HPACK huffman decode: Process 8 bits at a time

Makefile.am

@@ -43,6 +43,8 @@ EXTRA_DIST = nghttpx.conf.sample proxy.pac.sample android-config android-make \
 clang-format:
 	CLANGFORMAT=`git config --get clangformat.binary`; \
 	test -z $${CLANGFORMAT} && CLANGFORMAT="clang-format"; \
-	$${CLANGFORMAT} -i lib/*.{c,h} lib/includes/nghttp2/*.h \
+	$${CLANGFORMAT} -i \
+	`ls lib/*.{c,h} | grep -v nghttp2_hd_huffman_data.c` \
+	lib/includes/nghttp2/*.h \
 	src/*.{c,cc,h} src/includes/nghttp2/*.h examples/*.{c,cc} \
 	tests/*.{c,h}

lib/nghttp2_hd_huffman.c

@@ -197,22 +197,16 @@ ssize_t nghttp2_hd_huff_decode(nghttp2_hd_huff_decode_context *ctx,
   for (i = 0; i < srclen; ++i) {
     const nghttp2_huff_decode *t;
 
-    t = &huff_decode_table[ctx->state][src[i] >> 4];
+    t = &huff_decode_table[ctx->state][src[i]];
     if (t->flags & NGHTTP2_HUFF_FAIL) {
       return NGHTTP2_ERR_HEADER_COMP;
     }
-    if (t->flags & NGHTTP2_HUFF_SYM) {
+    if (t->flags & NGHTTP2_HUFF_SYM1) {
       /* this is macro, and may return from this function on error */
-      hd_huff_decode_sym_emit(bufs, t->sym, avail);
+      hd_huff_decode_sym_emit(bufs, t->sym[0], avail);
+      if (t->flags & NGHTTP2_HUFF_SYM2) {
+        hd_huff_decode_sym_emit(bufs, t->sym[1], avail);
       }
-
-    t = &huff_decode_table[t->state][src[i] & 0xf];
-    if (t->flags & NGHTTP2_HUFF_FAIL) {
-      return NGHTTP2_ERR_HEADER_COMP;
-    }
-    if (t->flags & NGHTTP2_HUFF_SYM) {
-      /* this is macro, and may return from this function on error */
-      hd_huff_decode_sym_emit(bufs, t->sym, avail);
     }
 
     ctx->state = t->state;

lib/nghttp2_hd_huffman.h

@@ -35,10 +35,12 @@ typedef enum {
   /* FSA accepts this state as the end of huffman encoding
      sequence. */
   NGHTTP2_HUFF_ACCEPTED = 1,
-  /* This state emits symbol */
-  NGHTTP2_HUFF_SYM = (1 << 1),
+  /* This state emits 1st symbol */
+  NGHTTP2_HUFF_SYM1 = (1 << 1),
+  /* This state emits 2nd symbol */
+  NGHTTP2_HUFF_SYM2 = (1 << 2),
   /* If state machine reaches this state, decoding fails. */
-  NGHTTP2_HUFF_FAIL = (1 << 2)
+  NGHTTP2_HUFF_FAIL = (1 << 3)
 } nghttp2_huff_decode_flag;
 
 typedef struct {
@@ -49,11 +51,15 @@ typedef struct {
   uint8_t state;
   /* bitwise OR of zero or more of the nghttp2_huff_decode_flag */
   uint8_t flags;
-  /* symbol if NGHTTP2_HUFF_SYM flag set */
-  uint8_t sym;
+  /* symbols if NGHTTP2_HUFF_SYM1 and optionally NGHTTP2_HUFF_SYM2
+     flag set.  If NGHTTP2_HUFF_SYM1 is set, sym[0] has the 1st
+     symbol.  Additionally, NGHTTP2_HUFF_SYM2 is set, sym[1] has the
+     2nd symbol.  Since maximum huffman code is 5 bits, we may get at
+     most 2 symbols in one transition. */
+  uint8_t sym[2];
 } nghttp2_huff_decode;
 
-typedef nghttp2_huff_decode huff_decode_table_type[16];
+typedef nghttp2_huff_decode huff_decode_table_type[256];
 
 typedef struct {
   /* Current huffman decoding state. We stripped leaf nodes, so the
@@ -72,6 +78,6 @@ typedef struct {
 } nghttp2_huff_sym;
 
 extern const nghttp2_huff_sym huff_sym_table[];
-extern const nghttp2_huff_decode huff_decode_table[][16];
+extern const nghttp2_huff_decode huff_decode_table[][256];
 
