| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <OS0PR01MB5922B9698903E199C211C57A86AE9@OS0PR01MB5922.jpnprd01.prod.outlook.com>
Date: Mon, 4 Oct 2021 14:56:33 +0000
From: Biju Das <biju.das.jz@...renesas.com>
To: Sergey Shtylyov <s.shtylyov@....ru>,
"David S. Miller" <davem@...emloft.net>,
Jakub Kicinski <kuba@...nel.org>
CC: Prabhakar Mahadev Lad <prabhakar.mahadev-lad.rj@...renesas.com>,
Andrew Lunn <andrew@...n.ch>,
Sergei Shtylyov <sergei.shtylyov@...il.com>,
Geert Uytterhoeven <geert+renesas@...der.be>,
Adam Ford <aford173@...il.com>,
Yoshihiro Shimoda <yoshihiro.shimoda.uh@...esas.com>,
"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
"linux-renesas-soc@...r.kernel.org"
<linux-renesas-soc@...r.kernel.org>,
Chris Paterson <Chris.Paterson2@...esas.com>,
Biju Das <biju.das@...renesas.com>
Subject: RE: [RFC/PATCH 16/18] ravb: Add Packet receive function for Gigabit
Ethernet
Hi Sergey,
Thanks for the feedback.
> Subject: Re: [RFC/PATCH 16/18] ravb: Add Packet receive function for
> Gigabit Ethernet
>
> On 9/23/21 5:08 PM, Biju Das wrote:
>
> > This patch series adds RX(packet receive) function for Gigabit
> > Ethernet found on RZ/G2L SoC.
> >
> > Signed-off-by: Biju Das <biju.das.jz@...renesas.com>
> > ---
> > drivers/net/ethernet/renesas/ravb.h | 1 +
> > drivers/net/ethernet/renesas/ravb_main.c | 157
> > ++++++++++++++++++++++-
> > 2 files changed, 156 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/net/ethernet/renesas/ravb.h
> > b/drivers/net/ethernet/renesas/ravb.h
> > index b0e067a6a8ee..85260f89e1cd 100644
> > --- a/drivers/net/ethernet/renesas/ravb.h
> > +++ b/drivers/net/ethernet/renesas/ravb.h
> > @@ -1092,6 +1092,7 @@ struct ravb_private {
> >
> > int duplex;
> > struct ravb_rx_desc *rgeth_rx_ring[NUM_RX_QUEUE];
> > + struct sk_buff *rxtop_skb;
>
> I'd prefer for this one declared earler in the *struct*, as well. And
> why not e.g 'rx_1st_skb'?
OK Will use 'rx_1st_skb' in the next RFC version.
>
> >
> > const struct ravb_hw_info *info;
> > struct reset_control *rstc;
> > diff --git a/drivers/net/ethernet/renesas/ravb_main.c
> > b/drivers/net/ethernet/renesas/ravb_main.c
> > index a08da7a37b92..867e180e6655 100644
> > --- a/drivers/net/ethernet/renesas/ravb_main.c
> > +++ b/drivers/net/ethernet/renesas/ravb_main.c
> > @@ -705,6 +705,23 @@ static void ravb_get_tx_tstamp(struct net_device
> *ndev)
> > }
> > }
> >
> > +static void ravb_rx_csum_rgeth(struct sk_buff *skb) {
> > + u8 *hw_csum;
> > +
> > + /* The hardware checksum is contained in sizeof(__sum16) (2) bytes
> > + * appended to packet data
> > + */
> > + if (unlikely(skb->len < sizeof(__sum16)))
> > + return;
> > + hw_csum = skb_tail_pointer(skb) - sizeof(__sum16);
> > +
> > + if (*hw_csum == 0)
> > + skb->ip_summed = CHECKSUM_UNNECESSARY;
> > + else
> > + skb->ip_summed = CHECKSUM_NONE;
>
> Mhm, what's the point of this whole function then? Why it can't be a
> copy of the R-Car analog?
This function should come after CSR initialization. Next RFC I will taken care of this.
*hwcsum == 0 means, there is no error in checksum. So it should be OK.
>
> [...]
> > @@ -720,11 +737,147 @@ static void ravb_rx_csum(struct sk_buff *skb)
> [...]
> > /* Packet receive function for Gigabit Ethernet */ static bool
> > ravb_rgeth_rx(struct net_device *ndev, int *quota, int q) {
> > - /* Place holder */
> > - return true;
> > + struct ravb_private *priv = netdev_priv(ndev);
> > + int entry = priv->cur_rx[q] % priv->num_rx_ring[q];
> > + int boguscnt = priv->dirty_rx[q] + priv->num_rx_ring[q] - priv-
> >cur_rx[q];
> > + struct net_device_stats *stats = &priv->stats[q];
>
> [q] should be dropped, as we've agreed...
nc_queue feature bit will take care of this. So the q will be always RVAB_BE.
