Don't allow enqueueing a request when not
enough free descriptors are available.
While here, also clarify some other
vq_desc_tail_idx code.
Change-Id: Ibcec2e31b2f609f2fd71f2426ce1a91728ed82f6
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/378154
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>
This was broken by commit 157969b60e.
However, this change doesn't fix any errors.
Neither vq->vq_free_cnt nor dxp->ndescs are used at the moment,
but this will change with future patches.
Fixes: 157969b60e3a ("bdev_virtio: added virtio_scsi_io_ctx")
Change-Id: I215d437c52f7060bf631bd193e7e59e2a4710368
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/378116
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: Daniel Verkamp <daniel.verkamp@intel.com>
This ensures that we don't run out of free descriptors
and queue entries.
Change-Id: I3c6960564032ffce6e6b6bf887e8c5e12fdf3205
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/378153
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
bdev_virtio I/O poller used to interfere with scan poller.
e.g. GPT read could be picked up by scan poller.
Even if such request is instantly rejected, this would cause
both GPT and scan to fail. Instead, defer bdev_register
until after the virtio_hw scan has finished.
I/O poller logic will still have to be greatly refactored to
add proper multiqueue/multibdev support. However, this patch
lifts scan processing responsibility from any I/O pollers.
Change-Id: I201de8aa0dc1db71ed836fd5e74d55604950f271
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/378064
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Currently GPT keeps the bdev it is examining open, even after determining there is no GPT on it.
It is due to spdk_gpt_base_free() not notifying bdev layer to free base bdev_part.
Additionally release of bdev module for that bdev was moved to bdev_part_free,
as the module is not claimed until bdev_part on base bdev are created.
Functional tests added to verify the changes.
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: Id75f88259267847e8ec476d19750dc1e2690f11a
Reviewed-on: https://review.gerrithub.io/378621
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>
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Reported-by: Cunyin Chang <cunyin.chang@intel.com>
Change-Id: I94ec3a5f43839f1fa4379d2f275c4e19f81e1a52
Reviewed-on: https://review.gerrithub.io/378003
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: John Kariuki <John.K.Kariuki@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
This moves towards the ability to queue ios while
a reset is in progress.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ibb8efbcec3931f966eb43030ea91945dfaaa5a77
Reviewed-on: https://review.gerrithub.io/377827
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
First, improve packing of spdk_bdev_io. Move any
fields which must be zeroed into the first cacheline.
Also change type and status from enums (needing 4 bytes)
to int16_t which is still a far bigger range than
needed.
Next, modify spdk_bdev_get_io to only zero the
first cacheline (actually a bit less).
SPDK_BDEV_IO_TYPE_INVALID is also added, making it
explicit that 0 is an invalid value for the IO type.
Previously this was somewhat inferred.
There are still additional improvements that can
be made to this area - primarily combining
spdk_bdev_get_io() and spdk_bdev_io_init() into
a single function.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I1916d6d5db02c93622b9725ec1095148e3f384d8
Reviewed-on: https://review.gerrithub.io/377799
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
spdk_bdev_io_submit() always returns 0, so the return
value is unnecessary. This eliminates a slew of checks
on that return value.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I493ecefc7b72d3b3a44bd92df70bd5c653453ff0
Reviewed-on: https://review.gerrithub.io/377627
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
We no longer support resubmitting an existing I/O to
a lower bdev. This enables accurate tracking of I/O
at each bdev layer for purposes of reset handling and
QoS.
__submit_request() in bdev.c is now only called
from spdk_bdev_io_submit(), so just collapse these
two functions together while we are here.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I1b46df42b1a6476f9dbd8c61dd3380dbd145315d
Reviewed-on: https://review.gerrithub.io/377626
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
This follows similar recent changes to bdev/split.
Future patches will move quite a bit of duplicated
code between these modules to a set of helper functions,
so doing this renaming will make those future patches
more clear.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ic67ee03d68b3080b7083ba3d91505f0ddf06da7b
Reviewed-on: https://review.gerrithub.io/376443
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Cunyin Chang <cunyin.chang@intel.com>
Reviewed-by: Ziye Yang <optimistyzy@gmail.com>
split, gpt and error all have a lot of similar code, and
are based on the presumption of one or more "partition" bdevs
on a "base" bdev. This results in quite a bit of duplicated
code between these three bdev modules. The error bdev
module does follow this same model - it just always has only
a one-to-one mapping between the error bdev and its base bdev.
