RE: [PATCH v2 2/2] net/intel: add Tx time queue
On Wed, Feb 12, 2025 at 09:47:11PM +, Soumyadeep Hore wrote:
> Enabling Tx timestamp queue for supporting Tx time based scheduling of
> packets.
>
Can you provide more details of this feature and how it can be used, how it is
enabled/disabled etc.
See also comments inline below.
/Bruce
> Signed-off-by: Soumyadeep Hore
> ---
> drivers/net/intel/common/tx.h | 5 +
> drivers/net/intel/ice/base/ice_lan_tx_rx.h | 1 +
> drivers/net/intel/ice/ice_ethdev.h | 1 +
> drivers/net/intel/ice/ice_rxtx.c | 174 +
> drivers/net/intel/ice/ice_rxtx.h | 5 +
> 5 files changed, 186 insertions(+)
>
> diff --git a/drivers/net/intel/common/tx.h
> b/drivers/net/intel/common/tx.h index d9cf4474fc..f3777fa9e7 100644
> --- a/drivers/net/intel/common/tx.h
> +++ b/drivers/net/intel/common/tx.h
> @@ -35,6 +35,7 @@ struct ci_tx_queue {
> volatile struct i40e_tx_desc *i40e_tx_ring;
> volatile struct iavf_tx_desc *iavf_tx_ring;
> volatile struct ice_tx_desc *ice_tx_ring;
> + volatile struct ice_ts_desc *ice_tstamp_ring;
This code looks a bit strange to me, can you please check my understanding of
it is correct.
This is a union, so the time stamp ring here is replacing the whole descriptor
ring for the queue? Therefore, we will have some queues which have regular
descriptors and others which have only timestamp descriptors.
Is that correct?
Another minor point is that this union has the elements in alphabetical order,
so ice_ts_desc needs to come before ice_tx_desc.
> volatile union ixgbe_adv_tx_desc *ixgbe_tx_ring;
> };
> volatile uint8_t *qtx_tail; /* register address of tail */
> @@ -76,6 +77,10 @@ struct ci_tx_queue {
> union {
> struct { /* ICE driver specific values */
> uint32_t q_teid; /* TX schedule node id. */
> + uint16_t nb_tstamp_desc;/* number of Timestamp
> descriptors */
> + volatile uint8_t *tstamp_tail; /* value of timestamp
> tail register */
> + rte_iova_t tstamp_ring_dma; /* Timestamp ring DMA
> address */
> + uint16_t next_tstamp_id;
You are adding lots of holes into the structure here, please reorder the fields
to reduce the space used by the structure.
Also, do you need the field tstamp_tail? Since ice_tstamp_ring is replacing
ice_tx_ring in the union above, you should be able to just reuse the existing
tail register for this, no?
Similarly for tstamp_ring_dma, can the existing dma address field not be used.
OVerall, I think rather than expanding out our common tx queue structure, it
may be better to have the queue structure hold a pointer to another separate tx
timestamp structure, allocated separately.
Hi Bruce only one Tx ring is available for multiple Tx queues so we need
separate DMA and tail.
> };
> struct { /* I40E driver specific values */
> uint8_t dcb_tc;
> diff --git a/drivers/net/intel/ice/base/ice_lan_tx_rx.h
> b/drivers/net/intel/ice/base/ice_lan_tx_rx.h
> index 940c6843d9..edd1137114 100644
> --- a/drivers/net/intel/ice/base/ice_lan_tx_rx.h
> +++ b/drivers/net/intel/ice/base/ice_lan_tx_rx.h
> @@ -1279,6 +1279,7 @@ struct ice_ts_desc {
> #define ICE_SET_TXTIME_MAX_Q_AMOUNT 127
> #define ICE_OP_TXTIME_MAX_Q_AMOUNT 2047
> #define ICE_TXTIME_FETCH_TS_DESC_DFLT8
> +#define ICE_TXTIME_FETCH_PROFILE_CNT 16
>
> /* Tx Time queue context data
> *
This base code update, adding a define, should be in the previous patch where
all the other base code defines are added.
> diff --git a/drivers/net/intel/ice/ice_ethdev.h
> b/drivers/net/intel/ice/ice_ethdev.h
> index afe8dae497..9649456771 100644
> --- a/drivers/net/intel/ice/ice_ethdev.h
> +++ b/drivers/net/intel/ice/ice_ethdev.h
> @@ -299,6 +299,7 @@ struct ice_vsi {
> uint8_t enabled_tc; /* The traffic class enabled */
> uint8_t vlan_anti_spoof_on; /* The VLAN anti-spoofing enabled */
> uint8_t vlan_filter_on; /* The VLAN filter enabled */
> + uint8_t enabled_txpp; /* TXPP support enabled */
While I realise that "enabled_txpp" matches the "enabled_tc" variable above, it
would read better as "txpp_enabled". You could also have it align to the
previous two members, perhaps: would it work calling it "txpp_on".
> /* information about rss configuration */
> u32 rss_key_size;
> u32 rss_lut_size;
> diff --git a/drivers/net/intel/ice/ice_rxtx.c
> b/drivers/net/intel/ice/ice_rxtx.c
> index 8dd8644b16..f043ae3aa6 100644
> --- a/drivers/net/intel/ice/ice_rxtx.c
> +++ b/drivers/net/intel/ice/ice_rxtx.c
> @@ -5,6 +5,7 @@
> #include
> #include
> #include
> +#include
>
> #include "ice_rxtx.h"
> #include "ice_rxtx_vec_common.h"
> @@ -741,6 +742,87 @@ ice_rx_queue_stop(struct rte_eth_dev *dev, uint16_t
> rx_queue_id)
>
RedHat QE's test result against DPDK release candidate 25.03-rc1
I tested below 18 scenarios on RHEL 9.4 hosts and didn't find any new dpdk issues. - VM with device assignment(PF) throughput testing(1G hugepage size): PASS - VM with device assignment(PF) throughput testing(2M hugepage size) : PASS - VM with device assignment(VF) throughput testing: PASS - PVP (host dpdk testpmd as vswitch) 1Q: throughput testing: PASS - PVP vhost-user 2Q throughput testing: PASS - PVP vhost-user 1Q - cross numa node throughput testing: PASS - VM with vhost-user 2 queues throughput testing: PASS - vhost-user reconnect with dpdk-client, qemu-server(qemu reconnect): PASS - vhost-user reconnect with dpdk-client, qemu-server(ovs reconnect): PASS - PVP reconnect with dpdk-client, qemu-server: PASS - PVP 1Q live migration testing: PASS - PVP 1Q cross numa node live migration testing: PASS - VM with ovs+dpdk+vhost-user 1Q live migration testing: PASS - VM with ovs+dpdk+vhost-user 1Q live migration testing (2M): PASS - VM with ovs+dpdk+vhost-user 2Q live migration testing: PASS - VM with ovs+dpdk+vhost-user 4Q live migration testing: PASS - Host PF + DPDK testing: PASS - Host VF + DPDK testing: PASS Test Versions: - qemu-kvm-8.2.0 - kernel 5.14 - libvirt 10.0 - ovs 3.3 - git log Commit e5176f23ae8b31437c3e5eb875c81f95bf3a9942 Author: Thomas Monjalon Date: Wed Feb 12 17:13:27 2025 +0100 version: 25.03-rc1 Signed-off-by: Thomas Monjalon < tho...@monjalon.net> - Test device : X540-AT2 NIC(ixgbe, 10G) Tested-by: Yanghang Liu
RE: [EXTERNAL] [PATCH] net/netvsc: add stats counters from VF
> -Original Message-
> From: lon...@linuxonhyperv.com
> Sent: Wednesday, February 19, 2025 5:13 AM
> To: Stephen Hemminger ; Wei Hu
>
> Cc: dev@dpdk.org; Long Li ; sta...@dpdk.org
> Subject: [EXTERNAL] [PATCH] net/netvsc: add stats counters from VF
>
> From: Long Li
>
> The netvsc driver should add per-queue and rx_nombuf counters from VF.
>
> Fixes: 4e9c73e96e83 ("net/netvsc: add Hyper-V network device")
> Cc: sta...@dpdk.org
> Signed-off-by: Long Li
> ---
> drivers/net/netvsc/hn_ethdev.c | 10 +-
> 1 file changed, 5 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/net/netvsc/hn_ethdev.c b/drivers/net/netvsc/hn_ethdev.c
> index 988c51928d..069d603d90 100644
> --- a/drivers/net/netvsc/hn_ethdev.c
> +++ b/drivers/net/netvsc/hn_ethdev.c
> @@ -824,8 +824,8 @@ static int hn_dev_stats_get(struct rte_eth_dev *dev,
> stats->oerrors += txq->stats.errors;
>
> if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
> - stats->q_opackets[i] = txq->stats.packets;
> - stats->q_obytes[i] = txq->stats.bytes;
> + stats->q_opackets[i] += txq->stats.packets;
> + stats->q_obytes[i] += txq->stats.bytes;
> }
> }
>
> @@ -841,12 +841,12 @@ static int hn_dev_stats_get(struct rte_eth_dev
> *dev,
> stats->imissed += rxq->stats.ring_full;
>
> if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
> - stats->q_ipackets[i] = rxq->stats.packets;
> - stats->q_ibytes[i] = rxq->stats.bytes;
> + stats->q_ipackets[i] += rxq->stats.packets;
> + stats->q_ibytes[i] += rxq->stats.bytes;
> }
> }
>
> - stats->rx_nombuf = dev->data->rx_mbuf_alloc_failed;
> + stats->rx_nombuf += dev->data->rx_mbuf_alloc_failed;
> return 0;
> }
>
> --
> 2.34.1
Reviewed-by: Wei Hu
[PATCH v2 4/5] net: update use of versioning macros
When compiling with MSVC the error below shows up due to function
versioning:
../lib/net/rte_net_crc.c(418): error C2061: syntax error:
identifier '__attribute__'
MSVC allows alias function names to be exported, but the mechanism is
different than the one used by gcc. It was considered to enhance the logic
in the existing version.map files but that file is also passed to other
tools on Linux, making this challenging. A simpler approach is to have an
optional version.map file to be used only when Microsoft's linker is to be
used. This optional map file is only necessary for libraries that have
versioned code.
Signed-off-by: Andre Muezerie
---
lib/net/rte_net_crc.c | 2 ++
lib/net/version_ms_linker.map | 23 +++
2 files changed, 25 insertions(+)
create mode 100644 lib/net/version_ms_linker.map
diff --git a/lib/net/rte_net_crc.c b/lib/net/rte_net_crc.c
index 2fb3eec231..5b41085a36 100644
--- a/lib/net/rte_net_crc.c
+++ b/lib/net/rte_net_crc.c
@@ -418,6 +418,7 @@ BIND_DEFAULT_SYMBOL(rte_net_crc_set_alg, _v26, 26);
MAP_STATIC_SYMBOL(struct rte_net_crc *rte_net_crc_set_alg(
enum rte_net_crc_alg alg, enum rte_net_crc_type type),
rte_net_crc_set_alg_v26);
+MAP_STATIC_BASE_SYMBOL(rte_net_crc_set_alg, rte_net_crc_set_alg_v26)
void rte_net_crc_free(struct rte_net_crc *crc)
{
@@ -449,6 +450,7 @@ BIND_DEFAULT_SYMBOL(rte_net_crc_calc, _v26, 26);
MAP_STATIC_SYMBOL(uint32_t rte_net_crc_calc(const struct rte_net_crc *ctx,
const void *data, const uint32_t data_len),
rte_net_crc_calc_v26);
+MAP_STATIC_BASE_SYMBOL(rte_net_crc_calc, rte_net_crc_calc_v26)
/* Call initialisation helpers for all crc algorithm handlers */
RTE_INIT(rte_net_crc_init)
diff --git a/lib/net/version_ms_linker.map b/lib/net/version_ms_linker.map
new file mode 100644
index 00..9503c669d1
--- /dev/null
+++ b/lib/net/version_ms_linker.map
@@ -0,0 +1,23 @@
+DPDK_25 {
+ global:
+
+ rte_eth_random_addr;
+ rte_ether_format_addr;
+ rte_ether_unformat_addr;
+ rte_net_crc_calc=rte_net_crc_calc_v26;
+ rte_net_crc_free;
+ rte_net_crc_set_alg=rte_net_crc_set_alg_v26;
+ rte_net_get_ptype;
+ rte_net_make_rarp_packet;
+ rte_net_skip_ip6_ext;
+
+ local: *;
+};
+
+DPDK_26 {
+ global:
+
+ rte_net_crc_calc=rte_net_crc_calc_v26;
+ rte_net_crc_set_alg=rte_net_crc_set_alg_v26;
+
+} DPDK_25;
--
2.48.1.vfs.0.0
[PATCH v2 2/5] buildtools: update map_to_win.py to use optional map file
When compiling with MSVC the error below shows up due to function
versioning:
../lib/net/rte_net_crc.c(418): error C2061: syntax error:
identifier '__attribute__'
MSVC allows alias function names to be exported, but the mechanism is
different than the one used by gcc. It was considered to enhance the logic
in the existing version.map files but that file is also passed to other
tools on Linux, making this challenging. A simpler approach is to have an
optional version.map file to be used only when Microsoft's linker is to be
used. This optional map file is only necessary for libraries that have
versioned code.
Script map_to_win.py was updated to look for this optional file and use it
when available.
Signed-off-by: Andre Muezerie
---
buildtools/map_to_win.py | 16 +++-
1 file changed, 15 insertions(+), 1 deletion(-)
diff --git a/buildtools/map_to_win.py b/buildtools/map_to_win.py
index aa1752cacd..f75db3842b 100644
--- a/buildtools/map_to_win.py
+++ b/buildtools/map_to_win.py
@@ -2,6 +2,7 @@
# SPDX-License-Identifier: BSD-3-Clause
# Copyright(c) 2019 Intel Corporation
+import os
import sys
@@ -27,7 +28,20 @@ def main(args):
# This works taking indented lines only which end with a ";" and which don't
# have a colon in them, i.e. the lines defining functions only.
else:
-with open(args[1]) as f_in:
+input_map = args[1]
+
+# When an optional map file for Microsoft's linker exists, use it.
Function aliases can
+# be used in these optional files. They end up in the exports.def file:
+# EXPORTS
+# rte_net_crc_set_alg=rte_net_crc_set_alg_v26;
+# More details about the export file syntax accepted by Microsoft's
linker can be found
+# here:
+#
https://learn.microsoft.com/en-us/cpp/build/reference/exports?view=msvc-170
+optional_input_map = input_map.removesuffix('.map') + '_ms_linker.map'
+if os.path.exists(optional_input_map):
+input_map = optional_input_map
+
+with open(input_map) as f_in:
functions = [ln[:-2] + '\n' for ln in sorted(f_in.readlines())
if is_function_line(ln)]
functions = ["EXPORTS\n"] + functions
--
2.48.1.vfs.0.0
[PATCH v2 3/5] eal: update versioning macros
When compiling with MSVC the error below shows up due to function versioning: ../lib/net/rte_net_crc.c(418): error C2061: syntax error: identifier '__attribute__' MSVC allows alias function names to be exported, but the mechanism is different than the one used by gcc. It was considered to enhance the logic in the existing version.map files but that file is also passed to other tools on Linux, making this challenging. A simpler approach is to have an optional version.map file to be used only when Microsoft's linker is to be used. This optional map file is only necessary for libraries that have versioned code. Signed-off-by: Andre Muezerie --- lib/eal/include/rte_function_versioning.h | 19 +++ 1 file changed, 19 insertions(+) diff --git a/lib/eal/include/rte_function_versioning.h b/lib/eal/include/rte_function_versioning.h index eb6dd2bc17..aa3165a659 100644 --- a/lib/eal/include/rte_function_versioning.h +++ b/lib/eal/include/rte_function_versioning.h @@ -82,6 +82,13 @@ */ #define MAP_STATIC_SYMBOL(f, p) +/* + * MAP_STATIC_BASE_SYMBOL + * Has the same purpose as MAP_STATIC_SYMBOL, but takes a base function name + * instead of the whole function prototype. It is used to support MSVC. + */ +#define MAP_STATIC_BASE_SYMBOL(b, p) + #else /* * No symbol versioning in use @@ -90,7 +97,19 @@ #define VERSION_SYMBOL_EXPERIMENTAL(b, e) #define __vsym #define BIND_DEFAULT_SYMBOL(b, e, n) + +#ifdef RTE_TOOLCHAIN_MSVC +#define MAP_STATIC_SYMBOL(f, p) +#define MAP_STATIC_BASE_SYMBOL(b, p) __pragma(comment(linker, "/alternatename:" #b "=" #p)) +/* + * version.map file should also be updated with "b=p;", like + * rte_net_crc_set_alg=rte_net_crc_set_alg_v26; + */ +#else #define MAP_STATIC_SYMBOL(f, p) f __attribute__((alias(RTE_STR(p +#define MAP_STATIC_BASE_SYMBOL(b, p) +#endif + /* * RTE_BUILD_SHARED_LIB=n */ -- 2.48.1.vfs.0.0
[PATCH v2 5/5] mbuf: enable to be compiled with MSVC
Now that the issues preventing this lib from be compiled with MSVC are fixed it can be included in the compilation. The "net" library will automatically get compiled as well as it has a dependency on "mbuf" which will now get fulfilled. Signed-off-by: Andre Muezerie Acked-by: Bruce Richardson --- lib/mbuf/meson.build | 6 -- 1 file changed, 6 deletions(-) diff --git a/lib/mbuf/meson.build b/lib/mbuf/meson.build index 2cee9057a5..0435c5e628 100644 --- a/lib/mbuf/meson.build +++ b/lib/mbuf/meson.build @@ -1,12 +1,6 @@ # SPDX-License-Identifier: BSD-3-Clause # Copyright(c) 2017 Intel Corporation -if is_ms_compiler -build = false -reason = 'not supported building with Visual Studio Toolset' -subdir_done() -endif - sources = files( 'rte_mbuf.c', 'rte_mbuf_ptype.c', -- 2.48.1.vfs.0.0
[PATCH v2 0/5] mbuf: enable to be compiled with MSVC
When compiling with MSVC the error below shows up due to function versioning: ../lib/net/rte_net_crc.c(418): error C2061: syntax error: identifier '__attribute__' MSVC allows alias function names to be exported, but the mechanism is different than the one used by gcc. It was considered to enhance the logic in the existing version.map files but that file is also passed to other tools on Linux, making this challenging. A simpler approach is to have an optional version.map file to be used only when Microsoft's linker is to be used. This optional map file is only necessary for libraries that have versioned code. v2: - updated versioning macros so that the latest mbuf/net code can compile with MSVC. V1 did not have these changes because by the time that series was sent out the versioning problem had not surfaced yet. - for testing purposes, dpdk-test was build with msvc and crc_autotest was run successfully. This test is known to call the versioned function which raised the issue. Andre Muezerie (5): doc: update guide on versioning macros buildtools: update map_to_win.py to use optional map file eal: update versioning macros net: update use of versioning macros mbuf: enable to be compiled with MSVC buildtools/map_to_win.py | 16 ++- doc/guides/contributing/abi_versioning.rst | 6 ++ lib/eal/include/rte_function_versioning.h | 19 ++ lib/mbuf/meson.build | 6 -- lib/net/rte_net_crc.c | 2 ++ lib/net/version_ms_linker.map | 23 ++ 6 files changed, 65 insertions(+), 7 deletions(-) create mode 100644 lib/net/version_ms_linker.map -- 2.48.1.vfs.0.0
[PATCH v2 1/5] doc: update guide on versioning macros
Added explanation about new macro MAP_STATIC_BASE_SYMBOL. Signed-off-by: Andre Muezerie --- doc/guides/contributing/abi_versioning.rst | 6 ++ 1 file changed, 6 insertions(+) diff --git a/doc/guides/contributing/abi_versioning.rst b/doc/guides/contributing/abi_versioning.rst index 7afd1c1886..1179445da3 100644 --- a/doc/guides/contributing/abi_versioning.rst +++ b/doc/guides/contributing/abi_versioning.rst @@ -151,6 +151,11 @@ The macros exported are: fully qualified function ``p``, so that if a symbol becomes versioned, it can still be mapped back to the public symbol name. +* ``MAP_STATIC_BASE_SYMBOL(b, p)``: Declare the base symbol ``b``, and map it + to the fully qualified function ``p``, so that if a symbol becomes versioned, + it can still be mapped back to the public symbol name. Used for compatibility + with MSVC. + * ``__vsym``: Annotation to be used in a declaration of the internal symbol ``be`` to signal that it is being used as an implementation of a particular version of symbol ``b``. @@ -397,6 +402,7 @@ defined, we add this ... } MAP_STATIC_SYMBOL(struct rte_acl_ctx *rte_acl_create(const struct rte_acl_param *param, int debug), rte_acl_create_v22); + MAP_STATIC_BASE_SYMBOL(rte_acl_create, rte_acl_create_v22); That tells the compiler that, when building a static library, any calls to the symbol ``rte_acl_create`` should be linked to ``rte_acl_create_v22`` -- 2.48.1.vfs.0.0
Re: [PATCH v3 01/10] bus/auxiliary: remove weak symbols
Acked-by: Chengwen Feng On 2025/2/18 23:37, David Marchand wrote: > Rather than use weak symbols, expose stubs symbols when needed. > > Signed-off-by: David Marchand
Re: [PATCH v3 05/10] net/hns3: remove weak symbols
Acked-by: Chengwen Feng On 2025/2/18 23:37, David Marchand wrote: > Rather than use weak symbols, expose stubs symbols when needed. > > Signed-off-by: David Marchand
Re: [PATCH v3 10/10] eal: deprecate weak symbols
Acked-by: Chengwen Feng On 2025/2/18 23:37, David Marchand wrote: > Mark __rte_weak as deprecated. > It will avoid having to support such feature with other linkers, > plus this was never really needed. > > Signed-off-by: David Marchand
[PATCH v13 00/28] [v13]drivers/net Add Support mucse N10 Pmd Driver
For This patchset just to support the basic chip init work and user can just found the eth_dev, but can't control more. For Now just support 2*10g nic,the chip can support 2*10g,4*10g,4*1g,8*1g,8*10g. The Feature rx side can support rx-cksum-offload,rss,vlan-filter flow_clow,uncast_filter,mcast_filter,1588,Jumbo-frame The Feature tx side can support tx-cksum-offload,tso,vxlan-tso flow director base on ntuple pattern of tcp/udp/ip/ eth_hdr->type for sriov is also support. Because of the chip design defect, for multiple-port mode one pci-bdf will have multiple-port (max can have four ports) so this code must be care of one bdf init multiple-port. v13: * Supplementary document about n10 network card characteristics. * update release_25_03.rst. * fixed the code style advisea Stephen Hemminger. * fixed the code issue check PVS-stdio for Static compilation error. v12: * fixed __rte_packed __deprecated__ compile issue. v11: * fixed array-bounds issue when used rte_memcpy src addr is * not enough to hold align dst. * improve efficient_code advised by Stephen v10: * fixed mingw windows meson issue * rnp not support windows for now. v9: * fixed commit log format check by devtools. * fixed code compile issue. v8: * fixed codespell issue. * fixed MAINTAINERS file v7: * add support nic basic feature such as rss vlan strip/filter, * mtu-change recv/send scater-recv/mutltiple-send. * fixed code rationality, advised by Ferruh Yigit. v6: * fixed the doc(rst) format problem advise by Thomas Monjalon v5: * fixed the symbol name require by the style documentation v4: * one patch has been forgot to upload :( v3: * fixed http://dpdk.org/patch/129830 FreeBSD 13 compile Issue * change iobar type to void suggest by Stephen Hemminger * add KMOD_DEP support for vfio-pci * change run-cmd argument parse check for invalid extra_args v2: * fixed MAINTAIN maillist fullname format * fixed driver/net/meson the order issue of new driver to driver list * improve virtual point function usage suggest by Stephen Hemminger Wenbo Cao (28): net/rnp: add skeleton net/rnp: add ethdev probe and remove net/rnp: add log net/rnp: support mailbox basic operate net/rnp: add device init and uninit net/rnp: add get device information operation net/rnp: add support MAC promisc mode net/rnp: add queue setup and release operations net/rnp: add queue stop and start operations net/rnp: add support device start stop operations net/rnp: add RSS support operations net/rnp: add support link update operations net/rnp: add support link setup operations net/rnp: add Rx burst simple support net/rnp: add Tx burst simple support net/rnp: add MTU set operation net/rnp: add Rx scatter segment version net/rnp: add Tx multiple segment version net/rnp: add support basic stats operation net/rnp: add support xstats operation net/rnp: add unicast MAC filter operation net/rnp: add supported packet types net/rnp: add support Rx checksum offload net/rnp: add support Tx TSO offload net/rnp: support VLAN offloads net/rnp: add support VLAN filters operations net/rnp: add queue info operation net/rnp: support Rx/Tx burst mode info .mailmap |1 + MAINTAINERS|6 + doc/guides/nics/features/rnp.ini | 33 + doc/guides/nics/img/mucse_nic_port.svg | 4023 doc/guides/nics/index.rst |1 + doc/guides/nics/rnp.rst| 124 + doc/guides/rel_notes/release_25_03.rst |5 + drivers/net/meson.build|1 + drivers/net/rnp/base/meson.build | 17 + drivers/net/rnp/base/rnp_bdq_if.c | 398 +++ drivers/net/rnp/base/rnp_bdq_if.h | 154 + drivers/net/rnp/base/rnp_bitrev.h | 64 + drivers/net/rnp/base/rnp_common.c | 103 + drivers/net/rnp/base/rnp_common.h | 17 + drivers/net/rnp/base/rnp_crc32.c | 37 + drivers/net/rnp/base/rnp_crc32.h | 10 + drivers/net/rnp/base/rnp_dma_regs.h| 68 + drivers/net/rnp/base/rnp_eth_regs.h| 91 + drivers/net/rnp/base/rnp_fw_cmd.c | 162 + drivers/net/rnp/base/rnp_fw_cmd.h | 357 +++ drivers/net/rnp/base/rnp_hw.h | 136 + drivers/net/rnp/base/rnp_mac.c | 358 +++ drivers/net/rnp/base/rnp_mac.h | 34 + drivers/net/rnp/base/rnp_mac_regs.h| 208 ++ drivers/net/rnp/base/rnp_mbx.c | 510 +++ drivers/net/rnp/base/rnp_mbx.h | 58 + drivers/net/rnp/base/rnp_mbx_fw.c | 493 +++ drivers/net/rnp/base/rnp_mbx_fw.h | 24 + drivers/net/rnp/base/rnp_osdep.h | 172 + drivers/net/rnp/meson.build| 27 + drivers/net/rnp/rnp.h | 258 ++ drivers/net/rnp/rnp_ethdev.c | 1845 +++ drivers/net/rnp/rnp_link.c | 439 +++ drivers/net/rnp/rnp_link.h | 52 + drivers/net/rnp/rnp_logs.h | 36 + drive
[PATCH v13 03/28] net/rnp: add log
add log function for trace or debug
Signed-off-by: Wenbo Cao
---
drivers/net/rnp/rnp_ethdev.c | 2 ++
drivers/net/rnp/rnp_logs.h | 36
2 files changed, 38 insertions(+)
create mode 100644 drivers/net/rnp/rnp_logs.h
diff --git a/drivers/net/rnp/rnp_ethdev.c b/drivers/net/rnp/rnp_ethdev.c
index e2eb80ba69..7fe93edcf6 100644
--- a/drivers/net/rnp/rnp_ethdev.c
+++ b/drivers/net/rnp/rnp_ethdev.c
@@ -64,6 +64,8 @@ static struct rte_pci_driver rte_rnp_pmd = {
.remove = rnp_pci_remove,
};
+RTE_LOG_REGISTER_SUFFIX(rnp_init_logtype, init, NOTICE);
+
RTE_PMD_REGISTER_PCI(net_rnp, rte_rnp_pmd);
RTE_PMD_REGISTER_PCI_TABLE(net_rnp, pci_id_rnp_map);
RTE_PMD_REGISTER_KMOD_DEP(net_rnp, "igb_uio | uio_pci_generic | vfio-pci");
diff --git a/drivers/net/rnp/rnp_logs.h b/drivers/net/rnp/rnp_logs.h
new file mode 100644
index 00..c68a5c1028
--- /dev/null
+++ b/drivers/net/rnp/rnp_logs.h
@@ -0,0 +1,36 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(C) 2023 Mucse IC Design Ltd.
+ */
+
+#ifndef __RNP_LOGS_H__
+#define __RNP_LOGS_H__
+
+#include
+
+extern int rnp_init_logtype;
+#define RTE_LOGTYPE_RNP_INIT rnp_init_logtype
+#define RNP_PMD_INIT_LOG(level, ...) \
+ RTE_LOG_LINE_PREFIX(level, RNP_INIT, "%s(): ", __func__, __VA_ARGS__)
+#define PMD_INIT_FUNC_TRACE() RNP_PMD_INIT_LOG(DEBUG, " >>")
+#define RNP_PMD_DRV_LOG(level, ...) \
+ RTE_LOG_LINE_PREFIX(level, RNP_INIT, \
+ "%s(): ", __func__, __VA_ARGS__)
+#define RNP_PMD_LOG(level, ...) \
+ RTE_LOG_LINE_PREFIX(level, RNP_INIT, \
+ "rnp_net: (%d) ", __LINE__, __VA_ARGS__)
+#define RNP_PMD_ERR(fmt, ...) \
+ RNP_PMD_LOG(ERR, fmt, ## __VA_ARGS__)
+#define RNP_PMD_WARN(fmt, ...) \
+ RNP_PMD_LOG(WARNING, fmt, ## __VA_ARGS__)
+#define RNP_PMD_INFO(fmt, ...) \
+ RNP_PMD_LOG(INFO, fmt, ## __VA_ARGS__)
+
+#ifdef RTE_LIBRTE_RNP_REG_DEBUG
+#define RNP_PMD_REG_LOG(level, ...) \
+ RTE_LOG_LINE_PREFIX(level, RNP_INIT, \
+ "%s(): ", __func__, __VA_ARGS__)
+#else
+#define RNP_PMD_REG_LOG(...) do { } while (0)
+#endif
+
+#endif /* __RNP_LOGS_H__ */
--
2.25.1
[PATCH v13 09/28] net/rnp: add queue stop and start operations
support rx/tx queue stop/start,for rx queue stop
need to reset a queue,must stop all rx queue
during reset this queue.
Signed-off-by: Wenbo Cao
---
doc/guides/nics/features/rnp.ini | 1 +
doc/guides/nics/rnp.rst | 4 +
drivers/net/rnp/base/rnp_common.c | 3 +
drivers/net/rnp/rnp_rxtx.c| 167 ++
drivers/net/rnp/rnp_rxtx.h| 9 ++
5 files changed, 184 insertions(+)
diff --git a/doc/guides/nics/features/rnp.ini b/doc/guides/nics/features/rnp.ini
index 65f1ed3da0..fd7d4b9d8d 100644
--- a/doc/guides/nics/features/rnp.ini
+++ b/doc/guides/nics/features/rnp.ini
@@ -5,6 +5,7 @@
;
[Features]
Speed capabilities = Y
+Queue start/stop = Y
Promiscuous mode = Y
Allmulticast mode= Y
Linux= Y
diff --git a/doc/guides/nics/rnp.rst b/doc/guides/nics/rnp.rst
index 99b96e9b8e..c3547c38b6 100644
--- a/doc/guides/nics/rnp.rst
+++ b/doc/guides/nics/rnp.rst
@@ -71,6 +71,10 @@ Listed below are the rte_eth functions supported:
* ``rte_eth_dev_close``
* ``rte_eth_dev_stop``
* ``rte_eth_dev_infos_get``
+* ``rte_eth_dev_rx_queue_start``
+* ``rte_eth_dev_rx_queue_stop``
+* ``rte_eth_dev_tx_queue_start``
+* ``rte_eth_dev_tx_queue_stop``
* ``rte_eth_promiscuous_disable``
* ``rte_eth_promiscuous_enable``
* ``rte_eth_allmulticast_enable``
diff --git a/drivers/net/rnp/base/rnp_common.c
b/drivers/net/rnp/base/rnp_common.c
index 5655126ae0..58de3bde03 100644
--- a/drivers/net/rnp/base/rnp_common.c
+++ b/drivers/net/rnp/base/rnp_common.c
@@ -65,6 +65,9 @@ int rnp_init_hw(struct rnp_hw *hw)
/* setup mac resiger ctrl base */
for (idx = 0; idx < hw->max_port_num; idx++)
hw->mac_base[idx] = (u8 *)hw->e_ctrl + RNP_MAC_BASE_OFFSET(idx);
+ /* tx all hw queue must be started */
+ for (idx = 0; idx < RNP_MAX_RX_QUEUE_NUM; idx++)
+ RNP_E_REG_WR(hw, RNP_TXQ_START(idx), true);
return 0;
}
diff --git a/drivers/net/rnp/rnp_rxtx.c b/drivers/net/rnp/rnp_rxtx.c
index d370948d6b..e65bc06d36 100644
--- a/drivers/net/rnp/rnp_rxtx.c
+++ b/drivers/net/rnp/rnp_rxtx.c
@@ -86,6 +86,7 @@ rnp_rx_queue_reset(struct rnp_eth_port *port,
struct rte_eth_txconf def_conf;
struct rnp_hw *hw = port->hw;
struct rte_mbuf *m_mbuf[2];
+ bool tx_origin_e = false;
bool tx_new = false;
uint16_t index;
int err = 0;
@@ -121,6 +122,9 @@ rnp_rx_queue_reset(struct rnp_eth_port *port,
return -ENOMEM;
}
rnp_rxq_flow_disable(hw, index);
+ tx_origin_e = txq->txq_started;
+ rte_io_wmb();
+ txq->txq_started = false;
rte_mbuf_refcnt_set(m_mbuf[0], 1);
rte_mbuf_refcnt_set(m_mbuf[1], 1);
m_mbuf[0]->data_off = RTE_PKTMBUF_HEADROOM;
@@ -139,6 +143,7 @@ rnp_rx_queue_reset(struct rnp_eth_port *port,
rnp_tx_queue_reset(port, txq);
rnp_tx_queue_sw_reset(txq);
}
+ txq->txq_started = tx_origin_e;
}
rte_mempool_put_bulk(adapter->reset_pool, (void **)m_mbuf, 2);
rnp_rxq_flow_enable(hw, index);
@@ -367,6 +372,7 @@ rnp_tx_queue_sw_reset(struct rnp_tx_queue *txq)
txq->nb_tx_free = txq->attr.nb_desc - 1;
txq->tx_next_dd = txq->tx_rs_thresh - 1;
txq->tx_next_rs = txq->tx_rs_thresh - 1;
+ txq->tx_tail = 0;
size = (txq->attr.nb_desc + RNP_TX_MAX_BURST_SIZE);
for (idx = 0; idx < size * sizeof(struct rnp_tx_desc); idx++)
@@ -469,3 +475,164 @@ rnp_tx_queue_setup(struct rte_eth_dev *dev,
return err;
}
+
+int rnp_tx_queue_stop(struct rte_eth_dev *eth_dev, uint16_t qidx)
+{
+ struct rnp_eth_port *port = RNP_DEV_TO_PORT(eth_dev);
+ struct rte_eth_dev_data *data = eth_dev->data;
+ struct rnp_tx_queue *txq;
+
+ PMD_INIT_FUNC_TRACE();
+ txq = eth_dev->data->tx_queues[qidx];
+ if (!txq) {
+ RNP_PMD_ERR("TX queue %u is null or not setup", qidx);
+ return -EINVAL;
+ }
+ if (data->tx_queue_state[qidx] == RTE_ETH_QUEUE_STATE_STARTED) {
+ txq->txq_started = 0;
+ /* wait for tx burst process stop traffic */
+ rte_delay_us(10);
+ rnp_tx_queue_release_mbuf(txq);
+ rnp_tx_queue_reset(port, txq);
+ rnp_tx_queue_sw_reset(txq);
+ data->tx_queue_state[qidx] = RTE_ETH_QUEUE_STATE_STOPPED;
+ }
+
+ return 0;
+}
+
+int rnp_tx_queue_start(struct rte_eth_dev *eth_dev, uint16_t qidx)
+{
+ struct rte_eth_dev_data *data = eth_dev->data;
+ struct rnp_tx_queue *txq;
+
+ PMD_INIT_FUNC_TRACE();
+
+ txq = data->tx_queues[qidx];
+ if (!txq) {
+ RNP_PMD_ERR("Can't start tx queue %d it's not setup by "
+ "tx_queue_setup API", qidx);
+ return -EINVAL;
+ }
+ if (data->tx_queue_state[qidx] == RTE_ETH_QUEUE_STATE_STOPPED)
[PATCH v13 08/28] net/rnp: add queue setup and release operations
support tx/rx queue setup and release add hw bd
queue reset,sw queue reset.
