Several errors have occurred where the adapter stops or fails but does not
raise the register values for the driver to detect failure. Thus driver is
unaware of the failure. The failure typically results in I/O timeouts, the
I/O timeout handler failing (after several seconds), and the error handler
escalating recovery policy and resulting in more errors. Eventually, the
driver is in a position where things have spiraled and it can't do recovery
because other recovery ops are still outstanding and it becomes unusable.
Resolve the situation by having the I/O timeout handler (actually a els,
SCSI I/O, NVMe ls, or NVMe I/O timeout), in addition to aborting the I/O,
perform a mailbox command and look for a response from the hardware. If
the mailbox command fails, it will mark the adapter offline and then invoke
the adapter reset handler to clean up.
The new I/O timeout test will be limited to a test every 5s. If there are
multiple I/O timeouts concurrently, only the 1st I/O timeout will generate
the mailbox command. Further testing will only occur once a timeout occurs
after a 5s delay from the last mailbox command has expired.
Link: https://lore.kernel.org/r/20210104180240.46824-14-jsmart2021@gmail.com
Co-developed-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
If a mailbox command times out, the SLI port is deemed in error and the
port is reset. The HBA cleanup is not returning I/Os to the NVMe layer
before the port is unregistered. This is due to the HBA being marked
offline (!SLI_ACTIVE) and cleanup being done by the mailbox timeout handler
rather than an general adapter reset routine. The mailbox timeout handler
mailbox handler only cleaned up SCSI I/Os.
Fix by reworking the mailbox handler to:
- After handling the mailbox error, detect the board is already in
failure (may be due to another error), and leave cleanup to the
other handler.
- If the mailbox command timeout is initial detector of the port error,
continue with the board cleanup and marking the adapter offline
(!SLI_ACTIVE). Remove the SCSI-only I/O cleanup routine. The generic
reset adapter routine that is subsequently invoked, will clean up the
I/Os.
- Have the reset adapter routine flush all NVMe and SCSI I/Os if the
adapter has been marked failed (!SLI_ACTIVE).
- Rework the NVMe I/O terminate routine to take a status code to fail the
I/O with and update so that cleaned up I/O calls the wqe completion
routine. Currently it is bypassing the wqe cleanup and calling the NVMe
I/O completion directly. The wqe completion routine will take care of
data structure and node cleanup then call the NVMe I/O completion
handler.
Link: https://lore.kernel.org/r/20210104180240.46824-11-jsmart2021@gmail.com
Co-developed-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Update Copyright in files changed by the 12.8.0.6 patch set to 2020
Link: https://lore.kernel.org/r/20201115192646.12977-18-james.smart@broadcom.com
Co-developed-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
This patch reworks the abort interfaces such that SLI-3 retains the
iocb-based formatting and completions and SLI-4 now uses native WQEs and
completion routines.
The following changes are made:
- The code is refactored from a confusing 2 routine sequence of
xx_abort_iotag_issue(), which creates/formats and abort cmd, and
xx_issue_abort_tag(), which then issues and handles the completion of
the abort cmd - into a single interface of xx_issue_abort_iotag(). The
new interface will determine whether SLI-3 or SLI-4 and then call the
appropriate handler. A completion handler can now be specified to
address the differences in completion handling. Note: original code is
all iocb based, with SLI-4 converting to SLI-3 for the SCSI/ELS path,
and NVMe natively using wqes.
- The SLI-3 side is refactored:
The older iocb-base lpfc_sli_issue_abort_iotag() routine is combined
with the logic of lpfc_sli_abort_iotag_issue() as well as the
iocb-specific code in lpfc_abort_handler() and lpfc_sli_abort_iocb() to
create the new single SLI-3 abort routine that formats and issues the
iocb.
- The SLI-4 side is refactored and added to:
The native WQE abort code in NVMe is moved to the new SLI-4
issue_abort_iotag() routine. Items in SCSI that set fields not set by
NVMe is migrated into the new routine. Thus the routine supports NVMe
and SCSI initiators. The nvmet block (target) formats the abort slightly
different (like the old NVMe initiator) thus it has its own prep routine
stolen from NVMe initiator and it retains the current code it has for
issuing the WQE (does not use the commonized routine the initiators
do). SLI-4 completion handlers were also added.
