Commit Graph

28 Commits

Author SHA1 Message Date
Ramachandra Rao Gajula fa08a39664 NVMe: Add MSI support
Some devices only have support for MSI, not MSI-X.  While MSI is more
limited, it still provides better performance than line-based interrupts.

Signed-off-by: Ramachandra Gajula <rama@fastorsystems.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-31 11:45:52 -04:00
Matthew Wilcox cf9f123b38 NVMe: Use dma_set_mask() correctly
In some circumstances setting a 64-bit DMA mask can fail, as explained
in Documentation/DMA-API-HOWTO.txt.  Use the recommended code sequence
to set a 32-bit DMA mask if setting a 64-bit DMA mask fails.

Reported-by: Chayan Biswas <Chayan.Biswas@sandisk.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-28 16:46:46 -04:00
Chayan Biswas cf90bc4830 Return the result from user admin command IOCTL even in case of failure
We copy the result to user if the command is completed from the
controller even if it completes with failure (non-zero) status.
A return status of < 0 indicates the command was not completed
by the controller. The user application may expect the error code
in the result field in case of failure.

Signed-off-by: Chayan Biswas <Chayan.Biswas@sandisk.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-23 13:38:59 -04:00
Keith Busch 053ab702cc NVMe: Do not cancel command multiple times
Cancelling an already cancelled command does not do anything, so check
the command context before cancelling it, continuing if had already been
cancelled so we do not log the same problem every second if a device
stops responding.

Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-17 09:18:38 -04:00
Wei Yongjun 1287dabd34 NVMe: fix error return code in nvme_submit_bio_queue()
nvme_submit_flush_data() might overwrite the initialisation of the
return value with 0, so move the -ENOMEM setting close to the usage.

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-17 09:13:18 -04:00
Dan Carpenter 5460fc0310 NVMe: check for integer overflow in nvme_map_user_pages()
You need to have CAP_SYS_ADMIN to trigger this overflow but it makes the
static checkers complain so we should fix it.  The worry is that
"length" comes from copy_from_user() so we need to check that "length +
offset" can't overflow.

I also changed the min_t() cast to be unsigned instead of signed.  Now
that we cap "length" to INT_MAX it doesn't make a difference, but it's a
little easier for reviewers to know that large values aren't cast to
negative.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-17 09:11:03 -04:00
Keith Busch 94f370cab6 NVMe: Use user defined admin ioctl timeout
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-09 16:03:50 -04:00
Matthew Wilcox 44af146a84 NVMe: Only clear the enable bit when disabling controller
Many of the bits in the Controller Configuration register may only be
modified when the Enable bit is clear.  Clearing them at the same time
as the Enable bit might be OK, but let's play it safe and only touch the
Enable bit.

Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
Reviewed-by: Keith Busch <keith.busch@intel.com>
2013-05-08 09:54:31 -04:00
Matthew Wilcox ba47e3865e NVMe: Wait for device to acknowledge shutdown
A recent update to the specification makes it clear that the host
is expected to wait for the device to acknowledge the Enable bit
transitioning to 0 as well as waiting for the device to acknowledge a
transition to 1.

Reported-by: Khosrow Panah <Khosrow.Panah@idt.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
Reviewed-by: Keith Busch <keith.busch@intel.com>
2013-05-08 09:53:49 -04:00
Keith Busch 78f8d2577b NVMe: Schedule timeout for sync commands
Schedule a timeout on sync commands in case the command times out and
the device is not being polled for timeouts. This prevents device removal
from hanging forever if the device has stopped responding.

Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-02 15:36:02 -04:00
Keith Busch f410c680b5 NVMe: Meta-data support in NVME_IOCTL_SUBMIT_IO
This adds support for namespaces with separate meta-data formats in the
submit io ioctl. The meta-data buffer has to be a contiguous, so such
a buffer is allocated and the mapped user pages are copied to/from this
buffer for write/read commands.

Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-02 15:35:09 -04:00
Keith Busch 159b67d7ae NVMe: Device specific stripe size handling
We have an nvme device that has a concept of a stripe size. IO requests
that do not transfer data crossing a stripe boundary has greater
performance compared to IO that does cross it. This patch sets the
stripe size for the device if the device and vendor ids match one with
this feature and splits IO requests that cross the stripe boundary.

Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-02 14:41:05 -04:00
Keith Busch 427e970801 NVMe: Split non-mergeable bio requests
It is possible a bio request can not be submitted as a single NVMe IO
command if the bio_vec is not mergeable with the NVMe PRP alignement
constraints. This condition was handled by submitting an IO for the
mergeable portion then submitting a follow on IO for the remaining data
after the previous IO completes. The remainder to be sent was tracked
by manipulating the bio->bi_idx and bio->bi_sector. This patch splits
the request as many times as necessary and submits the bios together.

Since submitting the bio may cause it to be requeued on split,
nvme_resubmit_bios had to be modified to remove the wait queue when
the bio list is empty prior to submitting the bio since a split would
have added the wait queue a second time, corrupting the wait queue head
task list.

