[PATCH] virtio_net: lower limit on buffer size

[PATCH] virtio_net: lower limit on buffer size

[Michael S. Tsirkin]

commit d85b758f72b0 "virtio_net: fix support for small rings"
was supposed to increase the buffer size for small rings
but had an unintentional side effect of decreasing
it for large rings. This seems to break some setups -
it's not yet clear why, but increasing buffer size
back to what it was before helps.

Fixes: d85b758f72b0 "virtio_net: fix support for small rings"
Reported-by: Mikulas Patocka <mpatocka@redhat.com>
Reported-by: "J. Bruce Fields" <bfields@fieldses.org>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
---

Can reporters please confirm whether the following helps?
If it does I think we should queue it even though I expect I'll need to
investigate the exact reasons for the failure down the road - probably
a hypervisor bug.

 drivers/net/virtio_net.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 87b5c20..60abb5d 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -842,7 +842,7 @@ static unsigned int get_mergeable_buf_len(struct receive_queue *rq,
 	unsigned int len;
 
 	len = hdr_len + clamp_t(unsigned int, ewma_pkt_len_read(avg_pkt_len),
-				rq->min_buf_len - hdr_len, PAGE_SIZE - hdr_len);
+				rq->min_buf_len, PAGE_SIZE - hdr_len);
 	return ALIGN(len, L1_CACHE_BYTES);
 }
 
@@ -2039,7 +2039,8 @@ static unsigned int mergeable_min_buf_len(struct virtnet_info *vi, struct virtqu
 	unsigned int buf_len = hdr_len + ETH_HLEN + VLAN_HLEN + packet_len;
 	unsigned int min_buf_len = DIV_ROUND_UP(buf_len, rq_size);
 
-	return max(min_buf_len, hdr_len);
+	return max(max(min_buf_len, hdr_len) - hdr_len,
+		   (unsigned int)GOOD_PACKET_LEN);
 }
 
 static int virtnet_find_vqs(struct virtnet_info *vi)
-- 
MST

Re: [PATCH] virtio_net: lower limit on buffer size

[J. Bruce Fields]

On Fri, Jun 02, 2017 at 02:56:04AM +0300, Michael S. Tsirkin wrote:
> commit d85b758f72b0 "virtio_net: fix support for small rings"
> was supposed to increase the buffer size for small rings
> but had an unintentional side effect of decreasing
> it for large rings. This seems to break some setups -
> it's not yet clear why, but increasing buffer size
> back to what it was before helps.
> 
> Fixes: d85b758f72b0 "virtio_net: fix support for small rings"
> Reported-by: Mikulas Patocka <mpatocka@redhat.com>
> Reported-by: "J. Bruce Fields" <bfields@fieldses.org>
> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> ---
> 
> Can reporters please confirm whether the following helps?

Works for me.

--b.

> If it does I think we should queue it even though I expect I'll need to
> investigate the exact reasons for the failure down the road - probably
> a hypervisor bug.
> 
>  drivers/net/virtio_net.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 87b5c20..60abb5d 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -842,7 +842,7 @@ static unsigned int get_mergeable_buf_len(struct receive_queue *rq,
>  	unsigned int len;
>  
>  	len = hdr_len + clamp_t(unsigned int, ewma_pkt_len_read(avg_pkt_len),
> -				rq->min_buf_len - hdr_len, PAGE_SIZE - hdr_len);
> +				rq->min_buf_len, PAGE_SIZE - hdr_len);
>  	return ALIGN(len, L1_CACHE_BYTES);
>  }
>  
> @@ -2039,7 +2039,8 @@ static unsigned int mergeable_min_buf_len(struct virtnet_info *vi, struct virtqu
>  	unsigned int buf_len = hdr_len + ETH_HLEN + VLAN_HLEN + packet_len;
>  	unsigned int min_buf_len = DIV_ROUND_UP(buf_len, rq_size);
>  
> -	return max(min_buf_len, hdr_len);
> +	return max(max(min_buf_len, hdr_len) - hdr_len,
> +		   (unsigned int)GOOD_PACKET_LEN);
>  }
>  
>  static int virtnet_find_vqs(struct virtnet_info *vi)
> -- 
> MST

Re: [PATCH] virtio_net: lower limit on buffer size

[Sergei Shtylyov]

Hello!

On 6/2/2017 2:56 AM, Michael S. Tsirkin wrote:

> commit d85b758f72b0 "virtio_net: fix support for small rings"

    Commit d85b758f72b0 ("virtio_net: fix support for small rings")

> was supposed to increase the buffer size for small rings
> but had an unintentional side effect of decreasing
> it for large rings. This seems to break some setups -
> it's not yet clear why, but increasing buffer size
> back to what it was before helps.
>
> Fixes: d85b758f72b0 "virtio_net: fix support for small rings"

Fixes: d85b758f72b0 ("virtio_net: fix support for small rings")

> Reported-by: Mikulas Patocka <mpatocka@redhat.com>
> Reported-by: "J. Bruce Fields" <bfields@fieldses.org>
> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
[...]

