ARM enum sizes diagnostic

ARM enum sizes diagnostic

[Mark Shinwell]

The diagnostic seen below should be marked as a warning rather
than an error; it is possible for the diagnostic to arise when compiling
with GCC and -fshort-enums in a situation where the values of such
short enumerations cannot escape the compilation unit.  It is in
fact already treated as a warning rather than an error as seen below.

OK to apply?

Mark

--


2007-02-15  Mark Shinwell  <shinwell@codesourcery.com>

         bfd/
         * elf32-arm.c (elf32_arm_merge_eabi_attributes): Upon discovery
         of incompatible EABI attributes for the sizes of enumerations,
         diagnose "Warning" not "ERROR".


--- elf32-arm.c 29 Jan 2007 16:29:21 -0000      1.104
+++ elf32-arm.c 15 Feb 2007 11:37:26 -0000
@@ -7201,7 +7201,7 @@ elf32_arm_merge_eabi_attributes (bfd *ib
                        && out_attr[i].i != in_attr[i].i)
                 {
                   _bfd_error_handler
-                   (_("ERROR: %B: Conflicting enum sizes"), ibfd);
+                   (_("Warning: %B: Conflicting enum sizes"), ibfd);
                 }
             }
           break;

Re: ARM enum sizes diagnostic

[Nick Clifton]

Hi Mark,

> The diagnostic seen below should be marked as a warning rather
> than an error; it is possible for the diagnostic to arise when compiling
> with GCC and -fshort-enums in a situation where the values of such
> short enumerations cannot escape the compilation unit.  It is in
> fact already treated as a warning rather than an error as seen below.

This is OK, but if you are going to fix this message then you might as 
well go the whole hog.  Specifically:

   * Report the two enum sizes.
   * Explain to the user why this might not be a problem.

So maybe something like this:

   warning: <foo.o> places 'enum' values into <N> bytes whereas
    <bar.o> places them into <M> bytes.  This will cause problems
    if the two files attempt to share enum values.

You may also want to think about having a way of disabling this warning 
message for users that do not want it.

Cheers
   Nick

Re: ARM enum sizes diagnostic

[Mark Shinwell]

Nick Clifton wrote:
> Hi Mark,
> 
>> The diagnostic seen below should be marked as a warning rather
>> than an error; it is possible for the diagnostic to arise when compiling
>> with GCC and -fshort-enums in a situation where the values of such
>> short enumerations cannot escape the compilation unit.  It is in
>> fact already treated as a warning rather than an error as seen below.
> 
> This is OK, but if you are going to fix this message then you might as 
> well go the whole hog.  Specifically:
> 
>   * Report the two enum sizes.
>   * Explain to the user why this might not be a problem.
> 
> So maybe something like this:
> 
>   warning: <foo.o> places 'enum' values into <N> bytes whereas
>    <bar.o> places them into <M> bytes.  This will cause problems
>    if the two files attempt to share enum values.
> 
> You may also want to think about having a way of disabling this warning 
> message for users that do not want it.

Agreed on both counts; I'll work on that.

Mark

Re: ARM enum sizes diagnostic

[Mark Shinwell]

Nick Clifton wrote:
>> The diagnostic seen below should be marked as a warning rather
>> than an error; it is possible for the diagnostic to arise when compiling
>> with GCC and -fshort-enums in a situation where the values of such
>> short enumerations cannot escape the compilation unit.  It is in
>> fact already treated as a warning rather than an error as seen below.
> 
> This is OK, but if you are going to fix this message then you might as 
> well go the whole hog.  Specifically:
> 
>   * Report the two enum sizes.
>   * Explain to the user why this might not be a problem.
> 
> So maybe something like this:
> 
>   warning: <foo.o> places 'enum' values into <N> bytes whereas
>    <bar.o> places them into <M> bytes.  This will cause problems
>    if the two files attempt to share enum values.
> 
> You may also want to think about having a way of disabling this warning 
> message for users that do not want it.

The following (which has taken rather more time than it was worth,
I think :-) is about as good as we can do, since we can't pinpoint
N and M precisely.  It behaves as expected in manual testing.

