On Thu, Apr 25, 2019 at 05:33:04PM -0700, Dan Williams wrote:
On Thu, Apr 25, 2019 at 12:32 AM Jan Kara jack@suse.cz wrote:
We also call vmf_insert_pfn_pmd() in dax_insert_pfn_mkwrite() -- does that need to change too?
It wasn't clear to me that it was a problem. I think that one already happens to be pmd-aligned.
Why would it need to be? The address is taken from vmf->address and that's set up in __handle_mm_fault() like .address = address & PAGE_MASK. So I don't see anything forcing PMD alignment of the virtual address...
True. So now I'm wondering if the masking should be done internal to the routine. Given it's prefixed vmf_ it seems to imply the api is prepared to take raw 'struct vm_fault' parameters. I think I'll go that route unless someone sees a reason to require the caller to handle this responsibility.
The vmf_ prefix was originally used to indicate 'returns a vm_fault_t' instead of 'returns an errno'. That said, I like the interpretation you're coming up with here, and it makes me wonder if we shouldn't change vmf_insert_pfn_pmd() to take (vmf, pfn, write) as arguments instead of separate vma, address & pmd arguments.