* [PATCH net] virtio_net: fix missing dma unmap for resize
@ 2023-11-06 8:18 Xuan Zhuo
2023-11-09 11:26 ` Paolo Abeni
2023-11-09 12:06 ` Michael S. Tsirkin
0 siblings, 2 replies; 6+ messages in thread
From: Xuan Zhuo @ 2023-11-06 8:18 UTC (permalink / raw)
To: netdev
Cc: Michael S. Tsirkin, Jason Wang, Xuan Zhuo, David S. Miller,
Eric Dumazet, Jakub Kicinski, Paolo Abeni, virtualization
For rq, we have three cases getting buffers from virtio core:
1. virtqueue_get_buf{,_ctx}
2. virtqueue_detach_unused_buf
3. callback for virtqueue_resize
But in commit 295525e29a5b("virtio_net: merge dma operations when
filling mergeable buffers"), I missed the dma unmap for the #3 case.
That will leak some memory, because I did not release the pages referred
by the unused buffers.
If we do such script, we will make the system OOM.
while true
do
ethtool -G ens4 rx 128
ethtool -G ens4 rx 256
free -m
done
Fixes: 295525e29a5b ("virtio_net: merge dma operations when filling mergeable buffers")
Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
---
drivers/net/virtio_net.c | 43 ++++++++++++++++++++--------------------
1 file changed, 22 insertions(+), 21 deletions(-)
diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index d16f592c2061..6423a3a007ce 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -408,6 +408,17 @@ static struct page *get_a_page(struct receive_queue *rq, gfp_t gfp_mask)
return p;
}
+static void virtnet_rq_free_buf(struct virtnet_info *vi,
+ struct receive_queue *rq, void *buf)
+{
+ if (vi->mergeable_rx_bufs)
+ put_page(virt_to_head_page(buf));
+ else if (vi->big_packets)
+ give_pages(rq, buf);
+ else
+ put_page(virt_to_head_page(buf));
+}
+
static void enable_delayed_refill(struct virtnet_info *vi)
{
spin_lock_bh(&vi->refill_lock);
@@ -634,17 +645,6 @@ static void *virtnet_rq_get_buf(struct receive_queue *rq, u32 *len, void **ctx)
return buf;
}
-static void *virtnet_rq_detach_unused_buf(struct receive_queue *rq)
-{
- void *buf;
-
- buf = virtqueue_detach_unused_buf(rq->vq);
- if (buf && rq->do_dma)
- virtnet_rq_unmap(rq, buf, 0);
-
- return buf;
-}
-
static void virtnet_rq_init_one_sg(struct receive_queue *rq, void *buf, u32 len)
{
struct virtnet_rq_dma *dma;
@@ -1764,7 +1764,7 @@ static void receive_buf(struct virtnet_info *vi, struct receive_queue *rq,
if (unlikely(len < vi->hdr_len + ETH_HLEN)) {
pr_debug("%s: short packet %i\n", dev->name, len);
DEV_STATS_INC(dev, rx_length_errors);
- virtnet_rq_free_unused_buf(rq->vq, buf);
+ virtnet_rq_free_buf(vi, rq, buf);
return;
}
@@ -4034,14 +4034,15 @@ static void virtnet_sq_free_unused_buf(struct virtqueue *vq, void *buf)
static void virtnet_rq_free_unused_buf(struct virtqueue *vq, void *buf)
{
struct virtnet_info *vi = vq->vdev->priv;
+ struct receive_queue *rq;
int i = vq2rxq(vq);
- if (vi->mergeable_rx_bufs)
- put_page(virt_to_head_page(buf));
- else if (vi->big_packets)
- give_pages(&vi->rq[i], buf);
- else
- put_page(virt_to_head_page(buf));
+ rq = &vi->rq[i];