MBR, Sergei

Re: [PATCH] virtio_net: lower limit on buffer size

[Mikulas Patocka]

On Fri, 2 Jun 2017, Michael S. Tsirkin wrote:

> commit d85b758f72b0 "virtio_net: fix support for small rings"
> was supposed to increase the buffer size for small rings
> but had an unintentional side effect of decreasing
> it for large rings. This seems to break some setups -
> it's not yet clear why, but increasing buffer size
> back to what it was before helps.
> 
> Fixes: d85b758f72b0 "virtio_net: fix support for small rings"
> Reported-by: Mikulas Patocka <mpatocka@redhat.com>
> Reported-by: "J. Bruce Fields" <bfields@fieldses.org>
> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> ---
> 
> Can reporters please confirm whether the following helps?
> If it does I think we should queue it even though I expect I'll need to
> investigate the exact reasons for the failure down the road - probably
> a hypervisor bug.

Hi

This patch fixes the problem for me.

Mikulas

>  drivers/net/virtio_net.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 87b5c20..60abb5d 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -842,7 +842,7 @@ static unsigned int get_mergeable_buf_len(struct receive_queue *rq,
>  	unsigned int len;
>  
>  	len = hdr_len + clamp_t(unsigned int, ewma_pkt_len_read(avg_pkt_len),
> -				rq->min_buf_len - hdr_len, PAGE_SIZE - hdr_len);
> +				rq->min_buf_len, PAGE_SIZE - hdr_len);
>  	return ALIGN(len, L1_CACHE_BYTES);
>  }
>  
> @@ -2039,7 +2039,8 @@ static unsigned int mergeable_min_buf_len(struct virtnet_info *vi, struct virtqu
>  	unsigned int buf_len = hdr_len + ETH_HLEN + VLAN_HLEN + packet_len;
>  	unsigned int min_buf_len = DIV_ROUND_UP(buf_len, rq_size);
>  
> -	return max(min_buf_len, hdr_len);
> +	return max(max(min_buf_len, hdr_len) - hdr_len,
> +		   (unsigned int)GOOD_PACKET_LEN);
>  }
>  
>  static int virtnet_find_vqs(struct virtnet_info *vi)
> -- 
> MST
> 

Re: [PATCH] virtio_net: lower limit on buffer size

[J. Bruce Fields]

On Fri, Jun 02, 2017 at 12:34:57PM +0300, Sergei Shtylyov wrote:
> Hello!
> 
> On 6/2/2017 2:56 AM, Michael S. Tsirkin wrote:
> 
> >commit d85b758f72b0 "virtio_net: fix support for small rings"
> 
>    Commit d85b758f72b0 ("virtio_net: fix support for small rings")
> 
> >was supposed to increase the buffer size for small rings
> >but had an unintentional side effect of decreasing
> >it for large rings. This seems to break some setups -
> >it's not yet clear why, but increasing buffer size
> >back to what it was before helps.
> >
> >Fixes: d85b758f72b0 "virtio_net: fix support for small rings"
> 
> Fixes: d85b758f72b0 ("virtio_net: fix support for small rings")

I may be bikeshedding, but, personally I never do the parens--they're
redundant given the quotes, and space is often tight.

--b.

> 
> >Reported-by: Mikulas Patocka <mpatocka@redhat.com>
> >Reported-by: "J. Bruce Fields" <bfields@fieldses.org>
> >Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> [...]
> 
> MBR, Sergei

Re: [PATCH] virtio_net: lower limit on buffer size

[Sergei Shtylyov]

On 06/02/2017 11:25 PM, J. Bruce Fields wrote:

>>> commit d85b758f72b0 "virtio_net: fix support for small rings"
>>
>>    Commit d85b758f72b0 ("virtio_net: fix support for small rings")
>>
>>> was supposed to increase the buffer size for small rings
>>> but had an unintentional side effect of decreasing
>>> it for large rings. This seems to break some setups -
>>> it's not yet clear why, but increasing buffer size
>>> back to what it was before helps.
>>>
>>> Fixes: d85b758f72b0 "virtio_net: fix support for small rings"
>>
>> Fixes: d85b758f72b0 ("virtio_net: fix support for small rings")
>
> I may be bikeshedding, but, personally I never do the parens--they're
> redundant given the quotes, and space is often tight.

    Just see Documetation/process/submitting-patches.rst.

MBR, Sergei

Re: [PATCH] virtio_net: lower limit on buffer size

[J. Bruce Fields]

On Sat, Jun 03, 2017 at 11:17:30PM +0300, Sergei Shtylyov wrote:
> On 06/02/2017 11:25 PM, J. Bruce Fields wrote:
> 
> >>>commit d85b758f72b0 "virtio_net: fix support for small rings"
> >>
> >>   Commit d85b758f72b0 ("virtio_net: fix support for small rings")
> >>
> >>>was supposed to increase the buffer size for small rings
> >>>but had an unintentional side effect of decreasing
> >>>it for large rings. This seems to break some setups -
> >>>it's not yet clear why, but increasing buffer size
> >>>back to what it was before helps.
> >>>
> >>>Fixes: d85b758f72b0 "virtio_net: fix support for small rings"
> >>
> >>Fixes: d85b758f72b0 ("virtio_net: fix support for small rings")
> >
> >I may be bikeshedding, but, personally I never do the parens--they're
> >redundant given the quotes, and space is often tight.
> 
>    Just see Documetation/process/submitting-patches.rst.

Yeah, I know, I claim it's a bad rule (but I'm too lazy to send a patch,
so, weight my opinion accordingly).

--b.