In many architectures loads may be reordered with older stores to
different locations. In the nvme driver the following two operations
could be reordered:
- Write shadow doorbell (dbbuf_db) into memory.
- Read EventIdx (dbbuf_ei) from memory.
This can result in a potential race condition between driver and VM host
processing requests (if given virtual NVMe controller has a support for
shadow doorbell). If that occurs, then the NVMe controller may decide to
wait for MMIO doorbell from guest operating system, and guest driver may
decide not to issue MMIO doorbell on any of subsequent commands.
This issue is purely timing-dependent one, so there is no easy way to
reproduce it. Currently the easiest known approach is to run "Oracle IO
Numbers" (orion) that is shipped with Oracle DB:
orion -run advanced -num_large 0 -size_small 8 -type rand -simulate \
concat -write 40 -duration 120 -matrix row -testname nvme_test
Where nvme_test is a .lun file that contains a list of NVMe block
devices to run test against. Limiting number of vCPUs assigned to given
VM instance seems to increase chances for this bug to occur. On test
environment with VM that got 4 NVMe drives and 1 vCPU assigned the
virtual NVMe controller hang could be observed within 10-20 minutes.
That correspond to about 400-500k IO operations processed (or about
100GB of IO read/writes).
Orion tool was used as a validation and set to run in a loop for 36
hours (equivalent of pushing 550M IO operations). No issues were
observed. That suggest that the patch fixes the issue.
Fixes: f9f38e3338 ("nvme: improve performance for virtual NVMe devices")
Signed-off-by: Michal Wnukowski <wnukowski@google.com>
Reviewed-by: Keith Busch <keith.busch@intel.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
[hch: updated changelog and comment a bit]
Signed-off-by: Christoph Hellwig <hch@lst.de>
rdma.git merge resolution for the 4.19 merge window
Conflicts:
drivers/infiniband/core/rdma_core.c
- Use the rdma code and revise with the new spelling for
atomic_fetch_add_unless
drivers/nvme/host/rdma.c
- Replace max_sge with max_send_sge in new blk code
drivers/nvme/target/rdma.c
- Use the blk code and revise to use NULL for ib_post_recv when
appropriate
- Replace max_sge with max_recv_sge in new blk code
net/rds/ib_send.c
- Use the net code and revise to use NULL for ib_post_recv when
appropriate
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
-----BEGIN PGP SIGNATURE-----
iQFSBAABCAA8FiEEq68RxlopcLEwq+PEeb4+QwBBGIYFAltwm2geHHRvcnZhbGRz
QGxpbnV4LWZvdW5kYXRpb24ub3JnAAoJEHm+PkMAQRiGITkH/iSzkVhT2OxHoir0
mLVzTi7/Z17L0e/ELl7TvAC0iLFlWZKdlGR0g3b4/QpXLPmNK4HxiDRTQuWn8ke0
qDZyDq89HqLt+mpeFZ43PCd9oqV8CH2xxK3iCWReqv6bNnowGnRpSStlks4rDqWn
zURC/5sUh7TzEG4s997RrrpnyPeQWUlf/Mhtzg2/WvK2btoLWgu5qzjX1uFh3s7u
vaF2NXVJ3X03gPktyxZzwtO1SwLFS1jhwUXWBZ5AnoJ99ywkghQnkqS/2YpekNTm
wFk80/78sU+d91aAqO8kkhHj8VRrd+9SGnZ4mB2aZHwjZjGcics4RRtxukSfOQ+6
L47IdXo=
=sJkt
-----END PGP SIGNATURE-----
Merge tag 'v4.18' into rdma.git for-next
Resolve merge conflicts from the -rc cycle against the rdma.git tree:
Conflicts:
drivers/infiniband/core/uverbs_cmd.c
- New ifs added to ib_uverbs_ex_create_flow in -rc and for-next
- Merge removal of file->ucontext in for-next with new code in -rc
drivers/infiniband/core/uverbs_main.c
- for-next removed code from ib_uverbs_write() that was modified
in for-rc
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
-----BEGIN PGP SIGNATURE-----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=b9ib
-----END PGP SIGNATURE-----
Merge tag 'for-4.19/block-20180812' of git://git.kernel.dk/linux-block
Pull block updates from Jens Axboe:
"First pull request for this merge window, there will also be a
followup request with some stragglers.
