thr3ads.net - Linux Virtualization - [PATCH v2 10/15] virtio: add API to enable VQs early [Oct 2014]

If this information is useful, please help other people find it:
Share via:

Michael S. Tsirkin

2014-Oct-06 15:10 UTC

[PATCH v2 00/15] virtio: fix spec compliance issues

Rusty, I have a mind to include this patchset for this merge window.
Any input on this?

This fixes the following virtio spec compliance issues:
1. on restore, drivers use device before setting ACKNOWLEDGE and DRIVER bits
2. on probe, drivers aren't prepared to handle config interrupts
   arriving before probe returns
3. on probe, drivers use device before DRIVER_OK it set

Note that 1 is a clear violation of virtio spec 0.9 and 1.0,
so I am proposing the fix for stable. OTOH 2 is against 1.0 rules but
is a known documented bug in many drivers, so let's fix it going
forward, but it does not seem to be worth it to backport
the changes.

An error handling bugfix for virtio-net is also included.

2 is merely a theoretical race condition, but it seems important
to address to make sure that changes to address 3 do not introduce
instability.

Michael S. Tsirkin (15):
  virtio_pci: fix virtio spec compliance on restore
  virtio: unify config_changed handling
  virtio-pci: move freeze/restore to virtio core
  virtio: defer config changed notifications
  virtio_blk: drop config_enable
  virtio-blk: drop config_mutex
  virtio_net: drop config_enable
  virtio-net: drop config_mutex
  virtio_net: minor cleanup
  virtio: add API to enable VQs early
  virtio_net: enable VQs early
  virtio_blk: enable VQs early
  virtio_console: enable VQs early
  9p/trans_virtio: enable VQs early
  virtio_net: fix use after free on allocation failure

 include/linux/virtio.h             |  14 +++++
 include/linux/virtio_config.h      |  17 +++++++
 drivers/block/virtio_blk.c         |  31 ++----------
 drivers/char/virtio_console.c      |   2 +
 drivers/misc/mic/card/mic_virtio.c |   6 +--
 drivers/net/virtio_net.c           |  42 ++++-----------
 drivers/s390/kvm/kvm_virtio.c      |   9 +---
 drivers/s390/kvm/virtio_ccw.c      |   6 +--
 drivers/virtio/virtio.c            | 101 +++++++++++++++++++++++++++++++++++++
 drivers/virtio/virtio_mmio.c       |   7 +--
 drivers/virtio/virtio_pci.c        |  33 ++----------
 net/9p/trans_virtio.c              |   2 +
 12 files changed, 159 insertions(+), 111 deletions(-)

-- 
MST

Michael S. Tsirkin

2014-Oct-06 15:10 UTC

head link

[PATCH v2 01/15] virtio_pci: fix virtio spec compliance on restore

On restore, virtio pci does the following:
+ set features
+ init vqs etc - device can be used at this point!
+ set ACKNOWLEDGE,DRIVER and DRIVER_OK status bits

This is in violation of the virtio spec, which
requires the following order:
- ACKNOWLEDGE
- DRIVER
- init vqs
- DRIVER_OK

This behaviour will break with hypervisors that assume spec compliant
behaviour.  It seems like a good idea to have this patch applied to
stable branches to reduce the support butden for the hypervisors.

Cc: stable at vger.kernel.org
Cc: Amit Shah <amit.shah at redhat.com>
Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
---
 drivers/virtio/virtio_pci.c | 36 ++++++++++++++++++++++++++++++++----
 1 file changed, 32 insertions(+), 4 deletions(-)

diff --git a/drivers/virtio/virtio_pci.c b/drivers/virtio/virtio_pci.c
index 3d1463c..0f2db51 100644
--- a/drivers/virtio/virtio_pci.c
+++ b/drivers/virtio/virtio_pci.c
@@ -789,6 +789,7 @@ static int virtio_pci_restore(struct device *dev)
 	struct pci_dev *pci_dev = to_pci_dev(dev);
 	struct virtio_pci_device *vp_dev = pci_get_drvdata(pci_dev);
 	struct virtio_driver *drv;
+	unsigned status = 0;
 	int ret;
 
 	drv = container_of(vp_dev->vdev.dev.driver,
@@ -799,14 +800,41 @@ static int virtio_pci_restore(struct device *dev)
 		return ret;
 
 	pci_set_master(pci_dev);
+	/* We always start by resetting the device, in case a previous
+	 * driver messed it up. */
+	vp_reset(&vp_dev->vdev);
+
+	/* Acknowledge that we've seen the device. */
+	status |= VIRTIO_CONFIG_S_ACKNOWLEDGE;
+	vp_set_status(&vp_dev->vdev, status);
+
+	/* Maybe driver failed before freeze.
+	 * Restore the failed status, for debugging. */
+	status |= vp_dev->saved_status & VIRTIO_CONFIG_S_FAILED;
+	vp_set_status(&vp_dev->vdev, status);
+
+	if (!drv)
+		return 0;
+
+	/* We have a driver! */
+	status |= VIRTIO_CONFIG_S_DRIVER;
+	vp_set_status(&vp_dev->vdev, status);
+
 	vp_finalize_features(&vp_dev->vdev);
 
-	if (drv && drv->restore)
-		ret = drv->restore(&vp_dev->vdev);
+	if (!drv->restore)
+		return 0;
+
+	ret = drv->restore(&vp_dev->vdev);
+	if (ret) {
+		status |= VIRTIO_CONFIG_S_FAILED;
+		vp_set_status(&vp_dev->vdev, status);
+		return ret;
+	}
 
 	/* Finally, tell the device we're all set */
-	if (!ret)
-		vp_set_status(&vp_dev->vdev, vp_dev->saved_status);
+	status |= VIRTIO_CONFIG_S_DRIVER_OK;
+	vp_set_status(&vp_dev->vdev, status);
 
 	return ret;
 }
-- 
MST

Michael S. Tsirkin

2014-Oct-06 15:10 UTC

head link

[PATCH v2 02/15] virtio: unify config_changed handling

Replace duplicated code in all transports with a single wrapper in
virtio.c.

The only functional change is in virtio_mmio.c: if a buggy device sends
us an interrupt before driver is set, we previously returned IRQ_NONE,
now we return IRQ_HANDLED.

As this must not happen in practice, this does not look like a big deal.

See also commit 3fff0179e33cd7d0a688dab65700c46ad089e934
	virtio-pci: do not oops on config change if driver not loaded.
for the original motivation behind the driver check.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
---
 include/linux/virtio.h             | 2 ++
 drivers/misc/mic/card/mic_virtio.c | 6 +-----
 drivers/s390/kvm/kvm_virtio.c      | 9 +--------
 drivers/s390/kvm/virtio_ccw.c      | 6 +-----
 drivers/virtio/virtio.c            | 9 +++++++++
 drivers/virtio/virtio_mmio.c       | 7 ++-----
 drivers/virtio/virtio_pci.c        | 6 +-----
 7 files changed, 17 insertions(+), 28 deletions(-)

diff --git a/include/linux/virtio.h b/include/linux/virtio.h
index b46671e..3c19bd3 100644
--- a/include/linux/virtio.h
+++ b/include/linux/virtio.h
@@ -108,6 +108,8 @@ void unregister_virtio_device(struct virtio_device *dev);
 
 void virtio_break_device(struct virtio_device *dev);
 
