linux.git/drivers/nvme/target, branch v5.5

nvmet: fix per feat data len for get_feature

2020-01-10T15:55:50+00:00

The existing implementation for the get_feature admin-cmd does not
use per-feature data len. This patch introduces a new helper function
nvmet_feat_data_len(), which is used to calculate per feature data len.
Right now we only set data len for fid 0x81 (NVME_FEAT_HOST_ID).

Fixes: commit e9061c397839 ("nvmet: Remove the data_len field from the nvmet_req struct")

Reviewed-by: Christoph Hellwig 
Signed-off-by: Amit Engel 
[endiness, naming, and kernel style fixes]
Signed-off-by: Chaitanya Kulkarni 
Signed-off-by: Keith Busch 
Signed-off-by: Jens Axboe

nvme_fc: add module to ops template to allow module references

2019-11-26T17:48:27+00:00

In nvme-fc: it's possible to have connected active controllers
and as no references are taken on the LLDD, the LLDD can be
unloaded.  The controller would enter a reconnect state and as
long as the LLDD resumed within the reconnect timeout, the
controller would resume.  But if a namespace on the controller
is the root device, allowing the driver to unload can be problematic.
To reload the driver, it may require new io to the boot device,
and as it's no longer connected we get into a catch-22 that
eventually fails, and the system locks up.

Fix this issue by taking a module reference for every connected
controller (which is what the core layer did to the transport
module). Reference is cleared when the controller is removed.

Acked-by: Himanshu Madhani 
Reviewed-by: Christoph Hellwig 
Signed-off-by: James Smart 
Signed-off-by: Keith Busch

nvmet-loop: Avoid preallocating big SGL for data

2019-11-26T17:14:19+00:00

nvme_loop_create_io_queues() preallocates a big buffer for the IO SGL based
on SG_CHUNK_SIZE.

Modern DMA engines are often capable of dealing with very big segments so
the SG_CHUNK_SIZE is often too big. SG_CHUNK_SIZE results in a static 4KB
SGL allocation per command.

If a controller has lots of deep queues, preallocation for the sg list can
consume substantial amounts of memory. For nvmet-loop, nr_hw_queues can be
128 and each queue's depth 128. This means the resulting preallocation
for the data SGL is 128*128*4K = 64MB per controller.

Switch to runtime allocation for SGL for lists longer than 2 entries. This
is the approach used by NVMe PCI so it should be reasonable for NVMeOF as
well. Runtime SGL allocation has always been the case for the legacy I/O
path so this is nothing new.

Tested-by: Chaitanya Kulkarni 
Reviewed-by: Christoph Hellwig 
Reviewed-by: Chaitanya Kulkarni 
Reviewed-by: Max Gurtovoy 
Signed-off-by: Israel Rukshin 
Signed-off-by: Keith Busch

Merge tag 'for-5.5/drivers-20191121' of git://git.kernel.dk/linux-block

2019-11-25T19:15:41+00:00

Pull block driver updates from Jens Axboe:
 "Here are the main block driver updates for 5.5. Nothing major in here,
  mostly just fixes. This contains:

   - a set of bcache changes via Coly

   - MD changes from Song

   - loop unmap write-zeroes fix (Darrick)

   - spelling fixes (Geert)

   - zoned additions cleanups to null_blk/dm (Ajay)

   - allow null_blk online submit queue changes (Bart)

   - NVMe changes via Keith, nothing major here either"

* tag 'for-5.5/drivers-20191121' of git://git.kernel.dk/linux-block: (56 commits)
  Revert "bcache: fix fifo index swapping condition in journal_pin_cmp()"
  drivers/md/raid5-ppl.c: use the new spelling of RWH_WRITE_LIFE_NOT_SET
  drivers/md/raid5.c: use the new spelling of RWH_WRITE_LIFE_NOT_SET
  bcache: don't export symbols
  bcache: remove the extra cflags for request.o
  bcache: at least try to shrink 1 node in bch_mca_scan()
  bcache: add idle_max_writeback_rate sysfs interface
  bcache: add code comments in bch_btree_leaf_dirty()
  bcache: fix deadlock in bcache_allocator
  bcache: add code comment bch_keylist_pop() and bch_keylist_pop_front()
  bcache: deleted code comments for dead code in bch_data_insert_keys()
  bcache: add more accurate error messages in read_super()
  bcache: fix static checker warning in bcache_device_free()
  bcache: fix a lost wake-up problem caused by mca_cannibalize_lock
  bcache: fix fifo index swapping condition in journal_pin_cmp()
  md/raid10: prevent access of uninitialized resync_pages offset
  md: avoid invalid memory access for array sb->dev_roles
  md/raid1: avoid soft lockup under high load
  null_blk: add zone open, close, and finish support
  dm: add zone open, close and finish support
  ...

