On Sat, May 11, 2024 at 4:20 AM Daniel Jurgens wrote:
>
> TX queue stop and wake are counted by some drivers.
> Support reporting these via netdev-genl queue stats.
>
> Signed-off-by: Daniel Jurgens
> Reviewed-by: Jiri Pirko
Good to see this finally :)
Revie
> It wouldn't be trivial. The stats are queried from the driver.
So are page pool stats, with the increments happening in the page pool
code, not the driver.
Andrew
> From: Andrew Lunn
> Sent: Friday, May 10, 2024 7:59 AM
> To: Dan Jurgens
> Cc: netdev@vger.kernel.org; m...@redhat.com; jasow...@redhat.com;
> xuanz...@linux.alibaba.com; virtualizat...@lists.linux.dev;
> da...@davemloft.net; eduma...@google.com; k...@kernel.org;
> pa
TX queue stop and wake are counted by some drivers.
Support reporting these via netdev-genl queue stats.
Signed-off-by: Daniel Jurgens
Reviewed-by: Jiri Pirko
---
Documentation/netlink/specs/netdev.yaml | 14 ++
include/net/netdev_queues.h | 3 +++
include/uapi/linux
On Thu, May 09, 2024 at 09:19:52PM +, Dan Jurgens wrote:
> > From: Andrew Lunn
> > Sent: Thursday, May 9, 2024 3:47 PM
> > To: Dan Jurgens
> > Subject: Re: [PATCH net-next 1/2] netdev: Add queue stats for TX stop and
> > wake
> >
> > On Thu, May 09
> From: Jakub Kicinski
> Sent: Thursday, May 9, 2024 8:31 PM
> To: Dan Jurgens
> Cc: netdev@vger.kernel.org; m...@redhat.com; jasow...@redhat.com;
> xuanz...@linux.alibaba.com; virtualizat...@lists.linux.dev;
> da...@davemloft.net; eduma...@google.com; pab...@redhat.com; Jiri
On Thu, 9 May 2024 11:32:15 -0500 Daniel Jurgens wrote:
> diff --git a/tools/include/uapi/linux/netdev.h
> b/tools/include/uapi/linux/netdev.h
> index cf24f1d9adf8..ccf6976b1693 100644
> --- a/tools/include/uapi/linux/netdev.h
> +++ b/tools/include/uapi/linux/netdev.h
> @@ -164,7 +164,8 @@ enum {
> From: Andrew Lunn
> Sent: Thursday, May 9, 2024 3:47 PM
> To: Dan Jurgens
> Subject: Re: [PATCH net-next 1/2] netdev: Add queue stats for TX stop and
> wake
>
> On Thu, May 09, 2024 at 11:32:15AM -0500, Daniel Jurgens wrote:
> > TX queue stop and wake are counted b
On Thu, May 09, 2024 at 11:32:15AM -0500, Daniel Jurgens wrote:
> TX queue stop and wake are counted by some drivers.
> Support reporting these via netdev-genl queue stats.
>
> Signed-off-by: Daniel Jurgens
> Reviewed-by: Jiri Pirko
> ---
> Documentation/netlink/s
TX queue stop and wake are counted by some drivers.
Support reporting these via netdev-genl queue stats.
Signed-off-by: Daniel Jurgens
Reviewed-by: Jiri Pirko
---
Documentation/netlink/specs/netdev.yaml | 10 ++
include/net/netdev_queues.h | 3 +++
include/uapi/linux
On Fri, 26 Apr 2024 11:39:27 +0800 Xuan Zhuo wrote:
> These stats are commonly. Support reporting those via netdev-genl queue
> stats.
