[U-Boot] [PATCH v2 0/2] dm: rpi: Move Raspberry Pi to use driver model

[U-Boot] [PATCH v2 0/2] dm: rpi: Move Raspberry Pi to use driver model

[Simon Glass]

This series adds driver model support to the GPIO and serial drivers used
by Raspberry Pi, and moves Raspberry Pi over to driver model.

This requires adding driver model support to the pl01x serial driver, and
replacing the bcm2835 GPIO driver with a driver model version (since there
are no longer clients that don't use driver model).

See u-boot-dm.git branch rpi-working for the tree this is based on.

Changes in v2:
- Adjust header file include to dm/platform_data/...
- Add new patch to move script load address to 0x8000

Simon Glass (2):
  dm: rpi: Move serial to driver model
  dm: rpi: Move script load address to 0x8000

 board/raspberrypi/rpi/rpi.c | 12 ++++++++++++
 include/configs/rpi.h       |  8 ++++----
 2 files changed, 16 insertions(+), 4 deletions(-)

-- 
2.1.0.rc2.206.gedb03e5

[U-Boot] [PATCH v2 1/2] dm: rpi: Move serial to driver model

[Simon Glass]

Adjust the configuration to use the driver model version of the pl01x
serial driver. Add the required platform data.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

Changes in v2:
- Adjust header file include to dm/platform_data/...

 board/raspberrypi/rpi/rpi.c | 12 ++++++++++++
 include/configs/rpi.h       |  6 +++---
 2 files changed, 15 insertions(+), 3 deletions(-)

diff --git a/board/raspberrypi/rpi/rpi.c b/board/raspberrypi/rpi/rpi.c
index aaded88..5ebafd7 100644
--- a/board/raspberrypi/rpi/rpi.c
+++ b/board/raspberrypi/rpi/rpi.c
@@ -24,6 +24,7 @@
 #include <asm/arch/mbox.h>
 #include <asm/arch/sdhci.h>
 #include <asm/global_data.h>
+#include <dm/platform_data/serial_pl01x.h>
 
 DECLARE_GLOBAL_DATA_PTR;
 
@@ -36,6 +37,17 @@ U_BOOT_DEVICE(bcm2835_gpios) = {
 	.platdata = &gpio_platdata,
 };
 
+static const struct pl01x_serial_platdata serial_platdata = {
+	.base = 0x20201000,
+	.type = TYPE_PL011,
+	.clock = 3000000,
+};
+
+U_BOOT_DEVICE(bcm2835_serials) = {
+	.name = "serial_pl01x",
+	.platdata = &serial_platdata,
+};
+
 struct msg_get_arm_mem {
 	struct bcm2835_mbox_hdr hdr;
 	struct bcm2835_mbox_tag_get_arm_mem get_arm_mem;
diff --git a/include/configs/rpi.h b/include/configs/rpi.h
index 71d605c..bef6062 100644
--- a/include/configs/rpi.h
+++ b/include/configs/rpi.h
@@ -35,6 +35,7 @@
 #define CONFIG_DM
 #define CONFIG_CMD_DM
 #define CONFIG_DM_GPIO
+#define CONFIG_DM_SERIAL
 
 /* Memory layout */
 #define CONFIG_NR_DRAM_BANKS		1
@@ -52,6 +53,7 @@
 					 CONFIG_SYS_SDRAM_SIZE - \
 					 GENERATED_GBL_DATA_SIZE)
 #define CONFIG_SYS_MALLOC_LEN		SZ_4M
+#define CONFIG_SYS_MALLOC_F_LEN		(1 << 10)
 #define CONFIG_SYS_MEMTEST_START	0x00100000
 #define CONFIG_SYS_MEMTEST_END		0x00200000
 #define CONFIG_LOADADDR			0x00200000
@@ -93,9 +95,7 @@
 #endif
 
 /* Console UART */
-#define CONFIG_PL011_SERIAL
-#define CONFIG_PL011_CLOCK		3000000
-#define CONFIG_PL01x_PORTS		{ (void *)0x20201000 }
+#define CONFIG_PL01X_SERIAL
 #define CONFIG_CONS_INDEX		0
 #define CONFIG_BAUDRATE			115200
 
-- 
2.1.0.rc2.206.gedb03e5

[U-Boot] [PATCH v2 1/2] dm: rpi: Move serial to driver model

[Stephen Warren]

On 11/24/2014 06:08 PM, Simon Glass wrote:
> Adjust the configuration to use the driver model version of the pl01x
> serial driver. Add the required platform data.

This basically seems to work, but I'll withhold any tested-by/acked-by
until we can work out why writes to address 0 cause U-Boot to fail, with
this patch applied.

