[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID:
<TYCPR01MB1126998DEC9DC91F64993EDAF86462@TYCPR01MB11269.jpnprd01.prod.outlook.com>
Date: Tue, 6 Feb 2024 17:07:18 +0000
From: Biju Das <biju.das.jz@...renesas.com>
To: Sergey Shtylyov <s.shtylyov@....ru>, "David S. Miller"
<davem@...emloft.net>, Eric Dumazet <edumazet@...gle.com>, Jakub Kicinski
<kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>
CC: Claudiu Beznea <claudiu.beznea.uj@...renesas.com>, Yoshihiro Shimoda
<yoshihiro.shimoda.uh@...esas.com>, Wolfram Sang
<wsa+renesas@...g-engineering.com>, nikita.yoush
<nikita.yoush@...entembedded.com>, "netdev@...r.kernel.org"
<netdev@...r.kernel.org>, "linux-renesas-soc@...r.kernel.org"
<linux-renesas-soc@...r.kernel.org>, Geert Uytterhoeven
<geert+renesas@...der.be>, Prabhakar Mahadev Lad
<prabhakar.mahadev-lad.rj@...renesas.com>, biju.das.au
<biju.das.au@...il.com>
Subject: RE: [PATCH v4 net-next 1/2] ravb: Add Rx checksum offload support for
GbEth
Hi Sergey Shtylyov,
Thanks for the feedback.
> -----Original Message-----
> From: Sergey Shtylyov <s.shtylyov@....ru>
> Sent: Monday, February 5, 2024 7:01 PM
> Subject: Re: [PATCH v4 net-next 1/2] ravb: Add Rx checksum offload support
> for GbEth
>
> On 2/3/24 5:25 PM, Biju Das wrote:
>
> > TOE has hardware support for calculating IP header and TCP/UDP/ICMP
> > checksum for both IPv4 and IPv6.
> >
> > Add Rx checksum offload supported by TOE for IPv4 and TCP/UDP protocols.
> >
> > For Rx, the 4-byte result of checksum calculation is attached to the
> > Ethernet frames.First 2-bytes is result of IPv4 header checksum and
> > next 2-bytes is TCP/UDP/ICMP checksum.
> >
> > If a frame does not have checksum error, 0x0000 is attached as
> > checksum calculation result. For unsupported frames 0xFFFF is attached
> > as checksum calculation result. In case of an IPv6 packet, IPv4
> > checksum is always set to 0xFFFF.
> >
> > We can test this functionality by the below commands
> >
> > ethtool -K eth0 rx on --> to turn on Rx checksum offload ethtool -K
> > eth0 rx off --> to turn off Rx checksum offload
> >
> > Signed-off-by: Biju Das <biju.das.jz@...renesas.com>
> [...]
>
> > diff --git a/drivers/net/ethernet/renesas/ravb.h
> > b/drivers/net/ethernet/renesas/ravb.h
> > index e0f8276cffed..64bf29d01ad0 100644
> > --- a/drivers/net/ethernet/renesas/ravb.h
> > +++ b/drivers/net/ethernet/renesas/ravb.h
> > @@ -205,7 +205,10 @@ enum ravb_reg {
> > TLFRCR = 0x0758,
> > RFCR = 0x0760,
> > MAFCR = 0x0778,
> > - CSR0 = 0x0800, /* RZ/G2L only */
> > +
> > + /* RZ/G2L TOE registers */
>
> Thanks. Though I think I'd prefer /* TOE registers (RZ/G2L only) */...
OK.
>
> [...]
