On Wed, 15 Jan 2014 14:57:56 -0800 (PST) David Rientjes <rientjes@xxxxxxxxxx> wrote:
> On Wed, 15 Jan 2014, Geert Uytterhoeven wrote:
>
> > {,set}page_address() are macros if WANT_PAGE_VIRTUAL.
> > If !WANT_PAGE_VIRTUAL, they're plain C functions.
> >
> > If someone calls them with a void *, this pointer is auto-converted to
> > struct page * if !WANT_PAGE_VIRTUAL, but causes a build failure on
> > architectures using WANT_PAGE_VIRTUAL (arc, m68k and sparc):
>
> s/sparc/sparc64/
>
> >
> > drivers/md/bcache/bset.c: In function _____btree_sort___:
> > drivers/md/bcache/bset.c:1190: warning: dereferencing ___void *___ pointer
> > drivers/md/bcache/bset.c:1190: error: request for member ___virtual___ in something not a structure or union
> >
> > Convert them to static inline functions to fix this. There are already
> > plenty of users of struct page members inside <linux/mm.h>, so there's no
> > reason to keep them as macros.
> >
> > Signed-off-by: Geert Uytterhoeven <geert@xxxxxxxxxxxxxx>
>
> Tested-by: David Rientjes <rientjes@xxxxxxxxxx>
>
> I'm cringing at the page_address(mempool_alloc(..., GFP_NOIO)) in
> drivers/md/bcache/bset.c, though.
Yes, I was staring suspiciously at that as well. Ended up deciding
that I wouldn't have coded it that way, but we should support the
casting.
> It's relying on that fact that
> mempool_alloc() can never return NULL if __GFP_WAIT but I think this could
> have all been avoided with
>
> struct page *page = mempool_alloc(state->pool, GFP_NOIO);
> out = page_address(page);
>
> instead of burying the mempool_alloc() in page_address() for what I think
> is cleaner code. Owell, it fixes the issue.
And that would make the later virt_to_page() unnecessary.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>