Skip to contentSkip to conversation

doc: update TAP device features

PATCH In review Stephen Hemminger 4 participants 15 messages Sep 4, 2024 – Oct 9, 2024

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 doc: update TAP device features

The TAP device does have per-queue stats and handles multi-process.

---
 doc/guides/nics/features/tap.ini | 2 ++
 1 file changed, 2 insertions(+)
doc/guides/nics/features/tap.ini +2 -0
index f26355e57f..f2ea5cd833 100644
@@ -14,10 +14,12 @@ Basic stats          = Y
 L3 checksum offload  = Y
 L4 checksum offload  = Y
 MTU update           = Y
+Multiprocess aware   = Y
 Multicast MAC filter = Y
 Unicast MAC filter   = Y
 Packet type parsing  = Y
 Flow control         = Y
+Stats per queue      = Y
 Linux                = Y
 ARMv7                = Y
 ARMv8                = Y
Signature · 1 line
2.45.2
#1 lore raw
Quoted 12 lines · Stephen Hemminger

The TAP device does have per-queue stats and handles multi-process.

---
 doc/guides/nics/features/tap.ini | 2 ++
 1 file changed, 2 insertions(+)
doc/guides/nics/features/tap.ini +0 -0
index f26355e57f..f2ea5cd833 100644

doc/guides/nics/features/tap.ini Basic stats = Y · line 18

Quoted 5 lines
doc/guides/nics/features/tap.ini +1 -0
@@ -14,10 +14,12 @@ Basic stats          = Y
 L3 checksum offload  = Y
 L4 checksum offload  = Y
 MTU update           = Y
+Multiprocess aware   = Y
 

ack

doc/guides/nics/features/tap.ini Basic stats = Y

Quoted 6 lines
doc/guides/nics/features/tap.ini +1 -0
 Multicast MAC filter = Y
 Unicast MAC filter   = Y
 Packet type parsing  = Y
 Flow control         = Y
+Stats per queue      = Y
 

This feature name is misleading,
it is for 'rte_eth_dev_set_[rt]x_queue_stats_mapping()' API, which is indeed for covering limitation for some drivers.
Tap does support getting stats per queue, but doesn't support above documented feature.

Quoted 3 lines
doc/guides/nics/features/tap.ini +0 -0
 Linux                = Y

ARMv7 = Y
ARMv8 = Y

#2 lore raw
Quoted 28 lines · On Fri, 4 Oct 2024 02:48:21 +0100 Ferruh Yigit <[email protected]> wrote:
Quoted 12 lines · Stephen Hemminger

The TAP device does have per-queue stats and handles multi-process.

---
 doc/guides/nics/features/tap.ini | 2 ++
 1 file changed, 2 insertions(+)
doc/guides/nics/features/tap.ini +0 -0
index f26355e57f..f2ea5cd833 100644

doc/guides/nics/features/tap.ini Basic stats = Y · line 18

Quoted 5 lines
doc/guides/nics/features/tap.ini +1 -0
@@ -14,10 +14,12 @@ Basic stats          = Y
 L3 checksum offload  = Y
 L4 checksum offload  = Y
 MTU update           = Y
+Multiprocess aware   = Y
 

ack

doc/guides/nics/features/tap.ini Basic stats = Y

Quoted 6 lines
doc/guides/nics/features/tap.ini +1 -0
 Multicast MAC filter = Y
 Unicast MAC filter   = Y
 Packet type parsing  = Y
 Flow control         = Y
+Stats per queue      = Y
 
Quoted 5 lines

This feature name is misleading,
it is for 'rte_eth_dev_set_[rt]x_queue_stats_mapping()' API, which is indeed for covering limitation for some drivers.
Tap does support getting stats per queue, but doesn't support above documented feature.

The stats queue mapping was a feature that was hinted at being removed. It only exists because of HW limitations on Intel ixgbe NIC and SW limitations from RTE_ETHDEV_QUEUE_STAT_CNTRS.

Perhaps there should be a generic SW emulation for this the mapping?

#3 lore raw
Quoted 37 lines · Stephen Hemminger
Quoted 28 lines · On Fri, 4 Oct 2024 02:48:21 +0100 Ferruh Yigit <[email protected]> wrote:
Quoted 12 lines · Stephen Hemminger

The TAP device does have per-queue stats and handles multi-process.

---
 doc/guides/nics/features/tap.ini | 2 ++
 1 file changed, 2 insertions(+)
doc/guides/nics/features/tap.ini +0 -0
index f26355e57f..f2ea5cd833 100644

doc/guides/nics/features/tap.ini Basic stats = Y · line 18

Quoted 5 lines
doc/guides/nics/features/tap.ini +1 -0
@@ -14,10 +14,12 @@ Basic stats          = Y
 L3 checksum offload  = Y
 L4 checksum offload  = Y
 MTU update           = Y
+Multiprocess aware   = Y
 

ack

doc/guides/nics/features/tap.ini Basic stats = Y

Quoted 6 lines
doc/guides/nics/features/tap.ini +1 -0
 Multicast MAC filter = Y
 Unicast MAC filter   = Y
 Packet type parsing  = Y
 Flow control         = Y
+Stats per queue      = Y
 
Quoted 5 lines

This feature name is misleading,
it is for 'rte_eth_dev_set_[rt]x_queue_stats_mapping()' API, which is indeed for covering limitation for some drivers.
Tap does support getting stats per queue, but doesn't support above documented feature.

Quoted 4 lines

The stats queue mapping was a feature that was hinted at being removed. It only exists because of HW limitations on Intel ixgbe NIC and SW limitations from RTE_ETHDEV_QUEUE_STAT_CNTRS.

We have a plan to remove 'RTE_ETHDEV_QUEUE_STAT_CNTRS', by moving queue stats to xstats.

But ixgbe limitation is there.

Quoted 2 lines

Perhaps there should be a generic SW emulation for this the mapping?

Ack, cc'ed Bruce.
But I am not sure ROI of the effort at this stage.

#4 lore raw
Quoted 47 lines · Ferruh Yigit
Quoted 37 lines · Stephen Hemminger
Quoted 28 lines · On Fri, 4 Oct 2024 02:48:21 +0100 Ferruh Yigit <[email protected]> wrote:
Quoted 12 lines · Stephen Hemminger

The TAP device does have per-queue stats and handles multi-process.

---
 doc/guides/nics/features/tap.ini | 2 ++
 1 file changed, 2 insertions(+)
doc/guides/nics/features/tap.ini +0 -0
index f26355e57f..f2ea5cd833 100644

doc/guides/nics/features/tap.ini Basic stats = Y · line 18

Quoted 5 lines
doc/guides/nics/features/tap.ini +1 -0
@@ -14,10 +14,12 @@ Basic stats          = Y
 L3 checksum offload  = Y
 L4 checksum offload  = Y
 MTU update           = Y
+Multiprocess aware   = Y
 

ack

doc/guides/nics/features/tap.ini Basic stats = Y

Quoted 6 lines
doc/guides/nics/features/tap.ini +1 -0
 Multicast MAC filter = Y
 Unicast MAC filter   = Y
 Packet type parsing  = Y
 Flow control         = Y
+Stats per queue      = Y
 
Quoted 5 lines

This feature name is misleading,
it is for 'rte_eth_dev_set_[rt]x_queue_stats_mapping()' API, which is indeed for covering limitation for some drivers.
Tap does support getting stats per queue, but doesn't support above documented feature.

Quoted 4 lines

The stats queue mapping was a feature that was hinted at being removed. It only exists because of HW limitations on Intel ixgbe NIC and SW limitations from RTE_ETHDEV_QUEUE_STAT_CNTRS.

We have a plan to remove 'RTE_ETHDEV_QUEUE_STAT_CNTRS', by moving queue stats to xstats.

Quoted 7 lines

But ixgbe limitation is there.

