The mx35lf1ge4ab_get_eccsr() function uses an SPI DMA operation to
read the eccsr, hence the buffer should not be on stack. Since commit
380583227c0c7f ("spi: spi-mem: Add extra sanity checks on the op param")
the kernel emmits a warning and blocks such operations.
Use the scratch buffer to get eccsr instead of trying to directly read
into a stack-allocated variable.
Signed-off-by: Daniel Golle <[email protected]>
---
drivers/mtd/nand/spi/macronix.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/mtd/nand/spi/macronix.c b/drivers/mtd/nand/spi/macronix.c
index dce835132a1e2..722a9738ba370 100644
--- a/drivers/mtd/nand/spi/macronix.c
+++ b/drivers/mtd/nand/spi/macronix.c
@@ -83,9 +83,10 @@ static int mx35lf1ge4ab_ecc_get_status(struct spinand_device *spinand,
* in order to avoid forcing the wear-leveling layer to move
* data around if it's not necessary.
*/
- if (mx35lf1ge4ab_get_eccsr(spinand, &eccsr))
+ if (mx35lf1ge4ab_get_eccsr(spinand, spinand->scratchbuf))
return nanddev_get_ecc_conf(nand)->strength;
+ eccsr = *spinand->scratchbuf;
if (WARN_ON(eccsr > nanddev_get_ecc_conf(nand)->strength ||
!eccsr))
return nanddev_get_ecc_conf(nand)->strength;
--
2.39.1
Hi Daniel,
On 19/01/23 09:15, Daniel Golle wrote:
> The mx35lf1ge4ab_get_eccsr() function uses an SPI DMA operation to
> read the eccsr, hence the buffer should not be on stack. Since commit
> 380583227c0c7f ("spi: spi-mem: Add extra sanity checks on the op param")
> the kernel emmits a warning and blocks such operations.
Good catch!
>
> Use the scratch buffer to get eccsr instead of trying to directly read
> into a stack-allocated variable.
>
> Signed-off-by: Daniel Golle <[email protected]>
> ---
Reviewed-by: Dhruva Gole <[email protected]>
> drivers/mtd/nand/spi/macronix.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/mtd/nand/spi/macronix.c b/drivers/mtd/nand/spi/macronix.c
> index dce835132a1e2..722a9738ba370 100644
> --- a/drivers/mtd/nand/spi/macronix.c
> +++ b/drivers/mtd/nand/spi/macronix.c
> @@ -83,9 +83,10 @@ static int mx35lf1ge4ab_ecc_get_status(struct spinand_device *spinand,
> * in order to avoid forcing the wear-leveling layer to move
> * data around if it's not necessary.
> */
> - if (mx35lf1ge4ab_get_eccsr(spinand, &eccsr))
> + if (mx35lf1ge4ab_get_eccsr(spinand, spinand->scratchbuf))
> return nanddev_get_ecc_conf(nand)->strength;
>
> + eccsr = *spinand->scratchbuf;
> if (WARN_ON(eccsr > nanddev_get_ecc_conf(nand)->strength ||
> !eccsr))
> return nanddev_get_ecc_conf(nand)->strength;
--
Thanks and Regards,
Dhruva Gole
Hi Daniel,
[email protected] wrote on Thu, 19 Jan 2023 03:45:43 +0000:
> The mx35lf1ge4ab_get_eccsr() function uses an SPI DMA operation to
> read the eccsr, hence the buffer should not be on stack. Since commit
> 380583227c0c7f ("spi: spi-mem: Add extra sanity checks on the op param")
> the kernel emmits a warning and blocks such operations.
>
> Use the scratch buffer to get eccsr instead of trying to directly read
> into a stack-allocated variable.
>
> Signed-off-by: Daniel Golle <[email protected]>
> ---
Looks like my scripts did not properly sent the "applied" e-mail, so
here it is, I applied this patch to nand/next.
> drivers/mtd/nand/spi/macronix.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/mtd/nand/spi/macronix.c b/drivers/mtd/nand/spi/macronix.c
> index dce835132a1e2..722a9738ba370 100644
> --- a/drivers/mtd/nand/spi/macronix.c
> +++ b/drivers/mtd/nand/spi/macronix.c
> @@ -83,9 +83,10 @@ static int mx35lf1ge4ab_ecc_get_status(struct spinand_device *spinand,
> * in order to avoid forcing the wear-leveling layer to move
> * data around if it's not necessary.
> */
> - if (mx35lf1ge4ab_get_eccsr(spinand, &eccsr))
> + if (mx35lf1ge4ab_get_eccsr(spinand, spinand->scratchbuf))
> return nanddev_get_ecc_conf(nand)->strength;
>
> + eccsr = *spinand->scratchbuf;
> if (WARN_ON(eccsr > nanddev_get_ecc_conf(nand)->strength ||
> !eccsr))
> return nanddev_get_ecc_conf(nand)->strength;
Thanks,
Miquèl