Re: configure problem - not building df - coreutils 5.2.1

[Charles Notley]

I tried hardcoding DF_PROG=df in the Makefiles, but it didn't work.  I 
also tried compiling df by itself:
address@hidden src]$ mipsel-unknown-linux-gnu-gcc -DHAVE_CONFIG_H -I. -I.. 
-I../lib 
-I/opt/crosstool/gcc-3.2.3-glibc-2.2.3/mipsel-unknown-linux-gnu/mipsel-unknown-linux-gnu/lib 
-g -O2 -MT df.o -MD -MP -MF ".deps/df.Po" -c -o df.o df.c
address@hidden src]$ mipsel-unknown-linux-gnu-gcc -g -O2 -o df df.o 
../lib/libfetish.a
df.o(.text+0x78c): In function `show_dev':
/home/chn/coreutils-5.2.1/src/df.c:298: undefined reference to 
`get_fs_usage'
collect2: ld returned 1 exit status

I don't understand what get_fs_usage is.  I'm guessing that it relates 
to the origional code you told me to look at:

# These tests are for df.
  AC_REQUIRE([gl_FSUSAGE])
  AC_REQUIRE([gl_MOUNTLIST])
  if test $gl_cv_list_mounted_fs = yes && test $gl_cv_fs_space = yes; then
    DF_PROG='df$(EXEEXT)'
  fi
  AC_REQUIRE([jm_AC_DOS])
  AC_REQUIRE([AC_FUNC_CANONICALIZE_FILE_NAME])

Any suggestions on where I should look to better understand this problem?

Paul Eggert wrote:
> Charles Notley <address@hidden> writes:
>
>
> > I set the --host argument because I'm cross-compiling.
>
>
> If you're cross-compiling, then I'm afraid we assume some expertise
> on your part so that you can debug the inevitable glitches.
> Please see m4/jm-macros.m4 to get a feel for why "configure"
> decided that df wasn't worth building.   Look for DF_PROG.