This correctly puts all of the I/O channels on the right
thread during shutdown.
Change-Id: I432c8cb981a68b40639f8876dae42597a9e6a85f
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/389637
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
When device was reinitialized, the old
virtqueue memory used to leak. While
here, also added a doc for virtio_dev_init.
Change-Id: I9bb8dbfa5d68deeb73e71567a7e51b82f869537a
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/388297
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
The vbdev_* functions that were declared in lvol.h are all internal to
the lvol vbdev implementation; they don't need to be in the public
header.
vbdev_get_lvs_bdev_by_lvs() is moved to the top of the file so it
doesn't need a forward declaration (no code change).
Change-Id: I98282d3cb0008da5c5b5059eee335219dc929f1c
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/389894
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Modify the vbdev_get_lvol_store_by_uuid() API function to take the UUID
as a string instead of a uuid_t, since this simplifies the calling code.
This also allows us to remove #include <uuid/uuid.h> from the public
API, giving us the freedom to change the underlying UUID implementation
later if necessary without breaking API.
Change-Id: Ib360281f384f95722c5ab64e75dcf3603f826e4c
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/389893
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Maciej Szwed <maciej.szwed@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
All of the call sites already have a lvs pointer; there's no need to
look up the lvs by UUID again.
Change-Id: Icef54d84e7c89f682ea47f10b282839ab4237ce0
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/389892
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
The array of message strings is now only used in vhost_user.c, so it
doesn't need to be exported.
Change-Id: I40163b95a52a1762eb3639db5553814bf1299959
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/389652
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
The bdev module get_ctx_size callback is only necessary if it returns a
non-zero size.
Change-Id: I00ca366b8a1949bdb091c5e8aefb19cb606481d0
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/389900
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Rather than iterating over the g_nvme_bdevs list to determine if a bdev
is an NVMe bdev, we can just compare the module pointer.
Also, the function can take a spdk_bdev pointer directly rather than
casting from a void pointer.
Change-Id: Iccf3c9b3263e9b2d67d07d037dc00d1cbc0f22a1
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/389899
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
This patch will make sure the nvme bdev work in correct way when we delete
bdev by using rpc method and then exit the app.
Change-Id: I19172440d46d96c7396282e1bef07d98ab601390
Signed-off-by: Cunyin Chang <cunyin.chang@intel.com>
Reviewed-on: https://review.gerrithub.io/389930
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
replace the bdev_nvme_destruct() with spdk_bdev_unregister(),
clean all the resource in the bdev layer.
Change-Id: Ic4b85dee6225f42babe868796a8c24dec5c22af6
Signed-off-by: Cunyin Chang <cunyin.chang@intel.com>
Reviewed-on: https://review.gerrithub.io/389855
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Currently SPDK NVMeoF will allocate data buffer from mempool, however,
some data buffers may not block size aligned, this is okay for NVMe
backend but will fail for AIO backend. 4KiB aligned data buffers
should be okay for most AIO backends.
Change-Id: If7b9c2962d716addbd755e80e59620377f2ba488
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/389245
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
The 'm' variable was only used in sizeof expressions; m was a struct
vhost_user_msg, and we already have one of those called 'msg' in the
function where it was needed.
A previous commit removed a RTE_USED() marking on m, which introduced a
warning on clang, since it can now see that m is unused.
Fixes: 73bfcede86 ("rte_virtio: remove rte_memzone from vq")
Change-Id: I63eb0d82c7c750b4181b6312bbda82ef3213c205
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/389648
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Also removed vdev->started field.
Used vdev->status field to determine
if a device has been `started`.
This is simply a cleanup.
Change-Id: Ieb06b03b3bd60bd5906b41c987c418604906c913
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/388296
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Now that vtpci_internal has been removed,
the vdev->id and g_ctrlr_counter are
obsolete and can be removed as well.
Change-Id: Ia1e83cdc81552173f4b57283f83c5791e2b90daa
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/388301
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Removed another direct DPDK dependency.
Change-Id: I12d9a49fb16fbefad42c09f46c280f6f7be094d1
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/388300
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Use standard spdk_dma_malloc instead.
