[U-Boot-Users] FADS and ICU862 Environment problems

[U-Boot-Users] FADS and ICU862 Environment problems

[Nye Liu]

The current FADS code does not seem to restore the environment properly,
and i can't figure out what it SHOULD do.

Right now, relocate_code only relocates up to the end of text
(_init_end), and since CFG_MONITOR_BASE + CONFIG_MONITOR_LEN is after
CFG_ENV_ADDR, ENV_IS_EMBEDDED is set.

Since ENV_IS_EMBEDDED is set, env_reloca_spec in flash.c does NOT
memcpy(env_addr,CFG_ENV_ADDR,CFG_ENV_SIZE).

There are several refereces to environment[] in start.S that are
ifdefed out if CONFIG_FADS or CONFIG_ICU862, which don't look right.

My fix is to remove those ifdefs, and make sure that ENV_IS_EMBEDDED
is NOT set by reducing CFG_MONITOR_LEN to not include the environment.

What is the correct behavior?

TIA

-- 
Nye Liu
nliu at mrv.com
(818) 772-6235x248

"Who would be stupid enough to quote a fictitious character?"
	-- Don Quixote

[U-Boot-Users] FADS and ICU862 Environment problems

[Wolfgang Denk]

In message <20030620014530.GA21955@zumanetworks.com> you wrote:
> The current FADS code does not seem to restore the environment properly,
> and i can't figure out what it SHOULD do.

It _should_ work, of course :-)

> Right now, relocate_code only relocates up to the end of text
> (_init_end), and since CFG_MONITOR_BASE + CONFIG_MONITOR_LEN is after
> CFG_ENV_ADDR, ENV_IS_EMBEDDED is set.

This  is  probably  a  result  of  our  attempt   to   get   rid   of
CONFIG_MONITOR_LEN  as  far  as  possible.  Now  it  is  ONLY used to
determine if the environment is embedded, or not.

> Since ENV_IS_EMBEDDED is set, env_reloca_spec in flash.c does NOT
> memcpy(env_addr,CFG_ENV_ADDR,CFG_ENV_SIZE).

Correcnt.

> There are several refereces to environment[] in start.S that are
> ifdefed out if CONFIG_FADS or CONFIG_ICU862, which don't look right.

Why not? I cannot speak for the FADS, of course. Won't touch no F***
ADS...

> My fix is to remove those ifdefs, and make sure that ENV_IS_EMBEDDED
> is NOT set by reducing CFG_MONITOR_LEN to not include the environment.
> 
> What is the correct behavior?

Setting CFG_MONITOR_LEN correctly is correct. But why removing  which
#ifdefs ?

Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
In the beginning, I was made. I didn't ask to be made. No one consul-
ted with me or considered my feelings  in  this  matter.  But  if  it
brought  some  passing fancy to some lowly humans as they haphazardly
pranced their way through life's mournful jungle, then so be it.
- Marvin the Paranoid Android

[U-Boot-Users] FADS and ICU862 Environment problems

[Nye Liu]

On Sat, Jun 21, 2003 at 02:21:36AM +0200, Wolfgang Denk wrote:
 
> Setting CFG_MONITOR_LEN correctly is correct. But why removing  which
> #ifdefs ?

if ENV_IS_EMBEDDED is set, nothing in environment.c gets compiled,
and cpu/mpc8xx/start.S has a few 

#if defined(CONFIG_FADS) || defined(CONFIG_ICU862)

directives which surround GOT_ENTRY(environment) and GOT(environment)
references...

safe to remove these?

btw. wolfgang, this patch is growing larger and larger every week, 
I'm not sure how easy it is going to be for me to split it up
into digestible chunks for you...

-- 
Nye Liu
nliu at mrv.com
(818) 772-6235x248

"Who would be stupid enough to quote a fictitious character?"
	-- Don Quixote

[U-Boot-Users] FADS and ICU862 Environment problems

[Wolfgang Denk]

In message <20030621002854.GA27308@zumanetworks.com> you wrote:
> On Sat, Jun 21, 2003 at 02:21:36AM +0200, Wolfgang Denk wrote:
>  
> > Setting CFG_MONITOR_LEN correctly is correct. But why removing  which
> > #ifdefs ?
> 
> if ENV_IS_EMBEDDED is set, nothing in environment.c gets compiled,
> and cpu/mpc8xx/start.S has a few 
> 
> #if defined(CONFIG_FADS) || defined(CONFIG_ICU862)
> 
> directives which surround GOT_ENTRY(environment) and GOT(environment)
> references...
> 
> safe to remove these?