>
> name: rx-hw-drops
> name: rx-hw-drop-overruns
> name: rx-csum-unnecessary
> name: rx-csum-none
> name: rx-csum-bad
> name: rx-hw-gr
These stats are commonly. Support reporting those via netdev-genl queue
stats.
name: rx-hw-drops
name: rx-hw-drop-overruns
name: rx-csum-unnecessary
name: rx-csum-none
name: rx-csum-bad
name: rx-hw-gro-packets
name: rx-hw-gro-bytes
name: rx-hw-gro-wire-packets
name: rx-hw-gro-wire-bytes
name: rx
These stats are commonly. Support reporting those via netdev-genl queue
stats.
name: rx-hw-drops
name: rx-hw-drop-overruns
name: rx-csum-unnecessary
name: rx-csum-none
name: rx-csum-bad
name: rx-hw-gro-packets
name: rx-hw-gro-bytes
name: rx-hw-gro-wire-packets
name: rx-hw-gro-wire-bytes
name: rx
These stats are commonly. Support reporting those via netdev-genl queue
stats.
name: rx-hw-drops
name: rx-hw-drop-overruns
name: rx-csum-unnecessary
name: rx-csum-none
name: rx-csum-bad
name: rx-hw-gro-packets
name: rx-hw-gro-bytes
name: rx-hw-gro-wire-packets
name: rx-hw-gro-wire-bytes
name: rx
On Thu, 14 Mar 2024 16:21:42 -0700, Jakub Kicinski wrote:
> On Thu, 14 Mar 2024 16:54:58 +0800 Xuan Zhuo wrote:
> > + -
> > +name: rx-hw-drops
> > +doc: |
> > + Number of packets that arrived at the device but never left it,
> > + encompassing packets dropped
On Thu, 14 Mar 2024 16:54:58 +0800 Xuan Zhuo wrote:
> + -
> +name: rx-hw-drops
> +doc: |
> + Number of packets that arrived at the device but never left it,
> + encompassing packets dropped for reasons such as insufficient
> buffer
s/encompassing/including/
These stats are commonly. Support reporting those via netdev-genl queue
stats.
name: rx-hw-drops
name: rx-hw-drop-overruns
name: rx-csum-unnecessary
name: rx-csum-none
name: rx-csum-bad
name: rx-hw-gro-packets
name: rx-hw-gro-bytes
name: rx-hw-gro-wire-packets
name: rx-hw-gro-wire-bytes
name: rx
On Wed, 14 Apr 2021 17:30:01 -0700 Tony Nguyen wrote:
> From: Anirudh Venkataramanan
>
> Add two new VSI states, one to track if a netdev for the VSI has been
> allocated and the other to track if the netdev has been registered.
> Call unregister_netdev/free_netdev only when th
On 15.04.2021 20:42:36, Vincent MAILHOL wrote:
> On Thu. 15 Apr 2021 at 18:04, Marc Kleine-Budde wrote:
> > On 15.04.2021 09:47:23, Colin King wrote:
> > > From: Colin Ian King
> > >
> > > There is an assignment to *netdev tha
On Thu. 15 Apr 2021 at 18:04, Marc Kleine-Budde wrote:
> On 15.04.2021 09:47:23, Colin King wrote:
> > From: Colin Ian King
> >
> > There is an assignment to *netdev that is can potentially be null but the
^^
Typo: that is can
On 15.04.2021 09:47:23, Colin King wrote:
> From: Colin Ian King
>
> There is an assignment to *netdev that is can potentially be null but the
> null check is checking netdev and not *netdev as intended. Fix this by
> adding in the missing * operator.
>
> Addresses-Co
From: Colin Ian King
There is an assignment to *netdev that is can potentially be null but the
null check is checking netdev and not *netdev as intended. Fix this by
adding in the missing * operator.
Addresses-Coverity: ("Dereference before null check")
Fixes: 8537257874e9 ("
From: Anirudh Venkataramanan
Add two new VSI states, one to track if a netdev for the VSI has been
allocated and the other to track if the netdev has been registered.
Call unregister_netdev/free_netdev only when the corresponding state
bits are set.
Signed-off-by: Anirudh Venkataramanan
Tested
Hi,
I'm seeing some odd behaviour with the netdev trigger and some WiFi
interfaces.
