[PATCH RESEND] spi: erase pointer to drvdata on removal

[PATCH RESEND] spi: erase pointer to drvdata on removal

[Vivien Didelot]

Without this patch, a SPI device may keep its drvdata whereas it was unbound
from its driver. This may result in accessing an invalid pointer.

As for i2c-core, let the SPI core handle the removal of the device's drvdata,
after a remove(), or a probe() failure.

This is a resent of the previous patch https://lkml.org/lkml/2012/11/1/314

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
---
 drivers/spi/spi.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/drivers/spi/spi.c b/drivers/spi/spi.c
index 84c2861..fe636fe 100644
--- a/drivers/spi/spi.c
+++ b/drivers/spi/spi.c
@@ -233,15 +233,25 @@ EXPORT_SYMBOL_GPL(spi_bus_type);
 static int spi_drv_probe(struct device *dev)
 {
 	const struct spi_driver		*sdrv = to_spi_driver(dev->driver);
+	struct spi_device		*sdev = to_spi_device(dev);
+	int				status;
 
-	return sdrv->probe(to_spi_device(dev));
+	status = sdrv->probe(sdev);
+	if (status)
+		spi_set_drvdata(sdev, NULL);
+	return status;
 }
 
 static int spi_drv_remove(struct device *dev)
 {
 	const struct spi_driver		*sdrv = to_spi_driver(dev->driver);
+	struct spi_device		*sdev = to_spi_device(dev);
+	int				status;
 
-	return sdrv->remove(to_spi_device(dev));
+	status = sdrv->remove(sdev);
+	if (status == 0)
+		spi_set_drvdata(sdev, NULL);
+	return status;
 }
 
 static void spi_drv_shutdown(struct device *dev)
-- 
1.7.11.7

Re: [PATCH RESEND] spi: erase pointer to drvdata on removal

[Grant Likely]

On Mon, Dec 3, 2012 at 8:28 PM, Vivien Didelot
<vivien.didelot@savoirfairelinux.com> wrote:
> Without this patch, a SPI device may keep its drvdata whereas it was unbound
> from its driver. This may result in accessing an invalid pointer.
>
> As for i2c-core, let the SPI core handle the removal of the device's drvdata,
> after a remove(), or a probe() failure.
>
> This is a resent of the previous patch https://lkml.org/lkml/2012/11/1/314
>
> Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>

Are you seeing a real bug without this patch? If so, then that is a
driver bug. I don't want to mask over bugs silently. The kernel should
complain.

There are actually lots of things that drivers should/should-not do
that could be checked by the probe/remove hooks, and not just in the
SPI code. spi_set_drvdata() is actually just a wrapper around
dev_set_drvdata().

Instead of clearing the pointer only in the spi code, perhaps the
checks should be in really_probe() and __device_release_driver() so it
covers all bus types. Also, don't clear the pointer. Just use
dev_err() to report on the driver bug so that it gets fixed properly.

g.

> ---
>  drivers/spi/spi.c | 14 ++++++++++++--
>  1 file changed, 12 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/spi/spi.c b/drivers/spi/spi.c
> index 84c2861..fe636fe 100644
> --- a/drivers/spi/spi.c
> +++ b/drivers/spi/spi.c
> @@ -233,15 +233,25 @@ EXPORT_SYMBOL_GPL(spi_bus_type);
>  static int spi_drv_probe(struct device *dev)
>  {
>         const struct spi_driver         *sdrv = to_spi_driver(dev->driver);
> +       struct spi_device               *sdev = to_spi_device(dev);
> +       int                             status;
>
> -       return sdrv->probe(to_spi_device(dev));
> +       status = sdrv->probe(sdev);
> +       if (status)
> +               spi_set_drvdata(sdev, NULL);
> +       return status;
>  }
>
>  static int spi_drv_remove(struct device *dev)
>  {
>         const struct spi_driver         *sdrv = to_spi_driver(dev->driver);
> +       struct spi_device               *sdev = to_spi_device(dev);
> +       int                             status;
>
> -       return sdrv->remove(to_spi_device(dev));
> +       status = sdrv->remove(sdev);
> +       if (status == 0)
> +               spi_set_drvdata(sdev, NULL);
> +       return status;
>  }
>
>  static void spi_drv_shutdown(struct device *dev)
> --
> 1.7.11.7
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/



--
Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.

Re: [PATCH RESEND] spi: erase pointer to drvdata on removal

[Vivien Didelot]

Hi Grant,

On Tue, 2012-12-04 at 14:56 +0000, Grant Likely wrote:
> Instead of clearing the pointer only in the spi code, perhaps the
> checks should be in really_probe() and __device_release_driver() so it
> covers all bus types. Also, don't clear the pointer. Just use
> dev_err() to report on the driver bug so that it gets fixed properly. 

I agree with you. I just figured out that on 2012-05-23, Hans de Goede
added a patch (0998d063) to enforce drvdata to be NULL after
device_release or on probe failure. Good to know.

Thanks,
Vivien