ADPATCH rewrite

Different patches can now easily refer to each other. Handy when there's not enough room for a simple patch. Updated & expanded Miles and LucasArts patches.
pdewacht · Apr 29, 2019 · 64a3654 · 64a3654
1 parent f4a4081
commit 64a3654
Show file tree

Hide file tree

Showing 16 changed files with 885 additions and 466 deletions.
diff --git a/adpatch/.gitignore b/adpatch/.gitignore
@@ -1,5 +1,6 @@
 adpatch
 adpatch.exe
 patterns.rl
-patch.c
+patch.inc
+scan.c
 *.o
diff --git a/adpatch/adpatch.h b/adpatch/adpatch.h
@@ -0,0 +1,41 @@
+#define MAX_MATCHES 8
+#define MAX_RELOCATIONS 2
+#define MAX_EXPORTS 2
+
+enum relocation_type {
+  R_386_32,
+  R_386_PC32,
+  R_386_16,
+  R_386_PC16
+};
+
+struct relocation {
+  long offset;
+  int symbol;
+  enum relocation_type type;
+};
+
+struct export {
+  int symbol;
+  int offset;
+};
+
+struct patch {
+  const char *name;
+  const char *code;
+  int code_len;
+  struct relocation rel[MAX_RELOCATIONS];
+  struct export export[MAX_EXPORTS];
+};
+
+struct match {
+  const struct patch *patch;
+  long offset;
+  int len;
+};
+
+
+int apply_patches(FILE *in, FILE *out, int lpt_port);
+int scan_and_copy(FILE *in, FILE *out, struct match *matches);
+
+extern const struct patch patch_data[];
diff --git a/adpatch/build.sh b/adpatch/build.sh
@@ -1,14 +1,15 @@
 #!/bin/sh
 set -e
 
-VERSION=20181030
+VERSION=20190429
 
 CC='wcc -bt=dos -zq -oxhs'
 DEFS="-dVERSION=$VERSION"
 
 set -x
-python3 genpat.py > patterns.rl
-ragel -s -G2 patch.rl
+python3 genpat.py
+ragel -s -G2 scan.rl
 $CC $DEFS main.c
 $CC $DEFS patch.c
-wlink name adpatch system dos file main,patch option quiet
+$CC $DEFS scan.c
+wlink name adpatch system dos file main,patch,scan option quiet
diff --git a/adpatch/buildgcc.sh b/adpatch/buildgcc.sh
@@ -4,6 +4,6 @@ set -e
 eval $(grep ^VERSION= build.sh)
 DEFS="-DVERSION=$VERSION"
 
-python3 genpat.py > patterns.rl
-ragel -s -G2 patch.rl
-cc -Wall -O $DEFS -o adpatch main.c patch.c
+python3 genpat.py
+ragel -s -G2 scan.rl
+cc -Wall $DEFS -o adpatch -fsanitize=undefined,address main.c patch.c scan.c
diff --git a/adpatch/clean.sh b/adpatch/clean.sh
@@ -1,2 +1,2 @@
 #!/bin/bash
-rm -f adpatch adpatch.exe main.o patch.c patch.o patterns.rl
+rm -f adpatch adpatch.exe main.o patch.o patch.inc scan.c scan.o patterns.rl
diff --git a/adpatch/elf.py b/adpatch/elf.py
@@ -0,0 +1,105 @@
+from collections import namedtuple
+import struct
+
+
+ElfHeader = namedtuple(
+    'ElfHeader',
+    'e_ident e_type e_machine e_version e_entry e_phoff e_shoff e_flags '
+    'e_ehsize e_phentsize e_phnum e_shentsize e_shnum e_shstrndx')
+
+
+SectionHeader = namedtuple(
+    'SectionHeader',
+    'sh_name sh_type sh_flags sh_addr sh_offset sh_size sh_link sh_info '
+    'sh_addralign sh_entsize')
+
+
+SymbolTableEntry = namedtuple(
+    'SymbolTableEntry',
+    'st_name st_value st_size st_info st_other st_shndx')
+
+
+RelocationEntry = namedtuple(
+    'RelocationEntry',
+    'r_offset r_type r_sym')
+
+
+def parse_elf_header(bs, offset=0):
+    return ElfHeader(*struct.unpack_from('<16s 2H 5I 6H', bs, offset))
+
+
+def parse_section_header(bs, offset):
+    return SectionHeader(*struct.unpack_from('<10I', bs, offset))
+
+
+def parse_symbol_table_entry(bs, offset):
+    return SymbolTableEntry(*struct.unpack_from('<3I 2B H', bs, offset))
+
+
+def parse_relocation_entry(bs, offset):
+    return RelocationEntry(*struct.unpack(
+        '<IBI', bs[offset:offset+8] + b'\x00'))
+
+
+def asciiz(bs, offset):
+    return bs[offset:bs.index(b'\x00', offset)].decode('ascii')
+
+
+RELOCATION_TYPES = {
+    1: 'R_386_32',
+    2: 'R_386_PC32',
+    20: 'R_386_16',
+    21: 'R_386_PC16',
+}
+
+
+def parse(bs):
+    elf = parse_elf_header(bs)
+    section_list = [
+        parse_section_header(bs, elf.e_shoff + i * elf.e_shentsize)
+        for i in range(elf.e_shnum)
+    ]
+
+    def section_name(n):
+        return asciiz(bs, section_list[elf.e_shstrndx].sh_offset + n)
+
+    sections = {section_name(s.sh_name): s for s in section_list}
+
+    s = sections['.symtab']
+    symbols = [
+        parse_symbol_table_entry(bs, s.sh_offset + i)
+        for i in range(0, s.sh_size, s.sh_entsize)
+    ]
+
+    def symbol_name(sym):
+        return asciiz(bs, sections['.strtab'].sh_offset + sym.st_name)
+
+    exports = {
+        symbol_name(sym): sym.st_value
+        for sym in symbols
+        if sym.st_info & 16 != 0
+        if section_name(sym.st_shndx) == '.text'
+    }
+
+    if '.rel.text' not in sections:
+        relocations = []
+    else:
+        s = sections['.rel.text']
+        relocations = [
+            parse_relocation_entry(bs, s.sh_offset + i)
+            for i in range(0, s.sh_size, s.sh_entsize)
+        ]
+
+        relocations = [
+            RelocationEntry(
+                r_offset=rel.r_offset,
+                r_type=RELOCATION_TYPES[rel.r_type],
+                r_sym=symbol_name(symbols[rel.r_sym])
+            )
+            for rel in relocations
+        ]
+
+    s = sections['.text']
+    text = bs[s.sh_offset:s.sh_offset+s.sh_size]
+
+    return (text, exports, relocations)