Add size and offset to immediate and memory operand structs (#489)

assets/porting-guide-v4-v5.md

@@ -1,9 +1,24 @@
 # Porting Guide v4 -> v5
 
-# Encoder
+### Encoder
 
 - `ZydisRegisterGetLargestEnclosing` will now return the given register itself for registers that don't have
   an enclosing register. Previously it would return `ZYDIS_REGISTER_NONE` in these cases.
 - `ZydisEncoderDecodedInstructionToEncoderRequest` now expects exactly `instruction->operand_count_visible` to be
   passed, not `operand_count_visible` at maximum. Passing a lower value was previously allowed but didn't really
-  make much sense at all.
+  make much sense at all.
+
+### Decoder
+
+- `ZydisDecodedOperandImm` struct was changed
+  - Added field `offset`
+    - Contains the offset of the immediate data, relative to the beginning of the instruction, in bytes.
+  - Added field `size`
+    - Contains the physical immediate size, in bits.
+- `ZydisDecodedOperandMemDisp_` struct was changed
+  - Added field `offset`
+    - Contains the offset of the immediate data, relative to the beginning of the instruction, in bytes.
+  - Added field `size`
+    - Contains the physical displacement size, in bits.
+  - Removed field `has_displacement`
+    - A `size` of 0 indicates that there is no displacement, effectively replacing the need for `has_displacement`.

include/Zydis/DecoderTypes.h

@@ -149,14 +149,19 @@ typedef struct ZydisDecodedOperandMem_
      */
     struct ZydisDecodedOperandMemDisp_
     {
-        /**
-         * Signals, if the displacement value is used.
-         */
-        ZyanBool has_displacement;
         /**
          * The displacement value
          */
         ZyanI64 value;
+        /**
+         * The offset of the displacement data, relative to the beginning of the
+         * instruction, in bytes.
+         */
+        ZyanU8 offset;
+        /**
+         * The physical displacement size, in bits.
+         */
+        ZyanU8 size;
     } disp;
 } ZydisDecodedOperandMem;
 
@@ -191,6 +196,15 @@ typedef struct ZydisDecodedOperandImm_
         ZyanU64 u;
         ZyanI64 s;
     } value;
+    /**
+      * The offset of the immediate data, relative to the beginning of the
+      * instruction, in bytes.
+      */
+    ZyanU8 offset;
+    /**
+     * The physical immediate size, in bits.
+     */
+    ZyanU8 size;
 } ZydisDecodedOperandImm;
 
 /**

src/Decoder.c

@@ -1433,8 +1433,9 @@ static ZyanStatus ZydisDecodeOperandMemory(const ZydisDecoderContext* context,
     if (displacement_size)
     {
         ZYAN_ASSERT(instruction->raw.disp.size == displacement_size);
-        operand->mem.disp.has_displacement = ZYAN_TRUE;
         operand->mem.disp.value = instruction->raw.disp.value;
+        operand->mem.disp.size = displacement_size;
+        operand->mem.disp.offset = instruction->raw.disp.offset;
     }
     return ZYAN_STATUS_SUCCESS;
 }
@@ -1829,7 +1830,8 @@ static ZyanStatus ZydisDecodeOperands(const ZydisDecoder* decoder, const ZydisDe
             ZYAN_ASSERT(instruction->raw.disp.size);
             operands[i].type = ZYDIS_OPERAND_TYPE_MEMORY;
             operands[i].mem.type = ZYDIS_MEMOP_TYPE_MEM;
-            operands[i].mem.disp.has_displacement = ZYAN_TRUE;
+            operands[i].mem.disp.size = instruction->raw.disp.size;
+            operands[i].mem.disp.offset = instruction->raw.disp.offset;
             operands[i].mem.disp.value = instruction->raw.disp.value;
             break;
         case ZYDIS_SEMANTIC_OPTYPE_MIB:
@@ -1877,6 +1879,8 @@ static ZyanStatus ZydisDecodeOperands(const ZydisDecoder* decoder, const ZydisDe
             {
                 operands[i].imm.value.u = instruction->raw.imm[imm_id].value.u;
             }
+            operands[i].imm.offset = instruction->raw.imm->offset;
+            operands[i].imm.size = instruction->raw.imm->size;
             operands[i].imm.is_signed = instruction->raw.imm[imm_id].is_signed;
             operands[i].imm.is_relative = instruction->raw.imm[imm_id].is_relative;
             ++imm_id;

