On Tue, Jul 20, 2021 at 08:57:06AM +0200, Christoph Hellwig wrote:

On Mon, Jul 19, 2021 at 07:39:52PM +0100, Matthew Wilcox (Oracle) wrote:

quoted

All but one caller already has the iomap_page, and we can avoid getting
it again.

Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>

In general reloading it from a single pointer should be faster than
passing an extra argument.  What is the rationale here?

I'm not sure we can make that determination in general; it's going to
depend on the number of registers the CPU has, the calling conventions,
how much inlining the compiler chooses to do, etc.

The compiler is actually _forced_ to reload the iop from the page
by the test_bit() in PagePrivate() / folio_test_private().  I want
to kill off that bit, but until I have time to work on that, we
don't have a good workaround.