similar to: performance regression in virtio-net in 2.6.32-rc4

First, I apologize if this message has been received multiple times. I''m having problems subscribing to this mailing list: Hi xen-users, I am trying to decide whether I should run a game server inside a Xen domain. My primary reason for wanting to virtualize is because I want to isolate this environment from the rest of my server. I really like the idea of isolating the game server

box. I'll send an updated KVM tools patch in a bit as well. Before: # netperf -H 192.168.33.4,ipv4 -t TCP_RR MIGRATED TCP REQUEST/RESPONSE TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.33.4 (192.168.33.4) port 0 AF_INET : first burst 0 Local /Remote Socket Size Request Resp. Elapsed Trans. Send Recv Size Size Time Rate bytes Bytes bytes bytes

box. I'll send an updated KVM tools patch in a bit as well. Before: # netperf -H 192.168.33.4,ipv4 -t TCP_RR MIGRATED TCP REQUEST/RESPONSE TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.33.4 (192.168.33.4) port 0 AF_INET : first burst 0 Local /Remote Socket Size Request Resp. Elapsed Trans. Send Recv Size Size Time Rate bytes Bytes bytes bytes

Commit 2613af0ed18a11d5c566a81f9a6510b73180660a (virtio_net: migrate mergeable rx buffers to page frag allocators) try to increase the payload/truesize for MTU-sized traffic. But this will introduce the extra overhead for GSO packets received because of the frag list. This commit tries to reduce this issue by coalesce the possible rx frags when possible during rx. Test result shows the about 15%

Commit 2613af0ed18a11d5c566a81f9a6510b73180660a (virtio_net: migrate mergeable rx buffers to page frag allocators) try to increase the payload/truesize for MTU-sized traffic. But this will introduce the extra overhead for GSO packets received because of the frag list. This commit tries to reduce this issue by coalesce the possible rx frags when possible during rx. Test result shows the about 15%

Commit 2613af0ed18a11d5c566a81f9a6510b73180660a (virtio_net: migrate mergeable rx buffers to page frag allocators) try to increase the payload/truesize for MTU-sized traffic. But this will introduce the extra overhead for GSO packets received because of the frag list. This commit tries to reduce this issue by coalesce the possible rx frags when possible during rx. Test result shows the about 15%

Folks, I''m playing with maxbw on links (as opposed to flows) in Crossbow, and I have a couple of questions. First, the limts seem only advisory. The first example has the main host talking to a zone that has 172.16.17.100 configured on znic0. When there is no maxbw, the throughtput is as expected; when maxbw is 55M the throughput only drops to 76 Mbps: # netperf -H

Currently if VIRTIO_RING_F_INDIRECT_DESC is enabled we will use indirect descriptors even if we have plenty of space in the ring. This means that we take a performance hit at all times due to the overhead of creating indirect descriptors. With this patch, we will use indirect descriptors only if we have less than either 16, or 12% of the total amount of descriptors available. I did basic

Currently if VIRTIO_RING_F_INDIRECT_DESC is enabled we will use indirect descriptors even if we have plenty of space in the ring. This means that we take a performance hit at all times due to the overhead of creating indirect descriptors. With this patch, we will use indirect descriptors only if we have less than either 16, or 12% of the total amount of descriptors available. I did basic

On Wed, Jun 10, 2020 at 06:18:32PM +0200, Eugenio Perez Martin wrote: > On Wed, Jun 10, 2020 at 5:13 PM Michael S. Tsirkin <mst at redhat.com> wrote: > > > > On Wed, Jun 10, 2020 at 02:37:50PM +0200, Eugenio Perez Martin wrote: > > > > +/* This function returns a value > 0 if a descriptor was found, or 0 if none were found. > > > > + * A negative

For small packets we can simplify xmit processing by linearizing buffers with the header: most packets seem to have enough head room we can use for this purpose. Since some older hypervisors (e.g. qemu before version 1.5) required that header is the first s/g element, we need a feature bit for this. Signed-off-by: Michael S. Tsirkin <mst at redhat.com> --- This is a repost of my old

For small packets we can simplify xmit processing by linearizing buffers with the header: most packets seem to have enough head room we can use for this purpose. Since some older hypervisors (e.g. qemu before version 1.5) required that header is the first s/g element, we need a feature bit for this. Signed-off-by: Michael S. Tsirkin <mst at redhat.com> --- This is a repost of my old

On Tue, Jun 16, 2020 at 05:23:43PM +0200, Eugenio Perez Martin wrote: > On Mon, Jun 15, 2020 at 6:05 PM Eugenio P??rez <eperezma at redhat.com> wrote: > > > > On Thu, 2020-06-11 at 07:30 -0400, Michael S. Tsirkin wrote: > > > On Wed, Jun 10, 2020 at 06:18:32PM +0200, Eugenio Perez Martin wrote: > > > > On Wed, Jun 10, 2020 at 5:13 PM Michael S. Tsirkin

Since commit 39e6c8208d7b ("net: solve a NAPI race") napi has been able to be rescheduled within napi_complete_done() even in non-busypoll case, but virtnet_poll() always enabled interrupts before complete, and when napi was rescheduled within napi_complete_done() it did not disable interrupts. This caused more interrupts when event idx is disabled. According to commit cbdadbbf0c79

Since commit 39e6c8208d7b ("net: solve a NAPI race") napi has been able to be rescheduled within napi_complete_done() even in non-busypoll case, but virtnet_poll() always enabled interrupts before complete, and when napi was rescheduled within napi_complete_done() it did not disable interrupts. This caused more interrupts when event idx is disabled. According to commit cbdadbbf0c79

On 06/06/2013 05:55 PM, Michael S. Tsirkin wrote: > For small packets we can simplify xmit processing by linearizing buffers > with the header: most packets seem to have enough head room we can use > for this purpose. > > Since some older hypervisors (e.g. qemu before version 1.5) > required that header is the first s/g element, > we need a feature bit for this. > >

Sometimes we need to coalesce the rx frags to avoid frag list. One example is virtio-net driver which tries to use small frags for both MTU sized packet and GSO packet. So this patch introduce skb_coalesce_rx_frag() to do this. Cc: Rusty Russell <rusty at rustcorp.com.au> Cc: Michael S. Tsirkin <mst at redhat.com> Cc: Michael Dalton <mwdalton at google.com> Cc: Eric Dumazet

Sometimes we need to coalesce the rx frags to avoid frag list. One example is virtio-net driver which tries to use small frags for both MTU sized packet and GSO packet. So this patch introduce skb_coalesce_rx_frag() to do this. Cc: Rusty Russell <rusty at rustcorp.com.au> Cc: Michael S. Tsirkin <mst at redhat.com> Cc: Michael Dalton <mwdalton at google.com> Cc: Eric Dumazet

Sometimes we need to coalesce the rx frags to avoid frag list. One example is virtio-net driver which tries to use small frags for both MTU sized packet and GSO packet. So this patch introduce skb_coalesce_rx_frag() to do this. Cc: Rusty Russell <rusty at rustcorp.com.au> Cc: Michael S. Tsirkin <mst at redhat.com> Cc: Michael Dalton <mwdalton at google.com> Cc: Eric Dumazet

Sometimes we need to coalesce the rx frags to avoid frag list. One example is virtio-net driver which tries to use small frags for both MTU sized packet and GSO packet. So this patch introduce skb_coalesce_rx_frag() to do this. Cc: Rusty Russell <rusty at rustcorp.com.au> Cc: Michael S. Tsirkin <mst at redhat.com> Cc: Michael Dalton <mwdalton at google.com> Cc: Eric Dumazet