Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
lib/rcu/rte_rcu_qsbr.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/rcu/rte_r
Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
lib/eal/include/generic/rte_rwlock.h | 8
lib/eal/ppc/include/rte_atomic.h | 16 --
Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
lib/mbuf/rte_mbuf.h | 12 ++--
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/lib
Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
lib/ipsec/ipsec_sqn.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/ipsec/ipse
Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
drivers/net/mlx5/linux/mlx5_verbs.c | 2 +-
drivers/net/mlx5/mlx5.c | 4 ++--
drivers/net/
Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
lib/bbdev/rte_bbdev.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/lib/bbdev/
Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
drivers/net/af_xdp/rte_eth_af_xdp.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --gi
Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
drivers/net/cxgbe/clip_tbl.c | 2 +-
drivers/net/cxgbe/mps_tcam.c | 2 +-
2 files changed, 2 insertions(
Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
drivers/net/octeontx/octeontx_ethdev.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git
Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
drivers/raw/ifpga/ifpga_rawdev.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/dri
Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
drivers/net/iavf/iavf_vchnl.c | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git
Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
drivers/net/cnxk/cn10k_tx.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/driv
Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
drivers/common/mlx5/linux/mlx5_nl.c | 2 +-
drivers/common/mlx5/mlx5_common_mr.c| 8
dr
Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
drivers/event/cnxk/cnxk_tim_worker.h | 2 +-
drivers/event/dsw/dsw_event.c | 4 ++--
drivers/ev
Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
app/test/test_ring_perf.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/app/test/t
This series replaces uses of __atomic_{add,and,or,sub,xor}_fetch with
__atomic_fetch_{add,and,or,sub,xor} intrinsics where the return value
is used.
This series is being separated from the other similar series in
an effort to reduce the chance of mistakes being spotted in review
since the usages i
Use __atomic_fetch_{add,and,or,sub,xor} instead of
__atomic_{add,and,or,sub,xor}_fetch adding the necessary code to
allow consumption of the resulting value.
Signed-off-by: Tyler Retzlaff
---
drivers/common/cnxk/roc_ae.c| 2 +-
drivers/common/cnxk/roc_ae_fpm_tables.c | 2 +-
drivers/
On 3/10/2023 4:41 AM, Kaiwen Deng wrote:
> The bond port is failed to forward packets in Link Aggregation
> 802.3AD(4) mode. Because bond devices get LACP packets from
> rx_ring instead of tx_ring.
>
> This commit will get LACP packets from tx_ring instead.
>
> Fixes: 112891cd27e5 ("net/bonding:
On 3/10/2023 9:35 AM, Dongdong Liu wrote:
> This patchset is to do some code refactor for hns3 RSS.
> 1. [PATCH 01/16]-[PATCH 02/16] fix some hns3 RSS bug.
> 2. [PATCH 03/16]-[PATCH 14/16] refactoring hns3 RSS code.
> 3. [PATCH 15/16]-[PATCH 16/16] reimplement hash flow function.
>
>
> Huisong Li
Hi Team,
Would like to know if dpdk has a select(fd)/epoll(fd) like mechanism to
wake-up the application (say rx-pthread) asynchronously upon packet
reception without enabling rxq interrupts?
I can't afford to lose a cpu-core polling continuously to check for packets
using rte_eth_rx_burst.And I
On Fri, Mar 10, 2023 at 10:07:17AM +0100, David Marchand wrote:
> Caught by UNH lab, Windows compilation is broken because of symbol
> exports:
> rte_mldev_exports.def : error LNK2001: unresolved external symbol
> rte_ml_io_bfloat16_to_float32
> rte_mldev_exports.def : error LNK2001: unresolv
On 3/10/2023 9:35 AM, Dongdong Liu wrote:
> From: Huisong Li
>
> Currently, the setting of hash algorithm comes from the
> default configuration in driver and the rte_flow interface.
> The hash algorithm that is set to hardware in both ways is
> saved in hw->rss_info.conf.func.
>
> But the 'func
On 3/10/2023 9:35 AM, Dongdong Liu wrote:
> From: Huisong Li
>
> The hash key length of hns3 driver is obtained from firmware. If the
> length isn't a multiple of HNS3_RSS_HASH_KEY_NUM (16), the last part
> of hash key will be truncated.
