doc: update RSS action with best effort
PATCH 1/1 doc: update RSS action with best effort
Using the rte_flow action RSS types field,
may result in an undefined outcome.
For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.
Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode.
--- doc/guides/prog_guide/rte_flow.rst | 10 ++++++++++ 1 file changed, 10 insertions(+)
doc/guides/prog_guide/rte_flow.rst +10 -0
@@ -1735,6 +1735,16 @@ unspecified "best-effort" settings from the underlying PMD, which depending on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS. +Best effort will be used, in case there is a conflict inside the rule. +The conflict can be the result of: + +- Conflicting RSS types, for example setting both UDP and TCP. + +- Conflicting between RSS types and the requested pattern to match, + for example matching on UDP and hashing RSS on TCP. + +- Hashing on types that are not supported by the PMD. + Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Signature · 1 line
1.8.3.1
Quoted 21 lines · Ori Kam
Using the
rte_flowaction RSS types field,
may result in an undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode.
Signed-off-by Ori Kam <[email protected]>--- doc/guides/prog_guide/rte_flow.rst | 10 ++++++++++ 1 file changed, 10 insertions(+)doc/guides/prog_guide/rte_flow.rst +0
-0
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1744
Quoted 10 lines
doc/guides/prog_guide/rte_flow.rst +7
-0@@ -1735,6 +1735,16 @@ unspecified "best-effort" settings from the underlying PMD, which depending on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS. +Best effort will be used, in case there is a conflict inside the rule. +The conflict can be the result of: + +- Conflicting RSS types, for example setting both UDP and TCP. + +- Conflicting between RSS types and the requested pattern to match, + for example matching on UDP and hashing RSS on TCP.
IMHO it is not a problem at all, but good to clarify anyway.
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1746
Quoted 2 lines
doc/guides/prog_guide/rte_flow.rst +2
-0+ +- Hashing on types that are not supported by the PMD.
Shouldn't it return error to the caller?
Quoted 5 lines
doc/guides/prog_guide/rte_flow.rst +1
-0+ Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Hi Andrew,
Quoted 30 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 11 lines · Ori Kam
Using the
rte_flowaction RSS types field,
may result in an undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode.
Review
Quoted 6 lines
Signed-off-by Ori Kam <[email protected]>--- doc/guides/prog_guide/rte_flow.rst | 10 ++++++++++ 1 file changed, 10 insertions(+)(quoted diff) +0
-0
b/doc/guides/prog_guide/rte_flow.rstdoc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the
Quoted 4 lines
index 3e5cd1e..d000560 100644
doc/guides/prog_guide/rte_flow.rst +0
-0@@ -1735,6 +1735,16 @@ unspecified "best-effort" settings from theunderlying PMD, which depending
Quoted 3 lines
doc/guides/prog_guide/rte_flow.rst +0
-0on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS.
Quoted 10 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1744
Quoted 7 lines
doc/guides/prog_guide/rte_flow.rst +7
-0+Best effort will be used, in case there is a conflict inside the rule. +The conflict can be the result of: + +- Conflicting RSS types, for example setting both UDP and TCP. + +- Conflicting between RSS types and the requested pattern to match, + for example matching on UDP and hashing RSS on TCP.IMHO it is not a problem at all, but good to clarify anyway.
Agree this patch is just for clarification.
Quoted 5 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the · line 1746
Quoted 2 lines
doc/guides/prog_guide/rte_flow.rst +2
-0+ +- Hashing on types that are not supported by the PMD.Shouldn't it return error to the caller?
That’s depends, if for example the application requested eth and IPv4, and the PMD can do only IPv4 so according to this, the PMD will just do IPv4.
Quoted 5 lines
Quoted 5 lines
doc/guides/prog_guide/rte_flow.rst +1
-0+ Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Quoted 41 lines · Ori Kam
Hi Andrew,
Quoted 30 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 11 lines · Ori Kam
Using the
rte_flowaction RSS types field,
may result in an undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode.
Review
Quoted 6 lines
Signed-off-by Ori Kam <[email protected]>--- doc/guides/prog_guide/rte_flow.rst | 10 ++++++++++ 1 file changed, 10 insertions(+)(quoted diff) +0
-0
b/doc/guides/prog_guide/rte_flow.rstdoc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the
Quoted 4 lines
index 3e5cd1e..d000560 100644
doc/guides/prog_guide/rte_flow.rst +0
-0@@ -1735,6 +1735,16 @@ unspecified "best-effort" settings from theunderlying PMD, which depending
Quoted 3 lines
doc/guides/prog_guide/rte_flow.rst +0
-0on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS.Quoted 9 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1744
Quoted 7 lines
doc/guides/prog_guide/rte_flow.rst +7
-0+Best effort will be used, in case there is a conflict inside the rule. +The conflict can be the result of: + +- Conflicting RSS types, for example setting both UDP and TCP. + +- Conflicting between RSS types and the requested pattern to match, + for example matching on UDP and hashing RSS on TCP.IMHO it is not a problem at all, but good to clarify anyway.
Quoted 8 lines
Agree this patch is just for clarification.
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the · line 1746
Quoted 2 lines
doc/guides/prog_guide/rte_flow.rst +2
-0+ +- Hashing on types that are not supported by the PMD.Shouldn't it return error to the caller?
