[U-Boot] at91: smartweb board fails with current HEAD

[U-Boot] at91: smartweb board fails with current HEAD

[Heiko Schocher]

Hello Wenyou,

since your patch:

commit f8b7fff1d5c5297d290acab1e398bd410b0c2e68
Author: Wenyou Yang <wenyou.yang@atmel.com>
Date:   Fri Apr 14 15:01:28 2017 +0800

     serial: atmel_usart: Add clk support

smartweb board fails booting (and I fear at last all other at91 based
siemens boards also). Reverting your patch and U-Boot boots fine again.

First problem is, that I need now a bigger MALLOC_F_LEN space, but the
value is for SPL and U-Boot ... but in SPL I cannot increase this value,
as I have only 4k init ram.

So I need to configure a lower value in SPL as in U-Boot ... this is not
possible at the moment... or?

Second problem: init of the uart seems to go now wrong, I have to add
this patch:

pollux:u-boot-at91sam9g45 hs [master] $ git diff
diff --git a/arch/arm/dts/at91sam9260-smartweb.dts b/arch/arm/dts/at91sam9260-smartweb.dts
index faed763..e59781b 100644
--- a/arch/arm/dts/at91sam9260-smartweb.dts
+++ b/arch/arm/dts/at91sam9260-smartweb.dts
@@ -18,6 +18,7 @@
         compatible = "atmel,at91sam9260", "atmel,at91sam9";

         chosen {
+               u-boot,dm-pre-reloc;
                 stdout-path = &dbgu;
         };

@@ -48,6 +49,7 @@
                         };

                         dbgu: serial at fffff200 {
+                               u-boot,dm-pre-reloc;
                                 status = "okay";
                         };

diff --git a/configs/smartweb_defconfig b/configs/smartweb_defconfig
index 48c8781..a769fc5 100644
--- a/configs/smartweb_defconfig
+++ b/configs/smartweb_defconfig
@@ -7,6 +7,7 @@ CONFIG_TARGET_SMARTWEB=y
  CONFIG_SPL_GPIO_SUPPORT=y
  CONFIG_SPL_LIBCOMMON_SUPPORT=y
  CONFIG_SPL_LIBGENERIC_SUPPORT=y
+CONFIG_SYS_MALLOC_F_LEN=0x2000
  CONFIG_SPL_NAND_SUPPORT=y
  CONFIG_DEFAULT_DEVICE_TREE="at91sam9260-smartweb"
  CONFIG_FIT=y
@@ -33,6 +34,8 @@ CONFIG_CMD_FAT=y
  # CONFIG_DOS_PARTITION is not set
  CONFIG_OF_CONTROL=y
  CONFIG_OF_EMBED=y
+CONFIG_CLK=y
+CONFIG_CLK_AT91=y
  CONFIG_DFU_NAND=y
  # CONFIG_MMC is not set
  CONFIG_USB=y

May the other (none atmel) at91 based boards needs similiar fixes?

Any hint is welcome.

bye,
Heiko
-- 
DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany

[U-Boot] at91: smartweb board fails with current HEAD

[Wenyou.Yang at microchip.com]

> -----Original Message-----
> From: Heiko Schocher [mailto:hs at denx.de]
> Sent: 2017年6月20日 22:05
> To: Wenyou Yang - A41535 <Wenyou.Yang@microchip.com>
> Cc: u-boot at lists.denx.de; Andreas Bießmann <biessmann@corscience.de>
> Subject: at91: smartweb board fails with current HEAD
> 
> Hello Wenyou,
> 
> since your patch:
> 
> commit f8b7fff1d5c5297d290acab1e398bd410b0c2e68
> Author: Wenyou Yang <wenyou.yang@atmel.com>
> Date:   Fri Apr 14 15:01:28 2017 +0800
> 
>      serial: atmel_usart: Add clk support
> 
> smartweb board fails booting (and I fear at last all other at91 based siemens
> boards also). Reverting your patch and U-Boot boots fine again.
> 
> First problem is, that I need now a bigger MALLOC_F_LEN space, but the value is
> for SPL and U-Boot ... but in SPL I cannot increase this value, as I have only 4k
> init ram.
> 
> So I need to configure a lower value in SPL as in U-Boot ... this is not possible at
> the moment... or?