>
I am not sure if the explanation is correct, according
On 3/10/2023 9:35 AM, Dongdong Liu wrote:
> From: Huisong Li
>
> The size of the redirection table is obtained from firmware. If the size
> isn't a multiple of HNS3_RSS_CFG_TBL_SIZE, the redirection table from
> user will be truncated.
>
> Fixes: c37ca66f2b27 ("net/hns3: support RSS")
> Cc: sta.
On Fri, 10 Mar 2023 18:18:36 +
Bruce Richardson wrote:
> On Alpine linux, the telemetry_data_autotest was failing for the
> test where we had dictionaries embedded in other dictionaries up
> to three levels deep. Indications are that this issue is due to
> excess data being stored on the stac
On Alpine linux, the telemetry_data_autotest was failing for the
test where we had dictionaries embedded in other dictionaries up
to three levels deep. Indications are that this issue is due to
excess data being stored on the stack, so replace stack-allocated
buffer data with dynamically allocated
21/02/2023 08:49, Gerry Gribbon:
> We were writing a value that should represent the number of items to be
> processed by hardware. The value being written was off by 1 (N*4)+3;
> The value should be (N*4) + 4 simplified to (N+1)*4
>
> Fixes: 5dfa003db53f ("common/mlx5: fix post doorbell barrier")
> >
> > +/* Note: Port_queue_id in xstats APIs is 8 bits, so we have a
> > +maximum of
> > + * 256 ports and queues for event_Dev */ #define MAX_PORTS_QUEUES
> > +256
> > +
>
> >There is a goal of removing the per-queue stats part in generic stats.
> >And in the process removing that limit
21/02/2023 08:47, Gerry Gribbon:
> Fix overflow of free QP mask.
> Regex used 64 QPs and used a bitmask to select a free QP for use.
> The bitmask in use was only 32 bits so did not allow half of the QPs
> to be utilised.
> Upgraded to 64 bit mask and using ffsll now instead of ffs.
>
> Fixes: 270
On Fri, 10 Mar 2023 17:35:31 +
"Sevincer, Abdullah" wrote:
> >
> > +/* Note: Port_queue_id in xstats APIs is 8 bits, so we have a maximum
> > +of
> > + * 256 ports and queues for event_Dev */ #define MAX_PORTS_QUEUES
> > +256
> > +
>
> >There is a goal of removing the per-queue stats
28/02/2023 15:50, Bruce Richardson:
> In case there are issues when configuring a device instance using the
> dpdk_idxd_cfg.py script, add a "verbose" mode option to it. This allows
> the user to see what changes are being made, and therefore which
> operation, if any, within the config fails.
>
>
From: Stephen Hemminger
Sent: Thursday, March 9, 2023 12:32 PM
To: Sevincer, Abdullah
Cc: dev@dpdk.org; jer...@marvell.com; Pattan, Reshma
Subject: Re: [PATCH v7] app/procinfo: display eventdev xstats
On Thu, 9 Mar 2023 12:27:58 -0600
Abdullah Sevincer wrote:
>
> +/* Note: Port_queue_id
https://bugs.dpdk.org/show_bug.cgi?id=1178
Bug ID: 1178
Summary: Deadlock with heap lock
Product: DPDK
Version: unspecified
Hardware: All
OS: All
Status: UNCONFIRMED
Severity: normal
Priority: No
Once allocated, a switch port list entry always stays there,
even after unplugging the ethdev that created it. Currently,
the entry's ethdev ID is not cleared on unplug. Referencing
the ethdev ID of a detached representor from a flow rule is
going to succeed, which is a bug. Also, if the user unplu
09/03/2023 14:58, Yogesh Jangra:
> The incorrect variable for the number of groups was used, so in the
> case of values not power of 2 the incorrect result was produced.
>
> Fixes: f7598a62d11 (table: support selector table)
Please use "git fixline" with this in your .gitconfig:
[alias]
The driver supports representor functionality. In it,
packets coming from VFs to the dedicated back-end Rx
queue get demultiplexed into front-end Rx queues of
representor ethdevs as per the per-packet metadata
indicating logical HW ingress ports. On transmit,
packets are provided with symmetrical m
02/03/2023 08:08, Yogesh Jangra:
> Currently, the rss object is configured after action, which leads
> to rss instruction failure, when rss instruction is called from
> action block. To resolve the error, rss object is configured
> before action and apply block.