+void virtio_config_changed(struct virtio_device *dev);
+
 /**
  * virtio_driver - operations for a virtio I/O driver
  * @driver: underlying device driver (populate name and owner).
diff --git a/drivers/misc/mic/card/mic_virtio.c
b/drivers/misc/mic/card/mic_virtio.c
index f14b600..e647947 100644
--- a/drivers/misc/mic/card/mic_virtio.c
+++ b/drivers/misc/mic/card/mic_virtio.c
@@ -462,16 +462,12 @@ static void mic_handle_config_change(struct
mic_device_desc __iomem *d,
 	struct mic_device_ctrl __iomem *dc
 		= (void __iomem *)d + mic_aligned_desc_size(d);
 	struct mic_vdev *mvdev = (struct mic_vdev *)ioread64(&dc->vdev);
-	struct virtio_driver *drv;
 
 	if (ioread8(&dc->config_change) != MIC_VIRTIO_PARAM_CONFIG_CHANGED)
 		return;
 
 	dev_dbg(mdrv->dev, "%s %d\n", __func__, __LINE__);
-	drv = container_of(mvdev->vdev.dev.driver,
-				struct virtio_driver, driver);
-	if (drv->config_changed)
-		drv->config_changed(&mvdev->vdev);
+	virtio_config_changed(&mvdev->vdev);
 	iowrite8(1, &dc->guest_ack);
 }
 
diff --git a/drivers/s390/kvm/kvm_virtio.c b/drivers/s390/kvm/kvm_virtio.c
index a134965..6431290 100644
--- a/drivers/s390/kvm/kvm_virtio.c
+++ b/drivers/s390/kvm/kvm_virtio.c
@@ -406,15 +406,8 @@ static void kvm_extint_handler(struct ext_code ext_code,
 
 	switch (param) {
 	case VIRTIO_PARAM_CONFIG_CHANGED:
-	{
-		struct virtio_driver *drv;
-		drv = container_of(vq->vdev->dev.driver,
-				   struct virtio_driver, driver);
-		if (drv->config_changed)
-			drv->config_changed(vq->vdev);
-
+		virtio_config_changed(vq->vdev);
 		break;
-	}
 	case VIRTIO_PARAM_DEV_ADD:
 		schedule_work(&hotplug_work);
 		break;
diff --git a/drivers/s390/kvm/virtio_ccw.c b/drivers/s390/kvm/virtio_ccw.c
index d2c0b44..6cbe6ef 100644
--- a/drivers/s390/kvm/virtio_ccw.c
+++ b/drivers/s390/kvm/virtio_ccw.c
@@ -940,11 +940,7 @@ static void virtio_ccw_int_handler(struct ccw_device *cdev,
 		vring_interrupt(0, vq);
 	}
 	if (test_bit(0, &vcdev->indicators2)) {
-		drv = container_of(vcdev->vdev.dev.driver,
-				   struct virtio_driver, driver);
-
-		if (drv && drv->config_changed)
-			drv->config_changed(&vcdev->vdev);
+		virtio_config_changed(&vcdev->vdev);
 		clear_bit(0, &vcdev->indicators2);
 	}
 }
diff --git a/drivers/virtio/virtio.c b/drivers/virtio/virtio.c
index fed0ce1..3980687 100644
--- a/drivers/virtio/virtio.c
+++ b/drivers/virtio/virtio.c
@@ -239,6 +239,15 @@ void unregister_virtio_device(struct virtio_device *dev)
 }
 EXPORT_SYMBOL_GPL(unregister_virtio_device);
 
+void virtio_config_changed(struct virtio_device *dev)
+{
+	struct virtio_driver *drv = drv_to_virtio(dev->dev.driver);
+
+	if (drv && drv->config_changed)
+		drv->config_changed(dev);
+}
+EXPORT_SYMBOL_GPL(virtio_config_changed);
+
 static int virtio_init(void)
 {
 	if (bus_register(&virtio_bus) != 0)
diff --git a/drivers/virtio/virtio_mmio.c b/drivers/virtio/virtio_mmio.c
index c600ccf..ef9a165 100644
--- a/drivers/virtio/virtio_mmio.c
+++ b/drivers/virtio/virtio_mmio.c
@@ -234,8 +234,6 @@ static irqreturn_t vm_interrupt(int irq, void *opaque)
 {
 	struct virtio_mmio_device *vm_dev = opaque;
 	struct virtio_mmio_vq_info *info;
-	struct virtio_driver *vdrv = container_of(vm_dev->vdev.dev.driver,
-			struct virtio_driver, driver);
 	unsigned long status;
 	unsigned long flags;
 	irqreturn_t ret = IRQ_NONE;
@@ -244,9 +242,8 @@ static irqreturn_t vm_interrupt(int irq, void *opaque)
 	status = readl(vm_dev->base + VIRTIO_MMIO_INTERRUPT_STATUS);
 	writel(status, vm_dev->base + VIRTIO_MMIO_INTERRUPT_ACK);
 
-	if (unlikely(status & VIRTIO_MMIO_INT_CONFIG)
-			&& vdrv && vdrv->config_changed) {
-		vdrv->config_changed(&vm_dev->vdev);
+	if (unlikely(status & VIRTIO_MMIO_INT_CONFIG)) {
+		virtio_config_changed(&vm_dev->vdev);
 		ret = IRQ_HANDLED;
 	}
 
diff --git a/drivers/virtio/virtio_pci.c b/drivers/virtio/virtio_pci.c
index 0f2db51..58cbf6e 100644
--- a/drivers/virtio/virtio_pci.c
+++ b/drivers/virtio/virtio_pci.c
@@ -211,12 +211,8 @@ static bool vp_notify(struct virtqueue *vq)
 static irqreturn_t vp_config_changed(int irq, void *opaque)
 {
 	struct virtio_pci_device *vp_dev = opaque;
-	struct virtio_driver *drv;
-	drv = container_of(vp_dev->vdev.dev.driver,
-			   struct virtio_driver, driver);
 
-	if (drv && drv->config_changed)
-		drv->config_changed(&vp_dev->vdev);
+	virtio_config_changed(&vp_dev->vdev);
 	return IRQ_HANDLED;
 }
 
-- 
MST

Michael S. Tsirkin

2014-Oct-06 15:10 UTC

head link

[PATCH v2 03/15] virtio-pci: move freeze/restore to virtio core

This is in preparation to extending config changed event handling
in core.
Wrapping these in an API also seems to make for a cleaner code.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
---
 include/linux/virtio.h      |  6 +++++
 drivers/virtio/virtio.c     | 53 +++++++++++++++++++++++++++++++++++++++++++
 drivers/virtio/virtio_pci.c | 55 ++-------------------------------------------
 3 files changed, 61 insertions(+), 53 deletions(-)

diff --git a/include/linux/virtio.h b/include/linux/virtio.h
index 3c19bd3..8df7ba8 100644
--- a/include/linux/virtio.h
+++ b/include/linux/virtio.h
@@ -78,6 +78,7 @@ bool virtqueue_is_broken(struct virtqueue *vq);
 /**
  * virtio_device - representation of a device using virtio
  * @index: unique position on the virtio bus
+ * @failed: saved value for CONFIG_S_FAILED bit (for restore)
  * @dev: underlying device.
  * @id: the device type identification (used to match it with a driver).
  * @config: the configuration ops for this device.
@@ -88,6 +89,7 @@ bool virtqueue_is_broken(struct virtqueue *vq);
  */
 struct virtio_device {
 	int index;
+	bool failed;
 	struct device dev;
 	struct virtio_device_id id;
 	const struct virtio_config_ops *config;
@@ -109,6 +111,10 @@ void unregister_virtio_device(struct virtio_device *dev);
 void virtio_break_device(struct virtio_device *dev);
 
 void virtio_config_changed(struct virtio_device *dev);
+#ifdef CONFIG_PM_SLEEP
+int virtio_device_freeze(struct virtio_device *dev);
+int virtio_device_restore(struct virtio_device *dev);
+#endif
 
