Skip to contentSkip to conversation

virtio: Improve queue_reset polarity to match to default reset state

PATCH In review Parav Pandit 3 participants 6 messages Apr 25, 2022 – Apr 26, 2022

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 virtio: Improve queue_reset polarity to match to default reset state

Currently when driver initiates a queue reset, device is expected to communicate reset status to the driver by changing the value of the queue_reset register twice. First to return value other than 1 when reset is ongoing, later to return 1 when queue reset is completed.

However initially during the device reset time the queue reset value is zero. queue_reset changes the value of the register to a different value on reset completion. Yet another time queue_reset value is expected to change when queue_select is reprogrammed.

For example in below flow, a created virtqueue, which is disabled by driver leaves the queue state as
queue_enable = 0, queue_reset = 1.

example flow:
a) 0,0 -> device init time value
b) 1,0 -> vq is enabled by driver and working
c) 1,1 -> vq is enabled, driver initiated reset
d) 1,0 -> vq is enabled, but device is busy doing the reset
e) 0,1 -> vq reset is completed in the device and VQ is now disabled (conflict with initial queue reset state #a above)

On next iteration, when queue_select selects the same VQ again, without enablement, device is confused to return 1 or 0 because it was reset once before via queue_reset register.

This demands complex device implementation to understand what should be returned for a VQ that is reset using queue_reset register vs other means.

Instead, it is better and efficient to maintain the same VQ state on the device when queue reset is completed.

new proposed flow:
q_enable, q_reset
A) 0, 0 -> default, device init time
B) 1, 0 -> driver has enabled vq
C) 1, 1 -> driver started q reset
D) 1, 1 -> q_reset stays 1 until device is busy resetting vq (device communicates that its working on resetting VQ, consistent with #C)
E) 0, 0 -> q_reset by device is completed, q got disabled (consistent with device init time #A)

Otherwise a device needs to maintain complex state as,

Current device side implementation requires something like, device.vq_reset_register_read()
{

  if (vq.reset_started) {
    if (vq.reset_completed)
      return queue_reset = 1;
    else if (vq.reset_in_progress)
      return queue_reset = 0;
  } else {
      when vq.queue_select, and vq.queue_enable = 1,
      and if vq.reset_completed. {
        do extra queue_reset = 0;
        vq.reset_completed = 0; /* so that it can go back to original state */
      }
     return queue_reset = 0;

}
}

Compare to above, it is more efficient for device to do:

device.vq_reset_register_read()
{
/* return current state of queue_reset;
* if reset is ongoing, return 1.
* if reset is completed or not done ever, return 0. */
return vq.queue_reset;
}

Hence, this patch proposes a simple change to have reset register polarity to be same as that of initial reset value.

Fixes 12998e738621 ("virtio: pci support virtqueue reset")
---
 content.tex | 19 +++++++++++++------
 1 file changed, 13 insertions(+), 6 deletions(-)
content.tex +13 -6
index 060bdab..77c1c66 100644
@@ -1069,14 +1069,18 @@ \subsubsection{Common configuration structure layout}\label{sec:Virtio Transport
 The driver MUST NOT write a 0 to \field{queue_enable}.
 
 If VIRTIO_F_RING_RESET has been negotiated, after the driver writes 1 to
-\field{queue_reset} to reset the queue, it MUST verify that the queue
-has been reset by reading back \field{queue_reset} and ensuring that it
-is 1. The driver MAY re-enable the queue by writing a 1 to
+\field{queue_reset} to reset the queue, driver MUST verify that the queue
+has been reset by reading back \field{queue_reset} until device returns a value of 0.
+Device continue to report value of 1 for the \field{queue_reset} until device finishes
+the queue reset request. When the device completes the queue reset, \field{queue_reset}
+and \field{queue_enable} set to zero, indicating
+that specified queue is ready to be enabled again. The driver MAY re-enable the queue by writing a 1 to
 \field{queue_enable} after ensuring that the other virtqueue fields have
 been set up correctly. The driver MAY set driver-writeable queue configuration
 values to different values than those that were used before the queue reset.
 (see \ref{sec:Basic Facilities of a Virtio Device / Virtqueues / Virtqueue Reset}).
 