No.

If you look at the u-boot.map file when building for the ICU862 board
you can see this:

...
                0x40f22b3c                __init_end = .
                0x40f22b3c                __bss_start = .
    
.bss            0x40f22c00     0x417c  
...
                0x40f26c30                flash_info
                0x40f40000                . =  ALIGN (0x40000)

.ppcenv         0x40f40000     0x4000
 common/environment.o(.ppcenv)
 .ppcenv        0x40f40000     0x4000 common/environment.o
                0x40f40000                environment
                0x40f44000                _end = .
                0x40f44000                PROVIDE (end, .)

The #ifdef in "cpu/mpc8xx/start.S" makes sure that  clear_bss()  does
not attempt to clear any memory that belongs to the environment.


> btw. wolfgang, this patch is growing larger and larger every week, 
> I'm not sure how easy it is going to be for me to split it up
> into digestible chunks for you...

Split it now, and submit any stable parts of it.


Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
When the ax entered the forest, the trees said, "The handle is one of
us!"                                               -- Turkish proverb

[U-Boot-Users] FADS and ICU862 Environment problems

[Nye Liu]

On Sun, Jun 22, 2003 at 06:22:03PM +0200, Wolfgang Denk wrote:
> In message <20030621002854.GA27308@zumanetworks.com> you wrote:
> > On Sat, Jun 21, 2003 at 02:21:36AM +0200, Wolfgang Denk wrote:
> >  
> > > Setting CFG_MONITOR_LEN correctly is correct. But why removing  which
> > > #ifdefs ?
> > 
> > if ENV_IS_EMBEDDED is set, nothing in environment.c gets compiled,
> > and cpu/mpc8xx/start.S has a few 
> > 
> > #if defined(CONFIG_FADS) || defined(CONFIG_ICU862)
> > 
> > directives which surround GOT_ENTRY(environment) and GOT(environment)
> > references...
> > 
> > safe to remove these?
> 
> No.
> 
> If you look at the u-boot.map file when building for the ICU862 board
> you can see this:
> 
> ...
>                 0x40f22b3c                __init_end = .
>                 0x40f22b3c                __bss_start = .
>     
> .bss            0x40f22c00     0x417c  
> ...
>                 0x40f26c30                flash_info
>                 0x40f40000                . =  ALIGN (0x40000)
> 
> .ppcenv         0x40f40000     0x4000
>  common/environment.o(.ppcenv)
>  .ppcenv        0x40f40000     0x4000 common/environment.o
>                 0x40f40000                environment
>                 0x40f44000                _end = .
>                 0x40f44000                PROVIDE (end, .)
> 

shouldn't the ifdef depend on ENV_IS_EMBEDDED? If that is set, even 
in the ICU862 case, nothing in environment.c gets compiled (ALL of 
the code is ifdef'ed out), and "environment" will be unresolved..

> The #ifdef in "cpu/mpc8xx/start.S" makes sure that  clear_bss()  does
> not attempt to clear any memory that belongs to the environment.

but only if ENV_IS_EMBEDDED is NOT set... unless i'm missing something
obvious?

> > btw. wolfgang, this patch is growing larger and larger every week, 
> > I'm not sure how easy it is going to be for me to split it up
> > into digestible chunks for you...
> 
> Split it now, and submit any stable parts of it.

Too much is tangled together, i will try, but it doesn't look good.

TIA

-- 
Nye Liu
nliu at mrv.com
(818) 772-6235x248

"Who would be stupid enough to quote a fictitious character?"
	-- Don Quixote

[U-Boot-Users] FADS and ICU862 Environment problems

[Wolfgang Denk]

In message <20030623212553.GA10209@zumanetworks.com> you wrote:
>
> shouldn't the ifdef depend on ENV_IS_EMBEDDED? If that is set, even 
> in the ICU862 case, nothing in environment.c gets compiled (ALL of 
> the code is ifdef'ed out), and "environment" will be unresolved..

ENV_IS_EMBEDDED?  is  not  set  on  the  ICU862  board,   since   the
environment  is  located  after  the end of the U-Boot image, but not
embedded within it.

