linux-stable.git/drivers/nvme/host, branch v5.4.78

nvme-tcp: avoid repeated request completion

2020-11-18T18:20:24+00:00

[ Upstream commit 0a8a2c85b83589a5c10bc5564b796836bf4b4984 ]

The request may be executed asynchronously, and rq->state may be
changed to IDLE. To avoid repeated request completion, only
MQ_RQ_COMPLETE of rq->state is checked in nvme_tcp_complete_timed_out.
It is not safe, so need adding check IDLE for rq->state.

Signed-off-by: Sagi Grimberg 
Signed-off-by: Chao Leng 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin

nvme-rdma: avoid repeated request completion

2020-11-18T18:20:24+00:00

[ Upstream commit fdf58e02adecbef4c7cbb2073d8ea225e6fd5f26 ]

The request may be executed asynchronously, and rq->state may be
changed to IDLE. To avoid repeated request completion, only
MQ_RQ_COMPLETE of rq->state is checked in nvme_rdma_complete_timed_out.
It is not safe, so need adding check IDLE for rq->state.

Signed-off-by: Sagi Grimberg 
Signed-off-by: Chao Leng 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin

nvme-tcp: avoid race between time out and tear down

2020-11-18T18:20:24+00:00

[ Upstream commit d6f66210f4b1aa2f5944f0e34e0f8db44f499f92 ]

Now use teardown_lock to serialize for time out and tear down. This may
cause abnormal: first cancel all request in tear down, then time out may
complete the request again, but the request may already be freed or
restarted.

To avoid race between time out and tear down, in tear down process,
first we quiesce the queue, and then delete the timer and cancel
the time out work for the queue. At the same time we need to delete
teardown_lock.

Signed-off-by: Chao Leng 
Reviewed-by: Sagi Grimberg 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin

nvme-rdma: avoid race between time out and tear down

2020-11-18T18:20:24+00:00

[ Upstream commit 3017013dcc82a4862bd1e140f8b762cfc594008d ]

Now use teardown_lock to serialize for time out and tear down. This may
cause abnormal: first cancel all request in tear down, then time out may
complete the request again, but the request may already be freed or
restarted.

To avoid race between time out and tear down, in tear down process,
first we quiesce the queue, and then delete the timer and cancel
the time out work for the queue. At the same time we need to delete
teardown_lock.

Signed-off-by: Chao Leng 
Reviewed-by: Sagi Grimberg 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin

nvme: introduce nvme_sync_io_queues

2020-11-18T18:20:24+00:00

[ Upstream commit 04800fbff4764ab7b32c49d19628605a5d4cb85c ]

Introduce sync io queues for some scenarios which just only need sync
io queues not sync all queues.

Signed-off-by: Chao Leng 
Reviewed-by: Sagi Grimberg 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin

nvme-rdma: handle unexpected nvme completion data length

2020-11-10T11:37:30+00:00

[ Upstream commit 25c1ca6ecaba3b751d3f7ff92d5cddff3b05f8d0 ]

Receiving a zero length message leads to the following warnings because
the CQE is processed twice:

refcount_t: underflow; use-after-free.
WARNING: CPU: 0 PID: 0 at lib/refcount.c:28

RIP: 0010:refcount_warn_saturate+0xd9/0xe0
Call Trace:
 
 nvme_rdma_recv_done+0xf3/0x280 [nvme_rdma]
 __ib_process_cq+0x76/0x150 [ib_core]
 ...

Sanity check the received data length, to avoids this.

Thanks to Chao Leng & Sagi for suggestions.

Signed-off-by: zhenwei pi 
Reviewed-by: Sagi Grimberg 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin

nvme-rdma: fix crash when connect rejected

2020-11-05T10:43:22+00:00

[ Upstream commit 43efdb8e870ee0f58633fd579aa5b5185bf5d39e ]

A crash can happened when a connect is rejected.   The host establishes
the connection after received ConnectReply, and then continues to send
the fabrics Connect command.  If the controller does not receive the
ReadyToUse capsule, host may receive a ConnectReject reply.

Call nvme_rdma_destroy_queue_ib after the host received the
RDMA_CM_EVENT_REJECTED event.  Then when the fabrics Connect command
times out, nvme_rdma_timeout calls nvme_rdma_complete_rq to fail the
request.  A crash happenes due to use after free in
nvme_rdma_complete_rq.

nvme_rdma_destroy_queue_ib is redundant when handling the
RDMA_CM_EVENT_REJECTED event as nvme_rdma_destroy_queue_ib is already
called in connection failure handler.

