Arena allocator

This allocator allocates a 4MiB arena into which all allocations are
made, and then increasingly larger arenas as earlier ones are used up.
Freeing memory in the arena is a no-op: clean all memory with
cmark_arena_reset().

In order to support realloc, we store the size of each allocation in a
size_t before the returned pointer.

The speedup is over 25% on large (benchmark-sized) inputs -- we pay a
small increase in maximum RSS (~10%) for this.

Author: Yuki Izumi

man/man3/cmark.3

@@ -115,6 +115,27 @@ typedef struct cmark_mem {
 Defines the memory allocation functions to be used by CMark when parsing
 and allocating a document tree
 
+.PP
+\fIcmark_mem *\f[] \fBcmark_get_default_mem_allocator\f[](\fI\f[])
+
+.PP
+The default memory allocator; uses the system's calloc, realloc and
+free.
+
+.PP
+\fIcmark_mem *\f[] \fBcmark_get_arena_mem_allocator\f[](\fI\f[])
+
+.PP
+An arena allocator; uses system calloc to allocate large slabs of
+memory. Memory in these slabs is not reused at all.
+
+.PP
+\fIvoid\f[] \fBcmark_arena_reset\f[](\fIvoid\f[])
+
+.PP
+Resets the arena allocator, quickly returning all used memory to the
+operating system.
+
 .SS
 Creating and Destroying Nodes
 
@@ -646,6 +667,13 @@ Rendering
 Render a \f[I]node\f[] tree as XML. It is the caller's responsibility to
 free the returned buffer.
 
+.PP
+\fIchar *\f[] \fBcmark_render_xml_with_mem\f[](\fIcmark_node *root\f[], \fIint options\f[], \fIcmark_mem *mem\f[])
+
+.PP
+As for \f[I]cmark_render_xml\f[], but specifying the allocator to use
+for the resulting string.
+
 .PP
 \fIchar *\f[] \fBcmark_render_html\f[](\fIcmark_node *root\f[], \fIint options\f[])
 
@@ -654,27 +682,55 @@ Render a \f[I]node\f[] tree as an HTML fragment. It is up to the user to
 add an appropriate header and footer. It is the caller's responsibility
 to free the returned buffer.
 
+.PP
+\fIchar *\f[] \fBcmark_render_html_with_mem\f[](\fIcmark_node *root\f[], \fIint options\f[], \fIcmark_mem *mem\f[])
+
+.PP
+As for \f[I]cmark_render_html\f[], but specifying the allocator to use
+for the resulting string.
+
 .PP
 \fIchar *\f[] \fBcmark_render_man\f[](\fIcmark_node *root\f[], \fIint options\f[], \fIint width\f[])
 
 .PP
 Render a \f[I]node\f[] tree as a groff man page, without the header. It
 is the caller's responsibility to free the returned buffer.
 
+.PP
+\fIchar *\f[] \fBcmark_render_man_with_mem\f[](\fIcmark_node *root\f[], \fIint options\f[], \fIint width\f[], \fIcmark_mem *mem\f[])
+
+.PP
+As for \f[I]cmark_render_man\f[], but specifying the allocator to use
+for the resulting string.
+
 .PP
 \fIchar *\f[] \fBcmark_render_commonmark\f[](\fIcmark_node *root\f[], \fIint options\f[], \fIint width\f[])
 
 .PP
 Render a \f[I]node\f[] tree as a commonmark document. It is the caller's
 responsibility to free the returned buffer.
 
+.PP
+\fIchar *\f[] \fBcmark_render_commonmark_with_mem\f[](\fIcmark_node *root\f[], \fIint options\f[], \fIint width\f[], \fIcmark_mem *mem\f[])
+
+.PP
+As for \f[I]cmark_render_commonmark\f[], but specifying the allocator to
+use for the resulting string.
+
 .PP
 \fIchar *\f[] \fBcmark_render_latex\f[](\fIcmark_node *root\f[], \fIint options\f[], \fIint width\f[])
 
 .PP
 Render a \f[I]node\f[] tree as a LaTeX document. It is the caller's
 responsibility to free the returned buffer.
 
