On Tue 04-10-16 09:39:48, Ross Zwisler wrote:

quoted

The gfp_mask that propagates from __do_fault() or do_page_mkwrite() is fine
because at that point it is correct. But once we grab filesystem locks
which are not reclaim safe, we should update vmf->gfp_mask we pass further
down into DAX code to not contain __GFP_FS (that's a bug we apparently have
there). And inside DAX code, we definitely are not generally safe to add
__GFP_FS to mapping_gfp_mask(). Maybe we'd be better off propagating struct
vm_fault into this function, using passed gfp_mask there and make sure
callers update gfp_mask as appropriate.

Yep, that makes sense to me.  In reviewing your set it also occurred to me that
we might want to stick a struct vm_area_struct *vma pointer in the vmf, since
you always need a vma when you are using a vmf, but we pass them as a pair
everywhere.

Actually, vma pointer will be in struct vm_fault after my DAX
write-protection series. So once that lands, we can clean up whatever
duplicit function parameters...

								Honza

-- 
Jan Kara [off-list ref]
SUSE Labs, CR