[ECOS] Conflicts in building ecos for Atmel evaluation board (EB40A)

[ECOS] Conflicts in building ecos for Atmel evaluation board (EB40A)

[sandeep]

Hi,

This status is wrt to ecos cvs updated an hour back.
I selected "Atmel evaluation board (EB40A)" as Hardware and "default" as 
Packages via Build->Templates.

This gave me following messages :

Hardware selection: warning
     The target specification lists a package `CYGPKG_HAL_ARM_AT91' which is not 
present in the component repository.
Hardware selection: warning
     The target specification lists a package `CYGPKG_HAL_ARM_AT91_EB40A' which 
is not present in the component repository.
Hardware selection: warning
     The target specification lists a package `CYGPKG_DEVS_FLASH_EB40A' which is 
not present in the component repository.
Hardware selection: warning
     The target specification lists a package `CYGPKG_DEVS_FLASH_ATMEL_AT49XXXX' 
which is not present in the component repository.

and following conflicts :
------------------------------------------------------------------------------
* Item            * Conflict    * Property
------------------------------------------------------------------------------
CYGPKG_LIBC_STDIO - Unsatisfied - Requires CYGBLD_ISO_STDIO_FILEPOS_HEADER == 
"<cyg/libc/stdio/stdio.h>"
CYGPKG_LIBC_STDIO - Unsatisfied - Requires CYGBLD_ISO_STDIO_FILEOPS_HEADER == 
"<cyg/libc/stdio/stdio.h>"
CYGHWR_HAL_ARM_CPU_FAMILY - Illegal - Legal Values (CYGINT_HAL_ARM_ARCH_ARM7 != 
0) ? "ARM7" : "" ... =snipped=

Out of these three, the two unsatisfied conflicts come, even when configtool is 
just fired.

Looks like something has broken cvs, in past couple of days commits??

-- 
regards
sandeep
--------------------------------------------------------------------------
I am so optimistic about beef prices that I've just leased a pot roast
with an option to buy.
--------------------------------------------------------------------------


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] Unsatisfied conflicts related to CYGPKG_LIBC_STDIO in configuring eCos.

[sandeep]

Hi,
I made a mistake in updating arm related hal directories. and missed out "-d" 
with update command to cvs. sorry about the false alarm on arm. :(
To make sure that I didn't make slip in updates to other sections of ecos source 
tree, I did fresh updation 'cvs update -d' in ecos source root.

All the warnings I mentioned in my earlier mail "Subject:Conflicts in building 
ecos for Atmel evaluation board (EB40A)" and the Illegal conflict related to ARM 
are gone now.
However rest of the status still remains same.

> ------------------------------------------------------------------------------ 
> * Item            * Conflict    * Property
> ------------------------------------------------------------------------------ 
> CYGPKG_LIBC_STDIO - Unsatisfied - Requires CYGBLD_ISO_STDIO_FILEPOS_HEADER == "<cyg/libc/stdio/stdio.h>"
> CYGPKG_LIBC_STDIO - Unsatisfied - Requires CYGBLD_ISO_STDIO_FILEOPS_HEADER == "<cyg/libc/stdio/stdio.h>"
> 
> the two unsatisfied conflicts come, even when configtool is just fired.
> 
> Looks like something has broken cvs, in past couple of days commits??


I think, I should be able to post a patch to this problem, if it is confirmed by 
anyone else on the list.

-- 
regards
sandeep
--------------------------------------------------------------------------
"Life, loathe it or ignore it, you can't like it."
		-- Marvin, "Hitchhiker's Guide to the Galaxy"
--------------------------------------------------------------------------


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] Conflicts in building ecos for Atmel evaluation board (EB40A)

[Nick Garnett]

sandeep <sandeep@codito.com> writes:

