[U-Boot] SPL broken for pandaboard with v2011.09?

[U-Boot] SPL broken for pandaboard with v2011.09?

[Steve Sakoman]

I'm considering using SPL instead of x-load for an upcoming omap project.

To begin learning about SPL I tried building and running SPL for
pandaboard using v2011.09.

With my initial build I got the following output:

 U-Boot SPL 2011.09 (Oct 11 2011 - 10:18:27)
Texas Instruments OMAP4430 ES2.0
SDRAM: identified size not same as expected size identified: 4
expected: 20000000

MMC Device 0 not found
spl: mmc device not found!!
### ERROR ### Please RESET the board ###

Searching the list to see if others were having trouble, I found this
patch and applied it: [U-Boot] [PATCH] omap: spl: fix build break due
to changes in FAT

This eliminated the MMC error, but spl still seems broken:

U-Boot SPL 2011.09 (Oct 11 2011 - 13:42:38)
Texas Instruments OMAP4430 ES2.0
SDRAM: identified size not same as expected size identified: 4
expected: 40000000
????: 738198048,

Has anyone else been able to successfully use SPL on pandaboard with v2011.09?

Steve

[U-Boot] SPL broken for pandaboard with v2011.09?

[John Rigby]

On Tue, Oct 11, 2011 at 3:00 PM, Steve Sakoman <sakoman@gmail.com> wrote:
> I'm considering using SPL instead of x-load for an upcoming omap project.
>
> To begin learning about SPL I tried building and running SPL for
> pandaboard using v2011.09.
>
> With my initial build I got the following output:
>
> ?U-Boot SPL 2011.09 (Oct 11 2011 - 10:18:27)
> Texas Instruments OMAP4430 ES2.0
> SDRAM: identified size not same as expected size identified: 4
> expected: 20000000
>
> MMC Device 0 not found
> spl: mmc device not found!!
> ### ERROR ### Please RESET the board ###
>
> Searching the list to see if others were having trouble, I found this
> patch and applied it: [U-Boot] [PATCH] omap: spl: fix build break due
> to changes in FAT
>
> This eliminated the MMC error, but spl still seems broken:
>
> U-Boot SPL 2011.09 (Oct 11 2011 - 13:42:38)
> Texas Instruments OMAP4430 ES2.0
> SDRAM: identified size not same as expected size identified: 4
> expected: 40000000
> ????: 738198048,
>
> Has anyone else been able to successfully use SPL on pandaboard with v2011.09?
>
You may want to see if the spl is oversize.  Sometime ago the size was
pushed up to 38K but the base was not moved to  accommodate the extra
size so there has been the possibility of problems since then.  I work
around it in my own tree by moving CONFIG_SPL_TEXT_BASE down but that
breaks HS devices.  The patches that start with this email
http://lists.denx.de/pipermail/u-boot/2011-September/102518.html set
the size back to 32K.

--john

[U-Boot] SPL broken for pandaboard with v2011.09?

[Steve Sakoman]

On Wed, Oct 12, 2011 at 3:00 PM, John Rigby <john.rigby@linaro.org> wrote:

>> Has anyone else been able to successfully use SPL on pandaboard with v2011.09?
>>
> You may want to see if the spl is oversize. ?Sometime ago the size was
> pushed up to 38K but the base was not moved to ?accommodate the extra
> size so there has been the possibility of problems since then. ?I work
> around it in my own tree by moving CONFIG_SPL_TEXT_BASE down but that
> breaks HS devices. ?The patches that start with this email
> http://lists.denx.de/pipermail/u-boot/2011-September/102518.html set
> the size back to 32K.

I saw those patches and have those changes incorporated too.

The spl I built is less than 32K.

It seems that the issue is in RAM setup -- the RAM test fails after
one word (hence it reporting 4 byte RAM size).

Is your tree based on v2011.09?  Is it public?

Steve

[U-Boot] SPL broken for pandaboard with v2011.09?

[John Rigby]

On Wed, Oct 12, 2011 at 4:08 PM, Steve Sakoman <sakoman@gmail.com> wrote:
> On Wed, Oct 12, 2011 at 3:00 PM, John Rigby <john.rigby@linaro.org> wrote:
>
>>> Has anyone else been able to successfully use SPL on pandaboard with v2011.09?
>>>
>> You may want to see if the spl is oversize. ?Sometime ago the size was
>> pushed up to 38K but the base was not moved to ?accommodate the extra
>> size so there has been the possibility of problems since then. ?I work
>> around it in my own tree by moving CONFIG_SPL_TEXT_BASE down but that
>> breaks HS devices. ?The patches that start with this email
>> http://lists.denx.de/pipermail/u-boot/2011-September/102518.html set
>> the size back to 32K.
>
> I saw those patches and have those changes incorporated too.
>
> The spl I built is less than 32K.
>
> It seems that the issue is in RAM setup -- the RAM test fails after
> one word (hence it reporting 4 byte RAM size).

My tree has the precalculated ddr timing patch:

U-Boot/SPL: omap4: Make ddr pre-calculated timings as default.

All ddr configurations(geometry/timings) are done automatically
by detecting the device connected at run time. Though this
is a useful feature, making this as a default setting increases
the code size by about 2K bytes. This is quite big, especially
in the case of SPL which runs from a smaller SRAM. So do not
use this feature as the default setting, instead use the
precalculated tables.

Signed-off-by: sricharan <r.sricharan@ti.com>

But you probably picked that up with the spl size fix from the same author.

>
> Is your tree based on v2011.09? ?Is it public?

