Move unwind info to read-only section on AIX

Move unwind info to read-only section on AIX

[Andrew Dixie]

Hi,

The following patch moves dwarf unwind information from the data
section to the read-only/text section on AIX.  This means the memory
for the unwind information can be shared across multiple processes.

The frame tables are still registered through collect2 scanning for
"_GLOBAL_F" symbol names, however these symbols now exist in the
eh_frame section.

Previously all the unwind pointers were absolute (requiring
relocations), these have been changed to PC-relative and data-relative
offsets.

The DW_EH_PE_datarel encoding doesn't appear to have previously been
used in the compiler, but it is already implemented in the runtime
unwind code.
The data relative encoding is needed as a PC relative offset cannot
point from the text section to the data section on AIX.
For the base for the data relative offset, I couldn't find any way to
get a reference to the start of the data segment, so this patch uses
the symbol __dso_handle as an arbitrary base for the data relative
offsets.
I'm considering changing the base to be the TOC.
I'm Interested in thoughts or ideas on a better base for the data
relative offsets?

Thanks,
Andrew

2014-07-25  Andrew Dixie  <andrewd@gentrack.com>

       Move exception tables to read-only memory on AIX.
       * collect2.c (write_c_file_stat): Provide dbase on AIX.
       (scan_prog_file): Don't output __dso_handle.
       * config/rs6000/aix.h (ASM_PREFERRED_EH_DATA_FORMAT): define.
       (EH_TABLES_CAN_BE_READ_ONLY): define.
       (ASM_OUTPUT_DWARF_PCREL): define.
       (ASM_OUTPUT_DWARF_DATAREL): define.
       (EH_FRAME_IN_DATA_SECTION): undefine.
       * config/rs6000/rs6000-aix.c: new file.
       (rs6000_aix_asm_output_dwarf_pcrel): new function.
       (rs6000_aix_asm_output_dwarf_datarel): new function.
       * config/rs6000/rs6000.c (rs6000_xcoff_asm_init_sections): remove
       assignment of exception_section.
       * dwarf2asm.c (dw2_asm_output_encoded_addr_rtx): Add call to
       ASM_OUTPUT_DWARF_DATAREL.
       * dwarf2out.c (switch_to_eh_frame_section): Add call to
       EMIT_COLLECT2_LABELS_FOR_EH_FRAMES.

diff -Nrup orig/gcc-4.10-20140706/gcc/collect2.c
gcc-4.10-20140706/gcc/collect2.c
--- orig/gcc-4.10-20140706/gcc/collect2.c       2014-06-26
21:16:28.000000000 +1200
+++ gcc-4.10-20140706/gcc/collect2.c    2014-07-28 12:41:54.852926208 +1200
@@ -2101,12 +2110,23 @@ write_c_file_stat (FILE *stream, const c
       fprintf (stream, "  struct object *next;\n");
       fprintf (stream, "};\n");

+      fprintf (stream, "extern void __register_frame_info_table_bases
(void *, struct object *, void *tbase, void *dbase);\n");
       fprintf (stream, "extern void __register_frame_info_table (void
*, struct object *);\n");
       fprintf (stream, "extern void *__deregister_frame_info (void *);\n");
+#ifdef TARGET_AIX_VERSION
+      fprintf (stream, "extern void *__dso_handle;\n");
+#endif

       fprintf (stream, "static void reg_frame () {\n");
       fprintf (stream, "\tstatic struct object ob;\n");
+#ifdef TARGET_AIX_VERSION
+      /* As per config/rs6000/rs6000-aix.c, we use __dso_handle as
the dbase on AIX.
+        This might not be the start of the segment, we assume signed offsets.
+       */
+      fprintf (stream, "\t__register_frame_info_table_bases
(frame_table, &ob, (void *)0, &__dso_handle);\n");
+#else
       fprintf (stream, "\t__register_frame_info_table (frame_table, &ob);\n");
+#endif
       fprintf (stream, "\t}\n");