When the WiFi interface has never been brought up (so is in an
operationally disabled state), if I bind a LED to the netdev
trigger, setting the device_name to the WiFi interface name and
enable the
In ipa_modem_start(), we set endpoint netdev pointers before the
network device is registered. If registration fails, we don't undo
those assignments. Instead, wait to assign the netdev pointer until
after registration succeeds.
Set these endpoint netdev pointers to NULL in ipa_modem
From: Arkadiusz Kubalewski
Remove vsi->netdev->name from the trace.
This is redundant information. With the devinfo trace, the adapter
is already identifiable.
Previously following error was produced when compiling against sparse.
i40e_main.c:2571 i40e_sync_vsi_filters() error:
From: Vu Pham
Dynamic allocate vlan table in mlx5e_priv for EN netdev
when needed. Don't allocate it for representor netdev.
Signed-off-by: Vu Pham
Signed-off-by: Saeed Mahameed
---
.../net/ethernet/mellanox/mlx5/core/en/fs.h | 18 +--
.../net/ethernet/mellanox/mlx5/core/en_fs.c
From: Vu Pham
Dynamic allocate arfs table in mlx5e_priv for EN netdev
when needed. Don't allocate it for representor netdev.
Signed-off-by: Vu Pham
Signed-off-by: Saeed Mahameed
---
.../net/ethernet/mellanox/mlx5/core/en/fs.h | 28 +-
.../net/ethernet/mellanox/mlx5/core/en_arfs.c
From: Anirudh Venkataramanan
Once a netdev is registered, the corresponding network interface can
be immediately used by userspace utilities (like say NetworkManager).
This can be problematic if the driver technically isn't fully up yet.
Move netdev registration to the end of probe, as by
On Wed, 2021-03-24 at 12:15 +0800, we...@ucloud.cn wrote:
> From: wenxu
>
> Ipsec and tls netdev features build should be done after the
> mlx5e_init_ipesc/tls which finishs the init for the ipsec/tls
> in the driver.
>
> Fixes: 3ef14e463f6e ("net/mlx5e: Separate
_secs sysctl to make
> the timeout configurable for automated testing systems.
> Lowering the timeout may also be useful for e.g. manual bisection.
> The default value matches the current behavior.
>
> Signed-off-by: Dmitry Vyukov
> Fixes: https://bugzilla.kernel.org/show_bug.cg
From: wenxu
Ipsec and tls netdev features build should be done after the
mlx5e_init_ipesc/tls which finishs the init for the ipsec/tls
in the driver.
Fixes: 3ef14e463f6e ("net/mlx5e: Separate between netdev objects and mlx5e
profiles initialization")
Signed-off-by: wenxu
---
d
ed testing systems.
> Lowering the timeout may also be useful for e.g. manual bisection.
> The default value matches the current behavior.
>
> Signed-off-by: Dmitry Vyukov
> Fixes: https://bugzilla.kernel.org/show_bug.cgi?id=211877
> Cc: netdev@vger.kernel.org
> Cc: linux-ker..
Add a function to set the dynamic queue limit minimum value.
Some specific drivers might have legitimate reasons to configure
dql.min_limit to a given value. Typically, this is the case when the
PDU of the protocol is smaller than the packet size to used to
carry those frames to the device.
Concr
omated testing systems.
> Lowering the timeout may also be useful for e.g. manual bisection.
> The default value matches the current behavior.
>
> Signed-off-by: Dmitry Vyukov
> Fixes: https://bugzilla.kernel.org/show_bug.cgi?id=211877
> Cc: netdev@vger.kernel.org
> Cc: li
Hello:
This patch was applied to netdev/net-next.git (refs/heads/master):
On Sun, 21 Mar 2021 01:04:45 +0200 you wrote:
> From: Vladimir Oltean
>
> This reverts commit e9bf96943b408e6c99dd13fb01cb907335787c61.