>
> Fixes: 8ba342ce6f0 (pipeline: a
09/03/2023 14:51, Devendra Singh Rawat:
> Add new maintainer and remove old one from list for qede PMD.
>
> Signed-off-by: Devendra Singh Rawat
> ---
> MAINTAINERS | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/MAINTAINERS b/MAINTAINERS
> index 878204c93b..bdd5c311bc 1
01/03/2023 20:59, Tomasz Duszynski:
> Platform bus is a software bus under Linux that manages devices which
> generally do not have built-in discovery mechanisms. Linux normally
> learns about platform devices directly from device-tree during
> boot-up phase.
>
> Up to this point, whenever some us
On Fri, 10 Mar 2023 13:58:52 +
Ferruh Yigit wrote:
> >>
> >> Why not fix the misbehaving drivers, instead of working around for
> >> softnic, as Stephen suggested?
> >>
> >> Is there a list of problematic drivers?
> >>
> >
> > Ferruh, I think this is not a reasonable request. We don't have
From: Arkadiusz Kusztal
Individual variables are needed for tracking the remaining data for
compression and decompression.
Fixes: 83cc3b90ad7a ("app/compress-perf: fix testing single operation")
Cc: michae...@nvidia.com
Signed-off-by: Arkadiusz Kusztal
Signed-off-by: Ciara Power
---
app/test
On 3/10/2023 2:22 PM, Andrew Rybchenko wrote:
> On 2/23/23 06:10, Stephen Hemminger wrote:
>> The static LOGTYPE PMD will go away in the future.
>> Replace its use in the sfc driver with the same static
>> type as is already used in RTE_LOG_REGISTER macro.
>>
>> These drivers are trying to do thing
On 3/10/2023 2:20 PM, Andrew Rybchenko wrote:
> On 3/9/23 07:28, Ivan Malov wrote:
>> Previously, DPDK sfc driver received support for tunnel
>> offload. In it, MAE needs to set intermediate mark from
>> an outer rule (OR) recirculation ID in order to help
>> the driver identify packets that hit th
10/03/2023 09:19, Srikanth Yalavarthi:
> Marvell ML CNXK Driver
> --
>
> This patch series implements common Machine Learning (ML) ROC code
> and driver for Marvell Octeon 10 (cnxk) platform. ML inferencing is
> supported on cnxk platform through an integrated ML inferencing
>
Hey Kai,
> -Original Message-
> From: Kai Ji
> Sent: Wednesday 8 March 2023 14:48
> To: dev@dpdk.org
> Cc: gak...@marvell.com; Ji, Kai
> Subject: [dpdk-dev v1] crypto/openssl: fix of ASAN heap-use-after-free
>
> fix of ASAN report on heap-use-after-free error on tmp buffer.
>
> Fixes:
10/03/2023 15:36, Dumitrescu, Cristian:
> From: David Marchand
> > On Fri, Mar 10, 2023 at 2:48 PM Dumitrescu, Cristian
> > wrote:
> > > > Why not fix the misbehaving drivers, instead of working around for
> > > > softnic, as Stephen suggested?
> > > >
> > > > Is there a list of problematic drive
On 3/10/2023 2:18 PM, Andrew Rybchenko wrote:
> On 3/9/23 07:11, Ivan Malov wrote:
>> The driver supports representor functionality. In it,
>> packets coming from VFs to the dedicated back-end Rx
>> queue get demultiplexed into front-end Rx queues of
>> representor ethdevs as per the per-packet met
On Fri, Mar 10, 2023 at 3:37 PM Dumitrescu, Cristian
wrote:
>
>
>
> > -Original Message-
> > From: David Marchand
> > Sent: Friday, March 10, 2023 1:49 PM
> > To: Dumitrescu, Cristian
> > Cc: Ferruh Yigit ; Stephen Hemminger
> > ; Jangra, Yogesh ;
> > Singh, Aman Deep ; Zhang, Yuying
> >
> -Original Message-
> From: David Marchand
> Sent: Friday, March 10, 2023 1:49 PM
> To: Dumitrescu, Cristian
> Cc: Ferruh Yigit ; Stephen Hemminger
> ; Jangra, Yogesh ;
> Singh, Aman Deep ; Zhang, Yuying
> ; dev@dpdk.org; R, Kamalakannan
> ; Suresh Narayane, Harshad
>
> Subject: Re: [
ASan reported issues in this driver.
rte_rawdev_obj_t context object points at a uint16_t.
skeleton_rawdev_enqueue_bufs() and skeleton_rawdev_dequeue_bufs() were
incorrectly casting to an int.
