is patch landed in today's linux-next (next-20141015).
> include/linux/virtio.h | 6 +
> drivers/virtio/virtio.c | 54
> +
> drivers/virtio/virtio_pci.c | 54
> ++---
> 3 file
On 10/15/2014 07:06 AM, Michael S. Tsirkin wrote:
> On Tue, Oct 14, 2014 at 02:53:27PM -0400, David Miller wrote:
>> > From: Jason Wang
>> > Date: Sat, 11 Oct 2014 15:16:43 +0800
>> >
>>> > > We free old transmitted packets in ndo_start_xmit() currently, so any
>>> > > packet must be orphaned als
This patch returns the number of packets sent in free_old_xmit_skbs(), this is
necessary for calling this function in napi.
Cc: Rusty Russell
Cc: Michael S. Tsirkin
Signed-off-by: Jason Wang
---
drivers/net/virtio_net.c |4 +++-
1 files changed, 3 insertions(+), 1 deletions(-)
diff --git
Accumulate the sent packets and sent bytes in local variables and perform a
single u64_stats_update_begin/end() after.
Cc: Rusty Russell
Cc: Michael S. Tsirkin
Signed-off-by: Jason Wang
---
drivers/net/virtio_net.c | 12
1 files changed, 8 insertions(+), 4 deletions(-)
diff --g
This patch introduces virtio_enable_cb_avail() to publish avail idx
and used event. This could be used by batched buffer submitting to
reduce the number of tx interrupts.
Cc: Rusty Russell
Cc: Michael S. Tsirkin
Signed-off-by: Jason Wang
---
drivers/virtio/virtio_ring.c | 22
According to David, proper accounting and queueing (at all levels, not
just TCP sockets) is more important than trying to skim a bunch of
cycles by avoiding TX interrupts. Having an event to free the SKB is
absolutely essential for the stack to operate correctly.
This series tries to enable tx int
With the help of xmit_more and virtqueue_enable_cb_avail(), this patch
enable tx interrupt only for the final skb in the chain if
needed. This will help to mitigate tx interrupts.
Cc: Rusty Russell
Cc: Michael S. Tsirkin
Signed-off-by: Jason Wang
---
drivers/net/virtio_net.c | 10 +++---
Orphan skb in ndo_start_xmit() breaks socket accounting and packet
queuing. This in fact breaks lots of things such as pktgen and several
TCP optimizations. And also make BQL can't be implemented for
virtio-net.
This patch tries to solve this issue by enabling tx interrupt. To
avoid introducing ex
This patch checks the new event idx to make sure used event idx never
goes back. This is used to synchronize the calls between
virtqueue_enable_cb_delayed() and virtqueue_enable_cb().
Cc: Rusty Russell
Cc: Michael S. Tsirkin
Signed-off-by: Jason Wang
---
drivers/virtio/virtio_ring.c |7 +++
On Wed, Oct 15, 2014 at 03:25:26PM +0800, Jason Wang wrote:
> This patch introduces virtio_enable_cb_avail() to publish avail idx
> and used event. This could be used by batched buffer submitting to
> reduce the number of tx interrupts.
>
> Cc: Rusty Russell
> Cc: Michael S. Tsirkin
> Signed-off
On Wed, Oct 15, 2014 at 03:25:25PM +0800, Jason Wang wrote:
> This patch checks the new event idx to make sure used event idx never
> goes back. This is used to synchronize the calls between
> virtqueue_enable_cb_delayed() and virtqueue_enable_cb().
>
> Cc: Rusty Russell
> Cc: Michael S. Tsirkin
On Wed, Oct 15, 2014 at 03:25:27PM +0800, Jason Wang wrote:
> Accumulate the sent packets and sent bytes in local variables and perform a
> single u64_stats_update_begin/end() after.
>
> Cc: Rusty Russell
> Cc: Michael S. Tsirkin
> Signed-off-by: Jason Wang
Not sure how much it's worth but sin
On Wed, 2014-10-15 at 15:25 +0800, Jason Wang wrote:
> Accumulate the sent packets and sent bytes in local variables and perform a
> single u64_stats_update_begin/end() after.
>
> Cc: Rusty Russell
> Cc: Michael S. Tsirkin
> Signed-off-by: Jason Wang
> ---
> drivers/net/virtio_net.c | 12 +++
On Wed, 2014-10-15 at 15:25 +0800, Jason Wang wrote:
...
> +static int free_old_xmit_skbs(struct send_queue *sq, int budget)
> +{
> + struct sk_buff *skb;
> + unsigned int len;
> + struct virtnet_info *vi = sq->vq->vdev->priv;
> + struct virtnet_stats *stats = this_cpu_ptr(vi->sta
From: Of Michael S. Tsirkin
> On Wed, Oct 15, 2014 at 03:25:27PM +0800, Jason Wang wrote:
> > Accumulate the sent packets and sent bytes in local variables and perform a
> > single u64_stats_update_begin/end() after.
> >
> > Cc: Rusty Russell
> > Cc: Michael S. Tsirkin
> > Signed-off-by: Jason Wa
On 10/15/2014 05:34 PM, Michael S. Tsirkin wrote:
> On Wed, Oct 15, 2014 at 03:25:25PM +0800, Jason Wang wrote:
>> This patch checks the new event idx to make sure used event idx never
>> goes back. This is used to synchronize the calls between
>> virtqueue_enable_cb_delayed() and virtqueue_enable_
On Wed, Oct 15, 2014 at 03:25:29PM +0800, Jason Wang wrote:
> Orphan skb in ndo_start_xmit() breaks socket accounting and packet
> queuing. This in fact breaks lots of things such as pktgen and several
> TCP optimizations. And also make BQL can't be implemented for
> virtio-net.
>
> This patch tri
On 10/15/2014 05:28 PM, Michael S. Tsirkin wrote:
> On Wed, Oct 15, 2014 at 03:25:26PM +0800, Jason Wang wrote:
>> This patch introduces virtio_enable_cb_avail() to publish avail idx
>> and used event. This could be used by batched buffer submitting to
>> reduce the number of tx interrupts.
>>
>> C
On Wed, Oct 15, 2014 at 03:25:30PM +0800, Jason Wang wrote:
> With the help of xmit_more and virtqueue_enable_cb_avail(), this patch
> enable tx interrupt only for the final skb in the chain if
> needed. This will help to mitigate tx interrupts.
>
> Cc: Rusty Russell
> Cc: Michael S. Tsirkin
> S
On 10/15/2014 05:37 PM, Eric Dumazet wrote:
> On Wed, 2014-10-15 at 15:25 +0800, Jason Wang wrote:
>
> ...
>
>> +static int free_old_xmit_skbs(struct send_queue *sq, int budget)
>> +{
>> +struct sk_buff *skb;
>> +unsigned int len;
>> +struct virtnet_info *vi = sq->vq->vdev->priv;
>> +
On Wed, Oct 15, 2014 at 03:25:24PM +0800, Jason Wang wrote:
> According to David, proper accounting and queueing (at all levels, not
> just TCP sockets) is more important than trying to skim a bunch of
> cycles by avoiding TX interrupts.
He also mentioned we should find other ways to batch
> Havi
On 10/15/2014 06:18 PM, Michael S. Tsirkin wrote:
> On Wed, Oct 15, 2014 at 03:25:29PM +0800, Jason Wang wrote:
>> > Orphan skb in ndo_start_xmit() breaks socket accounting and packet
>> > queuing. This in fact breaks lots of things such as pktgen and several
>> > TCP optimizations. And also make B
On Wed, Oct 15, 2014 at 06:13:19PM +0800, Jason Wang wrote:
> On 10/15/2014 05:34 PM, Michael S. Tsirkin wrote:
> > On Wed, Oct 15, 2014 at 03:25:25PM +0800, Jason Wang wrote:
> >> This patch checks the new event idx to make sure used event idx never
> >> goes back. This is used to synchronize the
On 10/15/2014 06:22 PM, Michael S. Tsirkin wrote:
> On Wed, Oct 15, 2014 at 03:25:30PM +0800, Jason Wang wrote:
>> With the help of xmit_more and virtqueue_enable_cb_avail(), this patch
>> enable tx interrupt only for the final skb in the chain if
>> needed. This will help to mitigate tx interrupts
On Wed, Oct 15, 2014 at 06:19:15PM +0800, Jason Wang wrote:
> On 10/15/2014 05:28 PM, Michael S. Tsirkin wrote:
> > On Wed, Oct 15, 2014 at 03:25:26PM +0800, Jason Wang wrote:
> >> This patch introduces virtio_enable_cb_avail() to publish avail idx
> >> and used event. This could be used by batched
