[ECOS] powerpc-eabi on cygwin

[ECOS] powerpc-eabi on cygwin

[chandrag]

Hi,

  I am trying to compile ecos for a powerpc target in cygwin environment.
Make exits by printing the following errors.  I downloaded the gnu tools
for cygwin from ecoscentric.  There is no problem in building the same
ecos sources under linux.

  What could be the problem?

Thanks in advance,
chandra
----------------------------------------------------------------------------
----
powerpc-eabi-gcc -c  -I/home/chandra/ecos-app/app_install/include
-I/ecos_5th_Fe
b2003/ecos/packages/hal/common/current
-I/ecos_5th_Feb2003/ecos/packages/hal/com
mon/current/src -I/ecos_5th_Feb2003/ecos/packages/hal/common/current/tests
-I. -
I/ecos_5th_Feb2003/ecos/packages/hal/common/current/src/ -msoft-float
-mcpu=860
-Wall -Wpointer-arith -Wstrict-prototypes -Winline -Wundef
-Woverloaded-virtual
-g -O2 -ffunction-sections -fdata-sections -fno-rtti -fno-exceptions
-fvtable-gc
 -finit-priority -Wp,-MD,src/drv_api.tmp -o src/hal_common_drv_api.o
/ecos_5th_F
eb2003/ecos/packages/hal/common/current/src/drv_api.c
cc1: format string has invalid operand number
cc1: format string has invalid operand number
cc1: format string has invalid operand number
cc1: format string has invalid operand number
cc1: format string has invalid operand number
cc1: format string has invalid operand number
cc1: variable-size type declared outside of any function
/ecos_5th_Feb2003/ecos/packages/hal/common/current/src/drv_api.c:1: confused
by
earlier errors, bailing out
make[1]: *** [src/drv_api.o.d] Error 1
make[1]: Leaving directory
`/home/chandra/ecos-app/app_build/hal/common/current'

make: *** [build] Error 2

----------------------------------------------------------------------------
----

The following command at bash$ prompt also fails with error messages.

----------------------------------------------------------------------------
----
$ powerpc-eabi-gcc -c test.c
cc1: Internal compiler error in size_binop, at fold-const.c:2014
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions.
----------------------------------------------------------------------------
-----

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

Re: [ECOS] powerpc-eabi on cygwin

[Jonathan Larmour]

chandrag@danlawinc.com wrote:
> 
> The following command at bash$ prompt also fails with error messages.
> 
> ----------------------------------------------------------------------------
> ----
> $ powerpc-eabi-gcc -c test.c
> cc1: Internal compiler error in size_binop, at fold-const.c:2014
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions.
> ----------------------------------------------------------------------------

I've built and tested all of eCos successfully with powerpc-eabi tools in 
cygwin without these problems, although I don't know what test.c is.

Do you have something like GCC_EXEC_PREFIX set in your environment?

Jifl
-- 
eCosCentric    http://www.eCosCentric.com/    The eCos and RedBoot experts
--[ "You can complain because roses have thorns, or you ]--
--[  can rejoice because thorns have roses." -Lincoln   ]-- Opinions==mine


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

Re: [ECOS] powerpc-eabi on cygwin

[Paul Randall]

Hi,

I am having a similar problem when making the eCos library using the current CVS
sources.  I am also using cygwin and the prebuilt powerpc-eabi toolchain.  The
error below occurs when compiling hal_mk_defs.c.  I don't get the error when
using a powerpc-eabi toolchain I built about a year ago.