- lpfc_abort_handler now becomes a wrapper that determines whether
SLI-3 or SLI-4 and calls the proper abort handler.
Link: https://lore.kernel.org/r/20201115192646.12977-16-james.smart@broadcom.com
Co-developed-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
To set up common use by the SCSI and NVMe I/O paths, create a new routine
that issues FCP I/O commands which can be used by either protocol. The new
routine addresses SLI-3 vs SLI-4 differences within its implementation.
Replace the (SLI-3 centric) iocb routine in the SCSI path with this new
WQE-centric common routine.
Link: https://lore.kernel.org/r/20201115192646.12977-13-james.smart@broadcom.com
Co-developed-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The driver is currently using SLI-4 WQE templates only for NVMe. Refactor
the template and the placement of the service routine so that it can be
used by both SCSI and NVMe.
Link: https://lore.kernel.org/r/20201115192646.12977-12-james.smart@broadcom.com
Co-developed-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
When a remote port is disconnected and disappears, its node structure
(ndlp) stays allocated and on a vport node list. While on the list it can
be matched, thus requires validation checks on state to be added in
numerous code paths. If the node comes back, its possible for there to be
multiple node structures for the same device on the vport node list. There
is no reason to keep the node structure around after it is no longer in
existence, and the current implementation creates problems for itself
(multiple nodes) and lots of unnecessary code for state validation.
Additionally, the reference taking on the node structure didn't follow the
normal model used by the kernel kref api. It included lots of odd logic to
match state with reference count. The combination of this odd logic plus
the way it was implicitly used in the discovery engine made its reference
taking implementation suspect and extremely hard to follow.
Change the driver such that the reference taking routines are now normal
ref increments/decrements and callout on refcount=0.
With this in place, the rework can be done such that the node structure is
fully removed and deallocated when the remote port no longer exists and all
references are removed. This removal logic, and the basic ref counting are
intrically tied, thus in a single patch.
Link: https://lore.kernel.org/r/20201115192646.12977-2-james.smart@broadcom.com
Co-developed-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The current logging methods typically end up requesting a reproduction with
a different logging level set to figure out what happened. This was mainly
by design to not clutter the kernel log messages with things that were
typically not interesting and the messages themselves could cause other
issues.
When looking to make a better system, it was seen that in many cases when
more data was wanted was when another message, usually at KERN_ERR level,
was logged. And in most cases, what the additional logging that was then
enabled was typically. Most of these areas fell into the discovery machine.
Based on this summary, the following design has been put in place: The
driver will maintain an internal log (256 elements of 256 bytes). The
"additional logging" messages that are usually enabled in a reproduction
will be changed to now log all the time to the internal log. A new logging
level is defined - LOG_TRACE_EVENT. When this level is set (it is not by
default) and a message marked as KERN_ERR is logged, all the messages in
the internal log will be dumped to the kernel log before the KERN_ERR
message is logged.
There is a timestamp on each message added to the internal log. However,
this timestamp is not converted to wall time when logged. The value of the
timestamp is solely to give a crude time reference for the messages.
Link: https://lore.kernel.org/r/20200630215001.70793-14-jsmart2021@gmail.com
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
As the nvmet layer does not have the concept of a remoteport object, which
can be used to identify the entity on the other end of the fabric that is
to receive an LS, the hosthandle was introduced. The driver passes the
hosthandle, a value representative of the remote port, with a ls request
receive. The LS request will create the association. The transport will
remember the hosthandle for the association, and if there is a need to
initiate a LS request to the remote port for the association, the
hosthandle will be used. When the driver loses connectivity with the
remote port, it needs to notify the transport that the hosthandle is no
longer valid, allowing the transport to terminate associations related to
the hosthandle.
This patch adds support to the driver for the hosthandle. The driver will
use the ndlp pointer of the remote port for the hosthandle in calls to
nvmet_fc_rcv_ls_req(). The discovery engine is updated to invalidate the
hosthandle whenever connectivity with the remote port is lost.
Signed-off-by: Paul Ely <paul.ely@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
In preparation for supporting both intiator mode and target mode
receiving NVME LS's, commonize the existing NVME LS request receive
handling found in the base driver and in the nvmet side.
