On 2023/11/1 21:36, Ferruh Yigit wrote:
On 11/1/2023 7:40 AM, Jie Hai wrote:
Currently, rte_eth_rss_conf supports configuring and querying
RSS hash functions, rss key and it's length, but not RSS hash
algorithm.
The structure ``rte_eth_dev_info`` is extended by adding a new
field "rss_algo_capa
Hi,
> -Original Message-
> From: Qi Zhang
> Sent: Thursday, November 2, 2023 8:05 PM
> To: Xing, Beilei
> Cc: dev@dpdk.org; Zhang, Qi Z ; sta...@dpdk.org
> Subject: [PATCH] net/iavf: fix Tx preparation
>
> 1. check nb_segs > Tx ring size for TSO case.
> 2. report nb_mtu_seg_max and nb_
Hi,
> -Original Message-
> From: Zhang, Qi Z
> Sent: Thursday, November 2, 2023 10:22 PM
> To: Yang, Qiming
> Cc: dev@dpdk.org; Zhang, Qi Z ; sta...@dpdk.org
> Subject: [PATCH v2] net/ice: fix Tx preparation
>
> 1. Check nb_segs > 8 for NO TSO case
> 2. Check nb_segs > Tx ring size for
From: Zerun Fu
Add support for configuring RX/TX pause frame switch mode.
Signed-off-by: Zerun Fu
Reviewed-by: Long Wu
Reviewed-by: Peng Zhang
Reviewed-by: Chaoyong He
---
drivers/net/nfp/nfp_ethdev.c | 1 +
drivers/net/nfp/nfp_net_common.c | 75 +++
d
This patch series add the support of RX and TX pause frame switch mode.
Zerun Fu (2):
net/nfp: support getting pause frame switch mode
net/nfp: support setting pause frame switch mode
drivers/net/nfp/nfp_ethdev.c | 2 +
drivers/net/nfp/nfp_net_common.c | 116
From: Zerun Fu
Add support for querying current RX/TX pause frame
switch mode from the device.
Signed-off-by: Zerun Fu
Reviewed-by: Long Wu
Reviewed-by: Peng Zhang
Reviewed-by: Chaoyong He
---
drivers/net/nfp/nfp_ethdev.c | 1 +
drivers/net/nfp/nfp_net_common.c | 41 +
1. Check nb_segs > 8 for NO TSO case
2. Check nb_segs > Tx ring size for TSO case
3. report nb_mtu_seg_max and nb_seg_max in dev_info.
Fixes: 17c7d0f9d6a4 ("net/ice: support basic Rx/Tx")
Cc: sta...@dpdk.org
Signed-off-by: Qi Zhang
---
drivers/net/ice/ice_ethdev.c | 2 ++
drivers/net/ice/ice_r
From: Qin Ke
Add support of dynamic config of allmulticast mode for NFP
by adding enable and disable callback function.
Signed-off-by: Qin Ke
Reviewed-by: Chaoyong He
---
doc/guides/nics/features/nfp.ini | 1 +
drivers/common/nfp/nfp_common_ctrl.h | 1 +
drivers/net/nfp/nfp_ethdev.c
When AVX2 is forcibly selected and outer checksum
offload is configured, the basic Tx path will be
selected. Also, the txq mbuf release API is incorrectly
set to iavf_tx_queue_release_mbufs_sse. This causes
coredump.
This commit selects release_txq_mbufs to releasing
txq mbufs when selecting the b
> -Original Message-
> From: Zeng, ZhichaoX
> Sent: Thursday, November 2, 2023 9:51 AM
> To: dev@dpdk.org
> Cc: sta...@dpdk.org; Liao, TingtingX ; Zeng,
> ZhichaoX ; Yang, Qiming
> ; Zhang, Qi Z ; Wang,
> Haiyue
> Subject: [PATCH v2] net/ice: fix DCF port statistics not cleared
>
> Ca
> -Original Message-
> From: Ye, MingjinX
> Sent: Thursday, November 2, 2023 10:39 AM
> To: Zhang, Qi Z ; dev@dpdk.org
> Cc: Yang, Qiming ; Zhou, YidingX
> ; sta...@dpdk.org
> Subject: RE: [PATCH v3] net/ice: fix crash on closing representor ports
>
>
>
> > -Original Message-
1. check nb_segs > Tx ring size for TSO case.