Quoted 2 lines

Perhaps there should be a generic SW emulation for this the mapping?

Ack, cc'ed Bruce.
But I am not sure ROI of the effort at this stage.

Not sure what the specific ask for me is here. :-) Overall, I think moving queue stats to xstats is the best way to go.

/Bruce

#5 lore raw
Quoted 53 lines · On Fri, 4 Oct 2024 08:54:43 +0100 Bruce Richardson…
Quoted 52 lines · Ferruh Yigit
Quoted 37 lines · Stephen Hemminger
Quoted 28 lines · On Fri, 4 Oct 2024 02:48:21 +0100 Ferruh Yigit <[email protected]> wrote:
Quoted 12 lines · Stephen Hemminger

The TAP device does have per-queue stats and handles multi-process.

---
 doc/guides/nics/features/tap.ini | 2 ++
 1 file changed, 2 insertions(+)
doc/guides/nics/features/tap.ini +0 -0
index f26355e57f..f2ea5cd833 100644

doc/guides/nics/features/tap.ini Basic stats = Y · line 18

Quoted 5 lines
doc/guides/nics/features/tap.ini +1 -0
@@ -14,10 +14,12 @@ Basic stats          = Y
 L3 checksum offload  = Y
 L4 checksum offload  = Y
 MTU update           = Y
+Multiprocess aware   = Y
 

ack

doc/guides/nics/features/tap.ini Basic stats = Y

Quoted 6 lines
doc/guides/nics/features/tap.ini +1 -0
 Multicast MAC filter = Y
 Unicast MAC filter   = Y
 Packet type parsing  = Y
 Flow control         = Y
+Stats per queue      = Y
 
Quoted 5 lines

This feature name is misleading,
it is for 'rte_eth_dev_set_[rt]x_queue_stats_mapping()' API, which is indeed for covering limitation for some drivers.
Tap does support getting stats per queue, but doesn't support above documented feature.

Quoted 4 lines

The stats queue mapping was a feature that was hinted at being removed. It only exists because of HW limitations on Intel ixgbe NIC and SW limitations from RTE_ETHDEV_QUEUE_STAT_CNTRS.

We have a plan to remove 'RTE_ETHDEV_QUEUE_STAT_CNTRS', by moving queue stats to xstats.

But ixgbe limitation is there.

Quoted 2 lines

Perhaps there should be a generic SW emulation for this the mapping?

Quoted 7 lines
Quoted 2 lines

Ack, cc'ed Bruce.
But I am not sure ROI of the effort at this stage.

Not sure what the specific ask for me is here. :-) Overall, I think moving queue stats to xstats is the best way to go.

/Bruce

The problem is that xstats are free form, so it is not possible for application to use them in device independent manner.

#6 lore raw
In reply to Bruce Richardson
Quoted 50 lines · Bruce Richardson
Quoted 49 lines · Ferruh Yigit
Quoted 37 lines · Stephen Hemminger
Quoted 28 lines · On Fri, 4 Oct 2024 02:48:21 +0100 Ferruh Yigit <[email protected]> wrote:
Quoted 12 lines · Stephen Hemminger

The TAP device does have per-queue stats and handles multi-process.

---
 doc/guides/nics/features/tap.ini | 2 ++
 1 file changed, 2 insertions(+)
doc/guides/nics/features/tap.ini +0 -0
index f26355e57f..f2ea5cd833 100644

doc/guides/nics/features/tap.ini Basic stats = Y · line 18

Quoted 5 lines
doc/guides/nics/features/tap.ini +1 -0
@@ -14,10 +14,12 @@ Basic stats          = Y
 L3 checksum offload  = Y
 L4 checksum offload  = Y
 MTU update           = Y
+Multiprocess aware   = Y
 

ack

doc/guides/nics/features/tap.ini Basic stats = Y

Quoted 6 lines
doc/guides/nics/features/tap.ini +1 -0
 Multicast MAC filter = Y
 Unicast MAC filter   = Y
 Packet type parsing  = Y
 Flow control         = Y
+Stats per queue      = Y
 
Quoted 5 lines

This feature name is misleading,
it is for 'rte_eth_dev_set_[rt]x_queue_stats_mapping()' API, which is indeed for covering limitation for some drivers.
Tap does support getting stats per queue, but doesn't support above documented feature.

Quoted 4 lines

The stats queue mapping was a feature that was hinted at being removed. It only exists because of HW limitations on Intel ixgbe NIC and SW limitations from RTE_ETHDEV_QUEUE_STAT_CNTRS.

We have a plan to remove 'RTE_ETHDEV_QUEUE_STAT_CNTRS', by moving queue stats to xstats.

But ixgbe limitation is there.

Quoted 9 lines
Quoted 5 lines
Quoted 2 lines

Perhaps there should be a generic SW emulation for this the mapping?

Ack, cc'ed Bruce.
But I am not sure ROI of the effort at this stage.

Not sure what the specific ask for me is here. :-) Overall, I think moving queue stats to xstats is the best way to go.

cc'ed because of "generic SW emulation" comment.

I was thinking if this mapping can be done transparent to the user by driver mapping queue <-> stats_register before reading stats, but @Bruce let me know this won't work because the stats tracking only happens after the mapping.

@Stephen, do you have something specific in your mind for SW emulation for mapping?

#7 lore raw

Per queue stats

...

Quoted 23 lines
Quoted 11 lines
Quoted 11 lines
Quoted 4 lines

The stats queue mapping was a feature that was hinted at being removed. It only exists because of HW limitations on Intel ixgbe NIC and SW limitations from RTE_ETHDEV_QUEUE_STAT_CNTRS.

We have a plan to remove 'RTE_ETHDEV_QUEUE_STAT_CNTRS', by moving queue stats to xstats.

But ixgbe limitation is there.

Quoted 9 lines
Quoted 5 lines
Quoted 2 lines

Perhaps there should be a generic SW emulation for this the mapping?

Ack, cc'ed Bruce.
But I am not sure ROI of the effort at this stage.

Not sure what the specific ask for me is here. :-) Overall, I think moving queue stats to xstats is the best way to go.

cc'ed because of "generic SW emulation" comment.

Quoted 7 lines

I was thinking if this mapping can be done transparent to the user by driver mapping queue <-> stats_register before reading stats, but @Bruce let me know this won't work because the stats tracking only happens after the mapping.

@Stephen, do you have something specific in your mind for SW emulation for mapping?

Probably not worth doing mapping tables for SW drivers.

My preference would be something like:
1. Introduce new API's rte_ethdev_get_queue_stats() and rte_ethdev_reset_queue_stats(). with callbacks in ethdev ops.
if driver does not support this, then do a shim in ethdev that uses old stats fields.
2. Mark the stats mapping API's as deprecated and always return error. Making special case for ixgbe at this point really is not worth maintaining.
3. After all drivers converted over, drop the fields in ethdev stats for queues. And fix the examples, etc. Yes this would be a big API change.

Would take a year to get done (ie 25.11).

#8 lore raw

Re: Per queue stats

Quoted 43 lines · Stephen Hemminger

...

Quoted 23 lines
Quoted 11 lines
Quoted 11 lines
Quoted 4 lines

The stats queue mapping was a feature that was hinted at being removed. It only exists because of HW limitations on Intel ixgbe NIC and SW limitations from RTE_ETHDEV_QUEUE_STAT_CNTRS.

We have a plan to remove 'RTE_ETHDEV_QUEUE_STAT_CNTRS', by moving queue stats to xstats.

But ixgbe limitation is there.

Quoted 9 lines
Quoted 5 lines
Quoted 2 lines

Perhaps there should be a generic SW emulation for this the mapping?

Ack, cc'ed Bruce.
But I am not sure ROI of the effort at this stage.