+
+ if (rq->do_dma)
+ virtnet_rq_unmap(rq, buf, 0);
+
+ virtnet_rq_free_buf(vi, rq, buf);
}
static void free_unused_bufs(struct virtnet_info *vi)
@@ -4057,10 +4058,10 @@ static void free_unused_bufs(struct virtnet_info *vi)
}
for (i = 0; i < vi->max_queue_pairs; i++) {
- struct receive_queue *rq = &vi->rq[i];
+ struct virtqueue *vq = vi->rq[i].vq;
- while ((buf = virtnet_rq_detach_unused_buf(rq)) != NULL)
- virtnet_rq_free_unused_buf(rq->vq, buf);
+ while ((buf = virtqueue_detach_unused_buf(vq)) != NULL)
+ virtnet_rq_free_unused_buf(vq, buf);
cond_resched();
}
}
--
2.32.0.3.g01195cf9f
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH net] virtio_net: fix missing dma unmap for resize
2023-11-06 8:18 [PATCH net] virtio_net: fix missing dma unmap for resize Xuan Zhuo
@ 2023-11-09 11:26 ` Paolo Abeni
2023-11-09 12:06 ` Michael S. Tsirkin
1 sibling, 0 replies; 6+ messages in thread
From: Paolo Abeni @ 2023-11-09 11:26 UTC (permalink / raw)
To: Xuan Zhuo, netdev
Cc: Michael S. Tsirkin, Jason Wang, David S. Miller, Eric Dumazet,
Jakub Kicinski, virtualization
On Mon, 2023-11-06 at 16:18 +0800, Xuan Zhuo wrote:
> For rq, we have three cases getting buffers from virtio core:
>
> 1. virtqueue_get_buf{,_ctx}
> 2. virtqueue_detach_unused_buf
> 3. callback for virtqueue_resize
>
> But in commit 295525e29a5b("virtio_net: merge dma operations when
> filling mergeable buffers"), I missed the dma unmap for the #3 case.
>
> That will leak some memory, because I did not release the pages referred
> by the unused buffers.
>
> If we do such script, we will make the system OOM.
>
> while true
> do
> ethtool -G ens4 rx 128
> ethtool -G ens4 rx 256
> free -m
> done
>
> Fixes: 295525e29a5b ("virtio_net: merge dma operations when filling mergeable buffers")
> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
@Micheal, @Jason: this fix LGTM, but I guess it deserve an explicit ack
from you before merging, thanks!
Paolo
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH net] virtio_net: fix missing dma unmap for resize
2023-11-06 8:18 [PATCH net] virtio_net: fix missing dma unmap for resize Xuan Zhuo
2023-11-09 11:26 ` Paolo Abeni
@ 2023-11-09 12:06 ` Michael S. Tsirkin
2023-11-10 1:58 ` Xuan Zhuo
1 sibling, 1 reply; 6+ messages in thread
From: Michael S. Tsirkin @ 2023-11-09 12:06 UTC (permalink / raw)
To: Xuan Zhuo
Cc: netdev, Jason Wang, David S. Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, virtualization
On Mon, Nov 06, 2023 at 04:18:32PM +0800, Xuan Zhuo wrote:
> For rq, we have three cases getting buffers from virtio core:
>
> 1. virtqueue_get_buf{,_ctx}
> 2. virtqueue_detach_unused_buf
> 3. callback for virtqueue_resize
>
> But in commit 295525e29a5b("virtio_net: merge dma operations when
> filling mergeable buffers"), I missed the dma unmap for the #3 case.
>
> That will leak some memory, because I did not release the pages referred
> by the unused buffers.
>
> If we do such script, we will make the system OOM.