This pull request contains:
- Fix for a thundering heard issue in the wbt block code (Anchal
Agarwal)
- A few NVMe pull requests:
* Improved tracepoints (Keith)
* Larger inline data support for RDMA (Steve Wise)
* RDMA setup/teardown fixes (Sagi)
* Effects log suppor for NVMe target (Chaitanya Kulkarni)
* Buffered IO suppor for NVMe target (Chaitanya Kulkarni)
* TP4004 (ANA) support (Christoph)
* Various NVMe fixes
- Block io-latency controller support. Much needed support for
properly containing block devices. (Josef)
- Series improving how we handle sense information on the stack
(Kees)
- Lightnvm fixes and updates/improvements (Mathias/Javier et al)
- Zoned device support for null_blk (Matias)
- AIX partition fixes (Mauricio Faria de Oliveira)
- DIF checksum code made generic (Max Gurtovoy)
- Add support for discard in iostats (Michael Callahan / Tejun)
- Set of updates for BFQ (Paolo)
- Removal of async write support for bsg (Christoph)
- Bio page dirtying and clone fixups (Christoph)
- Set of bcache fix/changes (via Coly)
- Series improving blk-mq queue setup/teardown speed (Ming)
- Series improving merging performance on blk-mq (Ming)
- Lots of other fixes and cleanups from a slew of folks"
* tag 'for-4.19/block-20180812' of git://git.kernel.dk/linux-block: (190 commits)
blkcg: Make blkg_root_lookup() work for queues in bypass mode
bcache: fix error setting writeback_rate through sysfs interface
null_blk: add lock drop/acquire annotation
Blk-throttle: reduce tail io latency when iops limit is enforced
block: paride: pd: mark expected switch fall-throughs
block: Ensure that a request queue is dissociated from the cgroup controller
block: Introduce blk_exit_queue()
blkcg: Introduce blkg_root_lookup()
block: Remove two superfluous #include directives
blk-mq: count the hctx as active before allocating tag
block: bvec_nr_vecs() returns value for wrong slab
bcache: trivial - remove tailing backslash in macro BTREE_FLAG
bcache: make the pr_err statement used for ENOENT only in sysfs_attatch section
bcache: set max writeback rate when I/O request is idle
bcache: add code comments for bset.c
bcache: fix mistaken comments in request.c
bcache: fix mistaken code comments in bcache.h
bcache: add a comment in super.c
bcache: avoid unncessary cache prefetch bch_btree_node_get()
bcache: display rate debug parameters to 0 when writeback is not running
...
When the user supplies a ctrl_loss_tmo < 0, we warn them that this will
cause the fabrics layer to attempt reconnection forever. However, in
reality the fabrics layer never attempts to reconnect because the
condition to test whether we should reconnect is backwards in this case.
Signed-off-by: Tal Shorer <tal.shorer@gmail.com>
Reviewed-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
NVMe 1.3 TP 4005 introduces new filed (NSATTR). This field indicates
whether given namespace is write protected or not. This patch sets the
gendisk associated with the namespace to read only based on the identify
namespace nsattr field.
Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
When the initial discovery fails the subsystem hasn't been setup yet
in nvme_mpath_stop, and we can't dereference ctrl->subsys.
Fixes: 0d0b660f ("nvme: add ANA support")
Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
A minor version number increase should not break backwards
compatibility.
Fixes: 3cb98f84d3 ("lightnvm: add minor version to generic geometry")
Reviewed-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Pull NVMe changes from Christoph:
"This contains the support for TP4004, Asymmetric Namespace Access,
which makes NVMe multipathing usable in practice."
