IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
Some UFS devices may expose bLUQueueDepth field as zero indicating
that the queue depth depends on the number of resources available
for LUN at a particular instant to handle the outstanding transfer
requests. Currently, when response for SCSI command is TASK_FULL
the LLD decrements the queue depth but fails to increment when the
resources are available. The scsi mid-layer handles the change in
queue depth heuristically and offers simple interface with
->change_queue_depth.
Signed-off-by: Sujit Reddy Thumma <sthumma@codeaurora.org>
Signed-off-by: Dolev Raviv <draviv@codeaurora.org>
Acked-by: Santosh Y <santoshsy@gmail.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Some of the UFS devices may support different number of commands
that can be queued per LU. At the current implementation,
SW configure each of the UFS devices LU's according to the
controller capability.
In this patch the queue depth available per LU is read and updated in
the LU's SW structure.
Signed-off-by: Dolev Raviv <draviv@codeaurora.org>
Signed-off-by: Raviv Shvili <rshvili@codeaurora.org>
Acked-by: Santosh Y <santoshsy@gmail.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Check query response status before copying the response.
Add descriptor query response size check, before copying it to buffer.
Signed-off-by: Dolev Raviv <draviv@codeaurora.org>
Signed-off-by: Raviv Shvili <rshvili@codeaurora.org>
Acked-by: Santosh Y <santoshsy@gmail.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Introduces the API for sending queries with descriptors.
A descriptor is a block or page of parameters that describe the device.
The descriptors are classified into types and can range in size
from 2 bytes through 255 bytes.
All descriptors have a length value as their first element, and a type
identification element as their second byte.
All descriptors are readable and some may be write once.
They are accessed using their type, index and selector.
Signed-off-by: Dolev Raviv <draviv@codeaurora.org>
Signed-off-by: Raviv Shvili <rshvili@codeaurora.org>
Acked-by: Santosh Y <santoshsy@gmail.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
The if_info pointer is not released by the mgmt_set_ip() function
Signed-off-by: Maurizio Lombardi <mlombard@redhat.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
change_queue_depth allows changing per-target queue depth via sysfs.
It also allows the SCSI midlayer to ramp down the number of concurrent
inflight requests in response to a SCSI BUSY status response and allows
the midlayer to ramp the count back up to the device maximum when the
BUSY condition has resolved.
Signed-off-by: Venkatesh Srinivas <venkateshs@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
The spinlock of tgt_lock is only for serializing read and write
req_vq, one lockless seqcount is enough for the purpose.
On one 16core VM with vhost-scsi backend, the patch can improve
IOPS with 3% on random read test.
Signed-off-by: Ming Lei <ming.lei@canonical.com>
[Add initialization in virtscsi_target_alloc. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
When copy_from_user fails, return -EFAULT, not -ENOMEM
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Reported-by: Robert Elliott <elliott@hp.com>
Reviewed-by: Joe Handzik <joseph.t.handzik@hp.com>
Reviewed-by: Scott Teel <scott.teel@hp.com>
Reviewed by: Mike MIller <michael.miller@canonical.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>
When devices come on line, they should be removed from the list of
offline devices that are monitored.
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Reviewed-by: Scott Teel <scott.teel@hp.com>
Reviewed-by: Joe Handzik <joseph.t.handzik@hp.com>
Reviewed by: Mike MIller <michael.miller@canonical.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
commit 28e1344647 "[SCSI] hpsa: enable unit attention reporting"
turns on unit attention notifications, but got the change wrong for
all architectures other than x86, which now store an uninitialized
value into the device register.
Gcc helpfully warns about this:
../drivers/scsi/hpsa.c: In function 'hpsa_set_driver_support_bits':
../drivers/scsi/hpsa.c:6373:17: warning: 'driver_support' is used uninitialized in this function [-Wuninitialized]
driver_support |= ENABLE_UNIT_ATTN;
^
This moves the #ifdef so only the prefetch-enable is conditional
on x86, not also reading the initial register contents.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Fixes: 28e1344647 "[SCSI] hpsa: enable unit attention reporting"
Cc: stable@vger.kernel.org # v3.14+
Acked-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Robert Elliott <elliott@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Reviewed-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
a 6-byte READ/WRITE CDB with a 0 block data transfer really
means a 256 block data transfer. The RAID mapping code failed
to handle this case. For 10/12/16 byte READ/WRITEs, 0 just means
no data should be transferred, and should not trigger BUG_ON.
