Fix archinfo for c64x for invalid and unaligned instructions ##arch (#24076)

Author: trufae

libr/arch/p/tms320/plugin.c

@@ -388,9 +388,14 @@ static bool decode(RArchSession *as, RAnalOp *op, RAnalOpMask mask) {
 	const ut64 addr = op->addr;
 	const ut8 *buf = op->bytes;
 	const int len = op->size;
-	op->size = 1;
+	op->size = 2;
 	const char *cpu = as->config->cpu;
+#ifdef CAPSTONE_TMS320C64X_H
+	// default to c64x if available
+	TMS_ANAL_OP_FN aop = tms320c64x_analop;
+#else
 	TMS_ANAL_OP_FN aop = tms320_c55x_op;
+#endif
 	if (R_STR_ISNOTEMPTY (cpu)) {
 		tms320_dasm_t *engine = tms320_engine_for_session (as);
 		if (!r_str_casecmp (cpu, "c64x")) {
@@ -410,15 +415,16 @@ static bool decode(RArchSession *as, RAnalOp *op, RAnalOpMask mask) {
 			aop = tms320_c55x_plus_op;
 		}
 	}
+	ut8 mbuf[4];
+	const ut8 *lbuf = buf;
 	if (len > 3 && as->config->big_endian & R_SYS_ENDIAN_BIG) {
-		ut8 mbuf[4];
 		mbuf[0] = buf[3];
 		mbuf[1] = buf[2];
 		mbuf[2] = buf[1];
 		mbuf[3] = buf[0];
-		return aop (as, op, addr, mbuf, len, mask) > 0;
+		lbuf = mbuf;
 	}
-	return aop (as, op, addr, buf, len, mask) > 0;
+	return aop (as, op, addr, lbuf, len, mask) > 0;
 }
 
 static bool tms320_init(RArchSession *as) {
@@ -465,16 +471,18 @@ static char *mnemonics(RArchSession *s, int id, bool json) {
 
 static int archinfo(RArchSession *as, ut32 q) {
 	switch (q) {
+	case R_ARCH_INFO_DATA_ALIGN:
+		return 2;
 	case R_ARCH_INFO_CODE_ALIGN:
-		return 0;
+		return 2;
 	case R_ARCH_INFO_MAXOP_SIZE:
 		return 8;
 	case R_ARCH_INFO_INVOP_SIZE:
-		return 1;
+		return 2;
 	case R_ARCH_INFO_MINOP_SIZE:
-		return 1;
+		return 2;
 	}
-	return -1;
+	return 4;
 }
 
