[PATCH] s5pv210: Change the base ram address to 0x3000'0000

[PATCH] s5pv210: Change the base ram address to 0x3000'0000

[Kyungmin Park]

s5pc110 (aka s5pv210) has 2 DRAM port and used it both usually.
Assume DMC0 starts with 0x2000'0000 with 128MiB.
DMC1 starts with 0x4000'0000 with 128MiB.
Note that DMC1 has to start address 0x4000'0000 at least.

Then there's too much memory hole 0x1800'0000 (128MiB + 256MiB)

To reduce memory waste, the DMC0 start with 0x3000'0000.

Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
---
diff --git a/arch/arm/mach-s5pv210/Makefile.boot b/arch/arm/mach-s5pv210/Makefile.boot
index ff90aa1..b0909e3 100644
--- a/arch/arm/mach-s5pv210/Makefile.boot
+++ b/arch/arm/mach-s5pv210/Makefile.boot
@@ -1,2 +1,2 @@
-   zreladdr-y	:= 0x20008000
-params_phys-y	:= 0x20000100
+   zreladdr-y	:= 0x30008000
+params_phys-y	:= 0x30000100
diff --git a/arch/arm/mach-s5pv210/include/mach/map.h b/arch/arm/mach-s5pv210/include/mach/map.h
index 34eb168..9f4c368 100644
--- a/arch/arm/mach-s5pv210/include/mach/map.h
+++ b/arch/arm/mach-s5pv210/include/mach/map.h
@@ -74,7 +74,7 @@
 #define S5PV210_PA_VIC3		(0xF2300000)
 #define S5P_PA_VIC3		S5PV210_PA_VIC3
 
-#define S5PV210_PA_SDRAM	(0x20000000)
+#define S5PV210_PA_SDRAM	(0x30000000)
 #define S5P_PA_SDRAM		S5PV210_PA_SDRAM
 
 /* I2S */
diff --git a/arch/arm/mach-s5pv210/include/mach/memory.h b/arch/arm/mach-s5pv210/include/mach/memory.h
index 379117e..07a15d8 100644
--- a/arch/arm/mach-s5pv210/include/mach/memory.h
+++ b/arch/arm/mach-s5pv210/include/mach/memory.h
@@ -13,7 +13,7 @@
 #ifndef __ASM_ARCH_MEMORY_H
 #define __ASM_ARCH_MEMORY_H
 
-#define PHYS_OFFSET		UL(0x20000000)
+#define PHYS_OFFSET		UL(0x30000000)
 #define CONSISTENT_DMA_SIZE	(SZ_8M + SZ_4M + SZ_2M)
 
 /* Maximum of 256MiB in one bank */

[PATCH] s5pv210: Change the base ram address to 0x3000'0000

[Ben Dooks]

On Fri, May 28, 2010 at 11:35:59AM +0900, Kyungmin Park wrote:
> s5pc110 (aka s5pv210) has 2 DRAM port and used it both usually.
> Assume DMC0 starts with 0x2000'0000 with 128MiB.
> DMC1 starts with 0x4000'0000 with 128MiB.
> Note that DMC1 has to start address 0x4000'0000 at least.
> 
> Then there's too much memory hole 0x1800'0000 (128MiB + 256MiB)
> 
> To reduce memory waste, the DMC0 start with 0x3000'0000.

We should be using SPARSEMEM now, is it not reducing any wastage
in virtual memory space?

How will the user change their board to deal with this change to hte
kernel?
 
> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
> ---
> diff --git a/arch/arm/mach-s5pv210/Makefile.boot b/arch/arm/mach-s5pv210/Makefile.boot
> index ff90aa1..b0909e3 100644
> --- a/arch/arm/mach-s5pv210/Makefile.boot
> +++ b/arch/arm/mach-s5pv210/Makefile.boot
> @@ -1,2 +1,2 @@
> -   zreladdr-y	:= 0x20008000
> -params_phys-y	:= 0x20000100
> +   zreladdr-y	:= 0x30008000
> +params_phys-y	:= 0x30000100
> diff --git a/arch/arm/mach-s5pv210/include/mach/map.h b/arch/arm/mach-s5pv210/include/mach/map.h
> index 34eb168..9f4c368 100644
> --- a/arch/arm/mach-s5pv210/include/mach/map.h
> +++ b/arch/arm/mach-s5pv210/include/mach/map.h
> @@ -74,7 +74,7 @@
>  #define S5PV210_PA_VIC3		(0xF2300000)
>  #define S5P_PA_VIC3		S5PV210_PA_VIC3
>  
> -#define S5PV210_PA_SDRAM	(0x20000000)
> +#define S5PV210_PA_SDRAM	(0x30000000)
>  #define S5P_PA_SDRAM		S5PV210_PA_SDRAM
>  
>  /* I2S */
> diff --git a/arch/arm/mach-s5pv210/include/mach/memory.h b/arch/arm/mach-s5pv210/include/mach/memory.h
> index 379117e..07a15d8 100644
> --- a/arch/arm/mach-s5pv210/include/mach/memory.h
> +++ b/arch/arm/mach-s5pv210/include/mach/memory.h
> @@ -13,7 +13,7 @@
>  #ifndef __ASM_ARCH_MEMORY_H
>  #define __ASM_ARCH_MEMORY_H
>  
> -#define PHYS_OFFSET		UL(0x20000000)
> +#define PHYS_OFFSET		UL(0x30000000)
>  #define CONSISTENT_DMA_SIZE	(SZ_8M + SZ_4M + SZ_2M)
>  
>  /* Maximum of 256MiB in one bank */

-- 
-- 
Ben

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

[PATCH] s5pv210: Change the base ram address to 0x3000'0000

[Kyungmin Park]

On Fri, May 28, 2010 at 12:53 PM, Ben Dooks <ben-linux@fluff.org> wrote:
> On Fri, May 28, 2010 at 11:35:59AM +0900, Kyungmin Park wrote:
>> s5pc110 (aka s5pv210) has 2 DRAM port and used it both usually.
>> Assume DMC0 starts with 0x2000'0000 with 128MiB.
>> DMC1 starts with 0x4000'0000 with 128MiB.
>> Note that DMC1 has to start address 0x4000'0000 at least.
>>
>> Then there's too much memory hole 0x1800'0000 (128MiB + 256MiB)
>>
>> To reduce memory waste, the DMC0 start with 0x3000'0000.