[U-Boot] [PATCH v2 2/2] dm: rpi: Move script load address to 0x8000

[Simon Glass]

The current load address of 0 seems to cause problems when driver model
is used. There doesn't seem to be any reason why the script needs to load
to 0, so move it.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

Changes in v2:
- Add new patch to move script load address to 0x8000

 include/configs/rpi.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/configs/rpi.h b/include/configs/rpi.h
index bef6062..deef450 100644
--- a/include/configs/rpi.h
+++ b/include/configs/rpi.h
@@ -171,7 +171,7 @@
  *   for the FDT/DTB to be up to 1M, which is hopefully plenty.
  */
 #define ENV_MEM_LAYOUT_SETTINGS \
-	"scriptaddr=0x00000000\0" \
+	"scriptaddr=0x00008000\0" \
 	"pxefile_addr_r=0x00100000\0" \
 	"kernel_addr_r=0x01000000\0" \
 	"fdt_addr_r=0x02000000\0" \
-- 
2.1.0.rc2.206.gedb03e5

[U-Boot] [PATCH v2 2/2] dm: rpi: Move script load address to 0x8000

[Stephen Warren]

On 11/24/2014 06:08 PM, Simon Glass wrote:
> The current load address of 0 seems to cause problems when driver model
> is used. There doesn't seem to be any reason why the script needs to load
> to 0, so move it.

I'm not particularly tied to the specific $scriptaddr value, but I think
we should find out why writing to address 0 causes failures, rather than
just papering over the issue.

[U-Boot] [PATCH v2 2/2] dm: rpi: Move script load address to 0x8000

[Simon Glass]

Hi Stephen,

On 24 November 2014 at 21:06, Stephen Warren <swarren@wwwdotorg.org> wrote:
> On 11/24/2014 06:08 PM, Simon Glass wrote:
>> The current load address of 0 seems to cause problems when driver model
>> is used. There doesn't seem to be any reason why the script needs to load
>> to 0, so move it.
>
> I'm not particularly tied to the specific $scriptaddr value, but I think
> we should find out why writing to address 0 causes failures, rather than
> just papering over the issue.

I had the same thought, but I have no documentation and have spent
quite a bit of time on it. Do you have any suggestions? Where are the
CPU vectors kept on this platform?

Regards,
Simon

[U-Boot] [PATCH v2 2/2] dm: rpi: Move script load address to 0x8000

[Stephen Warren]

On 11/24/2014 09:10 PM, Simon Glass wrote:
> Hi Stephen,
> 
> On 24 November 2014 at 21:06, Stephen Warren <swarren@wwwdotorg.org> wrote:
>> On 11/24/2014 06:08 PM, Simon Glass wrote:
>>> The current load address of 0 seems to cause problems when driver model
>>> is used. There doesn't seem to be any reason why the script needs to load
>>> to 0, so move it.
>>
>> I'm not particularly tied to the specific $scriptaddr value, but I think
>> we should find out why writing to address 0 causes failures, rather than
>> just papering over the issue.
> 
> I had the same thought, but I have no documentation and have spent
> quite a bit of time on it. Do you have any suggestions? Where are the
> CPU vectors kept on this platform?

I think I found the issue; see my latest message in thread "Re:
Raspberry Pi with driver model".

(I should mention I originally started looking at that code since I
dumped the memory at address 0 immediately after boot, and noticed the
UART address was there for some unexpected reason)

[U-Boot] [PATCH v2 2/2] dm: rpi: Move script load address to 0x8000

[Simon Glass]

Hi Stephen,

On 24 November 2014 at 21:22, Stephen Warren <swarren@wwwdotorg.org> wrote:
> On 11/24/2014 09:10 PM, Simon Glass wrote:
>> Hi Stephen,
>>
>> On 24 November 2014 at 21:06, Stephen Warren <swarren@wwwdotorg.org> wrote:
>>> On 11/24/2014 06:08 PM, Simon Glass wrote:
>>>> The current load address of 0 seems to cause problems when driver model
>>>> is used. There doesn't seem to be any reason why the script needs to load
>>>> to 0, so move it.
>>>
>>> I'm not particularly tied to the specific $scriptaddr value, but I think
>>> we should find out why writing to address 0 causes failures, rather than
>>> just papering over the issue.
>>
>> I had the same thought, but I have no documentation and have spent
>> quite a bit of time on it. Do you have any suggestions? Where are the
>> CPU vectors kept on this platform?
>
> I think I found the issue; see my latest message in thread "Re:
> Raspberry Pi with driver model".
>
> (I should mention I originally started looking at that code since I
> dumped the memory at address 0 immediately after boot, and noticed the
> UART address was there for some unexpected reason)
>

Ah! Well good that you know the platform. I just saw what looked like
vectors and thought nothing of it.

Regards,
Simon