Skip to contentSkip to conversation

Add support for Eukrea CPU9260 SBC

PATCH New Eric Benard 3 participants 10 messages Jul 16, 2009 – Jul 18, 2009

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Add support for Eukrea CPU9260 SBC

CPUAT91 is built around Atmel's AT91SAM9260 and has up to 64MB of NOR flash, up to 128MB of SDRAM, up to 2GB of NAND and includes a 10/100 Ethernet PHY in RMII mode.

---
 MAINTAINERS                        |    4 +
 MAKEALL                            |    1 +
 Makefile                           |    3 +
 board/eukrea/cpu9260/Makefile      |   59 +++++++
 board/eukrea/cpu9260/config.mk     |    1 +
 board/eukrea/cpu9260/cpu9260.c     |  209 +++++++++++++++++++++++
 board/eukrea/cpu9260/cpu9260_led.c |   46 +++++
 cpu/arm926ejs/at91/lowlevel_init.S |    2 +-
 include/configs/cpu9260.h          |  326 ++++++++++++++++++++++++++++++++++++
 9 files changed, 650 insertions(+), 1 deletions(-)
 create mode 100644 board/eukrea/cpu9260/Makefile
 create mode 100644 board/eukrea/cpu9260/config.mk
 create mode 100644 board/eukrea/cpu9260/cpu9260.c
 create mode 100644 board/eukrea/cpu9260/cpu9260_led.c
 create mode 100644 include/configs/cpu9260.h
MAINTAINERS +4 -0
index 575a7ec..b8d701a 100644
@@ -515,6 +515,10 @@ Dirk Behme <[email protected]>
 
 	omap3_beagle	ARM CORTEX-A8 (OMAP3530 SoC)
 
+Eric Benard <[email protected]>
+
+	cpu9260	ARM926EJS (AT91SAM9260 SoC)
+
 Rishi Bhattacharya <[email protected]>
 
 	omap5912osk	ARM926EJS
MAKEALL +1 -0
index 020ff73..bdc80fa 100755
@@ -592,6 +592,7 @@ LIST_at91="			\
 	at91sam9m10g45ek	\
 	at91sam9rlek		\
 	cmc_pu2			\
+	cpu9260			\
 	csb637			\
 	kb9202			\
 	meesc			\
Makefile +3 -0
index 090e645..3b7f427 100644
@@ -2813,6 +2813,9 @@ at91sam9rlek_config	:	unconfig
 	fi;
 	@$(MKCONFIG) -a at91sam9rlek arm arm926ejs at91sam9rlek atmel at91
 
+cpu9260_config	:	unconfig
+	@$(MKCONFIG) $(@:_config=) arm arm926ejs cpu9260 eukrea at91
+
 meesc_config	:	unconfig
 	@$(MKCONFIG) $(@:_config=) arm arm926ejs meesc esd at91
 
board/eukrea/cpu9260/Makefile new file +59 -0

59 lines not on this page · raw message

board/eukrea/cpu9260/config.mk new file +1 -0
new file mode 100644
index 0000000..9ce161e
@@ -0,0 +1 @@
+TEXT_BASE = 0x21f00000
board/eukrea/cpu9260/cpu9260.c new file +209 -0

209 lines not on this page · raw message

board/eukrea/cpu9260/cpu9260_led.c new file +46 -0

46 lines not on this page · raw message

cpu/arm926ejs/at91/lowlevel_init.S +1 -1
index 5ed518c..538d3e6 100644
@@ -194,7 +194,7 @@ SMRDATA:
 	.word CONFIG_SYS_PIOD_PPUDR_VAL
 	.word (AT91_BASE_SYS + AT91_PIOD + PIO_ASR)
 	.word CONFIG_SYS_PIOD_PPUDR_VAL
-#elif defined(CONFIG_AT91SAM9261)
+#elif defined(CONFIG_AT91SAM9261) || defined(CONFIG_AT91SAM9260)
 	.word (AT91_BASE_SYS + AT91_PIOC + PIO_PDR)
 	.word CONFIG_SYS_PIOC_PDR_VAL1
 	.word (AT91_BASE_SYS + AT91_PIOC + PIO_PUDR)
include/configs/cpu9260.h new file +326 -0

326 lines not on this page · raw message

Signature · 1 line
1.6.0.4
#1 lore raw

PATCH 1/1 Add support for Eukrea CPU9260 SBC

Dear Eric Benard,

Quoted 3 lines · In message <[email protected]> you wrote:

CPUAT91 is built around Atmel's AT91SAM9260 and has up to 64MB of NOR flash, up to 128MB of SDRAM, up to 2GB of NAND and includes a 10/100 Ethernet PHY in RMII mode.

...

board/eukrea/cpu9260/cpu9260.c line 170

Quoted 6 lines
board/eukrea/cpu9260/cpu9260.c new file +6 -0
+int dram_init(void)
+{
+	gd->bd->bi_dram[0].start = PHYS_SDRAM;
+	gd->bd->bi_dram[0].size = PHYS_SDRAM_SIZE;
+	return 0;
+}

How about using get_ram_size() for auto-sizing and testing?

cpu/arm926ejs/at91/lowlevel_init.S SMRDATA: · line 197

Quoted 10 lines
cpu/arm926ejs/at91/lowlevel_init.S +1 -1
index 5ed518c..538d3e6 100644
@@ -194,7 +194,7 @@ SMRDATA:
 	.word CONFIG_SYS_PIOD_PPUDR_VAL
 	.word (AT91_BASE_SYS + AT91_PIOD + PIO_ASR)
 	.word CONFIG_SYS_PIOD_PPUDR_VAL
-#elif defined(CONFIG_AT91SAM9261)
+#elif defined(CONFIG_AT91SAM9261) || defined(CONFIG_AT91SAM9260)

Please swap to keep sorted:

#elif defined(CONFIG_AT91SAM9260) || defined(CONFIG_AT91SAM9261)

Best regards,

Wolfgang Denk

Signature · 6 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Common sense and a sense of humor  are  the  same  thing,  moving  at
different speeds.  A sense of humor is just common sense, dancing.
                                                        - Clive James
#2 lore raw

PATCH 1/1 Add support for Eukrea CPU9260 SBC

Dear Wolfgang Denk,

Wolfgang Denk a ?crit :

Quoted 5 lines
Quoted 3 lines · In message <[email protected]> you wrote:

CPUAT91 is built around Atmel's AT91SAM9260 and has up to 64MB of NOR flash, up to 128MB of SDRAM, up to 2GB of NAND and includes a 10/100 Ethernet PHY in RMII mode.

...

Quoted 9 lines

board/eukrea/cpu9260/cpu9260.c line 170

Quoted 6 lines
board/eukrea/cpu9260/cpu9260.c new file +6 -0
+int dram_init(void)
+{
+	gd->bd->bi_dram[0].start = PHYS_SDRAM;
+	gd->bd->bi_dram[0].size = PHYS_SDRAM_SIZE;
+	return 0;
+}

How about using get_ram_size() for auto-sizing and testing?

the RAM needs to be properly initialized in low level before being able to test it so its geometry (rom/columns) and timings must be known before being able to detect its size using get_ram_size.
I'm not sure it's possible to use a generic geometry/timings there else u-boot may never relocate.
How would you do this ?

Quoted 13 lines

cpu/arm926ejs/at91/lowlevel_init.S SMRDATA: · line 197

Quoted 10 lines
cpu/arm926ejs/at91/lowlevel_init.S +1 -1
index 5ed518c..538d3e6 100644
@@ -194,7 +194,7 @@ SMRDATA:
 	.word CONFIG_SYS_PIOD_PPUDR_VAL
 	.word (AT91_BASE_SYS + AT91_PIOD + PIO_ASR)
 	.word CONFIG_SYS_PIOD_PPUDR_VAL
-#elif defined(CONFIG_AT91SAM9261)
+#elif defined(CONFIG_AT91SAM9261) || defined(CONFIG_AT91SAM9260)

Please swap to keep sorted:

Quoted 2 lines

#elif defined(CONFIG_AT91SAM9260) || defined(CONFIG_AT91SAM9261)

OK will do.

Thanks
Eric

#3 lore raw

PATCH 1/1 Add support for Eukrea CPU9260 SBC

Dear =?ISO-8859-1?Q?Eric_B=E9nard?=,

Quoted 6 lines · In message <[email protected]> you wrote:
Quoted 3 lines

How about using get_ram_size() for auto-sizing and testing?

the RAM needs to be properly initialized in low level before being able to test it so its geometry (rom/columns) and timings must be known before being able to detect its size using get_ram_size.

Indeed. That's the same on all board - including the many that successfully use get_ram_size()

Quoted 3 lines

I'm not sure it's possible to use a generic geometry/timings there else u-boot may never relocate.
How would you do this ?

Well, normally you have 2 or 3 row/col combinations that have to be considered; you try these in turn until you find one where get_ram_size() returns a non-zero size; if several work, you use that which returns the biggest size.

See for example "board/tqc/tqm8xx/tqm8xx.c" - it tries 8, 9 and 10 column mode...

Best regards,

Wolfgang Denk

Signature · 7 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Lady Nancy Astor:
    "Winston, if you were my husband, I'd put poison in your coffee."
Winston Churchill:
    "Nancy, if you were my wife, I'd drink it."
#4 lore raw

PATCH 1/1 Add support for Eukrea CPU9260 SBC

Dear Wolfgang,

Wolfgang Denk a ?crit :

Quoted 4 lines
Quoted 3 lines

I'm not sure it's possible to use a generic geometry/timings there else u-boot may never relocate.
How would you do this ?

Quoted 8 lines

Well, normally you have 2 or 3 row/col combinations that have to be considered; you try these in turn until you find one where get_ram_size() returns a non-zero size; if several work, you use that which returns the biggest size.

See for example "board/tqc/tqm8xx/tqm8xx.c" - it tries 8, 9 and 10 column mode...

