.com
Suggested-by: Daniel Wagner <dwagner@suse.de>
Suggested-by: Steven Rostedt <rostedt@goodmis.org>
Tested-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Reviewed-by: Daniel Wagner <dwagner@suse.de>
Signed-off-by: Arun Easi <aeasi@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
[ commit 8bfc149ba24c upstream ]
This message is helpful to troubleshoot missing LUNs/SAN boot errors. It'd
be nice to log it by default instead of only being enabled with debug.
This user had an accidental/forgotten file modprobe.d/qla2xxx.conf w/
option qlini_mode=disabled from experiments with FC target mode, and their
boot LUN didn't come up, as it skips SCSI scan, of course.
However, their boot log didn't provide any clues to help understand that.
The issue/message could be figured out w/ ql2xextended_error_logging, but
it would have been simpler (or even deflected/addressed by user) if it had
been there by default. And it also would help support/triage/deflection
tooling.
Expected change:
scsi host15: qla2xxx
+qla2xxx [0000:3b:00.0]-00fb:15: skipping scsi_scan_host() for non-initiator port
qla2xxx [0000:3b:00.0]-00fb:15: QLogic QLE2692 - QLE2692 Dual Port 16Gb FC to PCIe Gen3 x8 Adapter.
According to:
qla2x00_probe_one()
...
ret = scsi_add_host(...);
...
ql_log(ql_log_info, ...
"skipping scsi_scan_host() for non-initiator port\n");
...
ql_log(ql_log_info, ...
"QLogic %s - %s.\n", ha->model_number, ha->model_desc);
Link: https://lore.kernel.org/r/20220825120159.275051-1-mfo@canonical.com
Tested-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Mauricio Faria de Oliveira <mfo@canonical.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
[ commit eee8bb4a2b58 upstream ]
Currently qlt_stop_phase1() may fail to call flush_scheduled_work(), for
list_empty() may return true as soon as qlt_sess_work_fn() called
list_del(). In order to close this race window, check list_empty() after
calling flush_scheduled_work().
If this patch causes problems, please check commit c4f135d64382
("workqueue: Wrap flush_workqueue() using a macro"). We are on the way to
remove all flush_scheduled_work() calls from the kernel.
Link: https://lore.kernel.org/r/7f24469d-9e39-3398-d851-329b54c0b923@I-love.SAKURA.ne.jp
Tested-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
[ commit a4345557527f upstream ]
Any attempt to flush kernel-global WQs has possibility of deadlock
so we should simply stop using them, instead introduce scst_event_wq.
See also commit c4f135d64382 ("workqueue: Wrap flush_workqueue()
using a macro") # v5.19.
DID_TARGET_FAILURE is internal to the SCSI layer. Drivers must not use it
because:
1. It's not propagated upwards, so SG IO/passthrough users will not see an
error and think a command was successful.
2. There is no handling for it in scsi_decide_disposition() so it
results in entering SCSI error handling.
This has qla2xxx use DID_NO_CONNECT because it looks like we hit this error
when we can't find a port. It will give us the same hard error behavior and
it seems to match the error where we can't find the endpoint.
Link: https://lore.kernel.org/r/20220812010027.8251-7-michael.christie@oracle.com
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
[ commit a965d35c8741 upstream ]
Support for the following block layer changes in the Linux kernel v6.1:
- a4e1d0b76e7b ("block: Change the return type of blk_mq_map_queues() into void")
Support for the following block layer changes in the Linux kernel v6.1:
- de671d6116b5 ("block: change request end_io handler to pass back a return value")
Support for the following vfs file changes in the Linux kernel v6.1:
- 25885a35a720 ("Change calling conventions for filldir_t")
Support for the following dlm changes in the Linux kernel v6.1:
- 12cda13cfd53 ("fs: dlm: remove DLM_LSFL_FS from uapi")
Sending a REQUEST_SENSE with a buffer size 0 to the LUN that does not
exist causes the following kernel panic:
RIP: 0010:sg_init_table+0x1e/0x30
Call Trace:
scst_alloc_sg+0xc3/0x270 [scst]
scst_set_cmd_error+0x803/0xa40 [scst]
__scst_init_cmd+0x5c3/0xb80 [scst]
scst_cmd_init_done+0x142/0xae0 [scst]
cmnd_rx_start+0x7f5/0x13d0 [iscsi_scst]
isert_pdu_rx+0x54/0x140 [isert_scst]
isert_recv_completion_handler+0x498/0x580 [isert_scst]
isert_poll_cq+0x396/0x800 [isert_scst]
isert_cq_comp_work_cb+0x4a/0x120 [isert_scst]
process_one_work+0x1d1/0x410
worker_thread+0x2b/0x3d0
kthread+0x11a/0x130
ret_from_fork+0x1f/0x40
Hence set bufflen to 18 if a buffer size 0 was passed to avoid the
crash.
Reported-by: Lev Vainblat <lev@zadarastorage.com>
Sending an INQUIRY with a buffer size 0 to the LUN that does not exist
causes the following kernel panic:
RIP: 0010:sg_init_table+0x1e/0x30
Call Trace:
scst_alloc_sg+0xc3/0x270 [scst]
scst_set_cmd_error+0x8c9/0xa80 [scst]
__scst_init_cmd+0x5c3/0xb80 [scst]
scst_cmd_init_done+0x142/0xae0 [scst]
cmnd_rx_start+0x7f5/0x13d0 [iscsi_scst]
isert_pdu_rx+0x54/0x140 [isert_scst]
isert_recv_completion_handler+0x498/0x580 [isert_scst]
isert_poll_cq+0x396/0x800 [isert_scst]
isert_cq_comp_work_cb+0x4a/0x120 [isert_scst]
process_one_work+0x1d1/0x410
worker_thread+0x2b/0x3d0
kthread+0x11a/0x130
ret_from_fork+0x1f/0x40
Hence set bufflen to 36 if a buffer size 0 was passed to avoid the
crash.
Reported-by: Lev Vainblat <lev@zadarastorage.com>
Commit 8f394da36a36 ("scsi: qla2xxx: Drop TARGET_SCF_LOOKUP_LUN_FROM_TAG")
made the __qlt_24xx_handle_abts() function return early if
tcm_qla2xxx_find_cmd_by_tag() didn't find a command, but it missed to clean
up the allocated memory for the management command.
Link: https://lore.kernel.org/r/20220914024924.695604-1-rafaelmendsr@gmail.com
Fixes: 8f394da36a36 ("scsi: qla2xxx: Drop TARGET_SCF_LOOKUP_LUN_FROM_TAG")
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Rafael Mendonca <rafaelmendsr@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
[ commit 601be20fc6a1 upstream ]
SCST_VERSION_CODE has not been updated since SCST v3.4 because
the update-version script used the wrong file for parsing. Hence fix
the update-version script and SCST_VERSION_CODE.
This partially reverts commit d2b292c3f6fd ("scsi: qla2xxx: Enable ATIO
interrupt handshake for ISP27XX")
For some workloads where the host sends a batch of commands and then
pauses, ATIO interrupt coalesce can cause some incoming ATIO entries to be
ignored for extended periods of time, resulting in slow performance,
timeouts, and aborted commands.
Disable interrupt coalesce and re-enable the dedicated ATIO MSI-X
interrupt.
Link: https://lore.kernel.org/r/97dcf365-89ff-014d-a3e5-1404c6af511c@cybernetics.com
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Reviewed-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Tony Battersby <tonyb@cybernetics.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
[ commit 53661ded2460 upstream ]