src/Encoder.c

@@ -4705,7 +4705,7 @@ ZYDIS_EXPORT ZyanStatus ZydisEncoderDecodedInstructionToEncoderRequest(
             enc_op->mem.base = dec_op->mem.base;
             enc_op->mem.index = dec_op->mem.index;
             enc_op->mem.scale = dec_op->mem.type != ZYDIS_MEMOP_TYPE_MIB ? dec_op->mem.scale : 0;
-            if (dec_op->mem.disp.has_displacement)
+            if (dec_op->mem.disp.size)
             {
                 enc_op->mem.displacement = dec_op->mem.disp.value;
             }

src/FormatterATT.c

@@ -208,7 +208,7 @@ ZyanStatus ZydisFormatterATTFormatOperandMEM(const ZydisFormatter* formatter,
 
     const ZyanBool absolute = !formatter->force_relative_riprel &&
         (context->runtime_address != ZYDIS_RUNTIME_ADDRESS_NONE);
-    if (absolute && context->operand->mem.disp.has_displacement &&
+    if (absolute && context->operand->mem.disp.size &&
         (context->operand->mem.index == ZYDIS_REGISTER_NONE) &&
        ((context->operand->mem.base  == ZYDIS_REGISTER_NONE) ||
         (context->operand->mem.base  == ZYDIS_REGISTER_EIP ) ||
@@ -226,7 +226,7 @@ ZyanStatus ZydisFormatterATTFormatOperandMEM(const ZydisFormatter* formatter,
         if (neither_reg_nor_idx)
         {
             ZYAN_CHECK(formatter->func_print_address_abs(formatter, buffer, context));
-        } else if (context->operand->mem.disp.has_displacement && context->operand->mem.disp.value)
+        } else if (context->operand->mem.disp.size && context->operand->mem.disp.value)
         {
             ZYAN_CHECK(formatter->func_print_disp(formatter, buffer, context));
         }

src/FormatterIntel.c

@@ -212,7 +212,7 @@ ZyanStatus ZydisFormatterIntelFormatOperandMEM(const ZydisFormatter* formatter,
 
     const ZyanBool absolute = !formatter->force_relative_riprel &&
         (context->runtime_address != ZYDIS_RUNTIME_ADDRESS_NONE);
-    if (absolute && context->operand->mem.disp.has_displacement &&
+    if (absolute && context->operand->mem.disp.size &&
         (context->operand->mem.index == ZYDIS_REGISTER_NONE) &&
        ((context->operand->mem.base  == ZYDIS_REGISTER_NONE) ||
         (context->operand->mem.base  == ZYDIS_REGISTER_EIP ) ||
@@ -253,7 +253,7 @@ ZyanStatus ZydisFormatterIntelFormatOperandMEM(const ZydisFormatter* formatter,
         if (neither_reg_nor_idx)
         {
             ZYAN_CHECK(formatter->func_print_address_abs(formatter, buffer, context));
-        } else if (context->operand->mem.disp.has_displacement && context->operand->mem.disp.value)
+        } else if (context->operand->mem.disp.size && context->operand->mem.disp.value)
         {
             ZYAN_CHECK(formatter->func_print_disp(formatter, buffer, context));
         }

