Rewriting the Digest module causes linking errors

[Goswin von Brederlow <goswin-v-b@web.de>]

Hi,

I want to rewrite the Digest module to expose a more lowlevel interface
to the md5 digest and add support to digest Bigarrays. I've patched the
respective files involved and it all looks alright but when I try to
build ocaml I get the following error:

File "_none_", line 1, characters 0-1:
Error: Error while linking boot/stdlib.cma(Digest):
The external function `caml_md5_update_string' is not available
make[2]: *** [ocamlc] Error 2
make[2]: Leaving directory `/home/mrvn/src/debian/ocaml/ocaml-3.11.2'
make[1]: *** [world] Error 2
make[1]: Leaving directory `/home/mrvn/src/debian/ocaml/ocaml-3.11.2'
make: *** [build-stamp] Error 2

Can anyone explain why the new functions are not available?

MfG
        Goswin

PS: The patch is based on Debians 3.11.2-1 source and the patch below
can be added to debian/patches/ as is.

--
From: Goswin von Brederlow <goswin-v-b@web.de>
Date: Wed, 17 Mar 2010 05:56:56 +0200
Subject: [PATCH] Rewrite digest module

---
 byterun/md5.c                       |   30 ++++++++++++++++++++++++++++++
 byterun/md5.h                       |    6 +++++-
 otherlibs/bigarray/bigarray.h       |    2 ++
 otherlibs/bigarray/bigarray.ml      |   11 +++++++++++
 otherlibs/bigarray/bigarray.mli     |   10 ++++++++++
 otherlibs/bigarray/bigarray_stubs.c |   13 +++++++++++++
 stdlib/digest.ml                    |   13 +++++++++++++
 stdlib/digest.mli                   |   28 ++++++++++++++++++++++++++++
 8 files changed, 112 insertions(+), 1 deletion(-)

Index: ocaml-3.11.2/stdlib/digest.ml
===================================================================
--- ocaml-3.11.2.orig/stdlib/digest.ml	2010-03-17 08:17:24.000000000 +0100
+++ ocaml-3.11.2/stdlib/digest.ml	2010-03-17 08:19:46.000000000 +0100
@@ -16,8 +16,13 @@
 (* Message digest (MD5) *)
 
 type t = string
+type context
 
+external context: unit -> context = "caml_md5_context"
 external unsafe_string: string -> int -> int -> t = "caml_md5_string"
+external unsafe_update_string: context -> string -> int -> int -> unit
+  = "caml_md5_update_string"
+external final: context -> t = "caml_md5_final"
 external channel: in_channel -> int -> t = "caml_md5_chan"
 
 let string str =
@@ -28,6 +33,14 @@
   then invalid_arg "Digest.substring"
   else unsafe_string str ofs len
 
+let update_string context str =
+  unsafe_update_string context str 0 (String.length str)
+
+let update_substring context str ofs len =
+  if ofs < 0 || len < 0 || ofs > String.length str - len
+  then invalid_arg "Digest.update_substring"
+  else unsafe_update_string context str ofs len
+
 let file filename =
   let ic = open_in_bin filename in
   let d = channel ic (-1) in
Index: ocaml-3.11.2/stdlib/digest.mli
===================================================================
--- ocaml-3.11.2.orig/stdlib/digest.mli	2010-03-17 08:17:24.000000000 +0100
+++ ocaml-3.11.2/stdlib/digest.mli	2010-03-17 08:19:56.000000000 +0100
@@ -24,6 +24,34 @@
 type t = string
 (** The type of digests: 16-character strings. *)
 
+type context
+(** The type of a digest context. *)
+
+external context : unit -> context = "caml_md5_context"
+(** Return a fresh digest context. *)
+
+external unsafe_update_string : context -> string -> int -> int -> unit
+  = "caml_md5_update_string"
+(** [Digest.unsafe_update_string ctx s ofs len] updates the context
+    to include the substring of [s] starting at character number [ofs]
+    and containing [len] characters. *)
+
+external unsafe_string: string -> int -> int -> t = "caml_md5_string"
+(** [Digest.unsafe_string s ofs len] returns the digest of the
+    substring of [s] starting at character number [ofs] and containing
+    [len] characters. *)
+
+val update_string : context -> string -> unit
+(** [Digest.update_string ctx s ] updates the context to include [s]. *)
+
+val update_substring : context -> string -> int -> int -> unit
+(** [Digest.update_substring ctx s ofs len] updates the context
+    to include the substring of [s] starting at character number [ofs]
+    and containing [len] characters. *)
+
+external final : context -> t = "caml_md5_final"
+(** [Digest.final ctx] computs the final digest from [ctx]. *)
+
 val string : string -> t
 (** Return the digest of the given string. *)
 
Index: ocaml-3.11.2/byterun/md5.c
===================================================================
--- ocaml-3.11.2.orig/byterun/md5.c	2010-03-17 08:17:24.000000000 +0100
+++ ocaml-3.11.2/byterun/md5.c	2010-03-17 08:17:43.000000000 +0100
@@ -24,6 +24,36 @@
 
