Various improvements are made for collecting congestion statistics:
- Pre-existing logic is replaced with use of an hrtimer for increased
reporting accuracy.
- Congestion timestamp information is reorganized into a single struct.
- Common statistic collection logic is refactored into a helper routine.
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-8-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
There is mishandling of SLI-4 CQE status values larger than what is allowed
by the LPFC_IOCB_STATUS_MASK of 4 bits. The LPFC_IOCB_STATUS_MASK is a
leftover SLI-3 construct and serves no purpose in SLI-4 path.
Remove the LPFC_IOCB_STATUS_MASK and clean up general CQE status handling
in SLI-4 completion paths.
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-7-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
A firmware upgrade does not necessitate dumping of phba->dbg_log[] to kmsg
via LOG_TRACE_EVENT. A simple KERN_NOTICE log message should suffice to
notify the user of successful or unsuccessful firmware upgrade. As such,
firmware upgrade log messages are updated to use KERN_NOTICE instead of
LOG_TRACE_EVENT. Additionally, in order to notify the user of reset type
for instantiating newly downloaded firmware, lpfc_log_msg's default
KERN_LEVEL is updated to 5 or KERN_NOTICE.
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-6-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
When NPIV ports are zoned to devices that support both initiator and target
mode, a remote device's initiated PRLI results in unintended final kref
clean up of the device's ndlp structure. This disrupts NPIV ports'
discovery for target devices that support both initiator and target mode.
Modify the NPIV lpfc_drop_node clause such that we allow the ndlp to live
so long as it was in NLP_STE_PLOGI_ISSUE, NLP_STE_REG_LOGIN_ISSUE, or
NLP_STE_PRLI_ISSUE nlp_state. This allows lpfc's issued PRLI completion
routine to determine if the final kref clean up should execute rather than
a remote device's issued PRLI.
Fixes: db651ec225 ("scsi: lpfc: Correct used_rpi count when devloss tmo fires with no recovery")
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-5-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Pre-existing device loss recovery logic via the NLP_IN_RECOV_POST_DEV_LOSS
flag only handled Fabric Port Login, Fabric Controller, Management, and
Name Server addresses.
Fabric domain controllers fall under the same category for usage of the
NLP_IN_RECOV_POST_DEV_LOSS flag. Add a default case statement to mark an
ndlp for device loss recovery.
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-4-justintee8345@gmail.com
Acked-by: Martin Wilck <mwilck@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
In dev_loss_tmo callback routine, we early return if the ndlp is in a state
of rediscovery. This occurs when a target proactively PLOGIs or PRLIs
after an RSCN before the dev_loss_tmo callback routine is scheduled to run.
Move clear of the NLP_IN_DEV_LOSS flag before the ndlp state check in such
cases.
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-3-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Due to a target port D_ID swap, it is possible for the
lpfc_register_remote_port() routine to touch post mortem fc_rport memory
when trying to access fc_rport->dd_data.
The D_ID swap causes a simultaneous call to lpfc_unregister_remote_port(),
where fc_remote_port_delete() reclaims fc_rport memory.
Remove the fc_rport->dd_data->pnode NULL assignment because the following
line reassigns ndlp->rport with an fc_rport object from
fc_remote_port_add() anyways. The pnode nullification is superfluous.
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-2-justintee8345@gmail.com
Acked-by: Martin Wilck <mwilck@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Bart Van Assche <bvanassche@acm.org> says:
In the traces we recorded while testing zoned storage we noticed that UFS
commands are requeued while the clock is being ungated. Command requeueing
makes it harder than necessary to preserve the command order. Hence this
patch series that modifies the SCSI core and also the UFS driver such that
clock ungating does not trigger command requeueing.
Link: https://lore.kernel.org/r/20230529202640.11883-1-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Ungating the clock asynchronously causes ufshcd_queuecommand() to return
SCSI_MLQUEUE_HOST_BUSY and hence causes commands to be requeued. This is
suboptimal. Allow ufshcd_queuecommand() to sleep such that clock ungating
does not trigger command requeuing. Remove the ufshcd_scsi_block_requests()
and ufshcd_scsi_unblock_requests() calls because these are no longer
needed. The flush_work(&hba->clk_gating.ungate_work) call is sufficient to
make the SCSI core wait for clock ungating to complete.