Hi,

>
> We should be using SPARSEMEM now, is it not reducing any wastage
> in virtual memory space?

Please consider the older kernel also. and most systems are designed
for 0x3000'0000.

>
> How will the user change their board to deal with this change to hte
> kernel?

Now we got the these memory
DMC0 has 128MiB and DMC1 has 512MiB.
Then it uses the memory from 0x2000'0000 to 0x5fff'ffff. it used 1GiB
memory address space.
Of course we can enable thie HIGHMEM but older kernel doesn't support it fully.
Also I hope to use the 3G/1G memory scheme.

How to you think?

Thank you,
Kyungmin Park

[PATCH] s5pv210: Change the base ram address to 0x3000'0000

[Ben Dooks]

On Fri, May 28, 2010 at 01:02:24PM +0900, Kyungmin Park wrote:
> On Fri, May 28, 2010 at 12:53 PM, Ben Dooks <ben-linux@fluff.org> wrote:
> > On Fri, May 28, 2010 at 11:35:59AM +0900, Kyungmin Park wrote:
> >> s5pc110 (aka s5pv210) has 2 DRAM port and used it both usually.
> >> Assume DMC0 starts with 0x2000'0000 with 128MiB.
> >> DMC1 starts with 0x4000'0000 with 128MiB.
> >> Note that DMC1 has to start address 0x4000'0000 at least.
> >>
> >> Then there's too much memory hole 0x1800'0000 (128MiB + 256MiB)
> >>
> >> To reduce memory waste, the DMC0 start with 0x3000'0000.

but you just said DMC0 starts at 0x2... ? Is this reconfigurable via
software, if so can it be moved once initialisation si done? how fixed
is this?
 
> Hi,
> 
> >
> > We should be using SPARSEMEM now, is it not reducing any wastage
> > in virtual memory space?
> 
> Please consider the older kernel also. and most systems are designed
> for 0x3000'0000.

Ok, will a kernel with the base-address set to 0x3... work when loaded
on a system with RAM starting at 0x2... ? I'm not sure the answer to this
question.

I'd rather try and get one kernel working on all systems with the S5PV210
if possible instead of havign to pick which ones have certain base addresses


> >
> > How will the user change their board to deal with this change to hte
> > kernel?
> 
> Now we got the these memory
> DMC0 has 128MiB and DMC1 has 512MiB.
> Then it uses the memory from 0x2000'0000 to 0x5fff'ffff. it used 1GiB
> memory address space.
> Of course we can enable thie HIGHMEM but older kernel doesn't support it fully.
> Also I hope to use the 3G/1G memory scheme.

Given you can only get 1G of real memory into a S5PV210, is it really
necessary to go to 3G/1G? What problems are there with 2/2 split?

> How to you think?

I thought that when the s5pv210 was submitted we enabled the SPARSMEM
support so that whatever DRAM configuration we could handle it within
a single kernel.

-- 
Ben

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

[PATCH] s5pv210: Change the base ram address to 0x3000'0000

[Kyungmin Park]

On Fri, May 28, 2010 at 1:20 PM, Ben Dooks <ben-linux@fluff.org> wrote:
> On Fri, May 28, 2010 at 01:02:24PM +0900, Kyungmin Park wrote:
>> On Fri, May 28, 2010 at 12:53 PM, Ben Dooks <ben-linux@fluff.org> wrote:
>> > On Fri, May 28, 2010 at 11:35:59AM +0900, Kyungmin Park wrote:
>> >> s5pc110 (aka s5pv210) has 2 DRAM port and used it both usually.
>> >> Assume DMC0 starts with 0x2000'0000 with 128MiB.
>> >> DMC1 starts with 0x4000'0000 with 128MiB.
>> >> Note that DMC1 has to start address 0x4000'0000 at least.
>> >>
>> >> Then there's too much memory hole 0x1800'0000 (128MiB + 256MiB)
>> >>
>> >> To reduce memory waste, the DMC0 start with 0x3000'0000.
>
> but you just said DMC0 starts at 0x2... ? Is this reconfigurable via
> software, if so can it be moved once initialisation si done? how fixed
> is this?

DMC0 can set the its start address at bootloader, it can range from
0x2000'0000 to 0x3fff'ffff.

My goal is that reduce the 256MiB memory area and can use the full
memory without HIGHMEM and 2G/2G split configuration.

>
>> Hi,
>>
>> >
>> > We should be using SPARSEMEM now, is it not reducing any wastage
>> > in virtual memory space?
>>
>> Please consider the older kernel also. and most systems are designed
>> for 0x3000'0000.
>
> Ok, will a kernel with the base-address set to 0x3... work when loaded
> on a system with RAM starting at 0x2... ? I'm not sure the answer to this
> question.

Then can't detect last 2Gib memory, only detect first 2Gib memory at
kernel 2.6.29.

>
> I'd rather try and get one kernel working on all systems with the S5PV210
> if possible instead of havign to pick which ones have certain base addresses
>
>
>> >
>> > How will the user change their board to deal with this change to hte
>> > kernel?
>>
>> Now we got the these memory
>> DMC0 has 128MiB and DMC1 has 512MiB.
>> Then it uses the memory from 0x2000'0000 to 0x5fff'ffff. it used 1GiB
>> memory address space.
>> Of course we can enable thie HIGHMEM but older kernel doesn't support it fully.
>> Also I hope to use the 3G/1G memory scheme.
>
> Given you can only get 1G of real memory into a S5PV210, is it really
> necessary to go to 3G/1G? What problems are there with 2/2 split?

Right it's best but can't. It's hardware problem. DMC0 has different
memory with 166MHz and DMC1 has 200MHz
It can't change it.