That’s depends, if for example the application requested eth and IPv4, and the PMD can do only IPv4 so according to this, the PMD will just do IPv4.
I think it is a bad behaviour. Application has required information to provide correct parameters and therefore incorrect should be rejected. Am I missing something?
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
doc/guides/prog_guide/rte_flow.rst +1
-0+ Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Quoted 51 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 45 lines · Ori Kam
Hi Andrew,
Quoted 30 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 11 lines · Ori Kam
Using the
rte_flowaction RSS types field,
may result in an undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode.
Review
Quoted 6 lines
Signed-off-by Ori Kam <[email protected]>--- doc/guides/prog_guide/rte_flow.rst | 10 ++++++++++ 1 file changed, 10 insertions(+)(quoted diff) +0
-0
b/doc/guides/prog_guide/rte_flow.rstdoc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the
Quoted 4 lines
index 3e5cd1e..d000560 100644
doc/guides/prog_guide/rte_flow.rst +0
-0@@ -1735,6 +1735,16 @@ unspecified "best-effort" settings from theunderlying PMD, which depending
Quoted 3 lines
doc/guides/prog_guide/rte_flow.rst +0
-0on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS.Quoted 9 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1744
Quoted 7 lines
doc/guides/prog_guide/rte_flow.rst +7
-0+Best effort will be used, in case there is a conflict inside the rule. +The conflict can be the result of: + +- Conflicting RSS types, for example setting both UDP and TCP. + +- Conflicting between RSS types and the requested pattern to match, + for example matching on UDP and hashing RSS on TCP.IMHO it is not a problem at all, but good to clarify anyway.
Agree this patch is just for clarification.
Quoted 2 lines
Quoted 2 lines
doc/guides/prog_guide/rte_flow.rst +2
-0+ +- Hashing on types that are not supported by the PMD.
Quoted 9 lines
Quoted 4 lines
Quoted 2 lines
Shouldn't it return error to the caller?
That’s depends, if for example the application requested eth and IPv4, and the PMD can do only IPv4 so according to this, the PMD will just do IPv4.
I think it is a bad behaviour. Application has required information to provide correct parameters and therefore incorrect should be rejected. Am I missing something?
In RTE flow you can't know what are the parameters that are supported.
One option is to fail the rule, but this means failing the rule even if some of them are supported. Or we can choose to fail if all the types are unsupported, but this will result in miss match between adding two types and adding just one type. For example the PMD doesn't support RSS on eth and the application request RSS on ETH and IPv4 so the PMD will do IPv4 RSS,
while if the user selects only ETH the rule will fail.
Same this is exactly the same case as miss match the application requests something that the PMD can't supply.
In any case I think this is the current behavior of the PMD so this just clarify it.
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
doc/guides/prog_guide/rte_flow.rst +1
-0+ Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Quoted 60 lines · Ori Kam
Quoted 51 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 45 lines · Ori Kam
Hi Andrew,
Quoted 30 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 11 lines · Ori Kam
Using the
rte_flowaction RSS types field,
may result in an undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode.
Review
Quoted 6 lines
Signed-off-by Ori Kam <[email protected]>--- doc/guides/prog_guide/rte_flow.rst | 10 ++++++++++ 1 file changed, 10 insertions(+)(quoted diff) +0
-0
b/doc/guides/prog_guide/rte_flow.rstdoc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the
Quoted 4 lines
index 3e5cd1e..d000560 100644
doc/guides/prog_guide/rte_flow.rst +0
-0@@ -1735,6 +1735,16 @@ unspecified "best-effort" settings from theunderlying PMD, which depending
Quoted 3 lines
doc/guides/prog_guide/rte_flow.rst +0
-0on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS.Quoted 9 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1744
Quoted 7 lines
doc/guides/prog_guide/rte_flow.rst +7
-0+Best effort will be used, in case there is a conflict inside the rule. +The conflict can be the result of: + +- Conflicting RSS types, for example setting both UDP and TCP. + +- Conflicting between RSS types and the requested pattern to match, + for example matching on UDP and hashing RSS on TCP.IMHO it is not a problem at all, but good to clarify anyway.
Agree this patch is just for clarification.
Quoted 2 lines
Quoted 2 lines
doc/guides/prog_guide/rte_flow.rst +2
-0+ +- Hashing on types that are not supported by the PMD.Quoted 9 lines
Quoted 4 lines
Quoted 2 lines
Shouldn't it return error to the caller?
That’s depends, if for example the application requested eth and IPv4, and the PMD can do only IPv4 so according to this, the PMD will just do IPv4.
I think it is a bad behaviour. Application has required information to provide correct parameters and therefore incorrect should be rejected. Am I missing something?
Quoted 9 lines
In RTE flow you can't know what are the parameters that are supported.
One option is to fail the rule, but this means failing the rule even if some of them are supported. Or we can choose to fail if all the types are unsupported, but this will result in miss match between adding two types and adding just one type. For example the PMD doesn't support RSS on eth and the application request RSS on ETH and IPv4 so the PMD will do IPv4 RSS,
while if the user selects only ETH the rule will fail.
Same this is exactly the same case as miss match the application requests something that the PMD can't supply.
In any case I think this is the current behavior of the PMD so this just clarify it.
I'm sorry, but the PMD is hardly applicable here. Do you mean the "mlx5" PMD? Or have you reviewed all PMDs?
No I understand that the topic is more complicated, but covering complex cases should not spoil simple.
