Re: [kvm-devel] [virtio-net][PATCH] Don't arm tx hrtimer with a constant 500us each transmit
Rusty Russell wrote: On Wednesday 12 December 2007 23:54:00 Dor Laor wrote: commit 763769621d271d92204ed27552d75448587c1ac0 Author: Dor Laor [EMAIL PROTECTED] Date: Wed Dec 12 14:52:00 2007 +0200 [virtio-net][PATCH] Don't arm tx hrtimer with a constant 50us each transmit The current start_xmit sets 500us hrtimer to kick the host. The problem is that if another xmit happens before the timer was fired then the first xmit will have to wait additional 500us. This patch does not re-arm the timer if there is existing one. This will shorten the latency for tx. Hi Dor! Yes, I pondered this when I wrote the code. On the one hand, it's a low-probability pathological corner case, on the other, your patch reduces the number of timer reprograms in the normal case. One thing that came up in our discussions is to let the host do the timer processing instead of the guest. When tx exit mitigation is enabled, the guest bumps the queue pointer, but carefully refrains from kicking the host. The host polls the tx pointer using a timer, kicking itself periodically; if polling yields no packets it disables tx exit mitigation. This saves the guest the bother of programming the timer, which presumably requires an exit if the timer is the closest one to expiration. [btw, this can be implemented in virtqueue rather than virtio-net, no?] -- Any sufficiently difficult bug is indistinguishable from a feature. -- To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [kvm-devel] [virtio-net][PATCH] Don't arm tx hrtimer with a constant 500us each transmit
On Tuesday 18 December 2007 16:30:08 Avi Kivity wrote: Rusty Russell wrote: Yes, I pondered this when I wrote the code. On the one hand, it's a low-probability pathological corner case, on the other, your patch reduces the number of timer reprograms in the normal case. One thing that came up in our discussions is to let the host do the timer processing instead of the guest. When tx exit mitigation is enabled, the guest bumps the queue pointer, but carefully refrains from kicking the host. The host polls the tx pointer using a timer, kicking itself periodically; if polling yields no packets it disables tx exit mitigation. This saves the guest the bother of programming the timer, which presumably requires an exit if the timer is the closest one to expiration. [btw, this can be implemented in virtqueue rather than virtio-net, no?] Yes, the current patch is a hack (look at the hardcoded constant); wanted to see how much it helps, if any. More sophisticated timer management would be a definite win... funny, I have a patch here which helps that Rusty. -- To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [kvm-devel] [virtio-net][PATCH] Don't arm tx hrtimer with a constant 500us each transmit
Christian Borntraeger wrote: Am Mittwoch, 12. Dezember 2007 schrieb Dor Laor: Christian Borntraeger wrote: Am Mittwoch, 12. Dezember 2007 schrieb Dor Laor: --- a/drivers/net/virtio_net.c +++ b/drivers/net/virtio_net.c @@ -406,10 +405,10 @@ again: Hmm, while I agree in general with the patch, I fail to find the proper version of virtio_net where this patch applies. I tried kvm.git and linux-2.6.git from kernel.org. Can you give me a pointer to the repository where you work on virtio? Sorry for that, I added some debug prints of my one. Here it is: *git clone git*://kvm.*qumranet*.com/home/*dor*/src/linux-2.6-nv use branch 'virtio'. Ah, ok. I will look into that branch. BTW: what git repository do you use? I use Avis git from kernel.org: git://git.kernel.org/pub/scm/linux/kernel/git/avi/kvm I patch it with Anthony's http://hg.codemonkey.ws/linux-virtio/ Over that I send the patches. One can use my repository directly. Christian -- To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [kvm-devel] [virtio-net][PATCH] Don't arm tx hrtimer with a constant 500us each transmit
Am Mittwoch, 12. Dezember 2007 schrieb Dor Laor: --- a/drivers/net/virtio_net.c +++ b/drivers/net/virtio_net.c @@ -406,10 +405,10 @@ again: Hmm, while I agree in general with the patch, I fail to find the proper version of virtio_net where this patch applies. I tried kvm.git and linux-2.6.git from kernel.org. Can you give me a pointer to the repository where you work on virtio? Christian -- To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [kvm-devel] [virtio-net][PATCH] Don't arm tx hrtimer with a constant 500us each transmit
Christian Borntraeger wrote: Am Mittwoch, 12. Dezember 2007 schrieb Dor Laor: --- a/drivers/net/virtio_net.c +++ b/drivers/net/virtio_net.c @@ -406,10 +405,10 @@ again: Hmm, while I agree in general with the patch, I fail to find the proper version of virtio_net where this patch applies. I tried kvm.git and linux-2.6.git from kernel.org. Can you give me a pointer to the repository where you work on virtio? Sorry for that, I added some debug prints of my one. Here it is: *git clone git*://kvm.*qumranet*.com/home/*dor*/src/linux-2.6-nv use branch 'virtio'. BTW: what git repository do you use? This patch improves my tx performance from 720Mbps to 900Mbps. Dor Christian -- To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [kvm-devel] [virtio-net][PATCH] Don't arm tx hrtimer with a constant 500us each transmit
Am Mittwoch, 12. Dezember 2007 schrieb Dor Laor: Christian Borntraeger wrote: Am Mittwoch, 12. Dezember 2007 schrieb Dor Laor: --- a/drivers/net/virtio_net.c +++ b/drivers/net/virtio_net.c @@ -406,10 +405,10 @@ again: Hmm, while I agree in general with the patch, I fail to find the proper version of virtio_net where this patch applies. I tried kvm.git and linux-2.6.git from kernel.org. Can you give me a pointer to the repository where you work on virtio? Sorry for that, I added some debug prints of my one. Here it is: *git clone git*://kvm.*qumranet*.com/home/*dor*/src/linux-2.6-nv use branch 'virtio'. Ah, ok. I will look into that branch. BTW: what git repository do you use? I use Avis git from kernel.org: git://git.kernel.org/pub/scm/linux/kernel/git/avi/kvm Christian -- To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html