[U-Boot] Issue with running commands

[U-Boot] Issue with running commands

[Sughosh Ganu]

hi,
While testing on hawkboard with the latest commit, i hit an issue of
commands not being accepted.

hawkboard > reset
Unknown command '?????' - try 'help'
hawkboard >

Running git bisect showed that this is caused due to commit 054ea170f271:
cmd_mem: cmp: unify size code paths. Has anyone seen this issue -- i don't
think this is board/arch specific.

-sughosh

[U-Boot] Issue with running commands

[Wolfgang Denk]

Dear Sughosh Ganu,

In message <CAM6zRufYHPfuSNxEAtOaoCTyKpYRGFn0U6+YxkNCtD6_cL2j6g@mail.gmail.com> you wrote:
>
> While testing on hawkboard with the latest commit, i hit an issue of
> commands not being accepted.
>
> hawkboard > reset
> Unknown command '?????' - try 'help> '
> hawkboard >
>
> Running git bisect showed that this is caused due to commit 054ea170f271:
> cmd_mem: cmp: unify size code paths. Has anyone seen this issue -- i don't
> think this is board/arch specific.

You must have a different kind of problem with your board.  Commit
054ea170f271 is only changing the code of common/cmd_mem.c, which is
not used in any way by the reset command or by the command handler or
so - this cannot be causing such an effect; it could only trigger
another problem by changing code size or alignment or such.

Best regards,

Wolfgang Denk

-- 
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
It is easier to write an incorrect program than understand a  correct
one.

[U-Boot] Issue with running commands

[Christian Riesch]

Hi Sughosh,

On Mon, Jul 2, 2012 at 9:24 PM, Sughosh Ganu <urwithsughosh@gmail.com> wrote:
> hi,
> While testing on hawkboard with the latest commit, i hit an issue of
> commands not being accepted.
>
> hawkboard > reset
> Unknown command '?????' - try 'help'
> hawkboard >
>
> Running git bisect showed that this is caused due to commit 054ea170f271:
> cmd_mem: cmp: unify size code paths. Has anyone seen this issue -- i don't
> think this is board/arch specific.

I tried to reproduce this on the calimain board (AM1808 SoC), but
without success. The reset command works fine with current mainline
u-boot.

Regards, Christian

[U-Boot] Issue with running commands

[Sughosh Ganu]

hi Christian,

On Tue, Jul 3, 2012 at 11:52 AM, Christian Riesch <
christian.riesch@omicron.at> wrote:

> Hi Sughosh,
>
> On Mon, Jul 2, 2012 at 9:24 PM, Sughosh Ganu <urwithsughosh@gmail.com>
> wrote:
> > hi,
> > While testing on hawkboard with the latest commit, i hit an issue of
> > commands not being accepted.
> >
> > hawkboard > rese
> > Unknown command '?????' - try 'help'
> > hawkboard >
> >
> > Running git bisect showed that this is caused due to commit 054ea170f271:
> > cmd_mem: cmp: unify size code paths. Has anyone seen this issue -- i
> don't
> > think this is board/arch specific.
>
> I tried to reproduce this on the calimain board (AM1808 SoC), but
> without success. The reset command works fine with current mainline
> u-boot.
>

Thanks for testing this. This looks to be some kind of an alignment issue
on my board, which is getting introduced due to the commit i mentioned. The
said commit is not directly responsible for the problem, but it introduces
some kind of misalignment. I am looking into this -- at such times i so
dearly wish i had a jtag debugger :)

-sughosh

[U-Boot] Issue with running commands

[Graeme Russ]

Hi Sughosh

On Wed, Jul 4, 2012 at 4:35 AM, Sughosh Ganu <urwithsughosh@gmail.com> wrote:
> hi Christian,
>
> On Tue, Jul 3, 2012 at 11:52 AM, Christian Riesch <
> christian.riesch at omicron.at> wrote:
>
>> Hi Sughosh,
>>
>> On Mon, Jul 2, 2012 at 9:24 PM, Sughosh Ganu <urwithsughosh@gmail.com>
>> wrote:
>> > hi,
>> > While testing on hawkboard with the latest commit, i hit an issue of
>> > commands not being accepted.
>> >
>> > hawkboard > rese
>> > Unknown command '?????' - try 'help'
>> > hawkboard >
>> >
>> > Running git bisect showed that this is caused due to commit 054ea170f271:
>> > cmd_mem: cmp: unify size code paths. Has anyone seen this issue -- i
>> don't
>> > think this is board/arch specific.
>>
>> I tried to reproduce this on the calimain board (AM1808 SoC), but
>> without success. The reset command works fine with current mainline
>> u-boot.
>>
>
> Thanks for testing this. This looks to be some kind of an alignment issue
> on my board, which is getting introduced due to the commit i mentioned. The
> said commit is not directly responsible for the problem, but it introduces
> some kind of misalignment. I am looking into this -- at such times i so
> dearly wish i had a jtag debugger :)

I may be totally off track, but I saw a patch recently 'arm: armv7:
add compile option -mno-unaligned-access if available' - I wonder if
its related

Regards,

Graeme

[U-Boot] Issue with running commands

[Sughosh Ganu]

hi Graeme,
On Wed Jul 04, 2012 at 09:28:07AM +1000, Graeme Russ wrote:

