diff --git a/amd64.c b/amd64.c
index ab62bd92..664730a1 100644
--- a/amd64.c
+++ b/amd64.c
@@ -129,10 +129,11 @@ cothread_t co_derive(void* memory, unsigned int size, void (*entrypoint)(void))
   if(!co_active_handle) co_active_handle = &co_active_buffer;
 
   if(handle = (cothread_t)memory) {
-    long long *p = (long long*)((char*)handle + size);  /* seek to top of stack */
-    *--p = (long long)crash;                            /* crash if entrypoint returns */
-    *--p = (long long)entrypoint;                       /* start of function */
-    *(long long*)handle = (long long)p;                 /* stack pointer */
+    unsigned int offset = (size & ~15) - 32;
+    long long *p = (long long*)((char*)handle + offset);  /* seek to top of stack */
+    *--p = (long long)crash;                              /* crash if entrypoint returns */
+    *--p = (long long)entrypoint;                         /* start of function */
+    *(long long*)handle = (long long)p;                   /* stack pointer */
   }
 
   return handle;
@@ -145,14 +146,13 @@ cothread_t co_create(unsigned int size, void (*entrypoint)(void)) {
     co_swap = (void (*)(cothread_t, cothread_t))co_swap_function;
   }
   if(!co_active_handle) co_active_handle = &co_active_buffer;
-  size += 512;  /* allocate additional space for storage */
-  size &= ~15;  /* align stack to 16-byte boundary */
 
   if(handle = (cothread_t)malloc(size)) {
-    long long *p = (long long*)((char*)handle + size);  /* seek to top of stack */
-    *--p = (long long)crash;                            /* crash if entrypoint returns */
-    *--p = (long long)entrypoint;                       /* start of function */
-    *(long long*)handle = (long long)p;                 /* stack pointer */
+    unsigned int offset = (size & ~15) - 32;
+    long long *p = (long long*)((char*)handle + offset);  /* seek to top of stack */
+    *--p = (long long)crash;                              /* crash if entrypoint returns */
+    *--p = (long long)entrypoint;                         /* start of function */
+    *(long long*)handle = (long long)p;                   /* stack pointer */
   }
 
   return handle;
diff --git a/libco.h b/libco.h
index b7f60852..a436957c 100644
--- a/libco.h
+++ b/libco.h
@@ -1,5 +1,5 @@
 /*
-  libco v19 (2019-02-18)
+  libco v20 (2019-10-14)
   author: byuu
   license: ISC
 */
diff --git a/x86.c b/x86.c
index d79306c6..3eb2e21e 100644
--- a/x86.c
+++ b/x86.c
@@ -83,10 +83,11 @@ cothread_t co_derive(void* memory, unsigned int size, void (*entrypoint)(void))
   if(!co_active_handle) co_active_handle = &co_active_buffer;
 
   if(handle = (cothread_t)memory) {
-    long *p = (long*)((char*)handle + size);  /* seek to top of stack */
-    *--p = (long)crash;                       /* crash if entrypoint returns */
-    *--p = (long)entrypoint;                  /* start of function */
-    *(long*)handle = (long)p;                 /* stack pointer */
+    unsigned int offset = (size & ~15) - 32;
+    long *p = (long*)((char*)handle + offset);  /* seek to top of stack */
+    *--p = (long)crash;                         /* crash if entrypoint returns */
+    *--p = (long)entrypoint;                    /* start of function */
+    *(long*)handle = (long)p;                   /* stack pointer */
   }
 
   return handle;
@@ -99,14 +100,13 @@ cothread_t co_create(unsigned int size, void (*entrypoint)(void)) {
     co_swap = (void (fastcall*)(cothread_t, cothread_t))co_swap_function;
   }
   if(!co_active_handle) co_active_handle = &co_active_buffer;
-  size += 256;  /* allocate additional space for storage */
-  size &= ~15;  /* align stack to 16-byte boundary */
 
   if(handle = (cothread_t)malloc(size)) {
-    long *p = (long*)((char*)handle + size);  /* seek to top of stack */
-    *--p = (long)crash;                       /* crash if entrypoint returns */
-    *--p = (long)entrypoint;                  /* start of function */
-    *(long*)handle = (long)p;                 /* stack pointer */
+    unsigned int offset = (size & ~15) - 32;
+    long *p = (long*)((char*)handle + offset);  /* seek to top of stack */
+    *--p = (long)crash;                         /* crash if entrypoint returns */
+    *--p = (long)entrypoint;                    /* start of function */
+    *(long*)handle = (long)p;                   /* stack pointer */
   }
 
   return handle;