* 'nvme-4.19' of git://git.infradead.org/nvme:
nvmet: use Retain Async Event bit to clear AEN
nvmet: support configuring ANA groups
nvmet: add minimal ANA support
nvmet: track and limit the number of namespaces per subsystem
nvmet: keep a port pointer in nvmet_ctrl
nvme: add ANA support
nvme: remove nvme_req_needs_failover
nvme: simplify the API for getting log pages
nvme.h: add ANA definitions
nvme.h: add support for the log specific field
Signed-off-by: Jens Axboe <axboe@kernel.dk>
-----BEGIN PGP SIGNATURE-----
iQFSBAABCAA8FiEEq68RxlopcLEwq+PEeb4+QwBBGIYFAltU8z0eHHRvcnZhbGRz
QGxpbnV4LWZvdW5kYXRpb24ub3JnAAoJEHm+PkMAQRiG5X8H/2fJr7m3k242+t76
sitwvx1eoPqTgryW59dRKm9IuXAGA+AjauvHzaz1QxomeQa50JghGWefD0eiJfkA
1AphQ/24EOiAbbVk084dAI/C2p122dE4D5Fy7CrfLnuouyrbFaZI5STbnrRct7sR
9deeYW0GDHO1Uenp4WDCj0baaqJqaevZ+7GG09DnWpya2nQtSkGBjqn6GpYmrfOU
mqFuxAX8mEOW6cwK16y/vYtnVjuuMAiZ63/OJ8AQ6d6ArGLwAsdn7f8Fn4I4tEr2
L0d3CRLUyegms4++Dmlu05k64buQu46WlPhjCZc5/Ts4kjrNxBuHejj2/jeSnUSt
vJJlibI=
=42a5
-----END PGP SIGNATURE-----
Merge tag 'v4.18-rc6' into for-4.19/block2
Pull in 4.18-rc6 to get the NVMe core AEN change to avoid a
merge conflict down the line.
Signed-of-by: Jens Axboe <axboe@kernel.dk>
Also moved the logic of the remapping to the nvme core driver instead
of implementing it in the nvme pci driver. This way all the other nvme
transport drivers will benefit from it (in case they'll implement metadata
support).
Suggested-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Acked-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Max Gurtovoy <maxg@mellanox.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Currently this function is implemented in the scsi layer, but it's
actual place should be the block layer since T10-PI is a general
data integrity feature that is used in the nvme protocol as well.
Suggested-by: Christoph Hellwig <hch@lst.de>
Cc: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Max Gurtovoy <maxg@mellanox.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
-----BEGIN PGP SIGNATURE-----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=2ogY
-----END PGP SIGNATURE-----
Merge tag 'for-linus-20180727' of git://git.kernel.dk/linux-block
Pull block fixes from Jens Axboe:
"Bigger than usual at this time, mostly due to the O_DIRECT corruption
issue and the fact that I was on vacation last week. This contains:
- NVMe pull request with two fixes for the FC code, and two target
fixes (Christoph)
- a DIF bio reset iteration fix (Greg Edwards)
- two nbd reply and requeue fixes (Josef)
- SCSI timeout fixup (Keith)
- a small series that fixes an issue with bio_iov_iter_get_pages(),
which ended up causing corruption for larger sized O_DIRECT writes
that ended up racing with buffered writes (Martin Wilck)"
* tag 'for-linus-20180727' of git://git.kernel.dk/linux-block:
block: reset bi_iter.bi_done after splitting bio
block: bio_iov_iter_get_pages: pin more pages for multi-segment IOs
blkdev: __blkdev_direct_IO_simple: fix leak in error case
block: bio_iov_iter_get_pages: fix size of last iovec
nvmet: only check for filebacking on -ENOTBLK
nvmet: fixup crash on NULL device path
scsi: set timed out out mq requests to complete
blk-mq: export setting request completion state
nvme: if_ready checks to fail io to deleting controller
nvmet-fc: fix target sgl list on large transfers
nbd: handle unexpected replies better
nbd: don't requeue the same request twice.
