py/persistentcode: Remove remaining native qstr linking support.

Support for architecture-specific qstr linking was removed in
d4d53e9e114d779523e382c4ea38f0398e880aae, where native code was changed to
access qstr values via qstr_table.  The only remaining use for the special
qstr link table in persistentcode.c is to support native module written in
C, linked via mpy_ld.py.  But native modules can also use the standard
module-level qstr_table (and obj_table) which was introduced in the .mpy
file reworking in f2040bfc7ee033e48acef9f289790f3b4e6b74e5.

This commit removes the remaining native qstr liking support in
persistentcode.c's load_raw_code function, and adds two new relocation
options for constants.qstr_table and constants.obj_table.  mpy_ld.py is
updated to use these relocations options instead of the native qstr link
table.

Signed-off-by: Damien George <damien@micropython.org>

1 files changed, 41 insertions, 50 deletions

diff --git a/tools/mpy_ld.py b/tools/mpy_ld.py
index b8652462d4..9fda49826f 100755
--- a/tools/mpy_ld.py
+++ b/tools/mpy_ld.py
@@ -36,6 +36,8 @@ import makeqstrdata as qstrutil
 
 # MicroPython constants
 MPY_VERSION = 6
+MP_CODE_BYTECODE = 2
+MP_CODE_NATIVE_VIPER = 4
 MP_NATIVE_ARCH_X86 = 1
 MP_NATIVE_ARCH_X64 = 2
 MP_NATIVE_ARCH_ARMV6M = 4
@@ -44,8 +46,7 @@ MP_NATIVE_ARCH_ARMV7EMSP = 7
 MP_NATIVE_ARCH_ARMV7EMDP = 8
 MP_NATIVE_ARCH_XTENSA = 9
 MP_NATIVE_ARCH_XTENSAWIN = 10
-MP_CODE_BYTECODE = 2
-MP_CODE_NATIVE_VIPER = 4
+MP_PERSISTENT_OBJ_STR = 5
 MP_SCOPE_FLAG_VIPERRELOC = 0x10
 MP_SCOPE_FLAG_VIPERRODATA = 0x20
 MP_SCOPE_FLAG_VIPERBSS = 0x40
@@ -110,21 +111,20 @@ def asm_jump_xtensa(entry):
 
 
 class ArchData:
-    def __init__(self, name, mpy_feature, qstr_entry_size, word_size, arch_got, asm_jump):
+    def __init__(self, name, mpy_feature, word_size, arch_got, asm_jump, *, separate_rodata=False):
         self.name = name
         self.mpy_feature = mpy_feature
-        self.qstr_entry_size = qstr_entry_size
+        self.qstr_entry_size = 2
         self.word_size = word_size
         self.arch_got = arch_got
         self.asm_jump = asm_jump
-        self.separate_rodata = name == "EM_XTENSA" and qstr_entry_size == 4
+        self.separate_rodata = separate_rodata
 