Using the original lpfc_nvmet_unsol_ls_event() and
lpfc_nvme_unsol_ls_buffer() routines as a templates, commonize the
reception of an NVME LS request. The common routine will validate the LS
request, that it was received from a logged-in node, and allocate a
lpfc_async_xchg_ctx that is used to manage the LS request. The role of
the port is then inspected to determine which handler is to receive the
LS - nvme or nvmet. As such, the nvmet handler is tied back in. A handler
is created in nvme and is stubbed out.
Signed-off-by: Paul Ely <paul.ely@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
To support FC-NVME-2 support (actually FC-NVME (rev 1) with Ammendment 1),
both the nvme (host) and nvmet (controller/target) sides will need to be
able to receive LS requests. Currently, this support is in the nvmet side
only. To prepare for both sides supporting LS receive, rename
lpfc_nvmet_rcv_ctx to lpfc_async_xchg_ctx and commonize the definition.
Signed-off-by: Paul Ely <paul.ely@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Currently driver ktime stats, measuring code paths, is NVME-specific.
Convert the stats routines such that the code paths are generic, providing
status for NVME and SCSI. Added ktime stat calls in SCSI queuecommand and
cmpl routines.
Link: https://lore.kernel.org/r/20200322181304.37655-11-jsmart2021@gmail.com
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
SCSI layer sends driver IOs with more s/g segments than driver can handle.
This results in "Too many sg segments from dma_map_sg. Config 64, seg_cnt
219" error messages from the lpfc_scsi_prep_dma_buf_s3() routine.
The was due to use the driver using individual templates for pport and
vport, host reset enabled or not, nvme vs scsi, etc. In the end, there was
a combination for a vport that didn't match the pport.
Rather than enumerating more templates and more discretionary assignments,
revert to a base template that is copied to a template specific to the
pport/vport. Then, based on role, attributes and sli type, modify the
fields that are different for that port. Added a log message to
lpfc_create_port to validate values.
Link: https://lore.kernel.org/r/20200322181304.37655-5-jsmart2021@gmail.com
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
This patch modifies lpfc to register for Link Integrity events via the use
of an RDF ELS and to perform Link Integrity FPIN logging.
Specifically, the driver was modified to:
- Format and issue the RDF ELS immediately following SCR registration.
This registers the ability of the driver to receive FPIN ELS.
- Adds decoding of the FPIN els into the received descriptors, with
logging of the Link Integrity event information. After decoding, the ELS
is delivered to the scsi fc transport to be delivered to any user-space
applications.
- To aid in logging, simple helpers were added to create enum to name
string lookup functions that utilize the initialization helpers from the
fc_els.h header.
- Note: base header definitions for the ELS's don't populate the
descriptor payloads. As such, lpfc creates it's own version of the
structures, using the base definitions (mostly headers) and additionally
declaring the descriptors that will complete the population of the ELS.
Link: https://lore.kernel.org/r/20200210173155.547-3-jsmart2021@gmail.com
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
There are reports of multiple ports on the same system displaying different
hostnames in fabric FDMI displays.
Currently, the driver registers the hostname at initialization and obtains
the hostname via init_utsname()->nodename queried at the time the FC link
comes up. Unfortunately, if the machine hostname is updated after
initialization, such as via DHCP or admin command, the value registered
initially will be incorrect.
Fix by having the driver save the hostname that was registered with FDMI.
The driver then runs a heartbeat action that will check the hostname. If
the name changes, reregister the FMDI data.
The hostname is used in RSNN_NN, FDMI RPA and FDMI RHBA.
Link: https://lore.kernel.org/r/20191218235808.31922-5-jsmart2021@gmail.com
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Compilation can fail due to having an inline function reference where the
function body is not present.
Fix by removing the inline tag.
Fixes: 93a4d6f401 ("scsi: lpfc: Add registration for CPU Offline/Online events")
Link: https://lore.kernel.org/r/20191111230401.12958-4-jsmart2021@gmail.com
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The recent affinitization didn't address cpu offlining/onlining. If an
interrupt vector is shared and the low order cpu owning the vector is
offlined, as interrupts are managed, the vector is taken offline. This
causes the other CPUs sharing the vector will hang as they can't get io
completions.
Correct by registering callbacks with the system for Offline/Online
events. When a cpu is taken offline, its eq, which is tied to an interrupt
vector is found. If the cpu is the "owner" of the vector and if the
eq/vector is shared by other CPUs, the eq is placed into a polled mode.