 /**
  * virtio_driver - operations for a virtio I/O driver
diff --git a/drivers/virtio/virtio.c b/drivers/virtio/virtio.c
index 3980687..657f817 100644
--- a/drivers/virtio/virtio.c
+++ b/drivers/virtio/virtio.c
@@ -248,6 +248,59 @@ void virtio_config_changed(struct virtio_device *dev)
 }
 EXPORT_SYMBOL_GPL(virtio_config_changed);
 
+#ifdef CONFIG_PM_SLEEP
+int virtio_device_freeze(struct virtio_device *dev)
+{
+	struct virtio_driver *drv = drv_to_virtio(dev->dev.driver);
+
+	dev->failed = dev->config->get_status(dev) &
VIRTIO_CONFIG_S_FAILED;
+
+	if (drv && drv->freeze)
+		return drv->freeze(dev);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(virtio_device_freeze);
+
+int virtio_device_restore(struct virtio_device *dev)
+{
+	struct virtio_driver *drv = drv_to_virtio(dev->dev.driver);
+
+	/* We always start by resetting the device, in case a previous
+	 * driver messed it up. */
+	dev->config->reset(dev);
+
+	/* Acknowledge that we've seen the device. */
+	add_status(dev, VIRTIO_CONFIG_S_ACKNOWLEDGE);
+
+	/* Maybe driver failed before freeze.
+	 * Restore the failed status, for debugging. */
+	if (dev->failed)
+		add_status(dev, VIRTIO_CONFIG_S_FAILED);
+
+	if (!drv)
+		return 0;
+
+	/* We have a driver! */
+	add_status(dev, VIRTIO_CONFIG_S_DRIVER);
+
+	dev->config->finalize_features(dev);
+
+	if (drv->restore) {
+		int ret = drv->restore(dev);
+		if (ret) {
+			add_status(dev, VIRTIO_CONFIG_S_FAILED);
+			return ret;
+		}
+		/* Finally, tell the device we're all set */
+		add_status(dev, VIRTIO_CONFIG_S_DRIVER_OK);
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(virtio_device_restore);
+#endif
+
 static int virtio_init(void)
 {
 	if (bus_register(&virtio_bus) != 0)
diff --git a/drivers/virtio/virtio_pci.c b/drivers/virtio/virtio_pci.c
index 58cbf6e..d34ebfa 100644
--- a/drivers/virtio/virtio_pci.c
+++ b/drivers/virtio/virtio_pci.c
@@ -57,9 +57,6 @@ struct virtio_pci_device
 	/* Vectors allocated, excluding per-vq vectors if any */
 	unsigned msix_used_vectors;
 
-	/* Status saved during hibernate/restore */
-	u8 saved_status;
-
 	/* Whether we have vector per vq */
 	bool per_vq_vectors;
 };
@@ -764,16 +761,9 @@ static int virtio_pci_freeze(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
 	struct virtio_pci_device *vp_dev = pci_get_drvdata(pci_dev);
-	struct virtio_driver *drv;
 	int ret;
 
-	drv = container_of(vp_dev->vdev.dev.driver,
-			   struct virtio_driver, driver);
-
-	ret = 0;
-	vp_dev->saved_status = vp_get_status(&vp_dev->vdev);
-	if (drv && drv->freeze)
-		ret = drv->freeze(&vp_dev->vdev);
+	ret = virtio_device_freeze(&vp_dev->vdev);
 
 	if (!ret)
 		pci_disable_device(pci_dev);
@@ -784,55 +774,14 @@ static int virtio_pci_restore(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
 	struct virtio_pci_device *vp_dev = pci_get_drvdata(pci_dev);
-	struct virtio_driver *drv;
-	unsigned status = 0;
 	int ret;
 
-	drv = container_of(vp_dev->vdev.dev.driver,
-			   struct virtio_driver, driver);
-
 	ret = pci_enable_device(pci_dev);
 	if (ret)
 		return ret;
 
 	pci_set_master(pci_dev);
-	/* We always start by resetting the device, in case a previous
-	 * driver messed it up. */
-	vp_reset(&vp_dev->vdev);
-
-	/* Acknowledge that we've seen the device. */
-	status |= VIRTIO_CONFIG_S_ACKNOWLEDGE;
-	vp_set_status(&vp_dev->vdev, status);
-
-	/* Maybe driver failed before freeze.
-	 * Restore the failed status, for debugging. */
-	status |= vp_dev->saved_status & VIRTIO_CONFIG_S_FAILED;
-	vp_set_status(&vp_dev->vdev, status);
-
-	if (!drv)
-		return 0;
-
-	/* We have a driver! */
-	status |= VIRTIO_CONFIG_S_DRIVER;
-	vp_set_status(&vp_dev->vdev, status);
-
-	vp_finalize_features(&vp_dev->vdev);
-
-	if (!drv->restore)
-		return 0;
-
-	ret = drv->restore(&vp_dev->vdev);
-	if (ret) {
-		status |= VIRTIO_CONFIG_S_FAILED;
-		vp_set_status(&vp_dev->vdev, status);
-		return ret;
-	}
-
-	/* Finally, tell the device we're all set */
-	status |= VIRTIO_CONFIG_S_DRIVER_OK;
-	vp_set_status(&vp_dev->vdev, status);
-
-	return ret;
+	return virtio_device_restore(&vp_dev->vdev);
 }
 
 static const struct dev_pm_ops virtio_pci_pm_ops = {
-- 
MST

Michael S. Tsirkin

2014-Oct-06 15:10 UTC

head link

[PATCH v2 04/15] virtio: defer config changed notifications

Defer config changed notifications that arrive during
probe/scan/freeze/restore.

This will allow drivers to set DRIVER_OK earlier, without worrying about
racing with config change interrupts.

This change will also benefit old hypervisors (before 2009)
that send interrupts without checking DRIVER_OK: previously,
the callback could race with driver-specific initialization.

This will also help simplify drivers.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
---
 include/linux/virtio.h  |  6 ++++++
 drivers/virtio/virtio.c | 57 +++++++++++++++++++++++++++++++++++++++++--------
 2 files changed, 54 insertions(+), 9 deletions(-)

diff --git a/include/linux/virtio.h b/include/linux/virtio.h
index 8df7ba8..5636b11 100644
--- a/include/linux/virtio.h
+++ b/include/linux/virtio.h
@@ -79,6 +79,9 @@ bool virtqueue_is_broken(struct virtqueue *vq);
  * virtio_device - representation of a device using virtio
  * @index: unique position on the virtio bus
  * @failed: saved value for CONFIG_S_FAILED bit (for restore)
+ * @config_enabled: configuration change reporting enabled
+ * @config_changed: configuration change reported while disabled
+ * @config_lock: protects configuration change reporting
  * @dev: underlying device.
  * @id: the device type identification (used to match it with a driver).
  * @config: the configuration ops for this device.
@@ -90,6 +93,9 @@ bool virtqueue_is_broken(struct virtqueue *vq);
 struct virtio_device {
 	int index;
 	bool failed;
+	bool config_enabled;
+	bool config_changed;
+	spinlock_t config_lock;
 	struct device dev;
 	struct virtio_device_id id;
 	const struct virtio_config_ops *config;
diff --git a/drivers/virtio/virtio.c b/drivers/virtio/virtio.c
index 657f817..6214744 100644
--- a/drivers/virtio/virtio.c
+++ b/drivers/virtio/virtio.c
@@ -117,6 +117,42 @@ void virtio_check_driver_offered_feature(const struct
virtio_device *vdev,
 }
 EXPORT_SYMBOL_GPL(virtio_check_driver_offered_feature);
 