+
 \subsubsection{Notification structure layout}\label{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI Device Layout / Notification capability}
 
 The notification location is found using the VIRTIO_PCI_CAP_NOTIFY_CFG
@@ -2063,9 +2067,12 @@ \subsection{MMIO Device Register Layout}\label{sec:Virtio Transport Options / Vi
 it finishes handling an interrupt and MUST NOT set any of the undefined bits in the value.
 
 If VIRTIO_F_RING_RESET has been negotiated, after the driver writes 1 to
-\field{QueueReset} to reset the queue, it MUST verify that the queue
-has been reset by reading back \field{QueueReset} and ensuring that it
-is 1. The driver MAY re-enable the queue by writing a 1 to
+\field{QueueReset} to reset the queue, driver MUST verify that the queue
+has been reset by reading back \field{QueueReset} until device returns a value of 0.
+Device continue to report value of 1 for the \field{QueueReset} until device finishes
+the reset. When the device completes the queue reset, \field{QueueReset} and
+\field{QueueReady} set to zero, indicating that specified queue is ready to be
+enabled again. The driver MAY re-enable the queue by writing a 1 to
 \field{QueueReady} after ensuring that the other virtqueue fields have
 been set up correctly. The driver MAY set driver-writeable queue configuration
 values to different values than those that were used before the queue reset.
Signature · 1 line
1.8.3.1
#1 lore raw
Quoted 64 lines · Parav Pandit

Currently when driver initiates a queue reset, device is expected to communicate reset status to the driver by changing the value of the queue_reset register twice. First to return value other than 1 when reset is ongoing, later to return 1 when queue reset is completed.

However initially during the device reset time the queue reset value is zero. queue_reset changes the value of the register to a different value on reset completion. Yet another time queue_reset value is expected to change when queue_select is reprogrammed.

For example in below flow, a created virtqueue, which is disabled by driver leaves the queue state as
queue_enable = 0, queue_reset = 1.

example flow:
a) 0,0 -> device init time value
b) 1,0 -> vq is enabled by driver and working
c) 1,1 -> vq is enabled, driver initiated reset
d) 1,0 -> vq is enabled, but device is busy doing the reset
e) 0,1 -> vq reset is completed in the device and VQ is now disabled (conflict with initial queue reset state #a above)

On next iteration, when queue_select selects the same VQ again, without enablement, device is confused to return 1 or 0 because it was reset once before via queue_reset register.

This demands complex device implementation to understand what should be returned for a VQ that is reset using queue_reset register vs other means.

Instead, it is better and efficient to maintain the same VQ state on the device when queue reset is completed.

new proposed flow:
q_enable, q_reset
A) 0, 0 -> default, device init time
B) 1, 0 -> driver has enabled vq
C) 1, 1 -> driver started q reset
D) 1, 1 -> q_reset stays 1 until device is busy resetting vq (device communicates that its working on resetting VQ, consistent with #C)
E) 0, 0 -> q_reset by device is completed, q got disabled (consistent with device init time #A)

Otherwise a device needs to maintain complex state as,

Current device side implementation requires something like, device.vq_reset_register_read()
{

  if (vq.reset_started) {
    if (vq.reset_completed)
      return queue_reset = 1;
    else if (vq.reset_in_progress)
      return queue_reset = 0;
  } else {
      when vq.queue_select, and vq.queue_enable = 1,
      and if vq.reset_completed. {
        do extra queue_reset = 0;
        vq.reset_completed = 0; /* so that it can go back to original state */
      }
     return queue_reset = 0;

}
}
Compare to above, it is more efficient for device to do:

Quoted 8 lines

device.vq_reset_register_read()
{
/* return current state of queue_reset;
* if reset is ongoing, return 1.
* if reset is completed or not done ever, return 0. */
return vq.queue_reset;
}

That's a weird way to describe this. I guess the logic in the spec is simply a register bit that is set to != 1 when reset starts and set to 1 when reset completes.
You propose reversing its polarity. Fair enough.

I do think it's inelegant that the value is different when reset is through vq reset (1) and through device reset (0).

Quoted 16 lines

Hence, this patch proposes a simple change to have reset register polarity to be same as that of initial reset value.

