On Wed, Feb 26, 2020 at 09:04:25AM -0800, Darrick J. Wong wrote:

quoted

@@ -456,15 +435,8 @@ iomap_readpages(struct address_space *mapping, struct list_head *pages,
 			unlock_page(ctx.cur_page);
 		put_page(ctx.cur_page);
 	}
-
-	/*
-	 * Check that we didn't lose a page due to the arcance calling
-	 * conventions..
-	 */
-	WARN_ON_ONCE(!ret && !list_empty(ctx.pages));
-	return ret;

After all the discussion about "if we still have ctx.cur_page we should
just stop" in v7, I'm surprised that this patch now doesn't say much of
anything, not even a WARN_ON()?

The code quoted above puts the cur_page reference.  By dropping the
odd refactoring patch there is no need to check for cur_page being
left as a special condition as that still is the normal loop exit
state and properly handled, just as in the original iomap code.