From bff948aa337807260344c83ac9079d6386410094 Mon Sep 17 00:00:00 2001 From: Jakub Jelinek Date: Tue, 28 Jan 2020 08:46:23 +0100 Subject: [PATCH] i386: Fix ix86_fold_builtin shift folding [PR93418] The following testcase is miscompiled, because the variable shift left operand, { -1, -1, -1, -1 } is represented as a VECTOR_CST with VECTOR_CST_NPATTERNS 1 and VECTOR_CST_NELTS_PER_PATTERN 1, so when we call builder.new_unary_operation, builder.encoded_nelts () will be just 1 and thus we encode the resulting vector as if all the elements were the same. For non-masked is_vshift, we could perhaps call builder.new_binary_operation (TREE_TYPE (args[0]), args[0], args[1], false), but then there are masked shifts, for non-is_vshift we could perhaps call it too but with args[2] instead of args[1], but there is no builder.new_ternary_operation. All this stuff is primarily for aarch64 anyway, on x86 we don't have any variable length vectors, and it is not a big deal to compute all elements and just let builder.finalize () find the most efficient VECTOR_CST representation of the vector. So, instead of doing too much, this just keeps using new_unary_operation only if only one VECTOR_CST is involved (i.e. non-masked shift by constant) and for the rest just compute all elts. 2020-01-28 Jakub Jelinek PR target/93418 * config/i386/i386.c (ix86_fold_builtin) : If mask is not -1 or is_vshift is true, use new_vector with number of elts npatterns rather than new_unary_operation. * gcc.target/i386/avx2-pr93418.c: New test. --- gcc/ChangeLog | 5 +++++ gcc/config/i386/i386.c | 9 +++++++-- gcc/testsuite/ChangeLog | 3 +++ gcc/testsuite/gcc.target/i386/avx2-pr93418.c | 20 ++++++++++++++++++++ 4 files changed, 35 insertions(+), 2 deletions(-) create mode 100644 gcc/testsuite/gcc.target/i386/avx2-pr93418.c diff --git a/gcc/ChangeLog b/gcc/ChangeLog index 6db98ed94d0..3489428938c 100644 --- a/gcc/ChangeLog +++ b/gcc/ChangeLog @@ -1,5 +1,10 @@ 2020-01-28 Jakub Jelinek + PR target/93418 + * config/i386/i386.c (ix86_fold_builtin) : If mask is not + -1 or is_vshift is true, use new_vector with number of elts npatterns + rather than new_unary_operation. + PR tree-optimization/93454 * gimple-fold.c (fold_array_ctor_reference): Perform elt_size.to_uhwi () just once, instead of calling it in every diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c index ffe60baa72a..ffda3e8fd21 100644 --- a/gcc/config/i386/i386.c +++ b/gcc/config/i386/i386.c @@ -17278,8 +17278,13 @@ ix86_fold_builtin (tree fndecl, int n_args, countt = build_int_cst (integer_type_node, count); } tree_vector_builder builder; - builder.new_unary_operation (TREE_TYPE (args[0]), args[0], - false); + if (mask != HOST_WIDE_INT_M1U || is_vshift) + builder.new_vector (TREE_TYPE (args[0]), + TYPE_VECTOR_SUBPARTS (TREE_TYPE (args[0])), + 1); + else + builder.new_unary_operation (TREE_TYPE (args[0]), args[0], + false); unsigned int cnt = builder.encoded_nelts (); for (unsigned int i = 0; i < cnt; ++i) { diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog index 44d8e674465..6518f7f2307 100644 --- a/gcc/testsuite/ChangeLog +++ b/gcc/testsuite/ChangeLog @@ -1,5 +1,8 @@ 2020-01-28 Jakub Jelinek + PR target/93418 + * gcc.target/i386/avx2-pr93418.c: New test. + PR tree-optimization/93454 * gcc.dg/pr93454.c: New test. diff --git a/gcc/testsuite/gcc.target/i386/avx2-pr93418.c b/gcc/testsuite/gcc.target/i386/avx2-pr93418.c new file mode 100644 index 00000000000..67ed33ddf9d --- /dev/null +++ b/gcc/testsuite/gcc.target/i386/avx2-pr93418.c @@ -0,0 +1,20 @@ +/* PR target/93418 */ +/* { dg-do compile } */ +/* { dg-options "-O2 -mavx2 -fdump-tree-optimized" } */ +/* { dg-final { scan-tree-dump-not "link_error" "optimized" } } */ + +#include + +void link_error (void); + +void +foo (void) +{ + __m128i a = _mm_set1_epi32 (0xffffffffU); + __m128i b = _mm_setr_epi32 (16, 31, -34, 3); + __m128i c = _mm_sllv_epi32 (a, b); + __v4su d = (__v4su) c; + if (d[0] != 0xffff0000U || d[1] != 0x80000000U + || d[2] != 0 || d[3] != 0xfffffff8U) + link_error (); +} -- 2.30.2