This removes another direct DPDK
dependency. Yet, for multi-process
support we will more likely use a
memzone for the entire virtio device,
not it's virtqueues.
Change-Id: I57119ad830b573b718605b46cc3099b105651ee9
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385629
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Replaced them with spdk alternatives.
While here, also clarified the documentation
for virtio_*mb.
Change-Id: Ia604e9abd8eadc18fa3ea1a782e29e7da6b3c162
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385627
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Currently there are a bunch of asserts() on metadata
descriptors - change these to fail the blob parsing
instead.
While here also return -ENOMEM if any of the memory
allocations fail.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ie37b73c57b304d05a45d10a8d33bcc1d47e7a1be
Reviewed-on: https://review.gerrithub.io/388702
Reviewed-by: Maciej Szwed <maciej.szwed@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
data_ro means that write, write_zeroes and unmap operations are not
allowed.
md_ro means that resize, set_xattr and remove_xattr are not
allowed.
There is no code yet that can activate this - it is coming in a future
patch. Two usages are planned though:
1) a user explicitly marks a blob as read-only - this is persisted so that
future loads of the blob will ensure the blob cannot be modified - neither
metadata nor data
2) a future feature flag framework (how's that for alliteration) may allow
a blob to be opened, but not allow metadata modifications, if there are
feature flags set in the blob's or blobstore's metadata that the
application does not understand
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I247fd900430c56f7176edfb80dddd5a1a6c8dc87
Reviewed-on: https://review.gerrithub.io/388663
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
There are two ways to set stockets to nonblocking type:
- ioctl with FIONBIO
- fcntl with O_NONBLOCK
Those two should be equivalent for sockets used in SPDK.
During testing it was shown that VPP interprets only
the second type, so this patch changes all occurences of it.
When here, more descriptive error logs were set in case of
failure.
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: Ifa5b30e3a4fa04fe23f41fa2ae9dab4b01dd7d3c
Reviewed-on: https://review.gerrithub.io/388816
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Moved some vq-related functions
to virtio.c. It's the only place
they're used in.
This patch also removes the
virtqueue_full() function. It's
not used anywhere.
Note: all the functions content
has been moved 1:1.
Change-Id: Ib854fc0836378e6955c0a7358ecabcf2c3107d06
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385628
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Only multi-process shared controllers should be inserted into the shared
list in g_spdk_nvme_driver. To accomplish this, create a second
per-process global list of attached controllers (g_nvme_attached_ctrlrs)
and rename the driver struct field to shared_attached_ctrlrs to clarify
its purpose. Additionally, a new helper function, nvme_ctrlr_shared(),
returns whether a given controller should be on the shared or
per-process list.
Change-Id: I46d4e558ece8b7fc3d28868e32bb56d794f21aab
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/389190
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Simplify the PCIe transport by using an existing function to look up a
controller by transport ID.
Change-Id: I261865df1ba23069b052ca64944b7637d70c85ba
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/388701
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Each process should manage its own list of controllers that are
initializing; the list doesn't need to be shared between processes.
This is the first step toward preventing non-PCI controllers from being
added into the shared attached_ctrlrs list.
Change-Id: Ia6f85fe89e28a04f0950da5362bb2f49d1b76da9
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/388695
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
If the user calls spdk_rpc_close() multiple times, there would be a
double free. This is not normally possible when using the SPDK event
framework RPC subsystem, since it will only call spdk_rpc_close() once,
but it doesn't hurt to make the RPC library public API handle this case.
Fixes GitHub issue #219.
Change-Id: I80cff0ed871dec6b90bd2658e85b0cd0914b6ada
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/389394
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Change the return type of spdk_bdev_register related
functions and try to handle the duplicated name
issue.
Change-Id: I23af11583cf2050579d1624508306a35394bffde
Signed-off-by: Ziye Yang <optimistyzy@gmail.com>
Reviewed-on: https://review.gerrithub.io/388178
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Use the new one from io_channel.h.
Change-Id: I7bf6729caf6eeebcb58450a36119601957ad5da4
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/388290
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
This function will send a message to each allocated
thread asynchronously, then call a callback on
the originating thread.
Change-Id: I3ebe7c6c5b460a944a32487d1091b601a482a256
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/388041
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
This simplifies the public API and requirements for
user applications.