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Reported-by: Robert Elliott <elliott@hp.com>
Reviewed-by: Robert Elliott <elliott@hp.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Reviewed-by: Joe Handzik <joseph.t.handzik@hp.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
QLogic has acquired the NetXtremeII products and drivers from Broadcom.
This patch re-brands bnx2fc driver as a QLogic driver
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Acked-by: Eddie Wai <eddie.wai@broadcom.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
QLogic has acquired the NetXtremeII products and drivers from Broadcom.
This patch re-brands bnx2i driver as a QLogic driver
Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Acked-by: Eddie Wai <eddie.wai@broadcom.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
On systems with a non power-of-two CPU count the existing MSI-X grouping
code failed to distribute interrupts correctly. Rework the code to
handle arbitrary processor counts.
Also remove the hardcoded upper limit on the number of processors so we
can boot on large systems.
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Acked-by: Sreekanth Reddy <Sreekanth.reddy@avagotech.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
On systems with a non power-of-two CPU count the existing MSI-X grouping
code failed to distribute interrupts correctly. Rework the code to
handle arbitrary processor counts.
Also remove the hardcoded upper limit on the number of processors so we
can boot on large systems.
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Acked-by: Sreekanth Reddy <Sreekanth.reddy@avagotech.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Tack the firmware reply event_data payload to the end of its
corresponding struct fw_event_work allocation. This matches the
convention in the mptfusion driver and simplifies the code.
This avoids the following smatch warning:
drivers/scsi/mpt3sas/mpt3sas_scsih.c:2519
mpt3sas_send_trigger_data_event() warn: possible memory leak of
'fw_event'
Signed-off-by: Joe Lawrence <joe.lawrence@stratus.com>
Acked-by: Sreekanth Reddy <Sreekanth.Reddy@avagotech.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
In _scsih_{slave,target}_alloc, an incorrect structure type is passed
to sizeof() when allocating storage for hostdata. Luckily larger
structure types were used, so at least the wrong sizes were safe:
struct scsi_device (1784 bytes) > struct MPT3SAS_DEVICE (24 bytes)
struct scsi_target (760 bytes) > struct MPT3SAS_TARGET (32 bytes)
This fixes the following smatch warnings:
drivers/scsi/mpt3sas/mpt3sas_scsih.c:1166 _scsih_target_alloc()
warn: struct type mismatch 'MPT3SAS_TARGET vs scsi_target'
drivers/scsi/mpt3sas/mpt3sas_scsih.c:1280 _scsih_slave_alloc()
warn: struct type mismatch 'MPT3SAS_DEVICE vs scsi_device'
Signed-off-by: Joe Lawrence <joe.lawrence@stratus.com>
Acked-by: Sreekanth Reddy <Sreekanth.Reddy@avagotech.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
The MPT2SAS_ADAPTER reply_post_host_index[] holds calculated addresses
in memory mapped register space. Add an "__iomem" annotation to silence
the following sparse warnings:
drivers/scsi/mpt2sas/mpt2sas_base.c:1006:43:
warning: incorrect type in argument 2 (different address spaces)
expected void volatile [noderef] <asn:2>*addr
got unsigned long long [usertype] *<noident>
drivers/scsi/mpt2sas/mpt2sas_base.c:4299:22:
warning: cast removes address space of expression
drivers/scsi/mpt2sas/mpt2sas_base.c:4303:27:
warning: cast removes address space of expression
Signed-off-by: Joe Lawrence <joe.lawrence@stratus.com>
Acked-by: Sreekanth Reddy <Sreekanth.Reddy@avagotech.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Tack the firmware reply event_data payload to the end of its
corresponding struct fw_event_work allocation. This matches the
convention in the mptfusion driver and simplifies the code.
