https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123195
Rainer Orth <ro at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |rguenth at gcc dot gnu.org
--- Comment #1 from Rainer Orth <ro at gcc dot gnu.org> ---
When looking at the gcc.dg/vect/pr104116-ceil-div-2.c failure, I find that
the 'no optab' message is emitted here:
#0 dump_printf_loc (metadata=..., loc=..., format=0x5bc4f8 "no optab.\n")
at /vol/gcc/src/hg/master/local/gcc/dumpfile.cc:1385
#1 0x0158ddc0 in vectorizable_operation (vinfo=0x2adad58,
stmt_info=0x2ade1b0, gsi=0x0, slp_node=0x2b00350, cost_vec=0xffbfe1a8)
at /vol/gcc/src/hg/master/local/gcc/tree-vect-stmts.cc:6566
(gdb) p code
$1 = CEIL_DIV_EXPR
(gdb) p vectype
$2 = <vector_type 0xface28a0>
(gdb) pt $2
<vector_type face28a0
type <integer_type fac24360 int sizes-gimplified public SI
size <integer_cst fac0ae40 constant 32>
unit-size <integer_cst fac0ae58 constant 4>
align:32 warn_if_not_align:0 symtab:0 alias-set 1 canonical-type
fac24360 precision:32 min <integer_cst fac0b080 -2147483648> max <integer_cst
fac0b098 2147483647>
pointer_to_this <pointer_type fac24f60>>
V2SI
size <integer_cst fac0aea0 type <integer_type fac24060 bitsizetype>
constant 64>
unit-size <integer_cst fac0aeb8 type <integer_type fac24000 sizetype>
constant 8>
align:64 warn_if_not_align:0 symtab:0 alias-set -1 canonical-type face28a0
nunits:2>
I wonder if this is covered by existing effective-target keywords
(vect_call_{ceil,floor,round}) or needs new ones.
