[clang] blacklist some flags at the -### level

Summary: `clang -### ...` is allergic to a few flags.

Reviewed By: mbouaziz

Differential Revision: D6736820

fbshipit-source-id: d9b0e3c

infer/src/base/CommandLineOption.ml

@@ -771,18 +771,6 @@ let mk_subcommand command ?on_unknown_arg:(on_unknown = `Reject) ~name ?deprecat
   subcommand_actions := (name, switch) :: !subcommand_actions
 
 
-(* drop well-balanced first and last characters in [s] that satisfy the [drop] predicate; for
-   instance, [lrstrip ~drop:(function | 'a' | 'x' -> true | _ -> false) "xaabax"] returns "ab" *)
-let rec lrstrip ~drop s =
-  let n = String.length s in
-  if n < 2 then s
-  else
-    let first = String.unsafe_get s 0 in
-    if Char.equal first (String.unsafe_get s (n - 1)) && drop first then
-      lrstrip ~drop (String.slice s 1 (n - 1))
-    else s
-
-
 let args_from_argfile arg =
   let abs_fname =
     let fname = String.slice arg 1 (String.length arg) in
@@ -790,7 +778,7 @@ let args_from_argfile arg =
   in
   match In_channel.read_lines abs_fname with
   | lines ->
-      let strip = lrstrip ~drop:(function '"' | '\'' -> true | _ -> false) in
+      let strip = Utils.strip_balanced_once ~drop:(function '"' | '\'' -> true | _ -> false) in
       List.map ~f:strip lines
   | exception e ->
       raise (Arg.Bad ("Error reading argument file '" ^ abs_fname ^ "': " ^ Exn.to_string e))

infer/src/base/Utils.ml

@@ -364,3 +364,15 @@ let better_hash x = Marshal.to_string x [Marshal.No_sharing] |> Caml.Digest.stri
 let unlink_file_on_exit temp_file =
   "Cleaning temporary file " ^ temp_file
   |> Epilogues.register ~f:(fun () -> try Unix.unlink temp_file with _ -> ())
+
+
+(** drop at most one layer of well-balanced first and last characters satisfying [drop] from the
+   string; for instance, [strip_balanced ~drop:(function | 'a' | 'x' -> true | _ -> false) "xaabax"]
+   returns "aaba" *)
+let strip_balanced_once ~drop s =
+  let n = String.length s in
+  if n < 2 then s
+  else
+    let first = String.unsafe_get s 0 in
+    if Char.equal first (String.unsafe_get s (n - 1)) && drop first then String.slice s 1 (n - 1)
+    else s

infer/src/base/Utils.mli

@@ -110,3 +110,8 @@ val better_hash : 'a -> Caml.Digest.t
 
 val unlink_file_on_exit : string -> unit
 (** delete [temporary] file on exit *)
+
+val strip_balanced_once : drop:(char -> bool) -> string -> string
+(** drop at most one layer of well-balanced first and last characters satisfying [drop] from the
+   string; for instance, [strip_balanced ~drop:(function | 'a' | 'x' -> true | _ -> false) "xaabax"]
+   returns "aaba" *)

infer/src/clang/ClangCommand.ml

@@ -12,6 +12,9 @@ module L = Logging
 
 type t = {exec: string; argv: string list; orig_argv: string list; quoting_style: ClangQuotes.style}
 
+(** bad for every clang invocation *)
+let clang_blacklisted_flags = ["-fembed-bitcode-marker"; "-fno-canonical-system-headers"]
+
 let fcp_dir =
   Config.bin_dir ^/ Filename.parent_dir_name ^/ Filename.parent_dir_name
   ^/ "facebook-clang-plugins"
@@ -67,22 +70,54 @@ let file_arg_cmd_sanitizer cmd =
 
 let include_override_regex = Option.map ~f:Str.regexp Config.clang_include_to_override_regex
 