>
> The topic of the reverted patch is the support for switches wit
From: Vladimir Oltean
This reverts commit e9bf96943b408e6c99dd13fb01cb907335787c61.
The topic of the reverted patch is the support for switches with global
VLAN filtering, added by commit 061f6a505ac3 ("net: dsa: Add
ndo_vlan_rx_{add, kill}_vid implementation"). Be there a switch with 4
ports sw
From: Roi Dayan
Remove dedicated uplink rep netdev ndos and ethtools ops.
We will re-use the native NIC port net device instance and ethtool ops for
the Uplink representor.
Signed-off-by: Roi Dayan
Signed-off-by: Saeed Mahameed
---
.../net/ethernet/mellanox/mlx5/core/en_rep.c | 133
From: Roi Dayan
We will re-use the native NIC port net device instance for the Uplink
representor, hence same ndos must be used.
Signed-off-by: Roi Dayan
Reviewed-by: Saeed Mahameed
Signed-off-by: Saeed Mahameed
---
.../net/ethernet/mellanox/mlx5/core/en_main.c | 25 +++
.../
We are pleased to announce the opening of Call For
Submissions(CFS) for Netdev 0x15.
For overview of topics, submissions and requirements
please visit:
https://netdevconf.info/0x15/submit-proposal.html
For all submitted sessions, we employ a double blind
review process carried out by the
On Thu, 2021-03-11 at 13:33 -0400, Jason Gunthorpe wrote:
> On Wed, Mar 10, 2021 at 11:09:12PM -0800, Saeed Mahameed wrote:
> > From: Mark Bloch
> >
> > Now that a pointer to the managing E-Switch is stored in the
> > representor
> > use it.
> >
> > Signed-off-by: Mark Bloch
> > Reviewed-by: Sa
On 3/11/21 7:33 PM, Jason Gunthorpe wrote:
> On Wed, Mar 10, 2021 at 11:09:12PM -0800, Saeed Mahameed wrote:
>> From: Mark Bloch
>>
>> Now that a pointer to the managing E-Switch is stored in the representor
>> use it.
>>
>> Signed-off-by: Mark Bloch
>> Reviewed-by: Saeed Mahameed
>> Signed-off-
On Wed, Mar 10, 2021 at 11:09:12PM -0800, Saeed Mahameed wrote:
> From: Mark Bloch
>
> Now that a pointer to the managing E-Switch is stored in the representor
> use it.
>
> Signed-off-by: Mark Bloch
> Reviewed-by: Saeed Mahameed
> Signed-off-by: Saeed Mahameed
> ---
> drivers/infiniband/hw/
From: Mark Bloch
Now that a pointer to the managing E-Switch is stored in the representor
use it.
Signed-off-by: Mark Bloch
Reviewed-by: Saeed Mahameed
Signed-off-by: Saeed Mahameed
---
drivers/infiniband/hw/mlx5/fs.c | 2 +-
drivers/infiniband/hw/mlx5/ib_rep.c | 2 +-
drivers/infiniband
Add a function to set the dynamic queue limit minimum value.
This function is to be used by network drivers which are able to
prove, at least through empirical tests on several environment (with
other applications, heavy context switching, virtualization...), that
they constantly reach better perf
Hi!
> Add support for HW offloading of the netdev trigger.
>
> We need to change spinlock to mutex, because if spinlock is used, the
> trigger_offload() method cannot sleep, which can happen for ethernet
> PHYs.
Is that bugfix or just needed for offloading? Should be separate pa
On Fri 2020-10-30 12:44:30, Marek Behún wrote:
> Use bit fields members in struct led_netdev_data instead of one mode
> member and set_bit/clear_bit/test_bit functions. These functions are
> suitable for longer or variable length bit arrays.
