This patch will create a reply queue pool for each MSI-X index and will
provide an array of base addresses instead of the single address of
legacy mode. Using this new interface the driver can support higher
queue depths through scattered DMA pools.
If array mode is not supported driver will fall back to the legacy
method of reply pool allocation. This limits controller queue depth to
1K max. To enable a queue depth of more than 1K driver requires firmware
to support array mode and scratch_pad3 will provide the new queue depth
value.
When RDPQ is used, downgrading to an older firmware release should not
be permitted. This may cause firmware fault and is not supported.
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Firmware will fill out per-LD data to tell driver whether a particular
LD supports region lock bypass. If yes, then driver will send non-FP
LDIO to region lock bypass FIFO. With this change in driver, firmware
will optimize certain code to improve performance.
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
This patch adds task management for SCSI commands. Added functions are
task abort and target reset.
1. Currently, megaraid_sas driver performs controller reset when any IO
times out. With task management support added, task abort and target
reset will be tried to recover timed out IO. If task management fails,
then controller reset will be performaned. If the task management
request times out, fail the request and escalate to the next
level (controller reset).
2. mr_device_priv_data will be allocated for all generations of
controller, but is_tm_capable flag will never be set for
controllers (prior to Invader series) as firmware support is not
available for task management.
3. Task management capable firmware will set is_tm_capable flag in
firmware API.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Remove PCI id based checks and use instance->ctrl_context to decide
whether controller is MFI-based or a Fusion adapter. Additionally,
Fusion adapters are divided into two categories: Thunderbolt and
Invader.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Driver will expose max sge = 256 (earlier it was 64) if firmware
supports extended IO size (1M).
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Reviewed-by: Martin Petersen <martin.petersen@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Implemented JBOD map which will provide quick access for JBOD path and
also provide sequence number. This will help hardware to fail command
to the FW in case of any sequence mismatch.
Fast Path I/O for JBOD will refer JBOD map (which has sequence number
per JBOD device) instead of RAID map. Previously, the driver used RAID
map to get device handle for fast path I/O and this not have sequence
number information. Now, driver will use JBOD map instead. As part of
error handling, if JBOD map is failed/not supported by firmware, driver
will continue using legacy behavior.
Now there will be three IO paths for JBOD (syspd):
- JBOD map with sequence number (Fast Path)
- RAID map without sequence number (Fast Path)
- FW path via h/w exception queue deliberately setup devhandle
0xFFFF (FW path).
Relevant data structures:
- Driver send new DCMD MR_DCMD_SYSTEM_PD_MAP_GET_INFO for this purpose.
- struct MR_PD_CFG_SEQ- This structure represent map of single physical
device.
- struct MR_PD_CFG_SEQ_NUM_SYNC- This structure represent whole JBOD
map in general(size, count of sysPDs configured, struct MR_PD_CFG_SEQ
of syspD with 0 index).
- JBOD sequence map size is: sizeof(struct MR_PD_CFG_SEQ_NUM_SYNC)
+ (sizeof(struct MR_PD_CFG_SEQ) * (MAX_PHYSICAL_DEVICES - 1)) which
is allocated while setting up JBOD map at driver load time.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Reviewed-by: Martin Petersen <martin.petersen@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
This adds endianness annotations to all data structures, and a few
variables directly referencing them.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: James Bottomley <JBottomley@Odin.com>
megaraid_sas driver will use block layer provided tag for indexing internal
MPT frames to get any unique MPT frame tied with tag. Each IO request
submitted from SCSI mid layer will get associated MPT frame from MPT framepool
(retrieved and return back using spinlock inside megaraid_sas driver's
submission/completion call back). Getting MPT frame from MPT Frame pool is
very expensive operation because of associated spin lock operation (spinlock
overhead increase on multi NUMA node). This type of locking in driver is very
expensive call considering each IO request need - Acquire and Release of the
same lock.
With this support, in IO path driver will directly provide the unique command
index(which is based on block layer tag) and will get the MPT frame tied to
the tag and this way driver can get rid off lock, which synchronizes the
access to MPT frame pool while fetching and returning MPT frame from the pool.
This support in driver provides siginificant performance improvement(on multi
NUMA node system)on latest upstream with SCSI.MQ as well as on existing linux
distributions.
Here is the data for test executed at Avago-
- IO Tool- FIO
- 4 Socket SMC server. (4 NUMA node server)
- 12 SSDs in JBOD mode .
- 4K Rand READ, QD=32
- SCSI MQ x86_64 (Latest Upstream kernel)
- upto 300% Performance Improvement.
If IOs are running on single Node, perfromance gain is less, but as soon as
increase number of nodes, performance improvement is significant. IOs running
on all 4 NUMA nodes, with this patch applied IOPs observed was 1170K vs 344K
IOPs seen without this patch.
Logically, there are two parts of this patch- 1) Block layer tag support 2)
changes in calling convention of return_cmd. part 2 will revert the changes
done by patch- 90dc9d9 megaraid_sas : MFI MPT linked list corruption fix
because changes done in part 1 has fixed the problem of MFI MPT linked list
corruption. part 2 is very much dependent on part 1, so we decided to have
single patch for these two logical changes.
[jejb: remove chatty printk pointed out by hch]
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: James Bottomley <JBottomley@Odin.com>
This patch addresses below issues:
1) Few endianness bug fixes.
2) Break the iteration after (MAX_LOGICAL_DRIVES_EXT - 1)),
instead of MAX_LOGICAL_DRIVES_EXT.
3) Optimization in MFI INIT frame before firing.
4) MFI IO frame should be 256bytes aligned. Code is optimized to reduce
the size of frame for fusion adapters and make the MFI frame size
calculation a bit transparent and readable.
Cc: <stable@vger.kernel.org>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Chaitra Basappa <chaitra.basappa@avagotech.com>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
In OCR (Online Controller Reset) path, driver sets adapter state to
MEGASAS_HBA_OPERATIONAL before getting new RAID map. There will be a small
window where IO will come from OS with old RAID map. This patch will
update adapter state to MEGASAS_HBA_OPERATIONAL, only after driver has new
RAID map to avoid any IOs getting build using old RAID map.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Update MAINTAINERS list and copyright information for megaraid_sas driver.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Resending the patch. Addressed the review comments from Tomas Henzl.
Added comment for to-do work.
Problem statement:
MFI link list in megaraid_sas driver is used from mfi-mpt pass-through commands.
This list can be corrupted due to many possible race conditions in driver and
eventually we may see kernel panic.
One example -
MFI frame is freed from calling process as driver send command via polling method and interrupt
for that command comes after driver free mfi frame (actually even after some other context reuse
the mfi frame). When driver receive MPT frame in ISR, driver will be using the index of MFI and
access that MFI frame and finally in-used MFI frame’s list will be corrupted.
High level description of new solution -
Free MFI and MPT command from same context.
Free both the command either from process (from where mfi-mpt pass-through was called) or from
ISR context. Do not split freeing of MFI and MPT, because it creates the race condition which
will do MFI/MPT list corruption.
Renamed the cmd_pool_lock which is used in instance as well as fusion with below name.
mfi_pool_lock and mpt_pool_lock to add more code readability.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Resending the patch. Addressed the review comments from Tomas Henzl.
Current driver does fast path read load balancing between arm and mirror disk
for two Drive Raid-1 configuration only.
Now, Driver support fast path read load balancing for all (any number of disk) Raid-1 configuration.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Resending the patch. Addressed the review comments from Tomas Henzl.
reserved1 field(part of union) of Raid map struct was not required so it is removed.
Current MegaRAID firmware and hence the driver only supported 64VDs.
E.g: If the user wants to create more than 64VD on a controller,
it is not possible on current firmware/driver.
New feature and requirement to support upto 256VD, firmware/driver/apps need changes.
In addition to that there must be a backward compatibility of the new driver with the
older firmware and vice versa.
RAID map is the interface between Driver and FW to fetch all required
fields(attributes) for each Virtual Drives.
