Since further patches will be adding new descriptors
that are related to cluster layout throughout the blobstore,
add description for existing descriptor too.
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I722eb633445685789d5185ed59dfc910f76b109f
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/481724
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
This is an additional option that can be passed when creating
a blob.
When opts->enable_extent_pages is set to false (current default),
only EXTENT_RLE should be persisted on sync.
During blob load, when EXTENT_RLE is present in md,
blob->extent_rle_found is set to true.
When opts->enable_extent_pages is set to true,
only EXTENT_TABLE and EXTENT_PAGES should be persisted on sync.
During blob load, when EXTENT_TABLE is present in md,
blob->extent_table_found is set to true.
It is possible to find neither EXTENT_* descriptor when loading a blob.
This means that blob length is 0 and EXTENT_RLE was supposed to be used.
Yet none were persisted due to lack of clusters.
In such case blob->use_extent_table is set to true after finishing
blob load.
When parsing metadata ends, if extent_table_found is set - then
support for extent_table is enabled. All other cases disable it.
At this time path for Extent Pages is not implemented, so it should
not be used.
Later in the series, it will become the default path for serialization.
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I2146da6130a0645e686ab02a3b5d2d86a7d35a1f
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/479853
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
When a command arrives and no requests are available, the socket
recv state machine sits in the RECV_STATE_AWAIT_REQ state until another
network event occurs. If this I/O was the last one sent, this leaves the
target hung. To fix this, when a request is completed, kick the state
machine to make forward progress.
In practice, this can only occur once the pdu send acknowledgements are
asynchronous relative to arriving commands. That only begins happening
with the use of MSG_ZEROCOPY. When MSG_ZEROCOPY is turned on, it's
possible receive the next PDU in a chain for a command prior to seeing
the acknowledgement that the response that triggered that PDU actually
sent.
Change-Id: I556f31ad56970d36aa3538cfde375d35f3d4e551
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/480002
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Previously, the R2T was sent and if an H2C arrived prior
to seeing the R2T ack, it was processed anyway. Serialize
this process.
In practice, if the H2C arrives with a correctly functioning
initiator, that means the R2T already made it to the initiator.
But because the PDU hasn't been released yet, immediately processing the
PDU requires an extra PDU associated with the request. Basically, making
this change halves the worst-case number of PDUs required per
connection.
In the current sock layer implementations, it's not actually possible
for the R2T send ack to occur after that H2C arrives. But with the
upcoming addition of MSG_ZEROCOPY and other sock implementations, it's
best to fix this now.
Change-Id: Ifefaf48fcf2ff1dcc75e1686bbb9229b7ae3c219
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/479906
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
This function was only called from one spot.
Change-Id: I856f564d3ef6c6157be7a32a2cd812c702516a8d
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/482003
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This seems like a more descriptive name
Change-Id: Ia616865b3fb36d8f9ccc5fb2ca6185bdd8543cf8
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/482002
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
With our target design, there's no advantage to sending
multiple R2T PDUs per nvme command. This patch starts by
setting up the math so that at most 1 R2T PDU is required
per request. This can be guaranteed because the maximum
data transfer size (MDTS) is pre-negotiated in NVMe-oF
to a reasonable size at start up.
It then proceeds to simplify all of the logic around mapping
requests to PDUs. It turns out that the mapping is now always
1:1. There are two additional cases where there is no request
object at all but a PDU is still needed - the connection response
and termination request. Put an extra PDU on the queue object
for that purpose.
This is a major simplification.
Change-Id: I8d41f9bf95e70c354ece8fb786793624bec757ea
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/479905
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
We can always accept up to the maximum I/O size in an H2C,
so eliminate the #define.
Change-Id: I349dab5f9b6ec482a7c580b1396e03c8d30a250b
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/482278
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
The resources allocated to a queue pair do not need to be directly
correlated to the queue size requested by the initiator in NVMe-oF, as
long as enough resources are present. The RDMA transport, for instance,
does complex pooling of the resources behind the scenes when using a
shared receive queue.
Simplify the resource allocation for a TCP qpair to just always allocate
the max allowed queue size right away. This is a configurable parameter,
so system administrators can adjust for their needs. The initiator may
then request a queue size less than or equal to that, which will only be
enforced by queue depth counting and not impact the actual number of
resources allocated on the target.
This change relies on the MaxC2HSize being equal to the Maximum Data
Transfer Size (MDTS) reported. That is the default configuration, but
MDTS is configurable. Changing the MDTS with this patch to a value
larger than 128k will cause the target to break. This is addressed in
the next patch in this series.
Change-Id: Ibd4723785c6a4d8d444f9b7bbfa89f98de2320f5
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/479733
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
These values do not need to be negative.
Change-Id: Id9f798cf1c9da354448f9c6fbb90e599f877bb32
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/482277
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
By releasing the just-completed PDU prior to calling the callback,
for flows that immediately submit another PDU inside the callback,
the just-released PDU can be immediately reused. This reduces the number
of PDUs required in the pool to continue forward progress to half of the
previous value, while also making it more CPU cache friendly.