The enqueue/dequeue selftest had a leak on the enqueued string and was
wrong in passing a rte_rawdev_buf
The rawdev unit test can be run in the CI.
Signed-off-by: David Marchand
---
app/test/meson.build | 5 -
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/app/test/meson.build b/app/test/meson.build
index 5d78814878..6624b478b4 100644
--- a/app/test/meson.build
+++ b/app/test/meso
On 2/23/23 06:10, Stephen Hemminger wrote:
The static LOGTYPE PMD will go away in the future.
Replace its use in the sfc driver with the same static
type as is already used in RTE_LOG_REGISTER macro.
These drivers are trying to do things that other drivers are not
in creating log types. IMHO it
On 3/9/23 07:28, Ivan Malov wrote:
The currently used API is unfit for the purpose as it
checks the order in which the action is being added.
Use a dedicated API to request the reset.
Fixes: 012bf708c20f ("net/sfc: support group flows in tunnel offload")
Cc: sta...@dpdk.org
Signed-off-by: Ivan
On 3/9/23 07:28, Ivan Malov wrote:
Previously, DPDK sfc driver received support for tunnel
offload. In it, MAE needs to set intermediate mark from
an outer rule (OR) recirculation ID in order to help
the driver identify packets that hit the OR but miss
on action rule (AR) lookup. But, for packets
On 3/9/23 07:11, Ivan Malov wrote:
The driver supports representor functionality. In it,
packets coming from VFs to the dedicated back-end Rx
queue get demultiplexed into front-end Rx queues of
representor ethdevs as per the per-packet metadata
indicating logical HW ingress ports. On transmit,
pa
On Mon, Feb 20, 2023 at 12:14 AM Stephen Hemminger
wrote:
>
> While diagnosing some Windows cross build errors;
> noticed that lots of important DPDK libraries are not
> being built on Windows.
>
> Stephen Hemminger (8):
> net/null: build null PMD on Windows
> net/ring: build on Windows
> lp
>
>
> >
> >
> >
> > >
> > > Use lcore API to check if the lcore ID is valid. The runtime check
> > > does not add much value.
> >
> > From my perspective it adds a perfect value:
> > Only threads with valid lcore id have their own default mempool cache.
> The threads would call 'rte_lcore_id()
On 3/9/2023 4:11 AM, Ivan Malov wrote:
> The driver supports representor functionality. In it,
> packets coming from VFs to the dedicated back-end Rx
> queue get demultiplexed into front-end Rx queues of
> representor ethdevs as per the per-packet metadata
> indicating logical HW ingress ports. On
On 3/10/2023 1:47 PM, Dumitrescu, Cristian wrote:
>
>
>> -Original Message-
>> From: Ferruh Yigit
>> Sent: Friday, March 10, 2023 12:00 PM
>> To: Dumitrescu, Cristian ; Stephen Hemminger
>> ; Jangra, Yogesh ;
>> Singh, Aman Deep ; Zhang, Yuying
>>
>> Cc: dev@dpdk.org; R, Kamalakannan ;
On Fri, Mar 10, 2023 at 2:48 PM Dumitrescu, Cristian
wrote:
> > Why not fix the misbehaving drivers, instead of working around for
> > softnic, as Stephen suggested?
> >
> > Is there a list of problematic drivers?
> >
>
> Ferruh, I think this is not a reasonable request. We don't have the expertis
> -Original Message-
> From: Ferruh Yigit
> Sent: Friday, March 10, 2023 12:00 PM
> To: Dumitrescu, Cristian ; Stephen Hemminger
> ; Jangra, Yogesh ;
> Singh, Aman Deep ; Zhang, Yuying
>
> Cc: dev@dpdk.org; R, Kamalakannan ; Suresh
> Narayane, Harshad
> Subject: Re: [PATCH] app/testpmd
On Fri, Mar 10, 2023 at 1:00 PM Ferruh Yigit wrote:
> >> NAK
> >> No driver specific hacks please.
> >>
> >> Instead fix the driver design or bug please.