       fprintf (stream, "static void dereg_frame () {\n");
@@ -2868,7 +2888,15 @@ scan_prog_file (const char *prog_name, s
                             provides an explicit export list.  */
                          if (shared_obj && !is_shared
                              && which_pass == PASS_OBJ && !export_flag)
-                           add_to_list (&exports, name);
+                           {
+                                   /* Do not auto-export __dso_handle.
+                                      It is required to be local to
each module.
+                                    */
+                              if (strcmp(name, "__dso_handle") != 0)
+                                {
+                                  add_to_list (&exports, name);
+                                }
+                           }
 #endif
                          continue;
                        }
diff -Nrup orig/gcc-4.10-20140706/gcc/config/rs6000/rs6000-aix.c
gcc-4.10-20140706/gcc/config/rs6000/rs6000-aix.c
--- orig/gcc-4.10-20140706/gcc/config/rs6000/rs6000-aix.c
1970-01-01 12:00:00.000000000 +1200
+++ gcc-4.10-20140706/gcc/config/rs6000/rs6000-aix.c    2014-07-28
12:41:54.852926208 +1200
@@ -0,0 +1,54 @@
+/* Functions for AIX on PowerPC.
+   Copyright (C) 2014 Free Software Foundation, Inc.
+
+This file is part of GCC.
+
+GCC is free software; you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation; either version 3, or (at your option)
+any later version.
+
+GCC is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with GCC; see the file COPYING3.  If not see
+<http://www.gnu.org/licenses/>.  */
+
+#include "config.h"
+#include "system.h"
+#include "coretypes.h"
+#include "tm.h"
+#include "rtl.h"
+#include "tm_p.h"
+#include "output.h"
+
+/* The AIX assembler cannot handle the SYMBOL-. syntax.
+   Instead create a new symbol at the current point in the file and output
+   a relative reference to this new symbol. */
+void rs6000_asm_output_dwarf_pcrel(FILE *file, int size, const char *label)
+{
+       static int pcrel_no = 0;
+       ++pcrel_no;
+
+       fprintf(file, "AIXPCREL..%i:\n", pcrel_no);
+       fputs (integer_asm_op (size, FALSE), file);
+       assemble_name (file, label);
+       fprintf(file, "-AIXPCREL..%i", pcrel_no);
+}
+
+/* Output a symbol offset relative to the dbase for the current object.
+   We use __dso_handle as an arbitrary base for dbase and assume
signed offsets.
+
+   __dso_handle is embedded in all executables/libraries through
+   libgcc/config/rs6000/crtcxa.c
+ */
+void rs6000_asm_output_dwarf_datarel(FILE *file, int size, const char *label)
+{
+       fputs (integer_asm_op (size, FALSE), file);
+       assemble_name (file, label);
+       fputs("-__dso_handle", file);
+}
+
diff -Nrup orig/gcc-4.10-20140706/gcc/config/rs6000/rs6000.c
gcc-4.10-20140706/gcc/config/rs6000/rs6000.c
--- orig/gcc-4.10-20140706/gcc/config/rs6000/rs6000.c   2014-07-03
18:13:48.000000000 +1200
+++ gcc-4.10-20140706/gcc/config/rs6000/rs6000.c        2014-07-25
13:43:54.151118199 +1200
@@ -29251,7 +29251,6 @@ rs6000_xcoff_asm_init_sections (void)
     = get_unnamed_section (0, rs6000_xcoff_output_toc_section_asm_op, NULL);

   readonly_data_section = read_only_data_section;
-  exception_section = data_section;
 }

 static int
diff -Nrup orig/gcc-4.10-20140706/gcc/config/rs6000/rs6000-protos.h
gcc-4.10-20140706/gcc/config/rs6000/rs6000-protos.h
--- orig/gcc-4.10-20140706/gcc/config/rs6000/rs6000-protos.h
2014-06-26 02:46:08.000000000 +1200
+++ gcc-4.10-20140706/gcc/config/rs6000/rs6000-protos.h 2014-07-28
12:41:54.852926208 +1200
@@ -195,6 +195,8 @@ extern void rs6000_aix_asm_output_dwarf_
 extern void get_ppc476_thunk_name (char name[32]);
 extern bool rs6000_overloaded_builtin_p (enum rs6000_builtins);
 extern HOST_WIDE_INT rs6000_builtin_mask_calculate (void);
+extern void rs6000_asm_output_dwarf_pcrel(FILE *file, int size, const
char *label);
+extern void rs6000_asm_output_dwarf_datarel(FILE *file, int size,
const char *label);

