Add a few ps2 segments (#360)

* update spimdisasm version

* configure_disassembler_section

* lit4 and lit8

* ctor

* vtables

* black

* capitalization

* whoops

* a bit of docs on the new sections

* yeet

CHANGELOG.md

@@ -16,6 +16,13 @@
 * Fix disassembling segments that only have bss.
 * `ld_section_labels` was removed since it was a redundant list to `section_order`.
 * Give a proper error message for missing sections on `section_order` during linker script generation.
+* Allow configuring the `spimdisasm` section on extension segments (that inherit from `asm`, `data`, `rodata` or `bss` segments) before running the analisys on it.
+  * This is done by overriding the `configure_disassembler_section` method.
+* New ps2-specific segments:
+  * `lit4` and `lit8`: "Literal" sections that only contain `float`s and `double`s respectively.
+  * `ctor`: Data pointing to C++ global data initialization functions.
+  * `vtables`: C++ vtables
+* `spimdisasm` 1.23.0 or above is now required.
 
 ### 0.22.3
 

docs/Segments.md

@@ -275,6 +275,32 @@ While this kind of segment can be represented by other segment types ([`asm`](#a
 - [0x00B250, pad, nops_00B250]
 ```
 
+## PS2 exclusive segments
+
+### `lit4`
+
+`lit4` is a segment that only contains single-precision floats.
+
+splat will try to disassemble all the data from this segment as individual floats whenever possible.
+
+### `lit8`
+
+`lit8` is a segment that only contains double-precision floats.
+
+splat will try to disassemble all the data from this segment as individual doubles whenever possible.
+
+### `ctor`
+
+`ctor` is used by certain compilers (like MWCC) to store pointers to functions that initialize C++ global data objects.
+
+The disassembly of this section is tweaked to avoid confusing its data with other types of data, this is because the disassembler can sometimes get confused and disassemble a pointer as a float, string, etc.
+
+### `vtables`
+
+`vtables` is used by certain compilers (like MWCC) to store the virtual tables of C++ classes
+
+The disassembly of this section is tweaked to avoid confusing its data with other types of data, this is because the disassembler can sometimes get confused and disassemble a pointer as a float, string, etc.
+
 ## General segment options
 
 All splat's segments can be passed extra options for finer configuration. Note that those extra options require to rewrite the entry using the dictionary yaml notation instead of the list one.

pyproject.toml

@@ -20,7 +20,7 @@ dependencies = [
 
 [project.optional-dependencies]
 mips = [
-    "spimdisasm>=1.21.0,<2.0.0", # This value should be keep in sync with the version listed on disassembler/spimdisasm_disassembler.py
+    "spimdisasm>=1.23.0,<2.0.0", # This value should be keep in sync with the version listed on disassembler/spimdisasm_disassembler.py
     "rabbitizer>=1.8.0,<2.0.0",
     "pygfxd",
     "n64img>=0.1.4",

requirements.txt

@@ -3,8 +3,8 @@ pylibyaml
 tqdm
 intervaltree
 colorama
-# This value should be keep in sync with the version listed on disassembler/spimdisasm_disassembler.py
-spimdisasm>=1.21.0
+# This value should be keep in sync with the version listed on disassembler/spimdisasm_disassembler.py and pyproject.toml
+spimdisasm>=1.23.0
 rabbitizer>=1.8.0
 pygfxd
 n64img>=0.1.4

src/splat/disassembler/spimdisasm_disassembler.py

@@ -6,8 +6,8 @@
 
 
 class SpimdisasmDisassembler(disassembler.Disassembler):
-    # This value should be kept in sync with the version listed on requirements.txt
-    SPIMDISASM_MIN = (1, 21, 0)
+    # This value should be kept in sync with the version listed on requirements.txt and pyproject.toml
+    SPIMDISASM_MIN = (1, 23, 0)
 
     def configure(self):
         # Configure spimdisasm

src/splat/segtypes/common/bss.py

@@ -2,7 +2,7 @@
 
 from .data import CommonSegData
 
-from ...disassembler.disassembler_section import make_bss_section
+from ...disassembler.disassembler_section import DisassemblerSection, make_bss_section
 
 # If `options.opts.ld_bss_is_noload` is False, then this segment behaves like a `CommonSegData`
 
@@ -23,6 +23,13 @@ def is_noload() -> bool:
             return False
         return True
 
+    def configure_disassembler_section(
+        self, disassembler_section: DisassemblerSection
+    ) -> None:
+        "Allows to configure the section before running the analysis on it"
+
+        pass
+
     def disassemble_data(self, rom_bytes: bytes):
         if not options.opts.ld_bss_is_noload:
             super().disassemble_data(rom_bytes)
@@ -64,6 +71,8 @@ def disassemble_data(self, rom_bytes: bytes):
 
         assert self.spim_section is not None
 
+        self.configure_disassembler_section(self.spim_section)
+
         self.spim_section.analyze()
         self.spim_section.set_comment_offset(self.rom_start)
 

