Test Ipmi Payload

[justin0309]

openbmc code base: 2.14.0
auto-test : v4.0-stable

For the test items of Test Deactivate Payload,
I based on
openbmc/phosphor-net-ipmid@34a2c41
This commit fix, according to the description

//Deactivate SOL
~$ipmitool -I lanplus -H <BMC-IP> -U <UserID> -P <PWD> -C 17 raw 0x06
           0x49 0x01 0x01 0x00 0x00 0x00 0x00
Response: Unable to send RAW command (channel=0x0 netfn=0x6 lun=0x0
           cmd=0x49 rsp=0x80): Unknown (0x80)

But this behavior will still be considered as fail in auto-test, so should I modify the test process or the source code?

In addition,
Verify Set User Access Payload For Invalid User
'' does not contain 'rsp=0xcc'
Verify Get User Access Payload For Invalid User
' 02 00 00 00' does not contain 'rsp=0xcc'
Verify Get Channel Payload Support For Invalid Channel
'Unable to send RAW command (channel=0x0 netfn=0x6 lun=0x0 cmd=0x4e rsp=0xff): Unspecified error' does not contain 'rsp=0xcc'
I don’t know if these three response codes are code or auto-test that need to be modified.

[gkeishin]

@Ganesan149 @aravinth0510 @KothaiSrini Can your help respond to the above? or are you also seeing the same?

[Ganesan149]

@gkeishin
yes, seeing this issue.

[Gavin860606]

I've encountered the same failure based on openbmc v2.17.

sol session connection is handled by phosphor-net-ipmid, If we sends an ipmi payload actiavte command with instance id 1
(raw 0x06 0x48 0x01 0x01 0xc6 0x00 0x00 0x00)

The payload instance will treat as activated, but the session manager removed the session since it's a corrupted session.
However, the current design of phosphor-net-ipmid did not remove the sol payload instance as well.

The actual behavior of Test_Deactivate_Payload would looks like

1. raw 0x06 0x48 0x01 0x01 0xc6 0x00 0x00 0x00 (activate payload)
2. phosphor-net-ipmid/sol_manager handled the session and append the instance to the payloadMap 
3. After 1 second, phosphor-net-ipmid/session_manager removed the session since it's an idle session
4. raw 0x06 0x4a 0x01 (resp -> 01 01 00) the byte2 01 shows the payload instance 1 is still activated 
5. test suite continued to deactivate payload...

On the other hand, if you to use sol activate instance=1, you can see the payload was activated and the ssesion will not be removed since ipmitool/ipmi_sol.c wrapped the session connection behavior correctly...

My suggestion is to modify the ipmi payload validation logic and evaluate if we can use sol activate instance=1 sol deactivate instance=1 ipmi command instead.

[gkeishin]

@Ganesan149 thoughts on the above @Gavin860606 input ?