Several questions regarding ARCH_S3C2410

Several questions regarding ARCH_S3C2410

[Eric Miao]

Hi Ben,

While working towards a unified kernel, I found that ARCH_S3C2410 actually
includes multiple mach-* already.

linux-2.6$ git grep "ARCH_S3C2410" arch/arm/Makefile
arch/arm/Makefile:machine-$(CONFIG_ARCH_S3C2410)	:= s3c2410 s3c2400
s3c2412 s3c2416 s3c2440 s3c2443

There are several questions though:

1. It looks like only s3c2410 is significant, so all the common definitions
   like PHYS_OFFSET are actually same across all mach-s3c24* (except for
   s3c24a0) ? Yet I've found arch/arm/mach-s3c2400/include/mach/memory.h,
   where a different PHYS_OFFSET is defined, just feel a bit confused.

2. So except for s3c24a0, theretically, it's possible to combine all these
   mach into something mach-s3c24xx?

3. When trying building of s3c2410_defconfig, there are a lot complaints
   about re-definitions of {in,out}[s]{b,w,l}.  Taking a look into io.h,
   I'm wondering if it's possible to use those definitions in asm/io.h?
   (possibly with a customized __io())

In file included from linux-2.6/include/linux/scatterlist.h:8,
                 from linux-2.6/include/linux/dma-mapping.h:7,
                 from linux-2.6/include/linux/dmaengine.h:26,
                 from linux-2.6/include/linux/skbuff.h:30,
                 from linux-2.6/lib/kobject_uevent.c:24:
linux-2.6/arch/arm/include/asm/io.h:137: warning: "outb" redefined
linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:196: note: this is
the location of the previous definition
linux-2.6/arch/arm/include/asm/io.h:138: warning: "outw" redefined
linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:197: note: this is
the location of the previous definition
linux-2.6/arch/arm/include/asm/io.h:140: warning: "outl" redefined
linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:198: note: this is
the location of the previous definition
linux-2.6/arch/arm/include/asm/io.h:143: warning: "inb" redefined
linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:193: note: this is
the location of the previous definition
linux-2.6/arch/arm/include/asm/io.h:144: warning: "inw" redefined
linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:194: note: this is
the location of the previous definition
linux-2.6/arch/arm/include/asm/io.h:146: warning: "inl" redefined
linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:195: note: this is
the location of the previous definition
linux-2.6/arch/arm/include/asm/io.h:149: warning: "outsb" redefined
linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:207: note: this is
the location of the previous definition
linux-2.6/arch/arm/include/asm/io.h:150: warning: "outsw" redefined
linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:208: note: this is
the location of the previous definition
linux-2.6/arch/arm/include/asm/io.h:151: warning: "outsl" redefined
linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:209: note: this is
the location of the previous definition
linux-2.6/arch/arm/include/asm/io.h:153: warning: "insb" redefined
linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:203: note: this is
the location of the previous definition
linux-2.6/arch/arm/include/asm/io.h:154: warning: "insw" redefined
linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:204: note: this is
the location of the previous definition
linux-2.6/arch/arm/include/asm/io.h:155: warning: "insl" redefined
linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:205: note: this is
the location of the previous definition

Several questions regarding ARCH_S3C2410

[Kyungmin Park]