I think that information provided in dev_info should specify all supported protocols/fields.
Yes, it is acceptable to behave best effort if requested field is supported in general, but it not for the flow. Please, make it a bit clearer in the suggested description.
Quoted 7 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
doc/guides/prog_guide/rte_flow.rst +1
-0+ Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Hi Andrew,
Quoted 76 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 28 lines · Ori Kam
Quoted 28 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 22 lines · Ori Kam
Hi Andrew,
Quoted 20 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 16 lines · Ori Kam
Using the
rte_flowaction RSS types field,
may result in an undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode.
Signed-off-by Ori Kam <[email protected]>--- doc/guides/prog_guide/rte_flow.rst | 10 ++++++++++ 1 file changed, 10 insertions(+)Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Review
Quoted 1 line
(quoted diff) +0
-0
b/doc/guides/prog_guide/rte_flow.rstdoc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the
Quoted 4 lines
index 3e5cd1e..d000560 100644
doc/guides/prog_guide/rte_flow.rst +0
-0@@ -1735,6 +1735,16 @@ unspecified "best-effort" settings from theunderlying PMD, which depending
Quoted 1 line
doc/guides/prog_guide/rte_flow.rst +0
-0on the flow rule, may result in anything ranging from empty (singlequeue)
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1744
Quoted 9 lines
doc/guides/prog_guide/rte_flow.rst +7
-0to all-inclusive RSS. +Best effort will be used, in case there is a conflict inside the rule. +The conflict can be the result of: + +- Conflicting RSS types, for example setting both UDP and TCP. + +- Conflicting between RSS types and the requested pattern to match, + for example matching on UDP and hashing RSS on TCP.IMHO it is not a problem at all, but good to clarify anyway.
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Agree this patch is just for clarification.
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the · line 1746
Quoted 2 lines
doc/guides/prog_guide/rte_flow.rst +2
-0+ +- Hashing on types that are not supported by the PMD.Shouldn't it return error to the caller?
That’s depends, if for example the application requested eth and IPv4, and the PMD can do only IPv4 so according to this, the PMD will just do
IPv4.
Quoted 7 lines
Quoted 5 lines
I think it is a bad behaviour. Application has required information to provide correct parameters and therefore incorrect should be rejected. Am I missing something?
In RTE flow you can't know what are the parameters that are supported. One option is to fail the rule, but this means failing the rule even if some of
them are supported.
Quoted 1 line
Or we can choose to fail if all the types are unsupported, but this will result in
miss match between
Quoted 1 line
adding two types and adding just one type. For example the PMD doesn't
support RSS on eth
Quoted 3 lines
and the application request RSS on ETH and IPv4 so the PMD will do IPv4 RSS, while if the user selects only ETH the rule will fail.
Same this is exactly the same case as miss match the application requests
Quoted 7 lines
something
Quoted 2 lines
that the PMD can't supply.
In any case I think this is the current behavior of the PMD so this just clarify it.I'm sorry, but the PMD is hardly applicable here. Do you mean the "mlx5" PMD? Or have you reviewed all PMDs?
I assume that each PMD as its own limitations, I'm sure that no PMD can support all possible RSS. But maybe there is such perfect PMD.
Quoted 8 lines
No I understand that the topic is more complicated, but covering complex cases should not spoil simple.
I think that information provided indev_infoshould specify all supported protocols/fields.
Yes, it is acceptable to behave best effort if requested field is supported in general, but it not for the flow. Please, make it a bit clearer in the suggested description.
So what about the following wording: (after the current part) If requested RSS hash type is not supported, and no supported hash function is requested then the PMD may reject the flow.
Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
doc/guides/prog_guide/rte_flow.rst +1
-0+ Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Quoted 7 lines · Ori Kam
Quoted 5 lines
Quoted 2 lines
+
+- Hashing on types that are not supported by the PMD.Shouldn't it return error to the caller?
That’s depends, if for example the application requested eth and IPv4, and the PMD can do only IPv4 so according to this, the PMD will just do IPv4.
We should have some validation in ethdev to catch this. Is it not the case?
Signature · 1 line
David Marchand
Hi David,
Quoted 7 lines
-----Original Message-----
From: David Marchand <[email protected]>Quoted 3 lines · Ori Kam
Quoted 3 lines
Quoted 2 lines
+
+- Hashing on types that are not supported by the PMD.
Quoted 8 lines
Quoted 4 lines
Quoted 2 lines
Shouldn't it return error to the caller?
That’s depends, if for example the application requested eth and IPv4, and the PMD can do only IPv4 so according to this, the PMD will just do IPv4.
We should have some validation in ethdev to catch this. Is it not the case?
Like I said in my reply to Andrew, in rte_flow we don't have such caps. Maybe we should think about adding them for the RSS case, but again it is tricky What if one RSS type is supported depending on the flow or other types?
Quoted 3 lines
--
David Marchand
Quoted 10 lines · Ori Kam
Hi David,
Quoted 8 lines
-----Original Message-----
From: David Marchand <[email protected]>Quoted 4 lines · Ori Kam
Quoted 4 lines
Quoted 2 lines
+
+- Hashing on types that are not supported by the PMD.Shouldn't it return error to the caller?
Quoted 8 lines
Quoted 5 lines
Quoted 2 lines
That’s depends, if for example the application requested eth and IPv4, and the PMD can do only IPv4 so according to this, the PMD will just do IPv4.