Change-Id: Ibb0d25a7838a0fa683f39e79cb4fef78adf6aee8
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/388040
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Using pthead_detach to let system take back the resources owned
by SPDK created thread for nbd kernel module.
Change-Id: I39f3c0b60ecf57f3227a7b4733a0f144a6e169c5
Signed-off-by: Xiaodong Liu <xiaodong.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/388722
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: <shuhei.matsumoto.xt@hitachi.com>
Make this generic and not directly dependent on
the event framework. That way our libraries can
register pollers without adding a dependency.
Change-Id: I7ad7a7ddc131596ca1791a7b0f43dabfda050f5f
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/387690
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Signed-off-by: Maciej Szwed <maciej.szwed@intel.com>
Change-Id: Id0f7751d8889584fff1a947160ee89e2f158af1c
Reviewed-on: https://review.gerrithub.io/388943
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
With configuration where there are nested lvol stores we
encounter situation when hotremove and fini functions try
to unload the same lvol store second time (previous unloading
have not finished yet).
Signed-off-by: Maciej Szwed <maciej.szwed@intel.com>
Change-Id: Icb5d430d17024fbda0e6113b380d8f346a3e9a84
Reviewed-on: https://review.gerrithub.io/388790
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
It doesn't need to be accessed outside of bdev.c.
Change-Id: Ib1a1abe88b0993e391e0a39c5483781b2c0ad523
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/388508
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
The io_device_unregister callback might be
deferred and called on a different core, but
the virtio device has to be released from
the same thread that created it. Hence, once
the unregister callback is called, it has
to send yet another msg to the vdev-owning
thread. Since all virtio devices are currently
created on the same thread, no mutexes are
needed. They will need to be introduced once
we publish the API to connect to virtio
controllers.
This patch also defers bdev_virtio module-
-finish until all virtio_devs are destroyed.
Change-Id: Iaafceefa1e862b839b5db8c6c4975bf51441a083
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/388835
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
This will allow us to send a message
to a virtqueue via spdk_thread_send_msg.
Change-Id: I8502f27e74de107bc5e5ccf2939448896579acc2
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/388834
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
We previously used a separate io_device
for each bdev created from a virtio_dev.
The new approach makes us reuse the same
io_channel for different bdevs on the same
core. It also provides a built-in safety
check for freeing virtio_dev only when
all io_channels on this device have been
closed.
Change-Id: Iaae66964f90aa3cef95ed2e261c6f3dfb2f26caa
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/388505
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Currently on blobstore creation we use write zeros everytime.
Some drives does not support write zeros, but support unmap.
We should do write zeroes only on metadata and try to unmap
data clusters
Signed-off-by: Maciej Szwed <maciej.szwed@intel.com>
Change-Id: Iae36c1ccacc08340e79ad40c4c9a2c53dda920ba
Reviewed-on: https://review.gerrithub.io/387152
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Using pthread_create is much cleaner, and avoids race
conditions related to the forked process sharing memory
with the primary process. Specifically this fixes
a double spdk_event_call() of the shutdown event - one
from each process. The primary process may have already
recycled the shutdown event, and when the forked process
calls it again, hilarity ensues.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I3e4b77b8b380a8fa0227b770758099bcee0ef35b
Reviewed-on: https://review.gerrithub.io/388520
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
dev_fd => fd for /dev/nbdX device node
spdk_sp_fd => fd that SPDK will use to poll for new
I/O requests from the kernel nbd thread
kernel_sp_fd => fd passed to the kernel to send I/O
to SPDK
This enables an upcoming patch to switch from using
fork() to pthread_create().
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I0840a35eb1ba12810b7fc12775ee0df9ad3d99f9
Reviewed-on: https://review.gerrithub.io/388519
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
"OK/NG" may be only used in Japan. It may cause misunderstanding.
Hence remove "OK/NG" or change to "Allowed/Denied".
Access control of iSCSI target follows further refactoring.
Hence to save the amount of change, "OK/NG"s are mainly removed.
Change-Id: I1f2ef9bc2be21caa86a6ae98377b6d6b3f85386a
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/385075
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
The vdev->ctx conversion is now
straighforward and doesn't need
any macro.