Signed-off-by: Joe Lawrence <joe.lawrence@stratus.com>
Acked-by: Sreekanth Reddy <Sreekanth.Reddy@avagotech.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
In _scsih_{slave,target}_alloc, an incorrect structure type is passed
to sizeof() when allocating storage for hostdata. Luckily larger
structure types were used, so at least the wrong sizes were safe:
struct scsi_device (1784 bytes) > struct MPT2SAS_DEVICE (24 bytes)
struct scsi_target (760 bytes) > struct MPT2SAS_TARGET (40 bytes)
This fixes the following smatch warnings:
drivers/scsi/mpt2sas/mpt2sas_scsih.c:1295 _scsih_target_alloc()
warn: struct type mismatch 'MPT2SAS_TARGET vs scsi_target'
drivers/scsi/mpt2sas/mpt2sas_scsih.c:1409 _scsih_slave_alloc()
warn: struct type mismatch 'MPT2SAS_DEVICE vs scsi_device'
Signed-off-by: Joe Lawrence <joe.lawrence@stratus.com>
Acked-by: Sreekanth Reddy <Sreekanth.Reddy@avagotech.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Building an allmodconfig ARM kernel, I get multiple such
warnings because of a spinlock contained in packed structure
in the 3w-xxxx driver:
../drivers/scsi/3w-xxxx.c: In function 'tw_chrdev_ioctl':
../drivers/scsi/3w-xxxx.c:1001:68: warning: mis-aligned access used for structure member [-fstrict-volatile-bitfields]
timeout = wait_event_timeout(tw_dev->ioctl_wqueue, tw_dev->chrdev_request_id == TW_IOCTL_CHRDEV_FREE, timeout);
^
../drivers/scsi/3w-xxxx.c:1001:68: note: when a volatile object spans multiple type-sized locations, the compiler must choose between using a single mis-aligned access to preserve the volatility, or using multiple aligned accesses to avoid runtime faults; this code may fail at runtime if the hardware does not allow this access
The same bug apparently was present in 3w-sas and 3w-9xxx, but has been
fixed in the past. This patch uses the same fix by moving the pragma
in front of the TW_Device_Extension definition, so it only covers
hardware structures.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Adam Radford <aradford@gmail.com>
Cc: Adam Radford <linuxraid@lsi.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
The NCR53c406a scsi driver normally does not use DMA, unless
the USE_PIO macro is disabled by modifying the source code.
The call to free_dma() for some reason uses #ifdef USE_DMA,
which does not do the right thing, since USE_DMA is defined
as a boolean that is either 0 or 1, but always present.
One case where it gets in the way is randconfig builds on ARM,
which depending on the configuration does not provide a free_dma()
function, causing this build error:
drivers/scsi/NCR53c406a.c: In function 'NCR53c406a_release':
drivers/scsi/NCR53c406a.c:600:3: error: implicit declaration of function 'free_dma' [-Werror=implicit-function-declaration]
free_dma(shost->dma_channel);
^
This changes the code to use #if USE_DMA, to match the
rest of the file, which seems to be what the author intended.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Christoph Hellwig <hch@lst.de>
The qlogicfas scsi driver does not use DMA, and the call to free_dma()
in its exit function seems to have been copied incorrectly from
another driver but never caused trouble.
One case where it gets in the way is randconfig builds on ARM,
which depending on the configuration does not provide a free_dma()
function, causing this build error:
drivers/scsi/qlogicfas.c: In function 'qlogicfas_release':
drivers/scsi/qlogicfas.c:175:3: error: implicit declaration of function 'free_dma' [-Werror=implicit-function-declaration]
free_dma(shost->dma_channel);
^
Removing the incorrect function calls should be the obvious
fix for this, with no downsides.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Christoph Hellwig <hch@lst.de>
The pas16 scsi driver does not use DMA, and the call to free_dma()
in its exit function seems to have been copied incorrectly from
another driver but never caused trouble.
