[pull] master from php:master

.gitignore

@@ -177,6 +177,9 @@ php
 /ext/*/configure.ac
 /ext/*/run-tests.php
 
+# Generated by ./configure if libc might be musl
+/ext/gettext/tests/locale/en_US
+
 # ------------------------------------------------------------------------------
 # Generated by Windows build system
 # ------------------------------------------------------------------------------

ext/gettext/config.m4

@@ -23,6 +23,18 @@ if test "$PHP_GETTEXT" != "no"; then
     [AC_CHECK_LIB([c], [bindtextdomain], [
       GETTEXT_LIBS=
       GETTEXT_CHECK_IN_LIB=c
+
+      dnl If libintl.h is provided by libc, it's possible that libc is musl.
+      dnl The gettext family of functions under musl ignores the codeset
+      dnl suffix on directories like "en_US.UTF-8"; instead they look only
+      dnl in "en_US". To accomodate that, we symlink some test data from one
+      dnl to the other.
+      AC_MSG_NOTICE([symlinking en_US.UTF-8 messages to en_US in case you are on musl])
+      _linkdest="${srcdir%/}"/ext/gettext/tests/locale/en_US
+      AS_IF([test ! -e "${_linkdest}"],[
+        ln -s en_US.UTF-8 "${_linkdest}"
+      ])
+
       ],
       [AC_MSG_FAILURE([Unable to find required intl library for gettext.])])])
 

ext/gettext/gettext.c

@@ -180,7 +180,7 @@ PHP_FUNCTION(dcgettext)
 PHP_FUNCTION(bindtextdomain)
 {
 	zend_string *domain, *dir = NULL;
-	char *retval, dir_name[MAXPATHLEN];
+	char *retval, dir_name[MAXPATHLEN], *btd_result;
 
 	ZEND_PARSE_PARAMETERS_START(1, 2)
 		Z_PARAM_STR(domain)
@@ -191,7 +191,16 @@ PHP_FUNCTION(bindtextdomain)
 	PHP_GETTEXT_DOMAIN_LENGTH_CHECK(1, ZSTR_LEN(domain))
 
 	if (dir == NULL) {
-		RETURN_STRING(bindtextdomain(ZSTR_VAL(domain), NULL));
+		btd_result = bindtextdomain(ZSTR_VAL(domain), NULL);
+		if (btd_result == NULL) {
+			/* POSIX-compliant implementations can return
+			 * NULL if an error occured. On musl you will
+			 * also get NULL if the domain is not yet
+			 * bound, because musl has no default directory
+			 * to return in that case. */
+			 RETURN_FALSE;
+		}
+		RETURN_STRING(btd_result);
 	}
 
 	if (ZSTR_LEN(dir) != 0 && !zend_string_equals_literal(dir, "0")) {

ext/gettext/tests/bug53251.phpt

@@ -8,18 +8,28 @@ if (getenv('SKIP_REPEAT')) die('skip gettext leaks global state across requests'
 ?>
 --FILE--
 <?php
-var_dump(is_string(bindtextdomain('foo', null)));
+$results[] = is_string(bindtextdomain('foo', null));
 $dir = bindtextdomain('foo', '.');
-var_dump(bindtextdomain('foo', null) === $dir);
+$results[] = bindtextdomain('foo', null) === $dir;
+$results[] = bind_textdomain_codeset('foo', null);
+$results[] = bind_textdomain_codeset('foo', 'UTF-8');
+$results[] = bind_textdomain_codeset('foo', null);
 
-var_dump(bind_textdomain_codeset('foo', null));
-var_dump(bind_textdomain_codeset('foo', 'UTF-8'));
-var_dump(bind_textdomain_codeset('foo', null));
+$expected = [true, true, false, "UTF-8", "UTF-8"];
 
+// musl's bindtextdomain() has no default directory to return when
+// "foo" is unbound, so in the first call, you will get false instead
+// of a string.
+//
+// bind_textdomain_codeset() always returns false on musl
+// because musl only supports UTF-8. For more information:
+//
+//   * https://github.com/php/doc-en/issues/4311,
+//   * https://github.com/php/php-src/issues/17163
+//
+$expected_musl = [false, true, false, false, false];
+
+var_dump($results === $expected || $results === $expected_musl);
 ?>
 --EXPECT--
 bool(true)
-bool(true)
-bool(false)
-string(5) "UTF-8"
-string(5) "UTF-8"