OK to apply?

Mark

--


2007-03-18  Mark Shinwell  <shinwell@codesourcery.com>

	bfd/
	* bfd-in.h (bfd_elf32_arm_set_target_relocs): Add "bfd *"
	argument and extra last argument.
	* bfd-in2.h: Regenerate.
	* elf32-arm.c (elf32_arm_obj_tdata): Add no_enum_size_warning
	member.
	(bfd_elf32_arm_set_target_relocs): Add "bfd *" argument and
	extra last argument.  Set no_enum_size_warning appropriately.
	(elf32_arm_merge_eabi_attributes): Improve enum sizes
	diagnostic, suppressing it when no_enum_size_warning dictates.

	ld/
	* ld.texinfo: Document --no-enum-size-warning.
	* emultempl/armelf.em (no_enum_size_warning): New.
	(arm_elf_create_output_section_statements): Correct typo
	in comment.  Pass no_enum_size_warning to
	bfd_elf32_arm_set_target_relocs.
	(PARSE_AND_LIST_PROLOGUE): Define OPTION_NO_ENUM_SIZE_WARNING.
	(PARSE_AND_LIST_OPTIONS): Document --no-enum-size-warning.
	(PARSE_AND_LIST_ARGS_CASES): Add OPTION_NO_ENUM_SIZE_WARNING
	case.


Index: bfd/bfd-in.h
===================================================================
RCS file: /cvs/src/src/bfd/bfd-in.h,v
retrieving revision 1.123
diff -U3 -p -r1.123 bfd-in.h
--- bfd/bfd-in.h        29 Jan 2007 16:29:21 -0000      1.123
+++ bfd/bfd-in.h        18 Mar 2007 22:14:55 -0000
@@ -895,7 +895,8 @@ extern bfd_boolean bfd_elf32_arm_process
    (bfd *, struct bfd_link_info *);

  void bfd_elf32_arm_set_target_relocs
-  (struct bfd_link_info *, int, char *, int, int, bfd_arm_vfp11_fix);
+  (bfd *, struct bfd_link_info *, int, char *, int, int, bfd_arm_vfp11_fix,
+   int);

  extern bfd_boolean bfd_elf32_arm_get_bfd_for_interworking
    (bfd *, struct bfd_link_info *);
Index: bfd/bfd-in2.h
===================================================================
RCS file: /cvs/src/src/bfd/bfd-in2.h,v
retrieving revision 1.414
diff -U3 -p -r1.414 bfd-in2.h
--- bfd/bfd-in2.h       5 Feb 2007 19:50:11 -0000       1.414
+++ bfd/bfd-in2.h       18 Mar 2007 22:14:55 -0000
@@ -902,7 +902,8 @@ extern bfd_boolean bfd_elf32_arm_process
    (bfd *, struct bfd_link_info *);

  void bfd_elf32_arm_set_target_relocs
-  (struct bfd_link_info *, int, char *, int, int, bfd_arm_vfp11_fix);
+  (bfd *, struct bfd_link_info *, int, char *, int, int, bfd_arm_vfp11_fix,
+   int);

  extern bfd_boolean bfd_elf32_arm_get_bfd_for_interworking
    (bfd *, struct bfd_link_info *);
Index: bfd/elf32-arm.c
===================================================================
RCS file: /cvs/src/src/bfd/elf32-arm.c,v
retrieving revision 1.105
diff -U3 -p -r1.105 elf32-arm.c
--- bfd/elf32-arm.c     22 Feb 2007 17:03:59 -0000      1.105
+++ bfd/elf32-arm.c     18 Mar 2007 22:14:55 -0000
@@ -2064,6 +2064,9 @@ struct elf32_arm_obj_tdata

    aeabi_attribute known_eabi_attributes[NUM_KNOWN_ATTRIBUTES];
    aeabi_attribute_list *other_eabi_attributes;
+
+  /* Zero to warn when linking objects with incompatible enum sizes.  */
+  int no_enum_size_warning;
  };