One case where it gets in the way is randconfig builds on ARM,
which depending on the configuration does not provide a free_dma()
function, causing this build error:
drivers/scsi/pas16.c: In function 'pas16_release':
drivers/scsi/pas16.c:611:3: error: implicit declaration of function 'free_dma' [-Werror=implicit-function-declaration]
free_dma(shost->dma_channel);
Removing the incorrect function calls should be the obvious
fix for this, with no downsides.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Christoph Hellwig <hch@lst.de>
The advansys SCSI driver uses the dma_cache_sync function, which is
not available on the ARM architecture, and cannot be implemented
correctly, so we always get this build error:
drivers/scsi/advansys.c: In function 'advansys_get_sense_buffer_dma':
drivers/scsi/advansys.c:7882:2: error: implicit declaration of function 'dma_cache_sync' [-Werror=implicit-function-declaration]
dma_cache_sync(board->dev, scp->sense_buffer,
^
It seems nobody has missed this driver so far, so let's just
disable it for ARM to help randconfig builds.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Added big endian annotations to relevant data structure fields, and necessary
byte swappings to support little endian builds.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: Tyrel Datwyler <tyreld@linux.vnet.ibm.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Use the zeroing function instead of dma_alloc_coherent & memset(,0,)
Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Use macro definition
Signed-off-by: Fabian Frederick <fabf@skynet.be>
Acked-by: Anil Gurumurthy <anil.gurumurthy@qlogic.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
This patch remove variables that are initialized with a constant,
are never updated, and are only used as parameter of return.
Return the constant instead of using a variable.
Verified by compilation only.
The coccinelle script that find and fixes this issue is:
// <smpl>
@@
type T;
constant C;
identifier ret;
@@
- T ret = C;
... when != ret
when strict
return
- ret
+ C
;
// </smpl>
Signed-off-by: Peter Senna Tschudin <peter.senna@gmail.com>
Acked-by: Sudarsana Kalluru <Sudarsana.Kalluru@qlogic.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
bfa_swap_words() shifts its argument (assumed to be 64-bit) by 32 bits
each way. In two places the argument type is dma_addr_t, which may be
32-bit, in which case the effect of the bit shift is undefined:
drivers/scsi/bfa/bfa_fcpim.c: In function 'bfa_ioim_send_ioreq':
drivers/scsi/bfa/bfa_fcpim.c:2497:4: warning: left shift count >= width of type [enabled by default]
addr = bfa_sgaddr_le(sg_dma_address(sg));
^
drivers/scsi/bfa/bfa_fcpim.c:2497:4: warning: right shift count >= width of type [enabled by default]
drivers/scsi/bfa/bfa_fcpim.c:2509:4: warning: left shift count >= width of type [enabled by default]
addr = bfa_sgaddr_le(sg_dma_address(sg));
^
drivers/scsi/bfa/bfa_fcpim.c:2509:4: warning: right shift count >= width of type [enabled by default]
Avoid this by adding casts to u64 in bfa_swap_words().
Compile-tested only.
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Acked-by: Anil Gurumurthy <anil.gurumurthy@qlogic.com>
Cc: stable@vger.kernel.org
Fixes: f16a17507b ('[SCSI] bfa: remove all OS wrappers')
Signed-off-by: Christoph Hellwig <hch@lst.de>
Use the zeroing function instead of dma_alloc_coherent & memset(,0,)
Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Anil Gurumurthy <anil.gurumurthy@qlogic.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Use kstrdup when the goal of an allocation is copy a string into the
allocated region.
The Coccinelle semantic patch that makes this change is as follows:
// <smpl>
@@
expression from,to;
expression flag,E1,E2;
statement S;
@@
- to = kmalloc(strlen(from) + 1,flag);
+ to = kstrdup(from, flag);
... when != \(from = E1 \| to = E1 \)
if (to==NULL || ...) S
... when != \(from = E2 \| to = E2 \)
- strcpy(to, from);
// </smpl>
Signed-off-by: Himangi Saraogi <himangi774@gmail.com>
Acked-by: Julia Lawall <julia.lawall@lip6.fr>
Signed-off-by: Christoph Hellwig <hch@lst.de>
These speeds are to support the next generation of FCoE port speeds.