2. report nb_mtu_seg_max and nb_seg_max in dev_info.
Fixes: a2b29a7733ef ("net/avf: enable basic Rx Tx")
Cc: sta...@dpdk.org
Signed-off-by: Qi Zhang
---
drivers/net/iavf/iavf_ethdev.c | 2 ++
drivers/net/iavf/iavf_rxtx.c | 3 ++-
2 files changed,
On 2023/11/1 21:19, Ferruh Yigit wrote:
On 11/1/2023 7:40 AM, Jie Hai wrote:
In rte_eth_dev_rss_hash_conf_get(), the "rss_key_len" should be
greater than or equal to the "hash_key_size" which get from
rte_eth_dev_info_get() API. And the "rss_key" should contain at
least "hash_key_size" bytes. If
On 2023/11/1 21:13, Ferruh Yigit wrote:
On 11/1/2023 7:40 AM, Jie Hai wrote:
1. overwrite the comments of fields of 'rte_eth_rss_conf'.
2. Add comments for RTE_ETH_HASH_FUNCTION_DEFAULT.
Signed-off-by: Jie Hai
---
lib/ethdev/rte_ethdev.h | 34 +++---
lib/ethdev/r
Use __rte_atomic_thread_fence instead of directly using
__atomic_thread_fence builtin gcc intrinsic
Signed-off-by: Tyler Retzlaff
---
lib/eal/common/eal_common_trace.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/eal/common/eal_common_trace.c
b/lib/eal/common/eal_comm
Use __rte_atomic_thread_fence instead of directly using
__atomic_thread_fence builtin gcc intrinsic
Signed-off-by: Tyler Retzlaff
---
lib/stack/rte_stack_lf_c11.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/stack/rte_stack_lf_c11.h b/lib/stack/rte_stack_lf_c11.h
index
Use __rte_atomic_thread_fence instead of directly using
__atomic_thread_fence builtin gcc intrinsic
Signed-off-by: Tyler Retzlaff
---
lib/ring/rte_ring_c11_pvt.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/lib/ring/rte_ring_c11_pvt.h b/lib/ring/rte_ring_c11_pvt.h
inde
Replace use of __atomic_thread_fence with __rte_atomic_thread_fence.
It may be appropriate to use rte_atomic_thread_fence instead but it
will be up to maintainers to evaluate and make the change if appropriate.
Tyler Retzlaff (5):
distributor: use rte atomic thread fence
eal: use rte atomic t
Use __rte_atomic_thread_fence instead of directly using
__atomic_thread_fence builtin gcc intrinsic
Signed-off-by: Tyler Retzlaff
---
lib/distributor/rte_distributor.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/distributor/rte_distributor.c
b/lib/distributor/rte_dis
Use __rte_atomic_thread_fence instead of directly using
__atomic_thread_fence builtin gcc intrinsic
Signed-off-by: Tyler Retzlaff
---
lib/hash/rte_cuckoo_hash.c | 10 +-
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/lib/hash/rte_cuckoo_hash.c b/lib/hash/rte_cuckoo_hash.c
> -Original Message-
> From: Zhang, Qi Z
> Sent: Wednesday, November 1, 2023 6:49 PM
> To: Ye, MingjinX ; dev@dpdk.org
> Cc: Yang, Qiming ; Zhou, YidingX
> ; sta...@dpdk.org
> Subject: RE: [PATCH v3] net/ice: fix crash on closing representor ports
>
>
>
> > -Original Message-
From: Peng Zhang
For backward compatibility concern, the new application firmware
is designed to support both single PF scenario and multiple PF scenario.
Thus driver should inform application firmware which setup current
is. This should be done as early as possible since the setup may
affect som
From: Peng Zhang
In multiple PF scenario, management firmware is in charge of
application firmware unloading instead of driver by keepalive
mechanism.
A new NSP resource area is allocated for keepalive use with name
"nfp.beat". Driver sets periodically updates the PFs' corresponding
word in "nfp
From: Peng Zhang
Add a NSP command to check if any firmware have been loaded.
