Re: unchecked MSR access error: WRMSR to 0xd84 (tried to write 0x0000000000010003) at rIP: 0xffffffffa025a1b8 (snbep_uncore_msr_init_box+0x38/0x60 [intel_uncore])

[Borislav Petkov]

On Mon, Mar 04, 2024 at 02:22:50PM -0500, Liang, Kan wrote:
> The number of available CBOXs on a SNBEP machine is determined at boot
> time. It should not be larger than the maximum number of cores.
> The recent commit 89b0f15f408f ("x86/cpu/topology: Get rid of
> cpuinfo::x86_max_cores") change the boot_cpu_data.x86_max_cores to
> topology_num_cores_per_package().
> I guess the new function probably returns a different maximum number of
> cores on the machine. But I don't have a SNBEP on my hands. Could you
> please help to check whether a different maximum number of cores is
> returned?

Yeah, the topo rework looks at fault:

before:

online:              1
initial_apicid:      0
apicid:              0
pkg_id:              0
die_id:              0
cu_id:               255
core_id:             0
logical_pkg_id:      0
logical_die_id:      0
llc_id:              0
l2c_id:              0
max_cores:           4
max_die_per_pkg:     1
smp_num_siblings:    2

after:

online:              1
initial_apicid:      0
apicid:              0
pkg_id:              0
die_id:              0
cu_id:               255
core_id:             0
logical_pkg_id:      0
logical_die_id:      0
llc_id:              0
l2c_id:              0
amd_node_id:         0
amd_nodes_per_pkg:   0
num_threads:         32
num_cores:           16
max_dies_per_pkg:    1
max_threads_per_core:2

I'll let tglx poke at this.

Thx!

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette