Add mdio access type (clause 45/clause 22) attribute

[jiahua-wang]

Signed-off-by: Jiahua Wang jiahua.wang@broadcom.com

[kcudnik]

could there be other clauses more in the future ? maybe this should be enum?, i think enum would be better here, since bool value should be used for something that is enabled or not, and here you have explicit types 22 and 45 to set

[jiahua-wang]

MDIO was originally defined in Clause 22 of IEEE RFC802.3. Clause 45 of the 802.3ae specification provided the ability to access 65,536 registers in 32 different devices on 32 different ports. 802.3ae is 20 years old. There is no new MDIO access mode in the last 20 years. MDIO frame format starts with 32 bits of logical 1 preamble and 2 bits of start of frame. The start of frame is 01 for clause 22 and 00 for clause 45. If the first bit of the start frame has to be logical 0, then no new MDIO frame format will be possible.

[kcudnik]

i see, but it still would make sense to make it as enum, like for example you would have ipv4 and ipv6, it's better specify it as enum rather then bool, maybe lets ask community on this

[jiahua-wang]

Changed to attribute SAI_SWITCH_ATTR_MDIO_ACCESS_TYPE and enum sai_switch_mdio_clause_t

[lguohan]

@jiahua-wang , can you check why build is failing?

[prgeor]

@jiahua-wang , can you check why build is failing?

@jiahua-wang failure is due to your change, check the error log in build:-

2022-06-14T22:10:40.9552117Z ERROR: enum starts by object type SWITCH but not ending on MDIO_CLAUSE in SWITCH_MDIO_CLAUSE
2022-06-14T22:10:40.9552651Z ERROR: sai_switch_mdio_clause_t == SAI_SWITCH_ATTR_MDIO_ACCESS_TYPE not ending on enum name SWITCH_MDIO_CLAUSE
2022-06-14T22:10:40.9553117Z no need for END enum for SAI_IN_DROP_REASON_CUSTOM_RANGE_BASE
2022-06-14T22:10:40.9553526Z no need for END enum for SAI_OUT_DROP_REASON_CUSTOM_RANGE_BASE
2022-06-14T22:10:40.9553858Z non range base SAI_UDF_ATTR_BASE
2022-06-14T22:10:40.9554193Z make: *** [Makefile:114: saimetadata.h] Error 1
2022-06-14T22:10:40.9598190Z ##[error]Bash exited with code '2'.
2022-06-14T22:10:40.9681525Z ##[section]Finishing: Metadata check

[prgeor]

why can't the platform decide to choose between cl22 or cl45 for the mdio r/w? why should a higher SAI layer tell which one to choose? i am not able to understand the usecase. Can you add more to the PR description?

[kcudnik]

@jiahua-wang , can you check why build is failing?

@jiahua-wang failure is due to your change, check the error log in build:-

2022-06-14T22:10:40.9552117Z ERROR: enum starts by object type SWITCH but not ending on MDIO_CLAUSE in SWITCH_MDIO_CLAUSE
2022-06-14T22:10:40.9552651Z ERROR: sai_switch_mdio_clause_t == SAI_SWITCH_ATTR_MDIO_ACCESS_TYPE not ending on enum name SWITCH_MDIO_CLAUSE
2022-06-14T22:10:40.9553117Z no need for END enum for SAI_IN_DROP_REASON_CUSTOM_RANGE_BASE
2022-06-14T22:10:40.9553526Z no need for END enum for SAI_OUT_DROP_REASON_CUSTOM_RANGE_BASE
2022-06-14T22:10:40.9553858Z non range base SAI_UDF_ATTR_BASE
2022-06-14T22:10:40.9554193Z make: *** [Makefile:114: saimetadata.h] Error 1
2022-06-14T22:10:40.9598190Z ##[error]Bash exited with code '2'.
2022-06-14T22:10:40.9681525Z ##[section]Finishing: Metadata check

attr should be named SAI_SWITCH_ATTR_SWITCH_MDIO_CLAUSE or SAI_SWITCH_ATTR_MDIO_CLAUSE, 2nd preffered

[jiahua-wang]

With PR #1507 replacing this PR, I will close this PR.