Signed-off-by: Wenbo Cao
---
doc/guides/nics/rnp.rst | 3 +
drivers/net/rnp/base/meson.build| 1 +
drivers/net/rnp/base/rnp_bdq_if.c | 398 +++
drivers/net/rnp/base/rnp_bdq_if.h | 149 +
drivers/net/rnp/base/rnp_common.h | 4 +
drivers/net/rnp/base/rnp_dma_regs.h | 45 +++
drivers/net/rnp/base/rnp_eth_regs.h | 4 +
drivers/net/rnp/base/rnp_hw.h | 4 +
drivers/net/rnp/base/rnp_osdep.h| 12 +
drivers/net/rnp/meson.build | 1 +
drivers/net/rnp/rnp.h | 2 +
drivers/net/rnp/rnp_ethdev.c| 38 ++-
drivers/net/rnp/rnp_rxtx.c | 471
drivers/net/rnp/rnp_rxtx.h | 123
14 files changed, 1254 insertions(+), 1 deletion(-)
create mode 100644 drivers/net/rnp/base/rnp_bdq_if.c
create mode 100644 drivers/net/rnp/base/rnp_bdq_if.h
create mode 100644 drivers/net/rnp/rnp_rxtx.c
create mode 100644 drivers/net/rnp/rnp_rxtx.h
diff --git a/doc/guides/nics/rnp.rst b/doc/guides/nics/rnp.rst
index 5f6e22f287..99b96e9b8e 100644
--- a/doc/guides/nics/rnp.rst
+++ b/doc/guides/nics/rnp.rst
@@ -44,6 +44,7 @@ N10 has two functions, each function support muiple ports(1
to 8),which not same
Features
+- Multiple queues for TX and RX
- Promiscuous mode
Prerequisites and Pre-conditions
@@ -76,3 +77,5 @@ Listed below are the rte_eth functions supported:
* ``rte_eth_allmulticast_disable``
* ``rte_eth_promiscuous_get``
* ``rte_eth_allmulticast_get``
+* ``rte_eth_rx_queue_setup``
+* ``rte_eth_tx_queue_setup``
diff --git a/drivers/net/rnp/base/meson.build b/drivers/net/rnp/base/meson.build
index 5e1c8057fc..0c320e7116 100644
--- a/drivers/net/rnp/base/meson.build
+++ b/drivers/net/rnp/base/meson.build
@@ -7,6 +7,7 @@ sources = [
'rnp_mbx_fw.c',
'rnp_common.c',
'rnp_mac.c',
+'rnp_bdq_if.c',
]
base_lib = static_library('rnp_base', sources,
diff --git a/drivers/net/rnp/base/rnp_bdq_if.c
b/drivers/net/rnp/base/rnp_bdq_if.c
new file mode 100644
index 00..471c19efda
--- /dev/null
+++ b/drivers/net/rnp/base/rnp_bdq_if.c
@@ -0,0 +1,398 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(C) 2023 Mucse IC Design Ltd.
+ */
+
+#include "rnp_osdep.h"
+
+#include "../rnp.h"
+#include "rnp_dma_regs.h"
+#include "rnp_eth_regs.h"
+#include "rnp_bdq_if.h"
+#include "rnp_common.h"
+#include "../rnp_rxtx.h"
+
+static void
+rnp_read_mac_veb(struct rnp_hw *hw,
+ u16 nr_lane,
+ u16 vf_id,
+ struct rnp_veb_cfg *cfg)
+{
+ cfg->mac_lo = RNP_E_REG_RD(hw, RNP_VEB_MAC_LO(nr_lane, vf_id));
+ cfg->mac_hi = RNP_E_REG_RD(hw, RNP_VEB_MAC_HI(nr_lane, vf_id));
+ cfg->ring = RNP_E_REG_RD(hw, RNP_VEB_VF_RING(nr_lane, vf_id));
+}
+
+static void
+rnp_update_mac_veb(struct rnp_hw *hw,
+ u16 nr_lane,
+ u16 vf_id,
+ struct rnp_veb_cfg *cfg)
+{
+ u32 reg = cfg->ring;
+ u16 idx = 0;
+
+ idx = nr_lane;
+ wmb();
+ RNP_E_REG_WR(hw, RNP_VEB_MAC_LO(idx, vf_id), cfg->mac_lo);
+ RNP_E_REG_WR(hw, RNP_VEB_MAC_HI(idx, vf_id), cfg->mac_hi);
+ reg |= ((RNP_VEB_SWITCH_VF_EN | vf_id) << 8);
+ RNP_E_REG_WR(hw, RNP_VEB_VF_RING(idx, vf_id), reg);
+}
+
+void
+rnp_rxq_flow_disable(struct rnp_hw *hw,
+u16 hw_idx)
+{
+ u32 fc_ctrl;
+
+ spin_lock(&hw->rxq_reset_lock);
+ fc_ctrl = RNP_E_REG_RD(hw, RNP_RING_FC_EN(hw_idx));
+ wmb();
+ RNP_E_REG_WR(hw, RNP_RING_FC_THRESH(hw_idx), 0);
+ fc_ctrl |= 1 << (hw_idx % 32);
+ wmb();
+ RNP_E_REG_WR(hw, RNP_RING_FC_EN(hw_idx), fc_ctrl);
+}
+
+void
+rnp_rxq_flow_enable(struct rnp_hw *hw,
+ u16 hw_idx)
+{
+ u32 fc_ctrl;
+
+
+ fc_ctrl = RNP_E_REG_RD(hw, RNP_RING_FC_EN(hw_idx));
+ fc_ctrl &= ~(1 << (hw_idx % 32));
+ wmb();
+ RNP_E_REG_WR(hw, RNP_RING_FC_EN(hw_idx), fc_ctrl);
+
+ spin_unlock(&hw->rxq_reset_lock);
+}
+
+#define RNP_RXQ_RESET_PKT_LEN (64)
+
+static void
+rnp_reset_xmit(struct rnp_tx_queue *txq, u64 pkt_addr)
+{
+ volatile struct rnp_tx_desc *txbd;
+ struct rnp_txsw_entry *tx_entry;
+ u16 timeout = 0;
+ u16 tx_id;
+
+ tx_id = txq->tx_tail;
+ txbd = &txq->tx_bdr[tx_id];
+ tx_entry = &txq->sw_ring[tx_id];
+ memset(tx_entry, 0, sizeof(*tx_entry));
+
+ txbd->d.addr = pkt_addr;
+ txbd->d.blen = RNP_RXQ_RESET_PKT_LEN;
+ wmb();
+ txbd->d.cmd = cpu_to_le16(RNP_CMD_EOP | RNP_CMD_RS);
+ tx_id = (tx_id + 1) & txq->attr.nb_desc_mask;
+ wmb();
+ RNP_REG_WR(txq->tx_tailreg, 0, tx_id);
+ do {
+ if (txbd->d.cmd & RNP_CMD_DD)
+ break;
+ if (timeout == 1000)
+ RNP_PMD_ERR("rx queue %u reset send pkt is hang",
+
[PATCH v13 07/28] net/rnp: add support MAC promisc mode
add support two method of MAC unicast promisc
mulcast promisc broadcast promisc mode
Signed-off-by: Wenbo Cao
---
doc/guides/nics/features/rnp.ini| 2 +
doc/guides/nics/rnp.rst | 11 +++
drivers/net/rnp/base/rnp_common.c | 5 ++
drivers/net/rnp/base/rnp_eth_regs.h | 15
drivers/net/rnp/base/rnp_hw.h | 11 ++-
drivers/net/rnp/base/rnp_mac.c | 107 +++-
drivers/net/rnp/base/rnp_mac.h | 2 +
drivers/net/rnp/base/rnp_mac_regs.h | 39 ++
drivers/net/rnp/base/rnp_osdep.h| 5 ++
drivers/net/rnp/rnp_ethdev.c| 43 +++
10 files changed, 238 insertions(+), 2 deletions(-)
create mode 100644 drivers/net/rnp/base/rnp_mac_regs.h
diff --git a/doc/guides/nics/features/rnp.ini b/doc/guides/nics/features/rnp.ini
index 6766130a68..65f1ed3da0 100644
--- a/doc/guides/nics/features/rnp.ini
+++ b/doc/guides/nics/features/rnp.ini
@@ -5,5 +5,7 @@
;
[Features]
Speed capabilities = Y
+Promiscuous mode = Y
+Allmulticast mode= Y
Linux= Y
x86-64 = Y
diff --git a/doc/guides/nics/rnp.rst b/doc/guides/nics/rnp.rst
index ace3aeaa93..5f6e22f287 100644
--- a/doc/guides/nics/rnp.rst
+++ b/doc/guides/nics/rnp.rst
@@ -41,6 +41,11 @@ N10 has two functions, each function support muiple ports(1
to 8),which not same
rnp mucse nic port.
+Features
+
+
+- Promiscuous mode
+
Prerequisites and Pre-conditions
- Prepare the system as recommended by DPDK suite.
@@ -65,3 +70,9 @@ Listed below are the rte_eth functions supported:
* ``rte_eth_dev_close``
* ``rte_eth_dev_stop``
* ``rte_eth_dev_infos_get``
+* ``rte_eth_promiscuous_disable``
+* ``rte_eth_promiscuous_enable``
+* ``rte_eth_allmulticast_enable``
+* ``rte_eth_allmulticast_disable``
+* ``rte_eth_promiscuous_get``
+* ``rte_eth_allmulticast_get``
diff --git a/drivers/net/rnp/base/rnp_common.c
b/drivers/net/rnp/base/rnp_common.c
index 5cd4ec133e..5655126ae0 100644
--- a/drivers/net/rnp/base/rnp_common.c
+++ b/drivers/net/rnp/base/rnp_common.c
@@ -4,6 +4,7 @@
#include "rnp_osdep.h"
#include "rnp_hw.h"
+#include "rnp_mac_regs.h"
#include "rnp_eth_regs.h"
#include "rnp_dma_regs.h"
#include "rnp_common.h"
@@ -28,6 +29,7 @@ int rnp_init_hw(struct rnp_hw *hw)
struct rnp_eth_port *port = RNP_DEV_TO_PORT(hw->back->eth_dev);
u32 version = 0;
int ret = -1;
+ u32 idx = 0;
u32 state;
PMD_INIT_FUNC_TRACE();
@@ -60,6 +62,9 @@ int rnp_init_hw(struct rnp_hw *hw)
if (hw->nic_mode == RNP_DUAL_10G && hw->max_port_num == 2)
RNP_E_REG_WR(hw, RNP_TC_PORT_OFFSET(RNP_TARGET_TC_PORT),
RNP_PORT_OFF_QUEUE_NUM);
+ /* setup mac resiger ctrl base */
+ for (idx = 0; idx < hw->max_port_num; idx++)
+ hw->mac_base[idx] = (u8 *)hw->e_ctrl + RNP_MAC_BASE_OFFSET(idx);
return 0;
}
diff --git a/drivers/net/rnp/base/rnp_eth_regs.h
b/drivers/net/rnp/base/rnp_eth_regs.h
index 6957866d76..c4519ba1e5 100644
--- a/drivers/net/rnp/base/rnp_eth_regs.h
+++ b/drivers/net/rnp/base/rnp_eth_regs.h
@@ -10,6 +10,21 @@
#define RNP_E_FILTER_EN_ETH_(0x801c)
#define RNP_E_REDIR_EN _ETH_(0x8030)
+/* Mac Host Filter */
+#define RNP_MAC_FCTRL _ETH_(0x9110)
+#define RNP_MAC_FCTRL_MPE RTE_BIT32(8) /* Multicast Promiscuous En */
+#define RNP_MAC_FCTRL_UPE RTE_BIT32(9) /* Unicast Promiscuous En */
+#define RNP_MAC_FCTRL_BAM RTE_BIT32(10) /* Broadcast Accept Mode */
+#define RNP_MAC_FCTRL_BYPASS (\
+ RNP_MAC_FCTRL_MPE | \
+ RNP_MAC_FCTRL_UPE | \
+ RNP_MAC_FCTRL_BAM)
+/* Mucast unicast mac hash filter ctrl */
+#define RNP_MAC_MCSTCTRL _ETH_(0x9114)
+#define RNP_MAC_HASH_MASK RTE_GENMASK32(11, 0)
+#define RNP_MAC_MULTICASE_TBL_EN RTE_BIT32(2)
+#define RNP_MAC_UNICASE_TBL_EN RTE_BIT32(3)
+
#define RNP_TC_PORT_OFFSET(lane) _ETH_(0xe840 + 0x04 * (lane))
#endif /* _RNP_ETH_REGS_H */
diff --git a/drivers/net/rnp/base/rnp_hw.h b/drivers/net/rnp/base/rnp_hw.h
index 3d1d5cd87b..8e4ca54d45 100644
--- a/drivers/net/rnp/base/rnp_hw.h
+++ b/drivers/net/rnp/base/rnp_hw.h
@@ -60,9 +60,17 @@ struct rnp_mbx_info {
struct rnp_eth_port;
/* mac operations */
+enum rnp_mpf_modes {
+ RNP_MPF_MODE_NONE = 0,
+ RNP_MPF_MODE_ALLMULTI, /* Multitle Promisc */
+ RNP_MPF_MODE_PROMISC, /* Unicast Promisc */
+};
+
struct rnp_mac_ops {
- /* update mac packet filter mode */
+ /* get default mac address */
int (*get_macaddr)(struct rnp_eth_port *port, u8 *mac);
+ /* update mac packet filter mode */
+ int (*update_mpfm)(struct rnp_eth_port *port, u32 mode, bool en);
};
struct rnp_eth_adapter;
@@ -92,6 +100,7 @@ struct rnp_hw {
struct rnp_eth_adapter *back; /* backup to the adapter handle */
void __iomem *e_ctrl; /* eth
[PATCH v13 05/28] net/rnp: add device init and uninit
add firmware communic method and basic device
init, uninit and close resource function.
Signed-off-by: Wenbo Cao
Reviewed-by: Ferruh Yigit
---
doc/guides/nics/rnp.rst | 10 +
drivers/net/rnp/base/meson.build| 4 +
drivers/net/rnp/base/rnp_common.c | 73 ++
drivers/net/rnp/base/rnp_common.h | 12 +
drivers/net/rnp/base/rnp_dma_regs.h | 13 ++
drivers/net/rnp/base/rnp_eth_regs.h | 15 ++
drivers/net/rnp/base/rnp_fw_cmd.c | 75 +++
drivers/net/rnp/base/rnp_fw_cmd.h | 216 ++
drivers/net/rnp/base/rnp_hw.h | 39
drivers/net/rnp/base/rnp_mac.c | 28 +++
drivers/net/rnp/base/rnp_mac.h | 14 ++
drivers/net/rnp/base/rnp_mbx_fw.c | 335
drivers/net/rnp/base/rnp_mbx_fw.h | 18 ++
drivers/net/rnp/base/rnp_osdep.h| 91
drivers/net/rnp/meson.build | 1 +
drivers/net/rnp/rnp.h | 44
drivers/net/rnp/rnp_ethdev.c| 321 +-
17 files changed, 1300 insertions(+), 9 deletions(-)
create mode 100644 drivers/net/rnp/base/rnp_common.c
create mode 100644 drivers/net/rnp/base/rnp_common.h
create mode 100644 drivers/net/rnp/base/rnp_dma_regs.h
create mode 100644 drivers/net/rnp/base/rnp_eth_regs.h
create mode 100644 drivers/net/rnp/base/rnp_fw_cmd.c
create mode 100644 drivers/net/rnp/base/rnp_fw_cmd.h
create mode 100644 drivers/net/rnp/base/rnp_mac.c
create mode 100644 drivers/net/rnp/base/rnp_mac.h
create mode 100644 drivers/net/rnp/base/rnp_mbx_fw.c
create mode 100644 drivers/net/rnp/base/rnp_mbx_fw.h
diff --git a/doc/guides/nics/rnp.rst b/doc/guides/nics/rnp.rst
index cca249be78..1f860e3f18 100644
--- a/doc/guides/nics/rnp.rst
+++ b/doc/guides/nics/rnp.rst
@@ -54,3 +54,13 @@ Limitations or Known issues
---
X86-32, BSD, Armv7, RISC-V, Windows, are not supported yet.
+
+Supported APIs
+--
+
+rte_eth APIs
+
+
+Listed below are the rte_eth functions supported:
+* ``rte_eth_dev_close``
+* ``rte_eth_dev_stop``
diff --git a/drivers/net/rnp/base/meson.build b/drivers/net/rnp/base/meson.build
index 0430481fe4..5e1c8057fc 100644
--- a/drivers/net/rnp/base/meson.build
+++ b/drivers/net/rnp/base/meson.build
@@ -3,6 +3,10 @@
sources = [
'rnp_mbx.c',
+'rnp_fw_cmd.c',
+'rnp_mbx_fw.c',
+'rnp_common.c',
+'rnp_mac.c',
]
base_lib = static_library('rnp_base', sources,
diff --git a/drivers/net/rnp/base/rnp_common.c
b/drivers/net/rnp/base/rnp_common.c
new file mode 100644
index 00..5cd4ec133e
--- /dev/null
+++ b/drivers/net/rnp/base/rnp_common.c
@@ -0,0 +1,73 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(C) 2023 Mucse IC Design Ltd.
+ */
+
+#include "rnp_osdep.h"
+#include "rnp_hw.h"
+#include "rnp_eth_regs.h"
+#include "rnp_dma_regs.h"
+#include "rnp_common.h"
+#include "rnp_mbx_fw.h"
+#include "rnp_mac.h"
+#include "../rnp.h"
+
+static void
+rnp_hw_reset(struct rnp_hw *hw)
+{
+ PMD_INIT_FUNC_TRACE();
+
+ RNP_E_REG_WR(hw, RNP_NIC_RESET, 0);
+ /* hardware reset valid must be 0 -> 1 */
+ wmb();
+ RNP_E_REG_WR(hw, RNP_NIC_RESET, 1);
+ RNP_PMD_DRV_LOG(INFO, "PF[%d] reset nic finish", hw->mbx.pf_num);
+}
+
+int rnp_init_hw(struct rnp_hw *hw)
+{
+ struct rnp_eth_port *port = RNP_DEV_TO_PORT(hw->back->eth_dev);
+ u32 version = 0;
+ int ret = -1;
+ u32 state;
+
+ PMD_INIT_FUNC_TRACE();
+ version = RNP_E_REG_RD(hw, RNP_DMA_VERSION);
+ RNP_PMD_DRV_LOG(INFO, "nic hw version:0x%.2x", version);
+ rnp_fw_init(hw);
+ RNP_E_REG_WR(hw, RNP_DMA_HW_EN, FALSE);
+ do {
+ state = RNP_E_REG_RD(hw, RNP_DMA_HW_STATE);
+ } while (state == 0);
+ ret = rnp_mbx_fw_get_capability(port);
+ if (ret) {
+ RNP_PMD_ERR("mbx_get_capability error! errcode=%d", ret);
+ return ret;
+ }
+ rnp_hw_reset(hw);
+ rnp_mbx_fw_reset_phy(hw);
+ /* rx packet protocol engine bypass */
+ RNP_E_REG_WR(hw, RNP_E_ENG_BYPASS, FALSE);
+ /* enable host filter */
+ RNP_E_REG_WR(hw, RNP_E_FILTER_EN, TRUE);
+ /* enable vxlan parse */
+ RNP_E_REG_WR(hw, RNP_E_VXLAN_PARSE_EN, TRUE);
+ /* enable flow direct engine */
+ RNP_E_REG_WR(hw, RNP_E_REDIR_EN, TRUE);
+ /* enable dma engine */
+ RNP_E_REG_WR(hw, RNP_DMA_HW_EN, RNP_DMA_EN_ALL);
+#define RNP_TARGET_TC_PORT (2)
+#define RNP_PORT_OFF_QUEUE_NUM (2)
+ if (hw->nic_mode == RNP_DUAL_10G && hw->max_port_num == 2)
+ RNP_E_REG_WR(hw, RNP_TC_PORT_OFFSET(RNP_TARGET_TC_PORT),
+ RNP_PORT_OFF_QUEUE_NUM);
+
+ return 0;
+}
+
+int
+rnp_setup_common_ops(struct rnp_hw *hw)
+{
+ rnp_mac_ops_init(hw);
+
+ return 0;
+}
diff --git a/drivers/net/rnp/base/rnp_common.h
b/drivers/net/rnp/base/rnp_common.h
new file mode 100644
index 00..aaf77a
[PATCH v13 12/28] net/rnp: add support link update operations
This patch add support poll/irq link get mode.
Signed-off-by: Wenbo Cao
---
doc/guides/nics/features/rnp.ini | 2 +
doc/guides/nics/rnp.rst | 3 +
drivers/net/rnp/base/rnp_fw_cmd.c | 45
drivers/net/rnp/base/rnp_fw_cmd.h | 55 +
drivers/net/rnp/base/rnp_hw.h | 2 +-
drivers/net/rnp/base/rnp_mbx_fw.c | 72 ++-
drivers/net/rnp/base/rnp_mbx_fw.h | 4 +
drivers/net/rnp/meson.build | 1 +
drivers/net/rnp/rnp.h | 12 ++
drivers/net/rnp/rnp_ethdev.c | 116 +-
drivers/net/rnp/rnp_link.c| 340 ++
drivers/net/rnp/rnp_link.h| 50 +
12 files changed, 695 insertions(+), 7 deletions(-)
create mode 100644 drivers/net/rnp/rnp_link.c
create mode 100644 drivers/net/rnp/rnp_link.h
diff --git a/doc/guides/nics/features/rnp.ini b/doc/guides/nics/features/rnp.ini
index 2fc94825f5..695b9c0dba 100644
--- a/doc/guides/nics/features/rnp.ini
+++ b/doc/guides/nics/features/rnp.ini
@@ -5,6 +5,8 @@
;
[Features]
Speed capabilities = Y
+Link status = Y
+Link status event= Y
Queue start/stop = Y
Promiscuous mode = Y
Allmulticast mode= Y
diff --git a/doc/guides/nics/rnp.rst b/doc/guides/nics/rnp.rst
index 90af73d71f..e6a1aadefc 100644
--- a/doc/guides/nics/rnp.rst
+++ b/doc/guides/nics/rnp.rst
@@ -49,6 +49,7 @@ Features
Receiver Side Steering (RSS) on IPv4, IPv6, IPv4-TCP/UDP/SCTP,
IPv6-TCP/UDP/SCTP
Inner RSS is only support for vxlan/nvgre
- Promiscuous mode
+- Link state information
Prerequisites and Pre-conditions
@@ -92,3 +93,5 @@ Listed below are the rte_eth functions supported:
* ``rte_eth_allmulticast_get``
* ``rte_eth_rx_queue_setup``
* ``rte_eth_tx_queue_setup``
+* ``rte_eth_link_get``
+* ``rte_eth_link_get_nowait``
diff --git a/drivers/net/rnp/base/rnp_fw_cmd.c
b/drivers/net/rnp/base/rnp_fw_cmd.c
index 3891249499..a4a2171868 100644
--- a/drivers/net/rnp/base/rnp_fw_cmd.c
+++ b/drivers/net/rnp/base/rnp_fw_cmd.c
@@ -68,6 +68,45 @@ rnp_build_get_lane_status_req(struct rnp_mbx_fw_cmd_req *req,
arg->nr_lane = req_arg->param0;
}
+static void
+rnp_build_set_event_mask(struct rnp_mbx_fw_cmd_req *req,
+struct rnp_fw_req_arg *req_arg,
+void *cookie)
+{
+ struct rnp_set_pf_event_mask *arg =
+ (struct rnp_set_pf_event_mask *)req->data;
+
+ req->flags = 0;
+ req->opcode = RNP_SET_EVENT_MASK;
+ req->datalen = sizeof(*arg);
+ req->cookie = cookie;
+ req->reply_lo = 0;
+ req->reply_hi = 0;
+
+ arg->event_mask = req_arg->param0;
+ arg->event_en = req_arg->param1;
+}
+
+static void
+rnp_build_lane_evet_mask(struct rnp_mbx_fw_cmd_req *req,
+struct rnp_fw_req_arg *req_arg,
+void *cookie)
+{
+ struct rnp_set_lane_event_mask *arg =
+ (struct rnp_set_lane_event_mask *)req->data;
+
+ req->flags = 0;
+ req->opcode = RNP_SET_LANE_EVENT_EN;
+ req->datalen = sizeof(*arg);
+ req->cookie = cookie;
+ req->reply_lo = 0;
+ req->reply_hi = 0;
+
+ arg->nr_lane = req_arg->param0;
+ arg->event_mask = req_arg->param1;
+ arg->event_en = req_arg->param2;
+}
+
int rnp_build_fwcmd_req(struct rnp_mbx_fw_cmd_req *req,
struct rnp_fw_req_arg *arg,
void *cookie)
@@ -87,6 +126,12 @@ int rnp_build_fwcmd_req(struct rnp_mbx_fw_cmd_req *req,
case RNP_GET_LANE_STATUS:
rnp_build_get_lane_status_req(req, arg, cookie);
break;
+ case RNP_SET_EVENT_MASK:
+ rnp_build_set_event_mask(req, arg, cookie);
+ break;
+ case RNP_SET_LANE_EVENT_EN:
+ rnp_build_lane_evet_mask(req, arg, cookie);
+ break;
default:
err = -EOPNOTSUPP;
}
diff --git a/drivers/net/rnp/base/rnp_fw_cmd.h
b/drivers/net/rnp/base/rnp_fw_cmd.h
index 5f60ac6187..6c6fd1803e 100644
--- a/drivers/net/rnp/base/rnp_fw_cmd.h
+++ b/drivers/net/rnp/base/rnp_fw_cmd.h
@@ -6,6 +6,7 @@
#define _RNP_FW_CMD_H_
#include "rnp_osdep.h"
+#include "rnp_hw.h"
#define RNP_FW_LINK_SYNC (0x000c)
#define RNP_LINK_MAGIC_CODE(0xa5a4)
@@ -73,6 +74,22 @@ enum RNP_GENERIC_CMD {
RNP_SET_DDR_CSL = 0xFF11,
};
+struct rnp_port_stat {
+ u8 phy_addr; /* Phy MDIO address */
+
+ u8 duplex : 1; /* FIBRE is always 1,Twisted Pair 1 or 0 */
+ u8 autoneg : 1; /* autoned state */
+ u8 fec : 1;
+ u8 an_rev : 1;
+ u8 link_traing : 1;
+ u8 is_sgmii : 1; /* avild fw >= 0.5.0.17 */
+ u8 rsvd0: 2;
+ u16 speed; /* cur port linked speed */
+
+ u16 pause : 4;
+ u16 rsvd1 : 12;
+};
+
/* firmware -> driver reply */
struct __rte_aligned(4) __rte_packed_begin rnp_p
[PATCH v13 06/28] net/rnp: add get device information operation
add get device hardware capability function
Signed-off-by: Wenbo Cao
---
doc/guides/nics/features/rnp.ini | 1 +
doc/guides/nics/rnp.rst | 1 +
drivers/net/rnp/base/rnp_fw_cmd.c | 20 ++
drivers/net/rnp/base/rnp_fw_cmd.h | 80 +
drivers/net/rnp/base/rnp_mbx_fw.c | 55 ++
drivers/net/rnp/base/rnp_mbx_fw.h | 1 +
drivers/net/rnp/rnp.h | 73 ++-
drivers/net/rnp/rnp_ethdev.c | 115 +-
8 files changed, 344 insertions(+), 2 deletions(-)
diff --git a/doc/guides/nics/features/rnp.ini b/doc/guides/nics/features/rnp.ini
index 2ad04ee330..6766130a68 100644
--- a/doc/guides/nics/features/rnp.ini
+++ b/doc/guides/nics/features/rnp.ini
@@ -4,5 +4,6 @@
; Refer to default.ini for the full list of available PMD features.
;
[Features]
+Speed capabilities = Y
Linux= Y
x86-64 = Y
diff --git a/doc/guides/nics/rnp.rst b/doc/guides/nics/rnp.rst
index 1f860e3f18..ace3aeaa93 100644
--- a/doc/guides/nics/rnp.rst
+++ b/doc/guides/nics/rnp.rst
@@ -64,3 +64,4 @@ rte_eth APIs
Listed below are the rte_eth functions supported:
* ``rte_eth_dev_close``
* ``rte_eth_dev_stop``
+* ``rte_eth_dev_infos_get``
diff --git a/drivers/net/rnp/base/rnp_fw_cmd.c
b/drivers/net/rnp/base/rnp_fw_cmd.c
index 676c5a2521..3891249499 100644
--- a/drivers/net/rnp/base/rnp_fw_cmd.c
+++ b/drivers/net/rnp/base/rnp_fw_cmd.c
@@ -51,6 +51,23 @@ rnp_build_get_macaddress_req(struct rnp_mbx_fw_cmd_req *req,
arg->pfvf_num = req_arg->param1;
}
+static inline void
+rnp_build_get_lane_status_req(struct rnp_mbx_fw_cmd_req *req,
+ struct rnp_fw_req_arg *req_arg,
+ void *cookie)
+{
+ struct rnp_get_lane_st_req *arg = (struct rnp_get_lane_st_req
*)req->data;
+
+ req->flags = 0;
+ req->opcode = RNP_GET_LANE_STATUS;
+ req->datalen = sizeof(*arg);
+ req->cookie = cookie;
+ req->reply_lo = 0;
+ req->reply_hi = 0;
+
+ arg->nr_lane = req_arg->param0;
+}
+
int rnp_build_fwcmd_req(struct rnp_mbx_fw_cmd_req *req,
struct rnp_fw_req_arg *arg,
void *cookie)
@@ -67,6 +84,9 @@ int rnp_build_fwcmd_req(struct rnp_mbx_fw_cmd_req *req,
case RNP_GET_MAC_ADDRESS:
rnp_build_get_macaddress_req(req, arg, cookie);
break;
+ case RNP_GET_LANE_STATUS:
+ rnp_build_get_lane_status_req(req, arg, cookie);
+ break;
default:
err = -EOPNOTSUPP;
}
diff --git a/drivers/net/rnp/base/rnp_fw_cmd.h
b/drivers/net/rnp/base/rnp_fw_cmd.h
index 87f0823dba..5f60ac6187 100644
--- a/drivers/net/rnp/base/rnp_fw_cmd.h
+++ b/drivers/net/rnp/base/rnp_fw_cmd.h
@@ -129,6 +129,80 @@ struct rnp_mac_addr_rep {
u32 pcode;
};
+#define RNP_SPEED_CAP_UNKNOWN(0)
+#define RNP_SPEED_CAP_10M_FULL RTE_BIT32(2)
+#define RNP_SPEED_CAP_100M_FULL RTE_BIT32(3)
+#define RNP_SPEED_CAP_1GB_FULL RTE_BIT32(4)
+#define RNP_SPEED_CAP_10GB_FULL RTE_BIT32(5)
+#define RNP_SPEED_CAP_40GB_FULL RTE_BIT32(6)
+#define RNP_SPEED_CAP_25GB_FULL RTE_BIT32(7)
+#define RNP_SPEED_CAP_50GB_FULL RTE_BIT32(8)
+#define RNP_SPEED_CAP_100GB_FULL RTE_BIT32(9)
+#define RNP_SPEED_CAP_10M_HALF RTE_BIT32(10)
+#define RNP_SPEED_CAP_100M_HALF RTE_BIT32(11)
+#define RNP_SPEED_CAP_1GB_HALF RTE_BIT32(12)
+
+enum rnp_pma_phy_type {
+ RNP_PHY_TYPE_NONE = 0,
+ RNP_PHY_TYPE_1G_BASE_KX,
+ RNP_PHY_TYPE_SGMII,
+ RNP_PHY_TYPE_10G_BASE_KR,
+ RNP_PHY_TYPE_25G_BASE_KR,
+ RNP_PHY_TYPE_40G_BASE_KR4,
+ RNP_PHY_TYPE_10G_BASE_SR,
+ RNP_PHY_TYPE_40G_BASE_SR4,
+ RNP_PHY_TYPE_40G_BASE_CR4,
+ RNP_PHY_TYPE_40G_BASE_LR4,
+ RNP_PHY_TYPE_10G_BASE_LR,
+ RNP_PHY_TYPE_10G_BASE_ER,
+ RNP_PHY_TYPE_10G_TP,
+};
+
+struct rnp_lane_stat_rep {
+ u8 nr_lane; /* 0-3 cur port correspond with hw lane */
+ u8 pci_gen : 4; /* nic cur pci speed genX: 1,2,3 */
+ u8 pci_lanes: 4; /* nic cur pci x1 x2 x4 x8 x16 */
+ u8 pma_type;
+ u8 phy_type;/* interface media type */
+
+ u16 linkup : 1; /* cur port link state */
+ u16 duplex : 1; /* duplex state only RJ45 valid */
+ u16 autoneg : 1; /* autoneg state */
+ u16 fec : 1; /* fec state */
+ u16 rev_an : 1;
+ u16 link_traing : 1; /* link-traing state */
+ u16 media_available : 1;
+ u16 is_sgmii: 1; /* 1: Twisted Pair 0: FIBRE */
+ u16 link_fault : 4;
+#define RNP_LINK_LINK_FAULTRTE_BIT32(0)
+#define RNP_LINK_TX_FAULT RTE_BIT32(1)
+#define RNP_LINK_RX_FAULT RTE_BIT32(2)
+#define RNP_LINK_REMOTE_FAULT RTE_BIT32(3)
+ u16 is_backplane: 1; /* Backplane Mode */
+ u16 is_speed_10G_1G_auto_swit
[PATCH v13 02/28] net/rnp: add ethdev probe and remove
Add basic PCIe ethdev probe and remove.
Signed-off-by: Wenbo Cao
Reviewed-by: Stephen Hemminger
---
doc/guides/nics/rnp.rst | 17 ++
drivers/net/rnp/rnp.h| 16 +
drivers/net/rnp/rnp_ethdev.c | 66
3 files changed, 99 insertions(+)
create mode 100644 drivers/net/rnp/rnp.h
diff --git a/doc/guides/nics/rnp.rst b/doc/guides/nics/rnp.rst
index b76c60349c..cca249be78 100644
--- a/doc/guides/nics/rnp.rst
+++ b/doc/guides/nics/rnp.rst
@@ -14,6 +14,23 @@ Supported Chipsets and NICs
- MUCSE Ethernet Controller N10 Series for 10GbE or 40GbE (Dual-port)
+Support of Ethernet Network Controllers
+~~~
+
+PCIE NICS
+^
+
+* ``N10G-X2-DC Dual-port 10 Gigabit SFP Ethernet Adapter``
+* ``N10G-X4-QC Quad-port 10 Gigabit SFP Ethernet Adapter``
+* ``N10G-X8-DC Octa-port 10 Gigabit SFP Ethernet Adapter``
+
+OCP 3.0 NICs
+
+
+* ``N10G-X2-DCP Dual-port 10 Gigabit SFP Ethernet Adapter``
+* ``N10G-X4-DCP Dual-port 10 Gigabit SFP Ethernet Adapter``
+* ``N10G-X8-DCP Dual-port 10 Gigabit SFP Ethernet Adapter``
+
Chip Basic Overview
---
N10 has two functions, each function support muiple ports(1 to 8),which not
same as normal pcie network card(one pf for each port).
diff --git a/drivers/net/rnp/rnp.h b/drivers/net/rnp/rnp.h
new file mode 100644
index 00..4413a03d9d
--- /dev/null
+++ b/drivers/net/rnp/rnp.h
@@ -0,0 +1,16 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(C) 2023 Mucse IC Design Ltd.
+ */
+
+#ifndef __RNP_H__
+#define __RNP_H__
+
+#define PCI_VENDOR_ID_MUCSE(0x8848)
+#define RNP_DEV_ID_N10G_X2 (0x1000)
+#define RNP_DEV_ID_N10G_X4 (0x1020)
+#define RNP_DEV_ID_N10G_X8 (0x1060)
+
+struct rnp_eth_port {
+};
+
+#endif /* __RNP_H__ */
diff --git a/drivers/net/rnp/rnp_ethdev.c b/drivers/net/rnp/rnp_ethdev.c
index 9ce3c0b497..e2eb80ba69 100644
--- a/drivers/net/rnp/rnp_ethdev.c
+++ b/drivers/net/rnp/rnp_ethdev.c
@@ -1,3 +1,69 @@
/* SPDX-License-Identifier: BSD-3-Clause
* Copyright(C) 2023 Mucse IC Design Ltd.
*/
+
+#include
+#include
+
+#include "rnp.h"
+
+static int
+rnp_eth_dev_init(struct rte_eth_dev *eth_dev)
+{
+ RTE_SET_USED(eth_dev);
+
+ return -ENODEV;
+}
+
+static int
+rnp_eth_dev_uninit(struct rte_eth_dev *eth_dev)
+{
+ RTE_SET_USED(eth_dev);
+
+ return -ENODEV;
+}
+
+static int
+rnp_pci_remove(struct rte_pci_device *pci_dev)
+{
+ struct rte_eth_dev *eth_dev;
+
+ eth_dev = rte_eth_dev_allocated(pci_dev->device.name);
+
+ if (eth_dev)
+ /* Cleanup eth dev */
+ return rte_eth_dev_pci_generic_remove(pci_dev,
+ rnp_eth_dev_uninit);
+ return -ENODEV;
+}
+
+static int
+rnp_pci_probe(struct rte_pci_driver *pci_drv, struct rte_pci_device *pci_dev)
+{
+ int rc;
+
+ RTE_SET_USED(pci_drv);
+
+ rc = rte_eth_dev_pci_generic_probe(pci_dev, sizeof(struct rnp_eth_port),
+ rnp_eth_dev_init);
+
+ return rc;
+}
+
+static const struct rte_pci_id pci_id_rnp_map[] = {
+ { RTE_PCI_DEVICE(PCI_VENDOR_ID_MUCSE, RNP_DEV_ID_N10G_X2) },
+ { RTE_PCI_DEVICE(PCI_VENDOR_ID_MUCSE, RNP_DEV_ID_N10G_X4) },
+ { RTE_PCI_DEVICE(PCI_VENDOR_ID_MUCSE, RNP_DEV_ID_N10G_X8) },
+ { .vendor_id = 0, },
+};
+
+static struct rte_pci_driver rte_rnp_pmd = {
+ .id_table = pci_id_rnp_map,
+ .drv_flags = RTE_PCI_DRV_NEED_MAPPING | RTE_PCI_DRV_INTR_LSC,
+ .probe = rnp_pci_probe,
+ .remove = rnp_pci_remove,
+};
+
+RTE_PMD_REGISTER_PCI(net_rnp, rte_rnp_pmd);
+RTE_PMD_REGISTER_PCI_TABLE(net_rnp, pci_id_rnp_map);
+RTE_PMD_REGISTER_KMOD_DEP(net_rnp, "igb_uio | uio_pci_generic | vfio-pci");
--
2.25.1
[PATCH v13 04/28] net/rnp: support mailbox basic operate
This patch adds support for mailbox of rnp PMD driver,
mailbox is used for communication between pf with fw
and vf driver.
