It was found that this statistic is incorrectly reported
by HW and thus, useless.
Remove it.
Signed-off-by: Jan Sokolowski
---
drivers/net/ethernet/intel/ice/ice_ethtool.c | 1 -
drivers/net/ethernet/intel/ice/ice_main.c| 5 -
drivers/net/ethernet/intel/ice/ice_type.h| 1 -
3 files
On Thu, Nov 16, 2023 at 04:21:12PM +0100, Ivan Vecera wrote:
> Add two helpers i40e_(veb|vsi)_get_by_seid() to find corresponding
> VEB or VSI by their SEID value and use these helpers to replace
> existing open-coded loops.
>
> Reviewed-by: Wojciech Drewek
> Signed-off-by: Ivan Vecera
Hi Ivan,
From: Kunwu Chan
Date: Sun, 19 Nov 2023 23:12:09 +0800
> Hi Alexander,
> Thank you so much for your reply, I looked at the modification you
> mentioned, it's really cool. I'll definitely try it next time.
>
> But when using it, will it be easy to forget to free up memory?
You have a kfree() at
From: Nguyen, Anthony L
Sent: Friday, November 17, 2023 12:36 AM
>On 11/14/2023 1:10 AM, Jedrzej Jagielski wrote:
>> Currently ixgbe driver is notified of overheating events
>> via internal IXGBE_ERR_OVERTEMP erorr code.
>>
>> Change the approach to use is_overhaet variable
>> which set when su
Hi Alexander,
Thank you so much for your reply, I looked at the modification you
mentioned, it's really cool. I'll definitely try it next time.
But when using it, will it be easy to forget to free up memory?
Although 'kmalloc_track_caller' is used, according to my understanding,
it is also nec
Hi,
The bi-weekly netdev call at https://bbb.lwn.net/b/jak-wkr-seg-hjn
is scheduled tomorrow at 8:30 am (PT) / 5:30 pm (~EU).
So far the only agenda item is a minor update on CI,
please send other topics!
In terms of the review rotation - this week's reviewer is Intel.
_
On 20. 11. 23 12:42, Simon Horman wrote:
On Thu, Nov 16, 2023 at 04:21:12PM +0100, Ivan Vecera wrote:
Add two helpers i40e_(veb|vsi)_get_by_seid() to find corresponding
VEB or VSI by their SEID value and use these helpers to replace
existing open-coded loops.
Reviewed-by: Wojciech Drewek
Sig
Patch 1 modifies the set_rxh ethtool API to take a pointer to struct
ethtool_rxfh instead of individual params. This will allow future
changes to the struct without changing the API.
Patch 2 adds the support at the Kernel level, allowing the user to set a
symmetric-xor RSS hash for a netdevice vi
Pass a pointer to struct ethtool_rxfh instead of the hfunc values to the
drivers' set_rxfh and get_rxfh ethtool ops. This will allow us to add more
parameters to the struct without changing the APIs across all drivers.
In ethtool_get_rxfh(), copying struct ethtool_rxfh back to user-space is
moved
Symmetric RSS hash functions are beneficial in applications that monitor
both Tx and Rx packets of the same flow (IDS, software firewalls, ..etc).
Getting all traffic of the same flow on the same RX queue results in
higher CPU cache efficiency.
A NIC that supports "symmetric-xor" can achieve this
Fix the values of the ICE_AQ_VSI_Q_OPT_RSS_* registers. Shifting is
already done when the values are used, no need to double shift. Bug was
not discovered earlier since only ICE_AQ_VSI_Q_OPT_RSS_TPLZ (Zero) is
currently used.
Also, rename ICE_AQ_VSI_Q_OPT_RSS_XXX to ICE_AQ_VSI_Q_OPT_RSS_HASH_XXX
f
From: Qi Zhang
Refactor the driver to use a communication data structure for RSS
config. To do so we introduce the new ice_rss_hash_cfg struct, and then
pass it as an argument to several functions.
Also introduce enum ice_rss_cfg_hdr_type to specify a more granular and
flexible RSS configuration
The flow director and RSS blocks use separate methods to generate a
unique 64 bit ID for the flow. This is not extendable, especially for
the RSS that already uses all 64 bit space.
Refactor the flow generation API so that the ID is generated within
ice_flow_add_prof(). The FD and RSS blocks cache
From: Jeff Guo
Allow the user to set the symmetric Toeplitz hash function via:
# ethtool -X eth0 hfunc toeplitz symmetric-xor
All existing RSS configurations will be converted to symmetric unless they
have a non-symmetric field (other than IP src/dst and L4 src/dst ports)
used for hashing.
Allow the user to set the symmetric Toeplitz hash function via:
# ethtool -X eth0 hfunc toeplitz symmetric-xor
The driver will reject any new RSS configuration if a field other than
(IP src/dst and L4 src/dst ports) is requested for hashing.
