1. Patchew
  2. linux
  3. [PATCH net-next v9 0/9] Device memory TCP TX
  4. View patch

[PATCH net-next v9 7/9] gve: add netmem TX support to GVE DQO-RDA mode

Mina Almasry posted 9 patches 8 months ago
There is a newer version of this series
[PATCH net-next v9 7/9] gve: add netmem TX support to GVE DQO-RDA mode
Posted by Mina Almasry 8 months ago 
Use netmem_dma_*() helpers in gve_tx_dqo.c DQO-RDA paths to
enable netmem TX support in that mode.

Declare support for netmem TX in GVE DQO-RDA mode.

Signed-off-by: Mina Almasry <almasrymina@google.com>

---

v4:
- New patch
---
 drivers/net/ethernet/google/gve/gve_main.c   | 4 ++++
 drivers/net/ethernet/google/gve/gve_tx_dqo.c | 8 +++++---
 2 files changed, 9 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/google/gve/gve_main.c b/drivers/net/ethernet/google/gve/gve_main.c
index 8aaac9101377..430314225d4d 100644
--- a/drivers/net/ethernet/google/gve/gve_main.c
+++ b/drivers/net/ethernet/google/gve/gve_main.c
@@ -2665,6 +2665,10 @@ static int gve_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 
 	dev_info(&pdev->dev, "GVE version %s\n", gve_version_str);
 	dev_info(&pdev->dev, "GVE queue format %d\n", (int)priv->queue_format);
+
+	if (!gve_is_gqi(priv) && !gve_is_qpl(priv))
+		dev->netmem_tx = true;
+
 	gve_clear_probe_in_progress(priv);
 	queue_work(priv->gve_wq, &priv->service_task);
 	return 0;
diff --git a/drivers/net/ethernet/google/gve/gve_tx_dqo.c b/drivers/net/ethernet/google/gve/gve_tx_dqo.c
index 2eba868d8037..a27f1574a733 100644
--- a/drivers/net/ethernet/google/gve/gve_tx_dqo.c
+++ b/drivers/net/ethernet/google/gve/gve_tx_dqo.c
@@ -660,7 +660,8 @@ static int gve_tx_add_skb_no_copy_dqo(struct gve_tx_ring *tx,
 			goto err;
 
 		dma_unmap_len_set(pkt, len[pkt->num_bufs], len);
-		dma_unmap_addr_set(pkt, dma[pkt->num_bufs], addr);
+		netmem_dma_unmap_addr_set(skb_frag_netmem(frag), pkt,
+					  dma[pkt->num_bufs], addr);
 		++pkt->num_bufs;
 
 		gve_tx_fill_pkt_desc_dqo(tx, desc_idx, skb, len, addr,
@@ -1038,8 +1039,9 @@ static void gve_unmap_packet(struct device *dev,
 	dma_unmap_single(dev, dma_unmap_addr(pkt, dma[0]),
 			 dma_unmap_len(pkt, len[0]), DMA_TO_DEVICE);
 	for (i = 1; i < pkt->num_bufs; i++) {
-		dma_unmap_page(dev, dma_unmap_addr(pkt, dma[i]),
-			       dma_unmap_len(pkt, len[i]), DMA_TO_DEVICE);
+		netmem_dma_unmap_page_attrs(dev, dma_unmap_addr(pkt, dma[i]),
+					    dma_unmap_len(pkt, len[i]),
+					    DMA_TO_DEVICE, 0);
 	}
 	pkt->num_bufs = 0;
 }
-- 
2.49.0.805.g082f7c87e0-goog
Re: [PATCH net-next v9 7/9] gve: add netmem TX support to GVE DQO-RDA mode
Posted by Harshitha Ramamurthy 7 months, 3 weeks ago 
On Thu, Apr 17, 2025 at 4:15 PM Mina Almasry <almasrymina@google.com> wrote:
>
> Use netmem_dma_*() helpers in gve_tx_dqo.c DQO-RDA paths to
> enable netmem TX support in that mode.
>
> Declare support for netmem TX in GVE DQO-RDA mode.
>
> Signed-off-by: Mina Almasry <almasrymina@google.com>
>
> ---
>
> v4:
> - New patch
> ---
>  drivers/net/ethernet/google/gve/gve_main.c   | 4 ++++
>  drivers/net/ethernet/google/gve/gve_tx_dqo.c | 8 +++++---
>  2 files changed, 9 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/net/ethernet/google/gve/gve_main.c b/drivers/net/ethernet/google/gve/gve_main.c
> index 8aaac9101377..430314225d4d 100644
> --- a/drivers/net/ethernet/google/gve/gve_main.c
> +++ b/drivers/net/ethernet/google/gve/gve_main.c
> @@ -2665,6 +2665,10 @@ static int gve_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
>
>         dev_info(&pdev->dev, "GVE version %s\n", gve_version_str);
>         dev_info(&pdev->dev, "GVE queue format %d\n", (int)priv->queue_format);
> +
> +       if (!gve_is_gqi(priv) && !gve_is_qpl(priv))
> +               dev->netmem_tx = true;
> +

