SystemIO conflicts are not firmware bugs

Posted by haeseong 2 days ago

Counter28Comment6OpenOriginal

Comments

Comment by dataflow 3 hours ago

Two questions:

1. How does the kernel detect the conflict?

> you can certainly boot with acpi=off if you want to and no ACPI code will be executed at runtime

2. How does the kernel prevent the firmware from running anything?

Comment by mitxela 2 hours ago

ACPI firmware is not code that runs by itself. It is code that the kernel interprets. If the kernel doesn't interpret it, it won't run. SMM firmware runs by itself but also can't interact with ACPI locks anyway.

Comment by p_l 1 hour ago

For (1) - ACPI is how the kernel knows what addresses maps where. And when a driver claims the same address space, it knows there's a conflict

Comment by eqvinox 6 hours ago

> the internet will tell you that PC firmware developers are incompetent³

> ³ I’m not going to present an opinion on that here, merely say that this provides no supporting evidence for that assertion

If anything, isn't this evidence the vendor should've paid someone to write an ACPI driver for Linux? :-)

Comment by lmz 5 hours ago

No. Unless the hardware had Linux stickers on it.

Comment by mitxela 2 hours ago

Why should they? Would they sell more hardware if they did? And why Linux? Shouldn't they write an ACPI driver for Haiku first? Companies do what makes money, period. Activist companies have been almost completely wiped out by this stage of capitalism.