Fixes 12998e738621 ("virtio: pci support virtqueue reset")
---
 content.tex | 19 +++++++++++++------
 1 file changed, 13 insertions(+), 6 deletions(-)
content.tex +0 -0
index 060bdab..77c1c66 100644

content.tex \subsubsection{Common configuration structure layout}\label{sec:Virtio Transport · line 1074

Quoted 9 lines
content.tex +3 -3
@@ -1069,14 +1069,18 @@ \subsubsection{Common configuration structure layout}\label{sec:Virtio Transport
 The driver MUST NOT write a 0 to \field{queue_enable}.
 
 If VIRTIO_F_RING_RESET has been negotiated, after the driver writes 1 to
-\field{queue_reset} to reset the queue, it MUST verify that the queue
-has been reset by reading back \field{queue_reset} and ensuring that it
-is 1. The driver MAY re-enable the queue by writing a 1 to
+\field{queue_reset} to reset the queue, driver MUST verify that the queue
+has been reset by reading back \field{queue_reset} until device returns a value of 0.
+Device continue to report value of 1 for the \field{queue_reset} until device finishes

this MUST should move to conformance chapter BTW.

content.tex \subsubsection{Common configuration structure layout}\label{sec:Virtio Transport · line 1076

Quoted 2 lines
content.tex +2 -0
+the queue reset request. When the device completes the queue reset, \field{queue_reset}
+and \field{queue_enable} set to zero,

set-> "are set"

Quoted 2 lines

indicating
+that specified queue is ready to be enabled again. The driver MAY re-enable the queue by writing a 1 to

a 1 -> 1

Quoted 1 line

\field{queue_enable} after ensuring that the other virtqueue fields have

the other -> other (not your fault)

Quoted 5 lines

been set up correctly. The driver MAY set driver-writeable queue configuration values to different values than those that were used before the queue reset. (see \ref{sec:Basic Facilities of a Virtio Device / Virtqueues / Virtqueue Reset}).

+

drop this pls.

Quoted 7 lines

\subsubsection{Notification structure layout}\label{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI Device Layout / Notification capability}

The notification location is found using the VIRTIO_PCI_CAP_NOTIFY_CFG

content.tex +0 -0
@@ -2063,9 +2067,12 @@ \subsection{MMIO Device Register Layout}\label{sec:Virtio Transport Options / Vi
 it finishes handling an interrupt and MUST NOT set any of the undefined bits in the value.
 
 If VIRTIO_F_RING_RESET has been negotiated, after the driver writes 1 to

content.tex \subsection{MMIO Device Register Layout}\label{sec:Virtio Transport Options / Vi · line 2076

Quoted 10 lines
content.tex +6 -3
-\field{QueueReset} to reset the queue, it MUST verify that the queue
-has been reset by reading back \field{QueueReset} and ensuring that it
-is 1. The driver MAY re-enable the queue by writing a 1 to
+\field{QueueReset} to reset the queue, driver MUST verify that the queue
+has been reset by reading back \field{QueueReset} until device returns a value of 0.
+Device continue to report value of 1 for the \field{QueueReset} until device finishes
+the reset. When the device completes the queue reset, \field{QueueReset} and
+\field{QueueReady} set to zero, indicating that specified queue is ready to be
+enabled again. The driver MAY re-enable the queue by writing a 1 to
 \field{QueueReady} after ensuring that the other virtqueue fields have

same comments.

content.tex \subsection{MMIO Device Register Layout}\label{sec:Virtio Transport Options / Vi · line 2078

Quoted 2 lines
content.tex +0 -0
 been set up correctly. The driver MAY set driver-writeable queue configuration
 values to different values than those that were used before the queue reset.

ccw will have to be changed too.

Quoted 2 lines
Signature · 1 line
1.8.3.1
#2 lore raw
Quoted 4 lines

From: Michael S. Tsirkin <[email protected]>
Sent: Monday, April 25, 2022 7:11 PM

[..]

Quoted 5 lines

That's a weird way to describe this. I guess the logic in the spec is simply a register bit that is set to != 1 when reset starts and set to 1 when reset completes.
You propose reversing its polarity. Fair enough.

I will drop this part from the commit message as example clarifies it enough without device side code.

Quoted 10 lines

I do think it's inelegant that the value is different when reset is through vq reset (1) and through device reset (0).

