On Sun, Sep 28, 2025 at 10:37:47AM +0200, Jonathan Armani wrote:
> Hello Alex,
> 
> I still agree this is a really usefull feature :)
> 
> For PCI devices, maybe we could rely on a more generic approch with pcidevs instead of manually reverse with a switch ? 
> The only downside I see is that name will typically include "Audio" (or similar).
> 

Yes, in pcidevs, the " Audio" suffix is redundant and must be removed
as it makes the string longer than MAX_AUDIO_DEV_LEN. Using
pci_findproduct() or alike in cmpci(4) is more code than a quick
switch statement with 4 cases only.

> The USB with filtering and the audio part looks good and ok for me. 
> 
> While reviewing your diff I saw that some functions documented in
> audio.9 as declared as returning an int, but are actually unsigned
> int in the headers, is that expected ?

good catch! I'll fix these in another commit.