Paul Randall
Delta Information Systems

...
make -r -C io/wallclock/current headers
make[1]: Entering directory `/opt/testlib/io/wallclock/current'
make[1]: Leaving directory `/opt/testlib/io/wallclock/current'
make -r -C error/current headers
make[1]: Entering directory `/opt/testlib/error/current'
make[1]: Leaving directory `/opt/testlib/error/current'
headers finished
make -r -C hal/powerpc/arch/current
/ecos-c/cygwin/opt/testlib/install/include/cyg/hal/ppc_offsets.inc
make[1]: Entering directory `/opt/testlib/hal/powerpc/arch/current'
powerpc-eabi-gcc -msoft-float -mcpu=860 -Wall -Wpointer-arith
-Wstrict-prototypes -Winline -Wundef -Woverloaded-virtual -g -O2
-ffunction-sections -fdata-sections -fno-rtti -fno-exceptions
-fvtable-gc -finit-priority
-I/ecos-c/cygwin/opt/testlib/install/include
-I/ecos-c/cygwin/opt/ecoscvs/ecos/packages/hal/powerpc/arch/current
-I/ecos-c/cygwin/opt/ecoscvs/ecos/packages/hal/powerpc/arch/current/src
-I/ecos-c/cygwin/opt/ecoscvs/ecos/packages/hal/powerpc/arch/current/tests
-I.
 -Wp,-MD,ppc_offsets.tmp -o hal_mk_defs.tmp -S
/ecos-c/cygwin/opt/ecoscvs/ecos/packages/hal/powerpc/arch/current/src/hal_mk_defs.c

cc1: format string has invalid operand number
cc1: format string has invalid operand number
cc1: format string has invalid operand number
cc1: format string has invalid operand number
cc1: format string has invalid operand number
cc1: format string has invalid operand number
cc1: variable-size type declared outside of any function

/ecos-c/cygwin/opt/ecoscvs/ecos/packages/hal/powerpc/arch/current/src/hal_mk_defs.c:1:

confused by earlier errors, bailing out
make[1]: ***
[/ecos-c/cygwin/opt/testlib/install/include/cyg/hal/ppc_offsets.inc]
Error 1
make[1]: Leaving directory `/opt/testlib/hal/powerpc/arch/current'
make: *** [build] Error 2

chandrag@danlawinc.com wrote:

> Hi,
>
>   I am trying to compile ecos for a powerpc target in cygwin environment.
> Make exits by printing the following errors.  I downloaded the gnu tools
> for cygwin from ecoscentric.  There is no problem in building the same
> ecos sources under linux.
>
>   What could be the problem?
>
> Thanks in advance,
> chandra
> ----------------------------------------------------------------------------
> ----
> powerpc-eabi-gcc -c  -I/home/chandra/ecos-app/app_install/include
> -I/ecos_5th_Fe
> b2003/ecos/packages/hal/common/current
> -I/ecos_5th_Feb2003/ecos/packages/hal/com
> mon/current/src -I/ecos_5th_Feb2003/ecos/packages/hal/common/current/tests
> -I. -
> I/ecos_5th_Feb2003/ecos/packages/hal/common/current/src/ -msoft-float
> -mcpu=860
> -Wall -Wpointer-arith -Wstrict-prototypes -Winline -Wundef
> -Woverloaded-virtual
> -g -O2 -ffunction-sections -fdata-sections -fno-rtti -fno-exceptions
> -fvtable-gc
>  -finit-priority -Wp,-MD,src/drv_api.tmp -o src/hal_common_drv_api.o
> /ecos_5th_F
> eb2003/ecos/packages/hal/common/current/src/drv_api.c
> cc1: format string has invalid operand number
> cc1: format string has invalid operand number
> cc1: format string has invalid operand number
> cc1: format string has invalid operand number
> cc1: format string has invalid operand number
> cc1: format string has invalid operand number
> cc1: variable-size type declared outside of any function
> /ecos_5th_Feb2003/ecos/packages/hal/common/current/src/drv_api.c:1: confused
> by
> earlier errors, bailing out
> make[1]: *** [src/drv_api.o.d] Error 1
> make[1]: Leaving directory
> `/home/chandra/ecos-app/app_build/hal/common/current'
>
> make: *** [build] Error 2
>
> ----------------------------------------------------------------------------
> ----
>
> The following command at bash$ prompt also fails with error messages.
>
> ----------------------------------------------------------------------------
> ----
> $ powerpc-eabi-gcc -c test.c
> cc1: Internal compiler error in size_binop, at fold-const.c:2014
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions.
> ----------------------------------------------------------------------------
> -----
>
> --
> Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
> and search the list archive: http://sources.redhat.com/ml/ecos-discuss


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

Re: [ECOS] powerpc-eabi on cygwin

[Jonathan Larmour]

Paul Randall wrote:
> Hi,
> 
> I am having a similar problem when making the eCos library using the current CVS
> sources.  I am also using cygwin and the prebuilt powerpc-eabi toolchain.  The
> error below occurs when compiling hal_mk_defs.c.  I don't get the error when
> using a powerpc-eabi toolchain I built about a year ago.