>
>> How to you think?
>
> I thought that when the s5pv210 was submitted we enabled the SPARSMEM
> support so that whatever DRAM configuration we could handle it within
> a single kernel.

I also hope to use single kernel for s5p series. and that's reason I
hope to change it.
With some reason. mobile division has different memory types and DMC0
should be used fixed memory.
and only DMC1 can modify form 2Gib to 4Gib.

Thank you,
Kyungmin Park

[PATCH] s5pv210: Change the base ram address to 0x3000'0000

[Ben Dooks]

On Fri, May 28, 2010 at 02:25:41PM +0900, Kyungmin Park wrote:
> On Fri, May 28, 2010 at 1:20 PM, Ben Dooks <ben-linux@fluff.org> wrote:
> > On Fri, May 28, 2010 at 01:02:24PM +0900, Kyungmin Park wrote:
> >> On Fri, May 28, 2010 at 12:53 PM, Ben Dooks <ben-linux@fluff.org> wrote:
> >> > On Fri, May 28, 2010 at 11:35:59AM +0900, Kyungmin Park wrote:
> >> >> s5pc110 (aka s5pv210) has 2 DRAM port and used it both usually.
> >> >> Assume DMC0 starts with 0x2000'0000 with 128MiB.
> >> >> DMC1 starts with 0x4000'0000 with 128MiB.
> >> >> Note that DMC1 has to start address 0x4000'0000 at least.
> >> >>
> >> >> Then there's too much memory hole 0x1800'0000 (128MiB + 256MiB)
> >> >>
> >> >> To reduce memory waste, the DMC0 start with 0x3000'0000.
> >
> > but you just said DMC0 starts at 0x2... ? Is this reconfigurable via
> > software, if so can it be moved once initialisation si done? how fixed
> > is this?
> 
> DMC0 can set the its start address at bootloader, it can range from
> 0x2000'0000 to 0x3fff'ffff.
> 
> My goal is that reduce the 256MiB memory area and can use the full
> memory without HIGHMEM and 2G/2G split configuration.

If I remember correctly, te device has 4x256MiB banks, of which the
population is external on S5PV210 or packaging with the S5PC1xx series.

On the split situation, is it really that bad to have a 2G/2G split,
are you currently running into a situation where a single process
really needs that much address space (especially given the fact that
there can only ever be 1GiB of real memory in the system)?

On the higmem issue, I don't think we've ever had to bother with HIGMEM
even on a SMDKV210 with 1GiB of memory fitted when we first did the support.

> >
> >> Hi,
> >>
> >> >
> >> > We should be using SPARSEMEM now, is it not reducing any wastage
> >> > in virtual memory space?
> >>
> >> Please consider the older kernel also. and most systems are designed
> >> for 0x3000'0000.
> >
> > Ok, will a kernel with the base-address set to 0x3... work when loaded
> > on a system with RAM starting at 0x2... ? I'm not sure the answer to this
> > question.
> 
> Then can't detect last 2Gib memory, only detect first 2Gib memory at
> kernel 2.6.29.

Not sure how to deal with this one.
 
> >
> > I'd rather try and get one kernel working on all systems with the S5PV210
> > if possible instead of havign to pick which ones have certain base addresses
> >
> >
> >> >
> >> > How will the user change their board to deal with this change to hte
> >> > kernel?
> >>
> >> Now we got the these memory
> >> DMC0 has 128MiB and DMC1 has 512MiB.
> >> Then it uses the memory from 0x2000'0000 to 0x5fff'ffff. it used 1GiB
> >> memory address space.
> >> Of course we can enable thie HIGHMEM but older kernel doesn't support it fully.
> >> Also I hope to use the 3G/1G memory scheme.
> >
> > Given you can only get 1G of real memory into a S5PV210, is it really
> > necessary to go to 3G/1G? What problems are there with 2/2 split?
> 
> Right it's best but can't. It's hardware problem. DMC0 has different
> memory with 166MHz and DMC1 has 200MHz
> It can't change it.

Yes, but that's an allocation issue not a layout issue. Whatever is
allocating pages needs to know to prefer the DMC1 memory.
 
> >
> >> How to you think?
> >
> > I thought that when the s5pv210 was submitted we enabled the SPARSMEM
> > support so that whatever DRAM configuration we could handle it within
> > a single kernel.
> 
> I also hope to use single kernel for s5p series. and that's reason I
> hope to change it.
> With some reason. mobile division has different memory types and DMC0
> should be used fixed memory.
> and only DMC1 can modify form 2Gib to 4Gib.
> 
> Thank you,
> Kyungmin Park

-- 
-- 
Ben

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

[PATCH] s5pv210: Change the base ram address to 0x3000'0000

[Kyungmin Park]

On Fri, May 28, 2010 at 3:02 PM, Ben Dooks <ben-linux@fluff.org> wrote:
> On Fri, May 28, 2010 at 02:25:41PM +0900, Kyungmin Park wrote:
>> On Fri, May 28, 2010 at 1:20 PM, Ben Dooks <ben-linux@fluff.org> wrote:
>> > On Fri, May 28, 2010 at 01:02:24PM +0900, Kyungmin Park wrote:
>> >> On Fri, May 28, 2010 at 12:53 PM, Ben Dooks <ben-linux@fluff.org> wrote:
>> >> > On Fri, May 28, 2010 at 11:35:59AM +0900, Kyungmin Park wrote:
>> >> >> s5pc110 (aka s5pv210) has 2 DRAM port and used it both usually.
>> >> >> Assume DMC0 starts with 0x2000'0000 with 128MiB.
>> >> >> DMC1 starts with 0x4000'0000 with 128MiB.
>> >> >> Note that DMC1 has to start address 0x4000'0000 at least.
>> >> >>
>> >> >> Then there's too much memory hole 0x1800'0000 (128MiB + 256MiB)
>> >> >>
>> >> >> To reduce memory waste, the DMC0 start with 0x3000'0000.
>> >
>> > but you just said DMC0 starts at 0x2... ? Is this reconfigurable via
>> > software, if so can it be moved once initialisation si done? how fixed
>> > is this?
>>
>> DMC0 can set the its start address at bootloader, it can range from
>> 0x2000'0000 to 0x3fff'ffff.
>>
>> My goal is that reduce the 256MiB memory area and can use the full
>> memory without HIGHMEM and 2G/2G split configuration.
>
> If I remember correctly, te device has 4x256MiB banks, of which the
> population is external on S5PV210 or packaging with the S5PC1xx series.
>
> On the split situation, is it really that bad to have a 2G/2G split,
> are you currently running into a situation where a single process
> really needs that much address space (especially given the fact that
> there can only ever be 1GiB of real memory in the system)?

