[PATCH] Improve pattern recognizer for division by constant (PR tree-optimization/51581)
Richard Guenther
richard.guenther@gmail.com
Fri Jun 15 09:57:00 GMT 2012
On Thu, Jun 14, 2012 at 10:58 PM, Jakub Jelinek <jakub@redhat.com> wrote:
> Hi!
>
> This patch performs some of the expand_divmod tricks to expand
> integer division by constant using widening vector multiplication,
> some shifts and/or additions/subtractions.
>
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
Ok.
Maybe you can factor some of the code for use in already vector-form
division lowering in tree-vect-generic.c? Not sure if that is easily possible
though.
Thanks,
Richard.
> 2012-06-14 Jakub Jelinek <jakub@redhat.com>
>
> PR tree-optimization/51581
> * expr.h (choose_multiplier): New prototype.
> * expmed.c (choose_multiplier): No longer static.
> Change multiplier_ptr from rtx * to UHWI *.
> (expand_divmod): Adjust callers.
> * tree-vect-patterns.c (vect_recog_sdivmod_pow2_pattern):
> Renamed to...
> (vect_recog_divmod_pattern): ... this. Pass bb_vinfo as last
> argument to new_stmt_vec_info. Attempt to optimize also divisions
> by non-pow2 constants if integer vector division isn't supported.
> * tree-vect-stmts.c (vect_analyze_stmt): If node != NULL,
> don't look at pattern stmts and sequences.
>
> * gcc.c-torture/execute/pr51581-1.c: New test.
> * gcc.c-torture/execute/pr51581-2.c: New test.
> * gcc.dg/vect/pr51581-1.c: New test.
> * gcc.dg/vect/pr51581-2.c: New test.
> * gcc.dg/vect/pr51581-3.c: New test.
> * gcc.target/i386/avx-pr51581-1.c: New test.
> * gcc.target/i386/avx-pr51581-2.c: New test.
> * gcc.target/i386/avx2-pr51581-1.c: New test.
> * gcc.target/i386/avx2-pr51581-2.c: New test.
> * gcc.dg/vect/slp-26.c (main1): Divide by 0x8031 instead of 3.
>
> --- gcc/expr.h.jj 2012-01-02 20:39:59.000000000 +0100
> +++ gcc/expr.h 2012-06-13 12:02:26.498269223 +0200
> @@ -243,6 +243,13 @@ extern rtx emit_store_flag (rtx, enum rt
> /* Like emit_store_flag, but always succeeds. */
> extern rtx emit_store_flag_force (rtx, enum rtx_code, rtx, rtx,
> enum machine_mode, int, int);
> +
> +/* Choose a minimal N + 1 bit approximation to 1/D that can be used to
> + replace division by D, and put the least significant N bits of the result
> + in *MULTIPLIER_PTR and return the most significant bit. */
> +extern unsigned HOST_WIDE_INT choose_multiplier (unsigned HOST_WIDE_INT, int,
> + int, unsigned HOST_WIDE_INT *,
> + int *, int *);
>
> /* Functions from builtins.c: */
> extern rtx expand_builtin (tree, rtx, rtx, enum machine_mode, int);
> --- gcc/expmed.c.jj 2012-06-04 11:18:44.000000000 +0200
> +++ gcc/expmed.c 2012-06-13 14:51:52.343232937 +0200
> @@ -2363,8 +2363,6 @@ static bool choose_mult_variant (enum ma
> struct algorithm *, enum mult_variant *, int);
> static rtx expand_mult_const (enum machine_mode, rtx, HOST_WIDE_INT, rtx,
> const struct algorithm *, enum mult_variant);
> -static unsigned HOST_WIDE_INT choose_multiplier (unsigned HOST_WIDE_INT, int,
> - int, rtx *, int *, int *);
> static unsigned HOST_WIDE_INT invert_mod2n (unsigned HOST_WIDE_INT, int);
> static rtx extract_high_half (enum machine_mode, rtx);
> static rtx expand_mult_highpart (enum machine_mode, rtx, rtx, rtx, int, int);
> @@ -3293,10 +3291,10 @@ ceil_log2 (unsigned HOST_WIDE_INT x)
> Using this function, x/D will be equal to (x * m) >> (*POST_SHIFT_PTR),
> where m is the full HOST_BITS_PER_WIDE_INT + 1 bit multiplier. */
>
> -static
> unsigned HOST_WIDE_INT
> choose_multiplier (unsigned HOST_WIDE_INT d, int n, int precision,
> - rtx *multiplier_ptr, int *post_shift_ptr, int *lgup_ptr)
> + unsigned HOST_WIDE_INT *multiplier_ptr,
> + int *post_shift_ptr, int *lgup_ptr)
> {
> HOST_WIDE_INT mhigh_hi, mlow_hi;
> unsigned HOST_WIDE_INT mhigh_lo, mlow_lo;
> @@ -3368,12 +3366,12 @@ choose_multiplier (unsigned HOST_WIDE_IN
> if (n < HOST_BITS_PER_WIDE_INT)
> {
> unsigned HOST_WIDE_INT mask = ((unsigned HOST_WIDE_INT) 1 << n) - 1;
> - *multiplier_ptr = GEN_INT (mhigh_lo & mask);
> + *multiplier_ptr = mhigh_lo & mask;
> return mhigh_lo >= mask;
> }
> else
> {
> - *multiplier_ptr = GEN_INT (mhigh_lo);
> + *multiplier_ptr = mhigh_lo;
> return mhigh_hi;
> }
> }
> @@ -4053,10 +4051,9 @@ expand_divmod (int rem_flag, enum tree_c
> {
> if (unsignedp)
> {
> - unsigned HOST_WIDE_INT mh;
> + unsigned HOST_WIDE_INT mh, ml;
> int pre_shift, post_shift;
> int dummy;
> - rtx ml;
> unsigned HOST_WIDE_INT d = (INTVAL (op1)
> & GET_MODE_MASK (compute_mode));
>
> @@ -4118,7 +4115,8 @@ expand_divmod (int rem_flag, enum tree_c
> = (shift_cost[speed][compute_mode][post_shift - 1]
> + shift_cost[speed][compute_mode][1]
> + 2 * add_cost[speed][compute_mode]);
> - t1 = expand_mult_highpart (compute_mode, op0, ml,
> + t1 = expand_mult_highpart (compute_mode, op0,
> + GEN_INT (ml),
> NULL_RTX, 1,
> max_cost - extra_cost);
> if (t1 == 0)
> @@ -4149,7 +4147,8 @@ expand_divmod (int rem_flag, enum tree_c
> extra_cost
> = (shift_cost[speed][compute_mode][pre_shift]
> + shift_cost[speed][compute_mode][post_shift]);
> - t2 = expand_mult_highpart (compute_mode, t1, ml,
> + t2 = expand_mult_highpart (compute_mode, t1,