Signed-off-by: Wenbo Cao
Reviewed-by: Stephen Hemminger
---
drivers/net/rnp/base/meson.build | 11 +
drivers/net/rnp/base/rnp_hw.h| 77 +
drivers/net/rnp/base/rnp_mbx.c | 510 +++
drivers/net/rnp/base/rnp_mbx.h | 58
drivers/net/rnp/base/rnp_osdep.h | 51
drivers/net/rnp/meson.build | 5 +
drivers/net/rnp/rnp.h| 19 ++
7 files changed, 731 insertions(+)
create mode 100644 drivers/net/rnp/base/meson.build
create mode 100644 drivers/net/rnp/base/rnp_hw.h
create mode 100644 drivers/net/rnp/base/rnp_mbx.c
create mode 100644 drivers/net/rnp/base/rnp_mbx.h
create mode 100644 drivers/net/rnp/base/rnp_osdep.h
diff --git a/drivers/net/rnp/base/meson.build b/drivers/net/rnp/base/meson.build
new file mode 100644
index 00..0430481fe4
--- /dev/null
+++ b/drivers/net/rnp/base/meson.build
@@ -0,0 +1,11 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(C) 2023 Mucse IC Design Ltd.
+
+sources = [
+'rnp_mbx.c',
+]
+
+base_lib = static_library('rnp_base', sources,
+dependencies: [static_rte_eal, static_rte_net, static_rte_ethdev],
+c_args: c_args)
+base_objs = base_lib.extract_all_objects(recursive: true)
diff --git a/drivers/net/rnp/base/rnp_hw.h b/drivers/net/rnp/base/rnp_hw.h
new file mode 100644
index 00..326b850560
--- /dev/null
+++ b/drivers/net/rnp/base/rnp_hw.h
@@ -0,0 +1,77 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(C) 2023 Mucse IC Design Ltd.
+ */
+
+#ifndef _RNP_HW_H_
+#define _RNP_HW_H_
+
+#include "rnp_osdep.h"
+
+struct rnp_hw;
+/* Mailbox Operate Info */
+enum RNP_MBX_ID {
+ RNP_MBX_PF = 0,
+ RNP_MBX_VF,
+ RNP_MBX_FW = 64,
+};
+
+struct rnp_mbx_ops {
+ int (*read)(struct rnp_hw *hw,
+ u32 *msg,
+ u16 size,
+ enum RNP_MBX_ID);
+ int (*write)(struct rnp_hw *hw,
+ u32 *msg,
+ u16 size,
+ enum RNP_MBX_ID);
+ int (*read_posted)(struct rnp_hw *hw,
+ u32 *msg,
+ u16 size,
+ enum RNP_MBX_ID);
+ int (*write_posted)(struct rnp_hw *hw,
+ u32 *msg,
+ u16 size,
+ enum RNP_MBX_ID);
+ int (*check_for_msg)(struct rnp_hw *hw, enum RNP_MBX_ID);
+ int (*check_for_ack)(struct rnp_hw *hw, enum RNP_MBX_ID);
+ int (*check_for_rst)(struct rnp_hw *hw, enum RNP_MBX_ID);
+};
+
+struct rnp_mbx_sync {
+ u16 req;
+ u16 ack;
+};
+
+struct rnp_mbx_info {
+ const struct rnp_mbx_ops *ops;
+ u32 usec_delay; /* retry interval delay time */
+ u32 timeout;/* retry ops timeout limit */
+ u16 size; /* data buffer size*/
+ u16 vf_num; /* Virtual Function num */
+ u16 pf_num; /* Physical Function num */
+ u16 sriov_st; /* Sriov state */
+ u16 en_vfs; /* user enabled vf num */
+ bool is_pf;
+
+ struct rnp_mbx_sync syncs[RNP_MBX_FW + 1];
+};
+
+struct rnp_eth_adapter;
+
+/* hw device description */
+struct rnp_hw {
+ struct rnp_eth_adapter *back; /* backup to the adapter handle */
+ void __iomem *e_ctrl; /* ethernet control bar */
+ void __iomem *c_ctrl; /* crypto control bar */
+ u32 c_blen; /* crypto bar size */
+
+ /* pci device info */
+ u16 device_id;
+ u16 vendor_id;
+ u16 max_vfs;/* device max support vf */
+
+ u16 pf_vf_num;
+ struct rnp_mbx_info mbx;
+};
+
+#endif /* _RNP_HW_H_ */
diff --git a/drivers/net/rnp/base/rnp_mbx.c b/drivers/net/rnp/base/rnp_mbx.c
new file mode 100644
index 00..0862f49912
--- /dev/null
+++ b/drivers/net/rnp/base/rnp_mbx.c
@@ -0,0 +1,510 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(C) 2023 Mucse IC Design Ltd.
+ */
+
+#include
+
+#include "rnp_hw.h"
+#include "rnp_mbx.h"
+#include "../rnp.h"
+
+/PF MBX OPS/
+static inline u16
+rnp_mbx_get_req(struct rnp_hw *hw, enum RNP_MBX_ID mbx_id)
+{
+ u32 reg = 0;
+
+ if (mbx_id == RNP_MBX_FW)
+ reg = RNP_FW2PF_SYNC;
+ else
+ reg = RNP_VF2PF_SYNC(mbx_id);
+ mb();
+
+ return RNP_E_REG_RD(hw, reg) & RNP_MBX_SYNC_REQ_MASK;
+}
+
+static inline u16
+rnp_mbx_get_ack(struct rnp_hw *hw, enum RNP_MBX_ID mbx_id)
+{
+ u32 reg = 0;
+ u32 v = 0;
+
+ if (mbx_id == RNP_MBX_FW)
+ reg = RNP_FW2PF_SYNC;
+ else
+ reg = RNP_VF2PF_SYNC(mbx_id);
+ mb();
+ v = RNP_E_REG_RD(hw, reg);
+
+ return (v & RNP_MBX_SYNC_ACK_MASK) >> RNP_MBX_SYNC_ACK_S;
+}
+
[PATCH v13 11/28] net/rnp: add RSS support operations
add support rss reta updata/query rss hash update/get
dev_configure add rss conf check.
Signed-off-by: Wenbo Cao
---
doc/guides/nics/features/rnp.ini| 4 +
doc/guides/nics/rnp.rst | 7 +
drivers/net/rnp/base/rnp_eth_regs.h | 16 ++
drivers/net/rnp/meson.build | 1 +
drivers/net/rnp/rnp.h | 7 +
drivers/net/rnp/rnp_ethdev.c| 23 ++
drivers/net/rnp/rnp_rss.c | 367
drivers/net/rnp/rnp_rss.h | 44
8 files changed, 469 insertions(+)
create mode 100644 drivers/net/rnp/rnp_rss.c
create mode 100644 drivers/net/rnp/rnp_rss.h
diff --git a/doc/guides/nics/features/rnp.ini b/doc/guides/nics/features/rnp.ini
index fd7d4b9d8d..2fc94825f5 100644
--- a/doc/guides/nics/features/rnp.ini
+++ b/doc/guides/nics/features/rnp.ini
@@ -8,5 +8,9 @@ Speed capabilities = Y
Queue start/stop = Y
Promiscuous mode = Y
Allmulticast mode= Y
+RSS hash = Y
+RSS key update = Y
+RSS reta update = Y
+Inner RSS= Y
Linux= Y
x86-64 = Y
diff --git a/doc/guides/nics/rnp.rst b/doc/guides/nics/rnp.rst
index 66ff7f4090..90af73d71f 100644
--- a/doc/guides/nics/rnp.rst
+++ b/doc/guides/nics/rnp.rst
@@ -45,6 +45,9 @@ Features
- Multiple queues for TX and RX
+- Receiver Side Steering (RSS)
+ Receiver Side Steering (RSS) on IPv4, IPv6, IPv4-TCP/UDP/SCTP,
IPv6-TCP/UDP/SCTP
+ Inner RSS is only support for vxlan/nvgre
- Promiscuous mode
Prerequisites and Pre-conditions
@@ -77,6 +80,10 @@ Listed below are the rte_eth functions supported:
* ``rte_eth_dev_tx_queue_stop``
* ``rte_eth_dev_start``
* ``rte_eth_dev_stop``
+* ``rte_eth_dev_rss_hash_conf_get``
+* ``rte_eth_dev_rss_hash_update``
+* ``rte_eth_dev_rss_reta_query``
+* ``rte_eth_dev_rss_reta_update``
* ``rte_eth_promiscuous_disable``
* ``rte_eth_promiscuous_enable``
* ``rte_eth_allmulticast_enable``
diff --git a/drivers/net/rnp/base/rnp_eth_regs.h
b/drivers/net/rnp/base/rnp_eth_regs.h
index 60766d2035..be7ed5b886 100644
--- a/drivers/net/rnp/base/rnp_eth_regs.h
+++ b/drivers/net/rnp/base/rnp_eth_regs.h
@@ -32,7 +32,23 @@
#define RNP_MAC_MULTICASE_TBL_EN RTE_BIT32(2)
#define RNP_MAC_UNICASE_TBL_EN RTE_BIT32(3)
/* rss function ctrl */
+#define RNP_RSS_INNER_CTRL _ETH_(0x805c)
+#define RNP_INNER_RSS_EN (1)
+#define RNP_INNER_RSS_DIS (0)
#define RNP_RSS_REDIR_TB(n, id) _ETH_(0xe000 + ((n) * 0x200) + ((id) * 0x4))
+#define RNP_RSS_MRQC_ADDR _ETH_(0x92a0)
+/* RSS policy */
+#define RNP_RSS_HASH_CFG_MASK (0x3F3)
+#define RNP_RSS_HASH_IPV4_TCP RTE_BIT32(16)
+#define RNP_RSS_HASH_IPV4 RTE_BIT32(17)
+#define RNP_RSS_HASH_IPV6 RTE_BIT32(20)
+#define RNP_RSS_HASH_IPV6_TCP RTE_BIT32(21)
+#define RNP_RSS_HASH_IPV4_UDP RTE_BIT32(22)
+#define RNP_RSS_HASH_IPV6_UDP RTE_BIT32(23)
+#define RNP_RSS_HASH_IPV4_SCTP RTE_BIT32(24)
+#define RNP_RSS_HASH_IPV6_SCTP RTE_BIT32(25)
+/* rss hash key */
+#define RNP_RSS_KEY_TABLE(idx) _ETH_(0x92d0 + ((idx) * 0x4))
#define RNP_TC_PORT_OFFSET(lane) _ETH_(0xe840 + 0x04 * (lane))
diff --git a/drivers/net/rnp/meson.build b/drivers/net/rnp/meson.build
index f88fa89a53..e20cc88497 100644
--- a/drivers/net/rnp/meson.build
+++ b/drivers/net/rnp/meson.build
@@ -22,4 +22,5 @@ includes += include_directories('base')
sources = files(
'rnp_ethdev.c',
'rnp_rxtx.c',
+'rnp_rss.c',
)
diff --git a/drivers/net/rnp/rnp.h b/drivers/net/rnp/rnp.h
index c38c31c128..17a6309632 100644
--- a/drivers/net/rnp/rnp.h
+++ b/drivers/net/rnp/rnp.h
@@ -111,6 +111,13 @@ struct rnp_eth_port {
struct rnp_tx_queue *tx_queues[RNP_MAX_RX_QUEUE_NUM];
struct rnp_hw *hw;
+ struct rte_eth_rss_conf rss_conf;
+ uint16_t last_rx_num;
+ bool rxq_num_changed;
+ bool reta_has_cfg;
+ bool hw_rss_en;
+ uint32_t indirtbl[RNP_RSS_INDIR_SIZE];
+
rte_spinlock_t rx_mac_lock;
bool port_stopped;
};
diff --git a/drivers/net/rnp/rnp_ethdev.c b/drivers/net/rnp/rnp_ethdev.c
index 43de704c90..75bc57746c 100644
--- a/drivers/net/rnp/rnp_ethdev.c
+++ b/drivers/net/rnp/rnp_ethdev.c
@@ -18,6 +18,7 @@
#include "base/rnp_dma_regs.h"
#include "base/rnp_mac_regs.h"
#include "rnp_rxtx.h"
+#include "rnp_rss.h"
static struct rte_eth_dev *
rnp_alloc_eth_port(struct rte_pci_device *pci, char *name)
@@ -235,6 +236,9 @@ static int rnp_dev_start(struct rte_eth_dev *eth_dev)
}
/* disable eth rx flow */
RNP_RX_ETH_DISABLE(hw, lane);
+ ret = rnp_dev_rss_configure(eth_dev);
+ if (ret)
+ return ret;
ret = rnp_rx_scattered_setup(eth_dev);
if (ret)
return ret;
@@ -302,6 +306,19 @@ static int rnp_disable_all_tx_queue(struct rte_eth_dev
*dev)
return ret;
}
+static int rnp_dev_configure(struct rte_eth_dev *eth_dev)
+{
+ struct rnp_eth_port *port = RNP_DEV_TO_PORT(eth_dev);
+
+
[PATCH v13 10/28] net/rnp: add support device start stop operations
add basic support for device to start/stop function.
Signed-off-by: Wenbo Cao
---
doc/guides/nics/rnp.rst | 2 +
drivers/net/rnp/base/rnp_common.c | 22 +++
drivers/net/rnp/base/rnp_common.h | 1 +
drivers/net/rnp/base/rnp_dma_regs.h | 10 +
drivers/net/rnp/base/rnp_eth_regs.h | 5 +
drivers/net/rnp/base/rnp_hw.h | 1 +
drivers/net/rnp/base/rnp_mac.h | 14 ++
drivers/net/rnp/base/rnp_mac_regs.h | 42
drivers/net/rnp/rnp.h | 3 +
drivers/net/rnp/rnp_ethdev.c| 284 +++-
10 files changed, 383 insertions(+), 1 deletion(-)
diff --git a/doc/guides/nics/rnp.rst b/doc/guides/nics/rnp.rst
index c3547c38b6..66ff7f4090 100644
--- a/doc/guides/nics/rnp.rst
+++ b/doc/guides/nics/rnp.rst
@@ -75,6 +75,8 @@ Listed below are the rte_eth functions supported:
* ``rte_eth_dev_rx_queue_stop``
* ``rte_eth_dev_tx_queue_start``
* ``rte_eth_dev_tx_queue_stop``
+* ``rte_eth_dev_start``
+* ``rte_eth_dev_stop``
* ``rte_eth_promiscuous_disable``
* ``rte_eth_promiscuous_enable``
* ``rte_eth_allmulticast_enable``
diff --git a/drivers/net/rnp/base/rnp_common.c
b/drivers/net/rnp/base/rnp_common.c
index 58de3bde03..2dacb5113e 100644
--- a/drivers/net/rnp/base/rnp_common.c
+++ b/drivers/net/rnp/base/rnp_common.c
@@ -79,3 +79,25 @@ rnp_setup_common_ops(struct rnp_hw *hw)
return 0;
}
+
+int rnp_clock_valid_check(struct rnp_hw *hw, u16 nr_lane)
+{
+ uint16_t timeout = 0;
+
+ do {
+ RNP_E_REG_WR(hw, RNP_RSS_REDIR_TB(nr_lane, 0), 0x7f);
+ udelay(10);
+ timeout++;
+ if (timeout >= 1000)
+ break;
+ } while (RNP_E_REG_RD(hw, RNP_RSS_REDIR_TB(nr_lane, 0)) != 0x7f);
+
+ if (timeout >= 1000) {
+ RNP_PMD_ERR("ethernet[%d] eth reg can't be write", nr_lane);
+ return -EPERM;
+ }
+ /* clear the dirty value */
+ RNP_E_REG_WR(hw, RNP_RSS_REDIR_TB(nr_lane, 0), 0);
+
+ return 0;
+}
diff --git a/drivers/net/rnp/base/rnp_common.h
b/drivers/net/rnp/base/rnp_common.h
index cf1996a675..91e2996398 100644
--- a/drivers/net/rnp/base/rnp_common.h
+++ b/drivers/net/rnp/base/rnp_common.h
@@ -12,5 +12,6 @@
((macaddr[4] << 8)) | (macaddr[5]))
int rnp_init_hw(struct rnp_hw *hw);
int rnp_setup_common_ops(struct rnp_hw *hw);
+int rnp_clock_valid_check(struct rnp_hw *hw, u16 nr_lane);
#endif /* _RNP_COMMON_H_ */
diff --git a/drivers/net/rnp/base/rnp_dma_regs.h
b/drivers/net/rnp/base/rnp_dma_regs.h
index 7c17741c55..8bc6fe581a 100644
--- a/drivers/net/rnp/base/rnp_dma_regs.h
+++ b/drivers/net/rnp/base/rnp_dma_regs.h
@@ -6,9 +6,19 @@
#define _RNP_DMA_REGS_H_
#define RNP_DMA_VERSION(0)
+#define RNP_DMA_CTRL (0x4)
+/* 1bit <-> 16 bytes dma addr size */
+#define RNP_DMA_SCATTER_MEM_MASK RTE_GENMASK32(31, 16)
+#define RNP_DMA_SCATTER_MEN_S (16)
+#define RNP_DMA_RX_MEM_PAD_EN RTE_BIT32(8)
+#define RTE_DMA_VEB_BYPASS RTE_BIT32(4)
+#define RNP_DMA_TXRX_LOOP RTE_BIT32(1)
+#define RNP_DMA_TXMRX_LOOP RTE_BIT32(0)
+
#define RNP_DMA_HW_EN (0x10)
#define RNP_DMA_EN_ALL (0b)
#define RNP_DMA_HW_STATE (0x14)
+
/* --- queue register --- */
/* queue enable */
#define RNP_RXQ_START(qid) _RING_(0x0010 + 0x100 * (qid))
diff --git a/drivers/net/rnp/base/rnp_eth_regs.h
b/drivers/net/rnp/base/rnp_eth_regs.h
index 10e3d954b8..60766d2035 100644
--- a/drivers/net/rnp/base/rnp_eth_regs.h
+++ b/drivers/net/rnp/base/rnp_eth_regs.h
@@ -10,6 +10,9 @@
#define RNP_E_FILTER_EN_ETH_(0x801c)
#define RNP_E_REDIR_EN _ETH_(0x8030)
+#define RNP_RX_ETH_F_CTRL(n) _ETH_(0x8070 + ((n) * 0x8))
+#define RNP_RX_ETH_F_OFF (0x7ff)
+#define RNP_RX_ETH_F_ON(0x270)
/* rx queue flow ctrl */
#define RNP_RX_FC_ENABLE _ETH_(0x8520)
#define RNP_RING_FC_EN(n) _ETH_(0x8524 + ((0x4) * ((n) / 32)))
@@ -28,6 +31,8 @@
#define RNP_MAC_HASH_MASK RTE_GENMASK32(11, 0)
#define RNP_MAC_MULTICASE_TBL_EN RTE_BIT32(2)
#define RNP_MAC_UNICASE_TBL_EN RTE_BIT32(3)
+/* rss function ctrl */
+#define RNP_RSS_REDIR_TB(n, id) _ETH_(0xe000 + ((n) * 0x200) + ((id) * 0x4))
#define RNP_TC_PORT_OFFSET(lane) _ETH_(0xe840 + 0x04 * (lane))
diff --git a/drivers/net/rnp/base/rnp_hw.h b/drivers/net/rnp/base/rnp_hw.h
index 9541b4e9be..548e3a4468 100644
--- a/drivers/net/rnp/base/rnp_hw.h
+++ b/drivers/net/rnp/base/rnp_hw.h
@@ -120,6 +120,7 @@ struct rnp_hw {
bool lane_is_sgmii[RNP_MAX_PORT_OF_PF];
struct rnp_mbx_info mbx;
struct rnp_fw_info fw_info;
+ u16 min_dma_size;
spinlock_t rxq_reset_lock; /* reset op isn't thread safe */
spinlock_t txq_reset_lock; /* reset op isn't thread safe */
diff --git a/drivers/net/rnp/base/rnp_mac.h b/drivers/net/rnp/base/rnp_mac.h
index 57cbd9e3d5..1dac903396 100644
--- a/drivers/
[PATCH v13 17/28] net/rnp: add Rx scatter segment version
add support scatter multi segment received.
Signed-off-by: Wenbo Cao
---
doc/guides/nics/features/rnp.ini | 1 +
doc/guides/nics/rnp.rst | 2 +
drivers/net/rnp/rnp_rxtx.c | 132 ++-
drivers/net/rnp/rnp_rxtx.h | 2 +
4 files changed, 135 insertions(+), 2 deletions(-)
diff --git a/doc/guides/nics/features/rnp.ini b/doc/guides/nics/features/rnp.ini
index 6d13370e8f..6453762745 100644
--- a/doc/guides/nics/features/rnp.ini
+++ b/doc/guides/nics/features/rnp.ini
@@ -15,5 +15,6 @@ RSS hash = Y
RSS key update = Y
RSS reta update = Y
Inner RSS= Y
+Scattered Rx = Y
Linux= Y
x86-64 = Y
diff --git a/doc/guides/nics/rnp.rst b/doc/guides/nics/rnp.rst
index 5b58a6941c..b3d6d4d425 100644
--- a/doc/guides/nics/rnp.rst
+++ b/doc/guides/nics/rnp.rst
@@ -51,6 +51,8 @@ Features
- Promiscuous mode
- Link state information
- MTU update
+- Jumbo frames
+- Scatter-Gather IO support
Prerequisites and Pre-conditions
diff --git a/drivers/net/rnp/rnp_rxtx.c b/drivers/net/rnp/rnp_rxtx.c
index 260c34113a..c012782e1d 100644
--- a/drivers/net/rnp/rnp_rxtx.c
+++ b/drivers/net/rnp/rnp_rxtx.c
@@ -826,7 +826,6 @@ rnp_xmit_simple(void *_txq, struct rte_mbuf **tx_pkts,
uint16_t nb_pkts)
if (txq->tx_next_rs > txq->attr.nb_desc)
txq->tx_next_rs = txq->tx_rs_thresh - 1;
}
-
txq->tx_tail = i;
rte_wmb();
@@ -835,9 +834,138 @@ rnp_xmit_simple(void *_txq, struct rte_mbuf **tx_pkts,
uint16_t nb_pkts)
return start;
}
+static int
+rnp_rxq_bulk_alloc(struct rnp_rx_queue *rxq,
+ volatile struct rnp_rx_desc *rxbd,
+ struct rnp_rxsw_entry *rxe,
+ bool bulk_alloc)
+{
+ struct rte_eth_dev_data *data;
+ struct rte_mbuf *nmb = NULL;
+ uint16_t update_tail;
+
+ if (!bulk_alloc) {
+ data = rte_eth_devices[rxq->attr.port_id].data;
+ nmb = rte_mbuf_raw_alloc(rxq->mb_pool);
+ if (unlikely(!nmb)) {
+ data->rx_mbuf_alloc_failed++;
+ return -ENOMEM;
+ }
+ rxbd->d.pkt_addr = 0;
+ rxbd->d.cmd = 0;
+ rxe->mbuf = NULL;
+ rxe->mbuf = nmb;
+ rxbd->d.pkt_addr = rnp_get_dma_addr(&rxq->attr, nmb);
+ }
+ if (rxq->rxrearm_nb > rxq->rx_free_thresh) {
+ rxq->rxrearm_nb -= rxq->rx_free_thresh;
+ rxq->rxrearm_start += rxq->rx_free_thresh;
+ if (rxq->rxrearm_start >= rxq->attr.nb_desc)
+ rxq->rxrearm_start = 0;
+ update_tail = (uint16_t)((rxq->rxrearm_start == 0) ?
+ (rxq->attr.nb_desc - 1) : (rxq->rxrearm_start -
1));
+ rte_io_wmb();
+ RNP_REG_WR(rxq->rx_tailreg, 0, update_tail);
+ }
+
+ return 0;
+}
+
+static __rte_always_inline uint16_t
+rnp_scattered_rx(void *rx_queue, struct rte_mbuf **rx_pkts,
+uint16_t nb_pkts)
+{
+ struct rnp_rx_queue *rxq = (struct rnp_rx_queue *)rx_queue;
+ volatile struct rnp_rx_desc *bd_ring = rxq->rx_bdr;
+ struct rte_mbuf *first_seg = rxq->pkt_first_seg;
+ struct rte_mbuf *last_seg = rxq->pkt_last_seg;
+ struct rnp_rxsw_entry *sw_ring = rxq->sw_ring;
+ volatile struct rnp_rx_desc *rxbd;
+ volatile struct rnp_rx_desc rxd;
+ struct rnp_rxsw_entry *rxe;
+ struct rte_mbuf *rxm;
+ uint16_t rx_pkt_len;
+ uint16_t nb_rx = 0;
+ uint16_t rx_status;
+ uint16_t rx_id;
+
+ if (unlikely(!rxq->rxq_started || !rxq->rx_link))
+ return 0;
+ rx_id = rxq->rx_tail;
+ while (nb_rx < nb_pkts) {
+ rxbd = &bd_ring[rx_id];
+ rx_status = rxbd->wb.qword1.cmd;
+ if (!(rx_status & rte_cpu_to_le_16(RNP_CMD_DD)))
+ break;
+ rte_atomic_thread_fence(rte_memory_order_acquire);
+ rxd = *rxbd;
+ rxe = &sw_ring[rx_id];
+ rxm = rxe->mbuf;
+ if (rnp_rxq_bulk_alloc(rxq, rxbd, rxe, false))
+ break;
+ rx_id = (rx_id + 1) & rxq->attr.nb_desc_mask;
+ rte_prefetch0(sw_ring[rx_id].mbuf);
+ if ((rx_id & 0x3) == 0) {
+ rte_prefetch0(&bd_ring[rx_id]);
+ rte_prefetch0(&sw_ring[rx_id]);
+ }
+ rx_pkt_len = rxd.wb.qword1.lens;
+ rxm->data_len = rx_pkt_len;
+ rxm->data_off = RTE_PKTMBUF_HEADROOM;
+ if (!first_seg) {
+ /* first segment pkt */
+ first_seg = rxm;
+ first_seg->nb_segs = 1;
+ first_seg->pkt_len = rx_pkt_len;
+ } else
[PATCH v13 13/28] net/rnp: add support link setup operations
add set link_down/link_up implement
Signed-off-by: Wenbo Cao
---
doc/guides/nics/rnp.rst | 2 +
drivers/net/rnp/base/rnp_eth_regs.h | 3 +
drivers/net/rnp/base/rnp_fw_cmd.c | 22 +++
drivers/net/rnp/base/rnp_fw_cmd.h | 6 ++
drivers/net/rnp/base/rnp_mbx_fw.c | 33 ++
drivers/net/rnp/base/rnp_mbx_fw.h | 1 +
drivers/net/rnp/rnp_ethdev.c| 4 ++
drivers/net/rnp/rnp_link.c | 99 +
drivers/net/rnp/rnp_link.h | 2 +
9 files changed, 172 insertions(+)
diff --git a/doc/guides/nics/rnp.rst b/doc/guides/nics/rnp.rst
index e6a1aadefc..f5a79492bb 100644
--- a/doc/guides/nics/rnp.rst
+++ b/doc/guides/nics/rnp.rst
@@ -85,6 +85,8 @@ Listed below are the rte_eth functions supported:
* ``rte_eth_dev_rss_hash_update``
* ``rte_eth_dev_rss_reta_query``
* ``rte_eth_dev_rss_reta_update``
+* ``rte_eth_dev_set_link_down``
+* ``rte_eth_dev_set_link_up``
* ``rte_eth_promiscuous_disable``
* ``rte_eth_promiscuous_enable``
* ``rte_eth_allmulticast_enable``
diff --git a/drivers/net/rnp/base/rnp_eth_regs.h
b/drivers/net/rnp/base/rnp_eth_regs.h
index be7ed5b886..c74886e70c 100644
--- a/drivers/net/rnp/base/rnp_eth_regs.h
+++ b/drivers/net/rnp/base/rnp_eth_regs.h
@@ -5,6 +5,9 @@
#ifndef _RNP_ETH_REGS_H
#define _RNP_ETH_REGS_H
+#define RNP_ETH_TX_FIFO_STATE _ETH_(0x0330)
+#define RNP_ETH_TX_FIFO_EMPT(lane) ((1 << (lane)) | (1 << ((lane) + 4)))
+
#define RNP_E_ENG_BYPASS _ETH_(0x8000)
#define RNP_E_VXLAN_PARSE_EN _ETH_(0x8004)
#define RNP_E_FILTER_EN_ETH_(0x801c)
diff --git a/drivers/net/rnp/base/rnp_fw_cmd.c
b/drivers/net/rnp/base/rnp_fw_cmd.c
index a4a2171868..ed0eddfd1f 100644
--- a/drivers/net/rnp/base/rnp_fw_cmd.c
+++ b/drivers/net/rnp/base/rnp_fw_cmd.c
@@ -107,6 +107,25 @@ rnp_build_lane_evet_mask(struct rnp_mbx_fw_cmd_req *req,
arg->event_en = req_arg->param2;
}
+static void
+rnp_build_ifup_down(struct rnp_mbx_fw_cmd_req *req,
+ struct rnp_fw_req_arg *req_arg,
+ void *cookie)
+{
+ struct rnp_ifup_down_req *arg =
+ (struct rnp_ifup_down_req *)req->data;
+
+ req->flags = 0;
+ req->opcode = RNP_IFUP_DOWN;
+ req->datalen = sizeof(*arg);
+ req->cookie = cookie;
+ req->reply_lo = 0;
+ req->reply_hi = 0;
+
+ arg->nr_lane = req_arg->param0;
+ arg->up = req_arg->param1;
+}
+
int rnp_build_fwcmd_req(struct rnp_mbx_fw_cmd_req *req,
struct rnp_fw_req_arg *arg,
void *cookie)
@@ -132,6 +151,9 @@ int rnp_build_fwcmd_req(struct rnp_mbx_fw_cmd_req *req,
case RNP_SET_LANE_EVENT_EN:
rnp_build_lane_evet_mask(req, arg, cookie);
break;
+ case RNP_IFUP_DOWN:
+ rnp_build_ifup_down(req, arg, cookie);
+ break;
default:
err = -EOPNOTSUPP;
}
diff --git a/drivers/net/rnp/base/rnp_fw_cmd.h
b/drivers/net/rnp/base/rnp_fw_cmd.h
index 6c6fd1803e..26db07ad36 100644
--- a/drivers/net/rnp/base/rnp_fw_cmd.h
+++ b/drivers/net/rnp/base/rnp_fw_cmd.h
@@ -309,6 +309,12 @@ struct rnp_link_stat_req {
struct rnp_port_stat states[RNP_MAX_PORT_OF_PF];
};
+struct rnp_ifup_down_req {
+ u32 nr_lane;
+ u32 up;
+ u8 rsvd[24];
+};
+
struct rnp_mbx_fw_cmd_req {
u16 flags;
u16 opcode;
diff --git a/drivers/net/rnp/base/rnp_mbx_fw.c
b/drivers/net/rnp/base/rnp_mbx_fw.c
index d4aa4a1626..4f351ce960 100644
--- a/drivers/net/rnp/base/rnp_mbx_fw.c
+++ b/drivers/net/rnp/base/rnp_mbx_fw.c
@@ -458,3 +458,36 @@ rnp_rcv_msg_from_fw(struct rnp_eth_adapter *adapter, u32
*msgbuf)
return 0;
}
+
+static void rnp_link_stat_reset(struct rnp_hw *hw, u16 lane)
+{
+ u32 state;
+
+ spin_lock(&hw->link_sync);
+ state = RNP_E_REG_RD(hw, RNP_FW_LINK_SYNC);
+ state &= ~RNP_LINK_MAGIC_MASK;
+ state |= RNP_LINK_MAGIC_CODE;
+ state &= ~RTE_BIT32(lane);
+
+ RNP_E_REG_WR(hw, RNP_FW_LINK_SYNC, state);
+ rte_spinlock_unlock(&hw->link_sync);
+}
+
+int rnp_mbx_fw_ifup_down(struct rnp_eth_port *port, bool up)
+{
+ u16 nr_lane = port->attr.nr_lane;
+ struct rnp_hw *hw = port->hw;
+ struct rnp_fw_req_arg arg;
+ int err;
+
+ memset(&arg, 0, sizeof(arg));
+ arg.opcode = RNP_IFUP_DOWN;
+ arg.param0 = nr_lane;
+ arg.param1 = up;
+
+ err = rnp_fw_send_norep_cmd(port, &arg);
+ /* force firmware send irq event to dpdk */
+ if (!err && up)
+ rnp_link_stat_reset(hw, nr_lane);
+ return err;
+}
diff --git a/drivers/net/rnp/base/rnp_mbx_fw.h
b/drivers/net/rnp/base/rnp_mbx_fw.h
index 159a0237be..397d2ec8c4 100644
--- a/drivers/net/rnp/base/rnp_mbx_fw.h
+++ b/drivers/net/rnp/base/rnp_mbx_fw.h
@@ -19,5 +19,6 @@ int rnp_fw_init(struct rnp_hw *hw);
int rnp_rcv_msg_from_fw(struct rnp_eth_adapter *adapter, u32 *msgbuf);
int rnp_fw_mbx_ifup_do
[PATCH v13 14/28] net/rnp: add Rx burst simple support
add only support simple recv pkts.
Signed-off-by: Wenbo Cao
---
drivers/net/rnp/rnp_ethdev.c | 7 ++
drivers/net/rnp/rnp_rxtx.c | 130 +++
drivers/net/rnp/rnp_rxtx.h | 5 ++
3 files changed, 142 insertions(+)
diff --git a/drivers/net/rnp/rnp_ethdev.c b/drivers/net/rnp/rnp_ethdev.c
index cd704d7a90..03bf91e74e 100644
--- a/drivers/net/rnp/rnp_ethdev.c
+++ b/drivers/net/rnp/rnp_ethdev.c
@@ -329,6 +329,8 @@ static int rnp_dev_start(struct rte_eth_dev *eth_dev)
rnp_dev_set_link_up(eth_dev);
/* enable eth rx flow */
RNP_RX_ETH_ENABLE(hw, lane);
+ rnp_rx_func_select(eth_dev);
+ rnp_tx_func_select(eth_dev);
port->port_stopped = 0;
return 0;
@@ -589,6 +591,11 @@ static int rnp_dev_infos_get(struct rte_eth_dev *eth_dev,
/* default port configure */
dev_info->default_rxconf = (struct rte_eth_rxconf) {
.rx_drop_en = 0,
+ .rx_thresh = {
+ .pthresh = RNP_RX_DESC_FETCH_TH,
+ .hthresh = RNP_RX_DESC_FETCH_BURST,
+ },
+ .rx_free_thresh = RNP_DEFAULT_RX_FREE_THRESH,
.offloads = 0,
};
diff --git a/drivers/net/rnp/rnp_rxtx.c b/drivers/net/rnp/rnp_rxtx.c
index e65bc06d36..a742646c7e 100644
--- a/drivers/net/rnp/rnp_rxtx.c
+++ b/drivers/net/rnp/rnp_rxtx.c
@@ -636,3 +636,133 @@ int rnp_rx_queue_start(struct rte_eth_dev *eth_dev,
uint16_t qidx)
return 0;
}
+
+#define RNP_CACHE_FETCH_RX (4)
+static __rte_always_inline int
+rnp_refill_rx_ring(struct rnp_rx_queue *rxq)
+{
+ volatile struct rnp_rx_desc *rxbd;
+ struct rnp_rxsw_entry *rx_swbd;
+ struct rte_eth_dev_data *data;
+ struct rte_mbuf *mb;
+ uint16_t j, i;
+ uint16_t rx_id;
+ int ret;
+
+ rxbd = rxq->rx_bdr + rxq->rxrearm_start;
+ rx_swbd = &rxq->sw_ring[rxq->rxrearm_start];
+ ret = rte_mempool_get_bulk(rxq->mb_pool, (void *)rx_swbd,
+ rxq->rx_free_thresh);
+ data = rte_eth_devices[rxq->attr.port_id].data;
+ if (unlikely(ret != 0)) {
+ if (rxq->rxrearm_nb + rxq->rx_free_thresh >= rxq->attr.nb_desc)
{
+ for (i = 0; i < RNP_CACHE_FETCH_RX; i++) {
+ rx_swbd[i].mbuf = &rxq->fake_mbuf;
+ rxbd[i].d.pkt_addr = 0;
+ rxbd[i].d.cmd = 0;
+ }
+ }
+ data->rx_mbuf_alloc_failed += rxq->rx_free_thresh;
+ return 0;
+ }
+ for (j = 0; j < rxq->rx_free_thresh; ++j) {
+ mb = rx_swbd[j].mbuf;
+ rte_mbuf_refcnt_set(mb, 1);
+ mb->data_off = RTE_PKTMBUF_HEADROOM;
+ mb->port = rxq->attr.port_id;
+
+ rxbd[j].d.pkt_addr = rnp_get_dma_addr(&rxq->attr, mb);
+ rxbd[j].d.cmd = 0;
+ }
+ rxq->rxrearm_start += rxq->rx_free_thresh;
+ if (rxq->rxrearm_start >= rxq->attr.nb_desc - 1)
+ rxq->rxrearm_start = 0;
+ rxq->rxrearm_nb -= rxq->rx_free_thresh;
+
+ rx_id = (uint16_t)((rxq->rxrearm_start == 0) ?
