Merge pull request #52 from gbaraldi/gb/symbol_offset

staticfloat · web-flow · commit 975d584dcf61 · 2025-07-10T22:00:30.000-07:00
Add symbol offset function
diff --git a/Project.toml b/Project.toml
@@ -7,14 +7,14 @@ version = "0.4.4"
 Reexport = "189a3867-3050-52da-a836-e630ba90ab69"
 StructIO = "53d494c1-5632-5724-8f4c-31dff12d585f"
 
-
 [compat]
 Reexport = "0.2, 1.0"
 StructIO = "0.3"
 julia = "1.6"
 
 [extras]
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+Mmap = "a63ad114-7e13-5084-954f-fe012c677804"
 
 [targets]
-test = ["Test"]
+test = ["Test", "Mmap"]
diff --git a/src/Abstract/Symbol.jl b/src/Abstract/Symbol.jl
@@ -10,7 +10,7 @@ export Symbols,
 
 # Export SymtabEntry API
 export SymtabEntry,
-       deref, symbol_name, symbol_value, isundef, isglobal, islocal, isweak
+       deref, symbol_name, symbol_value, isundef, isglobal, islocal, isweak, symbol_section, symbol_offset
 
 # Export SymbolRef API
 export SymbolRef,
@@ -161,6 +161,19 @@ Return the value of the given symbol
 """
 @mustimplement symbol_value(sym::SymtabEntry)
 
+"""
+    symbol_section(sym::SymtabEntry)
+Return the section that this symbol refers to
+"""
+@mustimplement symbol_section(sym::SymtabEntry)
+
+"""
+    symbol_offset(sym::SymtabEntry)
+Return the offset into the file that this symbol refers to.
+"""
+@mustimplement symbol_offset(sym::SymtabEntry)
+
+
 """
     isundef(sym::SymtabEntry)
 
diff --git a/src/COFF/COFF.jl b/src/COFF/COFF.jl
@@ -17,7 +17,7 @@ import ObjectFile: DynamicLink, DynamicLinks, RPath, ObjectHandle, Section, Sect
                    findfirst, deref, section_name, section_size,
                    section_offset, section_address, section_number, segment_name, segment_offset,
                    segment_file_size, segment_memory_size, segment_address, strtab_lookup,
-                   symbol_name, symbol_value, isundef, isglobal, islocal, isweak, symbol_number
+                   symbol_name, symbol_value, isundef, isglobal, islocal, isweak, symbol_number, symbol_section, symbol_offset
 # Load in imported C #define constants
 include("constants.jl")
 
diff --git a/src/COFF/COFFSymbol.jl b/src/COFF/COFFSymbol.jl
@@ -80,7 +80,6 @@ function symbol_name(sym::COFFSymtabEntry)
 end
 
 symbol_value(sym::COFFSymtabEntry) = sym.Value
-symbol_section(sym::COFFSymtabEntry) = sym.SectionNumber
 symbol_type(sym::COFFSymtabEntry) = sym.Type
 function isundef(sym::COFFSymtabEntry)
     sym.StorageClass in (
@@ -112,6 +111,7 @@ end
 deref(sym::COFFSymbolRef) = sym.entry
 Symbols(sym::COFFSymbolRef) = sym.syms
 symbol_number(sym::COFFSymbolRef) = sym.idx
+symbol_section(sym::COFFSymbolRef) = Sections(handle(sym))[deref(sym).SectionNumber]
 @derefmethod symbol_type(sym::COFFSymbolRef)
 function symbol_name(sym::COFFSymbolRef)
     # COFF Symbols set the first four bytes of the name to zero to signify a
@@ -127,6 +127,10 @@ function symbol_name(sym::COFFSymbolRef)
     return unsafe_string(name)
 end
 
+function symbol_offset(sym::COFFSymbolRef)
+    # Return the offset into the file that this symbol refers to
+    return symbol_value(sym) + section_offset(symbol_section(sym))
+end
 
 # Symbol printing stuff
 """
@@ -161,7 +165,7 @@ function symbol_type_string(sym::COFFSymtabEntry)
     if isempty(type_string)
         string("Unknown Symbol Type (0x", string(sym.Type, base=16), ")")
     end
-    
+
     return type_string
 end
 @derefmethod symbol_type_string(s::COFFSymbolRef)
diff --git a/src/ELF/ELF.jl b/src/ELF/ELF.jl
@@ -17,7 +17,7 @@ import ObjectFile: DynamicLink, DynamicLinks, RPath, ObjectHandle, Section, Sect
                    findfirst, deref, section_name, section_size,
                    section_offset, section_address, section_number, segment_name, segment_offset,
                    segment_file_size, segment_memory_size, segment_address, strtab_lookup,
-                   symbol_name, symbol_value, isundef, isglobal, islocal, isweak, symbol_number
+                   symbol_name, symbol_value, isundef, isglobal, islocal, isweak, symbol_number, symbol_section, symbol_offset
 # Load in imported C #define constants
 include("constants.jl")
 
