how is PATH modified as crosstool runs?

[Robert P. J. Day]

  i want to clarify how the PATH variable should be modified as
crosstool processes each of its stages since this might let me
simplify a couple of the steps.  (or maybe not, who knows?)

  obviously, the first part of crosstool -- unpacking and patching the
sources and installing the kernel headers -- can be done with the
standard setting of the user's PATH variable, no problem there.  but
what happens with the next step?

  that next phase involves configuring, building and installing the
binutils utilities, which installs the following:

  ${PREFIX}/
    bin/
      ${TARGET}-addr2line
      ${TARGET}-ar
      ${TARGET}-as
      ... and so on ...
    ${TARGET}/
      bin/
        ar, as, ld, nm, objdump, ranlib, strip

so the executables under ${PREFIX}/bin are the TARGET-prepended
binutils to be used in future steps, while the executables under
${PREFIX}/${TARGET}/bin are exactly the same as their earlier
counterparts, they just have the regular names to be used in later
steps when the build goes looking for those names but expects to find
the cross-compile versions of those utilities.

  in any event, it seems that, once you finish the binutils step, it's
sufficient to add only that *first* directory to the search path,
right?  (you don't need the second one, at least at this point.)  so,
after the binutils step, it's sufficient to have:

  PATH=${PREFIX}/bin:${PATH}

  and, from what i've read, you can configure and install the glibc
headers in the next step with that same PATH value.

  so far, so good?  (it's the bootstrap gcc step that happens next
where things get a bit more complicated but i'll get to that later.)

rday



------
Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
Want to unsubscribe? Send a note to crossgcc-unsubscribe@sources.redhat.com