> >
> > Hi Stephen,
> >
> > This is not a Soft NIC driver-specific hack, this is required for working
> > around some of the ethdev drivers tha
> -Original Message-
> From: Singh, Aman Deep
> Sent: Friday, March 10, 2023 9:09 AM
> To: Stephen Hemminger ; Dumitrescu, Cristian
>
> Cc: Jangra, Yogesh ; dev@dpdk.org; R,
> Kamalakannan ; Suresh Narayane, Harshad
>
> Subject: Re: [PATCH] app/testpmd: fix closing softnic port before
A memory leak was detected using the AddressSanitizer tool,
when running the crypto-perf application with a test vector file.
The strdup function used returns a pointer to a null-terminated byte
string, which must be freed after use. This patch frees the pointer in
an error case, and also after a
-Original Message-
From: Dumitrescu, Cristian
Sent: Friday, March 10, 2023 1:40 AM
To: dev@dpdk.org
Cc: R, Kamalakannan ; Jangra, Yogesh
; Suresh Narayane, Harshad
Subject: [PATCH V3] doc: announce the pipeline legacy API deprecation
Announce the deprecation of the legacy pipeline,
On 3/9/2023 5:19 PM, Dumitrescu, Cristian wrote:
>
>
>> -Original Message-
>> From: Stephen Hemminger
>> Sent: Thursday, March 9, 2023 4:31 PM
>> To: Jangra, Yogesh
>> Cc: dev@dpdk.org; Dumitrescu, Cristian ; R,
>> Kamalakannan ; Suresh Narayane, Harshad
>>
>> Subject: Re: [PATCH] app/
On 3/10/2023 6:25 AM, Chaoyong He wrote:
> + /**
> + * Shift and mask nn_link_status so that it is effectively the value
> + * at offset NFP_NET_CFG_STS_NSP_LINK_RATE.
> + */
> + nn_link_status = (nn_link_status >> NFP_NET_CFG_STS_LINK_RATE_SHIFT) &
> + N
To ensure proper path tracking for files being used by the build, the
"files()" function should always be used. While meson currently assumes
that bare filenames passed to commands refer to paths in the current
directory, other reimplementations of meson, e.g. muon, require the
paths to be properly
According to the meson manual [1], the project language should be given
as 'c', rather than 'C' in the initial "project" function call. While
meson itself does not complain about this, the "muon" project, which
attempts to re-implement meson in C, does complain.
/home/bruce/dpdk/meson.build:4:17:
While experimenting with "muon" [1] as a replacement for meson, the
muon tool flagged a few issues that could do with being cleaned up
in our meson.build files. While currently muon is not in a position
to configure a DPDK build, it's not that far away from being able to
do so, so may be worth keep
> -Original Message-
> From: Thomas Monjalon
> Sent: 10 March 2023 15:01
> To: dev@dpdk.org
> Cc: Srikanth Yalavarthi ; Shivah Shankar Shankar
> Narayan Rao ; Jerin Jacob Kollanukkaran
> ; Anup Prabhu ; Prince Takkar
> ; Parijat Shukla
> Subject: [EXT] Re: [PATCH v6 00/39] Implementation
> -Original Message-
> From: Kaiwen Deng
> Sent: Friday, March 10, 2023 12:42 PM
> To: dev@dpdk.org
> Cc: sta...@dpdk.org; Yang, Qiming ; Zhou, YidingX
> ; Deng, KaiwenX ;
> Chas Williams ; Min Hu (Connor) ;
> Tomasz Kulasek ; Doherty, Declan
>
> Subject: [PATCH] net/bonding: fix forward
On 3/9/2023 8:09 PM, Cristian Dumitrescu wrote:
Announce the deprecation of the legacy pipeline, table and port
library API and gradual stabilization of the new API.
Signed-off-by: Cristian Dumitrescu
---
doc/guides/rel_notes/deprecation.rst | 18 ++
1 file changed, 18 inse
On Thu, Mar 09, 2023 at 08:09:49PM +, Cristian Dumitrescu wrote:
> Announce the deprecation of the legacy pipeline, table and port
> library API and gradual stabilization of the new API.
