2023-07-11 10:06:30

by Kubalewski, Arkadiusz

[permalink] [raw]
Subject: [PATCH net-next] tools: ynl-gen: fix parse multi-attr enum attribute

When attribute is enum type and marked as multi-attr, the netlink respond
is not parsed, fails with stack trace:
File "/root/arek/linux-dpll/tools/net/ynl/lib/ynl.py", line 600, in dump
return self._op(method, vals, dump=True)
File "/root/arek/linux-dpll/tools/net/ynl/lib/ynl.py", line 586, in _op
rsp_msg = self._decode(gm.raw_attrs, op.attr_set.name)
File "/root/arek/linux-dpll/tools/net/ynl/lib/ynl.py", line 453, in _decode
self._decode_enum(rsp, attr_spec)
File "/root/arek/linux-dpll/tools/net/ynl/lib/ynl.py", line 410, in _decode_enum
value = enum.entries_by_val[raw - i].name
TypeError: unsupported operand type(s) for -: 'list' and 'int'
error: 1

Allow succesfull parse of multi-attr enums by decoding and assigning their
names into response in the _decode_enum(..) function.

Signed-off-by: Arkadiusz Kubalewski <[email protected]>
---
tools/net/ynl/lib/ynl.py | 9 ++++++++-
1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/tools/net/ynl/lib/ynl.py b/tools/net/ynl/lib/ynl.py
index 3b343d6cbbc0..553d82dd6382 100644
--- a/tools/net/ynl/lib/ynl.py
+++ b/tools/net/ynl/lib/ynl.py
@@ -407,7 +407,14 @@ class YnlFamily(SpecFamily):
raw >>= 1
i += 1
else:
- value = enum.entries_by_val[raw - i].name
+ if attr_spec.is_multi:
+ for index in range(len(raw)):
+ if (type(raw[index]) == int):
+ enum_name = enum.entries_by_val[raw[index] - i].name
+ rsp[attr_spec['name']][index] = enum_name
+ return
+ else:
+ value = enum.entries_by_val[raw - i].name
rsp[attr_spec['name']] = value

def _decode_binary(self, attr, attr_spec):
--
2.37.3



2023-07-12 04:45:04

by Jakub Kicinski

[permalink] [raw]
Subject: Re: [PATCH net-next] tools: ynl-gen: fix parse multi-attr enum attribute

On Tue, 11 Jul 2023 11:53:23 +0200 Arkadiusz Kubalewski wrote:
> diff --git a/tools/net/ynl/lib/ynl.py b/tools/net/ynl/lib/ynl.py
> index 3b343d6cbbc0..553d82dd6382 100644
> --- a/tools/net/ynl/lib/ynl.py
> +++ b/tools/net/ynl/lib/ynl.py
> @@ -407,7 +407,14 @@ class YnlFamily(SpecFamily):
> raw >>= 1
> i += 1
> else:
> - value = enum.entries_by_val[raw - i].name
> + if attr_spec.is_multi:
> + for index in range(len(raw)):
> + if (type(raw[index]) == int):
> + enum_name = enum.entries_by_val[raw[index] - i].name
> + rsp[attr_spec['name']][index] = enum_name
> + return
> + else:
> + value = enum.entries_by_val[raw - i].name
> rsp[attr_spec['name']] = value

Two asks:

First this function stupidly looks at value-start. Best I can tell this
is a leftover from when enum set was an array, but potentially "indexed
with an offset" (ie. if value start = 10, first elem would have value
11, second 12 etc.). When we added support for sparse enums this was
carried forward, but it's actually incorrect. entries_by_val is indexed
with the real value, we should not subtract the start-value. So please
send a patch to set i to 0 at the start and ignore start-value here
(or LMK if I should send one).

Second, instead of fixing the value up here, after already putting it
in the rsp - can we call this function to decode the enum before?
A bit hard to explain, let me show you the diff of what I have in mind
for the call site:

diff --git a/tools/net/ynl/lib/ynl.py b/tools/net/ynl/lib/ynl.py
index 1b3a36fbb1c3..e2e8a8c5fb6b 100644
--- a/tools/net/ynl/lib/ynl.py
+++ b/tools/net/ynl/lib/ynl.py
@@ -466,15 +466,15 @@ genl_family_name_to_id = None
else:
raise Exception(f'Unknown {attr_spec["type"]} with name {attr_spec["name"]}')

+ if 'enum' in attr_spec:
+ decoded = self._decode_enum(rsp, attr_spec)
+
if not attr_spec.is_multi:
rsp[attr_spec['name']] = decoded
elif attr_spec.name in rsp:
rsp[attr_spec.name].append(decoded)
else:
rsp[attr_spec.name] = [decoded]
-
- if 'enum' in attr_spec:
- self._decode_enum(rsp, attr_spec)
return rsp

def _decode_extack_path(self, attrs, attr_set, offset, target):

Then _decode_enum() only has to ever deal with single values,
and the caller will take care of mutli_attr like it would for any other
type?
--
pw-bot: cr

