Hi, On Wed, Jul 14, 2010 at 1:58 AM, Sebastian Bauer <mail@xxxxxxxxxxxxxxxxxxx> wrote: > Trevor Scroggins <trevor.scroggins@xxxxxxxxx> schrieb: >> This is just a small patch to fix the assembler .[us]leb128 test in >> gcc/gcc/configure[.ac] for our version of Binutils. > > Could you explain the impact of this patch? I.e., what does it fix? Per configure, "Check if we have .[us]leb128, and support symbol arithmetic with it." GAS 2.11 reports that it support this, but according to the configure script, it does not. The script double-checks the GAS version after the initial test succeeds. The current test assumes the target `as --version` prints `GNU assembler [version]'. Our version of as prints `GNU assembler (GNU Binutils) 2.18', and subsequent sed and test expressions are invalid. I.e. `if test $as_major -eq 2 && test $as_minor -lt 11' returns an invalid number of arguments error. Because of the way the test is written, if test $as_major -eq 2 && test $as_minor -lt 11 then : else gcc_cv_as_leb128=yes fi a syntax error in either test statement will result in the else branch being taken, and the overall test will succeed. Luckily, this is correct for us (and many others). The mainline GCC trunk uses this, as_ver=`$gcc_cv_as --version 2>/dev/null | sed 1q` if echo "$as_ver" | grep GNU > /dev/null; then as_vers=`echo $as_ver | sed -n \ -e 's,^.*[ ]\([0-9][0-9]*\.[0-9][0-9]*.*\)$,\1,p'` as_major=`expr "$as_vers" : '\([0-9]*\)'` as_minor=`expr "$as_vers" : '[0-9]*\.\([0-9]*\)'` if test $as_major -eq 2 && test $as_minor -lt 11 then : else gcc_cv_as_leb128=yes fi fi which probably covers more cases. It was first fixed with http://gcc.gnu.org/viewcvs?view=revision&revision=127745, but it appears to have been modified again without comment. Your 4.4.3 branch already has the fix. Trev -- ______________________________________________________________________________ Amiga Development tools ML - //www.freelists.org/list/adtools Homepage...................: http://www.sourceforge.net/projects/adtools Listserver help............: mailto:adtools-request@xxxxxxxxxxxxx?Subject=HELP