  #define elf32_arm_tdata(abfd) \
@@ -3858,12 +3861,14 @@ bfd_elf32_arm_vfp11_fix_veneer_locations
  /* Set target relocation values needed during linking.  */

  void
-bfd_elf32_arm_set_target_relocs (struct bfd_link_info *link_info,
+bfd_elf32_arm_set_target_relocs (struct bfd *output_bfd,
+                                struct bfd_link_info *link_info,
                                  int target1_is_rel,
                                  char * target2_type,
                                   int fix_v4bx,
                                  int use_blx,
-                                 bfd_arm_vfp11_fix vfp11_fix)
+                                 bfd_arm_vfp11_fix vfp11_fix,
+                                int no_enum_warn)
  {
    struct elf32_arm_link_hash_table *globals;

@@ -3884,6 +3889,8 @@ bfd_elf32_arm_set_target_relocs (struct
    globals->fix_v4bx = fix_v4bx;
    globals->use_blx |= use_blx;
    globals->vfp11_fix = vfp11_fix;
+
+  elf32_arm_tdata (output_bfd)->no_enum_size_warning = no_enum_warn;
  }

  /* The thumb form of a long branch is a bit finicky, because the offset
@@ -7198,10 +7205,15 @@ elf32_arm_merge_eabi_attributes (bfd *ib
                   out_attr[i].i = in_attr[i].i;
                 }
               else if (in_attr[i].i != AEABI_enum_forced_wide
-                      && out_attr[i].i != in_attr[i].i)
+                      && out_attr[i].i != in_attr[i].i
+                      && !elf32_arm_tdata (obfd)->no_enum_size_warning)
                 {
+                 const char *aeabi_enum_names[] =
+                   { "", "variable-size", "32-bit", "" };
                   _bfd_error_handler
-                   (_("ERROR: %B: Conflicting enum sizes"), ibfd);
+                   (_("warning: %B uses %s enums yet the output is to use 
%s enums; use of enum values across objects may fail"),
+                    ibfd, aeabi_enum_names[in_attr[i].i],
+                    aeabi_enum_names[out_attr[i].i]);
                 }
             }
           break;
Index: ld/ld.texinfo
===================================================================
RCS file: /cvs/src/src/ld/ld.texinfo,v
retrieving revision 1.184
diff -U3 -p -r1.184 ld.texinfo
--- ld/ld.texinfo       29 Jan 2007 16:28:40 -0000      1.184
+++ ld/ld.texinfo       18 Mar 2007 22:14:56 -0000
@@ -5519,6 +5519,15 @@ instruction. The original instruction is
  the veneer. The extra cycles required to call and return from the veneer
  are sufficient to avoid the erratum in both the scalar and vector cases.

+@cindex NO_ENUM_SIZE_WARNING
+@kindex --no-enum-size-warning
+The @samp{--no-enum-size-warning} switch prevents the linker from
+warning when linking object files that specify incompatible EABI
+enumeration size attributes.  For example, with this switch enabled,
+linking of an object file using 32-bit enumeration values with another
+using enumeration values fitted into the smallest possible space will
+not be diagnosed.
+
  @ifclear GENERIC
  @lowersections
  @end ifclear
Index: ld/emultempl/armelf.em
===================================================================
RCS file: /cvs/src/src/ld/emultempl/armelf.em,v
retrieving revision 1.53
diff -U3 -p -r1.53 armelf.em
--- ld/emultempl/armelf.em      29 Jan 2007 16:28:40 -0000      1.53
+++ ld/emultempl/armelf.em      18 Mar 2007 22:14:56 -0000
@@ -36,6 +36,7 @@ static char *target2_type = "${TARGET2_T
  static int fix_v4bx = 0;
  static int use_blx = 0;
  static bfd_arm_vfp11_fix vfp11_denorm_fix = BFD_ARM_VFP11_FIX_DEFAULT;
+static int no_enum_size_warning = 0;

  static void
  gld${EMULATION_NAME}_before_parse (void)
@@ -233,13 +234,14 @@ arm_elf_finish (void)
            thumb_entry_symbol);
  }