>
> > + struct ravb_rx_desc *desc;
> > + struct sk_buff *skb;
> > + dma_addr_t dma_addr;
> > + u8 desc_status;
> > + u8 die_dt;
> > + u16 pkt_len;
> > + int limit;
> > +
> > + boguscnt = min(boguscnt, *quota);
> > + limit = boguscnt;
> > + desc = &priv->rgeth_rx_ring[q][entry];
> > + while (desc->die_dt != DT_FEMPTY) {
> > + /* Descriptor type must be checked before all other reads */
> > + dma_rmb();
> > + desc_status = desc->msc;
> > + pkt_len = le16_to_cpu(desc->ds_cc) & RX_DS;
> > +
> > + if (--boguscnt < 0)
> > + break;
> > +
> > + /* We use 0-byte descriptors to mark the DMA mapping errors */
> > + if (!pkt_len)
> > + continue;
> > +
> > + if (desc_status & MSC_MC)
> > + stats->multicast++;
> > +
> > + if (desc_status & (MSC_CRC | MSC_RFE | MSC_RTSF | MSC_RTLF |
> MSC_CEEF)) {
> > + stats->rx_errors++;
> > + if (desc_status & MSC_CRC)
> > + stats->rx_crc_errors++;
> > + if (desc_status & MSC_RFE)
> > + stats->rx_frame_errors++;
> > + if (desc_status & (MSC_RTLF | MSC_RTSF))
> > + stats->rx_length_errors++;
> > + if (desc_status & MSC_CEEF)
> > + stats->rx_missed_errors++;
> > + } else {
> > + die_dt = desc->die_dt & 0xF0;
> > + switch (die_dt) {
> > + case DT_FSINGLE:
> > + skb = ravb_get_skb_rgeth(ndev, q, entry, desc);
> > + skb_put(skb, pkt_len);
> > + skb->protocol = eth_type_trans(skb, ndev);
> > + if (ndev->features & NETIF_F_RXCSUM)
> > + ravb_rx_csum_rgeth(skb);
> > + napi_gro_receive(&priv->napi[q], skb);
> > + stats->rx_packets++;
> > + stats->rx_bytes += pkt_len;
> > + break;
> > + case DT_FSTART:
> > + priv->rxtop_skb = ravb_get_skb_rgeth(ndev, q,
> entry, desc);
>
> But don't you need to copy the data in this case?
It is not required, here.
>
> > + skb_put(priv->rxtop_skb, pkt_len);
> > + break;
> > + case DT_FMID:
> > + skb = ravb_get_skb_rgeth(ndev, q, entry, desc);
> > + skb_copy_to_linear_data_offset(priv->rxtop_skb,
> > + priv->rxtop_skb->len,
> > + skb->data,
> > + pkt_len);
> > + skb_put(priv->rxtop_skb, pkt_len);
> > + dev_kfree_skb(skb);
> > + break;
> > + case DT_FEND:
> > + skb = ravb_get_skb_rgeth(ndev, q, entry, desc);
> > + skb_copy_to_linear_data_offset(priv->rxtop_skb,
> > + priv->rxtop_skb->len,
> > + skb->data,
> > + pkt_len);
> > + skb_put(priv->rxtop_skb, pkt_len);
> > + dev_kfree_skb(skb);
> > + priv->rxtop_skb->protocol =
> > + eth_type_trans(priv->rxtop_skb, ndev);
> > + if (ndev->features & NETIF_F_RXCSUM)
> > + ravb_rx_csum_rgeth(skb);
> > + napi_gro_receive(&priv->napi[q],
> > + priv->rxtop_skb);
> > + stats->rx_packets++;
> > + stats->rx_bytes += priv->rxtop_skb->len;
> > + break;
> > + }
> > + }
> > +
> > + entry = (++priv->cur_rx[q]) % priv->num_rx_ring[q];
> > + desc = &priv->rgeth_rx_ring[q][entry];
> > + }
> > +
> > + /* Refill the RX ring buffers. */
> > + for (; priv->cur_rx[q] - priv->dirty_rx[q] > 0; priv->dirty_rx[q]++)
> {
> > + entry = priv->dirty_rx[q] % priv->num_rx_ring[q];
> > + desc = &priv->rgeth_rx_ring[q][entry];
> > + desc->ds_cc = cpu_to_le16(RGETH_RX_DESC_DATA_SIZE);
> > +
> > + if (!priv->rx_skb[q][entry]) {
> > + skb = netdev_alloc_skb(ndev,
> > + RGETH_RX_BUFF_MAX + RAVB_ALIGN - 1);
>
> ALIGN(RGETH_RX_BUFF_MAX, RAVB_ALIGN)?
It is taken care in next RFC.
Regards,
Biju
>
> [...]
>
> MBR, Sergey
Powered by blists - more mailing lists