Signed-off-by: Dick Kennedy <Dick.Kennedy@Emulex.Com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
The custom stats is an array with custom_length indicating the length
of the array. This patch fixes bnx2i and be2iscsi's setting of the
custom stats length. They both just have the one, eh_abort_cnt, so that should
be in the first entry of the custom array and custom_length should then
be one.
Reported-by: Rickard Strandqvist <rickard_strandqvist@spectrumdigital.se>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Acked-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Acked-by: Eddie Wai <eddie.wai@broadcom.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Remove two redundant casts from char * to char *.
Signed-off-by: Nick Black <nlb@google.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Christoph Hellwig <hch@lst.de>
The Kconfig symbols SCSI_TGT and SCSI_FC_TGT_ATTRS are unused since
"tgt: removal". Setting them has no effect. Remove these symbols.
Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Commit ID: 7e660100d8 added code to derive the
FLUSH_TIMEOUT from the basic I/O timeout. However, this patch did not use the
basic I/O timeout of the device. Fix this bug.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Despite supporting modern SCSI features some storage devices continue to
claim conformance to an older version of the SPC spec. This is done for
compatibility with legacy operating systems.
Linux by default will not attempt to read VPD pages on devices that
claim SPC-2 or older. Introduce a blacklist flag that can be used to
trigger VPD page inquiries on devices that are known to support them.
Reported-by: KY Srinivasan <kys@microsoft.com>
Tested-by: KY Srinivasan <kys@microsoft.com>
Reviewed-by: KY Srinivasan <kys@microsoft.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
CC: <stable@vger.kernel.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>
We currently set the field in common code based on the device type,
but then only use it in the cdrom driver which also overrides the
value previously set in the generic code.
Just leave this entirely to the CDROM driver to make everyones life
simpler.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Add two new device types, most importantly the zoned block device
one.
Split from an earlier patch by Hannes Reinecke.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Current the midlayer fakes up a struct request for the explicit reset
ioctls, and those don't have a tag allocated to them. The fnic driver pokes
into midlayer structures to paper over this design issue, but that won't
work for the blk-mq case.
Either someone who can actually test the hardware will have to come up with
a similar hack for the blk-mq case, or we'll have to bite the bullet and fix
the way the EH ioctls work for real, but until that happens we fail these
explicit requests here.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Webb Scales <webbnh@hp.com>
Acked-by: Jens Axboe <axboe@kernel.dk>
Tested-by: Bart Van Assche <bvanassche@acm.org>
Tested-by: Robert Elliott <elliott@hp.com>
Cc: Hiral Patel <hiralpat@cisco.com>
Cc: Suma Ramars <sramars@cisco.com>
Cc: Brian Uchino <buchino@cisco.com>
This patch adds support for an alternate I/O path in the scsi midlayer
which uses the blk-mq infrastructure instead of the legacy request code.
Use of blk-mq is fully transparent to drivers, although for now a host
template field is provided to opt out of blk-mq usage in case any unforseen
incompatibilities arise.
In general replacing the legacy request code with blk-mq is a simple and
mostly mechanical transformation. The biggest exception is the new code
that deals with the fact the I/O submissions in blk-mq must happen from
process context, which slightly complicates the I/O completion handler.
The second biggest differences is that blk-mq is build around the concept
of preallocated requests that also include driver specific data, which
in SCSI context means the scsi_cmnd structure. This completely avoids
dynamic memory allocations for the fast path through I/O submission.
Due the preallocated requests the MQ code path exclusively uses the
host-wide shared tag allocator instead of a per-LUN one. This only
affects drivers actually using the block layer provided tag allocator
instead of their own. Unlike the old path blk-mq always provides a tag,
although drivers don't have to use it.
