Commit Graph

508 Commits

Author SHA1 Message Date
Marvin Liu
6494b9a9b2 net/virtio: add vectorized packed ring Tx
Optimize packed ring Tx path like Rx path. Split Tx path into batch and
single Tx functions. Batch function is further optimized by AVX512
instructions.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-05-05 15:54:26 +02:00
Marvin Liu
77d66da838 net/virtio: add vectorized packed ring Rx
Optimize packed ring Rx path with SIMD instructions. Solution of
optimization is pretty like vhost, is that split path into batch and
single functions. Batch function is further optimized by AVX512
instructions.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-05-05 15:54:26 +02:00
Marvin Liu
57f90f8945 net/virtio: reuse packed ring functions
Move offload, xmit cleanup and packed xmit enqueue function to header
file. These functions will be reused by packed ring vectorized path.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-05-05 15:54:26 +02:00
Marvin Liu
6b7eefbceb net/virtio-user: add parameter to enable vectorized path
Add new devarg for virtio user device vectorized path selection.
By default vectorized path is disabled.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-05-05 15:54:26 +02:00
Marvin Liu
4710e16a4a net/virtio: add parameter to enable vectorized path
Previously, virtio split ring vectorized path was enabled by default.
This is not suitable for everyone because that path does not follow
virtio spec. Add new devarg for virtio vectorized path selection. By
default vectorized path is disabled.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-05-05 15:54:26 +02:00
Marvin Liu
f68a6fac2d net/virtio: enable in-order feature if negotiated
Ring initialization is different when inorder feature negotiated. This
action should dependent on negotiated feature bits.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-05-05 15:54:26 +02:00
Marvin Liu
1982462ead net/virtio: add Rx free threshold setting
Introduce free threshold setting in Rx queue, its default value is 32.
Limit the threshold size to multiple of four as only vectorized packed
Rx function will utilize it. Virtio driver will rearm Rx queue when
more than rx_free_thresh descs were dequeued.

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-05-05 15:54:26 +02:00
Ori Kam
f5bf02df31 eal/ppc: fix bool type after altivec include
The AltiVec header file breaks boolean type. [1] [2]

Currently the workaround was located only in mlx5 device.
Adding the trace module caused this issue to appear again, due to
order of includes, it keeps overriding the local fix.

This patch solves this issue by resetting the bool type, immediately
after it is being changed.

[1] https://mails.dpdk.org/archives/dev/2018-August/110281.html

[2]
In file included from
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool_trace_fp.h:18:0,
                 from
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool.h:54,
                 from
dpdk/drivers/common/mlx5/mlx5_common_mr.c:7:
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point.h: In
function '__rte_trace_point_fp_is_enabled':
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point.h:226:2:
error: incompatible types when returning type 'int' but '__vector __bool
int' was expected
  return false;
  ^
In file included from
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point.h:281:0,
                 from
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool_trace_fp.h:18,
                 from
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool.h:54,
                 from
dpdk/drivers/common/mlx5/mlx5_common_mr.c:7:
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool_trace_fp.h:
In function 'rte_mempool_trace_ops_dequeue_bulk':
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point_provider.h:104:6:
error: wrong type argument to unary exclamation mark
  if (!__rte_trace_point_fp_is_enabled()) \
      ^
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point.h:49:2:
note: in expansion of macro '__rte_trace_point_emit_header_fp'
  __rte_trace_point_emit_header_##_mode(&__##_tp); \
  ^
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point.h:99:2:
note: in expansion of macro '__RTE_TRACE_POINT'
  __RTE_TRACE_POINT(fp, tp, args, __VA_ARGS__)
  ^
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool_trace_fp.h:20:1:
note: in expansion of macro 'RTE_TRACE_POINT_FP'
 RTE_TRACE_POINT_FP(
 ^
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool_trace_fp.h:
In function 'rte_mempool_trace_ops_dequeue_contig_blocks':
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point_provider.h:104:6:
error: wrong type argument to unary exclamation mark
  if (!__rte_trace_point_fp_is_enabled()) \
      ^
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point.h:49:2:
note: in expansion of macro '__rte_trace_point_emit_header_fp'
  __rte_trace_point_emit_header_##_mode(&__##_tp); \
  ^
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point.h:99:2:
note: in expansion of macro '__RTE_TRACE_POINT'
  __RTE_TRACE_POINT(fp, tp, args, __VA_ARGS__)
  ^
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool_trace_fp.h:29:1:
note: in expansion of macro 'RTE_TRACE_POINT_FP'
 RTE_TRACE_POINT_FP(
 ^
dpdk/ppc_64-power8-linux-gcc/include/rte_mempool_trace_fp.h:
In function 'rte_mempool_trace_ops_enqueue_bulk':
dpdk/ppc_64-power8-linux-gcc/include/rte_trace_point_provider.h:104:6:
error: wrong type argument to unary exclamation mark
  if (!__rte_trace_point_fp_is_enabled()) \

