RE: RFC Proposal to enable MSI support in Linux kernel

[Nakajima, Jun]

> -----Original Message-----
> From: Zwane Mwaikambo [mailto:[e-mail]]
> Sent: Thursday, May 29, 2003 2:55 PM
> To: Nguyen, Tom L
> Cc: Jeff Garzik; Nakajima, Jun;
> Subject: RE: RFC Proposal to enable MSI support in Linux kernel
> 
> On Thu, 29 May 2003, Nguyen, Tom L wrote:
> 
> > I am working on an updated patch to 2.5.70 to incorporate the comments
> > from you and Jeff Garzik. Regarding the variable renaming of irq, I am
> > open to any suggestions. Please make a recommendation and I will
> > incorporate it into the next update.
> 
> Well essentially, if the function is being passed a vector, lets call the
> parameter 'vector', right now you have to look at the code paths careful
> to determine wether it's an irq number or a vector masquerading as an irq
> number.
Agree.

> 
> > Regarding platform_legacy_irq, I have not seen an edge-triggered
> > interrupt failure caused by platform_legacy_irq.
> 
> So we can go with Jeff's suggestion there and remove the unecessary check
> for wether to assign level or edge?

Not so fast. This part is #ifdef CONFIG_ACPI_PCI and legacy IRQs can be overridden by Interrupt Source Overrides. Since they can change polarity (as far as the spec is concerned), I think we should fix the code. Hint: Look at the mp_override_legacy_irq().

Jun

> 
> 	Zwane
> --
> function.linuxpower.ca