-let filter_and_replace_unsupported_args ?(replace_option_arg= fun _ s -> s)
+(** Filter arguments from [args], looking into argfiles too. [replace_options_arg prev arg] returns
+   [arg'], where [arg'] is the new version of [arg] given the preceding arguments (in reverse order) [prev]. *)
+let filter_and_replace_unsupported_args ?(replace_options_arg= fun _ s -> s)
     ?(blacklisted_flags= []) ?(blacklisted_flags_with_arg= []) ?(post_args= []) args =
-  let rec aux (prev, res_rev) args =
+  (* [prev] is the previously seen argument, [res_rev] is the reversed result, [changed] is true if
+     some change has been performed *)
+  let rec aux (prev_is_blacklisted_with_arg, res_rev, changed) args =
     match args with
     | [] ->
-        (* return non-reversed list *)
+        (prev_is_blacklisted_with_arg, res_rev, changed)
-        List.rev_append res_rev post_args
+    | _ :: tl when prev_is_blacklisted_with_arg ->
+        (* in the unlikely event that a blacklisted flag with arg sits as the last option in some
+           arg file, we need to remove its argument now *)
+        aux (false, res_rev, true) tl
+    | at_argfile :: tl when String.is_prefix at_argfile ~prefix:"@"
+      -> (
+        let argfile = String.slice at_argfile 1 (String.length at_argfile) in
+        match In_channel.read_lines argfile with
+        | lines ->
+            (* poor parsing of arguments with some stripping supported; hope that tools generating
+               argfiles more or less put one argument per line *)
+            let strip s =
+              String.strip s
+              |> Utils.strip_balanced_once ~drop:(function '"' | '\'' -> true | _ -> false)
+            in
+            let last_in_file_is_blacklisted, rev_res_with_file_args, changed_file =
+              List.map ~f:strip lines |> aux (prev_is_blacklisted_with_arg, res_rev, false)
+            in
+            if changed_file then aux (last_in_file_is_blacklisted, rev_res_with_file_args, true) tl
+            else
+              (* keep the same argfile if we haven't needed to change anything in it *)
+              aux (last_in_file_is_blacklisted, at_argfile :: res_rev, changed) tl
+        | exception e ->
+            L.external_warning "Error reading argument file '%s': %s@\n" at_argfile
+              (Exn.to_string e) ;
+            aux (false, at_argfile :: res_rev, changed) tl )
     | flag :: tl when List.mem ~equal:String.equal blacklisted_flags flag ->
-        aux (flag, res_rev) tl
+        aux (false, res_rev, true) tl
-    | flag1 :: flag2 :: tl when List.mem ~equal:String.equal blacklisted_flags_with_arg flag1 ->
+    | flag :: tl when List.mem ~equal:String.equal blacklisted_flags_with_arg flag ->
-        aux (flag2, res_rev) tl
+        (* remove the flag and its arg separately in case we are at the end of an argfile *)
+        aux (true, res_rev, true) tl
     | arg :: tl ->
-        let res_rev' = replace_option_arg prev arg :: res_rev in
+        let arg' = replace_options_arg res_rev arg in
-        aux (arg, res_rev') tl
+        aux (false, arg' :: res_rev, changed || not (phys_equal arg arg')) tl
   in
-  aux ("", []) args
+  match aux (false, [], false) args with _, res_rev, _ ->
+    (* return non-reversed list *)
+    List.rev_append res_rev post_args
 
 
 (* Work around various path or library issues occurring when one tries to substitute Apple's version
@@ -90,9 +125,10 @@ let filter_and_replace_unsupported_args ?(replace_option_arg= fun _ s -> s)
    fatal warnings. *)
 let clang_cc1_cmd_sanitizer cmd =
   (* command line options not supported by the opensource compiler or the plugins *)
-  let blacklisted_flags = ["-fembed-bitcode-marker"; "-fno-canonical-system-headers"] in
   let blacklisted_flags_with_arg = ["-mllvm"] in
-  let replace_option_arg option arg =
+  let replace_options_arg options arg =
+    match options with
+    | option :: _ ->
         if String.equal option "-arch" && String.equal arg "armv7k" then "armv7"
           (* replace armv7k arch with armv7 *)
         else if String.is_suffix arg ~suffix:"dep.tmp" then (
@@ -110,6 +146,8 @@ let clang_cc1_cmd_sanitizer cmd =
           | _ ->
               arg
         else arg
+    | [] ->
+        arg
   in
   let args_defines =
     if Config.bufferoverrun && not Config.biabduction then ["-D__INFER_BUFFEROVERRUN"] else []
@@ -124,8 +162,8 @@ let clang_cc1_cmd_sanitizer cmd =
        argv_cons "-Wno-everything"
   in
   let clang_arguments =
-    filter_and_replace_unsupported_args ~blacklisted_flags ~blacklisted_flags_with_arg
+    filter_and_replace_unsupported_args ~blacklisted_flags:clang_blacklisted_flags
-      ~replace_option_arg ~post_args:(List.rev post_args_rev) cmd.argv
+      ~blacklisted_flags_with_arg ~replace_options_arg ~post_args:(List.rev post_args_rev) cmd.argv
   in
   file_arg_cmd_sanitizer {cmd with argv= clang_arguments}
 
@@ -133,7 +171,10 @@ let clang_cc1_cmd_sanitizer cmd =
 let mk quoting_style ~prog ~args =
   (* Some arguments break the compiler so they need to be removed even before the normalization step *)
   let blacklisted_flags_with_arg = ["-index-store-path"] in
-  let sanitized_args = filter_and_replace_unsupported_args ~blacklisted_flags_with_arg args in
+  let sanitized_args =
+    filter_and_replace_unsupported_args ~blacklisted_flags:clang_blacklisted_flags
+      ~blacklisted_flags_with_arg args
+  in
   {exec= prog; orig_argv= sanitized_args; argv= sanitized_args; quoting_style}