Fixes: 725f5dd0bf ("net/mlx5: fix build on PPC64")

Signed-off-by: Ori Kam <orika@mellanox.com>
Signed-off-by: David Christensen <drc@linux.vnet.ibm.com>
Tested-by: David Christensen <drc@linux.vnet.ibm.com>
Tested-by: Raslan Darawsheh <rasland@mellanox.com>
Acked-by: Matan Azrad <matan@mellanox.com>
2020-05-06 11:45:13 +02:00
Ivan Dyukov
1357b4b362 net/virtio: support Virtio link speed feature
This patch adds a support of VIRTIO_NET_F_SPEED_DUPLEX feature
for virtio driver.

There are two ways to specify speed of the link:
   'speed' devarg
   negotiate speed from qemu via VIRTIO_NET_F_SPEED_DUPLEX
The highest priority is devarg. If devarg is not specified,
drivers tries to negotiate it from qemu.

Signed-off-by: Ivan Dyukov <i.dyukov@samsung.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-04-21 13:57:09 +02:00
Ivan Dyukov
b0db4bea00 net/virtio-user: adding link speed parameter
virtio driver already parses speed devarg. virtio-user should add
it to list of valid devargs and call eth_virtio_dev_init function
which init speed value.

eth_virtio_dev_init already is called from virtio_user_pmd_probe
function. The only change is required to enable speed devargs:
adding speed to list of valid devargs.

Signed-off-by: Ivan Dyukov <i.dyukov@samsung.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-04-21 13:57:09 +02:00
Ivan Dyukov
bc5b6c11bd net/virtio-user: fix devargs parsing
strtoull returns 0 if it fails to parse input string. It's ignored
in get_integer_arg.

This patch handles error cases for strtoull function.

Fixes: ce2eabdd43 ("net/virtio-user: add virtual device")
Cc: stable@dpdk.org

Signed-off-by: Ivan Dyukov <i.dyukov@samsung.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-04-21 13:57:09 +02:00
Ivan Dyukov
49119e3855 net/virtio: add link speed parameter
Some applications like pktgen use link speed to calculate
transmission rate. It limits outcome traffic to hardcoded 10G.

This patch adds speed devarg which allows to configure
link speed of virtio device.

Signed-off-by: Ivan Dyukov <i.dyukov@samsung.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-04-21 13:57:09 +02:00
Ivan Dyukov
44d7b2e87b net/virtio: refactor devargs parsing
Refactor vdpa specific devargs parsing to more generic way.

Signed-off-by: Ivan Dyukov <i.dyukov@samsung.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-04-21 13:57:09 +02:00
Xiaolong Ye
f6ac14f13b net/virtio: fix outdated comment
Fix comment that is no more correct as the code evolved.

Fixes: 9470427c88 ("net/virtio: do not store PCI device pointer at shared memory")
Cc: stable@dpdk.org

Signed-off-by: Xiaolong Ye <xiaolong.ye@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-04-21 13:57:07 +02:00
Stephen Hemminger
88fa5bb6b9 net/virtio: do not use PMD log type
Virtio driver has its own logtype and should not use legacy
PMD logtype.

Fixes: 32c118fd00 ("virtio: free mbuf's with threshold")
Fixes: e5f456a98d ("net/virtio: support in-order Rx and Tx")
Fixes: 1c8489da56 ("net/virtio-user: fix multi-process support")
Cc: stable@dpdk.org

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
2020-04-21 13:57:05 +02:00
Thomas Monjalon
ce6427ddca replace cold attributes
The new macro __rte_cold, for compiler hinting,
is now used where appropriate for consistency.

Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: David Christensen <drc@linux.vnet.ibm.com>
2020-04-16 18:30:58 +02:00
Thomas Monjalon
ef5baf3486 replace packed attributes
There is a common macro __rte_packed for packing structs,
which is now used where appropriate for consistency.

Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
2020-04-16 18:16:46 +02:00
Thomas Monjalon
f35e5b3e07 replace alignment attributes
There is a common macro __rte_aligned for alignment,
which is now used where appropriate for consistency.

Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: David Christensen <drc@linux.vnet.ibm.com>
2020-04-16 18:16:18 +02:00
Pavan Nikhilesh
acec04c4b2 build: disable experimental API check internally
Remove setting ALLOW_EXPERIMENTAL_API individually for each Makefile and
meson.build. Instead, enable ALLOW_EXPERIMENTAL_API flag across app, lib
and drivers.
This changes reduces the clutter across the project while still
maintaining the functionality of ALLOW_EXPERIMENTAL_API i.e. warning
external applications about experimental API usage.

Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Acked-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Signed-off-by: David Marchand <david.marchand@redhat.com>
2020-04-14 16:22:34 +02:00
Thomas Monjalon
dd45ed7e74 build: rename ppc sub-directories
The directories ppc_64 are renamed as ppc in
	- config/
	- lib/librte_eal/common/arch/
	- lib/librte_eal/common/include/arch/

The EAL directories are not really renamed, but symbolically linked,
because they will be moved with their new name in the next commits.

If ppc_32 needs to be supported, it can be in the same directory.
The arch directories arm and x86 are already covering both 32 and 64-bit
sub-architectures.

Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: David Christensen <drc@linux.vnet.ibm.com>
Acked-by: David Marchand <david.marchand@redhat.com>
2020-03-31 13:08:55 +02:00
Tiwei Bie
9277125731 net/virtio-user: do not reset virtqueues for split ring
Add missing braces to avoid resetting virtqueues unconditionally
during reconnection.

Fixes: 6ebbf4109f ("net/virtio-user: fix packed ring server mode")
Cc: stable@dpdk.org

Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-02-05 11:47:18 +01:00
Xuan Ding
6ebbf4109f net/virtio-user: fix packed ring server mode
This patch fixes the situation where data path does not work properly
when vhost reconnects to virtio in server mode with packed ring.

Currently, virtio and vhost share memory of vring. For split ring, vhost
can read the status of descriptors directly from the available ring and
the used ring during reconnection. Therefore, the data path can
continue.

But for packed ring, when reconnecting to virtio, vhost cannot get the
status of descriptors via the descriptor ring. By resetting Tx
and Rx queues, the data path can restart from the beginning.

Fixes: 4c3f5822eb ("net/virtio: add packed virtqueue defines")
Cc: stable@dpdk.org

Signed-off-by: Xuan Ding <xuan.ding@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-01-17 19:46:26 +01:00
Yunjian Wang
1ad2f882b3 net/virtio-user: check tap offload setting failure
The function vhost_kernel_tap_set_offload() could return errors,
the return value need to be checked. And there is no need to fail
when error is -ENOTSUP.

Fixes: 1db4d2330b ("net/virtio-user: check negotiated features before set")
Cc: stable@dpdk.org

Signed-off-by: Yunjian Wang <wangyunjian@huawei.com>
Reviewed-by: Tiwei Bie <tiwei.bie@intel.com>
2020-01-17 19:46:26 +01:00
Cheng Jiang
7e72f3ec1a net/virtio: support packed ring notification data
This patch supports the feature that the driver passes extra data
(besides identifying the virtqueue) in its device notifications,
expanding the notifications to include the avail index and avail
wrap counter.

Signed-off-by: Cheng Jiang <cheng1.jiang@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-01-17 19:46:26 +01:00
Tiwei Bie
47ac9661b6 net/virtio-user: do not close tap when disabling queue pairs
Do not close the tap fds when disabling queue pairs, instead,
we just need to unbind the backend. Otherwise, tap port can be
destroyed unexpectedly.

Fixes: e3b434818b ("net/virtio-user: support kernel vhost")
Cc: stable@dpdk.org

Reported-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2020-01-17 19:46:26 +01:00
Stephen Hemminger
b56c12acf2 net/virtio-user: check file descriptor before closing
Valgrind complains that virtio_user is calling close(-1).
Fix this by adding check in virtio that is similar to existing code.

