On 20-11-07 18:23, Talib Alim wrote:
To make things clear for me and anybody reading this post (and please confirm).
I should not use
virtual_address = phys_to_virt(page_to_pfn(page) << PAGE_SHIFT);
(I assume this is what you meant by home grown)
Indeed. If you'd need this effect, you'd use page_address(), which is the
same as the above for !CONFIG_HIGHMEM kernels and turns into to the above
for lowmem pages even on a CONFIG_HIGHMEM kernel.
and should use
virtual_address = kmap_atomic(page, KM_USER0);
and when done, use
kunmap_atomic(virtual_address, KM_USER0);
Either that or somehow arrange that you'll never be passed a highmem page to
begin with as highmem pages by their nature of not being permanently mapped
do not _have_ a virtual address until after they're mapped.
is KM_USER0 correct ?
Yes, AFAIA that should be correct. And remember -- you can't sleep while
holding the page.
Rene.
--
To unsubscribe from this list: send an email with
"unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
Please read the FAQ at http://kernelnewbies.org/FAQ
