2018-07-06 13:08:07

by Benjamin Gaignard

Subject: [PATCH] mtd: stm32-quadspi: replace "%p" with "%pK"

The format specifier "%p" can leak kernel addresses.
Use "%pK" instead.

Signed-off-by: Benjamin Gaignard <[email protected]>
---
drivers/mtd/spi-nor/stm32-quadspi.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/mtd/spi-nor/stm32-quadspi.c b/drivers/mtd/spi-nor/stm32-quadspi.c
index 72553506a00b..13e9fc961d3b 100644
--- a/drivers/mtd/spi-nor/stm32-quadspi.c
+++ b/drivers/mtd/spi-nor/stm32-quadspi.c
@@ -355,7 +355,7 @@ static int stm32_qspi_read_reg(struct spi_nor *nor,
struct device *dev = flash->qspi->dev;
struct stm32_qspi_cmd cmd;

- dev_dbg(dev, "read_reg: cmd:%#.2x buf:%p len:%#x\n", opcode, buf, len);
+ dev_dbg(dev, "read_reg: cmd:%#.2x buf:%pK len:%#x\n", opcode, buf, len);

memset(&cmd, 0, sizeof(cmd));
cmd.opcode = opcode;
@@ -376,7 +376,7 @@ static int stm32_qspi_write_reg(struct spi_nor *nor, u8 opcode,
struct device *dev = flash->qspi->dev;
struct stm32_qspi_cmd cmd;

- dev_dbg(dev, "write_reg: cmd:%#.2x buf:%p len:%#x\n", opcode, buf, len);
+ dev_dbg(dev, "write_reg: cmd:%#.2x buf:%pK len:%#x\n", opcode, buf, len);

memset(&cmd, 0, sizeof(cmd));
cmd.opcode = opcode;
@@ -398,7 +398,7 @@ static ssize_t stm32_qspi_read(struct spi_nor *nor, loff_t from, size_t len,
struct stm32_qspi_cmd cmd;
int err;

- dev_dbg(qspi->dev, "read(%#.2x): buf:%p from:%#.8x len:%#zx\n",
+ dev_dbg(qspi->dev, "read(%#.2x): buf:%pK from:%#.8x len:%#zx\n",
nor->read_opcode, buf, (u32)from, len);

memset(&cmd, 0, sizeof(cmd));
--
2.15.0

2018-07-06 14:45:40

by Richard Weinberger

Subject: Re: [PATCH] mtd: stm32-quadspi: replace "%p" with "%pK"

Am Freitag, 6. Juli 2018, 15:05:25 CEST schrieb Benjamin Gaignard:
> The format specifier "%p" can leak kernel addresses.
> Use "%pK" instead.

Does %pK really make sense for dev_dbg()?
When the driver is being debugged "leaking" is on purpose. :)

Thanks,
//richard

> Signed-off-by: Benjamin Gaignard <[email protected]>
> ---
> drivers/mtd/spi-nor/stm32-quadspi.c | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/mtd/spi-nor/stm32-quadspi.c b/drivers/mtd/spi-nor/stm32-quadspi.c
> index 72553506a00b..13e9fc961d3b 100644
> --- a/drivers/mtd/spi-nor/stm32-quadspi.c
> +++ b/drivers/mtd/spi-nor/stm32-quadspi.c
> @@ -355,7 +355,7 @@ static int stm32_qspi_read_reg(struct spi_nor *nor,
> struct device *dev = flash->qspi->dev;
> struct stm32_qspi_cmd cmd;
>
> - dev_dbg(dev, "read_reg: cmd:%#.2x buf:%p len:%#x\n", opcode, buf, len);
> + dev_dbg(dev, "read_reg: cmd:%#.2x buf:%pK len:%#x\n", opcode, buf, len);
>
> memset(&cmd, 0, sizeof(cmd));
> cmd.opcode = opcode;
> @@ -376,7 +376,7 @@ static int stm32_qspi_write_reg(struct spi_nor *nor, u8 opcode,
> struct device *dev = flash->qspi->dev;
> struct stm32_qspi_cmd cmd;
>
> - dev_dbg(dev, "write_reg: cmd:%#.2x buf:%p len:%#x\n", opcode, buf, len);
> + dev_dbg(dev, "write_reg: cmd:%#.2x buf:%pK len:%#x\n", opcode, buf, len);
>
> memset(&cmd, 0, sizeof(cmd));
> cmd.opcode = opcode;
> @@ -398,7 +398,7 @@ static ssize_t stm32_qspi_read(struct spi_nor *nor, loff_t from, size_t len,
> struct stm32_qspi_cmd cmd;
> int err;
>
> - dev_dbg(qspi->dev, "read(%#.2x): buf:%p from:%#.8x len:%#zx\n",
> + dev_dbg(qspi->dev, "read(%#.2x): buf:%pK from:%#.8x len:%#zx\n",
> nor->read_opcode, buf, (u32)from, len);
>
> memset(&cmd, 0, sizeof(cmd));
>

2018-07-06 14:50:39

by Boris Brezillon

Subject: Re: [PATCH] mtd: stm32-quadspi: replace "%p" with "%pK"

On Fri, 06 Jul 2018 16:44:44 +0200
Richard Weinberger <[email protected]> wrote:

> Am Freitag, 6. Juli 2018, 15:05:25 CEST schrieb Benjamin Gaignard:
> > The format specifier "%p" can leak kernel addresses.
> > Use "%pK" instead.
>
> Does %pK really make sense for dev_dbg()?
> When the driver is being debugged "leaking" is on purpose. :)

%p already gives you a hash of the pointer address. At least with %pK
you can turn it off through a sysctl :).