Additionally, code paths that perform io submission on the "sharing CPUs"
will check the eq state and poll for completion after submission of new io
to a wq that uses the eq.
Similarly, when a cpu comes back online and owns an offlined vector, the eq
is taken out of polled mode and rearmed to start driving interrupts for eq.
Link: https://lore.kernel.org/r/20191105005708.7399-9-jsmart2021@gmail.com
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
When the port, running as a nvme target, receives an ABTS, it submits
commands to the adapter to Abort i/o outstanding in the adapter. The Abort
command formatting routine left a command field set to zero, which
instructs the adapter to generate an ABTS on the wire as part of cleaning
up the I/O. This is common operation for an initiator, but not for a
target.
Fix the driver to check whether an ABTS had been received for the I/O, and
if so, change the Abort command formatting so that the ABTS generation is
disabled (IA=1). No need to ABTS it when the other side already has.
Also refactored the code such that there is a single routine being used for
nvme or nvmet ABORT requests, and IA is an argument.
Link: https://lore.kernel.org/r/20190922035906.10977-11-jsmart2021@gmail.com
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Capturing and downloading dif command data and dif data was done a dozen
years ago and no longer being used. Also creates a potential security hole.
Remove the debugfs buffer for dif debugging.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
CC: KyleMahlkuch <kmahlkuc@linux.vnet.ibm.com>
CC: Hannes Reinecke <hare@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Currently, each hardware queue, typically allocated per-cpu, consists of a
WQ/CQ pair per protocol. Meaning if both SCSI and NVMe are supported 2
WQ/CQ pairs will exist for the hardware queue. Separate queues are
unnecessary. The current implementation wastes memory backing the 2nd set
of queues, and the use of double the SLI-4 WQ/CQ's means less hardware
queues can be supported which means there may not always be enough to have
a pair per cpu. If there is only 1 pair per cpu, more cpu's may get their
own WQ/CQ.
Rework the implementation to use a single WQ/CQ pair by both protocols.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
As part of firmware download, the adapter is reset. On the adapter the
reset causes the function to stop and all outstanding io is terminated
(without responses). The reset path then starts teardown of the adapter,
starting with deregistration of the remote ports with the nvme-fc
transport. The local port is then deregistered and the driver waits for
local port deregistration. This never finishes.
The remote port deregistrations terminated the nvme controllers, causing
them to send aborts for all the outstanding io. The aborts were serviced in
the driver, but stalled due to its state. The nvme layer then stops to
reclaim it's outstanding io before continuing. The io must be returned
before the reset on the controller is deemed complete and the controller
delete performed. The remote port deregistration won't complete until all
the controllers are terminated. And the local port deregistration won't
complete until all controllers and remote ports are terminated. Thus things
hang.
The issue is the reset which stopped the adapter also stopped all the
responses that would drive i/o completions, and the aborts were also
stopped that stopped i/o completions. The driver, when resetting the
adapter like this, needs to be generating the completions as part of the
adapter reset so that I/O complete (in error), and any aborts are not
queued.
Fix by adding flush routines whenever the adapter port has been reset or
discovered in error. The flush routines will generate the completions for
the scsi and nvme outstanding io. The abort ios, if waiting, will be caught
and flushed as well.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
This is mostly update of the usual drivers: qla2xxx, hpsa, lpfc, ufs,
mpt3sas, ibmvscsi, megaraid_sas, bnx2fc and hisi_sas as well as the
removal of the osst driver (I heard from Willem privately that he
would like the driver removed because all his test hardware has
failed). Plus number of minor changes, spelling fixes and other
trivia.
Signed-off-by: James E.J. Bottomley <jejb@linux.ibm.com>
-----BEGIN PGP SIGNATURE-----
iJwEABMIAEQWIQTnYEDbdso9F2cI+arnQslM7pishQUCXSTl4yYcamFtZXMuYm90
dG9tbGV5QGhhbnNlbnBhcnRuZXJzaGlwLmNvbQAKCRDnQslM7pishdcxAQDCJVbd
fPUX76/V1ldupunF97+3DTharxxbst+VnkOnCwD8D4c0KFFFOI9+F36cnMGCPegE
fjy17dQLvsJ4GsidHy8=
=aS5B
-----END PGP SIGNATURE-----
Merge tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Pull SCSI updates from James Bottomley:
"This is mostly update of the usual drivers: qla2xxx, hpsa, lpfc, ufs,
mpt3sas, ibmvscsi, megaraid_sas, bnx2fc and hisi_sas as well as the
removal of the osst driver (I heard from Willem privately that he
would like the driver removed because all his test hardware has
failed). Plus number of minor changes, spelling fixes and other
trivia.