Add support for Asynchronous Namespace Access as specified in NVMe 1.3
TP 4004. With ANA each namespace attached to a controller belongs to an
ANA group that describes the characteristics of accessing the namespaces
through this controller. In the optimized and non-optimized states
namespaces can be accessed regularly, although in a multi-pathing
environment we should always prefer to access a namespace through a
controller where an optimized relationship exists. Namespaces in
Inaccessible, Permanent-Loss or Change state for a given controller
should not be accessed.
The states are updated through reading the ANA log page, which is read
once during controller initialization, whenever the ANA change notice
AEN is received, or when one of the ANA specific status codes that
signal a state change is received on a command.
The ANA state is kept in the nvme_ns structure, which makes the checks in
the fast path very simple. Updating the ANA state when reading the log
page is also very simple, the only downside is that finding the initial
ANA state when scanning for namespaces is a bit cumbersome.
The gendisk for a ns_head is only registered once a live path for it
exists. Without that the kernel would hang during partition scanning.
Includes fixes and improvements from Hannes Reinecke.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Keith Busch <keith.busch@intel.com>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Now that we just call out to blk_path_error there isn't really any good
reason to not merge it into the only caller.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Keith Busch <keith.busch@intel.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Merge nvme_get_log and nvme_get_log_ext into a single helper, which takes
a plain nsid instead of the nvme_ns pointer. Also add support for the
log specific field while we're at it.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Keith Busch <keith.busch@intel.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Instead of declaring and passing a dummy 'bad_wr' pointer, pass NULL
as third argument to ib_post_(send|recv|srq_recv)().
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
We follow the same queue teardown sequence in delete, reset and error
recovery. Centralize the logic. This patch does not change any
functionality.
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Centralize controller sequence to a single routine that correctly cleans
up after failures instead of having multiple apperances in several flows
(create, reset, reconnect).
One thing that we also gain here are the sanity/boundary checks also
when connecting back to a dynamic controller.
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
If the controller is going away, we need to unquiesce the IO queues so
that all pending request can fail gracefully before moving forward with
controller deletion. Do that before we destroy the IO queues so
blk_cleanup_queue won't block in freeze.
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
This will print the disk name to the nvme event trace for io requests so
a user can better distinguish traffic to different disks. This can be used
to create disk based filters. For example, to see only nvme0n2 traffic:
echo "disk == \"nvme0n2\"" > /sys/kernel/debug/tracing/events/nvme/filter
Signed-off-by: Keith Busch <keith.busch@intel.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
[hch: turned __assign_disk_name into an inline function]
Signed-off-by: Christoph Hellwig <hch@lst.de>
This appends the controller instance to the nvme trace buffer to
distinguish which controller is dispatching and completing a command.
Signed-off-by: Keith Busch <keith.busch@intel.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Christoph Hellwig <hch@lst.de>
The revised if_ready checks skipped over the case of returning error when
the controller is being deleted. Instead it was returning BUSY, which
caused the ios to retry, which caused the ns delete to hang waiting for
the ios to drain.
Stack trace of hang looks like:
kworker/u64:2 D 0 74 2 0x80000000
Workqueue: nvme-delete-wq nvme_delete_ctrl_work [nvme_core]
Call Trace:
? __schedule+0x26d/0x820
schedule+0x32/0x80
blk_mq_freeze_queue_wait+0x36/0x80
? remove_wait_queue+0x60/0x60
blk_cleanup_queue+0x72/0x160
nvme_ns_remove+0x106/0x140 [nvme_core]
nvme_remove_namespaces+0x7e/0xa0 [nvme_core]
nvme_delete_ctrl_work+0x4d/0x80 [nvme_core]
process_one_work+0x160/0x350
worker_thread+0x1c3/0x3d0
kthread+0xf5/0x130
? process_one_work+0x350/0x350
? kthread_bind+0x10/0x10
ret_from_fork+0x1f/0x30
Extend nvmf_fail_nonready_command() to supply the controller pointer so
that the controller state can be looked at. Fail any io to a controller
that is deleting.
