Skip to contentSkip to conversation

Sending I2C command via google-ipmi i2c

DISCUSSION George Hung (洪忠敬) 3 participants 4 messages Apr 12, 2019 – Apr 16, 2019

lore
Share & tools
lore
#0 lore raw

Hi all,

I’m trying to sending ipmi I2C command via google-ipmi-i2c extension, but I will get the response error, “Unspecified error” from ipmitool.

There are the command and response example below:

# ipmitool raw 0x2e 0x02 0x79 0x2b 0x00 1 0 0x98 0 1 0 0x99 0 2
Unable to send RAW command (channel=0x0 netfn=0x2e lun=0x0 cmd=0x2 rsp=0xff): Unspecified error

I check the error message from journal and it indicates that cannot open /dev/i2c-121 and I think it used the first byte of IANA number (0x79) as bus number.

Has anyone ever met this situation ? or something I have to configure for phosphor-ipmi-host / google-ipmi-i2c ?

Best Regards
George Hung

#1 lore raw
Quoted 15 lines · George Hung (洪忠敬)

Hi all,

I’m trying to sending ipmi I2C command via google-ipmi-i2c extension,

but I will get the response error, “Unspecified error” from ipmitool.

There are the command and response example below:

Quoted 7 lines
# ipmitool raw 0x2e 0x02 0x79 0x2b 0x00 1 0 0x98 0 1 0 0x99 0 2

Unable to send RAW command (channel=0x0 netfn=0x2e lun=0x0 cmd=0x2 rsp=0xff): Unspecified error

I check the error message from journal and it indicates that cannot open /dev/i2c-121 and I think it used the first byte of IANA number (0x79) as bus number.

I have a little breakout I use:
ipmitool -I dbus raw 0x2e 0x05 0xcf 0xc2 0x00 0x00 0x02 0x00 0x00 0x00

                                 |--|    |--|    |------------|
|----------------------|
                                 |       |        |              |   reqBuf
                                 |       |        | oem group
                                 |       | command
                                 | netfn for oem

So 0x2e is the netfn, properly parsed.
So 0x02 is the command, properly parsed.
So 0x79 0x2b 0x00 is the OEM group
So 0x00 1 0 0x98 0 1 0 0x99 0 2 is the command payload

As you say, it sounds like there's a bug in parsing the length of the OEM group number. We haven't rebased onto the newest ipmid downstream, so we haven't tested it ourselves. But this sounds like a bug in the new ipmid.

IIRC, they weren't properly handling OEM groups and recently fixed it. Please specify the SHA1 of the phosphor-host-ipmid source you're using to build this.

Quoted 12 lines

Has anyone ever met this situation ? or something I have to configure for phosphor-ipmi-host / google-ipmi-i2c ?

Best Regards

George Hung

#2 lore raw
Quoted 53 lines

-----Original Message-----
From: Patrick Venture [mailto:[email protected]]
Sent: Friday, April 12, 2019 10:23 PM
To: George Hung (洪忠敬)
Cc: OpenBMC Maillist ([email protected]); Patrick Venture ([email protected]); [email protected]; [email protected]; [email protected]; [email protected]
Subject: Re: Sending I2C command via google-ipmi i2c

Quoted 15 lines · George Hung (洪忠敬)

Hi all,

I’m trying to sending ipmi I2C command via google-ipmi-i2c extension,

but I will get the response error, “Unspecified error” from ipmitool.

There are the command and response example below:

Quoted 8 lines
# ipmitool raw 0x2e 0x02 0x79 0x2b 0x00 1 0 0x98 0 1 0 0x99 0 2

Unable to send RAW command (channel=0x0 netfn=0x2e lun=0x0 cmd=0x2 rsp=0xff): Unspecified error

I check the error message from journal and it indicates that cannot open

/dev/i2c-121 and I think it used the first byte of IANA number (0x79) as bus number.