In the earlier design driver was using the FW copy of RAID map where as
in the new design the Driver will keep the RAID map copy of its own; on which
it will operate for any raid map access in fast path.
Local driver raid map copy will provide ease of access through out the code
and provide generic interface for future FW raid map changes.
For the backward compatibility driver will notify FW that it supports 256VD
to the FW in driver capability field.
Based on the controller properly returned by the FW, the Driver will know
whether it supports 256VD or not and will copy the RAID map accordingly.
At any given time, driver will always have old or new Raid map.
So with this changes, driver can also work in host lock less mode. Please
see next patch which enable host lock less mode for megaraid_sas driver.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Resending the patch. Addressed the review comments from Tomas Henzl.
Current driver updates reply post host index to let firmware know that replies are processed,
while returning from ISR function, only if there is no oustanding replies in reply queue.
Driver will free the request frame immediately from ISR but reply post host index is not yet updated.
It means freed request can be used by submission path and there may be a tight loop in request/reply
path. In such condition, firmware may crash when it tries to post reply and there is no free
reply post descriptor.
Eventually two things needs to be change to avoid this issue.
Increase reply queue depth (double than request queue) to accommodate worst case scenario.
Update reply post host index to firmware once it reach to some pre-defined threshold value.
This change will make sure that firmware will always have some buffer of reply descriptor and
will never find empty reply descriptor in completion path.
Signed-off-by: Sumit Saxena <sumit.saxena@avagotech.com>
Signed-off-by: Kashyap Desai <kashyap.desai@avagotech.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
The following patch for megaraid_sas adds Dell PowerEdge VRTS SR-IOV VF
support (Device ID 0x002f).
This patch has some > 80 column lines that need to be left in place
for code readability purposes.
Signed-off-by: Adam Radford <aradford@gmail.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
This patch set is a set of driver updates (megaraid_sas, fnic, lpfc, ufs,
hpsa) we also have a couple of bug fixes (sd out of bounds and ibmvfc error
handling) and the first round of esas2r checker fixes and finally the much
anticipated big endian additions for megaraid_sas.
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.19 (GNU/Linux)
iQEcBAABAgAGBQJSNheiAAoJEDeqqVYsXL0MueMIAKD1kaB0oooRawE1+0vpKmyV
eE2M6trA8ofTeq0z1eNfRsVMkRsUuG9exW0CKS2z6mHiWwQ/zGbqT7ukveW+dMi3
mjKD0yO5ODk6bohWX/LiwZ6NGZSwC0dbIacXNy5ZsXKEizqwo1Jcc7qC/0AWn+o7
WpIL48XLPH0HqjQZ3dvgC6TWeFZOn9cKOWvQQq0S3ENALOx/eLZ+C7VrJLx5Magv
myNOUkTLzdlYglQfjaNO6et98k2oHTrzKwH7U2X6U75q7L8Pkj4RbNzce/Ge301V
u+R1w+BlbeTPdHopTBoTJupsvqDYBZxVwS7rr8nhSvfKduQppHnN6jX8yR4XNeM=
=RG3j
-----END PGP SIGNATURE-----
Merge tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Pull misc SCSI driver updates from James Bottomley:
"This patch set is a set of driver updates (megaraid_sas, fnic, lpfc,
ufs, hpsa) we also have a couple of bug fixes (sd out of bounds and
ibmvfc error handling) and the first round of esas2r checker fixes and
finally the much anticipated big endian additions for megaraid_sas"
* tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi: (47 commits)
[SCSI] fnic: fnic Driver Tuneables Exposed through CLI
[SCSI] fnic: Kernel panic while running sh/nosh with max lun cfg
[SCSI] fnic: Hitting BUG_ON(io_req->abts_done) in fnic_rport_exch_reset
[SCSI] fnic: Remove QUEUE_FULL handling code
[SCSI] fnic: On system with >1.1TB RAM, VIC fails multipath after boot up
[SCSI] fnic: FC stat param seconds_since_last_reset not getting updated
[SCSI] sd: Fix potential out-of-bounds access
[SCSI] lpfc 8.3.42: Update lpfc version to driver version 8.3.42
[SCSI] lpfc 8.3.42: Fixed issue of task management commands having a fixed timeout
[SCSI] lpfc 8.3.42: Fixed inconsistent spin lock usage.
