arch_initcall time

arch_initcall time

[Sean MacLennan]

I need to call i2c_register_board_info for the new i2c style ad7414 
driver. This needs to be called at arch initcall time. Currently I just 
do this:

static int __init warp_arch_init(void)
{
	i2c_register_board_info(0, warp_i2c_info, ARRAY_SIZE(warp_i2c_info));
	return 0;
}
arch_initcall(warp_arch_init);


It works, but is there a "better" place to put this? None of the other 
powerpc platforms make this call and I want to get it right, so that 
others don't blindly follow my example ;)

I kept the name vague rather than specific in case more drivers need to 
be setup this way in the future.

Cheers,
   Sean

Re: arch_initcall time

[Grant Likely]

On Feb 18, 2008 11:28 AM, Sean MacLennan <smaclennan@pikatech.com> wrote:
> I need to call i2c_register_board_info for the new i2c style ad7414
> driver. This needs to be called at arch initcall time. Currently I just
> do this:
>
> static int __init warp_arch_init(void)
> {
>         i2c_register_board_info(0, warp_i2c_info, ARRAY_SIZE(warp_i2c_info));
>         return 0;
> }
> arch_initcall(warp_arch_init);

Yes, this is the right thing to do, but use machine_arch_initcall()
instead so that it doesn't get called if it is not your board.

Cheers,
g.

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

Re: arch_initcall time

[Grant Likely]

On Feb 18, 2008 11:31 AM, Grant Likely <grant.likely@secretlab.ca> wrote:
> On Feb 18, 2008 11:28 AM, Sean MacLennan <smaclennan@pikatech.com> wrote:
> > I need to call i2c_register_board_info for the new i2c style ad7414
> > driver. This needs to be called at arch initcall time. Currently I just
> > do this:
> >
> > static int __init warp_arch_init(void)
> > {
> >         i2c_register_board_info(0, warp_i2c_info, ARRAY_SIZE(warp_i2c_info));
> >         return 0;
> > }
> > arch_initcall(warp_arch_init);
>
> Yes, this is the right thing to do, but use machine_arch_initcall()
> instead so that it doesn't get called if it is not your board.

That being said, I believe there is infrastructure to handle the
creation of your i2c board info from the device tree.  Your i2c board
info should not be hard coded.

Cheers,
g.

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

Re: arch_initcall time

[Olof Johansson]

On Mon, Feb 18, 2008 at 11:32:14AM -0700, Grant Likely wrote:
> On Feb 18, 2008 11:31 AM, Grant Likely <grant.likely@secretlab.ca> wrote:
> > On Feb 18, 2008 11:28 AM, Sean MacLennan <smaclennan@pikatech.com> wrote:
> > > I need to call i2c_register_board_info for the new i2c style ad7414
> > > driver. This needs to be called at arch initcall time. Currently I just
> > > do this:
> > >
> > > static int __init warp_arch_init(void)
> > > {
> > >         i2c_register_board_info(0, warp_i2c_info, ARRAY_SIZE(warp_i2c_info));
> > >         return 0;
> > > }
> > > arch_initcall(warp_arch_init);
> >
> > Yes, this is the right thing to do, but use machine_arch_initcall()
> > instead so that it doesn't get called if it is not your board.
> 
> That being said, I believe there is infrastructure to handle the
> creation of your i2c board info from the device tree.  Your i2c board
> info should not be hard coded.

Jon Smirl's patches? Not yet, unfortunately. It didn't make .25, but
maybe for .26.

(I will need to do it specifically on my platform, like fsl_soc already
does, as a stopgap until then).


-Olof

Re: arch_initcall time

[Josh Boyer]

On Mon, 18 Feb 2008 12:42:40 -0600
Olof Johansson <olof@lixom.net> wrote:

> On Mon, Feb 18, 2008 at 11:32:14AM -0700, Grant Likely wrote:
> > On Feb 18, 2008 11:31 AM, Grant Likely <grant.likely@secretlab.ca> wrote:
> > > On Feb 18, 2008 11:28 AM, Sean MacLennan <smaclennan@pikatech.com> wrote:
> > > > I need to call i2c_register_board_info for the new i2c style ad7414
> > > > driver. This needs to be called at arch initcall time. Currently I just
> > > > do this:
> > > >
> > > > static int __init warp_arch_init(void)
> > > > {
> > > >         i2c_register_board_info(0, warp_i2c_info, ARRAY_SIZE(warp_i2c_info));
> > > >         return 0;
> > > > }
> > > > arch_initcall(warp_arch_init);
> > >
> > > Yes, this is the right thing to do, but use machine_arch_initcall()
> > > instead so that it doesn't get called if it is not your board.
> > 
> > That being said, I believe there is infrastructure to handle the
> > creation of your i2c board info from the device tree.  Your i2c board
> > info should not be hard coded.
> 
> Jon Smirl's patches? Not yet, unfortunately. It didn't make .25, but
> maybe for .26.
> 
> (I will need to do it specifically on my platform, like fsl_soc already
> does, as a stopgap until then).

That, and Sean is still working on getting the iic device-tree-compliant driver through as
well :)

josh