X-Git-Url: https://wimlib.net/git/?a=blobdiff_plain;f=src%2Flzms_common.c;h=fb13a4358c5469a79d0c4c5e101c03ee2f9e59db;hb=831eeaad4513475ce947efaef16fc79ab3997444;hp=47c071178940732b8eb8e6a03883c770be5132cb;hpb=6bf4bba9f875d1815989a94c39a27f01961d82bb;p=wimlib

diff --git a/src/lzms_common.c b/src/lzms_common.c
index 47c07117..fb13a435 100644
--- a/src/lzms_common.c
+++ b/src/lzms_common.c
@@ -3,7 +3,7 @@
  */
 
 /*
- * Copyright (C) 2013, 2014 Eric Biggers
+ * Copyright (C) 2013-2016 Eric Biggers
  *
  * This file is free software; you can redistribute it and/or modify it under
  * the terms of the GNU Lesser General Public License as published by the Free
@@ -379,7 +379,7 @@ lzms_dilute_symbol_frequencies(u32 freqs[], unsigned num_syms)
 
 
 #ifdef __x86_64__
-static inline u8 *
+static forceinline u8 *
 find_next_opcode_sse4_2(u8 *p)
 {
 	const __v16qi potential_opcodes = (__v16qi) {0x48, 0x4C, 0xE8, 0xE9, 0xF0, 0xFF};
@@ -391,7 +391,11 @@ find_next_opcode_sse4_2(u8 *p)
 		"  pcmpestri $0x0, (%[p]), %[potential_opcodes]      \n"
 		"  jnc 1b                                            \n"
 		"2:                                                  \n"
+	#ifdef __ILP32__ /* x32 ABI (x86_64 with 32-bit pointers) */
+		"  add %%ecx, %[p]                                   \n"
+	#else
 		"  add %%rcx, %[p]                                   \n"
+	#endif
 		: [p] "+r" (p)
 		: [potential_opcodes] "x" (potential_opcodes), "a" (6), "d" (16)
 		: "rcx", "cc"
@@ -401,7 +405,7 @@ find_next_opcode_sse4_2(u8 *p)
 }
 #endif /* __x86_64__ */
 
-static inline u8 *
+static forceinline u8 *
 find_next_opcode_default(u8 *p)
 {
 	/*
@@ -433,7 +437,7 @@ find_next_opcode_default(u8 *p)
 	return p;
 }
 
-static inline u8 *
+static forceinline u8 *
 translate_if_needed(u8 *data, u8 *p, s32 *last_x86_pos,
 		    s32 last_target_usages[], bool undo)
 {
@@ -516,15 +520,15 @@ have_opcode:
 	p += opcode_nbytes;
 	if (undo) {
 		if (i - *last_x86_pos <= max_trans_offset) {
-			u32 n = get_unaligned_u32_le(p);
-			put_unaligned_u32_le(n - i, p);
+			u32 n = get_unaligned_le32(p);
+			put_unaligned_le32(n - i, p);
 		}
-		target16 = i + get_unaligned_u16_le(p);
+		target16 = i + get_unaligned_le16(p);
 	} else {
-		target16 = i + get_unaligned_u16_le(p);
+		target16 = i + get_unaligned_le16(p);
 		if (i - *last_x86_pos <= max_trans_offset) {
-			u32 n = get_unaligned_u32_le(p);
-			put_unaligned_u32_le(n + i, p);
+			u32 n = get_unaligned_le32(p);
+			put_unaligned_le32(n + i, p);
 		}
 	}