> Hi,
> 
> This status is wrt to ecos cvs updated an hour back.
> I selected "Atmel evaluation board (EB40A)" as Hardware and "default"
> as Packages via Build->Templates.
> 
> This gave me following messages :
> 
> Hardware selection: warning
>      The target specification lists a package `CYGPKG_HAL_ARM_AT91'
> which is not present in the component repository.
> Hardware selection: warning
>      The target specification lists a package
> `CYGPKG_HAL_ARM_AT91_EB40A' which is not present in the component
> repository.
> Hardware selection: warning
>      The target specification lists a package
> `CYGPKG_DEVS_FLASH_EB40A' which is not present in the component
> repository.
> Hardware selection: warning
>      The target specification lists a package
> `CYGPKG_DEVS_FLASH_ATMEL_AT49XXXX' which is not present in the
> component repository.
> 
> and following conflicts :
> ------------------------------------------------------------------------------
> * Item            * Conflict    * Property
> ------------------------------------------------------------------------------
> CYGPKG_LIBC_STDIO - Unsatisfied - Requires
> CYGBLD_ISO_STDIO_FILEPOS_HEADER == "<cyg/libc/stdio/stdio.h>"
> CYGPKG_LIBC_STDIO - Unsatisfied - Requires
> CYGBLD_ISO_STDIO_FILEOPS_HEADER == "<cyg/libc/stdio/stdio.h>"
> CYGHWR_HAL_ARM_CPU_FAMILY - Illegal - Legal Values
> (CYGINT_HAL_ARM_ARCH_ARM7 != 0) ? "ARM7" : "" ... =snipped=
> 
> Out of these three, the two unsatisfied conflicts come, even when
> configtool is just fired.
> 
> Looks like something has broken cvs, in past couple of days commits??

I've just tried configuring for the EB40A with a CVS repository I
updated at probably around the same time as you. I see no problems of
the sort you are seeing. Try updating with the -d and -P
options. Otherwise it looks like your copy of the repository is
somehow corrupt -- try doing a fresh checkout.

-- 
Nick Garnett                    eCos Kernel Architect
http://www.ecoscentric.com      The eCos and RedBoot experts


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] Unsatisfied conflicts related to CYGPKG_LIBC_STDIO in configuring eCos.

[Nick Garnett]

sandeep <sandeep@codito.com> writes:

> Hi,
> I made a mistake in updating arm related hal directories. and missed
> out "-d" with update command to cvs. sorry about the false alarm on
> arm. :(
> To make sure that I didn't make slip in updates to other sections of
> ecos source tree, I did fresh updation 'cvs update -d' in ecos source
> root.
> 
> All the warnings I mentioned in my earlier mail "Subject:Conflicts in
> building ecos for Atmel evaluation board (EB40A)" and the Illegal
> conflict related to ARM are gone now.
> However rest of the status still remains same.
> 
> > ------------------------------------------------------------------------------ 
> > * Item            * Conflict    * Property
> > ------------------------------------------------------------------------------ 
> > CYGPKG_LIBC_STDIO - Unsatisfied - Requires CYGBLD_ISO_STDIO_FILEPOS_HEADER == "<cyg/libc/stdio/stdio.h>"
> > CYGPKG_LIBC_STDIO - Unsatisfied - Requires CYGBLD_ISO_STDIO_FILEOPS_HEADER == "<cyg/libc/stdio/stdio.h>"
> > the two unsatisfied conflicts come, even when configtool is just
> > fired.
> > Looks like something has broken cvs, in past couple of days commits??
> 
> 
> I think, I should be able to post a patch to this problem, if it is
> confirmed by anyone else on the list.

Hmmm. I'm not sure what is going on here. I used the commandline
tool, and there these requirements get resolved automatically. I'm not
sure why Configtool doesn't do this too. There are many other config
options whose values are set by the same mechanism, but which don't
throw up conflicts here. Very strange.

This may have something to do with the checkin that Jifl made a month
ago, but I'll leave him to decide this.

-- 
Nick Garnett                    eCos Kernel Architect
http://www.ecoscentric.com      The eCos and RedBoot experts


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] Unsatisfied Conflicts . - kind of solved but some doubts

[sandeep]

Hi Nick,

thanks for your efforts in testing it out and replying. I replied in the middle 
of my further diagnosis of the problem to save concerned people from wasting 
their time, but i was bit late.

my further investigations seem to show some behaviour that i don't understand.

I am using 'eCos Configuration Tool 2.0(May 9 2003 09:51:10), Copyright (c) Red 
Hat, Inc. 1998-2002, Copyright (c) John Dallaway 2003' on linux,

what i find - if i just invoke the configtool, it shows the below mentioned 
conflicts (with default hardware it chooses to work with).now --
* i go ahead with build->template process for selecting EB40A hardware, it still 
shows that. but i choose to continue with default solution suggested by tool and 
  things proceed. conflicts are resolved.

after this i select another different template (to test), this time no conflicts 
  are shown. but if i didn't go through the saving (with suggested default 
conflict-resolutions) in previous step, the same conflict messages are shown for 
this template as well.

this is where i don't quite get what's going on.
moreover this sort of problem didn't come around 3 weeks back, when we similarly 
started with a fresh template.