 /* Declare functions in rs6000-c.c */

diff -Nrup orig/gcc-4.10-20140706/gcc/config/rs6000/t-aix52
gcc-4.10-20140706/gcc/config/rs6000/t-aix52
--- orig/gcc-4.10-20140706/gcc/config/rs6000/t-aix52    2014-01-03
11:23:26.000000000 +1300
+++ gcc-4.10-20140706/gcc/config/rs6000/t-aix52 2014-07-25
13:43:54.151118199 +1200
@@ -24,3 +24,7 @@ MULTILIB_OPTIONS      = pthread maix64
 MULTILIB_DIRNAMES      = pthread ppc64

 MULTILIB_MATCHES       =
+
+rs6000-aix.o: $(srcdir)/config/rs6000/rs6000-aix.c
+       $(COMPILE) $<
+       $(POSTCOMPILE)
diff -Nrup orig/gcc-4.10-20140706/gcc/config/rs6000/xcoff.h
gcc-4.10-20140706/gcc/config/rs6000/xcoff.h
--- orig/gcc-4.10-20140706/gcc/config/rs6000/xcoff.h    2014-06-26
02:46:08.000000000 +1200
+++ gcc-4.10-20140706/gcc/config/rs6000/xcoff.h 2014-07-28
12:48:43.361071782 +1200
@@ -299,8 +299,25 @@
 #define DATA_SECTION_ASM_OP \
   "\t.csect .data[RW]," XCOFF_CSECT_DEFAULT_ALIGNMENT_STR

+/* The eh_frames are put in the read-only text segment.
+   Local code labels/function will also be in the local text segment so use
+   PC relative addressing.
+   Global symbols must be in the data segment to allow loader relocations.
+   So use DW_EH_PE_indirect to allocate a slot in the local data segment.
+   There is no constant offset to this data segment from the text segment,
+   so use addressing relative to the data segment.
+ */
+#define ASM_PREFERRED_EH_DATA_FORMAT(CODE,GLOBAL) \
+       ((GLOBAL) ? (DW_EH_PE_indirect | DW_EH_PE_datarel |
DW_EH_PE_sdata4) : (DW_EH_PE_pcrel | DW_EH_PE_sdata4))

-/* Define to prevent DWARF2 unwind info in the data section rather
-   than in the .eh_frame section.  We do this because the AIX linker
-   would otherwise garbage collect these sections.  */
-#define EH_FRAME_IN_DATA_SECTION 1
+#define EMIT_COLLECT2_LABELS_FOR_EH_FRAMES 1
+
+#define TARGET_TERMINATE_DW2_EH_FRAME_INFO true
+
+#define EH_TABLES_CAN_BE_READ_ONLY 1
+
+#define ASM_OUTPUT_DWARF_PCREL(FILE, SIZE, LABEL)       \
+       rs6000_asm_output_dwarf_pcrel(FILE, SIZE, LABEL);
+
+#define ASM_OUTPUT_DWARF_DATAREL(FILE, SIZE, LABEL)       \
+       rs6000_asm_output_dwarf_datarel(FILE, SIZE, LABEL);
diff -Nrup orig/gcc-4.10-20140706/gcc/config.gcc
gcc-4.10-20140706/gcc/config.gcc
--- orig/gcc-4.10-20140706/gcc/config.gcc       2014-07-04
19:18:19.000000000 +1200
+++ gcc-4.10-20140706/gcc/config.gcc    2014-07-25 13:43:54.155118031 +1200
@@ -2369,6 +2369,7 @@ rs6000-ibm-aix5.1.* | powerpc-ibm-aix5.1
 rs6000-ibm-aix5.2.* | powerpc-ibm-aix5.2.*)
        tm_file="${tm_file} rs6000/aix.h rs6000/aix52.h rs6000/xcoff.h
rs6000/aix-stdint.h"
        tmake_file="rs6000/t-aix52 t-slibgcc"
+       extra_objs="$extra_objs rs6000-aix.o"
        extra_options="${extra_options} rs6000/aix64.opt"
        use_collect2=yes
        thread_file='aix'
@@ -2378,6 +2379,7 @@ rs6000-ibm-aix5.2.* | powerpc-ibm-aix5.2
 rs6000-ibm-aix5.3.* | powerpc-ibm-aix5.3.*)
        tm_file="${tm_file} rs6000/aix.h rs6000/aix53.h rs6000/xcoff.h
rs6000/aix-stdint.h"
        tmake_file="rs6000/t-aix52 t-slibgcc"
+       extra_objs="$extra_objs rs6000-aix.o"
        extra_options="${extra_options} rs6000/aix64.opt"
        use_collect2=yes
        thread_file='aix'
@@ -2387,6 +2389,7 @@ rs6000-ibm-aix5.3.* | powerpc-ibm-aix5.3
 rs6000-ibm-aix[6789].* | powerpc-ibm-aix[6789].*)
        tm_file="${tm_file} rs6000/aix.h rs6000/aix61.h rs6000/xcoff.h
rs6000/aix-stdint.h"
        tmake_file="rs6000/t-aix52 t-slibgcc"
+       extra_objs="$extra_objs rs6000-aix.o"
        extra_options="${extra_options} rs6000/aix64.opt"
        use_collect2=yes
        thread_file='aix'
diff -Nrup orig/gcc-4.10-20140706/gcc/doc/tm.texi.in
gcc-4.10-20140706/gcc/doc/tm.texi.in
--- orig/gcc-4.10-20140706/gcc/doc/tm.texi.in   2014-06-06
13:04:22.000000000 +1200
+++ gcc-4.10-20140706/gcc/doc/tm.texi.in        2014-07-28
12:41:54.852926208 +1200
@@ -6525,6 +6525,15 @@ Do not define this macro unless @code{TA
 also defined.
 @end defmac

