mmc: sd: optimize erase

To SD, there is no erase group, then the value erase_grp_size
will be default 1. When erasing SD blocks, the blocks will be
erased one by one, which is time consuming.

We use AU_SIZE as a group to speed up the erasing.

Erasing 4MB with a SD2.0 Card with AU_SIZE 4MB.
`time mmc erase 0x100000 0x2000`
time: 44.856 seconds (before optimization)
time: 0.335 seconds  (after optimization)

Signed-off-by: Peng Fan <peng.fan@nxp.com>
Cc: Jaehoon Chung <jh80.chung@samsung.com>
Cc: Simon Glass <sjg@chromium.org>
Cc: Bin Meng <bmeng.cn@gmail.com>
Cc: Stefan Wahren <stefan.wahren@i2se.com>
Cc: Clemens Gruber <clemens.gruber@pqgruber.com>
Cc: Kever Yang <kever.yang@rock-chips.com>
Cc: Eric Nelson <eric@nelint.com>
Cc: Stephen Warren <swarren@nvidia.com>
diff --git a/drivers/mmc/mmc_write.c b/drivers/mmc/mmc_write.c
index 0f8b5c7..2289640 100644
--- a/drivers/mmc/mmc_write.c
+++ b/drivers/mmc/mmc_write.c
@@ -100,8 +100,13 @@
 		       & ~(mmc->erase_grp_size - 1)) - 1);
 
 	while (blk < blkcnt) {
-		blk_r = ((blkcnt - blk) > mmc->erase_grp_size) ?
-			mmc->erase_grp_size : (blkcnt - blk);
+		if (IS_SD(mmc) && mmc->ssr.au) {
+			blk_r = ((blkcnt - blk) > mmc->ssr.au) ?
+				mmc->ssr.au : (blkcnt - blk);
+		} else {
+			blk_r = ((blkcnt - blk) > mmc->erase_grp_size) ?
+				mmc->erase_grp_size : (blkcnt - blk);
+		}
 		err = mmc_erase_t(mmc, start + blk, blk_r);
 		if (err)
 			break;