[PATCH v2] vp_vdpa: harden the logic of set status

[PATCH v2] vp_vdpa: harden the logic of set status

[Longpeng(Mike)]

From: Longpeng <longpeng2@huawei.com>

1. We should not set status to 0 when invoking vp_vdpa_set_status(),
   trigger a warning in that case.

2. The driver MUST wait for a read of device_status to return 0 before
   reinitializing the device. But we also don't want to keep us in an
   infinite loop forever, so wait for 5s if we try to reset the device.

Signed-off-by: Longpeng <longpeng2@huawei.com>
---
Changes v1->v2:
  - use WARN_ON instead of BUG_ON. [Stefano]
  - use "warning + failed" instead of "infinite loop". [Jason, Stefano]
  - use usleep_range instead of msleep (checkpatch). [Longpeng]

---
 drivers/vdpa/virtio_pci/vp_vdpa.c | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)

diff --git a/drivers/vdpa/virtio_pci/vp_vdpa.c b/drivers/vdpa/virtio_pci/vp_vdpa.c
index 13701c2a1963..a2d3b13ac646 100644
--- a/drivers/vdpa/virtio_pci/vp_vdpa.c
+++ b/drivers/vdpa/virtio_pci/vp_vdpa.c
@@ -214,6 +214,9 @@ static void vp_vdpa_set_status(struct vdpa_device *vdpa, u8 status)
 	struct virtio_pci_modern_device *mdev = vp_vdpa_to_mdev(vp_vdpa);
 	u8 s = vp_vdpa_get_status(vdpa);
 
+	/* We should never be setting status to 0. */
+	WARN_ON(status == 0);
+
 	if (status & VIRTIO_CONFIG_S_DRIVER_OK &&
 	    !(s & VIRTIO_CONFIG_S_DRIVER_OK)) {
 		vp_vdpa_request_irq(vp_vdpa);
@@ -222,14 +225,33 @@ static void vp_vdpa_set_status(struct vdpa_device *vdpa, u8 status)
 	vp_modern_set_status(mdev, status);
 }
 
+#define VP_VDPA_RESET_TMOUT_MS 5000 /* 5s */
+
 static int vp_vdpa_reset(struct vdpa_device *vdpa, bool clear)
 {
 	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
 	struct virtio_pci_modern_device *mdev = vp_vdpa_to_mdev(vp_vdpa);
 	u8 s = vp_vdpa_get_status(vdpa);
+	unsigned long timeout;
 
 	vp_modern_set_status(mdev, 0);
 
+	/*
+	 * As the virtio v1.1 spec (4.1.4.3.2) says: After writing 0 to
+	 * device_status, the driver MUST wait for a read of device_status
+	 * to return 0 before reinitializing the device.
+	 * To avoid keep us here forever, we only wait for 5 seconds.
+	 */
+	timeout = jiffies + msecs_to_jiffies(VP_VDPA_RESET_TMOUT_MS);
+	while (vp_modern_get_status(mdev)) {
+		usleep_range(1000, 1500);
+		if (time_after(jiffies, timeout)) {
+			dev_err(&mdev->pci_dev->dev,
+				"vp_vdpa: fail to reset device\n");
+			return -ETIMEDOUT;
+		}
+	}
+
 	if (s & VIRTIO_CONFIG_S_DRIVER_OK)
 		vp_vdpa_free_irq(vp_vdpa);
 
-- 
2.23.0

Re: [PATCH v2] vp_vdpa: harden the logic of set status

[Jason Wang]

