similar to: [PATCH RFC] virtio-pci: share config interrupt between virtio devices

On Mon, Sep 01, 2014 at 09:37:30AM +0300, Michael S. Tsirkin wrote: > Hi Michael, > On Mon, Sep 01, 2014 at 01:41:54PM +0800, Amos Kong wrote: > > One VM only has 128 msix interrupt, virtio-config interrupt > > has less workload. This patch shares one normal interrupt Thanks for your quick reply. > normal == INT#x? Please don't call it normal. The proper name is >

On Mon, Sep 01, 2014 at 09:37:30AM +0300, Michael S. Tsirkin wrote: > Hi Michael, > On Mon, Sep 01, 2014 at 01:41:54PM +0800, Amos Kong wrote: > > One VM only has 128 msix interrupt, virtio-config interrupt > > has less workload. This patch shares one normal interrupt Thanks for your quick reply. > normal == INT#x? Please don't call it normal. The proper name is >

On Mon, Sep 01, 2014 at 01:41:54PM +0800, Amos Kong wrote: > One VM only has 128 msix interrupt, virtio-config interrupt > has less workload. This patch shares one normal interrupt normal == INT#x? Please don't call it normal. The proper name is "legacy INT#x". So you are trying to use legacy INT#x at the same time with MSI-X? This does not work: the PCI spec says: While

This refactors find_vqs, making it more readable and robust, and fixing two regressions from 2.6.30: - double free_irq causing BUG_ON on device removal - probe failure when vq can't be assigned to msi-x vector (reported on old host kernels) An older version of this patch was tested by Amit Shah. Reported-by: Amit Shah <amit.shah at redhat.com> Signed-off-by: Michael S. Tsirkin

This refactors find_vqs, making it more readable and robust, and fixing two regressions from 2.6.30: - double free_irq causing BUG_ON on device removal - probe failure when vq can't be assigned to msi-x vector (reported on old host kernels) An older version of this patch was tested by Amit Shah. Reported-by: Amit Shah <amit.shah at redhat.com> Signed-off-by: Michael S. Tsirkin

From: Anthony Liguori <aliguori at amazon.com> If MSI-X initialization fails after setting msix_enabled = 1, then the device is left in an inconsistent state. This would normally only happen if there was a bug in the device emulation but it still should be handled correctly. Cc: Matt Wilson <msw at amazon.com> Cc: Rusty Russell <rusty at rustcorp.com.au> Cc: Michael Tsirkin

From: Anthony Liguori <aliguori at amazon.com> If MSI-X initialization fails after setting msix_enabled = 1, then the device is left in an inconsistent state. This would normally only happen if there was a bug in the device emulation but it still should be handled correctly. Cc: Matt Wilson <msw at amazon.com> Cc: Rusty Russell <rusty at rustcorp.com.au> Cc: Michael Tsirkin

Here's the patch series I ended up with. I haven't coded up the QEMU side yet, so no idea if the new driver works. Questions: (1) Do we win from separating ISR, NOTIFY and COMMON? (2) I used a "u8 bar"; should I use a bir and pack it instead? BIR seems a little obscure (noone else in the kernel source seems to refer to it). Cheers, Rusty.

Here's the patch series I ended up with. I haven't coded up the QEMU side yet, so no idea if the new driver works. Questions: (1) Do we win from separating ISR, NOTIFY and COMMON? (2) I used a "u8 bar"; should I use a bir and pack it instead? BIR seems a little obscure (noone else in the kernel source seems to refer to it). Cheers, Rusty.

Hi Michael, this series contains a couple cleanups for the virtio_pci interrupt handling code, including a switch to the new pci_irq_alloc_vectors helper. All these are in preparation of taking advantage of the new PCI layer / core IRQ interrupt affinity handling, for which I will send out a series once this and some core interrupt handling changes are in.

Hi Michael, this series contains a couple cleanups for the virtio_pci interrupt handling code, including a switch to the new pci_irq_alloc_vectors helper. All these are in preparation of taking advantage of the new PCI layer / core IRQ interrupt affinity handling, for which I will send out a series once this and some core interrupt handling changes are in.

Hi Michael, On Mon, Sep 15, 2014 at 1:32 AM, Michael S. Tsirkin <mst at redhat.com> wrote: > On Sun, Sep 14, 2014 at 08:23:26PM -0700, Anthony Liguori wrote: >> From: Anthony Liguori <aliguori at amazon.com> >> >> If MSI-X initialization fails after setting msix_enabled = 1, then >> the device is left in an inconsistent state. This would normally >>

Hi Michael, On Mon, Sep 15, 2014 at 1:32 AM, Michael S. Tsirkin <mst at redhat.com> wrote: > On Sun, Sep 14, 2014 at 08:23:26PM -0700, Anthony Liguori wrote: >> From: Anthony Liguori <aliguori at amazon.com> >> >> If MSI-X initialization fails after setting msix_enabled = 1, then >> the device is left in an inconsistent state. This would normally >>

This implements optional MSI-X support in virtio_pci. MSI-X is used whenever the host supports at least 2 MSI-X vectors: 1 for configuration changes and 1 for virtqueues. Per-virtqueue vectors are allocated if enough vectors available. Signed-off-by: Michael S. Tsirkin <mst at redhat.com> --- drivers/virtio/virtio_pci.c | 227 +++++++++++++++++++++++++++++++++++++++----

This implements optional MSI-X support in virtio_pci. MSI-X is used whenever the host supports at least 2 MSI-X vectors: 1 for configuration changes and 1 for virtqueues. Per-virtqueue vectors are allocated if enough vectors available. Signed-off-by: Michael S. Tsirkin <mst at redhat.com> --- drivers/virtio/virtio_pci.c | 227 +++++++++++++++++++++++++++++++++++++++----

Hi all, I'm working on this item in Upstream Networking todolist: | - Sharing config interrupts | Support more devices by sharing a single msi vector | between multiple virtio devices. | (Applies to virtio-blk too). I have this solution here, and only have draft patches of Solution 1 & 2, let's discuss if solution 3 is feasible. * We should not introduce perf regression in

Hi all, I'm working on this item in Upstream Networking todolist: | - Sharing config interrupts | Support more devices by sharing a single msi vector | between multiple virtio devices. | (Applies to virtio-blk too). I have this solution here, and only have draft patches of Solution 1 & 2, let's discuss if solution 3 is feasible. * We should not introduce perf regression in

I've renewed this again, with some comments from HPA. I've tried to keep the new patches separate, so you can see the changes since we last discussed this (and so it's easy to back it out if we decide it's insane). I haven't even looked at the QEMU side so this is completely untested. Comments gratefully received! Rusty. Michael S Tsirkin (1): pci: add pci_iomap_range

I've renewed this again, with some comments from HPA. I've tried to keep the new patches separate, so you can see the changes since we last discussed this (and so it's easy to back it out if we decide it's insane). I haven't even looked at the QEMU side so this is completely untested. Comments gratefully received! Rusty. Michael S Tsirkin (1): pci: add pci_iomap_range

Hi Michael, this series contains a couple cleanups for the virtio_pci interrupt handling code, including a switch to the new pci_irq_alloc_vectors helper, and support for automatic affinity by the PCI layer if the consumers ask for it. It then converts over virtio_blk to use this functionality so that it's blk-mq queues are aligned to the MSI-X vector routing. I have a similar patch in the