This is Win9x? Try adding --with-gc=simple to the GCC flags and tell us 
what happens.

Jifl
-- 
eCosCentric    http://www.eCosCentric.com/    The eCos and RedBoot experts
--[ "You can complain because roses have thorns, or you ]--
--[  can rejoice because thorns have roses." -Lincoln   ]-- Opinions==mine


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

Re: [ECOS] powerpc-eabi on cygwin

[Jonathan Larmour]

Jonathan Larmour wrote:
> This is Win9x? Try adding [snip]

Sorry, for clarity...  rebuild GCC according to 
http://sources.redhat.com/ecos/build-toolchain.html but add a configure 
option of --with-gc=simple

If at all possible, build it on NT/W2K. Apologies if it doesn't work on 
W98, but given how many people it potentially affects, it's likely we'll 
have to update the prebuilt tools.

Jifl
-- 
eCosCentric    http://www.eCosCentric.com/    The eCos and RedBoot experts
--[ "You can complain because roses have thorns, or you ]--
--[  can rejoice because thorns have roses." -Lincoln   ]-- Opinions==mine


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

Re: [ECOS] powerpc-eabi on cygwin

[Christopher Faylor]

On Thu, Apr 10, 2003 at 08:35:52PM +0100, Bart Veer wrote:
>>>>>> "Cgf" == Christopher Faylor <cgf@redhat.com> writes:
>
>    >>>> And now we've also determined it appears to be Win9x specific.
>    >>> 
>    >>> mmap problem maybe? Does adding "--with-gc=simple" to the
>    >>> command line make things work "better"? That should be the
>    >>> defaulton Windows 9x but if something isn't causing this to be
>    >>> considered the default, all sorts of breakage can happen.
>    >> 
>    >> Ah, is this determined by autoconf at toolchain build time? The
>    >> tools are built on NT/XP (since that should be more reliable
>    >> than Win9x in general). If so, that would certainly explain it
>    >> then.
>
>    cgf> No, it's supposed to be a runtime thing so that a gcc built
>    cgf> on NT will still work on 98.
>
>    cgf> I'm grasping at straws, really, but usually when there are
>    cgf> problems with gcc on Windows 98, it's been due to mmap
>    cgf> issues. At least that has been the case recently.
>
>    cgf> Has anyone tried this --with-gc=simple on 98 just to see if
>    cgf> this all a mad delusion on my part?
>
>--with-gc seems to be a configure-time option only, not a run-time
>option. gcc's Makefile.in includes just a single object file to
>implement garbage collection, e.g. ggc-simple.c or ggc-page.c, decided
>by the configure script.

Oh.  Sorry.  I just did a grep on my email logs to see what we'd been
recommending.  I must have misinterpreted.  Sorry for the confusion.

cgf

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

Re: [ECOS] powerpc-eabi on cygwin

[Bart Veer]

>>>>> "Cgf" == Christopher Faylor <cgf@redhat.com> writes:

    >>>> And now we've also determined it appears to be Win9x specific.
    >>> 
    >>> mmap problem maybe? Does adding "--with-gc=simple" to the
    >>> command line make things work "better"? That should be the
    >>> defaulton Windows 9x but if something isn't causing this to be
    >>> considered the default, all sorts of breakage can happen.
    >> 
    >> Ah, is this determined by autoconf at toolchain build time? The
    >> tools are built on NT/XP (since that should be more reliable
    >> than Win9x in general). If so, that would certainly explain it
    >> then.

    cgf> No, it's supposed to be a runtime thing so that a gcc built
    cgf> on NT will still work on 98.

    cgf> I'm grasping at straws, really, but usually when there are
    cgf> problems with gcc on Windows 98, it's been due to mmap
    cgf> issues. At least that has been the case recently.

    cgf> Has anyone tried this --with-gc=simple on 98 just to see if
    cgf> this all a mad delusion on my part?

--with-gc seems to be a configure-time option only, not a run-time
option. gcc's Makefile.in includes just a single object file to
implement garbage collection, e.g. ggc-simple.c or ggc-page.c, decided
by the configure script.

Unfortunately with current cygwin I am unable to rebuild a toolchain
under W98 - the configure hangs during a sed invocation. Instead I'll
have to do the rebuild on another machine which can be booted into XP.

