On Mon, May 15 2017 at 11:28am -0400,
Christoph Hellwig [off-list ref] wrote:

Instead just turn the macro into a helper for the warning message.
This removes an unessecary assignment in this patch, and will allow
to fix a place where -EIO is the wrong error value in th next.

Signed-off-by: Christoph Hellwig <hch@lst.de>

Something clearlyy got screwed up in my rebase.  Because I specifically
recall casting the multipath_clone_and_map()'s !pgpath call with:
(void) dm_report_EIO(m);

Very weird.  Anyway, thanks for catching these issues.

Mike