+ (rxq->attr.nb_desc - 1) : (rxq->rxrearm_start - 1));
+ rte_wmb();
+ RNP_REG_WR(rxq->rx_tailreg, 0, rx_id);
+
+ return j;
+}
+
+static __rte_always_inline uint16_t
+rnp_recv_pkts(void *_rxq, struct rte_mbuf **rx_pkts, uint16_t nb_pkts)
+{
+ struct rnp_rx_queue *rxq = (struct rnp_rx_queue *)_rxq;
+ struct rnp_rxsw_entry *rx_swbd;
+ uint32_t state_cmd[RNP_CACHE_FETCH_RX];
+ uint32_t pkt_len[RNP_CACHE_FETCH_RX] = {0};
+ volatile struct rnp_rx_desc *rxbd;
+ struct rte_mbuf *nmb;
+ int nb_dd, nb_rx = 0;
+ int i, j;
+
+ if (unlikely(!rxq->rxq_started || !rxq->rx_link))
+ return 0;
+ nb_pkts = RTE_ALIGN_FLOOR(nb_pkts, RNP_CACHE_FETCH_RX);
+ rxbd = &rxq->rx_bdr[rxq->rx_tail];
+ rte_prefetch0(rxbd);
+ if (rxq->rxrearm_nb > rxq->rx_free_thresh)
+ rnp_refill_rx_ring(rxq);
+
+ if (!(rxbd->wb.qword1.cmd & RNP_CMD_DD))
+ return 0;
+
+ rx_swbd = &rxq->sw_ring[rxq->rx_tail];
+ for (i = 0; i < nb_pkts;
+ i += RNP_CACHE_FETCH_RX, rxbd += RNP_CACHE_FETCH_RX,
+ rx_swbd += RNP_CACHE_FETCH_RX) {
+ for (j = 0; j < RNP_CACHE_FETCH_RX; j++)
+ state_cmd[j] = rxbd[j].wb.qword1.cmd;
+ rte_atomic_thread_fence(rte_memory_order_acquire);
+
+ for (nb_dd = 0; nb_dd < RNP_CACHE_FETCH_RX &&
+ (state_cmd[nb_dd] &
rte_cpu_to_le_16(RNP_CMD_DD));
+ nb_dd++)
+ ;
+ for (j = 0; j < nb_dd; j++)
+ pkt_len[j] = rxbd[j].wb.qword1.lens;
+
+ for (j = 0; j < nb_dd; ++j) {
+
[PATCH v13 15/28] net/rnp: add Tx burst simple support
add only support simple send pkts.
Signed-off-by: Wenbo Cao
---
drivers/net/rnp/rnp_ethdev.c | 6 +++
drivers/net/rnp/rnp_rxtx.c | 85 +++-
drivers/net/rnp/rnp_rxtx.h | 1 +
3 files changed, 91 insertions(+), 1 deletion(-)
diff --git a/drivers/net/rnp/rnp_ethdev.c b/drivers/net/rnp/rnp_ethdev.c
index 03bf91e74e..21f6fed241 100644
--- a/drivers/net/rnp/rnp_ethdev.c
+++ b/drivers/net/rnp/rnp_ethdev.c
@@ -600,6 +600,12 @@ static int rnp_dev_infos_get(struct rte_eth_dev *eth_dev,
};
dev_info->default_txconf = (struct rte_eth_txconf) {
+ .tx_thresh = {
+ .pthresh = RNP_TX_DESC_FETCH_TH,
+ .hthresh = RNP_TX_DESC_FETCH_BURST,
+ },
+ .tx_free_thresh = RNP_DEFAULT_TX_FREE_THRESH,
+ .tx_rs_thresh = RNP_DEFAULT_TX_RS_THRESH,
.offloads = 0,
};
diff --git a/drivers/net/rnp/rnp_rxtx.c b/drivers/net/rnp/rnp_rxtx.c
index a742646c7e..260c34113a 100644
--- a/drivers/net/rnp/rnp_rxtx.c
+++ b/drivers/net/rnp/rnp_rxtx.c
@@ -752,6 +752,89 @@ rnp_recv_pkts(void *_rxq, struct rte_mbuf **rx_pkts,
uint16_t nb_pkts)
return nb_rx;
}
+static __rte_always_inline int
+rnp_clean_tx_ring(struct rnp_tx_queue *txq)
+{
+ volatile struct rnp_tx_desc *txbd;
+ struct rnp_txsw_entry *tx_swbd;
+ struct rte_mbuf *m;
+ uint16_t next_dd;
+ uint16_t i;
+
+ txbd = &txq->tx_bdr[txq->tx_next_dd];
+ if (!(txbd->d.cmd & RNP_CMD_DD))
+ return 0;
+ *txbd = txq->zero_desc;
+ next_dd = txq->tx_next_dd - (txq->tx_free_thresh - 1);
+ tx_swbd = &txq->sw_ring[next_dd];
+
+ for (i = 0; i < txq->tx_rs_thresh; ++i, ++tx_swbd) {
+ if (tx_swbd->mbuf) {
+ m = tx_swbd->mbuf;
+ rte_pktmbuf_free_seg(m);
+ tx_swbd->mbuf = NULL;
+ }
+ }
+ txq->nb_tx_free = (txq->nb_tx_free + txq->tx_rs_thresh);
+ txq->tx_next_dd = (txq->tx_next_dd + txq->tx_rs_thresh) &
+ txq->attr.nb_desc_mask;
+
+ return 0;
+}
+
+static __rte_always_inline uint16_t
+rnp_xmit_simple(void *_txq, struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
+{
+ struct rnp_tx_queue *txq = (struct rnp_tx_queue *)_txq;
+ volatile struct rnp_tx_desc *txbd;
+ struct rnp_txsw_entry *tx_swbd;
+ uint64_t phy;
+ uint16_t start;
+ uint16_t i;
+
+ if (unlikely(!txq->txq_started || !txq->tx_link))
+ return 0;
+
+ if (txq->nb_tx_free < txq->tx_free_thresh)
+ rnp_clean_tx_ring(txq);
+
+ nb_pkts = RTE_MIN(txq->nb_tx_free, nb_pkts);
+ if (!nb_pkts)
+ return 0;
+ start = nb_pkts;
+ i = txq->tx_tail;
+
+ while (nb_pkts--) {
+ txbd = &txq->tx_bdr[i];
+ tx_swbd = &txq->sw_ring[i];
+ tx_swbd->mbuf = *tx_pkts++;
+ phy = rnp_get_dma_addr(&txq->attr, tx_swbd->mbuf);
+ txbd->d.addr = phy;
+ if (unlikely(tx_swbd->mbuf->data_len > RNP_MAC_MAXFRM_SIZE))
+ tx_swbd->mbuf->data_len = 0;
+ txbd->d.blen = tx_swbd->mbuf->data_len;
+ txbd->d.cmd = RNP_CMD_EOP;
+
+ i = (i + 1) & txq->attr.nb_desc_mask;
+ }
+ txq->nb_tx_free -= start;
+ if (txq->tx_tail + start > txq->tx_next_rs) {
+ txbd = &txq->tx_bdr[txq->tx_next_rs];
+ txbd->d.cmd |= RNP_CMD_RS;
+ txq->tx_next_rs = (txq->tx_next_rs + txq->tx_rs_thresh);
+
+ if (txq->tx_next_rs > txq->attr.nb_desc)
+ txq->tx_next_rs = txq->tx_rs_thresh - 1;
+ }
+
+ txq->tx_tail = i;
+
+ rte_wmb();
+ RNP_REG_WR(txq->tx_tailreg, 0, i);
+
+ return start;
+}
+
int rnp_rx_func_select(struct rte_eth_dev *dev)
{
dev->rx_pkt_burst = rnp_recv_pkts;
@@ -761,7 +844,7 @@ int rnp_rx_func_select(struct rte_eth_dev *dev)
int rnp_tx_func_select(struct rte_eth_dev *dev)
{
- dev->tx_pkt_burst = rte_eth_pkt_burst_dummy;
+ dev->tx_pkt_burst = rnp_xmit_simple;
dev->tx_pkt_prepare = rte_eth_pkt_burst_dummy;
return 0;
diff --git a/drivers/net/rnp/rnp_rxtx.h b/drivers/net/rnp/rnp_rxtx.h
index 39e5184502..a8fd8d052f 100644
--- a/drivers/net/rnp/rnp_rxtx.h
+++ b/drivers/net/rnp/rnp_rxtx.h
@@ -89,6 +89,7 @@ struct rnp_tx_queue {
const struct rte_memzone *rz;
uint64_t ring_phys_addr; /* tx dma ring physical addr */
volatile struct rnp_tx_desc *tx_bdr; /* tx dma ring virtual addr */
+ volatile struct rnp_tx_desc zero_desc;
struct rnp_txsw_entry *sw_ring; /* tx software ring addr */
volatile void *tx_tailreg; /* hw desc tail register */
volatile void *tx_headreg; /* hw desc head register*/
--
2.25.1
[PATCH v13 18/28] net/rnp: add Tx multiple segment version
add support multiple segs mbuf send.
Signed-off-by: Wenbo Cao
---
drivers/net/rnp/rnp_rxtx.c | 127 -
drivers/net/rnp/rnp_rxtx.h | 3 +-
2 files changed, 127 insertions(+), 3 deletions(-)
diff --git a/drivers/net/rnp/rnp_rxtx.c b/drivers/net/rnp/rnp_rxtx.c
index c012782e1d..b97761d12e 100644
--- a/drivers/net/rnp/rnp_rxtx.c
+++ b/drivers/net/rnp/rnp_rxtx.c
@@ -369,9 +369,11 @@ rnp_tx_queue_sw_reset(struct rnp_tx_queue *txq)
sw_ring[prev].next_id = idx;
prev = idx;
}
+ txq->last_desc_cleaned = txq->attr.nb_desc - 1;
txq->nb_tx_free = txq->attr.nb_desc - 1;
txq->tx_next_dd = txq->tx_rs_thresh - 1;
txq->tx_next_rs = txq->tx_rs_thresh - 1;
+ txq->nb_tx_used = 0;
txq->tx_tail = 0;
size = (txq->attr.nb_desc + RNP_TX_MAX_BURST_SIZE);
@@ -857,6 +859,7 @@ rnp_rxq_bulk_alloc(struct rnp_rx_queue *rxq,
rxe->mbuf = nmb;
rxbd->d.pkt_addr = rnp_get_dma_addr(&rxq->attr, nmb);
}
+ rxq->rxrearm_nb++;
if (rxq->rxrearm_nb > rxq->rx_free_thresh) {
rxq->rxrearm_nb -= rxq->rx_free_thresh;
rxq->rxrearm_start += rxq->rx_free_thresh;
@@ -924,7 +927,6 @@ rnp_scattered_rx(void *rx_queue, struct rte_mbuf **rx_pkts,
first_seg->nb_segs++;
last_seg->next = rxm;
}
- rxq->rxrearm_nb++;
if (!(rx_status & rte_cpu_to_le_16(RNP_CMD_EOP))) {
last_seg = rxm;
continue;
@@ -947,6 +949,107 @@ rnp_scattered_rx(void *rx_queue, struct rte_mbuf
**rx_pkts,
return nb_rx;
}
+static __rte_always_inline uint16_t
+rnp_multiseg_clean_txq(struct rnp_tx_queue *txq)
+{
+ uint16_t last_desc_cleaned = txq->last_desc_cleaned;
+ struct rnp_txsw_entry *sw_ring = txq->sw_ring;
+ volatile struct rnp_tx_desc *txbd;
+ uint16_t desc_to_clean_to;
+ uint16_t nb_tx_to_clean;
+
+ desc_to_clean_to = (uint16_t)(last_desc_cleaned + txq->tx_rs_thresh);
+ desc_to_clean_to = desc_to_clean_to & (txq->attr.nb_desc - 1);
+
+ desc_to_clean_to = sw_ring[desc_to_clean_to].last_id;
+ txbd = &txq->tx_bdr[desc_to_clean_to];
+ if (!(txbd->d.cmd & RNP_CMD_DD))
+ return txq->nb_tx_free;
+
+ if (last_desc_cleaned > desc_to_clean_to)
+ nb_tx_to_clean = (uint16_t)((txq->attr.nb_desc -
+ last_desc_cleaned) + desc_to_clean_to);
+ else
+ nb_tx_to_clean = (uint16_t)(desc_to_clean_to -
+ last_desc_cleaned);
+
+ txbd->d.cmd = 0;
+
+ txq->last_desc_cleaned = desc_to_clean_to;
+ txq->nb_tx_free = (uint16_t)(txq->nb_tx_free + nb_tx_to_clean);
+
+ return txq->nb_tx_free;
+}
+
+static __rte_always_inline uint16_t
+rnp_multiseg_xmit_pkts(void *_txq, struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
+{
+ struct rnp_tx_queue *txq = (struct rnp_tx_queue *)_txq;
+ volatile struct rnp_tx_desc *txbd;
+ struct rnp_txsw_entry *txe, *txn;
+ struct rte_mbuf *tx_pkt, *m_seg;
+ uint16_t send_pkts = 0;
+ uint16_t nb_used_bd;
+ uint16_t tx_last;
+ uint16_t nb_tx;
+ uint16_t tx_id;
+
+ if (unlikely(!txq->txq_started || !txq->tx_link))
+ return 0;
+ if (txq->nb_tx_free < txq->tx_free_thresh)
+ rnp_multiseg_clean_txq(txq);
+ if (unlikely(txq->nb_tx_free == 0))
+ return 0;
+ tx_id = txq->tx_tail;
+ txbd = &txq->tx_bdr[tx_id];
+ txe = &txq->sw_ring[tx_id];
+ for (nb_tx = 0; nb_tx < nb_pkts; nb_tx++) {
+ tx_pkt = tx_pkts[nb_tx];
+ nb_used_bd = tx_pkt->nb_segs;
+ tx_last = (uint16_t)(tx_id + nb_used_bd - 1);
+ if (tx_last >= txq->attr.nb_desc)
+ tx_last = (uint16_t)(tx_last - txq->attr.nb_desc);
+ if (nb_used_bd > txq->nb_tx_free)
+ if (nb_used_bd > rnp_multiseg_clean_txq(txq))
+ break;
+ m_seg = tx_pkt;
+ do {
+ txbd = &txq->tx_bdr[tx_id];
+ *txbd = txq->zero_desc;
+ txn = &txq->sw_ring[txe->next_id];
+ if (txe->mbuf) {
+ rte_pktmbuf_free_seg(txe->mbuf);
+ txe->mbuf = NULL;
+ }
+ txe->mbuf = m_seg;
+ txe->last_id = tx_last;
+ txbd->d.addr = rnp_get_dma_addr(&txq->attr, m_seg);
+ txbd->d.blen = rte_cpu_to_le_32(m_seg->data_len);
+ txbd->d.cmd &= ~RNP_CMD_EOP;
+ txbd->d.cmd |= RNP_DATA_DESC;
+ m_seg = m_seg->next;
+ tx_i
Re: [PATCH 00/10] enable "app" to be compiled with MSVC
On Tue, Feb 11, 2025 at 11:02 PM Andre Muezerie wrote: > > This series fixes many issues that prevent the "app" directory > from being compiled with MSVC. > > Andre Muezerie (10): > eal: add workaround for __builtin_constant_p > test_alarm: avoid warning about different qualifiers > test-pmd: fix printf format string mismatch > test-pmd: do explicit 64-bit shift to avoid implicit conversion > test-pmd: avoid undefined behavior > test-pmd: avoid non-constant initializer > test-pmd: don't return value from void function > test-pmd: declare lcore_count atomic when using C11 memory model > test: add workaround for __builtin_constant_p in test_memcpy_perf > app: enable app directory to be compiled with MSVC > > app/meson.build | 4 -- > app/test-pmd/cmdline.c | 19 +++-- > app/test-pmd/cmdline_flow.c | 22 +++--- > app/test-pmd/csumonly.c | 12 ++-- > app/test-pmd/meson.build| 4 +- > app/test-pmd/testpmd.c | 2 +- > app/test-pmd/util.c | 2 +- > app/test/test_alarm.c | 12 ++-- > app/test/test_memcpy_perf.c | 106 ++-- > app/test/test_ring_perf.c | 6 +- > lib/eal/include/generic/rte_pause.h | 2 +- > lib/eal/include/rte_common.h| 12 > 12 files changed, 114 insertions(+), 89 deletions(-) The last patch of the series was not received on the ml. This prevents CI from testing it. Please resubmit the whole series. Thanks. -- David Marchand
Re: [PATCH] ci: build with MSVC in GHA
On Tue, Feb 18, 2025 at 03:10:13PM +0100, David Marchand wrote: > On Tue, Feb 18, 2025 at 2:32 PM Aaron Conole wrote: > > > > David Marchand writes: > > > > > Add a build target in Windows 2022 for MSVC. > > > > > > Loading MSVC environment and having successive commands in a single > > > (cmd or pwsh) shell does not seem to correctly report if one of the > > > commands failed. > > > For now, load MSVC environment and run commands in separate shells. > > > > > > Signed-off-by: David Marchand > > > --- > > > > LGTM. > > > > Acked-by: Aaron Conole > > > > How far back will this work (ie: can we backport this to older branches?) > > MSVC was added in 23.11. > The patch applies cleanly and runs on 23.11 (where only lib/kvargs, > lib/log and lib/telemetry are compiled) and 24.11. > > I can add Cc: stable if you think it is worth having it in LTS (UNH > provides some coverage in LTS releases). > > > -- > David Marchand LGTM. -- Andre Muezerie
[PATCH v3 10/10] eal: deprecate weak symbols
Mark __rte_weak as deprecated. It will avoid having to support such feature with other linkers, plus this was never really needed. Signed-off-by: David Marchand --- doc/guides/rel_notes/release_25_03.rst | 2 ++ lib/eal/include/rte_common.h | 6 +- 2 files changed, 7 insertions(+), 1 deletion(-) diff --git a/doc/guides/rel_notes/release_25_03.rst b/doc/guides/rel_notes/release_25_03.rst index ca67c17c5c..8867a4bd74 100644 --- a/doc/guides/rel_notes/release_25_03.rst +++ b/doc/guides/rel_notes/release_25_03.rst @@ -188,6 +188,8 @@ API Changes * eal: The ``__rte_packed`` macro for packing data is replaced with ``__rte_packed_begin`` / ``__rte_packed_end``. +* eal: The ``__rte_weak`` macro is deprecated and will be removed in a future release. + * net: Changed the API for CRC calculation to be thread-safe. An opaque context argument was introduced to the net CRC API containing the algorithm type and length. diff --git a/lib/eal/include/rte_common.h b/lib/eal/include/rte_common.h index 3f77b7624e..4be65376a5 100644 --- a/lib/eal/include/rte_common.h +++ b/lib/eal/include/rte_common.h @@ -179,7 +179,11 @@ typedef uint16_t unaligned_uint16_t; /** * Mark a function or variable to a weak reference. */ -#define __rte_weak __attribute__((__weak__)) +#ifdef RTE_TOOLCHAIN_MSVC +#define __rte_weak RTE_DEPRECATED(__rte_weak) +#else +#define __rte_weak RTE_DEPRECATED(__rte_weak) __attribute__((__weak__)) +#endif /** * Mark a function to be pure. -- 2.48.1
[PATCH v3 08/10] net/virtio: remove weak symbols
Rather than use weak symbols, expose stubs symbols when needed.
Signed-off-by: David Marchand
Reviewed-by: Maxime Coquelin
---
drivers/net/virtio/meson.build | 5 +
drivers/net/virtio/virtio_rxtx.c| 6 --
drivers/net/virtio/virtio_rxtx_simple.c | 4 +++-
3 files changed, 12 insertions(+), 3 deletions(-)
diff --git a/drivers/net/virtio/meson.build b/drivers/net/virtio/meson.build
index 4302595424..c1c4a85bea 100644
--- a/drivers/net/virtio/meson.build
+++ b/drivers/net/virtio/meson.build
@@ -28,6 +28,7 @@ cflags += no_wvla_cflag
if arch_subdir == 'x86'
if cc_has_avx512
cflags += ['-DCC_AVX512_SUPPORT']
+cflags += ['-DVIRTIO_RXTX_PACKED_VEC']
virtio_avx512_lib = static_library('virtio_avx512_lib',
'virtio_rxtx_packed.c',
dependencies: [static_rte_ethdev,
@@ -41,11 +42,15 @@ if arch_subdir == 'x86'
cflags += '-DVIRTIO_CLANG_UNROLL_PRAGMA'
endif
endif
+cflags += ['-DVIRTIO_RXTX_VEC']
sources += files('virtio_rxtx_simple_sse.c')
elif arch_subdir == 'ppc'
+cflags += ['-DVIRTIO_RXTX_VEC']
sources += files('virtio_rxtx_simple_altivec.c')
elif arch_subdir == 'arm' and dpdk_conf.get('RTE_ARCH_64')
+cflags += ['-DVIRTIO_RXTX_PACKED_VEC']
sources += files('virtio_rxtx_packed.c')
+cflags += ['-DVIRTIO_RXTX_VEC']
sources += files('virtio_rxtx_simple_neon.c')
endif
diff --git a/drivers/net/virtio/virtio_rxtx.c b/drivers/net/virtio/virtio_rxtx.c
index b67f063b31..2a0d1b6101 100644
--- a/drivers/net/virtio/virtio_rxtx.c
+++ b/drivers/net/virtio/virtio_rxtx.c
@@ -2045,7 +2045,8 @@ virtio_xmit_pkts_inorder(void *tx_queue,
return nb_tx;
}
-__rte_weak uint16_t
+#ifndef VIRTIO_RXTX_PACKED_VEC
+uint16_t
virtio_recv_pkts_packed_vec(void *rx_queue __rte_unused,
struct rte_mbuf **rx_pkts __rte_unused,
uint16_t nb_pkts __rte_unused)
@@ -2053,10 +2054,11 @@ virtio_recv_pkts_packed_vec(void *rx_queue __rte_unused,
return 0;
}
-__rte_weak uint16_t
+uint16_t
virtio_xmit_pkts_packed_vec(void *tx_queue __rte_unused,
struct rte_mbuf **tx_pkts __rte_unused,
uint16_t nb_pkts __rte_unused)
{
return 0;
}
+#endif /* DVIRTIO_RXTX_PACKED_VEC */
diff --git a/drivers/net/virtio/virtio_rxtx_simple.c
b/drivers/net/virtio/virtio_rxtx_simple.c
index 439e00a7e1..aa96c9c488 100644
--- a/drivers/net/virtio/virtio_rxtx_simple.c
+++ b/drivers/net/virtio/virtio_rxtx_simple.c
@@ -43,8 +43,9 @@ virtio_rxq_vec_setup(struct virtnet_rx *rxq)
return 0;
}
+#ifndef VIRTIO_RXTX_VEC
/* Stub for linkage when arch specific implementation is not available */
-__rte_weak uint16_t
+uint16_t
virtio_recv_pkts_vec(void *rx_queue __rte_unused,
struct rte_mbuf **rx_pkts __rte_unused,
uint16_t nb_pkts __rte_unused)
@@ -52,3 +53,4 @@ virtio_recv_pkts_vec(void *rx_queue __rte_unused,
rte_panic("Wrong weak function linked by linker\n");
return 0;
}
+#endif /* VIRTIO_RXTX_VEC */
--
2.48.1
[PATCH v3 06/10] net/fm10k: remove weak symbols
Rather than use weak symbols, expose stubs symbols when needed.
Signed-off-by: David Marchand
Acked-by: Bruce Richardson
---
drivers/net/intel/fm10k/fm10k_ethdev.c | 20 +++-
1 file changed, 11 insertions(+), 9 deletions(-)
diff --git a/drivers/net/intel/fm10k/fm10k_ethdev.c
b/drivers/net/intel/fm10k/fm10k_ethdev.c
index 747042d621..75ce2e19cf 100644
--- a/drivers/net/intel/fm10k/fm10k_ethdev.c
+++ b/drivers/net/intel/fm10k/fm10k_ethdev.c
@@ -129,14 +129,15 @@ fm10k_mbx_unlock(struct fm10k_hw *hw)
rte_spinlock_unlock(FM10K_DEV_PRIVATE_TO_MBXLOCK(hw->back));
}
-/* Stubs needed for linkage when vPMD is disabled */
-__rte_weak int
+#ifndef RTE_ARCH_X86
+/* Stubs for non x86 architectures. */
+int
fm10k_rx_vec_condition_check(__rte_unused struct rte_eth_dev *dev)
{
return -1;
}
-__rte_weak uint16_t
+uint16_t
fm10k_recv_pkts_vec(
__rte_unused void *rx_queue,
__rte_unused struct rte_mbuf **rx_pkts,
@@ -145,7 +146,7 @@ fm10k_recv_pkts_vec(
return 0;
}
-__rte_weak uint16_t
+uint16_t
fm10k_recv_scattered_pkts_vec(
__rte_unused void *rx_queue,
__rte_unused struct rte_mbuf **rx_pkts,
@@ -154,39 +155,40 @@ fm10k_recv_scattered_pkts_vec(
return 0;
}
-__rte_weak int
+int
fm10k_rxq_vec_setup(__rte_unused struct fm10k_rx_queue *rxq)
{
return -1;
}
-__rte_weak void
+void
fm10k_rx_queue_release_mbufs_vec(
__rte_unused struct fm10k_rx_queue *rxq)
{
return;
}
-__rte_weak void
+void
fm10k_txq_vec_setup(__rte_unused struct fm10k_tx_queue *txq)
{
return;
}
-__rte_weak int
+int
fm10k_tx_vec_condition_check(__rte_unused struct fm10k_tx_queue *txq)
{
return -1;
}
-__rte_weak uint16_t
+uint16_t
fm10k_xmit_fixed_burst_vec(__rte_unused void *tx_queue,
__rte_unused struct rte_mbuf **tx_pkts,
__rte_unused uint16_t nb_pkts)
{
return 0;
}
+#endif /* RTE_ARCH_X86 */
/*
* reset queue to initial state, allocate software buffers used when starting
--
2.48.1
[PATCH v3 09/10] app/compress-perf: remove weak symbols
Remove dead code, those symbols are never used as the "strong" symbols
are always linked in.
Signed-off-by: David Marchand
---
app/test-compress-perf/main.c | 62 ---
1 file changed, 62 deletions(-)
diff --git a/app/test-compress-perf/main.c b/app/test-compress-perf/main.c
index fa366123ed..70ce4316cc 100644
--- a/app/test-compress-perf/main.c
+++ b/app/test-compress-perf/main.c
@@ -521,65 +521,3 @@ main(int argc, char **argv)
}
return ret;
}
-
-__rte_weak void *
-cperf_cyclecount_test_constructor(uint8_t dev_id __rte_unused,
-uint16_t qp_id __rte_unused,
-struct comp_test_data *options __rte_unused)
-{
- RTE_LOG(INFO, USER1, "Cycle count test is not supported yet\n");
- return NULL;
-}
-
-__rte_weak void
-cperf_cyclecount_test_destructor(void *arg __rte_unused)
-{
- RTE_LOG(INFO, USER1, "Something wrong happened!!!\n");
-}
-
-__rte_weak int
-cperf_cyclecount_test_runner(void *test_ctx __rte_unused)
-{
- return 0;
-}
-
-__rte_weak void *
-cperf_throughput_test_constructor(uint8_t dev_id __rte_unused,
-uint16_t qp_id __rte_unused,
-struct comp_test_data *options __rte_unused)
-{
- RTE_LOG(INFO, USER1, "Benchmark test is not supported yet\n");
- return NULL;
-}
-
-__rte_weak void
-cperf_throughput_test_destructor(void *arg __rte_unused)
-{
-
-}
-
-__rte_weak int
-cperf_throughput_test_runner(void *test_ctx __rte_unused)
-{
- return 0;
-}
-__rte_weak void *
-cperf_verify_test_constructor(uint8_t dev_id __rte_unused,
-uint16_t qp_id __rte_unused,
-struct comp_test_data *options __rte_unused)
-{
- RTE_LOG(INFO, USER1, "Verify test is not supported yet\n");
- return NULL;
-}
-
-__rte_weak void
-cperf_verify_test_destructor(void *arg __rte_unused)
-{
-
-}
-
-__rte_weak int
-cperf_verify_test_runner(void *test_ctx __rte_unused)
-{
- return 0;
-}
--
2.48.1
[PATCH v2 03/10] test-pmd: fix printf format string mismatch
Compiling with MSVC results in warnings like the one below:
app/test-pmd/csumonly.c(1085): warning C4477: 'printf' : format string
'%d' requires an argument of type 'int',
but variadic argument 1 has type 'uint64_t'
Signed-off-by: Andre Muezerie
Signed-off-by: Chengwen Feng
---
app/test-pmd/csumonly.c | 23 ---
1 file changed, 12 insertions(+), 11 deletions(-)
diff --git a/app/test-pmd/csumonly.c b/app/test-pmd/csumonly.c
index d77a140641..8de5ad6ad9 100644
--- a/app/test-pmd/csumonly.c
+++ b/app/test-pmd/csumonly.c
@@ -1070,7 +1070,7 @@ pkt_burst_checksum_forward(struct fwd_stream *fs)
info.l2_len, rte_be_to_cpu_16(info.ethertype),
info.l3_len, info.l4_proto, info.l4_len, buf);
if (rx_ol_flags & RTE_MBUF_F_RX_LRO)
- printf("rx: m->lro_segsz=%u\n", m->tso_segsz);
+ printf("rx: m->lro_segsz=%" PRIu64 "\n",
(uint64_t)m->tso_segsz);
if (info.is_tunnel == 1)
printf("rx: outer_l2_len=%d outer_ethertype=%x "
"outer_l3_len=%d\n", info.outer_l2_len,
@@ -1082,28 +1082,29 @@ pkt_burst_checksum_forward(struct fwd_stream *fs)
RTE_ETH_TX_OFFLOAD_TCP_CKSUM |
RTE_ETH_TX_OFFLOAD_SCTP_CKSUM)) ||
info.tso_segsz != 0)
- printf("tx: m->l2_len=%d m->l3_len=%d "
- "m->l4_len=%d\n",
- m->l2_len, m->l3_len, m->l4_len);
+ printf("tx: m->l2_len=%" PRIu64 " m->l3_len=%"
PRIu64
+ " m->l4_len=%" PRIu64 "\n",
+ (uint64_t)m->l2_len,
(uint64_t)m->l3_len,
+ (uint64_t)m->l4_len);
if (info.is_tunnel == 1) {
if ((tx_offloads &
RTE_ETH_TX_OFFLOAD_OUTER_IPV4_CKSUM) ||
(tx_offloads &
RTE_ETH_TX_OFFLOAD_OUTER_UDP_CKSUM) ||
(tx_ol_flags & RTE_MBUF_F_TX_OUTER_IPV6))
- printf("tx: m->outer_l2_len=%d "
- "m->outer_l3_len=%d\n",
- m->outer_l2_len,
- m->outer_l3_len);
+ printf("tx: m->outer_l2_len=%" PRIu64
+ " m->outer_l3_len=%" PRIu64
"\n",
+ (uint64_t)m->outer_l2_len,
+ (uint64_t)m->outer_l3_len);
if (info.tunnel_tso_segsz != 0 &&
(m->ol_flags &
(RTE_MBUF_F_TX_TCP_SEG |
RTE_MBUF_F_TX_UDP_SEG)))
- printf("tx: m->tso_segsz=%d\n",
- m->tso_segsz);
+ printf("tx: m->tso_segsz=%" PRIu64 "\n",
+ (uint64_t)m->tso_segsz);
} else if (info.tso_segsz != 0 &&
(m->ol_flags & (RTE_MBUF_F_TX_TCP_SEG |
RTE_MBUF_F_TX_UDP_SEG)))
- printf("tx: m->tso_segsz=%d\n", m->tso_segsz);
+ printf("tx: m->tso_segsz=%" PRIu64 "\n",
(uint64_t)m->tso_segsz);
rte_get_tx_ol_flag_list(m->ol_flags, buf, sizeof(buf));
printf("tx: flags=%s", buf);
printf("\n");
--
2.48.1.vfs.0.0
[PATCH v2 06/10] test-pmd: avoid non-constant initializer
Compiling with MSVC results in errors like the one below:
app/test-pmd/cmdline_flow.c(8819): error C2099: initializer
is not a constant
Signed-off-by: Andre Muezerie
Signed-off-by: Chengwen Feng
---
app/test-pmd/cmdline_flow.c | 10 +++---
1 file changed, 3 insertions(+), 7 deletions(-)
diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c
index e1720e54d7..24323d8891 100644
--- a/app/test-pmd/cmdline_flow.c
+++ b/app/test-pmd/cmdline_flow.c
@@ -8812,8 +8812,6 @@ parse_vc_spec(struct context *ctx, const struct token
*token,
return -1;
/* Parse parameter types. */
switch (ctx->curr) {
- static const enum index prefix[] = NEXT_ENTRY(COMMON_PREFIX);
-
case ITEM_PARAM_IS:
index = 0;
objmask = 1;
@@ -8828,7 +8826,7 @@ parse_vc_spec(struct context *ctx, const struct token
*token,
/* Modify next token to expect a prefix. */
if (ctx->next_num < 2)
return -1;
- ctx->next[ctx->next_num - 2] = prefix;
+ ctx->next[ctx->next_num - 2] = NEXT_ENTRY(COMMON_PREFIX);
/* Fall through. */
case ITEM_PARAM_MASK:
index = 2;
@@ -9270,7 +9268,6 @@ parse_vc_action_rss_type(struct context *ctx, const
struct token *token,
const char *str, unsigned int len,
void *buf, unsigned int size)
{
- static const enum index next[] = NEXT_ENTRY(ACTION_RSS_TYPE);
struct action_rss_data *action_rss_data;
unsigned int i;
@@ -9296,7 +9293,7 @@ parse_vc_action_rss_type(struct context *ctx, const
struct token *token,
/* Repeat token. */
if (ctx->next_num == RTE_DIM(ctx->next))
return -1;
- ctx->next[ctx->next_num++] = next;
+ ctx->next[ctx->next_num++] = NEXT_ENTRY(ACTION_RSS_TYPE);
if (!ctx->object)
return len;
action_rss_data = ctx->object;
@@ -9314,7 +9311,6 @@ parse_vc_action_rss_queue(struct context *ctx, const
struct token *token,
const char *str, unsigned int len,
void *buf, unsigned int size)
{
- static const enum index next[] = NEXT_ENTRY(ACTION_RSS_QUEUE);
struct action_rss_data *action_rss_data;
const struct arg *arg;
int ret;
@@ -9347,7 +9343,7 @@ parse_vc_action_rss_queue(struct context *ctx, const
struct token *token,
/* Repeat token. */
if (ctx->next_num == RTE_DIM(ctx->next))
return -1;
- ctx->next[ctx->next_num++] = next;
+ ctx->next[ctx->next_num++] = NEXT_ENTRY(ACTION_RSS_QUEUE);
end:
if (!ctx->object)
return len;
--
2.48.1.vfs.0.0
[PATCH v2 02/10] test_alarm: avoid warning about different qualifiers
Compiling with MSVC results in the warning below: app/test/test_alarm.c(54): warning C4090: 'function': different '_Atomic' qualifiers The fix is to use a macro to explicitly drop the qualifier. Signed-off-by: Andre Muezerie Signed-off-by: Chengwen Feng --- app/test/test_alarm.c | 12 ++-- 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/app/test/test_alarm.c b/app/test/test_alarm.c index 9ed8c6f72c..6445f713fe 100644 --- a/app/test/test_alarm.c +++ b/app/test/test_alarm.c @@ -51,12 +51,12 @@ test_alarm(void) "Expected rte_eal_alarm_cancel to fail with null callback parameter"); /* check if can set a alarm for one second */ - TEST_ASSERT_SUCCESS(rte_eal_alarm_set(US_PER_SEC, test_alarm_callback, &triggered), - "Setting one second alarm failed"); + TEST_ASSERT_SUCCESS(rte_eal_alarm_set(US_PER_SEC, test_alarm_callback, + RTE_PTR_UNQUAL(&triggered)), "Setting one second alarm failed"); /* set a longer alarm that will be canceled. */ - TEST_ASSERT_SUCCESS(rte_eal_alarm_set(10 * US_PER_SEC, test_alarm_callback, &later), - "Setting ten second alarm failed"); + TEST_ASSERT_SUCCESS(rte_eal_alarm_set(10 * US_PER_SEC, test_alarm_callback, + RTE_PTR_UNQUAL(&later)), "Setting ten second alarm failed"); /* wait for alarm to happen */ while (rte_atomic_load_explicit(&triggered, rte_memory_order_acquire) == false) @@ -65,11 +65,11 @@ test_alarm(void) TEST_ASSERT(!rte_atomic_load_explicit(&later, rte_memory_order_acquire), "Only one alarm should have fired."); - ret = rte_eal_alarm_cancel(test_alarm_callback, &triggered); + ret = rte_eal_alarm_cancel(test_alarm_callback, RTE_PTR_UNQUAL(&triggered)); TEST_ASSERT(ret == 0 && rte_errno == ENOENT, "Canceling alarm after run ret %d: %s", ret, rte_strerror(rte_errno)); - ret = rte_eal_alarm_cancel(test_alarm_callback, &later); + ret = rte_eal_alarm_cancel(test_alarm_callback, RTE_PTR_UNQUAL(&later)); TEST_ASSERT(ret == 1, "Canceling ten second alarm failed %d: %s", ret, rte_strerror(rte_errno)); -- 2.48.1.vfs.0.0
[PATCH v2 00/10] enable "app" to be compiled with MSVC
v2: - use lcore_count as atomic always, not only when RTE_USE_C11_MEM_MODEL is set. - use PRIu64 in csumonly.c instead of casting variables to int. This series fixes many issues that prevent the "app" directory from being compiled with MSVC. Andre Muezerie (10): eal: add workaround for __builtin_constant_p test_alarm: avoid warning about different qualifiers test-pmd: fix printf format string mismatch test-pmd: do explicit 64-bit shift to avoid implicit conversion test-pmd: avoid undefined behavior test-pmd: avoid non-constant initializer test-pmd: don't return value from void function test-pmd: declare lcore_count atomic test: add workaround for __builtin_constant_p in test_memcpy_perf app: enable app directory to be compiled with MSVC app/meson.build | 4 -- app/test-pmd/cmdline.c | 19 +++-- app/test-pmd/cmdline_flow.c | 22 +++--- app/test-pmd/csumonly.c | 23 +++--- app/test-pmd/meson.build| 4 +- app/test-pmd/testpmd.c | 2 +- app/test-pmd/util.c | 2 +- app/test/test_alarm.c | 12 ++-- app/test/test_memcpy_perf.c | 106 ++-- app/test/test_ring_perf.c | 6 +- lib/eal/include/generic/rte_pause.h | 2 +- lib/eal/include/rte_common.h| 12 12 files changed, 116 insertions(+), 98 deletions(-) -- 2.48.1.vfs.0.0
[PATCH v2 01/10] eal: add workaround for __builtin_constant_p
There's no MSVC equivalent for compiler extension __builtin_constant_p.
