Hi,
this is an upate of the string-merge section, it is based on the V2-STRING_CST
semantic patch series, which was finally installed yesterday.
It merges single-byte string constants with or without terminating NUL.
The patch has the same Ada and C test cases that were already in the V1 patch.
Thus there are no pre-requisite patches necessary at this time.
Bootstrapped and reg-tested on x86_64-pc-linux-gnu.
Is it OK for trunk?
Thanks
Bernd.
gcc:
2018-08-27 Bernd Edlinger <bernd.edlin...@hotmail.de>
* varasm.c (output_constant): Add new parameter merge_strings.
Make strings properly zero terminated in merge string sections.
(mergeable_string_section): Don't fail if the last char is non-zero.
(assemble_variable_contents): Handle merge string sections.
(assemble_variable): Likewise.
(assemble_constant_contents): Likewise.
(output_constant_def_contents): Likewise.
testsuite:
2018-08-27 Bernd Edlinger <bernd.edlin...@hotmail.de>
* gnat.dg/string_merge1.adb: New test.
* gnat.dg/string_merge2.adb: New test.
* gcc.dg/merge-all-constants-1.c: Adjust test.
* gcc.dg/merge-all-constants-2.c: New test.
diff -Npur gcc/varasm.c gcc/varasm.c
--- gcc/varasm.c 2018-08-26 15:02:43.157905415 +0200
+++ gcc/varasm.c 2018-08-26 17:57:26.488494866 +0200
@@ -111,7 +111,8 @@ static int compare_constant (const tree,
static void output_constant_def_contents (rtx);
static void output_addressed_constants (tree);
static unsigned HOST_WIDE_INT output_constant (tree, unsigned HOST_WIDE_INT,
- unsigned int, bool);
+ unsigned int, bool,
+ bool = false);
static void globalize_decl (tree);
static bool decl_readonly_section_1 (enum section_category);
#ifdef BSS_SECTION_ASM_OP
@@ -804,8 +805,8 @@ mergeable_string_section (tree decl ATTR
&& TREE_CODE (decl) == STRING_CST
&& TREE_CODE (TREE_TYPE (decl)) == ARRAY_TYPE
&& align <= 256
- && (len = int_size_in_bytes (TREE_TYPE (decl))) > 0
- && TREE_STRING_LENGTH (decl) >= len)
+ && (len = int_size_in_bytes (TREE_TYPE (decl))) >= 0
+ && TREE_STRING_LENGTH (decl) == len)
{
scalar_int_mode mode;
unsigned int modesize;
@@ -835,7 +836,7 @@ mergeable_string_section (tree decl ATTR
if (j == unit)
break;
}
- if (i == len - unit)
+ if (i == len - unit || (unit == 1 && i == len))
{
sprintf (name, "%s.str%d.%d", prefix,
modesize / 8, (int) (align / 8));
@@ -2117,7 +2118,7 @@ assemble_noswitch_variable (tree decl, c
static void
assemble_variable_contents (tree decl, const char *name,
- bool dont_output_data)
+ bool dont_output_data, bool merge_strings = false)
{
/* Do any machine/system dependent processing of the object. */
#ifdef ASM_DECLARE_OBJECT_NAME
@@ -2140,7 +2141,7 @@ assemble_variable_contents (tree decl, c
output_constant (DECL_INITIAL (decl),
tree_to_uhwi (DECL_SIZE_UNIT (decl)),
get_variable_align (decl),
- false);
+ false, merge_strings);
else
/* Leave space for it. */
assemble_zeros (tree_to_uhwi (DECL_SIZE_UNIT (decl)));
@@ -2316,7 +2317,9 @@ assemble_variable (tree decl, int top_le
switch_to_section (sect);
if (align > BITS_PER_UNIT)
ASM_OUTPUT_ALIGN (asm_out_file, floor_log2 (align / BITS_PER_UNIT));
- assemble_variable_contents (decl, name, dont_output_data);
+ assemble_variable_contents (decl, name, dont_output_data,
+ sect->common.flags & SECTION_MERGE
+ && sect->common.flags & SECTION_STRINGS);
if (asan_protected)
{
unsigned HOST_WIDE_INT int size
@@ -3471,7 +3474,8 @@ maybe_output_constant_def_contents (stru
constant's alignment in bits. */
static void
-assemble_constant_contents (tree exp, const char *label, unsigned int align)
+assemble_constant_contents (tree exp, const char *label, unsigned int align,
+ bool merge_strings = false)
{
HOST_WIDE_INT size;
@@ -3481,7 +3485,7 @@ assemble_constant_contents (tree exp, co
targetm.asm_out.declare_constant_name (asm_out_file, label, exp, size);
/* Output the value of EXP. */
- output_constant (exp, size, align, false);
+ output_constant (exp, size, align, false, merge_strings);
targetm.asm_out.decl_end ();
}
@@ -3522,10 +3526,13 @@ output_constant_def_contents (rtx symbol
|| (VAR_P (decl) && DECL_IN_CONSTANT_POOL (decl))
? DECL_ALIGN (decl)
: symtab_node::get (decl)->definition_alignment ());
- switch_to_section (get_constant_section (exp, align));
