[PATCH] mtd: spi-nor: fix page offset/remains calculation in spi_nor_write

[PATCH] mtd: spi-nor: fix page offset/remains calculation in spi_nor_write

[Bastian Krause]

nor->write() simply adds the number of written bytes to the pointer
given. Thus retval is incremented in each loop cycle for each
spi_nor_write() call without ever resetting it. This leads to wrong page
offset/remains calculations and an incorrect number of bytes written to
retlen.

This becomes apparant only if the calling function actually compares len
and retlen (e.g. mtd_peb_write() ). Otherwise wrong data is written:

  $ memcpy -s /dev/prng -d prng_data 0 0 10k
  $ erase /dev/mtd0.mypart
  $ cp prng_data /dev/mtd0.mypart
  $ memcmp -s prng_data -d /dev/mtd0.mypart 0 0

memcmp returned "files differ" before, with this patch it returns "OK".

Fixes: c8516869c4 ("spi: Extend the core to ease integration of SPI memory controllers")
Signed-off-by: Bastian Krause <bst@pengutronix.de>
---
 drivers/mtd/spi-nor/spi-nor.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/mtd/spi-nor/spi-nor.c b/drivers/mtd/spi-nor/spi-nor.c
index b2052ce0af..9dbc8302ed 100644
--- a/drivers/mtd/spi-nor/spi-nor.c
+++ b/drivers/mtd/spi-nor/spi-nor.c
@@ -972,6 +972,7 @@ static int spi_nor_write(struct mtd_info *mtd, loff_t to, size_t len,
 
 	for (i = 0; i < len; ) {
 		ssize_t written;
+		retval = 0;
 
 		page_offset = (to + i) & (nor->page_size - 1);
 		page_remain = min_t(size_t, nor->page_size - page_offset,
-- 
2.20.1


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH] mtd: spi-nor: fix page offset/remains calculation in spi_nor_write

[Bastian Krause]

On 8/1/19 11:19 AM, Bastian Krause wrote:
> nor->write() simply adds the number of written bytes to the pointer
> given. Thus retval is incremented in each loop cycle for each
> spi_nor_write() call without ever resetting it. This leads to wrong page
> offset/remains calculations and an incorrect number of bytes written to
> retlen.

Note: this happens only with the cadence-quadspi driver.

Bastian

> 
> This becomes apparant only if the calling function actually compares len
> and retlen (e.g. mtd_peb_write() ). Otherwise wrong data is written:
> 
>   $ memcpy -s /dev/prng -d prng_data 0 0 10k
>   $ erase /dev/mtd0.mypart
>   $ cp prng_data /dev/mtd0.mypart
>   $ memcmp -s prng_data -d /dev/mtd0.mypart 0 0
> 
> memcmp returned "files differ" before, with this patch it returns "OK".
> 
> Fixes: c8516869c4 ("spi: Extend the core to ease integration of SPI memory controllers")
> Signed-off-by: Bastian Krause <bst@pengutronix.de>
> ---
>  drivers/mtd/spi-nor/spi-nor.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/mtd/spi-nor/spi-nor.c b/drivers/mtd/spi-nor/spi-nor.c
> index b2052ce0af..9dbc8302ed 100644
> --- a/drivers/mtd/spi-nor/spi-nor.c
> +++ b/drivers/mtd/spi-nor/spi-nor.c
> @@ -972,6 +972,7 @@ static int spi_nor_write(struct mtd_info *mtd, loff_t to, size_t len,
>  
>  	for (i = 0; i < len; ) {
>  		ssize_t written;
> +		retval = 0;
>  
>  		page_offset = (to + i) & (nor->page_size - 1);
>  		page_remain = min_t(size_t, nor->page_size - page_offset,
> 


-- 
Pengutronix e.K.
Industrial Linux Solutions
http://www.pengutronix.de/
Peiner Str. 6-8, 31137 Hildesheim, Germany
Amtsgericht Hildesheim, HRA 2686

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH] mtd: spi-nor: fix page offset/remains calculation in spi_nor_write

[Sascha Hauer]

On Thu, Aug 01, 2019 at 11:19:50AM +0200, Bastian Krause wrote:
> nor->write() simply adds the number of written bytes to the pointer
> given. Thus retval is incremented in each loop cycle for each
> spi_nor_write() call without ever resetting it. This leads to wrong page
> offset/remains calculations and an incorrect number of bytes written to
> retlen.
> 
> This becomes apparant only if the calling function actually compares len
> and retlen (e.g. mtd_peb_write() ). Otherwise wrong data is written:
> 
>   $ memcpy -s /dev/prng -d prng_data 0 0 10k
>   $ erase /dev/mtd0.mypart
>   $ cp prng_data /dev/mtd0.mypart
>   $ memcmp -s prng_data -d /dev/mtd0.mypart 0 0
> 
> memcmp returned "files differ" before, with this patch it returns "OK".
> 
> Fixes: c8516869c4 ("spi: Extend the core to ease integration of SPI memory controllers")
> Signed-off-by: Bastian Krause <bst@pengutronix.de>
> ---

Applied, thanks

Sascha


-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox