RE: [PATCH v2 5/7] EDAC/ghes: Prevent chipset-specific edac from loading after ghes_edac is registered

[Kani, Toshi]

On Friday, August 19, 2022 12:30 PM, Borislav Petkov wrote:
> > This flag is a static variable, say ghes_present, which is set to
> > false initially. ghes_probe() sets it to true. ghes_edac_preferred()
> > (aka. ghes_get_device) checks this flag at beginning and returns false
> > if this flag is false. It does not get unlatched since ACPI GHES table
> > is static.
> 
> What is that flag needed for at all?

Because ghes_get_device() always returns &ghes_devs on Arm,
which is !NULL.  It does not check if GHES is present.

> There are two possibilities:
> 
> 1. ghes_probe() succeeds. ghes_edac loads properly and other drivers use
> ghes_get_devices() to know when to load.
> 
> 2. ghes_probe() fails and that is caught during platform testing of all
> those platforms who wish to use ghes_edac. BIOS is fixed and goto 1.
> 
> No need for funky flags whatsoever.

3. ghes_probe() is not called, but ghes_get_device() is called from
other edac drivers.

Toshi