Re: [PATCH v7 9/9] ACPI: APEI: EINJ: Update the documentation for EINJv2 support

[Luck, Tony]

On Sun, Jun 01, 2025 at 12:25:54PM +0200, Borislav Petkov wrote:
> Some questions inline...
> 
> On Sat, May 31, 2025 at 03:24:14PM -0700, Luck, Tony wrote:
> > EINJ V2 allows the user to perform multiple injections together.
> > 
> > The component_idN/component_syndromeN pairs of files direct the
> > "where" and the "what" of each injection.
> > 
> > But the kernel needs to know how many of these pairs to use
> > for an injection (to fill in a field in the structure passed
> > to the BIOS).
> 
> The kernel could realloc on each write. Or we could allocate the struct to max
> elems and trim it before passing it down to BIOS.

The actual structure passed to BIOS is the same each time. Just the
set_error_type_with_address::einjv2_struct::component_arr_count
changed to indicate how many errors to inject.  In theory the
driver could allocate and copy a correctly sized structure, but
Zaid's code here is simpler, an this is hardly a critical path.

> > User interface options:
> > 
> > 1) User can zero out the component_idN/component_syndromeN pairs
> > that they don't need and have the kernel count how many injections
> > are requested by looping to find the zero terminator.
> > 
> > 2) Kernel could zero all pairs after an injection to make the user
> > explicitly set the list of targets each time.
> > 
> > 3) User provides the count vis the nr_components file (perhaps
> > needs a better name?)
> 
> Yap, agree that the name is not optimal.

It can be dropped if we make the user zap previously supplied
component_idN/component_syndromeN pairs that are no longer
wanted.
> 
> User can inject into each component pairs file and the kernel can put that in
> the tracking struct. So you have:
> 
> # echo 4 > component_id0
> # echo A5A5A5A5 > component_syndrome0
> ... set other files and finish with usual
> # echo 1 > error_inject
> 
> <--- here, it goes through each component pair and builds the structure to
> pass down the BIOS.
> 
> And you track valid component pairs by setting the IDs to -1 or something else
> invalid.

This is just an improvement on my "option 1" (improved because all-ones
for the component ID is going to be invalid for sure, while all zeroes
could be a valid component).
> 
> All those component IDs which have remained invalid after the error_inject
> write happens, get ignored - you gather only those which are valid and inject.

Or just stop collecting on the first invalid one.

> And this way you can keep the old values too and gather them again and inject
> again, over and over again.
> 
> Right?

Yup.

-Tony