I've tried something very close to what is done in tqm8xx but I don't manage to get something reliable : either it hangs or I get data abort.

After checking the datasheet, I don't understand how we can change the geometry of this SDRAM controler while running from SDRAM : each time we increase the colum number, the address are modified thus we can't execute any more u-boot's code as it's no more where we are expecting it.

Eric

#5 lore raw

PATCH 1/1 Add support for Eukrea CPU9260 SBC

Quoted 5 lines

I've tried something very close to what is done in tqm8xx but I don't manage to get something reliable : either it hangs or I get data abort.

After checking the datasheet, I don't understand how we can change the geometry of this SDRAM controler while running from SDRAM

No, you can't. That part must be done while you run from flash. Actually, this is done in cpu/arm926ejs/at91/lowlevel_init.S, in the table SMRDATA1.

So, with the current code base, you can't autodetect ram size on the atmel 926x.

I have the same problem, as I have boards that ship as either 64M or 128M. I'd configure for 128M and look for aliases, reconfiguring for 64M if needed. This can be done in lowlevel_init.S or by setting up a temporary C environment with sp in static RAM and doing it in C.

In both cases, this doesn't fit the current code base and some refactoring would be needed to go mainline.

/alessandro

#6 lore raw

PATCH 1/1 Add support for Eukrea CPU9260 SBC

Hi Alessandro,

Alessandro Rubini a ?crit :

Quoted 11 lines

So, with the current code base, you can't autodetect ram size on the atmel 926x.

I have the same problem, as I have boards that ship as either 64M or 128M. I'd configure for 128M and look for aliases, reconfiguring for 64M if needed. This can be done in lowlevel_init.S or by setting up a temporary C environment with sp in static RAM and doing it in C.

In both cases, this doesn't fit the current code base and some refactoring would be needed to go mainline.

OK, in our case auto detection is not required so my updated patch gives the choice between both configurations through the Makefile.

Eric

#7 lore raw

PATCH 1/1 Add support for Eukrea CPU9260 SBC

In reply to Alessandro Rubini

Dear Alessandro Rubini,

Quoted 6 lines · In message <[email protected]> you wrote:
Quoted 5 lines

I've tried something very close to what is done in tqm8xx but I don't manage to get something reliable : either it hangs or I get data abort.

After checking the datasheet, I don't understand how we can change the geometry of this SDRAM controler while running from SDRAM

Quoted 6 lines

No, you can't. That part must be done while you run from flash. Actually, this is done in cpu/arm926ejs/at91/lowlevel_init.S, in the table SMRDATA1.

So, with the current code base, you can't autodetect ram size on the atmel 926x.

Argh. So this should be fixed.

Quoted 4 lines

I have the same problem, as I have boards that ship as either 64M or 128M. I'd configure for 128M and look for aliases, reconfiguring for 64M if needed. This can be done in lowlevel_init.S or by setting up a temporary C environment with sp in static RAM and doing it in C.

Other architectures (like PPC) do all this in C. This should be possible on ARM, too. It would be good to reuse existing and well-tested code like get_ram_size() which is fast and still a pretty reliable memory tester.

Quoted 2 lines

In both cases, this doesn't fit the current code base and some refactoring would be needed to go mainline.

Seems so :-(

Who is going to attack this - you, Alessandro?

Best regards,

Wolfgang Denk

Signature · 6 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Operating-system software is the program that  orchestrates  all  the
basic functions of a computer.
- The Wall Street Journal, Tuesday, September 15, 1987, page 40
#8 lore raw

PATCH 1/1 Add support for Eukrea CPU9260 SBC

Quoted 4 lines
Quoted 2 lines

So, with the current code base, you can't autodetect ram size on the atmel 926x.

Argh. So this should be fixed.

Yes.

Quoted 2 lines

Other architectures (like PPC) do all this in C. This should be possible on ARM, too.

Everything is possible. I fear it's not trivial to make a generic ARM implementation, since each platform has its own way to bring up stuff.

Quoted 1 line

Who is going to attack this - you, Alessandro?

I'm going to do it for 9263 next week, for a project I'm following. Then I'll be glad to study the general situation and see if I can suggest anything useful.

/alessandro

#9 lore raw

PATCH 1/1 Add support for Eukrea CPU9260 SBC

In reply to Eric Bénard

Dear =?ISO-8859-1?Q?Eric_B=E9nard?=,

Quoted 5 lines · In message <[email protected]> you wrote:

After checking the datasheet, I don't understand how we can change the geometry of this SDRAM controler while running from SDRAM : each time we increase the colum number, the address are modified thus we can't execute any more u-boot's code as it's no more where we are expecting it.

You cannot - that's obvious. You must not meddle with the SDRAM controler at all while running from this bank of memory. This code should be running from flash (or SRAM, or wherever you boot from).

Best regards,

Wolfgang Denk

Signature · 6 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
If programming was easy, they wouldn't need something as  complicated
as a human being to do it, now would they?
                       - L. Wall & R. L. Schwartz, _Programming Perl_