On Tue, Dec 6, 2022 at 10:13 AM Longpeng(Mike) <longpeng2@huawei.com> wrote:
>
> From: Longpeng <longpeng2@huawei.com>
>
> 1. We should not set status to 0 when invoking vp_vdpa_set_status(),
>    trigger a warning in that case.
>
> 2. The driver MUST wait for a read of device_status to return 0 before
>    reinitializing the device. But we also don't want to keep us in an
>    infinite loop forever, so wait for 5s if we try to reset the device.
>
> Signed-off-by: Longpeng <longpeng2@huawei.com>
> ---
> Changes v1->v2:
>   - use WARN_ON instead of BUG_ON. [Stefano]
>   - use "warning + failed" instead of "infinite loop". [Jason, Stefano]
>   - use usleep_range instead of msleep (checkpatch). [Longpeng]
>
> ---
>  drivers/vdpa/virtio_pci/vp_vdpa.c | 22 ++++++++++++++++++++++
>  1 file changed, 22 insertions(+)
>
> diff --git a/drivers/vdpa/virtio_pci/vp_vdpa.c b/drivers/vdpa/virtio_pci/vp_vdpa.c
> index 13701c2a1963..a2d3b13ac646 100644
> --- a/drivers/vdpa/virtio_pci/vp_vdpa.c
> +++ b/drivers/vdpa/virtio_pci/vp_vdpa.c
> @@ -214,6 +214,9 @@ static void vp_vdpa_set_status(struct vdpa_device *vdpa, u8 status)
>         struct virtio_pci_modern_device *mdev = vp_vdpa_to_mdev(vp_vdpa);
>         u8 s = vp_vdpa_get_status(vdpa);
>
> +       /* We should never be setting status to 0. */
> +       WARN_ON(status == 0);
> +
>         if (status & VIRTIO_CONFIG_S_DRIVER_OK &&
>             !(s & VIRTIO_CONFIG_S_DRIVER_OK)) {
>                 vp_vdpa_request_irq(vp_vdpa);
> @@ -222,14 +225,33 @@ static void vp_vdpa_set_status(struct vdpa_device *vdpa, u8 status)
>         vp_modern_set_status(mdev, status);
>  }
>
> +#define VP_VDPA_RESET_TMOUT_MS 5000 /* 5s */
> +
>  static int vp_vdpa_reset(struct vdpa_device *vdpa, bool clear)
>  {
>         struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>         struct virtio_pci_modern_device *mdev = vp_vdpa_to_mdev(vp_vdpa);
>         u8 s = vp_vdpa_get_status(vdpa);
> +       unsigned long timeout;
>
>         vp_modern_set_status(mdev, 0);
>
> +       /*
> +        * As the virtio v1.1 spec (4.1.4.3.2) says: After writing 0 to
> +        * device_status, the driver MUST wait for a read of device_status
> +        * to return 0 before reinitializing the device.
> +        * To avoid keep us here forever, we only wait for 5 seconds.

s/keep/keeping/

> +        */
> +       timeout = jiffies + msecs_to_jiffies(VP_VDPA_RESET_TMOUT_MS);
> +       while (vp_modern_get_status(mdev)) {
> +               usleep_range(1000, 1500);
> +               if (time_after(jiffies, timeout)) {
> +                       dev_err(&mdev->pci_dev->dev,
> +                               "vp_vdpa: fail to reset device\n");
> +                       return -ETIMEDOUT;
> +               }

Any chance to use readx_poll_timeout() here?

Thanks

> +       }
> +
>         if (s & VIRTIO_CONFIG_S_DRIVER_OK)
>                 vp_vdpa_free_irq(vp_vdpa);
>
> --
> 2.23.0
>

Re: [PATCH v2] vp_vdpa: harden the logic of set status

[Stefano Garzarella]