We should have some validation in ethdev to catch this. Is it not the case?
Like I said in my reply to Andrew, in
rte_flowwe don't have such caps. Maybe we should think about adding them for the RSS case, but again it is tricky What if one RSS type is supported depending on the flow or other types?
Also I'd like to add that ethdev layer is dummy for rte_flow API. It does not parse pattern/actions etc. May be should change it one day.
03/08/2020 17:55, Andrew Rybchenko:
Quoted 12 lines · Ori Kam
Quoted 11 lines
From: David Marchand <[email protected]>
Quoted 10 lines · Ori Kam
Quoted 6 lines
Quoted 4 lines
Quoted 2 lines
+
+- Hashing on types that are not supported by the PMD.Shouldn't it return error to the caller?
That’s depends, if for example the application requested eth and IPv4, and the PMD can do only IPv4 so according to this, the PMD will just do IPv4.
We should have some validation in ethdev to catch this. Is it not the case?
Quoted 6 lines
Quoted 3 lines
Like I said in my reply to Andrew, in
rte_flowwe don't have such caps. Maybe we should think about adding them for the RSS case, but again it is tricky What if one RSS type is supported depending on the flow or other types?Also I'd like to add that ethdev layer is dummy for
rte_flowAPI. It does not parse pattern/actions etc. May be should change it one day.
For now, what we can do is to have "best effort" (sic) checks. In the case we are 100% sure a rule cannot be fully supported, I think we should reject the rule.
If there are more complex cases to handle, we can accept the rule and support it with "best effort" handling.
The other appropriate action is to implement tests in DTS to check the behaviour of the PMDs, validating the compliance of the accepted rules with real flows.
Quoted 16 lines · Andrew Rybchenko
Quoted 15 lines · Ori Kam
Hi David,
Quoted 13 lines
-----Original Message-----
From: David Marchand <[email protected]>Quoted 6 lines · Ori Kam
Quoted 4 lines
Quoted 2 lines
+
+- Hashing on types that are not supported by the PMD.Shouldn't it return error to the caller?
That’s depends, if for example the application requested eth and IPv4, and the PMD can do only IPv4 so according to this, the PMD will just do IPv4.
We should have some validation in ethdev to catch this. Is it not the case?
Quoted 7 lines
Quoted 3 lines
Like I said in my reply to Andrew, in
rte_flowwe don't have such caps. Maybe we should think about adding them for the RSS case, but again it is tricky What if one RSS type is supported depending on the flow or other types?Also I'd like to add that ethdev layer is dummy for
rte_flowAPI. It does not parse pattern/actions etc. May be should change it one day.
As far as I remember not having capabilities was the design decision, PMD does the validation via 'rte_flow_validate()'.
Indeed it can be quite complex to have that kind of capability reporting, and without something like that there is not much to do in ethdev layer.
PATCH v2 1/1 doc: update RSS action with best effort
In reply to Ori KamUsing the rte_flow action RSS types field,
may result in undefined outcome.
For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.
Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode.
---
v2:
* Based on ML, update that using only unsupported hash type may cause the flow to be rejected by PMD.
--- doc/guides/prog_guide/rte_flow.rst | 11 +++++++++++ 1 file changed, 11 insertions(+)
doc/guides/prog_guide/rte_flow.rst +11 -0
@@ -1735,6 +1735,17 @@ unspecified "best-effort" settings from the underlying PMD, which depending on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS. +Best effort will be used, in case there is a conflict inside the rule. +The conflict can be the result of: + +- Conflicting RSS types, for example setting both UDP and TCP. + +- Conflicting between RSS types and the requested pattern to match, + for example matching on UDP and hashing RSS on TCP. + +If requested RSS hash type is not supported, +and no supported hash type is requested then the PMD may reject the flow. + Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Signature · 1 line
1.8.3.1
Hi All,
Can you please review and ack it? so it will be merged in 20.08 version.
Thanks,
Ori
Quoted 49 lines
-----Original Message-----
From: Ori Kam <[email protected]>Using the
rte_flowaction RSS types field,
may result in undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode.
Signed-off-by Ori Kam <[email protected]>---
v2:
* Based on ML, update that using only unsupported hash type may cause the flow to be rejected by PMD.--- doc/guides/prog_guide/rte_flow.rst | 11 +++++++++++ 1 file changed, 11 insertions(+)doc/guides/prog_guide/rte_flow.rst +0
-0@@ -1735,6 +1735,17 @@ unspecified "best-effort" settings from theunderlying PMD, which depending
on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS.doc/guides/prog_guide/rte_flow.rst +11
-1+Best effort will be used, in case there is a conflict inside the rule. +The conflict can be the result of: + +- Conflicting RSS types, for example setting both UDP and TCP. + +- Conflicting between RSS types and the requested pattern to match, + for example matching on UDP and hashing RSS on TCP. + +If requested RSS hash type is not supported, +and no supported hash type is requested then the PMD may reject the flow. + Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously. --1.8.3.1
Quoted 25 lines · Ori Kam
Using the
rte_flowaction RSS types field,
may result in undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode.