They also provide atomicity guarantees. If you can expla
;
>> This is the same with commit 0fda7600c2e1 ("geneve: move debug check
after
>> netdev unregister")
>>
>> Test commands:
>> ip netns del A
>> ip netns add A
>> ip netns add B
>>
>> ip netns exec B ip link add vx
On Sun, 21 Feb 2021 15:45:52 + Taehee Yoo wrote:
> The debug check must be done after unregister_netdevice_many() call --
> the hlist_del_rcu() for this is done inside .ndo_stop.
>
> This is the same with commit 0fda7600c2e1 ("geneve: move debug check after
> netdev un
The debug check must be done after unregister_netdevice_many() call --
the hlist_del_rcu() for this is done inside .ndo_stop.
This is the same with commit 0fda7600c2e1 ("geneve: move debug check after
netdev unregister")
Test commands:
ip netns del A
ip netns add A
ip n
On 2021/2/11 5:40 上午, Michael S. Tsirkin wrote:
On Wed, Jan 13, 2021 at 04:08:57PM +0800, Xuan Zhuo wrote:
The number of queues implemented by many virtio backends is limited,
especially some machines have a large number of CPUs. In this case, it
is often impossible to allocate a separate queu
The current driver calls netif_carrier_off() late in the link tear down
which can result in a netdev watchdog timeout.
Calling netif_carrier_off() immediately after netif_tx_stop_all_queues()
avoids the warning.
[ cut here ]
NETDEV WATCHDOG: enp3s0f2 (amd-xgbe
On 2/12/21 12:00 PM, Shyam Sundar S K wrote:
Current driver calls the netif_carrier_off() during the later point in
time to tear down the link which causes the netdev watchdog to timeout.
This is a bit confusing... how about:
The current driver calls netif_carrier_off() late in the link tear
Current driver calls the netif_carrier_off() during the later point in
time to tear down the link which causes the netdev watchdog to timeout.
Calling netif_carrier_off() immediately after netif_tx_stop_all_queues()
would avoids the warning.
[ cut here ]
NETDEV WATCHDOG
On Wed, 10 Feb 2021 16:40:41 -0500
"Michael S. Tsirkin" wrote:
> On Wed, Jan 13, 2021 at 04:08:57PM +0800, Xuan Zhuo wrote:
> > The number of queues implemented by many virtio backends is limited,
> > especially some machines have a large number of CPUs. In this case, it
> > is often impossible t
On Wed, Jan 13, 2021 at 04:08:57PM +0800, Xuan Zhuo wrote:
> The number of queues implemented by many virtio backends is limited,
> especially some machines have a large number of CPUs. In this case, it
> is often impossible to allocate a separate queue for XDP_TX.
>
> This patch allows XDP_TX to
Point out where patchwork bot's code lives, and that we don't want
people posting stuff that doesn't build.
Signed-off-by: Jakub Kicinski
---
Documentation/networking/netdev-FAQ.rst | 16
1 file changed, 16 insertions(+)
diff --git a/Documentation/networking/ne
On Wed, Feb 03, 2021 at 06:54:55PM +0200, Vadym Kochan wrote:
> Move handling of PRECHANGEUPPER event from prestera_switchdev to
> prestera_main which is responsible for basic netdev events handling
> and routing them to related module.
>
> Signed-off-by: Vadym Kochan
>
On Thu, Feb 4, 2021 at 3:06 AM Kevin Hao wrote:
>
> In the current implementation of {netdev,napi}_alloc_frag(), it doesn't
> have any align guarantee for the returned buffer address, But for some
> hardwares they do require the DMA buffer to be aligned correctly,
> so we wo
In the current implementation of {netdev,napi}_alloc_frag(), it doesn't
have any align guarantee for the returned buffer address, But for some
hardwares they do require the DMA buffer to be aligned correctly,
so we would have to use some workarounds like below if the buffers
allocated b
On Tue, Feb 02, 2021 at 08:26:19AM -0800, Alexander Duyck wrote:
> On Sun, Jan 31, 2021 at 12:17 AM Kevin Hao wrote:
> >
> > In the current implementation of {netdev,napi}_alloc_frag(), it doesn't
> > have any align guarantee for the returned buffer address, But for
Move handling of PRECHANGEUPPER event from prestera_switchdev to
prestera_main which is responsible for basic netdev events handling
and routing them to related module.