I have a little breakout I use:
ipmitool -I dbus raw 0x2e 0x05 0xcf 0xc2 0x00 0x00 0x02 0x00 0x00 0x00

                                 |--|    |--|    |------------|
|----------------------|
                                 |       |        |
|   reqBuf
                                 |       |        | oem group
                                 |       | command
                                 | netfn for oem

So 0x2e is the netfn, properly parsed.
So 0x02 is the command, properly parsed.
So 0x79 0x2b 0x00 is the OEM group
So 0x00 1 0 0x98 0 1 0 0x99 0 2 is the command payload

Quoted 7 lines

As you say, it sounds like there's a bug in parsing the length of the OEM group number. We haven't rebased onto the newest ipmid downstream, so we haven't tested it ourselves. But this sounds like a bug in the new ipmid.

IIRC, they weren't properly handling OEM groups and recently fixed it. Please specify the SHA1 of the phosphor-host-ipmid source you're using to build this.

SHA1 Revision: c46f6cd7dc47ccafcd2e86e160a32fe1be404e7e

Quoted 4 lines
Quoted 4 lines
Quoted 10 lines
Quoted 1 line

Has anyone ever met this situation ? or something I have to configure for

phosphor-ipmi-host / google-ipmi-i2c ?

Quoted 8 lines

Best Regards

George Hung

BRs

George Hung

#3 lore raw

OEM commands currently don't work in host-ipmid. I'm testing a bunch of random fixups to ipmid to get this kind of stuff working again
https://gerrit.openbmc-project.xyz/c/openbmc/phosphor-host-ipmid/+/20622

Quoted 83 lines · George Hung (洪忠敬)
Quoted 54 lines

-----Original Message-----
From: Patrick Venture [mailto:[email protected]]
Sent: Friday, April 12, 2019 10:23 PM
To: George Hung (洪忠敬)
Cc: OpenBMC Maillist ([email protected]); Patrick Venture ([email protected]); [email protected]; [email protected]; [email protected]; [email protected]
Subject: Re: Sending I2C command via google-ipmi i2c

Quoted 15 lines · George Hung (洪忠敬)

Hi all,

I’m trying to sending ipmi I2C command via google-ipmi-i2c extension,

but I will get the response error, “Unspecified error” from ipmitool.

There are the command and response example below:

Quoted 8 lines
# ipmitool raw 0x2e 0x02 0x79 0x2b 0x00 1 0 0x98 0 1 0 0x99 0 2

Unable to send RAW command (channel=0x0 netfn=0x2e lun=0x0 cmd=0x2 rsp=0xff): Unspecified error

I check the error message from journal and it indicates that cannot open

/dev/i2c-121 and I think it used the first byte of IANA number (0x79) as bus number.

I have a little breakout I use:
ipmitool -I dbus raw 0x2e 0x05 0xcf 0xc2 0x00 0x00 0x02 0x00 0x00 0x00

                                 |--|    |--|    |------------|
|----------------------|
                                 |       |        |
|   reqBuf
                                 |       |        | oem group
                                 |       | command
                                 | netfn for oem

So 0x2e is the netfn, properly parsed.
So 0x02 is the command, properly parsed.
So 0x79 0x2b 0x00 is the OEM group
So 0x00 1 0 0x98 0 1 0 0x99 0 2 is the command payload

Quoted 7 lines

As you say, it sounds like there's a bug in parsing the length of the OEM group number. We haven't rebased onto the newest ipmid downstream, so we haven't tested it ourselves. But this sounds like a bug in the new ipmid.

IIRC, they weren't properly handling OEM groups and recently fixed it. Please specify the SHA1 of the phosphor-host-ipmid source you're using to build this.

SHA1 Revision: c46f6cd7dc47ccafcd2e86e160a32fe1be404e7e

Quoted 4 lines
Quoted 4 lines
Quoted 10 lines
Quoted 1 line

Has anyone ever met this situation ? or something I have to configure for

phosphor-ipmi-host / google-ipmi-i2c ?

Quoted 8 lines

Best Regards

George Hung

BRs

George Hung