On Wed, Oct 15, 2014 at 06:25:25PM +0800, Jason Wang wrote:
> On 10/15/2014 06:18 PM, Michael S. Tsirkin wrote:
> > On Wed, Oct 15, 2014 at 03:25:29PM +0800, Jason Wang wrote:
> >> > Orphan skb in ndo_start_xmit() breaks socket accounting and packet
> >> > queuing. This in fact breaks lots of thing
On Wed, Oct 15, 2014 at 06:31:19PM +0800, Jason Wang wrote:
> On 10/15/2014 06:22 PM, Michael S. Tsirkin wrote:
> > On Wed, Oct 15, 2014 at 03:25:30PM +0800, Jason Wang wrote:
> >> With the help of xmit_more and virtqueue_enable_cb_avail(), this patch
> >> enable tx interrupt only for the final skb
On 10/15/2014 06:32 PM, Michael S. Tsirkin wrote:
> On Wed, Oct 15, 2014 at 06:13:19PM +0800, Jason Wang wrote:
>> On 10/15/2014 05:34 PM, Michael S. Tsirkin wrote:
>>> On Wed, Oct 15, 2014 at 03:25:25PM +0800, Jason Wang wrote:
This patch checks the new event idx to make sure used event idx n
On Wed, Oct 15, 2014 at 09:49:01AM +, David Laight wrote:
> From: Of Michael S. Tsirkin
> > On Wed, Oct 15, 2014 at 03:25:27PM +0800, Jason Wang wrote:
> > > Accumulate the sent packets and sent bytes in local variables and perform
> > > a
> > > single u64_stats_update_begin/end() after.
> > >
From: Michael S. Tsirkin
> On Wed, Oct 15, 2014 at 09:49:01AM +, David Laight wrote:
> > From: Of Michael S. Tsirkin
> > > On Wed, Oct 15, 2014 at 03:25:27PM +0800, Jason Wang wrote:
> > > > Accumulate the sent packets and sent bytes in local variables and
> > > > perform a
> > > > single u64_
On 10/15/2014 06:41 PM, Michael S. Tsirkin wrote:
> On Wed, Oct 15, 2014 at 06:19:15PM +0800, Jason Wang wrote:
>> On 10/15/2014 05:28 PM, Michael S. Tsirkin wrote:
>>> On Wed, Oct 15, 2014 at 03:25:26PM +0800, Jason Wang wrote:
This patch introduces virtio_enable_cb_avail() to publish avail i
On 10/15/2014 06:43 PM, Michael S. Tsirkin wrote:
> On Wed, Oct 15, 2014 at 06:25:25PM +0800, Jason Wang wrote:
>> > On 10/15/2014 06:18 PM, Michael S. Tsirkin wrote:
>>> > > On Wed, Oct 15, 2014 at 03:25:29PM +0800, Jason Wang wrote:
> > >> > Orphan skb in ndo_start_xmit() breaks socket accoun
On 10/15/2014 05:53 AM, Michael S. Tsirkin wrote:
> On newer hosts that support delayed tx interrupts,
> we probably don't have much to gain from orphaning
> packets early.
>
> Based on patch by Jason Wang.
>
> Note: this will likely degrade performance for hosts without event idx
> support. Vario
On 10/15/2014 06:25 PM, Michael S. Tsirkin wrote:
> On Wed, Oct 15, 2014 at 03:25:24PM +0800, Jason Wang wrote:
>> According to David, proper accounting and queueing (at all levels, not
>> just TCP sockets) is more important than trying to skim a bunch of
>> cycles by avoiding TX interrupts.
> He a
On Wed, Oct 15, 2014 at 06:44:41PM +0800, Jason Wang wrote:
> On 10/15/2014 06:32 PM, Michael S. Tsirkin wrote:
> > On Wed, Oct 15, 2014 at 06:13:19PM +0800, Jason Wang wrote:
> >> On 10/15/2014 05:34 PM, Michael S. Tsirkin wrote:
> >>> On Wed, Oct 15, 2014 at 03:25:25PM +0800, Jason Wang wrote:
>
On Wed, Oct 15, 2014 at 06:58:15PM +0800, Jason Wang wrote:
> On 10/15/2014 06:41 PM, Michael S. Tsirkin wrote:
> > On Wed, Oct 15, 2014 at 06:19:15PM +0800, Jason Wang wrote:
> >> On 10/15/2014 05:28 PM, Michael S. Tsirkin wrote:
> >>> On Wed, Oct 15, 2014 at 03:25:26PM +0800, Jason Wang wrote:
>
On Wed, Oct 15, 2014 at 07:04:20PM +0800, Jason Wang wrote:
> On 10/15/2014 05:53 AM, Michael S. Tsirkin wrote:
> > On newer hosts that support delayed tx interrupts,
> > we probably don't have much to gain from orphaning
> > packets early.
