On 4/18/24 13:25, Siddharth Vadapalli wrote:
On Thu, Apr 18, 2024 at 01:17:47PM +0200, Julien Panis wrote:
On 4/18/24 13:00, Siddharth Vadapalli wrote:
On 12-04-2024 21:08, Julien Panis wrote:
This patch adds XDP support to TI AM65 CPSW Ethernet driver.
The following features are implemented: NETDEV_XDP_ACT_BASIC,
NETDEV_XDP_ACT_REDIRECT, and NETDEV_XDP_ACT_NDO_XMIT.
Zero-copy and non-linear XDP buffer supports are NOT implemented.
Besides, the page pool memory model is used to get better performance.
Signed-off-by: Julien Panis <jpanis@xxxxxxxxxxxx>
Hello Julien,
This series crashes Linux on AM62ax SoC which also uses the
AM65-CPSW-NUSS driver:
https://gist.github.com/Siddharth-Vadapalli-at-TI/5ed0e436606001c247a7da664f75edee
Regards,
Siddharth.
Hello Siddharth.
Thanks for the log. I can read:
[ 1.966094] Missing net_device from driver
Did you check that nodes exist in the device tree for the net devices ?
Yes it exists. The device-tree used was also built with linux-next
tagged next-20240417. The node corresponding to eth0 is cpsw_port1 which
is present and enabled in the device-tree:
https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/tree/arch/arm64/boot/dts/ti/k3-am62a7-sk.dts?h=next-20240417#n644
Regards,
Siddharth.
I could reproduce the bug by disabling 'cpsw_port2' in my device tree,
which is 'k3-am625-sk.dts' for the board I use.
A condition is missing in am65_cpsw_create_xdp_rxqs() and
am65_cpsw_destroy_xdp_rxqs() functions.
For these 2 functions, the code which is in the for loop should be
run only when port ethX is enabled. That's why it crashes with
your device tree (cpsw_port2 is disabled, which is not the case by
default for the board I developed with).
I'll send a patch to fix the issue. Thanks for reporting it.
Julien
