A test scenario encountered an unload hang while an FLOGI ELS was in flight
when a link down condition occurred. The driver fails unload as it never
releases the fport node.
For most nodes, when the link drops, devloss tmo is started and the timeout
will cause the final node release. For the Fport, as it has not yet
registered with the SCSI transport, there is no devloss timer to be
started, so there is no final release. Additionally, the link down
sequence causes ABORTS to be issued for pending ELS's. The completions from
the ABORTS perform the release of node references. However, as the adapter
is being reset to be unloaded, those completions will never occur.
Fix by the following:
- In the ELS cleanup, recognize when unloading and place the ELS's on a
different list that immediately cleans up/completes the ELS's. It's
recognized that this condition primarily affects only the fport, with
other ports having normal clean up logic that handles things.
- Resolve the devloss issue by, when cleaning up nodes on after link down,
recognizing when the fabric node does not have a completed state (its
state is UNUSED) and removing a reference so the node can delete after
the ELS reference is released.
Link: https://lore.kernel.org/r/20210910233159.115896-5-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
A test scenario has a target issuing a TPLS after accepting the driver's
PRLI. TPLS is not supported by the driver so it rejects the ELS. However,
the reject was only happening on the primary N_Port. If the TPLS was to a
NPIV vport, not only would it reject the ELS, but it would act on the TPLS,
starting devloss, then unregister from the SCSI transport and release the
node. When devloss expired, it would access the node again and cause a page
faul.
Fix by altering the NPIV code to recognize that a correctly registered node
can reject unsolicited ELS I/O and to not unregister with the SCSI
transport and tear the node down. Add a check of the fc4_xpt_flags so that
only a zero value allows the unreg and teardown.
Link: https://lore.kernel.org/r/20210910233159.115896-4-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
In a rarely executed path, FLOGI failure, there is a refcounting error. If
FLOGI completed with an error, typically a timeout, the initial completion
handler would remove the job reference. However, the job completion isn't
the actual end of the job/exchange as the timeout usually initiates an
ABTS, and upon that ABTS completion, a final completion is sent. The driver
removes the reference again in the final completion. Thus the imbalance.
In the buggy cases, if there was a link bounce while the delayed response
is outstanding, the fport node may be referenced again but there was no
additional reference as it is already present. The delayed completion then
occurs and removes the last reference freeing the node and causing issues
in the link up processed that is using the node.
Fix this scenario by removing the snippet that removed the reference in the
initial FLOGI completion. The bad snippet was poorly trying to identify the
FLOGI as OK to do so by realizing the node was not registered with either
SCSI or NVMe transport.
Link: https://lore.kernel.org/r/20210910233159.115896-3-jsmart2021@gmail.com
Fixes: 618e2ee146 ("scsi: lpfc: Fix FLOGI failure due to accessing a freed node")
Cc: <stable@vger.kernel.org> # v5.13+
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
When parsing the txq list in lpfc_drain_txq(), the driver attempts to pass
the requests to the adapter. If such an attempt fails, a local "fail_msg"
string is set and a log message output. The job is then added to a
completions list for cancellation.
Processing of any further jobs from the txq list continues, but since
"fail_msg" remains set, jobs are added to the completions list regardless
of whether a wqe was passed to the adapter. If successfully added to
txcmplq, jobs are added to both lists resulting in list corruption.
Fix by clearing the fail_msg string after adding a job to the completions
list. This stops the subsequent jobs from being added to the completions
list unless they had an appropriate failure.
Link: https://lore.kernel.org/r/20210910233159.115896-2-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The kernel test robot reported the following sparse warning:
".../lpfc_els.c:3984:25: sparse: sparse: cast from restricted __be16"
For the error being flagged, using be32_to_cpu() on a be16 data type, it
was simple enough. But a review of other elements and warnings were also
evaluated.
This patch corrected several items in the original patch:
- Using be32_to_cpu() on a be16 data type
- cpu_to_le32() used on a std uint32_t (CPU) data type.
Note: This is a byte array, but stored in LE layout by hardware at
32-bit boundaries. So it possibly needed conversion.
- Using cpu_to_le32() on a std uint16_t and assigned to a char typeA
- Using le32_to_cpu() on a le16 type
- Missing cpu_to_le16() on an assignment
Link: https://lore.kernel.org/r/20210830231243.6227-1-jsmart2021@gmail.com
Fixes: 9064aeb2df ("scsi: lpfc: Add EDC ELS support")
Reported-by: kernel test robot <lkp@intel.com>
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Pull SCSI updates from James Bottomley:
"This series consists of the usual driver updates (ufs, qla2xxx,
target, smartpqi, lpfc, mpt3sas).
The core change causing the most churn was replacing the command
request field request with a macro, allowing us to offset map to it
and remove the redundant field; the same was also done for the tag
field.