After supporting the driver model and device tree, the size of SPL is beyond 4K.
SPL is not suitable in 4K internal SRAM

> 
> Second problem: init of the uart seems to go now wrong, I have to add this patch:
> 
> pollux:u-boot-at91sam9g45 hs [master] $ git diff diff --git
> a/arch/arm/dts/at91sam9260-smartweb.dts b/arch/arm/dts/at91sam9260-
> smartweb.dts
> index faed763..e59781b 100644
> --- a/arch/arm/dts/at91sam9260-smartweb.dts
> +++ b/arch/arm/dts/at91sam9260-smartweb.dts
> @@ -18,6 +18,7 @@
>          compatible = "atmel,at91sam9260", "atmel,at91sam9";
> 
>          chosen {
> +               u-boot,dm-pre-reloc;
>                  stdout-path = &dbgu;
>          };
> 
> @@ -48,6 +49,7 @@
>                          };
> 
>                          dbgu: serial at fffff200 {
> +                               u-boot,dm-pre-reloc;
>                                  status = "okay";
>                          };
> 
> diff --git a/configs/smartweb_defconfig b/configs/smartweb_defconfig index
> 48c8781..a769fc5 100644
> --- a/configs/smartweb_defconfig
> +++ b/configs/smartweb_defconfig
> @@ -7,6 +7,7 @@ CONFIG_TARGET_SMARTWEB=y
>   CONFIG_SPL_GPIO_SUPPORT=y
>   CONFIG_SPL_LIBCOMMON_SUPPORT=y
>   CONFIG_SPL_LIBGENERIC_SUPPORT=y
> +CONFIG_SYS_MALLOC_F_LEN=0x2000
>   CONFIG_SPL_NAND_SUPPORT=y
>   CONFIG_DEFAULT_DEVICE_TREE="at91sam9260-smartweb"
>   CONFIG_FIT=y
> @@ -33,6 +34,8 @@ CONFIG_CMD_FAT=y
>   # CONFIG_DOS_PARTITION is not set
>   CONFIG_OF_CONTROL=y
>   CONFIG_OF_EMBED=y
> +CONFIG_CLK=y
> +CONFIG_CLK_AT91=y
>   CONFIG_DFU_NAND=y
>   # CONFIG_MMC is not set
>   CONFIG_USB=y
> 
> May the other (none atmel) at91 based boards needs similiar fixes?

Yes, needs similar fixes.

> 
> Any hint is welcome.
> 
> bye,
> Heiko
> --
> DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
> HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany


Best Regards,
Wenyou Yang

[U-Boot] at91: smartweb board fails with current HEAD

[Heiko Schocher]

Hello Wenyou,

Am 21.06.2017 um 02:52 schrieb Wenyou.Yang at microchip.com:
>
>
>> -----Original Message-----
>> From: Heiko Schocher [mailto:hs at denx.de]
>> Sent: 2017年6月20日 22:05
>> To: Wenyou Yang - A41535 <Wenyou.Yang@microchip.com>
>> Cc: u-boot at lists.denx.de; Andreas Bießmann <biessmann@corscience.de>
>> Subject: at91: smartweb board fails with current HEAD
>>
>> Hello Wenyou,
>>
>> since your patch:
>>
>> commit f8b7fff1d5c5297d290acab1e398bd410b0c2e68
>> Author: Wenyou Yang <wenyou.yang@atmel.com>
>> Date:   Fri Apr 14 15:01:28 2017 +0800
>>
>>       serial: atmel_usart: Add clk support
>>
>> smartweb board fails booting (and I fear at last all other at91 based siemens
>> boards also). Reverting your patch and U-Boot boots fine again.
>>
>> First problem is, that I need now a bigger MALLOC_F_LEN space, but the value is
>> for SPL and U-Boot ... but in SPL I cannot increase this value, as I have only 4k
>> init ram.
>>
>> So I need to configure a lower value in SPL as in U-Boot ... this is not possible at
>> the moment... or?
>
> After supporting the driver model and device tree, the size of SPL is beyond 4K.
> SPL is not suitable in 4K internal SRAM

