This patch supports rte flow with mask for FDIR, including
eth/ipv4/ipv6/tcp/udp flow items, where src/dst for ipv4/ipv6
and sport/dport for tcp/udp are realized by switch filter.
This patch is based on DPDK v21.11.5 LTS
[4e50ad4469f7c037e32de5aa3535d1cd25de0741], for customer cherry-pick.
Signed
> -Original Message-
> From: Zeng, ZhichaoX
> Sent: Thursday, December 21, 2023 3:28 PM
> To: dev@dpdk.org
> Cc: Zhang, Qi Z ; Zeng, ZhichaoX
> ; Wu, Jingjing ; Xing, Beilei
>
> Subject: [PATCH v4 1/3] net/iavf: support Tx LLDP on scalar
>
> This patch adds an mbuf dynfield IAVF_TX_LL
P7 device family supports 16 byte Rx completions.
Enable SSE vector mode for compressed Rx CQE processing.
Signed-off-by: Ajit Khaparde
Reviewed-by: Damodharam Ammepalli
---
drivers/net/bnxt/bnxt_ethdev.c | 16 ++-
drivers/net/bnxt/bnxt_rxr.h | 2 +
drivers/net/bnxt/bnxt_rxtx_
P7 device family supports 16 byte Rx completions.
Add AVX2 vector mode for compressed Rx CQE.
Signed-off-by: Ajit Khaparde
Reviewed-by: Damodharam Ammepalli
---
drivers/net/bnxt/bnxt_ethdev.c| 5 +
drivers/net/bnxt/bnxt_rxr.h | 2 +
drivers/net/bnxt/bnxt_rxtx_vec_avx2.c |
From: Damodharam Ammepalli
P7 devices report speeds over speeds2 hsi fields. Adding required
support to capture the capability from phy_qcap and save the
speeds2 fields into driver priv structure.
In fixed mode update the link_speed from force_link_speeds2 field.
Updates to logging to provide mor
From: Damodharam Ammepalli
Add the driver support for HWRM_STAT_EXT_CTX_QUERY HWRM
msg. In this patch only P7 chipset is enabled for this HWRM
while P5 and previous generation remain with HWRM_STAT_CTX_QUERY.
Signed-off-by: Damodharam Ammepalli
Reviewed-by: Ajit Khaparde
---
drivers/net/bnxt/
From: Damodharam Ammepalli
This patch adds support for tunnel TPA type.
The tunnel TPA support is brought in by the updated bit_field
tnl_tpa_en(4) in hwrm_vnic_tpa_cfg_input->enables,
which is used by the firmware to indicate the capability
of the underlying hardware.
This patch updates hwrm HW
From: Jay Ding
When a VF undergoes an FLR, the firmware indicates this via
an async notification to the PF. Note that the PF driver
needs to register for the notification with the firmware.
Add support for VF_FLR async event handling when the driver
is running on a PF.
Signed-off-by: Jay Ding
S
Fallback to all multicast mode if FW rejects multicast filter
programming.
The firmware can reject the MC filter programming request if
it is running low in resources when there is a large number of functions.
The driver must be prepared to fallback to the all-multicast mode
if the original MC fil
Check if the firmware can support RSS based on these types and
program the hardware accordingly when requested when the
firmware indicates that the underlying hardware supports the
functionality.
Signed-off-by: Ajit Khaparde
Reviewed-by: Damodharam Ammepalli
---
drivers/net/bnxt/bnxt.h|
From: Damodharam Ammepalli
This patch addsbnxt query callback to rte_flow_ops in non TruFlow mode.
At this point only the RSS hash function type is displayed.
Signed-off-by: Damodharam Ammepalli
Reviewed-by: Ajit Khaparde
---
drivers/net/bnxt/bnxt_flow.c | 61 +
P7 adapters support XOR based and checksum based RSS hashing.
Add support for checksum and XOR based RSS hash for these adapters.
Signed-off-by: Ajit Khaparde
Reviewed-by: Damodharam Ammepalli
---
drivers/net/bnxt/bnxt.h| 15 +--
drivers/net/bnxt/bnxt_ethdev.c | 72 ++-
dr
Ring the Rx doorbell during the Rx ring flush processing only if
there is a valid completion.
Signed-off-by: Ajit Khaparde
Reviewed-by: Damodharam Ammepalli
---
drivers/net/bnxt/bnxt_rxr.c | 9 +
1 file changed, 5 insertions(+), 4 deletions(-)
diff --git a/drivers/net/bnxt/bnxt_rxr.c b
Normally the PMD allocates a new mbuf for every mbuf consumed.