> + GEN_INT (ml),
> NULL_RTX, 1,
> max_cost - extra_cost);
> if (t2 == 0)
> @@ -4262,8 +4261,7 @@ expand_divmod (int rem_flag, enum tree_c
> else if (size <= HOST_BITS_PER_WIDE_INT)
> {
> choose_multiplier (abs_d, size, size - 1,
> - &mlr, &post_shift, &lgup);
> - ml = (unsigned HOST_WIDE_INT) INTVAL (mlr);
> + &ml, &post_shift, &lgup);
> if (ml < (unsigned HOST_WIDE_INT) 1 << (size - 1))
> {
> rtx t1, t2, t3;
> @@ -4275,8 +4273,8 @@ expand_divmod (int rem_flag, enum tree_c
> extra_cost = (shift_cost[speed][compute_mode][post_shift]
> + shift_cost[speed][compute_mode][size - 1]
> + add_cost[speed][compute_mode]);
> - t1 = expand_mult_highpart (compute_mode, op0, mlr,
> - NULL_RTX, 0,
> + t1 = expand_mult_highpart (compute_mode, op0,
> + GEN_INT (ml), NULL_RTX, 0,
> max_cost - extra_cost);
> if (t1 == 0)
> goto fail1;
> @@ -4356,10 +4354,9 @@ expand_divmod (int rem_flag, enum tree_c
> /* We will come here only for signed operations. */
> if (op1_is_constant && HOST_BITS_PER_WIDE_INT >= size)
> {
> - unsigned HOST_WIDE_INT mh;
> + unsigned HOST_WIDE_INT mh, ml;
> int pre_shift, lgup, post_shift;
> HOST_WIDE_INT d = INTVAL (op1);
> - rtx ml;
>
> if (d > 0)
> {
> @@ -4399,8 +4396,8 @@ expand_divmod (int rem_flag, enum tree_c
> extra_cost = (shift_cost[speed][compute_mode][post_shift]
> + shift_cost[speed][compute_mode][size - 1]
> + 2 * add_cost[speed][compute_mode]);
> - t3 = expand_mult_highpart (compute_mode, t2, ml,
> - NULL_RTX, 1,
> + t3 = expand_mult_highpart (compute_mode, t2,
> + GEN_INT (ml), NULL_RTX, 1,
> max_cost - extra_cost);
> if (t3 != 0)
> {
> --- gcc/tree-vect-patterns.c.jj 2012-05-08 14:16:31.000000000 +0200
> +++ gcc/tree-vect-patterns.c 2012-06-14 13:22:27.123090754 +0200
> @@ -53,8 +53,8 @@ static gimple vect_recog_widen_shift_pat
> tree *, tree *);
> static gimple vect_recog_vector_vector_shift_pattern (VEC (gimple, heap) **,
> tree *, tree *);
> -static gimple vect_recog_sdivmod_pow2_pattern (VEC (gimple, heap) **,
> - tree *, tree *);
> +static gimple vect_recog_divmod_pattern (VEC (gimple, heap) **,
> + tree *, tree *);
> static gimple vect_recog_mixed_size_cond_pattern (VEC (gimple, heap) **,
> tree *, tree *);
> static gimple vect_recog_bool_pattern (VEC (gimple, heap) **, tree *, tree *);
> @@ -66,7 +66,7 @@ static vect_recog_func_ptr vect_vect_rec
> vect_recog_widen_shift_pattern,
> vect_recog_over_widening_pattern,
> vect_recog_vector_vector_shift_pattern,
> - vect_recog_sdivmod_pow2_pattern,
> + vect_recog_divmod_pattern,
> vect_recog_mixed_size_cond_pattern,
> vect_recog_bool_pattern};
>
> @@ -1585,29 +1585,30 @@ vect_recog_vector_vector_shift_pattern (
> return pattern_stmt;
> }
>
> -/* Detect a signed division by power of two constant that wouldn't be
> +/* Detect a signed division by a constant that wouldn't be
> otherwise vectorized:
>
> type a_t, b_t;
>
> S1 a_t = b_t / N;
>
> - where type 'type' is a signed integral type and N is a constant positive
> - power of two.
> + where type 'type' is an integral type and N is a constant.
>
> - Similarly handle signed modulo by power of two constant:
> + Similarly handle modulo by a constant:
>
> S4 a_t = b_t % N;
>
> Input/Output:
>
> * STMTS: Contains a stmt from which the pattern search begins,
> - i.e. the division stmt. S1 is replaced by:
> + i.e. the division stmt. S1 is replaced by if N is a power
> + of two constant and type is signed:
> S3 y_t = b_t < 0 ? N - 1 : 0;
> S2 x_t = b_t + y_t;
> S1' a_t = x_t >> log2 (N);
>
> - S4 is replaced by (where *_T temporaries have unsigned type):
> + S4 is replaced if N is a power of two constant and
> + type is signed by (where *_T temporaries have unsigned type):
> S9 y_T = b_t < 0 ? -1U : 0U;
> S8 z_T = y_T >> (sizeof (type_t) * CHAR_BIT - log2 (N));
> S7 z_t = (type) z_T;
> @@ -1625,16 +1626,22 @@ vect_recog_vector_vector_shift_pattern (
> S1 or modulo S4 stmt. */
>
> static gimple
> -vect_recog_sdivmod_pow2_pattern (VEC (gimple, heap) **stmts,
> - tree *type_in, tree *type_out)
> +vect_recog_divmod_pattern (VEC (gimple, heap) **stmts,
> + tree *type_in, tree *type_out)
> {
> gimple last_stmt = VEC_pop (gimple, *stmts);
> - tree oprnd0, oprnd1, vectype, itype, cond;
> + tree oprnd0, oprnd1, vectype, itype, witype, vecwtype, cond;
> gimple pattern_stmt, def_stmt;
> enum tree_code rhs_code;
> stmt_vec_info stmt_vinfo = vinfo_for_stmt (last_stmt);
> loop_vec_info loop_vinfo = STMT_VINFO_LOOP_VINFO (stmt_vinfo);
> + bb_vec_info bb_vinfo = STMT_VINFO_BB_VINFO (stmt_vinfo);
> optab optab;
> + tree dummy, q;
> + enum tree_code dummy_code;
> + int dummy_int, prec;
> + VEC (tree, heap) *dummy_vec;
> + stmt_vec_info def_stmt_vinfo;
>
> if (!is_gimple_assign (last_stmt))
> return NULL;
> @@ -1658,10 +1665,7 @@ vect_recog_sdivmod_pow2_pattern (VEC (gi
> if (TREE_CODE (oprnd0) != SSA_NAME
> || TREE_CODE (oprnd1) != INTEGER_CST
> || TREE_CODE (itype) != INTEGER_TYPE
> - || TYPE_UNSIGNED (itype)
> - || TYPE_PRECISION (itype) != GET_MODE_PRECISION (TYPE_MODE (itype))
> - || !integer_pow2p (oprnd1)
> - || tree_int_cst_sgn (oprnd1) != 1)
> + || TYPE_PRECISION (itype) != GET_MODE_PRECISION (TYPE_MODE (itype)))
> return NULL;