The most impactful change is the final removal of scsi_ioctl, which
has been deprecated for over a decade"
* tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi: (293 commits)
scsi: ufs: Fix ufshcd_request_sense_async() for Samsung KLUFG8RHDA-B2D1
scsi: ufs: ufs-exynos: Fix static checker warning
scsi: mpt3sas: Use the proper SCSI midlayer interfaces for PI
scsi: lpfc: Use the proper SCSI midlayer interfaces for PI
scsi: lpfc: Copyright updates for 14.0.0.1 patches
scsi: lpfc: Update lpfc version to 14.0.0.1
scsi: lpfc: Add bsg support for retrieving adapter cmf data
scsi: lpfc: Add cmf_info sysfs entry
scsi: lpfc: Add debugfs support for cm framework buffers
scsi: lpfc: Add support for maintaining the cm statistics buffer
scsi: lpfc: Add rx monitoring statistics
scsi: lpfc: Add support for the CM framework
scsi: lpfc: Add cmfsync WQE support
scsi: lpfc: Add support for cm enablement buffer
scsi: lpfc: Add cm statistics buffer support
scsi: lpfc: Add EDC ELS support
scsi: lpfc: Expand FPIN and RDF receive logging
scsi: lpfc: Add MIB feature enablement support
scsi: lpfc: Add SET_HOST_DATA mbox cmd to pass date/time info to firmware
scsi: fc: Add EDC ELS definition
...
Complete the enablement of the cm framework feature in the adapter. Perform
the following:
- Detect the presence of the congestion management framework feature.
When the cm framework is present:
- Issue the SET_FEATURE command to enable the feature.
- Register the cm statistics buffer with the adapter.
- Read the cm enablement buffer to determine the cm framework state for cm
management.
When cm management is enabled:
- Monitor all FPIN and congestion signalling events, incrementing
counters.
- Regularly sync with the adapter to communicate congestion events and to
receive an rx request limit.
- Monitor requests for rx data and ensure that no more than the
adapter prescribed limit is issued on the link. If the limit is
exceeded, SCSI and/or NVMe traffic is temporarily suspended.
- Maintain the minute, hourly, daily statistics buffer.
- Monitor for congestion enablement change events, causing a reread of the
enablement buffer and acting on any change in enablement.
And:
- Add teardown logic, including buffer deregistration, on adapter
detachment or reset.
Link: https://lore.kernel.org/r/20210816162901.121235-10-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
When congestion mgmt is enabled, cmf has the driver regularly issue a
command to synchronize reporting of congestion mgmt events such as fpin and
signal delivery.
This patch adds the definition of the CMF_SYNC WQE and its CQE fields as
well as support for issuing the command. The patch also adds the few
remaining cmf-related SLI additions, such as feature definition for
enablement of CMF and notifications to the driver if the cm enablement mode
changes.
Link: https://lore.kernel.org/r/20210816162901.121235-9-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
As part of the cmf framework, the firmware maintains a table with
congestion related state information, specifically whether enabled and if
enabled, whether monitoring or actively managing congestion.
Add definition of the table and add support to read the table from the
adapter and determine if it is enabled. In support of this, the READ_OBJECT
mailbox command definition is added to the driver.
Link: https://lore.kernel.org/r/20210816162901.121235-8-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The cmf framework requires the driver to maintain a cm statistics table,
accessible inband, of congestion related statistics that are reported per
minute, rolled up to per hour, and rolled up again per day. Several days
worth may be maintained. The table is registered with the adapter when the
MIB feature is enabled.
Add definition of the table and add support to register the table with the
adapter. Includes definition and initialization of event counters that are
later added to the statistics table.
Link: https://lore.kernel.org/r/20210816162901.121235-7-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
When congestion management is enabled, issue EDC ELS to register congestion
signaling capabilities with the fabric. The response handling will process
the fabric parameters and set the reporting parameters.
Similarly, add support for receiving an EDC request from the fabric
generating a corresponding response.
Implement handlers for congestion signals from the fabric and maintain
statistics for them.
Link: https://lore.kernel.org/r/20210816162901.121235-6-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
MIB support is currently limited to detecting support in the adapter and
ensuring FDMI support is enabled if present. For the new framework MIB
support also requires active enablement of support via the SET_FEATURES
command with the firmware.
Rework the MIB detection and enablement for the following:
- Move detection away from the get_sli4_parameters routine, and into the
hba_setup path. get_sli4_parameters is only called once at attachment
while hba_setup is called as part of any SLI port reset path. This
ensures detection after firmware download.
- Update SET_FEATURES mbx command for the MIB enablement feature and add
support for the feature.
- Create the cmf_setup routine to encapsulate the detection of MIB support
and perform the enablement of the MIB support feature.
