Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230321065826.2044-1-wsa+renesas@sang-engineering.com> <79d945a4-e105-4bc4-3e73-64971731660e@omp.ru>
In-Reply-To: <79d945a4-e105-4bc4-3e73-64971731660e@omp.ru>
From:   Geert Uytterhoeven <geert@linux-m68k.org>
Date:   Thu, 23 Mar 2023 09:32:27 +0100
Message-ID: <CAMuHMdUt_kTH3tnrdF=oKBLyjrstei8PLsyr+dFXVoPEyxTLAA@mail.gmail.com>
Subject: Re: [PATCH net-next] sh_eth: remove open coded netif_running()
To:     Sergey Shtylyov <s.shtylyov@omp.ru>
Cc:     Wolfram Sang <wsa+renesas@sang-engineering.com>,
        netdev@vger.kernel.org, linux-renesas-soc@vger.kernel.org,
        Geert Uytterhoeven <geert+renesas@glider.be>,
        "David S. Miller" <davem@davemloft.net>,
        Eric Dumazet <edumazet@google.com>,
        Jakub Kicinski <kuba@kernel.org>,
        Paolo Abeni <pabeni@redhat.com>, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 8BIT
Precedence: bulk

Hi Sergey,

On Wed, Mar 22, 2023 at 9:54 PM Sergey Shtylyov <s.shtylyov@omp.ru> wrote:
> On 3/21/23 9:58 AM, Wolfram Sang wrote:
> > It had a purpose back in the days, but today we have a handy helper.
>
>    Well, the is_opened flag doesn't get set/cleared at the same time as
> __LINK_STATE_START. I'm not sure they are interchangeable...
>
> > Reported-by: Geert Uytterhoeven <geert+renesas@glider.be>
> > Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>

> > --- a/drivers/net/ethernet/renesas/sh_eth.c
> > +++ b/drivers/net/ethernet/renesas/sh_eth.c
> > @@ -2441,8 +2441,6 @@ static int sh_eth_open(struct net_device *ndev)
> >
> >       netif_start_queue(ndev);
> >
> > -     mdp->is_opened = 1;
> > -
>
>    __LINK_STATE_START gets set before the ndo_open() method call, so
> before the RPM call that enbales the clocks...
>
> >       return ret;
> >
> >  out_free_irq:
> > @@ -2565,7 +2563,7 @@ static struct net_device_stats *sh_eth_get_stats(struct net_device *ndev)
> >       if (mdp->cd->no_tx_cntrs)
> >               return &ndev->stats;
> >
> > -     if (!mdp->is_opened)
> > +     if (!netif_running(ndev))
> >               return &ndev->stats;
>
>    I guess mdp->is_opened is checked here to avoid reading the counter
> regs when RPM hasn't been called yet to enable the clocks...

Exactly, cfr. commit 7fa2955ff70ce453 ("sh_eth: Fix sleeping function
called from invalid context").

So you mean sh_eth_get_stats() can now be called after setting
__LINK_STATE_START, but before RPM has enabled the clocks?
Is there some protection against parallel execution of ndo_open()
and get_stats()?

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds