ANDROID: dm-bow: ignore discard in CHECKPOINT state

In the CHECKPOINT state, the original driver behavior was to pass discard
requests through to the underlying device. This creates a severe data
integrity issue, as it could physically erase data in ranges the driver
considers `UNCHANGED` and is actively protecting.

This pass-through behavior causes the driver's internal state map to
become desynchronized from the actual physical state of the storage,
undermining the purpose of the checkpoint.

This commit modifies the I/O mapping logic in `dm_bow_map` to intercept
discard requests received during the CHECKPOINT state. Instead of passing
the request down, the driver now immediately completes the `bio` with a
success status (`bio_endio`) and returns `DM_MAPIO_SUBMITTED`.

This effectively and safely ignores the discard, preventing the unintended
erasure of protected data and ensuring the checkpoint's integrity is
maintained.

Bug: 463896763
Bug: 433815205
Change-Id: Ic534f9729bd6bf73df41c2952853c75cee3b5c96
(cherry picked from commit a4a94617e6941915942489956fdc8a1bf2cc4729)
Signed-off-by: Lianwei Wang <lianwei.wang@gm.com>

drivers/md/dm-bow.c

diff --git a/drivers/md/dm-bow.c b/drivers/md/dm-bow.c
index 6ce39fa..9fa79b6 100644
--- a/drivers/md/dm-bow.c
+++ b/drivers/md/dm-bow.c
@@ -1163,7 +1163,16 @@ static int dm_bow_map(struct dm_target *ti, struct bio *bio)
 		} else if (state == CHECKPOINT) {
 			if (bio->bi_iter.bi_sector == 0)
 				ret = handle_sector0(bc, bio);
-			else if (bio_data_dir(bio) == WRITE)
+			else if (bio_op(bio) == REQ_OP_DISCARD) {
+				/*
+				 * Ignore discard requests in CHECKPOINT state.
+				 * Passing them through would physically erase data that we
+				 * are trying to protect, creating a state mismatch.
+				 * We complete the bio with success and stop processing.
+				 */
+				bio_endio(bio);
+				ret = DM_MAPIO_SUBMITTED;
+			} else if (bio_data_dir(bio) == WRITE)
 				ret = queue_write(bc, bio);
 			/* else pass-through */
 		}