 const RArchPlugin r_arch_plugin_tms320 = {

test/db/anal/tms320.c64x_32

@@ -50,123 +50,113 @@ pi~=
 EOF
 EXPECT=<<EOF
 a17 = (half) 0x13cf
-a28 = 0 ext a0 .. 0
-b23:b22 = (word)*a1++[12]
-a4 = 2 ext a0 .. 0x1a
-a26 = 0x4988
-a4 = 0x12 .bitset a20 .. 0x12
-a13:a12 = (word)*-a18[a12]
-a17 = a31 avg a0
+a0 = 2 .bitset a0 .. 2
+a0 = (half)*+b15[0x5a48]
+b9 = b17 - 1
+*-a18[a4] = (word)b4
+a19 = 0xf ext a2 .. 5
+a7 = 7 ext a25 .. 0x10
 b22 = (word)*+b14[0x3da6]
 a4 = (half)*+b15[0x49]
 a25 = 8 ext a21 .. 0x10
 a = a8 ~ a21
-a16 = a23 * a25
-b17 = (half)*+b15[0xd15]
+a0 = (half)*+b15[0x48]
 a16 = (half)*+b14[0x0]
 *+b15[0x6c41] = (word)b19
 b8 = (byte)*+b14[0x660f]
 b12 = (byte)*+b14[0x2e0f]
 b12 = (byte)*+b14[0x2e0f]
-a0 = 0x1a .bitset a20 .. 0x1a
-a22 = 9 .bitclear a17 .. 9
-a10 = 0xe ext a12 .. 0
-a7 = 0 ext a27 .. 0x10
-a22 = 0x18 ext a29 .. 8
-*+b14[0x488b] = (byte)b0
+b6 = b4 cmpgtu b18
+*-a12[a14] = (half)a11:a10
+b3 = b2 * b16
+b24 = 2 .bitset b4 .. 2
+b14 = (half)*-a2[a2]
+a17 = 0x19 ext a2 .. 0x1f
 a14 = 1 ext a17 .. 4
 a17 = (byte)*a3--[10]
 b16 = (half) -0x2c00
 b4 = (half)*+b15[0x488d]
-a6 = (half) 1
+*-a0[a7] = (word)a0
 a30 = 0 ext a0 .. 0
 a23:a22 = 1 + a20:a19
 a20 = (half)*-a23[a18]
-b8 += 0x2463
-*-a1[14] = (word)a17
-a16 = 0x1a ext a0 .. 0x11
-a23 = 0x1c ext a16 .. 0x10
-*+b14[0x5b1] = (word)b26
-a4 = 2 ext a0 .. 0x13
-b19 = 0x1a .bitset b30 .. 0x1a
-b9 = (half)*+b15[0xb05]
-a10 = 0x10 ext a8 .. 0
+a4 = 0xe ext a12 .. 0x12
+b8 = 0x1a .bitset b0 .. 0x1a
+a0 += -0x6ffa
+b16 += -0x6eee
+*+a18[4] = (half)a1:a0
+b30 += 0x4400
+a17 = (half)*+b14[0x9fb]
+b4 = 2 .bitset b0 .. 2
 a26 = (byte)*-a5[15]
 b0 = 2 .bitset b0 .. 2
 a30 = (word)*+b15[0x4883]
-*+b15[0x47c1] = (word)b17
-*+b15[0x7fff] = (word)b31
-a1 = a16 * a15
-*-a3[a7] = (word)b1
-a10 = 8 ext a0 .. 0
-a18 = max(a24, a16)
-b20 = (half)*+b14[0x5bc3]
-b30 = (word)*-a16[0]
+*+b15[0x41ff] = (word)b15
+a0 = (half) 0x119
+a12 = a8 * a13
+a17 = 0xa ext a2 .. 3
+b16 = b22 - 0x12
+b0 = (byte)*+b14[0x66]
 a0 = 0x10 ext a15 .. 0
-a2 = 8 ext a29 .. 0x18
-a23:a22 = (word)*-a1[12]
-b4 = 2 .bitset b0 .. 2
+b10 = 2 .bitset b6 .. 2
+a0 = 4 ext a18 .. 0xb
 *+b14[0xd05] = (half)b17
 a14 = 8 ext a0 .. 0
 b11 = (half)*+b14[0x4948]
 a9 = (half)*+b15[0x4848]
-a17 = a24 * a29
-a1:a0 = a0 * a0
-a26 = a0 * a1
-b30 = (word)*-a7[24]
+*+b15[0x100] = (byte)b14
+*+b14[0x40e9] = (word)a3
+b8 = (half)*+b14[0x100]
+b30 = (byte)*+b14[0x66]
 a17 = 0x18 ext a2 .. 5
 a10 = 0x10 ext a8 .. 0
 a19 = (half)*+b15[0x5048]
-a15 = a24 * a29
-a17 = 0xf ext a2 .. 0x11
-*a13--(a9) = (half)b27:b26
-b2 = 5 * b0
-a0 = 0 ext a0 .. 0
-a22 = (half)*-a0[a2]
-a30 = 6 ext a17 .. 2