Change-Id: I8031b8f9f57ac05f261d96433d9899fe5e31d318
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/479904
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ziye Yang <ziye.yang@intel.com>
Reviewed-by: Or Gerlitz <gerlitz.or@gmail.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
When we unlock a range, we remove the range from the
locked bdev list before doing the for_each_channel
iteration to remove the range from each channel.
But at the same time, right after removing from the
locked list, a new lock on that range could start.
In that case, we also do a for_each_channel to add
the range to each channel, and that will race with
the for_each_channel remove. When the lock start
wins, it finds the range already in the channel,
but doesn't set the owner_range which results in
a seg fault when the for_each_channel completes.
The fix is actually rather simple. We just add the
locked_ctx to the comparison when checking if the
range is already in the channel. If the locked_ctx
matches, then we know it was added as part of
initializing a new channel. If it doesn't, then
we create a new range object pointing to the new
locked_ctx. The first one will get removed when
the remove for_each_channel catches up.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I94f8b20376dd437f404add35744d42fc148303ff
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/482620
Community-CI: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Maciej Szwed <maciej.szwed@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
If a locking operation has to wait because of an
existing lock, we queue the lock context. When the
existing lock finishes unlocking, we restart the
queued lock context. But we have to make sure
we restart the lock context on the same thread it
was originally submitted, since it has a channel
associated with it.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I555515f3adfc3c13a86584c601ed541d605980b7
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/482463
Community-CI: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com>
Reviewed-by: Maciej Szwed <maciej.szwed@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
For ACWU we always set value 1 because bdev holds
information specific for namespace only. This value
actually does not matter because we also set NACWU
which makes ACWU irrelevant. We set ACWU because
NVMe specs requires ACWU != 0 if fused commands
are supported.
Signed-off-by: Maciej Szwed <maciej.szwed@intel.com>
Change-Id: Ida4357026d3b32677fc824b3cd878e7ad8ef2680
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/477915
Community-CI: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This function is required for NVMf implementation
for compare and write fused command.
Signed-off-by: Maciej Szwed <maciej.szwed@intel.com>
Change-Id: If41611f5c0b8e4ed8eec66f09858c724f1800d59
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/477914
Community-CI: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Add call for spdk_nvmf_bdev_ctrlr_compare_and_write_cmd
function in spdk_nvmf_ctrlr_process_io_cmd function
when fused command is discovered.
This patch also removes redundant defines for fused flags.
Signed-off-by: Maciej Szwed <maciej.szwed@intel.com>
Change-Id: I61971a56577ab32b52e1fde1e572f718a9a2d9aa
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/476621
Community-CI: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Move fused cmd related code from spdk_nvmf_ctrlr_process_io_cmd
to separate function.
Signed-off-by: Maciej Szwed <maciej.szwed@intel.com>
Change-Id: Ic662a968b054f05db7f6e1cf4fa9aa13f6fb7c40
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/481942
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This patch introduces new spdk_nvmf_bdev_ctrlr_compare_cmd
function which implements support for compare operation.
Signed-off-by: Maciej Szwed <maciej.szwed@intel.com>
Change-Id: Iadf402a6441a78ea0e6468f1066c6b0e10e63b9b
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/477782
Community-CI: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This patch introduces new function that is a part of
upcoming support for fused commands.
Signed-off-by: Maciej Szwed <maciej.szwed@intel.com>
Change-Id: I019c587bee7fd0f745ec17c141baf4cb7bf86645
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/476611
Community-CI: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Added new function for getting NVMe specific return code
for fused commands. Also changed one of the return codes
in fused commands so that we could distinguish error
cases.
Signed-off-by: Maciej Szwed <maciej.szwed@intel.com>
Change-Id: I86417ea4f5b8f3e6496162be3d6c6128076e35d4
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/481666
Community-CI: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
This change fixes a merge incompatibility between commits
50cb6a04acf3f77863cc7fe7753dabd79beaab57 and
708ed4fb6e8c41d6033ce26b349171aa77703061.
Change-Id: I5bc71a3c214667f01de66857cf61b9eb25f6cf6b
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/482586
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
In the event that we have more than one event outstanding for a qpair
at the time of destruction, we need to ack all of the events, Luckily
the synchronization is already there in the form of the ctrlr lock.
Signed-off-by: Seth Howell <seth.howell@intel.com>
Change-Id: Ib297598f2e28d9b9bd83e904f950795a61fa883a
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/479171
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
A new API was added `spdk_bdev_io_get_aux_buf` allowing the caller to request
an auxiliary buffer for its own private use. The API is used in the same manner that
`spdk_bdev_io_get_buf` is used and the length of the buffer is always the same as the
bdev_io primary buffer. 'spdk_bdev_io_put_aux_buf' is called to free the
auxiliary buffer.
The initial use case is crypto, in the next patch in series it is used. No UT were
added as the logic isn't that complicated and it is fully tested with each run
of crypto.