The big merge conflict this time around is the SPDX licence tags.
Following discussion on linux-next, we believe our version to be more
accurate than the one in the tree, so the resolution is to take our
version for all the SPDX conflicts"
Note on the SPDX license tag conversion conflicts: the SCSI tree had
done its own SPDX conversion, which in some cases conflicted with the
treewide ones done by Thomas & co.
In almost all cases, the conflicts were purely syntactic: the SCSI tree
used the old-style SPDX tags ("GPL-2.0" and "GPL-2.0+") while the
treewide conversion had used the new-style ones ("GPL-2.0-only" and
"GPL-2.0-or-later").
In these cases I picked the new-style one.
In a few cases, the SPDX conversion was actually different, though. As
explained by James above, and in more detail in a pre-pull-request
thread:
"The other problem is actually substantive: In the libsas code Luben
Tuikov originally specified gpl 2.0 only by dint of stating:
* This file is licensed under GPLv2.
In all the libsas files, but then muddied the water by quoting GPLv2
verbatim (which includes the or later than language). So for these
files Christoph did the conversion to v2 only SPDX tags and Thomas
converted to v2 or later tags"
So in those cases, where the spdx tag substantially mattered, I took the
SCSI tree conversion of it, but then also took the opportunity to turn
the old-style "GPL-2.0" into a new-style "GPL-2.0-only" tag.
Similarly, when there were whitespace differences or other differences
to the comments around the copyright notices, I took the version from
the SCSI tree as being the more specific conversion.
Finally, in the spdx conversions that had no conflicts (because the
treewide ones hadn't been done for those files), I just took the SCSI
tree version as-is, even if it was old-style. The old-style conversions
are perfectly valid, even if the "-only" and "-or-later" versions are
perhaps more descriptive.
* tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi: (185 commits)
scsi: qla2xxx: move IO flush to the front of NVME rport unregistration
scsi: qla2xxx: Fix NVME cmd and LS cmd timeout race condition
scsi: qla2xxx: on session delete, return nvme cmd
scsi: qla2xxx: Fix kernel crash after disconnecting NVMe devices
scsi: megaraid_sas: Update driver version to 07.710.06.00-rc1
scsi: megaraid_sas: Introduce various Aero performance modes
scsi: megaraid_sas: Use high IOPS queues based on IO workload
scsi: megaraid_sas: Set affinity for high IOPS reply queues
scsi: megaraid_sas: Enable coalescing for high IOPS queues
scsi: megaraid_sas: Add support for High IOPS queues
scsi: megaraid_sas: Add support for MPI toolbox commands
scsi: megaraid_sas: Offload Aero RAID5/6 division calculations to driver
scsi: megaraid_sas: RAID1 PCI bandwidth limit algorithm is applicable for only Ventura
scsi: megaraid_sas: megaraid_sas: Add check for count returned by HOST_DEVICE_LIST DCMD
scsi: megaraid_sas: Handle sequence JBOD map failure at driver level
scsi: megaraid_sas: Don't send FPIO to RL Bypass queue
scsi: megaraid_sas: In probe context, retry IOC INIT once if firmware is in fault
scsi: megaraid_sas: Release Mutex lock before OCR in case of DCMD timeout
scsi: megaraid_sas: Call disable_irq from process IRQ poll
scsi: megaraid_sas: Remove few debug counters from IO path
...
This patch updates RSCN receive processing to check for the remote
port being an NVME port, and if so, invoke the nvme_fc callback to
rescan the remote port. The rescan will generate a discovery udev
event.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Arun Easi <aeasi@marvell.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
This patch adds general RSCN support:
- The ability to transmit an RSCN to the port on the other end of
the link (regular port if pt2pt, or fabric controller if fabric).
- And general recognition of an RSCN ELS when an ELS is received.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Arun Easi <aeasi@marvell.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Currently the driver is notified of new command frame receipt by CQEs. As
part of the CQE processing, the driver upcalls the nvmet_fc transport to
deliver the command. nvmet_fc, as part of receiving the command builds out
a context for it, where one of the first steps is to allocate memory for
the io.
