Faster huffman decoding

lib/nghttp2_hd_huffman.c

@@ -104,41 +104,37 @@ int nghttp2_hd_huff_encode(nghttp2_bufs *bufs, const uint8_t *src,
 }
 
 void nghttp2_hd_huff_decode_context_init(nghttp2_hd_huff_decode_context *ctx) {
-  ctx->state = 0;
+  ctx->fstate = NGHTTP2_HUFF_ACCEPTED;
-  ctx->accept = 1;
 }
 
 ssize_t nghttp2_hd_huff_decode(nghttp2_hd_huff_decode_context *ctx,
                                nghttp2_buf *buf, const uint8_t *src,
                                size_t srclen, int final) {
-  size_t i;
+  const uint8_t *end = src + srclen;
+  nghttp2_huff_decode node = {ctx->fstate, 0};
+  const nghttp2_huff_decode *t = &node;
+  uint8_t c;
 
   /* We use the decoding algorithm described in
      http://graphics.ics.uci.edu/pub/Prefix.pdf */
-  for (i = 0; i < srclen; ++i) {
+  for (; src != end;) {
-    const nghttp2_huff_decode *t;
+    c = *src++;
-
+    t = &huff_decode_table[t->fstate & 0x1ff][c >> 4];
-    t = &huff_decode_table[ctx->state][src[i] >> 4];
+    if (t->fstate & NGHTTP2_HUFF_SYM) {
-    if (t->flags & NGHTTP2_HUFF_FAIL) {
-      return NGHTTP2_ERR_HEADER_COMP;
-    }
-    if (t->flags & NGHTTP2_HUFF_SYM) {
       *buf->last++ = t->sym;
     }
 
-    t = &huff_decode_table[t->state][src[i] & 0xf];
+    t = &huff_decode_table[t->fstate & 0x1ff][c & 0xf];
-    if (t->flags & NGHTTP2_HUFF_FAIL) {
+    if (t->fstate & NGHTTP2_HUFF_SYM) {
-      return NGHTTP2_ERR_HEADER_COMP;
-    }
-    if (t->flags & NGHTTP2_HUFF_SYM) {
       *buf->last++ = t->sym;
     }
-
-    ctx->state = t->state;
-    ctx->accept = (t->flags & NGHTTP2_HUFF_ACCEPTED) != 0;
   }
-  if (final && !ctx->accept) {
+
+  ctx->fstate = t->fstate;
+
+  if (final && !(ctx->fstate & NGHTTP2_HUFF_ACCEPTED)) {
     return NGHTTP2_ERR_HEADER_COMP;
   }
-  return (ssize_t)i;
+
+  return (ssize_t)srclen;
 }

lib/nghttp2_hd_huffman.h

@@ -34,21 +34,20 @@
 typedef enum {
   /* FSA accepts this state as the end of huffman encoding
      sequence. */
-  NGHTTP2_HUFF_ACCEPTED = 1,
+  NGHTTP2_HUFF_ACCEPTED = 1 << 14,
   /* This state emits symbol */
-  NGHTTP2_HUFF_SYM = (1 << 1),
+  NGHTTP2_HUFF_SYM = 1 << 15,
-  /* If state machine reaches this state, decoding fails. */
-  NGHTTP2_HUFF_FAIL = (1 << 2)
 } nghttp2_huff_decode_flag;
 
 typedef struct {
-  /* huffman decoding state, which is actually the node ID of internal
+  /* fstate is the current huffman decoding state, which is actually
-     huffman tree.  We have 257 leaf nodes, but they are identical to
+     the node ID of internal huffman tree with
-     root node other than emitting a symbol, so we have 256 internal
+     nghttp2_huff_decode_flag OR-ed.  We have 257 leaf nodes, but they
-     nodes [1..255], inclusive. */
+     are identical to root node other than emitting a symbol, so we
-  uint8_t state;
+     have 256 internal nodes [1..255], inclusive.  The node ID 256 is
-  /* bitwise OR of zero or more of the nghttp2_huff_decode_flag */
+     a special node and it is a terminal state that means decoding
-  uint8_t flags;
+     failed. */
+  uint16_t fstate;
   /* symbol if NGHTTP2_HUFF_SYM flag set */
   uint8_t sym;
 } nghttp2_huff_decode;
@@ -56,12 +55,8 @@ typedef struct {
 typedef nghttp2_huff_decode huff_decode_table_type[16];
 
 typedef struct {
-  /* Current huffman decoding state. We stripped leaf nodes, so the
+  /* fstate is the current huffman decoding state. */
-     value range is [0..255], inclusive. */
+  uint16_t fstate;
-  uint8_t state;
-  /* nonzero if we can say that the decoding process succeeds at this
-     state */
-  uint8_t accept;
 } nghttp2_hd_huff_decode_context;
 
 typedef struct {

mkhufftbl.py

@@ -357,9 +357,8 @@ def _build_transition_table(ctx, node):
 def huffman_tree_build_transition_table(ctx):
     _build_transition_table(ctx, ctx.root)
 