diff --git a/src/ELF/ELFSymbol.jl b/src/ELF/ELFSymbol.jl
@@ -27,7 +27,7 @@ function Symbols(oh::H) where {H <: ELFHandle}
     if !isempty(dyn_sections)
         return ELFSymbols(first(dyn_sections))
     end
-    
+
     dyn_sections = findall(sections, ".dynsym")
     if !isempty(dyn_sections)
         return ELFSymbols(first(dyn_sections))
@@ -176,3 +176,10 @@ function symbol_value(sym::ELFSymbolRef)
     # Return our ill-gotten goods
     return value
 end
+
+function symbol_offset(sym::ELFSymbolRef)
+    addr = symbol_value(sym)
+    sect = symbol_section(sym)
+    virtual_offset =  section_address(sect) - section_offset(sect)
+    return addr - virtual_offset
+end
diff --git a/src/MachO/MachO.jl b/src/MachO/MachO.jl
@@ -17,7 +17,7 @@ import ObjectFile: DynamicLink, DynamicLinks, RPath, ObjectHandle, Section, Sect
                    findfirst, deref, section_name, section_size,
                    section_offset, section_address, section_number, segment_name, segment_offset,
                    segment_file_size, segment_memory_size, segment_address, strtab_lookup,
-                   symbol_name, symbol_value, isundef, isglobal, islocal, isweak, symbol_number
+                   symbol_name, symbol_value, isundef, isglobal, islocal, isweak, symbol_number, symbol_section, symbol_offset
 # Load in imported C #define constants
 include("constants.jl")
 
diff --git a/src/MachO/MachOSymbol.jl b/src/MachO/MachOSymbol.jl
@@ -25,7 +25,6 @@ end
 symbol_name(sym::MachOSymtabEntry) = string("strtab@", sym.n_strx)
 symbol_value(sym::MachOSymtabEntry) = sym.n_value
 symbol_type(sym::MachOSymtabEntry) = sym.n_type
-symbol_section(sym::MachOSymtabEntry) = sym.n_sect
 symbol_description(sym::MachOSymtabEntry) = sym.n_desc
 
 function isglobal(sym::MachOSymtabEntry)
@@ -39,7 +38,7 @@ function isweak(sym::MachOSymtabEntry)
 end
 function isundef(sym::MachOSymtabEntry)
     return (symbol_type(sym) == N_UNDF) ||
-           (isglobal(sym) && symbol_section(sym) == NO_SECT)
+           (isglobal(sym) && sym.n_sect == NO_SECT)
 end
 
 
@@ -115,6 +114,23 @@ end
 deref(sym::MachOSymbolRef) = sym.entry
 Symbols(sym::MachOSymbolRef) = sym.syms
 symbol_number(sym::MachOSymbolRef) = sym.idx
+
+function symbol_section(sym::MachOSymbolRef)
+    # MachO uses 0 to indicate no section, so we use NO_SECT
+    if deref(sym).n_sect == 0
+        return NO_SECT
+    end
+    return Sections(handle(sym))[deref(sym).n_sect]
+end
+
 function symbol_name(sym::MachOSymbolRef)
     return strtab_lookup(StrTab(Symbols(sym)), sym.entry.n_strx)
-end
+end
+
+function symbol_offset(sym::MachOSymbolRef)
+    addr = symbol_value(sym)
+    sect = symbol_section(sym)
+    virtual_offset =  section_address(sect) - section_offset(sect)
+    return addr - virtual_offset
+end
+
diff --git a/test/libjulias/libjulia.1.10.10.dylib b/test/libjulias/libjulia.1.10.10.dylib
diff --git a/test/libjulias/libjulia.dll b/test/libjulias/libjulia.dll
diff --git a/test/libjulias/libjulia.so b/test/libjulias/libjulia.so
diff --git a/test/runtests.jl b/test/runtests.jl
@@ -278,3 +278,23 @@ end
     @test "KERNEL32.dll" in dynamic_links
     @test "libstdc++-6.dll" in dynamic_links
 end
+
+using Mmap
+@testset "Finding dep_libs" begin
+    function find_dep_libs(file)
+        obj = only(readmeta(open(file, "r")))
+        syms = collect(Symbols(obj))
+        syms_names = symbol_name.(syms)
+        sym = syms[findfirst(syms_names .== mangle_symbol_name(obj, "dep_libs"))]
+        offset = symbol_offset(sym)
+        filem = Mmap.mmap(file)
+        data = String(filem[offset: (offset + 255)])
+        @test contains(data, "libjulia-internal")
+        @test contains(data, "libjulia-codegen")
+        @test contains(data, "libopenlibm")
+    end
+    for file in readdir("./libjulias")
+        find_dep_libs(joinpath("./libjulias", file))
+    end
+end
+