Fixes: 3bc32bb118 ("nvme-fabrics: refactor queue ready check")
Fixes: 35897b920c ("nvme-fabrics: fix and refine state checks in __nvmf_check_ready")
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Tested-by: Ewan D. Milne <emilne@redhat.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
We can not match a command to its completion based on the command
id alone. We need the submitting queue identifier to pair with the
completion, so this patch adds that to the trace buffer.
This patch is also collapsing the admin and IO submission traces into a
single one so we don't need to duplicate this and creating unnecessary
code branches: we know if the command is an admin vs IO based on the qid.
And since we're here, the patch fixes code formatting in the area.
Signed-off-by: Keith Busch <keith.busch@intel.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
[hch: move the qid helper to nvme.h and made it an inline function]
Signed-off-by: Christoph Hellwig <hch@lst.de>
We will need to reference the controller in the setup and completion
time for tracing and future traffic based keep alive support.
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Allow up to 4 segments of inline data for NVMF WRITE operations. This
reduces latency for small WRITEs by removing the need for the target to
issue a READ WR for IB, or a REG_MR + READ WR chain for iWarp.
Also cap the inline segments used based on the limitations of the
device.
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Currently, the code initializes the keep alive work item whenever
nvme_start_keep_alive() is called. However, this routine is called
several times while reconnecting, etc. Although it's hoped that keep
alive is always disabled and not scheduled when start is called,
re-initing if it were scheduled or completing can have very bad
side effects. There's no need for re-initialization.
Move the keep_alive work item and cmd struct initialization to
controller init.
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
The old code in nvme_user_cmd() passed the userspace virtual address
from nvme_passthru_cmd.metadata as the length of the metadata buffer
as well as the address to nvme_submit_user_cmd().
Fixes: 63263d60 ("nvme: Use metadata for passthrough commands")
Signed-off-by: Roland Dreier <roland@purestorage.com>
Reviewed-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Avoid excuting set_feature command if there is no supported bit in
Optional Asynchronous Events Supported (OAES).
Fixes: c0561f82 ("nvme: submit AEN event configuration on startup")
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Weiping Zhang <zhangweiping@didichuxing.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
For devices that does not specify a limit on its transfer size, the
get_chk_meta command may send down a single I/O retrieving the full
chunk metadata table. Resulting in large 2-4MB I/O requests. Instead,
split up the I/Os to a maximum of 256KB and issue them separately to
reduce memory requirements.
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
Reviewed-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Since both blk_old_get_request() and blk_mq_alloc_request() initialize
rq->__data_len to zero, it is not necessary to initialize that member
in nvme_nvm_alloc_request(). Hence remove the rq->__data_len
initialization from nvme_nvm_alloc_request().
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The nvme driver specific structures need to be initialized prior to
enabling the generic controller so we can unwind on failure with out
using the reference counting callbacks so that 'probe' and 'remove'
can be symmetric.
The newly added iod_mempool is the only resource that was being
allocated out of order, and a failure there would leak the generic
controller memory. This patch just moves that allocation above the
controller initialization.
Fixes: 943e942e62 ("nvme-pci: limit max IO size and segments to avoid high order allocations")
Reported-by: Weiping Zhang <zwp10758@gmail.com>
Signed-off-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
If reconnect/reset failed where the controller async event buffer
was freed, we might end up freeing it again as we call
nvme_rdma_destroy_admin_queue again in the remove path. Given that
the sequence is guaranteed to serialize by .ctrl_stop, we simply
set ctrl->async_event_sqe.data to NULL and don't free it in future
visits.
Reported-by: Max Gurtovoy <maxg@mellanox.com>
Tested-by: Max Gurtovoy <maxg@mellanox.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
nvme requires an sg table allocation for each request. If the request
is large, then the allocation can become quite large. For instance,
with our default software settings of 1280KB IO size, we'll need
10248 bytes of sg table. That turns into a 2nd order allocation,
which we can't always guarantee. If we fail the allocation, blk-mq
will retry it later. But there's no guarantee that we'll EVER be
able to allocate that much contigious memory.