ext/gettext/tests/gettext_bind_textdomain_codeset-retval.phpt

@@ -15,14 +15,22 @@ gettext
     } catch (ValueError $e) {
 	    echo $e->getMessage() . PHP_EOL;
     }
-    var_dump(bind_textdomain_codeset('messages', "UTF-8"));
+
+    // bind_textdomain_codeset() always returns false on musl
+    // because musl only supports UTF-8. For more information:
+    //
+    //   * https://github.com/php/doc-en/issues/4311,
+    //   * https://github.com/php/php-src/issues/17163
+    //
+    $result = bind_textdomain_codeset('messages', "UTF-8");
+    var_dump($result === false || $result === "UTF-8");
 
     echo "Done\n";
 ?>
 --EXPECT--
 bind_textdomain_codeset(): Argument #1 ($domain) must not be empty
 bind_textdomain_codeset(): Argument #1 ($domain) must not be empty
-string(5) "UTF-8"
+bool(true)
 Done
 --CREDITS--
 Florian Holzhauer [email protected]

ext/opcache/jit/ir/ir.c

@@ -330,6 +330,7 @@ static ir_ref ir_next_const(ir_ctx *ctx)
 
 static void ir_grow_top(ir_ctx *ctx)
 {
+	ir_ref old_insns_limit = ctx->insns_limit;
 	ir_insn *buf = ctx->ir_base - ctx->consts_limit;
 
 	if (ctx->insns_limit < 1024 * 4) {
@@ -341,6 +342,12 @@ static void ir_grow_top(ir_ctx *ctx)
 	}
 	buf = ir_mem_realloc(buf, (ctx->consts_limit + ctx->insns_limit) * sizeof(ir_insn));
 	ctx->ir_base = buf + ctx->consts_limit;
+
+	if (ctx->use_lists) {
+		ctx->use_lists = ir_mem_realloc(ctx->use_lists, ctx->insns_limit * sizeof(ir_use_list));
+		memset(ctx->use_lists + old_insns_limit, 0,
+			(ctx->insns_limit - old_insns_limit) * sizeof(ir_use_list));
+	}
 }
 
 static ir_ref ir_next_insn(ir_ctx *ctx)
@@ -1152,7 +1159,7 @@ void ir_build_def_use_lists(ir_ctx *ctx)
 	ir_ref n, i, j, *p, def;
 	ir_insn *insn;
 	uint32_t edges_count;
-	ir_use_list *lists = ir_mem_calloc(ctx->insns_count, sizeof(ir_use_list));
+	ir_use_list *lists = ir_mem_calloc(ctx->insns_limit, sizeof(ir_use_list));
 	ir_ref *edges;
 	ir_use_list *use_list;
 
@@ -1207,7 +1214,7 @@ void ir_build_def_use_lists(ir_ctx *ctx)
 	ir_ref n, i, j, *p, def;
 	ir_insn *insn;
 	size_t linked_lists_size, linked_lists_top = 0, edges_count = 0;
-	ir_use_list *lists = ir_mem_calloc(ctx->insns_count, sizeof(ir_use_list));
+	ir_use_list *lists = ir_mem_calloc(ctx->insns_limit, sizeof(ir_use_list));
 	ir_ref *edges;
 	ir_use_list *use_list;
 	ir_ref *linked_lists;

ext/opcache/jit/ir/ir.h

@@ -528,11 +528,12 @@ void ir_strtab_free(ir_strtab *strtab);
 #define IR_OPT_INLINE          (1<<16)
 #define IR_OPT_FOLDING         (1<<17)
 #define IR_OPT_CFG             (1<<18) /* merge BBs, by remove END->BEGIN nodes during CFG construction */
-#define IR_OPT_CODEGEN         (1<<19)
-#define IR_GEN_NATIVE          (1<<20)
-#define IR_GEN_CODE            (1<<21) /* C or LLVM */
+#define IR_OPT_MEM2SSA         (1<<19)
+#define IR_OPT_CODEGEN         (1<<20)
+#define IR_GEN_NATIVE          (1<<21)
+#define IR_GEN_CODE            (1<<22) /* C or LLVM */
 
-#define IR_GEN_CACHE_DEMOTE    (1<<22) /* Demote the generated code from closest CPU caches */
+#define IR_GEN_CACHE_DEMOTE    (1<<23) /* Demote the generated code from closest CPU caches */
 
 /* debug related */
 #ifdef IR_DEBUG
@@ -751,13 +752,15 @@ ir_ref ir_binding_find(const ir_ctx *ctx, ir_ref ref);
 /* Def -> Use lists */
 void ir_build_def_use_lists(ir_ctx *ctx);
 