> > The #ifdef in "cpu/mpc8xx/start.S" makes sure that  clear_bss()  does
> > not attempt to clear any memory that belongs to the environment.
> 
> but only if ENV_IS_EMBEDDED is NOT set... unless i'm missing something
> obvious?

I see no discrepancy - ENV_IS_EMBEDDED is NOT set for ICU862

> > Split it now, and submit any stable parts of it.
> 
> Too much is tangled together, i will try, but it doesn't look good.

I know how that feels. Split splittable parts, then.


Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
Calm down, it's *__only* ones and zeroes.

[U-Boot-Users] FADS and ICU862 Environment problems

[Nye Liu]

On Mon, Jun 23, 2003 at 11:35:48PM +0200, Wolfgang Denk wrote:
> In message <20030623212553.GA10209@zumanetworks.com> you wrote:
> >
> > shouldn't the ifdef depend on ENV_IS_EMBEDDED? If that is set, even 
> > in the ICU862 case, nothing in environment.c gets compiled (ALL of 
> > the code is ifdef'ed out), and "environment" will be unresolved..
> 
> ENV_IS_EMBEDDED?  is  not  set  on  the  ICU862  board,   since   the
> environment  is  located  after  the end of the U-Boot image, but not
> embedded within it.
> 
> > > The #ifdef in "cpu/mpc8xx/start.S" makes sure that  clear_bss()  does
> > > not attempt to clear any memory that belongs to the environment.
> > 
> > but only if ENV_IS_EMBEDDED is NOT set... unless i'm missing something
> > obvious?
> 
> I see no discrepancy - ENV_IS_EMBEDDED is NOT set for ICU862

I'm sorry, i must be dense, i'm still not getting this (i do not know
anything about the ICU board) ...

So, leave in the ifdef ICU862 case, but what about FADS (and other
boards) which may or may not have ENV_IS_EMBEDDED set?

shouldn't they also NOT initialize past the bss into environment.o if
ENV_IS_EMBEDDED is not set, but not refer to the "environment" symbol if
ENV_IS_EMBEDDED IS set?

i.e.

#if !defined(ENV_IS_EMBEDDED) || defined(CONFIG_ICU862)

Sorry that this is taking so long to explain, but now i'm hopelessly
confused ;/

-- 
Nye Liu
nliu at mrv.com
(818) 772-6235x248

"Who would be stupid enough to quote a fictitious character?"
	-- Don Quixote

[U-Boot-Users] FADS and ICU862 Environment problems

[Wolfgang Denk]

In message <20030623214211.GA10366@zumanetworks.com> you wrote:
>
> > I see no discrepancy - ENV_IS_EMBEDDED is NOT set for ICU862
> 
> I'm sorry, i must be dense, i'm still not getting this (i do not know
> anything about the ICU board) ...
> 
> So, leave in the ifdef ICU862 case, but what about FADS (and other
> boards) which may or may not have ENV_IS_EMBEDDED set?

I don't know anything about the FADS board. I tried to use one  once.
Probably I will never touch one again.

> shouldn't they also NOT initialize past the bss into environment.o if
> ENV_IS_EMBEDDED is not set, but not refer to the "environment" symbol if
> ENV_IS_EMBEDDED IS set?

It depends on the flash layout,  i.  e.  where  your  environment  is
located.

> #if !defined(ENV_IS_EMBEDDED) || defined(CONFIG_ICU862)

No. "!defined(ENV_IS_EMBEDDED)" might include  situations  where  the
environment is in EEPROM or so.


> Sorry that this is taking so long to explain, but now i'm hopelessly
> confused ;/

Me too. What's wrong with the existing  code?  Where  does  it  cause
problems?


Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
Cleverness and Style have no place in getting a project completed.
                                                  -- Tom Christiansen

[U-Boot-Users] FADS and ICU862 Environment problems

[Nye Liu]

On Tue, Jun 24, 2003 at 08:24:30PM +0200, Wolfgang Denk wrote:
> Me too. What's wrong with the existing  code?  Where  does  it  cause
> problems?

in FADS, where ENV_IS_EMBEDDED may or may not be set, depending on how
big MONITOR_LENGTH is... (i.e. past FLASH_ENV_BASE or not)

-- 
Nye Liu
nliu at mrv.com
(818) 772-6235x248

"Who would be stupid enough to quote a fictitious character?"
	-- Don Quixote

[U-Boot-Users] FADS and ICU862 Environment problems

[Wolfgang Denk]

In message <20030624202853.GA15860@zumanetworks.com> you wrote:
> On Tue, Jun 24, 2003 at 08:24:30PM +0200, Wolfgang Denk wrote:
> > Me too. What's wrong with the existing  code?  Where  does  it  cause
> > problems?
> 
> in FADS, where ENV_IS_EMBEDDED may or may not be set, depending on how
> big MONITOR_LENGTH is... (i.e. past FLASH_ENV_BASE or not)

But this is exactly the point of defining  MONITOR_LENGTH:  _if_  the
environment  shall  be  embedded in the U-Boot image (ENV_IS_EMBEDDED
will be set), then MONITOR_LENGTH must be big enough to  include  the
environment;  and  if  the  environment  shall be in a separate flash
sector, then MONITOR_LENGTH must be small enough NOT to  include  the
environment.

I don't know who designed it that way (because I don't care about the
FADS board), but according to "board/fads/u-boot.lds" the environment
on this board is aligned on  a  256kB  boundary  _after_  the  U-Boot
image.  So  MONITOR_LENGTH  must  not  include  the  environment, and
ENV_IS_EMBEDDED shall not be set.


Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
This is now.  Later is later.

[U-Boot-Users] FADS and ICU862 Environment problems

[Nye Liu]

On Tue, Jun 24, 2003 at 10:38:22PM +0200, Wolfgang Denk wrote:
> In message <20030624202853.GA15860@zumanetworks.com> you wrote:
> > On Tue, Jun 24, 2003 at 08:24:30PM +0200, Wolfgang Denk wrote:
> > > Me too. What's wrong with the existing  code?  Where  does  it  cause
> > > problems?
> > 
> > in FADS, where ENV_IS_EMBEDDED may or may not be set, depending on how
> > big MONITOR_LENGTH is... (i.e. past FLASH_ENV_BASE or not)
> 
> But this is exactly the point of defining  MONITOR_LENGTH:  _if_  the
> environment  shall  be  embedded in the U-Boot image (ENV_IS_EMBEDDED
> will be set), then MONITOR_LENGTH must be big enough to  include  the
> environment;  and  if  the  environment  shall be in a separate flash
> sector, then MONITOR_LENGTH must be small enough NOT to  include  the
> environment.
> 
> I don't know who designed it that way (because I don't care about the
> FADS board), but according to "board/fads/u-boot.lds" the environment
> on this board is aligned on  a  256kB  boundary  _after_  the  U-Boot
> image.  So  MONITOR_LENGTH  must  not  include  the  environment, and
> ENV_IS_EMBEDDED shall not be set.

In which case NOTHING in environment.c is compiled, and "environment"
goes unresolved....

should it JUST be #ifdef CONFIG_ICU862?

-- 
Nye Liu
nliu at mrv.com
(818) 772-6235x248

"Who would be stupid enough to quote a fictitious character?"
	-- Don Quixote

[U-Boot-Users] FADS and ICU862 Environment problems

[Nye Liu]

On Sun, Jun 22, 2003 at 06:22:03PM +0200, Wolfgang Denk wrote:
> No.
> 
> If you look at the u-boot.map file when building for the ICU862 board
> you can see this:
> 
> ...
>                 0x40f22b3c                __init_end = .
>                 0x40f22b3c                __bss_start = .
>     
> .bss            0x40f22c00     0x417c  
> ...
>                 0x40f26c30                flash_info
>                 0x40f40000                . =  ALIGN (0x40000)
> 
> .ppcenv         0x40f40000     0x4000
>  common/environment.o(.ppcenv)
>  .ppcenv        0x40f40000     0x4000 common/environment.o
>                 0x40f40000                environment
>                 0x40f44000                _end = .
>                 0x40f44000                PROVIDE (end, .)
> 
> The #ifdef in "cpu/mpc8xx/start.S" makes sure that  clear_bss()  does
> not attempt to clear any memory that belongs to the environment.

but later, you write:

> I see no discrepancy - ENV_IS_EMBEDDED is NOT set for ICU862

how does environment.o have any data; it is totally ifdef'ed out
if ENV_IS_EMBEDDED is not set...?