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20230529202640.11883-6-bvanassche@acm.org
Reviewed-by: Bean Huo <beanhuo@micron.com>
Reviewed-by: Bao D. Nguyen <quic_nguyenb@quicinc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Niklas Cassel <nks@flawful.org> says:
This series adds support for Command Duration Limits.
The series is based on linux tag: v6.4-rc1
The series can also be found in git: https://github.com/floatious/linux/commits/cdl-v7
=================
CDL in ATA / SCSI
=================
Command Duration Limits is defined in:
T13 ATA Command Set - 5 (ACS-5) and
T10 SCSI Primary Commands - 6 (SPC-6) respectively
(a simpler version of CDL is defined in T10 SPC-5).
CDL defines Duration Limits Descriptors (DLD).
7 DLDs for read commands and 7 DLDs for write commands.
Simply put, a DLD contains a limit and a policy.
A command can specify that a certain limit should be applied by setting
the DLD index field (3 bits, so 0-7) in the command itself.
The DLD index points to one of the 7 DLDs.
DLD index 0 means no descriptor, so no limit.
DLD index 1-7 means DLD 1-7.
A DLD can have a few different policies, but the two major ones are:
-Policy 0xF (abort), command will be completed with command aborted error
(ATA) or status CHECK CONDITION (SCSI), with sense data indicating that
the command timed out.
-Policy 0xD (complete-unavailable), command will be completed without
error (ATA) or status GOOD (SCSI), with sense data indicating that the
command timed out. Note that the command will not have transferred any
data to/from the device when the command timed out, even though the
command returned success.
Regardless of the CDL policy, in case of a CDL timeout, the I/O will
result in a -ETIME error to user-space.
The DLDs are defined in the CDL log page(s) and are readable and writable.
Reading and writing the CDL DLDs are outside the scope of the kernel.
If a user wants to read or write the descriptors, they can do so using a
user-space application that sends passthrough commands, such as cdl-tools:
https://github.com/westerndigitalcorporation/cdl-tools
================================
The introduction of ioprio hints
================================
What the kernel does provide, is a method to let I/O use one of the CDL DLDs
defined in the device. Note that the kernel will simply forward the DLD index
to the device, so the kernel currently does not know, nor does it need to know,
how the DLDs are defined inside the device.
The way that the CDL DLD index is supplied to the kernel is by introducing a
new 10 bit "ioprio hint" field within the existing 16 bit ioprio definition.
Currently, only 6 out of the 16 ioprio bits are in use, the remaining 10 bits
are unused, and are currently explicitly disallowed to be set by the kernel.
For now, we only add ioprio hints representing CDL DLD index 1-7. Additional
ioprio hints for other QoS features could be defined in the future.
A theoretical future work could be to make an I/O scheduler aware of these
hints. E.g. for CDL, an I/O scheduler could make use of the duration limit
in each descriptor, and take that information into account while scheduling
commands. Right now, the ioprio hints will be ignored by the I/O schedulers.
==============================
How to use CDL from user-space
==============================
Since CDL is mutually exclusive with NCQ priority
(see ncq_prio_enable and sas_ncq_prio_enable in
Documentation/ABI/testing/sysfs-block-device),
CDL has to be explicitly enabled using:
echo 1 > /sys/block/$bdev/device/cdl_enable
Since the ioprio hints are supplied through the existing I/O priority API,
it should be simple for an application to make use of the ioprio hints.
It simply has to reuse one of the new macros defined in
include/uapi/linux/ioprio.h: IOPRIO_PRIO_HINT() or IOPRIO_PRIO_VALUE_HINT(),
and supply one of the new hints defined in include/uapi/linux/ioprio.h:
IOPRIO_HINT_DEV_DURATION_LIMIT_[1-7], which indicates that the I/O should
use the corresponding CDL DLD index 1-7.
By reusing the I/O priority API, the user can both define a DLD to use per
AIO (io_uring sqe->ioprio or libaio iocb->aio_reqprio) or per-thread
(ioprio_set()).