Signed-off-by Ori Kam <[email protected]>---
v2:
* Based on ML, update that using only unsupported hash type may cause the flow to be rejected by PMD.--- doc/guides/prog_guide/rte_flow.rst | 11 +++++++++++ 1 file changed, 11 insertions(+)doc/guides/prog_guide/rte_flow.rst +0
-0
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1741
Quoted 7 lines
doc/guides/prog_guide/rte_flow.rst +4
-0@@ -1735,6 +1735,17 @@ unspecified "best-effort" settings from the underlying PMD, which depending on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS. +Best effort will be used, in case there is a conflict inside the rule. +The conflict can be the result of: + +- Conflicting RSS types, for example setting both UDP and TCP.
Thinking more about it, I see no conflict at all. It is common to specify both TCP and UDP in hash function if user wants to take TCP ports for TCP packets and UDP ports for UDP into account.
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1744
Quoted 3 lines
doc/guides/prog_guide/rte_flow.rst +3
-0+ +- Conflicting between RSS types and the requested pattern to match, + for example matching on UDP and hashing RSS on TCP.
I'd not name it a conflict either. It is just common rule when applicable fields are used only.
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1748
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst +4
-0+ +If requested RSS hash type is not supported, +and no supported hash type is requested then the PMD may reject the flow. +
I disagree with such description. If requested RSS hash type is not supported (not present in dev_info.flow_type_rss_offloads), it must be rejected and error returned.
If requested RSS hash type is not supported for a packet matching the rule, but supported in general (present in dev_info.flow_type_rss_offloads), part of RSS hash type specification may be ignored and only applicable bits are used. If result is empty, PMD may reject the flow rule.
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst +0
-0Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Hi Andrew,
Quoted 34 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 16 lines · Ori Kam
Using the
rte_flowaction RSS types field,
may result in undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode.
Signed-off-by Ori Kam <[email protected]>---
v2:
* Based on ML, update that using only unsupported hash type may cause the flow to be rejected by PMD.Review
Quoted 5 lines
--- doc/guides/prog_guide/rte_flow.rst | 11 +++++++++++ 1 file changed, 11 insertions(+)(quoted diff) +0
-0
b/doc/guides/prog_guide/rte_flow.rstdoc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the
Quoted 4 lines
index 3e5cd1e..d730b66 100644
doc/guides/prog_guide/rte_flow.rst +0
-0@@ -1735,6 +1735,17 @@ unspecified "best-effort" settings from theunderlying PMD, which depending
Quoted 3 lines
doc/guides/prog_guide/rte_flow.rst +0
-0on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS.
Quoted 10 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1741
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst +4
-0+Best effort will be used, in case there is a conflict inside the rule. +The conflict can be the result of: + +- Conflicting RSS types, for example setting both UDP and TCP.Thinking more about it, I see no conflict at all. It is common to specify both TCP and UDP in hash function if user wants to take TCP ports for TCP packets and UDP ports for UDP into account.
I fully agree with you that it is common to use both UDP and TCP and expect it to work based on the traffic. this is the point of this patch. To clarify that this is valid input and the PMD will work in best effort mode.
Quoted 7 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1744
Quoted 3 lines
doc/guides/prog_guide/rte_flow.rst +3
-0+ +- Conflicting between RSS types and the requested pattern to match, + for example matching on UDP and hashing RSS on TCP.I'd not name it a conflict either. It is just common rule when applicable fields are used only.
Just like my comment from above.
Quoted 5 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1748
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst +4
-0+ +If requested RSS hash type is not supported, +and no supported hash type is requested then the PMD may reject the flow. +
Quoted 9 lines
I disagree with such description. If requested RSS hash type is not supported (not present in dev_info.flow_type_rss_offloads), it must be rejected and error returned.
If requested RSS hash type is not supported for a packet matching the rule, but supported in general (present in dev_info.flow_type_rss_offloads), part of RSS hash type specification may be ignored and only applicable bits are used. If result is empty, PMD may reject the flow rule.
The flow should be rejected even if it is used with some type that is supported?
Quoted 4 lines
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst +0
-0Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Quoted 46 lines · Ori Kam
Hi Andrew,
Quoted 34 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 16 lines · Ori Kam
Using the
rte_flowaction RSS types field,
may result in undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode.
Signed-off-by Ori Kam <[email protected]>---
v2:
* Based on ML, update that using only unsupported hash type may cause the flow to be rejected by PMD.Review
Quoted 5 lines
--- doc/guides/prog_guide/rte_flow.rst | 11 +++++++++++ 1 file changed, 11 insertions(+)(quoted diff) +0
-0
b/doc/guides/prog_guide/rte_flow.rstdoc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the
Quoted 4 lines
index 3e5cd1e..d730b66 100644
doc/guides/prog_guide/rte_flow.rst +0
-0@@ -1735,6 +1735,17 @@ unspecified "best-effort" settings from theunderlying PMD, which depending
Quoted 3 lines
doc/guides/prog_guide/rte_flow.rst +0
-0on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS.Quoted 10 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1741
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst +4
-0+Best effort will be used, in case there is a conflict inside the rule. +The conflict can be the result of: + +- Conflicting RSS types, for example setting both UDP and TCP.Thinking more about it, I see no conflict at all. It is common to specify both TCP and UDP in hash function if user wants to take TCP ports for TCP packets and UDP ports for UDP into account.
Quoted 3 lines
I fully agree with you that it is common to use both UDP and TCP and expect it to work based on the traffic. this is the point of this patch. To clarify that this is valid input and the PMD will work in best effort mode.