Signed-off-by: Peng Zhang
Reviewed-by: Chaoyong He
Reviewed-by: Long Wu
---
drivers/net/nfp/nfpcore/nfp_nsp.c | 6 ++
drivers/net/nfp/nfpcore/nfp_nsp.h | 1 +
2 files changed, 7 insertions(+)
diff --git a/driv
From: Peng Zhang
In single PF scenario, the format symbols just is related
with PF ID 0. In multiple PF scenario, the format symbols
should be related with PF ID. So this commit adds the
PF ID used to format symbols.
Signed-off-by: Peng Zhang
Reviewed-by: Chaoyong He
Reviewed-by: Long Wu
---
From: Peng Zhang
When firmware load failed, it doesn't have any notice.
So add the check about the firmware load and add an exit
point when the firmware load process fail.
Signed-off-by: Peng Zhang
Reviewed-by: Chaoyong He
Reviewed-by: Long Wu
---
drivers/net/nfp/nfp_ethdev.c | 7 ++-
1
From: Peng Zhang
If the firmware supports multiple PFs each PF is represented by a single
physical port. While if the firmware only supports a single PF there
might be one or more physical ports represented by a single PF.
Adjust the check to handle both single and multiple PFs firmware.
Signed
From: Peng Zhang
The commands sent to the NSP take the NSP major version into account. Up
until now only NSP major version 0 have been supported and the value
have been hard-coded to 0.
In preparation to add support for both NSP version 0.x and 1.x, extend
the command to take the running NSP ver
From: Peng Zhang
Support for multiple PFs have been added to the NFP3800 firmware. This
can be detected by reading the NSP major version, which was bumped to 1
when support was added.
Add a flag and detecting method to record if the current device is
cabable to support multiple PFs. This will be
From: Shihong Wang
When DPDK application exit abnormally, there might have DMA error,
and which will cause the load of firmware failed.
Fix this by force the physical port down to clear the possible DMA error.
Fixes: 896c265ef954 ("net/nfp: use new CPP interface")
Cc: sta...@dpdk.org
Signed-of
From: Peng Zhang
In rare cases, when DPDK application exit, the interrupt handler was not
processed the interrupt in time, resulting in the LSC interrupt mask bit
not being cleared. So when the DPDK application start again, the newly
coming LSC interrupts cannot be received and processed properly
Up to now, the NFP card using only one PF (or BDF) for multiple physical
ports, this force the PMD import the difference logic for 'PF' and
'physical port'. Which is not easy to understand and also not compatible
with some DPDK applications.
This patch series add the support of multiple PF, which w
The probe logic of the secondary process of PF PMD now is not very
similarly with the logic of the primary process, which cause we need two
different logics when we add new feature in some case.
Refactor the probe logic of the secondary process to solve this problem.
Signed-off-by: Chaoyong He
S
On Tue, Oct 31, 2023 at 08:49:28PM -0400, David Young wrote:
> Merged windows_install_build.rst into building_from_sources.rst
[...]
> -Download the DPDK source code from the official repository
> -``https://fast.dpdk.org/rel/``.
> +Windows System Requirements
> +^^^
>
Hi Amit,
I prefer not use static variable to control it because it introduce many
coupling.
Suggest add one function which prepare the test_m2d_auto_free, like
prepare_m2d_auto_free
if ((info.dev_capa & RTE_DMA_CAPA_M2D_AUTO_FREE) &&
dma_add_test[TEST_M2D_AUTO_FREE].ena
在 2023/11/2 0:08, Stephen Hemminger 写道:
On Wed, 1 Nov 2023 10:36:07 +0800
"lihuisong (C)" wrote:
Do we need to report this size? It's a common feature for all PMDs.
It would make sense then to have max_rx_bufsize set to 16K by default
in ethdev, and PMD could then raise/lower based on hardwa
1. Check nb_segs > 8 for NO TSO case
2. Check nb_segs > Tx ring size for TSO case
3. report nb_mtu_seg_max and nb_seg_max in dev_info.
Fixes: 17c7d0f9d6a4 ("net/ice: support basic Rx/Tx")
Cc: sta...@dpdk.org
Signed-off-by: Qi Zhang
---
drivers/net/ice/ice_ethdev.c | 2 ++
drivers/net/ice/ice_r
Call 'ice_dcf_stats_reset' during the initialization of the DCF port in
order to clear any statistics that may exist from the last use of the DCF
and to avoid statistics errors.