>> ------------------------------------------------------------------------------ 
>> * Item            * Conflict    * Property
>> ------------------------------------------------------------------------------ 
>> CYGPKG_LIBC_STDIO - Unsatisfied - Requires 
>> CYGBLD_ISO_STDIO_FILEPOS_HEADER == "<cyg/libc/stdio/stdio.h>"
>> CYGPKG_LIBC_STDIO - Unsatisfied - Requires 
>> CYGBLD_ISO_STDIO_FILEOPS_HEADER == "<cyg/libc/stdio/stdio.h>"
>>
>> the two unsatisfied conflicts come, even when configtool is just fired.

-- 
regards
sandeep
--------------------------------------------------------------------------
"Life, loathe it or ignore it, you can't like it."
		-- Marvin, "Hitchhiker's Guide to the Galaxy"
--------------------------------------------------------------------------


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] Unsatisfied Conflicts . - kind of solved but some doubts

[Nick Garnett]

sandeep <sandeep@codito.com> writes:

> Hi Nick,
> 
> thanks for your efforts in testing it out and replying. I replied in
> the middle of my further diagnosis of the problem to save concerned
> people from wasting their time, but i was bit late.
> 
> my further investigations seem to show some behaviour that i don't understand.
> 
> I am using 'eCos Configuration Tool 2.0(May 9 2003 09:51:10),
> Copyright (c) Red Hat, Inc. 1998-2002, Copyright (c) John Dallaway
> 2003' on linux,
> 
> what i find - if i just invoke the configtool, it shows the below
> mentioned conflicts (with default hardware it chooses to work
> with).now --
> * i go ahead with build->template process for selecting EB40A
> hardware, it still shows that. but i choose to continue with default
> solution suggested by tool and things proceed. conflicts are resolved.
> 
> after this i select another different template (to test), this time no
> conflicts are shown. but if i didn't go through the saving (with
> suggested default conflict-resolutions) in previous step, the same
> conflict messages are shown for this template as well.
> 
> this is where i don't quite get what's going on.
> moreover this sort of problem didn't come around 3 weeks back, when we
> similarly started with a fresh template.

There is definitely something weird going on. I'm not enough of an
expert in the inner workings of libcdl to be able to analyze what the
exact problem is. Hopefully, Bart will be along soon to put us right.

-- 
Nick Garnett                    eCos Kernel Architect
http://www.ecoscentric.com      The eCos and RedBoot experts


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] Unsatisfied Conflicts . - kind of solved but some doubts

[Bart Veer]

>>>>> "Nick" == Nick Garnett <nickg@ecoscentric.com> writes:

    Nick> sandeep <sandeep@codito.com> writes:

    <snip>

    >> this is where i don't quite get what's going on. moreover this
    >> sort of problem didn't come around 3 weeks back, when we
    >> similarly started with a fresh template.

    Nick> There is definitely something weird going on. I'm not enough
    Nick> of an expert in the inner workings of libcdl to be able to
    Nick> analyze what the exact problem is. Hopefully, Bart will be
    Nick> along soon to put us right.

The problem is caused by jifl's check-in of 2004-03-12. He had added
some new requires to CYGPKG_LIBC_STDIO but had failed to update the
templates. This is no problem for ecosconfig which will resolve such
things automatically, but the configtool handles inference somewhat
differently. I believe the templates will be fixed shortly.

Bart

-- 
Bart Veer                       eCos Configuration Architect
http://www.ecoscentric.com/     The eCos and RedBoot experts

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

[ECOS] Which gcc should we use for the latest CVS ECOS?

[Linux HA]

Oops, sorry Bart I didn't mean to send it directly to you.

I'm putting together a project that uses the ARM (EB40
like board). I'm using the latest full ecos from the
CVS. What is the 'best' stable release of the gnu tools
I should use?

-- 
Linux Home Automation                 Neil Cherry
http://home.comcast.net/~ncherry/     (Text only)
http://hcs.sourceforge.net/           (HCS II & HCS_C)


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] Which gcc should we use for the latest CVS ECOS?

