[PATCH v5 2/4] mtd: mtdconcat: don't use mtd_{suspend,resume}()

[Sean Nyekjaer]

From: Boris Brezillon <boris.brezillon at collabora.com>

The MTD suspend logic will soon be adjusted to automatically wait for
device wake-up before issuing IOs. In order to do that a new read-write
lock will be added and taken in write-mode in the
mtd_{suspend,resume}() path. Since mtdconcat.c itself is an MTD device,
calling mtd_suspend/resume() on subdevices from the mtdconcat
->_{suspend,resume}() hook will lead to a nested lock, which lockdep
will complain about if we don't add a proper annotation. Let's keep
things simple and replace those mtd_{suspend,resume}(subdev) calls by
subdev->_{suspend,resume}() ones to avoid this situation.

Tested-by: Sean Nyekjaer <sean at geanix.com>
Signed-off-by: Boris Brezillon <boris.brezillon at collabora.com>
Signed-off-by: Sean Nyekjaer <sean at geanix.com>
---
 drivers/mtd/mtdconcat.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

diff --git a/drivers/mtd/mtdconcat.c b/drivers/mtd/mtdconcat.c
index f685a581df48..f4a4274489b4 100644
--- a/drivers/mtd/mtdconcat.c
+++ b/drivers/mtd/mtdconcat.c
@@ -566,9 +566,15 @@ static int concat_suspend(struct mtd_info *mtd)
 
 	for (i = 0; i < concat->num_subdev; i++) {
 		struct mtd_info *subdev = concat->subdev[i];
-		if ((rc = mtd_suspend(subdev)) < 0)
+		/*
+		 * Call the MTD hook directly to avoid a nested lock
+		 * on ->suspend_lock.
+		 */
+		rc = subdev->_suspend ? subdev->_suspend(subdev) : 0;
+		if (rc < 0)
 			return rc;
 	}
+
 	return rc;
 }
 
@@ -579,7 +585,12 @@ static void concat_resume(struct mtd_info *mtd)
 
 	for (i = 0; i < concat->num_subdev; i++) {
 		struct mtd_info *subdev = concat->subdev[i];
-		mtd_resume(subdev);
+		/*
+		 * Call the MTD hook directly to avoid a nested lock
+		 * on ->resume_lock.
+		 */
+		if (subdev->_resume)
+			subdev->_resume(subdev);
 	}
 }
 
-- 
2.33.0