Re: [Intel-wired-lan] [PATCH iwl-next v3] igb: Retrieve Tx timestamp directly from interrupt for i210

[Vadim Fedorenko]

On 05/02/2026 10:37, Loktionov, Aleksandr wrote:
> > -----Original Message-----
> > From: Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx>
> > Sent: Thursday, February 5, 2026 11:04 AM
> > To: Loktionov, Aleksandr <aleksandr.loktionov@xxxxxxxxx>
> > Cc: Kurt Kanzenbach <kurt@xxxxxxxxxxxxx>; Nguyen, Anthony L
> > <anthony.l.nguyen@xxxxxxxxx>; Kitszel, Przemyslaw
> > <przemyslaw.kitszel@xxxxxxxxx>; Paul Menzel <pmenzel@xxxxxxxxxxxxx>;
> > Vadim Fedorenko <vadim.fedorenko@xxxxxxxxx>; Gomes, Vinicius
> > <vinicius.gomes@xxxxxxxxx>; netdev@xxxxxxxxxxxxxxx; Richard Cochran
> > <richardcochran@xxxxxxxxx>; linux-kernel@xxxxxxxxxxxxxxx; Andrew Lunn
> > <andrew+netdev@xxxxxxx>; Eric Dumazet <edumazet@xxxxxxxxxx>; intel-
> > wired-lan@xxxxxxxxxxxxxxxx; Keller, Jacob E
> > <jacob.e.keller@xxxxxxxxx>; Jakub Kicinski <kuba@xxxxxxxxxx>; Paolo
> > Abeni <pabeni@xxxxxxxxxx>; David S. Miller <davem@xxxxxxxxxxxxx>
> > Subject: Re: RE: [Intel-wired-lan] [PATCH iwl-next v3] igb: Retrieve
> > Tx timestamp directly from interrupt for i210
> >
> > On 2026-02-05 09:47:14 [+0000], Loktionov, Aleksandr wrote:
> > …
> > > > --- a/drivers/net/ethernet/intel/igb/igb_ptp.c
> > > > +++ b/drivers/net/ethernet/intel/igb/igb_ptp.c
> > > > @@ -796,6 +796,28 @@ static int igb_ptp_verify_pin(struct
> > …
> > > > +	igb_ptp_tx_hwtstamp(adapter); <-Calls existing function
> > designed for work queue!
> > > skb_tstamp_tx() can sleep
> > > Smells like sleep-in-atomic isn't it?
> >
> > How or where can it sleep?
> >
> > Sebastian
>
> igb_ptp_tx_hwtstamp() -> https://elixir.bootlin.com/linux/v6.19-rc5/C/ident/skb_tstamp_tx -> https://elixir.bootlin.com/linux/v6.19-rc5/C/ident/__skb_complete_tx_timestamp -> https://elixir.bootlin.com/linux/v6.19-rc5/C/ident/sock_queue_err_skb -> https://elixir.bootlin.com/linux/v6.19-rc5/C/ident/skb_queue_tail -> https://elixir.bootlin.com/linux/v6.19-rc5/source/net/core/skbuff.c#L4075
>
> spin_lock_irqsave(&wq_head->lock, flags);  <- RT mutex can sleep

Hmm... that actually means we have some drivers broken for RT kernels if
they are processing TX timestamps within a single irq vector:
- hisilicon/hns3
- intel/i40e (and ice probably)
- marvell/mvpp2

For igb/igc/i40e it's still OK to process TX timestamps directly in
MSI-X configuration, as ring processing has separate vector, right?

But in general skb_tstamp_tx should be moved to BH processing (NAPI poll
callback).