>
> vectype = get_vectype_for_scalar_type (itype);
> @@ -1680,104 +1684,438 @@ vect_recog_sdivmod_pow2_pattern (VEC (gi
> return NULL;
> }
>
> - /* Pattern detected. */
> - if (vect_print_dump_info (REPORT_DETAILS))
> - fprintf (vect_dump, "vect_recog_sdivmod_pow2_pattern: detected: ");
> -
> - cond = build2 (LT_EXPR, boolean_type_node, oprnd0, build_int_cst (itype, 0));
> - if (rhs_code == TRUNC_DIV_EXPR)
> + prec = TYPE_PRECISION (itype);
> + if (integer_pow2p (oprnd1))
> {
> - tree var = vect_recog_temp_ssa_var (itype, NULL);
> - def_stmt
> - = gimple_build_assign_with_ops3 (COND_EXPR, var, cond,
> - fold_build2 (MINUS_EXPR, itype,
> - oprnd1,
> - build_int_cst (itype,
> - 1)),
> - build_int_cst (itype, 0));
> - new_pattern_def_seq (stmt_vinfo, def_stmt);
> - var = vect_recog_temp_ssa_var (itype, NULL);
> - def_stmt
> - = gimple_build_assign_with_ops (PLUS_EXPR, var, oprnd0,
> - gimple_assign_lhs (def_stmt));
> - append_pattern_def_seq (stmt_vinfo, def_stmt);
> + if (TYPE_UNSIGNED (itype) || tree_int_cst_sgn (oprnd1) != 1)
> + return NULL;
>
> - pattern_stmt
> - = gimple_build_assign_with_ops (RSHIFT_EXPR,
> - vect_recog_temp_ssa_var (itype, NULL),
> - var,
> - build_int_cst (itype,
> - tree_log2 (oprnd1)));
> + /* Pattern detected. */
> + if (vect_print_dump_info (REPORT_DETAILS))
> + fprintf (vect_dump, "vect_recog_divmod_pattern: detected: ");
> +
> + cond = build2 (LT_EXPR, boolean_type_node, oprnd0,
> + build_int_cst (itype, 0));
> + if (rhs_code == TRUNC_DIV_EXPR)
> + {
> + tree var = vect_recog_temp_ssa_var (itype, NULL);
> + tree shift;
> + def_stmt
> + = gimple_build_assign_with_ops3 (COND_EXPR, var, cond,
> + fold_build2 (MINUS_EXPR, itype,
> + oprnd1,
> + build_int_cst (itype,
> + 1)),
> + build_int_cst (itype, 0));
> + new_pattern_def_seq (stmt_vinfo, def_stmt);
> + var = vect_recog_temp_ssa_var (itype, NULL);
> + def_stmt
> + = gimple_build_assign_with_ops (PLUS_EXPR, var, oprnd0,
> + gimple_assign_lhs (def_stmt));
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> +
> + shift = build_int_cst (itype, tree_log2 (oprnd1));
> + pattern_stmt
> + = gimple_build_assign_with_ops (RSHIFT_EXPR,
> + vect_recog_temp_ssa_var (itype,
> + NULL),
> + var, shift);
> + }
> + else
> + {
> + tree signmask;
> + STMT_VINFO_PATTERN_DEF_SEQ (stmt_vinfo) = NULL;
> + if (compare_tree_int (oprnd1, 2) == 0)
> + {
> + signmask = vect_recog_temp_ssa_var (itype, NULL);
> + def_stmt
> + = gimple_build_assign_with_ops3 (COND_EXPR, signmask, cond,
> + build_int_cst (itype, 1),
> + build_int_cst (itype, 0));
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> + }
> + else
> + {
> + tree utype
> + = build_nonstandard_integer_type (prec, 1);
> + tree vecutype = get_vectype_for_scalar_type (utype);
> + tree shift
> + = build_int_cst (utype, GET_MODE_BITSIZE (TYPE_MODE (itype))
> + - tree_log2 (oprnd1));
> + tree var = vect_recog_temp_ssa_var (utype, NULL);
> +
> + def_stmt
> + = gimple_build_assign_with_ops3 (COND_EXPR, var, cond,
> + build_int_cst (utype, -1),
> + build_int_cst (utype, 0));
> + def_stmt_vinfo
> + = new_stmt_vec_info (def_stmt, loop_vinfo, bb_vinfo);
> + set_vinfo_for_stmt (def_stmt, def_stmt_vinfo);
> + STMT_VINFO_VECTYPE (def_stmt_vinfo) = vecutype;
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> + var = vect_recog_temp_ssa_var (utype, NULL);
> + def_stmt
> + = gimple_build_assign_with_ops (RSHIFT_EXPR, var,
> + gimple_assign_lhs (def_stmt),
> + shift);
> + def_stmt_vinfo
> + = new_stmt_vec_info (def_stmt, loop_vinfo, bb_vinfo);
> + set_vinfo_for_stmt (def_stmt, def_stmt_vinfo);
> + STMT_VINFO_VECTYPE (def_stmt_vinfo) = vecutype;
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> + signmask = vect_recog_temp_ssa_var (itype, NULL);
> + def_stmt
> + = gimple_build_assign_with_ops (NOP_EXPR, signmask, var,
> + NULL_TREE);
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> + }
> + def_stmt
> + = gimple_build_assign_with_ops (PLUS_EXPR,
> + vect_recog_temp_ssa_var (itype,
> + NULL),
> + oprnd0, signmask);
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> + def_stmt
> + = gimple_build_assign_with_ops (BIT_AND_EXPR,
> + vect_recog_temp_ssa_var (itype,
> + NULL),
> + gimple_assign_lhs (def_stmt),
> + fold_build2 (MINUS_EXPR, itype,
> + oprnd1,
> + build_int_cst (itype,
> + 1)));
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> +
> + pattern_stmt
> + = gimple_build_assign_with_ops (MINUS_EXPR,
> + vect_recog_temp_ssa_var (itype,
> + NULL),
> + gimple_assign_lhs (def_stmt),
> + signmask);
> + }
> +
> + if (vect_print_dump_info (REPORT_DETAILS))
> + print_gimple_stmt (vect_dump, pattern_stmt, 0, TDF_SLIM);
> +
> + VEC_safe_push (gimple, heap, *stmts, last_stmt);
> +
> + *type_in = vectype;
> + *type_out = vectype;
> + return pattern_stmt;
> }
> - else
> +
> + if (!host_integerp (oprnd1, TYPE_UNSIGNED (itype))
> + || integer_zerop (oprnd1)
> + || prec > HOST_BITS_PER_WIDE_INT)
> + return NULL;
> +
> + witype = build_nonstandard_integer_type (prec * 2,
> + TYPE_UNSIGNED (itype));
> + vecwtype = get_vectype_for_scalar_type (witype);
> + if (vecwtype == NULL_TREE)
> + return NULL;
> +