Change-Id: I88b0cf3fa5e95b93e5f69a0a3ae9508f9fac3683
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385625
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This is a single header file that
provides access to all virtio initiator
capabilities. This is a preparation
to exporting rte_virtio as a separate
spdk lib. Also renamed virtio_dev.c.
Change-Id: If45a0cadc61943b6ffd87cf70a59321a83b238a6
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385624
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
It's contents have been moved to
virtio_user/vhost.h and virtio_dev.h.
The virtio_user_dev struct can't be hidden
in virtio_user.c (similarly to the virtio_hw),
due to another abstraction layer in virtio_user
- vhost-user and vhost-kernel. Even though only
vhost-user is implemented now, let's not close
the possibility to implement vhost-kernel in
the future. The vhost-user/vhost-kernel code
should probably be placed in separate files,
with the virtio_user_dev struct in common
vhost.h header.
Change-Id: I36ee96eff30b398dd129953a874513e727619f98
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385622
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Moved it's contents to virtio_user.c.
Having this logic separated across
multiple files doesn't make much sense.
Change-Id: Ib2daaa56e1b3d755d9476abca247c7cbcfa7096b
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385621
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Renamed all vtpci functions.
While here, also added some documentation.
Change-Id: Id24aaab8c5b556b969c60ee1c95b9d46936ea13d
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385426
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
virtio_hw struct definition is now hidden
in virtio_pci.c. The API to enumerate all
PCI devices is available through virtio_dev.h.
Change-Id: Ib9c968732d9f78204e9664c426ad99542cfe8c93
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385425
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
The vtpci layer is about to be removed.
This is only 1:1 code move, all functions
will be renamed/refactored in future patches.
Change-Id: I63c1b30fdaf6eff2e5d8bf60e62c87c2cb9f2b15
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385424
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Now that legacy PCI support has
been dropped, we can safely assume
there's only one backend type
per device. So the vtpci_internal
- containing process-local data -
can be removed.
Change-Id: I15d87a280b74318c7a44ee60ce3408756d1f5a69
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385423
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This patch initiates the removal
of the vtpci layer from rte_virtio.
The general idea is to provide separate
virtio_dev_construct and virtio_dev_init
functions. construct would just allocate
and setup the struct data. init would
be the one to start the device via
MMIO/vhost-user. virtio_dev_construct
takes 2 params - backend ops and backend
context. The context is kept as void*
and can be cast to proper structs inside
backend ops callbacks.
Change-Id: I1d6d92e1cf09b9d0c9fa4fd2cb70203e3fc7e65b
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/387553
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
We currently pass virtio_user_dev as
an argument to various function in belief
they will access the virtio_dev via
virtio_user_dev->vdev field. However,
this field is about to be removed soon.
This patch is a mid-step towards a bigger
change. See the next patch for details.
Change-Id: Ie684aa1614d9777d79d350bbfda79effe2a2d5f6
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385422
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
It's been only used in 1 place.
Change-Id: I0b7c9cdb809a6946eda5ee361362dec86e57b4ef
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385623
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This is required for future changes.
The general idea is to provide separate
virtio_dev_construct and virtio_dev_init
functions. construct would just allocate
and setup the struct data, while init
would be the one to start the device via
MMIO/vhost-user. If we want to destruct
a non-initialized device, there's no
need to do a reset.
Change-Id: Ib8ce56169e86e43fba8fcd1bf6c2a5b12ed99966
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385421
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Even though config change intrrupt
handlers are still required, we need
them only for real PCI devices.
They should be handled internally
in vhost_pci.
However, current QEMU implementation
doesn't ever change SCSI PCI config,
so no interrupt handler is required
for now.
We might want to add it together with
Virtio-Blk, as PCI config will change
there with each resize.
Change-Id: Iece024cd31c788857ce900385e289758ca3f5948
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385304
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
It's been only used in 1 place.
Change-Id: I3e933d3fb0816ea30f983fb3de69ce49238c3b72
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385303
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
When there are no lvols present on lvol store
spdk_bdev_module_examine_done function is not called
when tasting is done.