I am totally confused now ;/

-- 
Nye Liu
nliu at mrv.com
(818) 772-6235x248

"Who would be stupid enough to quote a fictitious character?"
	-- Don Quixote

[U-Boot-Users] FADS and ICU862 Environment problems

[Wolfgang Denk]

In message <20030624235953.GA16810@zumanetworks.com> you wrote:
>
> > I don't know who designed it that way (because I don't care about the
> > FADS board), but according to "board/fads/u-boot.lds" the environment
> > on this board is aligned on  a  256kB  boundary  _after_  the  U-Boot
> > image.  So  MONITOR_LENGTH  must  not  include  the  environment, and
> > ENV_IS_EMBEDDED shall not be set.
> 
> In which case NOTHING in environment.c is compiled, and "environment"
> goes unresolved....

But this is normal for a non-embedded environment. If you don;t embed
it with the U-Boot image it will not be initialized after  installing
the  image  anyway  (because  it is not included), so there is little
sense initializing it at compile time.

There are several other options: in cases where  the  environment  is
directly  adjecent  to  the  U-Boot image (it looks as if the FADS is
such a configuration) you can "include" the environment, producing  a
bigger image which then contains an initialized environment.

It is up for the board maintainer to decide what he wants.

Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
All I ask is a chance to prove that money can't make me happy.

[U-Boot-Users] FADS and ICU862 Environment problems

[Wolfgang Denk]

In message <20030625002749.GA17011@zumanetworks.com> you wrote:
>
> > I see no discrepancy - ENV_IS_EMBEDDED is NOT set for ICU862
> 
> how does environment.o have any data; it is totally ifdef'ed out
> if ENV_IS_EMBEDDED is not set...?

It does not. On the ICU862 the U-Boot image is at the  lower  end  of
the enfironment, while the environment is at the upper end (IIRC this
board  uses  top  boot sector flash memory). The environment does NOT
get installed when installing the U-Boot image. So there is  no  need
to put any initialized data in it.

It will be initialized as soon as the user types  "saveenv"  for  the
first time. Until then, the default environment is used.

> I am totally confused now ;/

I'm sorry.


Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
"Whoever undertakes to set himself up as a judge of Truth  and  Know-
ledge is shipwrecked by the laughter of the gods."  - Albert Einstein

[U-Boot-Users] FADS and ICU862 Environment problems

[Nye Liu]

On Thu, Jun 26, 2003 at 08:50:27AM +0200, Wolfgang Denk wrote:
> In message <20030624235953.GA16810@zumanetworks.com> you wrote:
> >
> > > I don't know who designed it that way (because I don't care about the
> > > FADS board), but according to "board/fads/u-boot.lds" the environment
> > > on this board is aligned on  a  256kB  boundary  _after_  the  U-Boot
> > > image.  So  MONITOR_LENGTH  must  not  include  the  environment, and
> > > ENV_IS_EMBEDDED shall not be set.
> > 
> > In which case NOTHING in environment.c is compiled, and "environment"
> > goes unresolved....

unresolved, meaning:

ld -Bstatic -T /home/nyet/src/local/u-boot/board/fads/u-boot.lds -Ttext 0x02800000 -n cpu/mpc8xx/start.o \
                --start-group board/fads/libfads.a cpu/mpc8xx/libmpc8xx.a lib_ppc/libppc.a fs/jffs2/libjffs2.a fs/fdos/libfdos.a net/libnet.a disk/libdisk.a rtc/librtc.a dtt/libdtt.a drivers/libdrivers.a post/libpost.a post/cpu/libcpu.a common/libcommon.a lib_generic/libgeneric.a --end-group \
                -Map u-boot.map -o u-boot
cpu/mpc8xx/start.o(.got2+0x24):/local/nyet-build/u-boot-mrv/cpu/mpc8xx/start.S:99: undefined reference to `environment'
make: *** [u-boot] Error 1

> 
> But this is normal for a non-embedded environment. If you don;t embed
> it with the U-Boot image it will not be initialized after  installing
> the  image  anyway  (because  it is not included), so there is little
> sense initializing it at compile time.
> 
> There are several other options: in cases where  the  environment  is
> directly  adjecent  to  the  U-Boot image (it looks as if the FADS is
> such a configuration) you can "include" the environment, producing  a
> bigger image which then contains an initialized environment.
> 
> It is up for the board maintainer to decide what he wants.
> 
> Best regards,
> 
> Wolfgang Denk
> 
> -- 
> Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
> Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
> All I ask is a chance to prove that money can't make me happy.