> + if (!supportable_widening_operation (WIDEN_MULT_EXPR, last_stmt,
> + vecwtype, vectype,
> + &dummy, &dummy, &dummy_code,
> + &dummy_code, &dummy_int, &dummy_vec))
> + return NULL;
> +
> + STMT_VINFO_PATTERN_DEF_SEQ (stmt_vinfo) = NULL;
> +
> + if (TYPE_UNSIGNED (itype))
> {
> - tree signmask;
> - STMT_VINFO_PATTERN_DEF_SEQ (stmt_vinfo) = NULL;
> - if (compare_tree_int (oprnd1, 2) == 0)
> + unsigned HOST_WIDE_INT mh, ml;
> + int pre_shift, post_shift;
> + unsigned HOST_WIDE_INT d = tree_low_cst (oprnd1, 1)
> + & GET_MODE_MASK (TYPE_MODE (itype));
> + tree t1, t2, t3, t4, t5, t6;
> +
> + if (d >= ((unsigned HOST_WIDE_INT) 1 << (prec - 1)))
> + /* FIXME: Can transform this into oprnd0 >= oprnd1 ? 1 : 0. */
> + return NULL;
> +
> + /* Find a suitable multiplier and right shift count
> + instead of multiplying with D. */
> + mh = choose_multiplier (d, prec, prec, &ml, &post_shift, &dummy_int);
> +
> + /* If the suggested multiplier is more than SIZE bits, we can do better
> + for even divisors, using an initial right shift. */
> + if (mh != 0 && (d & 1) == 0)
> + {
> + pre_shift = floor_log2 (d & -d);
> + mh = choose_multiplier (d >> pre_shift, prec, prec - pre_shift,
> + &ml, &post_shift, &dummy_int);
> + gcc_assert (!mh);
> + }
> + else
> + pre_shift = 0;
> +
> + if (mh != 0)
> {
> - signmask = vect_recog_temp_ssa_var (itype, NULL);
> + if (post_shift - 1 >= prec)
> + return NULL;
> +
> + /* t1 = oprnd0 w* ml;
> + t2 = t1 >> prec;
> + t3 = (type) t2;
> + t4 = oprnd0 - t3;
> + t5 = t4 >> 1;
> + t6 = t3 + t5;
> + q = t6 >> (post_shift - 1); */
> + t1 = vect_recog_temp_ssa_var (witype, NULL);
> def_stmt
> - = gimple_build_assign_with_ops3 (COND_EXPR, signmask, cond,
> - build_int_cst (itype, 1),
> - build_int_cst (itype, 0));
> + = gimple_build_assign_with_ops (WIDEN_MULT_EXPR, t1, oprnd0,
> + build_int_cst (itype, ml));
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> + def_stmt_vinfo = new_stmt_vec_info (def_stmt, loop_vinfo, bb_vinfo);
> + set_vinfo_for_stmt (def_stmt, def_stmt_vinfo);
> + STMT_VINFO_VECTYPE (def_stmt_vinfo) = vecwtype;
> +
> + t2 = vect_recog_temp_ssa_var (witype, NULL);
> + def_stmt
> + = gimple_build_assign_with_ops (RSHIFT_EXPR, t2, t1,
> + build_int_cst (itype, prec));
> append_pattern_def_seq (stmt_vinfo, def_stmt);
> + def_stmt_vinfo = new_stmt_vec_info (def_stmt, loop_vinfo, bb_vinfo);
> + set_vinfo_for_stmt (def_stmt, def_stmt_vinfo);
> + STMT_VINFO_VECTYPE (def_stmt_vinfo) = vecwtype;
> +
> + t3 = vect_recog_temp_ssa_var (itype, NULL);
> + def_stmt
> + = gimple_build_assign_with_ops (NOP_EXPR, t3, t2, NULL_TREE);
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> +
> + t4 = vect_recog_temp_ssa_var (itype, NULL);
> + def_stmt
> + = gimple_build_assign_with_ops (MINUS_EXPR, t4, oprnd0, t3);
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> +
> + t5 = vect_recog_temp_ssa_var (itype, NULL);
> + def_stmt
> + = gimple_build_assign_with_ops (RSHIFT_EXPR, t5, t4,
> + integer_one_node);
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> +
> + t6 = vect_recog_temp_ssa_var (itype, NULL);
> + def_stmt
> + = gimple_build_assign_with_ops (PLUS_EXPR, t6, t3, t5);
> +
> + if (post_shift != 1)
> + {
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> +
> + q = vect_recog_temp_ssa_var (witype, NULL);
> + pattern_stmt
> + = gimple_build_assign_with_ops (RSHIFT_EXPR, q, t6,
> + build_int_cst (itype,
> + post_shift
> + - 1));
> + }
> + else
> + {
> + q = t6;
> + pattern_stmt = def_stmt;
> + }
> }
> else
> {
> - tree utype
> - = build_nonstandard_integer_type (TYPE_PRECISION (itype), 1);
> - tree vecutype = get_vectype_for_scalar_type (utype);
> - tree shift
> - = build_int_cst (utype, GET_MODE_BITSIZE (TYPE_MODE (itype))
> - - tree_log2 (oprnd1));
> - tree var = vect_recog_temp_ssa_var (utype, NULL);
> - stmt_vec_info def_stmt_vinfo;
> + if (pre_shift >= prec || post_shift >= prec)
> + return NULL;
>
> + /* t1 = oprnd0 >> pre_shift;
> + t2 = t1 w* ml;
> + t3 = t2 >> (prec + post_shift);
> + q = (type) t3; */
> + if (pre_shift)
> + {
> + t1 = vect_recog_temp_ssa_var (itype, NULL);
> + def_stmt
> + = gimple_build_assign_with_ops (RSHIFT_EXPR, t1, oprnd0,
> + build_int_cst (NULL,
> + pre_shift));
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> + }
> + else
> + t1 = oprnd0;
> +
> + t2 = vect_recog_temp_ssa_var (witype, NULL);
> def_stmt
> - = gimple_build_assign_with_ops3 (COND_EXPR, var, cond,
> - build_int_cst (utype, -1),
> - build_int_cst (utype, 0));
> - def_stmt_vinfo = new_stmt_vec_info (def_stmt, loop_vinfo, NULL);
> + = gimple_build_assign_with_ops (WIDEN_MULT_EXPR, t2, t1,
> + build_int_cst (itype, ml));
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> + def_stmt_vinfo = new_stmt_vec_info (def_stmt, loop_vinfo, bb_vinfo);
> set_vinfo_for_stmt (def_stmt, def_stmt_vinfo);
> - STMT_VINFO_VECTYPE (def_stmt_vinfo) = vecutype;
> + STMT_VINFO_VECTYPE (def_stmt_vinfo) = vecwtype;
> +
> + t3 = vect_recog_temp_ssa_var (witype, NULL);
> + def_stmt
> + = gimple_build_assign_with_ops (RSHIFT_EXPR, t3, t2,
> + build_int_cst (itype, post_shift
> + + prec));
> append_pattern_def_seq (stmt_vinfo, def_stmt);