Limit the IO size such that we never need more than a single page
of memory. That's a lot faster and more reliable. Then back that
allocation with a mempool, so that we know we'll always be able
to succeed the allocation at some point.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Acked-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
There is race between nvme_remove and nvme_reset_work that can
lead to io hang.
nvme_remove nvme_reset_work
-> nvme_remove_dead_ctrl
-> nvme_dev_disable
-> quiesce request_queue
-> queue remove_work
-> cancel_work_sync reset_work
-> nvme_remove_namespaces
-> splice ctrl->namespaces
nvme_remove_dead_ctrl_work
-> nvme_kill_queues
-> nvme_ns_remove do nothing
-> blk_cleanup_queue
-> blk_freeze_queue
Finally, the request_queue is quiesced state when wait freeze,
we will get io hang here. To fix it, move the nvme_kill_queues
from nvme_remove_dead_ctrl_work to nvme_remove_dead_ctrl.
Suggested-by: Keith Busch <keith.busch@linux.intel.com>
Signed-off-by: Jianchao Wang <jianchao.w.wang@oracle.com>
Reviewed-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Rather than leaving io queues quiesced after tearing down an association,
restart them. This allows ios to be replayed, with fastfail ios terminating
and non-fastfail getting into loops of retry.
This follows rdma's lead.
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Sagi Grimberg <sagi@grimber.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
That is user argument, and theoretically controller limits can change
over time (over reconnects/resets). Instead, use the sqsize controller
attribute to check queue depth boundaries and use it to the tagset
allocation.
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
The race is between completing the request at error recovery work and
rdma completions. If we cancel the request before getting the good
rdma completion we get a NULL deref of the request MR at
nvme_rdma_process_nvme_rsp().
When Canceling the request we return its mr to the mr pool (set mr to
NULL) and also unmap its data. Canceling the requests while the rdma
queues are active is not safe. Because rdma queues are active and we
get good rdma completions that can use the mr pointer which may be NULL.
Completing the request too soon may lead also to performing DMA to/from
user buffers which might have been already unmapped.
The commit fixes the race by draining the QP before starting the abort
commands mechanism.
Signed-off-by: Israel Rukshin <israelr@mellanox.com>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
If nvme_rdma_configure_admin_queue fails before we allocated
the async event buffer, we will falsly free it because
nvme_rdma_free_queue is freeing it. Fix it by allocating the buffer right
after nvme_rdma_alloc_queue and free it right before nvme_rdma_queue_free
to maintain orderly reverse cleanup sequence.
Reported-by: Israel Rukshin <israelr@mellanox.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Failures after nvme_init_ctrl will defer resource cleanups to .free_ctrl
when the reference is released, hence we should not free the controller
queues for these failures.
Fix that by moving controller queues allocation before controller
initialization and correctly freeing them for failures before
initialization and skip them for failures after initialization.
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Pull NVMe fixes from Christoph:
"Fix various little regressions introduced in this merge window, plus
a rework of the fibre channel connect and reconnect path to share the
code instead of having separate sets of bugs. Last but not least a
trivial trace point addition from Hannes."