Signed-off-by: Vadym Kochan
---
.../ethernet/marvell/prestera/prestera_main.c | 29 +--
.../marvell/prestera
Hello:
This series was applied to netdev/net-next.git (refs/heads/master):
On Mon, 1 Feb 2021 22:54:44 -0800 you wrote:
> From: Saeed Mahameed
>
> 1) Initialize netdevice features and structures on netdevice allocation
>and outside of the mlx5e profile.
>
> 2) As n
On Sun, Jan 31, 2021 at 12:17 AM Kevin Hao wrote:
>
> In the current implementation of {netdev,napi}_alloc_frag(), it doesn't
> have any align guarantee for the returned buffer address, But for some
> hardwares they do require the DMA buffer to be aligned correctly,
> so we wo
netdevices, and will allow replacing mlx5e profiles on the fly to reuse
the same netdevice for multiple profiles.
e.g. for uplink representor profile as shown in the following patch
4) Profile callbacks are not allowed to touch netdev->features directly
anymore, since in downstream patch we w
In the current implementation of {netdev,napi}_alloc_frag(), it doesn't
have any align guarantee for the returned buffer address, But for some
hardwares they do require the DMA buffer to be aligned correctly,
so we would have to use some workarounds like below if the buffers
allocated b
Hello:
This patch was applied to netdev/net.git (refs/heads/master):
On Tue, 26 Jan 2021 03:02:14 +0300 you wrote:
> On building the route there is an assumption that the destination
> could be local. In this case loopback_dev is used to get the address.
> If the address is still
On building the route there is an assumption that the destination
could be local. In this case loopback_dev is used to get the address.
If the address is still cannot be retrieved dn_route_output_slow
returns EADDRNOTAVAIL with loopback_dev reference taken.
Cannot find hash for the fixes tag becau
In the current implementation of {netdev,napi}_alloc_frag(), it doesn't
have any align guarantee for the returned buffer address, But for some
hardwares they do require the DMA buffer to be aligned correctly,
so we would have to use some workarounds like below if the buffers
allocated b
Introduce a new netdev feature, NETIF_F_GRO_UDP_FWD, to allow user
to turn UDP GRO on and off for forwarding.
Defaults to off to not change current datapath.
Suggested-by: Paolo Abeni
Signed-off-by: Alexander Lobakin
---
include/linux/netdev_features.h | 4 +++-
net/ethtool/common.c
Introduce a new netdev_feature, NETIF_F_GRO_UDP, to allow user
to turn UDP GRO on and off when there is no socket, e.g. when
forwarding.
Defaults to off to not change current datapath.
Suggested-by: Paolo Abeni
Signed-off-by: Alexander Lobakin
---
include/linux/netdev_features.h | 4 +++-
net/e
DP_TX to run by reuse the existing SQ with
> __netif_tx_lock() hold when there are not enough queues.
>
> Signed-off-by: Xuan Zhuo
> Reviewed-by: Dust Li
Since reviews are not coming in let me share some of mine.
nit: please put [PATCH net-next] not [PATCH netdev]
> -static str
The number of queues implemented by many virtio backends is limited,
especially some machines have a large number of CPUs. In this case, it
is often impossible to allocate a separate queue for XDP_TX.
This patch allows XDP_TX to run by reuse the existing SQ with
__netif_tx_lock() hold when there a
From: Vladimir Oltean
In the effort of making .ndo_get_stats64 be able to sleep, we need to
ensure the callers of dev_get_stats do not use atomic context.
In the case of the appldata driver, an RCU read-side critical section is
used to ensure the integrity of the list of network interfaces, beca
From: Vladimir Oltean
In the effort of making .ndo_get_stats64 be able to sleep, we need to
ensure the callers of dev_get_stats do not use atomic context.