Signed-off-by: Maciej Szwed <maciej.szwed@intel.com>
Change-Id: Ia6c559f392148e3b4444dbc547df462bbe02356d
Reviewed-on: https://review.gerrithub.io/388402
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
To add initiators to an existing target, manage name and netmask
of initiator group by linked list. It is not easy for array to
detect duplication and change configuration.
Change-Id: Id9e77740e24ac309a33d8de1be37663681160d94
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/379932
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ziye Yang <optimistyzy@gmail.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
spdk_iscsi_init_grp_create_from_configfile() returns 0 even if
creating an initiator group (IG) is failed due to empty netmask or
initiator name. This will be unexpected behavior for end users.
Hence change the return code from 0 to -ENVAL for this failure.
Besides do the following.
Change other error return codes in the function by using errno too.
Current IG don't have any UT code in repository. UT code to confirm
this fix is also submitted.
Change-Id: Ic918feb5c6a3e772b325664d881b2673b553a217
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/384408
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Signed-off-by: Maciej Szwed <maciej.szwed@intel.com>
Change-Id: Ic11a56c5616098ff42f31f78aa96289af5965118
Reviewed-on: https://review.gerrithub.io/387611
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Signed-off-by: Maciej Szwed <maciej.szwed@intel.com>
Change-Id: If89275acfb1560982e332148a99ed3c83f8cb34f
Reviewed-on: https://review.gerrithub.io/387609
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
We used to unlink the unix domain socket
regardless of who created it. Even we
got the "socket already exists" error
on init, we would still unlink the socket
on shutdown.
Change-Id: Ib14382d6d7e127d8f51f6544cbb063594e040879
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/388284
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Deinitialize only those subsystem that
have been fully initialized.
Change-Id: I86cdd4cda0a80c107956843ed8198072e898a743
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/386892
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
spdk_bdev_nvme_io_passthru_md() is verified on QEMU NVMe
and will be verified on Cosmos mini OpenSSD.
Change-Id: Ib759b6b6095deaa4ae7cf746f3a241f678295605
Signed-off-by: Young Tack Jin <youngtack.jin@circuitblvd.com>
Reviewed-on: https://review.gerrithub.io/387114
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
- for environments that don't support rlimits
Change-Id: I6dde1eb9e1cb2274462a450625d8bc70a3889163
Signed-off-by: John Meneghini <johnm@netapp.com>
Reviewed-on: https://review.gerrithub.io/386355
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
We probably do not allocate the mgmt_channel when we destroy the
rdma queue pair, add the judgment before put the io channnel.
Change-Id: I528c9686907c04c5025ff9d3872f24f2996e9d33
Signed-off-by: Cunyin Chang <cunyin.chang@intel.com>
Reviewed-on: https://review.gerrithub.io/388159
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Change the return type of spdk_channel_msg from
void to int. If this msg executed in a failure,
we do not need continue sending the message to other
threads, we can just tell the original thread, and
let the orgiginal thread call the spdk_channel_for_each_cpl
call back.
Thus we can track the qpair creation/destroy case for bdev
reset in nvme bdev module;
Change-Id: Ide9dffd1f84a29fcf61d8339a9ece2a0245d968d
Signed-off-by: Ziye Yang <optimistyzy@gmail.com>
Reviewed-on: https://review.gerrithub.io/387284
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I8ea8c59a0c67176c0c0c39abf807afad61ff3828
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/387689
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Change-Id: I864cd8a6c206dfbe62fcb3f72275c1ae51aa4ed7
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/387688
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
We always unregister pollers from the core they're running on,
so we don't need the async completion event when unregistering
them. To avoid changing the logic, create events instead.
Change-Id: Ib50d93e1578684bea69200d89a5e0d383acc8cfe
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/387685
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Register all pollers to the current core. If necessary, send
an event to the correct core before registering.
Change-Id: Ie34cc8b11143a58c0f621c87c409a3d09d929648
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/387682
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
This is the inverse operation of spdk_io_channel_get_ctx.
