[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CADEbmW0r4BCU_qWHRrvAPQ-kwA-xMDj2YD_OdiGotRnfEMtpoQ@mail.gmail.com>
Date: Mon, 4 Nov 2024 16:50:59 +0100
From: Michal Schmidt <mschmidt@...hat.com>
To: Aleksandr Loktionov <aleksandr.loktionov@...el.com>
Cc: intel-wired-lan@...ts.osuosl.org, anthony.l.nguyen@...el.com,
netdev@...r.kernel.org, Jan Sokolowski <jan.sokolowski@...el.com>,
Padraig J Connolly <padraig.j.connolly@...el.com>, maciej.fijalkowski@...el.com,
Przemek Kitszel <przemyslaw.kitszel@...el.com>
Subject: Re: [PATCH iwl-next v4444] i40e: add ability to reset VF for Tx and
Rx MDD events
On Mon, Nov 4, 2024 at 4:40 PM Michal Schmidt <mschmidt@...hat.com> wrote:
>
> On Tue, Oct 29, 2024 at 1:36 PM Aleksandr Loktionov
> <aleksandr.loktionov@...el.com> wrote:
[...]
> > +
> > + i40e_vc_reset_vf(vf, true);
> > }
> > }
> >
> > /* re-enable mdd interrupt cause */
> > clear_bit(__I40E_MDD_EVENT_PENDING, pf->state);
>
> Can you remove this 2nd clearing of the __I40E_MDD_EVENT_PENDING bit?
> If the interrupt handler detects a MDD event while we're still
> printing the message about the previous one, we don't want to forget
> it by clearing it here.
>
> Michal
Well, I suppose the race I described cannot happen because the
unmasking of ..._MAL_DETECT_MASK happens after this.
But it's still redundant to clear the bit twice.
Michal
> > reg = rd32(hw, I40E_PFINT_ICR0_ENA);
> > reg |= I40E_PFINT_ICR0_ENA_MAL_DETECT_MASK;
> > wr32(hw, I40E_PFINT_ICR0_ENA, reg);
> > i40e_flush(hw);
> > +
> > + i40e_print_vfs_mdd_events(pf);
> > }
> >
> > /**
Powered by blists - more mailing lists