Fixed a comment typo also (not mine for once).
Signed-off-by: paul luse <paul.e.luse@intel.com>
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ib1939fcbc8e5db36fd909ef26771a725a551e8e6
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/478383
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
To avoid partial write issue of this PDU.
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: Id9b22da844c75ae53c6881850d192b40ac4098ac
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/481948
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Purpose: To prepare for the further patch submission.
Since we do not need to keep this variable too late.
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: Ibaa100925e1ea317253d4fe7e560917e063fcf6b
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/482290
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Since only after DATAIN pdu sending out, we can have
free slot to handle queued data in tasks.
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: I49a52597e8660453ea90c5960d020eb53f81265d
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/482048
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This is prepared for the further call back usage.
Change-Id: Iccf304c87e67debfb4e7c330acc9cc233cc3ec48
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/481917
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This patch eliminates the flushing logic and simplies
the writev logic. And this patch can also improve the performance.
We support async write for PDUs other than login response, logout response,
and text response in this patch. We will support async write also for them
later in this patch series.
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: I243f598f297d594da0bb18466bc47dab918ed3ee
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/481686
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
We will not enable the live recovery feature for SPDK internal vhost library,
so we unmask the protocol flag for internal vhost library.
For the purpose to make it can be compiled with latest DPDK version, some
mandatory APIs are required, so add them here.
Change-Id: I34fab7ed90c86a0fb612852a47f6cadeb8a072f3
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/482069
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
There is a spdk_nvmf_tgt_listen() which opens a port for specified
transport (trid) which opens possibility to accept new connections
from initiators. However there is no counterpart of this function
(i.e. spdk_nvmf_tgt_stop_listen()), which would stop listening.
Instead the current code relies on spdk_nvmf_subsystem_destroy()
to stop the listener, which seems to be wrong.
Fixes#1129
Change-Id: I6e73d8c234dc451f0fee8394132eae34cd4f4756
Signed-off-by: Jan Kryl <jan.kryl@mayadata.io>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/479873
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Amortize the writev syscall cost by using the writev_async socket API.
This allows the socket layer to batch writes into one system call
and also apply further optimizations such as posix's MSG_ZEROCOPY
when they are available. As part of doing so we remove the error
return in the socket layer writev_async implementation for sockets
that don't have a poll group.
Doing so eliminates the send queue processing.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Change-Id: I5432ae322afaff7b96c22269fc06b75f9ae60b81
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/475420
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Initiator drivers (e.g nvme/tcp) don't use poll groups but rather directly
poll the qpair. In this case we want to allow the polling function (e.g
_qpair_process_completions()) to flush async writes pending on the socket.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Change-Id: Ibd8c73691213d58e287b7110d0f5a381a89a64d0
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/475419
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Replaying md through _spdk_bs_load_replay_md_cpl() starts with
md page 0 in search of first valid md page starting a chain
for particular blob.
When it is found, next pages read are from the current pages
`next` page - next in chain.
After whole chain is read, it goes back to first page in chain
and starts search for next valid chain from there.
This patch adds separation between reading particular chain,
and moving to the next one.
Moving on to the next one happens in _spdk_bs_load_replay_md_chain_cpl().
Further in the series, extent pages will be added in the metadata.
Those are not within any particular blobs chain of metadata,
but spread out over the md region.
It is not enough to read all md and read extent pages.
In case of power failure, only extent pages known to be valid
are the ones which are pointed to by some valid md chain.
In futher patches, a step will be added after reading particular
valid md chain to go read extent pages pointed by it.
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I6e7cd64af66ce5db0abd2ad5962d604ac2b30994
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/481900
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Moved finishing of unloading to separate function,
which is now called on every failure and success when unloading
the blobstore.
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I34539b78c5cc63a6fe5891014cba89b9eb62d4df
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/482009
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Before this change it was possible to fail at
writing out some of used md pages.
bserrno output of those was not verified.
This patch adds it at every step.
With that two function don't need (and never needed)
to pass the bserrno:
_spdk_bs_load_write_used_md()
spdk_bs_load_complete()
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I1a61763f03665ba1b00e5949ef0cf37eefaaf08f
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/482008
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
This is simplification of load path.
seq is save in ctx already, no need to pass it to the function.
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: Ief0ddc1826c461adbad71ba1a3897c510ec2a971
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/482007
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Not inlining all host to controller operations breaks the target within
the context of fused commands. This issue was discovered when enabling
the compare-and-write fused command. Only the write command buffer was
being inlined which caused the write to jump the compare in the
transport specific state machine on the target side before our fused
command checks in the generic code.
Change-Id: I9e52ae6160e01ffd36d20429ffc8459491c729ef
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/482001
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
We should check the thread's state at the end of message callback, or
we may leak the message memory in case the thread was set to exit state.
Change-Id: Ifb67c3b5c39440c411eca1d045c11e8aa6c514cc
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/482206
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
