[PATCH] mtd: nand-imx: repair reading the oob area

[Uwe Kleine-König]

After sending the READOOB command the core (in nand_read_oob_std()) uses
the read_buf callback to fetch the result. So in the imx driver it is
necessary to emit the NFC_OUTPUT command to make the controller actually
read the requested data into the internal buffer and then copy the data
to the location where read_buf expects it.

This fixes reading from /dev/nand0.oob and also (somewhat) reading bad
block markers in the absence of a bad block table.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig at pengutronix.de>
---
Hello,

I failed to find the commit that broke reading from /dev/nand0.oob but
failed to identify it. I think this bug is quite old. I didn't test, but
I think this was already broken in 2010.09.0 where the nand support was
moved from drivers/nand to drivers/mtd/nand. (This is where gitk output
ended :-)

Best regards
Uwe

 drivers/mtd/nand/nand_imx.c | 15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)

diff --git a/drivers/mtd/nand/nand_imx.c b/drivers/mtd/nand/nand_imx.c
index d69a012f017b..f9d3c2e34a99 100644
--- a/drivers/mtd/nand/nand_imx.c
+++ b/drivers/mtd/nand/nand_imx.c
@@ -996,14 +996,23 @@ static void imx_nand_command(struct mtd_info *mtd, unsigned command,
 		else
 			host->buf_start = column + mtd->writesize;
 
-		command = NAND_CMD_READ0;
-
-		host->send_cmd(host, command);
+		host->send_cmd(host, NAND_CMD_READ0);
 		mxc_do_addr_cycle(mtd, column, page_addr);
 
 		if (host->pagesize_2k)
 			/* send read confirm command */
 			host->send_cmd(host, NAND_CMD_READSTART);
+
+		/*
+		 * After the core issued READOOB the result is read using
+		 * .read_buf, so we have to make sure the data is actually
+		 * there.
+		 */
+		if (command == NAND_CMD_READOOB) {
+			host->send_page(host, NFC_OUTPUT);
+			copy_spare(mtd, 1, host->data_buf + mtd->writesize);
+		}
+
 		break;
 
 	case NAND_CMD_SEQIN:
-- 
2.28.0