Fixes: 7564d5509611 ("net/ice: add DCF hardware initialization")
Cc: sta...@dpdk.org
Signed-off-by: Zhichao Zeng
---
Reviewed-by: John Daley
Thanks,
John
From: Weiguo Li
Date: Wednesday, November 1, 2023 at 12:28 AM
To: John Daley (johndale)
Cc: dev@dpdk.org , sta...@dpdk.org , Weiguo Li
Subject: [PATCH] net/enic: avoid extra unlock when setting MTU in enic
The 'set_mtu_done' goto statement is being execut
> On 10/28/2023 7:53 AM, Chaoyong He wrote:
> > The data structure 'struct nfp_net_adapter' has only one data field
> > and we won't extend it in the future, which makes this abstraction
> > unneeded, so remove this data structure and the related macro
> > 'NFP_NET_DEV_PRIVATE_TO_HW'.
> >
>
> Ment
On 8/24/2023 8:18 AM, David Marchand wrote:
> The L4 checksum offloading API does not require l4_len to be set.
> Make the driver discover the L4 headers size by itself.
>
> Fixes: 6546e76056e3 ("net/tap: calculate checksums of multi segs packets")
> Cc: sta...@dpdk.org
>
> Signed-off-by: David M
Use rte_ctz64 instead of __builtin_ctzl
Fixes: 18898c4d06f9 ("eal: use abstracted bit count functions")
Signed-off-by: Tyler Retzlaff
---
lib/rcu/rte_rcu_qsbr.c | 4 ++--
lib/rcu/rte_rcu_qsbr.h | 2 +-
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/lib/rcu/rte_rcu_qsbr.c b/lib/r
Use rte_ctz64 instead of __builtin_ctzl
Fixes: 18898c4d06f9 ("eal: use abstracted bit count functions")
Signed-off-by: Tyler Retzlaff
---
lib/member/rte_member_vbf.c | 12 ++--
lib/member/rte_member_x86.h | 6 +++---
2 files changed, 9 insertions(+), 9 deletions(-)
diff --git a/lib/me
Use rte_ctz64 instead of __builtin_ctzl
Fixes: 18898c4d06f9 ("eal: use abstracted bit count functions")
Signed-off-by: Tyler Retzlaff
---
lib/hash/rte_cuckoo_hash.c | 16
1 file changed, 8 insertions(+), 8 deletions(-)
diff --git a/lib/hash/rte_cuckoo_hash.c b/lib/hash/rte_cuc
The first set of conversions missed the long 'l' versions of the
builtins that were being used. This series completes the conversion
of remaining libraries from __builtin_ctzl and __builtin_clzl.
Tyler Retzlaff (5):
table: use abstracted bit count functions
distributor: use abstracted bit coun
Use rte_ctz64 instead of __builtin_ctzl
Fixes: 18898c4d06f9 ("eal: use abstracted bit count functions")
Signed-off-by: Tyler Retzlaff
---
lib/distributor/rte_distributor_single.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/distributor/rte_distributor_single.c
b/lib/
Use rte_clz64 instead of __builtin_clzl
Use rte_ctz64 instead of __builtin_ctzl
Fixes: 18898c4d06f9 ("eal: use abstracted bit count functions")
Signed-off-by: Tyler Retzlaff
---
lib/table/rte_lru_arm64.h | 2 +-
lib/table/rte_swx_table_em.c | 4 ++--
lib/table/rte_table_hash_ext.c | 4 ++
timesync_read_rx_timestamp
On 9/21/2023 12:59 PM, lihuisong (C) wrote:
> add ice & igc maintainers
>
> 在 2023/9/21 19:06, Ferruh Yigit 写道:
>> On 9/21/2023 11:02 AM, lihuisong (C) wrote:
>>> Hi Ferruh,
>>>
>>> Sorry for my delay reply because of taking a look at all PMDs
>>> implementation.
>>>
>>>
On 9/21/2023 12:17 PM, Hemant Agrawal wrote:
> HI Ferruh,
>
>> On 9/21/2023 11:02 AM, lihuisong (C) wrote:
>>> Hi Ferruh,
>>>
>>> Sorry for my delay reply because of taking a look at all PMDs
>>> implementation.
