From 7921a90e334117206c6bb78bad57e07fb242214c Mon Sep 17 00:00:00 2001 From: Richard Biener Date: Wed, 17 Jul 2019 10:26:25 +0000 Subject: [PATCH] re PR tree-optimization/91178 (Infinite recursion in split_constant_offset in slp after r260289) 2019-07-17 Richard Biener PR tree-optimization/91178 * tree-vect-stmts.c (get_group_load_store_type): For SLP loads with a gap larger than the vector size always use VMAT_STRIDED_SLP. (vectorizable_load): For VMAT_STRIDED_SLP with a permutation avoid loading vectors that are only contained in the gap and thus are not needed. * gcc.dg/torture/pr91178.c: New testcase. From-SVN: r273549 --- gcc/ChangeLog | 10 ++++++++++ gcc/testsuite/ChangeLog | 5 +++++ gcc/testsuite/gcc.dg/torture/pr91178.c | 11 +++++++++++ gcc/tree-vect-stmts.c | 22 ++++++++++++++++++++-- 4 files changed, 46 insertions(+), 2 deletions(-) create mode 100644 gcc/testsuite/gcc.dg/torture/pr91178.c diff --git a/gcc/ChangeLog b/gcc/ChangeLog index 877de1982b8..b15db2d089e 100644 --- a/gcc/ChangeLog +++ b/gcc/ChangeLog @@ -1,3 +1,13 @@ +2019-07-17 Richard Biener + + PR tree-optimization/91178 + * tree-vect-stmts.c (get_group_load_store_type): For SLP + loads with a gap larger than the vector size always use + VMAT_STRIDED_SLP. + (vectorizable_load): For VMAT_STRIDED_SLP with a permutation + avoid loading vectors that are only contained in the gap + and thus are not needed. + 2019-07-17 Richard Biener PR tree-optimization/91180 diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog index f9ee4825f17..2d7a6bfc654 100644 --- a/gcc/testsuite/ChangeLog +++ b/gcc/testsuite/ChangeLog @@ -1,3 +1,8 @@ +2019-07-17 Richard Biener + + PR tree-optimization/91178 + * gcc.dg/torture/pr91178.c: New testcase. + 2019-07-17 Richard Biener PR tree-optimization/91180 diff --git a/gcc/testsuite/gcc.dg/torture/pr91178.c b/gcc/testsuite/gcc.dg/torture/pr91178.c new file mode 100644 index 00000000000..b7a2dbe9e7f --- /dev/null +++ b/gcc/testsuite/gcc.dg/torture/pr91178.c @@ -0,0 +1,11 @@ +/* { dg-do compile } */ + +int a; +extern int f[10][91125]; +int b[50]; +void c() +{ + for (int d = 6; d <= a; d++) + for (int e = 16; e <= 24; e++) + b[e] -= f[d][d]; +} diff --git a/gcc/tree-vect-stmts.c b/gcc/tree-vect-stmts.c index 601a6f55fbf..5d05e108ede 100644 --- a/gcc/tree-vect-stmts.c +++ b/gcc/tree-vect-stmts.c @@ -2267,6 +2267,14 @@ get_group_load_store_type (stmt_vec_info stmt_info, tree vectype, bool slp, / vect_get_scalar_dr_size (first_dr_info))) overrun_p = false; + /* If the gap at the end of the group exceeds a whole vector + in size use the strided SLP code which can skip code-generation + for the gap. */ + if (vls_type == VLS_LOAD && known_gt (gap, nunits)) + *memory_access_type = VMAT_STRIDED_SLP; + else + *memory_access_type = VMAT_CONTIGUOUS; + /* If the gap splits the vector in half and the target can do half-vector operations avoid the epilogue peeling by simply loading half of the vector only. Usually @@ -2274,7 +2282,8 @@ get_group_load_store_type (stmt_vec_info stmt_info, tree vectype, bool slp, dr_alignment_support alignment_support_scheme; scalar_mode elmode = SCALAR_TYPE_MODE (TREE_TYPE (vectype)); machine_mode vmode; - if (overrun_p + if (*memory_access_type == VMAT_CONTIGUOUS + && overrun_p && !masked_p && (((alignment_support_scheme = vect_supportable_dr_alignment (first_dr_info, false))) @@ -2297,7 +2306,6 @@ get_group_load_store_type (stmt_vec_info stmt_info, tree vectype, bool slp, "Peeling for outer loop is not supported\n"); return false; } - *memory_access_type = VMAT_CONTIGUOUS; } } else @@ -8732,6 +8740,7 @@ vectorizable_load (stmt_vec_info stmt_info, gimple_stmt_iterator *gsi, /* Checked by get_load_store_type. */ unsigned int const_nunits = nunits.to_constant (); unsigned HOST_WIDE_INT cst_offset = 0; + unsigned int group_gap = 0; gcc_assert (!LOOP_VINFO_FULLY_MASKED_P (loop_vinfo)); gcc_assert (!nested_in_vect_loop); @@ -8749,6 +8758,7 @@ vectorizable_load (stmt_vec_info stmt_info, gimple_stmt_iterator *gsi, if (slp && grouped_load) { group_size = DR_GROUP_SIZE (first_stmt_info); + group_gap = DR_GROUP_GAP (first_stmt_info); ref_type = get_group_alias_ptr_type (first_stmt_info); } else @@ -8892,6 +8902,14 @@ vectorizable_load (stmt_vec_info stmt_info, gimple_stmt_iterator *gsi, if (nloads > 1) vec_alloc (v, nloads); stmt_vec_info new_stmt_info = NULL; + if (slp && slp_perm + && (group_el % group_size) > group_size - group_gap + && (group_el % group_size) + nloads * lnel < group_size) + { + dr_chain.quick_push (NULL_TREE); + group_el += nloads * lnel; + continue; + } for (i = 0; i < nloads; i++) { tree this_off = build_int_cst (TREE_TYPE (alias_off), -- 2.30.2