>
> Signed-off-by: Cristian Dumitrescu
Presumably the marking the new APIs as stable shouldn't need to be
https://bugs.dpdk.org/show_bug.cgi?id=1177
Bug ID: 1177
Summary: telemetry unit test crashes with musl
Product: DPDK
Version: unspecified
Hardware: All
OS: All
Status: UNCONFIRMED
Severity: normal
https://bugs.dpdk.org/show_bug.cgi?id=1176
Bug ID: 1176
Summary: errno_autotest failing with musl
Product: DPDK
Version: unspecified
Hardware: All
OS: All
Status: UNCONFIRMED
Severity: normal
Pri
From: Huisong Li
Currently, hns3 driver supports setting mulitiple rte flow RSS rule,
but only the last is valid. This implementation is different from
the mainstream usage of rte flow hash in the community. Please see
the discussion threads [1] and [2].
This patch sets RSS hash feature complete
From: Huisong Li
Add the verification of RSS types from ethdev ops and rte flow without
pattern attribute. The following cases are invalid:
1. types contains unsupported RSS type but hasn't type driver support.
2. types has L3 src/dst but hasn't supported packet type.
3. types has L4 src/dst but
From: Huisong Li
Some RSS fields in struct hns3_rss_conf (e.g. conf, queue,
valid) are only used when create RSS flow rule, which is
unnecessary for RSS configuration information from ethdev
ops. This patch removes these fields from hns3_rss_conf
and add a new struct hns3_flow_rss_conf as rte flo
From: Huisong Li
Save hash algo from rte flow RSS rule to RSS filter list node
instead of struct hns3_rss_conf.
Cc: sta...@dpdk.org
Signed-off-by: Huisong Li
Signed-off-by: Dongdong Liu
---
drivers/net/hns3/hns3_flow.c | 19 ++-
1 file changed, 10 insertions(+), 9 deletions(-
From: Huisong Li
Remove redundant assignment and directly use new RSS rule to configure
hardware. Additionally, considering that the new rule configuration may
need to be modified, the 'const' of input parameter about it is removed.
Cc: sta...@dpdk.org
Signed-off-by: Huisong Li
Signed-off-by:
From: Huisong Li
Separate the setting and clearing of RSS rule.
Cc: sta...@dpdk.org
Signed-off-by: Huisong Li
Signed-off-by: Dongdong Liu
---
drivers/net/hns3/hns3_flow.c | 46 +---
1 file changed, 22 insertions(+), 24 deletions(-)
diff --git a/drivers/net/hn
From: Huisong Li
The maximum queue number from RSS flow rule allowed depends on
the maximum queue number (512) under one TC instead of the one
of Rx/Tx queue so as to eliminate restrictions on user usage.
Fixes: c37ca66f2b27 ("net/hns3: support RSS")
Cc: sta...@dpdk.org
Signed-off-by: Huisong L
From: Huisong Li
The settings of RSS types comes from the ethdev ops (like, dev_configure
and rss_hash_update) and rte_flow API. For the ethdev ops, driver has to
save it so as to it can be restored when reset is triggered.
While rte_flow API no need to maintain this field, it can be restored by
From: Huisong Li
The settings of redirection table comes from the ethdev ops (like,
dev_configure and rss_hash_update) and rte_flow API. For the ethdev
ops, driver has to save it to rss_info::rss_indirection_tbl in hns3_hw
structure so as to it can be restored when reset is triggered.
While rte_f
From: Huisong Li
The settings of hash key comes from the ethdev ops (like, dev_configure
and rss_hash_update) and rte_flow API. For the ethdev ops, driver has
to save it to rss_info::key in hns3_hw structure so as to it can be
restored when reset is triggered. While rte_flow API no need to use
th
From: Huisong Li
Currently, the setting of hash algorithm comes from the
default configuration in driver and the rte_flow interface.
The hash algorithm that is set to hardware in both ways is
saved in hw->rss_info.conf.func.
But the 'func' in struct rte_flow_action_rss is usually used
in rte flo
From: Huisong Li
Currently, reta_query() API reports the redirection table from software.
This patch uses the one in hardware to report.
Fixes: c37ca66f2b27 ("net/hns3: support RSS")
Cc: sta...@dpdk.org
Signed-off-by: Huisong Li
Signed-off-by: Dongdong Liu
---
drivers/net/hns3/hns3_cmd.h |
From: Huisong Li
Use the configuration in hardware to report hash types instead
of data maintained in software.
