Re: PCI Device Driver / remap_pfn_range()

[Nick Piggin]

Alan Cox wrote:
> On Maw, 2006-04-18 at 04:18 +1000, Nick Piggin wrote:
>
> > I'm pretty sure you can't remap_pfn_range vmalloced memory because
> > it doesn't use contiguous page frames.
>
>
> To remap vmalloc memory you need something like this. Note that vmalloc
> memory may not be DMA accessible, vmalloc_32 memory maybe. Alternatively
> you can build your own scatter gather  lists from pages subject to
> hardware limits.
>
> The following GPL code from various drivers shows how to do vmalloc
> mapping into an application. Having a common helper for this is a
> discussion/todo item when that area of the vm gets future adjustments
> but for now this code should do the trick:

Yep, that would be a good option, thanks Alan.

You can look through the tree at how drivers remap their rvmalloced
memory -- (here's a snippet from drivers/media/video/meye.c):

while (size > 0) {
    page = vmalloc_to_pfn((void *)pos);
    if (remap_pfn_range(vma, start, page, PAGE_SIZE, PAGE_SHARED)) {
        mutex_unlock(&meye.lock);
        return -EAGAIN;
    }
    start += PAGE_SIZE;
    pos += PAGE_SIZE;
    if (size > PAGE_SIZE)
        size -= PAGE_SIZE;
    else
        size = 0;
}

The trick is to get the underlying pfns and remap one at a time.

You could also tinker with vmalloc_to_page+vm_insert_page, although
that might not yet be the best option for an out of tree driver.

--
SUSE Labs, Novell Inc.
Send instant messages to your online friends http://au.messenger.yahoo.com 
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/