On 2/20/20 6:51 AM, Corey Minyard wrote:
> On Wed, Feb 19, 2020 at 04:09:55PM -0600, Stuart Hayes wrote:
>> Hello!
>>
>> I have a question.
>>
>> Right now, it appears that the acpi_ipmi module in the linux kernel
>> (allowing access to the IPMI operation region in ACPI) does not get loaded
>> automatically, which can result in undesired results or kernel log messages
>> that don't make it obvious to most users what the issue even is.
>>
>> Would adding "MODULE_SOFTDEP("post: acpi_ipmi");" to ipmi_msghandler.c be an
>> acceptable solution?
>
> I'm not sure, that doesn't seem optimal for most systems. What aboud
> adding a request_module() call to acpi_ipmi_probe() in
> drivers/char/ipmi/ipmi_si_platform.c? That way you only request the
> module if there is an ACPI IPMI device.
>
Thanks... I agree, that seems better, and it seems to work.
Would you consider a patch to that effect?
>>
>> Right now ipmi_msghandler already has softdep for ipmi_devintf, and adding
>> it for acpi_ipmi seems to fix the issues I'm seeing on a system that needs
>> it (the system needs it for both the power_meter device, as well as for an
>> ACPI _DSM method that controls PCIe SSD LEDs (this is a very new PCI ECN)).
>>
>> I'd be happy to submit a kernel patch to do this if there's no objection (or
>> better solution that I've missed).
>
> That would be great if you could try what I suggested.
>
> Thanks,
>
> -corey
>
>>
>> Thanks!
>>
_______________________________________________
Openipmi-developer mailing list
Openipmi-developer@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/openipmi-developer