By reduce the 256MiB memory from 0x2000'0000 to 0x3000'0000,
we can use 128MiB + (128MiB hole) + 512MiB without any memory changes
and configurations.

we got several boards which has 2Gib, 3Gib, and 4Gib. So to use these board
with one kernel. it's best way to use 0x3000'0000 and we used it from
first time.
but the problem is that mainline codes are committed as 0x2000'0000.

>
> On the higmem issue, I don't think we've ever had to bother with HIGMEM
> even on a SMDKV210 with 1GiB of memory fitted when we first did the support.
>
>> >
>> >> Hi,
>> >>
>> >> >
>> >> > We should be using SPARSEMEM now, is it not reducing any wastage
>> >> > in virtual memory space?
>> >>
>> >> Please consider the older kernel also. and most systems are designed
>> >> for 0x3000'0000.
>> >
>> > Ok, will a kernel with the base-address set to 0x3... work when loaded
>> > on a system with RAM starting at 0x2... ? I'm not sure the answer to this
>> > question.
>>
>> Then can't detect last 2Gib memory, only detect first 2Gib memory at
>> kernel 2.6.29.
>
> Not sure how to deal with this one.
>
>> >
>> > I'd rather try and get one kernel working on all systems with the S5PV210
>> > if possible instead of havign to pick which ones have certain base addresses
>> >
>> >
>> >> >
>> >> > How will the user change their board to deal with this change to hte
>> >> > kernel?
>> >>
>> >> Now we got the these memory
>> >> DMC0 has 128MiB and DMC1 has 512MiB.
>> >> Then it uses the memory from 0x2000'0000 to 0x5fff'ffff. it used 1GiB
>> >> memory address space.
>> >> Of course we can enable thie HIGHMEM but older kernel doesn't support it fully.
>> >> Also I hope to use the 3G/1G memory scheme.
>> >
>> > Given you can only get 1G of real memory into a S5PV210, is it really
>> > necessary to go to 3G/1G? What problems are there with 2/2 split?
>>
>> Right it's best but can't. It's hardware problem. DMC0 has different
>> memory with 166MHz and DMC1 has 200MHz
>> It can't change it.
>
> Yes, but that's an allocation issue not a layout issue. Whatever is
> allocating pages needs to know to prefer the DMC1 memory.

No it's hardware issue. when connect the 166MHz memory we need to
async bridge to work on 200MHz.
In this case, we can't use DMC0 CS1 for 200MHz memory device. that's
reason we put the 4Gib memory to DMC1.

Thank you,
Kyungmin Park

>
>> >
>> >> How to you think?
>> >
>> > I thought that when the s5pv210 was submitted we enabled the SPARSMEM
>> > support so that whatever DRAM configuration we could handle it within
>> > a single kernel.
>>
>> I also hope to use single kernel for s5p series. and that's reason I
>> hope to change it.
>> With some reason. mobile division has different memory types and DMC0
>> should be used fixed memory.
>> and only DMC1 can modify form 2Gib to 4Gib.

[PATCH] s5pv210: Change the base ram address to 0x3000'0000

[Ben Dooks]

On Fri, May 28, 2010 at 03:19:43PM +0900, Kyungmin Park wrote:
> On Fri, May 28, 2010 at 3:02 PM, Ben Dooks <ben-linux@fluff.org> wrote:
> > On Fri, May 28, 2010 at 02:25:41PM +0900, Kyungmin Park wrote:
> >> On Fri, May 28, 2010 at 1:20 PM, Ben Dooks <ben-linux@fluff.org> wrote:
> >> > On Fri, May 28, 2010 at 01:02:24PM +0900, Kyungmin Park wrote:
> >> >> On Fri, May 28, 2010 at 12:53 PM, Ben Dooks <ben-linux@fluff.org> wrote:
> >> >> > On Fri, May 28, 2010 at 11:35:59AM +0900, Kyungmin Park wrote:
> >> >> >> s5pc110 (aka s5pv210) has 2 DRAM port and used it both usually.
> >> >> >> Assume DMC0 starts with 0x2000'0000 with 128MiB.
> >> >> >> DMC1 starts with 0x4000'0000 with 128MiB.
> >> >> >> Note that DMC1 has to start address 0x4000'0000 at least.
> >> >> >>
> >> >> >> Then there's too much memory hole 0x1800'0000 (128MiB + 256MiB)
> >> >> >>
> >> >> >> To reduce memory waste, the DMC0 start with 0x3000'0000.
> >> >
> >> > but you just said DMC0 starts at 0x2... ? Is this reconfigurable via
> >> > software, if so can it be moved once initialisation si done? how fixed
> >> > is this?
> >>
> >> DMC0 can set the its start address at bootloader, it can range from
> >> 0x2000'0000 to 0x3fff'ffff.
> >>
> >> My goal is that reduce the 256MiB memory area and can use the full
> >> memory without HIGHMEM and 2G/2G split configuration.
> >
> > If I remember correctly, te device has 4x256MiB banks, of which the
> > population is external on S5PV210 or packaging with the S5PC1xx series.
> >
> > On the split situation, is it really that bad to have a 2G/2G split,
> > are you currently running into a situation where a single process
> > really needs that much address space (especially given the fact that
> > there can only ever be 1GiB of real memory in the system)?
> 
> By reduce the 256MiB memory from 0x2000'0000 to 0x3000'0000,
> we can use 128MiB + (128MiB hole) + 512MiB without any memory changes
> and configurations.
> 
> we got several boards which has 2Gib, 3Gib, and 4Gib. So to use these board
> with one kernel. it's best way to use 0x3000'0000 and we used it from
> first time.
> but the problem is that mainline codes are committed as 0x2000'0000.

ok, but if we change the base to 0x3... then will we end up breaking the
kernel for someone else? ie, booting on an smdkv210?