+.PP
+\fIchar *\f[] \fBcmark_render_latex_with_mem\f[](\fIcmark_node *root\f[], \fIint options\f[], \fIint width\f[], \fIcmark_mem *mem\f[])
+
+.PP
+As for \f[I]cmark_render_latex\f[], but specifying the allocator to use
+for the resulting string.
+
 .SS
 Options
 

src/CMakeLists.txt

@@ -40,6 +40,7 @@ set(LIBRARY_SOURCES
   houdini_html_e.c
   houdini_html_u.c
   cmark_ctype.c
+  arena.c
   ${HEADERS}
   )
 
@@ -64,7 +65,7 @@ set_target_properties(${PROGRAM} PROPERTIES
   COMPILE_FLAGS -DCMARK_STATIC_DEFINE)
 
 # Check integrity of node structure when compiled as debug:
-set(CMAKE_C_FLAGS_DEBUG "${CMAKE_C_FLAGS_DEBUG} -DCMARK_DEBUG_NODES")
+set(CMAKE_C_FLAGS_DEBUG "${CMAKE_C_FLAGS_DEBUG} -DCMARK_DEBUG_NODES -DDEBUG")
 set(CMAKE_LINKER_DEBUG "${CMAKE_LINKER_FLAGS_DEBUG}")
 
 set(CMAKE_C_FLAGS_PROFILE "${CMAKE_C_FLAGS_RELEASE} -pg")

src/arena.c

@@ -0,0 +1,72 @@
+#include <stdlib.h>
+#include <string.h>
+#include <stdint.h>
+#include "cmark.h"
+
+static struct arena_chunk {
+  size_t sz, used;
+  void *ptr;
+  struct arena_chunk *prev;
+} *A = NULL;
+
+static struct arena_chunk *alloc_arena_chunk(size_t sz, struct arena_chunk *prev) {
+  struct arena_chunk *c = calloc(1, sizeof(*c));
+  if (!c)
+    abort();
+  c->sz = sz;
+  c->ptr = calloc(1, sz);
+  c->prev = prev;
+  return c;
+}
+
+static void init_arena(void) {
+  A = alloc_arena_chunk(4 * 1048576, NULL);
+}
+
+void cmark_arena_reset(void) {
+  while (A) {
+    free(A->ptr);
+    struct arena_chunk *n = A->prev;
+    free(A);
+    A = n;
+  }
+}
+
+static void *arena_calloc(size_t nmem, size_t size) {
+  if (!A)
+    init_arena();
+
+  size_t sz = nmem * size + sizeof(size_t);
+  if (sz > A->sz) {
+    A->prev = alloc_arena_chunk(sz, A->prev);
+    return (uint8_t *) A->prev->ptr + sizeof(size_t);
+  }
+  if (sz > A->sz - A->used) {
+    A = alloc_arena_chunk(A->sz + A->sz / 2, A);
+  }
+  void *ptr = (uint8_t *) A->ptr + A->used;
+  A->used += sz;
+  *((size_t *) ptr) = nmem * size;
+  return (uint8_t *) ptr + sizeof(size_t);
+}
+
+static void *arena_realloc(void *ptr, size_t size) {
+  if (!A)
+    init_arena();
+
+  void *new_ptr = arena_calloc(1, size);
+  if (ptr)
+    memcpy(new_ptr, ptr, ((size_t *) ptr)[-1]);
+  return new_ptr;
+}
+
+static void arena_free(void *ptr) {
+  (void) ptr;
+  /* no-op */
+}
+
+cmark_mem CMARK_ARENA_MEM_ALLOCATOR = {arena_calloc, arena_realloc, arena_free};
+
+cmark_mem *cmark_get_arena_mem_allocator() {
+  return &CMARK_ARENA_MEM_ALLOCATOR;
+}

src/blocks.c

@@ -108,8 +108,8 @@ cmark_parser *cmark_parser_new_with_mem(int options, cmark_mem *mem) {
 }
 
 cmark_parser *cmark_parser_new(int options) {
-  extern cmark_mem DEFAULT_MEM_ALLOCATOR;
-  return cmark_parser_new_with_mem(options, &DEFAULT_MEM_ALLOCATOR);
+  extern cmark_mem CMARK_DEFAULT_MEM_ALLOCATOR;
+  return cmark_parser_new_with_mem(options, &CMARK_DEFAULT_MEM_ALLOCATOR);
 }
 
 void cmark_parser_free(cmark_parser *parser) {

src/cmark.c

@@ -28,7 +28,11 @@ static void *xrealloc(void *ptr, size_t size) {
   return new_ptr;
 }
 
-cmark_mem DEFAULT_MEM_ALLOCATOR = {xcalloc, xrealloc, free};
+cmark_mem CMARK_DEFAULT_MEM_ALLOCATOR = {xcalloc, xrealloc, free};
+
+cmark_mem *cmark_get_default_mem_allocator() {
+  return &CMARK_DEFAULT_MEM_ALLOCATOR;
+}
 
 char *cmark_markdown_to_html(const char *text, size_t len, int options) {
   cmark_node *doc;

src/cmark.h

@@ -100,6 +100,24 @@ typedef struct cmark_mem {
   void (*free)(void *);
 } cmark_mem;
 