Just don't name it "Conflicting RSS types", may be "Non-applicable RSS types", but it collapses two bullets into one.
Quoted 24 lines
Quoted 8 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1744
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst +3
-0+ +- Conflicting between RSS types and the requested pattern to match, + for example matching on UDP and hashing RSS on TCP.I'd not name it a conflict either. It is just common rule when applicable fields are used only.
Just like my comment from above.
Quoted 5 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1748
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst +4
-0+ +If requested RSS hash type is not supported, +and no supported hash type is requested then the PMD may reject the flow. +Quoted 9 lines
I disagree with such description. If requested RSS hash type is not supported (not present in dev_info.flow_type_rss_offloads), it must be rejected and error returned.
If requested RSS hash type is not supported for a packet matching the rule, but supported in general (present in dev_info.flow_type_rss_offloads), part of RSS hash type specification may be ignored and only applicable bits are used. If result is empty, PMD may reject the flow rule.
Quoted 1 line
The flow should be rejected even if it is used with some type that is supported?
Yes, if user ask for something what is not supported at all, it is a problem and user should be notified. May be it is too restrictive, but I'd prefer this way.
User has all required information in dev_info and can remove unsupported hash types from request. IMHO, it should enforce user check the result to be not empty and handle it properly.
Quoted 7 lines
Quoted 6 lines
Quoted 6 lines
doc/guides/prog_guide/rte_flow.rst +0
-0Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Hi Andrew,
Quoted 49 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 46 lines · Ori Kam
Hi Andrew,
Quoted 44 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 16 lines · Ori Kam
Using the
rte_flowaction RSS types field,
may result in undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode.
Signed-off-by Ori Kam <[email protected]>---
v2:
* Based on ML, update that using only unsupported hash type may cause the flow to be rejected by PMD.Review
Quoted 5 lines
--- doc/guides/prog_guide/rte_flow.rst | 11 +++++++++++ 1 file changed, 11 insertions(+)(quoted diff) +0
-0
b/doc/guides/prog_guide/rte_flow.rstdoc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the
Quoted 4 lines
index 3e5cd1e..d730b66 100644
doc/guides/prog_guide/rte_flow.rst +0
-0@@ -1735,6 +1735,17 @@ unspecified "best-effort" settings from theunderlying PMD, which depending
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1741
Quoted 7 lines
doc/guides/prog_guide/rte_flow.rst +4
-0on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS. +Best effort will be used, in case there is a conflict inside the rule. +The conflict can be the result of: + +- Conflicting RSS types, for example setting both UDP and TCP.Thinking more about it, I see no conflict at all. It is common to specify both TCP and UDP in hash function if user wants to take TCP ports for TCP packets and UDP ports for UDP into account.
Quoted 7 lines
Quoted 3 lines
I fully agree with you that it is common to use both UDP and TCP and expect it to work based on the traffic. this is the point of this patch. To clarify that this is valid input and the PMD will work in best effort mode.
Just don't name it "Conflicting RSS types", may be "Non-applicable RSS types", but it collapses two bullets into one.
Will rephrase.
Quoted 29 lines
Quoted 6 lines
Quoted 6 lines
Quoted 5 lines
doc/guides/prog_guide/rte_flow.rst +3
-0+ +- Conflicting between RSS types and the requested pattern to match, + for example matching on UDP and hashing RSS on TCP.Quoted 8 lines
Quoted 3 lines
I'd not name it a conflict either. It is just common rule when applicable fields are used only.
Just like my comment from above.
Quoted 3 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending
Quoted 3 lines
doc/guides/prog_guide/rte_flow.rst +3
-0+ +If requested RSS hash type is not supported, +and no supported hash type is requested then the PMD may reject theflow.
Quoted 11 lines
Quoted 11 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the · line 1748
Quoted 1 line
doc/guides/prog_guide/rte_flow.rst +1
-0+I disagree with such description. If requested RSS hash type is not supported (not present in dev_info.flow_type_rss_offloads), it must be rejected and error returned.
If requested RSS hash type is not supported for a packet matching the rule, but supported in general (present in dev_info.flow_type_rss_offloads), part of RSS hash type specification may be ignored and only applicable bits are used. If result is empty, PMD may reject the flow rule.Quoted 1 line
The flow should be rejected even if it is used with some type that is
supported?
Quoted 8 lines
Yes, if user ask for something what is not supported at all, it is a problem and user should be notified. May be it is too restrictive, but I'd prefer this way.
User has all required information indev_infoand can remove unsupported hash types from request. IMHO, it should enforce user check the result to be not empty and handle it properly.
Will update the doc.
Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 6 lines
doc/guides/prog_guide/rte_flow.rst +0
-0Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Quoted 46 lines · Ori Kam
Hi Andrew,
Quoted 34 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 16 lines · Ori Kam
Using the
rte_flowaction RSS types field,
may result in undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode.