Quoted 6 lines

If VIRTIO_F_RING_RESET has been negotiated, after the driver writes 1
to -\field{queue_reset} to reset the queue, it MUST verify that the queue -has been reset by reading back \field{queue_reset} and ensuring that it -is 1. The driver MAY re-enable the queue by writing a 1 to +\field{queue_reset} to reset the queue, driver MUST verify that the +queue has been reset by reading back \field{queue_reset} until device

Quoted 5 lines

returns a value of 0.

Review

Quoted 2 lines
(quoted diff) +2 -0
+Device continue to report value of 1 for the \field{queue_reset}
+until device finishes

this MUST should move to conformance chapter BTW.

In this proposal it is covered in the section 4.1.4.3.2 " Driver Requirements: Common configuration structure layout". This section is listed in 7.2 " Clause 2: PCI Driver Conformance".

To me it appears that this is covered in the conformance. What am I missing?

Quoted 6 lines
Quoted 3 lines

+the queue reset request. When the device completes the queue reset, +\field{queue_reset} and \field{queue_enable} set to zero,

set-> "are set"

Will fix.

Quoted 5 lines

Review

Quoted 3 lines

indicating

(quoted diff) +2 -0
+that specified queue is ready to be enabled again. The driver MAY
+re-enable the queue by writing a 1 to

a 1 -> 1

Will fix.

Quoted 6 lines
Quoted 3 lines

\field{queue_enable} after ensuring that the other virtqueue fields
have

the other -> other (not your fault)

Should this be pre-patch?
Or small enough editorial change to be part of this?

Quoted 9 lines
Quoted 2 lines

been set up correctly. The driver MAY set driver-writeable queue
configuration values to different values than those that were used before

the queue reset.

Quoted 1 line

(see \ref{sec:Basic Facilities of a Virtio Device / Virtqueues / Virtqueue

Reset}).

Quoted 2 lines

+

drop this pls.

My bad. Will fix it.

Quoted 2 lines

ccw will have to be changed too.

My weak area. Didn't find the vq reset section in below 3 commits.

a4ce81a virtio: mmio support virtqueue reset 12998e7 virtio: pci support virtqueue reset 3b5378d virtio: introduce virtqueue reset as basic facility

In a previous email by Cornelia, she replied " For ccw, we have not yet added queue reset;". So, can we please do this later in a some a non_bug_fix patch?

I will revise v1 for the comments that I can fix. Based on v1 review, we can split patches and improve further.

#3 lore raw
Quoted 29 lines · Parav Pandit
Quoted 4 lines

From: Michael S. Tsirkin <[email protected]>
Sent: Monday, April 25, 2022 7:11 PM

[..]

Quoted 5 lines

That's a weird way to describe this. I guess the logic in the spec is simply a register bit that is set to != 1 when reset starts and set to 1 when reset completes.
You propose reversing its polarity. Fair enough.

I will drop this part from the commit message as example clarifies it enough without device side code.

Quoted 10 lines

I do think it's inelegant that the value is different when reset is through vq reset (1) and through device reset (0).

Quoted 6 lines

If VIRTIO_F_RING_RESET has been negotiated, after the driver writes 1
to -\field{queue_reset} to reset the queue, it MUST verify that the queue -has been reset by reading back \field{queue_reset} and ensuring that it -is 1. The driver MAY re-enable the queue by writing a 1 to +\field{queue_reset} to reset the queue, driver MUST verify that the +queue has been reset by reading back \field{queue_reset} until device

Quoted 5 lines

returns a value of 0.

Review

Quoted 2 lines
(quoted diff) +2 -0
+Device continue to report value of 1 for the \field{queue_reset}
+until device finishes

this MUST should move to conformance chapter BTW.

Quoted 5 lines

In this proposal it is covered in the section 4.1.4.3.2 " Driver Requirements: Common configuration structure layout". This section is listed in 7.2 " Clause 2: PCI Driver Conformance".

To me it appears that this is covered in the conformance. What am I missing?

My bad. git diff did something I did not expect here showing \subsection instead of \drivernormative. I'll have to learn how it decides which line to show as function name.

Quoted 16 lines
Quoted 6 lines
Quoted 3 lines

+the queue reset request. When the device completes the queue reset, +\field{queue_reset} and \field{queue_enable} set to zero,

set-> "are set"

Will fix.