+ section *sect = get_constant_section (exp, align);
+ switch_to_section (sect);
if (align > BITS_PER_UNIT)
ASM_OUTPUT_ALIGN (asm_out_file, floor_log2 (align / BITS_PER_UNIT));
- assemble_constant_contents (exp, XSTR (symbol, 0), align);
+ assemble_constant_contents (exp, XSTR (symbol, 0), align,
+ sect->common.flags & SECTION_MERGE
+ && sect->common.flags & SECTION_STRINGS);
if (asan_protected)
{
HOST_WIDE_INT size = get_constant_size (exp);
@@ -4838,7 +4845,7 @@ output_constructor (tree, unsigned HOST_
static unsigned HOST_WIDE_INT
output_constant (tree exp, unsigned HOST_WIDE_INT size, unsigned int align,
- bool reverse)
+ bool reverse, bool merge_strings /* = false */)
{
enum tree_code code;
unsigned HOST_WIDE_INT thissize;
@@ -4966,8 +4973,11 @@ output_constant (tree exp, unsigned HOST
case CONSTRUCTOR:
return output_constructor (exp, size, align, reverse, NULL);
case STRING_CST:
- thissize
- = MIN ((unsigned HOST_WIDE_INT)TREE_STRING_LENGTH (exp), size);
+ thissize = (unsigned HOST_WIDE_INT)TREE_STRING_LENGTH (exp);
+ if (merge_strings
+ && (thissize == 0
+ || TREE_STRING_POINTER (exp) [thissize - 1] != '\0'))
+ thissize++;
gcc_checking_assert (check_string_literal (exp, size));
assemble_string (TREE_STRING_POINTER (exp), thissize);
break;
diff -Npur gcc/testsuite/gcc.dg/merge-all-constants-1.c gcc/testsuite/gcc.dg/merge-all-constants-1.c
--- gcc/testsuite/gcc.dg/merge-all-constants-1.c 2018-08-16 17:28:11.000000000 +0200
+++ gcc/testsuite/gcc.dg/merge-all-constants-1.c 2018-08-26 16:31:12.650271931 +0200
@@ -1,8 +1,8 @@
/* { dg-do compile } */
/* { dg-options "-w -O2 -fmerge-all-constants" } */
-const char str1[36] = "0123456789abcdefghijklmnopqrstuvwxyz";
+const char str1[36] = "\000123456789abcdefghijklmnopqrstuvwxyz";
const char str2[38] = "0123456789abcdefghijklmnopqrstuvwxyz";
-const char str3[10] = "0123456789abcdefghijklmnopqrstuvwxyz";
+const char str3[10] = "\000123456789abcdefghijklmnopqrstuvwxyz";
-/* { dg-final { scan-assembler-not "\.rodata\.str" } } */
+/* { dg-final { scan-assembler-not "\\.rodata\\.str" } } */
diff -Npur gcc/testsuite/gcc.dg/merge-all-constants-2.c gcc/testsuite/gcc.dg/merge-all-constants-2.c
--- gcc/testsuite/gcc.dg/merge-all-constants-2.c 1970-01-01 01:00:00.000000000 +0100
+++ gcc/testsuite/gcc.dg/merge-all-constants-2.c 2018-08-26 16:31:12.650271931 +0200
@@ -0,0 +1,8 @@
+/* { dg-do compile } */
+/* { dg-options "-w -O2 -fmerge-all-constants" } */
+
+const char str1[36] = "0123456789abcdefghijklmnopqrstuvwxyz";
+const char str2[37] = "0123456789abcdefghijklmnopqrstuvwxyz";
+const char str3[10] = "0123456789abcdefghijklmnopqrstuvwxyz";
+
+/* { dg-final { scan-assembler-not "\\.rodata\[\n\r\]" } } */
diff -Npur gcc/testsuite/gnat.dg/string_merge1.adb gcc/testsuite/gnat.dg/string_merge1.adb
--- gcc/testsuite/gnat.dg/string_merge1.adb 1970-01-01 01:00:00.000000000 +0100
+++ gcc/testsuite/gnat.dg/string_merge1.adb 2018-08-26 16:31:12.650271931 +0200
@@ -0,0 +1,19 @@
+-- { dg-do compile }
+-- { dg-options "-O1 -fmerge-all-constants" }
+
+procedure String_Merge1 is
+ procedure Process (X : String);
+ pragma Import (Ada, Process);
+begin
+ Process ("ABCD");
+end;
+
+-- We expect something like:
+
+-- .section .rodata.str1.1,"aMS",@progbits,1
+-- .LC1:
+-- .string "ABCD"
+
+-- { dg-final { scan-assembler-times "\\.rodata\\.str" 1 } }
+-- { dg-final { scan-assembler-times "\\.string" 1 } }
+-- { dg-final { scan-assembler-times "\"ABCD\"" 1 } }
diff -Npur gcc/testsuite/gnat.dg/string_merge2.adb gcc/testsuite/gnat.dg/string_merge2.adb
--- gcc/testsuite/gnat.dg/string_merge2.adb 1970-01-01 01:00:00.000000000 +0100
+++ gcc/testsuite/gnat.dg/string_merge2.adb 2018-08-26 16:31:12.650271931 +0200
@@ -0,0 +1,19 @@
+-- { dg-do compile }
+-- { dg-options "-O1 -fmerge-all-constants" }
+
+procedure String_Merge2 is
+ procedure Process (X : String);
+ pragma Import (Ada, Process);
+begin
+ Process ("ABCD" & Ascii.NUL);
+end;
+
+-- We expect something like:
+
+-- .section .rodata.str1.1,"aMS",@progbits,1
+-- .LC1:
+-- .string "ABCD"
+
+-- { dg-final { scan-assembler-times "\\.rodata\\.str" 1 } }
+-- { dg-final { scan-assembler-times "\\.string" 1 } }
+-- { dg-final { scan-assembler-times "\"ABCD\"" 1 } }