>
> while true
> do
> ethtool -G ens4 rx 128
> ethtool -G ens4 rx 256
> free -m
> done
>
> Fixes: 295525e29a5b ("virtio_net: merge dma operations when filling mergeable buffers")
> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> ---
> drivers/net/virtio_net.c | 43 ++++++++++++++++++++--------------------
> 1 file changed, 22 insertions(+), 21 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index d16f592c2061..6423a3a007ce 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -408,6 +408,17 @@ static struct page *get_a_page(struct receive_queue *rq, gfp_t gfp_mask)
> return p;
> }
>
> +static void virtnet_rq_free_buf(struct virtnet_info *vi,
> + struct receive_queue *rq, void *buf)
> +{
> + if (vi->mergeable_rx_bufs)
> + put_page(virt_to_head_page(buf));
> + else if (vi->big_packets)
> + give_pages(rq, buf);
> + else
> + put_page(virt_to_head_page(buf));
> +}
> +
> static void enable_delayed_refill(struct virtnet_info *vi)
> {
> spin_lock_bh(&vi->refill_lock);
> @@ -634,17 +645,6 @@ static void *virtnet_rq_get_buf(struct receive_queue *rq, u32 *len, void **ctx)
> return buf;
> }
>
> -static void *virtnet_rq_detach_unused_buf(struct receive_queue *rq)
> -{
> - void *buf;
> -
> - buf = virtqueue_detach_unused_buf(rq->vq);
> - if (buf && rq->do_dma)
> - virtnet_rq_unmap(rq, buf, 0);
> -
> - return buf;
> -}
> -
> static void virtnet_rq_init_one_sg(struct receive_queue *rq, void *buf, u32 len)
> {
> struct virtnet_rq_dma *dma;
> @@ -1764,7 +1764,7 @@ static void receive_buf(struct virtnet_info *vi, struct receive_queue *rq,
> if (unlikely(len < vi->hdr_len + ETH_HLEN)) {
> pr_debug("%s: short packet %i\n", dev->name, len);
> DEV_STATS_INC(dev, rx_length_errors);
> - virtnet_rq_free_unused_buf(rq->vq, buf);
> + virtnet_rq_free_buf(vi, rq, buf);
> return;
> }
>
> @@ -4034,14 +4034,15 @@ static void virtnet_sq_free_unused_buf(struct virtqueue *vq, void *buf)
> static void virtnet_rq_free_unused_buf(struct virtqueue *vq, void *buf)
> {
> struct virtnet_info *vi = vq->vdev->priv;
> + struct receive_queue *rq;
> int i = vq2rxq(vq);
>
> - if (vi->mergeable_rx_bufs)
> - put_page(virt_to_head_page(buf));
> - else if (vi->big_packets)
> - give_pages(&vi->rq[i], buf);
> - else
> - put_page(virt_to_head_page(buf));
> + rq = &vi->rq[i];
> +
> + if (rq->do_dma)
> + virtnet_rq_unmap(rq, buf, 0);
> +
> + virtnet_rq_free_buf(vi, rq, buf);
> }
>
So we have virtnet_rq_free_buf which sounds like it should free any
buf, and we have virtnet_rq_free_unused_buf which is only for unused.
Or so it would seem from names but this is not true.
Better function names?
> static void free_unused_bufs(struct virtnet_info *vi)
> @@ -4057,10 +4058,10 @@ static void free_unused_bufs(struct virtnet_info *vi)
> }
>
> for (i = 0; i < vi->max_queue_pairs; i++) {
> - struct receive_queue *rq = &vi->rq[i];
> + struct virtqueue *vq = vi->rq[i].vq;
>
> - while ((buf = virtnet_rq_detach_unused_buf(rq)) != NULL)
> - virtnet_rq_free_unused_buf(rq->vq, buf);
> + while ((buf = virtqueue_detach_unused_buf(vq)) != NULL)
> + virtnet_rq_free_unused_buf(vq, buf);
> cond_resched();
> }
> }
> --
> 2.32.0.3.g01195cf9f
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH net] virtio_net: fix missing dma unmap for resize
2023-11-09 12:06 ` Michael S. Tsirkin
@ 2023-11-10 1:58 ` Xuan Zhuo
2023-11-10 5:37 ` Michael S. Tsirkin
0 siblings, 1 reply; 6+ messages in thread
From: Xuan Zhuo @ 2023-11-10 1:58 UTC (permalink / raw)
To: Michael S. Tsirkin
Cc: netdev, Jason Wang, David S. Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, virtualization
On Thu, 9 Nov 2023 07:06:16 -0500, "Michael S. Tsirkin" <mst@redhat.com> wrote:
> On Mon, Nov 06, 2023 at 04:18:32PM +0800, Xuan Zhuo wrote:
> > For rq, we have three cases getting buffers from virtio core:
> >
> > 1. virtqueue_get_buf{,_ctx}
> > 2. virtqueue_detach_unused_buf
> > 3. callback for virtqueue_resize
> >
> > But in commit 295525e29a5b("virtio_net: merge dma operations when
> > filling mergeable buffers"), I missed the dma unmap for the #3 case.