Not sure what the specific ask for me is here. :-) Overall, I think moving queue stats to xstats is the best way to go.

cc'ed because of "generic SW emulation" comment.

Quoted 7 lines

I was thinking if this mapping can be done transparent to the user by driver mapping queue <-> stats_register before reading stats, but @Bruce let me know this won't work because the stats tracking only happens after the mapping.

@Stephen, do you have something specific in your mind for SW emulation for mapping?

Probably not worth doing mapping tables for SW drivers.

My preference would be something like:
1. Introduce new API's rte_ethdev_get_queue_stats() and rte_ethdev_reset_queue_stats(). with callbacks in ethdev ops.
if driver does not support this, then do a shim in ethdev that uses old stats fields.
2. Mark the stats mapping API's as deprecated and always return error. Making special case for ixgbe at this point really is not worth maintaining.
3. After all drivers converted over, drop the fields in ethdev stats for queues. And fix the examples, etc. Yes this would be a big API change.

Quoted 2 lines

Would take a year to get done (ie 25.11).

The problem with existing a API, 'rte_eth_stats_get()', is it has queue stats fixed sized array which limits us also for some cases causes unnecessary memory consumption.

Our solution for now is move queue stats to xstats, as it is flexible this solves the problem, but as you said in other thread xstats is free form and may be hard for the application to use it.

One option is we can add some more data to the xstats, to help application to consume it easier, like number of queues just before the queue stats and define queue stats formatting syntax.

But if we will go with API option, instead of adding new queue stats ('rte_ethdev_get_queue_stats()'), we can update the existing 'rte_eth_stats_get()' to return dynamic number of queue stats.

And we need to find someone to own this task.

Above is for queue stats, it does not solve the mapping issue, as we need it for a few drivers and we already have the APIs, I am for just keeping them as it is, they are not causing any burden to us.

#9 lore raw

RE: Per queue stats

Quoted 72 lines

From: Ferruh Yigit [mailto:[email protected]]
Sent: Sunday, 6 October 2024 22.51

Quoted 2 lines · Stephen Hemminger

...

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

The stats queue mapping was a feature that was hinted at being

removed.

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

It only exists because of HW limitations on Intel ixgbe NIC and

SW

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 2 lines

limitations from RTE_ETHDEV_QUEUE_STAT_CNTRS.

We have a plan to remove 'RTE_ETHDEV_QUEUE_STAT_CNTRS', by moving

queue

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

stats to xstats.

But ixgbe limitation is there.

Quoted 1 line

Perhaps there should be a generic SW emulation for this the

mapping?

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 4 lines
Quoted 1 line

Ack, cc'ed Bruce.
But I am not sure ROI of the effort at this stage.

Not sure what the specific ask for me is here. :-) Overall, I think

moving

Quoted 6 lines
Quoted 6 lines
Quoted 2 lines

queue stats to xstats is the best way to go.

cc'ed because of "generic SW emulation" comment.

I was thinking if this mapping can be done transparent to the user

by

Quoted 1 line
Quoted 1 line

driver mapping queue <-> stats_register before reading stats, but

@Bruce

Quoted 4 lines
Quoted 4 lines

let me know this won't work because the stats tracking only happens after the mapping.

@Stephen, do you have something specific in your mind for SW

emulation

Quoted 6 lines
Quoted 1 line

for mapping?

Probably not worth doing mapping tables for SW drivers.

My preference would be something like:
1. Introduce new API's rte_ethdev_get_queue_stats() and

rte_ethdev_reset_queue_stats().

Quoted 2 lines

with callbacks in ethdev ops.
if driver does not support this, then do a shim in ethdev

that uses old stats fields.

Quoted 1 line

2. Mark the stats mapping API's as deprecated and always

return error.

Quoted 1 line

Making special case for ixgbe at this point really is not

worth maintaining.

Quoted 1 line

3. After all drivers converted over, drop the fields in

ethdev stats for queues.

Quoted 1 line

And fix the examples, etc. Yes this would be a big API

change.

Quoted 3 lines

Would take a year to get done (ie 25.11).