CCing Mr. Kim

On Fri, Jun 18, 2010 at 12:43 PM, Eric Miao <eric.y.miao@gmail.com> wrote:
> Hi Ben,
>
> While working towards a unified kernel, I found that ARCH_S3C2410 actually
> includes multiple mach-* already.
>
> linux-2.6$ git grep "ARCH_S3C2410" arch/arm/Makefile
> arch/arm/Makefile:machine-$(CONFIG_ARCH_S3C2410) ? ? ? ?:= s3c2410 s3c2400
> s3c2412 s3c2416 s3c2440 s3c2443
>
> There are several questions though:
>
> 1. It looks like only s3c2410 is significant, so all the common definitions
> ? like PHYS_OFFSET are actually same across all mach-s3c24* (except for
> ? s3c24a0) ? Yet I've found arch/arm/mach-s3c2400/include/mach/memory.h,
> ? where a different PHYS_OFFSET is defined, just feel a bit confused.
>
> 2. So except for s3c24a0, theretically, it's possible to combine all these
> ? mach into something mach-s3c24xx?
>
> 3. When trying building of s3c2410_defconfig, there are a lot complaints
> ? about re-definitions of {in,out}[s]{b,w,l}. ?Taking a look into io.h,
> ? I'm wondering if it's possible to use those definitions in asm/io.h?
> ? (possibly with a customized __io())
>
> In file included from linux-2.6/include/linux/scatterlist.h:8,
> ? ? ? ? ? ? ? ? from linux-2.6/include/linux/dma-mapping.h:7,
> ? ? ? ? ? ? ? ? from linux-2.6/include/linux/dmaengine.h:26,
> ? ? ? ? ? ? ? ? from linux-2.6/include/linux/skbuff.h:30,
> ? ? ? ? ? ? ? ? from linux-2.6/lib/kobject_uevent.c:24:
> linux-2.6/arch/arm/include/asm/io.h:137: warning: "outb" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:196: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:138: warning: "outw" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:197: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:140: warning: "outl" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:198: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:143: warning: "inb" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:193: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:144: warning: "inw" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:194: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:146: warning: "inl" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:195: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:149: warning: "outsb" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:207: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:150: warning: "outsw" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:208: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:151: warning: "outsl" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:209: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:153: warning: "insb" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:203: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:154: warning: "insw" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:204: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:155: warning: "insl" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:205: note: this is
> the location of the previous definition
>
> _______________________________________________
> linux-arm-kernel mailing list
> linux-arm-kernel at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
>

Several questions regarding ARCH_S3C2410

[Ben Dooks]

On Fri, Jun 18, 2010 at 11:43:37AM +0800, Eric Miao wrote:
> Hi Ben,
> 
> While working towards a unified kernel, I found that ARCH_S3C2410 actually
> includes multiple mach-* already.
> 
> linux-2.6$ git grep "ARCH_S3C2410" arch/arm/Makefile
> arch/arm/Makefile:machine-$(CONFIG_ARCH_S3C2410)	:= s3c2410 s3c2400
> s3c2412 s3c2416 s3c2440 s3c2443
> 
> There are several questions though:
> 
> 1. It looks like only s3c2410 is significant, so all the common definitions
>    like PHYS_OFFSET are actually same across all mach-s3c24* (except for
>    s3c24a0) ? Yet I've found arch/arm/mach-s3c2400/include/mach/memory.h,
>    where a different PHYS_OFFSET is defined, just feel a bit confused.

You can safely ignore the s3c24a0 for the moment, there's no-one interested
in actually getting it submitted.
 
> 2. So except for s3c24a0, theretically, it's possible to combine all these
>    mach into something mach-s3c24xx?

Theoretically yes, practically leave it alone.

> 3. When trying building of s3c2410_defconfig, there are a lot complaints
>    about re-definitions of {in,out}[s]{b,w,l}.  Taking a look into io.h,
>    I'm wondering if it's possible to use those definitions in asm/io.h?
>    (possibly with a customized __io())
> 
> In file included from linux-2.6/include/linux/scatterlist.h:8,
>                  from linux-2.6/include/linux/dma-mapping.h:7,
>                  from linux-2.6/include/linux/dmaengine.h:26,
>                  from linux-2.6/include/linux/skbuff.h:30,
>                  from linux-2.6/lib/kobject_uevent.c:24:
> linux-2.6/arch/arm/include/asm/io.h:137: warning: "outb" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:196: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:138: warning: "outw" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:197: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:140: warning: "outl" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:198: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:143: warning: "inb" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:193: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:144: warning: "inw" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:194: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:146: warning: "inl" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:195: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:149: warning: "outsb" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:207: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:150: warning: "outsw" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:208: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:151: warning: "outsl" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:209: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:153: warning: "insb" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:203: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:154: warning: "insw" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:204: note: this is
> the location of the previous definition
> linux-2.6/arch/arm/include/asm/io.h:155: warning: "insl" redefined
> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:205: note: this is
> the location of the previous definition

that's something for you to sort out.

-- 
Ben

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

Several questions regarding ARCH_S3C2410

[Ben Dooks]

On Fri, Jun 18, 2010 at 04:44:59PM +0900, Kyungmin Park wrote:
> CCing Mr. Kim

not sure it really needs him to be involved at-all

-- 
Ben

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

Several questions regarding ARCH_S3C2410

[Eric Miao]

On Fri, Jun 18, 2010 at 5:00 PM, Ben Dooks <ben-linux@fluff.org> wrote:
> On Fri, Jun 18, 2010 at 11:43:37AM +0800, Eric Miao wrote:
>> Hi Ben,
>>
>> While working towards a unified kernel, I found that ARCH_S3C2410 actually
>> includes multiple mach-* already.
>>
>> linux-2.6$ git grep "ARCH_S3C2410" arch/arm/Makefile
>> arch/arm/Makefile:machine-$(CONFIG_ARCH_S3C2410) ? ? ?:= s3c2410 s3c2400
>> s3c2412 s3c2416 s3c2440 s3c2443
>>
>> There are several questions though:
>>
>> 1. It looks like only s3c2410 is significant, so all the common definitions
>> ? ?like PHYS_OFFSET are actually same across all mach-s3c24* (except for
>> ? ?s3c24a0) ? Yet I've found arch/arm/mach-s3c2400/include/mach/memory.h,
>> ? ?where a different PHYS_OFFSET is defined, just feel a bit confused.
>
> You can safely ignore the s3c24a0 for the moment, there's no-one interested
> in actually getting it submitted.

Grrr... indeed.Looking into mach-s3c24a0, it's almost empty except for some
header files.

>
>> 2. So except for s3c24a0, theretically, it's possible to combine all these
>> ? ?mach into something mach-s3c24xx?
>
> Theoretically yes, practically leave it alone.
>
>> 3. When trying building of s3c2410_defconfig, there are a lot complaints
>> ? ?about re-definitions of {in,out}[s]{b,w,l}. ?Taking a look into io.h,
>> ? ?I'm wondering if it's possible to use those definitions in asm/io.h?
>> ? ?(possibly with a customized __io())
>>
>> In file included from linux-2.6/include/linux/scatterlist.h:8,
>> ? ? ? ? ? ? ? ? ?from linux-2.6/include/linux/dma-mapping.h:7,
>> ? ? ? ? ? ? ? ? ?from linux-2.6/include/linux/dmaengine.h:26,
>> ? ? ? ? ? ? ? ? ?from linux-2.6/include/linux/skbuff.h:30,
>> ? ? ? ? ? ? ? ? ?from linux-2.6/lib/kobject_uevent.c:24:
>> linux-2.6/arch/arm/include/asm/io.h:137: warning: "outb" redefined
>> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:196: note: this is
>> the location of the previous definition
>> linux-2.6/arch/arm/include/asm/io.h:138: warning: "outw" redefined
>> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:197: note: this is
>> the location of the previous definition
>> linux-2.6/arch/arm/include/asm/io.h:140: warning: "outl" redefined
>> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:198: note: this is
>> the location of the previous definition
>> linux-2.6/arch/arm/include/asm/io.h:143: warning: "inb" redefined
>> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:193: note: this is
>> the location of the previous definition
>> linux-2.6/arch/arm/include/asm/io.h:144: warning: "inw" redefined
>> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:194: note: this is
>> the location of the previous definition
>> linux-2.6/arch/arm/include/asm/io.h:146: warning: "inl" redefined
>> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:195: note: this is
>> the location of the previous definition
>> linux-2.6/arch/arm/include/asm/io.h:149: warning: "outsb" redefined
>> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:207: note: this is
>> the location of the previous definition
>> linux-2.6/arch/arm/include/asm/io.h:150: warning: "outsw" redefined
>> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:208: note: this is
>> the location of the previous definition
>> linux-2.6/arch/arm/include/asm/io.h:151: warning: "outsl" redefined
>> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:209: note: this is
>> the location of the previous definition
>> linux-2.6/arch/arm/include/asm/io.h:153: warning: "insb" redefined
>> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:203: note: this is
>> the location of the previous definition
>> linux-2.6/arch/arm/include/asm/io.h:154: warning: "insw" redefined
>> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:204: note: this is
>> the location of the previous definition
>> linux-2.6/arch/arm/include/asm/io.h:155: warning: "insl" redefined
>> linux-2.6/arch/arm/mach-s3c2410/include/mach/io.h:205: note: this is
>> the location of the previous definition
>
> that's something for you to sort out.
>

Come on, you must be kidding me. I have no idea how that crap was
being formed :-)

Several questions regarding ARCH_S3C2410

[Kukjin Kim]

Ben Dooks wrote:
> 
> On Fri, Jun 18, 2010 at 04:44:59PM +0900, Kyungmin Park wrote:
> > CCing Mr. Kim
> 
> not sure it really needs him to be involved at-all
> 
I'd like to focus on S5P SoCs and Samsung common(plat-samsung) :-)

And maybe Ben is more interested in S3C SoCs and Samsung common.

Thanks.

Best regards,
Kgene.
--
Kukjin Kim <kgene.kim@samsung.com>, Senior Engineer,
SW Solution Development Team, Samsung Electronics Co., Ltd.