Fixes: c37ca66f2b27 ("net/hns3: support RSS")
Cc: sta...@dpdk.org
Signed-off-by: Huisong Li
Signed-off-by: Dongdong Liu
---
drivers/net/hns3/hns3_rss.c | 260 ++
From: Huisong Li
The size of the redirection table is obtained from firmware. If the size
isn't a multiple of HNS3_RSS_CFG_TBL_SIZE, the redirection table from
user will be truncated.
Fixes: c37ca66f2b27 ("net/hns3: support RSS")
Cc: sta...@dpdk.org
Signed-off-by: Huisong Li
Signed-off-by: Don
From: Huisong Li
Currently, hns3_dev_rss_hash_conf_get() interface reports RSS key from
the key maintained by driver. It's better to report the key from hardware,
which is more realistic.
Fixes: c37ca66f2b27 ("net/hns3: support RSS")
Cc: sta...@dpdk.org
Signed-off-by: Huisong Li
Signed-off-by:
From: Huisong Li
The hash key length of hns3 driver is obtained from firmware. If the
length isn't a multiple of HNS3_RSS_HASH_KEY_NUM (16), the last part
of hash key will be truncated.
Fixes: 4a7384e3c34d ("net/hns3: refactor set RSS hash algorithm and key
interface")
Fixes: c37ca66f2b27 ("net
This patchset is to do some code refactor for hns3 RSS.
1. [PATCH 01/16]-[PATCH 02/16] fix some hns3 RSS bug.
2. [PATCH 03/16]-[PATCH 14/16] refactoring hns3 RSS code.
3. [PATCH 15/16]-[PATCH 16/16] reimplement hash flow function.
Huisong Li (16):
net/hns3: fix possible truncation of hash key w
On Thu, Mar 9, 2023 at 10:21 PM Stephen Hemminger
wrote:
>
> On Thu, 9 Mar 2023 22:10:03 +0100
> David Marchand wrote:
>
> > > diff --git a/drivers/net/ring/rte_eth_ring.c
> > > b/drivers/net/ring/rte_eth_ring.c
> > > index bd5a47dd9017..e8bc9b627102 100644
> > > --- a/drivers/net/ring/rte_eth_r
10/03/2023 10:23, David Marchand:
> Caught by UNH lab, Windows compilation is broken because of symbol
> exports:
> rte_mldev_exports.def : error LNK2001: unresolved external symbol
> rte_ml_io_bfloat16_to_float32
> rte_mldev_exports.def : error LNK2001: unresolved external symbol
> rte
10/03/2023 09:57, David Marchand:
> Caught by UNH lab, cross compiling for aarch32:
>
> ../lib/mldev/mldev_utils_neon.c: In function
> '__float32_to_int8_neon_s8x8':
> ../lib/mldev/mldev_utils_neon.c:34:10: error: implicit declaration
> of function 'vcvtaq_s32_f32'; did you mean 'vcvtq
> -Original Message-
> From: Thomas Monjalon
> Sent: 10 March 2023 14:59
> To: Srikanth Yalavarthi
> Cc: dev@dpdk.org; Shivah Shankar Shankar Narayan Rao
> ; Jerin Jacob Kollanukkaran
> ; Anup Prabhu ; Prince Takkar
> ; Parijat Shukla
> Subject: Re: [EXT] Re: [PATCH v5 02/39] ml/cnxk: ad
10/03/2023 09:19, Srikanth Yalavarthi:
> Marvell ML CNXK Driver
> --
>
> This patch series implements common Machine Learning (ML) ROC code
> and driver for Marvell Octeon 10 (cnxk) platform. ML inferencing is
> supported on cnxk platform through an integrated ML inferencing
>
10/03/2023 09:25, Srikanth Yalavarthi:
> From: Thomas Monjalon
> > 07/02/2023 17:06, Srikanth Yalavarthi:
> > > --- a/doc/guides/rel_notes/release_23_03.rst
> > > +++ b/doc/guides/rel_notes/release_23_03.rst
> > > +* **Implementation of Marvell CNXK machine learning driver for .**
> >
> > It seem
Caught by UNH lab, Windows compilation is broken because of symbol
exports:
rte_mldev_exports.def : error LNK2001: unresolved external symbol
rte_ml_io_bfloat16_to_float32
rte_mldev_exports.def : error LNK2001: unresolved external symbol
rte_ml_io_float16_to_float32
rte_mldev_export
1 - 100 of 162 matches
Mail list logo