+/** The default memory allocator; uses the system's calloc,
+ * realloc and free.
+ */
+CMARK_EXPORT
+cmark_mem *cmark_get_default_mem_allocator();
+
+/** An arena allocator; uses system calloc to allocate large
+ * slabs of memory.  Memory in these slabs is not reused at all.
+ */
+CMARK_EXPORT
+cmark_mem *cmark_get_arena_mem_allocator();
+
+/** Resets the arena allocator, quickly returning all used memory
+ * to the operating system.
+ */
+CMARK_EXPORT
+void cmark_arena_reset(void);
+
 /**
  * ## Creating and Destroying Nodes
  */
@@ -507,31 +525,61 @@ cmark_node *cmark_parse_file(FILE *f, int options);
 CMARK_EXPORT
 char *cmark_render_xml(cmark_node *root, int options);
 
+/** As for 'cmark_render_xml', but specifying the allocator to use for
+ * the resulting string.
+ */
+CMARK_EXPORT
+char *cmark_render_xml_with_mem(cmark_node *root, int options, cmark_mem *mem);
+
 /** Render a 'node' tree as an HTML fragment.  It is up to the user
  * to add an appropriate header and footer. It is the caller's
  * responsibility to free the returned buffer.
  */
 CMARK_EXPORT
 char *cmark_render_html(cmark_node *root, int options);
 
+/** As for 'cmark_render_html', but specifying the allocator to use for
+ * the resulting string.
+ */
+CMARK_EXPORT
+char *cmark_render_html_with_mem(cmark_node *root, int options, cmark_mem *mem);
+
 /** Render a 'node' tree as a groff man page, without the header.
  * It is the caller's responsibility to free the returned buffer.
  */
 CMARK_EXPORT
 char *cmark_render_man(cmark_node *root, int options, int width);
 
+/** As for 'cmark_render_man', but specifying the allocator to use for
+ * the resulting string.
+ */
+CMARK_EXPORT
+char *cmark_render_man_with_mem(cmark_node *root, int options, int width, cmark_mem *mem);
+
 /** Render a 'node' tree as a commonmark document.
  * It is the caller's responsibility to free the returned buffer.
  */
 CMARK_EXPORT
 char *cmark_render_commonmark(cmark_node *root, int options, int width);
 
+/** As for 'cmark_render_commonmark', but specifying the allocator to use for
+ * the resulting string.
+ */
+CMARK_EXPORT
+char *cmark_render_commonmark_with_mem(cmark_node *root, int options, int width, cmark_mem *mem);
+
 /** Render a 'node' tree as a LaTeX document.
  * It is the caller's responsibility to free the returned buffer.
  */
 CMARK_EXPORT
 char *cmark_render_latex(cmark_node *root, int options, int width);
 
+/** As for 'cmark_render_latex', but specifying the allocator to use for
+ * the resulting string.
+ */
+CMARK_EXPORT
+char *cmark_render_latex_with_mem(cmark_node *root, int options, int width, cmark_mem *mem);
+
 /**
  * ## Options
  */

src/commonmark.c

@@ -466,10 +466,14 @@ static int S_render_node(cmark_renderer *renderer, cmark_node *node,
 }
 
 char *cmark_render_commonmark(cmark_node *root, int options, int width) {
+  return cmark_render_commonmark_with_mem(root, options, width, cmark_node_mem(root));
+}
+
+char *cmark_render_commonmark_with_mem(cmark_node *root, int options, int width, cmark_mem *mem) {
   if (options & CMARK_OPT_HARDBREAKS) {
     // disable breaking on width, since it has
     // a different meaning with OPT_HARDBREAKS
     width = 0;
   }
-  return cmark_render(root, options, width, outc, S_render_node);
+  return cmark_render(mem, root, options, width, outc, S_render_node);
 }

src/html.c

@@ -323,8 +323,12 @@ static int S_render_node(cmark_node *node, cmark_event_type ev_type,
 }
 
 char *cmark_render_html(cmark_node *root, int options) {
+  return cmark_render_html_with_mem(root, options, cmark_node_mem(root));
+}
+
+char *cmark_render_html_with_mem(cmark_node *root, int options, cmark_mem *mem) {
   char *result;
-  cmark_strbuf html = CMARK_BUF_INIT(cmark_node_mem(root));
+  cmark_strbuf html = CMARK_BUF_INIT(mem);
   cmark_event_type ev_type;
   cmark_node *cur;
   struct render_state state = {&html, NULL};

src/latex.c

@@ -449,5 +449,9 @@ static int S_render_node(cmark_renderer *renderer, cmark_node *node,
 }
 
 char *cmark_render_latex(cmark_node *root, int options, int width) {
-  return cmark_render(root, options, width, outc, S_render_node);
+  return cmark_render_latex_with_mem(root, options, width, cmark_node_mem(root));
+}
+
+char *cmark_render_latex_with_mem(cmark_node *root, int options, int width, cmark_mem *mem) {
+  return cmark_render(mem, root, options, width, outc, S_render_node);
 }