+/* SSA Construction */
+int ir_mem2ssa(ir_ctx *ctx);
+
 /* CFG - Control Flow Graph (implementation in ir_cfg.c) */
 int ir_build_cfg(ir_ctx *ctx);
-int ir_remove_unreachable_blocks(ir_ctx *ctx);
 int ir_build_dominators_tree(ir_ctx *ctx);
 int ir_find_loops(ir_ctx *ctx);
 int ir_schedule_blocks(ir_ctx *ctx);
-void ir_build_prev_refs(ir_ctx *ctx);
+void ir_reset_cfg(ir_ctx *ctx);
 
 /* SCCP - Sparse Conditional Constant Propagation (implementation in ir_sccp.c) */
 int ir_sccp(ir_ctx *ctx);
@@ -929,7 +932,7 @@ IR_ALWAYS_INLINE void *ir_jit_compile(ir_ctx *ctx, int opt_level, size_t *size)
 		}
 
 		return ir_emit_code(ctx, size);
-	} else if (opt_level == 1 || opt_level == 2) {
+	} else if (opt_level > 0) {
 		if (!(ctx->flags & IR_OPT_FOLDING)) {
 			// IR_ASSERT(0 && "IR_OPT_FOLDING must be set in ir_init() for -O1 and -O2");
 			return NULL;
@@ -938,14 +941,29 @@ IR_ALWAYS_INLINE void *ir_jit_compile(ir_ctx *ctx, int opt_level, size_t *size)
 
 		ir_build_def_use_lists(ctx);
 
-		if (opt_level == 2
-		 && !ir_sccp(ctx)) {
-			return NULL;
+		if (ctx->flags & IR_OPT_MEM2SSA) {
+			if (!ir_build_cfg(ctx)
+			 || !ir_build_dominators_tree(ctx)
+			 || !ir_mem2ssa(ctx)) {
+				return NULL;
+			}
 		}
 
-		if (!ir_build_cfg(ctx)
-		 || !ir_build_dominators_tree(ctx)
-		 || !ir_find_loops(ctx)
+		if (opt_level > 1) {
+			ir_reset_cfg(ctx);
+			if (!ir_sccp(ctx)) {
+				return NULL;
+			}
+		}
+
+		if (!ctx->cfg_blocks) {
+			if (!ir_build_cfg(ctx)
+			 || !ir_build_dominators_tree(ctx)) {
+				return NULL;
+			}
+		}
+
+		if (!ir_find_loops(ctx)
 		 || !ir_gcm(ctx)
 		 || !ir_schedule(ctx)
 		 || !ir_match(ctx)

ext/opcache/jit/ir/ir_aarch64.dasc

@@ -1095,6 +1095,8 @@ binop_fp:
 				}
 			}
 			return IR_SKIPPED | IR_NOP;
+		case IR_NOP:
+			return IR_SKIPPED | IR_NOP;
 		default:
 			break;
 	}
@@ -5603,7 +5605,8 @@ static void ir_allocate_unique_spill_slots(ir_ctx *ctx)
 						} else if (def_flags & IR_USE_MUST_BE_IN_REG) {
 							if (insn->op == IR_VLOAD
 							 && ctx->live_intervals[ctx->vregs[i]]
-							 && ctx->live_intervals[ctx->vregs[i]]->stack_spill_pos != -1) {
+							 && ctx->live_intervals[ctx->vregs[i]]->stack_spill_pos != -1
+							 && ir_is_same_mem_var(ctx, i, ctx->ir_base[insn->op2].op3)) {
 								/* pass */
 							} else if (insn->op != IR_PARAM) {
 								reg = ir_get_free_reg(insn->type, available);
@@ -5704,7 +5707,7 @@ static void ir_allocate_unique_spill_slots(ir_ctx *ctx)
 								if (reg != IR_REG_NONE && IR_REGSET_IN(available, reg)) {
 									IR_REGSET_EXCL(available, reg);
 									ctx->regs[i][j] = reg | IR_REG_SPILL_LOAD;
-								} else if (j > 1 && input == insn->op1 && ctx->regs[i][1] != IR_REG_NONE) {
+								} else if (IR_IS_FOLDABLE_OP(insn->op) && j > 1 && input == insn->op1 && ctx->regs[i][1] != IR_REG_NONE) {
 									ctx->regs[i][j] = ctx->regs[i][1];
 								} else if (use_flags & IR_USE_MUST_BE_IN_REG) {
 									reg = ir_get_free_reg(ctx->ir_base[input].type, available);

ext/opcache/jit/ir/ir_cfg.c

@@ -8,6 +8,8 @@
 #include "ir.h"
 #include "ir_private.h"
 