> >
> > That will leak some memory, because I did not release the pages referred
> > by the unused buffers.
> >
> > If we do such script, we will make the system OOM.
> >
> > while true
> > do
> > ethtool -G ens4 rx 128
> > ethtool -G ens4 rx 256
> > free -m
> > done
> >
> > Fixes: 295525e29a5b ("virtio_net: merge dma operations when filling mergeable buffers")
> > Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> > ---
> > drivers/net/virtio_net.c | 43 ++++++++++++++++++++--------------------
> > 1 file changed, 22 insertions(+), 21 deletions(-)
> >
> > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > index d16f592c2061..6423a3a007ce 100644
> > --- a/drivers/net/virtio_net.c
> > +++ b/drivers/net/virtio_net.c
> > @@ -408,6 +408,17 @@ static struct page *get_a_page(struct receive_queue *rq, gfp_t gfp_mask)
> > return p;
> > }
> >
> > +static void virtnet_rq_free_buf(struct virtnet_info *vi,
> > + struct receive_queue *rq, void *buf)
> > +{
> > + if (vi->mergeable_rx_bufs)
> > + put_page(virt_to_head_page(buf));
> > + else if (vi->big_packets)
> > + give_pages(rq, buf);
> > + else
> > + put_page(virt_to_head_page(buf));
> > +}
> > +
>
> > static void enable_delayed_refill(struct virtnet_info *vi)
> > {
> > spin_lock_bh(&vi->refill_lock);
> > @@ -634,17 +645,6 @@ static void *virtnet_rq_get_buf(struct receive_queue *rq, u32 *len, void **ctx)
> > return buf;
> > }
> >
> > -static void *virtnet_rq_detach_unused_buf(struct receive_queue *rq)
> > -{
> > - void *buf;
> > -
> > - buf = virtqueue_detach_unused_buf(rq->vq);
> > - if (buf && rq->do_dma)
> > - virtnet_rq_unmap(rq, buf, 0);
> > -
> > - return buf;
> > -}
> > -
> > static void virtnet_rq_init_one_sg(struct receive_queue *rq, void *buf, u32 len)
> > {
> > struct virtnet_rq_dma *dma;
> > @@ -1764,7 +1764,7 @@ static void receive_buf(struct virtnet_info *vi, struct receive_queue *rq,
> > if (unlikely(len < vi->hdr_len + ETH_HLEN)) {
> > pr_debug("%s: short packet %i\n", dev->name, len);
> > DEV_STATS_INC(dev, rx_length_errors);
> > - virtnet_rq_free_unused_buf(rq->vq, buf);
> > + virtnet_rq_free_buf(vi, rq, buf);
> > return;
> > }
> >
> > @@ -4034,14 +4034,15 @@ static void virtnet_sq_free_unused_buf(struct virtqueue *vq, void *buf)
> > static void virtnet_rq_free_unused_buf(struct virtqueue *vq, void *buf)
> > {
> > struct virtnet_info *vi = vq->vdev->priv;
> > + struct receive_queue *rq;
> > int i = vq2rxq(vq);
> >
> > - if (vi->mergeable_rx_bufs)
> > - put_page(virt_to_head_page(buf));
> > - else if (vi->big_packets)
> > - give_pages(&vi->rq[i], buf);
> > - else
> > - put_page(virt_to_head_page(buf));
> > + rq = &vi->rq[i];
> > +
> > + if (rq->do_dma)
> > + virtnet_rq_unmap(rq, buf, 0);
> > +
> > + virtnet_rq_free_buf(vi, rq, buf);
> > }
> >
>
> So we have virtnet_rq_free_buf which sounds like it should free any
> buf, and we have virtnet_rq_free_unused_buf which is only for unused.
