[PATCH] i2c-mv64xxx: use subsys_initcall for early initialization

[PATCH] i2c-mv64xxx: use subsys_initcall for early initialization

[Mike Rapoport]

Signed-off-by: Mike Rapoport <mike-UTxiZqZC01RS1MOuV/RT9w@public.gmane.org>
---
 drivers/i2c/busses/i2c-mv64xxx.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/i2c/busses/i2c-mv64xxx.c b/drivers/i2c/busses/i2c-mv64xxx.c
index 1624206..6f3d084 100644
--- a/drivers/i2c/busses/i2c-mv64xxx.c
+++ b/drivers/i2c/busses/i2c-mv64xxx.c
@@ -592,7 +592,7 @@ mv64xxx_i2c_exit(void)
 	platform_driver_unregister(&mv64xxx_i2c_driver);
 }
 
-module_init(mv64xxx_i2c_init);
+subsys_initcall(mv64xxx_i2c_init);
 module_exit(mv64xxx_i2c_exit);
 
 MODULE_AUTHOR("Mark A. Greer <mgreer-Igf4POYTYCDQT0dZR+AlfA@public.gmane.org>");
-- 
1.6.6.2

Re: [PATCH] i2c-mv64xxx: use subsys_initcall for early initialization

[Ben Dooks]

On Sun, Sep 12, 2010 at 11:35:29AM +0200, Mike Rapoport wrote:
> Signed-off-by: Mike Rapoport <mike-UTxiZqZC01RS1MOuV/RT9w@public.gmane.org>
> ---
>  drivers/i2c/busses/i2c-mv64xxx.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)

We really should have an i2c_initcall, but it may be difficult to move
everyone over to it. What would be lovely is some form of dependency
based initialisation, but that's probably a way off...

I'll consider this for the next merge window.
 
> diff --git a/drivers/i2c/busses/i2c-mv64xxx.c b/drivers/i2c/busses/i2c-mv64xxx.c
> index 1624206..6f3d084 100644
> --- a/drivers/i2c/busses/i2c-mv64xxx.c
> +++ b/drivers/i2c/busses/i2c-mv64xxx.c
> @@ -592,7 +592,7 @@ mv64xxx_i2c_exit(void)
>  	platform_driver_unregister(&mv64xxx_i2c_driver);
>  }
>  
> -module_init(mv64xxx_i2c_init);
> +subsys_initcall(mv64xxx_i2c_init);
>  module_exit(mv64xxx_i2c_exit);
>  
>  MODULE_AUTHOR("Mark A. Greer <mgreer-Igf4POYTYCDQT0dZR+AlfA@public.gmane.org>");

-- 
Ben

Q:      What's a light-year?
A:      One-third less calories than a regular year.

Re: [PATCH] i2c-mv64xxx: use subsys_initcall for early initialization

[Mike Rapoport]

Ben Dooks wrote:
> On Sun, Sep 12, 2010 at 11:35:29AM +0200, Mike Rapoport wrote:
>> Signed-off-by: Mike Rapoport <mike-UTxiZqZC01RS1MOuV/RT9w@public.gmane.org>
>> ---
>>  drivers/i2c/busses/i2c-mv64xxx.c |    2 +-
>>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> We really should have an i2c_initcall, but it may be difficult to move
> everyone over to it. What would be lovely is some form of dependency
> based initialisation, but that's probably a way off...

Yeah, that would be nice...

> I'll consider this for the next merge window.

Thanks

>> diff --git a/drivers/i2c/busses/i2c-mv64xxx.c b/drivers/i2c/busses/i2c-mv64xxx.c
>> index 1624206..6f3d084 100644
>> --- a/drivers/i2c/busses/i2c-mv64xxx.c
>> +++ b/drivers/i2c/busses/i2c-mv64xxx.c
>> @@ -592,7 +592,7 @@ mv64xxx_i2c_exit(void)
>>  	platform_driver_unregister(&mv64xxx_i2c_driver);
>>  }
>>  
>> -module_init(mv64xxx_i2c_init);
>> +subsys_initcall(mv64xxx_i2c_init);
>>  module_exit(mv64xxx_i2c_exit);
>>  
>>  MODULE_AUTHOR("Mark A. Greer <mgreer-Igf4POYTYCDQT0dZR+AlfA@public.gmane.org>");
> 


-- 
Sincerely yours,
Mike.