+static void __virtio_config_changed(struct virtio_device *dev)
+{
+	struct virtio_driver *drv = drv_to_virtio(dev->dev.driver);
+
+	if (!dev->config_enabled)
+		dev->config_changed = true;
+	else if (drv && drv->config_changed)
+		drv->config_changed(dev);
+}
+
+void virtio_config_changed(struct virtio_device *dev)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&dev->config_lock, flags);
+	__virtio_config_changed(dev);
+	spin_unlock_irqrestore(&dev->config_lock, flags);
+}
+EXPORT_SYMBOL_GPL(virtio_config_changed);
+
+static void virtio_config_disable(struct virtio_device *dev)
+{
+	spin_lock_irq(&dev->config_lock);
+	dev->config_enabled = false;
+	spin_unlock_irq(&dev->config_lock);
+}
+
+static void virtio_config_enable(struct virtio_device *dev)
+{
+	spin_lock_irq(&dev->config_lock);
+	dev->config_enabled = true;
+	__virtio_config_changed(dev);
+	dev->config_changed = false;
+	spin_unlock_irq(&dev->config_lock);
+}
+
 static int virtio_dev_probe(struct device *_d)
 {
 	int err, i;
@@ -153,6 +189,8 @@ static int virtio_dev_probe(struct device *_d)
 		add_status(dev, VIRTIO_CONFIG_S_DRIVER_OK);
 		if (drv->scan)
 			drv->scan(dev);
+
+		virtio_config_enable(dev);
 	}
 
 	return err;
@@ -163,6 +201,8 @@ static int virtio_dev_remove(struct device *_d)
 	struct virtio_device *dev = dev_to_virtio(_d);
 	struct virtio_driver *drv = drv_to_virtio(dev->dev.driver);
 
+	virtio_config_disable(dev);
+
 	drv->remove(dev);
 
 	/* Driver should have reset device. */
@@ -211,6 +251,10 @@ int register_virtio_device(struct virtio_device *dev)
 	dev->index = err;
 	dev_set_name(&dev->dev, "virtio%u", dev->index);
 
+	spin_lock_init(&dev->config_lock);
+	dev->config_enabled = false;
+	dev->config_changed = false;
+
 	/* We always start by resetting the device, in case a previous
 	 * driver messed it up.  This also tests that code path a little. */
 	dev->config->reset(dev);
@@ -239,20 +283,13 @@ void unregister_virtio_device(struct virtio_device *dev)
 }
 EXPORT_SYMBOL_GPL(unregister_virtio_device);
 
-void virtio_config_changed(struct virtio_device *dev)
-{
-	struct virtio_driver *drv = drv_to_virtio(dev->dev.driver);
-
-	if (drv && drv->config_changed)
-		drv->config_changed(dev);
-}
-EXPORT_SYMBOL_GPL(virtio_config_changed);
-
 #ifdef CONFIG_PM_SLEEP
 int virtio_device_freeze(struct virtio_device *dev)
 {
 	struct virtio_driver *drv = drv_to_virtio(dev->dev.driver);
 
+	virtio_config_disable(dev);
+
 	dev->failed = dev->config->get_status(dev) &
VIRTIO_CONFIG_S_FAILED;
 
 	if (drv && drv->freeze)
@@ -296,6 +333,8 @@ int virtio_device_restore(struct virtio_device *dev)
 		add_status(dev, VIRTIO_CONFIG_S_DRIVER_OK);
 	}
 
+	virtio_config_enable(dev);
+
 	return 0;
 }
 EXPORT_SYMBOL_GPL(virtio_device_restore);
-- 
MST

Michael S. Tsirkin

2014-Oct-06 15:10 UTC

head link

[PATCH v2 05/15] virtio_blk: drop config_enable

Now that virtio core ensures config changes don't
arrive during probing, drop config_enable flag
in virtio blk.
On removal, flush is now sufficient to guarantee that
no change work is queued.

This help simplify the driver, and will allow
setting DRIVER_OK earlier without losing config
change notifications.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
---
 drivers/block/virtio_blk.c | 23 ++++-------------------
 1 file changed, 4 insertions(+), 19 deletions(-)

diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
index 0a58140..91272f1a 100644
--- a/drivers/block/virtio_blk.c
+++ b/drivers/block/virtio_blk.c
@@ -44,9 +44,6 @@ struct virtio_blk
 	/* Lock for config space updates */
 	struct mutex config_lock;
 
-	/* enable config space updates */
-	bool config_enable;
-
 	/* What host tells us, plus 2 for header & tailer. */
 	unsigned int sg_elems;
 
@@ -348,8 +345,6 @@ static void virtblk_config_changed_work(struct work_struct
*work)
 	u64 capacity, size;
 
 	mutex_lock(&vblk->config_lock);
-	if (!vblk->config_enable)
-		goto done;
 
 	/* Host must always specify the capacity. */
 	virtio_cread(vdev, struct virtio_blk_config, capacity, &capacity);
@@ -374,7 +369,7 @@ static void virtblk_config_changed_work(struct work_struct
*work)
 	set_capacity(vblk->disk, capacity);
 	revalidate_disk(vblk->disk);
 	kobject_uevent_env(&disk_to_dev(vblk->disk)->kobj, KOBJ_CHANGE,
envp);
-done:
+
 	mutex_unlock(&vblk->config_lock);
 }
 
@@ -609,7 +604,6 @@ static int virtblk_probe(struct virtio_device *vdev)
 	mutex_init(&vblk->config_lock);
 
 	INIT_WORK(&vblk->config_work, virtblk_config_changed_work);
-	vblk->config_enable = true;
 
 	err = init_vq(vblk);
 	if (err)
@@ -771,10 +765,8 @@ static void virtblk_remove(struct virtio_device *vdev)
 	int index = vblk->index;
 	int refc;
 
-	/* Prevent config work handler from accessing the device. */
-	mutex_lock(&vblk->config_lock);
-	vblk->config_enable = false;
-	mutex_unlock(&vblk->config_lock);
+	/* Make sure no work handler is accessing the device. */
+	flush_work(&vblk->config_work);
 
 	del_gendisk(vblk->disk);
 	blk_cleanup_queue(vblk->disk->queue);
@@ -784,8 +776,6 @@ static void virtblk_remove(struct virtio_device *vdev)
 	/* Stop all the virtqueues. */
 	vdev->config->reset(vdev);
 
-	flush_work(&vblk->config_work);
-
 	refc = atomic_read(&disk_to_dev(vblk->disk)->kobj.kref.refcount);
 	put_disk(vblk->disk);
 	vdev->config->del_vqs(vdev);
@@ -805,11 +795,7 @@ static int virtblk_freeze(struct virtio_device *vdev)
 	/* Ensure we don't receive any more interrupts */
 	vdev->config->reset(vdev);
 
-	/* Prevent config work handler from accessing the device. */
-	mutex_lock(&vblk->config_lock);
-	vblk->config_enable = false;
-	mutex_unlock(&vblk->config_lock);
-
+	/* Make sure no work handler is accessing the device. */
 	flush_work(&vblk->config_work);
 
 	blk_mq_stop_hw_queues(vblk->disk->queue);
@@ -823,7 +809,6 @@ static int virtblk_restore(struct virtio_device *vdev)
 	struct virtio_blk *vblk = vdev->priv;
 	int ret;
 
-	vblk->config_enable = true;
 	ret = init_vq(vdev->priv);
 	if (!ret)
 		blk_mq_start_stopped_hw_queues(vblk->disk->queue, true);
-- 
MST

Michael S. Tsirkin

2014-Oct-06 15:11 UTC

head link

[PATCH v2 06/15] virtio-blk: drop config_mutex

config_mutex served two purposes: prevent multiple concurrent config
change handlers, and synchronize access to config_enable flag.

Since commit dbf2576e37da0fcc7aacbfbb9fd5d3de7888a3c1
    workqueue: make all workqueues non-reentrant
all workqueues are non-reentrant, and config_enable
is now gone.

Get rid of the unnecessary lock.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck at de.ibm.com>
---
 drivers/block/virtio_blk.c | 8 --------
 1 file changed, 8 deletions(-)

diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
index 91272f1a..89ba8d6 100644
--- a/drivers/block/virtio_blk.c
+++ b/drivers/block/virtio_blk.c
@@ -41,9 +41,6 @@ struct virtio_blk
 	/* Process context for config space updates */
 	struct work_struct config_work;
 
-	/* Lock for config space updates */
-	struct mutex config_lock;
-
 	/* What host tells us, plus 2 for header & tailer. */
 	unsigned int sg_elems;
 
@@ -344,8 +341,6 @@ static void virtblk_config_changed_work(struct work_struct
*work)
 	char *envp[] = { "RESIZE=1", NULL };
 	u64 capacity, size;
 
-	mutex_lock(&vblk->config_lock);
-
 	/* Host must always specify the capacity. */
 	virtio_cread(vdev, struct virtio_blk_config, capacity, &capacity);
 
@@ -369,8 +364,6 @@ static void virtblk_config_changed_work(struct work_struct
*work)
 	set_capacity(vblk->disk, capacity);
 	revalidate_disk(vblk->disk);
 	kobject_uevent_env(&disk_to_dev(vblk->disk)->kobj, KOBJ_CHANGE,
envp);
-
-	mutex_unlock(&vblk->config_lock);
 }
 
 static void virtblk_config_changed(struct virtio_device *vdev)
@@ -601,7 +594,6 @@ static int virtblk_probe(struct virtio_device *vdev)
 
 	vblk->vdev = vdev;
 	vblk->sg_elems = sg_elems;
-	mutex_init(&vblk->config_lock);
 
 	INIT_WORK(&vblk->config_work, virtblk_config_changed_work);
 
-- 
MST

Michael S. Tsirkin

2014-Oct-06 15:11 UTC

head link

[PATCH v2 07/15] virtio_net: drop config_enable

Now that virtio core ensures config changes don't arrive during probing,
drop config_enable flag in virtio net.
On removal, flush is now sufficient to guarantee that no change work is
queued.

This help simplify the driver, and will allow setting DRIVER_OK earlier
without losing config change notifications.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
---
 drivers/net/virtio_net.c | 27 ++++-----------------------
 1 file changed, 4 insertions(+), 23 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 59caa06..743fb04 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -123,9 +123,6 @@ struct virtnet_info {
 	/* Host can handle any s/g split between our header and packet data */
 	bool any_header_sg;
 
-	/* enable config space updates */
-	bool config_enable;
-
 	/* Active statistics */
 	struct virtnet_stats __percpu *stats;
 
@@ -1408,9 +1405,6 @@ static void virtnet_config_changed_work(struct work_struct
*work)
 	u16 v;
 
 	mutex_lock(&vi->config_lock);
-	if (!vi->config_enable)
-		goto done;
-
 	if (virtio_cread_feature(vi->vdev, VIRTIO_NET_F_STATUS,
 				 struct virtio_net_config, status, &v) < 0)
 		goto done;
@@ -1758,7 +1752,6 @@ static int virtnet_probe(struct virtio_device *vdev)
 	}
 
 	mutex_init(&vi->config_lock);
-	vi->config_enable = true;
 	INIT_WORK(&vi->config_work, virtnet_config_changed_work);
 
 	/* If we can receive ANY GSO packets, we must allocate large ones. */
@@ -1875,17 +1868,13 @@ static void virtnet_remove(struct virtio_device *vdev)
 
 	unregister_hotcpu_notifier(&vi->nb);
 
-	/* Prevent config work handler from accessing the device. */
-	mutex_lock(&vi->config_lock);
-	vi->config_enable = false;
-	mutex_unlock(&vi->config_lock);
+	/* Make sure no work handler is accessing the device. */
+	flush_work(&vi->config_work);
 
 	unregister_netdev(vi->dev);
 
 	remove_vq_common(vi);
 
-	flush_work(&vi->config_work);
-
 	free_percpu(vi->stats);
 	free_netdev(vi->dev);
 }
@@ -1898,10 +1887,8 @@ static int virtnet_freeze(struct virtio_device *vdev)
 
 	unregister_hotcpu_notifier(&vi->nb);
 
-	/* Prevent config work handler from accessing the device */
-	mutex_lock(&vi->config_lock);
-	vi->config_enable = false;
-	mutex_unlock(&vi->config_lock);
+	/* Make sure no work handler is accessing the device */
+	flush_work(&vi->config_work);
 
 	netif_device_detach(vi->dev);
 	cancel_delayed_work_sync(&vi->refill);
@@ -1916,8 +1903,6 @@ static int virtnet_freeze(struct virtio_device *vdev)
 
 	remove_vq_common(vi);
 
-	flush_work(&vi->config_work);
-
 	return 0;
 }
 
@@ -1941,10 +1926,6 @@ static int virtnet_restore(struct virtio_device *vdev)
 
 	netif_device_attach(vi->dev);
 
-	mutex_lock(&vi->config_lock);
-	vi->config_enable = true;
-	mutex_unlock(&vi->config_lock);
-
 	rtnl_lock();
 	virtnet_set_queues(vi, vi->curr_queue_pairs);
 	rtnl_unlock();
-- 
MST

Michael S. Tsirkin

2014-Oct-06 15:11 UTC

head link

[PATCH v2 08/15] virtio-net: drop config_mutex

config_mutex served two purposes: prevent multiple concurrent config
change handlers, and synchronize access to config_enable flag.

Since commit dbf2576e37da0fcc7aacbfbb9fd5d3de7888a3c1
    workqueue: make all workqueues non-reentrant
all workqueues are non-reentrant, and config_enable
is now gone.

Get rid of the unnecessary lock.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck at de.ibm.com>
---
 drivers/net/virtio_net.c | 7 +------
 1 file changed, 1 insertion(+), 6 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 743fb04..23e4a69 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -132,9 +132,6 @@ struct virtnet_info {
 	/* Work struct for config space updates */
 	struct work_struct config_work;
 
-	/* Lock for config space updates */
-	struct mutex config_lock;
-
 	/* Does the affinity hint is set for virtqueues? */
 	bool affinity_hint_set;
 
@@ -1404,7 +1401,6 @@ static void virtnet_config_changed_work(struct work_struct
*work)
 		container_of(work, struct virtnet_info, config_work);
 	u16 v;
 
-	mutex_lock(&vi->config_lock);
 	if (virtio_cread_feature(vi->vdev, VIRTIO_NET_F_STATUS,
 				 struct virtio_net_config, status, &v) < 0)
 		goto done;
@@ -1430,7 +1426,7 @@ static void virtnet_config_changed_work(struct work_struct
*work)
 		netif_tx_stop_all_queues(vi->dev);
 	}
 done:
-	mutex_unlock(&vi->config_lock);
+	return;
 }
 
 static void virtnet_config_changed(struct virtio_device *vdev)
@@ -1751,7 +1747,6 @@ static int virtnet_probe(struct virtio_device *vdev)
 		u64_stats_init(&virtnet_stats->rx_syncp);
 	}
 
-	mutex_init(&vi->config_lock);
 	INIT_WORK(&vi->config_work, virtnet_config_changed_work);
 
 	/* If we can receive ANY GSO packets, we must allocate large ones. */
-- 
MST

Michael S. Tsirkin

2014-Oct-06 15:11 UTC

head link

[PATCH v2 09/15] virtio_net: minor cleanup

goto done;
done:
	return;
is ugly, it was put there to make diff review easier.
replace by open-coded return.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
Acked-by: Cornelia Huck <cornelia.huck at de.ibm.com>
---
 drivers/net/virtio_net.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 23e4a69..ef04d23 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -1403,7 +1403,7 @@ static void virtnet_config_changed_work(struct work_struct
*work)
 
 	if (virtio_cread_feature(vi->vdev, VIRTIO_NET_F_STATUS,
 				 struct virtio_net_config, status, &v) < 0)
-		goto done;
+		return;
 
 	if (v & VIRTIO_NET_S_ANNOUNCE) {
 		netdev_notify_peers(vi->dev);
@@ -1414,7 +1414,7 @@ static void virtnet_config_changed_work(struct work_struct
*work)
 	v &= VIRTIO_NET_S_LINK_UP;
 
 	if (vi->status == v)
-		goto done;
+		return;
 
 	vi->status = v;
 
@@ -1425,8 +1425,6 @@ static void virtnet_config_changed_work(struct work_struct
*work)
 		netif_carrier_off(vi->dev);
 		netif_tx_stop_all_queues(vi->dev);
 	}
-done:
-	return;
 }
 
 static void virtnet_config_changed(struct virtio_device *vdev)
-- 
MST

Michael S. Tsirkin

2014-Oct-06 15:11 UTC

head link

[PATCH v2 10/15] virtio: add API to enable VQs early

virtio spec 0.9.X requires DRIVER_OK to be set before
VQs are used, but some drivers use VQs before probe
function returns.
Since DRIVER_OK is set after probe, this violates the spec.

Even though under virtio 1.0 transitional devices support this
behaviour, we want to make it possible for those early callers to become
spec compliant and eventually support non-transitional devices.

Add API for drivers to call before using VQs.

Sets DRIVER_OK internally.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
---
 include/linux/virtio_config.h | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/include/linux/virtio_config.h b/include/linux/virtio_config.h
index e8f8f71..e36403b 100644
--- a/include/linux/virtio_config.h
+++ b/include/linux/virtio_config.h
@@ -109,6 +109,23 @@ struct virtqueue *virtio_find_single_vq(struct
virtio_device *vdev,
 	return vq;
 }
 
+/**
+ * virtio_enable_vqs_early - enable vq use in probe function
+ * @vdev: the device
+ *
+ * Driver must call this to use vqs in the probe function.
+ *
+ * Note: vqs are enabled automatically after probe returns.
+ */
+static inline
+void virtio_enable_vqs_early(struct virtio_device *dev)
+{
+	unsigned status = dev->config->get_status(dev);
+
+	BUG_ON(status & VIRTIO_CONFIG_S_DRIVER_OK);
+	dev->config->set_status(dev, status | VIRTIO_CONFIG_S_DRIVER_OK);
+}
+
 static inline
 const char *virtio_bus_name(struct virtio_device *vdev)
 {
-- 
MST

Michael S. Tsirkin

2014-Oct-06 15:11 UTC

head link

[PATCH v2 11/15] virtio_net: enable VQs early

virtio spec requires drivers to set DRIVER_OK before using VQs.
This is set automatically after probe returns, virtio net violated this
rule by using receive VQs within probe.

To fix, call virtio_enable_vqs_early before using VQs.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
---
 drivers/net/virtio_net.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index ef04d23..430f3ae 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -1792,6 +1792,8 @@ static int virtnet_probe(struct virtio_device *vdev)
 		goto free_vqs;
 	}
 
+	virtio_enable_vqs_early(vdev);
+
 	/* Last of all, set up some receive buffers. */
 	for (i = 0; i < vi->curr_queue_pairs; i++) {
 		try_fill_recv(&vi->rq[i], GFP_KERNEL);
-- 
MST

Michael S. Tsirkin

2014-Oct-06 15:11 UTC

head link

[PATCH v2 12/15] virtio_blk: enable VQs early

virtio spec requires drivers to set DRIVER_OK before using VQs.
This is set automatically after probe returns, virtio block violated this
rule by calling add_disk, which causes the VQ to be used directly within
probe.

To fix, call virtio_enable_vqs_early before using VQs.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
---
 drivers/block/virtio_blk.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
index 89ba8d6..46b04bf 100644
--- a/drivers/block/virtio_blk.c
+++ b/drivers/block/virtio_blk.c
@@ -719,6 +719,8 @@ static int virtblk_probe(struct virtio_device *vdev)
 	if (!err && opt_io_size)
 		blk_queue_io_opt(q, blk_size * opt_io_size);
 
+	virtio_enable_vqs_early(vdev);
+
 	add_disk(vblk->disk);
 	err = device_create_file(disk_to_dev(vblk->disk), &dev_attr_serial);
 	if (err)
-- 
MST

Michael S. Tsirkin

2014-Oct-06 15:11 UTC

head link

[PATCH v2 13/15] virtio_console: enable VQs early

virtio spec requires drivers to set DRIVER_OK before using VQs.
This is set automatically after probe returns, virtio console violated this
rule by adding inbufs, which causes the VQ to be used directly within
probe.

To fix, call virtio_enable_vqs_early before using VQs.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
---
 drivers/char/virtio_console.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/char/virtio_console.c b/drivers/char/virtio_console.c
index b585b47..6ebe8f6 100644
--- a/drivers/char/virtio_console.c
+++ b/drivers/char/virtio_console.c
@@ -1449,6 +1449,8 @@ static int add_port(struct ports_device *portdev, u32 id)
 	spin_lock_init(&port->outvq_lock);
 	init_waitqueue_head(&port->waitqueue);
 
+	virtio_enable_vqs_early(portdev->vdev);
+
 	/* Fill the in_vq with buffers so the host can send us data. */
 	nr_added_bufs = fill_queue(port->in_vq, &port->inbuf_lock);
 	if (!nr_added_bufs) {
-- 
MST

Michael S. Tsirkin

2014-Oct-06 15:11 UTC

head link

[PATCH v2 14/15] 9p/trans_virtio: enable VQs early

virtio spec requires drivers to set DRIVER_OK before using VQs.
This is set automatically after probe returns, but virtio 9p device
adds self to channel list within probe, at which point VQ can be
used in violation of the spec.

To fix, call virtio_enable_vqs_early before using VQs.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
---
 net/9p/trans_virtio.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/net/9p/trans_virtio.c b/net/9p/trans_virtio.c
index 6940d8f..766ba48 100644
--- a/net/9p/trans_virtio.c
+++ b/net/9p/trans_virtio.c
@@ -575,6 +575,8 @@ static int p9_virtio_probe(struct virtio_device *vdev)
 	/* Ceiling limit to avoid denial of service attacks */
 	chan->p9_max_pages = nr_free_buffer_pages()/4;
 
+	virtio_enable_vqs_early(vdev);
+
 	mutex_lock(&virtio_9p_lock);
 	list_add_tail(&chan->chan_list, &virtio_chan_list);
 	mutex_unlock(&virtio_9p_lock);
-- 
MST

Michael S. Tsirkin

2014-Oct-06 15:11 UTC

head link

[PATCH v2 15/15] virtio_net: fix use after free on allocation failure

In the extremely unlikely event that driver initialization fails after
RX buffers are added, virtio net frees RX buffers while VQs are
still active, potentially causing device to use a freed buffer.