> Or so it would seem from names but this is not true.
> Better function names?
Sorry. not get it.
virtnet_rq_free_buf() that free the buf passed in. That is called by
virtnet_rq_free_unused_buf or receive_buf to free the buffer. I think
the name is right.
virtnet_rq_free_unused_buf is called by free_unused_bufs() and the
virtqueue_resize() to free the unused bufs. I think this name is right also.
So I do not get your mean.
Are there any details I've overlooked?
Thanks.
>
> > static void free_unused_bufs(struct virtnet_info *vi)
> > @@ -4057,10 +4058,10 @@ static void free_unused_bufs(struct virtnet_info *vi)
> > }
> >
> > for (i = 0; i < vi->max_queue_pairs; i++) {
> > - struct receive_queue *rq = &vi->rq[i];
> > + struct virtqueue *vq = vi->rq[i].vq;
> >
> > - while ((buf = virtnet_rq_detach_unused_buf(rq)) != NULL)
> > - virtnet_rq_free_unused_buf(rq->vq, buf);
> > + while ((buf = virtqueue_detach_unused_buf(vq)) != NULL)
> > + virtnet_rq_free_unused_buf(vq, buf);
> > cond_resched();
> > }
> > }
> > --
> > 2.32.0.3.g01195cf9f
>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH net] virtio_net: fix missing dma unmap for resize
2023-11-10 1:58 ` Xuan Zhuo
@ 2023-11-10 5:37 ` Michael S. Tsirkin
2023-11-10 5:44 ` Xuan Zhuo
0 siblings, 1 reply; 6+ messages in thread
From: Michael S. Tsirkin @ 2023-11-10 5:37 UTC (permalink / raw)
To: Xuan Zhuo
Cc: netdev, Jason Wang, David S. Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, virtualization
On Fri, Nov 10, 2023 at 09:58:45AM +0800, Xuan Zhuo wrote:
> On Thu, 9 Nov 2023 07:06:16 -0500, "Michael S. Tsirkin" <mst@redhat.com> wrote:
> > On Mon, Nov 06, 2023 at 04:18:32PM +0800, Xuan Zhuo wrote:
> > > For rq, we have three cases getting buffers from virtio core:
> > >
> > > 1. virtqueue_get_buf{,_ctx}
> > > 2. virtqueue_detach_unused_buf
> > > 3. callback for virtqueue_resize
> > >
> > > But in commit 295525e29a5b("virtio_net: merge dma operations when
> > > filling mergeable buffers"), I missed the dma unmap for the #3 case.
> > >
> > > That will leak some memory, because I did not release the pages referred
> > > by the unused buffers.
> > >
> > > If we do such script, we will make the system OOM.
> > >
> > > while true
> > > do
> > > ethtool -G ens4 rx 128
> > > ethtool -G ens4 rx 256
> > > free -m
> > > done
> > >
> > > Fixes: 295525e29a5b ("virtio_net: merge dma operations when filling mergeable buffers")
> > > Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> > > ---
> > > drivers/net/virtio_net.c | 43 ++++++++++++++++++++--------------------
> > > 1 file changed, 22 insertions(+), 21 deletions(-)
> > >
> > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > > index d16f592c2061..6423a3a007ce 100644
> > > --- a/drivers/net/virtio_net.c
> > > +++ b/drivers/net/virtio_net.c
> > > @@ -408,6 +408,17 @@ static struct page *get_a_page(struct receive_queue *rq, gfp_t gfp_mask)
> > > return p;
> > > }
> > >
