mlx5_ext_rxq_get() and mlx5_ext_txq_get() functions did not return
NULL value if query index was not referencing external queue.
As a result, calling functions did not expect the NULL on return.
External Rx queue:
- In mlx5_ext_rxq_get() remove assert and return NULL if a queue index
does not
-Original Message-
From: Ivan Malov
Sent: Thursday, July 31, 2025 11:14 AM
To: Wani, Shaiq
Cc: dev@dpdk.org; Richardson, Bruce ; Singh, Aman
Deep ; Shaw, Jeffrey B
Subject: Re: [PATCH 2/2] net/ice: count drop- all filter stats in Rx stats
Hi,
On Thu, 31 Jul 2025, Shaiq Wani wrote:
Hi,
On Thu, 31 Jul 2025, Shaiq Wani wrote:
Packets dropped in the receive direction are counted as
"rx_unknown_protocol" (GLSWID_RUPP),these packets need to be added
to the "ipackets" and the GLV_GORCH/GLV_GORCL counters
need to be added to rx_bytes.
Signed-off-by: Jeff Shaw
Signed-off-by: Sh
Packets dropped in the receive direction are counted as
"rx_unknown_protocol" (GLSWID_RUPP),these packets need to be added
to the "ipackets" and the GLV_GORCH/GLV_GORCL counters
need to be added to rx_bytes.
Signed-off-by: Jeff Shaw
Signed-off-by: Shaiq Wani
---
drivers/net/intel/ice/ice_ethdev
The tx packets were dropped when creating drop any rule
for ingress direction only. Added a direction lookup flag
to fix the issue.
Signed-off-by: Jeff Shaw
Signed-off-by: Shaiq Wani
---
drivers/net/intel/ice/base/ice_switch.c | 8
drivers/net/intel/ice/ice_switch_filter.c | 2 ++
2
Patch 1.Tx packets were dropped when a drop-any rule was applied
only on ingress. A direction lookup flag is added to resolve this.
Patch 2.Added accounting for Rx drop packets in ipackets and ibytes.
Shaiq Wani (2):
net/ice: add flag for direction lookup to occur
net/ice: count drop-
Maybe that's not a problem, because ACL can handle idle
packets normally, so I can't reproduce this issue with
external test cases at the moment, but I'm just describing
what idle nomatch node has to do if it wants to implement
an effect pointing to itself:
The reason why RTE_ACL_NODE_SINGLE node
Hello,
The DTS group has met and agreed on the top level goals for 25.11 DTS
development.
DTS Roadmap:
* DTS API
* Provide a single point of entry into the framework for all test
suites, allowing stable usage across releases and user friendly
discoverability of dts functionality
* Pac
#
July 24, 2025
Attendees
1. Patrick Robb
2. Luca Vizzarro
3. Dean Marx
4. Adam Hassick
5. Paul Szczepanek
#
Minutes
> From: Bruce Richardson [mailto:bruce.richard...@intel.com]
> Sent: Wednesday, 30 July 2025 19.29
>
> On Mon, Jul 28, 2025 at 07:48:53PM +0200, Morten Brørup wrote:
> > > From: Bruce Richardson [mailto:bruce.richard...@intel.com]
> > > Sent: Monday, 28 July 2025 17.11
> > >
> > > On Mon, Jul 28,
https://bugs.dpdk.org/show_bug.cgi?id=1762
Bug ID: 1762
Summary: Unable to use TB-mode when code blocks have different
length (ACC100 / VRB drivers)
Product: DPDK
Version: unspecified
Hardware: All
OS: All
On Mon, Jul 28, 2025 at 07:48:53PM +0200, Morten Brørup wrote:
> > From: Bruce Richardson [mailto:bruce.richard...@intel.com]
> > Sent: Monday, 28 July 2025 17.11
> >
> > On Mon, Jul 28, 2025 at 04:51:30PM +0200, Morten Brørup wrote:
> > > > From: Dean Marx [mailto:dm...@iol.unh.edu]
> > > > Sent:
Hi Thomas,
On Wed, 30 Jul 2025, Thomas Wilks wrote:
Port over the rss_key_update, pmd_rss_reta and pmd_rss_hash
test suites from old DTS into one file including all of the
helper functions that are required by all of the test suites
to work.
