aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--gcc/ChangeLog6
-rw-r--r--gcc/testsuite/ChangeLog5
-rw-r--r--gcc/testsuite/g++.dg/torture/pr44900.C76
-rw-r--r--gcc/tree-sra.c12
4 files changed, 93 insertions, 6 deletions
diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 4b70451..517ff74 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,9 @@
+2010-07-21 Martin Jambor <mjambor@suse.cz>
+
+ PR tree-optimization/44900
+ * tree-sra.c (load_assign_lhs_subreplacements): Updated comments.
+ (sra_modify_assign): Move gsi to the next statmenent unconditionally.
+
2010-07-21 Bernd Schmidt <bernds@codesourcery.com>
PR middle-end/44738
diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog
index 56aba5f..a6555a8 100644
--- a/gcc/testsuite/ChangeLog
+++ b/gcc/testsuite/ChangeLog
@@ -1,3 +1,8 @@
+2010-07-21 Martin Jambor <mjambor@suse.cz>
+
+ PR tree-optimization/44900
+ * g++.dg/torture/pr44900.C: New test.
+
2010-07-21 Bernd Schmidt <bernds@codesourcery.com>
PR middle-end/44738
diff --git a/gcc/testsuite/g++.dg/torture/pr44900.C b/gcc/testsuite/g++.dg/torture/pr44900.C
new file mode 100644
index 0000000..5c0efcb
--- /dev/null
+++ b/gcc/testsuite/g++.dg/torture/pr44900.C
@@ -0,0 +1,76 @@
+/* { dg-do run { target i?86-*-* x86_64-*-* } } */
+/* { dg-options "-msse" } */
+/* { dg-require-effective-target sse } */
+
+typedef float __m128 __attribute__ ((__vector_size__ (16), __may_alias__));
+typedef float __v4sf __attribute__ ((__vector_size__ (16)));
+
+extern __inline __m128 __attribute__((__gnu_inline__, __always_inline__,
+__artificial__))
+_mm_set_ps (const float __Z, const float __Y, const float __X, const float __W)
+{
+ return __extension__ (__m128)(__v4sf){ __W, __X, __Y, __Z };
+}
+
+struct vec
+{
+ union {
+ __m128 v;
+ float e[4];
+ };
+
+ static const vec & zero()
+ {
+ static const vec v = _mm_set_ps(0, 0, 0, 0);
+ return v;
+ }
+
+ vec() {}
+ vec(const __m128 & a) : v(a) {}
+
+ operator const __m128&() const { return v; }
+};
+
+struct vec2
+{
+ vec _v1;
+ vec _v2;
+
+ vec2() {}
+ vec2(const vec & a, const vec & b) : _v1(a), _v2(b) {}
+
+ static vec2 load(const float * a)
+ {
+ return vec2(
+ __builtin_ia32_loadups(&a[0]),
+ __builtin_ia32_loadups(&a[4]));
+ }
+
+ const vec & v1() const { return _v1; }
+ const vec & v2() const { return _v2; }
+};
+
+extern "C" void abort(void);
+
+
+inline bool operator==(const vec & a, const vec & b)
+{ return 0xf == __builtin_ia32_movmskps(__builtin_ia32_cmpeqps(a, b)); }
+
+int main( int argc, char * argv[] )
+{
+ __attribute__((aligned(16))) float data[] =
+ { 16, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5 };
+
+ float * p = &data[2];
+ vec2 a;
+
+ a = vec2::load(p);
+
+ vec v1 = a.v1();
+ vec v2 = a.v2();
+
+ if (v2.e[3] != 7.0)
+ abort();
+
+ return 0;
+}
diff --git a/gcc/tree-sra.c b/gcc/tree-sra.c
index e4971d2..9fd6d2c 100644
--- a/gcc/tree-sra.c
+++ b/gcc/tree-sra.c
@@ -2416,9 +2416,11 @@ handle_unscalarized_data_in_subtree (struct access *top_racc, tree lhs,
(sub)tree. If that is not possible, refresh the TOP_RACC base aggregate and
load the accesses from it. LEFT_OFFSET is the offset of the left whole
subtree being copied, RIGHT_OFFSET is the same thing for the right subtree.
- GSI is stmt iterator used for statement insertions. *REFRESHED is true iff
- the rhs top aggregate has already been refreshed by contents of its scalar
- reductions and is set to true if this function has to do it. */
+ NEW_GSI is stmt iterator used for statement insertions after the original
+ assignment, OLD_GSI is used to insert statements before the assignment.
+ *REFRESHED keeps the information whether we have needed to refresh
+ replacements of the LHS and from which side of the assignments this takes
+ place. */
static void
load_assign_lhs_subreplacements (struct access *lacc, struct access *top_racc,
@@ -2722,9 +2724,7 @@ sra_modify_assign (gimple *stmt, gimple_stmt_iterator *gsi)
&orig_gsi, gsi, &refreshed, lhs);
if (refreshed != SRA_UDH_RIGHT)
{
- if (*stmt == gsi_stmt (*gsi))
- gsi_next (gsi);
-
+ gsi_next (gsi);
unlink_stmt_vdef (*stmt);
gsi_remove (&orig_gsi, true);
sra_stats.deleted++;