The following tests (new since GCC 3.3) cause GCC to ICE for ia64-linux while processing objects of type "_Complex float" and _Complex long double". These are not regressions, the problems have apparently existed since _Complex was added. FAIL: gcc.dg/compat/scalar-by-value-3 c_compat_x_tst.o compile FAIL: gcc.dg/compat/scalar-by-value-4 c_compat_x_tst.o compile FAIL: gcc.dg/compat/scalar-return-3 c_compat_x_tst.o compile FAIL: gcc.dg/compat/struct-by-value-16 c_compat_x_tst.o compile FAIL: gcc.dg/compat/struct-by-value-16 c_compat_y_tst.o compile FAIL: gcc.dg/compat/struct-by-value-18 c_compat_x_tst.o compile FAIL: gcc.dg/compat/struct-by-value-5 c_compat_x_tst.o compile The attached minimized test cases duplicate the ICEs from the original tests. Output with gcc version 3.4 20030617 (experimental): --- bug2.c --- bug2.c: In function `testitcld': bug2.c:8: internal compiler error: in gen_movtf, at insn-emit.c:2556 Please submit a full bug report, with preprocessed source if appropriate. See <URL:http://gcc.gnu.org/bugs.html> for instructions. --- bug3a.c --- bug3a.c: In function `testScf7': bug3a.c:9: internal compiler error: in emit_move_insn, at expr.c:3262 Please submit a full bug report, with preprocessed source if appropriate. See <URL:http://gcc.gnu.org/bugs.html> for instructions. --- bug3b.c --- bug3b.c: In function `test2_cf': bug3b.c:9: internal compiler error: in emit_move_insn, at expr.c:3262 Please submit a full bug report, with preprocessed source if appropriate. See <URL:http://gcc.gnu.org/bugs.html> for instructions. --- bug4.c --- bug4.c: In function `checkScf7': bug4.c:7: internal compiler error: in convert_move, at expr.c:573 Please submit a full bug report, with preprocessed source if appropriate. See <URL:http://gcc.gnu.org/bugs.html> for instructions. Two of the ICE messages are the same, but the test cases might get there by different routes so they're both included.
Created attachment 4238 [details] test case
Created attachment 4239 [details] test case 2
Created attachment 4240 [details] test case 3
Created attachment 4241 [details] test case 4
I can confirm that testcases 2-4 ICE on ia64-hpux with gcc 3.2.3, 3.3 branch and mainline. The first testcase however does not ICE for me. I wonder why...
This is caused by the same problem that created bug 11226 and is fixed by the same patch. This patch is on the ToT and the 3.3 branch. I tested ToT and 3.3.2 with the patch from http://gcc.gnu.org/ml/gcc-patches/2003-12/msg00145.html and all the test cases worked so this defect should be closed out.
Fixed according to Steve.