[PATCH] bswap: Fix up recent vector CONSTRUCTOR optimization [PR98568]
Jakub Jelinek
jakub@redhat.com
Thu Jan 7 08:31:47 GMT 2021
Hi!
As the testcase shows, bswap can match even byte-swapping or indentity
from low part of some wider SSA_NAME.
For bswap replacement other than for vector CONSTRUCTOR the code has been
using NOP_EXPR casts if the types weren't compatible, but for vectors
we need to use VIEW_CONVERT_EXPR. The problem with the latter is that
we require that it has the same size, which isn't guaranteed, so this patch
in those cases first adds a narrowing NOP_EXPR cast and only afterwards
does a VIEW_CONVERT_EXPR.
Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux, ok for
trunk?
2021-01-06 Jakub Jelinek <jakub@redhat.com>
PR tree-optimization/98568
* gimple-ssa-store-merging.c (bswap_view_convert): New function.
(bswap_replace): Use it.
* g++.dg/torture/pr98568.C: New test.
--- gcc/gimple-ssa-store-merging.c.jj 2021-01-05 16:15:58.965337667 +0100
+++ gcc/gimple-ssa-store-merging.c 2021-01-06 19:20:37.225578412 +0100
@@ -978,6 +978,25 @@ public:
}; // class pass_optimize_bswap
+/* Helper function for bswap_replace. Build VIEW_CONVERT_EXPR from
+ VAL to TYPE. If VAL has different type size, emit a NOP_EXPR cast
+ first. */
+
+static tree
+bswap_view_convert (gimple_stmt_iterator *gsi, tree type, tree val)
+{
+ gcc_assert (INTEGRAL_TYPE_P (TREE_TYPE (val)));
+ if (TYPE_SIZE (type) != TYPE_SIZE (TREE_TYPE (val)))
+ {
+ HOST_WIDE_INT prec = TREE_INT_CST_LOW (TYPE_SIZE (type));
+ tree itype = build_nonstandard_integer_type (prec, 1);
+ gimple *g = gimple_build_assign (make_ssa_name (itype), NOP_EXPR, val);
+ gsi_insert_before (gsi, g, GSI_SAME_STMT);
+ val = gimple_assign_lhs (g);
+ }
+ return build1 (VIEW_CONVERT_EXPR, type, val);
+}
+
/* Perform the bswap optimization: replace the expression computed in the rhs
of gsi_stmt (GSI) (or if NULL add instead of replace) by an equivalent
bswap, load or load + bswap expression.
@@ -1100,7 +1119,7 @@ bswap_replace (gimple_stmt_iterator gsi,
gimple_set_vuse (load_stmt, n->vuse);
gsi_insert_before (&gsi, load_stmt, GSI_SAME_STMT);
if (conv_code == VIEW_CONVERT_EXPR)
- val_tmp = build1 (VIEW_CONVERT_EXPR, TREE_TYPE (tgt), val_tmp);
+ val_tmp = bswap_view_convert (&gsi, TREE_TYPE (tgt), val_tmp);
gimple_assign_set_rhs_with_ops (&gsi, conv_code, val_tmp);
update_stmt (cur_stmt);
}
@@ -1144,7 +1163,7 @@ bswap_replace (gimple_stmt_iterator gsi,
if (!is_gimple_val (src))
return NULL_TREE;
if (conv_code == VIEW_CONVERT_EXPR)
- src = build1 (VIEW_CONVERT_EXPR, TREE_TYPE (tgt), src);
+ src = bswap_view_convert (&gsi, TREE_TYPE (tgt), src);
g = gimple_build_assign (tgt, conv_code, src);
}
else if (cur_stmt)
@@ -1227,7 +1246,7 @@ bswap_replace (gimple_stmt_iterator gsi,
tmp = make_temp_ssa_name (bswap_type, NULL, "bswapdst");
tree atmp = tmp;
if (conv_code == VIEW_CONVERT_EXPR)
- atmp = build1 (VIEW_CONVERT_EXPR, TREE_TYPE (tgt), tmp);
+ atmp = bswap_view_convert (&gsi, TREE_TYPE (tgt), tmp);
convert_stmt = gimple_build_assign (tgt, conv_code, atmp);
gsi_insert_after (&gsi, convert_stmt, GSI_SAME_STMT);
}
--- gcc/testsuite/g++.dg/torture/pr98568.C.jj 2021-01-06 19:19:25.940377456 +0100
+++ gcc/testsuite/g++.dg/torture/pr98568.C 2021-01-06 19:19:16.608482074 +0100
@@ -0,0 +1,37 @@
+// PR tree-optimization/98568
+// { dg-do compile }
+
+char a[2];
+char b[4];
+
+void
+foo (int x)
+{
+ a[1] = x >> 8;
+ a[0] = x;
+}
+
+void
+bar (long long x)
+{
+ b[3] = x >> 24;
+ b[2] = x >> 16;
+ b[1] = x >> 8;
+ b[0] = x;
+}
+
+void
+baz (int x)
+{
+ a[0] = x >> 8;
+ a[1] = x;
+}
+
+void
+qux (long long x)
+{
+ b[0] = x >> 24;
+ b[1] = x >> 16;
+ b[2] = x >> 8;
+ b[3] = x;
+}
Jakub
More information about the Gcc-patches
mailing list