Re: [PATCH] i2c-mv64xxx: use subsys_initcall for early initialization

[Mark Brown]

On Wed, Sep 15, 2010 at 03:43:08PM +0200, Mike Rapoport wrote:
> Ben Dooks wrote:

>>>  drivers/i2c/busses/i2c-mv64xxx.c |    2 +-
>>>  1 files changed, 1 insertions(+), 1 deletions(-)

>> We really should have an i2c_initcall, but it may be difficult to move
>> everyone over to it. What would be lovely is some form of dependency
>> based initialisation, but that's probably a way off...

> Yeah, that would be nice...

Grant Likely was muttering about contributing something here.

Re: [PATCH] i2c-mv64xxx: use subsys_initcall for early initialization

[Linus Walleij]

2010/9/15 Ben Dooks <ben-i2c-elnMNo+KYs3YtjvyW6yDsg@public.gmane.org>:
> On Sun, Sep 12, 2010 at 11:35:29AM +0200, Mike Rapoport wrote:
>> Signed-off-by: Mike Rapoport <mike-UTxiZqZC01RS1MOuV/RT9w@public.gmane.org>
>> ---
>>  drivers/i2c/busses/i2c-mv64xxx.c |    2 +-
>>  1 files changed, 1 insertions(+), 1 deletions(-)
>
> We really should have an i2c_initcall, but it may be difficult to move
> everyone over to it. What would be lovely is some form of dependency
> based initialisation, but that's probably a way off...

SPI, SDIO and other external busses (Dallas 1-wire? w1) may all have
the same "needs to be up early because we need that before that"
type of problem.

Makes me think of crazy things like putting a lightweight version of
Lennars Pötterings "systemd" scheme into the kernel for resolving
these initcalls.

Yours,
Linus Walleij

Re: [PATCH] i2c-mv64xxx: use subsys_initcall for early initialization

[Mark Brown]

On Thu, Sep 23, 2010 at 09:11:00AM +0200, Linus Walleij wrote:

> Makes me think of crazy things like putting a lightweight version of
> Lennars P?tterings "systemd" scheme into the kernel for resolving
> these initcalls.

We pretty much need that; it's not just the initcalls - we also need the
probes to go through in the right order, we're just using the initcall
fiddling to bodge that for now with particularly critical system
components.

Re: [PATCH] i2c-mv64xxx: use subsys_initcall for early initialization

[Ben Dooks]

On Thu, Sep 23, 2010 at 09:11:00AM +0200, Linus Walleij wrote:
> 2010/9/15 Ben Dooks <ben-i2c-elnMNo+KYs3YtjvyW6yDsg@public.gmane.org>:
> > On Sun, Sep 12, 2010 at 11:35:29AM +0200, Mike Rapoport wrote:
> >> Signed-off-by: Mike Rapoport <mike-UTxiZqZC01RS1MOuV/RT9w@public.gmane.org>
> >> ---
> >>  drivers/i2c/busses/i2c-mv64xxx.c |    2 +-
> >>  1 files changed, 1 insertions(+), 1 deletions(-)
> >
> > We really should have an i2c_initcall, but it may be difficult to move
> > everyone over to it. What would be lovely is some form of dependency
> > based initialisation, but that's probably a way off...
> 
> SPI, SDIO and other external busses (Dallas 1-wire? w1) may all have
> the same "needs to be up early because we need that before that"
> type of problem.
> 
> Makes me think of crazy things like putting a lightweight version of
> Lennars Pötterings "systemd" scheme into the kernel for resolving
> these initcalls.

Either that, or build stuff as modules and use an initrd...

Not sure what the best solution is...

-- 
Ben

Q:      What's a light-year?
A:      One-third less calories than a regular year.