[OpenAFS-devel] Re: Linux readpage handler

[Andrew Deason]

On Thu, 26 May 2011 18:00:25 +0100
Simon Wilkinson <sxw@inf.ed.ac.uk> wrote:

> I haven't verified this, but my reading of the write code path
> suggests that iov_iter_copy_from_user_atomic() will only copy data
> that is available and we'll go round the loop in
> generic_perform_write(), firstly paging in the data, then beginning
> the write, until we can copy all of the requested data. So I would
> hope that this isn't an issue on modern systems.

Yeah, I see it now. Even if we don't manage to copy any data out, it's
just another iteration of the loop, so we keep retrying until it
succeeds.

That's what I wanted to know; thanks!

-- 
Andrew Deason
adeason@sinenomine.net