This patch reverts commit 047b93a35961f7a6561e6f5dcb040738f822b892 which breaks
MAX7301 GPIO driver because that commit was dependant on a rejected patch that
was implementing selection of SPI speed from the Device Tree.
Signed-off-by: Christophe Leroy <[email protected]>
--- linux-3.11-rc6/drivers/gpio/gpio-max7301.c 2013-08-17 21:09:17.000000000 +0200
+++ linux/drivers/gpio/gpio-max7301.c 2013-08-18 06:31:52.000000000 +0200
@@ -56,8 +56,7 @@
int ret;
/* bits_per_word cannot be configured in platform data */
- if (spi->dev.platform_data)
- spi->bits_per_word = 16;
+ spi->bits_per_word = 16;
ret = spi_setup(spi);
if (ret < 0)
return ret;
On 08/20/2013 08:29 AM, Christophe Leroy wrote:
> This patch reverts commit 047b93a35961f7a6561e6f5dcb040738f822b892 which breaks
> MAX7301 GPIO driver because that commit was dependant on a rejected patch that
> was implementing selection of SPI speed from the Device Tree.
>
> Signed-off-by: Christophe Leroy <[email protected]>
Acked-by: Roland Stigge <[email protected]>
Good, thanks!
>
> --- linux-3.11-rc6/drivers/gpio/gpio-max7301.c 2013-08-17 21:09:17.000000000 +0200
> +++ linux/drivers/gpio/gpio-max7301.c 2013-08-18 06:31:52.000000000 +0200
> @@ -56,8 +56,7 @@
> int ret;
>
> /* bits_per_word cannot be configured in platform data */
> - if (spi->dev.platform_data)
> - spi->bits_per_word = 16;
> + spi->bits_per_word = 16;
> ret = spi_setup(spi);
> if (ret < 0)
> return ret;
>
On Tue, Aug 20, 2013 at 8:29 AM, Christophe Leroy
<[email protected]> wrote:
> This patch reverts commit 047b93a35961f7a6561e6f5dcb040738f822b892 which breaks
> MAX7301 GPIO driver because that commit was dependant on a rejected patch that
> was implementing selection of SPI speed from the Device Tree.
>
> Signed-off-by: Christophe Leroy <[email protected]>
Patch applied with Roland's ACK.
But seriously, this is the kind of stuff that scares me a lot,
when developers merge dependent patches into two different
trees, that is just a recipe for chaos and me getting flamed
by other kernel maintainers.
There is *no* mention of this dependency in the other
commit.
Yours,
Linus Walleij
Le 23/08/2013 19:47, Linus Walleij a ?crit :
> On Tue, Aug 20, 2013 at 8:29 AM, Christophe Leroy
> <[email protected]> wrote:
>
>> This patch reverts commit 047b93a35961f7a6561e6f5dcb040738f822b892 which breaks
>> MAX7301 GPIO driver because that commit was dependant on a rejected patch that
>> was implementing selection of SPI speed from the Device Tree.
>>
>> Signed-off-by: Christophe Leroy <[email protected]>
> Patch applied with Roland's ACK.
>
> But seriously, this is the kind of stuff that scares me a lot,
> when developers merge dependent patches into two different
> trees, that is just a recipe for chaos and me getting flamed
> by other kernel maintainers.
>
> There is *no* mention of this dependency in the other
> commit.
I'm very sorry for this mishap. If I didn't mention it in the commit,
this is because when I developped the change I submitted you a few
months back, I was not aware of that other patch. It looks like it has
been in my company's kernel tree for years, therefore in my mind it was
a standard feature of the kernel. It looks like nobody had tried and
submitted it for inclusion previously.
That's thanks to Roland that I discovered it was indeed not a standard
feature, then I tried to submit that patch a week ago and it was
rejected by Stephen Warren for good reason.
For me it is a lesson learnt, and I'll make sure it doesn't happen again.
Regards
Christophe