Declaring interrupt clear routines as inline is bogus as they are used as
an indirect pointer.
Remove the inline references.
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>
You can't declare a function inline in a header if it doesn't have a body
available to the compiler. So realistically you either don't declare it
inline or you make it a static inline in the header. I think the latter
applies in this case, so this should be the fix
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Acked-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Change the SLI4 queue creation code to use NUMA node based memory
allocation based on the cpu the queues will be related to.
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>
Currently the driver maintains a sideband structure which has a pointer for
each queue element. However, at 8 bytes per pointer, and up to 4k elements
per queue, and 100s of queues, this can take up a lot of memory.
Convert the driver to using an access routine that calculates the element
address based on its index rather than using the pointer table.
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 in trunking mode, the adapter can be placed into diagnostic mode and
each link in the trunk tested via loopback.
Add support to the driver to perform per-link loopback testing when in
trunking mode.
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>
Driver is using uint16_t and is encountering an overflow of the 16bits when
calculating link speed.
Fix by using a u32 type.
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>
The work done to date utilized the number of present cpus when sizing
per-cpu structures. Structures should have been sized based on the max
possible cpu count.
Convert the driver over to possible cpu count for sizing allocation.
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>
When driving high iop counts, auto_imax coalescing kicks in and drives the
performance to extremely small iops levels.
There are two issues:
1) auto_imax is enabled by default. The auto algorithm, when iops gets
high, divides the iops by the hdwq count and uses that value to
calculate EQ_Delay. The EQ_Delay is set uniformly on all EQs whether
they have load or not. The EQ_delay is only manipulated every 5s (a
long time). Thus there were large 5s swings of no interrupt delay
followed by large/maximum delay, before repeating.
2) When processing a CQ, the driver got mixed up on the rate of when
to ring the doorbell to keep the chip appraised of the eqe or cqe
consumption as well as how how long to sit in the thread and
process queue entries. Currently, the driver capped its work at
64 entries (very small) and exited/rearmed the CQ. Thus, on heavy
loads, additional overheads were taken to exit and re-enter the
interrupt handler. Worse, if in the large/maximum coalescing
windows,k it could be a while before getting back to servicing.
The issues are corrected by the following:
- A change in defaults. Auto_imax is turned OFF and fcp_imax is set
to 0. Thus all interrupts are immediate.
- Cleanup of field names and their meanings. Existing names were
non-intuitive or used for duplicate things.
- Added max_proc_limit field, to control the length of time the
handlers would service completions.
- Reworked EQ handling:
Added common routine that walks eq, applying notify interval and max
processing limits. Use queue_claimed to claim ownership of the queue
while processing. Always rearm the queue whenever the common routine
is called.
Rework queue element processing, namely to eliminate hba_index vs
host_index. Only one index is necessary. The queue entry can be
marked invalid and the host_index updated immediately after eqe
processing.
After rework, xx_release routines are now DB write functions. Renamed
the routines as such.
Moved lpfc_sli4_eq_flush(), which does similar action, to same area.
Replaced the 2 individual loops that walk an eq with a call to the
common routine.
Slightly revised lpfc_sli4_hba_handle_eqe() calling syntax.
Added per-cpu counters to detect interrupt rates and scale
interrupt coalescing values.
- Reworked CQ handling:
Added common routine that walks cq, applying notify interval and max
processing limits. Use queue_claimed to claim ownership of the queue
while processing. Always rearm the queue whenever the common routine
is called.
Rework queue element processing, namely to eliminate hba_index vs
host_index. Only one index is necessary. The queue entry can be
marked invalid and the host_index updated immediately after cqe
processing.
After rework, xx_release routines are now DB write functions. Renamed
the routines as such.
Replaced the 3 individual loops that walk a cq with a call to the
common routine.
Redefined lpfc_sli4_sp_handle_mcqe() to commong handler definition with
queue reference. Add increment for mbox completion to handler.
- Added a new module/sysfs attribute: lpfc_cq_max_proc_limit To allow
dynamic changing of the CQ max_proc_limit value being used.
Although this leaves an EQ as an immediate interrupt, that interrupt will
only occur if a CQ bound to it is in an armed state and has cqe's to
process. By staying in the cq processing routine longer, high loads will
avoid generating more interrupts as they will only rearm as the processing
thread exits. The immediately interrupt is also beneficial to idle or
lower-processing CQ's as they get serviced immediately without being
penalized by sharing an EQ with a more loaded CQ.
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>
Review of the eq coalescing logic showed the code was a bit fragmented.
Sometimes it would save/set via an interrupt max value, while in others it
would do so via a usdelay. There were also two places changing eq delay,
one place that issued mailbox commands, and another that changed via
register writes if supported.
