Re: [PATCH] HPET driver

[Vojtech Pavlik <vojtech@suse.cz>]

On Thu, May 13, 2004 at 07:49:22PM -0400, Jeff Garzik wrote:
> Andrew Morton wrote:
> >Jeff Garzik <jgarzik@pobox.com> wrote:
> >
> >>>+    vma->vm_flags |= (VM_IO | VM_SHM | VM_LOCKED);
> >>>+    vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
> >>>+    addr = __pa(addr);
> >>
> >>where did these flags come from?  don't you just want VM_RESERVED?
> >
> >
> >VM_IO is the way to mark mmapped I/O devices. 
> >
> >	vma->vm_flags |= VM_IO;
> >
> >should be sufficient here.
> >
> >hm, I'm trying to decrypt how the driver accesses the hardware.  It's
> >taking copies of kernel virtual addresses based off hpet_virt_address, but
> >there are no readl's or writel's in there.  Is the actual device access
> >done over in time_hpet.c?
> 
> 
> HPET writes into RAM at magic addresses, so it's not really a bus address.

Since it lives in the bridge (either north or south), IMO it is a bus
address ... at least on AMD machines it's the same space where MMIO
comes from.

> Thus I think only VM_RESERVED is needed...
> 
> 	Jeff

-- 
Vojtech Pavlik
SuSE Labs, SuSE CR