Change-Id: Ica6593240dfb05cb53dc7ec910bc0a78270e81c0
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/387679
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Change-Id: Ib1e184f831eef82432308279c6d8ea43088748ef
Signed-off-by: Liang Yan <liang.z.yan@intel.com>
Reviewed-on: https://review.gerrithub.io/387273
Reviewed-by: Ziye Yang <optimistyzy@gmail.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Use the env abstraction PCI functions to compare PCI addresses so that
details like whether or not domain is specified or whether 0-padding is
present don't affect the comparison.
For example, 0000:01:00.0 should compare equal to 01:00.0.
Change-Id: I9f3aaeb5f8fdbf3e246e31a41b4c09151288015e
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/387202
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Change-Id: Ibe31b961fec36644cf64716c512429d68e873e88
Signed-off-by: Ed rodriguez <edwinr@netapp.com>
Reviewed-on: https://review.gerrithub.io/386170
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Currently when one of lvols fail during loading we end up
with empty lvol store loaded, where we actually should
unload lvol store as well.
Signed-off-by: Maciej Szwed <maciej.szwed@intel.com>
Change-Id: I90ce3b79baba0db0d875b106c94924be394aada9
Reviewed-on: https://review.gerrithub.io/385950
Reviewed-by: Piotr Pelpliński <piotr.pelplinski@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
This patch make the cache buffer shift of blobfs configurable. We
can configure the cache buffer shift according workload, it will
help to improve the performance.
Change-Id: Ied1c2e5d6fd9eaa7aa0759c03c654fcf2e77aa23
Signed-off-by: Cunyin Chang <cunyin.chang@intel.com>
Reviewed-on: https://review.gerrithub.io/383381
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ziye Yang <optimistyzy@gmail.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
This helps enforce uniqueness of the subsystem id and
allows for quick lookup by that id.
Change-Id: I4d2948892839d42b8c2b2a406682848415ad2bd6
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/376251
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
DPDK 17.11-rc3 removes pci_probe*
and pci_detach functions. It introduces
different ones - rte_eal_dev_attach/detach.
Those have a slightly different signature.
Change-Id: Iadde9ff37c64190dad41929997f9ff78379f36e1
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/387656
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
DPDK 17.11-rc1 moved a part of it's
eal code into two separate libraries.
Change-Id: Ibcea448487a001c30abb68447e3fe2104ba9a7c3
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/387655
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
DPDK 17.11-rc3 switched to IOVA-centric
addressing. The old API is still available,
but is deprecated.
Change-Id: I7b54a1e626e39368acd2190dec725dbf30c7a5de
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/387654
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
DPDK 17.11 removed all rte_config.h includes
from rte_*.h headers, meaning we should either
use gcc param -include rte_config.h (just
like DPDK does), or include this file before
each other rte_*.h include. Since we're using
the latter approach in many places already,
I decided to follow it.
While here, also removed rte_vdev.h dependency
from rte_virtio/virtio_user.c. It's not used
anyway.
Change-Id: I865ee9f828211c03a60fd0446f7a418d5dddd140
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/387653
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Legacy PCI code is hard to maintain
and compiles with DPDK 17.05 only.
If we ever need this functionality
in SPDK, we should first add proper
env/pci layer for direct UIO/VFIO
communication, and then revisit this
topic.
Change-Id: I25622c16f467633e01003c9edbb7e6f297c68a86
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/385302
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Because ppc64 has less strict memory ordering behaviour
than x86 we need to introduce a memory barrier when
polling for completions due to possible reordering
of tracker and cpl access.
Change-Id: Id17116c38b2ba69154c175c539fc97c60897deb0
Signed-off-by: Jonas Pfefferle <jpf@zurich.ibm.com>
Reviewed-on: https://review.gerrithub.io/383728
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
When an iSCSI connection is closed, spdk_iscsi_conn_stop_poller()
is called in spdk_iscsi_conn_destruct() or
spdk_iscsi_conn_check_shutdown().
However, through both paths, g_num_connections[] is decremented
twice:
(1) in spdk_iscsi_conn_stop_poller()
(2) in spdk_iscsi_conn_free()
This behavior makes g_num_connections[] negative unexpectedly and
for some cases load balancing does not work and the same CPU is
always selected.
How to fix:
spdk_iscsi_conn_stop_poller() is called in the following functions:
- spdk_iscsi_conn_check_shutdown()
- spdk_iscsi_conn_destruct()
- spdk_iscsi_conn_handle_idle().