On Tue, Dec 06, 2022 at 10:13:21AM +0800, Longpeng(Mike) wrote:
>From: Longpeng <longpeng2@huawei.com>
>
>1. We should not set status to 0 when invoking vp_vdpa_set_status(),
>   trigger a warning in that case.
>
>2. The driver MUST wait for a read of device_status to return 0 before
>   reinitializing the device. But we also don't want to keep us in an
>   infinite loop forever, so wait for 5s if we try to reset the device.
>
>Signed-off-by: Longpeng <longpeng2@huawei.com>
>---
>Changes v1->v2:
>  - use WARN_ON instead of BUG_ON. [Stefano]
>  - use "warning + failed" instead of "infinite loop". [Jason, Stefano]
>  - use usleep_range instead of msleep (checkpatch). [Longpeng]
>
>---
> drivers/vdpa/virtio_pci/vp_vdpa.c | 22 ++++++++++++++++++++++
> 1 file changed, 22 insertions(+)
>
>diff --git a/drivers/vdpa/virtio_pci/vp_vdpa.c b/drivers/vdpa/virtio_pci/vp_vdpa.c
>index 13701c2a1963..a2d3b13ac646 100644
>--- a/drivers/vdpa/virtio_pci/vp_vdpa.c
>+++ b/drivers/vdpa/virtio_pci/vp_vdpa.c
>@@ -214,6 +214,9 @@ static void vp_vdpa_set_status(struct vdpa_device *vdpa, u8 status)
> 	struct virtio_pci_modern_device *mdev = vp_vdpa_to_mdev(vp_vdpa);
> 	u8 s = vp_vdpa_get_status(vdpa);
>
>+	/* We should never be setting status to 0. */
>+	WARN_ON(status == 0);
>+
> 	if (status & VIRTIO_CONFIG_S_DRIVER_OK &&
> 	    !(s & VIRTIO_CONFIG_S_DRIVER_OK)) {
> 		vp_vdpa_request_irq(vp_vdpa);
>@@ -222,14 +225,33 @@ static void vp_vdpa_set_status(struct vdpa_device *vdpa, u8 status)
> 	vp_modern_set_status(mdev, status);
> }
>
>+#define VP_VDPA_RESET_TMOUT_MS 5000 /* 5s */

What about moving this define on top of this file?
Near the other macro.

And I would use TIMEOUT entirely.

>+
> static int vp_vdpa_reset(struct vdpa_device *vdpa, bool clear)
> {
> 	struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
> 	struct virtio_pci_modern_device *mdev = vp_vdpa_to_mdev(vp_vdpa);
> 	u8 s = vp_vdpa_get_status(vdpa);
>+	unsigned long timeout;
>
> 	vp_modern_set_status(mdev, 0);
>
>+	/*
>+	 * As the virtio v1.1 spec (4.1.4.3.2) says: After writing 0 to

I think we can refer to the lates v1.2 (the section should be the same).

>+	 * device_status, the driver MUST wait for a read of device_status
>+	 * to return 0 before reinitializing the device.
>+	 * To avoid keep us here forever, we only wait for 5 seconds.
>+	 */
>+	timeout = jiffies + msecs_to_jiffies(VP_VDPA_RESET_TMOUT_MS);
>+	while (vp_modern_get_status(mdev)) {
>+		usleep_range(1000, 1500);
>+		if (time_after(jiffies, timeout)) {
>+			dev_err(&mdev->pci_dev->dev,
>+				"vp_vdpa: fail to reset device\n");
>+			return -ETIMEDOUT;
>+		}
>+	}
>+
> 	if (s & VIRTIO_CONFIG_S_DRIVER_OK)
> 		vp_vdpa_free_irq(vp_vdpa);
>
>-- 
>2.23.0
>

The rest LGTM!

Thanks,
Stefano

Re: [PATCH v2] vp_vdpa: harden the logic of set status

[Longpeng (Mike, Cloud Infrastructure Service Product Dept.)]