2023-07-12 10:36:07

by Kubalewski, Arkadiusz

[permalink] [raw]
Subject: RE: [PATCH net-next] tools: ynl-gen: fix parse multi-attr enum attribute

>From: Jakub Kicinski <[email protected]>
>Sent: Wednesday, July 12, 2023 6:00 AM
>
>On Tue, 11 Jul 2023 11:53:23 +0200 Arkadiusz Kubalewski wrote:
>> diff --git a/tools/net/ynl/lib/ynl.py b/tools/net/ynl/lib/ynl.py index
>> 3b343d6cbbc0..553d82dd6382 100644
>> --- a/tools/net/ynl/lib/ynl.py
>> +++ b/tools/net/ynl/lib/ynl.py
>> @@ -407,7 +407,14 @@ class YnlFamily(SpecFamily):
>> raw >>= 1
>> i += 1
>> else:
>> - value = enum.entries_by_val[raw - i].name
>> + if attr_spec.is_multi:
>> + for index in range(len(raw)):
>> + if (type(raw[index]) == int):
>> + enum_name = enum.entries_by_val[raw[index] -
>>i].name
>> + rsp[attr_spec['name']][index] = enum_name
>> + return
>> + else:
>> + value = enum.entries_by_val[raw - i].name
>> rsp[attr_spec['name']] = value
>
>Two asks:
>
>First this function stupidly looks at value-start. Best I can tell this is
>a leftover from when enum set was an array, but potentially "indexed with
>an offset" (ie. if value start = 10, first elem would have value 11, second
>12 etc.). When we added support for sparse enums this was carried forward,
>but it's actually incorrect. entries_by_val is indexed with the real value,
>we should not subtract the start-value. So please send a patch to set i to
>0 at the start and ignore start-value here (or LMK if I should send one).
>
>Second, instead of fixing the value up here, after already putting it in
>the rsp - can we call this function to decode the enum before?
>A bit hard to explain, let me show you the diff of what I have in mind for
>the call site:
>
>diff --git a/tools/net/ynl/lib/ynl.py b/tools/net/ynl/lib/ynl.py index
>1b3a36fbb1c3..e2e8a8c5fb6b 100644
>--- a/tools/net/ynl/lib/ynl.py
>+++ b/tools/net/ynl/lib/ynl.py
>@@ -466,15 +466,15 @@ genl_family_name_to_id = None
> else:
> raise Exception(f'Unknown {attr_spec["type"]} with name
>{attr_spec["name"]}')
>
>+ if 'enum' in attr_spec:
>+ decoded = self._decode_enum(rsp, attr_spec)
>+
> if not attr_spec.is_multi:
> rsp[attr_spec['name']] = decoded
> elif attr_spec.name in rsp:
> rsp[attr_spec.name].append(decoded)
> else:
> rsp[attr_spec.name] = [decoded]
>-
>- if 'enum' in attr_spec:
>- self._decode_enum(rsp, attr_spec)
> return rsp
>
> def _decode_extack_path(self, attrs, attr_set, offset, target):
>
>Then _decode_enum() only has to ever deal with single values, and the
>caller will take care of mutli_attr like it would for any other type?

Sure, I will try to implement your proposal and send update here.

Thank you!
Arkadiusz

>--
>pw-bot: cr

2023-07-12 17:04:23

by Jakub Kicinski

[permalink] [raw]
Subject: Re: [PATCH net-next] tools: ynl-gen: fix parse multi-attr enum attribute

On Wed, 12 Jul 2023 09:47:43 +0000 Kubalewski, Arkadiusz wrote:
> >+ if 'enum' in attr_spec:
> >+ decoded = self._decode_enum(rsp, attr_spec)

To be clear - this is just a quick mock up, you'll need to change
the arguments here, obviously. Probably to decoded and attr_spec?

2023-07-13 09:53:32

by Kubalewski, Arkadiusz

[permalink] [raw]
Subject: RE: [PATCH net-next] tools: ynl-gen: fix parse multi-attr enum attribute

>From: Jakub Kicinski <[email protected]>
>Sent: Wednesday, July 12, 2023 6:24 PM
>To: Kubalewski, Arkadiusz <[email protected]>
>Cc: [email protected]; [email protected];
>[email protected]; [email protected]; [email protected];
>[email protected]
>Subject: Re: [PATCH net-next] tools: ynl-gen: fix parse multi-attr enum
>attribute
>
>On Wed, 12 Jul 2023 09:47:43 +0000 Kubalewski, Arkadiusz wrote:
>> >+ if 'enum' in attr_spec:
>> >+ decoded = self._decode_enum(rsp, attr_spec)
>
>To be clear - this is just a quick mock up, you'll need to change
>the arguments here, obviously. Probably to decoded and attr_spec?

Well I did something that works for me:
("[PATCH net-next 0/2] tools: ynl-gen: fix parse multi-attr enum attribute")

But I am pretty sure it could break the other _decode_enum call
(from _decode_binary(..)), wasn't able to test it yet, as it seems to be used
only with ovs_flow.yaml spec (binary + struct type attr).

If you could take a look would be great.

Thank you!
Arkadiusz