Signed-off-by: Chao Leng 
Reviewed-by: Sagi Grimberg 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Sasha Levin

nvme-pci: disable the write zeros command for Intel 600P/P3100

2020-10-29T08:57:25+00:00

commit ce4cc3133dc72c31bd49ddcf22d0f9eeff47a761 upstream.

The write zeros command does not work with 4k range.

bash-4.4# ./blkdiscard /dev/nvme0n1p2
bash-4.4# strace -efallocate xfs_io -c "fzero 536895488 2048" /dev/nvme0n1p2
fallocate(3, FALLOC_FL_ZERO_RANGE, 536895488, 2048) = 0
+++ exited with 0 +++
bash-4.4# dd bs=1 if=/dev/nvme0n1p2 skip=536895488 count=512 | hexdump -C
00000000  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
*
00000200

bash-4.4# ./blkdiscard /dev/nvme0n1p2
bash-4.4# strace -efallocate xfs_io -c "fzero 536895488 4096" /dev/nvme0n1p2
fallocate(3, FALLOC_FL_ZERO_RANGE, 536895488, 4096) = 0
+++ exited with 0 +++
bash-4.4# dd bs=1 if=/dev/nvme0n1p2 skip=536895488 count=512 | hexdump -C
00000000  5c 61 5c b0 96 21 1b 5e  85 0c 07 32 9c 8c eb 3c  |\a\..!.^...2...<|
00000010  4a a2 06 ca 67 15 2d 8e  29 8d a8 a0 7e 46 8c 62  |J...g.-.)...~F.b|
00000020  bb 4c 6c c1 6b f5 ae a5  e4 a9 bc 93 4f 60 ff 7a  |.Ll.k.......O`.z|

Reported-by: Eric Sandeen 
Signed-off-by: David Milburn 
Tested-by: Eric Sandeen 
Signed-off-by: Christoph Hellwig 
[ Fix-up for 5.4 since NVME_QUIRK_NO_TEMP_THRESH_CHANGE doesn't exist ]
Signed-off-by: Jason Andryuk 
Signed-off-by: Greg Kroah-Hartman

nvme-tcp: check page by sendpage_ok() before calling kernel_sendpage()

2020-10-14T08:33:01+00:00

commit 7d4194abfc4de13a2663c7fee6891de8360f7a52 upstream.

Currently nvme_tcp_try_send_data() doesn't use kernel_sendpage() to
send slab pages. But for pages allocated by __get_free_pages() without
__GFP_COMP, which also have refcount as 0, they are still sent by
kernel_sendpage() to remote end, this is problematic.

The new introduced helper sendpage_ok() checks both PageSlab tag and
page_count counter, and returns true if the checking page is OK to be
sent by kernel_sendpage().

This patch fixes the page checking issue of nvme_tcp_try_send_data()
with sendpage_ok(). If sendpage_ok() returns true, send this page by
kernel_sendpage(), otherwise use sock_no_sendpage to handle this page.

Signed-off-by: Coly Li 
Cc: Chaitanya Kulkarni 
Cc: Christoph Hellwig 
Cc: Hannes Reinecke 
Cc: Jan Kara 
Cc: Jens Axboe 
Cc: Mikhail Skorzhinskii 
Cc: Philipp Reisner 
Cc: Sagi Grimberg 
Cc: Vlastimil Babka 
Cc: stable@vger.kernel.org
Signed-off-by: David S. Miller 
Signed-off-by: Greg Kroah-Hartman

nvme-core: put ctrl ref when module ref get fail

2020-10-14T08:33:00+00:00

commit 4bab69093044ca81f394bd0780be1b71c5a4d308 upstream.

When try_module_get() fails in the nvme_dev_open() it returns without
releasing the ctrl reference which was taken earlier.

Put the ctrl reference which is taken before calling the
try_module_get() in the error return code path.

Fixes: 52a3974feb1a "nvme-core: get/put ctrl and transport module in nvme_dev_open/release()"
Signed-off-by: Chaitanya Kulkarni 
Reviewed-by: Logan Gunthorpe 
Signed-off-by: Christoph Hellwig 
Signed-off-by: Greg Kroah-Hartman