Re: [net PATCH] net: macb: Relocate mog_init_rings() callback from macb_mac_link_up() to macb_open()

[Paolo Abeni]

On 11/28/25 11:36 AM, Xiaolei Wang wrote:
> In the non-RT kernel, local_bh_disable() merely disables preemption,
> whereas it maps to an actual spin lock in the RT kernel. Consequently,
> when attempting to refill RX buffers via netdev_alloc_skb() in
> macb_mac_link_up(), a deadlock scenario arises as follows:
>   The dependency chain caused by macb_mac_link_up():
>   &bp->lock --> (softirq_ctrl.lock) --> _xmit_ETHER#2

I'm sorry, but I can't see how this dependency chain is caused by
mog_init_rings(), please extend the above info pin pointing the
function/code effectively acquiring the lock and how it's reached.

>   The dependency chain caused by macb_start_xmit():
>   _xmit_ETHER#2 --> &bp->lock
> 
> Notably, invoking the mog_init_rings() callback upon link establishment
> is unnecessary. Instead, we can exclusively call mog_init_rings() within
> the ndo_open() callback. This adjustment resolves the deadlock issue.
> Given that mog_init_rings() is only applicable to
> non-MACB_CAPS_MACB_IS_EMAC cases, we can simply move it to macb_open()
> and simultaneously eliminate the MACB_CAPS_MACB_IS_EMAC check.
> 
> Suggested-by: Kevin Hao <kexin.hao@xxxxxxxxxxxxx>
> Signed-off-by: Xiaolei Wang <xiaolei.wang@xxxxxxxxxxxxx>

Please include a suitable fixes tag.

Thanks,

Paolo