Signed-off-by Ori Kam <[email protected]>---
v2:
* Based on ML, update that using only unsupported hash type may cause the flow to be rejected by PMD.Review
Quoted 5 lines
--- doc/guides/prog_guide/rte_flow.rst | 11 +++++++++++ 1 file changed, 11 insertions(+)(quoted diff) +0
-0
b/doc/guides/prog_guide/rte_flow.rstdoc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the
Quoted 4 lines
index 3e5cd1e..d730b66 100644
doc/guides/prog_guide/rte_flow.rst +0
-0@@ -1735,6 +1735,17 @@ unspecified "best-effort" settings from theunderlying PMD, which depending
Quoted 3 lines
doc/guides/prog_guide/rte_flow.rst +0
-0on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS.Quoted 10 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1741
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst +4
-0+Best effort will be used, in case there is a conflict inside the rule. +The conflict can be the result of: + +- Conflicting RSS types, for example setting both UDP and TCP.Thinking more about it, I see no conflict at all. It is common to specify both TCP and UDP in hash function if user wants to take TCP ports for TCP packets and UDP ports for UDP into account.
Quoted 4 lines
I fully agree with you that it is common to use both UDP and TCP and expect it to work based on the traffic. this is the point of this patch. To clarify that this is valid input and the PMD will work in best effort mode.
I think confusion occurs when the the RSS hash function set part of flow rule rss action.
Than providing a UDP flow and setting UDP & TCP RSS hash types looks confusing/wrong.
Does mlx support selecting hash function per flow?
Can it be possible to set RSS functions separately from any specific flow?
Quoted 32 lines
Quoted 8 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1744
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst +3
-0+ +- Conflicting between RSS types and the requested pattern to match, + for example matching on UDP and hashing RSS on TCP.I'd not name it a conflict either. It is just common rule when applicable fields are used only.
Just like my comment from above.
Quoted 5 lines
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1748
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst +4
-0+ +If requested RSS hash type is not supported, +and no supported hash type is requested then the PMD may reject the flow. +Quoted 9 lines
I disagree with such description. If requested RSS hash type is not supported (not present in dev_info.flow_type_rss_offloads), it must be rejected and error returned.
If requested RSS hash type is not supported for a packet matching the rule, but supported in general (present in dev_info.flow_type_rss_offloads), part of RSS hash type specification may be ignored and only applicable bits are used. If result is empty, PMD may reject the flow rule.The flow should be rejected even if it is used with some type that is supported?
Quoted 4 lines
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst +0
-0Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
PATCH v3 1/1 doc: update RSS action with best effort
In reply to Ori KamUsing the rte_flow action RSS types field,
may result in undefined outcome.
For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.
Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode, and will fail in case the requested type is not supported.
---
v3:
* Address ML comments.
v2:
* Address ML comments.
--- doc/guides/prog_guide/rte_flow.rst | 12 ++++++++++++ 1 file changed, 12 insertions(+)
doc/guides/prog_guide/rte_flow.rst +12 -0
@@ -1735,6 +1735,18 @@ unspecified "best-effort" settings from the underlying PMD, which depending on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS. +In case there are Non-applicable RSS types in the rule, +Best effort will be used. +This may be the result of: + +- Setting both UDP and TCP on the same rule. + +- Setting RSS types that don't match the requested pattern, + for example, matching on UDP and hashing RSS on TCP. + +If requested RSS hash type is not supported in ``dev_info``. +The flow creation will fail. + Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Signature · 1 line
1.8.3.1
Quoted 33 lines · Ori Kam
Using the
rte_flowaction RSS types field,
may result in undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode, and will fail in case the requested type is not supported.
Signed-off-by Ori Kam <[email protected]>---
v3:
* Address ML comments.v2:
* Address ML comments.--- doc/guides/prog_guide/rte_flow.rst | 12 ++++++++++++ 1 file changed, 12 insertions(+)doc/guides/prog_guide/rte_flow.rst +2
-0@@ -1735,6 +1735,18 @@ unspecified "best-effort" settings from the underlying PMD, which depending on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS. +In case there are Non-applicable RSS types in the rule, +Best effort will be used.
doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1749
Quoted 10 lines
doc/guides/prog_guide/rte_flow.rst +10
-0+This may be the result of: + +- Setting both UDP and TCP on the same rule. + +- Setting RSS types that don't match the requested pattern, + for example, matching on UDP and hashing RSS on TCP. + +If requested RSS hash type is not supported in ``dev_info``. +The flow creation will fail. +
Consider:
If non-applicable for matching packets RSS types are requested, these RSS types are simply ignored. For example, it happens if:
- Hashing of both TCP and UDP ports is requested (only one can present in a packet).
- Requested RSS types contradict to flow rule pattern (e.g. pattern has UDP item, but RSS types contain TCP).
If requested RSS types are not supported by the Ethernet device at all (not reported in ``dev_info.flow_type_rss_offloads``), the flow creation will fail.
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst +0
-0Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Hi Andrew,
Quoted 57 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 17 lines · Ori Kam
Using the
rte_flowaction RSS types field,
may result in undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode, and will fail in case the requested type is not supported.
