diff options
Diffstat (limited to '0097-Improve-non-loop-disambiguation.patch')
| -rw-r--r-- | 0097-Improve-non-loop-disambiguation.patch | 101 |
1 files changed, 101 insertions, 0 deletions
diff --git a/0097-Improve-non-loop-disambiguation.patch b/0097-Improve-non-loop-disambiguation.patch new file mode 100644 index 0000000..ae609d2 --- /dev/null +++ b/0097-Improve-non-loop-disambiguation.patch @@ -0,0 +1,101 @@ +From 6de2e0d400cbe46da482a672810c37b1832c408c Mon Sep 17 00:00:00 2001 +From: =?UTF-8?q?=E9=83=91=E6=99=A8=E5=8D=89?= <zhengchenhui1@huawei.com> +Date: Thu, 25 Jul 2024 19:45:43 +0800 +Subject: [PATCH] Improve non-loop disambiguation + +This optimization is brought from https://gcc.gnu.org/git/gitweb.cgi?p=gcc.git;h=038b077689bb5310386b04d40a2cea234f01e6aa. + +When dr_may_alias_p is called without a loop context, it tries +to use the tree-affine interface to calculate the difference +between the two addresses and use that difference to check whether +the gap between the accesses is known at compile time. However, as the +example in the PR shows, this doesn't expand SSA_NAMEs and so can easily +be defeated by things like reassociation. + +One fix would have been to use aff_combination_expand to expand the +SSA_NAMEs, but we'd then need some way of maintaining the associated +cache. This patch instead reuses the innermost_loop_behavior fields +(which exist even when no loop context is provided). + +It might still be useful to do the aff_combination_expand thing too, +if an example turns out to need it. +--- + gcc/common.opt | 4 ++++ + gcc/testsuite/gcc.dg/vect/bb-slp-pr106019.c | 16 +++++++++++++++ + gcc/tree-data-ref.cc | 22 +++++++++++++++++++++ + 3 files changed, 42 insertions(+) + create mode 100644 gcc/testsuite/gcc.dg/vect/bb-slp-pr106019.c + +diff --git a/gcc/common.opt b/gcc/common.opt +index b18f0b944..75bf9c9c1 100644 +--- a/gcc/common.opt ++++ b/gcc/common.opt +@@ -3217,6 +3217,10 @@ ftree-loop-vectorize + Common Var(flag_tree_loop_vectorize) Optimization EnabledBy(ftree-vectorize) + Enable loop vectorization on trees. + ++falias-analysis-expand-ssa ++Common Var(flag_alias_analysis_expand_ssa) Init(0) ++Enable expanded SSA name analysis during alias analysis. ++ + ftree-slp-vectorize + Common Var(flag_tree_slp_vectorize) Optimization EnabledBy(ftree-vectorize) + Enable basic block vectorization (SLP) on trees. +diff --git a/gcc/testsuite/gcc.dg/vect/bb-slp-pr106019.c b/gcc/testsuite/gcc.dg/vect/bb-slp-pr106019.c +new file mode 100644 +index 000000000..5ff8a8a62 +--- /dev/null ++++ b/gcc/testsuite/gcc.dg/vect/bb-slp-pr106019.c +@@ -0,0 +1,16 @@ ++/* { dg-do compile } */ ++/* { dg-additional-options "-falias-analysis-expand-ssa" } */ ++ ++void f(double *p, long i) ++{ ++ p[i+0] += 1; ++ p[i+1] += 1; ++} ++void g(double *p, long i) ++{ ++ double *q = p + i; ++ q[0] += 1; ++ q[1] += 1; ++} ++ ++/* { dg-final { scan-tree-dump-not "can't determine dependence" slp2 } } */ +diff --git a/gcc/tree-data-ref.cc b/gcc/tree-data-ref.cc +index e6ae9e847..a05073c51 100644 +--- a/gcc/tree-data-ref.cc ++++ b/gcc/tree-data-ref.cc +@@ -2993,6 +2993,28 @@ dr_may_alias_p (const struct data_reference *a, const struct data_reference *b, + disambiguation. */ + if (!loop_nest) + { ++ if (flag_alias_analysis_expand_ssa) ++ { ++ tree tree_size_a = TYPE_SIZE_UNIT (TREE_TYPE (DR_REF (a))); ++ tree tree_size_b = TYPE_SIZE_UNIT (TREE_TYPE (DR_REF (b))); ++ ++ if (DR_BASE_ADDRESS (a) ++ && DR_BASE_ADDRESS (b) ++ && operand_equal_p (DR_BASE_ADDRESS (a), DR_BASE_ADDRESS (b)) ++ && operand_equal_p (DR_OFFSET (a), DR_OFFSET (b)) ++ && poly_int_tree_p (tree_size_a) ++ && poly_int_tree_p (tree_size_b) ++ && !ranges_maybe_overlap_p (wi::to_widest (DR_INIT (a)), ++ wi::to_widest (tree_size_a), ++ wi::to_widest (DR_INIT (b)), ++ wi::to_widest (tree_size_b))) ++ { ++ gcc_assert (integer_zerop (DR_STEP (a)) ++ && integer_zerop (DR_STEP (b))); ++ return false; ++ } ++ } ++ + aff_tree off1, off2; + poly_widest_int size1, size2; + get_inner_reference_aff (DR_REF (a), &off1, &size1); +-- +2.33.0 + |