-NGHTTP2_HUFF_ACCEPTED = 1
+NGHTTP2_HUFF_ACCEPTED = 1 << 14
-NGHTTP2_HUFF_SYM = 1 << 1
+NGHTTP2_HUFF_SYM = 1 << 15
-NGHTTP2_HUFF_FAIL = 1 << 2
 
 def _print_transition_table(node):
     if node.term is not None:
@@ -374,8 +373,7 @@ def _print_transition_table(node):
             out = sym
             flags |= NGHTTP2_HUFF_SYM
         if nd is None:
-            id = 0
+            id = 256
-            flags |= NGHTTP2_HUFF_FAIL
         else:
             id = nd.id
             if id is None:
@@ -384,13 +382,32 @@ def _print_transition_table(node):
                 flags |= NGHTTP2_HUFF_ACCEPTED
             elif nd.accept:
                 flags |= NGHTTP2_HUFF_ACCEPTED
-        print '  {{{}, 0x{:02x}, {}}},'.format(id, flags, out)
+        print '  {{0x{:02x}, {}}},'.format(id | flags, out)
     print '},'
     _print_transition_table(node.left)
     _print_transition_table(node.right)
 
 def huffman_tree_print_transition_table(ctx):
     _print_transition_table(ctx.root)
+    print '/* 256 */'
+    print '{'
+    print '  {0x100, 0},'
+    print '  {0x100, 0},'
+    print '  {0x100, 0},'
+    print '  {0x100, 0},'
+    print '  {0x100, 0},'
+    print '  {0x100, 0},'
+    print '  {0x100, 0},'
+    print '  {0x100, 0},'
+    print '  {0x100, 0},'
+    print '  {0x100, 0},'
+    print '  {0x100, 0},'
+    print '  {0x100, 0},'
+    print '  {0x100, 0},'
+    print '  {0x100, 0},'
+    print '  {0x100, 0},'
+    print '  {0x100, 0},'
+    print '},'
 
 if __name__ == '__main__':
     ctx = Context()
@@ -436,14 +453,12 @@ const nghttp2_huff_sym huff_sym_table[] = {'''
 enum {{
   NGHTTP2_HUFF_ACCEPTED = {},
   NGHTTP2_HUFF_SYM = {},
-  NGHTTP2_HUFF_FAIL = {},
 }} nghttp2_huff_decode_flag;
-'''.format(NGHTTP2_HUFF_ACCEPTED, NGHTTP2_HUFF_SYM, NGHTTP2_HUFF_FAIL)
+'''.format(NGHTTP2_HUFF_ACCEPTED, NGHTTP2_HUFF_SYM)
 
     print '''\
 typedef struct {
-  uint8_t state;
+  uint16_t fstate;
-  uint8_t flags;
   uint8_t sym;
 } nghttp2_huff_decode;
 '''

tests/main.c

@@ -402,6 +402,7 @@ int main() {
                    test_nghttp2_hd_deflate_hd_vec) ||
       !CU_add_test(pSuite, "hd_decode_length", test_nghttp2_hd_decode_length) ||
       !CU_add_test(pSuite, "hd_huff_encode", test_nghttp2_hd_huff_encode) ||
+      !CU_add_test(pSuite, "hd_huff_decode", test_nghttp2_hd_huff_decode) ||
       !CU_add_test(pSuite, "adjust_local_window_size",
                    test_nghttp2_adjust_local_window_size) ||
       !CU_add_test(pSuite, "check_header_name",

tests/nghttp2_hd_test.c

@@ -1538,3 +1538,32 @@ void test_nghttp2_hd_huff_encode(void) {
 
   nghttp2_bufs_free(&bufs);
 }
+
+void test_nghttp2_hd_huff_decode(void) {
+  const uint8_t e[] = {0x1f, 0xff, 0xff, 0xff, 0xff, 0xff};
+  nghttp2_hd_huff_decode_context ctx;
+  nghttp2_buf outbuf;
+  uint8_t b[256];
+  ssize_t len;
+
+  nghttp2_buf_wrap_init(&outbuf, b, sizeof(b));
+  nghttp2_hd_huff_decode_context_init(&ctx);
+  len = nghttp2_hd_huff_decode(&ctx, &outbuf, e, 1, 1);
+
+  CU_ASSERT(1 == len);
+  CU_ASSERT(0 == memcmp("a", outbuf.pos, 1));
+
+  /* Premature sequence must elicit decoding error */
+  nghttp2_buf_wrap_init(&outbuf, b, sizeof(b));
+  nghttp2_hd_huff_decode_context_init(&ctx);
+  len = nghttp2_hd_huff_decode(&ctx, &outbuf, e, 2, 1);
+
+  CU_ASSERT(NGHTTP2_ERR_HEADER_COMP == len);
+
+  /* Fully decoding EOS is error */
+  nghttp2_buf_wrap_init(&outbuf, b, sizeof(b));
+  nghttp2_hd_huff_decode_context_init(&ctx);
+  len = nghttp2_hd_huff_decode(&ctx, &outbuf, e, 2, 6);
+
+  CU_ASSERT(NGHTTP2_ERR_HEADER_COMP == len);
+}

tests/nghttp2_hd_test.h

@@ -50,5 +50,6 @@ void test_nghttp2_hd_public_api(void);
 void test_nghttp2_hd_deflate_hd_vec(void);
 void test_nghttp2_hd_decode_length(void);
 void test_nghttp2_hd_huff_encode(void);
+void test_nghttp2_hd_huff_decode(void);
 
 #endif /* NGHTTP2_HD_TEST_H */