Fixes: 37a7eb2ae8 ("net/virtio-user: add device emulation layer")
Cc: stable@dpdk.org

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Reviewed-by: Tiwei Bie <tiwei.bie@intel.com>
2020-01-17 19:46:02 +01:00
Tiwei Bie
2286291df1 net/virtio-user: drop attribute unused for memory callback
The "addr" param has been used since the event callbacks are enabled
for external memory. So the "__rte_unused" should be dropped.

Besides, slightly refine the coding style by consistently assuming
tabs are 8 characters.

Fixes: f32c7c9de9 ("malloc: enable event callbacks for external memory")
Cc: stable@dpdk.org

Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-11-26 18:05:15 +01:00
Pawel Modrak
85ff364f3b build: align symbols with global ABI version
Merge all versions in linker version script files to DPDK_20.0.

This commit was generated by running the following command:

:~/DPDK$ buildtools/update-abi.sh 20.0

Signed-off-by: Pawel Modrak <pawelx.modrak@intel.com>
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Acked-by: Thomas Monjalon <thomas@monjalon.net>
2019-11-20 23:05:39 +01:00
Anatoly Burakov
fbaf943887 build: remove individual library versions
Since the library versioning for both stable and experimental ABI's is
now managed globally, the LIBABIVER and version variables no longer
serve any useful purpose, and can be removed.

The replacement in Makefiles was done using the following regex:

	^(#.*\n)?LIBABIVER\s*:=\s*\d+\n(\s*\n)?

(LIBABIVER := numbers, optionally preceded by a comment and optionally
succeeded by an empty line)

The replacement for meson files was done using the following regex:

	^(#.*\n)?version\s*=\s*\d+\n(\s*\n)?

(version = numbers, optionally preceded by a comment and optionally
succeeded by an empty line)

[David]: those variables are manually removed for the files:
- drivers/common/qat/Makefile
- lib/librte_eal/meson.build
[David]: the LIBABIVER is restored for the external ethtool example
library.

Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Thomas Monjalon <thomas@monjalon.net>
2019-11-20 23:05:39 +01:00
Marvin Liu
a76552d48f net/virtio-user: fix setting filters
As doc mentioned, Rx/Mac/vlan filters are all supported by best effort.
These control commands should return success.

Fixes: f9b9d1a557 ("net/virtio-user: add multiple queues in device emulation")
Cc: stable@dpdk.org

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-11-08 23:15:05 +01:00
Marvin Liu
df42dde55b net/virtio: do not limit packed ring size
Virtio spec only set rule that packed ring maximum size is up to 2^15
entries. Should not limit packed ring size to power of two.

Fixes: aea29aa5d3 ("net/virtio: enable packed virtqueues by default")
Cc: stable@dpdk.org

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-11-08 23:15:05 +01:00
Andrew Rybchenko
1526dd0532 net/virtio: fix Tx checksum offloads
Missing parenthesis around expression before type cast to struct
virtio_net_hdr pointer makes the arithmetic to be in
sizeof(struct virtio_net_hdr) units.

Use rte_pktmbuf_mtod_offset() to fix the problem.

Type of head_size is changed to signed since some compilers bark
on unary minus applied to unsigned.

Fixes: 1ae55ad38e ("net/virtio: fix mbuf data and packet length mismatch")
Cc: stable@dpdk.org

Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-10-25 19:23:23 +02:00
Andrew Rybchenko
9fc963acc2 net/virtio: reject unsupported Tx multi-queue modes
This driver supports none of DCB or VMDQ modes, therefore must
check and return error if configured incorrectly.

Fixes: c1f86306a0 ("virtio: add new driver")
Cc: stable@dpdk.org

Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-10-25 19:23:22 +02:00
Dilshod Urazov
13b3137f3b net/virtio: reject unsupported Rx multi-queue modes
This driver supports none of DCB, RSS or VMDQ modes, therefore must
check and return error if configured incorrectly.

Virtio can distribute Rx packets across multi-queue, but there is
no controls (algorithm, redirection table, hash function) except
number of Rx queues and ETH_MQ_RX_NONE is the best fit meaning
no method is enforced on how to route packets to MQs.

Fixes: c1f86306a0 ("virtio: add new driver")
Cc: stable@dpdk.org

Signed-off-by: Dilshod Urazov <dilshod.urazov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-10-25 19:23:22 +02:00
Dilshod Urazov
c0e03310ee net/virtio: reject deferred Tx start
Deferred start Tx queue is not supported by the driver.