 /* MD5 message digest */
 
+CAMLprim value caml_md5_context(void)
+{
+  value context;
+  struct MD5Context *ctx = malloc(sizeof(struct MD5Context));
+  if (ctx == NULL) {
+    caml_raise_out_of_memory();
+  }
+  caml_MD5Init(ctx);
+  context = caml_alloc_small(1, Abstract_tag);
+  Store_field(context, 0, (value)ctx);
+  return context;
+}
+
+CAMLprim value caml_md5_update_string(value context,
+				      value str, value ofs, value len)
+{
+  struct MD5Context *ctx = (struct MD5Context *)Data_custom_val(context);
+  caml_MD5Update(ctx, &Byte_u(str, Long_val(ofs)), Long_val(len));
+  return Val_unit;
+}
+
+CAMLprim value caml_md5_final(value context)
+{
+  struct MD5Context ctx = *(struct MD5Context *)Data_custom_val(context);
+  value res;
+  res = caml_alloc_string(16);
+  caml_MD5Final(&Byte_u(res, 0), &ctx);
+  return res;
+}
+
 CAMLprim value caml_md5_string(value str, value ofs, value len)
 {
   struct MD5Context ctx;
Index: ocaml-3.11.2/byterun/md5.h
===================================================================
--- ocaml-3.11.2.orig/byterun/md5.h	2010-03-17 08:17:24.000000000 +0100
+++ ocaml-3.11.2/byterun/md5.h	2010-03-17 08:17:43.000000000 +0100
@@ -22,6 +22,10 @@
 #include "mlvalues.h"
 #include "io.h"
 
+CAMLextern value caml_md5_context (void);
+CAMLextern value caml_md5_update_string (value context,
+					 value str, value ofs, value len);
+CAMLextern value caml_md5_final (value context);
 CAMLextern value caml_md5_string (value str, value ofs, value len);
 CAMLextern value caml_md5_chan (value vchan, value len);
 
@@ -32,7 +36,7 @@
 };
 
 CAMLextern void caml_MD5Init (struct MD5Context *context);
-CAMLextern void caml_MD5Update (struct MD5Context *context, unsigned char *buf, 
+CAMLextern void caml_MD5Update (struct MD5Context *context, unsigned char *buf,
                                 uintnat len);
 CAMLextern void caml_MD5Final (unsigned char *digest, struct MD5Context *ctx);
 CAMLextern void caml_MD5Transform (uint32 *buf, uint32 *in);
Index: ocaml-3.11.2/otherlibs/bigarray/bigarray.h
===================================================================
--- ocaml-3.11.2.orig/otherlibs/bigarray/bigarray.h	2010-03-17 08:17:43.000000000 +0100
+++ ocaml-3.11.2/otherlibs/bigarray/bigarray.h	2010-03-17 08:17:43.000000000 +0100
@@ -93,4 +93,6 @@
                                  ... /*dimensions, with type intnat */);
 CAMLBAextern uintnat caml_ba_byte_size(struct caml_ba_array * b);
 
+CAMLprim value caml_md5_update_bigarray (value context, value array);
+
 #endif
Index: ocaml-3.11.2/otherlibs/bigarray/bigarray.ml
===================================================================
--- ocaml-3.11.2.orig/otherlibs/bigarray/bigarray.ml	2010-03-17 08:17:43.000000000 +0100
+++ ocaml-3.11.2/otherlibs/bigarray/bigarray.ml	2010-03-17 08:17:43.000000000 +0100
@@ -124,6 +124,17 @@
     ba
   let map_file fd ?pos kind layout shared dim =
     Genarray.map_file fd ?pos kind layout shared [|dim|]
+
+  module Digest = struct
+    external update_bigarray: context -> ('a, 'b, 'c) Bigarray.Array1.t -> unit
+      = "caml_md5_update_bigarray" "noalloc"
+
+    let bigarray arr =
+      let context = context () in
+      let () = update_bigarray context arr
+      in
+	final context
+  end
 end
 
 module Array2 = struct
Index: ocaml-3.11.2/otherlibs/bigarray/bigarray.mli
===================================================================
--- ocaml-3.11.2.orig/otherlibs/bigarray/bigarray.mli	2010-03-17 08:17:43.000000000 +0100
+++ ocaml-3.11.2/otherlibs/bigarray/bigarray.mli	2010-03-17 08:17:43.000000000 +0100
@@ -507,6 +507,16 @@
       Use with caution and only when the program logic guarantees that
       the access is within bounds. *)
 
+  module Digest : sig
+    external update_bigarray : context -> ('a, 'b, 'c) t -> unit
+      = "caml_md5_update_bigarray" "noalloc"
+    (** Updates the context to include the bigarray. This function
+	runs concurrent with other threads. *)
+
+    val bigarray : ('a, 'b, 'c) t -> Digest.t
+    (** Return the digest of the given bigarray. This function
+	runs concurrent with other threads. *)
+  end
 end
 
 
Index: ocaml-3.11.2/otherlibs/bigarray/bigarray_stubs.c
===================================================================
--- ocaml-3.11.2.orig/otherlibs/bigarray/bigarray_stubs.c	2010-03-17 08:17:43.000000000 +0100
+++ ocaml-3.11.2/otherlibs/bigarray/bigarray_stubs.c	2010-03-17 08:17:43.000000000 +0100
@@ -23,6 +23,8 @@
 #include "intext.h"
 #include "memory.h"
 #include "mlvalues.h"
+#include "md5.h"
+#include "signals.h"
 
 #define int8 caml_ba_int8
 #define uint8 caml_ba_uint8
@@ -1097,3 +1099,14 @@
   caml_register_custom_operations(&caml_ba_ops);
   return Val_unit;
 }
+
+CAMLprim value caml_md5_update_bigarray(value context, value array)
+{
+  struct MD5Context *ctx = (struct MD5Context *)Data_custom_val(context);
+  void* data = Caml_ba_data_val(array);
+  size_t len = Caml_ba_byte_size(Caml_ba_array_val(array));
+  caml_leave_blocking_section();
+  caml_MD5Update(ctx, data, len);
+  caml_enter_blocking_section();
+  return Val_unit;
+}