Clean this up by:
- Standardizing the operation of lpfc_modify_hba_eq_delay() routine so
that it is always told of a us delay to impose. The routine then chooses
the best way to set that - via register or via mbx.
- Rather than two value types stored in eq->q_mode (usdelay if change via
register, imax if change via mbox) - q_mode always contains usdelay.
Before any value change, old vs new value is compared and only if
different is a change done.
- Revised the dmult calculation. dmult is not set based on overall imax
divided by hardware queues - instead imax applies to a single cpu and
the value will be replicated to all cpus.
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 desired affinity for the hardware queue behavior is for hdwq 0 to be
affinitized with cpu 0, hdwq 1 to cpu 1, and so on. The implementation so
far does not do this if the number of cpus is greater than the number of
hardware queues (e.g. hardware queue allocation was administratively
reduced or hardware queue resources could not scale to the cpu count).
Correct the queue affinitization logic when queue count is less than
cpu count.
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>
Many io statistics were being sampled and saved using adapter-based data
structures. This was creating a lot of contention and cache thrashing in
the I/O path.
Move the statistics to the hardware queue data structures. Given the
per-queue data structures, use of atomic types is lessened.
Add new sysfs and debugfs stat routines to collate the per hardware queue
values and report at an adapter level.
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>
Similar to the io execution path that reports cpu context information, the
debugfs routines for cpu information needs to be aligned with new hardware
queue implementation.
Convert debugfs cnd nvme cpucheck statistics to report information per
Hardware Queue.
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>
Currently, both nvme and fcp each have their own concept of an io_channel,
which is a combination wq/cq and associated msix. Different cpus would
share an io_channel.
The driver is now moving to per-cpu wq/cq pairs and msix vectors. The
driver will still use separate wq/cq pairs per protocol on each cpu, but
the protocols will share the msix vector.
Given the elimination of the nvme and fcp io channels, the module
parameters will be removed. A new parameter, lpfc_hdw_queue is added which
allows the wq/cq pair allocation per cpu to be overridden and allocated to
lesser value. If lpfc_hdw_queue is zero, the number of pairs allocated will
be based on the number of cpus. If non-zero, the parameter specifies the
number of queues to allocate. At this time, the maximum non-zero value is
64.
To manage this new paradigm, a new hardware queue structure is created to
track queue activity and relationships.
As MSIX vector allocation must be known before setting up the
relationships, msix allocation now occurs before queue datastructures are
allocated. If the number of vectors allocated is less than the desired
hardware queues, the hardware queue counts will be reduced to the number of
vectors
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>
Add trunking support to the driver. Trunking is found on more recent
asics. In general, trunking appears as a single "port" to the driver
and overall behavior doesn't differ. Link speed is reported as an
aggregate value, while link speed control is done on a per-physical
link basis with all links in the trunk symmetrical. Some commands
returning port information are updated to additionally provide
trunking information. And new ACQEs are generated to report physical
link events relative to the trunk.
This patch contains the following modifications:
- Added link speed settings of 128GB and 256GB.
- Added handling of trunk-related ACQEs, mainly logging and trapping
of physical link statuses.
- Added additional bsg interface to query trunk state by applications.
- Augment link_state sysfs attribtute to display trunk link status
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>
On FCoE adapters, when running link bounce test in a loop, initiator
failed to login with switch switch and required driver reload to
recover. Switch reached a point where all subsequent FLOGIs would be
LS_RJT'd. Further testing showed the condition to be related to not
performing FCF discovery between FLOGI's.
Fix by monitoring FLOGI failures and once a repeated error is seen
repeat FCF discovery.
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>
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>
Current implementation missed setting the duration field. Correct the code
to set the field.
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>
Traditional SLI4 required the driver to clear Valid bits on
EQEs and CQEs after consuming them.
The new if_type=6 hardware will cycle the value for what is
valid on each queue itteration. The driver no longer has to
touch the valid bits. This also means all the cpu cache
dirtying and perhaps flush/refill's done by the hardware
in accessing the EQ/CQ elements is eliminated.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
New if_type=6 adapters support an additional BAR that provides
apertures to allow direct WQE to adapter push support - termed
Direct Packet Push (DPP). WQ creation differs slightly to ask for
a WQ to be DPP-ized. When submitting a WQE to a DPP WQ, it is
submitted to the host memory for the WQ normally, but is also
written by the host cpu directly to a BAR aperture. Write buffer
coalescing in hardware is (hopefully) turned on, enabling single
pci write operation support. The doorbell is thing rung to indicate
the WQE is available and was pushed to the aperture.