Fixes: 0748be2cf9 ("ethdev: queue start and stop")
Cc: stable@dpdk.org

Signed-off-by: Dilshod Urazov <dilshod.urazov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-10-25 19:23:22 +02:00
Dilshod Urazov
955d7c1f03 net/virtio: reject deferred Rx start
Deferred start Rx queue is not supported by the driver.

Fixes: 0748be2cf9 ("ethdev: queue start and stop")
Cc: stable@dpdk.org

Signed-off-by: Dilshod Urazov <dilshod.urazov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-10-25 19:23:22 +02:00
Andrew Rybchenko
be048a1aaa net/virtio: fix descriptor addressed in Tx
Previous fix removes usage of rte_pktmbuf_prepend() to get pointer
to virtio net header which changes mbuf data_off and data_len.
Size of virtio net header is added to segment length when Tx descriptor
is composed, but segment address (calculated using data_off) is not
adjusted to take size of virtio net header into account.

Fixes: 1ae55ad38e ("net/virtio: fix mbuf data and packet length mismatch")
Cc: stable@dpdk.org

Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Tiwei Bie <tiwei.bie@intel.com>
2019-10-23 16:43:09 +02:00
David Christensen
a44c4e1e3a net/virtio: fix Rx AltiVec path by getting all packets
The loop to read packets does not take all packets as the number of
available packets (nb_used) is decremented in the loop.

Fixes: 52b5a707e6 ("net/virtio: add Altivec Rx")

Signed-off-by: David Christensen <drc@linux.vnet.ibm.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-10-23 16:43:09 +02:00
David Christensen
86d1afa12c net/virtio: fix Rx stats on AltiVec path
With vectorized functions, only the rx stats for number of packets is
incremented.
Update also the other statistics.

Fixes: 52b5a707e6 ("net/virtio: add Altivec Rx")

Signed-off-by: David Christensen <drc@linux.vnet.ibm.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-10-23 16:43:09 +02:00
Joyce Kong
2c661d418e net/virtio: improve perf via one-way barriers on used flag
In case VIRTIO_F_ORDER_PLATFORM(36) is not negotiated, then the frontend
and backend are assumed to be implemented in software, that is they can
run on identical CPUs in an SMP configuration.
Thus a weak form of memory barriers like rte_smp_r/wmb, other than
rte_cio_r/wmb, is sufficient for this case(vq->hw->weak_barriers == 1)
and yields better performance.
For the above case, this patch helps yielding even better performance
by replacing the two-way barriers with C11 one-way barriers for used
flags in packed ring.

Signed-off-by: Joyce Kong <joyce.kong@arm.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Reviewed-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-10-23 16:43:09 +02:00
Joyce Kong
6094557de0 net/virtio: improve perf via one-way barrier on avail flag
In case VIRTIO_F_ORDER_PLATFORM(36) is not negotiated, then the frontend
and backend are assumed to be implemented in software, that is they can
run on identical CPUs in an SMP configuration.
Thus a weak form of memory barriers like rte_smp_r/wmb, other than
rte_cio_r/wmb, is sufficient for this case(vq->hw->weak_barriers == 1)
and yields better performance.
For the above case, this patch helps yielding even better performance
by replacing the two-way barriers with C11 one-way barriers for avail
flags in packed ring.

Meanwhile, a read barrier is required to ensure ordering between
descriptor's flags and content reads [1]. With C11, load-acquire can
enforce the ordering instead of rmb barrier.

[1] https://patchwork.dpdk.org/patch/49109/

Signed-off-by: Joyce Kong <joyce.kong@arm.com>
Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Reviewed-by: Phil Yang <phil.yang@arm.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-10-23 16:43:09 +02:00
David Marchand
da4880609c net/virtio: do not require IO permissions
On x86, iopl permissions are only available to root user (or users that
have the CAP_SYS_RAWIO capability).
But those permissions are only needed when the virtio devices accesses
are done with inb/outb instructions, which is when the device is bound
to a UIO kernel module.

So far, the virtio driver was refusing to register based on the check
on IO permissions.
This check does not make sense when binding the device to vfio.

Now that the check on IO permissions has been abstracted in the ioport
API, we can remove it on virtio side.