 #endif /* NGHTTP2_HD_HUFFMAN_H */

mkhufftbl.py

@@ -324,12 +324,12 @@ def _set_node_id(ctx, node, prefix):
 def huffman_tree_set_node_id(ctx):
     _set_node_id(ctx, ctx.root, [])
 
-def _traverse(node, sym, start_node, root, left):
+def _traverse(node, syms, start_node, root, left):
     if left == 0:
-        if sym == 256:
-            sym = None
+        if syms and syms[0] == 256:
+            syms = []
             node = None
-        start_node.trans.append((node, sym))
+        start_node.trans.append((node, syms))
         return
 
     if node.term is not None:
@@ -337,12 +337,12 @@ def _traverse(node, sym, start_node, root, left):
 
     def go(node):
         if node.term is not None:
-            assert sym is None
-            nsym = node.term
+            assert len(syms) <= 1
+            nsyms = syms + [node.term]
         else:
-            nsym = sym
+            nsyms = syms
 
-        _traverse(node, nsym, start_node, root, left - 1)
+        _traverse(node, nsyms, start_node, root, left - 1)
 
     go(node.left)
     go(node.right)
@@ -350,7 +350,7 @@ def _traverse(node, sym, start_node, root, left):
 def _build_transition_table(ctx, node):
     if node is None:
         return
-    _traverse(node, None, node, ctx.root, 4)
+    _traverse(node, [], node, ctx.root, 8)
     _build_transition_table(ctx, node.left)
     _build_transition_table(ctx, node.right)
 
@@ -358,21 +358,26 @@ def huffman_tree_build_transition_table(ctx):
     _build_transition_table(ctx, ctx.root)
 
 NGHTTP2_HUFF_ACCEPTED = 1
-NGHTTP2_HUFF_SYM = 1 << 1
-NGHTTP2_HUFF_FAIL = 1 << 2
+NGHTTP2_HUFF_SYM1 = 1 << 1
+NGHTTP2_HUFF_SYM2 = 1 << 2
+NGHTTP2_HUFF_FAIL = 1 << 3
 
 def _print_transition_table(node):
     if node.term is not None:
         return
     print '    /* {} */'.format(node.id)
     print '    {'
-    for nd, sym in node.trans:
+    for nd, syms in node.trans:
         flags = 0
-        if sym is None:
-            out = 0
+        if len(syms) == 0:
+            out = [0, 0]
         else:
-            out = sym
-            flags |= NGHTTP2_HUFF_SYM
+            out = syms
+            flags |= NGHTTP2_HUFF_SYM1
+            if len(out) == 2:
+                flags |= NGHTTP2_HUFF_SYM2
+            else:
+                out = out + [0]
         if nd is None:
             id = 0
             flags |= NGHTTP2_HUFF_FAIL
@@ -384,7 +389,7 @@ 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[0], out[1])
     print '    },'
     _print_transition_table(node.left)
     _print_transition_table(node.right)
@@ -432,20 +437,21 @@ const nghttp2_huff_sym huff_sym_table[] = {'''
     print '''\
 enum {{
   NGHTTP2_HUFF_ACCEPTED = {},
-  NGHTTP2_HUFF_SYM = {},
+  NGHTTP2_HUFF_SYM1 = {},
+  NGHTTP2_HUFF_SYM2 = {},
   NGHTTP2_HUFF_FAIL = {},
 }} nghttp2_huff_decode_flag;
-'''.format(NGHTTP2_HUFF_ACCEPTED, NGHTTP2_HUFF_SYM, NGHTTP2_HUFF_FAIL)
+'''.format(NGHTTP2_HUFF_ACCEPTED, NGHTTP2_HUFF_SYM1, NGHTTP2_HUFF_SYM2, NGHTTP2_HUFF_FAIL)
 
     print '''\
 typedef struct {
   uint8_t state;
   uint8_t flags;
-  uint8_t sym;
+  uint8_t sym[2];
 } nghttp2_huff_decode;
 '''
 
     print '''\
-const nghttp2_huff_decode huff_decode_table[][16] = {'''
+const nghttp2_huff_decode huff_decode_table[][256] = {'''
     huffman_tree_print_transition_table(ctx)
     print '};'