diff --git a/getopt.c b/getopt.c
index ca2d79d..7139bfd 100644
--- a/getopt.c
+++ b/getopt.c
@@ -11,12 +11,16 @@ int optind=1, opterr=1, optopt, __optpos, optreset=0;
 static void __getopt_msg(const char *a, const char *b, const char *c, size_t l)
 {
 	FILE *f = stderr;
+#if !defined(WIN32) && !defined(_WIN32)
 	flockfile(f);
+#endif
 	fputs(a, f);
 	fwrite(b, strlen(b), 1, f);
 	fwrite(c, 1, l, f);
 	fputc('\n', f);
+#if !defined(WIN32) && !defined(_WIN32)
 	funlockfile(f);
+#endif
 }
 
 int getopt(int argc, char * const argv[], const char *optstring)
diff --git a/kthread.c b/kthread.c
index 6004435..04909c8 100644
--- a/kthread.c
+++ b/kthread.c
@@ -3,9 +3,7 @@
 #include <limits.h>
 #include <stdint.h>
 
-#if defined(WIN32) || defined(_WIN32)
-#include <malloc.h>
-#define alloca _alloca
+#if (defined(WIN32) || defined(_WIN32)) && defined(_MSC_VER)
 #define __sync_fetch_and_add(ptr, addend)     _InterlockedExchangeAdd((void*)ptr, addend)
 #endif
 
@@ -59,12 +57,13 @@ void kt_for(int n_threads, void (*func)(void*,long,int), void *data, long n)
 		kt_for_t t;
 		pthread_t *tid;
 		t.func = func, t.data = data, t.n_threads = n_threads, t.n = n;
-		t.w = (ktf_worker_t*)alloca(n_threads * sizeof(ktf_worker_t));
-		tid = (pthread_t*)alloca(n_threads * sizeof(pthread_t));
+		t.w = (ktf_worker_t*)calloc(n_threads, sizeof(ktf_worker_t));
+		tid = (pthread_t*)calloc(n_threads, sizeof(pthread_t));
 		for (i = 0; i < n_threads; ++i)
 			t.w[i].t = &t, t.w[i].i = i;
 		for (i = 0; i < n_threads; ++i) pthread_create(&tid[i], 0, ktf_worker, &t.w[i]);
 		for (i = 0; i < n_threads; ++i) pthread_join(tid[i], 0);
+		free(tid); free(t.w);
 	} else {
 		long j;
 		for (j = 0; j < n; ++j) func(data, j, 0);
diff --git a/sketch.c b/sketch.c
index ab3bcc8..9bc5eba 100644
--- a/sketch.c
+++ b/sketch.c
@@ -5,11 +5,6 @@
 #include "kvec.h"
 #include "minimap.h"
 
-#if defined(WIN32) || defined(_WIN32)
-#include <malloc.h>
-#define alloca _alloca
-#endif
-
 unsigned char seq_nt4_table[256] = {
 	0, 1, 2, 3,  4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4, 
 	4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4,  4, 4, 4, 4, 
@@ -86,7 +81,7 @@ void mm_sketch(void *km, const char *str, int len, int w, int k, uint32_t rid, i
 	tiny_queue_t tq;
 
 	assert(len > 0 && w > 0 && k > 0 && k <= 28); // 56 bits for k-mer; could use long k-mers, but 28 enough in practice
-	buf = (mm128_t*)alloca(w * 16);
+	buf = (mm128_t*)calloc(w, 16);
 	memset(buf, 0xff, w * 16);
 	memset(&tq, 0, sizeof(tiny_queue_t));
 	kv_resize(mm128_t, km, *p, p->n + len/w);
@@ -145,4 +140,5 @@ void mm_sketch(void *km, const char *str, int len, int w, int k, uint32_t rid, i
 	}
 	if (min.x != UINT64_MAX)
 		kv_push(mm128_t, km, *p, min);
+	free(buf);
 }