RE: Error building glibc-2.3.2 for my arm-9tdmi-linux cross compiler on cygwin

["Dave Korn" <dave dot korn at artimi dot com>]

On 18 January 2008 12:26, J Johnson wrote:

  A minor confusion appears to have developed in this thread:

>> J Johnson wrote:
>>> The problem seems to be in the generated
>>> build-glibc-headers/sysd-syscalls. 
>> The line causing the problem seems to be related to the definition of "IFS"
>> which spans two lines the last of which contains just a sinlge quote. See a
>> snip from the file below:
>>> 
>>> ------------- BEGIN sysd-syscalls -----------------
>>> 
>>> HOSTTYPE=i686
>>> IFS='
>>> '
>>> JAVA_HOME='"C:\Program Files\Java\jdk1.5.0_13"'
>>> 
>>> ---------------------- END --------------------------------

>> I have *NOT* looked directly at that shipppit of script code in the
>> original context.
>> While strange, the above is not uncommon. I believe, in effect, the IFS
>> (input field separator) is a newline.
>> Having done *LOTS* of work with cygwin, a *VERY* common thing is the
>> default End of Line.
>> When you installed cygwin, it asks... what to use. All scripts - like
>> this *require* unix EOL.
>> 
>> Cygwin must know how to handle "\n" in IO operations.
>> 
>> Welcome to Unix Emulation, Enjoy the show :-)

  This is back-to-front!  Cygwin knows exactly how to deal with Unix EOL
("\n"); in fact, it's none to happy with anything else.  For the above script
to make any sense, the character after IFS=' would have to be a CR, not a
newline; the idea is for bash to treat the CR as a separator, so that it
interprest the CR-LF dos-style line end as a blank followed by a newline.

>> at the cygwin bash$ prompt, type "mount" and see what it says:
>> 
>> In my case, it looks like this:
>> 
>> duane@desk ~
>> $ mount
>> C:\cygwin\bin on /usr/bin type system (binmode)
>> C:\cygwin\lib on /usr/lib type system (binmode)
>> C:\cygwin on / type system (binmode)
>> c: on /cygdrive/c type system (binmode,noumount)
>> e: on /cygdrive/e type system (binmode,noumount)
>> 
>> In my case, "binmode" is correct.
>> 
>> It is OK to use non-bin mode, if you are careful (ie: your own stuff)
>> But... it is *NOT* ok for things like 'autoconfig' scripts
>> (configuration scripts and the ilk)
>> 
>> I personally gave up on CR/LF.

  I'd also recommend doing so.  It's nothing but a pain; if you're being
unixy, be proper unixy and use the unixy line-ending!

>> Also - you sometimes have problems with EDITORS.... converting, or
>> inserting crud.

> Has anyone performed a recent build on cygwin for arm-9tdmi-linux  or
> arm-xscale-linux?  I'd like to confirm that it is in fact something wrong
> with my config and not the toolchain.  

  No, I haven't, but I can say with a high degree of certainty that your
original problem:


-----------------------------------quote-----------------------------------
sdeps/ieee754/flt-32 sysdeps/ieee754/dbl-64 sysdeps/ieee754
sysdeps/generic/elf sysdeps/generic; do \
          test -f $dir/syscalls.list && \
          { /bin/sh sysdeps/unix/make-syscalls.sh $dir || exit 1; }; \
          test $dir = sysdeps/unix && break; \
        done >
/ecos-c/CNAV/tmp2/crosstool-0.43/build/arm-9tdmi-linux-gnu/gcc-4.1.0-glibc-2.3
.2/build-glibc-headers/sysd
-syscallsT
-rThe system cannot find the file specified.
-rThe system cannot find the file specified.
-rThe system cannot find the file specified.
-rThe system cannot find the file specified.
-----------------------------------quote-----------------------------------

is caused by the wrong kind of line-endings.

  Did you use a win32 GUI program to unpack the tarballs?  That's one possible
explanation; another would be if you were editing any of the contained files
using notepad or wordpad.



    cheers,
      DaveK
-- 
Can't think of a witty .sigline today....


--
For unsubscribe information see http://sourceware.org/lists.html#faq