 
 ARCH_DATA = {
     "x86": ArchData(
         "EM_386",
         MP_NATIVE_ARCH_X86 << 2,
-        2,
         4,
         (R_386_PC32, R_386_GOT32, R_386_GOT32X),
         asm_jump_x86,
@@ -132,7 +132,6 @@ ARCH_DATA = {
     "x64": ArchData(
         "EM_X86_64",
         MP_NATIVE_ARCH_X64 << 2,
-        2,
         8,
         (R_X86_64_GOTPCREL, R_X86_64_REX_GOTPCRELX),
         asm_jump_x86,
@@ -140,7 +139,6 @@ ARCH_DATA = {
     "armv6m": ArchData(
         "EM_ARM",
         MP_NATIVE_ARCH_ARMV6M << 2,
-        2,
         4,
         (R_ARM_GOT_BREL,),
         asm_jump_thumb,
@@ -148,7 +146,6 @@ ARCH_DATA = {
     "armv7m": ArchData(
         "EM_ARM",
         MP_NATIVE_ARCH_ARMV7M << 2,
-        2,
         4,
         (R_ARM_GOT_BREL,),
         asm_jump_thumb2,
@@ -156,7 +153,6 @@ ARCH_DATA = {
     "armv7emsp": ArchData(
         "EM_ARM",
         MP_NATIVE_ARCH_ARMV7EMSP << 2,
-        2,
         4,
         (R_ARM_GOT_BREL,),
         asm_jump_thumb2,
@@ -164,7 +160,6 @@ ARCH_DATA = {
     "armv7emdp": ArchData(
         "EM_ARM",
         MP_NATIVE_ARCH_ARMV7EMDP << 2,
-        2,
         4,
         (R_ARM_GOT_BREL,),
         asm_jump_thumb2,
@@ -172,7 +167,6 @@ ARCH_DATA = {
     "xtensa": ArchData(
         "EM_XTENSA",
         MP_NATIVE_ARCH_XTENSA << 2,
-        2,
         4,
         (R_XTENSA_32, R_XTENSA_PLT),
         asm_jump_xtensa,
@@ -181,9 +175,9 @@ ARCH_DATA = {
         "EM_XTENSA",
         MP_NATIVE_ARCH_XTENSAWIN << 2,
         4,
-        4,
         (R_XTENSA_32, R_XTENSA_PLT),
         asm_jump_xtensa,
+        separate_rodata=True,
     ),
 }
 
@@ -435,8 +429,8 @@ def populate_got(env):
 
     # Create a relocation for each GOT entry
     for got_entry in got_list:
-        if got_entry.name == "mp_fun_table":
-            dest = "mp_fun_table"
+        if got_entry.name in ("mp_native_qstr_table", "mp_native_obj_table", "mp_fun_table"):
+            dest = got_entry.name
         elif got_entry.name.startswith("mp_fun_table+0x"):
             dest = int(got_entry.name.split("+")[1], 16) // env.arch.word_size
         elif got_entry.sec_name.startswith(".text"):
@@ -754,19 +748,19 @@ def link_objects(env, native_qstr_vals_len, native_qstr_objs_len):
         env.lit_section = Section("LIT", bytearray(lit_size), env.arch.word_size)
         env.sections.insert(1, env.lit_section)
 
-    # Create section to contain mp_native_qstr_val_table
-    env.qstr_val_section = Section(
-        ".text.QSTR_VAL",
+    # Create section to contain mp_native_qstr_table
+    env.qstr_table_section = Section(
+        ".external.qstr_table",
         bytearray(native_qstr_vals_len * env.arch.qstr_entry_size),
         env.arch.qstr_entry_size,
     )
-    env.sections.append(env.qstr_val_section)
 
-    # Create section to contain mp_native_qstr_obj_table
-    env.qstr_obj_section = Section(
-        ".text.QSTR_OBJ", bytearray(native_qstr_objs_len * env.arch.word_size), env.arch.word_size
+    # Create section to contain mp_native_obj_table
+    env.obj_table_section = Section(
+        ".external.obj_table",
+        bytearray(native_qstr_objs_len * env.arch.word_size),
+        env.arch.word_size,
     )
-    env.sections.append(env.qstr_obj_section)
 
     # Resolve unknown symbols
     mp_fun_table_sec = Section(".external.mp_fun_table", b"", 0)
@@ -796,10 +790,10 @@ def link_objects(env, native_qstr_vals_len, native_qstr_objs_len):
             pass
         elif sym.name == "mp_fun_table":
             sym.section = Section(".external", b"", 0)
-        elif sym.name == "mp_native_qstr_val_table":
-            sym.section = env.qstr_val_section
-        elif sym.name == "mp_native_qstr_obj_table":
-            sym.section = env.qstr_obj_section
+        elif sym.name == "mp_native_qstr_table":
+            sym.section = env.qstr_table_section
+        elif sym.name == "mp_native_obj_table":
+            sym.section = env.obj_table_section
         elif sym.name in env.known_syms:
             sym.resolved = env.known_syms[sym.name]
         else:
@@ -923,13 +917,21 @@ def build_mpy(env, entry_offset, fmpy, native_qstr_vals, native_qstr_objs):
     out.write_bytes(bytearray([ord("M"), MPY_VERSION, env.arch.mpy_feature, MP_SMALL_INT_BITS]))
 
