From c54547dad90de7f05ceaf590f017f0f7d76f44d8 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Th=C3=A9ophile=20Bastian?= <contact@tobast.fr>
Date: Thu, 26 Sep 2019 17:22:40 +0200
Subject: [PATCH] Add asm acquisition boilerplate

---
 src/asm_acquire.ml | 131 +++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 131 insertions(+)
 create mode 100644 src/asm_acquire.ml

diff --git a/src/asm_acquire.ml b/src/asm_acquire.ml
new file mode 100644
index 0000000..fb3c21b
--- /dev/null
+++ b/src/asm_acquire.ml
@@ -0,0 +1,131 @@
+(** Module to acquire the ASM of a given ELF object by its path
+
+    Uses `objdump -d` internally, and parses the output
+*)
+
+module StrMap = Map.Make(String)
+
+(** A memory address *)
+type addr_t = int
+
+(** A single asm instruction *)
+type asm_instr_t = {
+  instr_addr:   addr_t;   (** Memory location of this instruction *)
+  instr_bytes:  Bytes.t;  (** Binary representation of the instruction *)
+  instr_asm:    string;   (** Asm for the instruction (eg `movq …`) *)
+}
+type asm_t = asm_instr_t list
+type asm_sub_t = {
+  sub_section:  string;
+  sub_name:     string;
+  sub_addr:     addr_t;
+  sub_asm:      asm_t;
+}
+type asm_info_t = asm_sub_t StrMap.t
+
+exception ParseError of string
+
+(** Pretty printers *)
+let pp_hex_bytes ppx bytes_array =
+  Bytes.iter (fun byte -> Format.fprintf ppx "%02x" (int_of_char byte))
+    bytes_array
+
+let pp_asm_instr ppx asm_instr =
+  Format.fprintf ppx "%04x:\t%a\t%s@."
+    asm_instr.instr_addr
+    pp_hex_bytes asm_instr.instr_bytes
+    asm_instr.instr_asm
+
+let pp_asm ppx asm_instrs = List.iter (pp_asm_instr ppx) asm_instrs
+
+let pp_asm_sub ppx asm_sub =
+  Format.fprintf ppx "%016x <%s> {%s}:@.@[%a@]@."
+    asm_sub.sub_addr
+    asm_sub.sub_name
+    asm_sub.sub_section
+    pp_asm asm_sub.sub_asm
+
+let pp_asm_info ppx asm_info =
+  StrMap.iter (fun _ asm_sub -> Format.fprintf ppx "%a" pp_asm_sub asm_sub)
+    asm_info
+
+(** Reads the whole content of a Unix file descriptor, returning it as a Bytes
+    sequence *)
+let read_all_fd fd =
+  Unix.set_nonblock fd ;
+
+  let rec do_read accu cur_size =
+    let n_accu = Bytes.extend accu 0 256 in
+    let bytes_read = (
+      try
+        Unix.read fd n_accu cur_size 256
+      with
+      | Unix.Unix_error(Unix.EAGAIN, _, _)
+      | Unix.Unix_error(Unix.EWOULDBLOCK, _, _) ->
+        0
+    ) in
+
+    if bytes_read = 256 then
+      do_read n_accu (cur_size + 256)
+    else (
+      Bytes.extend n_accu 0 (bytes_read - 256)
+    )
+  in
+  do_read (Bytes.create 0) 0
+
+
+(** Raised when the call to objdump failed. An exception
+   [ObjdumpFailed (status_code, stderr)] will contain the status code of the
+   objdump process, and its stderr contents as a string. *)
+exception ObjdumpFailed of int * string
+
+
+(** Runs `objdump -d prog_path` and returns its stdout as a string *)
+let get_objdump prog_path =
+  (* Setup process *)
+  let stdin_read, stdin = Unix.pipe () in
+  let stdout, stdout_write = Unix.pipe () in
+  let stderr, stderr_write = Unix.pipe () in
+
+  (* Run the process *)
+  Format.eprintf "Running objdump...@." ;
+  let objdump_pid = Unix.create_process
+      "objdump"
+      [| "objdump" ; "-d" ; prog_path |]
+      stdin_read stdout_write stderr_write
+  in
+  (* Close stdin already: we won't send anything and make it clear *)
+  Unix.close stdin ;
+  Format.eprintf "\tPID = %d@." objdump_pid;
+
+  let pid, status = Unix.waitpid [] objdump_pid in
+  if pid <> objdump_pid then
+    raise (Failure ("Could not properly wait on objdump")) ;
+
+  Format.eprintf "Run objdump: done. Status: %a@." (fun ppx sc -> match sc with
+      | Unix.WEXITED ex_code -> Format.fprintf ppx "%d" ex_code
+      | _ -> Format.fprintf ppx "Not exited") status ;
+
+  (* Extract stdout, stderr data *)
+  let stdout_content = read_all_fd stdout in
+  let stderr_content = read_all_fd stderr in
+
+  (* Properly close pipes *)
+  Unix.close stdout ;
+  Unix.close stderr ;
+
+  (* Check status code *)
+  match status with
+  | Unix.WEXITED 0 ->
+    Bytes.to_string stdout_content
+  | Unix.WEXITED err_code ->
+    raise (ObjdumpFailed (err_code, (Bytes.to_string stderr_content)))
+  | _ ->
+    raise (ObjdumpFailed (-1, (Bytes.to_string stderr_content)))
+
+(** Interprets `objdump -d` output and yield a list of functions, alongside
+    with their addresses, symbol names, asm, … *)
+let interpret_objdump objdump_out : asm_info_t =
+  (* TODO *)
+  ignore objdump_out;
+  assert false