* 'nvme-4.18' of git://git.infradead.org/nvme:
nvme-fabrics: fix and refine state checks in __nvmf_check_ready
nvme-fabrics: handle the admin-only case properly in nvmf_check_ready
nvme-fabrics: refactor queue ready check
blk-mq: remove blk_mq_tagset_iter
nvme: remove nvme_reinit_tagset
nvme-fc: fix nulling of queue data on reconnect
nvme-fc: remove reinit_request routine
nvme-fc: change controllers first connect to use reconnect path
nvme: don't rely on the changed namespace list log
nvmet: free smart-log buffer after use
nvme-rdma: fix error flow during mapping request data
nvme: add bio remapping tracepoint
nvme: fix NULL pointer dereference in nvme_init_subsystem
- make sure we only allow internally generates commands in any non-live
state
- only allow connect commands on non-live queues when actually in the
new or connecting states
- treat all other non-live, non-dead states the same as a default
cach-all
This fixes a regression where we could not shutdown a controller
orderly as we didn't allow the internal generated Property Set
command, and also ensures we don't accidentally let a Connect command
through in the wrong state.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: James Smart <james.smart@broadcom.com>
In the ADMIN_ONLY state we don't have any I/O queues, but we should accept
all admin commands without further checks.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: James Smart <james.smart@broadcom.com>
Move the is_connected check to the fibre channel transport, as it has no
meaning for other transports. To facilitate this split out a new
nvmf_fail_nonready_command helper that is called by the transport when
it is asked to handle a command on a queue that is not ready.
Also avoid a function call for the queue live fast path by inlining
the check.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: James Smart <james.smart@broadcom.com>
The reconnect path is calling the init routines to clear a queue
structure. But the queue structure has state that perhaps needs
to persist as long as the controller is live.
Remove the nvme_fc_init_queue() calls on reconnect.
The nvme_fc_free_queue() calls will clear state bits and reset
any relevant queue state for a new connection.
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
The reinit_request routine is not necessary. Remove support for the
op callback.
As all that nvme_reinit_tagset() does is itterate and call the
reinit routine, it too has no purpose. Remove the call.
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Current code follows the framework that has been in the transports
from the beginning where initial link-side controller connect occurs
as part of "creating the controller". Thus that first connect fully
talks to the controller and obtains values that can then be used in
for blk-mq setup, etc. It also means that everything about the
controller is fully know before the "create controller" call returns.
This has several weaknesses:
- The initial create_ctrl call made by the cli will block for a long
time as wire transactions are performed synchronously. This delay
becomes longer if errors occur or connectivity is lost and retries
need to be performed.
- Code wise, it means there is a separate connect path for initial
controller connect vs the (same) steps used in the reconnect path.
- And as there's separate paths, it means there's separate error
handling and retry logic. It also plays havoc with the NEW state
(should transition out of it after successful initial connect) vs
the RESETTING and CONNECTING (reconnect) states that want to be
transitioned to on error.
- As there's separate paths, to recover from errors and disruptions,
it requires separate recovery/retry paths as well and can severely
convolute the controller state.
This patch reworks the fc transport to use the same connect paths
for the initial connection as it uses for reconnect. This makes a
single path for error recovery and handling.
This patch:
- Removes the driving of the initial connect and replaces it with
a state transition to CONNECTING and initiating the reconnect
thread. A dummy state transition of RESETTING had to be traversed
as a direct transtion of NEW->CONNECTING is not allowed. Given
that the controller is "new", the RESETTING transition is a simple
no-op. Once in the reconnecting thread, the normal behaviors of
ctrl_loss_tmo (max_retries * connect_delay) and dev_loss_tmo will
apply before the controller is torn down.
- Only if the state transitions couldn't be traversed and the
reconnect thread not scheduled, will the controller be torn down
while in create_ctrl.
- The prior code used the controller state of NEW to indicate
whether request queues had been initialized or not. For the admin
queue, the request queue is always created, so there's no need to
check a state. For IO queues, change to tracking whether a successful
io request queue create has occurred (e.g. 1st successful connect).
- The initial controller id is initialized to the dynamic controller
id used in the initial connect message. It will be overwritten by
the real controller id once the controller is connected on the wire.
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Michal Wnukowski
Jason Gunthorpe
Linus Torvalds
Tal Shorer
Chaitanya Kulkarni
Hannes Reinecke
Matias Bjørling
Jens Axboe
Max Gurtovoy
Christoph Hellwig
Bart Van Assche
Sagi Grimberg
Gustavo A. R. Silva
Keith Busch
James Smart
Steve Wise
Roland Dreier
Weiping Zhang
Scott Bauer
Jianchao Wang
Israel Rukshin