[Idea 1]
Remove the code to decrement g_num_connections[] from
spdk_iscsi_conn_free().
[Idea 2]
Remove the code to decrement g_num_connections[] from
spdk_iscsi_conn_stop_poller(). Add the code to decrement
g_num_connections[] to __add_idle_conn().
[Idea 1] is simple because just only one line is deleted.
[Idea 2] may be more symmetric than [Idea 1] but to find the right
place to add the code in __add_idle_conn() is not clear and
difficult.
Hence [Idea 1] is proposed.
[Idea 1] may not be the optimal one and more refactoring may be
done.
Change-Id: I81050f4a0e7b3ddd40896f46ab2eb8de14bbcb3a
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/384026
Reviewed-by: Cunyin Chang <cunyin.chang@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Change-Id: I2524b7881885a0784492a8189bf22a0b8c92d470
Signed-off-by: Ed rodriguez <edwinr@netapp.com>
Reviewed-on: https://review.gerrithub.io/386169
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
The following functions returns 1 and 0 for succcess and error,
respectively:
- spdk_iscsi_tgt_node_allow_ipv6()
- spdk_iscsi_tgt_node_allow_ipv4()
- spdk_iscsi_tgt_node_allow_netmask()
- spdk_iscsi_tgt_node_access()
Using bool for this purpose will avoid our misunderstanding.
Change-Id: I927876e0503c0eee5364e829a4713f9a345996f6
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/383664
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
In the Appendix C. SendTargets Operation in the iSCSI Specification
(RFC7143), there is the following description:
In a Discovery session, a target MUST return all path information
(IP address-port pairs and Target Portal Group Tags) for the
targets on the target Network Entity that the requesting
initiator is authorized to access.
However no description such that Target Portal Group Tag is used
to check authorization of the requesting initiator.
Moreover, according to the iSCSI Specification (RFC7143),
Target Portal Group Tag is unique within an iSCSI Target Node.
Tarrget Portal Group Tag identifies a portal group within an
iSCSI node.
However, current SPDK uses Target Portal Group Tag to check
authorization of the requesting initiator by mistake. This
implementation does not have any compatibility with other storage.
Hence delete the code related with Target Portal Group Tag from
SendTargets function.
Change-Id: If4b79f0d8fe8f5892a798e4f6716045fd18cf0a4
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/383660
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
SPDK have used "*" and "[*]" as wildcard address
for iSCSI network portal. These abbreviations may be convenient
but unusual. If "*" or "[*]" is used, output warning message
and forcefully overwrite "*" or "[*]" by "0.0.0.0" or "[::]"
respectively.
After confirming that the usage of "*" and "[*]" is not used any
more, the code for "*" and "[*]" will be deleted.
Change-Id: Ibb8fb0d4598ffa1f7b1690ad6afd10d5ae162612
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/383556
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
There is case like no memory to create the channel.
Needs to handle this properly.
Change-Id: I5d13d18037e6aa8f057769b1ef345f45597b22af
Signed-off-by: GangCao <gang.cao@intel.com>
Reviewed-on: https://review.gerrithub.io/386016
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
There are two separate abstraction layers:
* vsocket - which represents a unix domain socket
* virtio_net - which represents a vsocket connection
There can be many connections
on the same socket. vsocket
provides an API to enable/disable
particular virtio features on
the fly, but it's the virtio_net
that uses these features.
virtio_net used to rely on
vsocket->features during
feature negotiation, breaking
the layer encapsulation (and
yet causing a deadlock - two
locks were being locked in a
separate order). Now each
virtio_net device has it's own
copy of vsocket features, created
at the time of virtio_net creation.
vsocket->features have to be
still present, as features can be
enabled/disabled while no
virtio_net device has been
created yet.
Fixes#214
Change-Id: Ic4b2dd8cae6050813fc9a420b2ed30bc5ae60393
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/386294
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
The process of spdk_bdev_finish will always result in an
early call to the user's completion notification - prior
to the actual shutdown events happening.
Change-Id: Ib53393deb794b616d2b7ec94cb41502305bd99b7
Reviewed-on: https://review.gerrithub.io/386707
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