I may have been confused with getting the sparsemem support in, as it
definetly deal with the case of memory holes, I also thought it would
allow the kernel to run with a non-hole'd virtual memory space.

The current kernel defconfig has 2:2, so not sure why the importance
of 1:3 split.

> >
> > On the higmem issue, I don't think we've ever had to bother with HIGMEM
> > even on a SMDKV210 with 1GiB of memory fitted when we first did the support.
> >
> >> >
> >> >> Hi,
> >> >>
> >> >> >
> >> >> > We should be using SPARSEMEM now, is it not reducing any wastage
> >> >> > in virtual memory space?
> >> >>
> >> >> Please consider the older kernel also. and most systems are designed
> >> >> for 0x3000'0000.
> >> >
> >> > Ok, will a kernel with the base-address set to 0x3... work when loaded
> >> > on a system with RAM starting at 0x2... ? I'm not sure the answer to this
> >> > question.
> >>
> >> Then can't detect last 2Gib memory, only detect first 2Gib memory at
> >> kernel 2.6.29.
> >
> > Not sure how to deal with this one.
> >
> >> >
> >> > I'd rather try and get one kernel working on all systems with the S5PV210
> >> > if possible instead of havign to pick which ones have certain base addresses
> >> >
> >> >
> >> >> >
> >> >> > How will the user change their board to deal with this change to hte
> >> >> > kernel?
> >> >>
> >> >> Now we got the these memory
> >> >> DMC0 has 128MiB and DMC1 has 512MiB.
> >> >> Then it uses the memory from 0x2000'0000 to 0x5fff'ffff. it used 1GiB
> >> >> memory address space.
> >> >> Of course we can enable thie HIGHMEM but older kernel doesn't support it fully.
> >> >> Also I hope to use the 3G/1G memory scheme.
> >> >
> >> > Given you can only get 1G of real memory into a S5PV210, is it really
> >> > necessary to go to 3G/1G? What problems are there with 2/2 split?
> >>
> >> Right it's best but can't. It's hardware problem. DMC0 has different
> >> memory with 166MHz and DMC1 has 200MHz
> >> It can't change it.
> >
> > Yes, but that's an allocation issue not a layout issue. Whatever is
> > allocating pages needs to know to prefer the DMC1 memory.
> 
> No it's hardware issue. when connect the 166MHz memory we need to
> async bridge to work on 200MHz.
> In this case, we can't use DMC0 CS1 for 200MHz memory device. that's
> reason we put the 4Gib memory to DMC1.

Ok, but you're still giving all the memory to the kernel? If i've
calculated correctly then the kernel is still in the DMC0 area
with the 166MHz memory?
 
> Thank you,
> Kyungmin Park
> 
> >
> >> >
> >> >> How to you think?
> >> >
> >> > I thought that when the s5pv210 was submitted we enabled the SPARSMEM
> >> > support so that whatever DRAM configuration we could handle it within
> >> > a single kernel.
> >>
> >> I also hope to use single kernel for s5p series. and that's reason I
> >> hope to change it.
> >> With some reason. mobile division has different memory types and DMC0
> >> should be used fixed memory.
> >> and only DMC1 can modify form 2Gib to 4Gib.

-- 
-- 
Ben

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

[PATCH] s5pv210: Change the base ram address to 0x3000'0000

[Kyungmin Park]

On Fri, May 28, 2010 at 4:09 PM, Ben Dooks <ben-linux@fluff.org> wrote:
> On Fri, May 28, 2010 at 03:19:43PM +0900, Kyungmin Park wrote:
>> On Fri, May 28, 2010 at 3:02 PM, Ben Dooks <ben-linux@fluff.org> wrote:
>> > On Fri, May 28, 2010 at 02:25:41PM +0900, Kyungmin Park wrote:
>> >> On Fri, May 28, 2010 at 1:20 PM, Ben Dooks <ben-linux@fluff.org> wrote:
>> >> > On Fri, May 28, 2010 at 01:02:24PM +0900, Kyungmin Park wrote:
>> >> >> On Fri, May 28, 2010 at 12:53 PM, Ben Dooks <ben-linux@fluff.org> wrote:
>> >> >> > On Fri, May 28, 2010 at 11:35:59AM +0900, Kyungmin Park wrote:
>> >> >> >> s5pc110 (aka s5pv210) has 2 DRAM port and used it both usually.
>> >> >> >> Assume DMC0 starts with 0x2000'0000 with 128MiB.
>> >> >> >> DMC1 starts with 0x4000'0000 with 128MiB.
>> >> >> >> Note that DMC1 has to start address 0x4000'0000 at least.
>> >> >> >>
>> >> >> >> Then there's too much memory hole 0x1800'0000 (128MiB + 256MiB)
>> >> >> >>
>> >> >> >> To reduce memory waste, the DMC0 start with 0x3000'0000.
>> >> >
>> >> > but you just said DMC0 starts at 0x2... ? Is this reconfigurable via
>> >> > software, if so can it be moved once initialisation si done? how fixed
>> >> > is this?
>> >>
>> >> DMC0 can set the its start address at bootloader, it can range from
>> >> 0x2000'0000 to 0x3fff'ffff.
>> >>
>> >> My goal is that reduce the 256MiB memory area and can use the full
>> >> memory without HIGHMEM and 2G/2G split configuration.
>> >
>> > If I remember correctly, te device has 4x256MiB banks, of which the
>> > population is external on S5PV210 or packaging with the S5PC1xx series.
>> >
>> > On the split situation, is it really that bad to have a 2G/2G split,
>> > are you currently running into a situation where a single process
>> > really needs that much address space (especially given the fact that
>> > there can only ever be 1GiB of real memory in the system)?
>>
>> By reduce the 256MiB memory from 0x2000'0000 to 0x3000'0000,
>> we can use 128MiB + (128MiB hole) + 512MiB without any memory changes
>> and configurations.
>>
>> we got several boards which has 2Gib, 3Gib, and 4Gib. So to use these board
>> with one kernel. it's best way to use 0x3000'0000 and we used it from
>> first time.
>> but the problem is that mainline codes are committed as 0x2000'0000.
>
> ok, but if we change the base to 0x3... then will we end up breaking the
> kernel for someone else? ie, booting on an smdkv210?
>
> I may have been confused with getting the sparsemem support in, as it
> definetly deal with the case of memory holes, I also thought it would
> allow the kernel to run with a non-hole'd virtual memory space.
>
> The current kernel defconfig has 2:2, so not sure why the importance
> of 1:3 split.
>
>> >
>> > On the higmem issue, I don't think we've ever had to bother with HIGMEM
>> > even on a SMDKV210 with 1GiB of memory fitted when we first did the support.
>> >
>> >> >
>> >> >> Hi,
>> >> >>
>> >> >> >
>> >> >> > We should be using SPARSEMEM now, is it not reducing any wastage
>> >> >> > in virtual memory space?
>> >> >>
>> >> >> Please consider the older kernel also. and most systems are designed
>> >> >> for 0x3000'0000.
>> >> >
>> >> > Ok, will a kernel with the base-address set to 0x3... work when loaded
>> >> > on a system with RAM starting at 0x2... ? I'm not sure the answer to this
>> >> > question.
>> >>
>> >> Then can't detect last 2Gib memory, only detect first 2Gib memory at
>> >> kernel 2.6.29.
>> >
>> > Not sure how to deal with this one.
>> >
>> >> >
>> >> > I'd rather try and get one kernel working on all systems with the S5PV210
>> >> > if possible instead of havign to pick which ones have certain base addresses
>> >> >
>> >> >
>> >> >> >
>> >> >> > How will the user change their board to deal with this change to hte
>> >> >> > kernel?
>> >> >>
>> >> >> Now we got the these memory
>> >> >> DMC0 has 128MiB and DMC1 has 512MiB.
>> >> >> Then it uses the memory from 0x2000'0000 to 0x5fff'ffff. it used 1GiB
>> >> >> memory address space.
>> >> >> Of course we can enable thie HIGHMEM but older kernel doesn't support it fully.
>> >> >> Also I hope to use the 3G/1G memory scheme.
>> >> >
>> >> > Given you can only get 1G of real memory into a S5PV210, is it really
>> >> > necessary to go to 3G/1G? What problems are there with 2/2 split?
>> >>
>> >> Right it's best but can't. It's hardware problem. DMC0 has different
>> >> memory with 166MHz and DMC1 has 200MHz
>> >> It can't change it.
>> >
>> > Yes, but that's an allocation issue not a layout issue. Whatever is
>> > allocating pages needs to know to prefer the DMC1 memory.
>>
>> No it's hardware issue. when connect the 166MHz memory we need to
>> async bridge to work on 200MHz.
>> In this case, we can't use DMC0 CS1 for 200MHz memory device. that's
>> reason we put the 4Gib memory to DMC1.
>
> Ok, but you're still giving all the memory to the kernel? If i've
> calculated correctly then the kernel is still in the DMC0 area
> with the 166MHz memory?

Right, kernel located at 0x30xx'xxxx, Is it possible to locate the
kernel at 0x40xx'xxxx and use the 0x30xx'xxxx for free memory. I mean
does the kernel locate in the middle of memory?
If possible, it's best.

One note it's impossible to swap the DMC0 memory to DMC1. it's hardware wired.

Thank you,
Kyungmin Park
>
>> Thank you,
>> Kyungmin Park
>>
>> >
>> >> >
>> >> >> How to you think?
>> >> >
>> >> > I thought that when the s5pv210 was submitted we enabled the SPARSMEM
>> >> > support so that whatever DRAM configuration we could handle it within
>> >> > a single kernel.
>> >>
>> >> I also hope to use single kernel for s5p series. and that's reason I
>> >> hope to change it.
>> >> With some reason. mobile division has different memory types and DMC0
>> >> should be used fixed memory.
>> >> and only DMC1 can modify form 2Gib to 4Gib.
>
> --
> --
> Ben
>
> Q: ? ? ?What's a light-year?
> A: ? ? ?One-third less calories than a regular year.
>
>

[PATCH] s5pv210: Change the base ram address to 0x3000'0000

[Kukjin Kim]

Kyungmin Park wrote:
> 
> s5pc110 (aka s5pv210) has 2 DRAM port and used it both usually.
> Assume DMC0 starts with 0x2000'0000 with 128MiB.
> DMC1 starts with 0x4000'0000 with 128MiB.
> Note that DMC1 has to start address 0x4000'0000 at least.

Hi,

Following is memory map of S5PV210 and S5PC110.

0x80000000  -------------------
            |          |
0x70000000  |          |
            |          |
0x60000000  |  DMC 1  |  up to 1GB
            |          |
0x50000000  |          |
            |          |
0x40000000  ----------------- 
            |          |
0x30000000  |  DMC 0  |  up to 512MB
            |          |
0x20000000  -------------------

According to memory map, S5PV210 and S5PC110 can support up to 1.5GB memory.
So, the base address of system memory should be 0x20000000 to support 1.5GB.

The oneDRAM in S5PC110 MCP(Multi Chip Package) has to map to DMC0, but
mDDR(or SDR) in the MCP without oneDRAM can be mapped to DMC0.
Also, whatever memory types, the base address must be at 0x20000000, in case
the memory size is over 256MB.

That is, it has problem to move base address of system memory to 0x30000000
only to support specific S5PC110 MCP type.

> Then there's too much memory hole 0x1800'0000 (128MiB + 256MiB)
> 
> To reduce memory waste, the DMC0 start with 0x3000'0000.
> 
> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
> ---
> diff --git a/arch/arm/mach-s5pv210/Makefile.boot b/arch/arm/mach-
> s5pv210/Makefile.boot
> index ff90aa1..b0909e3 100644
> --- a/arch/arm/mach-s5pv210/Makefile.boot
> +++ b/arch/arm/mach-s5pv210/Makefile.boot
> @@ -1,2 +1,2 @@
> -   zreladdr-y	:= 0x20008000
> -params_phys-y	:= 0x20000100
> +   zreladdr-y	:= 0x30008000
> +params_phys-y	:= 0x30000100
> diff --git a/arch/arm/mach-s5pv210/include/mach/map.h b/arch/arm/mach-
> s5pv210/include/mach/map.h
> index 34eb168..9f4c368 100644
> --- a/arch/arm/mach-s5pv210/include/mach/map.h
> +++ b/arch/arm/mach-s5pv210/include/mach/map.h
> @@ -74,7 +74,7 @@
>  #define S5PV210_PA_VIC3		(0xF2300000)
>  #define S5P_PA_VIC3		S5PV210_PA_VIC3
> 
> -#define S5PV210_PA_SDRAM	(0x20000000)
> +#define S5PV210_PA_SDRAM	(0x30000000)
>  #define S5P_PA_SDRAM		S5PV210_PA_SDRAM
> 
>  /* I2S */
> diff --git a/arch/arm/mach-s5pv210/include/mach/memory.h b/arch/arm/mach-
> s5pv210/include/mach/memory.h
> index 379117e..07a15d8 100644
> --- a/arch/arm/mach-s5pv210/include/mach/memory.h
> +++ b/arch/arm/mach-s5pv210/include/mach/memory.h
> @@ -13,7 +13,7 @@
>  #ifndef __ASM_ARCH_MEMORY_H
>  #define __ASM_ARCH_MEMORY_H
> 
> -#define PHYS_OFFSET		UL(0x20000000)
> +#define PHYS_OFFSET		UL(0x30000000)
>  #define CONSISTENT_DMA_SIZE	(SZ_8M + SZ_4M + SZ_2M)
> 
>  /* Maximum of 256MiB in one bank */
> 
Thanks.

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

[PATCH] s5pv210: Change the base ram address to 0x3000'0000

[Ben Dooks]

On Fri, May 28, 2010 at 11:35:59AM +0900, Kyungmin Park wrote:
> s5pc110 (aka s5pv210) has 2 DRAM port and used it both usually.
> Assume DMC0 starts with 0x2000'0000 with 128MiB.
> DMC1 starts with 0x4000'0000 with 128MiB.
> Note that DMC1 has to start address 0x4000'0000 at least.
> 
> Then there's too much memory hole 0x1800'0000 (128MiB + 256MiB)
> 
> To reduce memory waste, the DMC0 start with 0x3000'0000.
> 
> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
> ---
> diff --git a/arch/arm/mach-s5pv210/Makefile.boot b/arch/arm/mach-s5pv210/Makefile.boot
> index ff90aa1..b0909e3 100644
> --- a/arch/arm/mach-s5pv210/Makefile.boot
> +++ b/arch/arm/mach-s5pv210/Makefile.boot
> @@ -1,2 +1,2 @@
> -   zreladdr-y	:= 0x20008000
> -params_phys-y	:= 0x20000100
> +   zreladdr-y	:= 0x30008000
> +params_phys-y	:= 0x30000100

Apart from the technical comments that have been in this thread, I would
like to share notes about making changes like these that affect the entire
SoC kernel build.

Firstly, changing this effects the rest of the users, not everyone is
using the same device with same layout. This configuration boots on
both S5PC110 and S5PV210 devices. Some of these _have_ to support 1GiB
of memory. Some of them have their bootloaders programemd to start at
0x20... So please think of how this will affect everyone else.

Secondly, if you have thought about point 1, then please let us know in
the commit messages.  The commit message isn't just a log of what the
patch does, it can also be used to show any necessary working about what
is happening and side-effects of it.

Thirdly, the kernel's default configuration for a SoC/arch isn't meant to 
be the best-fit for your board(s) or development practicies. It is meant
to cover as many devices as possible, to allow build testing and to have
something that people can test with. If it is not the most efficient on
the board, then make your own defconfig that is a best fit for your
purposes.

So, sometimes when this sort of change is required there may be a case of
making it configurable so that your own defconfig/boardconfig can select
it for your own projects.


Please keep in mind that other people have/are using the kernel and that
some changes have an affect on other projects as well as your own. This
is why user-space facing changes are the most debated and difficult to
do, kernel internals shift quite easily, whilst the data exported to
userspace is kept as stable as possible.

-- 
Ben

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

[PATCH] s5pv210: Change the base ram address to 0x3000'0000

[Kyungmin Park]