As all of the modules move to allocating their own bdev_io
rather than allowing for adjusting an existing bdev_io and
resubmitting it, there will be even more duplicated code
between these modules.
So this patch adds a set of helper functions in the common
bdev library to eliminate all of this duplicated code.
This patch also moves the split module to use it - future
patches will also convert gpt and error.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Id8032331b46c32da9fca18a077580ccb274c6204
Reviewed-on: https://review.gerrithub.io/376423
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
This removes some functions that were only called from
one place, to reduce the number of functions that will
need to get moved to the common bdev library in a future
patch.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ieaaf9f1fdc907591902ce543024d547a70d00b7f
Reviewed-on: https://review.gerrithub.io/376422
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: Cunyin Chang <cunyin.chang@intel.com>
Reviewed-by: Ziye Yang <optimistyzy@gmail.com>
Make sure DEBUG macro is not defined while processing log macros,
otherwise DEBUG (which should be passed as is down to RTE_LOG and
concatenated with RTE_LOG_ prefix) gets substituted too early:
In file included from /spdk/dpdk/build/include/rte_debug.h:46:0,
from /spdk/dpdk/build/include/rte_pci.h:85,
from rte_virtio/virtio_dev.c:48:
rte_virtio/virtio_dev.c: In function ‘virtio_init_vring’:
/spdk/dpdk/build/include/rte_log.h:333:11: error: ‘RTE_LOG_1’ undeclared (first use in this function)
rte_log(RTE_LOG_ ## l, \
^
rte_virtio/virtio_logs.h:41:2: note: in expansion of macro ‘RTE_LOG’
RTE_LOG(level, PMD, "%s(): " fmt "\n", __func__, ## args)
^
rte_virtio/virtio_logs.h:42:31: note: in expansion of macro ‘PMD_INIT_LOG’
#define PMD_INIT_FUNC_TRACE() PMD_INIT_LOG(DEBUG, " >>")
^
rte_virtio/virtio_dev.c:82:2: note: in expansion of macro ‘PMD_INIT_FUNC_TRACE’
PMD_INIT_FUNC_TRACE();
While at this, change format string type specifier to the correct one.
Change-Id: I7f7bda105aa00b99bea4ea5f6a9bff268940a557
Signed-off-by: Pavel Borzenkov <pavel.borzenkov@gmail.com>
Reviewed-on: https://review.gerrithub.io/377945
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>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
I inadvertently misordered the arguments in a call to spdk_bdev_unmap.
This call is unused as of this patch, but is used by future patches.
Change-Id: I6042a3f4e05bded21b0ceb4d3b104856bcbf8c12
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/377961
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 remove callback function will remove the nvme_bdev from the global
list, we do not need to remove the device in the function bdev_nvme_destruct().
just make sure to remove it from list when the app exit.
Change-Id: I1859bfd696ed9c0ca3ac1cd8ffadfd9488df0fcd
Signed-off-by: Cunyin Chang <cunyin.chang@intel.com>
Reviewed-on: https://review.gerrithub.io/375941
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: Daniel Verkamp <daniel.verkamp@intel.com>
There's no need to keep a copy of that file.
Use system include instead.
Change-Id: I07bfb7c5de57efaa420a9fe38408c8ef230110f3
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/376350
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 field was set, but unused. virtio_hw already has
max_queues field. Since SPDK rte_virtio does not provide
any virtqueue abstraction nb_tx_queues and it's setup
functions are not needed.
Change-Id: Ib9bbdc86e6b6cede13a6cc251331590d6985db66
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/376345
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Save us some allocations and fix
scan_target() error handling path
Change-Id: I29fe18a6e592c7fd32d0486f831f6b053f59b561
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/375604
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Add new versions of all of the I/O calls that take parameters in blocks
instead of bytes. These are intended to replace the old APIs, but
we'll keep them for now to preserve compatibility.
Change-Id: I85ab665c653e8c697016c628837d49aa0c3bfcd0
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/376255
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: John Kariuki <John.K.Kariuki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
The bdev modules now take all read, write, unmap, and flush requests in
terms of blocks rather than bytes.
The public bdev APIs still accept offset and length in bytes for now.
Change-Id: I57f0955d52272f57755f0ff4dbc56721fdc2ef51
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/376037
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>
Target enumeration now uses separate poller to receive virtio
response messages. A special virtio_scsi_scan_base struct has
been added to keep track of how many targets we still have to
receive response from.