This patch:
- Updates the WQ Create commands for the DPP options
- Adds the bar mapping for if_type=6 DPP bar
- Adds the WQE pushing to the DDP aperture received from WQ create
- Adds a new module parameter to disable DPP operation if desired.
Default is enabled.
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>
New hardware supports a SLI-4 interface, but with a new if_type
variant of 6.
If_type=6 has a different PCI BAR map, separate EQ/CQ doorbells,
and some changes in doorbell formats.
Add the changes for the if_type into headers, adapter initialization
and control flows. Add new eq and cq handlers.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Up until now, all SLI-4 devices had the same doorbells at the same
bar locations. With newer hardware, there are now independent EQ and
CQ doorbells and the bar locations differ.
Prepare the code for new hardware by separating the eq/cq doorbell into
separate components. The components can be set based on if_type.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Up until now, an SLI-4 device had no variance in the way it handled
its EQs and CQs. With newer hardware, there are now differences in
doorbells and some differences in how entries are valid.
Prepare the code for new hardware by creating a sli4-based callout
table that can be set based on if_type.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
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>
Increased CQ and WQ sizes for SCSI FCP, matching those used for NVMe
development.
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>
Increased the sizes of the SCSI WQ's and CQ's so that SCSI operation is
similar to that used by NVME. However, size increase restricted only to
those newer adapters that can support the larger WQE size, thus bigger
queue sizes.
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>
XRI_ABORTED_CQE completions were not being handled in the fast path.
They were being queued and deferred to the lpfc worker thread for
processing. This is an artifact of the driver design prior to moving
queue processing out of the isr and into a workq element. Now that queue
processing is already in a deferred context, remove this artifact and
process them directly.
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>
Hardware queues are a fast staging area to push commands into the
adapter. The adapter should drain them extremely quickly. However,
under heavy io load, the host cpu is pushing commands faster than the
drain rate of the adapter causing the driver to resource busy commands.
Enlarge the hardware queue (wq & cq) to support a larger number of queue
entries (4x the prior size) before backpressure. Enlarging the queue
requires larger contiguous buffers (16k) per logical page for the
hardware. This changed calling sequences that were expecting 4K page
sizes that now must pass a parameter with the page sizes. It also
required use of a new version of an adapter command that can vary the
page size values.
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>
Under heavy target nvme load duration, the lpfc irq handler is
encountering cpu lockup warnings.
Convert the driver to a shortened ISR handler which identifies the
interrupting condition then schedules a workq thread to process the
completion queue the interrupt was for. This moves all the real work
into the workq element.
As nvmet_fc upcalls are no longer in ISR context, don't set the feature
flags
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Add Buffer to buffer credit recovery support to the driver. This is a
negotiated feature with the peer that allows for both sides to detect
dropped RRDY's and FC Frames and recover credit.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-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>
Various oops being seen on being in the ISR too long and cpu lockups,
when under heavy load.
The amount of work being posted off of completion queues kept the ISR
running almost all the time
Correct the issue by limiting the amount of work per iteration.
[mkp: 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>
Since unsol rcv ISR and command cmpl ISR both access/lock this list,
separate get/put lists will reduce contention.
Replaced
struct list_head lpfc_nvmet_ctx_list;
with
struct list_head lpfc_nvmet_ctx_get_list;
struct list_head lpfc_nvmet_ctx_put_list;
and all correpsonding locks and counters.
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>
Administrator intervention is currently required to get good numbers
when switching from running latency tests to IOPS tests.
The configured interrupt coalescing values will greatly effect the
results of these tests. Currently, the driver has a single coalescing
value set by values of the module attribute. This patch changes the
driver to support auto-configuration of the coalescing value based on
the total number of outstanding IOs and average number of CQEs processed
per interrupt for an EQ. Values are checked every 5 seconds.
The driver defaults to the automatic selection. Automatic selection can
be disabled by the new lpfc_auto_imax module_parameter.
Older hardware can only change interrupt coalescing by mailbox
command. Newer hardware supports change via a register. The patch
support both.
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>
System panic with general protection fault during driver load
The driver uses a static array sli4_hba.handler_name to store the irq
handler names. If the io_channel_irqs exceeds the pre-allocated size
(32+1), then the driver will overwrite other fields of sli4_hba.
Fix: Dynamically allocate handler_name.
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>
Too many work items being processed in IRQ context take a lot of CPU
time and cause problems.
With a recent change, we get out of the ISR after hitting entry_repost
work items on a queue. However, the actual values for entry repost are
still high. EQ is 128 and CQ is 128, this could translate into
processing 128 * 128 (16384) work items under IRQ context.
Set entry_repost in the actual queue creation routine now. Limit EQ
repost to 8 and CQ repost to 64 to further limit the amount of time
spent in the IRQ.