> > > +static void virtnet_rq_free_buf(struct virtnet_info *vi,
> > > + struct receive_queue *rq, void *buf)
> > > +{
> > > + if (vi->mergeable_rx_bufs)
> > > + put_page(virt_to_head_page(buf));
> > > + else if (vi->big_packets)
> > > + give_pages(rq, buf);
> > > + else
> > > + put_page(virt_to_head_page(buf));
> > > +}
> > > +
> >
> > > static void enable_delayed_refill(struct virtnet_info *vi)
> > > {
> > > spin_lock_bh(&vi->refill_lock);
> > > @@ -634,17 +645,6 @@ static void *virtnet_rq_get_buf(struct receive_queue *rq, u32 *len, void **ctx)
> > > return buf;
> > > }
> > >
> > > -static void *virtnet_rq_detach_unused_buf(struct receive_queue *rq)
> > > -{
> > > - void *buf;
> > > -
> > > - buf = virtqueue_detach_unused_buf(rq->vq);
> > > - if (buf && rq->do_dma)
> > > - virtnet_rq_unmap(rq, buf, 0);
> > > -
> > > - return buf;
> > > -}
> > > -
> > > static void virtnet_rq_init_one_sg(struct receive_queue *rq, void *buf, u32 len)
> > > {
> > > struct virtnet_rq_dma *dma;
> > > @@ -1764,7 +1764,7 @@ static void receive_buf(struct virtnet_info *vi, struct receive_queue *rq,
> > > if (unlikely(len < vi->hdr_len + ETH_HLEN)) {
> > > pr_debug("%s: short packet %i\n", dev->name, len);
> > > DEV_STATS_INC(dev, rx_length_errors);
> > > - virtnet_rq_free_unused_buf(rq->vq, buf);
> > > + virtnet_rq_free_buf(vi, rq, buf);
> > > return;
> > > }
> > >
> > > @@ -4034,14 +4034,15 @@ static void virtnet_sq_free_unused_buf(struct virtqueue *vq, void *buf)
> > > static void virtnet_rq_free_unused_buf(struct virtqueue *vq, void *buf)
> > > {
> > > struct virtnet_info *vi = vq->vdev->priv;
> > > + struct receive_queue *rq;
> > > int i = vq2rxq(vq);
> > >
> > > - if (vi->mergeable_rx_bufs)
> > > - put_page(virt_to_head_page(buf));
> > > - else if (vi->big_packets)
> > > - give_pages(&vi->rq[i], buf);
> > > - else
> > > - put_page(virt_to_head_page(buf));
> > > + rq = &vi->rq[i];
> > > +
> > > + if (rq->do_dma)
> > > + virtnet_rq_unmap(rq, buf, 0);
> > > +
> > > + virtnet_rq_free_buf(vi, rq, buf);
> > > }
> > >
> >
> > So we have virtnet_rq_free_buf which sounds like it should free any
> > buf, and we have virtnet_rq_free_unused_buf which is only for unused.
> > Or so it would seem from names but this is not true.
> > Better function names?
>
> Sorry. not get it.
>
> virtnet_rq_free_buf() that free the buf passed in. That is called by
> virtnet_rq_free_unused_buf or receive_buf to free the buffer. I think
> the name is right.
>
> virtnet_rq_free_unused_buf is called by free_unused_bufs() and the
> virtqueue_resize() to free the unused bufs. I think this name is right also.
>
> So I do not get your mean.
> Are there any details I've overlooked?
>
> Thanks.
Bad function names - they are too similar. Function name should
say what it does not where it's called from.
What is the difference? That virtnet_rq_free_unused_buf unmaps
and frees and virtnet_rq_free_buf just frees memory?
