diff -ru a/src/DES_bs.c b/src/DES_bs.c
--- a/src/DES_bs.c	2012-07-14 09:36:44.000000000 -0400
+++ b/src/DES_bs.c	2016-08-07 15:08:48.011369100 -0400
@@ -52,7 +52,7 @@
 };
 
 #if DES_BS_ASM
-extern void DES_bs_init_asm(void);
+extern void DES_INTERNAL_CALL DES_bs_init_asm(void);
 #endif
 
 void DES_bs_init(int LM, int cpt)
diff -ru a/src/DES_bs.h b/src/DES_bs.h
--- a/src/DES_bs.h	2013-05-06 02:19:49.000000000 -0400
+++ b/src/DES_bs.h	2016-08-07 17:57:00.097603400 -0400
@@ -35,6 +35,16 @@
 #define DES_bs_vector			ARCH_WORD
 #endif
 
+#ifndef HAVE_INTERNAL_ATTR
+#define HAVE_INTERNAL_ATTR 0
+#endif
+
+#if     HAVE_INTERNAL_ATTR
+#define DES_INTERNAL_CALL  __attribute__((visibility("internal")))
+#else
+#define DES_INTERNAL_CALL
+#endif
+
 /*
  * All bitslice DES parameters combined into one struct for more efficient
  * cache usage. Don't re-order unless you know what you're doing, as there
@@ -112,7 +122,7 @@
 #else
 #define DES_bs_mt			0
 #define DES_bs_cpt			1
-extern DES_bs_combined DES_bs_all;
+extern  DES_INTERNAL_CALL DES_bs_combined DES_bs_all;
 #define for_each_t(n)
 #define init_t()
 #endif
@@ -120,7 +130,7 @@
 /*
  * Initializes the internal structures.
  */
-extern void DES_bs_init(int LM, int cpt);
+extern void DES_INTERNAL_CALL DES_bs_init(int LM, int cpt);
 
 /*
  * Sets a salt for DES_bs_crypt().
@@ -139,17 +149,17 @@
 /*
  * Almost generic implementation: 24-bit salts, variable iteration count.
  */
-extern void DES_bs_crypt(int count, int keys_count);
+extern void DES_INTERNAL_CALL DES_bs_crypt(int count, int keys_count);
 
 /*
  * A simplified special-case implementation: 12-bit salts, 25 iterations.
  */
-extern void DES_bs_crypt_25(int keys_count);
+extern void DES_INTERNAL_CALL DES_bs_crypt_25(int keys_count);
 
 /*
  * Another special-case version: a non-zero IV, no salts, no iterations.
  */
-extern int DES_bs_crypt_LM(int *keys_count, struct db_salt *salt);
+extern int DES_INTERNAL_CALL DES_bs_crypt_LM(int *keys_count, struct db_salt *salt);
 
 /*
  * Converts an ASCII ciphertext to binary to be used with one of the
diff -ru a/src/john.c b/src/john.c
--- a/src/john.c	2013-05-29 19:27:25.000000000 -0400
+++ b/src/john.c	2016-08-07 17:03:35.227295300 -0400
@@ -61,8 +61,12 @@
 #include "batch.h"
 
 #if CPU_DETECT
+#if defined(HAVE_INTERNAL_ATTR) && HAVE_INTERNAL_ATTR
+extern int __attribute__((visibility("internal"))) CPU_detect(void);
+#else
 extern int CPU_detect(void);
 #endif
+#endif
 
 extern struct fmt_main fmt_DES, fmt_BSDI, fmt_MD5, fmt_BF;
 extern struct fmt_main fmt_AFS, fmt_LM;
diff -ru a/src/Makefile b/src/Makefile
--- a/src/Makefile	2013-05-29 19:21:25.000000000 -0400
+++ b/src/Makefile	2016-08-07 20:17:30.889817100 -0400
@@ -104,6 +104,8 @@
 #	@echo "linux-ppc64-altivec      Linux, PowerPC 64-bit w/AltiVec"
 	@echo "linux-ppc64              Linux, PowerPC 64-bit"
 	@echo "linux-ia64               Linux, IA-64"
+	@echo "midipix-x86-64-avx       Midipix, x86-64 with AVX (2011+ Intel CPUs)"
+	@echo "midipix-x86-64           Midipix, x86-64 with SSE2 (most common)"
 	@echo "freebsd-x86-64           FreeBSD, x86-64 with SSE2 (best)"
 	@echo "freebsd-x86-sse2         FreeBSD, x86 with SSE2 (best if 32-bit)"
 	@echo "freebsd-x86-mmx          FreeBSD, x86 with MMX"
@@ -306,6 +308,22 @@
 		CFLAGS="$(CFLAGS) -DHAVE_CRYPT" \
 		LDFLAGS="$(LDFLAGS) -lcrypt"
 
+midipix-x86-64-avx:
+	$(LN) x86-64.h arch.h
+	$(MAKE) $(PROJ) \
+		JOHN_OBJS="$(JOHN_OBJS) c3_fmt.o x86-64.o" \
+		CFLAGS_MAIN="$(CFLAGS) -DJOHN_AVX -DHAVE_CRYPT" \
+		CFLAGS="$(CFLAGS) -mavx -DHAVE_CRYPT -DHAVE_INTERNAL_ATTR" \
+		ASFLAGS="$(ASFLAGS) -mavx" \
+		LDFLAGS="$(LDFLAGS) -lcrypt"
+
+midipix-x86-64:
+	$(LN) x86-64.h arch.h
+	$(MAKE) $(PROJ) \
+		JOHN_OBJS="$(JOHN_OBJS) c3_fmt.o x86-64.o" \
+		CFLAGS="$(CFLAGS) -DHAVE_CRYPT -DHAVE_INTERNAL_ATTR" \
+		LDFLAGS="$(LDFLAGS) -lcrypt"
+
 freebsd-x86-64:
 	$(LN) x86-64.h arch.h
 	$(MAKE) $(PROJ) \
diff -ru a/src/x86-64.S b/src/x86-64.S
--- a/src/x86-64.S	2012-07-21 09:08:57.000000000 -0400
+++ b/src/x86-64.S	2016-08-07 19:20:55.800629100 -0400
@@ -963,9 +963,27 @@
 
