Re: Add support for checking before-the-fact whether an IRQ is

Re: Add support for checking before-the-fact whether an IRQ is

[Arjan van de Ven]

[-- Attachment #1: Type: text/plain, Size: 616 bytes --]

On Mon, 2003-12-29 at 19:18, Linux Kernel Mailing List wrote:
> ChangeSet 1.1552, 2003/12/29 10:18:11-08:00, torvalds@home.osdl.org
> 
> 	Add support for checking before-the-fact whether an IRQ is
> 	already registered or not. The x86 PCI layer wants this for
> 	its availability testing.
> 	
> 	Doing a request_irq()/free_irq() pair to check this condition
> 	like we used to do can lock the machine if the irq happens to
> 	be screaming.


question; which lock prevents someone else claiming the irq and making
it unsharable/unclaimable between can_request_irq() and the eventual
request_irq() ????

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: Add support for checking before-the-fact whether an IRQ is

[Linus Torvalds]

On Mon, 29 Dec 2003, Arjan van de Ven wrote:
> On Mon, 2003-12-29 at 19:18, Linux Kernel Mailing List wrote:
> > ChangeSet 1.1552, 2003/12/29 10:18:11-08:00, torvalds@home.osdl.org
> > 
> > 	Add support for checking before-the-fact whether an IRQ is
> > 	already registered or not. The x86 PCI layer wants this for
> > 	its availability testing.
> > 	
> > 	Doing a request_irq()/free_irq() pair to check this condition
> > 	like we used to do can lock the machine if the irq happens to
> > 	be screaming.
> 
> 
> question; which lock prevents someone else claiming the irq and making
> it unsharable/unclaimable between can_request_irq() and the eventual
> request_irq() ????

Nothing. It never did. This is basically a heuristic: "find the irq that 
looks the least used".

I just fixed the heuristic so that it shouldn't cause problems if the 
system comes up with the BIOS having done something silly like leaving an 
irq pending..

		Linus

Re: Add support for checking before-the-fact whether an IRQ is

[Rusty Russell]

On Mon, 29 Dec 2003 11:43:29 -0800 (PST)
Linus Torvalds <torvalds@osdl.org> wrote:
> > question; which lock prevents someone else claiming the irq and making
> > it unsharable/unclaimable between can_request_irq() and the eventual
> > request_irq() ????
> 
> Nothing. It never did. This is basically a heuristic: "find the irq that 
> looks the least used".

I just worry about other people using it.  I'd prefer "irq_looks_free()",
but since it's x86-specific it's not a big issue.

Rusty.
-- 
   there are those who do and those who hang on and you don't see too
   many doers quoting their contemporaries.  -- Larry McVoy