> - var = vect_recog_temp_ssa_var (utype, NULL);
> + def_stmt_vinfo = new_stmt_vec_info (def_stmt, loop_vinfo, bb_vinfo);
> + set_vinfo_for_stmt (def_stmt, def_stmt_vinfo);
> + STMT_VINFO_VECTYPE (def_stmt_vinfo) = vecwtype;
> +
> + q = vect_recog_temp_ssa_var (itype, NULL);
> + pattern_stmt
> + = gimple_build_assign_with_ops (NOP_EXPR, q, t3, NULL_TREE);
> + }
> + }
> + else
> + {
> + unsigned HOST_WIDE_INT ml;
> + int lgup, post_shift;
> + HOST_WIDE_INT d = tree_low_cst (oprnd1, 0);
> + unsigned HOST_WIDE_INT abs_d;
> + bool add = false;
> + tree uwitype = NULL, vecuwtype = NULL;
> + tree t1, t2, t3, t4, t5, t6, t7;
> +
> + /* Give up for -1. */
> + if (d == -1)
> + return NULL;
> +
> + if (!vect_supportable_shift (RSHIFT_EXPR, witype))
> + {
> + uwitype = build_nonstandard_integer_type (prec * 2, 1);
> + vecuwtype = get_vectype_for_scalar_type (uwitype);
> + if (vecuwtype == NULL_TREE)
> + return NULL;
> + }
> +
> + /* Since d might be INT_MIN, we have to cast to
> + unsigned HOST_WIDE_INT before negating to avoid
> + undefined signed overflow. */
> + abs_d = (d >= 0
> + ? (unsigned HOST_WIDE_INT) d
> + : - (unsigned HOST_WIDE_INT) d);
> +
> + /* n rem d = n rem -d */
> + if (rhs_code == TRUNC_MOD_EXPR && d < 0)
> + {
> + d = abs_d;
> + oprnd1 = build_int_cst (itype, abs_d);
> + }
> + else if (HOST_BITS_PER_WIDE_INT >= prec
> + && abs_d == (unsigned HOST_WIDE_INT) 1 << (prec - 1))
> + /* This case is not handled correctly below. */
> + return NULL;
> +
> + choose_multiplier (abs_d, prec, prec - 1, &ml, &post_shift, &lgup);
> + if (ml >= (unsigned HOST_WIDE_INT) 1 << (prec - 1))
> + {
> + add = true;
> + ml |= (~(unsigned HOST_WIDE_INT) 0) << (prec - 1);
> + }
> + if (post_shift >= prec)
> + return NULL;
> +
> + /* t1 = oprnd1 w* ml; */
> + t1 = vect_recog_temp_ssa_var (witype, NULL);
> + def_stmt
> + = gimple_build_assign_with_ops (WIDEN_MULT_EXPR, t1, oprnd0,
> + build_int_cst (itype, ml));
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> + def_stmt_vinfo = new_stmt_vec_info (def_stmt, loop_vinfo, bb_vinfo);
> + set_vinfo_for_stmt (def_stmt, def_stmt_vinfo);
> + STMT_VINFO_VECTYPE (def_stmt_vinfo) = vecwtype;
> +
> + if (vecuwtype != NULL)
> + {
> + /* t2 = (uwtype) t1; */
> + t2 = vect_recog_temp_ssa_var (uwitype, NULL);
> def_stmt
> - = gimple_build_assign_with_ops (RSHIFT_EXPR, var,
> - gimple_assign_lhs (def_stmt),
> - shift);
> - def_stmt_vinfo = new_stmt_vec_info (def_stmt, loop_vinfo, NULL);
> + = gimple_build_assign_with_ops (NOP_EXPR, t2, t1, NULL_TREE);
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> + def_stmt_vinfo = new_stmt_vec_info (def_stmt, loop_vinfo, bb_vinfo);
> set_vinfo_for_stmt (def_stmt, def_stmt_vinfo);
> - STMT_VINFO_VECTYPE (def_stmt_vinfo) = vecutype;
> + STMT_VINFO_VECTYPE (def_stmt_vinfo) = vecuwtype;
> + }
> + else
> + t2 = t1;
> +
> + /* t3 = t2 >> prec; or t3 = t2 >> (prec + post_shift); */
> + t3 = vect_recog_temp_ssa_var (vecuwtype ? uwitype : witype, NULL);
> + def_stmt
> + = gimple_build_assign_with_ops (RSHIFT_EXPR, t3, t2,
> + build_int_cst (itype,
> + prec
> + + (!add
> + && vecuwtype == NULL
> + ? post_shift : 0)));
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> + def_stmt_vinfo = new_stmt_vec_info (def_stmt, loop_vinfo, bb_vinfo);
> + set_vinfo_for_stmt (def_stmt, def_stmt_vinfo);
> + STMT_VINFO_VECTYPE (def_stmt_vinfo) = vecuwtype ? vecuwtype : vecwtype;
> +
> + /* t4 = (type) t3; */
> + t4 = vect_recog_temp_ssa_var (itype, NULL);
> + def_stmt
> + = gimple_build_assign_with_ops (NOP_EXPR, t4, t3, NULL_TREE);
> + append_pattern_def_seq (stmt_vinfo, def_stmt);
> +
> + if (add)
> + {
> + /* t5 = t4 + oprnd0; */
> + t5 = vect_recog_temp_ssa_var (itype, NULL);
> + def_stmt
> + = gimple_build_assign_with_ops (PLUS_EXPR, t5, t4, oprnd0);
> append_pattern_def_seq (stmt_vinfo, def_stmt);
> - signmask = vect_recog_temp_ssa_var (itype, NULL);
> + }
> + else
> + t5 = t4;
> +
> + if ((add || vecuwtype != NULL) && post_shift)
> + {
> + /* t6 = t5 >> post_shift; */
> + t6 = vect_recog_temp_ssa_var (itype, NULL);
> def_stmt
> - = gimple_build_assign_with_ops (NOP_EXPR, signmask, var,
> - NULL_TREE);
> + = gimple_build_assign_with_ops (RSHIFT_EXPR, t6, t5,
> + build_int_cst (itype, post_shift));
> append_pattern_def_seq (stmt_vinfo, def_stmt);
> }
> + else
> + t6 = t5;
> +
> + /* t7 = oprnd0 >> (prec - 1); */
> + t7 = vect_recog_temp_ssa_var (itype, NULL);
> def_stmt
> - = gimple_build_assign_with_ops (PLUS_EXPR,
> - vect_recog_temp_ssa_var (itype, NULL),
> - oprnd0, signmask);
> + = gimple_build_assign_with_ops (RSHIFT_EXPR, t7, oprnd0,
> + build_int_cst (itype, prec - 1));
> append_pattern_def_seq (stmt_vinfo, def_stmt);
> +
> + /* q = t6 - t7; or q = t7 - t6; */
> + q = vect_recog_temp_ssa_var (itype, NULL);
> + pattern_stmt
> + = gimple_build_assign_with_ops (MINUS_EXPR, q, d < 0 ? t7 : t6,
> + d < 0 ? t6 : t7);
> + }
> +
> + if (rhs_code == TRUNC_MOD_EXPR)
> + {
> + tree r, t1;
> +
> + /* We divided. Now finish by:
> + t1 = q * oprnd1;
> + r = oprnd0 - t1; */
> + append_pattern_def_seq (stmt_vinfo, pattern_stmt);
> +
> + t1 = vect_recog_temp_ssa_var (itype, NULL);
> def_stmt
