Re: [Freeipmi-users] problem with freeipmi 0.5.3 and intel SE7221BK1-E

[Bala.A]

Hi Al,

I think, there is no such device address 42 and 84.  I suspect a bug in
SSIF device driver code.

Regards,
Bala

>>
>> address@hidden:~# bmc-info --no-probing -D SSIF --driver-address=42
>> --driver-device=/dev/i2c-0 --register-spacing=1
>> ipmi-ssif-api-udm.c: 81: ipmi_ssif_cmd_udm: errnum (4): expression
>> failed
>> ipmi_cmd_get_device_id: permission denied
>>
>> address@hidden:~# bmc-info --no-probing -D SSIF --driver-address=84
>> --driver-device=/dev/i2c-0 --register-spacing=1
>> ipmi-ssif-api-udm.c: 81: ipmi_ssif_cmd_udm: errnum (4): expression
>> failed
>> ipmi_cmd_get_device_id: permission denied
>
> Bala, FYI line 81 is a call to 'ipmi_ssif_write'.
>
> errnum 4 in the ssif lib is IPMI_SSIF_CTX_ERR_PERMISSION.
>
> the errnum seems to be  set via this code chunk:
>
>   if (buf_len <= IPMI_I2C_SMBUS_BLOCK_MAX)
>     {
>       count = ipmi_ssif_single_part_write (ctx->device_fd,
>                                            buf,
>                                            buf_len);
>     }
>   else
>     {
>       count = ipmi_ssif_multi_part_write (ctx->device_fd,
>                                           buf,
>                                           buf_len);
>     }
>
>   if (count < 0)
>     {
>       if (errno == EACCES || errno == EPERM)
>         ctx->errnum = IPMI_SSIF_CTX_ERR_PERMISSION;
>       else
>         ERR_LOG(ctx->errnum = IPMI_SSIF_CTX_ERR_SYSTEM_ERROR);
>       goto cleanup;
>     }
>
> tracing this into the deeper write calls, the errno is coming from the
> core ioctl() call for ssif.
>
> It looks like Chris is root above, so EACCES is not the likely returned
> errno, EPERM = "Operation Not Permitted" is.  So the question is why is
> the ioctl not allowed.
>