On Mon, 22 Aug 2016, Richard Biener wrote: > On Sat, Aug 20, 2016 at 1:27 AM, Patrick Palka <patr...@parcs.ath.cx> wrote: > > * For SSA_NAME: Print the ssa name's def stmt on its own line. > > > > Before: > > <ssa_name 0x7ffff6a26ca8 > > type <vector_type 0x7ffff6a09bd0 > > type <integer_type 0x7ffff688a888 unsigned int sizes-gimplified > > public unsigned SI > > size <integer_cst 0x7ffff6886ca8 constant 32> > > unit size <integer_cst 0x7ffff6886cc0 constant 4> > > align 32 symtab 0 alias set -1 canonical type 0x7ffff688a888 > > precision 32 min <integer_cst 0x7ffff6886f30 0> max <integer_cst > > 0x7ffff6886f18 4294967295> context <translation_unit_decl 0x7ffff6893348 > > D.2203> > > pointer_to_this <pointer_type 0x7ffff68aae70>> > > unsigned V8SI > > size <integer_cst 0x7ffff68a2168 constant 256> > > unit size <integer_cst 0x7ffff68a21f8 constant 32> > > align 256 symtab 0 alias set 1 canonical type 0x7ffff6a09bd0 nunits > > 8 > > pointer_to_this <pointer_type 0x7ffff6a09d20>> > > visited var <var_decl 0x7ffff6a046c0 vect__2.15>def_stmt vect__2.15_101 > > = vect__1.14_99 & { 1, 1, 1, 1, 1, 1, 1, 1 }; > > > > version 101> > > > > After: > > (gdb) print debug_tree (op0) > > <ssa_name 0x7ffff6a26ca8 > > type <vector_type 0x7ffff6a09bd0 > > type <integer_type 0x7ffff688a888 unsigned int sizes-gimplified > > public unsigned SI > > size <integer_cst 0x7ffff6886ca8 constant 32> > > unit size <integer_cst 0x7ffff6886cc0 constant 4> > > align 32 symtab 0 alias set -1 canonical type 0x7ffff688a888 > > precision 32 min <integer_cst 0x7ffff6886f30 0> max <integer_cst > > 0x7ffff6886f18 4294967295> context <translation_unit_decl 0x7ffff6893348 > > D.2203> > > pointer_to_this <pointer_type 0x7ffff68aae70>> > > unsigned V8SI > > size <integer_cst 0x7ffff68a2168 constant 256> > > unit size <integer_cst 0x7ffff68a21f8 constant 32> > > align 256 symtab 0 alias set 1 canonical type 0x7ffff6a09bd0 nunits > > 8 > > pointer_to_this <pointer_type 0x7ffff6a09d20>> > > visited var <var_decl 0x7ffff6a046c0 vect__2.15> > > def_stmt vect__2.15_101 = vect__1.14_99 & { 1, 1, 1, 1, 1, 1, 1, 1 }; > > > > version 101> > > > > * For VECTOR_CST: Coalesce the output of identical consecutive elt values. > > > > Before: > > <vector_cst 0x7ffff6a23870 > > type <vector_type 0x7ffff6a09bd0 > > type <integer_type 0x7ffff688a888 unsigned int sizes-gimplified > > public unsigned SI > > size <integer_cst 0x7ffff6886ca8 constant 32> > > unit size <integer_cst 0x7ffff6886cc0 constant 4> > > align 32 symtab 0 alias set -1 canonical type 0x7ffff688a888 > > precision 32 min <integer_cst 0x7ffff6886f30 0> max <integer_cst > > 0x7ffff6886f18 4294967295> context <translation_unit_decl 0x7ffff6893348 > > D.2203> > > pointer_to_this <pointer_type 0x7ffff68aae70>> > > unsigned V8SI > > size <integer_cst 0x7ffff68a2168 constant 256> > > unit size <integer_cst 0x7ffff68a21f8 constant 32> > > align 256 symtab 0 alias set 1 canonical type 0x7ffff6a09bd0 nunits > > 8 > > pointer_to_this <pointer_type 0x7ffff6a09d20>> > > constant > > elt0: <integer_cst 0x7ffff69ec840 type <integer_type 0x7ffff69db3f0 > > ui> constant 0> elt1: <integer_cst 0x7ffff69ec840 0> elt2: <integer_cst > > 0x7ffff69ec840 0> elt3: <integer_cst 0x7ffff69ec840 0> elt4: <integer_cst > > 0x7ffff69ec840 0> elt5: <integer_cst 0x7ffff69ec840 0> elt6: <integer_cst > > 0x7ffff69ec840 0> elt7: <integer_cst 0x7ffff69ec840 0>> > > > > After: > > (gdb) print debug_tree (op1) > > <vector_cst 0x7ffff6a23870 > > type <vector_type 0x7ffff6a09bd0 > > type <integer_type 0x7ffff688a888 unsigned int sizes-gimplified > > public unsigned SI > > size <integer_cst 0x7ffff6886ca8 constant 32> > > unit size <integer_cst 0x7ffff6886cc0 constant 4> > > align 32 symtab 0 alias set -1 canonical type 0x7ffff688a888 > > precision 32 min <integer_cst 0x7ffff6886f30 0> max <integer_cst > > 0x7ffff6886f18 4294967295> context <translation_unit_decl 0x7ffff6893348 > > D.2203> > > pointer_to_this <pointer_type 0x7ffff68aae70>> > > unsigned V8SI > > size <integer_cst 0x7ffff68a2168 constant 256> > > unit size <integer_cst 0x7ffff68a21f8 constant 32> > > align 256 symtab 0 alias set 1 canonical type 0x7ffff6a09bd0 nunits > > 8 > > pointer_to_this <pointer_type 0x7ffff6a09d20>> > > constant > > elt0...elt7: <integer_cst 0x7ffff69ec840 type <integer_type > > 0x7ffff69db3f0 ui> constant 0>> > > > > (I also tested the change on non-uniform VECTOR_CSTs.) > > > > Does this look OK to commit after bootstrap + regtesting? > > Ok - but can you please try to avoid the excessive newline after the > def_stmt: line? > I think the only way to do this currently is to inline > print_gimple_stmt and adjust it > accordingly. Or add a param to 'indent_to' whether to emit the newline.
How does this look? gcc/ChangeLog: * print-tree.c (print_node) [SSA_NAME]: When printing the node's def stmt, replace the call to print_gimple_stmt() with its inlined body and adjust it to avoid printing an unwanted newline by calling pp_flush() instead of pp_newline_and_flush() and by not setting pp_needs_newline. --- gcc/print-tree.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/gcc/print-tree.c b/gcc/print-tree.c index 2baf34a..d775491 100644 --- a/gcc/print-tree.c +++ b/gcc/print-tree.c @@ -882,7 +882,12 @@ print_node (FILE *file, const char *prefix, tree node, int indent) print_node_brief (file, "var", SSA_NAME_VAR (node), indent + 4); indent_to (file, indent + 4); fprintf (file, "def_stmt "); - print_gimple_stmt (file, SSA_NAME_DEF_STMT (node), indent + 4, 0); + { + pretty_printer buffer; + buffer.buffer->stream = file; + pp_gimple_stmt_1 (&buffer, SSA_NAME_DEF_STMT (node), indent + 4, 0); + pp_flush (&buffer); + } indent_to (file, indent + 4); fprintf (file, "version %u", SSA_NAME_VERSION (node)); -- 2.10.0.rc1.53.gfcb49dc