Re: [PATCH v10 1/3] ipmi: ssif_bmc: Add SSIF BMC driver

[Quan Nguyen <quan@os.amperecomputing.com>]

On 14/10/2022 20:05, Graeme Gregory wrote:
> On Mon, Oct 10, 2022 at 12:08:24PM +0100, Graeme Gregory wrote:
>>
>> On 10/10/2022 02:28, Quan Nguyen wrote:
>>>
>>>
>>> On 07/10/2022 20:26, Graeme Gregory wrote:
>>>>
>>>> On 04/10/2022 10:31, Quan Nguyen wrote:
>>>>> The SMBus system interface (SSIF) IPMI BMC driver can be used to
>>>>> perform
>>>>> in-band IPMI communication with their host in management (BMC) side.
>>>>>
>>>>> Thanks Dan for the copy_from_user() fix in the link below.
>>>>>
>>>>> Link: https://lore.kernel.org/linux-arm-kernel/20220310114119.13736-4-quan@os.amperecomputing.com/
>>>>> Signed-off-by: Quan Nguyen <quan@os.amperecomputing.com>
>>>>
>>>> I have been testing this on our hardware and I don't seem to be able
>>>> to get this driver working. I was using an older version.
>>>>
>>>> I have hacked ssifbridged to deal with the change in len from u8 to
>>>> unsigned int.
>>>>
>>>> It works as long as I only ever send SSIF commands, any attempt to
>>>> read a response crashes the state machine and the driver never
>>>> recovers. No further SSIF comms is possible! (slave doesnt even ACK
>>>> writes).
>>>>
>>>> A couple of comments below on possible state machine errors.
>>>>
>>>> Its possible I am doing something wrong!
>>>>
>>>
>>> Thanks Graeme for the test and the comments.
>>>
>>> What's your testing hardware?
>>>
>>> This was tested with Aspeed ast2500 and ast2600 with the patch series
>>> [1] below applied.
>>>
>>> If you use the same hw, could you pick the series and see if any thing
>>> improve ?
>>>
>>> [1] https://lore.kernel.org/all/20210616031046.2317-1-quan@os.amperecomputing.com/
>>>
>> Thanks, that patch series does stop the state machine crashing.
>>
>> I am testing on AST2600EVB with A1 rev, but we also have our own DC-SCM with
>> A3 chip.
>>
>> Responses are still not working for me, but I think that may be an error in
>> my ssifbridged hacks.
>>
>> Oct 10 10:54:55 qcom-evb-proto-ccf37d18ea0c ssifbridged[335]: Read ssif
>> request message with len=13 netfn=44 lun=0 cmd=2
>>
>> Oct 10 10:54:55 qcom-evb-proto-ccf37d18ea0c ipmid[312]: BootCode:
>> 000000000000000000
>>
>> Oct 10 10:54:55 qcom-evb-proto-ccf37d18ea0c ssifbridged[335]: Send ssif
>> respond message with len=4 netfn=45 lun=0 cmd=2 cc=0
>>
>> Oct 10 10:54:55 qcom-evb-proto-ccf37d18ea0c kernel: ipmi-ssif-host 0-0010:
>> Warn: on_read_requested_event unexpected READ REQUESTED in state=SSIF_READY
>>
>> Oct 10 10:54:55 qcom-evb-proto-ccf37d18ea0c kernel: ipmi-ssif-host 0-0010:
>> Warn: on_stop_event unexpected SLAVE STOP in state=SSIF_ABORTING
>>
> Just to close the loop on this, I have now fully tested this driver on
> our setup internally.
> 
> Final hitch turned out to be an aardvark not issuing STOP events between
> reads/writes.
> 
> Thanks for the work.
> 

Thanks a lot Graeme, for the test.

- Quan