Re: [PATCH v2 4/4] kvm tools: Fix virtio console hangs by removing IRQ injection for tx path
On Sun, 2011-05-08 at 21:09 +0800, Asias He wrote: As virtio spec says: Because this is high importance and low bandwidth, the current Linux implementation polls for the buffer to be used, rather than waiting for an interrupt, simplifying the implementation signicantly. drivers/char/virtio_console.c send_buf() { ... /* Tell Host to go! */ virtqueue_kick(out_vq); ... while (!virtqueue_get_buf(out_vq, len)) cpu_relax(); ... } The console hangs can simply be reproduced by yes command which gives tremendous console IOs and IRQs. Sasha, does this fix the hangs you were seeing? We should re-enable virtio console unconditionally if this does - that increases test coverage for virtio console. Pekka -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH v2 4/4] kvm tools: Fix virtio console hangs by removing IRQ injection for tx path
On Sun, 2011-05-08 at 20:05 +0300, Pekka Enberg wrote: On Sun, 2011-05-08 at 21:09 +0800, Asias He wrote: As virtio spec says: Because this is high importance and low bandwidth, the current Linux implementation polls for the buffer to be used, rather than waiting for an interrupt, simplifying the implementation signicantly. drivers/char/virtio_console.c send_buf() { ... /* Tell Host to go! */ virtqueue_kick(out_vq); ... while (!virtqueue_get_buf(out_vq, len)) cpu_relax(); ... } The console hangs can simply be reproduced by yes command which gives tremendous console IOs and IRQs. Sasha, does this fix the hangs you were seeing? We should re-enable virtio console unconditionally if this does - that increases test coverage for virtio console. I'm seeing no more hangs, but why enable it unconditionally? Maybe enable it by default, but we shouldn't force the activation of virtio modules if the user doesn't want them. -- Sasha. -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH v2 4/4] kvm tools: Fix virtio console hangs by removing IRQ injection for tx path
On Sun, May 8, 2011 at 8:21 PM, Sasha Levin levinsasha...@gmail.com wrote: I'm seeing no more hangs, but why enable it unconditionally? Maybe enable it by default, but we shouldn't force the activation of virtio modules if the user doesn't want them. I meant enabling the device on PCI bus like we did before. -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH v2 4/4] kvm tools: Fix virtio console hangs by removing IRQ injection for tx path
On Sun, 2011-05-08 at 20:28 +0300, Pekka Enberg wrote: On Sun, May 8, 2011 at 8:21 PM, Sasha Levin levinsasha...@gmail.com wrote: I'm seeing no more hangs, but why enable it unconditionally? Maybe enable it by default, but we shouldn't force the activation of virtio modules if the user doesn't want them. I meant enabling the device on PCI bus like we did before. Thats what I've meant too. virtio-console is the only device which got initialized even if it wasn't requested (even when '-c serial' was passed specifically). -- Sasha. -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH v2 4/4] kvm tools: Fix virtio console hangs by removing IRQ injection for tx path
On Sun, May 8, 2011 at 8:29 PM, Sasha Levin levinsasha...@gmail.com wrote: On Sun, 2011-05-08 at 20:28 +0300, Pekka Enberg wrote: On Sun, May 8, 2011 at 8:21 PM, Sasha Levin levinsasha...@gmail.com wrote: I'm seeing no more hangs, but why enable it unconditionally? Maybe enable it by default, but we shouldn't force the activation of virtio modules if the user doesn't want them. I meant enabling the device on PCI bus like we did before. Thats what I've meant too. virtio-console is the only device which got initialized even if it wasn't requested (even when '-c serial' was passed specifically). The more options we have, the more combinations we need to test. What's the downside of enabling virtio console by default? The upside is that it's less likely to break. Btw, we should probably do that for virtio rng as well. -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH v2 4/4] kvm tools: Fix virtio console hangs by removing IRQ injection for tx path
On Sun, 2011-05-08 at 20:34 +0300, Pekka Enberg wrote: On Sun, May 8, 2011 at 8:29 PM, Sasha Levin levinsasha...@gmail.com wrote: On Sun, 2011-05-08 at 20:28 +0300, Pekka Enberg wrote: On Sun, May 8, 2011 at 8:21 PM, Sasha Levin levinsasha...@gmail.com wrote: I'm seeing no more hangs, but why enable it unconditionally? Maybe enable it by default, but we shouldn't force the activation of virtio modules if the user doesn't want them. I meant enabling the device on PCI bus like we did before. Thats what I've meant too. virtio-console is the only device which got initialized even if it wasn't requested (even when '-c serial' was passed specifically). The more options we have, the more combinations we need to test. What's the downside of enabling virtio console by default? The upside is that it's less likely to break. Btw, we should probably do that for virtio rng as well. I fully support enabling it by default, I'm against enabling it even if the user asked to have it disabled. -- Sasha. -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH v2 4/4] kvm tools: Fix virtio console hangs by removing IRQ injection for tx path
On Sun, 2011-05-08 at 20:35 +0300, Sasha Levin wrote: I fully support enabling it by default, I'm against enabling it even if the user asked to have it disabled. Sure. -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html