-- 
Nye Liu
nliu at mrv.com
(818) 772-6235x248

"Who would be stupid enough to quote a fictitious character?"
	-- Don Quixote

[U-Boot-Users] FADS and ICU862 Environment problems

[Wolfgang Denk]

In message <20030626214057.GA28486@zumanetworks.com> you wrote:
> 
> unresolved, meaning:
> 
> ld -Bstatic -T /home/nyet/src/local/u-boot/board/fads/u-boot.lds -Ttext 0x02800000 -n cpu/mpc8xx/start.o \
>                 --start-group board/fads/libfads.a cpu/mpc8xx/libmpc8xx.a lib_ppc/libppc.a fs/jffs2/libjffs2.a fs/fdos/libfdos.a net/libnet.a disk/libdisk.a rtc/librtc.a dtt/libdtt.a drivers/libdrivers.a post/libpost.a post/cpu/libcpu.a common/libcommon
> .a lib_generic/libgeneric.a --end-group \
>                 -Map u-boot.map -o u-boot
> cpu/mpc8xx/start.o(.got2+0x24):/local/nyet-build/u-boot-mrv/cpu/mpc8xx/start.S:99: undefined reference to `environment'
> make: *** [u-boot] Error 1

Strange. I see no such problems:

-> MAKEALL FADS823 FADS850SAR FADS860T ICU862
rm -f include/config.h include/config.mk
Configuring for FADS823 board...
   text    data     bss     dec     hex filename
  95364  323632   16372  435368   6a4a8 u-boot
rm -f include/config.h include/config.mk
Configuring for FADS850SAR board...
   text    data     bss     dec     hex filename
  87028  284488   16296  387812   5eae4 u-boot
rm -f include/config.h include/config.mk
Configuring for FADS860T board...
   text    data     bss     dec     hex filename
  87352   38776   16336  142464   22c80 u-boot
rm -f include/config.h include/config.mk
Configuring for ICU862 board...
   text    data     bss     dec     hex filename
 106712   47928   16764  171404   29d8c u-boot

No errors, no warnings.

I think I lost you. Please let's restart from zero.

What exactly is your problem with the current code?


Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
"We Americans, we're a simple people... but piss us  off,  and  we'll
bomb your cities."           - Robin Williams, _Good Morning Vietnam_

[U-Boot-Users] FADS and ICU862 Environment problems

[Nye Liu]

On Fri, Jun 27, 2003 at 12:03:19AM +0200, Wolfgang Denk wrote:
> In message <20030626214057.GA28486@zumanetworks.com> you wrote:
> > 
> > unresolved, meaning:
> > 
> > ld -Bstatic -T /home/nyet/src/local/u-boot/board/fads/u-boot.lds -Ttext 0x02800000 -n cpu/mpc8xx/start.o \
> >                 --start-group board/fads/libfads.a cpu/mpc8xx/libmpc8xx.a lib_ppc/libppc.a fs/jffs2/libjffs2.a fs/fdos/libfdos.a net/libnet.a disk/libdisk.a rtc/librtc.a dtt/libdtt.a drivers/libdrivers.a post/libpost.a post/cpu/libcpu.a common/libcommon
> > .a lib_generic/libgeneric.a --end-group \
> >                 -Map u-boot.map -o u-boot
> > cpu/mpc8xx/start.o(.got2+0x24):/local/nyet-build/u-boot-mrv/cpu/mpc8xx/start.S:99: undefined reference to `environment'
> > make: *** [u-boot] Error 1
> 
> Strange. I see no such problems:
> 
> -> MAKEALL FADS823 FADS850SAR FADS860T ICU862
> rm -f include/config.h include/config.mk
> Configuring for FADS823 board...
>    text    data     bss     dec     hex filename
>   95364  323632   16372  435368   6a4a8 u-boot
> rm -f include/config.h include/config.mk
> Configuring for FADS850SAR board...
>    text    data     bss     dec     hex filename
>   87028  284488   16296  387812   5eae4 u-boot
> rm -f include/config.h include/config.mk
> Configuring for FADS860T board...
>    text    data     bss     dec     hex filename
>   87352   38776   16336  142464   22c80 u-boot
> rm -f include/config.h include/config.mk
> Configuring for ICU862 board...
>    text    data     bss     dec     hex filename
>  106712   47928   16764  171404   29d8c u-boot
> 
> No errors, no warnings.
> 
> I think I lost you. Please let's restart from zero.
> 
> What exactly is your problem with the current code?