On Mon, May 31, 2010 at 9:39 AM, Ben Dooks <ben-linux@fluff.org> wrote:
> On Fri, May 28, 2010 at 11:35:59AM +0900, Kyungmin Park wrote:
>> s5pc110 (aka s5pv210) has 2 DRAM port and used it both usually.
>> Assume DMC0 starts with 0x2000'0000 with 128MiB.
>> DMC1 starts with 0x4000'0000 with 128MiB.
>> Note that DMC1 has to start address 0x4000'0000 at least.
>>
>> Then there's too much memory hole 0x1800'0000 (128MiB + 256MiB)
>>
>> To reduce memory waste, the DMC0 start with 0x3000'0000.
>>
>> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
>> ---
>> diff --git a/arch/arm/mach-s5pv210/Makefile.boot b/arch/arm/mach-s5pv210/Makefile.boot
>> index ff90aa1..b0909e3 100644
>> --- a/arch/arm/mach-s5pv210/Makefile.boot
>> +++ b/arch/arm/mach-s5pv210/Makefile.boot
>> @@ -1,2 +1,2 @@
>> - ? zreladdr-y ? ? ? ?:= 0x20008000
>> -params_phys-y ? ? ? ?:= 0x20000100
>> + ? zreladdr-y ? ? ? ?:= 0x30008000
>> +params_phys-y ? ? ? ?:= 0x30000100
>
> Apart from the technical comments that have been in this thread, I would
> like to share notes about making changes like these that affect the entire
> SoC kernel build.
>
> Firstly, changing this effects the rest of the users, not everyone is
> using the same device with same layout. This configuration boots on
> both S5PC110 and S5PV210 devices. Some of these _have_ to support 1GiB
> of memory. Some of them have their bootloaders programemd to start at
> 0x20... So please think of how this will affect everyone else.
>
> Secondly, if you have thought about point 1, then please let us know in
> the commit messages. ?The commit message isn't just a log of what the
> patch does, it can also be used to show any necessary working about what
> is happening and side-effects of it.
>
> Thirdly, the kernel's default configuration for a SoC/arch isn't meant to
> be the best-fit for your board(s) or development practicies. It is meant
> to cover as many devices as possible, to allow build testing and to have
> something that people can test with. If it is not the most efficient on
> the board, then make your own defconfig that is a best fit for your
> purposes.
>
> So, sometimes when this sort of change is required there may be a case of
> making it configurable so that your own defconfig/boardconfig can select
> it for your own projects.
>
>
> Please keep in mind that other people have/are using the kernel and that
> some changes have an affect on other projects as well as your own. This
> is why user-space facing changes are the most debated and difficult to
> do, kernel internals shift quite easily, whilst the data exported to
> userspace is kept as stable as possible.

Definitely I agree with your opinions. and that's why I change it to
0x3000'0000.
when we start the project. chip vendor guides us to use the
0x3000'0000. at that time there's some reason to use this address. As
I know. but some time later they change it to 0x2000'0000.
But we already setup and develop the all environment as 0x3000'0000.
it's also difficult to change it to 0x2000'0000.

Thank you,
Kyungmin Park

[PATCH] s5pv210: Change the base ram address to 0x3000'0000

[Kyungmin Park]

On Mon, May 31, 2010 at 9:47 AM, Kyungmin Park <kmpark@infradead.org> wrote:
> On Mon, May 31, 2010 at 9:39 AM, Ben Dooks <ben-linux@fluff.org> wrote:
>> On Fri, May 28, 2010 at 11:35:59AM +0900, Kyungmin Park wrote:
>>> s5pc110 (aka s5pv210) has 2 DRAM port and used it both usually.
>>> Assume DMC0 starts with 0x2000'0000 with 128MiB.
>>> DMC1 starts with 0x4000'0000 with 128MiB.
>>> Note that DMC1 has to start address 0x4000'0000 at least.
>>>
>>> Then there's too much memory hole 0x1800'0000 (128MiB + 256MiB)
>>>
>>> To reduce memory waste, the DMC0 start with 0x3000'0000.
>>>
>>> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
>>> ---
>>> diff --git a/arch/arm/mach-s5pv210/Makefile.boot b/arch/arm/mach-s5pv210/Makefile.boot
>>> index ff90aa1..b0909e3 100644
>>> --- a/arch/arm/mach-s5pv210/Makefile.boot
>>> +++ b/arch/arm/mach-s5pv210/Makefile.boot
>>> @@ -1,2 +1,2 @@
>>> - ? zreladdr-y ? ? ? ?:= 0x20008000
>>> -params_phys-y ? ? ? ?:= 0x20000100
>>> + ? zreladdr-y ? ? ? ?:= 0x30008000
>>> +params_phys-y ? ? ? ?:= 0x30000100
>>
>> Apart from the technical comments that have been in this thread, I would
>> like to share notes about making changes like these that affect the entire
>> SoC kernel build.
>>
>> Firstly, changing this effects the rest of the users, not everyone is
>> using the same device with same layout. This configuration boots on
>> both S5PC110 and S5PV210 devices. Some of these _have_ to support 1GiB
>> of memory. Some of them have their bootloaders programemd to start at
>> 0x20... So please think of how this will affect everyone else.
>>
>> Secondly, if you have thought about point 1, then please let us know in
>> the commit messages. ?The commit message isn't just a log of what the
>> patch does, it can also be used to show any necessary working about what
>> is happening and side-effects of it.
>>
>> Thirdly, the kernel's default configuration for a SoC/arch isn't meant to
>> be the best-fit for your board(s) or development practicies. It is meant
>> to cover as many devices as possible, to allow build testing and to have
>> something that people can test with. If it is not the most efficient on
>> the board, then make your own defconfig that is a best fit for your
>> purposes.
>>
>> So, sometimes when this sort of change is required there may be a case of
>> making it configurable so that your own defconfig/boardconfig can select
>> it for your own projects.
>>
>>
>> Please keep in mind that other people have/are using the kernel and that
>> some changes have an affect on other projects as well as your own. This
>> is why user-space facing changes are the most debated and difficult to
>> do, kernel internals shift quite easily, whilst the data exported to
>> userspace is kept as stable as possible.
>
> Definitely I agree with your opinions. and that's why I change it to
> 0x3000'0000.
> when we start the project. chip vendor guides us to use the
> 0x3000'0000. at that time there's some reason to use this address. As
> I know. but some time later they change it to 0x2000'0000.
> But we already setup and develop the all environment as 0x3000'0000.
> it's also difficult to change it to 0x2000'0000.

Please see the committed date.
http://git.denx.de/?p=u-boot.git;a=commitdiff;h=8bc4ee9e8213abe4031ea1720aa02fa98d4402ad

+#
+# Copyright (C) 2008 # Samsung Elecgtronics
+# Kyungmin Park <kyungmin.park@samsung.com>
+#
+
+# On S5PC100 we use the 128 MiB OneDRAM bank at
+#
+# 0x30000000 to 0x35000000 (80MiB)
+# 0x38000000 to 0x40000000 (128MiB)
+#
+# On S5PC110 we use the 128 MiB OneDRAM bank at
+#
+# 0x30000000 to 0x35000000 (80MiB)
+# 0x40000000 to 0x48000000 (128MiB)
+#
+TEXT_BASE = 0x34800000

Thank you,
Kyungmin Park