>>>
>>>
>>> 在 2023/9/16 1:46, Ferruh Yigit 写道:
On 8/17/2023 9:42 AM, Huisong Li w
On 10/20/2023 4:58 AM, lihuisong (C) wrote:
>
> 在 2023/9/21 19:17, Hemant Agrawal 写道:
>> HI Ferruh,
>>
>>> On 9/21/2023 11:02 AM, lihuisong (C) wrote:
Hi Ferruh,
Sorry for my delay reply because of taking a look at all PMDs
implementation.
在 2023/9/16 1:46, Ferru
Provide an alternate RTE_TRACE_POINT_REGISTER macro when building with
MSVC that allocates segments for the trace point using MSVC specific
features
Signed-off-by: Tyler Retzlaff
---
lib/eal/include/rte_trace_point_register.h | 14 ++
1 file changed, 14 insertions(+)
diff --git a/li
add missing __extension__ keyword to RTE_ALIGN_MUL_NEAR statement
expression to be consistent with other macros using statement
expressions
Signed-off-by: Tyler Retzlaff
---
lib/eal/include/rte_common.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/eal/include/rte_commo
On 10/31/2023 12:23 PM, Jie Hai wrote:
> This patchset contains some bugfix for hns3 pmd.
>
> Huisong Li (2):
> net/hns3: fix setting DCB capability
> net/hns3: fix LRO offload to report
>
> Jie Hai (4):
> net/hns3: fix return value
> net/hns3: fix some error log
> net/hns3: do not expo
On 11/1/2023 7:28 AM, Weiguo Li wrote:
> The 'set_mtu_done' goto statement is being executed in a context
> where the 'mtu_lock' has not been previously locked.
>
> To avoid the extra unlocking operation, replace the goto statement
> with a return statement.
>
> Fixes: c3e09182bcd6 ("net/enic: su
>++ PCIe maintainers.
>+I will leave this up to @David Marchand / @Thomas as this patch has common
>code changes and needs to come via main tree.
>+Also in this case, The comment was given very early(Back in June 7) for the
>same.
>+https://patches.dpdk.org/project/dpdk/patch/20230607210050.1
On 10/28/2023 7:53 AM, Chaoyong He wrote:
> This patch series clean up the NFP PMD, by:
> - Using the DPDK macro and API to replace the user defined ones.
> - Remove the unneeded macro and logic.
> - Remove the duplicated logic.
>
> ---
> v2:
> * Fix the compile error.
> * Fix one check script war
On Wed, 1 Nov 2023 15:20:53 +0800
Ruifeng Wang wrote:
> From: Ruifeng Wang
> To: Kevin Laatz , Robin Jarry ,
> Morten Brørup
> Cc: dev@dpdk.org, honnappa.nagaraha...@arm.com, n...@arm.com, Ruifeng Wang
> , sta...@dpdk.org
> Subject: [PATCH] eal: stop iteration after lcore info is processed
Test compresses and decompresses pointers with various combinations
of memory regions and alignments and verifies the pointers are
recovered correctly.
Signed-off-by: Paul Szczepanek
---
app/test/meson.build | 1 +
app/test/test_eal_ptr_compress.c | 108
Add a test that runs a zero copy burst enqueue and dequeue on a ring
of raw pointers and compressed pointers at different burst sizes to
showcase performance benefits of newly added pointer compression APIs.
Refactored threading code to pass more parameters to threads to
reuse existing code. Added
Documentation added in the EAL guide for the new
utility functions for pointer compression
showing example code and potential usecases.
Signed-off-by: Paul Szczepanek
Reviewed-by: Honnappa Nagarahalli
---
.../prog_guide/env_abstraction_layer.rst | 142 ++
1 file changed, 14
Add a new utility header for compressing pointers. The provided
functions can store pointers in 32-bit offsets.
The compression takes advantage of the fact that pointers are
usually located in a limited memory region (like a mempool).
We can compress them by converting them to offsets from a base
This patchset is proposing adding a new EAL header with utility functions
that allow compression of arrays of pointers.
