Hi,
I ran part of the testsuite on the dataflow branch for spu-elf. I
only ran compile.exp as I don't have access to hardware (at home) and I
have not figured out how to use the "free" simulator from IBM yet.
There is one regression:
gcc.c-torture/compile/930513-2.c ICEs in find_reg_note when compiled
with -O3 -fomit-frame-pointer -funroll-all-loops -finline-functions.
SPU has libcalls for double compares.
The backtrace is:
#1 0x08499f85 in find_reg_note (insn=0x0, kind=REG_LIBCALL, datum=0x0)
at ../../gcc/rtlanal.c:1593
#2 0x082fa876 in emit_copy_of_insn_after (insn=0xb7c4e4b0,
after=0xb7c4b210) at ../../gcc/emit-rtl.c:5326
#3 0x0826a3da in duplicate_insn_chain (from=0xb7c433a8, to=0xb7c4b18c)
at ../../gcc/cfglayout.c:971
#4 0x0826a4d6 in cfg_layout_duplicate_bb (bb=0xb7c44168)
at ../../gcc/cfglayout.c:1038
#5 0x0852535a in duplicate_block (bb=0xb7c44168, e=0x0,
after=0xb7c44480) at ../../gcc/cfghooks.c:804
#6 0x0826a914 in copy_bbs (bbs=0xa600cc8, n=2, new_bbs=0xa5e6b58,
edges=0xbfd5283c, num_edges=2, new_edges=0xbfd52834,
base=0xa60aa30, after=0xb7c44480) at ../../gcc/cfglayout.c:1221
#7 0x0827152d in duplicate_loop_to_header_edge (loop=0xa60aa30,
e=0xb7c4e780, loops=0xa5e6358, ndupl=7,
wont_exit=0xa5e5038, orig=0x0, to_remove=0x0, n_to_remove=0x0,
flags=3) at ../../gcc/cfgloopmanip.c:990
#8 0x08277508 in unroll_loop_stupid (loops=0xa5e6358, loop=0xa60aa30)
at ../../gcc/loop-unroll.c:1452
#9 0x08274839 in unroll_and_peel_loops (loops=0xa5e6358, flags=6)
at ../../gcc/loop-unroll.c:190
Also there is a progressions in that gcc.c-torture/compile/20000804-1.c
no longer ICEs at -O3 -funroll-loops.
I will run the full testsuite tomorrow when I get to work as I have
access to hardware there.
Thanks,
Andrew Pinski
