Re: [GIT PULL] ACPI fix for v6.16-rc3

[Rafael J. Wysocki]

On Sat, Jun 21, 2025 at 5:35 PM Linus Torvalds
<torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
>
> On Sat, 21 Jun 2025 at 04:31, Rafael J. Wysocki <rafael@xxxxxxxxxx> wrote:
> >
> > This fixes a crash in ACPICA while attempting to evaluate a control
> > method that expects more arguments than are being passed to it, which
> > was exposed by a defective firmware update from a prominent OEM on
> > multiple systems.
>
> Christ. Reading the ACPI issues page makes me go "D'oh".
>
> Does anybody know what the heck Windows does in this situation? Does
> it just happen to work because it uses random arguments and happily
> dereferences bogus things without realizing, or does it do the "zero
> out missing arguments" thing?

Saket said: "I didn't run into this same bug on Windows though and the
interpreter just aborted out with a different error message saying
that this method already exists elsewhere. Maybe Windows thinks that
when RUCC is called with 2 args instead of 3, it is perhaps referring
to a different method with the same name, but warns that this name
already exists (AE_ALREADY_EXISTS)."

> Because clearly that firmware bug must have passed entirely unnoticed
> by people testing that thing on Windows...

Well, given the above, I'm wondering how it has been tested on
Windows.  It looks like somebody set a really low bar for the
verification of it.

Cheers, Rafael