nvmet: stop using bio_set_op_attrs

2019-11-04T17:56:42+00:00

bio_set_op_attrs has been long deprecated, replace it with a direct
assignment of the flags to bio->bi_opf.

Reviewed-by: Chaitanya Kulkarni 
Reviewed-by: Johannes Thumshirn 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Keith Busch 
Signed-off-by: Jens Axboe

nvmet: add plugging for read/write when ns is bdev

2019-11-04T17:56:42+00:00

With reference to the following issue reported on the mailing list :-
http://lists.infradead.org/pipermail/linux-nvme/2019-October/027604.html
this patch adds plugging for the bdev-ns under nvmet_bdev_execute_rw().

We can see the following performance improvement in random write
workload I/Os with the setup described in the link when device_path
configured as /dev/md0.

Without this patch :-

  write: IOPS=40.8k, BW=159MiB/s (167MB/s)(4777MiB/30002msec)
  write: IOPS=41.2k, BW=161MiB/s (169MB/s)(4831MiB/30011msec)
    slat (usec): min=8,  max=10823, avg=15.64,  stdev=16.85
    slat (usec): min=8,  max=401,   avg=15.40,  stdev= 9.56
    clat (usec): min=54, max=2492,  avg=759.07, stdev=172.62
    clat (usec): min=56, max=1997,  avg=768.06, stdev=178.72

With this patch :-

  write: IOPS=123k, BW=480MiB/s (504MB/s)(14.1GiB/30011msec)
  write: IOPS=123k, BW=481MiB/s (504MB/s)(14.1GiB/30002msec)
    slat (usec): min=8,  max=9941,  avg=13.31,  stdev= 8.04
    slat (usec): min=8,  max=289,   avg=13.31,  stdev= 3.37
    clat (usec): min=43, max=17635, avg=245.46, stdev=171.23
    clat (usec): min=44, max=17751, avg=245.25, stdev=183.14

Reviewed-by: Sagi Grimberg 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Chaitanya Kulkarni 
Signed-off-by: Keith Busch 
Signed-off-by: Jens Axboe

nvmet: clean up command parsing a bit

2019-11-04T17:56:42+00:00

Move the special cases for fabrics commands and the discovery controller
to nvmet_parse_admin_cmd in preparation for adding passthrough support.

Reviewed-by: Sagi Grimberg 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Keith Busch 
Signed-off-by: Jens Axboe

nvmet: fill discovery controller sn, fr and mn correctly

2019-11-04T17:56:42+00:00

Discovery controllers need this information as well.

Signed-off-by: Sagi Grimberg 
Signed-off-by: Keith Busch 
Signed-off-by: Jens Axboe

nvmet: Open code nvmet_req_execute()

2019-11-04T17:56:42+00:00

Now that nvmet_req_execute does nothing, open code it.

Reviewed-by: Chaitanya Kulkarni 
Signed-off-by: Christoph Hellwig 
[split patch, update changelog]
Signed-off-by: Logan Gunthorpe 
Signed-off-by: Keith Busch 
Signed-off-by: Jens Axboe

nvmet: Remove the data_len field from the nvmet_req struct

2019-11-04T17:56:42+00:00

Instead of storing the expected length and checking it when it's
executed, just check the length inside the command themselves.

A new helper, nvmet_check_data_len() is created to help with this
check.

Signed-off-by: Christoph Hellwig 
[split patch, udpate changelog]
Signed-off-by: Logan Gunthorpe 
Signed-off-by: Keith Busch 
Signed-off-by: Jens Axboe