Hmm... no?

I have running a SPL (without DM/DTS support of course) from current mainline
source on the board.
Only problem is, that we need different MALLOC_F_LEN space in SPL and U-Boot
now!

added Simon Glass and Tom Rini, may I oversee some obvious config and/or
they have an idea?

>> Second problem: init of the uart seems to go now wrong, I have to add this patch:
>>
>> pollux:u-boot-at91sam9g45 hs [master] $ git diff diff --git
>> a/arch/arm/dts/at91sam9260-smartweb.dts b/arch/arm/dts/at91sam9260-
>> smartweb.dts
>> index faed763..e59781b 100644
>> --- a/arch/arm/dts/at91sam9260-smartweb.dts
>> +++ b/arch/arm/dts/at91sam9260-smartweb.dts
>> @@ -18,6 +18,7 @@
>>           compatible = "atmel,at91sam9260", "atmel,at91sam9";
>>
>>           chosen {
>> +               u-boot,dm-pre-reloc;
>>                   stdout-path = &dbgu;
>>           };
>>
>> @@ -48,6 +49,7 @@
>>                           };
>>
>>                           dbgu: serial at fffff200 {
>> +                               u-boot,dm-pre-reloc;
>>                                   status = "okay";
>>                           };
>>
>> diff --git a/configs/smartweb_defconfig b/configs/smartweb_defconfig index
>> 48c8781..a769fc5 100644
>> --- a/configs/smartweb_defconfig
>> +++ b/configs/smartweb_defconfig
>> @@ -7,6 +7,7 @@ CONFIG_TARGET_SMARTWEB=y
>>    CONFIG_SPL_GPIO_SUPPORT=y
>>    CONFIG_SPL_LIBCOMMON_SUPPORT=y
>>    CONFIG_SPL_LIBGENERIC_SUPPORT=y
>> +CONFIG_SYS_MALLOC_F_LEN=0x2000
>>    CONFIG_SPL_NAND_SUPPORT=y
>>    CONFIG_DEFAULT_DEVICE_TREE="at91sam9260-smartweb"
>>    CONFIG_FIT=y
>> @@ -33,6 +34,8 @@ CONFIG_CMD_FAT=y
>>    # CONFIG_DOS_PARTITION is not set
>>    CONFIG_OF_CONTROL=y
>>    CONFIG_OF_EMBED=y
>> +CONFIG_CLK=y
>> +CONFIG_CLK_AT91=y
>>    CONFIG_DFU_NAND=y
>>    # CONFIG_MMC is not set
>>    CONFIG_USB=y
>>
>> May the other (none atmel) at91 based boards needs similiar fixes?
>
> Yes, needs similar fixes.

:-(

So your patch broke all none atmel at91 boards?

Just trying on the at91 based corvus board, same problem, but the
same fixes did not yet help ... need more debug time here ...
The taurus board is currently down, after fixing the smartweb and corvus
boards, I hope to get it running again, but I think, same problem.

