2008-06-20 10:58:19

by Ben Dooks

[permalink] [raw]
Subject: MMC: card test driver should not be builtin with other card support

The mmc test driver and mmc block driver will attempt to bind
to any card present in the system, which means only one of these
drivers will end up with the card. If either one of these is selected
as builtin, ensure the other does not get built.

Signed-off-by: Ben Dooks <[email protected]>

Index: linux-2.6.26-rc6-sdmmc3/drivers/mmc/card/Kconfig
===================================================================
--- linux-2.6.26-rc6-sdmmc3.orig/drivers/mmc/card/Kconfig 2008-06-20 11:53:08.000000000 +0100
+++ linux-2.6.26-rc6-sdmmc3/drivers/mmc/card/Kconfig 2008-06-20 11:53:29.000000000 +0100
@@ -6,7 +6,7 @@ comment "MMC/SD Card Drivers"

config MMC_BLOCK
tristate "MMC block device driver"
- depends on BLOCK
+ depends on BLOCK && MMC_TEST != y
default y
help
Say Y here to enable the MMC block device driver support.
@@ -42,6 +42,7 @@ config SDIO_UART
config MMC_TEST
tristate "MMC host test driver"
default n
+ depends on MMC_BLOCK != y
help
Development driver that performs a series of reads and writes
to a memory card in order to expose certain well known bugs

--
Ben ([email protected], http://www.fluff.org/)

'a smiley only costs 4 bytes'


2008-06-20 11:01:29

by Ben Dooks

[permalink] [raw]
Subject: Re: MMC: card test driver should not be builtin with other card support

On Fri, Jun 20, 2008 at 11:58:02AM +0100, [email protected] wrote:
> The mmc test driver and mmc block driver will attempt to bind
> to any card present in the system, which means only one of these
> drivers will end up with the card. If either one of these is selected
> as builtin, ensure the other does not get built.
>
> Signed-off-by: Ben Dooks <[email protected]>
>
> Index: linux-2.6.26-rc6-sdmmc3/drivers/mmc/card/Kconfig
> ===================================================================
> --- linux-2.6.26-rc6-sdmmc3.orig/drivers/mmc/card/Kconfig 2008-06-20 11:53:08.000000000 +0100
> +++ linux-2.6.26-rc6-sdmmc3/drivers/mmc/card/Kconfig 2008-06-20 11:53:29.000000000 +0100
> @@ -6,7 +6,7 @@ comment "MMC/SD Card Drivers"
>
> config MMC_BLOCK
> tristate "MMC block device driver"
> - depends on BLOCK
> + depends on BLOCK && MMC_TEST != y
> default y
> help
> Say Y here to enable the MMC block device driver support.
> @@ -42,6 +42,7 @@ config SDIO_UART
> config MMC_TEST
> tristate "MMC host test driver"
> default n
> + depends on MMC_BLOCK != y
> help
> Development driver that performs a series of reads and writes
> to a memory card in order to expose certain well known bugs

This is what I would have liked to have done in Kconfig, but of course
the system fails with a circular dependency error. Is there any way
to get this done without some horrible hacks?

--
Ben ([email protected], http://www.fluff.org/)

'a smiley only costs 4 bytes'

2008-06-20 17:08:44

by Pierre Ossman

[permalink] [raw]
Subject: Re: MMC: card test driver should not be builtin with other card support

On Fri, 20 Jun 2008 12:01:15 +0100
Ben Dooks <[email protected]> wrote:

> On Fri, Jun 20, 2008 at 11:58:02AM +0100, [email protected] wrote:
> > The mmc test driver and mmc block driver will attempt to bind
> > to any card present in the system, which means only one of these
> > drivers will end up with the card. If either one of these is selected
> > as builtin, ensure the other does not get built.
> >
>
> This is what I would have liked to have done in Kconfig, but of course
> the system fails with a circular dependency error. Is there any way
> to get this done without some horrible hacks?
>

AFAIK, you can manually control the driver binding using sysfs. Should
be solvable for the rootfs case using an initrd.

Rgds
--
-- Pierre Ossman

Linux kernel, MMC maintainer http://www.kernel.org
rdesktop, core developer http://www.rdesktop.org

WARNING: This correspondence is being monitored by the
Swedish government. Use end-to-end encryption where
possible.