Re: AMD SMU driver (amdsmu)

From: obiwac <obiwac_at_freebsd.org>
Date: Fri, 25 Jul 2025 12:33:10 UTC
This is the PCI device ID of the root complex (usually hostb0). You
can check this with pciconf -lv in "device", e.g.

hostb0@pci0:0:0:0:    class=0x060000 rev=0x00 hdr=0x00 vendor=0x1022
device=0x14e8 subvendor=0xf111 subdevice=0x0006
    vendor     = 'Advanced Micro Devices, Inc. [AMD]'
    device     = 'Phoenix Root Complex'
    class      = bridge
    subclass   = HOST-PCI

On Fri, 25 Jul 2025 at 12:37, <cyric@mm.st> wrote:
>
> obiwac wrote:
> > Hi!
> >
> > I have added an amdsmu driver for the AMD SMU chip (which is necessary
> > for S0i3 entry as well as diagnostics), and I'm looking for people to
> > review and test these changes out.
> >
> > The three patches I'm interested in at the moment are:
> >
> > - https://reviews.freebsd.org/D48683 amdsmu: Initial work on a driver
> > for the AMD SMU
> > - https://reviews.freebsd.org/D48714 amdsmu: Get and expose sysctls
> > for metrics about last sleep
> > - https://reviews.freebsd.org/D51487 amdsmu: Read and dump idlemask
> >
> > There is one other amdsmu-related patch, but it depends on ACPI
> > changes for s2idle that haven't been committed yet, so it's less of a
> > priority:
> >
> > - https://reviews.freebsd.org/D48721 amdsmu: Sleep entry/exit hints for PMFW
> >
> > This is currently only expected to work on the following CPUs:
> >
> >     #define CPUID_AMD_REMBRANDT    0x14B5
> >     #define CPUID_AMD_PHOENIX    0x14E8
> >     #define CPUID_AMD_STRIX_POINT    0x14A4
>
> How does this map to e.g. dmesg cpu information?
>
> CPU: AMD Ryzen 9 9900X 12-Core Processor             (4391.75-MHz
> K8-class CPU)
>   Origin="AuthenticAMD"  Id=0xb40f40  Family=0x1a  Model=0x44  Stepping=0
>
> > And of these, I have only had the opportunity to test on Phoenix
> > (specifically the 7040 series).
>