From daffe8bbbbc78c847f15a25e598c445270bea67a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Florian=20M=C3=A4rkl?= <info@florianmaerkl.de>
Date: Sun, 21 May 2023 05:16:37 +0200
Subject: [PATCH] Fix arm64 adrp assembler (#3519)

Fixed the calculation of the immediate value and more spaces, e.g.
"adrp     x0, 0x1234000" are allowed now.
---
 librz/asm/arch/arm/armass64.c | 43 ++++++++++++-----------------------
 test/db/asm/arm_64            |  4 +++-
 2 files changed, 17 insertions(+), 30 deletions(-)

diff --git a/librz/asm/arch/arm/armass64.c b/librz/asm/arch/arm/armass64.c
index 837ae38c52..bac4035779 100644
--- a/librz/asm/arch/arm/armass64.c
+++ b/librz/asm/arch/arm/armass64.c
@@ -846,41 +846,26 @@ static ut32 logical(ArmOp *op, bool invert, LogicalOp opc) {
 	} else {
 		return UT32_MAX;
 	}
-
-	ut8 flip[4];
-	rz_write_le32(flip, data);
-	return rz_read_be32(flip);
+	return rz_swap_ut32(data);
 }
 
-static ut32 adrp(ArmOp *op, ut64 addr, ut32 k) { //, int reg, ut64 dst) {
-	ut64 at = 0LL;
-	ut32 data = k;
-	if (op->operands[0].type == ARM_GPR) {
-		data |= encode1reg(op);
-	} else {
+static ut32 adrp(ArmOp *op, ut64 addr) { //, int reg, ut64 dst) {
+	ut32 data = 0x90000000;
+	if (op->operands[0].type != ARM_GPR) {
 		RZ_LOG_ERROR("assembler: arm64: adrp: invalid assembly. valid usage: adrp x0, addr\n");
 		return UT32_MAX;
 	}
-	if (op->operands[1].type == ARM_CONSTANT) {
-		// XXX what about negative values?
-		at = op->operands[1].immediate - addr;
-		at /= 4;
-	} else {
+	data |= op->operands[0].reg & 0x1f;
+	if (op->operands[1].type != ARM_CONSTANT) {
 		RZ_LOG_ERROR("assembler: arm64: adrp: invalid assembly. valid usage: adrp x0, addr\n");
 		return UT32_MAX;
 	}
-	ut8 b0 = at;
-	ut8 b1 = (at >> 3) & 0xff;
-
-#if 0
-	ut8 b2 = (at >> (8 + 7)) & 0xff;
-	data += b0 << 29;
-	data += b1 << 16;
-	data += b2 << 24;
-#endif
-	data += b0 << 16;
-	data += b1 << 8;
-	return data;
+	ut64 imm = op->operands[1].immediate & ~0xfff;
+	imm -= addr & ~0xfff;
+	imm >>= 12;
+	data |= (imm & 3) << 29;
+	data |= ((imm >> 2) & rz_num_bitmask(19)) << 5;
+	return rz_swap_ut32(data);
 }
 
 static ut32 adr(ArmOp *op, int addr) {
@@ -1396,8 +1381,8 @@ bool arm64ass(const char *str, ut64 addr, ut32 *op) {
 		*op = arithmetic(&ops, 0x11);
 	} else if (!strncmp(str, "adr x", 5)) { // w
 		*op = adr(&ops, addr);
-	} else if (!strncmp(str, "adrp x", 6)) {
-		*op = adrp(&ops, addr, 0x00000090);
+	} else if (!strncmp(str, "adrp ", 5)) {
+		*op = adrp(&ops, addr);
 	} else if (!strncmp(str, "neg", 3)) {
 		*op = neg(&ops);
 	} else if (!strcmp(str, "isb")) {
diff --git a/test/db/asm/arm_64 b/test/db/asm/arm_64
index 8828604eed..36b0fb2366 100644
--- a/test/db/asm/arm_64
+++ b/test/db/asm/arm_64
@@ -42,6 +42,8 @@ a "eon w0, w13, w20" a001344a
 a "eon x0, x13, x20" a00134ca
 a "eon x0, x13, x20, ror 15" a03df4ca
 a "adrp x5, 0x0" 05000090
+ad "adrp x0, 0x11341000" e07fff90 0x12345678
+ad "adrp x0, 0x18343000" e0ff02d0 0x12345678
 a "add x0, x1, x2" 2000028b
 a "add x0, xzr, x30" e0031e8b
 a "b 0x0" 00000014
@@ -428,7 +430,7 @@ d "add w1, w2, w2, lsl 10" 4128020b 0x0 (set x1 (cast 64 false (+ (cast 32 false
 d "add w1, w2, w2, lsr 12" 4130420b 0x0 (set x1 (cast 64 false (+ (cast 32 false (var x2)) (>> (cast 32 false (var x2)) (bv 6 0xc) false))))
 d "add w1, w2, w2, asr 31" 417c820b 0x0 (set x1 (cast 64 false (+ (cast 32 false (var x2)) (>> (cast 32 false (var x2)) (bv 6 0x1f) (msb (cast 32 false (var x2)))))))
 d "adr x2, 0x10042" 02020050 0x10000 (set x2 (bv 64 0x10042))
-d "adrp x2, 0x4000" 020000b0 0x3210 (set x2 (bv 64 0x4000))
+ad "adrp x2, 0x4000" 020000b0 0x3210 (set x2 (bv 64 0x4000))
 d "and x1, x2, x3" 4100038a 0x0 (set x1 (& (var x2) (var x3)))
 d "and w1, w2, w3" 4100030a 0x0 (set x1 (cast 64 false (& (cast 32 false (var x2)) (cast 32 false (var x3)))))
 d "ands x1, x2, x3" 410003ea 0x0 (seq (set x1 (& (var x2) (var x3))) (set zf (is_zero (var x1))) (set nf (msb (var x1))) (set cf false) (set vf false))