When running with tests that do large ios (1MB), it was found on some
systems, the total number of outstanding I/O's, at 1MB per, completely
consumed the system's memory. Thus additional ios were getting blocked in
the memory allocator. Given that this blocked the lpfc thread processing
CQEs, there were lots of other commands that were received and which are
then held up, and given CQEs are serially processed, the aggregate delays
for an IO waiting behind the others became cummulative - enough so that the
initiator hit timeouts for the ios.
The basic fix is to avoid the direct upcall and instead schedule a work
item for each io as it is received. This allows the cq processing to
complete very quickly, and each io can then run or block on it's own.
However, this general solution hurts latency when there are few ios. As
such, implemented the fix such that the driver watches how many CQEs it has
processed sequentially in one run. As long as the count is below a
threshold, the direct nvmet_fc upcall will be made. Only when the count is
exceeded will it revert to work scheduling.
Given that debug of this showed a surprisingly long delay in cq processing,
the io timer stats were updated to better reflect the processing of the
different points.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
For files modified as part of 12.2.0.0 patches, update copyright to 2019
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
So far MSIX vector allocation assumed it would be 1:1 with hardware
queues. However, there are several reasons why fewer MSIX vectors may be
allocated than hardware queues such as the platform being out of vectors or
adapter limits being less than cpu count.
This patch reworks the MSIX/EQ relationships with the per-cpu hardware
queues so they can function independently. MSIX vectors will be equitably
split been cpu sockets/cores and then the per-cpu hardware queues will be
mapped to the vectors most efficient for them.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The XRI get/put lists were partitioned per hardware queue. However, the
adapter rarely had sufficient resources to give a large number of resources
per queue. As such, it became common for a cpu to encounter a lack of XRI
resource and request the upper io stack to retry after returning a BUSY
condition. This occurred even though other cpus were idle and not using
their resources.
Create as efficient a scheme as possible to move resources to the cpus that
need them. Each cpu maintains a small private pool which it allocates from
for io. There is a watermark that the cpu attempts to keep in the private
pool. The private pool, when empty, pulls from a global pool from the
cpu. When the cpu's global pool is empty it will pull from other cpu's
global pool. As there many cpu global pools (1 per cpu or hardware queue
count) and as each cpu selects what cpu to pull from at different rates and
at different times, it creates a radomizing effect that minimizes the
number of cpu's that will contend with each other when the steal XRI's from
another cpu's global pool.
On io completion, a cpu will push the XRI back on to its private pool. A
watermark level is maintained for the private pool such that when it is
exceeded it will move XRI's to the CPU global pool so that other cpu's may
allocate them.
On NVME, as heartbeat commands are critical to get placed on the wire, a
single expedite pool is maintained. When a heartbeat is to be sent, it will
allocate an XRI from the expedite pool rather than the normal cpu
private/global pools. On any io completion, if a reduction in the expedite
pools is seen, it will be replenished before the XRI is placed on the cpu
private pool.
Statistics are added to aid understanding the XRI levels on each cpu and
their behaviors.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
SLI4 nvme functions are passing the SLI3 ring number when posting wqe to
hardware. This should be indicating the hardware queue to use, not the ring
number.
Replace ring number with the hardware queue that should be used.
Note: SCSI avoided this issue as it utilized an older lfpc_issue_iocb
routine that properly adapts.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Once the IO buff allocations were made shared, there was a single XRI
buffer list shared by all hardware queues. A single list isn't great for
performance when shared across the per-cpu hardware queues.
Create a separate XRI IO buffer get/put list for each Hardware Queue. As
SGLs and associated IO buffers get allocated/posted to the firmware; round
robin their assignment across all available hardware Queues so that there
is an equitable assignment.
Modify SCSI and NVME IO submit code paths to use the Hardware Queue logic
for XRI allocation.
Add a debugfs interface to display hardware queue statistics
Added new empty_io_bufs counter to track if a cpu runs out of XRIs.
Replace common_ variables/names with io_ to make meanings clearer.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
There is a extra queue and msix vector for expresslane. Now that the driver
will be doing queues per cpu, this oddball queue is no longer needed.
Expresslane will utilize the normal per-cpu queues.