When passing caches full of pointers between threads, memory containing
the pointers is copied multiple times which is especially costly between
cores. A compression method will
Rebuild with commit
c0335cc197 (tap; rebuild and update the BPF flow program, 2023-10-31)
Signed-off-by: Stephen Hemminger
---
drivers/net/tap/tap_bpf_insns.h | 2959 ---
1 file changed, 1503 insertions(+), 1456 deletions(-)
diff --git a/drivers/net/tap/tap_bpf_insns
From: Madhuker Mythri
As per analysis on Tap PMD, the existing RSS algorithm considering
4-tuple(Src-IP, Dst-IP, Src-port and Dst-port) and identification of
fragment packets is not done, thus we are seeing all the fragmented
chunks of single packet differs in RSS hash value and distributed acros
Move the BPF program related code into a subdirectory.
And add a Makefile for building it.
The code depends on include files from iproute2.
But these are not public headers which iproute2 exports
as a package API. Therefore make a local copy here.
The standalone build was also broken because by
c
Update the documentation and tools to build the BPF program used by
tap device. And apply fix to the RSS algorithm to correctly
handle non-IP protocols.
v6 - cosmetic improvements to extract process
add better boilerplate and fix python lint warnings
Madhuker Mythri (1):
net/tap: Fixed RSS
On 10/28/2023 7:53 AM, Chaoyong He wrote:
> The data structure 'struct nfp_net_adapter' has only one data field and
> we won't extend it in the future, which makes this abstraction unneeded,
> so remove this data structure and the related macro
> 'NFP_NET_DEV_PRIVATE_TO_HW'.
>
Mentioned abstract
On 10/28/2023 7:31 AM, Chaoyong He wrote:
> From: Shihong Wang
>
> This patch resolves configuration error of ol_flags in the
> rte_security_capability. Currently ol_flags in the ingress direction
> of the SA, 'RTE_SECURITY_TX_OLOAD_NEED_MDATA' is configured. In fact,
> ol_flags only in the egres
On 11/1/2023 8:12 AM, Ori Kam wrote:
> Hi
>
>> -Original Message-
>> From: Ruifeng Wang
>> Sent: Wednesday, November 1, 2023 9:16 AM
>>
>> aarch32 build with gcc-13.0.1 generated following warning:
>>
>> In function 'memcpy',
>> inlined from 'rte_memcpy' at
>> ../lib/eal/arm/include/r
On 11/1/2023 3:32 AM, Jiawen Wu wrote:
> Refer to commit 85e46c532bc7 ("net/ixgbe: add proper memory barriers in
> Rx"). Fix the same issue as ixgbe.
>
> Segmentation fault has been observed while running the
> txgbe_recv_pkts_lro() function to receive packets on the Loongson 3A5000
> processor. I
On November 1, 2023 19:20:29 Morten Brørup wrote:
Add co-maintainer for Memory pool.
Suggested-by: Thomas Monjalon
Signed-off-by: Morten Brørup
Acked-by: Andrew Rybchenko
---
MAINTAINERS | 1 +
1 file changed, 1 insertion(+)
diff --git a/MAINTAINERS b/MAINTAINERS
index 4083658697..8a4
Add co-maintainer for Memory pool.
Suggested-by: Thomas Monjalon
Signed-off-by: Morten Brørup
---
MAINTAINERS | 1 +
1 file changed, 1 insertion(+)
diff --git a/MAINTAINERS b/MAINTAINERS
index 4083658697..8a4e9f0a9e 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -379,6 +379,7 @@ T: git://dpdk.o
On Wed, 1 Nov 2023 10:36:07 +0800
"lihuisong (C)" wrote:
> > Do we need to report this size? It's a common feature for all PMDs.
> > It would make sense then to have max_rx_bufsize set to 16K by default
> > in ethdev, and PMD could then raise/lower based on hardware.
> It is not appropriate t
On 11/1/2023 2:19 AM, Chaoyong He wrote:
> From: Long Wu
>
> CI found that overrunning array of 32 2-byte elements at
> element index 65535 (byte offset 131071) by dereferencing
> pointer "members + agg_new_idx".
>
> Coverity issue: 403099
> Fixes: 6d72657 ("net/bonding: add other aggregator mod
Keep unzipping if the next CQE is the miniCQE array in
rxq_cq_decompress_v() routine only for non-MPRQ scenario,
MPRQ requires buffer replenishment between the miniCQEs.
Restore the check for the initial compressed CQE for SPRQ
and check that the current CQE is not compressed before
copying it as
Keep unzipping if the next CQE is the miniCQE array in
rxq_cq_decompress_v() routine only for non-MPRQ scenario,
MPRQ requires buffer replenishment between the miniCQEs.