The problem with existing a API, 'rte_eth_stats_get()', is it has queue stats fixed sized array which limits us also for some cases causes unnecessary memory consumption.

Our solution for now is move queue stats to xstats, as it is flexible this solves the problem, but as you said in other thread xstats is free form and may be hard for the application to use it.

Quoted 7 lines

One option is we can add some more data to the xstats, to help application to consume it easier, like number of queues just before the queue stats and define queue stats formatting syntax.

But if we will go with API option, instead of adding new queue stats ('rte_ethdev_get_queue_stats()'), we can update the existing 'rte_eth_stats_get()' to return dynamic number of queue stats.

Some users might not care about queue stats, so let's remove them entirely from rte_eth_stats_get(), and put them elsewhere.

Quoted 7 lines

And we need to find someone to own this task.

Above is for queue stats, it does not solve the mapping issue, as we need it for a few drivers and we already have the APIs, I am for just keeping them as it is, they are not causing any burden to us.

#10 lore raw

Re: Per queue stats

Quoted 76 lines · Morten Brørup
Quoted 76 lines

From: Ferruh Yigit [mailto:[email protected]]
Sent: Sunday, 6 October 2024 22.51

Quoted 2 lines · Stephen Hemminger

...

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

The stats queue mapping was a feature that was hinted at being

removed.

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

It only exists because of HW limitations on Intel ixgbe NIC and

SW

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 2 lines

limitations from RTE_ETHDEV_QUEUE_STAT_CNTRS.

We have a plan to remove 'RTE_ETHDEV_QUEUE_STAT_CNTRS', by moving

queue

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

stats to xstats.

But ixgbe limitation is there.

Quoted 1 line

Perhaps there should be a generic SW emulation for this the

mapping?

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 4 lines
Quoted 1 line

Ack, cc'ed Bruce.
But I am not sure ROI of the effort at this stage.

Not sure what the specific ask for me is here. :-) Overall, I think

moving

Quoted 6 lines
Quoted 6 lines
Quoted 2 lines

queue stats to xstats is the best way to go.

cc'ed because of "generic SW emulation" comment.

I was thinking if this mapping can be done transparent to the user

by

Quoted 1 line
Quoted 1 line

driver mapping queue <-> stats_register before reading stats, but

@Bruce

Quoted 4 lines
Quoted 4 lines

let me know this won't work because the stats tracking only happens after the mapping.

@Stephen, do you have something specific in your mind for SW

emulation

Quoted 6 lines
Quoted 1 line

for mapping?

Probably not worth doing mapping tables for SW drivers.

My preference would be something like:
1. Introduce new API's rte_ethdev_get_queue_stats() and

rte_ethdev_reset_queue_stats().

Quoted 2 lines

with callbacks in ethdev ops.
if driver does not support this, then do a shim in ethdev

that uses old stats fields.

Quoted 1 line

2. Mark the stats mapping API's as deprecated and always

return error.

Quoted 1 line

Making special case for ixgbe at this point really is not

worth maintaining.

Quoted 1 line

3. After all drivers converted over, drop the fields in

ethdev stats for queues.

Quoted 1 line

And fix the examples, etc. Yes this would be a big API

change.

Quoted 3 lines

Would take a year to get done (ie 25.11).

The problem with existing a API, 'rte_eth_stats_get()', is it has queue stats fixed sized array which limits us also for some cases causes unnecessary memory consumption.

Our solution for now is move queue stats to xstats, as it is flexible this solves the problem, but as you said in other thread xstats is free form and may be hard for the application to use it.

One option is we can add some more data to the xstats, to help application to consume it easier, like number of queues just before the queue stats and define queue stats formatting syntax.

Quoted 6 lines
Quoted 3 lines

But if we will go with API option, instead of adding new queue stats ('rte_ethdev_get_queue_stats()'), we can update the existing 'rte_eth_stats_get()' to return dynamic number of queue stats.