EAL already had __rte_constant which was used as a first attempt to
workaround __builtin_constant_p when using MSVC. However, there are
pieces of code that would benefit from being able to provide a default
value to be used instead of it being always 0 like how it was done by
__rte_constant.
A new macro is added here allowing such default to be provided by the
caller.
Signed-off-by: Andre Muezerie
Signed-off-by: Chengwen Feng
---
lib/eal/include/generic/rte_pause.h | 2 +-
lib/eal/include/rte_common.h| 12
2 files changed, 13 insertions(+), 1 deletion(-)
diff --git a/lib/eal/include/generic/rte_pause.h
b/lib/eal/include/generic/rte_pause.h
index 968c0886d3..57e13807ea 100644
--- a/lib/eal/include/generic/rte_pause.h
+++ b/lib/eal/include/generic/rte_pause.h
@@ -130,7 +130,7 @@ rte_wait_until_equal_64(volatile uint64_t *addr, uint64_t
expected,
* rte_memory_order_acquire and rte_memory_order_relaxed.
*/
#define RTE_WAIT_UNTIL_MASKED(addr, mask, cond, expected, memorder) do { \
- RTE_BUILD_BUG_ON(!__builtin_constant_p(memorder)); \
+ RTE_BUILD_BUG_ON(!__rte_constant_with_default(memorder, 1)); \
RTE_BUILD_BUG_ON((memorder) != rte_memory_order_acquire && \
(memorder) != rte_memory_order_relaxed); \
typeof(*(addr)) expected_value = (expected); \
diff --git a/lib/eal/include/rte_common.h b/lib/eal/include/rte_common.h
index 3f77b7624e..6bdce70551 100644
--- a/lib/eal/include/rte_common.h
+++ b/lib/eal/include/rte_common.h
@@ -50,6 +50,18 @@ extern "C" {
#define __rte_constant(e) __extension__(__builtin_constant_p(e))
#endif
+/**
+ * Determine if an expression's value is constant at compile time.
+ * All compilers except MSVC will return 1 if the first argument is a
+ * compile-time constant, and 0 otherwise. MSVC will just return the second
+ * argument as a default value.
+ */
+#ifdef RTE_TOOLCHAIN_MSVC
+#define __rte_constant_with_default(e, def) def
+#else
+#define __rte_constant_with_default(e, def)
__extension__(__builtin_constant_p(e))
+#endif
+
/*
* RTE_TOOLCHAIN_GCC is defined if the target is built with GCC,
* while a host application (like pmdinfogen) may have another compiler.
--
2.48.1.vfs.0.0
Re: [PATCH v2 03/10] test-pmd: fix printf format string mismatch
On Tue, Feb 18, 2025 at 04:46:56PM +, Bruce Richardson wrote:
> On Tue, Feb 18, 2025 at 08:32:02AM -0800, Andre Muezerie wrote:
> > Compiling with MSVC results in warnings like the one below:
> >
> > app/test-pmd/csumonly.c(1085): warning C4477: 'printf' : format string
> > '%d' requires an argument of type 'int',
> > but variadic argument 1 has type 'uint64_t'
> >
> > Signed-off-by: Andre Muezerie
> > Signed-off-by: Chengwen Feng
> > ---
> > app/test-pmd/csumonly.c | 23 ---
> > 1 file changed, 12 insertions(+), 11 deletions(-)
> >
> > diff --git a/app/test-pmd/csumonly.c b/app/test-pmd/csumonly.c
> > index d77a140641..8de5ad6ad9 100644
> > --- a/app/test-pmd/csumonly.c
> > +++ b/app/test-pmd/csumonly.c
> > @@ -1070,7 +1070,7 @@ pkt_burst_checksum_forward(struct fwd_stream *fs)
> > info.l2_len, rte_be_to_cpu_16(info.ethertype),
> > info.l3_len, info.l4_proto, info.l4_len, buf);
> > if (rx_ol_flags & RTE_MBUF_F_RX_LRO)
> > - printf("rx: m->lro_segsz=%u\n", m->tso_segsz);
> > + printf("rx: m->lro_segsz=%" PRIu64 "\n",
> > (uint64_t)m->tso_segsz);
>
> tso_segsz is already uint64_t, so no need for the cast.
The compilers differ in behavior here. tso_segsz only uses 16 bits of the
uint64_t,
and gcc tries to be smart about it and implicitly converts tso_segsz into an int
(since it fits into an int). Msvc does not do that, and keeps the type for
tso_segsz
as uint64_t. To support both compilers it seems there's no way to avoid the
cast.
>
> > if (info.is_tunnel == 1)
> > printf("rx: outer_l2_len=%d outer_ethertype=%x "
> > "outer_l3_len=%d\n", info.outer_l2_len,
> > @@ -1082,28 +1082,29 @@ pkt_burst_checksum_forward(struct fwd_stream *fs)
> > RTE_ETH_TX_OFFLOAD_TCP_CKSUM |
> > RTE_ETH_TX_OFFLOAD_SCTP_CKSUM)) ||
> > info.tso_segsz != 0)
> > - printf("tx: m->l2_len=%d m->l3_len=%d "
> > - "m->l4_len=%d\n",
> > - m->l2_len, m->l3_len, m->l4_len);
> > + printf("tx: m->l2_len=%" PRIu64 " m->l3_len=%"
> > PRIu64
> > + " m->l4_len=%" PRIu64 "\n",
> > + (uint64_t)m->l2_len,
> > (uint64_t)m->l3_len,
> > + (uint64_t)m->l4_len);
>
> Same here, using casts and a changed print format seems wrong in the patch.
> Either we change the format string to match the variable type, or we cast
> the variables to match the format string. We should not do both, IMHO.
I also normally prefer to not see unnecessary casts in the code, but it looks
like
it is needed in this case, as explained above.
>
> > if (info.is_tunnel == 1) {
> > if ((tx_offloads &
> > RTE_ETH_TX_OFFLOAD_OUTER_IPV4_CKSUM) ||
> > (tx_offloads &
> > RTE_ETH_TX_OFFLOAD_OUTER_UDP_CKSUM) ||
> > (tx_ol_flags & RTE_MBUF_F_TX_OUTER_IPV6))
> > - printf("tx: m->outer_l2_len=%d "
> > - "m->outer_l3_len=%d\n",
> > - m->outer_l2_len,
> > - m->outer_l3_len);
> > + printf("tx: m->outer_l2_len=%" PRIu64
> > + " m->outer_l3_len=%" PRIu64
> > "\n",
> > + (uint64_t)m->outer_l2_len,
> > + (uint64_t)m->outer_l3_len);
> > if (info.tunnel_tso_segsz != 0 &&
> > (m->ol_flags &
> > (RTE_MBUF_F_TX_TCP_SEG |
> > RTE_MBUF_F_TX_UDP_SEG)))
> > - printf("tx: m->tso_segsz=%d\n",
> > - m->tso_segsz);
> > + printf("tx: m->tso_segsz=%" PRIu64 "\n",
> > + (uint64_t)m->tso_segsz);
> > } else if (info.tso_segsz != 0 &&
> > (m->ol_flags & (RTE_MBUF_F_TX_TCP_SEG |
> > RTE_MBUF_F_TX_UDP_SEG)))
> > - printf("tx: m->tso_segsz=%d\n", m->tso_segsz);
> > + printf("tx: m->tso_segsz=%" PRIu64 "\n",
> > (uint64_t)m->tso_segsz);
> > rte_get_tx_ol_flag_list(m->ol_flags, buf, sizeof(buf));
> >
RE: [PATCH v2 01/10] eal: add workaround for __builtin_constant_p
> From: Andre Muezerie [mailto:andre...@linux.microsoft.com] > Sent: Tuesday, 18 February 2025 17.32 > > There's no MSVC equivalent for compiler extension __builtin_constant_p. > EAL already had __rte_constant which was used as a first attempt to > workaround __builtin_constant_p when using MSVC. However, there are > pieces of code that would benefit from being able to provide a default > value to be used instead of it being always 0 like how it was done by > __rte_constant. > > A new macro is added here allowing such default to be provided by the > caller. NAK to the new macro. Please use the sizeof() hack suggested by Stephen to provide an implementation of __rte_constant(e) for MSVC. Here's a link to an example of the hack: https://news.ycombinator.com/item?id=16720172
Re: [PATCH v2 03/10] test-pmd: fix printf format string mismatch
On Tue, Feb 18, 2025 at 09:03:01AM -0800, Andre Muezerie wrote:
> On Tue, Feb 18, 2025 at 04:46:56PM +, Bruce Richardson wrote:
> > On Tue, Feb 18, 2025 at 08:32:02AM -0800, Andre Muezerie wrote:
> > > Compiling with MSVC results in warnings like the one below:
> > >
> > > app/test-pmd/csumonly.c(1085): warning C4477: 'printf' : format string
> > > '%d' requires an argument of type 'int',
> > > but variadic argument 1 has type 'uint64_t'
> > >
> > > Signed-off-by: Andre Muezerie
> > > Signed-off-by: Chengwen Feng
> > > ---
> > > app/test-pmd/csumonly.c | 23 ---
> > > 1 file changed, 12 insertions(+), 11 deletions(-)
> > >
> > > diff --git a/app/test-pmd/csumonly.c b/app/test-pmd/csumonly.c
> > > index d77a140641..8de5ad6ad9 100644
> > > --- a/app/test-pmd/csumonly.c
> > > +++ b/app/test-pmd/csumonly.c
> > > @@ -1070,7 +1070,7 @@ pkt_burst_checksum_forward(struct fwd_stream *fs)
> > > info.l2_len, rte_be_to_cpu_16(info.ethertype),
> > > info.l3_len, info.l4_proto, info.l4_len, buf);
> > > if (rx_ol_flags & RTE_MBUF_F_RX_LRO)
> > > - printf("rx: m->lro_segsz=%u\n", m->tso_segsz);
> > > + printf("rx: m->lro_segsz=%" PRIu64 "\n",
> > > (uint64_t)m->tso_segsz);
> >
> > tso_segsz is already uint64_t, so no need for the cast.
>
> The compilers differ in behavior here. tso_segsz only uses 16 bits of the
> uint64_t,
> and gcc tries to be smart about it and implicitly converts tso_segsz into an
> int
> (since it fits into an int). Msvc does not do that, and keeps the type for
> tso_segsz
> as uint64_t. To support both compilers it seems there's no way to avoid the
> cast.
>
Ok. Then can we just keep the %u and cast to either unsigned or uint16_t?
No need to update the format char if we are casting, right?
/Bruce
[PATCH v2 09/10] test: add workaround for __builtin_constant_p in test_memcpy_perf
There's no MSVC equivalent for compiler extension __builtin_constant_p,
so a workaround is needed.
Signed-off-by: Andre Muezerie
---
app/test/test_memcpy_perf.c | 106 ++--
1 file changed, 53 insertions(+), 53 deletions(-)
diff --git a/app/test/test_memcpy_perf.c b/app/test/test_memcpy_perf.c
index 5c05a84619..6091b6f9dd 100644
--- a/app/test/test_memcpy_perf.c
+++ b/app/test/test_memcpy_perf.c
@@ -167,66 +167,66 @@ do_uncached_write(uint8_t *dst, int is_dst_cached,
* Run a single memcpy performance test. This is a macro to ensure that if
* the "size" parameter is a constant it won't be converted to a variable.
*/
-#define SINGLE_PERF_TEST(dst, is_dst_cached, dst_uoffset, \
- src, is_src_cached, src_uoffset, size) \
-do {\
-unsigned int iter, t; \
-size_t dst_addrs[TEST_BATCH_SIZE], src_addrs[TEST_BATCH_SIZE]; \
-uint64_t start_time, total_time = 0;\
-uint64_t total_time2 = 0; \
-for (iter = 0; iter < (TEST_ITERATIONS / TEST_BATCH_SIZE); iter++) {\
-fill_addr_arrays(dst_addrs, is_dst_cached, dst_uoffset, \
- src_addrs, is_src_cached, src_uoffset);\
-start_time = rte_rdtsc(); \
-for (t = 0; t < TEST_BATCH_SIZE; t++) \
-rte_memcpy(dst+dst_addrs[t], src+src_addrs[t], size); \
-total_time += rte_rdtsc() - start_time; \
-} \
-for (iter = 0; iter < (TEST_ITERATIONS / TEST_BATCH_SIZE); iter++) {\
-fill_addr_arrays(dst_addrs, is_dst_cached, dst_uoffset, \
- src_addrs, is_src_cached, src_uoffset);\
-start_time = rte_rdtsc(); \
-for (t = 0; t < TEST_BATCH_SIZE; t++) \
-memcpy(dst+dst_addrs[t], src+src_addrs[t], size); \
-total_time2 += rte_rdtsc() - start_time;\
-} \
-printf("%3.0f -", (double)total_time / TEST_ITERATIONS); \
-printf("%3.0f", (double)total_time2 / TEST_ITERATIONS); \
-printf("(%6.2f%%) ", ((double)total_time - total_time2)*100/total_time2); \
+#define SINGLE_PERF_TEST(dst, is_dst_cached, dst_uoffset,
\
+src, is_src_cached, src_uoffset, size)
\
+do {
\
+ unsigned int iter, t;
\
+ size_t dst_addrs[TEST_BATCH_SIZE], src_addrs[TEST_BATCH_SIZE];
\
+ uint64_t start_time, total_time = 0;
\
+ uint64_t total_time2 = 0;
\
+ for (iter = 0; iter < (TEST_ITERATIONS / TEST_BATCH_SIZE); iter++) {
\
+ fill_addr_arrays(dst_addrs, is_dst_cached, dst_uoffset,
\
+src_addrs, is_src_cached, src_uoffset);
\
+ start_time = rte_rdtsc();
\
+ for (t = 0; t < TEST_BATCH_SIZE; t++)
\
+ rte_memcpy(dst+dst_addrs[t], src+src_addrs[t], size);
\
+ total_time += rte_rdtsc() - start_time;
\
+ }
\
+ for (iter = 0; iter < (TEST_ITERATIONS / TEST_BATCH_SIZE); iter++) {
\
+ fill_addr_arrays(dst_addrs, is_dst_cached, dst_uoffset,
\
+src_addrs, is_src_cached, src_uoffset);
\
+ start_time = rte_rdtsc();
\
+ for (t = 0; t < TEST_BATCH_SIZE; t++)
\
+ memcpy(dst+dst_addrs[t], src+src_addrs[t], size);
\
+ total_time2 += rte_rdtsc() - start_time;
\
+ }
\
+ printf("%3.0f -", (double)total_time / TEST_ITERATIONS);
\
+ printf("%3.0f", (double)total_time2 / TEST_ITERATIONS);
\
+ printf("(%6.2f%%) ", ((double)total_time -
total_time2)*100/total_time2); \
} while (0)
/* Run aligned memcpy tests for each cached/unc
[PATCH v2 05/10] test-pmd: avoid undefined behavior
Compiling with MSVC results in warnings like below:
app/test-pmd/cmdline.c(9023): warning C5101: use of preprocessor
directive in function-like macro argument list is undefined behavior
Signed-off-by: Andre Muezerie
Signed-off-by: Chengwen Feng
---
app/test-pmd/cmdline.c | 15 +--
1 file changed, 13 insertions(+), 2 deletions(-)
diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
index 2afcf916c0..4f0b0340c8 100644
--- a/app/test-pmd/cmdline.c
+++ b/app/test-pmd/cmdline.c
@@ -9011,6 +9011,18 @@ static void cmd_dump_parsed(void *parsed_result,
}
static cmdline_parse_token_string_t cmd_dump_dump =
+#ifdef RTE_EXEC_ENV_WINDOWS
+ TOKEN_STRING_INITIALIZER(struct cmd_dump_result, dump,
+ "dump_physmem#"
+ "dump_memzone#"
+ "dump_socket_mem#"
+ "dump_struct_sizes#"
+ "dump_ring#"
+ "dump_mempool#"
+ "dump_devargs#"
+ "dump_lcores#"
+ "dump_log_types");
+#else
TOKEN_STRING_INITIALIZER(struct cmd_dump_result, dump,
"dump_physmem#"
"dump_memzone#"
@@ -9020,10 +9032,9 @@ static cmdline_parse_token_string_t cmd_dump_dump =
"dump_mempool#"
"dump_devargs#"
"dump_lcores#"
-#ifndef RTE_EXEC_ENV_WINDOWS
"dump_trace#"
-#endif
"dump_log_types");
+#endif
static cmdline_parse_inst_t cmd_dump = {
.f = cmd_dump_parsed, /* function to call */
--
2.48.1.vfs.0.0
[PATCH v2 10/10] app: enable app directory to be compiled with MSVC
Enabled "app" directory to be compiled with MSVC along with all its
contents.
Signed-off-by: Andre Muezerie
---
app/meson.build | 4
app/test-pmd/meson.build | 4 +++-
2 files changed, 3 insertions(+), 5 deletions(-)
diff --git a/app/meson.build b/app/meson.build
index e2db888ae1..1798db3ae4 100644
--- a/app/meson.build
+++ b/app/meson.build
@@ -1,10 +1,6 @@
# SPDX-License-Identifier: BSD-3-Clause
# Copyright(c) 2017-2019 Intel Corporation
-if is_ms_compiler
-subdir_done()
-endif
-
disable_apps = ',' + get_option('disable_apps')
disable_apps = run_command(list_dir_globs, disable_apps, check:
true).stdout().split()
diff --git a/app/test-pmd/meson.build b/app/test-pmd/meson.build
index f1c36529b4..8e8863812a 100644
--- a/app/test-pmd/meson.build
+++ b/app/test-pmd/meson.build
@@ -3,7 +3,9 @@
# override default name to drop the hyphen
name = 'testpmd'
-cflags += '-Wno-deprecated-declarations'
+if cc.has_argument('-Wno-deprecated-declarations')
+cflags += '-Wno-deprecated-declarations'
+endif
sources = files(
'5tswap.c',
'cmdline.c',
--
2.48.1.vfs.0.0
[PATCH v2 07/10] test-pmd: don't return value from void function
Compiling with MSVC results in the warning below:
app/test-pmd/cmdline_flow.c(13964): warning C4098: 'cmd_set_raw_parsed':
'void' function returning a value
Signed-off-by: Andre Muezerie
Signed-off-by: Chengwen Feng
---
app/test-pmd/cmdline_flow.c | 12
1 file changed, 8 insertions(+), 4 deletions(-)
diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c
index 24323d8891..15a18db2c7 100644
--- a/app/test-pmd/cmdline_flow.c
+++ b/app/test-pmd/cmdline_flow.c
@@ -13952,11 +13952,15 @@ cmd_set_raw_parsed(const struct buffer *in)
int gtp_psc = -1; /* GTP PSC option index. */
const void *src_spec;
- if (in->command == SET_SAMPLE_ACTIONS)
- return cmd_set_raw_parsed_sample(in);
+ if (in->command == SET_SAMPLE_ACTIONS) {
+ cmd_set_raw_parsed_sample(in);
+ return;
+ }
else if (in->command == SET_IPV6_EXT_PUSH ||
-in->command == SET_IPV6_EXT_REMOVE)
- return cmd_set_ipv6_ext_parsed(in);
+in->command == SET_IPV6_EXT_REMOVE) {
+ cmd_set_ipv6_ext_parsed(in);
+ return;
+ }
RTE_ASSERT(in->command == SET_RAW_ENCAP ||
in->command == SET_RAW_DECAP);
if (in->command == SET_RAW_ENCAP) {
--
2.48.1.vfs.0.0
[PATCH v2 08/10] test-pmd: declare lcore_count atomic
Compiling with MSVC results in the error below:
app/test/test_ring_perf.c(197): error C7712: address argument to atomic
operation must be a pointer to an atomic integer,
'volatile unsigned int *' is not valid
The fix is to mark lcore_count as atomic.
Signed-off-by: Andre Muezerie
Signed-off-by: Chengwen Feng
---
app/test/test_ring_perf.c | 6 +-
1 file changed, 1 insertion(+), 5 deletions(-)
diff --git a/app/test/test_ring_perf.c b/app/test/test_ring_perf.c
index 57cd04a124..366e256323 100644
--- a/app/test/test_ring_perf.c
+++ b/app/test/test_ring_perf.c
@@ -34,7 +34,7 @@ struct lcore_pair {
unsigned c1, c2;
};
-static volatile unsigned lcore_count = 0;
+static RTE_ATOMIC(unsigned int) lcore_count;
static void
test_ring_print_test_string(unsigned int api_type, int esize,
@@ -193,11 +193,7 @@ enqueue_dequeue_bulk_helper(const unsigned int flag,
struct thread_params *p)
unsigned int n_remaining;
const unsigned int bulk_n = bulk_sizes[p->ring_params->bulk_sizes_i];
-#ifdef RTE_USE_C11_MEM_MODEL
if (rte_atomic_fetch_add_explicit(&lcore_count, 1,
rte_memory_order_relaxed) + 1 != 2)
-#else
- if (__sync_add_and_fetch(&lcore_count, 1) != 2)
-#endif
while(lcore_count != 2)
rte_pause();
--
2.48.1.vfs.0.0
[PATCH v2 04/10] test-pmd: do explicit 64-bit shift to avoid implicit conversion
Compiling with MSVC results in warnings like the one below:
app/test-pmd/util.c(201): warning C4334: '<<': result of 32-bit shift
implicitly converted to 64 bits (was 64-bit shift intended?)
Signed-off-by: Andre Muezerie
Signed-off-by: Chengwen Feng
---
app/test-pmd/cmdline.c | 4 ++--
app/test-pmd/testpmd.c | 2 +-
app/test-pmd/util.c| 2 +-
3 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
index 86d763b66a..2afcf916c0 100644
--- a/app/test-pmd/cmdline.c
+++ b/app/test-pmd/cmdline.c
@@ -12632,11 +12632,11 @@ cmd_config_dynf_specific_parsed(void *parsed_result,
}
old_port_flags = ports[res->port_id].mbuf_dynf;
if (!strcmp(res->value, "set")) {
- ports[res->port_id].mbuf_dynf |= 1UL << flag;
+ ports[res->port_id].mbuf_dynf |= RTE_BIT64(flag);
if (old_port_flags == 0)
add_tx_dynf_callback(res->port_id);
} else {
- ports[res->port_id].mbuf_dynf &= ~(1UL << flag);
+ ports[res->port_id].mbuf_dynf &= ~RTE_BIT64(flag);
if (ports[res->port_id].mbuf_dynf == 0)
remove_tx_dynf_callback(res->port_id);
}
diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
index 0520aba0db..0a5b999c3a 100644
--- a/app/test-pmd/testpmd.c
+++ b/app/test-pmd/testpmd.c
@@ -4152,7 +4152,7 @@ get_eth_dcb_conf(struct rte_eth_conf *eth_conf, enum
dcb_mode_enable dcb_mode,
for (i = 0; i < vmdq_rx_conf->nb_pool_maps; i++) {
vmdq_rx_conf->pool_map[i].vlan_id = vlan_tags[i];
vmdq_rx_conf->pool_map[i].pools =
- 1 << (i % vmdq_rx_conf->nb_queue_pools);
+ RTE_BIT64(i % vmdq_rx_conf->nb_queue_pools);
}
for (i = 0; i < RTE_ETH_DCB_NUM_USER_PRIORITIES; i++) {
vmdq_rx_conf->dcb_tc[i] = i % num_tcs;
diff --git a/app/test-pmd/util.c b/app/test-pmd/util.c
index 5fa05fad16..3934831b19 100644
--- a/app/test-pmd/util.c
+++ b/app/test-pmd/util.c
@@ -201,7 +201,7 @@ dump_pkt_burst(uint16_t port_id, uint16_t queue, struct
rte_mbuf *pkts[],
MKDUMPSTR(print_buf, buf_size, cur_len,
" - dynf %s: %d",
dynf_names[dynf_index],
- !!(ol_flags & (1UL << dynf_index)));
+ !!(ol_flags & RTE_BIT64(dynf_index)));
}
if (mb->packet_type) {
rte_get_ptype_name(mb->packet_type, buf, sizeof(buf));
--
2.48.1.vfs.0.0
Re: [PATCH] doc/dmadevs: clarify naming constraints for idxd devices
On 31/01/2025 14:49, Bruce Richardson wrote: The requirement to have DPDK in the DSA wq name is given in the idxd chapter section 5.3.3 [1]. However, that information should also be given in section 5.3.1 [2] where we list out the fields to be set and what values they should have. [1]https://doc.dpdk.org/guides-24.11/dmadevs/idxd.html#device-probing-and-initialization [2]https://doc.dpdk.org/guides-24.11/dmadevs/idxd.html#intelreg-dsa-devices-using-idxd-kernel-driver Signed-off-by: Bruce Richardson --- doc/guides/dmadevs/idxd.rst | 4 1 file changed, 4 insertions(+) Acked-by: Kevin Laatz
[PATCH v3 03/10] drivers: remove weak symbols in Nitrox drivers
Make compress and crypto drivers register to the common driver.
Remove (unneeded) include_directories().
Signed-off-by: David Marchand
---
drivers/common/nitrox/meson.build | 3 --
drivers/common/nitrox/nitrox_device.c | 75 ++-
drivers/common/nitrox/nitrox_device.h | 16 ++
drivers/common/nitrox/version.map | 1 +
drivers/compress/nitrox/meson.build | 2 -
drivers/compress/nitrox/nitrox_comp.c | 6 +++
drivers/crypto/nitrox/meson.build | 2 -
drivers/crypto/nitrox/nitrox_sym.c| 6 +++
8 files changed, 57 insertions(+), 54 deletions(-)
diff --git a/drivers/common/nitrox/meson.build
b/drivers/common/nitrox/meson.build
index f3cb42f006..115dd8de4d 100644
--- a/drivers/common/nitrox/meson.build
+++ b/drivers/common/nitrox/meson.build
@@ -14,6 +14,3 @@ sources += files(
'nitrox_logs.c',
'nitrox_qp.c',
)
-
-includes += include_directories('../../crypto/nitrox')
-includes += include_directories('../../compress/nitrox')
diff --git a/drivers/common/nitrox/nitrox_device.c
b/drivers/common/nitrox/nitrox_device.c
index 39edc440a7..6cd57faaa4 100644
--- a/drivers/common/nitrox/nitrox_device.c
+++ b/drivers/common/nitrox/nitrox_device.c
@@ -6,8 +6,6 @@
#include "nitrox_device.h"
#include "nitrox_hal.h"
-#include "nitrox_sym.h"
-#include "nitrox_comp.h"
#define PCI_VENDOR_ID_CAVIUM 0x177d
#define NITROX_V_PCI_VF_DEV_ID 0x13
@@ -63,11 +61,21 @@ ndev_release(struct nitrox_device *ndev)
rte_free(ndev);
}
+TAILQ_HEAD(ndrv_list, nitrox_driver);
+static struct ndrv_list ndrv_list = TAILQ_HEAD_INITIALIZER(ndrv_list);
+
+void
+nitrox_register_driver(struct nitrox_driver *ndrv)
+{
+ TAILQ_INSERT_TAIL(&ndrv_list, ndrv, next);
+}
+
static int
nitrox_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
struct rte_pci_device *pdev)
{
struct nitrox_device *ndev;
+ struct nitrox_driver *ndrv;
int err = -1;
/* Nitrox CSR space */
@@ -79,19 +87,21 @@ nitrox_pci_probe(struct rte_pci_driver *pci_drv
__rte_unused,
return -ENOMEM;
ndev_init(ndev, pdev);
- err = nitrox_sym_pmd_create(ndev);
- if (err)
- goto sym_pmd_err;
-
- err = nitrox_comp_pmd_create(ndev);
- if (err)
- goto comp_pmd_err;
+ TAILQ_FOREACH(ndrv, &ndrv_list, next) {
+ err = ndrv->create(ndev);
+ if (err)
+ goto drv_err;
+ }
return 0;
-comp_pmd_err:
- nitrox_sym_pmd_destroy(ndev);
-sym_pmd_err:
+drv_err:
+ ndrv = TAILQ_PREV(ndrv, ndrv_list, next);
+ while (ndrv != NULL) {
+ ndrv->destroy(ndev);
+ ndrv = TAILQ_PREV(ndrv, ndrv_list, next);
+ }
+
ndev_release(ndev);
return err;
}
@@ -100,19 +110,18 @@ static int
nitrox_pci_remove(struct rte_pci_device *pdev)
{
struct nitrox_device *ndev;
+ struct nitrox_driver *ndrv;
int err;
ndev = find_ndev(pdev);
if (!ndev)
return -ENODEV;
- err = nitrox_sym_pmd_destroy(ndev);
- if (err)
- return err;
-
- err = nitrox_comp_pmd_destroy(ndev);
- if (err)
- return err;
+ TAILQ_FOREACH(ndrv, &ndrv_list, next) {
+ err = ndrv->destroy(ndev);
+ if (err)
+ return err;
+ }
ndev_release(ndev);
return 0;
@@ -133,33 +142,5 @@ static struct rte_pci_driver nitrox_pmd = {
.remove = nitrox_pci_remove,
};
-__rte_weak int
-nitrox_sym_pmd_create(struct nitrox_device *ndev)
-{
- RTE_SET_USED(ndev);
- return 0;
-}
-
-__rte_weak int
-nitrox_sym_pmd_destroy(struct nitrox_device *ndev)
-{
- RTE_SET_USED(ndev);
- return 0;
-}
-
-__rte_weak int
-nitrox_comp_pmd_create(struct nitrox_device *ndev)
-{
- RTE_SET_USED(ndev);
- return 0;
-}
-
-__rte_weak int
-nitrox_comp_pmd_destroy(struct nitrox_device *ndev)
-{
- RTE_SET_USED(ndev);
- return 0;
-}
-
RTE_PMD_REGISTER_PCI(nitrox, nitrox_pmd);
RTE_PMD_REGISTER_PCI_TABLE(nitrox, pci_id_nitrox_map);
diff --git a/drivers/common/nitrox/nitrox_device.h
b/drivers/common/nitrox/nitrox_device.h
index 877bccb321..b74b71808e 100644
--- a/drivers/common/nitrox/nitrox_device.h
+++ b/drivers/common/nitrox/nitrox_device.h
@@ -6,6 +6,7 @@
#define _NITROX_DEVICE_H_
#include
+#include
struct nitrox_sym_device;
struct nitrox_comp_device;
@@ -21,4 +22,19 @@ struct nitrox_device {
uint16_t nr_queues;
};
+struct nitrox_driver {
+ TAILQ_ENTRY(nitrox_driver) next;
+ int (*create)(struct nitrox_device *ndev);
+ int (*destroy)(struct nitrox_device *ndev);
+};
+
+__rte_internal
+void nitrox_register_driver(struct nitrox_driver *ndrv);
+
+#define NITROX_REGISTER_DRIVER(ndrv) \
+RTE_INIT(ndrv ## _register) \
+{ \
+ nitrox_register_driver(&ndrv); \
+}
+
#endif /* _NITROX_DEVICE_H
Re: [PATCH v2 1/2] net/intel: add E830 ETF offload timestamp resolution
On Wed, Feb 12, 2025 at 09:47:10PM +, Soumyadeep Hore wrote:
> From: Paul Greenwalt
>
> Update E830 ETF offload time stamp resolution to 128ns along
I'm not sure the word "update" is applicable here, since I just see a
single new define being added. The timestamp resolution wasn't previously
being set in the code, was it, or am I missing something?
> with certain HW defines for TXTIME_PROFILE register.
This commit title and log message should probably just be "add defines for
Tx time stamping" or something similar to that.
>
> Signed-off-by: Soumyadeep Hore
> Signed-off-by: Paul Greenwalt
> ---
> drivers/net/intel/ice/base/ice_lan_tx_rx.h | 4
> 1 file changed, 4 insertions(+)
>
> diff --git a/drivers/net/intel/ice/base/ice_lan_tx_rx.h
> b/drivers/net/intel/ice/base/ice_lan_tx_rx.h
> index f92382346f..940c6843d9 100644
> --- a/drivers/net/intel/ice/base/ice_lan_tx_rx.h
> +++ b/drivers/net/intel/ice/base/ice_lan_tx_rx.h
> @@ -1278,6 +1278,8 @@ struct ice_ts_desc {
> #define ICE_TXTIME_MAX_QUEUE 2047
> #define ICE_SET_TXTIME_MAX_Q_AMOUNT 127
> #define ICE_OP_TXTIME_MAX_Q_AMOUNT 2047
> +#define ICE_TXTIME_FETCH_TS_DESC_DFLT8
> +
> /* Tx Time queue context data
> *
> * The sizes of the variables may be larger than needed due to crossing byte
> @@ -1303,6 +1305,8 @@ struct ice_txtime_ctx {
> u8 drbell_mode_32;
> #define ICE_TXTIME_CTX_DRBELL_MODE_321
> u8 ts_res;
> +#define ICE_TXTIME_CTX_FETCH_PROF_ID_0 0
> +#define ICE_TXTIME_CTX_RESOLUTION_128NS 7
> u8 ts_round_type;
> u8 ts_pacing_slot;
> u8 merging_ena;
> --
> 2.43.0
>
Re: [PATCH v2 05/10] test-pmd: avoid undefined behavior
On Tue, Feb 18, 2025 at 08:32:04AM -0800, Andre Muezerie wrote: > Compiling with MSVC results in warnings like below: > > app/test-pmd/cmdline.c(9023): warning C5101: use of preprocessor > directive in function-like macro argument list is undefined behavior > > Signed-off-by: Andre Muezerie > Signed-off-by: Chengwen Feng > --- > app/test-pmd/cmdline.c | 15 +-- > 1 file changed, 13 insertions(+), 2 deletions(-) > > diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c > index 2afcf916c0..4f0b0340c8 100644 > --- a/app/test-pmd/cmdline.c > +++ b/app/test-pmd/cmdline.c > @@ -9011,6 +9011,18 @@ static void cmd_dump_parsed(void *parsed_result, > } > > static cmdline_parse_token_string_t cmd_dump_dump = > +#ifdef RTE_EXEC_ENV_WINDOWS > + TOKEN_STRING_INITIALIZER(struct cmd_dump_result, dump, > + "dump_physmem#" > + "dump_memzone#" > + "dump_socket_mem#" > + "dump_struct_sizes#" > + "dump_ring#" > + "dump_mempool#" > + "dump_devargs#" > + "dump_lcores#" > + "dump_log_types"); > +#else > TOKEN_STRING_INITIALIZER(struct cmd_dump_result, dump, > "dump_physmem#" > "dump_memzone#" > @@ -9020,10 +9032,9 @@ static cmdline_parse_token_string_t cmd_dump_dump = > "dump_mempool#" > "dump_devargs#" > "dump_lcores#" > -#ifndef RTE_EXEC_ENV_WINDOWS > "dump_trace#" > -#endif > "dump_log_types"); > +#endif > Rather than defining two separate lists, is a better fix not to provide a dummy implementation of dump_trace for windows, that returns e.g. TEST_SKIPPED. /Bruce
[PATCH] config/arm: update NVIDIA BlueField-3 configuration
ARM configuration requires explicit `mcpu` specifications in the
implementor description.
The patch adds `mcpu` and `flags` description for the NVIDIA
BlueField-3 configuration.
Signed-off-by: Gregory Etelson
---
config/arm/meson.build | 5 +
1 file changed, 5 insertions(+)
diff --git a/config/arm/meson.build b/config/arm/meson.build
index 5302861e2c..74f4dd4e1f 100644
--- a/config/arm/meson.build
+++ b/config/arm/meson.build
@@ -85,6 +85,11 @@ part_number_config_arm = {
},
'0xd42': {
'march': 'armv8.4-a',
+'mcpu' : 'cortex-a78ae',
+'flags': [
+['RTE_MAX_LCORE', 16],
+['RTE_MAX_NUMA_NODES', 1]
+],
},
'0xd49': {
'mcpu': 'neoverse-n2',
--
2.45.2
Re: [PATCH] ci: build with MSVC in GHA
David Marchand writes:
> Add a build target in Windows 2022 for MSVC.
>
> Loading MSVC environment and having successive commands in a single
> (cmd or pwsh) shell does not seem to correctly report if one of the
> commands failed.
> For now, load MSVC environment and run commands in separate shells.
>
> Signed-off-by: David Marchand
> ---
LGTM.
Acked-by: Aaron Conole
How far back will this work (ie: can we backport this to older branches?)
> .github/workflows/build.yml | 38 +
> 1 file changed, 38 insertions(+)
>
> diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
> index dcafb4a8f5..aef692ca1f 100644
> --- a/.github/workflows/build.yml
> +++ b/.github/workflows/build.yml
> @@ -174,6 +174,44 @@ jobs:
>build/meson-logs/meson-log.txt
>build/meson-logs/testlog.txt
>
> + windows-vm-builds:
> +name: ${{ join(matrix.config.*, '-') }}
> +runs-on: ${{ matrix.config.os }}
> +strategy:
> + fail-fast: false
> + matrix:
> +config:
> + - os: windows-2022
> +compiler: msvc
> +
> +steps:
> +- name: Checkout sources
> + uses: actions/checkout@v4
> +- name: Set up Python
> + uses: actions/setup-python@v5
> + with:
> +python-version: '3.x'
> +- name: Install dependencies
> + run: python -m pip install meson==1.6.1 ninja
> +- name: Configure
> + shell: cmd
> + run: |
> +call "C:\Program Files\Microsoft Visual
> Studio\2022\Enterprise\Common7\Tools\VsDevCmd.bat" -host_arch=amd64
> -arch=amd64
> +meson setup -Denable_stdatomic=true build
> +- name: Build
> + shell: cmd
> + run: |
> +call "C:\Program Files\Microsoft Visual
> Studio\2022\Enterprise\Common7\Tools\VsDevCmd.bat" -host_arch=amd64
> -arch=amd64
> +meson compile -C build
> +- name: Upload logs on failure
> + if: failure()
> + uses: actions/upload-artifact@v4
> + with:
> +name: meson-logs-${{ join(matrix.config.*, '-') }}
> +path: |
> + build/.ninja_log
> + build/meson-logs/meson-log.txt
> +
>prepare-container-images:
> name: ${{ join(matrix.config.*, '-') }}
> runs-on: ubuntu-latest
[PATCH] ethdev: fix get_reg_info
'width' and 'offset' are input parameters when dumping the register
info of an Ethernet device. They should be copied in the new request
before calling the device callback function.