[Andrew Lunn]

On Wed, Apr 14, 2004 at 11:29:34PM -0400, Linux HA wrote:
> Oops, sorry Bart I didn't mean to send it directly to you.
> 
> I'm putting together a project that uses the ARM (EB40
> like board). I'm using the latest full ecos from the
> CVS. What is the 'best' stable release of the gnu tools
> I should use?

gcc 3.2.x is known to be broken on ARM. It produces bad code for
jffs2, so avoid that version. I think gcc 3.3 fixes that problem, any
some people have reported success.

Having said all that i would recommend the eCosCentric binaries. They
have been well tested with eCos. So these are probably you best bet.

     Andrew

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] new warning message related to ecos.db with today's updates

[sandeep]

Thanks eCos maintainers,
After updating local cvs to get recent commits by jifl
those problems are gone. however a new warning has cropped up (that stays 
irrespective of - whether configtool is up with it's default template or
I use my customised target .ecc). I carefully went over my cvs update process 
and verified that I didn't miss anything.

ecos.db, package CYGPKG_HAL_ARM_PID: warning
Version subdirectory `tests' does not have a CDL script `hal_arm_pid.cdl'.

I guess this warning is harmless enough that it can be ignored ????


This warning comes even with ecosconfig..

[blackmagic:/compiling/ecos/builds]ecosconfig --config=eb40atry.ecc  check
ecos.db, package CYGPKG_HAL_ARM_PID: warning
     Version subdirectory `tests' does not have a CDL script `hal_arm_pid.cdl'.
Target: eb40a
Template: default
Added:
  CYGPKG_FS_JFFS2
  CYGPKG_IO_FILEIO
  CYGPKG_LINUX_COMPAT
  CYGPKG_CRC
  CYGPKG_IO_FLASH
No conflicts
-- 
regards
sandeep
--------------------------------------------------------------------------
If God didn't mean for us to juggle, tennis balls wouldn't come three
to a can.
--------------------------------------------------------------------------


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] new warning message related to ecos.db with today's updates

[Nick Garnett]

sandeep <sandeep@codito.com> writes:

> Thanks eCos maintainers,
> After updating local cvs to get recent commits by jifl
> those problems are gone. however a new warning has cropped up (that
> stays irrespective of - whether configtool is up with it's default
> template or
> I use my customised target .ecc). I carefully went over my cvs update
> process and verified that I didn't miss anything.
> 
> ecos.db, package CYGPKG_HAL_ARM_PID: warning
> Version subdirectory `tests' does not have a CDL script `hal_arm_pid.cdl'.
> 
> I guess this warning is harmless enough that it can be ignored ????
> 
> 
> This warning comes even with ecosconfig..
> 
> [blackmagic:/compiling/ecos/builds]ecosconfig --config=eb40atry.ecc  check
> ecos.db, package CYGPKG_HAL_ARM_PID: warning
>      Version subdirectory `tests' does not have a CDL script `hal_arm_pid.cdl'.

This looks like there is a "tests" directory at the same level as the
"current" directory. I'm not sure where that came from. Try doing a
"cvs update -P" in the hal/arm/pid directory to see if CVS will prune
it out. Alternatively, just remove it.


-- 
Nick Garnett                    eCos Kernel Architect
http://www.ecoscentric.com      The eCos and RedBoot experts


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] new warning message related to ecos.db with today's updates - fixed

[sandeep]

> This looks like there is a "tests" directory at the same level as the
> "current" directory. I'm not sure where that came from. Try doing a
> "cvs update -P" in the hal/arm/pid directory to see if CVS will prune
> it out. Alternatively, just remove it.
Thanks Nick. it was indeed the problem. don't remember how it came there.

something might have got messed up last night.


-- 
regards
sandeep
--------------------------------------------------------------------------
Graduate life: It's not just a job.  It's an indenture.
--------------------------------------------------------------------------


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] Which gcc should we use for the latest CVS ECOS?

[Linux HA]

Andrew Lunn wrote:
> On Wed, Apr 14, 2004 at 11:29:34PM -0400, Linux HA wrote:

>>I'm putting together a project that uses the ARM (EB40
>>like board). I'm using the latest full ecos from the
>>CVS. What is the 'best' stable release of the gnu tools
>>I should use?

> Having said all that i would recommend the eCosCentric binaries. They
> have been well tested with eCos. So these are probably you best bet.

Thanks, the eCosCentric binaries it is then.

-- 
Linux Home Automation                 Neil Cherry
http://home.comcast.net/~ncherry/     (Text only)
http://hcs.sourceforge.net/           (HCS II & HCS_C)

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss