diff --git a/cbits/sha3.c b/cbits/sha3.c
--- a/cbits/sha3.c
+++ b/cbits/sha3.c
@@ -124,8 +124,8 @@
 		ctx->bufindex = 0;
 	}
 
-	/* process as much ctx->bufsz-block as possible except the last one in case we finalize */
-	for (; len > ctx->bufsz; len -= ctx->bufsz, data += ctx->bufsz)
+	/* process as much ctx->bufsz-block */
+	for (; len >= ctx->bufsz; len -= ctx->bufsz, data += ctx->bufsz)
 		sha3_do_chunk(ctx->state, (uint64_t *) data, ctx->bufsz / 8);
 
 	/* append data into buf */
diff --git a/cbits/skein256.c b/cbits/skein256.c
--- a/cbits/skein256.c
+++ b/cbits/skein256.c
@@ -124,6 +124,9 @@
 {
 	uint32_t to_fill;
 
+	if (!len)
+		return;
+
 	to_fill = 32 - ctx->bufindex;
 
 	if (ctx->bufindex == 32) {
@@ -131,8 +134,9 @@
 		ctx->bufindex = 0;
 	}
 
-	/* process partial buffer if there's enough data to make a block */
-	if (ctx->bufindex && len >= to_fill) {
+	/* process partial buffer if there's enough data to make a block
+	 * and there's without doubt further blocks */
+	if (ctx->bufindex && len > to_fill) {
 		memcpy(ctx->buf + ctx->bufindex, data, to_fill);
 		skein256_do_chunk(ctx, (uint64_t *) ctx->buf, 32);
 		len -= to_fill;
diff --git a/cbits/skein512.c b/cbits/skein512.c
--- a/cbits/skein512.c
+++ b/cbits/skein512.c
@@ -142,6 +142,9 @@
 {
 	uint32_t to_fill;
 
+	if (!len)
+		return;
+
 	to_fill = 64 - ctx->bufindex;
 
 	if (ctx->bufindex == 64) {
@@ -149,8 +152,9 @@
 		ctx->bufindex = 0;
 	}
 
-	/* process partial buffer if there's enough data to make a block */
-	if (ctx->bufindex && len >= to_fill) {
+	/* process partial buffer if there's enough data to make a block
+	 * and there's without doubt further blocks */
+	if (ctx->bufindex && len > to_fill) {
 		memcpy(ctx->buf + ctx->bufindex, data, to_fill);
 		skein512_do_chunk(ctx, (uint64_t *) ctx->buf, 64);
 		len -= to_fill;
diff --git a/cryptohash.cabal b/cryptohash.cabal
--- a/cryptohash.cabal
+++ b/cryptohash.cabal
@@ -1,5 +1,5 @@
 Name:                cryptohash
-Version:             0.11.2
+Version:             0.11.3
 Description:
     A collection of crypto hashes, with a practical incremental and one-pass, pure APIs,
     with performance close to the fastest implementations available in other languages.