> - = gimple_build_assign_with_ops (BIT_AND_EXPR,
> - vect_recog_temp_ssa_var (itype, NULL),
> - gimple_assign_lhs (def_stmt),
> - fold_build2 (MINUS_EXPR, itype,
> - oprnd1,
> - build_int_cst (itype,
> - 1)));
> + = gimple_build_assign_with_ops (MULT_EXPR, t1, q, oprnd1);
> append_pattern_def_seq (stmt_vinfo, def_stmt);
>
> + r = vect_recog_temp_ssa_var (itype, NULL);
> pattern_stmt
> - = gimple_build_assign_with_ops (MINUS_EXPR,
> - vect_recog_temp_ssa_var (itype, NULL),
> - gimple_assign_lhs (def_stmt),
> - signmask);
> + = gimple_build_assign_with_ops (MINUS_EXPR, r, oprnd0, t1);
> }
>
> + /* Pattern detected. */
> + if (vect_print_dump_info (REPORT_DETAILS))
> + fprintf (vect_dump, "vect_recog_divmod_pattern: detected: ");
> +
> if (vect_print_dump_info (REPORT_DETAILS))
> print_gimple_stmt (vect_dump, pattern_stmt, 0, TDF_SLIM);
>
> --- gcc/tree-vect-stmts.c.jj 2012-06-01 14:41:05.000000000 +0200
> +++ gcc/tree-vect-stmts.c 2012-06-14 14:35:10.494173297 +0200
> @@ -5401,7 +5401,9 @@ vect_analyze_stmt (gimple stmt, bool *ne
>
> Pattern statement needs to be analyzed instead of the original statement
> if the original statement is not relevant. Otherwise, we analyze both
> - statements. */
> + statements. In basic blocks we are called from some SLP instance
> + traversal, don't analyze pattern stmts instead, the pattern stmts
> + already will be part of SLP instance. */
>
> pattern_stmt = STMT_VINFO_RELATED_STMT (stmt_info);
> if (!STMT_VINFO_RELEVANT_P (stmt_info)
> @@ -5430,6 +5432,7 @@ vect_analyze_stmt (gimple stmt, bool *ne
> }
> }
> else if (STMT_VINFO_IN_PATTERN_P (stmt_info)
> + && node == NULL
> && pattern_stmt
> && (STMT_VINFO_RELEVANT_P (vinfo_for_stmt (pattern_stmt))
> || STMT_VINFO_LIVE_P (vinfo_for_stmt (pattern_stmt))))
> @@ -5446,6 +5449,7 @@ vect_analyze_stmt (gimple stmt, bool *ne
> }
>
> if (is_pattern_stmt_p (stmt_info)
> + && node == NULL
> && (pattern_def_seq = STMT_VINFO_PATTERN_DEF_SEQ (stmt_info)))
> {
> gimple_stmt_iterator si;
> --- gcc/testsuite/gcc.c-torture/execute/pr51581-1.c.jj 2012-06-13 21:15:46.675726547 +0200
> +++ gcc/testsuite/gcc.c-torture/execute/pr51581-1.c 2012-06-13 21:06:55.000000000 +0200
> @@ -0,0 +1,157 @@
> +/* PR tree-optimization/51581 */
> +
> +extern void abort (void);
> +
> +#define N 4096
> +int a[N], c[N];
> +unsigned int b[N], d[N];
> +
> +__attribute__((noinline, noclone)) void
> +f1 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + c[i] = a[i] / 3;
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f2 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + d[i] = b[i] / 3;
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f3 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + c[i] = a[i] / 18;
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f4 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + d[i] = b[i] / 18;
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f5 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + c[i] = a[i] / 19;
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f6 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + d[i] = b[i] / 19;
> +}
> +
> +#if __SIZEOF_INT__ == 4 && __SIZEOF_LONG_LONG__ == 8
> +__attribute__((noinline, noclone)) void
> +f7 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + c[i] = (int) ((unsigned long long) (a[i] * 0x55555556LL) >> 32) - (a[i] >> 31);
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f8 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + d[i] = ((unsigned int) ((b[i] * 0xaaaaaaabULL) >> 32) >> 1);
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f9 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + c[i] = (((int) ((unsigned long long) (a[i] * 0x38e38e39LL) >> 32)) >> 2) - (a[i] >> 31);
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f10 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + d[i] = (unsigned int) ((b[i] * 0x38e38e39ULL) >> 32) >> 2;
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f11 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + c[i] = (((int) ((unsigned long long) (a[i] * 0x6bca1af3LL) >> 32)) >> 3) - (a[i] >> 31);
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f12 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + {
> + unsigned int tmp = (b[i] * 0xaf286bcbULL) >> 32;
> + d[i] = (((b[i] - tmp) >> 1) + tmp) >> 4;
> + }
> +}
> +#endif
> +
> +int
> +main ()
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + {
> + asm ("");
> + a[i] = i - N / 2;
> + b[i] = i;
> + }
> + a[0] = -__INT_MAX__ - 1;
> + a[1] = -__INT_MAX__;
> + a[N - 1] = __INT_MAX__;
> + b[N - 1] = ~0;
> + f1 ();
> + f2 ();
> + for (i = 0; i < N; i++)
> + if (c[i] != a[i] / 3 || d[i] != b[i] / 3)
> + abort ();
> + f3 ();
> + f4 ();
> + for (i = 0; i < N; i++)
> + if (c[i] != a[i] / 18 || d[i] != b[i] / 18)
> + abort ();
> + f5 ();
> + f6 ();
> + for (i = 0; i < N; i++)
> + if (c[i] != a[i] / 19 || d[i] != b[i] / 19)
> + abort ();
> +#if __SIZEOF_INT__ == 4 && __SIZEOF_LONG_LONG__ == 8
> + f7 ();
> + f8 ();
> + for (i = 0; i < N; i++)
> + if (c[i] != a[i] / 3 || d[i] != b[i] / 3)
> + abort ();
> + f9 ();
> + f10 ();
> + for (i = 0; i < N; i++)