Bart

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

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

Re: [ECOS] powerpc-eabi on cygwin

[Christopher Faylor]

On Wed, Apr 09, 2003 at 12:46:11PM +0100, Jonathan Larmour wrote:
>Christopher Faylor wrote:
>>On Tue, Apr 08, 2003 at 05:20:03PM +0100, Jonathan Larmour wrote:
>>>
>>>And now we've also determined it appears to be Win9x specific.
>>
>>
>>mmap problem maybe?  Does adding "--with-gc=simple" to the command line
>>make things work "better"?  That should be the defaulton Windows 9x but
>>if something isn't causing this to be considered the default, all sorts
>>of breakage can happen.
>
>Ah, is this determined by autoconf at toolchain build time? The tools are 
>built on NT/XP (since that should be more reliable than Win9x in general). 
>If so, that would certainly explain it then.

No, it's supposed to be a runtime thing so that a gcc built on NT will still
work on 98.

I'm grasping at straws, really, but usually when there are problems with gcc on
Windows 98, it's been due to mmap issues.  At least that has been the case
recently.

Has anyone tried this --with-gc=simple on 98 just to see if this all a
mad delusion on my part?

cgf

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

Re: [ECOS] powerpc-eabi on cygwin

[Jonathan Larmour]

Christopher Faylor wrote:
> On Tue, Apr 08, 2003 at 05:20:03PM +0100, Jonathan Larmour wrote:
>>
>>And now we've also determined it appears to be Win9x specific.
> 
> 
> mmap problem maybe?  Does adding "--with-gc=simple" to the command line
> make things work "better"?  That should be the defaulton Windows 9x but
> if something isn't causing this to be considered the default, all sorts
> of breakage can happen.

Ah, is this determined by autoconf at toolchain build time? The tools are 
built on NT/XP (since that should be more reliable than Win9x in general). 
If so, that would certainly explain it then.

Thanks!

Jifl
-- 
eCosCentric    http://www.eCosCentric.com/    The eCos and RedBoot experts
--[ "You can complain because roses have thorns, or you ]--
--[  can rejoice because thorns have roses." -Lincoln   ]-- Opinions==mine


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

Re: [ECOS] powerpc-eabi on cygwin

[Christopher Faylor]

On Tue, Apr 08, 2003 at 05:20:03PM +0100, Jonathan Larmour wrote:
>Jonathan Larmour wrote:
>>chandrag@danlawinc.com wrote:
>>>>$ powerpc-eabi-gcc -c test.c
>>>>cc1: Internal compiler error in size_binop, at fold-const.c:2014
>>>>Please submit a full bug report,
>>>>with preprocessed source if appropriate.
>>>>See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions.
>>
>>
>>I just updated my cygwin installation, and sure enough now I see the 
>>problem. I tried downgrading the DLL version from 1.3.22 to 1.3.21, but 
>>there was no change.
>
>And now we've also determined it appears to be Win9x specific.

mmap problem maybe?  Does adding "--with-gc=simple" to the command line
make things work "better"?  That should be the defaulton Windows 9x but
if something isn't causing this to be considered the default, all sorts
of breakage can happen.

cgf

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

Re: [ECOS] powerpc-eabi on cygwin

[Jonathan Larmour]

Jonathan Larmour wrote:
> chandrag@danlawinc.com wrote:
>>> $ powerpc-eabi-gcc -c test.c
>>> cc1: Internal compiler error in size_binop, at fold-const.c:2014
>>> Please submit a full bug report,
>>> with preprocessed source if appropriate.
>>> See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions.
> 
> 
> I just updated my cygwin installation, and sure enough now I see the 
> problem. I tried downgrading the DLL version from 1.3.22 to 1.3.21, but 
> there was no change.

And now we've also determined it appears to be Win9x specific. Use NT/XP 
or W2K instead and you shouldn't have problems, although we have heard a 
report of problems on W2K.

