From: Frode Solheim <frode@fs-uae.net>
Date: Tue, 22 Sep 2015 18:37:09 +0000 (+0200)
Subject: JIT: Do not use uae_p32 for an intentional case of 32-bit wraparound
X-Git-Tag: 3200~55^2
X-Git-Url: https://git.unchartedbackwaters.co.uk/w/?a=commitdiff_plain;h=f46a79831f79911947d2b738f307c6463ae8ab06;p=francis%2Fwinuae.git

JIT: Do not use uae_p32 for an intentional case of 32-bit wraparound
---

diff --git a/jit/compemu_fpp.cpp b/jit/compemu_fpp.cpp
index d025fdd7..a76ac929 100644
--- a/jit/compemu_fpp.cpp
+++ b/jit/compemu_fpp.cpp
@@ -548,6 +548,9 @@ void comp_fbcc_opp (uae_u32 opcode)
 		return;
 	}
 
+	// comp_pc_p is expected to be bound to 32-bit addresses
+	assert((uintptr) comp_pc_p <= 0xffffffffUL);
+
 	if (opcode & 0x20) {  /* only cc from 00 to 1f are defined */
 		FAIL (1);
 		return;
@@ -558,9 +561,13 @@ void comp_fbcc_opp (uae_u32 opcode)
 	else {
 		off = comp_get_ilong ((m68k_pc_offset += 4) - 4);
 	}
-	mov_l_ri (S1, uae_p32(
-			  comp_pc_p + off - (m68k_pc_offset - start_68k_offset)));
-	mov_l_ri (PC_P, uae_p32(comp_pc_p));
+
+	/* Note, "off" will sometimes be (unsigned) "negative", so the following
+         * uintptr can be > 0xffffffff, but the result will be correct due to
+         * wraparound when truncated to 32 bit in the call to mov_l_ri. */
+	mov_l_ri(S1, (uintptr)
+		(comp_pc_p + off - (m68k_pc_offset - start_68k_offset)));
+	mov_l_ri(PC_P, (uintptr) comp_pc_p);
 
 	/* Now they are both constant. Might as well fold in m68k_pc_offset */
 	add_l_ri (S1, m68k_pc_offset);