Updated debugfs sli4 queue output to go along with the change
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Currently, both NVME and SCSI get their IO buffers from separate
pools. XRI's are associated 1:1 with IO buffers, so XRI's are also split
between protocols.
Eliminate the independent pools and use a single pool. Each buffer
structure now has a common section and a protocol section. Per protocol
routines for SGL initialization are removed and replaced by common
routines. Initialization of the buffers is only done on the common area.
All other fields, which are protocol specific, are initialized when the
buffer is allocated for use in the per-protocol allocation routine.
In the past, the SCSI side allocated IO buffers as part of slave_alloc
calls until the maximum XRIs for SCSI was reached. As all XRIs are now
common and may be used for either protocol, allocation for everything is
done as part of adapter initialization and the scsi side has no action in
slave alloc.
As XRI's are no longer split, the lpfc_xri_split module parameter is
removed.
Adapters based on SLI3 will continue to use the older scsi_buf_list_get/put
routines. All SLI4 adapters utilize the new IO buffer scheme
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
This patch adds a "pci_bus_reset" option to the board_mode sysfs attribute.
This option uses the pci_reset_bus() api to reset the PCIe link the adapter
is on, which will reset the chip/adapter. Prior to issuing this option,
all functions on the same chip must be placed in the offline state by the
admin. After the reset, all of the instances may be brought online again.
The primary purpose of this functionality is to support cases where
firmware update required a chip reset but the admin did not want to reboot
the machine in order to instantiate the firmware update.
Sanity checks take place prior to the reset to ensure the adapter is the
sole entity on the PCIe bus and that all functions are in the offline
state.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
On driver termination, after the driver stops fw logging by writing a
register on the chip, the driver immediately unmaps and frees the logging
buffer, without confirming in any way that the chip has received the write
and terminated the logging. As termination on the chip is not immediate,
the chip may issue a dma request to the now unmapped dma buffer, resulting
in a iommu fault.
Change the driver to receive a confirmation that logging ahs been
terminated. As the driver always issues an SLI reset with the device as
part of shutdown, and as part of that is receiving confirmation that the
reset is complete - the driver was modified to perform the write to disable
fw logging prior to the SLI reset and only free the fw log buffer after the
SLI reset is complete. That guarantees use of the fw log buffer is fully
terminated when it is unmapped.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The driver is getting hit with 100s of RSCNs during remote port address
changes. Each of those RSCN's ends up generating UNREG_RPI and REG_PRI
mailbox commands. The discovery engine within the driver doesn't wait for
the mailbox command completions. Instead it sets state flags and moves
forward. At some point, there's a massive backlog of mailbox commands which
take time for the adapter to process. Additionally, it appears there were
duplicate events from the switch so the driver generated duplicate mailbox
commands for the same remote port. During this window, failures on PLOGI
and PRLI ELS's are see as the adapter is rejecting them as they are for
remote ports that still have pending mailbox commands.
Streamline the discovery engine so that PLOGI log checks for outstanding
UNREG_RPIs and defer the processing until the commands complete. This
better synchronizes the ELS transmission vs the RPI registrations.
Filter out multiple UNREG_RPIs being queued up for the same remote port.
Beef up log messages in this area.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The switches seem to respond faster to GID_PT vs GID_FT NameServer
queries. Add support for GID_PT to be used over GID_FT to enable
faster storage failover detection. Includes addition of new module
parameter to select between GID_PT and GID_FT (GID_FT is default).
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Currently, PLOGI failures are infinitely delayed/retried. There have
been some fabric situations where the PLOGI's were to the nameserver
and it stopped responding. The retries would never clear up. A better
resolution in this situation is to retry a couple of times, then drop
the link and reinit. This brings back connectivity to the nameserver.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
This patch adds the ability to read firmware logs from the adapter. The driver
registers a buffer with the adapter that is then written to by the adapter.
The adapter posts CQEs to indicate content updates in the buffer. While the
adapter is writing to the buffer in a circular fashion, an application will
poll the driver to read the next amount of log data from the buffer.
Driver log buffer size is configurable via the ras_fwlog_buffsize sysfs
attribute. Verbosity to be used by firmware when logging to host memory is
controlled through the ras_fwlog_level attribute. The ras_fwlog_func
attribute enables or disables loggy by firmware.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Remove the ScsiResult macro and open code it on all call sites.
This will make subsequent refactoring in this area easier.
