Re: [PATCH v2] genirq: move clear of kstat_irqs to free_desc()

[Thomas Gleixner]

On Tue, Jan 13 2026 at 10:21, Luigi Rizzo wrote:
> On Tue, Jan 13, 2026, 10:09 Thomas Gleixner <tglx@xxxxxxxxxx> wrote:
>> Right, though it really does not matter unless you deeply care about:
>>
>> alpha, arc, hexagon, m68k, mips, parisc, sparc, xtensa and random old
>> ARM32 platforms :)
>>
>
> Purely academic question: I suppose the lock is just to make sure that
> there isn't anyone currently using the descriptor before marking it
> disabled; but is it really needed, or by the time the function is called it
> is already known that it is not in use anymore?

It's still accessible through the array and while it's probably
irrelevant, I have no intention to audit all the legazy architecture
code for random abuse. At least the removed itanium code did nasty
things with them behind the core.

Thanks,

        tglx