On 27/09/18 19:13, Daniel Mack wrote:

Hi,

On 27/9/2018 1:42 AM, Chris Packham wrote:

quoted

On 27/09/18 11:33, Chris Packham wrote:

quoted

On 27/09/18 09:24, Daniel Mack wrote:

quoted

quoted

quoted

diff --git a/drivers/mtd/nand/raw/marvell_nand.c b/drivers/mtd/nand/raw/marvell_nand.c
index 666f34b58dec..e96ec7b9a152 100644
--- a/drivers/mtd/nand/raw/marvell_nand.c
+++ b/drivers/mtd/nand/raw/marvell_nand.c

@@ -613,7 +613,8 @@ static int marvell_nfc_wait_cmdd(struct nand_chip *chip)
     static int marvell_nfc_wait_op(struct nand_chip *chip, unsigned int timeout_ms)
     {
     	struct marvell_nfc *nfc = to_marvell_nfc(chip->controller);
-	int ret;
+	int ret = -EALREADY;

Won't this cause us to hit the if (!ret) below if we don't end up
calling wait_for_completion_timeout()?

Nope boolean logic fail.

Yup :)

quoted

quoted

In fact I did just hit something
like this on my Armada-385 based board

marvell-nfc f10d0000.nand-controller: Timeout waiting for RB signal
ubi0 warning: do_sync_erase: error -5 while erasing PEB 755, retry

But this still might be something else, it doesn't happen regularly. For
what it's worth

Tested-by: Chris Packham <chris.packham@alliedtelesis.co.nz>

Thanks. Did you see the issue that I described without that patch?

Yes. I've just tried a test without your patch and I see the same thing. 
I wonder if my R/B configuration is correct.