working static strings migration

macho-go/pkg/ios/macho/edit.go

@@ -10,6 +10,8 @@ import (
 	"strings"
 	"time"
 
+  "unsafe"
+
 	log "github.com/sirupsen/logrus"
 
 	. "ios-wrapper/pkg/ios"
@@ -17,6 +19,7 @@ import (
 )
 
 // #include "fixups.h"
+// #include "arm.h"
 import "C"
 
 func rewriteLoadcommandsWithoutCodesignature(mc *MachoContext) {
@@ -342,6 +345,7 @@ func (mc *MachoContext) RemoveBindSymbols() {
 			value := C.MakeRebaseFixupOpcode(C.int(symbol.next), C.ulonglong(target), C.ulonglong(high8))
 			v := make([]byte, 8)
 			mc.byteorder.PutUint64(v, uint64(value))
+      fmt.Printf("change to rebase at %x\n", symbol.file_address)
 			mc.file.WriteAt(v, int64(symbol.file_address))
 		}
 	}
@@ -582,10 +586,6 @@ func (mc *MachoContext) RemoveStrings() {
   // their reference down too,
   // symtab and dysymtab can be ignored, by removing them lmao
 
-  fmt.Println("TODO: Removing strings")
-
-  // identify the cstring region
-
   var cstring *Section64;
 	for _, command := range mc.commands {
 		switch command.(type) {
@@ -605,7 +605,6 @@ func (mc *MachoContext) RemoveStrings() {
 		}
 	}
 
-  // no section __cstring in __TEXT
   if cstring == nil {
     return
   }
@@ -661,7 +660,7 @@ func (mc *MachoContext) RemoveStrings() {
   // size of the section and segment is defined by the total
   // space for strings required
 
-  section := Section64{
+  new_cstring_section := Section64{
     sectname: secname,
     segname: segname,
     addr: secstart,
@@ -686,7 +685,7 @@ func (mc *MachoContext) RemoveStrings() {
     initprot: 3, // read/write
     nsects: 1,
     flags: 0,
-    sections: []*Section64{&section},
+    sections: []*Section64{&new_cstring_section},
 	}
 
   // rewrite the header to be correct
@@ -793,19 +792,88 @@ func (mc *MachoContext) RemoveStrings() {
   // so we should be very free, because very few program goes this far
   // but if this happens, god bless you
 
-  // for the demo, we can try to hard code, (addresses are virtual)
-  // text should be in 0x100003f70
-  // string to reference is 0x100008000
-
   // encoding ADRP is actually hard hmmge?
 
-  // revert to offset
+  // this part uses file offsets for calculations
-  mc.file.WriteAt([]byte{0x20, 0x00, 0x00, 0xb0}, int64(0x000003f70))
+
-  mc.file.WriteAt([]byte{0x00, 0x00, 0x00, 0x91}, int64(0x000003f74))
+  in_cstring := func (offset uint64) bool {
+    cstring_start := uint64(cstring.Offset())
+    cstring_end := cstring_start + cstring.Size()
+    return (offset >= cstring_start) && (offset < cstring_end)
+  }
+
+  text := mc.segments[1]
+  text_start := text.Fileoff()
+  text_end := text_start + text.Filesize()
+
+  inst := make([]byte, 4)
+  for addr := text_start; addr < text_end; addr = addr + 4 {
+    mc.file.ReadAt(inst, int64(addr))
+    inst_adrp := binary.LittleEndian.Uint32(inst)
+    mc.file.ReadAt(inst, int64(addr + 4))
+    inst_add := binary.LittleEndian.Uint32(inst)
+
+    if !(C.is_adrp(C.uint(inst_adrp)) != 0 && C.is_add(C.uint(inst_add)) != 0) {
+      continue
+    }
+
+    base := (addr >> 12) << 12
+
+    // calculate the old string reference
+    ref_base := C.adrp_imm_get(C.uint(inst_adrp))
+    ref_offset := C.add_imm_get(C.uint(inst_add))
+    ref := base + uint64(ref_base + ref_offset)
+
+    if (!in_cstring(ref)) {
+      continue
+    }
+
+    oldstr := uint64(ref)
+    oldstr_relative := oldstr - uint64(cstring.Offset())
+
+    // find the new string address
+    // using oldstr relative address to cstring section
+    newstr := uint64(new_cstring_section.Offset()) + oldstr_relative
+    newstr_base := (newstr >> 12) << 12 // to calculate new offset in adrp
+    newstr_offset := newstr - newstr_base // to calculate new offset in add
+
+    C.adrp_imm_set((*C.uint32_t)(unsafe.Pointer(&inst_adrp)), C.uint(newstr_base - base))
+    C.add_imm_set((*C.uint32_t)(unsafe.Pointer(&inst_add)), C.uint(newstr_offset))
+
+    binary.LittleEndian.PutUint32(inst, inst_adrp)
+    mc.file.WriteAt(inst, int64(addr))
+    binary.LittleEndian.PutUint32(inst, inst_add)
+    mc.file.WriteAt(inst, int64(addr + 4))
+  }
 
   // modify the rebase table (for both opcode and fixups chain versions)
   // this is for pointer references
 
+	isModernSymbol := mc.dyldinfo == nil
+	isLegacySymbol := !isModernSymbol
+	for _, symbol := range mc.CollectBindSymbols() {
+		if isLegacySymbol {
+		} else {
+			// (high8 << 56 | target) - mach_header
+
+      ref := uint64(symbol.high8 << 56 | symbol.target)
+      if (!in_cstring(ref)) {
+        continue
+      }
+
+      oldstr := ref
+      oldstr_relative := oldstr - uint64(cstring.Offset())
+      newstr := uint64(new_cstring_section.Offset()) + oldstr_relative
+
+			target := newstr & 0x00FFFFFFFFFFFFFF
+			high8 := newstr >> 56
+			value := C.MakeRebaseFixupOpcode(C.int(symbol.next), C.ulonglong(target), C.ulonglong(high8))
+			v := make([]byte, 8)
+			mc.byteorder.PutUint64(v, uint64(value))
+      fmt.Printf("change to rebase at %x\n", symbol.file_address)
+			mc.file.WriteAt(v, int64(symbol.file_address))
+		}
+	}
 }
 
 func (mc *MachoContext) AddSection(segname string, name string, size int) Section {