Jifl
-- 
eCosCentric    http://www.eCosCentric.com/    The eCos and RedBoot experts
--[ "You can complain because roses have thorns, or you ]--
--[  can rejoice because thorns have roses." -Lincoln   ]-- Opinions==mine


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

Re: [ECOS] powerpc-eabi on cygwin

[Chandra Sekhar, G]

Jonathan Larmour wrote:

> chandrag@danlawinc.com wrote:
> > test.c is a simple program.  The same error occurs any for other
> > file (even with drv_api.c, when compiled single) for the command
> > 'powerpc-eabi-gcc'.
> >
> > There is no environment variable like GCC_EXEC_PREFIX.
> [snip]
> >>$ powerpc-eabi-gcc -c test.c
> >>cc1: Internal compiler error in size_binop, at fold-const.c:2014
> >>Please submit a full bug report,
> >>with preprocessed source if appropriate.
> >>See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions.
>
> I just updated my cygwin installation, and sure enough now I see the
> problem. I tried downgrading the DLL version from 1.3.22 to 1.3.21, but
> there was no change.
>
> So it looks like this is _probably_ another cygwin bug of some sort to
> track down unfortunately, although looking at where the error has
> happened, it's very unclear as to what it could be as there don't appear
> to be any DLL calls in the vicinity.
>
> This is going to have to be investigated before the 2.0 final release.
> Bart/JLD take note! I would try and submit it to bugzilla, but bugzilla
> appears to be down.
>
> Jifl
> --
> eCosCentric    http://www.eCosCentric.com/    The eCos and RedBoot experts
> --[ "You can complain because roses have thorns, or you ]--
> --[  can rejoice because thorns have roses." -Lincoln   ]-- Opinions==mine

Is there any older version of powerpc tools for cygwin which do not has this
problem?  Please give a link.

Thanks and regards,
chandra


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

Re: [ECOS] powerpc-eabi on cygwin

[Jonathan Larmour]

chandrag@danlawinc.com wrote:
> test.c is a simple program.  The same error occurs any for other
> file (even with drv_api.c, when compiled single) for the command
> 'powerpc-eabi-gcc'.
> 
> There is no environment variable like GCC_EXEC_PREFIX.
[snip]
>>$ powerpc-eabi-gcc -c test.c
>>cc1: Internal compiler error in size_binop, at fold-const.c:2014
>>Please submit a full bug report,
>>with preprocessed source if appropriate.
>>See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions.

I just updated my cygwin installation, and sure enough now I see the 
problem. I tried downgrading the DLL version from 1.3.22 to 1.3.21, but 
there was no change.

So it looks like this is _probably_ another cygwin bug of some sort to 
track down unfortunately, although looking at where the error has 
happened, it's very unclear as to what it could be as there don't appear 
to be any DLL calls in the vicinity.

This is going to have to be investigated before the 2.0 final release. 
Bart/JLD take note! I would try and submit it to bugzilla, but bugzilla 
appears to be down.

Jifl
-- 
eCosCentric    http://www.eCosCentric.com/    The eCos and RedBoot experts
--[ "You can complain because roses have thorns, or you ]--
--[  can rejoice because thorns have roses." -Lincoln   ]-- Opinions==mine


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

RE: [ECOS] powerpc-eabi on cygwin

[chandrag]

test.c is a simple program.  The same error occurs any for other
file (even with drv_api.c, when compiled single) for the command
'powerpc-eabi-gcc'.

There is no environment variable like GCC_EXEC_PREFIX.

thanks and regards,
chandra

-----Original Message-----
From: Jonathan Larmour [mailto:jifl@eCosCentric.com]
Sent: Tuesday, April 08, 2003 3:46 AM
To: chandrag@danlawinc.com
Cc: ecos-discuss@sources.redhat.com
Subject: Re: [ECOS] powerpc-eabi on cygwin


chandrag@danlawinc.com wrote:
> 
> The following command at bash$ prompt also fails with error messages.
> 
>
----------------------------------------------------------------------------
> ----
> $ powerpc-eabi-gcc -c test.c
> cc1: Internal compiler error in size_binop, at fold-const.c:2014
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions.
>
----------------------------------------------------------------------------

I've built and tested all of eCos successfully with powerpc-eabi tools in 
cygwin without these problems, although I don't know what test.c is.

Do you have something like GCC_EXEC_PREFIX set in your environment?

Jifl
-- 
eCosCentric    http://www.eCosCentric.com/    The eCos and RedBoot experts
--[ "You can complain because roses have thorns, or you ]--
--[  can rejoice because thorns have roses." -Lincoln   ]-- Opinions==mine

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