在 2022/12/6 19:00, Stefano Garzarella 写道:
> On Tue, Dec 06, 2022 at 10:13:21AM +0800, Longpeng(Mike) wrote:
>> From: Longpeng <longpeng2@huawei.com>
>>
>> 1. We should not set status to 0 when invoking vp_vdpa_set_status(),
>>   trigger a warning in that case.
>>
>> 2. The driver MUST wait for a read of device_status to return 0 before
>>   reinitializing the device. But we also don't want to keep us in an
>>   infinite loop forever, so wait for 5s if we try to reset the device.
>>
>> Signed-off-by: Longpeng <longpeng2@huawei.com>
>> ---
>> Changes v1->v2:
>>  - use WARN_ON instead of BUG_ON. [Stefano]
>>  - use "warning + failed" instead of "infinite loop". [Jason, Stefano]
>>  - use usleep_range instead of msleep (checkpatch). [Longpeng]
>>
>> ---
>> drivers/vdpa/virtio_pci/vp_vdpa.c | 22 ++++++++++++++++++++++
>> 1 file changed, 22 insertions(+)
>>
>> diff --git a/drivers/vdpa/virtio_pci/vp_vdpa.c 
>> b/drivers/vdpa/virtio_pci/vp_vdpa.c
>> index 13701c2a1963..a2d3b13ac646 100644
>> --- a/drivers/vdpa/virtio_pci/vp_vdpa.c
>> +++ b/drivers/vdpa/virtio_pci/vp_vdpa.c
>> @@ -214,6 +214,9 @@ static void vp_vdpa_set_status(struct vdpa_device 
>> *vdpa, u8 status)
>>     struct virtio_pci_modern_device *mdev = vp_vdpa_to_mdev(vp_vdpa);
>>     u8 s = vp_vdpa_get_status(vdpa);
>>
>> +    /* We should never be setting status to 0. */
>> +    WARN_ON(status == 0);
>> +
>>     if (status & VIRTIO_CONFIG_S_DRIVER_OK &&
>>         !(s & VIRTIO_CONFIG_S_DRIVER_OK)) {
>>         vp_vdpa_request_irq(vp_vdpa);
>> @@ -222,14 +225,33 @@ static void vp_vdpa_set_status(struct 
>> vdpa_device *vdpa, u8 status)
>>     vp_modern_set_status(mdev, status);
>> }
>>
>> +#define VP_VDPA_RESET_TMOUT_MS 5000 /* 5s */
> 
> What about moving this define on top of this file?
> Near the other macro.
> 
> And I would use TIMEOUT entirely.
> 
Okay.

>> +
>> static int vp_vdpa_reset(struct vdpa_device *vdpa, bool clear)
>> {
>>     struct vp_vdpa *vp_vdpa = vdpa_to_vp(vdpa);
>>     struct virtio_pci_modern_device *mdev = vp_vdpa_to_mdev(vp_vdpa);
>>     u8 s = vp_vdpa_get_status(vdpa);
>> +    unsigned long timeout;
>>
>>     vp_modern_set_status(mdev, 0);
>>
>> +    /*
>> +     * As the virtio v1.1 spec (4.1.4.3.2) says: After writing 0 to
> 
> I think we can refer to the lates v1.2 (the section should be the same).
> 
Okay, will do in next version, thanks.

>> +     * device_status, the driver MUST wait for a read of device_status
>> +     * to return 0 before reinitializing the device.
>> +     * To avoid keep us here forever, we only wait for 5 seconds.
>> +     */
>> +    timeout = jiffies + msecs_to_jiffies(VP_VDPA_RESET_TMOUT_MS);
>> +    while (vp_modern_get_status(mdev)) {
>> +        usleep_range(1000, 1500);
>> +        if (time_after(jiffies, timeout)) {
>> +            dev_err(&mdev->pci_dev->dev,
>> +                "vp_vdpa: fail to reset device\n");
>> +            return -ETIMEDOUT;
>> +        }
>> +    }
>> +
>>     if (s & VIRTIO_CONFIG_S_DRIVER_OK)
>>         vp_vdpa_free_irq(vp_vdpa);
>>
>> -- 
>> 2.23.0
>>
> 
> The rest LGTM!
> 
> Thanks,
> Stefano
> 
> 
> .