Signed-off-by Ori Kam <[email protected]>---
v3:
* Address ML comments.Review
Quoted 7 lines
v2:
* Address ML comments.--- doc/guides/prog_guide/rte_flow.rst | 12 ++++++++++++ 1 file changed, 12 insertions(+)(quoted diff) +0
-0
b/doc/guides/prog_guide/rte_flow.rstdoc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the
Quoted 4 lines
index 3e5cd1e..eada283 100644
doc/guides/prog_guide/rte_flow.rst +0
-0@@ -1735,6 +1735,18 @@ unspecified "best-effort" settings from theunderlying PMD, which depending
Quoted 5 lines
doc/guides/prog_guide/rte_flow.rst +2
-0on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS. +In case there are Non-applicable RSS types in the rule, +Best effort will be used.doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1749
Quoted 10 lines
doc/guides/prog_guide/rte_flow.rst +10
-0+This may be the result of: + +- Setting both UDP and TCP on the same rule. + +- Setting RSS types that don't match the requested pattern, + for example, matching on UDP and hashing RSS on TCP. + +If requested RSS hash type is not supported in ``dev_info``. +The flow creation will fail. +Consider:
If non-applicable for matching packets RSS types are requested, these RSS types are simply ignored. For example, it happens if:- Hashing of both TCP and UDP ports is requested (only one can present in a packet).
Quoted 7 lines
- Requested RSS types contradict to flow rule pattern (e.g. pattern has UDP item, but RSS types contain TCP).
If requested RSS types are not supported by the Ethernet device at all (not reported in ``dev_info.flow_type_rss_offloads``), the flow creation will fail.
Sure sounds good, but since this is just copying your words, I will add your Signed-off-by to this patch, is that O.K?
Quoted 4 lines
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst +0
-0Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Quoted 62 lines · Ori Kam
Hi Andrew,
Quoted 59 lines
-----Original Message-----
From: Andrew Rybchenko <[email protected]>Quoted 17 lines · Ori Kam
Using the
rte_flowaction RSS types field,
may result in undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode, and will fail in case the requested type is not supported.
Signed-off-by Ori Kam <[email protected]>---
v3:
* Address ML comments.Review
Quoted 7 lines
v2:
* Address ML comments.--- doc/guides/prog_guide/rte_flow.rst | 12 ++++++++++++ 1 file changed, 12 insertions(+)(quoted diff) +0
-0
b/doc/guides/prog_guide/rte_flow.rstdoc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the
Quoted 4 lines
index 3e5cd1e..eada283 100644
doc/guides/prog_guide/rte_flow.rst +0
-0@@ -1735,6 +1735,18 @@ unspecified "best-effort" settings from theunderlying PMD, which depending
Quoted 5 lines
doc/guides/prog_guide/rte_flow.rst +2
-0on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS. +In case there are Non-applicable RSS types in the rule, +Best effort will be used.doc/guides/prog_guide/rte_flow.rst unspecified "best-effort" settings from the underlying PMD, which depending · line 1749
Quoted 10 lines
doc/guides/prog_guide/rte_flow.rst +10
-0+This may be the result of: + +- Setting both UDP and TCP on the same rule. + +- Setting RSS types that don't match the requested pattern, + for example, matching on UDP and hashing RSS on TCP. + +If requested RSS hash type is not supported in ``dev_info``. +The flow creation will fail. +Consider:
If non-applicable for matching packets RSS types are requested, these RSS types are simply ignored. For example, it happens if:- Hashing of both TCP and UDP ports is requested (only one can present in a packet).
- Requested RSS types contradict to flow rule pattern (e.g. pattern has UDP item, but RSS types contain TCP).
Quoted 6 lines
Quoted 4 lines
If requested RSS types are not supported by the Ethernet device at all (not reported in ``dev_info.flow_type_rss_offloads``), the flow creation will fail.
Sure sounds good, but since this is just copying your words, I will add your Signed-off-by to this patch, is that O.K?
OK
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
doc/guides/prog_guide/rte_flow.rst +0
-0Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
PATCH v4 1/1 doc: update RSS action with best effort
In reply to Ori KamUsing the rte_flow action RSS types field,
may result in undefined outcome.
For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.
Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode, and will fail in case the requested type is not supported.
--- doc/guides/prog_guide/rte_flow.rst | 13 +++++++++++++ 1 file changed, 13 insertions(+)
doc/guides/prog_guide/rte_flow.rst +13 -0
@@ -1735,6 +1735,19 @@ unspecified "best-effort" settings from the underlying PMD, which depending on the flow rule, may result in anything ranging from empty (single queue) to all-inclusive RSS. +If non-applicable for matching packets RSS types are requested, +these RSS types are simply ignored. For example, it happens if: + +- Hashing of both TCP and UDP ports is requested + (only one can present in a packet). + +- Requested RSS types contradict to flow rule pattern + (e.g. pattern has UDP item, but RSS types contain TCP). + +If requested RSS hash types are not supported by the Ethernet device at all +(not reported in ``dev_info.flow_tpe_rss_offloads``), +the flow creation will fail. + Note: RSS hash result is stored in the ``hash.rss`` mbuf field which overlaps ``hash.fdir.lo``. Since `Action: MARK`_ sets the ``hash.fdir.hi`` field only, both can be requested simultaneously.
Signature · 1 line
1.8.3.1
Ferruh Yigit
AppliedQuoted 7 lines · Ori Kam
Using the
rte_flowaction RSS types field,
may result in undefined outcome.For example selecting both UDP and TCP,
selecting TCP RSS type but the pattern is targeting UDP traffic. another option is that the PMD doesn't support all requested types.
Quoted 7 lines
Until now, it wasn't clear what will happen in such cases. This commit clarify this issue by stating that the PMD will work in the best-effort mode, and will fail in case the requested type is not supported.
Signed-off-by Ori Kam <[email protected]>Signed-off-by Andrew Rybchenko <[email protected]>
Applied to dpdk-next-net/main, thanks.