The LED driver for HP-PARISC workstations uses a workqueue to
periodically check for updates in network interface statistics, and
flicker whe
From: Vladimir Oltean
In the effort of making .ndo_get_stats64 be able to sleep, we need to
ensure the callers of dev_get_stats do not use atomic context.
In the case of the appldata driver, an RCU read-side critical section is
used to ensure the integrity of the list of network interfaces, beca
From: Vladimir Oltean
In the effort of making .ndo_get_stats64 be able to sleep, we need to
ensure the callers of dev_get_stats do not use atomic context.
The LED driver for HP-PARISC workstations uses a workqueue to
periodically check for updates in network interface statistics, and
flicker whe
From: Vladimir Oltean
In the effort of making .ndo_get_stats64 be able to sleep, we need to
ensure the callers of dev_get_stats do not use atomic context.
The LED driver for HP-PARISC workstations uses a workqueue to
periodically check for updates in network interface statistics, and
flicker whe
From: Vladimir Oltean
In the effort of making .ndo_get_stats64 be able to sleep, we need to
ensure the callers of dev_get_stats do not use atomic context.
In the case of the appldata driver, an RCU read-side critical section is
used to ensure the integrity of the list of network interfaces, beca
From: Vladimir Oltean
In the effort of making .ndo_get_stats64 be able to sleep, we need to
ensure the callers of dev_get_stats do not use atomic context.
The LED driver for HP-PARISC workstations uses a workqueue to
periodically check for updates in network interface statistics, and
flicker whe
From: Vladimir Oltean
In the effort of making .ndo_get_stats64 be able to sleep, we need to
ensure the callers of dev_get_stats do not use atomic context.
In the case of the appldata driver, an RCU read-side critical section is
used to ensure the integrity of the list of network interfaces, beca
On 2021/1/6 上午10:55, Xuan Zhuo wrote:
On Wed, 6 Jan 2021 10:46:43 +0800, Jason Wang wrote:
On 2021/1/5 下午8:42, Xuan Zhuo wrote:
On Tue, 5 Jan 2021 17:32:19 +0800, Jason Wang wrote:
On 2021/1/5 下午5:11, Xuan Zhuo wrote:
The first patch made some adjustments to xsk.
Thanks a lot for the wor
On 2021/1/5 下午8:42, Xuan Zhuo wrote:
On Tue, 5 Jan 2021 17:32:19 +0800, Jason Wang wrote:
On 2021/1/5 下午5:11, Xuan Zhuo wrote:
The first patch made some adjustments to xsk.
Thanks a lot for the work. It's rather interesting.
The second patch itself can be used as an independent patch to
From: Vladimir Oltean
In the effort of making .ndo_get_stats64 be able to sleep, we need to
ensure the callers of dev_get_stats do not use atomic context.
The LED driver for HP-PARISC workstations uses a workqueue to
periodically check for updates in network interface statistics, and
flicker whe
From: Vladimir Oltean
In the effort of making .ndo_get_stats64 be able to sleep, we need to
ensure the callers of dev_get_stats do not use atomic context.
In the case of the appldata driver, an RCU read-side critical section is
used to ensure the integrity of the list of network interfaces, beca
Hi Xuan,
Thank you for the patch! Perhaps something to improve:
[auto build test WARNING on ipvs/master]
[also build test WARNING on linus/master v5.11-rc2 next-20210104]
[cannot apply to sparc-next/master]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submittin
Hi Xuan,
url:
https://github.com/0day-ci/linux/commits/Xuan-Zhuo/virtio-net-support-xdp-socket-zero-copy-xmit/20210105-171505
base: https://git.kernel.org/pub/scm/linux/kernel/git/horms/ipvs.git master
config: i386-randconfig-m021-20210105 (attached as .config)
compiler: gcc-9 (Debian 9.3.0
Hi Xuan,
Thank you for the patch! Yet something to improve:
[auto build test ERROR on ipvs/master]
[also build test ERROR on linus/master v5.11-rc2 next-20210104]
[cannot apply to sparc-next/master]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch,
On Tue, Jan 05, 2021 at 05:11:43PM +0800, Xuan Zhuo wrote:
> Virtio net support xdp socket.