Fixes: 083db2ed9e9 ('ethdev: add report of register names and filter')
Cc: sta...@dpdk.org
Signed-off-by: Thierry Herbelot
---
lib/ethdev/rte_ethdev.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/lib/ethdev/rte_ethdev.c b/lib/ethdev/rte_ethdev.c
index 6413c54e3b39..073a3bcf5c0b 100644
--- a/lib/ethdev/rte_ethdev.c
+++ b/lib/ethdev/rte_ethdev.c
@@ -6511,6 +6511,8 @@ rte_eth_dev_get_reg_info(uint16_t port_id, struct
rte_dev_reg_info *info)
}
reg_info.length = info->length;
+ reg_info.width = info->width;
+ reg_info.offset = info->offset;
reg_info.data = info->data;
ret = rte_eth_dev_get_reg_info_ext(port_id, ®_info);
--
2.39.2
[Patch v2] net/mana: use mana_local_data for tracking usage data for primary process
From: Long Li
The driver uses mana_shared_data for tracking usage count for primary
process. This is not correct as the mana_shared_data is allocated
by the primary and is meant to track usage of secondary process by the
primary process. And it creates a race condition when the device is
removed because the counter is no longer available if this shared
memory is freed.
Move the usage count tracking to mana_local_data and fix the race
condition in mana_pci_remove().
Fixes: 517ed6e2d590 ("net/mana: add basic driver with build environment")
Signed-off-by: Long Li
---
Changes:
v2: use atomic variable to track the secondary_cnt in shared memory, remove the
spinlock for shared memory
drivers/net/mana/mana.c | 99 +++--
drivers/net/mana/mana.h | 6 +--
drivers/net/mana/mp.c | 2 +-
3 files changed, 57 insertions(+), 50 deletions(-)
diff --git a/drivers/net/mana/mana.c b/drivers/net/mana/mana.c
index c37c4e3444..69cb7ab13f 100644
--- a/drivers/net/mana/mana.c
+++ b/drivers/net/mana/mana.c
@@ -23,9 +23,14 @@
#include "mana.h"
/* Shared memory between primary/secondary processes, per driver */
-/* Data to track primary/secondary usage */
struct mana_shared_data *mana_shared_data;
-static struct mana_shared_data mana_local_data;
+
+/* Local data to track device instance usage for primary/secondary processes */
+static struct mana_local_data {
+ int init_done;
+ unsigned int primary_cnt;
+ unsigned int secondary_cnt;
+} mana_local_data;
/* The memory region for the above data */
static const struct rte_memzone *mana_shared_mz;
@@ -1167,8 +1172,12 @@ mana_init_shared_data(void)
rte_spinlock_lock(&mana_shared_data_lock);
/* Skip if shared data is already initialized */
- if (mana_shared_data)
+ if (mana_shared_data) {
+ DRV_LOG(INFO, "shared data is already initialized");
goto exit;
+ }
+
+ memset(&mana_local_data, 0, sizeof(mana_local_data));
if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
mana_shared_mz = rte_memzone_reserve(MZ_MANA_SHARED_DATA,
@@ -1181,8 +1190,7 @@ mana_init_shared_data(void)
}
mana_shared_data = mana_shared_mz->addr;
- memset(mana_shared_data, 0, sizeof(*mana_shared_data));
- rte_spinlock_init(&mana_shared_data->lock);
+ rte_atomic32_set(&mana_shared_data->secondary_cnt, 0);
} else {
secondary_mz = rte_memzone_lookup(MZ_MANA_SHARED_DATA);
if (!secondary_mz) {
@@ -1192,7 +1200,6 @@ mana_init_shared_data(void)
}
mana_shared_data = secondary_mz->addr;
- memset(&mana_local_data, 0, sizeof(mana_local_data));
}
exit:
@@ -1213,11 +1220,11 @@ mana_init_once(void)
if (ret)
return ret;
- rte_spinlock_lock(&mana_shared_data->lock);
+ rte_spinlock_lock(&mana_shared_data_lock);
switch (rte_eal_process_type()) {
case RTE_PROC_PRIMARY:
- if (mana_shared_data->init_done)
+ if (mana_local_data.init_done)
break;
ret = mana_mp_init_primary();
@@ -1225,7 +1232,7 @@ mana_init_once(void)
break;
DRV_LOG(ERR, "MP INIT PRIMARY");
- mana_shared_data->init_done = 1;
+ mana_local_data.init_done = 1;
break;
case RTE_PROC_SECONDARY:
@@ -1248,7 +1255,7 @@ mana_init_once(void)
break;
}
- rte_spinlock_unlock(&mana_shared_data->lock);
+ rte_spinlock_unlock(&mana_shared_data_lock);
return ret;
}
@@ -1319,11 +1326,6 @@ mana_probe_port(struct ibv_device *ibdev, struct
ibv_device_attr_ex *dev_attr,
eth_dev->tx_pkt_burst = mana_tx_burst;
eth_dev->rx_pkt_burst = mana_rx_burst;
- rte_spinlock_lock(&mana_shared_data->lock);
- mana_shared_data->secondary_cnt++;
- mana_local_data.secondary_cnt++;
- rte_spinlock_unlock(&mana_shared_data->lock);
-
rte_eth_copy_pci_info(eth_dev, pci_dev);
rte_eth_dev_probing_finish(eth_dev);
@@ -1406,10 +1408,6 @@ mana_probe_port(struct ibv_device *ibdev, struct
ibv_device_attr_ex *dev_attr,
goto failed;
}
- rte_spinlock_lock(&mana_shared_data->lock);
- mana_shared_data->primary_cnt++;
- rte_spinlock_unlock(&mana_shared_data->lock);
-
eth_dev->device = &pci_dev->device;
DRV_LOG(INFO, "device %s at port %u", name, eth_dev->data->port_id);
@@ -1552,13 +1550,37 @@ mana_pci_probe(struct rte_pci_driver *pci_drv
__rte_unused,
count = mana_pci_probe_mac(pci_dev, NULL);
}
+ /* If no device is found, clean up resources if this is the last one */
if (!count) {
- rte_memzon
[PATCH] net/netvsc: add stats counters from VF
From: Long Li
The netvsc driver should add per-queue and rx_nombuf counters from VF.
Fixes: 4e9c73e96e83 ("net/netvsc: add Hyper-V network device")
Cc: sta...@dpdk.org
Signed-off-by: Long Li
---
drivers/net/netvsc/hn_ethdev.c | 10 +-
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/drivers/net/netvsc/hn_ethdev.c b/drivers/net/netvsc/hn_ethdev.c
index 988c51928d..069d603d90 100644
--- a/drivers/net/netvsc/hn_ethdev.c
+++ b/drivers/net/netvsc/hn_ethdev.c
@@ -824,8 +824,8 @@ static int hn_dev_stats_get(struct rte_eth_dev *dev,
stats->oerrors += txq->stats.errors;
if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
- stats->q_opackets[i] = txq->stats.packets;
- stats->q_obytes[i] = txq->stats.bytes;
+ stats->q_opackets[i] += txq->stats.packets;
+ stats->q_obytes[i] += txq->stats.bytes;
}
}
@@ -841,12 +841,12 @@ static int hn_dev_stats_get(struct rte_eth_dev *dev,
stats->imissed += rxq->stats.ring_full;
if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
- stats->q_ipackets[i] = rxq->stats.packets;
- stats->q_ibytes[i] = rxq->stats.bytes;
+ stats->q_ipackets[i] += rxq->stats.packets;
+ stats->q_ibytes[i] += rxq->stats.bytes;
}
}
- stats->rx_nombuf = dev->data->rx_mbuf_alloc_failed;
+ stats->rx_nombuf += dev->data->rx_mbuf_alloc_failed;
return 0;
}
--
2.34.1
[DPDK/ethdev Bug 1664] rte_ipv6_addr_is_v4mapped wrong logic
https://bugs.dpdk.org/show_bug.cgi?id=1664
Bug ID: 1664
Summary: rte_ipv6_addr_is_v4mapped wrong logic
Product: DPDK
Version: unspecified
Hardware: All
OS: All
Status: UNCONFIRMED
Severity: normal
Priority: Normal
Component: ethdev
Assignee: dev@dpdk.org
Reporter: sunnyla...@gmail.com
Target Milestone: ---
rte_ipv6_addr_is_v4mapped() is implemented by comparing the first 32 bits of an
IPv6 address with the :::0.0.0.0/96 prefix.
https://git.dpdk.org/dpdk/tree/lib/net/rte_ip6.h?id=fba9875559906e04eaeb74532f4cfd51194259a2#n313
The following snippet prints "1", but the loopback address is not a v4mapped
address.
#include
#include
int main() {
struct rte_ipv6_addr a = RTE_IPV6_ADDR_LOOPBACK;
printf("%d\n", rte_ipv6_addr_is_v4mapped(&a));
}
The function should compare the first 96 bits instead.
rte_ipv6_addr_is_v4compat() has the same mistake.
--
You are receiving this mail because:
You are the assignee for the bug.
[PATCH v3 04/10] net/enic: remove weak symbols
Rather than use weak symbols, expose stubs symbols when needed.
Signed-off-by: David Marchand
---
drivers/net/enic/enic_main.c | 8 +++-
drivers/net/enic/meson.build | 1 +
2 files changed, 4 insertions(+), 5 deletions(-)
diff --git a/drivers/net/enic/enic_main.c b/drivers/net/enic/enic_main.c
index b755b15d92..5ee2ae555d 100644
--- a/drivers/net/enic/enic_main.c
+++ b/drivers/net/enic/enic_main.c
@@ -509,15 +509,13 @@ static void enic_prep_wq_for_simple_tx(struct enic *enic,
uint16_t queue_idx)
}
}
-/*
- * The 'strong' version is in enic_rxtx_vec_avx2.c. This weak version is used
- * used when that file is not compiled.
- */
-__rte_weak bool
+#ifndef ENIC_RXTX_VEC
+bool
enic_use_vector_rx_handler(__rte_unused struct rte_eth_dev *eth_dev)
{
return false;
}
+#endif /* ENIC_RXTX_VEC */
void enic_pick_rx_handler(struct rte_eth_dev *eth_dev)
{
diff --git a/drivers/net/enic/meson.build b/drivers/net/enic/meson.build
index 00f8348348..1e26338350 100644
--- a/drivers/net/enic/meson.build
+++ b/drivers/net/enic/meson.build
@@ -33,6 +33,7 @@ includes += include_directories('base')
# may not. This is to support users who build for the min supported machine
# and need to run the binary on newer CPUs too.
if dpdk_conf.has('RTE_ARCH_X86_64')
+cflags += '-DENIC_RXTX_VEC'
enic_avx2_lib = static_library('enic_avx2_lib',
'enic_rxtx_vec_avx2.c',
dependencies: [static_rte_ethdev, static_rte_bus_pci],
--
2.48.1
[PATCH v3 01/10] bus/auxiliary: remove weak symbols
Rather than use weak symbols, expose stubs symbols when needed.
Signed-off-by: David Marchand
---
drivers/bus/auxiliary/auxiliary_common.c | 6 --
drivers/bus/auxiliary/meson.build| 1 +
2 files changed, 5 insertions(+), 2 deletions(-)
diff --git a/drivers/bus/auxiliary/auxiliary_common.c
b/drivers/bus/auxiliary/auxiliary_common.c
index e6cbc4d356..b444eb43ef 100644
--- a/drivers/bus/auxiliary/auxiliary_common.c
+++ b/drivers/bus/auxiliary/auxiliary_common.c
@@ -36,12 +36,13 @@ auxiliary_devargs_lookup(const char *name)
return NULL;
}
+#ifndef AUXILIARY_OS_SUPPORTED
/*
* Test whether the auxiliary device exist.
*
* Stub for OS not supporting auxiliary bus.
*/
-__rte_weak bool
+bool
auxiliary_dev_exists(const char *name)
{
RTE_SET_USED(name);
@@ -53,11 +54,12 @@ auxiliary_dev_exists(const char *name)
*
* Stub for OS not supporting auxiliary bus.
*/
-__rte_weak int
+int
auxiliary_scan(void)
{
return 0;
}
+#endif /* AUXILIARY_OS_SUPPORTED */
/*
* Update a device's devargs being scanned.
diff --git a/drivers/bus/auxiliary/meson.build
b/drivers/bus/auxiliary/meson.build
index 10468fd130..38d2f05d4b 100644
--- a/drivers/bus/auxiliary/meson.build
+++ b/drivers/bus/auxiliary/meson.build
@@ -7,6 +7,7 @@ sources = files(
'auxiliary_params.c',
)
if is_linux
+cflags += '-DAUXILIARY_OS_SUPPORTED'
sources += files(
'linux/auxiliary.c',
)
--
2.48.1
[PATCH v3 00/10] Remove weak symbols
This is an alternative to André series: https://inbox.dpdk.org/dev/1735009552-31906-1-git-send-email-andre...@linux.microsoft.com/ Weak symbols can be easily replaced with some linking updates, and make it clearer which symbols are actually ending up in the final binary. -- David Marchand Changes since v2: - rebased, Changes since v1: - fix net/hns3 build for Arm64 without SVE, David Marchand (10): bus/auxiliary: remove weak symbols common/qat: remove weak symbols drivers: remove weak symbols in Nitrox drivers net/enic: remove weak symbols net/hns3: remove weak symbols net/fm10k: remove weak symbols net/nfp: remove weak symbols net/virtio: remove weak symbols app/compress-perf: remove weak symbols eal: deprecate weak symbols app/test-compress-perf/main.c| 62 doc/guides/rel_notes/release_25_03.rst | 2 + drivers/bus/auxiliary/auxiliary_common.c | 6 +- drivers/bus/auxiliary/meson.build| 1 + drivers/common/nitrox/meson.build| 3 - drivers/common/nitrox/nitrox_device.c| 75 +--- drivers/common/nitrox/nitrox_device.h| 16 + drivers/common/nitrox/version.map| 1 + drivers/common/qat/qat_qp.c | 8 --- drivers/common/qat/qat_qp.h | 5 -- drivers/compress/nitrox/meson.build | 2 - drivers/compress/nitrox/nitrox_comp.c| 6 ++ drivers/crypto/nitrox/meson.build| 2 - drivers/crypto/nitrox/nitrox_sym.c | 6 ++ drivers/net/enic/enic_main.c | 8 +-- drivers/net/enic/meson.build | 1 + drivers/net/hns3/hns3_rxtx.c | 22 --- drivers/net/intel/fm10k/fm10k_ethdev.c | 20 --- drivers/net/nfp/meson.build | 7 ++- drivers/net/nfp/nfdk/nfp_nfdk_vec_stub.c | 2 +- drivers/net/nfp/nfp_rxtx_vec_stub.c | 4 +- drivers/net/virtio/meson.build | 5 ++ drivers/net/virtio/virtio_rxtx.c | 6 +- drivers/net/virtio/virtio_rxtx_simple.c | 4 +- lib/eal/include/rte_common.h | 6 +- 25 files changed, 119 insertions(+), 161 deletions(-) -- 2.48.1
[PATCH v3 07/10] net/nfp: remove weak symbols
Rather than use weak symbols, link stubs code when needed.
Signed-off-by: David Marchand
Acked-by: Chaoyong He
---
drivers/net/nfp/meson.build | 7 +--
drivers/net/nfp/nfdk/nfp_nfdk_vec_stub.c | 2 +-
drivers/net/nfp/nfp_rxtx_vec_stub.c | 4 ++--
3 files changed, 8 insertions(+), 5 deletions(-)
diff --git a/drivers/net/nfp/meson.build b/drivers/net/nfp/meson.build
index 4052846dc2..39762bd45a 100644
--- a/drivers/net/nfp/meson.build
+++ b/drivers/net/nfp/meson.build
@@ -17,7 +17,6 @@ sources = files(
'flower/nfp_flower_service.c',
'nfd3/nfp_nfd3_dp.c',
'nfdk/nfp_nfdk_dp.c',
-'nfdk/nfp_nfdk_vec_stub.c',
'nfpcore/nfp_cppcore.c',
'nfpcore/nfp_crc.c',
'nfpcore/nfp_elf.c',
@@ -45,7 +44,6 @@ sources = files(
'nfp_net_flow.c',
'nfp_net_meta.c',
'nfp_rxtx.c',
-'nfp_rxtx_vec_stub.c',
'nfp_service.c',
)
@@ -67,6 +65,11 @@ if arch_subdir == 'x86'
)
objs += nfp_avx2_lib.extract_all_objects(recursive: true)
+else
+sources += files(
+'nfp_rxtx_vec_stub.c',
+'nfdk/nfp_nfdk_vec_stub.c',
+)
endif
deps += ['hash', 'security', 'common_nfp']
diff --git a/drivers/net/nfp/nfdk/nfp_nfdk_vec_stub.c
b/drivers/net/nfp/nfdk/nfp_nfdk_vec_stub.c
index 146ec21d51..4f905bce6b 100644
--- a/drivers/net/nfp/nfdk/nfp_nfdk_vec_stub.c
+++ b/drivers/net/nfp/nfdk/nfp_nfdk_vec_stub.c
@@ -5,7 +5,7 @@
#include "nfp_nfdk_vec.h"
-uint16_t __rte_weak
+uint16_t
nfp_net_nfdk_vec_avx2_xmit_pkts(__rte_unused void *tx_queue,
__rte_unused struct rte_mbuf **tx_pkts,
__rte_unused uint16_t nb_pkts)
diff --git a/drivers/net/nfp/nfp_rxtx_vec_stub.c
b/drivers/net/nfp/nfp_rxtx_vec_stub.c
index c480f61ef0..201965afbe 100644
--- a/drivers/net/nfp/nfp_rxtx_vec_stub.c
+++ b/drivers/net/nfp/nfp_rxtx_vec_stub.c
@@ -10,13 +10,13 @@
#include "nfp_rxtx_vec.h"
-bool __rte_weak
+bool
nfp_net_get_avx2_supported(void)
{
return false;
}
-uint16_t __rte_weak
+uint16_t
nfp_net_vec_avx2_recv_pkts(__rte_unused void *rx_queue,
__rte_unused struct rte_mbuf **rx_pkts,
__rte_unused uint16_t nb_pkts)
--
2.48.1
[PATCH v3 05/10] net/hns3: remove weak symbols
Rather than use weak symbols, expose stubs symbols when needed.
Signed-off-by: David Marchand
---
Changes since v1:
- fix build for Arm64 without SVE,
---
drivers/net/hns3/hns3_rxtx.c | 22 +++---
1 file changed, 15 insertions(+), 7 deletions(-)
diff --git a/drivers/net/hns3/hns3_rxtx.c b/drivers/net/hns3/hns3_rxtx.c
index 09e39cb673..bb7ffee12c 100644
--- a/drivers/net/hns3/hns3_rxtx.c
+++ b/drivers/net/hns3/hns3_rxtx.c
@@ -2787,32 +2787,36 @@ hns3_recv_scattered_pkts(void *rx_queue,
return nb_rx;
}
-void __rte_weak
+#ifndef RTE_ARCH_ARM64
+void
hns3_rxq_vec_setup(__rte_unused struct hns3_rx_queue *rxq)
{
}
-int __rte_weak
+int
hns3_rx_check_vec_support(__rte_unused struct rte_eth_dev *dev)
{
return -ENOTSUP;
}
-uint16_t __rte_weak
+uint16_t
hns3_recv_pkts_vec(__rte_unused void *rx_queue,
__rte_unused struct rte_mbuf **rx_pkts,
__rte_unused uint16_t nb_pkts)
{
return 0;
}
+#endif /* RTE_ARCH_ARM64 */
-uint16_t __rte_weak
+#ifndef RTE_HAS_SVE_ACLE
+uint16_t
hns3_recv_pkts_vec_sve(__rte_unused void *rx_queue,
__rte_unused struct rte_mbuf **rx_pkts,
__rte_unused uint16_t nb_pkts)
{
return 0;
}
+#endif /* RTE_HAS_SVE_ACLE */
int
hns3_rx_burst_mode_get(struct rte_eth_dev *dev, __rte_unused uint16_t queue_id,
@@ -4256,27 +4260,31 @@ hns3_xmit_pkts(void *tx_queue, struct rte_mbuf
**tx_pkts, uint16_t nb_pkts)
return nb_tx;
}
-int __rte_weak
+#ifndef RTE_ARCH_ARM64
+int
hns3_tx_check_vec_support(__rte_unused struct rte_eth_dev *dev)
{
return -ENOTSUP;
}
-uint16_t __rte_weak
+uint16_t
hns3_xmit_pkts_vec(__rte_unused void *tx_queue,
__rte_unused struct rte_mbuf **tx_pkts,
__rte_unused uint16_t nb_pkts)
{
return 0;
}
+#endif /* RTE_ARCH_ARM64 */
-uint16_t __rte_weak
+#ifndef RTE_HAS_SVE_ACLE
+uint16_t
hns3_xmit_pkts_vec_sve(void __rte_unused * tx_queue,
struct rte_mbuf __rte_unused **tx_pkts,
uint16_t __rte_unused nb_pkts)
{
return 0;
}
+#endif /* RTE_HAS_SVE_ACLE */
int
hns3_tx_burst_mode_get(struct rte_eth_dev *dev, __rte_unused uint16_t queue_id,
--
2.48.1
[PATCH v3 02/10] common/qat: remove weak symbols
Remove dead code, those symbols are never used as the "strong" symbols
are always linked in.
Signed-off-by: David Marchand
---
drivers/common/qat/qat_qp.c | 8
drivers/common/qat/qat_qp.h | 5 -
2 files changed, 13 deletions(-)
diff --git a/drivers/common/qat/qat_qp.c b/drivers/common/qat/qat_qp.c
index 4bf9bac23e..0d2bbdb8a5 100644
--- a/drivers/common/qat/qat_qp.c
+++ b/drivers/common/qat/qat_qp.c
@@ -944,11 +944,3 @@ qat_cq_get_fw_cipher_crc_cap(struct qat_qp *qp)
return ret;
}
#endif
-
-__rte_weak int
-qat_comp_process_response(void **op __rte_unused, uint8_t *resp __rte_unused,
- void *op_cookie __rte_unused,
- uint64_t *dequeue_err_count __rte_unused)
-{
- return 0;
-}
diff --git a/drivers/common/qat/qat_qp.h b/drivers/common/qat/qat_qp.h
index f0ea907503..5ccaedefa7 100644
--- a/drivers/common/qat/qat_qp.h
+++ b/drivers/common/qat/qat_qp.h
@@ -156,11 +156,6 @@ int
qat_cq_get_fw_cipher_crc_cap(struct qat_qp *qp);
#endif
-/* Needed for weak function*/
-int
-qat_comp_process_response(void **op __rte_unused, uint8_t *resp __rte_unused,
- void *op_cookie __rte_unused,
- uint64_t *dequeue_err_count __rte_unused);
int
qat_read_qp_config(struct qat_pci_device *qat_dev);
--
2.48.1
RE: [PATCH] net/mlx5: fix leak in HWS flow counter action
> -Original Message-
> From: David Marchand
> Sent: Monday, January 27, 2025 11:54
> To: dev@dpdk.org
> Cc: sta...@dpdk.org; Dariusz Sosnowski ; Slava
> Ovsiienko ; Bing Zhao ; Ori Kam
> ; Suanming Mou ; Matan Azrad
> ; Jack Min
> Subject: [PATCH] net/mlx5: fix leak in HWS flow counter action
>
> External email: Use caution opening links or attachments
>
>
> This was caught by our internal covscan.
> mp_name can be leaked in case of errors.
>
> Fixes: 4d368e1da3a4 ("net/mlx5: support flow counter action for HWS")
> Cc: sta...@dpdk.org
>
> Signed-off-by: David Marchand
Acked-by: Dariusz Sosnowski
Best regards,
Dariusz Sosnowski
Re: [PATCH v2 2/2] net/intel: add Tx time queue
On Wed, Feb 12, 2025 at 09:47:11PM +, Soumyadeep Hore wrote:
> Enabling Tx timestamp queue for supporting Tx time based
> scheduling of packets.
>
Can you provide more details of this feature and how it can be used, how it
is enabled/disabled etc.
See also comments inline below.
/Bruce
> Signed-off-by: Soumyadeep Hore
> ---
> drivers/net/intel/common/tx.h | 5 +
> drivers/net/intel/ice/base/ice_lan_tx_rx.h | 1 +
> drivers/net/intel/ice/ice_ethdev.h | 1 +
> drivers/net/intel/ice/ice_rxtx.c | 174 +
> drivers/net/intel/ice/ice_rxtx.h | 5 +
> 5 files changed, 186 insertions(+)
>
> diff --git a/drivers/net/intel/common/tx.h b/drivers/net/intel/common/tx.h
> index d9cf4474fc..f3777fa9e7 100644
> --- a/drivers/net/intel/common/tx.h
> +++ b/drivers/net/intel/common/tx.h
> @@ -35,6 +35,7 @@ struct ci_tx_queue {
> volatile struct i40e_tx_desc *i40e_tx_ring;
> volatile struct iavf_tx_desc *iavf_tx_ring;
> volatile struct ice_tx_desc *ice_tx_ring;
> + volatile struct ice_ts_desc *ice_tstamp_ring;
This code looks a bit strange to me, can you please check my understanding
of it is correct.
This is a union, so the time stamp ring here is replacing the whole
descriptor ring for the queue? Therefore, we will have some queues which
have regular descriptors and others which have only timestamp descriptors.
Is that correct?
Another minor point is that this union has the elements in alphabetical
order, so ice_ts_desc needs to come before ice_tx_desc.
> volatile union ixgbe_adv_tx_desc *ixgbe_tx_ring;
> };
> volatile uint8_t *qtx_tail; /* register address of tail */
> @@ -76,6 +77,10 @@ struct ci_tx_queue {
> union {
> struct { /* ICE driver specific values */
> uint32_t q_teid; /* TX schedule node id. */
> + uint16_t nb_tstamp_desc;/* number of Timestamp
> descriptors */
> + volatile uint8_t *tstamp_tail; /* value of timestamp
> tail register */
> + rte_iova_t tstamp_ring_dma; /* Timestamp ring DMA
> address */
> + uint16_t next_tstamp_id;
You are adding lots of holes into the structure here, please reorder the
fields to reduce the space used by the structure.
Also, do you need the field tstamp_tail? Since ice_tstamp_ring is replacing
ice_tx_ring in the union above, you should be able to just reuse the
existing tail register for this, no?
Similarly for tstamp_ring_dma, can the existing dma address field not be
used.
OVerall, I think rather than expanding out our common tx queue structure,
it may be better to have the queue structure hold a pointer to another
separate tx timestamp structure, allocated separately.
> };
> struct { /* I40E driver specific values */
> uint8_t dcb_tc;
> diff --git a/drivers/net/intel/ice/base/ice_lan_tx_rx.h
> b/drivers/net/intel/ice/base/ice_lan_tx_rx.h
> index 940c6843d9..edd1137114 100644
> --- a/drivers/net/intel/ice/base/ice_lan_tx_rx.h
> +++ b/drivers/net/intel/ice/base/ice_lan_tx_rx.h
> @@ -1279,6 +1279,7 @@ struct ice_ts_desc {
> #define ICE_SET_TXTIME_MAX_Q_AMOUNT 127
> #define ICE_OP_TXTIME_MAX_Q_AMOUNT 2047
> #define ICE_TXTIME_FETCH_TS_DESC_DFLT8
> +#define ICE_TXTIME_FETCH_PROFILE_CNT 16
>
> /* Tx Time queue context data
> *
This base code update, adding a define, should be in the previous patch
where all the other base code defines are added.
> diff --git a/drivers/net/intel/ice/ice_ethdev.h
> b/drivers/net/intel/ice/ice_ethdev.h
> index afe8dae497..9649456771 100644
> --- a/drivers/net/intel/ice/ice_ethdev.h
> +++ b/drivers/net/intel/ice/ice_ethdev.h
> @@ -299,6 +299,7 @@ struct ice_vsi {
> uint8_t enabled_tc; /* The traffic class enabled */
> uint8_t vlan_anti_spoof_on; /* The VLAN anti-spoofing enabled */
> uint8_t vlan_filter_on; /* The VLAN filter enabled */
> + uint8_t enabled_txpp; /* TXPP support enabled */
While I realise that "enabled_txpp" matches the "enabled_tc" variable
above, it would read better as "txpp_enabled". You could also have it align
to the previous two members, perhaps: would it work calling it "txpp_on".
> /* information about rss configuration */
> u32 rss_key_size;
> u32 rss_lut_size;
> diff --git a/drivers/net/intel/ice/ice_rxtx.c
> b/drivers/net/intel/ice/ice_rxtx.c
> index 8dd8644b16..f043ae3aa6 100644
> --- a/drivers/net/intel/ice/ice_rxtx.c
> +++ b/drivers/net/intel/ice/ice_rxtx.c
> @@ -5,6 +5,7 @@
> #include
> #include
> #include
> +#include
>
> #include "ice_rxtx.h"
> #include "ice_rxtx_vec_common.h"
> @@ -741,6 +742,87 @@ ice_rx_queue_stop(struct rte_eth_dev *dev, uint16_t
> rx_queue_id)
> return 0;
> }
>
> +/**
> + * ice_setup_txtime_ctx - setup a struct ice_txtime_ctx instance
> + * @ring: The
Re: [PATCH v2 07/10] test-pmd: don't return value from void function
On Tue, Feb 18, 2025 at 08:32:06AM -0800, Andre Muezerie wrote: > Compiling with MSVC results in the warning below: > > app/test-pmd/cmdline_flow.c(13964): warning C4098: 'cmd_set_raw_parsed': > 'void' function returning a value > > Signed-off-by: Andre Muezerie > Signed-off-by: Chengwen Feng > --- Acked-by: Bruce Richardson
Re: [PATCH v2 03/10] test-pmd: fix printf format string mismatch
On Tue, Feb 18, 2025 at 08:32:02AM -0800, Andre Muezerie wrote:
> Compiling with MSVC results in warnings like the one below:
>
> app/test-pmd/csumonly.c(1085): warning C4477: 'printf' : format string
> '%d' requires an argument of type 'int',
> but variadic argument 1 has type 'uint64_t'
>
> Signed-off-by: Andre Muezerie
> Signed-off-by: Chengwen Feng
> ---
> app/test-pmd/csumonly.c | 23 ---
> 1 file changed, 12 insertions(+), 11 deletions(-)
>
> diff --git a/app/test-pmd/csumonly.c b/app/test-pmd/csumonly.c
> index d77a140641..8de5ad6ad9 100644
> --- a/app/test-pmd/csumonly.c
> +++ b/app/test-pmd/csumonly.c
> @@ -1070,7 +1070,7 @@ pkt_burst_checksum_forward(struct fwd_stream *fs)
> info.l2_len, rte_be_to_cpu_16(info.ethertype),
> info.l3_len, info.l4_proto, info.l4_len, buf);
> if (rx_ol_flags & RTE_MBUF_F_RX_LRO)
> - printf("rx: m->lro_segsz=%u\n", m->tso_segsz);
> + printf("rx: m->lro_segsz=%" PRIu64 "\n",
> (uint64_t)m->tso_segsz);
tso_segsz is already uint64_t, so no need for the cast.
> if (info.is_tunnel == 1)
> printf("rx: outer_l2_len=%d outer_ethertype=%x "
> "outer_l3_len=%d\n", info.outer_l2_len,
> @@ -1082,28 +1082,29 @@ pkt_burst_checksum_forward(struct fwd_stream *fs)
> RTE_ETH_TX_OFFLOAD_TCP_CKSUM |
> RTE_ETH_TX_OFFLOAD_SCTP_CKSUM)) ||
> info.tso_segsz != 0)
> - printf("tx: m->l2_len=%d m->l3_len=%d "
> - "m->l4_len=%d\n",
> - m->l2_len, m->l3_len, m->l4_len);
> + printf("tx: m->l2_len=%" PRIu64 " m->l3_len=%"
> PRIu64
> + " m->l4_len=%" PRIu64 "\n",
> + (uint64_t)m->l2_len,
> (uint64_t)m->l3_len,
> + (uint64_t)m->l4_len);
Same here, using casts and a changed print format seems wrong in the patch.
Either we change the format string to match the variable type, or we cast
the variables to match the format string. We should not do both, IMHO.
> if (info.is_tunnel == 1) {
> if ((tx_offloads &
> RTE_ETH_TX_OFFLOAD_OUTER_IPV4_CKSUM) ||
> (tx_offloads &
> RTE_ETH_TX_OFFLOAD_OUTER_UDP_CKSUM) ||
> (tx_ol_flags & RTE_MBUF_F_TX_OUTER_IPV6))
> - printf("tx: m->outer_l2_len=%d "
> - "m->outer_l3_len=%d\n",
> - m->outer_l2_len,
> - m->outer_l3_len);
> + printf("tx: m->outer_l2_len=%" PRIu64
> + " m->outer_l3_len=%" PRIu64
> "\n",
> + (uint64_t)m->outer_l2_len,
> + (uint64_t)m->outer_l3_len);
> if (info.tunnel_tso_segsz != 0 &&
> (m->ol_flags &
> (RTE_MBUF_F_TX_TCP_SEG |
> RTE_MBUF_F_TX_UDP_SEG)))
> - printf("tx: m->tso_segsz=%d\n",
> - m->tso_segsz);
> + printf("tx: m->tso_segsz=%" PRIu64 "\n",
> + (uint64_t)m->tso_segsz);
> } else if (info.tso_segsz != 0 &&
> (m->ol_flags & (RTE_MBUF_F_TX_TCP_SEG |
> RTE_MBUF_F_TX_UDP_SEG)))
> - printf("tx: m->tso_segsz=%d\n", m->tso_segsz);
> + printf("tx: m->tso_segsz=%" PRIu64 "\n",
> (uint64_t)m->tso_segsz);
> rte_get_tx_ol_flag_list(m->ol_flags, buf, sizeof(buf));
> printf("tx: flags=%s", buf);
> printf("\n");
> --
> 2.48.1.vfs.0.0
>
Re: [PATCH v2 10/10] app: enable app directory to be compiled with MSVC
On Tue, Feb 18, 2025 at 08:32:09AM -0800, Andre Muezerie wrote:
> Enabled "app" directory to be compiled with MSVC along with all its
> contents.
>
> Signed-off-by: Andre Muezerie
> ---
> app/meson.build | 4
> app/test-pmd/meson.build | 4 +++-
> 2 files changed, 3 insertions(+), 5 deletions(-)
>
> diff --git a/app/meson.build b/app/meson.build
> index e2db888ae1..1798db3ae4 100644
> --- a/app/meson.build
> +++ b/app/meson.build
> @@ -1,10 +1,6 @@
> # SPDX-License-Identifier: BSD-3-Clause
> # Copyright(c) 2017-2019 Intel Corporation
>
> -if is_ms_compiler
> -subdir_done()
> -endif
> -
> disable_apps = ',' + get_option('disable_apps')
> disable_apps = run_command(list_dir_globs, disable_apps, check:
> true).stdout().split()
>
> diff --git a/app/test-pmd/meson.build b/app/test-pmd/meson.build
> index f1c36529b4..8e8863812a 100644
> --- a/app/test-pmd/meson.build
> +++ b/app/test-pmd/meson.build
> @@ -3,7 +3,9 @@
>
> # override default name to drop the hyphen
> name = 'testpmd'
> -cflags += '-Wno-deprecated-declarations'
> +if cc.has_argument('-Wno-deprecated-declarations')
> +cflags += '-Wno-deprecated-declarations'
> +endif
> sources = files(
> '5tswap.c',
> 'cmdline.c',
> --
Acked-by: Bruce Richardson
Re: [PATCH v2 04/10] test-pmd: do explicit 64-bit shift to avoid implicit conversion
On Tue, Feb 18, 2025 at 08:32:03AM -0800, Andre Muezerie wrote: > Compiling with MSVC results in warnings like the one below: > > app/test-pmd/util.c(201): warning C4334: '<<': result of 32-bit shift > implicitly converted to 64 bits (was 64-bit shift intended?) > > Signed-off-by: Andre Muezerie > Signed-off-by: Chengwen Feng > --- Acked-by: Bruce Richardson
Re: [PATCH] ci: build with MSVC in GHA
David Marchand writes: > On Tue, Feb 18, 2025 at 2:32 PM Aaron Conole wrote: >> >> David Marchand writes: >> >> > Add a build target in Windows 2022 for MSVC. >> > >> > Loading MSVC environment and having successive commands in a single >> > (cmd or pwsh) shell does not seem to correctly report if one of the >> > commands failed. >> > For now, load MSVC environment and run commands in separate shells. >> > >> > Signed-off-by: David Marchand >> > --- >> >> LGTM. >> >> Acked-by: Aaron Conole >> >> How far back will this work (ie: can we backport this to older branches?) > > MSVC was added in 23.11. > The patch applies cleanly and runs on 23.11 (where only lib/kvargs, > lib/log and lib/telemetry are compiled) and 24.11. > > I can add Cc: stable if you think it is worth having it in LTS (UNH > provides some coverage in LTS releases). I usually prefer to back-port CI related changes because it helps us catch issues / improve the stable release cycle.
Re: [PATCH v6 1/8] build: introduce optional internal dependencies
On Tue, Feb 18, 2025 at 11:15:27AM +, Anatoly Burakov wrote:
> Allow specifying internal dependencies as either mandatory or optional.