> >
> > > static void free_unused_bufs(struct virtnet_info *vi)
> > > @@ -4057,10 +4058,10 @@ static void free_unused_bufs(struct virtnet_info *vi)
> > > }
> > >
> > > for (i = 0; i < vi->max_queue_pairs; i++) {
> > > - struct receive_queue *rq = &vi->rq[i];
> > > + struct virtqueue *vq = vi->rq[i].vq;
> > >
> > > - while ((buf = virtnet_rq_detach_unused_buf(rq)) != NULL)
> > > - virtnet_rq_free_unused_buf(rq->vq, buf);
> > > + while ((buf = virtqueue_detach_unused_buf(vq)) != NULL)
> > > + virtnet_rq_free_unused_buf(vq, buf);
> > > cond_resched();
> > > }
> > > }
> > > --
> > > 2.32.0.3.g01195cf9f
> >
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH net] virtio_net: fix missing dma unmap for resize
2023-11-10 5:37 ` Michael S. Tsirkin
@ 2023-11-10 5:44 ` Xuan Zhuo
0 siblings, 0 replies; 6+ messages in thread
From: Xuan Zhuo @ 2023-11-10 5:44 UTC (permalink / raw)
To: Michael S. Tsirkin
Cc: netdev, Jason Wang, David S. Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, virtualization
On Fri, 10 Nov 2023 00:37:32 -0500, "Michael S. Tsirkin" <mst@redhat.com> wrote:
> On Fri, Nov 10, 2023 at 09:58:45AM +0800, Xuan Zhuo wrote:
> > On Thu, 9 Nov 2023 07:06:16 -0500, "Michael S. Tsirkin" <mst@redhat.com> wrote:
> > > On Mon, Nov 06, 2023 at 04:18:32PM +0800, Xuan Zhuo wrote:
> > > > For rq, we have three cases getting buffers from virtio core:
> > > >
> > > > 1. virtqueue_get_buf{,_ctx}
> > > > 2. virtqueue_detach_unused_buf
> > > > 3. callback for virtqueue_resize
> > > >
> > > > But in commit 295525e29a5b("virtio_net: merge dma operations when
> > > > filling mergeable buffers"), I missed the dma unmap for the #3 case.
> > > >
> > > > That will leak some memory, because I did not release the pages referred
> > > > by the unused buffers.
> > > >
> > > > If we do such script, we will make the system OOM.
> > > >
> > > > while true
> > > > do
> > > > ethtool -G ens4 rx 128
> > > > ethtool -G ens4 rx 256
> > > > free -m
> > > > done
> > > >
> > > > Fixes: 295525e29a5b ("virtio_net: merge dma operations when filling mergeable buffers")
> > > > Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> > > > ---
> > > > drivers/net/virtio_net.c | 43 ++++++++++++++++++++--------------------
> > > > 1 file changed, 22 insertions(+), 21 deletions(-)
> > > >
> > > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > > > index d16f592c2061..6423a3a007ce 100644
> > > > --- a/drivers/net/virtio_net.c
> > > > +++ b/drivers/net/virtio_net.c
> > > > @@ -408,6 +408,17 @@ static struct page *get_a_page(struct receive_queue *rq, gfp_t gfp_mask)
> > > > return p;
> > > > }
> > > >
> > > > +static void virtnet_rq_free_buf(struct virtnet_info *vi,
> > > > + struct receive_queue *rq, void *buf)
> > > > +{
> > > > + if (vi->mergeable_rx_bufs)
> > > > + put_page(virt_to_head_page(buf));
> > > > + else if (vi->big_packets)
> > > > + give_pages(rq, buf);
> > > > + else
> > > > + put_page(virt_to_head_page(buf));
> > > > +}
> > > > +
> > >
> > > > static void enable_delayed_refill(struct virtnet_info *vi)
> > > > {
> > > > spin_lock_bh(&vi->refill_lock);
> > > > @@ -634,17 +645,6 @@ static void *virtnet_rq_get_buf(struct receive_queue *rq, u32 *len, void **ctx)
> > > > return buf;
> > > > }
> > > >
> > > > -static void *virtnet_rq_detach_unused_buf(struct receive_queue *rq)
> > > > -{
> > > > - void *buf;
> > > > -
> > > > - buf = virtqueue_detach_unused_buf(rq->vq);
> > > > - if (buf && rq->do_dma)
> > > > - virtnet_rq_unmap(rq, buf, 0);
> > > > -
> > > > - return buf;
> > > > -}