-/* This is a convenitent point to tell BFD about target specific flags.
+/* This is a convenient point to tell BFD about target specific flags.
     After the output has been created, but before inputs are read.  */
  static void
  arm_elf_create_output_section_statements (void)
  {
-  bfd_elf32_arm_set_target_relocs (&link_info, target1_is_rel, target2_type,
-                                   fix_v4bx, use_blx, vfp11_denorm_fix);
+  bfd_elf32_arm_set_target_relocs (output_bfd, &link_info, target1_is_rel,
+                                  target2_type, fix_v4bx, use_blx,
+                                  vfp11_denorm_fix, no_enum_size_warning);
  }

  EOF
@@ -256,6 +258,7 @@ PARSE_AND_LIST_PROLOGUE='
  #define OPTION_FIX_V4BX                        306
  #define OPTION_USE_BLX                 307
  #define OPTION_VFP11_DENORM_FIX                308
+#define OPTION_NO_ENUM_SIZE_WARNING    309
  '

  PARSE_AND_LIST_SHORTOPTS=p
@@ -270,6 +273,7 @@ PARSE_AND_LIST_LONGOPTS='
    { "fix-v4bx", no_argument, NULL, OPTION_FIX_V4BX},
    { "use-blx", no_argument, NULL, OPTION_USE_BLX},
    { "vfp11-denorm-fix", required_argument, NULL, OPTION_VFP11_DENORM_FIX},
+  { "no-enum-size-warning", no_argument, NULL, OPTION_NO_ENUM_SIZE_WARNING},
  '

  PARSE_AND_LIST_OPTIONS='
@@ -281,6 +285,7 @@ PARSE_AND_LIST_OPTIONS='
    fprintf (file, _("     --fix-v4bx               Rewrite BX rn as MOV 
pc, rn for ARMv4\n"));
    fprintf (file, _("     --use-blx                Enable use of BLX 
instructions\n"));
    fprintf (file, _("     --vfp11-denorm-fix       Specify how to fix 
VFP11 denorm erratum\n"));
+  fprintf (file, _("     --no-enum-size-warning   Don'\''t warn about 
objects with incompatible enum sizes\n"));
  '

  PARSE_AND_LIST_ARGS_CASES='
@@ -326,6 +331,10 @@ PARSE_AND_LIST_ARGS_CASES='
        else
          einfo (_("Unrecognized VFP11 fix type '\''%s'\''.\n"), optarg);
        break;
+
+    case OPTION_NO_ENUM_SIZE_WARNING:
+      no_enum_size_warning = 1;
+      break;
  '

  # We have our own after_open and before_allocation functions, but they call

Re: ARM enum sizes diagnostic

[Nick Clifton]

Hi Mark,

> 2007-03-18  Mark Shinwell  <shinwell@codesourcery.com>
> 
>     bfd/
>     * bfd-in.h (bfd_elf32_arm_set_target_relocs): Add "bfd *"
>     argument and extra last argument.
>     * bfd-in2.h: Regenerate.
>     * elf32-arm.c (elf32_arm_obj_tdata): Add no_enum_size_warning
>     member.
>     (bfd_elf32_arm_set_target_relocs): Add "bfd *" argument and
>     extra last argument.  Set no_enum_size_warning appropriately.
>     (elf32_arm_merge_eabi_attributes): Improve enum sizes
>     diagnostic, suppressing it when no_enum_size_warning dictates.
> 
>     ld/
>     * ld.texinfo: Document --no-enum-size-warning.
>     * emultempl/armelf.em (no_enum_size_warning): New.
>     (arm_elf_create_output_section_statements): Correct typo
>     in comment.  Pass no_enum_size_warning to
>     bfd_elf32_arm_set_target_relocs.
>     (PARSE_AND_LIST_PROLOGUE): Define OPTION_NO_ENUM_SIZE_WARNING.
>     (PARSE_AND_LIST_OPTIONS): Document --no-enum-size-warning.
>     (PARSE_AND_LIST_ARGS_CASES): Add OPTION_NO_ENUM_SIZE_WARNING
>     case.

Approved - please apply.

Cheers
   Nick