Quoted 5 lines

Review

Quoted 3 lines

indicating

(quoted diff) +2 -0
+that specified queue is ready to be enabled again. The driver MAY
+re-enable the queue by writing a 1 to

a 1 -> 1

Will fix.

Quoted 1 line
Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

\field{queue_enable} after ensuring that the other virtqueue fields
have

the other -> other (not your fault)

Should this be pre-patch?
Or small enough editorial change to be part of this?

ideally a separate patch and issue.

Quoted 13 lines
Quoted 9 lines
Quoted 2 lines

been set up correctly. The driver MAY set driver-writeable queue
configuration values to different values than those that were used before

the queue reset.

Quoted 1 line

(see \ref{sec:Basic Facilities of a Virtio Device / Virtqueues / Virtqueue

Reset}).

Quoted 2 lines

+

drop this pls.

My bad. Will fix it.

Quoted 2 lines

ccw will have to be changed too.

Quoted 8 lines

My weak area. Didn't find the vq reset section in below 3 commits.

a4ce81a virtio: mmio support virtqueue reset 12998e7 virtio: pci support virtqueue reset 3b5378d virtio: introduce virtqueue reset as basic facility

In a previous email by Cornelia, she replied " For ccw, we have not yet added queue reset;". So, can we please do this later in a some a non_bug_fix patch?

Oh right, I was confused. maybe mention in commit log so others are not confused.

Quoted 2 lines

I will revise v1 for the comments that I can fix. Based on v1 review, we can split patches and improve further.

#4 lore raw
In reply to Parav Pandit
Quoted 74 lines · Parav Pandit

Currently when driver initiates a queue reset, device is expected to communicate reset status to the driver by changing the value of the queue_reset register twice. First to return value other than 1 when reset is ongoing, later to return 1 when queue reset is completed.

However initially during the device reset time the queue reset value is zero. queue_reset changes the value of the register to a different value on reset completion. Yet another time queue_reset value is expected to change when queue_select is reprogrammed.

For example in below flow, a created virtqueue, which is disabled by driver leaves the queue state as
queue_enable = 0, queue_reset = 1.

example flow:
a) 0,0 -> device init time value
b) 1,0 -> vq is enabled by driver and working
c) 1,1 -> vq is enabled, driver initiated reset
d) 1,0 -> vq is enabled, but device is busy doing the reset
e) 0,1 -> vq reset is completed in the device and VQ is now disabled (conflict with initial queue reset state #a above)

On next iteration, when queue_select selects the same VQ again, without enablement, device is confused to return 1 or 0 because it was reset once before via queue_reset register.

This demands complex device implementation to understand what should be returned for a VQ that is reset using queue_reset register vs other means.

Instead, it is better and efficient to maintain the same VQ state on the device when queue reset is completed.

new proposed flow:
q_enable, q_reset
A) 0, 0 -> default, device init time
B) 1, 0 -> driver has enabled vq
C) 1, 1 -> driver started q reset
D) 1, 1 -> q_reset stays 1 until device is busy resetting vq (device communicates that its working on resetting VQ, consistent with #C)
E) 0, 0 -> q_reset by device is completed, q got disabled (consistent with device init time #A)

Otherwise a device needs to maintain complex state as,

Current device side implementation requires something like, device.vq_reset_register_read()
{

  if (vq.reset_started) {
    if (vq.reset_completed)
      return queue_reset = 1;
    else if (vq.reset_in_progress)
      return queue_reset = 0;
  } else {
      when vq.queue_select, and vq.queue_enable = 1,
      and if vq.reset_completed. {
        do extra queue_reset = 0;
        vq.reset_completed = 0; /* so that it can go back to original state */
      }
     return queue_reset = 0;

}
}

Compare to above, it is more efficient for device to do:

device.vq_reset_register_read()
{
/* return current state of queue_reset;
* if reset is ongoing, return 1.
* if reset is completed or not done ever, return 0. */
return vq.queue_reset;
}

Quoted 6 lines

Hence, this patch proposes a simple change to have reset register polarity to be same as that of initial reset value.

Fixes 12998e738621 ("virtio: pci support virtqueue reset")

BTW post this to virtio-comment pls (cross post to virtio-dev is ok).