a nit: but it would fit in better and be more uniform if this is set
earlier in the function where other features are set for the
net_device.

>         gve_clear_probe_in_progress(priv);
>         queue_work(priv->gve_wq, &priv->service_task);
>         return 0;
> diff --git a/drivers/net/ethernet/google/gve/gve_tx_dqo.c b/drivers/net/ethernet/google/gve/gve_tx_dqo.c
> index 2eba868d8037..a27f1574a733 100644
> --- a/drivers/net/ethernet/google/gve/gve_tx_dqo.c
> +++ b/drivers/net/ethernet/google/gve/gve_tx_dqo.c
> @@ -660,7 +660,8 @@ static int gve_tx_add_skb_no_copy_dqo(struct gve_tx_ring *tx,
>                         goto err;
>
>                 dma_unmap_len_set(pkt, len[pkt->num_bufs], len);
> -               dma_unmap_addr_set(pkt, dma[pkt->num_bufs], addr);
> +               netmem_dma_unmap_addr_set(skb_frag_netmem(frag), pkt,
> +                                         dma[pkt->num_bufs], addr);
>                 ++pkt->num_bufs;
>
>                 gve_tx_fill_pkt_desc_dqo(tx, desc_idx, skb, len, addr,
> @@ -1038,8 +1039,9 @@ static void gve_unmap_packet(struct device *dev,
>         dma_unmap_single(dev, dma_unmap_addr(pkt, dma[0]),
>                          dma_unmap_len(pkt, len[0]), DMA_TO_DEVICE);
>         for (i = 1; i < pkt->num_bufs; i++) {
> -               dma_unmap_page(dev, dma_unmap_addr(pkt, dma[i]),
> -                              dma_unmap_len(pkt, len[i]), DMA_TO_DEVICE);
> +               netmem_dma_unmap_page_attrs(dev, dma_unmap_addr(pkt, dma[i]),
> +                                           dma_unmap_len(pkt, len[i]),
> +                                           DMA_TO_DEVICE, 0);
>         }
>         pkt->num_bufs = 0;
>  }
> --
> 2.49.0.805.g082f7c87e0-goog
>
Re: [PATCH net-next v9 7/9] gve: add netmem TX support to GVE DQO-RDA mode
Posted by Mina Almasry 7 months, 3 weeks ago 
On Tue, Apr 22, 2025 at 10:43 AM Harshitha Ramamurthy
<hramamurthy@google.com> wrote:
>
> On Thu, Apr 17, 2025 at 4:15 PM Mina Almasry <almasrymina@google.com> wrote:
> >
> > Use netmem_dma_*() helpers in gve_tx_dqo.c DQO-RDA paths to
> > enable netmem TX support in that mode.
> >
> > Declare support for netmem TX in GVE DQO-RDA mode.
> >
> > Signed-off-by: Mina Almasry <almasrymina@google.com>
> >
> > ---
> >
> > v4:
> > - New patch
> > ---
> >  drivers/net/ethernet/google/gve/gve_main.c   | 4 ++++
> >  drivers/net/ethernet/google/gve/gve_tx_dqo.c | 8 +++++---
> >  2 files changed, 9 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/net/ethernet/google/gve/gve_main.c b/drivers/net/ethernet/google/gve/gve_main.c
> > index 8aaac9101377..430314225d4d 100644
> > --- a/drivers/net/ethernet/google/gve/gve_main.c
> > +++ b/drivers/net/ethernet/google/gve/gve_main.c
> > @@ -2665,6 +2665,10 @@ static int gve_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
> >
> >         dev_info(&pdev->dev, "GVE version %s\n", gve_version_str);
> >         dev_info(&pdev->dev, "GVE queue format %d\n", (int)priv->queue_format);
> > +
> > +       if (!gve_is_gqi(priv) && !gve_is_qpl(priv))
> > +               dev->netmem_tx = true;
> > +
>
> a nit: but it would fit in better and be more uniform if this is set
> earlier in the function where other features are set for the
> net_device.
>