The rss_key_update test cases verify that setting a
https://bugs.dpdk.org/show_bug.cgi?id=1761
Bug ID: 1761
Summary: rte_raw_cksum_mbuf obtain incorrect cksum when a
packet in mbuf split mode.
Product: DPDK
Version: unspecified
Hardware: All
OS: Linux
Added fast mbuf release, re-using the existing mbuf pool pointer
in the queue structure.
Signed-off-by: Morten Brørup
---
v4:
* Force the generic function called by the separate tx_pkt_burst callbacks
to be inlined.
v3:
* Use separate tx_pkt_burst callbacks depending on per-device TX offload
Added fast mbuf release, re-using the existing mbuf pool pointer
in the queue structure.
Signed-off-by: Morten Brørup
---
v3:
* Use separate tx_pkt_burst callbacks depending on per-device TX offload
configuration. (Ivan Malov, Konstantin Ananyev)
* Check TX offload configuration for mutually ex
Hi Dean,
thank you so much for your patch. Looks mostly correct, just lots of
polishing suggestions for a more readable code.
On 17/07/2025 21:57, Dean Marx wrote:
+"""QinQ (802.1ad) Test Suite.
+
+This test suite verifies the correctness and capability of DPDK Poll Mode
Drivers (PMDs)
+in h
Port over the rss_key_update, pmd_rss_reta and pmd_rss_hash
test suites from old DTS into one file including all of the
helper functions that are required by all of the test suites
to work.
The rss_key_update test cases verify that setting a new hash
key when Receive Side Scaling (RSS) will result
From: Alex Chapman
This patch adds the required functionality for the RSS key_update,
RETA, and hash test suites. This includes:
The setting of custom RETA values for routing packets to specific
queues.
The setting of the RSS mode on all ports, to specify how to hash
the packets.
The updating of
Hi all,
This is v4 of the RSS test suites which addresses the issues with running the
hashing algorithm
test cases on ConnectX nics. The patches have also been rebased onto the latest
upstream changes.
Also thank you for your help and feedback with my previous RFC patch.
Changes in v4:
- Mer
Currently, the AF_XDP PMD attaches its XDP program using a fixed
set of flags, defaulting to `XDP_FLAGS_UPDATE_IF_NOEXIST`. This lacks
flexibility for users who might need to explicitly control the XDP
mode based on their specific hardware capabilities or performance
requirements.
This patch intro
As part of non-template implementation,
new default rule was added in HWS mode (dv_flow_en=2)
in case dv_xmeta_en=4 (32bits extended META for HWS mode).
In case the proxy port did not start, and it was not configured,
it failed to create this rule and failed to start port representors.
Added ignor
Looks good to me, just some minor comments.
On 17/07/2025 21:57, Dean Marx wrote:
+def set_vlan_extend(self, port: int, enable: bool, verify: bool = True) ->
None:
+if enable ^ (vlan_settings is not None and VLANOffloadFlag.EXTEND
in vlan_settings):
+self._logg
A previous patch addressing an ASAN global-buffer-overflow issue
inadvertently degraded the random mask value translation.
Specifically, changing the mask from 0x to 0x caused
the value to be shifted left by 16 bits, since Testpmd
defines arg::mask as a byte array.
Independent of the A
Thanks for cleaning up.
Series-acked-by: Morten Brørup
Added immediate return from ntb_dev_intr_handler()
if no doorbells.
Signed-off-by: Tofig Aliev
---
...ntb-immediate-return-if-no-doorbells.patch | 32 +++
drivers/raw/ntb/ntb.c | 4 ++-
2 files changed, 35 insertions(+), 1 deletion(-)
create mode 100644
Ethdev driver maintainers (CC: Ethdev API maintainers),
Your ethdev drivers support RTE_ETH_TX_OFFLOAD_MBUF_FAST_FREE, and probably
call rte_mempool_put_bulk() in the mempool lib when FAST_FREE'ing mbufs,
thereby bypassing the mbuf lib.
The appropriate mbuf lib function is rte_mbuf_raw_free_bulk
During asynchronous flow creation by index,
the items array was initialized with only one element,
but the table metadata did not update the item count accordingly.
This mismatch led to an out-of-bounds memcpy operation,
as the code attempted to copy more elements than were actually allocated.
To
28 matches
Mail list logo