> Specifying a dependency as optional will mean that the component being
s/optional/mandatory/
> built will be skipped if said dependency is not being built. At build
> time, the build system will resolve any optional dependencies and add
> them to the list of dependencies to be built. Any source files requiring
> said optional depepdencies will still have to be added explicitly to the
> build by the respective component (e.g. adding BPF-related files when BPF
> support is being built).
>
> Signed-off-by: Bruce Richardson
> Signed-off-by: Anatoly Burakov
> ---
>
> Notes:
> v4 -> v5:
> - Automatically handle optional dependencies based on Bruce's
> earlier patch [1]
>
> [1]
> https://patches.dpdk.org/project/dpdk/patch/20231220142152.492556-4-bruce.richard...@intel.com/
>
> app/meson.build | 14 ++-
> app/proc-info/meson.build | 4 +-
> app/test-bbdev/meson.build| 18 +++--
> app/test-crypto-perf/meson.build | 4 +-
> app/test-pmd/meson.build | 56 +--
> app/test/meson.build | 12 +++---
> drivers/meson.build | 13 ++-
> examples/ethtool/meson.build | 4 +-
> examples/l2fwd-crypto/meson.build | 4 +-
> examples/l3fwd/meson.build| 4 +-
> examples/meson.build | 13 ++-
> examples/vm_power_manager/meson.build | 16 +++-
> lib/meson.build | 13 ++-
> 13 files changed, 89 insertions(+), 86 deletions(-)
>
> diff --git a/app/meson.build b/app/meson.build
> index e2db888ae1..61202495bd 100644
> --- a/app/meson.build
> +++ b/app/meson.build
> @@ -66,6 +66,7 @@ foreach app:apps
> # external package/library requirements
> ext_deps = []
> deps = []
> +optional_deps = []
>
> if not enable_apps.contains(app)
> build = false
> @@ -83,10 +84,19 @@ foreach app:apps
> endif
> endif
>
> +# resolve any optional internal dependencies
> +def_lib = get_option('default_library')
> +foreach d: optional_deps
> +# if optional dependency is defined, add it to the deps list
> +if is_variable(def_lib + '_rte_' + d)
> +deps += [d]
> +endif
> +endforeach
> +
> if build
> dep_objs = []
> foreach d:deps
> -var_name = get_option('default_library') + '_rte_' + d
> +var_name = def_lib + '_rte_' + d
> if not is_variable(var_name)
> build = false
> reason = 'missing internal dependency, "@0@"'.format(d)
> @@ -111,7 +121,7 @@ foreach app:apps
>
> dpdk_apps_enabled += app
> link_libs = []
> -if get_option('default_library') == 'static'
> +if def_lib == 'static'
> link_libs = dpdk_static_libraries + dpdk_drivers
> endif
>
> diff --git a/app/proc-info/meson.build b/app/proc-info/meson.build
> index 316253cd5d..51315f72a1 100644
> --- a/app/proc-info/meson.build
> +++ b/app/proc-info/meson.build
> @@ -9,8 +9,6 @@ endif
>
> sources = files('main.c')
> deps += ['ethdev', 'security', 'eventdev']
> -if dpdk_conf.has('RTE_LIB_METRICS')
> -deps += 'metrics'
> -endif
> +optional_deps += 'metrics'
>
> cflags += no_wvla_cflag
> diff --git a/app/test-bbdev/meson.build b/app/test-bbdev/meson.build
> index 85b060edec..d41bd60396 100644
> --- a/app/test-bbdev/meson.build
> +++ b/app/test-bbdev/meson.build
> @@ -14,17 +14,11 @@ sources = files(
> 'test_bbdev_vector.c',
> )
> deps += ['bbdev', 'bus_vdev']
> -if dpdk_conf.has('RTE_BASEBAND_FPGA_LTE_FEC')
> -deps += ['baseband_fpga_lte_fec']
> -endif
> -if dpdk_conf.has('RTE_BASEBAND_FPGA_5GNR_FEC')
> -deps += ['baseband_fpga_5gnr_fec']
> -endif
> -if dpdk_conf.has('RTE_BASEBAND_ACC')
> -deps += ['baseband_acc']
> -endif
> -if dpdk_conf.has('RTE_BASEBAND_LA12XX')
> -deps += ['baseband_la12xx']
> -endif
> +optional_deps += [
> +'baseband_fpga_lte_fec',
> +'baseband_fpga_5gnr_fec',
> +'baseband_acc',
> +'baseband_la12xx',
> +]
>
> cflags += no_wvla_cflag
> diff --git a/app/test-crypto-perf/meson.build
> b/app/test-crypto-perf/meson.build
> index 87dd3bc5f1..fb48d9ec29 100644
> --- a/app/test-crypto-perf/meson.build
> +++ b/app/test-crypto-perf/meson.build
> @@ -20,8 +20,6 @@ sources = files(
> 'main.c',
> )
> deps += ['cryptodev', 'net', 'security']
> -if dpdk_conf.has('RTE_CRYPTO_SCHEDULER')
> -deps += 'crypto_scheduler'
> -endif
> +optional_deps += 'crypto_scheduler'
>
> cflags += no_wvla_cflag
> diff --git a/app/test-pmd/meson.build b/app/test-pmd/meson.build
> index f1c36529b4..bbc8d429f8 100644
> --- a/app/test-pmd/meson.build
> +++ b/app/test-pmd/meson.build
> @@ -34,47 +34,29 @@ sources = files(
> if dpdk_conf.h
Re: [PATCH] ci: build with MSVC in GHA
On Tue, Feb 18, 2025 at 2:32 PM Aaron Conole wrote: > > David Marchand writes: > > > Add a build target in Windows 2022 for MSVC. > > > > Loading MSVC environment and having successive commands in a single > > (cmd or pwsh) shell does not seem to correctly report if one of the > > commands failed. > > For now, load MSVC environment and run commands in separate shells. > > > > Signed-off-by: David Marchand > > --- > > LGTM. > > Acked-by: Aaron Conole > > How far back will this work (ie: can we backport this to older branches?) MSVC was added in 23.11. The patch applies cleanly and runs on 23.11 (where only lib/kvargs, lib/log and lib/telemetry are compiled) and 24.11. I can add Cc: stable if you think it is worth having it in LTS (UNH provides some coverage in LTS releases). -- David Marchand
Re: [PATCH v22 00/27] remove use of VLAs for Windows
On Fri, Feb 7, 2025 at 3:23 PM Konstantin Ananyev wrote: > > > As per guidance technical board meeting 2024/04/17. This series > > > removes the use of VLAs from code built for Windows for all 3 > > > toolchains. If there are additional opportunities to convert VLAs > > > to regular C arrays please provide the details for incorporation > > > into the series. > > > > > > MSVC does not support VLAs, replace VLAs with standard C arrays > > > or alloca(). alloca() is available for all toolchain/platform > > > combinations officially supported by DPDK. > > > > - I have one concern wrt patch 7. > > This changes the API/ABI of the RCU library. > > ABI can't be broken in the 25.03 release. > > > > Since MSVC builds do not include RCU yet, I skipped this change and > > adjusted this libray meson.build. > > > > Konstantin, do you think patch 7 could be rewritten to make use of > > alloca() and avoid an API change? > > https://patchwork.dpdk.org/project/dpdk/patch/1738805610-17507-8-git-send-email-andre...@linux.microsoft.com/ > > I am not big fan of alloca() approach, but yes it is surely possible. Can you please explain your reluctance? > BTW, why it is considered ad API/ABI change? > Because we introduce extra limit on max allowable size? Yes, this is what was mentionned in the commitlog. > If that would help somehow, we can make it even bigger: 1K or so. Strictly speaking, it is still an API change. -- David Marchand
Re: [EXTERNAL] [PATCH v3 4/6] trace: support dumping binary inside a struct
On Wed, Feb 12, 2025 at 6:14 AM Sunil Kumar Kori wrote:
>
> > On Tue, Feb 11, 2025 at 9:53 AM Sunil Kumar Kori
> > wrote:
> > >
> > > > diff --git a/lib/eal/common/eal_common_trace_ctf.c
> > > > b/lib/eal/common/eal_common_trace_ctf.c
> > > > index 6bc8bb9036..d9b307e076 100644
> > > > --- a/lib/eal/common/eal_common_trace_ctf.c
> > > > +++ b/lib/eal/common/eal_common_trace_ctf.c
> > > > @@ -378,6 +378,9 @@ char *trace_metadata_fixup_field(const char
> > *field)
> > > > "->",
> > > > "*",
> > > > " ",
> > > > + "&",
> > > > + "(",
> > > > + ")",
> > > Adding brackets makes token names a bit complex. Same name is used in
> > > metadata file to dump the traces to the user. With this complex name,
> > > user might not understand the purpose of that information.
> > >
> > > For example, _conf_src_port_pcie_sizeof_uint64_t_ is created in
> > > metafile and same will be dumped. But with this User might not get that
> > which information is provided.
> >
> > In practice, as there is no other documentation for a trace point
> > arguments, a
> > user needs to read the trace point definitions.
> > So it seems trivial to me to link a variable name in the trace point
> > emitter, and
> > the metadata in the trace files.
> >
> > >
> > > This is the reason; we followed the existing naming convention which is
> > > user
> > friendly.
> >
> > User friendly? I don't see how this is different with '.' and '->'.
> In general, structure fields are given a proper name to represent the purpose.
> When we use it directly in trace point using '.' or '->' then it remains a
> meaningful name.
> Adding more tokens in name, is making them complex and deviating from there
> meaning.
>
> I am not saying that the mentioned support should not be there. I am just
> trying to convey
> that If it is possible to make meaningful names, then that will be more
> helpful.
Hard to preserve such information given the limitations of the C
parser (which seems to apply to the CTF format).
I still think that interpretation of the metadata in the traces
require looking at the source code, which means that the "readability"
objection is weak.
Jerin, opinion please.
--
David Marchand
[PATCH 1/2] doc: fix mlx5 VXLAN matching limitations
P bit of VXLAN-GPE and G bit of VXLAN-GBP should be set when matching the
protocols.
Fixes: 49765b78c81b ("net/mlx5/hws: match all VXLAN fields")
Signed-off-by: Minggang Li(Gavin)
---
doc/guides/nics/mlx5.rst | 4
1 file changed, 4 insertions(+)
diff --git a/doc/guides/nics/mlx5.rst b/doc/guides/nics/mlx5.rst
index 690b46a974..caf711ce67 100644
--- a/doc/guides/nics/mlx5.rst
+++ b/doc/guides/nics/mlx5.rst
@@ -339,6 +339,10 @@ Limitations
- For ConnectX-5, the UDP destination port must be the standard one (4789).
- Default UDP destination is 4789 if not explicitly specified.
- Group zero's behavior may differ which depends on FW.
+ - User should set different flags when matching on VXLAN-GPE/GBP:
+
+ - for VXLAN-GPE - P flag
+ - for VXLAN-GBP - G flag
- Matching on VXLAN-GPE header fields:
--
2.34.1
[PATCH 2/2] app/testpmd: use flags from user if it is set
It's user's responsibility to set correct flags, especially the G bit, to
distinguish the protocols, VXLAN and VXLAN-GBP, sharing the same port 4789.
It will be set to 0x8 if user does not specify the flags.
Fixes: da118115d95c ("app/testpmd: support matching any VXLAN field")
Signed-off-by: Minggang Li(Gavin)
---
app/test-pmd/cmdline_flow.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c
index e1720e54d7..3947304b63 100644
--- a/app/test-pmd/cmdline_flow.c
+++ b/app/test-pmd/cmdline_flow.c
@@ -13598,7 +13598,8 @@ update_fields(uint8_t *buf, struct rte_flow_item *item,
uint16_t next_proto)
break;
case RTE_FLOW_ITEM_TYPE_VXLAN:
vxlan = (struct rte_vxlan_hdr *)buf;
- vxlan->vx_flags = 0x08;
+ if (!vxlan->flags)
+ vxlan->flags = 0x08;
break;
case RTE_FLOW_ITEM_TYPE_VXLAN_GPE:
gpe = (struct rte_vxlan_gpe_hdr *)buf;
--
2.34.1
RE: [PATCH v3 1/2] crypto/ipsec_mb: add SM4 GCM support
Hi Brian,
> -Original Message-
> From: Dooley, Brian
> Sent: Monday, January 13, 2025 6:10 PM
> Cc: dev@dpdk.org; gak...@marvell.com; Ji, Kai ; De Lara
> Guarch, Pablo ; Dooley, Brian
>
> Subject: [PATCH v3 1/2] crypto/ipsec_mb: add SM4 GCM support
>
> This patch introduces SM4 GCM algorithm support to the AESNI_MB PMD.
> SM4 GCM is available in the v2.0 release of Intel IPsec MB.
>
> Signed-off-by: Brian Dooley
> ---
A couple of comments below.
Thanks,
Pablo
...
> --- a/drivers/crypto/ipsec_mb/pmd_aesni_mb_priv.h
> +++ b/drivers/crypto/ipsec_mb/pmd_aesni_mb_priv.h
> @@ -826,6 +826,36 @@ static const struct rte_cryptodev_capabilities
> aesni_mb_capabilities[] = {
> }, }
> }, }
> },
> + { /* SM4 GCM */
> + .op = RTE_CRYPTO_OP_TYPE_SYMMETRIC,
> + {.sym = {
> + .xform_type = RTE_CRYPTO_SYM_XFORM_AEAD,
> + {.aead = {
> + .algo = RTE_CRYPTO_AEAD_SM4_GCM,
> + .block_size = 16,
> + .key_size = {
> + .min = 16,
> + .max = 16,
> + .increment = 0,
> + },
> + .digest_size = {
> + .min = 16,
> + .max = 16,
> + .increment = 0,
Digest size can be 1 to 16 bytes.
> + },
> + .aad_size = {
> + .min = 0,
> + .max = 65535,
> + .increment = 1,
> + },
> + .iv_size = {
> + .min = 12,
> + .max = 12,
> + .increment = 0,
> + }
> + }, }
> + }, }
> + },
> #endif
> RTE_CRYPTODEV_END_OF_CAPABILITIES_LIST()
> };
> diff --git a/lib/cryptodev/rte_crypto_sym.h b/lib/cryptodev/rte_crypto_sym.h
> index 505356ff44..b47e52f63b 100644
> --- a/lib/cryptodev/rte_crypto_sym.h
> +++ b/lib/cryptodev/rte_crypto_sym.h
> @@ -482,8 +482,10 @@ enum rte_crypto_aead_algorithm {
> /**< AES algorithm in CCM mode. */
> RTE_CRYPTO_AEAD_AES_GCM,
> /**< AES algorithm in GCM mode. */
> - RTE_CRYPTO_AEAD_CHACHA20_POLY1305
> + RTE_CRYPTO_AEAD_CHACHA20_POLY1305,
> /**< Chacha20 cipher with poly1305 authenticator */
> + RTE_CRYPTO_AEAD_SM4_GCM
> + /**< SM4 cipher with GCM mode */
To keep consistency, "in GCM mode".
> };
[PATCH v6 1/8] build: introduce optional internal dependencies
Allow specifying internal dependencies as either mandatory or optional.
Specifying a dependency as optional will mean that the component being
built will be skipped if said dependency is not being built. At build
time, the build system will resolve any optional dependencies and add
them to the list of dependencies to be built. Any source files requiring
said optional depepdencies will still have to be added explicitly to the
build by the respective component (e.g. adding BPF-related files when BPF
support is being built).
Signed-off-by: Bruce Richardson
Signed-off-by: Anatoly Burakov
---
Notes:
v4 -> v5:
- Automatically handle optional dependencies based on Bruce's
earlier patch [1]
[1]
https://patches.dpdk.org/project/dpdk/patch/20231220142152.492556-4-bruce.richard...@intel.com/
app/meson.build | 14 ++-
app/proc-info/meson.build | 4 +-
app/test-bbdev/meson.build| 18 +++--
app/test-crypto-perf/meson.build | 4 +-
app/test-pmd/meson.build | 56 +--
app/test/meson.build | 12 +++---
drivers/meson.build | 13 ++-
examples/ethtool/meson.build | 4 +-
examples/l2fwd-crypto/meson.build | 4 +-
examples/l3fwd/meson.build| 4 +-
examples/meson.build | 13 ++-
examples/vm_power_manager/meson.build | 16 +++-
lib/meson.build | 13 ++-
13 files changed, 89 insertions(+), 86 deletions(-)
diff --git a/app/meson.build b/app/meson.build
index e2db888ae1..61202495bd 100644
--- a/app/meson.build
+++ b/app/meson.build
@@ -66,6 +66,7 @@ foreach app:apps
# external package/library requirements
ext_deps = []
deps = []
+optional_deps = []
if not enable_apps.contains(app)
build = false
@@ -83,10 +84,19 @@ foreach app:apps
endif
endif
+# resolve any optional internal dependencies
+def_lib = get_option('default_library')
+foreach d: optional_deps
+# if optional dependency is defined, add it to the deps list
+if is_variable(def_lib + '_rte_' + d)
+deps += [d]
+endif
+endforeach
+
if build
dep_objs = []
foreach d:deps
-var_name = get_option('default_library') + '_rte_' + d
+var_name = def_lib + '_rte_' + d
if not is_variable(var_name)
build = false
reason = 'missing internal dependency, "@0@"'.format(d)
@@ -111,7 +121,7 @@ foreach app:apps
dpdk_apps_enabled += app
link_libs = []
-if get_option('default_library') == 'static'
+if def_lib == 'static'
link_libs = dpdk_static_libraries + dpdk_drivers
endif
diff --git a/app/proc-info/meson.build b/app/proc-info/meson.build
index 316253cd5d..51315f72a1 100644
--- a/app/proc-info/meson.build
+++ b/app/proc-info/meson.build
@@ -9,8 +9,6 @@ endif
sources = files('main.c')
deps += ['ethdev', 'security', 'eventdev']
-if dpdk_conf.has('RTE_LIB_METRICS')
-deps += 'metrics'
-endif
+optional_deps += 'metrics'
cflags += no_wvla_cflag
diff --git a/app/test-bbdev/meson.build b/app/test-bbdev/meson.build
index 85b060edec..d41bd60396 100644
--- a/app/test-bbdev/meson.build
+++ b/app/test-bbdev/meson.build
@@ -14,17 +14,11 @@ sources = files(
'test_bbdev_vector.c',
)
deps += ['bbdev', 'bus_vdev']
-if dpdk_conf.has('RTE_BASEBAND_FPGA_LTE_FEC')
-deps += ['baseband_fpga_lte_fec']
-endif
-if dpdk_conf.has('RTE_BASEBAND_FPGA_5GNR_FEC')
-deps += ['baseband_fpga_5gnr_fec']
-endif
-if dpdk_conf.has('RTE_BASEBAND_ACC')
-deps += ['baseband_acc']
-endif
-if dpdk_conf.has('RTE_BASEBAND_LA12XX')
-deps += ['baseband_la12xx']
-endif
+optional_deps += [
+'baseband_fpga_lte_fec',
+'baseband_fpga_5gnr_fec',
+'baseband_acc',
+'baseband_la12xx',
+]
cflags += no_wvla_cflag
diff --git a/app/test-crypto-perf/meson.build b/app/test-crypto-perf/meson.build
index 87dd3bc5f1..fb48d9ec29 100644
--- a/app/test-crypto-perf/meson.build
+++ b/app/test-crypto-perf/meson.build
@@ -20,8 +20,6 @@ sources = files(
'main.c',
)
deps += ['cryptodev', 'net', 'security']
-if dpdk_conf.has('RTE_CRYPTO_SCHEDULER')
-deps += 'crypto_scheduler'
-endif
+optional_deps += 'crypto_scheduler'
cflags += no_wvla_cflag
diff --git a/app/test-pmd/meson.build b/app/test-pmd/meson.build
index f1c36529b4..bbc8d429f8 100644
--- a/app/test-pmd/meson.build
+++ b/app/test-pmd/meson.build
@@ -34,47 +34,29 @@ sources = files(
if dpdk_conf.has('RTE_HAS_JANSSON')
ext_deps += jansson_dep
endif
-
-deps += ['ethdev', 'cmdline']
-if dpdk_conf.has('RTE_CRYPTO_SCHEDULER')
-deps += 'crypto_scheduler'
-endif
-if dpdk_conf.has('RTE_LIB_BITRATESTATS')
-deps += 'bitratestats'
-endif
if dpdk_conf.has('RTE_LIB_BPF')
sources += files('bpf_cmd.c')
-deps += 'bpf'
-endif
-if dpdk_conf.has('RTE_LIB_G
[PATCH v6 3/8] devtools: add script to flag unneeded dependencies
From: Bruce Richardson
While not a serious problem, DPDK components often list more
dependencies than are actually necessary to build, due to the use of
recursive dependencies. In extreme cases, such as with core libraries,
this can lead to longer configuration times due to meson having to
deduplicate long lists of dependencies. Therefore we can add a script to
identify when a component has got unnecessary dependencies listed.
Signed-off-by: Bruce Richardson
---
devtools/find-duplicate-deps.py | 62 +
1 file changed, 62 insertions(+)
create mode 100755 devtools/find-duplicate-deps.py
diff --git a/devtools/find-duplicate-deps.py b/devtools/find-duplicate-deps.py
new file mode 100755
index 00..cbd0ca7196
--- /dev/null
+++ b/devtools/find-duplicate-deps.py
@@ -0,0 +1,62 @@
+#! /usr/bin/env python3
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2024 Intel Corporation
+
+"""Identify any superfluous dependencies listed in DPDK deps graph."""
+
+import sys
+
+all_deps = {}
+
+
+class dep:
+"""Holds a component and its dependencies."""
+
+def __init__(self, name, dep_names):
+"""Create and process a component and its deps."""
+self.name = name.strip('" ')
+self.base_deps = [all_deps[dn.strip('" ')] for dn in dep_names]
+self.recursive_deps = []
+for d in self.base_deps:
+self.recursive_deps.extend(d.base_deps)
+self.recursive_deps.extend(d.recursive_deps)
+self.extra_deps = []
+for d in self.base_deps:
+if d in self.recursive_deps:
+self.extra_deps.append(d.name)
+if self.extra_deps:
+print(f'{self.name}: extra deps {self.extra_deps}')
+
+def dict_add(self, d):
+"""Add this object to a dictionary by name."""
+d[self.name] = self
+
+
+def remove_attrs(ln):
+"""Remove attributes from a line."""
+while ln.find("[") != -1:
+start = ln.find("[")
+end = ln.find("]")
+ln = ln[:start] + ln[end + 1 :]
+return ln.strip()
+
+
+def main(argv):
+"""Read the dependency tree from a dot file and process it."""
+if len(argv) != 2:
+print(f'Usage: {argv[0]} /deps.dot', file=sys.stderr)
+sys.exit(1)
+
+with open(argv[1]) as f:
+for ln in f.readlines():
+ln = remove_attrs(ln.strip())
+if '->' in ln:
+name, deps = ln.split('->')
+deps = deps.strip(' {}')
+dep(name, deps.split(',')).dict_add(all_deps)
+elif ln.startswith('"') and ln.endswith('"'):
+dep(ln.strip('"'), []).dict_add(all_deps)
+
+
+if __name__ == '__main__':
+main(sys.argv)
--
2.43.5
[PATCH v6 2/8] build: output a dependency log in build directory
From: Bruce Richardson
As meson processes our DPDK source tree it manages dependencies
specified by each individual driver. To enable future analysis of the
dependency links between components, log the dependencies of each DPDK
component as it gets processed. This could potentially allow other tools
to automatically enable or disable components based on the desired end
components to be built, e.g. if the user requests net/ice, ensure that
common/iavf is also enabled in the drivers.
The output file produced is in "dot" or "graphviz" format, which allows
producing a graphical representation of the DPDK dependency tree if so
desired. For example: "dot -Tpng -O build/deps.dot" to produce the
image file "build/deps.dot.png".
Signed-off-by: Bruce Richardson
Acked-by: Konstantin Ananyev
Signed-off-by: Anatoly Burakov
---
Notes:
v4 -> v5:
- Change output format to include display name
app/meson.build| 17 +++-
buildtools/log-deps.py | 94 ++
buildtools/meson.build | 2 +
drivers/meson.build| 14 +++
examples/meson.build | 18 +++-
lib/meson.build| 14 +++
6 files changed, 157 insertions(+), 2 deletions(-)
create mode 100644 buildtools/log-deps.py
diff --git a/app/meson.build b/app/meson.build
index 61202495bd..a971738d40 100644
--- a/app/meson.build
+++ b/app/meson.build
@@ -77,7 +77,22 @@ foreach app:apps
endif
if build
+# call into subdir may update name so record it here
+display_name = name
+
subdir(name)
+app_name = 'dpdk-' + name
+
+# log mandatory dependencies
+cmds = ['--type', 'app']
+cmds += ['--display-name', display_name]
+run_command([log_deps_cmd, cmds, app_name, deps], check: false)
+
+# log optional dependencies, if any
+if optional_deps.length() > 0
+cmds += ['--optional']
+run_command([log_deps_cmd, cmds, app_name, optional_deps], check:
false)
+endif
if not build and require_apps
error('Cannot build explicitly requested app "@0@".\n'.format(name)
+ '\tReason: ' + reason)
@@ -125,7 +140,7 @@ foreach app:apps
link_libs = dpdk_static_libraries + dpdk_drivers
endif
-exec = executable('dpdk-' + name,
+exec = executable(app_name,
[ sources, resources ],
c_args: cflags,
link_args: ldflags,
diff --git a/buildtools/log-deps.py b/buildtools/log-deps.py
new file mode 100644
index 00..7a99b129d7
--- /dev/null
+++ b/buildtools/log-deps.py
@@ -0,0 +1,94 @@
+#! /usr/bin/env python3
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2024 Intel Corporation
+
+"""Utility script to build up a list of dependencies from meson."""
+
+import os
+import sys
+import argparse
+import typing as T
+
+
+def file_to_list(filename: str) -> T.List[str]:
+"""Read file into a list of strings."""
+with open(filename, encoding="utf-8") as f:
+return f.readlines()
+
+
+def list_to_file(filename: str, lines: T.List[str]):
+"""Write a list of strings out to a file."""
+with open(filename, "w", encoding="utf-8") as f:
+f.writelines(lines)
+
+
+def gen_deps(
+component_type: str,
+optional: bool,
+component: str,
+display_name: T.Optional[str],
+deps: T.List[str],
+) -> str:
+"""Generate a dependency graph for meson."""
+dep_list_str = '", "'.join(deps)
+deps_str = "" if not deps else f' -> {{ "{dep_list_str}" }}'
+# we define custom attributes for the nodes
+attr_str = f'dpdk_componentType="{component_type}"'
+if optional:
+# we use a dotted line to represent optional dependencies
+attr_str += ',style="dotted"'
+if display_name is not None:
+attr_str += f',dpdk_displayName="{display_name}"'
+return f'"{component}"{deps_str} [{attr_str}]\n'
+
+
+def _main():
+depsfile = f'{os.environ["MESON_BUILD_ROOT"]}/deps.dot'
+
+# to reset the deps file on each build, the script is called without any
params
+if len(sys.argv) == 1:
+os.remove(depsfile)
+sys.exit(0)
+
+# we got arguments, parse them
+parser = argparse.ArgumentParser(
+description="Generate a dependency graph for meson."
+)
+# type is required
+parser.add_argument(
+"--type", required=True, help="Type of dependency (lib, examples,
etc.)"
+)
+parser.add_argument(
+"--optional", action="store_true", help="Whether the dependency is
optional"
+)
+parser.add_argument(
+"--display-name",
+help="Component name as it is used in the build system",
+)
+# component is required
+parser.add_argument("component", help="The component to add to the graph")
+parser.add_argument("deps", nargs="*", help="The dependencies of the
component")
+
+parsed = parser.parse_args()
+
+try:
+contents = file_to_list(depsfile)
+exce
[PATCH v6 0/8] Record and rework component dependencies
As part of the meson build, we can record the dependencies for each component as we process it, logging them to a file. This file can be used as input to a number of other scripts and tools, for example, to graph the dependencies, or to allow higher-level build-config tools to automatically enable component requirements, etc. The first patch of this set separates dependencies inside meson into optional or mandatory. The second patch of this set generates the basic dependency tree. The third patch does some processing of that dependency tree to identify cases where dependencies are being unnecessarily specified. Reducing these makes it easier to have readable dependency graphs in future, without affecting the build. The following 4 patches are based on the output of the second patch, and greatly cut down the number of direct dependency links between components. Even with the cut-down dependencies, the full dependency graph is nigh-unreadable, so the final patch adds a new script to generate dependency tree subgraphs, creating dot files for e.g. the dependencies of a particular component, or a component class such as mempool drivers. v6: - Fix build for common/qat - Rebase - Rename some patch headings to avoid check git log warnings v4 -> v5: - Improve optional dependency handling by making it (mostly) automatic - Fix CI issue from accidentally changing testpmd binary filename - Explicitly encode display name vs binary name information in the graph - Reduce some more dependencies - Update all scripts to handle new graph output format v3 -> v4: - Update to latest main v2 -> v3: - Split dependencies into optional and mandatory - Fixup graph scripts to read and generate graphs that encode optional dependencies into the graph - Python version fixes to avoid using features not available in minimum supported Python version - Formatting with Ruff, and PEP-484 compliance Anatoly Burakov (1): build: introduce optional internal dependencies Bruce Richardson (7): build: output a dependency log in build directory devtools: add script to flag unneeded dependencies drivers: remove kvargs from class dependencies lib: reduce library dependencies drivers: reduce driver dependencies app: reduce app dependencies devtools: add script to generate DPDK dependency graphs app/dumpcap/meson.build| 2 +- app/graph/meson.build | 2 +- app/meson.build| 31 ++- app/pdump/meson.build | 2 +- app/proc-info/meson.build | 6 +- app/test-bbdev/meson.build | 18 +- app/test-crypto-perf/meson.build | 6 +- app/test-fib/meson.build | 2 +- app/test-pmd/meson.build | 56 ++ app/test-sad/meson.build | 2 +- app/test/meson.build | 14 +- buildtools/log-deps.py | 94 + buildtools/meson.build | 2 + devtools/draw-dependency-graphs.py | 223 + devtools/find-duplicate-deps.py| 62 ++ drivers/baseband/fpga_5gnr_fec/meson.build | 2 +- drivers/baseband/fpga_lte_fec/meson.build | 2 +- drivers/baseband/la12xx/meson.build| 2 +- drivers/baseband/null/meson.build | 2 +- drivers/baseband/turbo_sw/meson.build | 2 +- drivers/bus/auxiliary/meson.build | 2 - drivers/bus/dpaa/meson.build | 2 +- drivers/bus/fslmc/meson.build | 2 +- drivers/bus/ifpga/meson.build | 2 +- drivers/bus/pci/meson.build| 4 +- drivers/bus/platform/meson.build | 1 - drivers/bus/uacce/meson.build | 2 - drivers/bus/vdev/meson.build | 2 - drivers/common/cnxk/meson.build| 4 +- drivers/common/cpt/meson.build | 2 +- drivers/common/mlx5/meson.build| 2 +- drivers/common/zsda/meson.build| 2 +- drivers/compress/mlx5/meson.build | 2 +- drivers/compress/nitrox/meson.build| 2 +- drivers/compress/octeontx/meson.build | 2 +- drivers/crypto/bcmfs/meson.build | 2 +- drivers/crypto/cnxk/meson.build| 2 +- drivers/crypto/dpaa_sec/meson.build| 2 +- drivers/crypto/ipsec_mb/meson.build| 2 +- drivers/crypto/mlx5/meson.build| 2 +- drivers/crypto/nitrox/meson.build | 2 +- drivers/dma/cnxk/meson.build | 2 +- drivers/dma/dpaa/meson.build | 2 +- drivers/dma/dpaa2/meson.build | 2 +- drivers/dma/odm/meson.build| 2 +- drivers/dma/skeleton/meson.build | 2 +- drivers/event/cnxk/meson.build | 2 +- drivers/event/dlb2/meson.build | 2 +- drivers/event/dpaa2/meson.build| 2 +- drivers/event/meson.build | 2 +- dr
[PATCH v6 4/8] drivers: remove kvargs from class dependencies
From: Bruce Richardson
The kvargs library is used by EAL, and therefore is implicitly a
dependency of every DPDK driver. Remove it from the minimum set of
dependencies for each driver class as it's unnecessary to call it out
there.
Signed-off-by: Bruce Richardson
---
drivers/event/meson.build | 2 +-
drivers/net/meson.build | 2 +-
drivers/regex/meson.build | 2 +-
drivers/vdpa/meson.build | 2 +-
4 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/event/meson.build b/drivers/event/meson.build
index d6706b57f7..2708833adf 100644
--- a/drivers/event/meson.build
+++ b/drivers/event/meson.build
@@ -19,4 +19,4 @@ if not (toolchain == 'gcc' and
cc.version().version_compare('<4.8.6') and
dpdk_conf.has('RTE_ARCH_ARM64'))
drivers += 'octeontx'
endif
-std_deps = ['eventdev', 'kvargs']
+std_deps = ['eventdev']
diff --git a/drivers/net/meson.build b/drivers/net/meson.build
index 460eb69e5b..246f99ec17 100644
--- a/drivers/net/meson.build
+++ b/drivers/net/meson.build
@@ -65,6 +65,6 @@ drivers = [
'xsc',
'zxdh',
]
-std_deps = ['ethdev', 'kvargs'] # 'ethdev' also pulls in mbuf, net, eal etc
+std_deps = ['ethdev'] # 'ethdev' also pulls in mbuf, net, eal etc
std_deps += ['bus_pci'] # very many PMDs depend on PCI, so make std
std_deps += ['bus_vdev']# same with vdev bus
diff --git a/drivers/regex/meson.build b/drivers/regex/meson.build
index ff2a8fea89..10192e7c77 100644
--- a/drivers/regex/meson.build
+++ b/drivers/regex/meson.build
@@ -5,4 +5,4 @@ drivers = [
'mlx5',
'cn9k',
]
-std_deps = ['ethdev', 'kvargs', 'regexdev'] # 'ethdev' also pulls in mbuf,
net, eal etc
+std_deps = ['ethdev', 'regexdev'] # 'ethdev' also pulls in mbuf, net, eal etc
diff --git a/drivers/vdpa/meson.build b/drivers/vdpa/meson.build
index 896e8e0304..e01c277b9e 100644
--- a/drivers/vdpa/meson.build
+++ b/drivers/vdpa/meson.build
@@ -11,5 +11,5 @@ drivers = [
'nfp',
'sfc',
]
-std_deps = ['bus_pci', 'kvargs']
+std_deps = ['bus_pci']
std_deps += ['vhost']
--
2.43.5
[PATCH v6 5/8] lib: reduce library dependencies
From: Bruce Richardson
Rather than having each library depend up on EAL + any extra libs, we
can take advantage of recursive dependency support in meson and
just assign the dependencies of each directory directly, rather than
appending to the array. For libraries which only depend upon EAL, keep
that as a default, but for libraries which depend upon even a single
extra lib, that EAL dependency is unnecessary.
Going further, we can identify using the find_duplicate_deps.py script
any unnecessary deps in each library's list, and remove them to slim the
dependency tree down.
Reducing number of dependencies means that meson takes less time
processing and deduplicating the dependency tree for each component, and
also shrinks the dependency graph for DPDK itself.