For now the blk-mq path is disable by defauly and must be enabled using
the "use_blk_mq" module parameter. Once the remaining work in the block
layer to make blk-mq more suitable for slow devices is complete I hope
to make it the default and eventually even remove the old code path.
Based on the earlier scsi-mq prototype by Nicholas Bellinger.
Thanks to Bart Van Assche and Robert Elliot for testing, benchmarking and
various sugestions and code contributions.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Webb Scales <webbnh@hp.com>
Acked-by: Jens Axboe <axboe@kernel.dk>
Tested-by: Bart Van Assche <bvanassche@acm.org>
Tested-by: Robert Elliott <elliott@hp.com>
Blk-mq drivers usually preallocate their S/G list as part of the request,
but if we want to support the very large S/G lists currently supported by
the SCSI code that would tie up a lot of memory in the preallocated request
pool. Add support to the scatterlist code so that it can initialize a
S/G list that uses a preallocated first chunks and dynamically allocated
additional chunks. That way the scsi-mq code can preallocate a first
page worth of S/G entries as part of the request, and dynamically extend
the S/G list when needed.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Webb Scales <webbnh@hp.com>
Acked-by: Jens Axboe <axboe@kernel.dk>
Tested-by: Bart Van Assche <bvanassche@acm.org>
Tested-by: Robert Elliott <elliott@hp.com>
Replace the calls to the various blk_end_request variants with opencode
equivalents. Blk-mq is using a model that gives the driver control
between the bio updates and the actual completion, and making the old
code follow that same model allows us to keep the code more similar for
both paths.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Webb Scales <webbnh@hp.com>
Acked-by: Jens Axboe <axboe@kernel.dk>
Tested-by: Bart Van Assche <bvanassche@acm.org>
Tested-by: Robert Elliott <elliott@hp.com>
This saves us an atomic operation for each I/O submission and completion
for the usual case where the driver doesn't set a per-target can_queue
value. Only a few iscsi hardware offload drivers set the per-target
can_queue value at the moment.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Webb Scales <webbnh@hp.com>
Acked-by: Jens Axboe <axboe@kernel.dk>
Tested-by: Bart Van Assche <bvanassche@acm.org>
Tested-by: Robert Elliott <elliott@hp.com>
Seems like these counters are missing any sort of synchronization for
updates, as a over 10 year old comment from me noted. Fix this by
using atomic counters, and while we're at it also make sure they are
in the same cacheline as the _busy counters and not needlessly stored
to in every I/O completion.
With the new model the _busy counters can temporarily go negative,
so all the readers are updated to check for > 0 values. Longer
term every successful I/O completion will reset the counters to zero,
so the temporarily negative values will not cause any harm.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Webb Scales <webbnh@hp.com>
Acked-by: Jens Axboe <axboe@kernel.dk>
Tested-by: Bart Van Assche <bvanassche@acm.org>
Tested-by: Robert Elliott <elliott@hp.com>
Avoid taking the queue_lock to check the per-device queue limit. Instead
we do an atomic_inc_return early on to grab our slot in the queue,
and if necessary decrement it after finishing all checks.
Unlike the host and target busy counters this doesn't allow us to avoid the
queue_lock in the request_fn due to the way the interface works, but it'll
allow us to prepare for using the blk-mq code, which doesn't use the
queue_lock at all, and it at least avoids a queue_lock round trip in
scsi_device_unbusy, which is still important given how busy the queue_lock
is.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Webb Scales <webbnh@hp.com>
Acked-by: Jens Axboe <axboe@kernel.dk>
Tested-by: Bart Van Assche <bvanassche@acm.org>
Tested-by: Robert Elliott <elliott@hp.com>
Avoid taking the host-wide host_lock to check the per-host queue limit.
Instead we do an atomic_inc_return early on to grab our slot in the queue,
and if necessary decrement it after finishing all checks.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Webb Scales <webbnh@hp.com>
Acked-by: Jens Axboe <axboe@kernel.dk>
Tested-by: Bart Van Assche <bvanassche@acm.org>
Tested-by: Robert Elliott <elliott@hp.com>