There are a few other benefits from doing this: it fixes a potential
issue with the previous handling of a non-mergeable bio as the requeuing
method could would use an unlocked nvme_queue if the callback isn't
invoked on the queue's associated cpu; it will be possible to retry a
failed bio if desired at some later time since it does not manipulate
the original bio; the bio integrity extensions require the bio to be in
its original condition for the checks to work correctly if we implement
the end-to-end data protection in the future.

Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-02 14:38:59 -04:00
Keith Busch cbb6218fd4 NVMe: Remove dead code in nvme_dev_add
There is no situation that could occur where we could error out of this
function and require cleaning up allocated namespaces.

Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-02 14:36:45 -04:00
Keith Busch a9ef4343af NVMe: Check for NULL memory in nvme_dev_add
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-02 14:35:44 -04:00
Keith Busch 68b8eca5f8 NVMe: Fix error clean-up on nvme_alloc_queue
The nvme_queue's depth is not set if we fail to allocate submission queue
entries, which was being used to determine how much coherent memory to
free on error. Use the depth variable instead.

Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-02 14:34:35 -04:00
Keith Busch 025c557a71 NVMe: Free admin queue on request_irq error
Fixes a potential memory leak if requesting the admin queue irq fails.

Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-02 14:33:53 -04:00
Arjan van de Ven 564a232c05 NVMe: Set TASK_INTERRUPTIBLE before processing queues
The kthread has two tasks; handling timeouts (for which it runs once per
second), and submitting queued BIOs.  If a BIO happens to be queued after
the thread has processed the queue but before it calls schedule_timeout(),
the thread will sleep for a second before submitting it, which can cause
performance problems in some rare cases (that will become more common in
a subsequent patch).

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Tested-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-05-01 16:40:27 -04:00
Keith Busch 5e82e952f0 NVMe: Add a character device for each nvme device
Registers a miscellaneous device for each nvme controller probed. This
creates character device files as /dev/nvmeN, where N is the device
instance, and supports nvme admin ioctl commands so devices without
namespaces can be managed.

Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-04-16 15:43:55 -04:00
Matthew Wilcox 1c9b52651d NVMe: Fix endian-related problems in user I/O submission path
When constructing the command, dsmgmt needs to be treated as a 32-bit
value, not a 16-bit value.  reftag, apptag and appmask all need to be
converted from native-endian to little-endian.  Again, sparse's bitwise
warnings caught this problem.  Thanks to Keith for pointing out the
correct way to fix the reftag.

Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
Acked-by: Keith Busch <keith.busch@intel.com>
2013-04-16 15:21:06 -04:00
Matthew Wilcox af2d9ca744 NVMe: Fix I/O cancellation status on big-endian machines
The sparse bitwise checks pointed out that I needed to shift the status
before changing its endianness, not after.

Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-04-16 15:18:30 -04:00
Matthew Wilcox 422ef0c7c8 NVMe: Don't fail initialisation unnecessarily
The nvme_dev_add() function currently returns the last error code that it
saw, which (if everything else succeeds) happens to be the result of an
optional command, so it can legitimately fail.  Looking at the error path
more closely reveals that we should return success from this function,
even if no device namespaces are added.  So once the queues are created
and the device has responded to Identify, make sure that this function
succeeds.

Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
Acked-by: Keith Busch <keith.busch@intel.com>
2013-04-16 15:13:41 -04:00
Matthew Wilcox 063cc6d559 NVMe: Abstract out sector to block number conversion
Introduce nvme_block_nr() to help convert sectors to block numbers.
This fixes an integer overflow in the SCSI conversion layer, and it's
slightly less typing than opencoding it.

Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
Acked-by: Keith Busch <keith.busch@intel.com>
2013-04-16 15:05:22 -04:00
Arjan van de Ven acb7aa0db0 NVMe: Use round_jiffies_relative() for the periodic, once-per-second timer
The nvme driver has a "once per second" event where the management kthread
wakes up the system and then reschedules itself for 1 second later.
For power efficiency reasons, I'd like this timer to happen together
with other wakeups in the system.

This patch makes the schedule_timeout() call in the kthread use
round_jiffies_relative(), causing the wakeup to at least align with other
"once per X seconds" events in the kernel.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Tested-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-04-16 15:05:16 -04:00
Vishal Verma 5d0f6131a7 NVMe: Add nvme-scsi.c
Translates SCSI commands in SG_IO ioctl to NVMe commands.
Uses the scsi-nvme translation spec from nvmexpress.org as reference.

Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-03-28 14:50:49 -04:00
Vishal Verma f8ebf8409a NVMe: Add definitions for format command
The SCSI emulation has the ability to send format commands, so we need
to add the definition of the command.  Also add a missing error code.

Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-03-27 07:17:49 -04:00
Vishal Verma 13c3b0fcc8 NVMe: Move structures & definitions to header file
nvme-scsi.c uses several data structures and definitions that were
previously private to nvme-core.c.  Move the definitions to nvme.h,
protected by __KERNEL__.

Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-03-27 07:05:42 -04:00
Vishal Verma 729dd1bd80 NVMe: Rename nvme.c to nvme-core.c
In preparation for adding nvme-scsi.c
It is preferable to retain the module name 'nvme'

Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
2013-03-26 14:24:56 -04:00