To fix, reset device first - same as we do on device removal.

Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck at de.ibm.com>
---
 drivers/net/virtio_net.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 430f3ae..3551417 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -1830,6 +1830,8 @@ static int virtnet_probe(struct virtio_device *vdev)
 	return 0;
 
 free_recv_bufs:
+	vi->vdev->config->reset(vdev);
+
 	free_receive_bufs(vi);
 	unregister_netdev(dev);
 free_vqs:
-- 
MST

Cornelia Huck

2014-Oct-06 15:16 UTC

head link

[PATCH v2 02/15] virtio: unify config_changed handling

On Mon, 6 Oct 2014 18:10:44 +0300
"Michael S. Tsirkin" <mst at redhat.com> wrote:
> Replace duplicated code in all transports with a single wrapper in
> virtio.c.
> 
> The only functional change is in virtio_mmio.c: if a buggy device sends
> us an interrupt before driver is set, we previously returned IRQ_NONE,
> now we return IRQ_HANDLED.
> 
> As this must not happen in practice, this does not look like a big deal.
> 
> See also commit 3fff0179e33cd7d0a688dab65700c46ad089e934
> 	virtio-pci: do not oops on config change if driver not loaded.
> for the original motivation behind the driver check.
> 
> Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
> ---
>  include/linux/virtio.h             | 2 ++
>  drivers/misc/mic/card/mic_virtio.c | 6 +-----
>  drivers/s390/kvm/kvm_virtio.c      | 9 +--------
>  drivers/s390/kvm/virtio_ccw.c      | 6 +-----
>  drivers/virtio/virtio.c            | 9 +++++++++
>  drivers/virtio/virtio_mmio.c       | 7 ++-----
>  drivers/virtio/virtio_pci.c        | 6 +-----
>  7 files changed, 17 insertions(+), 28 deletions(-)
Reviewed-by: Cornelia Huck <cornelia.huck at de.ibm.com>

Cornelia Huck

2014-Oct-06 15:20 UTC

head link

[PATCH v2 03/15] virtio-pci: move freeze/restore to virtio core

On Mon, 6 Oct 2014 18:10:51 +0300
"Michael S. Tsirkin" <mst at redhat.com> wrote:
> This is in preparation to extending config changed event handling
> in core.
> Wrapping these in an API also seems to make for a cleaner code.
> 
> Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
> ---
>  include/linux/virtio.h      |  6 +++++
>  drivers/virtio/virtio.c     | 53
+++++++++++++++++++++++++++++++++++++++++++
>  drivers/virtio/virtio_pci.c | 55
++-------------------------------------------
>  3 files changed, 61 insertions(+), 53 deletions(-)
> 
> diff --git a/include/linux/virtio.h b/include/linux/virtio.h
> index 3c19bd3..8df7ba8 100644
> --- a/include/linux/virtio.h
> +++ b/include/linux/virtio.h
> @@ -78,6 +78,7 @@ bool virtqueue_is_broken(struct virtqueue *vq);
>  /**
>   * virtio_device - representation of a device using virtio
>   * @index: unique position on the virtio bus
> + * @failed: saved value for CONFIG_S_FAILED bit (for restore)
Have you considered s/failed/saved_failed/ ?
>   * @dev: underlying device.
>   * @id: the device type identification (used to match it with a driver).
>   * @config: the configuration ops for this device.
Otherwise, my

Reviewed-by: Cornelia Huck <cornelia.huck at de.ibm.com>

still stands.

Cornelia Huck

2014-Oct-06 15:25 UTC

head link

[PATCH v2 04/15] virtio: defer config changed notifications

On Mon, 6 Oct 2014 18:10:55 +0300
"Michael S. Tsirkin" <mst at redhat.com> wrote:
> Defer config changed notifications that arrive during
> probe/scan/freeze/restore.
> 
> This will allow drivers to set DRIVER_OK earlier, without worrying about
> racing with config change interrupts.
> 
> This change will also benefit old hypervisors (before 2009)
> that send interrupts without checking DRIVER_OK: previously,
> the callback could race with driver-specific initialization.
> 
> This will also help simplify drivers.
> 
> Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
> ---
>  include/linux/virtio.h  |  6 ++++++
>  drivers/virtio/virtio.c | 57
+++++++++++++++++++++++++++++++++++++++++--------
>  2 files changed, 54 insertions(+), 9 deletions(-)
Reviewed-by: Cornelia Huck <cornelia.huck at de.ibm.com>

Cornelia Huck

2014-Oct-06 15:26 UTC

head link

[PATCH v2 05/15] virtio_blk: drop config_enable

On Mon, 6 Oct 2014 18:10:58 +0300
"Michael S. Tsirkin" <mst at redhat.com> wrote:
> Now that virtio core ensures config changes don't
> arrive during probing, drop config_enable flag
> in virtio blk.
> On removal, flush is now sufficient to guarantee that
> no change work is queued.
> 
> This help simplify the driver, and will allow
> setting DRIVER_OK earlier without losing config
> change notifications.
> 
> Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
> ---
>  drivers/block/virtio_blk.c | 23 ++++-------------------
>  1 file changed, 4 insertions(+), 19 deletions(-)
Reviewed-by: Cornelia Huck <cornelia.huck at de.ibm.com>

Cornelia Huck

2014-Oct-06 15:28 UTC

head link

[PATCH v2 07/15] virtio_net: drop config_enable

On Mon, 6 Oct 2014 18:11:05 +0300
"Michael S. Tsirkin" <mst at redhat.com> wrote:
> Now that virtio core ensures config changes don't arrive during
probing,
> drop config_enable flag in virtio net.
> On removal, flush is now sufficient to guarantee that no change work is
> queued.
> 
> This help simplify the driver, and will allow setting DRIVER_OK earlier
> without losing config change notifications.
> 
> Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
> ---
>  drivers/net/virtio_net.c | 27 ++++-----------------------
>  1 file changed, 4 insertions(+), 23 deletions(-)
Reviewed-by: Cornelia Huck <cornelia.huck at de.ibm.com>

Cornelia Huck

2014-Oct-06 15:29 UTC

head link

[PATCH v2 10/15] virtio: add API to enable VQs early

On Mon, 6 Oct 2014 18:11:16 +0300
"Michael S. Tsirkin" <mst at redhat.com> wrote:
> virtio spec 0.9.X requires DRIVER_OK to be set before
> VQs are used, but some drivers use VQs before probe
> function returns.
> Since DRIVER_OK is set after probe, this violates the spec.
> 
> Even though under virtio 1.0 transitional devices support this
> behaviour, we want to make it possible for those early callers to become
> spec compliant and eventually support non-transitional devices.
> 
> Add API for drivers to call before using VQs.
> 
> Sets DRIVER_OK internally.
> 
> Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
> ---
>  include/linux/virtio_config.h | 17 +++++++++++++++++
>  1 file changed, 17 insertions(+)
Reviewed-by: Cornelia Huck <cornelia.huck at de.ibm.com>

Cornelia Huck

2014-Oct-06 15:39 UTC

head link

[PATCH v2 11/15] virtio_net: enable VQs early

On Mon, 6 Oct 2014 18:11:19 +0300
"Michael S. Tsirkin" <mst at redhat.com> wrote:
> virtio spec requires drivers to set DRIVER_OK before using VQs.
> This is set automatically after probe returns, virtio net violated this
> rule by using receive VQs within probe.
> 
> To fix, call virtio_enable_vqs_early before using VQs.
> 
> Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
> ---
>  drivers/net/virtio_net.c | 2 ++
>  1 file changed, 2 insertions(+)
Reviewed-by: Cornelia Huck <cornelia.huck at de.ibm.com>

