Re: pci_alloc_consistent from interrupt == BAD

[Pete Zaitcev <zaitcev@redhat.com>]

> Date: Mon, 21 Jan 2002 16:08:53 -0800 (PST)
> From: "David S. Miller" <davem@redhat.com>
> 
>    From: Pete Zaitcev <zaitcev@redhat.com>
>    Date: Mon, 21 Jan 2002 18:52:05 -0500
> 
>    > It [GFP_HIGH] is a trivial fix whereas backing
>    > out this ability to call pci_alloc_consistent from interrupts is not
>    > a trivial change at all.
>    
>    David, would you make a statement about pci_free_consistent.
>    I find it annoying that it cannot be called from an interrupt.
>    
> It is not an unreasonable requirement.  Wasn't there some problem with
> the pci pool stuff if it free'd up a chunk in an interrupt?

Yes, the pci_pool_free is one of the reasons why I ask.
I understand that the fundamental difference is that
allocation can fail (given right GFP flag), so this is what
we do in an interrupt (and we drop a packet in the driver).
But frees cannot fail, so if that tries to slip then we are stuck.

-- Pete