>
> We should open the module param "napi_tx" for using this feature.
what does this imply exactly?
> In fact, various virtio implementations have some problems:
> 1. The tx interrupt may be lost
> 2. The
Hi Xuan,
Thank you for the patch! Yet something to improve:
[auto build test ERROR on ipvs/master]
[also build test ERROR on linus/master v5.11-rc2 next-20210104]
[cannot apply to sparc-next/master]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch,
On Tue, Jan 05, 2021 at 05:11:38PM +0800, Xuan Zhuo wrote:
> The first patch made some adjustments to xsk.
>
> The second patch itself can be used as an independent patch to solve the
> problem
> that XDP may fail to load when the number of queues is insufficient.
>
> The third to last patch imp
On 2021/1/5 下午5:11, Xuan Zhuo wrote:
The first patch made some adjustments to xsk.
Thanks a lot for the work. It's rather interesting.
The second patch itself can be used as an independent patch to solve the problem
that XDP may fail to load when the number of queues is insufficient.
xdp/xsk_buff_pool.c
@@ -167,12 +167,13 @@ static int __xp_assign_dev(struct xsk_buff_pool *pool,
bpf.command = XDP_SETUP_XSK_POOL;
bpf.xsk.pool = pool;
bpf.xsk.queue_id = queue_id;
+ bpf.xsk.need_dma = true;
err = netdev->netdev_ops->ndo_bpf(net
Virtio net support xdp socket.
We should open the module param "napi_tx" for using this feature.
In fact, various virtio implementations have some problems:
1. The tx interrupt may be lost
2. The tx interrupt may have a relatively large delay
This brings us to several questions:
1. Wakeup wakes
Split function free_old_xmit_skbs, add sub-function __free_old_xmit_ptr,
which is convenient to call with other statistical information, and
supports the parameter 'xsk_wakeup' required for processing xsk.
Use netif stop check as a function virtnet_sq_stop_check, which will be
used when adding xsk
If support xsk, a new ptr will be recovered during the
process of freeing the old ptr. In order to distinguish between ctx sent
by XDP_TX and ctx sent by xsk, a struct is added here to distinguish
between these two situations. virtnet_xdp_type.type It is used to
distinguish different ctx, and virtn
The number of queues implemented by many virtio backends is limited,
especially some machines have a large number of CPUs. In this case, it
is often impossible to allocate a separate queue for XDP_TX.
This patch allows XDP_TX to run by lock when not enough queue.
Signed-off-by: Xuan Zhuo
---
dr
The first patch made some adjustments to xsk.
The second patch itself can be used as an independent patch to solve the problem
that XDP may fail to load when the number of queues is insufficient.
The third to last patch implements support for xsk in virtio-net.
A practical problem with virtio is
While we bid goodbye to 2020, here's some good hopeful news for 2021.
cheers,
jamal
Forwarded Message
Subject: [NetDev-People] Announcing Netdev 0x15
Date: Sun, 27 Dec 2020 15:54:31 -0700
From: Tom Herbert via people
Reply-To: Tom Herbert
To: peo...@netdevconf.info
On Tue, 22 Dec 2020 21:54:21 -0500 Jeff Dike wrote:
> virtnet_set_channels can recursively call cpus_read_lock if CONFIG_XPS
> and CONFIG_HOTPLUG are enabled.
>
> The path is:
> virtnet_set_channels - calls get_online_cpus(), which is a trivial
> wrapper around cpus_read_lock()
> netif_set
1 - 100 of 2517 matches
Mail list logo