+static int ir_remove_unreachable_blocks(ir_ctx *ctx);
+
 IR_ALWAYS_INLINE void _ir_add_successors(const ir_ctx *ctx, ir_ref ref, ir_worklist *worklist)
 {
 	ir_use_list *use_list = &ctx->use_lists[ref];
@@ -57,6 +59,24 @@ IR_ALWAYS_INLINE void _ir_add_predecessors(const ir_insn *insn, ir_worklist *wor
 	}
 }
 
+void ir_reset_cfg(ir_ctx *ctx)
+{
+	ctx->cfg_blocks_count = 0;
+	ctx->cfg_edges_count = 0;
+	if (ctx->cfg_blocks) {
+		ir_mem_free(ctx->cfg_blocks);
+		ctx->cfg_blocks = NULL;
+		if (ctx->cfg_edges) {
+			ir_mem_free(ctx->cfg_edges);
+			ctx->cfg_edges = NULL;
+		}
+		if (ctx->cfg_map) {
+			ir_mem_free(ctx->cfg_map);
+			ctx->cfg_map = NULL;
+		}
+	}
+}
+
 int ir_build_cfg(ir_ctx *ctx)
 {
 	ir_ref n, *p, ref, start, end;
@@ -330,11 +350,15 @@ static void ir_remove_merge_input(ir_ctx *ctx, ir_ref merge, ir_ref from)
 		}
 	}
 	i--;
+	for (j = i + 1; j <= n; j++) {
+		ir_insn_set_op(insn, j, IR_UNUSED);
+	}
 	if (i == 1) {
 		insn->op = IR_BEGIN;
 		insn->inputs_count = 1;
 		use_list = &ctx->use_lists[merge];
 		if (use_list->count > 1) {
+			n++;
 			for (k = 0, p = &ctx->use_edges[use_list->refs]; k < use_list->count; k++, p++) {
 				use = *p;
 				use_insn = &ctx->ir_base[use];
@@ -347,22 +371,24 @@ static void ir_remove_merge_input(ir_ctx *ctx, ir_ref merge, ir_ref from)
 						if (ir_bitset_in(life_inputs, j - 1)) {
 							use_insn->op1 = ir_insn_op(use_insn, j);
 						} else if (input > 0) {
-							ir_use_list_remove_all(ctx, input, use);
+							ir_use_list_remove_one(ctx, input, use);
 						}
 					}
 					use_insn->op = IR_COPY;
-					use_insn->op2 = IR_UNUSED;
-					use_insn->op3 = IR_UNUSED;
+					use_insn->inputs_count = 1;
+					for (j = 2; j <= n; j++) {
+						ir_insn_set_op(use_insn, j, IR_UNUSED);
+					}
 					ir_use_list_remove_all(ctx, merge, use);
 				}
 			}
 		}
 	} else {
 		insn->inputs_count = i;
 
-		n++;
 		use_list = &ctx->use_lists[merge];
 		if (use_list->count > 1) {
+			n++;
 			for (k = 0, p = &ctx->use_edges[use_list->refs]; k < use_list->count; k++, p++) {
 				use = *p;
 				use_insn = &ctx->ir_base[use];
@@ -378,9 +404,13 @@ static void ir_remove_merge_input(ir_ctx *ctx, ir_ref merge, ir_ref from)
 							}
 							i++;
 						} else if (input > 0) {
-							ir_use_list_remove_all(ctx, input, use);
+							ir_use_list_remove_one(ctx, input, use);
 						}
 					}
+					use_insn->inputs_count = i - 1;
+					for (j = i; j <= n; j++) {
+						ir_insn_set_op(use_insn, j, IR_UNUSED);
+					}
 				}
 			}
 		}
@@ -390,7 +420,7 @@ static void ir_remove_merge_input(ir_ctx *ctx, ir_ref merge, ir_ref from)
 }
 
 /* CFG constructed after SCCP pass doesn't have unreachable BBs, otherwise they should be removed */
-int ir_remove_unreachable_blocks(ir_ctx *ctx)
+static int ir_remove_unreachable_blocks(ir_ctx *ctx)
 {
 	uint32_t b, *p, i;
 	uint32_t unreachable_count = 0;

ext/opcache/jit/ir/ir_check.c

@@ -362,6 +362,10 @@ bool ir_check(const ir_ctx *ctx)
 									break;
 								}
 							}
+							if (count == 0 && (insn->op == IR_END || insn->op == IR_LOOP_END)) {
+								/* Dead block */
+								break;
+							}
 							fprintf(stderr, "ir_base[%d].op (%s) must have 1 successor (%d)\n",
 								i, ir_op_name[insn->op], count);
 							ok = 0;