In case of mbuf alloc failure, that slot in the Rx or AGG ring remains
empty till a new mbuf is not allocated for that slot. If this happens
too frequently the Rx ring or the aggregation ring could be completely
drained of mbufs and can
From: Kalesh AP
On older firmware versions, HWRM_FUNC_QCAPS response is not
returning the maximum number of multicast filters that can be
supported by the function. As a result, memory allocation with
size 0 fails.
Bugzilla ID: 1309
Cc: sta...@dpdk.org
Signed-off-by: Kalesh AP
Reviewed-by: Aji
From: Kalesh AP
While forcing speed to 200G, driver sets the structure variable
"bp->link_info->link_signal_mode" value to BNXT_SIG_MODE_PAM4.
After that when the user forces the speed back to 25G, this
cached value is not set back to BNXT_SIG_MODE_NRZ which results
in issuing the HWRM_PORT_PHY_C
From: Kalesh AP
Thor based NICs can support PAM4 as well as NRZ link negotiation.
While PAM4 can negotiate speeds at 50G, 100G and 200G, the PMD will
use NRZ signaling for 50G and 100G speeds. PAM4 signaling will be
used only for 200G speed negotiations.
Driver has to check for NRZ speed support
From: Kalesh AP
While setting forced speed, the speed should have mapped to
macro "HWRM_PORT_PHY_CFG_INPUT_FORCE_xxx" instead of
"HWRM_PORT_PHY_CFG_INPUT_AUTO_xxx". We do not see any issue
as both these macros are defined to the same value.
Fixing it for better convey the intent.
Signed-off-by:
Thor2 supports compressed Rx completions instead of the
full featured 32-byte Rx completions.
Add support for these compressed CQEs in scalar mode.
Unlike in the typical Rx completions, the hardware does
not provide the opaque field to index into the aggregator
descriptor ring. So maintain the cons
UDP GSO aka UDP Fragmentation Offload allows an application or stack
to provide a data payload larger than the MTU. The application then
updates the mbuf ol_flags and sets the PKT_TX_UDP_SEG flag. Then based
on the tso_segs and tso_mss fields in the mbuf the PMD can indicate
the UDP GSO transmit re
This patchset contains more code fixes and reafctoring, apart from
extending more support for the newer P7 device family.
The patchset adds vector mode support for compressed Rx CQE.
This patchset includes support for checksum and XOR RSS hash mode.
It also adds support for forced speeds and autone
Document CLI for diagnose purpose.
Signed-off-by: Qi Zhang
---
doc/guides/nics/ice.rst | 36
1 file changed, 36 insertions(+)
diff --git a/doc/guides/nics/ice.rst b/doc/guides/nics/ice.rst
index 820a385b06..29309abe4d 100644
--- a/doc/guides/nics/ice.rst
+++
Added Testpmd CLI support for dumping Tx scheduling tree.
Usage:
testpmd>txsched dump
The output file is in "dot" format, which can be converted
into an image file using Graphviz.
- In "brief" mode, all scheduling nodes in the tree are displayed.
- In "detail" mode, each node's configuration
Dear Team,
I hope this message finds you well.
We have encountered a recurring deadlock issue within the function
rte_rwlock_write_lock in the DPDK version 22.11.3 LTS.
It appears to be related to a known matter addressed in
https://bugs.dpdk.org/show_bug.cgi?id=1277 and subsequently resolved
in
Document CLI for diagnose purpose.
Signed-off-by: Qi Zhang
---
doc/guides/nics/ice.rst | 36
1 file changed, 36 insertions(+)
diff --git a/doc/guides/nics/ice.rst b/doc/guides/nics/ice.rst
index 820a385b06..29309abe4d 100644
--- a/doc/guides/nics/ice.rst
+++
Added Testpmd CLI support for dumping Tx scheduling tree.
Usage:
testpmd>txsched dump
The output file is in "dot" format, which can be converted
into an image file using Graphviz.
- In "brief" mode, all scheduling nodes in the tree are displayed.
- In "detail" mode, each node's configuration
The only way to enable diagnostics for TX paths is to modify the
application source code. Making it difficult to diagnose faults.
In this patch, the devarg option "mbuf_check" is introduced and the
parameters are configured to enable the corresponding diagnostics.
supported cases: mbuf, size, seg
Hi Chuanyu,
> -Original Message-
> From: Chuanyu Xue
> Sent: Friday, December 22, 2023 11:04 AM
> To: Su, Simei
> Cc: Xing, Beilei ; chuanyu@uconn.edu;
> dev@dpdk.org; Zhang, Qi Z ; Lu, Wenzhuo
>
> Subject: RE: [PATCH] net/e1000: support launchtime feature
>
> Hi Simei,
> Thank you
28 matches
Mail list logo