src/splat/segtypes/common/codesubsegment.py

@@ -38,14 +38,29 @@ def __init__(self, *args, **kwargs):
             else None
         )
 
+        self.is_hasm = False
+
     @property
     def needs_symbols(self) -> bool:
         return True
 
     def get_linker_section(self) -> str:
         return ".text"
 
+    def configure_disassembler_section(
+        self, disassembler_section: DisassemblerSection
+    ) -> None:
+        "Allows to configure the section before running the analysis on it"
+
+        section = disassembler_section.get_section()
+
+        section.isHandwritten = self.is_hasm
+        section.instrCat = self.instr_category
+        section.detectRedundantFunctionEnd = self.detect_redundant_function_end
+
     def scan_code(self, rom_bytes, is_hasm=False):
+        self.is_hasm = is_hasm
+
         if not isinstance(self.rom_start, int):
             log.error(
                 f"Segment '{self.name}' (type '{self.type}') requires a rom_start. Got '{self.rom_start}'"
@@ -75,11 +90,7 @@ def scan_code(self, rom_bytes, is_hasm=False):
 
         assert self.spim_section is not None
 
-        self.spim_section.get_section().isHandwritten = is_hasm
-        self.spim_section.get_section().instrCat = self.instr_category
-        self.spim_section.get_section().detectRedundantFunctionEnd = (
-            self.detect_redundant_function_end
-        )
+        self.configure_disassembler_section(self.spim_section)
 
         self.spim_section.analyze()
         self.spim_section.set_comment_offset(self.rom_start)

src/splat/segtypes/common/data.py

@@ -1,12 +1,11 @@
 from pathlib import Path
 from typing import Optional
-
 from ...util import options, symbols, log
 
 from .codesubsegment import CommonSegCodeSubsegment
 from .group import CommonSegGroup
 
-from ...disassembler.disassembler_section import make_data_section
+from ...disassembler.disassembler_section import DisassemblerSection, make_data_section
 
 
 class CommonSegData(CommonSegCodeSubsegment, CommonSegGroup):
@@ -82,6 +81,22 @@ def get_linker_section(self) -> str:
     def get_linker_entries(self):
         return CommonSegCodeSubsegment.get_linker_entries(self)
 
+    def configure_disassembler_section(
+        self, disassembler_section: DisassemblerSection
+    ) -> None:
+        "Allows to configure the section before running the analysis on it"
+
+        section = disassembler_section.get_section()
+
+        # Set data string encoding
+        # First check the global configuration
+        if options.opts.data_string_encoding is not None:
+            section.stringEncoding = options.opts.data_string_encoding
+
+        # Then check the per-segment configuration in case we want to override the global one
+        if self.str_encoding is not None:
+            section.stringEncoding = self.str_encoding
+
     def disassemble_data(self, rom_bytes):
         if not isinstance(self.rom_start, int):
             log.error(
@@ -112,16 +127,7 @@ def disassemble_data(self, rom_bytes):
 
         assert self.spim_section is not None
 
-        # Set rodata string encoding
-        # First check the global configuration
-        if options.opts.data_string_encoding is not None:
-            self.spim_section.get_section().stringEncoding = (
-                options.opts.data_string_encoding
-            )
-
-        # Then check the per-segment configuration in case we want to override the global one
-        if self.str_encoding is not None:
-            self.spim_section.get_section().stringEncoding = self.str_encoding
+        self.configure_disassembler_section(self.spim_section)
 
         self.spim_section.analyze()
         self.spim_section.set_comment_offset(self.rom_start)

src/splat/segtypes/common/rodata.py

@@ -5,7 +5,10 @@
 
 from .data import CommonSegData
 
-from ...disassembler.disassembler_section import make_rodata_section
+from ...disassembler.disassembler_section import (
+    DisassemblerSection,
+    make_rodata_section,
+)
 
 
 class CommonSegRodata(CommonSegData):
@@ -41,6 +44,22 @@ def get_possible_text_subsegment_for_symbol(
             return None
         return text_segment, func
 
+    def configure_disassembler_section(
+        self, disassembler_section: DisassemblerSection
+    ) -> None:
+        "Allows to configure the section before running the analysis on it"
+
+        section = disassembler_section.get_section()
+
+        # Set rodata string encoding
+        # First check the global configuration
+        if options.opts.string_encoding is not None:
+            section.stringEncoding = options.opts.string_encoding
+
+        # Then check the per-segment configuration in case we want to override the global one
+        if self.str_encoding is not None:
+            section.stringEncoding = self.str_encoding
+
     def disassemble_data(self, rom_bytes):
         if not isinstance(self.rom_start, int):
             log.error(
@@ -71,16 +90,7 @@ def disassemble_data(self, rom_bytes):
 
         assert self.spim_section is not None
 
-        # Set rodata string encoding
-        # First check the global configuration
-        if options.opts.string_encoding is not None:
-            self.spim_section.get_section().stringEncoding = (
-                options.opts.string_encoding
-            )
-
-        # Then check the per-segment configuration in case we want to override the global one
-        if self.str_encoding is not None:
-            self.spim_section.get_section().stringEncoding = self.str_encoding
+        self.configure_disassembler_section(self.spim_section)
 
         self.spim_section.analyze()
         self.spim_section.set_comment_offset(self.rom_start)

src/splat/segtypes/ps2/ctor.py

@@ -0,0 +1,28 @@
+from typing import Optional
+
+from ..common.data import CommonSegData
+from ...disassembler.disassembler_section import DisassemblerSection
+
+
+class Ps2SegCtor(CommonSegData):
+    """Segment that contains pointers to C++ global data initialization functions"""
+
+    def get_linker_section(self) -> str:
+        return ".ctor"
+
+    def get_section_flags(self) -> Optional[str]:
+        return "a"
+
+    def configure_disassembler_section(
+        self, disassembler_section: DisassemblerSection
+    ) -> None:
+        "Allows to configure the section before running the analysis on it"
+
+        super().configure_disassembler_section(disassembler_section)
+
+        section = disassembler_section.get_section()
+
+        # We use s32 to make sure spimdisasm disassembles the data from this section as words/references to other symbols
+        section.enableStringGuessing = False
+        section.typeForOwnedSymbols = "s32"
+        section.sizeForOwnedSymbols = 4

src/splat/segtypes/ps2/lit4.py

@@ -0,0 +1,28 @@
+from typing import Optional
+
+from ..common.data import CommonSegData
+from ...disassembler.disassembler_section import DisassemblerSection
+
+
+class Ps2SegLit4(CommonSegData):
+    """Segment that only contains single-precision floats"""
+
+    def get_linker_section(self) -> str:
+        return ".lit4"
+
+    def get_section_flags(self) -> Optional[str]:
+        return "wa"
+
+    def configure_disassembler_section(
+        self, disassembler_section: DisassemblerSection
+    ) -> None:
+        "Allows to configure the section before running the analysis on it"
+
+        super().configure_disassembler_section(disassembler_section)
+
+        section = disassembler_section.get_section()
+
+        # Tell spimdisasm this section only contains floats
+        section.enableStringGuessing = False
+        section.typeForOwnedSymbols = "f32"
+        section.sizeForOwnedSymbols = 4

src/splat/segtypes/ps2/lit8.py

@@ -0,0 +1,28 @@
+from typing import Optional
+
+from ..common.data import CommonSegData
+from ...disassembler.disassembler_section import DisassemblerSection
+
+
+class Ps2SegLit8(CommonSegData):
+    """Segment that only contains double-precision floats"""
+
+    def get_linker_section(self) -> str:
+        return ".lit8"
+
+    def get_section_flags(self) -> Optional[str]:
+        return "wa"
+
+    def configure_disassembler_section(
+        self, disassembler_section: DisassemblerSection
+    ) -> None:
+        "Allows to configure the section before running the analysis on it"
+
+        super().configure_disassembler_section(disassembler_section)
+
+        section = disassembler_section.get_section()
+
+        # Tell spimdisasm this section only contains doubles
+        section.enableStringGuessing = False
+        section.typeForOwnedSymbols = "f64"
+        section.sizeForOwnedSymbols = 8

src/splat/segtypes/ps2/vtables.py

@@ -0,0 +1,27 @@
+from typing import Optional
+
+from ..common.data import CommonSegData
+from ...disassembler.disassembler_section import DisassemblerSection
+
+
+class Ps2SegVtables(CommonSegData):
+    """Segment that contains a pointer to C++ vtables"""
+
+    def get_linker_section(self) -> str:
+        return ".vtables"
+
+    def get_section_flags(self) -> Optional[str]:
+        return "a"
+
+    def configure_disassembler_section(
+        self, disassembler_section: DisassemblerSection
+    ) -> None:
+        "Allows to configure the section before running the analysis on it"
+
+        super().configure_disassembler_section(disassembler_section)
+
+        section = disassembler_section.get_section()
+
+        # We use s32 to make sure spimdisasm disassembles the data from this section as words/references to other symbols
+        section.enableStringGuessing = False
+        section.typeForOwnedSymbols = "s32"