> > > > -
> > > > static void virtnet_rq_init_one_sg(struct receive_queue *rq, void *buf, u32 len)
> > > > {
> > > > struct virtnet_rq_dma *dma;
> > > > @@ -1764,7 +1764,7 @@ static void receive_buf(struct virtnet_info *vi, struct receive_queue *rq,
> > > > if (unlikely(len < vi->hdr_len + ETH_HLEN)) {
> > > > pr_debug("%s: short packet %i\n", dev->name, len);
> > > > DEV_STATS_INC(dev, rx_length_errors);
> > > > - virtnet_rq_free_unused_buf(rq->vq, buf);
> > > > + virtnet_rq_free_buf(vi, rq, buf);
> > > > return;
> > > > }
> > > >
> > > > @@ -4034,14 +4034,15 @@ static void virtnet_sq_free_unused_buf(struct virtqueue *vq, void *buf)
> > > > static void virtnet_rq_free_unused_buf(struct virtqueue *vq, void *buf)
> > > > {
> > > > struct virtnet_info *vi = vq->vdev->priv;
> > > > + struct receive_queue *rq;
> > > > int i = vq2rxq(vq);
> > > >
> > > > - if (vi->mergeable_rx_bufs)
> > > > - put_page(virt_to_head_page(buf));
> > > > - else if (vi->big_packets)
> > > > - give_pages(&vi->rq[i], buf);
> > > > - else
> > > > - put_page(virt_to_head_page(buf));
> > > > + rq = &vi->rq[i];
> > > > +
> > > > + if (rq->do_dma)
> > > > + virtnet_rq_unmap(rq, buf, 0);
> > > > +
> > > > + virtnet_rq_free_buf(vi, rq, buf);
> > > > }
> > > >
> > >
> > > So we have virtnet_rq_free_buf which sounds like it should free any
> > > buf, and we have virtnet_rq_free_unused_buf which is only for unused.
> > > Or so it would seem from names but this is not true.
> > > Better function names?
> >
> > Sorry. not get it.
> >
> > virtnet_rq_free_buf() that free the buf passed in. That is called by
> > virtnet_rq_free_unused_buf or receive_buf to free the buffer. I think
> > the name is right.
> >
> > virtnet_rq_free_unused_buf is called by free_unused_bufs() and the
> > virtqueue_resize() to free the unused bufs. I think this name is right also.
> >
> > So I do not get your mean.
> > Are there any details I've overlooked?
> >
> > Thanks.
>
> Bad function names - they are too similar. Function name should
> say what it does not where it's called from.
> What is the difference? That virtnet_rq_free_unused_buf unmaps
> and frees and virtnet_rq_free_buf just frees memory?
Yes. virtnet_rq_free_buf frees memory.
For this patch, I think virtnet_rq_free_buf is ok.
virtnet_rq_free_buf is as your request.
virtnet_rq_free_unused_buf is named as it want to do. Or as you said.
Indeed these are similar.
Could you give some advices?
Thanks.
>
>
> > >
> > > > static void free_unused_bufs(struct virtnet_info *vi)
> > > > @@ -4057,10 +4058,10 @@ static void free_unused_bufs(struct virtnet_info *vi)
> > > > }
> > > >
> > > > for (i = 0; i < vi->max_queue_pairs; i++) {
> > > > - struct receive_queue *rq = &vi->rq[i];
> > > > + struct virtqueue *vq = vi->rq[i].vq;
> > > >
> > > > - while ((buf = virtnet_rq_detach_unused_buf(rq)) != NULL)
> > > > - virtnet_rq_free_unused_buf(rq->vq, buf);
> > > > + while ((buf = virtqueue_detach_unused_buf(vq)) != NULL)
> > > > + virtnet_rq_free_unused_buf(vq, buf);
> > > > cond_resched();
> > > > }
> > > > }
> > > > --
> > > > 2.32.0.3.g01195cf9f
> > >
>
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2023-11-10 6:22 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2023-11-06 8:18 [PATCH net] virtio_net: fix missing dma unmap for resize Xuan Zhuo
2023-11-09 11:26 ` Paolo Abeni
2023-11-09 12:06 ` Michael S. Tsirkin
2023-11-10 1:58 ` Xuan Zhuo
2023-11-10 5:37 ` Michael S. Tsirkin
2023-11-10 5:44 ` Xuan Zhuo
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).