Restore the check for the initial compressed CQE for SPRQ
and check that the current CQE is not compressed before
copying it as
> diff --git a/app/test/test_dmadev.c b/app/test/test_dmadev.c index
> 216f84b6bb..3d4cb37ee6 100644
> --- a/app/test/test_dmadev.c
> +++ b/app/test/test_dmadev.c
> @@ -49,6 +49,8 @@ struct dma_add_test dma_add_test[] = {
> [TEST_M2D_AUTO_FREE] = {.name = "m2d_auto_free", .enabled =
> false},
On 11/1/2023 7:40 AM, Jie Hai wrote:
> Add the command "show port X rss-hash algorithm" to display
> the RSS hash algorithms of port X. An example is shown:
>
> testpmd> show port 0 rss-hash algorithm
> RSS algorithm:
> toeplitz
>
> Signed-off-by: Jie Hai
> Acked-by: Huisong Li
> ---
> app/t
On 11/1/2023 7:40 AM, Jie Hai wrote:
> Currently, rte_eth_rss_conf supports configuring and querying
> RSS hash functions, rss key and it's length, but not RSS hash
> algorithm.
>
> The structure ``rte_eth_dev_info`` is extended by adding a new
> field "rss_algo_capa". Drivers are responsible for
On 11/1/2023 7:40 AM, Jie Hai wrote:
> In rte_eth_dev_rss_hash_conf_get(), the "rss_key_len" should be
> greater than or equal to the "hash_key_size" which get from
> rte_eth_dev_info_get() API. And the "rss_key" should contain at
> least "hash_key_size" bytes. If these requirements are not met,
>
On 11/1/2023 7:40 AM, Jie Hai wrote:
> 1. overwrite the comments of fields of 'rte_eth_rss_conf'.
> 2. Add comments for RTE_ETH_HASH_FUNCTION_DEFAULT.
>
> Signed-off-by: Jie Hai
> ---
> lib/ethdev/rte_ethdev.h | 34 +++---
> lib/ethdev/rte_flow.h | 1 +
> 2 files c
On 25/10/2023 13:57, Juraj Linkeš wrote:
The aarch32 with gcc13 fails with:
Compiler for C supports arguments -march=armv8-a: NO
../config/arm/meson.build:714:12: ERROR: Problem encountered: No
suitable armv8 march version found.
This is because we test -march=armv8-a alone (without the -mpf
On 10/28/2023 4:01 AM, lihuisong (C) wrote:
> With belows to changes,
> Acked-by: Huisong Li
>
>
> 在 2023/10/28 9:46, Jie Hai 写道:
>> Currently, rte_eth_rss_conf supports configuring and querying
>> RSS hash functions, rss key and it's length, but not RSS hash
>> algorithm.
>>
>> The structure ``
On 10/28/2023 3:41 AM, lihuisong (C) wrote:
>
> 在 2023/10/28 9:46, Jie Hai 写道:
>> 1. overwrite the comments of fields of 'rte_eth_rss_conf'.
>> 2. Add comments for RTE_ETH_HASH_FUNCTION_DEFAULT.
>>
>> Signed-off-by: Jie Hai
>> ---
>> lib/ethdev/rte_ethdev.h | 33 ++--
On 01/11/2023 07:42, Morten Brørup wrote:
From: Paul Szczepanek [mailto:paul.szczepa...@arm.com]
Sent: Tuesday, 31 October 2023 19.11
[...]
In a more realistic mock application running the l3 forwarding dpdk
example that works in pipeline mode this translated into a ~5%
throughput
increase on
Documentation added in the EAL guide for the new
utility functions for pointer compression
showing example code and potential usecases
Signed-off-by: Paul Szczepanek
Reviewed-by: Honnappa Nagarahalli
---
.../prog_guide/env_abstraction_layer.rst | 142 ++
1 file changed, 142
Add a new utility header for compressing pointers. The provided
functions can store pointers in 32-bit offsets.
The compression takes advantage of the fact that pointers are
usually located in a limited memory region (like a mempool).
We can compress them by converting them to offsets from a base
Test compresses and decompresses pointers with various combinations
of memory regions and alignments and verifies the pointers are
recovered correctly.