The symmetric RSS will not be supported on PFs not
On 11/20/2023 3:27 AM, Jan Sokolowski wrote:
> It was found that this statistic is incorrectly reported
> by HW and thus, useless.
>
> Remove it.
Is there any further information about what is reported incorrectly? Is
rx_length_errors part of netdev stats and standardized? Does HW fail to
repo
Add support for devlink loopback param. Supported values are "enabled",
"disabled" and "prioritized". Default configuration is set to "enabled.
By default loopback traffic BW is locked to PF configured BW. HW is
capable of higher speeds on loopback traffic. Loopback param set to
"prioritized" enab
Please disregard, I have sent wrong patch.
On 2023-11-20 19:28, Pawel Kaminski wrote:
Add support for devlink loopback param. Supported values are "enabled",
"disabled" and "prioritized". Default configuration is set to "enabled.
By default loopback traffic BW is locked to PF configured BW. HW
Supported number of ntuple filters affect also maximum location value that
can be provided to ethtool command. Update error message to provide info
about max supported value.
Fix double spaces in the error messages.
Reviewed-by: Jesse Brandeburg
Signed-off-by: Pawel Kaminski
---
drivers/net/et
This series adds vfio live migration support for Intel E810 VF devices
based on the v2 migration protocol definition series discussed here[0].
Steps to test:
1. Bind one or more E810 VF devices to the module ice-vfio-pci.ko
2. Assign the VFs to the virtual machine and enable device live migration
Export RX queue context get function which is consumed by linux live
migration driver to save and load device state.
Signed-off-by: Yahui Cao
Signed-off-by: Lingyu Liu
---
drivers/net/ethernet/intel/ice/ice_common.c | 268
drivers/net/ethernet/intel/ice/ice_common.h | 5 +
Export TX queue context get and set function which is consumed by linux
live migration driver to save and load device state.
TX queue context contains static fields which does not change during TX
traffic and dynamic fields which may change during TX traffic.
Signed-off-by: Yahui Cao
---
driver
From: Lingyu Liu
During migration device resume stage, part of device state is loaded by
replaying logged virtual channel message. By default, once virtual
channel message is processed successfully, PF will send message to VF.
In addition, PF will notify VF about link state while handling virtua
From: Lingyu Liu
Add basic entry point for live migration functionality initialization,
uninitialization and add helper function for vfio driver to reach pf
driver data.
Signed-off-by: Lingyu Liu
Reviewed-by: Michal Swiatkowski
Signed-off-by: Yahui Cao
---
drivers/net/ethernet/intel/ice/Make
From: Lingyu Liu
Save the virtual channel messages sent by VF on the source side during
runtime. The logged virtchnl messages will be transferred and loaded
into the device on the destination side during the device resume stage.
For the feature which can not be migrated yet, it must be disabled
From: Lingyu Liu
Add device state save/load function to adapter vfio migration stack
when device is in stop-copy/resume stage.
Device state saving handler is called by vfio driver in device stop copy
stage. It snapshots the device state, translates device state into device
specific data and fill
From: Lingyu Liu
VSI id is a resource id for each VF and it is an absolute hardware id
per PCI card. It is exposed to VF driver through virtual channel
messages at the VF-PF negotiation stage. It is constant during the whole
device lifecycle unless driver re-init.
Almost all of the virtual chann
From: Lingyu Liu
RX Queue head is a fundamental dma ring context which determines the
next RX descriptor to be fetched. However, RX Queue head is not visible
to VF while it is only visible in PF. As a result, PF needs to save and
load RX Queue Head explicitly.
Since network packets may come in a
From: Lingyu Liu
TX Queue head is a fundamental DMA ring context which determines the
next TX descriptor to be fetched. However, TX Queue head is not visible
to VF while it is only visible in PF. As a result, PF needs to save and
load TX Queue head explicitly.
Unfortunately, due to HW limitation
From: Lingyu Liu
Device suspend handler is called by vfio driver before saving device
state. Typical operation includes stopping TX/RX queue.
Signed-off-by: Lingyu Liu
Signed-off-by: Yahui Cao
---
.../net/ethernet/intel/ice/ice_migration.c| 69 +++
include/linux/net/intel/
In E800 device model, VF takes direct control over the context of
AdminQ, irq ctrl, TX tail and RX tail by accessing VF pci mmio. Rest of
all the state can only be setup by PF and the procedure is that VF sends
all these configuration to PF through virtual channel messages to setup
the rest of the
From: Lingyu Liu
Add a vendor-specific vfio_pci driver for E800 devices.
It uses vfio_pci_core to register to the VFIO subsystem and then
implements the E800 specific logic to support VF live migration.
It implements the device state transition flow for live
migration.
Signed-off-by: Lingyu Li
32 matches
Mail list logo