Cornelia Huck

2014-Oct-06 15:42 UTC

head link

[PATCH v2 12/15] virtio_blk: enable VQs early

On Mon, 6 Oct 2014 18:11:22 +0300
"Michael S. Tsirkin" <mst at redhat.com> wrote:
> virtio spec requires drivers to set DRIVER_OK before using VQs.
> This is set automatically after probe returns, virtio block violated this
> rule by calling add_disk, which causes the VQ to be used directly within
> probe.
> 
> To fix, call virtio_enable_vqs_early before using VQs.
> 
> Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
> ---
>  drivers/block/virtio_blk.c | 2 ++
>  1 file changed, 2 insertions(+)
Reviewed-by: Cornelia Huck <cornelia.huck at de.ibm.com>

Amit Shah

2014-Oct-07 12:05 UTC

head link

[PATCH v2 01/15] virtio_pci: fix virtio spec compliance on restore

On (Mon) 06 Oct 2014 [18:10:40], Michael S. Tsirkin
wrote:> On restore, virtio pci does the following:
> + set features
> + init vqs etc - device can be used at this point!
> + set ACKNOWLEDGE,DRIVER and DRIVER_OK status bits
> 
> This is in violation of the virtio spec, which
> requires the following order:
> - ACKNOWLEDGE
> - DRIVER
> - init vqs
> - DRIVER_OK
> 
> This behaviour will break with hypervisors that assume spec compliant
> behaviour.  It seems like a good idea to have this patch applied to
> stable branches to reduce the support butden for the hypervisors.
> 
> Cc: stable at vger.kernel.org
> Cc: Amit Shah <amit.shah at redhat.com>
> Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
I didn't see my previous questions answered from the initial posting
-- can you please respond to them?

		Amit

Michael S. Tsirkin

2014-Oct-07 12:53 UTC

head link

[PATCH v2 01/15] virtio_pci: fix virtio spec compliance on restore

On Mon, Oct 06, 2014 at 06:10:40PM +0300, Michael S. Tsirkin
wrote:> On restore, virtio pci does the following:
> + set features
> + init vqs etc - device can be used at this point!
> + set ACKNOWLEDGE,DRIVER and DRIVER_OK status bits
> 
> This is in violation of the virtio spec, which
> requires the following order:
> - ACKNOWLEDGE
> - DRIVER
> - init vqs
> - DRIVER_OK
> 
> This behaviour will break with hypervisors that assume spec compliant
> behaviour.  It seems like a good idea to have this patch applied to
> stable branches to reduce the support butden for the hypervisors.
Tested suspend to ram with virtio net and blk.
> 
> Cc: stable at vger.kernel.org
> Cc: Amit Shah <amit.shah at redhat.com>
> Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
> ---
>  drivers/virtio/virtio_pci.c | 36 ++++++++++++++++++++++++++++++++----
>  1 file changed, 32 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/virtio/virtio_pci.c b/drivers/virtio/virtio_pci.c
> index 3d1463c..0f2db51 100644
> --- a/drivers/virtio/virtio_pci.c
> +++ b/drivers/virtio/virtio_pci.c
> @@ -789,6 +789,7 @@ static int virtio_pci_restore(struct device *dev)
>  	struct pci_dev *pci_dev = to_pci_dev(dev);
>  	struct virtio_pci_device *vp_dev = pci_get_drvdata(pci_dev);
>  	struct virtio_driver *drv;
> +	unsigned status = 0;
>  	int ret;
>  
>  	drv = container_of(vp_dev->vdev.dev.driver,
> @@ -799,14 +800,41 @@ static int virtio_pci_restore(struct device *dev)
>  		return ret;
>  
>  	pci_set_master(pci_dev);
> +	/* We always start by resetting the device, in case a previous
> +	 * driver messed it up. */
> +	vp_reset(&vp_dev->vdev);
> +
> +	/* Acknowledge that we've seen the device. */
> +	status |= VIRTIO_CONFIG_S_ACKNOWLEDGE;
> +	vp_set_status(&vp_dev->vdev, status);
> +
> +	/* Maybe driver failed before freeze.
> +	 * Restore the failed status, for debugging. */
> +	status |= vp_dev->saved_status & VIRTIO_CONFIG_S_FAILED;
> +	vp_set_status(&vp_dev->vdev, status);
> +
> +	if (!drv)
> +		return 0;
> +
> +	/* We have a driver! */
> +	status |= VIRTIO_CONFIG_S_DRIVER;
> +	vp_set_status(&vp_dev->vdev, status);
> +
>  	vp_finalize_features(&vp_dev->vdev);
>  
> -	if (drv && drv->restore)
> -		ret = drv->restore(&vp_dev->vdev);
> +	if (!drv->restore)
> +		return 0;
As Amit points out, this is a bug: for a driver without restore
callback we should still set DRIVER_OK.
Will post v3.
> +
> +	ret = drv->restore(&vp_dev->vdev);
> +	if (ret) {
> +		status |= VIRTIO_CONFIG_S_FAILED;
> +		vp_set_status(&vp_dev->vdev, status);
> +		return ret;
> +	}
>  
>  	/* Finally, tell the device we're all set */
> -	if (!ret)
> -		vp_set_status(&vp_dev->vdev, vp_dev->saved_status);
> +	status |= VIRTIO_CONFIG_S_DRIVER_OK;
> +	vp_set_status(&vp_dev->vdev, status);
>  
>  	return ret;
>  }
> -- 
> MST
>

Possibly Parallel Threads

Search for more apparently analagous threads

Linux Virtualization - Oct 2014 - [PATCH v2 10/15] virtio: add API to enable VQs early

[PATCH v2 00/15] virtio: fix spec compliance issues

[PATCH v2 01/15] virtio_pci: fix virtio spec compliance on restore

[PATCH v2 02/15] virtio: unify config_changed handling

[PATCH v2 03/15] virtio-pci: move freeze/restore to virtio core

[PATCH v2 04/15] virtio: defer config changed notifications

[PATCH v2 05/15] virtio_blk: drop config_enable

[PATCH v2 06/15] virtio-blk: drop config_mutex

[PATCH v2 07/15] virtio_net: drop config_enable

[PATCH v2 08/15] virtio-net: drop config_mutex

[PATCH v2 09/15] virtio_net: minor cleanup

[PATCH v2 10/15] virtio: add API to enable VQs early

[PATCH v2 11/15] virtio_net: enable VQs early

[PATCH v2 12/15] virtio_blk: enable VQs early

[PATCH v2 13/15] virtio_console: enable VQs early

[PATCH v2 14/15] 9p/trans_virtio: enable VQs early

[PATCH v2 15/15] virtio_net: fix use after free on allocation failure

[PATCH v2 02/15] virtio: unify config_changed handling

[PATCH v2 03/15] virtio-pci: move freeze/restore to virtio core

[PATCH v2 04/15] virtio: defer config changed notifications

[PATCH v2 05/15] virtio_blk: drop config_enable

[PATCH v2 07/15] virtio_net: drop config_enable

[PATCH v2 10/15] virtio: add API to enable VQs early

[PATCH v2 11/15] virtio_net: enable VQs early

[PATCH v2 12/15] virtio_blk: enable VQs early

[PATCH v2 01/15] virtio_pci: fix virtio spec compliance on restore

[PATCH v2 01/15] virtio_pci: fix virtio spec compliance on restore

Possibly Parallel Threads