> > @@ -978,6 +981,21 @@ enum CSR0_BIT {
> > CSR0_RPE = 0x00000020,
> > };
> >
> > +enum CSR2_BIT {
> > + CSR2_RIP4 = 0x00000001,
> > + CSR2_RTCP4 = 0x00000010,
> > + CSR2_RUDP4 = 0x00000020,
> > + CSR2_RICMP4 = 0x00000040,
> > + CSR2_RTCP6 = 0x00100000,
> > + CSR2_RUDP6 = 0x00200000,
> > + CSR2_RICMP6 = 0x00400000,
> > + CSR2_RHOP = 0x01000000,
> > + CSR2_RROUT = 0x02000000,
> > + CSR2_RAHD = 0x04000000,
> > + CSR2_RDHD = 0x08000000,
> > + CSR2_ALL = 0x0F700071,
>
> I doubt we really need CSR2_ALL...
Will remove it.
>
> [...]
> > diff --git a/drivers/net/ethernet/renesas/ravb_main.c
> > b/drivers/net/ethernet/renesas/ravb_main.c
> > index 0e3731f50fc2..4f310bcee7c0 100644
> > --- a/drivers/net/ethernet/renesas/ravb_main.c
> > +++ b/drivers/net/ethernet/renesas/ravb_main.c
> > @@ -522,6 +522,24 @@ static int ravb_ring_init(struct net_device *ndev,
> int q)
> > return -ENOMEM;
> > }
> >
> > +static void ravb_csum_init_gbeth(struct net_device *ndev) {
> > + if (!(ndev->features & NETIF_F_RXCSUM))
> > + goto done;
> > +
> > + ravb_write(ndev, 0, CSR0);
> > + if (ravb_wait(ndev, CSR0, CSR0_RPE, 0)) {
> > + netdev_err(ndev, "Timeout enabling hardware checksum\n");
> > + ndev->features &= ~NETIF_F_RXCSUM;
> > + } else {
> > + ravb_write(ndev, CSR2_ALL & ~(CSR2_RTCP6 | CSR2_RUDP6 |
> > + CSR2_RICMP6), CSR2);
>
> With these 3 bits being 0, the bits 24...27 are ignored anyway, the
> manual says. So I think I'd prefer:
>
> ravb_write(ndev, CSR2_RIP4 | CSR2_RTCP4 | CSR2_RUDP4 |
> CSR2_RICMP4,
> CSR2);
OK.
>
> > + }
> > +
> > +done:
> > + ravb_write(ndev, CSR0_TPE | CSR0_RPE, CSR0);
>
> I think we shouldn't set CSR0.TPE yet at this point, as we n't setup
> CSR1 yet...
This will cause regression. As board won't boot as tx fails. So CSR0.{TPE,RPE}
is must.
>
> [...]
> > @@ -2334,11 +2381,48 @@ static void ravb_set_rx_csum(struct net_device
> > *ndev, bool enable)
> [...]
> > static int ravb_set_features_gbeth(struct net_device *ndev,
> > netdev_features_t features)
> > {
> > - /* Place holder */
> > - return 0;
> > + netdev_features_t changed = ndev->features ^ features;
> > + struct ravb_private *priv = netdev_priv(ndev);
> > + unsigned long flags;
> > + int ret = 0;
> > + u32 val;
> > +
> > + spin_lock_irqsave(&priv->lock, flags);
> > + if (changed & NETIF_F_RXCSUM) {
> > + if (features & NETIF_F_RXCSUM)
> > + val = CSR2_ALL & ~(CSR2_RTCP6 | CSR2_RUDP6 |
> CSR2_RICMP6);
>
> Likewise, I'd prefer:
>
> val = CSR2_RIP4 | CSR2_RTCP4 | CSR2_RUDP4 | CSR2_RICMP4;
Ok.
>
> > + else
> > + val = 0;
> > +
> > + ret = ravb_endisable_csum_gbeth(ndev, CSR2, val, CSR0_RPE);
> > + if (ret)
> > + goto done;
> > + }
> > +
> > + ndev->features = features;
> > +done:
> > + spin_unlock_irqrestore(&priv->lock, flags);
> > +
> > + return ret;
> > }
> >
> > static int ravb_set_features_rcar(struct net_device *ndev,
> [...]
>
> Otherwise LGTM. We're close! :-)
I will send v5, based on above.
Cheers,
Biju
Powered by blists - more mailing lists