My tree has a bunch of extra patches on top and is based on rc2.  I
know it works:

http://git.linaro.org/gitweb?p=boot/u-boot-linaro-stable.git;a=summary

My -next tree has had no testing because I only just put it together.
It is nearly identical to the above but rebased to v2011.09 final:

http://git.linaro.org/gitweb?p=boot/u-boot-linaro-next.git;a=summary

Let me know how it goes.

--john

[U-Boot] SPL broken for pandaboard with v2011.09?

[Steve Sakoman]

On Wed, Oct 12, 2011 at 3:23 PM, John Rigby <john.rigby@linaro.org> wrote:

> My tree has the precalculated ddr timing patch:
>
> U-Boot/SPL: omap4: Make ddr pre-calculated timings as default.
>
> All ddr configurations(geometry/timings) are done automatically
> by detecting the device connected at run time. Though this
> is a useful feature, making this as a default setting increases
> the code size by about 2K bytes. This is quite big, especially
> in the case of SPL which runs from a smaller SRAM. So do not
> use this feature as the default setting, instead use the
> precalculated tables.
>
> Signed-off-by: sricharan <r.sricharan@ti.com>
>
> But you probably picked that up with the spl size fix from the same author.

I did, and the mmc/fat fix also makes the same change so there is an
easily fixable conflict when applying all three patches.

>> Is your tree based on v2011.09? ?Is it public?
>
> My tree has a bunch of extra patches on top and is based on rc2. ?I
> know it works:
>
> http://git.linaro.org/gitweb?p=boot/u-boot-linaro-stable.git;a=summary

OK, I will give that a try first.

> My -next tree has had no testing because I only just put it together.
> It is nearly identical to the above but rebased to v2011.09 final:
>
> http://git.linaro.org/gitweb?p=boot/u-boot-linaro-next.git;a=summary

If your rc2 based branch works then I will try this one too to see if
something broke subsequently.

Thanks!

Steve

[U-Boot] SPL broken for pandaboard with v2011.09?

[Steve Sakoman]

On Wed, Oct 12, 2011 at 3:23 PM, John Rigby <john.rigby@linaro.org> wrote:

> My tree has a bunch of extra patches on top and is based on rc2.  I
> know it works:
>
> http://git.linaro.org/gitweb?p=boot/u-boot-linaro-stable.git;a=summary

Very strange - it doesn't work for me :-(

I build from the above repo, commit 5cb383736d0e26cf85336389e234b5372fb500dd

I get similar results:

U-Boot SPL 2011.09-rc2 (Oct 12 2011 - 15:48:01)
Texas Instruments OMAP4430 ES2.0
SDRAM: identified size not same as expected size identified: 4
expected: 40000000

MMC Device 0 not found
spl: mmc device not found!!
### ERROR ### Please RESET the board ###

X-load builds work just fine on my pandaboard, which is Rev EA1.
Perhaps it is a board rev issue.

Steve

[U-Boot] SPL broken for pandaboard with v2011.09?

[Aneesh V]

Hi Steve, John,

Sorry for the late reply.

On Thursday 13 October 2011 05:12 AM, Steve Sakoman wrote:
> On Wed, Oct 12, 2011 at 3:23 PM, John Rigby<john.rigby@linaro.org>  wrote:
>
>> My tree has a bunch of extra patches on top and is based on rc2.  I
>> know it works:
>>
>> http://git.linaro.org/gitweb?p=boot/u-boot-linaro-stable.git;a=summary
>
> Very strange - it doesn't work for me :-(
>
> I build from the above repo, commit 5cb383736d0e26cf85336389e234b5372fb500dd
>
> I get similar results:
>
> U-Boot SPL 2011.09-rc2 (Oct 12 2011 - 15:48:01)
> Texas Instruments OMAP4430 ES2.0
> SDRAM: identified size not same as expected size identified: 4
> expected: 40000000
>
> MMC Device 0 not found
> spl: mmc device not found!!
> ### ERROR ### Please RESET the board ###
>
> X-load builds work just fine on my pandaboard, which is Rev EA1.
> Perhaps it is a board rev issue.

Yes it is. SPL works on everything other than a Panda with OMAP4430
ES2.0. I could reproduce it on an ES2.0 Panda. Please note that an SDP
with OMAP4430 ES2.0 works too, which is rather strange. At the outset,
it looked like a memory issue so I used
CONFIG_SYS_EMIF_PRECALCULATED_TIMING_REGS, which removed memory check
failure. However, it still fails at the same point. The fact that
Lauterbach is not working on my board is not helping my investigations.
Also, the only ES2.0 Panda I could find here is also being used by
another person. So, I have to time-share it with him. I will try to
investigate it further today as soon as I get the board.

br,
Aneesh

>
> Steve

[U-Boot] SPL broken for pandaboard with v2011.09?

[Steve Sakoman]

On Thu, Oct 13, 2011 at 1:18 AM, Aneesh V <aneesh@ti.com> wrote:
>> Perhaps it is a board rev issue.
>
> Yes it is. SPL works on everything other than a Panda with OMAP4430
> ES2.0. I could reproduce it on an ES2.0 Panda. Please note that an SDP
> with OMAP4430 ES2.0 works too, which is rather strange. At the outset,
> it looked like a memory issue so I used
> CONFIG_SYS_EMIF_PRECALCULATED_TIMING_REGS, which removed memory check
> failure. However, it still fails at the same point. The fact that
> Lauterbach is not working on my board is not helping my investigations.
> Also, the only ES2.0 Panda I could find here is also being used by
> another person. So, I have to time-share it with him. I will try to
> investigate it further today as soon as I get the board.

Thanks for looking at this Aneesh!

I too tried both computed and precalculated timing and found that both failed.

X-loader works on my board so I have begun comparing the setup between
x-loader and u-boot-spl.  There are definitely differences in the
register setup.

Will spend some more time on this tomorrow.

Steve