[SCSI] lpfc 8.3.42: Fix driver's abort loop functionality to skip IOs already getting aborted
[SCSI] lpfc 8.3.42: Fixed failure to allocate SCSI buffer on PPC64 platform for SLI4 devices
[SCSI] lpfc 8.3.42: Fix WARN_ON when driver unloads
[SCSI] lpfc 8.3.42: Avoided making pci bar ioremap call during dual-chute WQ/RQ pci bar selection
[SCSI] lpfc 8.3.42: Fixed driver iocbq structure's iocb_flag field running out of space
[SCSI] lpfc 8.3.42: Fix crash on driver load due to cpu affinity logic
[SCSI] lpfc 8.3.42: Fixed logging format of setting driver sysfs attributes hard to interpret
[SCSI] lpfc 8.3.42: Fixed back to back RSCNs discovery failure.
[SCSI] lpfc 8.3.42: Fixed race condition between BSG I/O dispatch and timeout handling
[SCSI] lpfc 8.3.42: Fixed function mode field defined too small for not recognizing dual-chute mode
...
This patch will add big endian architecture support to megaraid_sas
driver. The support added is for LSI MegaRAID all generation controllers-
(3Gb/s, 6Gb/s and 12 Gb/s controllers).
We have done basic sanity test @ppc64 arch and @x86_64. Additional
testing/observations are welcome.
[jejb: fix up rejections]
Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: Sumit Saxena <sumit.saxena@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Fixed several typos in comments in megaraid_mbox.c, megaraid_mm.c
and megaraid_sas_fusion.h.
Signed-off-by: Matthias Schid <aircrach115@gmail.com>
Signed-off-by: Stefan Huber <steffhip@gmail.com>
Signed-off-by: Simon Puels <simon.puels@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
MegaRAID older Firmware does not support uneven span configuration for PRL11.
E.g User wants to create 34 Driver PRL11 config, it was not possible using old
firmware, since it was not supported configuration in old firmware
Old Firmware expect even number of Drives in each span and same number of
physical drives at each span. Considering above design, 17 Drives at Span-0
and 17 drives at span-1 was not possible.
Now, using this new feature Firmware and Driver both required changes. New
Firmware can allow user to create 16 Drives at span-0 and 18 Drives at
span-1. This will allow user to create 34 Drives Uneven span PRL11.
RAID map is interface between Driver and FW to fetch all required
fields(attributes) for each Virtual Drives. Since legacy RAID map consider
Even Span design, there was no place to keep Uneven span information in
existing Raid map. Because of this limitation, for Uneven span VD, driver can
not use RAID map.
This patch address the changes required in Driver to support Uneven span PRL11
support.
1. Driver will find if Firmware has UnevenSpanSupport or not by reading
Controller Info.
2. If Firmware has UnvenSpan PRL11 support, then Driver will inform about its
capability of handling UnevenSpan PRL11 to the firmware.
3. Driver will update its copy of span info on each time Raid map update is
called.
4. Follow different IO path if it is Uneven Span. (For Uneven Span, Driver
uses Span Set info to find relavent fields for that particular Virtual
Disk)
More verbose prints will be available by setting "SPAN_DEBUG" to 1 at
compilation time.
Signed-off-by: Sumit Saxena <sumit.saxena@lsi.com>
Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
This Driver will use more than 8 MSI-x support provided by Invader/Fury max
upto 128 MSI-x.
[jejb: fix checkpatch warning]
Signed-off-by: Sumit Saxena <sumit.saxena@lsi.com>
Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
The following patch for megaraid_sas will allow Fastpath T10PI/DIF
frame builds to work with 4k sector size.
Signed-off-by: Adam Radford <aradford@gmail.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
This patch adds MegaRAID 9265/9285 (Device id 0x5b) specific code
Signed-off-by: Adam Radford <aradford@gmail.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Sumit Saxena