Signed-off-by: Johannes Thumshirn <jthumshirn@suse.de>
Cc: James Smart <james.smart@broadcom.com>
Cc: Dick Kennedy <dick.kennedy@broadcom.com>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Change references from "Broadcom Limited" to "Broadcom Inc." in the
copyright message. Update copyright duration if not yet updated for 2018.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
To reduce latency when initializing WQE content, created templates for the
most common wqes. This reduces the number of operations taken to set the
content. It's not a lot of speed up, but every bit helps.
This patch updates the NVME target path.
[mkp: fixed typo]
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
To reduce latency when initializing WQE content, create templates for the
most common wqes. This reduces the number of operations taken to set the
content. It's not a lot of speed up, but every bit helps.
This patch updates the NVME initiator path.
[mkp: fixed typo]
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Updated Copyright in files updated 11.4.0.7
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
I/O conditions on the nvme target may have the driver submitting to a
full hardware wq. The hardware wq is a shared resource among all nvme
controllers. When the driver hit a full wq, it failed the io posting
back to the nvme-fc transport, which then escalated it into errors.
Correct by maintaining a sideband queue within the driver that is added
to when the WQ full condition is hit, and drained from as soon as new WQ
space opens up.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
During driver unload, the driver may crash due to NULL pointers. The
NULL pointers were due to the driver not protecting itself sufficiently
during some of the teardown paths. Additionally, the driver was not
waiting for and cleanup up nvme io resources. As such, the driver wasn't
making the callbacks to the transport, stalling the transports
association teardown.
This patch waits for io clean up before tearding down and adds checks
for possible NULL pointers.
Cc: <stable@vger.kernel.org> # 4.12+
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
In preparation for unconditionally passing the struct timer_list pointer to
all timer callbacks, switch to using the new timer_setup() and from_timer()
to pass the timer pointer explicitly.
Cc: James Smart <james.smart@broadcom.com>
Cc: Dick Kennedy <dick.kennedy@broadcom.com>
Cc: "James E.J. Bottomley" <jejb@linux.vnet.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org
Signed-off-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Various oops including cpu LOCKUPs were seen.
For asynchronously received ius where the driver must assign exchange
resources, the resources were on a single get (free) list and put list
(finished, waiting to be put on get list). As all cpus are sharing the
lists, an interrupt for a receive frame may have to wait for all the
other cpus to place their done work onto the put list before it can
acquire the lock to pull from the list.
Fix by breaking the resource lists into per-cpu lists or at least more
than 1 list with cpu's sharing the lists). A cpu would allocate from the
free list for its own cpu, and put its done work on the its own put list
- avoiding the contention. As cpu load may vary, when empty, a cpu may
grab from another cpu, thereby changing resource distribution. But
searching for a resource only occurs on 1 or a few cpus until a single
resource can be allocated. if the condition reoccurs, it starts looking
at a different cpu.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
We might have a NULL pring in lpfc_els_abort(), for example on error
recovery path, since queues are destroyed during error recovery
mechanism.
In this case, we should just drop the abort since the queues will be
recreated anyway. This patch just verifies for NULL pointer and stop the
abortion of the queue in case of a NULL pring.
Also, this patch converts return type of lpfc_els_abort() from int to
void, since it's not checked anywhere.
Reported-by: Harsha Thyagaraja <hathyaga@in.ibm.com>
Reported-by: Naresh Bannoth <nbannoth@in.ibm.com>
Tested-by: Raphael Silva <raphasil@linux.vnet.ibm.com>
Signed-off-by: Guilherme G. Piccoli <gpiccoli@linux.vnet.ibm.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Acked-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Previous logic would just drop the IO.
Added logic to queue the IO to wait for an IO context resource from an
IO thats already in progress.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Currently IO resources are mapped 1 to 1 with RQ buffers posted
Added logic to separate RQE buffers from IO op resources
(sgl/iocbq/context). During initialization, the driver will determine
how many SGLs it will allocate for NVMET (based on what the firmware
reports) and associate a NVMET IOCBq and NVMET context structure with
each one.
Now that hdr/data buffers are immediately reposted back to the RQ, 512
RQEs for each MRQ is sufficient. Also, since NVMET data buffers are now
128 bytes, lpfc_nvmet_mrq_post is not necessary anymore as we will
always post the max (512) buffers per NVMET MRQ.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
James Smart