Some users might not care about queue stats, so let's remove them entirely from rte_eth_stats_get(), and put them elsewhere.

+1 for this. I think separate APIs for queue stats would make more sense.

/Bruce

#11 lore raw
In reply to Ferruh Yigit
Quoted 8 lines · On Fri, 4 Oct 2024 02:48:21 +0100 Ferruh Yigit <[email protected]> wrote:
Quoted 2 lines

+Stats per queue = Y

This feature name is misleading,
it is for 'rte_eth_dev_set_[rt]x_queue_stats_mapping()' API, which is indeed for covering limitation for some drivers.
Tap does support getting stats per queue, but doesn't support above documented feature.

Just looked and the current state of the feature matrix is not correct for this part of the table. Although the feature intro says it is stats per queue mapping, that is not what other drivers are doing.

Only these drivers have queue_stats_mapping_set:
cnxk, enic, igc, ixgbe, txgbe, virtio

These drivers claim to have 'Stats per queue' but do not have stats mapping:
af_xdp, ark, atlantic, avp, bnxt, cxgbe, failsafe, fm10, hinic, hns3, ionic, mlx4, mlx5, netvsc, nfb, nfp, ngbe, qede, thunderx, vmxnet3

So the overwhelming practice is that the features.rst description is wrong!

#12 lore raw
Quoted 16 lines · Stephen Hemminger
Quoted 8 lines · On Fri, 4 Oct 2024 02:48:21 +0100 Ferruh Yigit <[email protected]> wrote:
Quoted 2 lines

+Stats per queue = Y

This feature name is misleading,
it is for 'rte_eth_dev_set_[rt]x_queue_stats_mapping()' API, which is indeed for covering limitation for some drivers.
Tap does support getting stats per queue, but doesn't support above documented feature.

Just looked and the current state of the feature matrix is not correct for this part of the table. Although the feature intro says it is stats per queue mapping, that is not what other drivers are doing.

Quoted 9 lines

Only these drivers have queue_stats_mapping_set:
cnxk, enic, igc, ixgbe, txgbe, virtio

These drivers claim to have 'Stats per queue' but do not have stats mapping:
af_xdp, ark, atlantic, avp, bnxt, cxgbe, failsafe, fm10, hinic, hns3, ionic, mlx4, mlx5, netvsc, nfb, nfp, ngbe, qede, thunderx, vmxnet3

So the overwhelming practice is that the features.rst description is wrong!

You are right.

What about update the 'features.rst', and extend the definition of 'Stats per queue' mapping + queue stats in basic stats or extended stats. This will be make name of the feature more intuitive.

#13 lore raw
Quoted 24 lines · On Sun, 6 Oct 2024 21:36:31 +0100 Ferruh Yigit <[email protected]> wrote:
Quoted 23 lines · Stephen Hemminger
Quoted 8 lines · On Fri, 4 Oct 2024 02:48:21 +0100 Ferruh Yigit <[email protected]> wrote:
Quoted 2 lines

+Stats per queue = Y

This feature name is misleading,
it is for 'rte_eth_dev_set_[rt]x_queue_stats_mapping()' API, which is indeed for covering limitation for some drivers.
Tap does support getting stats per queue, but doesn't support above documented feature.

Just looked and the current state of the feature matrix is not correct for this part of the table. Although the feature intro says it is stats per queue mapping, that is not what other drivers are doing.

Only these drivers have queue_stats_mapping_set:
cnxk, enic, igc, ixgbe, txgbe, virtio

These drivers claim to have 'Stats per queue' but do not have stats mapping:
af_xdp, ark, atlantic, avp, bnxt, cxgbe, failsafe, fm10, hinic, hns3, ionic, mlx4, mlx5, netvsc, nfb, nfp, ngbe, qede, thunderx, vmxnet3

Quoted 8 lines
Quoted 2 lines

So the overwhelming practice is that the features.rst description is wrong!

You are right.

What about update the 'features.rst', and extend the definition of 'Stats per queue' mapping + queue stats in basic stats or extended stats. This will be make name of the feature more intuitive.