src/Utils.c

@@ -49,7 +49,7 @@ ZyanStatus ZydisCalcAbsoluteAddress(const ZydisDecodedInstruction* instruction,
     switch (operand->type)
     {
     case ZYDIS_OPERAND_TYPE_MEMORY:
-        if (!operand->mem.disp.has_displacement)
+        if (!operand->mem.disp.size)
         {
             return ZYAN_STATUS_INVALID_ARGUMENT;
         }

tools/ZydisFuzzEncoder.c

@@ -132,7 +132,7 @@ void ZydisCompareRequestToInstruction(const ZydisEncoderRequest *request,
                 ZyanBool acceptable_mismatch = ZYAN_FALSE;
                 if (op1->mem.displacement != op2->mem.disp.value)
                 {
-                    if ((op2->mem.disp.has_displacement) &&
+                    if ((op2->mem.disp.size) &&
                         (op1->mem.index == ZYDIS_REGISTER_NONE) &&
                         ((op1->mem.base == ZYDIS_REGISTER_NONE) ||
                          (op1->mem.base == ZYDIS_REGISTER_EIP) ||

tools/ZydisFuzzShared.c

@@ -136,6 +136,20 @@ void ZydisPrintInstruction(const ZydisDecodedInstruction* instruction,
 
 #endif
 
+void ZydisValidateImmediateSize(ZyanU64 value) 
+{
+    if ((value != 0) &&
+        (value != 8) &&
+        (value != 16) &&
+        (value != 32) &&
+        (value != 64)) 
+    {
+        fprintf(stderr, "Value 0x%016" PRIX64 " does not match any of the expected "
+            "values (0, 8, 16, 32, 64).\n", value);
+        abort();
+    }
+}
+
 // NOTE: This function doesn't validate flag values, yet.
 void ZydisValidateEnumRanges(const ZydisDecodedInstruction* insn,
     const ZydisDecodedOperand* operands, ZyanU8 operand_count)
@@ -147,6 +161,7 @@ void ZydisValidateEnumRanges(const ZydisDecodedInstruction* insn,
             " = 0x%016" PRIX64 "\n", (ZyanU64)(value), (ZyanU64)(max));                            \
         abort();                                                                                   \
     }
+#   define ZYDIS_CHECK_MAX ZYDIS_CHECK_ENUM
 
     ZYDIS_CHECK_ENUM(insn->length, ZYDIS_MAX_INSTRUCTION_LENGTH);
 
@@ -175,11 +190,14 @@ void ZydisValidateEnumRanges(const ZydisDecodedInstruction* insn,
             ZYDIS_CHECK_ENUM(op->mem.segment, ZYDIS_REGISTER_MAX_VALUE);
             ZYDIS_CHECK_ENUM(op->mem.base, ZYDIS_REGISTER_MAX_VALUE);
             ZYDIS_CHECK_ENUM(op->mem.index, ZYDIS_REGISTER_MAX_VALUE);
-            ZYDIS_CHECK_ENUM(op->mem.disp.has_displacement, ZYAN_TRUE);
+            ZydisValidateImmediateSize(op->mem.disp.size);
+            ZYDIS_CHECK_MAX(op->mem.disp.offset + (op->mem.disp.size / 8), insn->length);
             break;
         case ZYDIS_OPERAND_TYPE_IMMEDIATE:
             ZYDIS_CHECK_ENUM(op->imm.is_signed, ZYAN_TRUE);
             ZYDIS_CHECK_ENUM(op->imm.is_relative, ZYAN_TRUE);
+            ZydisValidateImmediateSize(op->imm.size);
+            ZYDIS_CHECK_MAX(op->imm.offset + (op->imm.size / 8), insn->length);
             break;
         default:
             break;
@@ -216,6 +234,7 @@ void ZydisValidateEnumRanges(const ZydisDecodedInstruction* insn,
     }
 
 #   undef ZYDIS_CHECK_ENUM
+#   undef ZYDIS_CHECK_MAX
 }
 
 void ZydisValidateInstructionIdentity(const ZydisDecodedInstruction* insn1,