Ensure a crashing FDE still generates other FDEs

DwarfSynth/Main.ml

@@ -1,7 +1,9 @@
 open Std
 
 let main outfile proj =
-  let pre_dwarf = Simplest.of_proj proj in
+  let pre_dwarf = proj
+                  |> Simplest.of_proj
+                  |> Simplest.clean_lost_track_subs in
   Format.printf "%a" Frontend.pp_pre_dwarf_readelf pre_dwarf;
   let pre_c_dwarf = PreCBinding.convert_pre_c pre_dwarf in
   let fd = open_out_bin outfile in

DwarfSynth/PreCBinding.ml

@@ -1,6 +1,8 @@
 open Simplest
 open Std
 
+exception InvalidPreDwarf of string
+
 type pre_c_pre_dwarf_entry = {
   location: int64;
   cfa_offset: int64;
@@ -58,7 +60,9 @@ let convert_pre_c_entry loc entry : pre_c_pre_dwarf_entry =
   let offset, offset_reg = (match entry with
       | RspOffset off -> off, 7
       | RbpOffset off -> off, 6
-      | CfaLostTrack -> assert false (* Should be filtered out beforehand *)
+      | CfaLostTrack ->
+        raise (InvalidPreDwarf
+                 ("CfaLostTrack should be filtered out beforehand"))
     ) in
   {
     location = loc;
@@ -72,22 +76,33 @@ let convert_pre_c_entries entries : pre_c_pre_dwarf_entry array =
       (fun loc entry _ -> convert_pre_c_entry loc entry)
       entries empty_entry
 
-let convert_pre_c_fde name entry id : pre_c_pre_dwarf_fde =
+let convert_pre_c_fde name entry : pre_c_pre_dwarf_fde option =
-  {
+  try
+    Some {
       num = AddrMap.cardinal entry.cfa_changes_fde;
       initial_location = entry.beg_pos;
       end_location = entry.end_pos;
       name = name;
       entries = convert_pre_c_entries entry.cfa_changes_fde
     }
+  with InvalidPreDwarf reason -> (
+      Format.eprintf "FAILED subroutine %s: %s@." name reason ;
+      None
+    )
+
 
 module StrMapTool = MapTool(StrMap)
 let convert_pre_c (cfa_map: subroutine_cfa_map) : pre_c_pre_dwarf =
   (** Converts a `subroutine_cfa_map` to a `pre_c_pre_dwarf` type, in
       preparation for C coversion. *)
   let num_fde = StrMap.cardinal cfa_map in
-  let fdes = StrMapTool.fold_map_to_array
+  let fdes_list_with_none = StrMap.fold (fun name entry folded ->
-      convert_pre_c_fde cfa_map empty_fde in
+      (convert_pre_c_fde name entry) :: folded)
+      cfa_map [] in
+  let fdes_list = List.fold_left (fun folded elt -> match elt with
+      | Some x -> x::folded
+      | None -> folded) [] fdes_list_with_none in
+  let fdes = Array.of_list fdes_list in
   {
     num_fde = num_fde ;
     fdes = fdes

DwarfSynth/Simplest.ml

@@ -458,3 +458,13 @@ let of_proj proj : subroutine_cfa_map =
   (** Extracts the `cfa_changes` of a project *)
   let prog = BStd.Project.program proj in
   of_prog prog
+
+let clean_lost_track_subs pre_dwarf : subroutine_cfa_map =
+  (** Removes the subroutines on which we lost track from [pre_dwarf] *)
+  let sub_lost_track sub_name (sub: subroutine_cfa_data) =
+    not @@ AddrMap.exists (fun addr pos -> (match pos with
+        | RspOffset _ | RbpOffset _ -> false
+        | CfaLostTrack -> true))
+        sub.cfa_changes_fde
+  in
+  StrMap.filter sub_lost_track pre_dwarf

csmith/csmith_test_generated.sh

@@ -0,0 +1,91 @@
+#!/bin/bash
+
+USAGE="$0 nb_tests kept_directory"
+
+if [ "$#" -lt 1 ] ; then
+    >&2 echo -e "Missing argument(s). Usage:\n$USAGE"
+    exit 1
+fi
+
+NB_TESTS=$1
+KEPT_DIRECTORY=$2
+RUN_TIMEOUT=2
+
+tmpdir="$(mktemp -d)"
+error_count=0
+
+function statusline {
+    echo -ne "\r\e[1m[\e[32m$(printf "%03d/%03d" "$1" "$NB_TESTS")\e[39m] >>>\e[0m $2                              "
+}
+
+function reporterror {
+    echo -e "\n\e[1;31m[$(printf "%03d" $1)] Error: \e[0m $2"
+}
+
+function mkpath {
+    echo "$tmpdir/$1"
+}
+
+function failed_test {
+    path=$(mkpath $1)
+    cp "$path."* "$KEPT_DIRECTORY"
+    reporterror "$1" "$2"
+    error_count=$((error_count + 1))
+}
+
+function test_empty_dwarf {
+    path="$2"
+    statusline $1 "test DWARF emptyness"
+    if [ -z "$(readelf -wF "$path.eh.bin")" ]; then
+        return 1
+    else
+        return 0
+    fi
+}
+
+function test_run {
+    path="$2"
+    statusline $1 "running binaries"
+    orig="$(timeout $RUN_TIMEOUT "$path.orig.bin")"
+    orig_status=$?
+    synth="$(timeout $RUN_TIMEOUT "$path.eh.bin")"
+    synth_status=$?
+
+    if [ "$orig_status" -ne "$synth_status" ] || [ "$orig" != "$synth" ] ; then
+        return 1
+    else
+        return 0
+    fi
+}
+
+mkdir -p "$KEPT_DIRECTORY"
+
+statusline "" ""
+for num in $(seq 1 $NB_TESTS); do
+    ## Generation, compilation, synthesis
+    statusline $num "csmith"
+    path=$(mkpath $num)
+    csmith > "$path.c"
+    statusline $num "compiling"
+    gcc -O2 -I/usr/include/csmith-2.3.0/ -w "$path.c" -o "$path.orig.bin"
+    objcopy --remove-section '.eh_frame' --remove-section '.eh_frame_hdr' \
+        "$path.orig.bin" "$path.bin"
+    statusline $num "generating dwarf"
+    ../synthesize_dwarf.sh "$path.bin" "$path.eh.bin"
+
+    ## Testing
+    if ! test_empty_dwarf "$num" "$path"; then
+        failed_test "$num" "empty generated DWARF"
+    elif ! test_run "$num" "$path"; then
+        failed_test "$num" "different execution behaviour"
+    fi
+done
+
+echo ""
+if [ "$error_count" -eq "0" ] ; then
+    echo -e "\n== ALL TESTS PASSED =="
+else
+    echo -e "\n== FAILED TESTS: $error_count/$NB_TESTS =="
+fi
+
+rm -rf "$tmpdir"

synthesize_dwarf.sh

@@ -35,7 +35,7 @@ function find_ml_dwarf_write {
 }
 
 function bap_synth {
-    bap "$INPUT_FILE" -p dwarfsynth --dwarfsynth-output "$TMP_DIR/marshal" \
+    bap "$INPUT_FILE" --no-byteweight -p dwarfsynth --dwarfsynth-output "$TMP_DIR/marshal" \
         > /dev/null
     return $?
 }