BTW: This is again a good example how important it is to use automated
tests, may with tbot [1] I only detected this, because I have running
automated weekly U-Boot mainline tests. I think we have a lot of (not
only at91) dead/untested boards in code ...

bye,
Heiko
[1] https://github.com/hsdenx/tbot
     https://tbot.tools/
-- 
DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany

[U-Boot] at91: smartweb board fails with current HEAD

[Simon Glass]

Hi Heiko,

On 20 June 2017 at 22:26, Heiko Schocher <hs@denx.de> wrote:
> Hello Wenyou,
>
> Am 21.06.2017 um 02:52 schrieb Wenyou.Yang at microchip.com:
>>
>>
>>
>>> -----Original Message-----
>>> From: Heiko Schocher [mailto:hs at denx.de]
>>> Sent: 2017年6月20日 22:05
>>> To: Wenyou Yang - A41535 <Wenyou.Yang@microchip.com>
>>> Cc: u-boot at lists.denx.de; Andreas Bießmann <biessmann@corscience.de>
>>> Subject: at91: smartweb board fails with current HEAD
>>>
>>> Hello Wenyou,
>>>
>>> since your patch:
>>>
>>> commit f8b7fff1d5c5297d290acab1e398bd410b0c2e68
>>> Author: Wenyou Yang <wenyou.yang@atmel.com>
>>> Date:   Fri Apr 14 15:01:28 2017 +0800
>>>
>>>       serial: atmel_usart: Add clk support
>>>
>>> smartweb board fails booting (and I fear at last all other at91 based
>>> siemens
>>> boards also). Reverting your patch and U-Boot boots fine again.
>>>
>>> First problem is, that I need now a bigger MALLOC_F_LEN space, but the
>>> value is
>>> for SPL and U-Boot ... but in SPL I cannot increase this value, as I have
>>> only 4k
>>> init ram.
>>>
>>> So I need to configure a lower value in SPL as in U-Boot ... this is not
>>> possible at
>>> the moment... or?
>>
>>
>> After supporting the driver model and device tree, the size of SPL is
>> beyond 4K.
>> SPL is not suitable in 4K internal SRAM
>
>
> Hmm... no?
>
> I have running a SPL (without DM/DTS support of course) from current
> mainline
> source on the board.
> Only problem is, that we need different MALLOC_F_LEN space in SPL and U-Boot
> now!
>
> added Simon Glass and Tom Rini, may I oversee some obvious config and/or
> they have an idea?

I think it would be possible to add CONFIG_SPL_MALLOC_F_LEN as a new
Kconfig option. It could default to the same value as the main
version.

Regards,
Simon

[...]

[U-Boot] at91: smartweb board fails with current HEAD

[Wolfgang Denk]

Dear Wenyou,

In message <F9F4555C4E01D7469D37975B62D0EFBB60A0D3@CHN-SV-EXMX07.mchp-main.com> you wrote:
>
> After supporting the driver model and device tree, the size of SPL is beyond 4K.
> SPL is not suitable in 4K internal SRAM

If you are aware of this situation, you should not post patches that
break a number of existing boards.

Instead, we should search for a compromise.

We will have to do this now. But that should have been done before
upstreaming this code.

Thanks.

Wolfgang Denk

-- 
DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
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
Parkinson's Law:  Work expands to fill the time alloted it.

[U-Boot] at91: smartweb board fails with current HEAD

[Yang, Wenyou]

Hello Wolfgang,


On 2017/6/22 15:07, Wolfgang Denk wrote:
> Dear Wenyou,
>
> In message <F9F4555C4E01D7469D37975B62D0EFBB60A0D3@CHN-SV-EXMX07.mchp-main.com> you wrote:
>> After supporting the driver model and device tree, the size of SPL is beyond 4K.
>> SPL is not suitable in 4K internal SRAM
> If you are aware of this situation, you should not post patches that
> break a number of existing boards.
>
> Instead, we should search for a compromise.
It is my fault, I learned the lesson, I will more careful while posting 
patches.

Thank you for your reminder.
> We will have to do this now. But that should have been done before
> upstreaming this code.
>
> Thanks.
>
> Wolfgang Denk

Best Regards,
Wenyou Yang