diff --git a/emu/cpu.h b/emu/cpu.h
index e7e89b8b..9bd1eade 100644
--- a/emu/cpu.h
+++ b/emu/cpu.h
@@ -76,6 +76,8 @@ struct cpu_state {
 #define ZF_FLAG (1 << 6)
 #define DF_FLAG (1 << 10)
     };
+    // please pretend this doesn't exist
+    dword_t df_offset;
     // for maximum efficiency these are stored in bytes
     byte_t cf;
     byte_t of;
diff --git a/emu/gadgets-x86/string.S b/emu/gadgets-x86/string.S
index 96f195ab..6b0fcd4d 100644
--- a/emu/gadgets-x86/string.S
+++ b/emu/gadgets-x86/string.S
@@ -6,6 +6,10 @@
     andl $~DF_FLAG, CPU_eflags(%_cpu)
     gret
 
+.gadget std
+    orl $DF_FLAG, CPU_eflags(%_cpu)
+    gret
+
 .macro do_strop op, size, rep, s, ss, a
     # repnz is only a thing for cmps and scas
     .ifc \rep,repnz
@@ -20,11 +24,15 @@
             jz 2f
     1:
         .endif
+        movl $-(\size/8), CPU_df_offset(%_cpu)
+        testl $DF_FLAG, CPU_eflags(%_cpu)
+        jnz 3f
+        negl CPU_df_offset(%_cpu)
+    3:
         .ifc \op,stos
             movl %edi, %_addr
             write_prep
             mov\ss %\a, (%_addrq)
-
         .else; .ifc \op,movs
             movl %esi, %_addr
             read_prep
@@ -34,10 +42,10 @@
             mov\ss %tmp\s, (%_addrq)
         .endif; .endif
         .ifin(\op, lods,movs,cmps,scas)
-            addl $(\size/8), %esi
+            addl CPU_df_offset(%_cpu), %esi
         .endifin
         .ifin(\op, movs,stos,cmps)
-            addl $(\size/8), %edi
+            addl CPU_df_offset(%_cpu), %edi
         .endifin
         .ifc \rep,rep
             decl %ecx
diff --git a/emu/gen.c b/emu/gen.c
index be22efd9..74bc32d5 100644
--- a/emu/gen.c
+++ b/emu/gen.c
@@ -249,7 +249,7 @@ static inline bool gen_op(struct gen_state *state, gadget_t *gadgets, enum arg a
 #define POPF() UNDEFINED
 #define SAHF UNDEFINED
 #define CLD g(cld)
-#define STD UNDEFINED
+#define STD g(std)
 
 #define MUL18(val,z) MUL1(val,z)
 #define MUL1(val,z) load(val, z); gz(mul, z)
diff --git a/emu/offsets.c b/emu/offsets.c
index 53313319..3a8bd791 100644
--- a/emu/offsets.c
+++ b/emu/offsets.c
@@ -29,6 +29,7 @@ void cpu() {
     OFFSET(CPU, cpu_state, op1);
     OFFSET(CPU, cpu_state, op2);
     OFFSET(CPU, cpu_state, flags_res);
+    OFFSET(CPU, cpu_state, df_offset);
     MACRO(PF_RES);
     MACRO(ZF_RES);
     MACRO(SF_RES);