     # MPY: n_qstr
-    out.write_uint(1)
+    out.write_uint(1 + len(native_qstr_vals))
 
     # MPY: n_obj
-    out.write_uint(0)
+    out.write_uint(len(native_qstr_objs))
 
     # MPY: qstr table
     out.write_qstr(fmpy)  # filename
+    for q in native_qstr_vals:
+        out.write_qstr(q)
+
+    # MPY: object table
+    for q in native_qstr_objs:
+        out.write_bytes(bytearray([MP_PERSISTENT_OBJ_STR]))
+        out.write_uint(len(q))
+        out.write_bytes(bytes(q, "utf8") + b"\x00")
 
     # MPY: kind/len
     out.write_uint(len(env.full_text) << 3 | (MP_CODE_NATIVE_VIPER - MP_CODE_BYTECODE))
@@ -937,17 +939,6 @@ def build_mpy(env, entry_offset, fmpy, native_qstr_vals, native_qstr_objs):
     # MPY: machine code
     out.write_bytes(env.full_text)
 
-    # MPY: n_qstr_link (assumes little endian)
-    out.write_uint(len(native_qstr_vals) + len(native_qstr_objs))
-    for q in range(len(native_qstr_vals)):
-        off = env.qstr_val_section.addr + q * env.arch.qstr_entry_size
-        out.write_uint(off << 2)
-        out.write_qstr(native_qstr_vals[q])
-    for q in range(len(native_qstr_objs)):
-        off = env.qstr_obj_section.addr + q * env.arch.word_size
-        out.write_uint(off << 2 | 3)
-        out.write_qstr(native_qstr_objs[q])
-
     # MPY: scope_flags
     scope_flags = MP_SCOPE_FLAG_VIPERRELOC
     if len(env.full_rodata):
@@ -978,10 +969,14 @@ def build_mpy(env, entry_offset, fmpy, native_qstr_vals, native_qstr_objs):
                 kind = 0
         elif isinstance(kind, str) and kind.startswith(".bss"):
             kind = bss_const_table_idx
-        elif kind == "mp_fun_table":
+        elif kind == "mp_native_qstr_table":
             kind = 6
+        elif kind == "mp_native_obj_table":
+            kind = 7
+        elif kind == "mp_fun_table":
+            kind = 8
         else:
-            kind = 7 + kind
+            kind = 9 + kind
         assert addr % env.arch.word_size == 0, addr
         offset = addr // env.arch.word_size
         if kind == prev_kind and base == prev_base and offset == prev_offset + 1:
@@ -1023,18 +1018,14 @@ def do_preprocess(args):
         for i, q in enumerate(static_qstrs):
             print("#define %s (%u)" % (q, i + 1), file=f)
         for i, q in enumerate(sorted(qstr_vals)):
-            print("#define %s (mp_native_qstr_val_table[%d])" % (q, i), file=f)
+            print("#define %s (mp_native_qstr_table[%d])" % (q, i + 1), file=f)
         for i, q in enumerate(sorted(qstr_objs)):
             print(
-                "#define MP_OBJ_NEW_QSTR_%s ((mp_obj_t)mp_native_qstr_obj_table[%d])" % (q, i),
+                "#define MP_OBJ_NEW_QSTR_%s ((mp_obj_t)mp_native_obj_table[%d])" % (q, i),
                 file=f,
             )
-        if args.arch == "xtensawin":
-            qstr_type = "uint32_t"  # esp32 can only read 32-bit values from IRAM
-        else:
-            qstr_type = "uint16_t"
-        print("extern const {} mp_native_qstr_val_table[];".format(qstr_type), file=f)
-        print("extern const mp_uint_t mp_native_qstr_obj_table[];", file=f)
+        print("extern const uint16_t mp_native_qstr_table[];", file=f)
+        print("extern const mp_uint_t mp_native_obj_table[];", file=f)
 
 
 def do_link(args):