> >
> > Based on patch by Jason Wang.
> >
> > Note: this w
On Wed, Oct 15, 2014 at 07:14:14PM +0800, Jason Wang wrote:
> On 10/15/2014 06:25 PM, Michael S. Tsirkin wrote:
> > On Wed, Oct 15, 2014 at 03:25:24PM +0800, Jason Wang wrote:
> >> According to David, proper accounting and queueing (at all levels, not
> >> just TCP sockets) is more important than t
On Wed, Oct 15, 2014 at 10:51:32AM +, David Laight wrote:
> From: Michael S. Tsirkin
> > On Wed, Oct 15, 2014 at 09:49:01AM +, David Laight wrote:
> > > From: Of Michael S. Tsirkin
> > > > On Wed, Oct 15, 2014 at 03:25:27PM +0800, Jason Wang wrote:
> > > > > Accumulate the sent packets and
commit 0b725a2ca61bedc33a2a63d0451d528b268cf975
net: Remove ndo_xmit_flush netdev operation, use signalling instead.
added code that looks at skb->xmit_more after the skb has
been put in TX VQ. Since some paths process the ring and free the skb
immediately, this can cause use after free.
Fix
From: Michael S. Tsirkin
> commit 0b725a2ca61bedc33a2a63d0451d528b268cf975
> net: Remove ndo_xmit_flush netdev operation, use signalling instead.
>
> added code that looks at skb->xmit_more after the skb has
> been put in TX VQ. Since some paths process the ring and free the skb
> immediately,
On Wed, Oct 15, 2014 at 01:24:57PM +, David Laight wrote:
> From: Michael S. Tsirkin
> > commit 0b725a2ca61bedc33a2a63d0451d528b268cf975
> > net: Remove ndo_xmit_flush netdev operation, use signalling instead.
> >
> > added code that looks at skb->xmit_more after the skb has
> > been put i
On newer hosts that support delayed tx interrupts,
we probably don't have much to gain from orphaning
packets early.
Based on patch by Jason Wang.
Note: this might degrade performance for
hosts without event idx support.
Should be addressed by the next patch.
Signed-off-by: Michael S. Tsirkin
-
From: Jason Wang
We already have counters for sent packets and sent bytes.
Use them to reduce the number of u64_stats_update_begin/end().
Take care not to bother with stats update when called
speculatively.
Based on a patch by Jason Wang.
Cc: Rusty Russell
Signed-off-by: Jason Wang
Signed-of
Improve tx batching using byte queue limits.
Should be especially effective for MQ.
Signed-off-by: Michael S. Tsirkin
---
drivers/net/virtio_net.c | 20
1 file changed, 16 insertions(+), 4 deletions(-)
diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index a
From: Raushaniya Maksudova
This value would be useful in the next patch to provide the amount of
the freed memory for OOM killer.
Signed-off-by: Raushaniya Maksudova
Signed-off-by: Denis V. Lunev
CC: Rusty Russell
CC: Michael S. Tsirkin
---
drivers/virtio/virtio_balloon.c | 5 -
1 file
Excessive virtio_balloon inflation can cause invocation of OOM-killer, when
Linux is under severe memory pressure. Various mechanisms are responsible for
correct virtio_balloon memory management. Nevertheless it is often the case
that these control tools does not have enough time to react on fast c
From: Raushaniya Maksudova
Excessive virtio_balloon inflation can cause invocation of OOM-killer,
when Linux is under severe memory pressure. Various mechanisms are
responsible for correct virtio_balloon memory management. Nevertheless
it is often the case that these control tools does not have e
From: "Michael S. Tsirkin"
Date: Wed, 15 Oct 2014 16:23:28 +0300
> You used __netif_subqueue_stopped but that seems to use
> a slightly more expensive test_bit internally.
More expensive in what sense? It should be roughly the same
as "x & y" sans the volatile.
Anyways I'm ambivalent and I wan
49 matches
Mail list logo
