This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
[committed] Fix setting of objc_libgcc_s_path
- From: Richard Sandiford <rsandifo at nildram dot co dot uk>
- To: gcc-patches at gcc dot gnu dot org
- Date: Sat, 29 Dec 2007 15:17:47 +0000
- Subject: [committed] Fix setting of objc_libgcc_s_path
Every call to objc_init does:
global objc_libgcc_s_path
...
lappend objc_libgcc_s_path [gcc-set-multilib-library-path $OBJC_UNDER_TEST]
so objc_libgcc_s_path gets longer with each new *.exp file.
With a 3-ABI mips64-linux-gnu run, LD_LIBRARY_PATH was getting
too long during the 3rd run.
Fixed by changing the lappend to a set, just as treelang.exp does.
Bootstrapped & regression-tested on x86_64-linux-gnu and
regression-tested on mips64-linux-gnu. Installed as obvious.
Richard
[ Just for the record, in case anyone's wondering:
if { $objc_initialized == 1 } { return; }
seems to be dead code, as nothing ever sets objc_initialized.
The code build by:
objc_maybe_build_wrapper "${tmpdir}/objc-testglue.o"
depends on the multilib being tested, so a simple 0/1 guard
would be (and was presumably found to be) too simplistic. ]
gcc/testsuite/
* lib/objc.exp (objc_libgcc_s_path): Set objc_libgcc_s_path
instead of appending to it.
Index: gcc/testsuite/lib/objc.exp
===================================================================
--- gcc/testsuite/lib/objc.exp 2007-12-28 15:02:12.000000000 +0000
+++ gcc/testsuite/lib/objc.exp 2007-12-28 15:02:17.000000000 +0000
@@ -115,7 +115,7 @@ proc objc_init { args } {
objc_maybe_build_wrapper "${tmpdir}/objc-testglue.o"
- append objc_libgcc_s_path [gcc-set-multilib-library-path $OBJC_UNDER_TEST]
+ set objc_libgcc_s_path [gcc-set-multilib-library-path $OBJC_UNDER_TEST]
}
proc objc_target_compile { source dest type options } {