Re: Question on follow-up work for secondary interrupters

[Mathias Nyman <mathias.nyman@linux.intel.com>]

Hi

On 4/11/26 17:41, raoxu wrote:
> Hi Mathias,
> 
> I saw your recent xHCI refactoring series for usb-next.
> 
> In your earlier reply to my patch, you mentioned that after this
> refactoring it should be possible to use the secondary interrupter
> itself as the request_irq() data, instead of adding a separate
> array/context.
> 
> Would you say the code is now in a good state for me to continue the
> secondary interrupter work on top of it?

In general yes.
There are some small opens still regarding free_irq() that need attention.

The real benefit versus the regression risk of adding several xHC interrupts and
spreading them over several CPUs as the default xhci driver behavior is still open,
especially after the v11 patchseries [1] discussion.

1. https://lore.kernel.org/linux-usb/7F2D19446AC21BD0+20260202132936.2660641-1-raoxu@uniontech.com/

Can you share if there are some other specific use cases or other reason for this work?
With good justification we could add this support but limit it to those cases.

I'm myslef mostly interested in proper secondary interrupter support as enabling work
for xHCI VTIO and virtualization support

Thanks
Mathias