similar to: [PATCH] virtio_net: Unregister and re-register xdp_rxq across freeze/restore

On Fri, Jun 05, 2020 at 02:46:24PM -0700, Sean Christopherson wrote: > Unregister each queue's xdp_rxq during freeze, and re-register the new > instance during restore. All queues are released during free and > recreated during restore, i.e. the pre-freeze xdp_rxq will be lost. > > The bug is detected by WARNs in xdp_rxq_info_unreg() and > xdp_rxq_info_unreg_mem_model()

When initializing XDP in virtnet_open(), some rq xdp initialization may hit an error causing net device open failed. However, previous rqs have already initialized XDP and enabled NAPI, which is not the expected behavior. Need to roll back the previous rq initialization to avoid leaks in error unwinding of init code. Also extract helper functions of disable and enable queue pairs. Use newly

? 2023/5/9 09:43, Xuan Zhuo ??: > On Mon, 8 May 2023 11:00:10 -0400, Feng Liu <feliu at nvidia.com> wrote: >> >> On 2023-05-07 p.m.9:45, Xuan Zhuo wrote: >>> External email: Use caution opening links or attachments >>> >>> >>> On Sat, 6 May 2023 08:08:02 -0400, Feng Liu <feliu at nvidia.com> wrote: >>>> >>>> On

On Mon, 8 May 2023 11:00:10 -0400, Feng Liu <feliu at nvidia.com> wrote: > > > On 2023-05-07 p.m.9:45, Xuan Zhuo wrote: > > External email: Use caution opening links or attachments > > > > > > On Sat, 6 May 2023 08:08:02 -0400, Feng Liu <feliu at nvidia.com> wrote: > >> > >> > >> On 2023-05-05 p.m.10:33, Xuan Zhuo wrote: >

On Tue, 2 May 2023 20:35:25 -0400, Feng Liu <feliu at nvidia.com> wrote: > When initializing XDP in virtnet_open(), some rq xdp initialization > may hit an error causing net device open failed. However, previous > rqs have already initialized XDP and enabled NAPI, which is not the > expected behavior. Need to roll back the previous rq initialization > to avoid leaks in error

When initializing XDP in virtnet_open(), some rq xdp initialization may hit an error causing net device open failed. However, previous rqs have already initialized XDP and enabled NAPI, which is not the expected behavior. Need to roll back the previous rq initialization to avoid leaks in error unwinding of init code. Also extract a helper function of disable queue pairs, and use newly introduced

When initializing XDP in virtnet_open(), some rq xdp initialization may hit an error causing net device open failed. However, previous rqs have already initialized XDP and enabled NAPI, which is not the expected behavior. Need to roll back the previous rq initialization to avoid leaks in error unwinding of init code. Fixes: 754b8a21a96d ("virtio_net: setup xdp_rxq_info") Signed-off-by:

On 2023-05-05 p.m.10:33, Xuan Zhuo wrote: > External email: Use caution opening links or attachments > > > On Tue, 2 May 2023 20:35:25 -0400, Feng Liu <feliu at nvidia.com> wrote: >> When initializing XDP in virtnet_open(), some rq xdp initialization >> may hit an error causing net device open failed. However, previous >> rqs have already initialized XDP and

On Tue, May 02, 2023 at 01:41:34PM -0400, Feng Liu wrote: > When initializing XDP in virtnet_open(), some rq xdp initialization > may hit an error causing net device open failed. However, previous > rqs have already initialized XDP and enabled NAPI, which is not the > expected behavior. Need to roll back the previous rq initialization > to avoid leaks in error unwinding of init

When initializing XDP in virtnet_open(), some rq xdp initialization may hit an error causing net device open failed. However, previous rqs have already initialized XDP and enabled NAPI, which is not the expected behavior. Need to roll back the previous rq initialization to avoid leaks in error unwinding of init code. Also extract a helper function of disable queue pairs, and use newly introduced

The virtio_net driver doesn't dynamically change the RX-ring queue layout and backing pages, but instead reject XDP setup if all the conditions for XDP is not meet. Thus, the xdp_rxq_info also remains fairly static. This allow us to simply add the reg/unreg to net_device open/close functions. Driver hook points for xdp_rxq_info: * reg : virtnet_open * unreg: virtnet_close V3: - bugfix,

The virtio_net driver doesn't dynamically change the RX-ring queue layout and backing pages, but instead reject XDP setup if all the conditions for XDP is not meet. Thus, the xdp_rxq_info also remains fairly static. This allow us to simply add the init+reg/unreg to net_device open/close functions. Driver hook points for xdp_rxq_info: * init+reg: virtnet_open * unreg : virtnet_close Cc:

The virtio_net driver doesn't dynamically change the RX-ring queue layout and backing pages, but instead reject XDP setup if all the conditions for XDP is not meet. Thus, the xdp_rxq_info also remains fairly static. This allow us to simply add the init+reg/unreg to net_device open/close functions. Driver hook points for xdp_rxq_info: * init+reg: virtnet_open * unreg : virtnet_close Cc:

The virtio_net driver doesn't dynamically change the RX-ring queue layout and backing pages, but instead reject XDP setup if all the conditions for XDP is not meet. Thus, the xdp_rxq_info also remains fairly static. This allow us to simply add the reg/unreg to net_device open/close functions. Driver hook points for xdp_rxq_info: * reg : virtnet_open * unreg: virtnet_close V3: - bugfix,

The virtio_net driver doesn't dynamically change the RX-ring queue layout and backing pages, but instead reject XDP setup if all the conditions for XDP is not meet. Thus, the xdp_rxq_info also remains fairly static. This allow us to simply add the reg/unreg to net_device open/close functions. Driver hook points for xdp_rxq_info: * reg : virtnet_open * unreg: virtnet_close Cc:

Use newly introduced helper function that exactly does the same of closing the queue pairs. Signed-off-by: Feng Liu <feliu at nvidia.com> Reviewed-by: William Tu <witu at nvidia.com> Reviewed-by: Parav Pandit <parav at nvidia.com> --- drivers/net/virtio_net.c | 7 ++----- 1 file changed, 2 insertions(+), 5 deletions(-) diff --git a/drivers/net/virtio_net.c

Hi, This two small patches improves ndo_close() callback to follow the mirror sequence of ndo_open() callback. This improves the code auditing and also ensure that xdp rxq info is not unregistered while NAPI on RXQ is ongoing. Please review. Patch summary: patch-1 ensures that xdp rq info is unregistered after rq napi is disabled patch-2 keeps the mirror sequence for close() be mirror of open()

Cited commit in fixes tag frees rxq xdp info while RQ NAPI is still enabled and packet processing may be ongoing. Follow the mirror sequence of open() in the stop() callback. This ensures that when rxq info is unregistered, no rx packet processing is ongoing. Fixes: 754b8a21a96d ("virtio_net: setup xdp_rxq_info") Signed-off-by: Parav Pandit <parav at nvidia.com> ---

When sending control commands, virtio net sets up several buffers for DMA. The buffers are all part of the net device which means it's actually allocated by kvmalloc so it's in theory (on extreme memory pressure) possible to get a vmalloc'ed buffer which on some platforms means we can't DMA there. Fix up by moving the DMA buffers into a separate structure. Reported-by: Mikulas

After removing the reset function, the freeze and restore functions are now unused when CONFIG_PM_SLEEP is disabled: drivers/net/virtio_net.c:1881:12: error: 'virtnet_restore_up' defined but not used [-Werror=unused-function] static int virtnet_restore_up(struct virtio_device *vdev) drivers/net/virtio_net.c:1859:13: error: 'virtnet_freeze_down' defined but not used