On Fri, 2025-12-12 at 08:27 +0000, Khushit Shah wrote:
On 12 Dec 2025, at 1:49 PM, David Woodhouse dwmw2@infradead.org wrote:
On Fri, 2025-12-12 at 08:16 +0000, Khushit Shah wrote:
I thought the earlier discussion preferred kvm_lapic_ignore_suppress_eoi_broadcast(), but I’m not tied to it.
I think some of that earlier discussion was 'informed' by me typing code into my mailer, and managing to type an example whose name implied the exact opposite of what the code actually returned. :)
Got it :), How about kvm_lapic_respect_suppress_eoi_broadcast()? It avoids the double-negative and keeps the semantics clear.
Sure. That gives us fairly clear separate helpers (in a .h file) which control whether we *advertise* it, and whether we really *do* it.
With them giving consistent 'true' and 'false' for 'ENABLE' and 'DISABLE' modes respectively, and being *inconsistent* for the 'QUIRK' mode.
You could even use the same switch() structure for both, with only the default: case being different, to make it crystal clear what's going on.