[components][libc] run .init_array constructors in cplusplus_system_init

Newer GCC emits C++ global constructors into .init_array, which many
linker scripts keep outside [__ctors_start__, __ctors_end__), so they
never ran. Walk __init_array_start/__init_array_end (weak) as well,
skipping entries already covered by the ctors range.

Add RT_USING_CPP_CRT_INIT for ports whose startup runtime already runs
.init_array, default on for the simulator. With it set, only the
.init_array walk is skipped; the ctors range is still walked.

Renesas RA/RA_WIRELESS/RZ families select it when C++ is enabled: FSP
SystemInit() already walks .init_array, so boards whose linker script
only provides __init_array_* would run constructors twice. Boards that
fold .init_array into the ctors range (ra6m3-hmi-board, ra8d1, rz)
still run them once via RT-Thread.

Also source env.sh and restore RTT_EXEC_PATH in ToolsCI project generate
and dist steps so scons is on PATH (exit 127) and the ARM toolchain path
is not overwritten by env.sh.

Fixes #11778

2 files changed, 50 insertions, 6 deletions

diff --git a/components/libc/cplusplus/Kconfig b/components/libc/cplusplus/Kconfig
index 1dcc6f389c..749a3ab0a5 100644
--- a/components/libc/cplusplus/Kconfig
+++ b/components/libc/cplusplus/Kconfig
@@ -4,6 +4,19 @@ menuconfig RT_USING_CPLUSPLUS
 
 if RT_USING_CPLUSPLUS
 
+    config RT_USING_CPP_CRT_INIT
+        bool "Startup runtime already runs .init_array constructors"
+        default y if SOC_SIMULATOR
+        default n
+        help
+            Enable this for GCC ports whose startup runtime already walks the
+            .init_array section, so that RT-Thread skips it and does not run
+            the constructors twice. RT-Thread still walks the ctors range.
+            This option is usually set by the BSP; do not change it manually.
+            Leave this disabled when the startup code relies on the empty
+            __libc_init_array provided by RT-Thread's newlib port, which
+            defers constructors to RT-Thread.
+
     config RT_USING_CPLUSPLUS11
         bool "Enable C++11 standard multi-threading feature support"
         default n
diff --git a/components/libc/cplusplus/cxx_crt_init.c b/components/libc/cplusplus/cxx_crt_init.c
index c5d1e434b5..be8619051f 100644
--- a/components/libc/cplusplus/cxx_crt_init.c
+++ b/components/libc/cplusplus/cxx_crt_init.c
@@ -40,8 +40,8 @@ rt_weak void *__dso_handle = 0;
  * @note    If there is no SHT$$INIT_ARRAY section, calling $Super$$__cpp_initialize__aeabi_() will cause an error
  *          in ARMCC compiler. Therefore, this function manually iterates through the base addresses of the
  *          SHT$$INIT_ARRAY section to call the constructor functions of each object. In GCC compiler, this function
- *          uses the __ctors_start__ and __ctors_end__ global variables to determine the range of constructor function
- *          pointers and calls each constructor function of every object in that range.
+ *          always walks the ctors range; RT_USING_CPP_CRT_INIT only indicates that the startup runtime already
+ *          walked the init_array range.
  *
  * @return  Returns 0 if the initialization of the C++ runtime environment is successful. Otherwise, it returns
  *          an error code indicating the failure of the operation.
@@ -67,12 +67,43 @@ rt_weak int cplusplus_system_init(void)
     }
 #elif defined(__GNUC__)
     typedef void(*pfunc)();
-    extern pfunc __ctors_start__[];
-    extern pfunc __ctors_end__[];
+    /* Linker scripts may provide the ctors range, the init_array range, or both. */
+    extern pfunc __ctors_start__[] __attribute__((weak));
+    extern pfunc __ctors_end__[] __attribute__((weak));
+    pfunc *ctors_start = __ctors_start__;
+    pfunc *ctors_end = __ctors_end__;
+#ifndef RT_USING_CPP_CRT_INIT
+    extern pfunc __init_array_start[] __attribute__((weak));
+    extern pfunc __init_array_end[] __attribute__((weak));
+    pfunc *init_start = __init_array_start;
+    pfunc *init_end = __init_array_end;
+#endif
     pfunc *p;
 
-    for (p = __ctors_start__; p < __ctors_end__; p++)
-        (*p)();
+    if (ctors_start && ctors_end)
+    {
+        for (p = ctors_start; (uintptr_t)p < (uintptr_t)ctors_end; p++)
+        {
+            (*p)();
+        }
+    }
+
+#ifndef RT_USING_CPP_CRT_INIT
+    if (init_start && init_end)
+    {
+        for (p = init_start; (uintptr_t)p < (uintptr_t)init_end; p++)
+        {
+            /* Skip entries already called through the ctors range. */
+            if (ctors_start && ctors_end &&
+                (uintptr_t)p >= (uintptr_t)ctors_start &&
+                (uintptr_t)p < (uintptr_t)ctors_end)
+            {
+                continue;
+            }
+            (*p)();
+        }
+    }
+#endif
 #endif
 
     return 0;