Link: https://lore.kernel.org/r/20210816162901.121235-4-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The lpfc_sli4_nvmet_xri_aborted() routine takes out the abts_buf_list_lock
and traverses the buffer contexts to match the xri. Upon match, it then
takes the context lock before potentially removing the context from the
associated buffer list. This violates the lock hierarchy used elsewhere in
the driver of locking context, then the abts_buf_list_lock - thus a
possible deadlock.
Resolve by: after matching, release the abts_buf_list_lock, then take the
context lock, and if to be deleted from the list, retake the
abts_buf_list_lock, maintaining lock hierarchy. This matches same list lock
hierarchy as elsewhere in the driver
Link: https://lore.kernel.org/r/20210730163309.25809-1-jsmart2021@gmail.com
Reported-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
When a node moves to NPR state due to a device recovery event, the
nlp_fc4_types in the node are cleared. An ADISC received for a node in the
NPR state triggers an ADISC. Without fc4 types being known, the calls to
register with the transport are no-op'd, thus no additional references are
placed on the node by transport re-registrations. A subsequent RSCN could
trigger another unregister request, which will decrement the reference
counts, leading to the ref count hitting zero and the node being freed
while futher discovery on the node is being attempted by the RSCN event
handling.
Fix by skipping the trigger of an ADISC when in NPR state. The normal ADISC
process will kick off in the regular discovery path after receiving a
response from name server.
Link: https://lore.kernel.org/r/20210707184351.67872-19-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
During RSCN storms, some instances of LIP on SLI-3 adapters lead to a
situation where FLOGIs keep failing with firmware indicating an illegal
command error code. This situation was preceded by an ADISC completion
that was processed while the link was down. This path on SLI-3 performs a
CLEAR_LA and attempts to activate a VPI with REG_VPI. Later, as the FLOGI
completes, it's no longer in sync with the VPI state. In SLI-3 it is
illegal to have an active VPI during FLOGI.
Resolve by circumventing the SLI-3 path that performs the CLEAR_LA and
REG_VPI. The path will be taken after the FLOGI after the next Link Up.
Link: https://lore.kernel.org/r/20210707184351.67872-18-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
In the PLOGI and ADISC completion handling, the device removal event could
be skipped during some link errors. This could leave a stale node in UNUSED
state. Driver unload would hang for a long time waiting for this node to
be freed.
Resolve by taking the following steps:
- Always post ADISC completion events to discovery state machine upon
ADISC completion.
- In case of a completion error for PLOGI/ADISC, ensure that init refcount
is dropped if not registered with transport.
Link: https://lore.kernel.org/r/20210707184351.67872-17-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
On an RSCN event, the nodes specified in RSCN payload and in MAPPED state
are moved to NPR state in order to revalidate the login. This triggers an
immediate unregister from SCSI/NVMe backend. The assumption is that the
node may be missing. The re-registration with the backend happens after
either relogin (PLOGI/PRLI; if ADISC is disabled or login truly lost) or
when ADISC completes successfully (rediscover with ADISC enabled).
However, the NVMe-FC standard provides for an RSCN to be triggered when
the remote port supports a discovery controller and there was a change
of discovery log content. As the remote port typically also supports
storage subsystems, this unregister causes all storage controller
connections to fail and require reconnect.
Correct by reworking the code to ensure that the unregistration only occurs
when a login state is truly terminated, thereby leaving the NVMe storage
controllers in place.
The changes made are:
- Retain node state in ADISC_ISSUE when scheduling ADISC ELS retry.
- Do not clear wwpn/wwnn values upon ADISC failure.
- Move MAPPED nodes to NPR during RSCN processing, but do not unregister
with transport. On GIDFT completion, identify missing nodes (not marked
NLP_NPR_2B_DISC) and unregister them.
- Perform unregistration for nodes that will go through ADISC processing
if ADISC completion fails.
- Successful ADISC completion will move node back to MAPPED state.
Link: https://lore.kernel.org/r/20210707184351.67872-16-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Mailbox commands sent via ioctl/bsg from user applications may be
interrupted from processing by a concurrently triggered PCI function
reset. The command will not generate a completion due to the reset. This
results in a user application hang waiting for the mailbox command to
complete.
Resolve by changing the function reset handler to detect that there was an
outstanding mailbox command and simulate a mailbox completion. Add some
additional debug when a mailbox command times out.
Link: https://lore.kernel.org/r/20210707184351.67872-13-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
If a LOGO is received for a node that is in the NPR state, post a DEVICE_RM
event to allow clean up of the logged out node.
Clearing the NLP_NPR_2B_DISC flag upon receipt of a LOGO ACC may cause
skipping of processing outstanding PLOGIs triggered by parallel RSCN
events. If an outstanding PLOGI is being retried and receipt of a LOGO ACC
occurs, then allow the discovery state machine's PLOGI completion to clean
up the node.
Link: https://lore.kernel.org/r/20210707184351.67872-6-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