Quoted 48 lines
---
 content.tex | 19 +++++++++++++------
 1 file changed, 13 insertions(+), 6 deletions(-)
content.tex +13 -6
index 060bdab..77c1c66 100644
@@ -1069,14 +1069,18 @@ \subsubsection{Common configuration structure layout}\label{sec:Virtio Transport
 The driver MUST NOT write a 0 to \field{queue_enable}.
 
 If VIRTIO_F_RING_RESET has been negotiated, after the driver writes 1 to
-\field{queue_reset} to reset the queue, it MUST verify that the queue
-has been reset by reading back \field{queue_reset} and ensuring that it
-is 1. The driver MAY re-enable the queue by writing a 1 to
+\field{queue_reset} to reset the queue, driver MUST verify that the queue
+has been reset by reading back \field{queue_reset} until device returns a value of 0.
+Device continue to report value of 1 for the \field{queue_reset} until device finishes
+the queue reset request. When the device completes the queue reset, \field{queue_reset}
+and \field{queue_enable} set to zero, indicating
+that specified queue is ready to be enabled again. The driver MAY re-enable the queue by writing a 1 to
 \field{queue_enable} after ensuring that the other virtqueue fields have
 been set up correctly. The driver MAY set driver-writeable queue configuration
 values to different values than those that were used before the queue reset.
 (see \ref{sec:Basic Facilities of a Virtio Device / Virtqueues / Virtqueue Reset}).
 
+
 \subsubsection{Notification structure layout}\label{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI Device Layout / Notification capability}
 
 The notification location is found using the VIRTIO_PCI_CAP_NOTIFY_CFG
@@ -2063,9 +2067,12 @@ \subsection{MMIO Device Register Layout}\label{sec:Virtio Transport Options / Vi
 it finishes handling an interrupt and MUST NOT set any of the undefined bits in the value.
 
 If VIRTIO_F_RING_RESET has been negotiated, after the driver writes 1 to
-\field{QueueReset} to reset the queue, it MUST verify that the queue
-has been reset by reading back \field{QueueReset} and ensuring that it
-is 1. The driver MAY re-enable the queue by writing a 1 to
+\field{QueueReset} to reset the queue, driver MUST verify that the queue
+has been reset by reading back \field{QueueReset} until device returns a value of 0.
+Device continue to report value of 1 for the \field{QueueReset} until device finishes
+the reset. When the device completes the queue reset, \field{QueueReset} and
+\field{QueueReady} set to zero, indicating that specified queue is ready to be
+enabled again. The driver MAY re-enable the queue by writing a 1 to
 \field{QueueReady} after ensuring that the other virtqueue fields have
 been set up correctly. The driver MAY set driver-writeable queue configuration
 values to different values than those that were used before the queue reset.
Signature · 1 line
1.8.3.1
#5 lore raw
In reply to Parav Pandit
Quoted 31 lines · Parav Pandit

Currently when driver initiates a queue reset, device is expected to communicate reset status to the driver by changing the value of the queue_reset register twice. First to return value other than 1 when reset is ongoing, later to return 1 when queue reset is completed.

However initially during the device reset time the queue reset value is zero. queue_reset changes the value of the register to a different value on reset completion. Yet another time queue_reset value is expected to change when queue_select is reprogrammed.

For example in below flow, a created virtqueue, which is disabled by driver leaves the queue state as
queue_enable = 0, queue_reset = 1.

example flow:
a) 0,0 -> device init time value
b) 1,0 -> vq is enabled by driver and working
c) 1,1 -> vq is enabled, driver initiated reset
d) 1,0 -> vq is enabled, but device is busy doing the reset
e) 0,1 -> vq reset is completed in the device and VQ is now disabled (conflict with initial queue reset state #a above)

On next iteration, when queue_select selects the same VQ again, without enablement, device is confused to return 1 or 0 because it was reset once before via queue_reset register.

This demands complex device implementation to understand what should be returned for a VQ that is reset using queue_reset register vs other means.