Several questions regarding ARCH_S3C2410

[Nicolas Pitre]

On Fri, 18 Jun 2010, Eric Miao wrote:

> On Fri, Jun 18, 2010 at 5:00 PM, Ben Dooks <ben-linux@fluff.org> wrote:
> > On Fri, Jun 18, 2010 at 11:43:37AM +0800, Eric Miao wrote:
> >> Hi Ben,
> >>
> >> While working towards a unified kernel, I found that ARCH_S3C2410 actually
> >> includes multiple mach-* already.
> >>
> >> linux-2.6$ git grep "ARCH_S3C2410" arch/arm/Makefile
> >> arch/arm/Makefile:machine-$(CONFIG_ARCH_S3C2410) ? ? ?:= s3c2410 s3c2400
> >> s3c2412 s3c2416 s3c2440 s3c2443
> >>
> >> There are several questions though:
> >>
> >> 1. It looks like only s3c2410 is significant, so all the common definitions
> >> ? ?like PHYS_OFFSET are actually same across all mach-s3c24* (except for
> >> ? ?s3c24a0) ? Yet I've found arch/arm/mach-s3c2400/include/mach/memory.h,
> >> ? ?where a different PHYS_OFFSET is defined, just feel a bit confused.
> >
> > You can safely ignore the s3c24a0 for the moment, there's no-one interested
> > in actually getting it submitted.
> 
> Grrr... indeed.Looking into mach-s3c24a0, it's almost empty except for some
> header files.

What about simply removing that useless stuff from the tree then?

It could always be retrieved from the Git history if someone wants to 
revive it.


Nicolas

Several questions regarding ARCH_S3C2410

[Kyungmin Park]

Then how about to prepare the these activities.

Next time we will use the s5pc210 (or whatever) and I hope to use it
with s5pc110.
To to this how about to merge the s5p series?

I suggest to read the OMAP mailing list how to support OMAP2, 3, and 4
with single kernel.

Thank you,
Kyungmin Park

On Fri, Jun 18, 2010 at 8:00 PM, Kukjin Kim <kgene.kim@samsung.com> wrote:
> Ben Dooks wrote:
>>
>> On Fri, Jun 18, 2010 at 04:44:59PM +0900, Kyungmin Park wrote:
>> > CCing Mr. Kim
>>
>> not sure it really needs him to be involved at-all
>>
> I'd like to focus on S5P SoCs and Samsung common(plat-samsung) :-)
>
> And maybe Ben is more interested in S3C SoCs and Samsung common.
>
> Thanks.
>
> Best regards,
> Kgene.
> --
> Kukjin Kim <kgene.kim@samsung.com>, Senior Engineer,
> SW Solution Development Team, Samsung Electronics Co., Ltd.
>
>

Several questions regarding ARCH_S3C2410

[Eric Miao]

On Fri, Jun 18, 2010 at 10:21 PM, Kyungmin Park <kmpark@infradead.org> wrote:
> Then how about to prepare the these activities.
>
> Next time we will use the s5pc210 (or whatever) and I hope to use it
> with s5pc110.

Better merge them into one, no more mach-* is recommended unless that's
inevitable.

> To to this how about to merge the s5p series?
>
> I suggest to read the OMAP mailing list how to support OMAP2, 3, and 4
> with single kernel.
>

Not necessary maybe. Just post your difficulties when merging them
into a single kernel and we can discuss.

Several questions regarding ARCH_S3C2410

[Kyungmin Park]

On Fri, Jun 18, 2010 at 11:28 PM, Eric Miao <eric.y.miao@gmail.com> wrote:
> On Fri, Jun 18, 2010 at 10:21 PM, Kyungmin Park <kmpark@infradead.org> wrote:
>> Then how about to prepare the these activities.
>>
>> Next time we will use the s5pc210 (or whatever) and I hope to use it
>> with s5pc110.
>
> Better merge them into one, no more mach-* is recommended unless that's
> inevitable.

Agree.

>
>> To to this how about to merge the s5p series?
>>
>> I suggest to read the OMAP mailing list how to support OMAP2, 3, and 4
>> with single kernel.
>>
>
> Not necessary maybe. Just post your difficulties when merging them
> into a single kernel and we can discuss.
>

I hope so, before starting this work, I make a consensus with ben and kukjin.
I like the OMAP style, runtime detection. IMHO Samsung SoCs goes like
these approaches.

Thank you,
Kyungmin Park