<snip>

> >> On Mon, Jul 2, 2012 at 9:24 PM, Sughosh Ganu <urwithsughosh@gmail.com>
> >> wrote:
> >> > hi,
> >> > While testing on hawkboard with the latest commit, i hit an issue of
> >> > commands not being accepted.
> >> >
> >> > hawkboard > rese
> >> > Unknown command '?????' - try 'help'
> >> > hawkboard >
> >> >
> >> > Running git bisect showed that this is caused due to commit 054ea170f271:
> >> > cmd_mem: cmp: unify size code paths. Has anyone seen this issue -- i
> >> don't
> >> > think this is board/arch specific.
> >>
> >> I tried to reproduce this on the calimain board (AM1808 SoC), but
> >> without success. The reset command works fine with current mainline
> >> u-boot.
> >>
> >
> > Thanks for testing this. This looks to be some kind of an alignment issue
> > on my board, which is getting introduced due to the commit i mentioned. The
> > said commit is not directly responsible for the problem, but it introduces
> > some kind of misalignment. I am looking into this -- at such times i so
> > dearly wish i had a jtag debugger :)
> 
> I may be totally off track, but I saw a patch recently 'arm: armv7:
> add compile option -mno-unaligned-access if available' - I wonder if
> its related

  The SoC on my board uses a armv5 based core(arm926ejs), so this
  patch won't apply for my board.

-sughosh

[U-Boot] Issue with running commands

[Prabhakar Lad]

Hi Sughosh,

On Thu, Jul 5, 2012 at 12:18 PM, Sughosh Ganu <urwithsughosh@gmail.com> wrote:
> hi Graeme,
> On Wed Jul 04, 2012 at 09:28:07AM +1000, Graeme Russ wrote:
>
> <snip>
>
>> >> On Mon, Jul 2, 2012 at 9:24 PM, Sughosh Ganu <urwithsughosh@gmail.com>
>> >> wrote:
>> >> > hi,
>> >> > While testing on hawkboard with the latest commit, i hit an issue of
>> >> > commands not being accepted.
>> >> >
>> >> > hawkboard > rese
>> >> > Unknown command '?????' - try 'help'
>> >> > hawkboard >
>> >> >
>> >> > Running git bisect showed that this is caused due to commit 054ea170f271:
>> >> > cmd_mem: cmp: unify size code paths. Has anyone seen this issue -- i
>> >> don't
>> >> > think this is board/arch specific.
>> >>

Tested on Logic Pd's OMAPL138, and found no issues as you mentioned.

Thx,
--Prabhakar Lad

>> >> I tried to reproduce this on the calimain board (AM1808 SoC), but
>> >> without success. The reset command works fine with current mainline
>> >> u-boot.
>> >>
>> >
>> > Thanks for testing this. This looks to be some kind of an alignment issue
>> > on my board, which is getting introduced due to the commit i mentioned. The
>> > said commit is not directly responsible for the problem, but it introduces
>> > some kind of misalignment. I am looking into this -- at such times i so
>> > dearly wish i had a jtag debugger :)
>>
>> I may be totally off track, but I saw a patch recently 'arm: armv7:
>> add compile option -mno-unaligned-access if available' - I wonder if
>> its related
>
>   The SoC on my board uses a armv5 based core(arm926ejs), so this
>   patch won't apply for my board.
>
> -sughosh
> _______________________________________________
> U-Boot mailing list
> U-Boot at lists.denx.de
> http://lists.denx.de/mailman/listinfo/u-boot

[U-Boot] Issue with running commands

[Sughosh Ganu]

On Tue Jul 03, 2012 at 12:54:04AM +0530, Sughosh Ganu wrote:
> hi,
> While testing on hawkboard with the latest commit, i hit an issue of
> commands not being accepted.
> 
> hawkboard > reset
> Unknown command '?????' - try 'help'
> hawkboard >
> 
> Running git bisect showed that this is caused due to commit 054ea170f271:
> cmd_mem: cmp: unify size code paths. Has anyone seen this issue -- i don't
> think this is board/arch specific.

  I was trying to debug the issue, and i found different behaviour
  with different toolchains.

Codesourcery(gcc - 4.5.2, binutils - 2.20.51)

With commit 211e47549b668c7, which is the top of master, i don't hit
the issue anymore.

ELDK-5.2(gcc - 4.6.4, binutils - 2.22)

With this toolchain, i never hit the issue. But enabling DEBUG in
board.c, i get a hang after nand initialisation. I am not able to use
debug messages, as the board starts to boot with inclusion of even a
single debug message.

I am not sure now if this is something specific to my board, as i see
different results with different toolchains. It'd be great if
this change(DEBUG enabled in board.c with an eldk toolchain) can be
checked on other boards(preferably davinci boards). Thanks.

-sughosh

[U-Boot] Issue with running commands

[Sughosh Ganu]

On Tue Jul 03, 2012 at 12:54:04AM +0530, Sughosh Ganu wrote:
> hi,
> While testing on hawkboard with the latest commit, i hit an issue of
> commands not being accepted.
> 
> hawkboard > reset
> Unknown command '?????' - try 'help'
> hawkboard >
> 
> Running git bisect showed that this is caused due to commit 054ea170f271:
> cmd_mem: cmp: unify size code paths. Has anyone seen this issue -- i don't
> think this is board/arch specific.

Worked with my friend during the last week to figure out the
issue. That he had a jtag debugger with him actually helped us track
down the issue. The issue was that the board was using the wrong
version of the nand_read_page function in spl -- we are supposed to
use the page read version with oob read first, while the other version
was being used. This resulted in corruptions being introduced in the
u-boot image that was being copied from the nand.

-sughosh