> + if (c[i] != a[i] / 18 || d[i] != b[i] / 18)
> + abort ();
> + f11 ();
> + f12 ();
> + for (i = 0; i < N; i++)
> + if (c[i] != a[i] / 19 || d[i] != b[i] / 19)
> + abort ();
> +#endif
> + return 0;
> +}
> --- gcc/testsuite/gcc.c-torture/execute/pr51581-2.c.jj 2012-06-13 21:15:49.390710861 +0200
> +++ gcc/testsuite/gcc.c-torture/execute/pr51581-2.c 2012-06-13 21:12:49.000000000 +0200
> @@ -0,0 +1,173 @@
> +/* PR tree-optimization/51581 */
> +
> +extern void abort (void);
> +
> +#define N 4096
> +int a[N], c[N];
> +unsigned int b[N], d[N];
> +
> +__attribute__((noinline, noclone)) void
> +f1 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + c[i] = a[i] % 3;
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f2 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + d[i] = b[i] % 3;
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f3 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + c[i] = a[i] % 18;
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f4 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + d[i] = b[i] % 18;
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f5 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + c[i] = a[i] % 19;
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f6 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + d[i] = b[i] % 19;
> +}
> +
> +#if __SIZEOF_INT__ == 4 && __SIZEOF_LONG_LONG__ == 8
> +__attribute__((noinline, noclone)) void
> +f7 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + {
> + int x = (int) ((unsigned long long) (a[i] * 0x55555556LL) >> 32) - (a[i] >> 31);
> + c[i] = a[i] - x * 3;
> + }
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f8 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + {
> + unsigned int x = ((unsigned int) ((b[i] * 0xaaaaaaabULL) >> 32) >> 1);
> + d[i] = b[i] - x * 3;
> + }
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f9 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + {
> + int x = (((int) ((unsigned long long) (a[i] * 0x38e38e39LL) >> 32)) >> 2) - (a[i] >> 31);
> + c[i] = a[i] - x * 18;
> + }
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f10 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + {
> + unsigned int x = (unsigned int) ((b[i] * 0x38e38e39ULL) >> 32) >> 2;
> + d[i] = b[i] - x * 18;
> + }
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f11 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + {
> + int x = (((int) ((unsigned long long) (a[i] * 0x6bca1af3LL) >> 32)) >> 3) - (a[i] >> 31);
> + c[i] = a[i] - x * 19;
> + }
> +}
> +
> +__attribute__((noinline, noclone)) void
> +f12 (void)
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + {
> + unsigned int tmp = (b[i] * 0xaf286bcbULL) >> 32;
> + unsigned int x = (((b[i] - tmp) >> 1) + tmp) >> 4;
> + d[i] = b[i] - x * 19;
> + }
> +}
> +#endif
> +
> +int
> +main ()
> +{
> + int i;
> + for (i = 0; i < N; i++)
> + {
> + asm ("");
> + a[i] = i - N / 2;
> + b[i] = i;
> + }
> + a[0] = -__INT_MAX__ - 1;
> + a[1] = -__INT_MAX__;
> + a[N - 1] = __INT_MAX__;
> + b[N - 1] = ~0;
> + f1 ();
> + f2 ();
> + for (i = 0; i < N; i++)
> + if (c[i] != a[i] % 3 || d[i] != b[i] % 3)
> + abort ();
> + f3 ();
> + f4 ();
> + for (i = 0; i < N; i++)
> + if (c[i] != a[i] % 18 || d[i] != b[i] % 18)
> + abort ();
> + f5 ();
> + f6 ();
> + for (i = 0; i < N; i++)
> + if (c[i] != a[i] % 19 || d[i] != b[i] % 19)
> + abort ();
> +#if __SIZEOF_INT__ == 4 && __SIZEOF_LONG_LONG__ == 8
> + f7 ();
> + f8 ();
> + for (i = 0; i < N; i++)
> + if (c[i] != a[i] % 3 || d[i] != b[i] % 3)
> + abort ();
> + f9 ();
> + f10 ();
> + for (i = 0; i < N; i++)
> + if (c[i] != a[i] % 18 || d[i] != b[i] % 18)
> + abort ();
> + f11 ();
> + f12 ();
> + for (i = 0; i < N; i++)
> + if (c[i] != a[i] % 19 || d[i] != b[i] % 19)
> + abort ();
> +#endif
> + return 0;
> +}
> --- gcc/testsuite/gcc.dg/vect/pr51581-1.c.jj 2012-06-14 15:13:10.901637903 +0200
> +++ gcc/testsuite/gcc.dg/vect/pr51581-1.c 2012-06-14 15:22:07.358447332 +0200
> @@ -0,0 +1,18 @@
> +/* PR tree-optimization/51581 */
> +
> +#include "tree-vect.h"
> +
> +#define main main1
> +#include "../../gcc.c-torture/execute/pr51581-1.c"
> +#undef main
> +
> +int
> +main ()
> +{
> + int i;
> + check_vect ();
> + asm ("");
> + return main1 ();
> +}
> +
> +/* { dg-final { cleanup-tree-dump "vect" } } */
> --- gcc/testsuite/gcc.dg/vect/pr51581-2.c.jj 2012-06-14 15:13:10.000000000 +0200
> +++ gcc/testsuite/gcc.dg/vect/pr51581-2.c 2012-06-14 15:22:15.511398956 +0200
> @@ -0,0 +1,18 @@
> +/* PR tree-optimization/51581 */
> +
> +#include "tree-vect.h"
> +
> +#define main main1
> +#include "../../gcc.c-torture/execute/pr51581-2.c"
> +#undef main
> +
> +int
> +main ()
> +{
> + int i;
> + check_vect ();
> + asm ("");
> + return main1 ();
> +}
> +
> +/* { dg-final { cleanup-tree-dump "vect" } } */
> --- gcc/testsuite/gcc.dg/vect/pr51581-3.c.jj 2012-06-14 15:12:01.027052255 +0200
> +++ gcc/testsuite/gcc.dg/vect/pr51581-3.c 2012-06-14 15:11:44.000000000 +0200
> @@ -0,0 +1,118 @@
> +/* PR tree-optimization/51581 */
> +
> +#include "tree-vect.h"
> +
> +int a[8], b[8];
> +unsigned int c[8], d[8];
> +
> +void
> +f1 (void)
> +{
> + a[0] = b[0] / 8;