Signed-off-by: Bruce Richardson
---
lib/argparse/meson.build | 2 +-
lib/bbdev/meson.build| 2 +-
lib/bitratestats/meson.build | 2 +-
lib/bpf/meson.build | 2 +-
lib/cmdline/meson.build | 2 +-
lib/compressdev/meson.build | 2 +-
lib/cryptodev/meson.build| 2 +-
lib/dispatcher/meson.build | 2 +-
lib/distributor/meson.build | 2 +-
lib/dmadev/meson.build | 2 --
lib/eal/meson.build | 5 +
lib/efd/meson.build | 2 +-
lib/ethdev/meson.build | 2 +-
lib/eventdev/meson.build | 3 +--
lib/fib/meson.build | 4 +---
lib/gpudev/meson.build | 2 +-
lib/graph/meson.build| 2 +-
lib/gro/meson.build | 2 +-
lib/gso/meson.build | 2 +-
lib/hash/meson.build | 4 +---
lib/ip_frag/meson.build | 2 +-
lib/ipsec/meson.build| 2 +-
lib/kvargs/meson.build | 2 +-
lib/latencystats/meson.build | 2 +-
lib/lpm/meson.build | 4 +---
lib/mbuf/meson.build | 2 +-
lib/member/meson.build | 2 +-
lib/mempool/meson.build | 2 +-
lib/metrics/meson.build | 2 +-
lib/mldev/meson.build| 2 +-
lib/net/meson.build | 2 +-
lib/node/meson.build | 2 +-
lib/pcapng/meson.build | 2 +-
lib/pdcp/meson.build | 2 +-
lib/pdump/meson.build| 2 +-
lib/pipeline/meson.build | 2 +-
lib/port/meson.build | 2 +-
lib/power/meson.build| 3 +--
lib/rawdev/meson.build | 2 --
lib/rcu/meson.build | 2 +-
lib/regexdev/meson.build | 2 +-
lib/reorder/meson.build | 2 +-
lib/rib/meson.build | 2 +-
lib/ring/meson.build | 1 -
lib/sched/meson.build| 2 +-
lib/security/meson.build | 2 +-
lib/table/meson.build| 2 +-
lib/telemetry/meson.build| 2 +-
lib/vhost/meson.build| 2 +-
49 files changed, 46 insertions(+), 62 deletions(-)
diff --git a/lib/argparse/meson.build b/lib/argparse/meson.build
index b6a08ca049..96abc8766f 100644
--- a/lib/argparse/meson.build
+++ b/lib/argparse/meson.build
@@ -4,4 +4,4 @@
sources = files('rte_argparse.c')
headers = files('rte_argparse.h')
-deps += ['log']
+deps = ['log']
diff --git a/lib/bbdev/meson.build b/lib/bbdev/meson.build
index 07685e7578..2e68aa7873 100644
--- a/lib/bbdev/meson.build
+++ b/lib/bbdev/meson.build
@@ -11,4 +11,4 @@ sources = files('rte_bbdev.c')
headers = files('rte_bbdev.h',
'rte_bbdev_pmd.h',
'rte_bbdev_op.h')
-deps += ['mbuf']
+deps = ['mbuf']
diff --git a/lib/bitratestats/meson.build b/lib/bitratestats/meson.build
index ede7e0a579..8defcd53bf 100644
--- a/lib/bitratestats/meson.build
+++ b/lib/bitratestats/meson.build
@@ -3,4 +3,4 @@
sources = files('rte_bitrate.c')
headers = files('rte_bitrate.h')
-deps += ['ethdev', 'metrics']
+deps = ['metrics']
diff --git a/lib/bpf/meson.build b/lib/bpf/meson.build
index 28df7f469a..ba1d0cb1d0 100644
--- a/lib/bpf/meson.build
+++ b/lib/bpf/meson.build
@@ -33,7 +33,7 @@ headers = files('bpf_def.h',
'rte_bpf.h',
'rte_bpf_ethdev.h')
-deps += ['mbuf', 'net', 'ethdev']
+deps = ['ethdev']
dep = dependency('libelf', required: false, method: 'pkg-config')
if dep.found()
diff --git a/lib/cmdline/meson.build b/lib/cmdline/meson.build
index e38e05893a..b3b623156d 100644
--- a/lib/cmdline/meson.build
+++ b/lib/cmdline/meson.build
@@ -33,4 +33,4 @@ else
sources += files('cmdline_os_unix.c')
endif
-deps += ['net']
+deps = ['net']
diff --git a/lib/compressdev/meson.build b/lib/compressdev/meson.build
index c80295dc0d..4b86955baf 100644
--- a/lib/compressdev/meson.build
+++ b/lib/compressdev/meson.build
@@ -16,4 +16,4 @@ driver_sdk_headers = files(
'rte_compressdev_pmd.h',
'rte_compressdev_internal.h',
)
-deps += ['kvargs', 'mbuf']
+deps = ['mbuf']
diff --git a/lib/cryptodev/meson.build b/lib/cryptodev/meson.build
index 4734acf321..74e42ac700 100644
--- a/lib/cryptodev/meson.build
+++ b/lib/cryptodev/meson.build
@@ -20,4 +20,4 @@ driver_sdk_headers += files(
'cryptodev_pmd.h',
)
-deps += ['kvargs', 'mbuf', 'rcu', 'telemetry']
+deps = ['mbuf', 'rcu']
diff --git a/lib/dispatcher/meson.build b/lib/dispatcher/meson.build
index 5
[PATCH v6 8/8] devtools: add script to generate DPDK dependency graphs
From: Bruce Richardson
Rather than the single monolithic graph that would be output from the
deps.dot file in a build directory, we can post-process that to generate
simpler graphs for different tasks. This new "draw_dependency_graphs"
script takes the "deps.dot" as input and generates an output file that
has the nodes categorized, filtering them based off the requested node or
category. For example, use "--match net/ice" to show the dependency tree
from that driver, or "--match lib" to show just the library dependency
tree.
Signed-off-by: Bruce Richardson
Signed-off-by: Anatoly Burakov
---
devtools/draw-dependency-graphs.py | 223 +
1 file changed, 223 insertions(+)
create mode 100755 devtools/draw-dependency-graphs.py
diff --git a/devtools/draw-dependency-graphs.py
b/devtools/draw-dependency-graphs.py
new file mode 100755
index 00..4fb765498d
--- /dev/null
+++ b/devtools/draw-dependency-graphs.py
@@ -0,0 +1,223 @@
+#! /usr/bin/env python3
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2024 Intel Corporation
+
+import argparse
+import collections
+import sys
+import typing as T
+
+# typedef for dependency data types
+Deps = T.Set[str]
+DepData = T.Dict[str, T.Dict[str, T.Dict[bool, Deps]]]
+
+
+def parse_dep_line(line: str) -> T.Tuple[str, Deps, str, bool]:
+"""Parse digraph line into (component, {dependencies}, type, optional)."""
+# extract attributes first
+first, last = line.index("["), line.rindex("]")
+edge_str, attr_str = line[:first], line[first + 1 : last]
+# key=value, key=value, ...
+attrs = {
+key.strip('" '): value.strip('" ')
+for attr_kv in attr_str.split(",")
+for key, value in [attr_kv.strip().split("=", 1)]
+}
+# check if edge is defined as dotted line, meaning it's optional
+optional = "dotted" in attrs.get("style", "")
+try:
+component_type = attrs["dpdk_componentType"]
+except KeyError as _e:
+raise ValueError(f"Error: missing component type: {line}") from _e
+
+# now, extract component name and any of its dependencies
+deps: T.Set[str] = set()
+try:
+component, deps_str = edge_str.strip('" ').split("->", 1)
+component = component.strip().strip('" ')
+deps_str = deps_str.strip().strip("{}")
+deps = {d.strip('" ') for d in deps_str.split(",")}
+except ValueError as _e:
+component = edge_str.strip('" ')
+
+return component, deps, component_type, optional
+
+
+def gen_dep_line(component: str, deps: T.Set[str], optional: bool) -> str:
+"""Generate a dependency line for a component."""
+# we use dotted line to represent optional components
+attr_str = ' [style="dotted"]' if optional else ""
+dep_list_str = '", "'.join(deps)
+deps_str = "" if not deps else f' -> {{ "{dep_list_str}" }}'
+return f'"{component}"{deps_str}{attr_str}\n'
+
+
+def read_deps_list(lines: T.List[str]) -> DepData:
+"""Read a list of dependency lines into a dictionary."""
+deps_data: T.Dict[str, T.Any] = {}
+for ln in lines:
+if ln.startswith("digraph") or ln == "}":
+continue
+
+component, deps, component_type, optional = parse_dep_line(ln)
+
+# each component will have two sets of dependencies - required and
optional
+c_dict = deps_data.setdefault(component_type,
{}).setdefault(component, {})
+c_dict[optional] = deps
+return deps_data
+
+
+def create_classified_graph(deps_data: DepData) -> T.Iterator[str]:
+"""Create a graph of dependencies with components classified by type."""
+yield "digraph dpdk_dependencies {\n overlap=false\n model=subset\n"
+for n, deps_t in enumerate(deps_data.items()):
+component_type, component_dict = deps_t
+yield f' subgraph cluster_{n} {{\nlabel = "{component_type}"\n'
+for component, optional_d in component_dict.items():
+for optional, deps in optional_d.items():
+yield gen_dep_line(component, deps, optional)
+yield " }\n"
+yield "}\n"
+
+
+def parse_match(line: str, dep_data: DepData) -> T.List[str]:
+"""Extract list of components from a category string."""
+# if this is not a compound string, we have very few valid choices
+if "/" not in line:
+# is this a category?
+if line in dep_data:
+return list(dep_data[line].keys())
+# this isn't a category. maybe an app name?
+maybe_app_name = f"dpdk-{line}"
+if maybe_app_name in dep_data["app"]:
+return [maybe_app_name]
+if maybe_app_name in dep_data["examples"]:
+return [maybe_app_name]
+# this isn't an app name either, so just look for component with that
name
+for _, component_dict in dep_data.items():
+if line in component_dict:
+return [line]
+# nothing found still. one last try: maybe it's a driver? we have to
be ca
[PATCH v6 7/8] app: reduce app dependencies
From: Bruce Richardson
Remove any unnecessary dependencies from the app dependency lists.
This will give each app a near-minimum set of required dependencies.
Signed-off-by: Bruce Richardson
---
app/dumpcap/meson.build | 2 +-
app/graph/meson.build| 2 +-
app/pdump/meson.build| 2 +-
app/proc-info/meson.build| 2 +-
app/test-crypto-perf/meson.build | 2 +-
app/test-fib/meson.build | 2 +-
app/test-sad/meson.build | 2 +-
app/test/meson.build | 2 +-
8 files changed, 8 insertions(+), 8 deletions(-)
diff --git a/app/dumpcap/meson.build b/app/dumpcap/meson.build
index 69c016c780..6204cf051a 100644
--- a/app/dumpcap/meson.build
+++ b/app/dumpcap/meson.build
@@ -14,4 +14,4 @@ endif
ext_deps += pcap_dep
sources = files('main.c')
-deps += ['ethdev', 'pdump', 'pcapng', 'bpf']
+deps += ['pdump']
diff --git a/app/graph/meson.build b/app/graph/meson.build
index 344e4a418f..9c4cd080d9 100644
--- a/app/graph/meson.build
+++ b/app/graph/meson.build
@@ -9,7 +9,7 @@ if not build
subdir_done()
endif
-deps += ['graph', 'eal', 'lpm', 'ethdev', 'node', 'cmdline', 'net']
+deps += ['node', 'cmdline']
sources = files(
'cli.c',
'conn.c',
diff --git a/app/pdump/meson.build b/app/pdump/meson.build
index a4c7dd44e7..f84f43be51 100644
--- a/app/pdump/meson.build
+++ b/app/pdump/meson.build
@@ -8,6 +8,6 @@ if is_windows
endif
sources = files('main.c')
-deps += ['ethdev', 'kvargs', 'pdump']
+deps += ['pdump']
cflags += no_wvla_cflag
diff --git a/app/proc-info/meson.build b/app/proc-info/meson.build
index 51315f72a1..ee2a3f0317 100644
--- a/app/proc-info/meson.build
+++ b/app/proc-info/meson.build
@@ -8,7 +8,7 @@ if is_windows
endif
sources = files('main.c')
-deps += ['ethdev', 'security', 'eventdev']
+deps += ['security', 'eventdev']
optional_deps += 'metrics'
cflags += no_wvla_cflag
diff --git a/app/test-crypto-perf/meson.build b/app/test-crypto-perf/meson.build
index fb48d9ec29..f056431bf4 100644
--- a/app/test-crypto-perf/meson.build
+++ b/app/test-crypto-perf/meson.build
@@ -19,7 +19,7 @@ sources = files(
'cperf_test_verify.c',
'main.c',
)
-deps += ['cryptodev', 'net', 'security']
+deps += ['cryptodev']
optional_deps += 'crypto_scheduler'
cflags += no_wvla_cflag
diff --git a/app/test-fib/meson.build b/app/test-fib/meson.build
index eb36772cf3..25e2ea1a1d 100644
--- a/app/test-fib/meson.build
+++ b/app/test-fib/meson.build
@@ -8,4 +8,4 @@ if is_windows
endif
sources = files('main.c')
-deps += ['fib', 'lpm', 'net']
+deps += ['fib', 'lpm']
diff --git a/app/test-sad/meson.build b/app/test-sad/meson.build
index a1cbcc71d6..cd81638819 100644
--- a/app/test-sad/meson.build
+++ b/app/test-sad/meson.build
@@ -8,6 +8,6 @@ if is_windows
endif
sources = files('main.c')
-deps += ['ipsec', 'net']
+deps += ['ipsec']
cflags += no_wvla_cflag
diff --git a/app/test/meson.build b/app/test/meson.build
index eb0447e8c6..d480d39311 100644
--- a/app/test/meson.build
+++ b/app/test/meson.build
@@ -2,7 +2,7 @@
# Copyright(c) 2017-2023 Intel Corporation
# the main test files [test.c and commands.c] relies on these libraries
-deps += ['cmdline', 'ring', 'mempool', 'mbuf']
+deps += ['cmdline']
sources += files('commands.c', 'test.c')
# optional dependencies: some files may use these - and so we should link them
in -
--
2.43.5
[PATCH v6 6/8] drivers: reduce driver dependencies
From: Bruce Richardson
Remove any unnecessary dependencies from the driver dependency lists.
This will give each driver a near-minimum set of required dependencies.
Signed-off-by: Bruce Richardson
Reviewed-by: Rosen Xu
---
drivers/baseband/fpga_5gnr_fec/meson.build | 2 +-
drivers/baseband/fpga_lte_fec/meson.build | 2 +-
drivers/baseband/la12xx/meson.build| 2 +-
drivers/baseband/null/meson.build | 2 +-
drivers/baseband/turbo_sw/meson.build | 2 +-
drivers/bus/auxiliary/meson.build | 2 --
drivers/bus/dpaa/meson.build | 2 +-
drivers/bus/fslmc/meson.build | 2 +-
drivers/bus/ifpga/meson.build | 2 +-
drivers/bus/pci/meson.build| 4 +---
drivers/bus/platform/meson.build | 1 -
drivers/bus/uacce/meson.build | 2 --
drivers/bus/vdev/meson.build | 2 --
drivers/common/cnxk/meson.build| 4 ++--
drivers/common/cpt/meson.build | 2 +-
drivers/common/mlx5/meson.build| 2 +-
drivers/common/zsda/meson.build| 2 +-
drivers/compress/mlx5/meson.build | 2 +-
drivers/compress/nitrox/meson.build| 2 +-
drivers/compress/octeontx/meson.build | 2 +-
drivers/crypto/bcmfs/meson.build | 2 +-
drivers/crypto/cnxk/meson.build| 2 +-
drivers/crypto/dpaa_sec/meson.build| 2 +-
drivers/crypto/ipsec_mb/meson.build| 2 +-
drivers/crypto/mlx5/meson.build| 2 +-
drivers/crypto/nitrox/meson.build | 2 +-
drivers/dma/cnxk/meson.build | 2 +-
drivers/dma/dpaa/meson.build | 2 +-
drivers/dma/dpaa2/meson.build | 2 +-
drivers/dma/odm/meson.build| 2 +-
drivers/dma/skeleton/meson.build | 2 +-
drivers/event/cnxk/meson.build | 2 +-
drivers/event/dlb2/meson.build | 2 +-
drivers/event/dpaa2/meson.build| 2 +-
drivers/event/octeontx/meson.build | 3 +--
drivers/event/sw/meson.build | 2 +-
drivers/mempool/cnxk/meson.build | 2 +-
drivers/mempool/dpaa/meson.build | 2 +-
drivers/mempool/dpaa2/meson.build | 2 +-
drivers/mempool/octeontx/meson.build | 2 +-
drivers/ml/cnxk/meson.build| 2 +-
drivers/net/cnxk/meson.build | 4 +---
drivers/net/intel/iavf/meson.build | 2 +-
drivers/net/intel/ice/meson.build | 2 +-
drivers/net/mana/meson.build | 2 +-
drivers/net/mlx5/meson.build | 2 +-
drivers/net/sfc/meson.build| 2 +-
drivers/net/softnic/meson.build| 2 +-
drivers/net/virtio/meson.build | 2 +-
drivers/raw/cnxk_bphy/meson.build | 2 +-
drivers/raw/cnxk_gpio/meson.build | 2 +-
drivers/raw/gdtc/meson.build | 2 +-
drivers/raw/ntb/meson.build| 2 +-
drivers/raw/skeleton/meson.build | 2 +-
drivers/regex/cn9k/meson.build | 2 +-
drivers/regex/mlx5/meson.build | 2 +-
drivers/vdpa/ifc/meson.build | 2 +-
drivers/vdpa/meson.build | 3 +--
drivers/vdpa/mlx5/meson.build | 2 +-
drivers/vdpa/sfc/meson.build | 2 +-
60 files changed, 57 insertions(+), 70 deletions(-)
diff --git a/drivers/baseband/fpga_5gnr_fec/meson.build
b/drivers/baseband/fpga_5gnr_fec/meson.build
index c3678d23eb..31b9e92fbb 100644
--- a/drivers/baseband/fpga_5gnr_fec/meson.build
+++ b/drivers/baseband/fpga_5gnr_fec/meson.build
@@ -1,7 +1,7 @@
# SPDX-License-Identifier: BSD-3-Clause
# Copyright(c) 2020 Intel Corporation
-deps += ['bus_vdev', 'ring', 'pci', 'bus_pci']
+deps += ['bus_vdev', 'bus_pci']
sources = files('rte_fpga_5gnr_fec.c')
diff --git a/drivers/baseband/fpga_lte_fec/meson.build
b/drivers/baseband/fpga_lte_fec/meson.build
index 14e07826ef..fbf24755db 100644
--- a/drivers/baseband/fpga_lte_fec/meson.build
+++ b/drivers/baseband/fpga_lte_fec/meson.build
@@ -1,5 +1,5 @@
# SPDX-License-Identifier: BSD-3-Clause
# Copyright(c) 2019 Intel Corporation
-deps += ['bus_vdev', 'ring', 'pci', 'bus_pci']
+deps += ['bus_vdev', 'bus_pci']
sources = files('fpga_lte_fec.c')
diff --git a/drivers/baseband/la12xx/meson.build
b/drivers/baseband/la12xx/meson.build
index 7b7e41c961..e1dbdd0fa7 100644
--- a/drivers/baseband/la12xx/meson.build
+++ b/drivers/baseband/la12xx/meson.build
@@ -1,6 +1,6 @@
# SPDX-License-Identifier: BSD-3-Clause
# Copyright 2020-2021 NXP
-deps += ['bus_vdev', 'ring']
+deps += ['bus_vdev']
sources = files('bbdev_la12xx.c')
diff --git a/drivers/baseband/null/meson.build
b/drivers/baseband/null/meson.build
index 22863f0bd8..716d6c6fdb 100644
--- a/drivers/baseband/null/meson.build
+++ b/drivers/baseband/null/meson.build
@@ -1,5 +1,5 @@
# SPDX-License-Identifier: BSD-3-Clause
# Copyright(c) 2018 Luca Boccassi
-deps += ['bus_vdev', 'ring']
+deps += ['bus_vdev']
sources = file
RE: [PATCH 1/2] doc: fix mlx5 VXLAN matching limitations
> -Original Message-
> From: Minggang(Gavin) Li
> Sent: Tuesday, February 18, 2025 11:33
> To: Matan Azrad ; Slava Ovsiienko
> ; Ori Kam ; NBU-Contact-Thomas
> Monjalon (EXTERNAL) ; Dariusz Sosnowski
> ; Bing Zhao ; Suanming Mou
> ; Alex Vesker ; rongwei liu
>
> Cc: dev@dpdk.org; Raslan Darawsheh
> Subject: [PATCH 1/2] doc: fix mlx5 VXLAN matching limitations
>
> P bit of VXLAN-GPE and G bit of VXLAN-GBP should be set when matching the
> protocols.
>
> Fixes: 49765b78c81b ("net/mlx5/hws: match all VXLAN fields")
> Signed-off-by: Minggang Li(Gavin)
Acked-by: Dariusz Sosnowski
Best regards,
Dariusz Sosnowski
RE: [PATCH 2/2] app/testpmd: use flags from user if it is set
> -Original Message-
> From: Minggang Li(Gavin)
> Sent: Tuesday, February 18, 2025 11:33
> To: Matan Azrad ; Slava Ovsiienko
> ; Ori Kam ; NBU-Contact-Thomas
> Monjalon (EXTERNAL) ; Aman Singh
> ; Minggang(Gavin) Li
> Cc: dev@dpdk.org; Raslan Darawsheh
> Subject: [PATCH 2/2] app/testpmd: use flags from user if it is set
>
> External email: Use caution opening links or attachments
>
>
> It's user's responsibility to set correct flags, especially the G bit, to
> distinguish the
> protocols, VXLAN and VXLAN-GBP, sharing the same port 4789.
>
> It will be set to 0x8 if user does not specify the flags.
>
> Fixes: da118115d95c ("app/testpmd: support matching any VXLAN field")
> Signed-off-by: Minggang Li(Gavin)
Acked-by: Dariusz Sosnowski
Best regards,
Dariusz Sosnowski
<>
[PATCH] app/testpmd: fix fail to set queue number when attach device
Testpmd doesn't modify the Rx/Tx queue number of all ports after attach a
new device. The root cause is that "port config all rxq " would
check if all ports are stopped and the status of this port is still in
'RTE_PORT_HANDLING'. This status is changed to 'RTE_PORT_STOPPED' in
setup_attached_port() when start port.
In addition, this new port needs to be initialized in advance because these
information are depanded on many commands. Otherwise other commands like
"port config xxx" may also encounter similar issue. But the initialization
can't be in new event callback as discussed in following commit. So this
patch moves setup_attached_port() to the time point 100ms after received
the new event.
Bugzilla ID: 1663
Fixes: 994635edb2c0 ("app/testpmd: add port attach/detach for multiple process")
Signed-off-by: Huisong Li
---
app/test-pmd/testpmd.c | 13 -
1 file changed, 8 insertions(+), 5 deletions(-)
diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
index 19e2e37c11..b0ee39db52 100644
--- a/app/test-pmd/testpmd.c
+++ b/app/test-pmd/testpmd.c
@@ -705,7 +705,7 @@ eth_dev_set_mtu_mp(uint16_t port_id, uint16_t mtu)
}
/* Forward function declarations */
-static void setup_attached_port(portid_t pi);
+static void setup_attached_port(void *arg);
static void check_all_ports_link_status(uint32_t port_mask);
static int eth_event_callback(portid_t port_id,
enum rte_eth_event_type type,
@@ -2906,8 +2906,6 @@ start_port(portid_t pid)
at_least_one_port_exist = true;
port = &ports[pi];
- if (port->need_setup)
- setup_attached_port(pi);
if (port->port_status == RTE_PORT_STOPPED) {
port->port_status = RTE_PORT_HANDLING;
@@ -3441,7 +3439,7 @@ attach_port(char *identifier)
/* setup ports matching the devargs used for probing */
if (port_is_forwarding(pi))
continue; /* port was already attached before */
- setup_attached_port(pi);
+ setup_attached_port((void *)(intptr_t)pi);
}
out:
printf("Port %s is attached.\n", identifier);
@@ -3449,8 +3447,9 @@ attach_port(char *identifier)
}
static void
-setup_attached_port(portid_t pi)
+setup_attached_port(void *arg)
{
+ portid_t pi = (intptr_t)arg;
unsigned int socket_id;
int ret;
@@ -3786,6 +3785,10 @@ eth_event_callback(portid_t port_id, enum
rte_eth_event_type type, void *param,
ports[port_id].need_setup = 1;
ports[port_id].port_status = RTE_PORT_HANDLING;
}
+ /* Can't initialize port directly in new event. */
+ if (rte_eal_alarm_set(10, setup_attached_port,
+ (void *)(intptr_t)port_id))
+ fprintf(stderr, "Could not set up deferred task to
setup this attached port.\n");
break;
case RTE_ETH_EVENT_INTR_RMV:
if (port_id_is_invalid(port_id, DISABLED_WARN))
--
2.22.0
[PATCH 0/2] use flags from user if it is set
This patch series also fixed mlx5 release notes on VXLAN limitations. Minggang Li (Gavin) (2): doc: fix mlx5 VXLAN matching limitations app/testpmd: use flags from user if it is set app/test-pmd/cmdline_flow.c | 3 ++- doc/guides/nics/mlx5.rst| 4 2 files changed, 6 insertions(+), 1 deletion(-) -- 2.34.1
RE: [PATCH v3 2/2] app/test: add SM4 GCM tests
Hi Pablo,
> -Original Message-
> From: Dooley, Brian
> Sent: Monday, January 13, 2025 6:10 PM
> Cc: dev@dpdk.org; gak...@marvell.com; Ji, Kai ; De Lara
> Guarch, Pablo ; Dooley, Brian
>
> Subject: [PATCH v3 2/2] app/test: add SM4 GCM tests
>
> Added SM4-GCM tests for the AESNI MB PMD.
>
> Signed-off-by: Brian Dooley
> ---
> app/test/test_cryptodev.c | 158 +
> app/test/test_cryptodev_aead_test_vectors.h | 708
>
> 2 files changed, 866 insertions(+)
>
Just one comment below.
Thanks,
Pablo
...
> b/app/test/test_cryptodev_aead_test_vectors.h
> index 73bfb8dad4..aacefe5ea4 100644
> --- a/app/test/test_cryptodev_aead_test_vectors.h
> +++ b/app/test/test_cryptodev_aead_test_vectors.h
> @@ -50,6 +50,54 @@ static uint8_t ccm_aad_test_2[22] = {
> 0xA5, 0xB8, 0xFC, 0xBA, 0x00, 0x00
> };
>
> +static uint8_t sm4_gcm_aad_test_2[MAX_AAD_LENGTH] = {
> + 0x3f, 0x89, 0x42, 0x20
> +};
MAX_AAD_LENGTH is pretty much 64KB. This is allocating too much memory
unnecessarily, so I would limit to the actual size .
You could use [] with no defined size, and that should work.
> +
...
RE: [PATCH] net/mlx5: fix flow null ops seg fault
> -Original Message-
> From: Kevin Traynor
> Sent: Friday, February 14, 2025 11:06
> To: dev@dpdk.org
> Cc: Dariusz Sosnowski ; Maayan Kashani
> ; Raslan Darawsheh ; Ori Kam
> ; Kevin Traynor ; sta...@dpdk.org
> Subject: [PATCH] net/mlx5: fix flow null ops seg fault
>
> External email: Use caution opening links or attachments
>
>
> flow null driver ops may get selected and try to call .list_create causing a
> seg fault.
>
> Full details and back traces in bugzilla.
>
> Add a null ops implementation for .list_create.
>
> Bugzilla ID: 1661
> Fixes: e38776c36c8a ("net/mlx5: introduce HWS for non-template flow API")
> Cc: sta...@dpdk.org
>
> Signed-off-by: Kevin Traynor
For segfault fix:
Acked-by: Dariusz Sosnowski
Regarding why null ops where selected here, let's continue on Bugzilla.
Best regards,
Dariusz Sosnowski
Re: [RFC v2 08/10] net/virtio: remove weak symbols
Hi David, On 2/9/25 6:28 PM, David Marchand wrote: Rather than use weak symbols, expose stubs symbols when needed. Signed-off-by: David Marchand --- drivers/net/virtio/meson.build | 5 + drivers/net/virtio/virtio_rxtx.c| 6 -- drivers/net/virtio/virtio_rxtx_simple.c | 4 +++- 3 files changed, 12 insertions(+), 3 deletions(-) Reviewed-by: Maxime Coquelin Thanks, Maxime
RE: [PATCH] event/dlb2: fix event weight handling in SSE code path
> -Original Message-
> From: Pathak, Pravin
> Sent: Thursday, February 6, 2025 8:20 PM
> To: Richardson, Bruce ; dev@dpdk.org
> Cc: sta...@dpdk.org
> Subject: [EXTERNAL] RE: [PATCH] event/dlb2: fix event weight handling in SSE
> code path
>
> Reviewed-by: Pravin Pathak > -Original
> Message- > From: Richardson, Bruce > Sent:
> Tuesday, January 21, 2025 12: 12 PM > To: dev@ dpdk. org > Cc: Pathak, Pravin
> Reviewed-by: Pravin Pathak
Applied to dpdk-next-eventdev/for-main. Thanks
>
> > -Original Message-
> > From: Richardson, Bruce
> > Sent: Tuesday, January 21, 2025 12:12 PM
> > To: dev@dpdk.org
> > Cc: Pathak, Pravin ; Richardson, Bruce
> > ; sta...@dpdk.org
> > Subject: [PATCH] event/dlb2: fix event weight handling in SSE code
> > path
> >
> > Commit 6e2e98d6775b ("event/dlb2: support independent enqueue")
> > updated the avx512 code path with support for the event, or "qe" weights.
> > However, the SSE code path looks to have been missed in the update.
> > Update the SSE "build_hcws" function to match the changes made to the
> > AVX-512 version.
> >
> > Bugzilla ID: 1616
> > Fixes: 6e2e98d6775b ("event/dlb2: support independent enqueue")
> > Cc: sta...@dpdk.org
> >
> > Signed-off-by: Bruce Richardson
> > ---
> > drivers/event/dlb2/dlb2_sse.c | 11 ++-
> > 1 file changed, 6 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/event/dlb2/dlb2_sse.c
> > b/drivers/event/dlb2/dlb2_sse.c index 248d7519d5..f2e1f9fb7e 100644
> > --- a/drivers/event/dlb2/dlb2_sse.c
> > +++ b/drivers/event/dlb2/dlb2_sse.c
> > @@ -190,11 +190,11 @@ dlb2_event_build_hcws(struct dlb2_port
> *qm_port,
> > qe[3].data = ev[3].u64;
> >
> > /* will only be set for DLB 2.5 + */
> > - if (qm_port->cq_weight) {
> > - qe[0].weight = ev[0].impl_opaque & 3;
> > - qe[1].weight = ev[1].impl_opaque & 3;
> > - qe[2].weight = ev[2].impl_opaque & 3;
> > - qe[3].weight = ev[3].impl_opaque & 3;
> > + if (qm_port->dlb2->enable_cq_weight) {
> > + qe[0].weight =
> > RTE_PMD_DLB2_GET_QE_WEIGHT(&ev[0]);
> > + qe[1].weight =
> > RTE_PMD_DLB2_GET_QE_WEIGHT(&ev[1]);
> > + qe[2].weight =
> > RTE_PMD_DLB2_GET_QE_WEIGHT(&ev[2]);
> > + qe[3].weight =
> > RTE_PMD_DLB2_GET_QE_WEIGHT(&ev[3]);
> > }
> >
> > break;
> > @@ -205,6 +205,7 @@ dlb2_event_build_hcws(struct dlb2_port *qm_port,
> > qe[i].cmd_byte =
> > cmd_byte_map[qm_port-
> > >is_directed][ev[i].op];
> > qe[i].sched_type = sched_type[i];
> > + qe[i].weight =
> > RTE_PMD_DLB2_GET_QE_WEIGHT(&ev[i]);
> > qe[i].data = ev[i].u64;
> > qe[i].qid = queue_id[i];
> > qe[i].priority = EV_TO_DLB2_PRIO(ev[i].priority);
> > --
> > 2.43.0
Re: [RFC 00/11] Fix dev_ops dereferences
On Mon, Feb 17, 2025 at 09:24:04PM +0100, Morten Brørup wrote: > > From: Stephen Hemminger [mailto:step...@networkplumber.org] > > Sent: Monday, 17 February 2025 19.54 > > > > A common pattern in DPDK internals is to have an array of function > > pointers. When using these pointers an unnecessary derefence is done; > > this is allowed in C because of a historical quirk but unnecessary. > > > > The PVS Studio checker flags these as: > > https://pvs-studio.com/en/docs/warnings/v516/ > > > > This leads to a large number of warnings marked as High severity. > > This patchset simplifies the source and gets rid of these warnings. > > > > It should have no impact on the resulting binary. > > +1 to this RFC. > The code becomes much cleaner to read. > Agreed. Series-acked-by: Bruce Richardson
Re: [PATCH v1 1/1] usertools/devbind: add documentation for no-IOMMU mode
On 02/12/2024 10:35, Anatoly Burakov wrote: Support for `--noiommu-mode` flag was added, but documentation for it was not. Add documentation for the flag. Signed-off-by: Anatoly Burakov --- doc/guides/tools/devbind.rst | 5 + 1 file changed, 5 insertions(+) diff --git a/doc/guides/tools/devbind.rst b/doc/guides/tools/devbind.rst index 841615570f..e010d4b7df 100644 --- a/doc/guides/tools/devbind.rst +++ b/doc/guides/tools/devbind.rst @@ -62,6 +62,11 @@ OPTIONS Use this flag to change ownership to the specified user and group, so that devices bound to VFIO would be usable by unprivileged users. +* ``--noiommu-mode`` + + When using vfio-pci driver on a system with no IOMMU, this flag should be used to + enable unsafe no-IOMMU mode for vfio-pci driver. + .. warning:: Ping. -- Thanks, Anatoly
Re: [PATCH v3 0/2] Rewrite devbind
On 04/12/2024 10:47, Anatoly Burakov wrote: It has been suggested [1] that a major cleanup/rewrite of devbind would be beneficial in terms of long term maintainability of the code. I was in a coding mood over the weekend, and so I've went ahead and rewritten devbind. Note that this is one giant patch, rather than a series of patches adjusting existing code. Making it a patch series is possible, however the internal code architecture diverges quite significantly from the original devbind script due to its copious usage of string operations/pattern matching and global variables, so it is unclear whether subdividing this patch would be worth the effort. Instead, as has been suggested [2], the patchset now consists of creating a new file, followed by a removal of old file and rename of the new file. It is expected that this will be squashed on apply. The script has become slightly bigger - 1000 lines instead of 800, however I would argue that since most of that increase is infrastructure, comments, and sacrificing code golf for code readability (such as expanding one-liners into multiple lines), the trade-off between being able to read and reason about what happens in the script is worth the added line count. [1] https://patches.dpdk.org/project/dpdk/patch/c2bf00195c2d43833a831a9cc9346b4606d6ea2e.1723810613.git.anatoly.bura...@intel.com/ [2] https://patches.dpdk.org/project/dpdk/cover/cover.1733151400.git.anatoly.bura...@intel.com/ Anatoly Burakov (2): usertools/devbind: update coding style usertools/devbind: replace devbind doc/guides/tools/devbind.rst | 11 + usertools/dpdk-devbind.py| 1683 ++ 2 files changed, 924 insertions(+), 770 deletions(-) Ping. -- Thanks, Anatoly
[PATCH v2 1/3] eventdev/eth_rx: add API to burst add queues to Rx adapter
This patch introduces a new API, rte_event_eth_rx_adapter_queues_add(),
to allow bulk addition of multiple Rx queues in the eventdev Rx adapter.
The existing rte_event_eth_rx_adapter_queue_add() API supports adding
multiple queues by specifying rx_queue_id = -1, but it lacks the ability
to apply specific configurations to each of the added queues.
A new internal PMD operation, eventdev_eth_rx_adapter_queues_add_t, has
been introduced to enable this functionality. It takes an array of
receive queue IDs along with their corresponding queue configurations.
Signed-off-by: Shijith Thotton
---
.../eventdev/event_ethernet_rx_adapter.rst| 60 +--
lib/eventdev/eventdev_pmd.h | 34
lib/eventdev/eventdev_trace.h | 14 ++
lib/eventdev/eventdev_trace_points.c | 3 +
lib/eventdev/rte_event_eth_rx_adapter.c | 155 ++
lib/eventdev/rte_event_eth_rx_adapter.h | 33
lib/eventdev/version.map | 3 +
7 files changed, 292 insertions(+), 10 deletions(-)
diff --git a/doc/guides/prog_guide/eventdev/event_ethernet_rx_adapter.rst
b/doc/guides/prog_guide/eventdev/event_ethernet_rx_adapter.rst
index 2e68cca798..bae46cc7d7 100644
--- a/doc/guides/prog_guide/eventdev/event_ethernet_rx_adapter.rst
+++ b/doc/guides/prog_guide/eventdev/event_ethernet_rx_adapter.rst
@@ -96,16 +96,23 @@ when the adapter is created using the above-mentioned APIs.
Adding Rx Queues to the Adapter Instance
-Ethdev Rx queues are added to the instance using the
-``rte_event_eth_rx_adapter_queue_add()`` function. Configuration for the Rx
-queue is passed in using a ``struct rte_event_eth_rx_adapter_queue_conf``
-parameter. Event information for packets from this Rx queue is encoded in the
-``ev`` field of ``struct rte_event_eth_rx_adapter_queue_conf``. The
-servicing_weight member of the struct rte_event_eth_rx_adapter_queue_conf
-is the relative polling frequency of the Rx queue and is applicable when the
-adapter uses a service core function. The applications can configure queue
-event buffer size in ``struct
rte_event_eth_rx_adapter_queue_conf::event_buf_size``
-parameter.
+Ethdev Rx queues can be added to the instance using either the
+``rte_event_eth_rx_adapter_queue_add()`` function or
+``rte_event_eth_rx_adapter_queues_add()``. The former is used to add a single
Rx
+queue at a time, while the latter allows adding multiple Rx queues in a single
+call.
+
+Single Queue Addition
+^
+
+The ``rte_event_eth_rx_adapter_queue_add()`` API allows adding a single Rx
queue
+to the adapter instance. Configuration for the Rx queue is passed using a
+``struct rte_event_eth_rx_adapter_queue_conf`` parameter. Event information for
+packets from this Rx queue is encoded in the ``ev`` field of this struct. The
+``servicing_weight`` member of the struct determines the relative polling
+frequency of the Rx queue and is applicable when the adapter uses a service
core
+function. Applications can also configure the queue event buffer size using the
+``event_buf_size`` parameter in ``struct rte_event_eth_rx_adapter_queue_conf``.
.. code-block:: c
@@ -122,6 +129,39 @@ parameter.
eth_dev_id,
0, &queue_config);
+Bulk Queue Addition
+^^^
+
+The ``rte_event_eth_rx_adapter_queues_add()`` API allows the addition of
+multiple Rx queues in a single call. While
+``rte_event_eth_rx_adapter_queue_add()`` supports adding multiple queues by
+specifying ``rx_queue_id = -1``, it does not allow applying specific
+configurations to each queue individually. The
+``rte_event_eth_rx_adapter_queues_add()`` API accepts an array of receive queue
+IDs along with their corresponding configurations, enabling control over each
Rx
+queue's settings.
+
+.. code-block:: c
+
+struct rte_event_eth_rx_adapter_queue_conf queue_config[nb_rx_queues];
+int rx_queue_id[nb_rx_queues];
+
+for (int i = 0; i < nb_rx_queues; i++) {
+rx_queue_id[i] = i;
+queue_config[i].rx_queue_flags = 0;
+queue_config[i].ev.queue_id = i;
+queue_config[i].ev.sched_type = RTE_SCHED_TYPE_ATOMIC;
+queue_config[i].ev.priority = 0;
+queue_config[i].servicing_weight = 1;
+queue_config[i].event_buf_size = 1024;
+}
+
+err = rte_event_eth_rx_adapter_queues_add(id,
+ eth_dev_id,
+ rx_queue_id,
+ queue_config,
+ nb_rx_queues);
+
Querying Adapter Capabilities
~
diff --git a/lib/eventdev/eventdev_pmd.h b/lib/eventdev/eventdev_pmd.h
index 36148f8d86..ad13ba5b03 100644
--- a/lib/eventdev/eventdev_pmd.h
+++