Quoted 7 lines

Instead, it is better and efficient to maintain the same VQ state on the device when queue reset is completed.

new proposed flow:
q_enable, q_reset
A) 0, 0 -> default, device init time
B) 1, 0 -> driver has enabled vq

I wonder whether we it's more consistent with device reset than using write to clear if we do

B) 1, 1 -> driver has enabled vq
C) 1, 1 -> driver starts q rest by writing 0
D) 1, 1 -> device is resetting vq
E) 0, 0 -> vq is rest and disabled

Quoted 5 lines

C) 1, 1 -> driver started q reset
D) 1, 1 -> q_reset stays 1 until device is busy resetting vq (device communicates that its working on resetting VQ, consistent with #C)
E) 0, 0 -> q_reset by device is completed, q got disabled (consistent with device init time #A)

Is it too late to simply use a single queue_enable rather than introducing queue_rest?

Thanks

Quoted 92 lines

Otherwise a device needs to maintain complex state as,

Current device side implementation requires something like, device.vq_reset_register_read()
{

  if (vq.reset_started) {
    if (vq.reset_completed)
      return queue_reset = 1;
    else if (vq.reset_in_progress)
      return queue_reset = 0;
  } else {
      when vq.queue_select, and vq.queue_enable = 1,
      and if vq.reset_completed. {
        do extra queue_reset = 0;
        vq.reset_completed = 0; /* so that it can go back to original state */
      }
     return queue_reset = 0;

}
}

Compare to above, it is more efficient for device to do:

device.vq_reset_register_read()
{
/* return current state of queue_reset;
* if reset is ongoing, return 1.
* if reset is completed or not done ever, return 0. */
return vq.queue_reset;
}

Hence, this patch proposes a simple change to have reset register polarity to be same as that of initial reset value.

Fixes 12998e738621 ("virtio: pci support virtqueue reset")
---
 content.tex | 19 +++++++++++++------
 1 file changed, 13 insertions(+), 6 deletions(-)
content.tex +13 -7
index 060bdab..77c1c66 100644
@@ -1069,14 +1069,18 @@ \subsubsection{Common configuration structure layout}\label{sec:Virtio Transport
 The driver MUST NOT write a 0 to \field{queue_enable}.
 
 If VIRTIO_F_RING_RESET has been negotiated, after the driver writes 1 to
-\field{queue_reset} to reset the queue, it MUST verify that the queue
-has been reset by reading back \field{queue_reset} and ensuring that it
-is 1. The driver MAY re-enable the queue by writing a 1 to
+\field{queue_reset} to reset the queue, driver MUST verify that the queue
+has been reset by reading back \field{queue_reset} until device returns a value of 0.
+Device continue to report value of 1 for the \field{queue_reset} until device finishes
+the queue reset request. When the device completes the queue reset, \field{queue_reset}
+and \field{queue_enable} set to zero, indicating
+that specified queue is ready to be enabled again. The driver MAY re-enable the queue by writing a 1 to
 \field{queue_enable} after ensuring that the other virtqueue fields have
 been set up correctly. The driver MAY set driver-writeable queue configuration
 values to different values than those that were used before the queue reset.
 (see \ref{sec:Basic Facilities of a Virtio Device / Virtqueues / Virtqueue Reset}).
 
+
 \subsubsection{Notification structure layout}\label{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI Device Layout / Notification capability}
 
 The notification location is found using the VIRTIO_PCI_CAP_NOTIFY_CFG
@@ -2063,9 +2067,12 @@ \subsection{MMIO Device Register Layout}\label{sec:Virtio Transport Options / Vi
 it finishes handling an interrupt and MUST NOT set any of the undefined bits in the value.
 
 If VIRTIO_F_RING_RESET has been negotiated, after the driver writes 1 to
-\field{QueueReset} to reset the queue, it MUST verify that the queue
-has been reset by reading back \field{QueueReset} and ensuring that it
-is 1. The driver MAY re-enable the queue by writing a 1 to
+\field{QueueReset} to reset the queue, driver MUST verify that the queue
+has been reset by reading back \field{QueueReset} until device returns a value of 0.
+Device continue to report value of 1 for the \field{QueueReset} until device finishes
+the reset. When the device completes the queue reset, \field{QueueReset} and
+\field{QueueReady} set to zero, indicating that specified queue is ready to be
+enabled again. The driver MAY re-enable the queue by writing a 1 to
 \field{QueueReady} after ensuring that the other virtqueue fields have
 been set up correctly. The driver MAY set driver-writeable queue configuration
 values to different values than those that were used before the queue reset.
--

1.8.3.1

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]