Fix fof IRQ routines as well.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
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>
Large block writes to the nvme target were failing because the default
number of RQs posted was insufficient.
Expand the NVMET RQs to 2048 RQEs and ensure a minimum of 512 RQEs are
posted, no matter how many MRQs are configured.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
More debug messages added for nvme statistics.
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>
With 255 vports created a link trasition can casue a crash.
When going through discovery after a link bounce the driver is using
rpis before the cmd FCOE_POST_HDR_TEMPLATES completes. By doing that the
next rpi bumps the rpi range out of the boundary.
The fix it to increment the next_rpi only when the
FCOE_POST_HDR_TEMPLATE succeeds.
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>
The driver with nvme had this routine stubbed.
Right now XRI_ABORTED_CQE is not handled and the FC NVMET
Transport has a new API for the driver.
Missing code path, new NVME abort API
Update ABORT processing for NVMET
There are 3 new FC NVMET Transport API/ template routines for NVMET:
lpfc_nvmet_xmt_fcp_release
This NVMET template callback routine called to release context
associated with an IO This routine is ALWAYS called last, even
if the IO was aborted or completed in error.
lpfc_nvmet_xmt_fcp_abort
This NVMET template callback routine called to abort an exchange that
has an IO in progress
nvmet_fc_rcv_fcp_req
When the lpfc driver receives an ABTS, this NVME FC transport layer
callback routine is called. For this case there are 2 paths thru the
driver: the driver either has an outstanding exchange / context for the
XRI to be aborted or not. If not, a BA_RJT is issued otherwise a BA_ACC
NVMET Driver abort paths:
There are 2 paths for aborting an IO. The first one is we receive an IO and
decide not to process it because of lack of resources. An unsolicated ABTS
is immediately sent back to the initiator as a response.
lpfc_nvmet_unsol_fcp_buffer
lpfc_nvmet_unsol_issue_abort (XMIT_SEQUENCE_WQE)
The second one is we sent the IO up to the NVMET transport layer to
process, and for some reason the NVME Transport layer decided to abort the
IO before it completes all its phases. For this case there are 2 paths
thru the driver:
the driver either has an outstanding TSEND/TRECEIVE/TRSP WQE or no
outstanding WQEs are present for the exchange / context.
lpfc_nvmet_xmt_fcp_abort
if (LPFC_NVMET_IO_INP)
lpfc_nvmet_sol_fcp_issue_abort (ABORT_WQE)
lpfc_nvmet_sol_fcp_abort_cmp
else
lpfc_nvmet_unsol_fcp_issue_abort
lpfc_nvmet_unsol_issue_abort (XMIT_SEQUENCE_WQE)
lpfc_nvmet_unsol_fcp_abort_cmp
Context flags:
LPFC_NVMET_IOP - his flag signifies an IO is in progress on the exchange.
LPFC_NVMET_XBUSY - this flag indicates the IO completed but the firmware
is still busy with the corresponding exchange. The exchange should not be
reused until after a XRI_ABORTED_CQE is received for that exchange.
LPFC_NVMET_ABORT_OP - this flag signifies an ABORT_WQE was issued on the
exchange.
LPFC_NVMET_CTX_RLS - this flag signifies a context free was requested,
but we are deferring it due to an XBUSY or ABORT in progress.
A ctxlock is added to the context structure that is used whenever these
flags are set/read within the context of an IO.
The LPFC_NVMET_CTX_RLS flag is only set in the defer_relase routine when
the transport has resolved all IO associated with the buffer. The flag is
cleared when the CTX is associated with a new IO.
An exchange can has both an LPFC_NVMET_XBUSY and a LPFC_NVMET_ABORT_OP
condition active simultaneously. Both conditions must complete before the
exchange is freed.
When the abort callback (lpfc_nvmet_xmt_fcp_abort) is envoked:
If there is an outstanding IO, the driver will issue an ABORT_WQE. This
should result in 3 completions for the exchange:
1) IO cmpl with XB bit set
2) Abort WQE cmpl
3) XRI_ABORTED_CQE cmpl
For this scenerio, after completion #1, the NVMET Transport IO rsp
callback is called. After completion #2, no action is taken with respect
to the exchange / context. After completion #3, the exchange context is
free for re-use on another IO.
If there is no outstanding activity on the exchange, the driver will send a
ABTS to the Initiator. Upon completion of this WQE, the exchange / context
is freed for re-use on another IO.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
previous code did little more than log a message.
This patch adds abort path support, modeled after the SCSI code paths.
Currently addresses only the initiator path. Target path under
development, but stubbed out.
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>
James Smart