Thanks for taking a look. I actually thought about that while trying
to implement this, but AFAIU (correct if wrong), gve_is_gqi and
gve_is_qpl need priv to be initialized, so this feature set must be
performed after gve_init_priv in this function. I suppose this feature
checking maybe can be put before register_netdev. Do you prefer that?


-- 
Thanks,
Mina
Re: [PATCH net-next v9 7/9] gve: add netmem TX support to GVE DQO-RDA mode
Posted by Harshitha Ramamurthy 7 months, 3 weeks ago 
On Tue, Apr 22, 2025 at 2:30 PM Mina Almasry <almasrymina@google.com> wrote:
>
> On Tue, Apr 22, 2025 at 10:43 AM Harshitha Ramamurthy
> <hramamurthy@google.com> wrote:
> >
> > On Thu, Apr 17, 2025 at 4:15 PM Mina Almasry <almasrymina@google.com> wrote:
> > >
> > > Use netmem_dma_*() helpers in gve_tx_dqo.c DQO-RDA paths to
> > > enable netmem TX support in that mode.
> > >
> > > Declare support for netmem TX in GVE DQO-RDA mode.
> > >
> > > Signed-off-by: Mina Almasry <almasrymina@google.com>
> > >
> > > ---
> > >
> > > v4:
> > > - New patch
> > > ---
> > >  drivers/net/ethernet/google/gve/gve_main.c   | 4 ++++
> > >  drivers/net/ethernet/google/gve/gve_tx_dqo.c | 8 +++++---
> > >  2 files changed, 9 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/drivers/net/ethernet/google/gve/gve_main.c b/drivers/net/ethernet/google/gve/gve_main.c
> > > index 8aaac9101377..430314225d4d 100644
> > > --- a/drivers/net/ethernet/google/gve/gve_main.c
> > > +++ b/drivers/net/ethernet/google/gve/gve_main.c
> > > @@ -2665,6 +2665,10 @@ static int gve_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
> > >
> > >         dev_info(&pdev->dev, "GVE version %s\n", gve_version_str);
> > >         dev_info(&pdev->dev, "GVE queue format %d\n", (int)priv->queue_format);
> > > +
> > > +       if (!gve_is_gqi(priv) && !gve_is_qpl(priv))
> > > +               dev->netmem_tx = true;
> > > +
> >
> > a nit: but it would fit in better and be more uniform if this is set
> > earlier in the function where other features are set for the
> > net_device.
> >
>
> Thanks for taking a look. I actually thought about that while trying
> to implement this, but AFAIU (correct if wrong), gve_is_gqi and
> gve_is_qpl need priv to be initialized, so this feature set must be
> performed after gve_init_priv in this function. I suppose this feature
> checking maybe can be put before register_netdev. Do you prefer that?

Ah yes, you are right. Thanks for checking. That would be preferable.
Another option is to move it inside gve_init_priv() after the mode has
been set. Either is okay.

Thanks,
Harshitha
>
>
> --
> Thanks,
> Mina

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.