 .text
 
+#ifdef __PE__
+DES_PE_pop_ret:
+	pop %rsi
+	pop %rdi
+	ret
+#endif
+
+#ifdef __PE__
+.def DES_bs_init_asm; .scl 2; .type 32; .endef
+#endif
+
 DO_ALIGN(6)
 .globl DES_bs_init_asm
+
 DES_bs_init_asm:
+#ifdef __PE__
+	push %rdi
+	push %rsi
+	mov  %rcx,%rdi
+	mov  %rdx,%rsi
+#endif
 	pcmpeqd %xmm0,%xmm0
 	movdqa %xmm0,pnot
 	paddb %xmm0,%xmm0
@@ -985,19 +1003,34 @@
 	movdqa %xmm0,mask40
 	SHLB1(%xmm0)
 	movdqa %xmm0,mask80
+#ifdef __PE__
+	jmp DES_PE_pop_ret
+#else
 	ret
+#endif
 
 #define iterations			%edi
 #define rounds_and_swapped		%eax
 
+#ifdef __PE__
+.def DES_bs_crypt; .scl 2; .type 32; .endef
+#endif
+
 DO_ALIGN(6)
 .globl DES_bs_crypt
 DES_bs_crypt:
+#ifdef __PE__
+	push %rdi
+	push %rsi
+	mov  %rcx,%rdi
+	mov  %rdx,%rsi
+#endif
 	cmpl $0,DES_bs_all_keys_changed(%rip)
 	jz DES_bs_crypt_body
 	pushq %rdi
 	call DES_bs_finalize_keys
 	popq %rdi
+
 DES_bs_crypt_body:
 	pxor zero,zero
 	leaq DES_bs_all_KS_v(%rip),k_ptr
@@ -1046,17 +1079,36 @@
 	movl $0x108,rounds_and_swapped
 	subl $1,iterations
 	jnz DES_bs_crypt_swap
+#ifdef __PE__
+	jmp DES_PE_pop_ret
+#else
 	ret
+#endif
+
 DES_bs_crypt_next:
 	subq $nvec(0x300-48),k_ptr
 	movl $8,rounds_and_swapped
 	subl $1,iterations
 	jnz DES_bs_crypt_start
+#ifdef __PE__
+	jmp DES_PE_pop_ret
+#else
 	ret
+#endif
+
+#ifdef __PE__
+.def DES_bs_crypt_25; .scl 2; .type 32; .endef
+#endif
 
 DO_ALIGN(6)
 .globl DES_bs_crypt_25
 DES_bs_crypt_25:
+#ifdef __PE__
+	push %rdi
+	push %rsi
+	mov  %rcx,%rdi
+	mov  %rdx,%rsi
+#endif
 	cmpl $0,DES_bs_all_keys_changed(%rip)
 	jnz DES_bs_finalize_keys_25
 DES_bs_crypt_25_body:
@@ -1108,7 +1160,12 @@
 	movl $0x108,rounds_and_swapped
 	subl $1,iterations
 	jnz DES_bs_crypt_25_swap
+#ifdef __PE__
+	jmp DES_PE_pop_ret
+#else
 	ret
+#endif
+
 DES_bs_crypt_25_next:
 	subq $nvec(0x300-48),k_ptr
 	movl $8,rounds_and_swapped
@@ -1174,9 +1231,19 @@
 
 #define rounds				%eax
 
+#ifdef __PE__
+.def DES_bs_crypt_LM; .scl 2; .type 32; .endef
+#endif
+
 DO_ALIGN(6)
 .globl DES_bs_crypt_LM
 DES_bs_crypt_LM:
+#ifdef __PE__
+	push %rdi
+	push %rsi
+	mov  %rcx,%rdi
+	mov  %rdx,%rsi
+#endif
 	movl (%rdi),%r8d
 	movdqa mask01,%xmm7
 	movdqa mask02,%xmm8
@@ -1333,7 +1400,11 @@
 	subl $1,rounds
 	jnz DES_bs_crypt_LM_loop
 	xchgq %r8,%rax
+#ifdef __PE__
+	jmp DES_PE_pop_ret
+#else
 	ret
+#endif
 
 #endif
 
@@ -1350,8 +1421,19 @@
 #ifdef UNDERSCORES
 #define CPU_detect _CPU_detect
 #endif
+
+#ifdef __PE__
+.def CPU_detect; .scl 2; .type 32; .endef
+#endif
+
 .globl CPU_detect
 CPU_detect:
+#ifdef __PE__
+	push %rdi
+	push %rsi
+	mov  %rcx,%rdi
+	mov  %rdx,%rsi
+#endif
 	pushq %rbx
 	movl $1,%eax
 	cpuid
@@ -1376,12 +1458,21 @@
 #endif
 	movl $1,%eax
 	popq %rbx
+#ifdef __PE__
+	jmp DES_PE_pop_ret
+#else
 	ret
+#endif
+
 CPU_detect_fail:
 	xorl %eax,%eax
 	popq %rbx
+#ifdef __PE__
+	jmp DES_PE_pop_ret
+#else
 	ret
 #endif
+#endif
 
 #if defined(__ELF__) && defined(__linux__)
 .section .note.GNU-stack,"",@progbits