Ok. To summarize:

If ENV_IS_EMBEDDED is set in FADS, it does not get copied during
relocation due to common/env_flash.c

If ENV_IS_EMBEDDED is NOT set (by reducing MONITOR_LEN to not
include ENV), i get this error.

-- 
Nye Liu
nliu at mrv.com
(818) 772-6235x248

"Who would be stupid enough to quote a fictitious character?"
	-- Don Quixote

[U-Boot-Users] FADS and ICU862 Environment problems

[Wolfgang Denk]

In message <20030626221032.GA28649@zumanetworks.com> you wrote:
>
> Ok. To summarize:
> 
> If ENV_IS_EMBEDDED is set in FADS, it does not get copied during
> relocation due to common/env_flash.c
> 
> If ENV_IS_EMBEDDED is NOT set (by reducing MONITOR_LEN to not
> include ENV), i get this error.

Sorry, I don't know much about the FADS (and actually I really  don;t
want to care about broken hardware like this; but anyway).

Which specific board configuration are you using?

Checking "include/configs/FADS860T.h" I see:

#define CFG_MONITOR_LEN         (272 << 10)     /* Reserve 272 kB for Monitor */
...
#define CFG_ENV_IS_IN_FLASH     1
#define CFG_ENV_OFFSET          0x00040000
#define CFG_ENV_SIZE            0x4000  /* Total Size of Environment Sector */
#define CFG_ENV_SECT_SIZE       0x40000 /* see README - env sector total size */


After building the board "u-boot.map" says:

...
.text           0x02800000    0x15538 
 cpu/mpc8xx/start.o(.text)
 .text          0x02800000     0x2468 cpu/mpc8xx/start.o
                0x028001c8                _start_of_vectors
...
                0x0281b77c                __init_end = .
                0x0281b77c                __bss_start = .
...
.ppcenv         0x02840000     0x4000
 common/environment.o(.ppcenv)
 .ppcenv        0x02840000     0x4000 common/environment.o
                0x02840000                environment
                0x02844000                _end = .
                0x02844000                PROVIDE (end, .)
Address of section .text set to 0x2800000
...


So the text segment starts at 0x02800000, and one sector (or 256  kB)
later @ 0x02840000  the  environment  follows.  CFG_MONITOR_LEN is
bigger than __bss_start, so ENV_IS_EMBEDDED should be set.

You are right,  the  environment  will  NOT  be  copied  in  function
env_relocate_spec()  in "common/env_flash.c", but it _will_ be copied
in  function  env_relocate()   in   "common/env_common.c"   (assuming
gd->env_valid is zero).


I think this is correct. What am I missing?

Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
grep me no patterns and I'll tell you no lines.

[U-Boot-Users] FADS and ICU862 Environment problems

[Nye Liu]

On Fri, Jun 27, 2003 at 12:31:29AM +0200, Wolfgang Denk wrote:
> In message <20030626221032.GA28649@zumanetworks.com> you wrote:
> >
> > Ok. To summarize:
> > 
> > If ENV_IS_EMBEDDED is set in FADS, it does not get copied during
> > relocation due to common/env_flash.c
> > 
> > If ENV_IS_EMBEDDED is NOT set (by reducing MONITOR_LEN to not
> > include ENV), i get this error.
> 
....  snip
> You are right,  the  environment  will  NOT  be  copied  in  function
> env_relocate_spec()  in "common/env_flash.c", but it _will_ be copied
> in  function  env_relocate()   in   "common/env_common.c"   (assuming
> gd->env_valid is zero).
> 
> 
> I think this is correct. What am I missing?

but that is the default environment that is copied, not the saved
copy stored via saveenv? When is the saved environment restored?
relocate_code() never copies it since it is AFTER __init_end due
to the way u-boot.lds locates environment.o after .bss

-- 
Nye Liu
nliu at mrv.com
(818) 772-6235x248

"Who would be stupid enough to quote a fictitious character?"
	-- Don Quixote

[U-Boot-Users] FADS and ICU862 Environment problems

[Wolfgang Denk]

In message <20030626224147.GA28873@zumanetworks.com> you wrote:
>
> > I think this is correct. What am I missing?
> 
> but that is the default environment that is copied, not the saved
> copy stored via saveenv? When is the saved environment restored?
> relocate_code() never copies it since it is AFTER __init_end due
> to the way u-boot.lds locates environment.o after .bss

Argh... now I see. Well, I think you are right.

Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
Generally speaking, there are other ways to accomplish whatever it is
that you think you need ...                               - Doug Gwyn

[U-Boot-Users] a small problem

[Yabo Ding]

In post/memory.c file


int memory_post_test (int flags)
{
	int ret = 0;
	DECLARE_GLOBAL_DATA_PTR;
	bd_t *bd = gd->bd;
	unsigned long memsize = (bd->bi_memsize >= 256 << 20 ?
				 256 << 20 : bd->bi_memsize) - (1 << 20);


	if (flags & POST_POWERFAIL) {
		ret = memory_post_tests (CFG_SDRAM_BASE, memsize);
	} else {			/* POST_POWERNORMAL */

		unsigned long i;

		for (i = 0; i < (memsize >> 20) && ret == 0; i++) {
			if (ret == 0)
				ret = memory_post_tests (i << 20, 0x800);
			if (ret == 0)
				ret = memory_post_tests ((i << 20) + 0xff800, 0x800);
		}
	}

	return ret;
}


I think the line should be:

	unsigned long memsize = (bd->bi_memsize >= 257 << 20 ?
				 256 << 20 : bd->bi_memsize) - (1 << 20);





Yabo Ding
bobfree@bonbon.net

[U-Boot-Users] Re: a small problem

[Wolfgang Denk]

In message <3EFBBE6A.6000200@bonbon.net> you wrote:
> 
> In post/memory.c file
...
> 	unsigned long memsize = (bd->bi_memsize >= 256 << 20 ?
> 				 256 << 20 : bd->bi_memsize) - (1 << 20);
...
> I think the line should be:
> 
> 	unsigned long memsize = (bd->bi_memsize >= 257 << 20 ?
> 				 256 << 20 : bd->bi_memsize) - (1 << 20);

Will you please explain what makes you think so?


Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd@denx.de
...though his invention worked superbly -- his theory was a crock  of
sewage from beginning to end.         - Vernor Vinge, "The Peace War"

[U-Boot-Users] Re: a small problem

[Yabo Ding]

Sorry, I have a mistake about the parenthesis' position.


	Yabo Ding



Wolfgang Denk wrote:
> In message <3EFBBE6A.6000200@bonbon.net> you wrote:
> 
>>In post/memory.c file
> 
> ...
> 
>>	unsigned long memsize = (bd->bi_memsize >= 256 << 20 ?
>>				 256 << 20 : bd->bi_memsize) - (1 << 20);
> 
> ...
> 
>>I think the line should be:
>>
>>	unsigned long memsize = (bd->bi_memsize >= 257 << 20 ?
>>				 256 << 20 : bd->bi_memsize) - (1 << 20);
> 
> 
> Will you please explain what makes you think so?
> 
> 
> Best regards,
> 
> Wolfgang Denk
> 

[U-Boot-Users] FADS and ICU862 Environment problems

[Nye Liu]

On Fri, Jun 27, 2003 at 01:10:57AM +0200, Wolfgang Denk wrote:
> In message <20030626224147.GA28873@zumanetworks.com> you wrote:
> >
> > > I think this is correct. What am I missing?
> > 
> > but that is the default environment that is copied, not the saved
> > copy stored via saveenv? When is the saved environment restored?
> > relocate_code() never copies it since it is AFTER __init_end due
> > to the way u-boot.lds locates environment.o after .bss
> 
> Argh... now I see. Well, I think you are right.

let me know when you come up with a recommended solution; for now I
have removed those ifdef's in start.S and reduced MONITOR_LEN such that
ENV_IS_EMBEDDED is not set.

Suggestions for the ENV_IS_EMBEDDED case?

-- 
Nye Liu
nliu at mrv.com
(818) 772-6235x248

"Who would be stupid enough to quote a fictitious character?"
	-- Don Quixote