Making target scan async allows other pollers to run while
bdev_virtio still has to wait for a response.
Change-Id: Ia9af9d61fb4f880d0535df0b18aa443fff6dc9be
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/375132
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>
In preparations to make it asynchronous.
Change-Id: Iab651974841d83cf36b8e050e3905638f88ad992
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/375729
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>
bdev_virtio requires us to do SCSI inquiry for
each device and wait for a response. It currently
polls inline for this response, taking up entire
reactor exclusively. Making it async would allow
other pollers to run while bdev_virtio still has
to wait.
Change-Id: Iefc88e0a2efb5b791ffe23b2e623b7c50d759084
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/375573
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>
spdk_bdev_init_complete could have been called twice.
While here, also simplify bdev_initialize error path.
On allocation failure, bdev subsystem will now finish
with bdev_mgr->module_init_complete set to false.
Fixes: 7fefd60fabb8 ("bdev: make module init synchronous again")
Change-Id: Ia4b5d571d26beb34f0e6c0f4c7b3176215eec69f
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/375572
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>
Replace SPDK_TRACE_DEBUG with component-specific flags.
Change-Id: Iee7eafab5e6ac8713f247323a18552b5afb0e86a
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/375834
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>
This matches the name to the behavior and prepares for addition of a new
log macro for "info" log level.
Change-Id: I94ccd49face4309d3368e399528776ab140748c4
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/375833
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>
Issue each I/O to the base bdev as a new discrete I/O rather
than resubmitting the original I/O. This enables some
upcoming changes to better handle queueing and reset-related
work at the generic bdev layer, which will also deprecate
the resubmit I/O logic.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ic1caa89014981e948edfec5e8eaaacd41a73d49c
Reviewed-on: https://review.gerrithub.io/375495
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>
This makes it consistent it with other parts of this structure.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ie36488813b53ce20663c50a5c9f049c4c9723d3a
Reviewed-on: https://review.gerrithub.io/375494
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>
1) Change each split_disk to just point to a
split_base structure and drop the extra base_bdev
member. We can get the bdev from the split_base.
2) Simplify the names a bit for some of the members:
split_base::base_bdev to split_base::bdev and
split_disk::split_base to split_disk::base.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ia8621b19ad09324d939ed43d79900fecb18291e6
Reviewed-on: https://review.gerrithub.io/375493
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>
gencnt is no longer really used since there is no more
differentiation between "soft" and "hard" reset. The
original idea was that a hard reset would complete I/O
known to the bdev layer even if the underlying bdev
module had not actually completed it yet. We do not
actually support that concept, and it was a bit flawed
anyways.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ifa2c85bb474c7dd55eb7386d6cad5079f5edbccc
Reviewed-on: https://review.gerrithub.io/375484
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>
I/O requests are now allocated from bdev_io memory.
virtio_req->iov now points to raw payload, request and response
iovectors are available as separate fields. This solution
should apply for both vhost-scsi and blk.
Change-Id: I588fbdd7fc5442329aadbcb3e31b2f4a7118ec8f
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/375264
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>
Fixed various minor bugs and removed unused code.
Change-Id: I24d3f10a494b9f9c69f45e888c7e1511adc268bc
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/375004
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>
Fixes a GCC 7 warning:
rte_virtio/virtio_user/vhost_user.c: In function ‘vhost_user_setup’:
rte_virtio/virtio_user/vhost_user.c:439:46: error: ‘%s’ directive output
may be truncated writing up to 4095 bytes into a region of size 108
[-Werror=format-truncation=]
snprintf(un.sun_path, sizeof(un.sun_path), "%s", dev->path);
Change-Id: I147c9efe93cc6ce9370da6443f181f916457e3e6
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/375198
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>
Supports both PCI mode (for usage in guest VMs) and
vhost-user mode (for usage in host processes). The rte_virtio
subdirectory contains a lot of code lifted from the DPDK
virtio-net driver. Most of the PCI and vhost-user code is
reused almost exactly as-is, but the virtio code is drastically
rewritten as the DPDK code was very network specific.
Has been lightly tested with both the bdevio and bdevperf
applications in both PCI and vhost-user modes.
Still quite a bit of work needed - a list of todo
items is included in a README in the module's directory.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I85989d3de9ea89a87b719ececdb6d2ac16b77f53
Reviewed-on: https://review.gerrithub.io/374519
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>