>
> Thanks,
> //richard
>
> > Signed-off-by: Benjamin Gaignard <[email protected]>
> > ---
> > drivers/mtd/spi-nor/stm32-quadspi.c | 6 +++---
> > 1 file changed, 3 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/mtd/spi-nor/stm32-quadspi.c b/drivers/mtd/spi-nor/stm32-quadspi.c
> > index 72553506a00b..13e9fc961d3b 100644
> > --- a/drivers/mtd/spi-nor/stm32-quadspi.c
> > +++ b/drivers/mtd/spi-nor/stm32-quadspi.c
> > @@ -355,7 +355,7 @@ static int stm32_qspi_read_reg(struct spi_nor *nor,
> > struct device *dev = flash->qspi->dev;
> > struct stm32_qspi_cmd cmd;
> >
> > - dev_dbg(dev, "read_reg: cmd:%#.2x buf:%p len:%#x\n", opcode, buf, len);
> > + dev_dbg(dev, "read_reg: cmd:%#.2x buf:%pK len:%#x\n", opcode, buf, len);
> >
> > memset(&cmd, 0, sizeof(cmd));
> > cmd.opcode = opcode;
> > @@ -376,7 +376,7 @@ static int stm32_qspi_write_reg(struct spi_nor *nor, u8 opcode,
> > struct device *dev = flash->qspi->dev;
> > struct stm32_qspi_cmd cmd;
> >
> > - dev_dbg(dev, "write_reg: cmd:%#.2x buf:%p len:%#x\n", opcode, buf, len);
> > + dev_dbg(dev, "write_reg: cmd:%#.2x buf:%pK len:%#x\n", opcode, buf, len);
> >
> > memset(&cmd, 0, sizeof(cmd));
> > cmd.opcode = opcode;
> > @@ -398,7 +398,7 @@ static ssize_t stm32_qspi_read(struct spi_nor *nor, loff_t from, size_t len,
> > struct stm32_qspi_cmd cmd;
> > int err;
> >
> > - dev_dbg(qspi->dev, "read(%#.2x): buf:%p from:%#.8x len:%#zx\n",
> > + dev_dbg(qspi->dev, "read(%#.2x): buf:%pK from:%#.8x len:%#zx\n",
> > nor->read_opcode, buf, (u32)from, len);
> >
> > memset(&cmd, 0, sizeof(cmd));
> >
>
>

2018-07-06 14:51:47

by Richard Weinberger

Subject: Re: [PATCH] mtd: stm32-quadspi: replace "%p" with "%pK"

Am Freitag, 6. Juli 2018, 16:47:32 CEST schrieb Boris Brezillon:
> On Fri, 06 Jul 2018 16:44:44 +0200
> Richard Weinberger <[email protected]> wrote:
>
> > Am Freitag, 6. Juli 2018, 15:05:25 CEST schrieb Benjamin Gaignard:
> > > The format specifier "%p" can leak kernel addresses.
> > > Use "%pK" instead.
> >
> > Does %pK really make sense for dev_dbg()?
> > When the driver is being debugged "leaking" is on purpose. :)
>
> %p already gives you a hash of the pointer address. At least with %pK
> you can turn it off through a sysctl :).

Once again I mixed %p and %pK.
So then %pK is what we want. :D

Thanks,
//richard

2018-07-07 10:09:58

by Boris Brezillon

Subject: Re: [PATCH] mtd: stm32-quadspi: replace "%p" with "%pK"

On Fri, 6 Jul 2018 15:05:25 +0200
Benjamin Gaignard <[email protected]> wrote:

> The format specifier "%p" can leak kernel addresses.
> Use "%pK" instead.
>
> Signed-off-by: Benjamin Gaignard <[email protected]>

Applied.

Thanks,

Boris

> ---
> drivers/mtd/spi-nor/stm32-quadspi.c | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/mtd/spi-nor/stm32-quadspi.c b/drivers/mtd/spi-nor/stm32-quadspi.c
> index 72553506a00b..13e9fc961d3b 100644
> --- a/drivers/mtd/spi-nor/stm32-quadspi.c
> +++ b/drivers/mtd/spi-nor/stm32-quadspi.c
> @@ -355,7 +355,7 @@ static int stm32_qspi_read_reg(struct spi_nor *nor,
> struct device *dev = flash->qspi->dev;
> struct stm32_qspi_cmd cmd;
>
> - dev_dbg(dev, "read_reg: cmd:%#.2x buf:%p len:%#x\n", opcode, buf, len);
> + dev_dbg(dev, "read_reg: cmd:%#.2x buf:%pK len:%#x\n", opcode, buf, len);
>
> memset(&cmd, 0, sizeof(cmd));
> cmd.opcode = opcode;
> @@ -376,7 +376,7 @@ static int stm32_qspi_write_reg(struct spi_nor *nor, u8 opcode,
> struct device *dev = flash->qspi->dev;
> struct stm32_qspi_cmd cmd;
>
> - dev_dbg(dev, "write_reg: cmd:%#.2x buf:%p len:%#x\n", opcode, buf, len);
> + dev_dbg(dev, "write_reg: cmd:%#.2x buf:%pK len:%#x\n", opcode, buf, len);
>
> memset(&cmd, 0, sizeof(cmd));
> cmd.opcode = opcode;
> @@ -398,7 +398,7 @@ static ssize_t stm32_qspi_read(struct spi_nor *nor, loff_t from, size_t len,
> struct stm32_qspi_cmd cmd;
> int err;
>
> - dev_dbg(qspi->dev, "read(%#.2x): buf:%p from:%#.8x len:%#zx\n",
> + dev_dbg(qspi->dev, "read(%#.2x): buf:%pK from:%#.8x len:%#zx\n",
> nor->read_opcode, buf, (u32)from, len);
>
> memset(&cmd, 0, sizeof(cmd));