> + a[1] = b[1] / 8;
> + a[2] = b[2] / 8;
> + a[3] = b[3] / 8;
> + a[4] = b[4] / 8;
> + a[5] = b[5] / 8;
> + a[6] = b[6] / 8;
> + a[7] = b[7] / 8;
> +}
> +
> +void
> +f2 (void)
> +{
> + c[0] = d[0] / 3;
> + c[1] = d[1] / 3;
> + c[2] = d[2] / 3;
> + c[3] = d[3] / 3;
> + c[4] = d[4] / 3;
> + c[5] = d[5] / 3;
> + c[6] = d[6] / 3;
> + c[7] = d[7] / 3;
> +}
> +
> +void
> +f3 (void)
> +{
> + a[0] = b[0] / 8;
> + a[1] = b[1] / 4;
> + a[2] = b[2] / 8;
> + a[3] = b[3] / 4;
> + a[4] = b[4] / 8;
> + a[5] = b[5] / 4;
> + a[6] = b[6] / 8;
> + a[7] = b[7] / 4;
> +}
> +
> +void
> +f4 (void)
> +{
> + c[0] = d[0] / 3;
> + c[1] = d[1] / 5;
> + c[2] = d[2] / 3;
> + c[3] = d[3] / 5;
> + c[4] = d[4] / 3;
> + c[5] = d[5] / 5;
> + c[6] = d[6] / 3;
> + c[7] = d[7] / 5;
> +}
> +
> +void
> +f5 (void)
> +{
> + a[0] = b[0] / 14;
> + a[1] = b[1] / 15;
> + a[2] = b[2] / 14;
> + a[3] = b[3] / 15;
> + a[4] = b[4] / 14;
> + a[5] = b[5] / 15;
> + a[6] = b[6] / 14;
> + a[7] = b[7] / 15;
> +}
> +
> +void
> +f6 (void)
> +{
> + c[0] = d[0] / 6;
> + c[1] = d[1] / 5;
> + c[2] = d[2] / 6;
> + c[3] = d[3] / 5;
> + c[4] = d[4] / 6;
> + c[5] = d[5] / 5;
> + c[6] = d[6] / 13;
> + c[7] = d[7] / 5;
> +}
> +
> +int
> +main ()
> +{
> + int i;
> + check_vect ();
> + asm ("");
> + for (i = 0; i < 8; i++)
> + {
> + asm ("");
> + b[i] = i - 4;
> + d[i] = i - 4;
> + }
> + f1 ();
> + f2 ();
> + for (i = 0; i < 8; i++)
> + if (a[i] != b[i] / 8 || c[i] != d[i] / 3)
> + abort ();
> + f3 ();
> + f4 ();
> + for (i = 0; i < 8; i+= 2)
> + if (a[i] != b[i] / 8 || a[i + 1] != b[i + 1] / 4
> + || c[i] != d[i] / 3 || c[i + 1] != d[i + 1] / 5)
> + abort ();
> + f5 ();
> + f6 ();
> + for (i = 0; i < 8; i+= 2)
> + if (a[i] != b[i] / 14 || a[i + 1] != b[i + 1] / 15
> + || c[i] != d[i] / (i == 6 ? 13 : 6) || c[i + 1] != d[i + 1] / 5)
> + abort ();
> + return 0;
> +}
> +
> +/* { dg-final { cleanup-tree-dump "vect" } } */
> --- gcc/testsuite/gcc.target/i386/avx-pr51581-1.c.jj 2012-06-14 15:15:33.126792308 +0200
> +++ gcc/testsuite/gcc.target/i386/avx-pr51581-1.c 2012-06-14 15:21:45.810576497 +0200
> @@ -0,0 +1,23 @@
> +/* PR tree-optimization/51581 */
> +/* { dg-do run } */
> +/* { dg-options "-O2 -ftree-vectorize -mavx -fno-vect-cost-model" } */
> +/* { dg-require-effective-target avx } */
> +
> +#ifndef CHECK_H
> +#define CHECK_H "avx-check.h"
> +#endif
> +#ifndef TEST
> +#define TEST avx_test
> +#endif
> +
> +#define main main1
> +#include "../../gcc.c-torture/execute/pr51581-1.c"
> +#undef main
> +
> +#include CHECK_H
> +
> +static void
> +TEST (void)
> +{
> + main1 ();
> +}
> --- gcc/testsuite/gcc.target/i386/avx-pr51581-2.c.jj 2012-06-14 15:18:18.939808388 +0200
> +++ gcc/testsuite/gcc.target/i386/avx-pr51581-2.c 2012-06-14 15:21:52.088538756 +0200
> @@ -0,0 +1,23 @@
> +/* PR tree-optimization/51581 */
> +/* { dg-do run } */
> +/* { dg-options "-O2 -ftree-vectorize -mavx -fno-vect-cost-model" } */
> +/* { dg-require-effective-target avx } */
> +
> +#ifndef CHECK_H
> +#define CHECK_H "avx-check.h"
> +#endif
> +#ifndef TEST
> +#define TEST avx_test
> +#endif
> +
> +#define main main1
> +#include "../../gcc.c-torture/execute/pr51581-2.c"
> +#undef main
> +
> +#include CHECK_H
> +
> +static void
> +TEST (void)
> +{
> + main1 ();
> +}
> --- gcc/testsuite/gcc.target/i386/avx2-pr51581-1.c.jj 2012-06-14 15:18:33.394722358 +0200
> +++ gcc/testsuite/gcc.target/i386/avx2-pr51581-1.c 2012-06-14 15:19:31.813373637 +0200
> @@ -0,0 +1,9 @@
> +/* PR tree-optimization/51581 */
> +/* { dg-do run } */
> +/* { dg-options "-O2 -ftree-vectorize -mavx2 -fno-vect-cost-model" } */
> +/* { dg-require-effective-target avx2 } */
> +
> +#define CHECK_H "avx2-check.h"
> +#define TEST avx2_test
> +
> +#include "avx-pr51581-1.c"
> --- gcc/testsuite/gcc.target/i386/avx2-pr51581-2.c.jj 2012-06-14 15:19:48.027280349 +0200
> +++ gcc/testsuite/gcc.target/i386/avx2-pr51581-2.c 2012-06-14 15:19:56.621229017 +0200
> @@ -0,0 +1,9 @@
> +/* PR tree-optimization/51581 */
> +/* { dg-do run } */
> +/* { dg-options "-O2 -ftree-vectorize -mavx2 -fno-vect-cost-model" } */
> +/* { dg-require-effective-target avx2 } */
> +
> +#define CHECK_H "avx2-check.h"
> +#define TEST avx2_test
> +
> +#include "avx-pr51581-2.c"
> --- gcc/testsuite/gcc.dg/vect/slp-26.c.jj 2010-11-26 18:38:56.000000000 +0100
> +++ gcc/testsuite/gcc.dg/vect/slp-26.c 2012-06-14 22:53:08.000000000 +0200
> @@ -10,7 +10,7 @@ main1 ()
> {
> int i;
> unsigned short in[N*8] = {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63};
> - unsigned short out[N*8], a[N], b[N] = {3,6,9,12,15,18,21,24};
> + unsigned short out[N*8], a[N], b[N] = {3,0x8031,0x7fff,0x8032,0xffff,0,0x8030,0x8000};
>
> /* Partial SLP is not supported. */
> for (i = 0; i < N; i++)
> @@ -20,7 +20,7 @@ main1 ()
> out[i*4 + 2] = in[i*4 + 2];
> out[i*4 + 3] = in[i*4 + 3];
>
> - a[i] = b[i] / 3;
> + a[i] = b[i] / 0x8031;
> }
>
> /* check results: */
> @@ -30,7 +30,7 @@ main1 ()
> || out[i*4 + 1] != in[i*4 + 1]
> || out[i*4 + 2] != in[i*4 + 2]
> || out[i*4 + 3] != in[i*4 + 3]
> - || a[i] != b[i] / 3)
> + || a[i] != b[i] / 0x8031)
> abort ();
> }
>
>
> Jakub
More information about the Gcc-patches
mailing list