We still need to call rte_eal_iopl_init() in the virtio constructor so
that the interrupt thread inherits this permission in the case it could
be used with UIO later.

Signed-off-by: David Marchand <david.marchand@redhat.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Tiwei Bie <tiwei.bie@intel.com>
2019-10-25 11:54:10 +02:00
Marvin Liu
1ae55ad38e net/virtio: fix mbuf data and packet length mismatch
If reserve virtio header room by function rte_pktmbuf_prepend, both
segment data length and packet length of mbuf will be increased.
Data length will be equal to descriptor length, while packet length
should be decreased as virtio-net header won't be taken into packet.
Thus will cause mismatch in mbuf structure. Fix this issue by access
mbuf data directly and increase descriptor length if it is needed.

Fixes: 58169a9c81 ("net/virtio: support Tx checksum offload")
Fixes: 892dc798fa ("net/virtio: implement Tx path for packed queues")
Fixes: 4905ed3a52 ("net/virtio: optimize Tx enqueue for packed ring")
Fixes: e5f456a98d ("net/virtio: support in-order Rx and Tx")
Cc: stable@dpdk.org

Reported-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-10-07 15:00:57 +02:00
Thibaut Collet
016f56b5bc net/virtio: get all pending Rx packets in vectorized paths
The loop to read packets does not take all packets as the number of
available packets (nb_used) is decremented in the loop.
Take all available packets provides a performance improvement of 3%.

Fixes: fc3d66212f ("virtio: add vector Rx")
Cc: stable@dpdk.org

Signed-off-by: Thibaut Collet <thibaut.collet@6wind.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-10-07 15:00:57 +02:00
Thibaut Collet
6958e40dbc net/virtio: fix Rx stats with vectorized functions
With vectorized functions, only the rx stats for number of packets is
incremented.
Update also the other statistics.
Performance impact is about 2%

Fixes: fc3d66212f ("virtio: add vector Rx")
Cc: stable@dpdk.org

Signed-off-by: Thibaut Collet <thibaut.collet@6wind.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-10-07 15:00:57 +02:00
Marvin Liu
e76097f848 net/virtio: cleanup on demand when in-order Tx
Check whether space are enough before burst enqueue operation. If more
space is needed, will try to clean up used descriptors for space on
demand. It can give more chances to free used descriptors, thus will
help RFC2544 performance. Also deduct failed xmit packets from total
xmit number.

Fixes: e5f456a98d ("net/virtio: support in-order Rx and Tx")
Cc: stable@dpdk.org

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-10-07 15:00:57 +02:00
Marvin Liu
01689b1ed4 net/virtio: update stats when in-order Tx done
When doing xmit in-order enqueue, packets are buffered and then flushed
into avail ring. Buffered packets can be dropped due to insufficient
space. Moving stats update action just after successful avail ring
updates can guarantee correctness.

Fixes: e5f456a98d ("net/virtio: support in-order Rx and Tx")
Cc: stable@dpdk.org

Signed-off-by: Marvin Liu <yong.liu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
2019-10-07 15:00:57 +02:00
Ivan Ilchenko
ca041cd44f ethdev: change allmulticast callbacks to return status
Enabling/disabling of allmulticast mode is not always successful and
it should be taken into account to be able to handle it properly.

When correct return status is unclear from driver code, -EAGAIN is used.

Signed-off-by: Ivan Ilchenko <ivan.ilchenko@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Acked-by: Hyong Youb Kim <hyonkim@cisco.com>
2019-10-07 15:00:55 +02:00
Igor Romanov
9970a9ad07 ethdev: make stats and xstats reset callbacks return int
Change return value of the callbacks from void to int. Make
implementations across all drivers return negative errno
values in case of error conditions.

Both callbacks are updated together because a large number of
drivers assign the same function to both callbacks.

Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
2019-10-07 15:00:54 +02:00
Andrew Rybchenko
9039c81257 ethdev: change promiscuous callbacks to return status
Enabling/disabling of promiscuous mode is not always successful and
it should be taken into account to be able to handle it properly.

When correct return status is unclear from driver code, -EAGAIN is used.

Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Acked-by: Matan Azrad <matan@mellanox.com>
Acked-by: Hyong Youb Kim <hyonkim@cisco.com>
2019-10-07 15:00:54 +02:00