Like this?

doc/guides/nics/features.rst +2 -3
index cd0115ffb3..a351d7b3b5 100644
@@ -727,8 +727,6 @@ Basic stats
 Support basic statistics such as: ipackets, opackets, ibytes, obytes,
 imissed, ierrors, oerrors, rx_nombuf.
 
-And per queue stats: q_ipackets, q_opackets, q_ibytes, q_obytes, q_errors.
-
 These apply to all drivers.
 
 * **[implements] eth_dev_ops**: ``stats_get``, ``stats_reset``.
@@ -754,7 +752,8 @@ Supports Extended Statistics, changes from driver to driver.
 Stats per queue
 ---------------
 
-Supports configuring per-queue stat counter mapping.
+Supports per queue stats: q_ipackets, q_opackets, q_ibytes, q_obytes, q_errors.
+May also support configuring per-queue stat counter mapping.
 
 * **[implements] eth_dev_ops**: ``queue_stats_mapping_set``.
 * **[related]    API**: ``rte_eth_dev_set_rx_queue_stats_mapping()``,
#14 lore raw
Quoted 53 lines · Stephen Hemminger
Quoted 24 lines · On Sun, 6 Oct 2024 21:36:31 +0100 Ferruh Yigit <[email protected]> wrote:
Quoted 23 lines · Stephen Hemminger
Quoted 8 lines · On Fri, 4 Oct 2024 02:48:21 +0100 Ferruh Yigit <[email protected]> wrote:
Quoted 2 lines

+Stats per queue = Y

This feature name is misleading,
it is for 'rte_eth_dev_set_[rt]x_queue_stats_mapping()' API, which is indeed for covering limitation for some drivers.
Tap does support getting stats per queue, but doesn't support above documented feature.

Just looked and the current state of the feature matrix is not correct for this part of the table. Although the feature intro says it is stats per queue mapping, that is not what other drivers are doing.

Only these drivers have queue_stats_mapping_set:
cnxk, enic, igc, ixgbe, txgbe, virtio

These drivers claim to have 'Stats per queue' but do not have stats mapping:
af_xdp, ark, atlantic, avp, bnxt, cxgbe, failsafe, fm10, hinic, hns3, ionic, mlx4, mlx5, netvsc, nfb, nfp, ngbe, qede, thunderx, vmxnet3

Quoted 8 lines
Quoted 2 lines

So the overwhelming practice is that the features.rst description is wrong!

You are right.

What about update the 'features.rst', and extend the definition of 'Stats per queue' mapping + queue stats in basic stats or extended stats. This will be make name of the feature more intuitive.

Like this?

doc/guides/nics/features.rst +0 -2
index cd0115ffb3..a351d7b3b5 100644
@@ -727,8 +727,6 @@ Basic stats
 Support basic statistics such as: ipackets, opackets, ibytes, obytes,
 imissed, ierrors, oerrors, rx_nombuf.
 
-And per queue stats: q_ipackets, q_opackets, q_ibytes, q_obytes, q_errors.
-
 These apply to all drivers.
 
 * **[implements] eth_dev_ops**: ``stats_get``, ``stats_reset``.

doc/guides/nics/features.rst Supports Extended Statistics, changes from driver to driver. · line 759

Quoted 9 lines
doc/guides/nics/features.rst +2 -1
@@ -754,7 +752,8 @@ Supports Extended Statistics, changes from driver to driver.
 Stats per queue
 ---------------
 
-Supports configuring per-queue stat counter mapping.
+Supports per queue stats: q_ipackets, q_opackets, q_ibytes, q_obytes, q_errors.
+May also support configuring per-queue stat counter mapping.
 
 * **[implements] eth_dev_ops**: ``queue_stats_mapping_set``.
 * **[related]    API**: ``rte_eth_dev_set_rx_queue_stats_mapping()``,

Yes, plus 'implements' & 'related' fields needs to be extended.