=======
Testing
=======
With the following fio patches:
https://github.com/floatious/fio/commits/cdl
fio adds support for ioprio hints, such that CDL can be tested using e.g.:
fio --ioengine=io_uring --cmdprio_percentage=10 --cmdprio_hint=DLD_index
A simple way to test is to use a DLD with a very short duration limit,
and send large reads. Regardless of the CDL policy, in case of a CDL
timeout, the I/O will result in a -ETIME error to user-space.
We also provide a CDL test suite located in the cdl-tools repo, see:
https://github.com/westerndigitalcorporation/cdl-tools#testing-a-system-command-duration-limits-support
We have tested this patch series using:
-real hardware
-the following QEMU implementation:
https://github.com/floatious/qemu/tree/cdl
(NOTE: the QEMU implementation requires you to define the CDL policy at compile
time, so you currently need to recompile QEMU when switching between policies.)
===================
Further information
===================
For further information about CDL, see Damien's slides:
Presented at SDC 2021:
https://www.snia.org/sites/default/files/SDC/2021/pdfs/SNIA-SDC21-LeMoal-Be-On-Time-command-duration-limits-Feature-Support-in%20Linux.pdf
Presented at Lund Linux Con 2022:
https://drive.google.com/file/d/1I6ChFc0h4JY9qZdO1bY5oCAdYCSZVqWw/view?usp=sharing
================
Changes since V6
================
-Rebased series on v6.4-rc1.
-Picked up Reviewed-by tags from Hannes (Thank you Hannes!)
-Picked up Reviewed-by tag from Christoph (Thank you Christoph!)
-Changed KernelVersion from 6.4 to 6.5 for new sysfs attributes.
For older change logs, see previous patch series versions:
https://lore.kernel.org/linux-scsi/20230406113252.41211-1-nks@flawful.org/https://lore.kernel.org/linux-scsi/20230404182428.715140-1-nks@flawful.org/https://lore.kernel.org/linux-scsi/20230309215516.3800571-1-niklas.cassel@wdc.com/https://lore.kernel.org/linux-scsi/20230124190308.127318-1-niklas.cassel@wdc.com/https://lore.kernel.org/linux-scsi/20230112140412.667308-1-niklas.cassel@wdc.com/https://lore.kernel.org/linux-scsi/20221208105947.2399894-1-niklas.cassel@wdc.com/
Link: https://lore.kernel.org/r/20230511011356.227789-1-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
A CDL timeout for policy 0xF is defined as a NCQ error, just with a CDL
specific sk/asc/ascq in the sense data. Therefore, the existing code in
libata does not need to be modified to handle a policy 0xF CDL timeout.
For Command Duration Limits policy 0xD:
The device shall complete the command without error with the additional
sense code set to DATA CURRENTLY UNAVAILABLE.
Since a CDL timeout for policy 0xD is not an error, we cannot use the NCQ
Command Error log (10h).
Instead, we need to read the Sense Data for Successful NCQ Commands log
(0Fh).
In the success case, just like in the error case, we cannot simply read a
log page from the interrupt handler itself, since reading a log page
involves sending a READ LOG DMA EXT or READ LOG EXT command.
Therefore, we add a new EH action ATA_EH_GET_SUCCESS_SENSE. When a command
completes without error, and when the ATA_SENSE bit is set, this new action
is set as pending, and EH is scheduled.
This way, similar to the NCQ error case, the log page will be read from EH
context.
An alternative would have been to add a new kthread or workqueue to handle
this. However, extending EH can be done with minimal changes and avoids the
need to synchronize a new kthread/workqueue with EH.
Co-developed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-20-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
For devices supporting the command duration limits feature, translate the
dld field of read and write operation to set the command duration limit
index field of the command task file when the duration limit feature is
enabled.
The function ata_set_tf_cdl() is introduced to do this. For unqueued (non
NCQ) read and write operations, this function sets the command duration
limit index set as the lower 3 bits of the feature field. For queued NCQ
read/write commands, the index is set as the lower 3 bits of the auxiliary
field.
The flag ATA_QCFLAG_HAS_CDL is introduced to indicate that a command
taskfile has a non zero cdl field.
Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Igor Pylypiv <ipylypiv@google.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Co-developed-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-19-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