Signed-off-by: Paul Szczepanek
---
app/test/meson.build | 1 +
app/test/test_eal_ptr_compress.c | 108
Add a test that runs a zero copy burst enqueue and dequeue on a ring
of raw pointers and compressed pointers at different burst sizes to
showcase performance benefits of newly added pointer compression APIs.
Refactored threading code to pass more parameters to threads to
reuse existing code. Added
This patchset is proposing adding a new EAL header with utility functions
that allow compression of arrays of pointers.
When passing caches full of pointers between threads, memory containing
the pointers is copied multiple times which is especially costly between
cores. A compression method will
On 10/28/2023 7:26 AM, Chaoyong He wrote:
> From: James Hershaw
>
> Reduce the space reserved for layer 2 overhead by defining
> NFP_ETH_OVERHEAD.
>
> Previously, the overhead was not explicitly defined, only the
> NFP_FRAME_SIZE_MAX value and the maximum layer 3 MTU was read from
> hardware, wh
On 11/11/2022 05:27, Qiming Yang wrote:
Add recommended matching list for ice PMD in DPDK 22.07 and
i40e PMD in DPDK 22.07 and 22.11.
Signed-off-by: Qiming Yang
---
doc/guides/nics/i40e.rst | 8
doc/guides/nics/ice.rst | 2 ++
2 files changed, 10 insertions(+)
diff --git a/doc/gu
> -Original Message-
> From: Kaiwen Deng
> Sent: Wednesday, November 1, 2023 9:35 AM
> To: dev@dpdk.org
> Cc: sta...@dpdk.org; Yang, Qiming ; Zhou, YidingX
> ; Deng, KaiwenX ; Wu,
> Jingjing ; Xing, Beilei ; Zeng,
> ZhichaoX
> Subject: [PATCH v2] net/iavf: fix coredump when exiting tes
> -Original Message-
> From: Ye, MingjinX
> Sent: Wednesday, November 1, 2023 6:14 PM
> To: dev@dpdk.org
> Cc: Yang, Qiming ; Zhou, YidingX
> ; Ye, MingjinX ;
> sta...@dpdk.org; Zhang, Qi Z
> Subject: [PATCH v3] net/ice: fix crash on closing representor ports
>
> The data resource in
The data resource in struct rte_eth_dev is cleared and points to NULL
when the DCF port is closed.
If the DCF representor port is closed after the DCF port is closed,
a segmentation fault occurs because the representor port accesses
the data resource released by the DCF port.
This patch checks if
Buffer auto free test failed for more than 1 dma device as the device
initialization for the test was been done only for the first dma device.
This changeset fixes the same.
Fixes: 877cb3e37426 ("dmadev: add buffer auto free offload")
Signed-off-by: Amit Prakash Shukla
---
app/test/test_dmadev.
lgtm,
Acked-by: Huisong Li
在 2023/11/1 15:40, Jie Hai 写道:
1. overwrite the comments of fields of 'rte_eth_rss_conf'.
2. Add comments for RTE_ETH_HASH_FUNCTION_DEFAULT.
Signed-off-by: Jie Hai
---
lib/ethdev/rte_ethdev.h | 34 +++---
lib/ethdev/rte_flow.h | 1 +
lgtm,
Acked-by: Huisong Li
在 2023/11/1 15:40, Jie Hai 写道:
Display RSS hash algorithm with command show-port as below.
- RSS info
-- hash algorithm : toeplitz
Signed-off-by: Jie Hai
Signed-off-by: Dongdong Liu
Acked-by: Reshma Pattan
---
app/proc-info/main.c | 2 ++
1 file ch
在 2023/11/1 15:40, Jie Hai 写道:
In rte_eth_dev_rss_hash_conf_get(), the "rss_key_len" should be
greater than or equal to the "hash_key_size" which get from
rte_eth_dev_info_get() API. And the "rss_key" should contain at
least "hash_key_size" bytes. If these requirements are not met,
the query un
LGTM
Series-acked-by: Chengwen Feng
On 2023/11/1 15:40, Jie Hai wrote:
> This patchset is to support setting and querying RSS algorithms.
> For this purpose, field "rss_algo_capa" is added to ``rte_eth_dev_info``
> and field "algorithm" is added to ``rte_eth_rss_conf``.
> The drivers should repo
1 - 100 of 122 matches
Mail list logo