+a14 = 8 ext a0 .. 0
+a19 = (half)*+b15[0x7148]
+*+b15[0x100] = (byte)b14
+b0 = 2 .bitset b0 .. 2
+b11 = (half)*+b14[0x4048]
+a9 = 0x12 .bitset a16 .. 0x12
 a4 = (half) 0x6381
 a24 = a0 * a1
 b13 = (byte)*+b14[0x1066]
-*a16++[10] = (half)a0
-b19 += 0x13fc
-a17 = 0xe ext a2 .. 0x15
-a26 = 0xc * a3
-b0 = b2 * b0
-a24 = (byte)*+b14[0x508b]
+a2 = 0x12 ext a21 .. 0x15
+a19 = 0x12 ext a31 .. 0x13
+a19 = (half)*+b15[0x5548]
+a0 = 0 ext a0 .. 0
+b7 = 2 .bitset b27 .. 2
 a28 = 0x10 ext a8 .. 0
 a18 = 1 ext a31 .. 4
 a17 = (byte)*a2--[9]
-a26 = a1 * a17
-a28 = 1 ext a17 .. 4
-a17 = (byte)*a3--[10]
-a30 = (half)*a3++[24]
-a17 = (byte)*a3--[11]
-*+b14[0x488d] = (word)a16
-b8 = (half)*-a8[a2]
+*+b14[0x730f] = (word)b1
+a8 = 0 ext a9 .. 8
+b26 = (byte)*+b14[0x5424]
+a8 = (half)*+b15[0x2448]
+a8 = 1 ext a9 .. 8
+b0 = (half)*-a0[0]
 a18 = a1 * a17
 *+b15[0x63fc] = (word)b16
 *+a19[a4] = (half)a31:a30
 *-a1[14] = (word)a27
-a3 = (byte)*+b14[0x4039]
+a0 = 8 .bitclear a14 .. 8
 b26 = (half)*+b14[0x2200]
 b11 = b2 * b16
 *+b14[0x505b] = (half)a8
 b2 = (half)*+b14[0x5ec3]
-a8 = 0 ext a0 .. 0
+a0 = (half)*+b15[0x48]
 a14 = 0x10 ext a0 .. 0
 a19 = 0xa .bitset a26 .. 0xa
 *+b15[0x31c0] = (byte)b4
-a0 = (half) 1
-a14 = a24 * a0
-a17 += 0x788
-*+b14[0x405b] = (half)a3
-b3:b2 = (word)*+a23[22]
-b0 = (half)*+b14[0x662e]
-b0 = 0x1a .bitset b20 .. 0x1a
-a14 = 2 .bitset a4 .. 2
-b22 = b4 - b18
-a17 = 0x1b ext a2 .. 0x1f
-a3 = a18 * a20
+b0 = (byte)*-a0[a7]
+a6 = 8 ext a0 .. 0
+a8:a7 = a2 * a17
+a0 = a22 + 0x1a
+b12 = (byte)*+b14[0x2e0f]
+*+a17[a4] = (half)b7:b6
+b0 = b4 cmpgtu b18
+a17 = 0x19 ext a2 .. 0x13
+a = a24 ~ a31
+a4 = a12 - a16
 a31 = (half)*+a5[14]
 a3 = 9 .bitclear a18 .. 9
 a20 = 0 ext a0 .. 0
@@ -189,18 +179,35 @@ a15 = a25 - a26
 a20 = (half)*+b15[0x348]
 a16 = 0xe ext a0 .. 0
 a19 = (half)*-a16[a10]
-a2 = (half)*a3++[29]
-a26 = 2 .bitset a0 .. 2
-a12 = 0xe ext a29 .. 0x18
+a31 = (half)*+b14[0x3644]
+a0 += -0x6eed
+a11 = (half)*+b15[0x5849]
 a12 = 5 .bitclear a5 .. 5
 a0 = (half)*+b14[0x0]
 a16 = (half)*+b14[0x29f0]
 *+b14[0x7f48] = (byte)a19
+b0 = (half)*+b14[0x48]
+a15 = a20 - a16
+a16 = a16 - 0xc
+a4 = a2 - 4
+b17 += 0x4a63
+a0 = 3 ext a1 .. 0x14
+a0 = zero
+*+a16[0] = (word)a13
+a24 = (half)*+b14[0xfb7]
+a12 = (half)*+b14[0x6689]
+*-a24[0] = (half)a7:a6
+a14 = (half)*+b14[0xfb6]
+a24 = 0x1d .bitclear a5 .. 0x1d
+a0 = 0 ext a0 .. 0
+a26 = 0 ext a13 .. 7
+*a11++[a11] = (word)a26
+a2 = 0x90
+b4 = b4 - b18
+a16 += -0x6eed
 a0 = 0 ext a0 .. 0
 b30 = b30 * b17
-a16 = 2 .bitset a2 .. 2
-a0 = 0x10 .bitset a28 .. 0x10
-b3 = b15 * b16
-a7:a6 = a5 * a18
+b16 = b2 - 4
+a0 = (half)*-a0[a5]
 EOF
 RUN