+@defmac EMIT_COLLECT2_LABELS_FOR_EH_FRAMES
+If defined, gcc will output specially named labels for DWARF 2 frame unwind
+information, and collect2 will detect these labels and emit code to register
+the frames at runtime.
+
+If not defined, the target must register frame unwind tables
+through another mechanism.
+@end defmac
+
 @defmac EH_TABLES_CAN_BE_READ_ONLY
 Define this macro to 1 if your target is such that no frame unwind
 information encoding used with non-PIC code will ever require a
@@ -7067,6 +7076,11 @@ A C statement to issue assembly directiv
 reference to the given @var{label}, using an integer of the given @var{size}.
 @end defmac

+@defmac ASM_OUTPUT_DWARF_DATAREL (@var{stream}, @var{size}, @var{label})
+A C statement to issue assembly directives that create a reference to the
+given @var{label} relative to the dbase, using an integer of the
given @var{size}.
+@end defmac
+
 @defmac ASM_OUTPUT_DWARF_TABLE_REF (@var{label})
 A C statement to issue assembly directives that create a reference to
 the DWARF table identifier @var{label} from the current section.  This
diff -Nrup orig/gcc-4.10-20140706/gcc/dwarf2asm.c
gcc-4.10-20140706/gcc/dwarf2asm.c
--- orig/gcc-4.10-20140706/gcc/dwarf2asm.c      2014-01-03
11:23:26.000000000 +1300
+++ gcc-4.10-20140706/gcc/dwarf2asm.c   2014-07-25 14:07:57.050230529 +1200
@@ -992,6 +992,13 @@ dw2_asm_output_encoded_addr_rtx (int enc
          dw2_assemble_integer (size, addr);
          break;

+#ifdef ASM_OUTPUT_DWARF_DATAREL
+       case DW_EH_PE_datarel:
+         gcc_assert (GET_CODE (addr) == SYMBOL_REF);
+         ASM_OUTPUT_DWARF_DATAREL (asm_out_file, size, XSTR (addr, 0));
+         break;
+#endif
+
        case DW_EH_PE_pcrel:
          gcc_assert (GET_CODE (addr) == SYMBOL_REF);
 #ifdef ASM_OUTPUT_DWARF_PCREL
diff -Nrup orig/gcc-4.10-20140706/gcc/dwarf2out.c
gcc-4.10-20140706/gcc/dwarf2out.c
--- orig/gcc-4.10-20140706/gcc/dwarf2out.c      2014-07-02
09:35:41.000000000 +1200
+++ gcc-4.10-20140706/gcc/dwarf2out.c   2014-07-28 12:48:12.925359014 +1200
@@ -424,6 +424,11 @@ static void
 switch_to_eh_frame_section (bool back)
 {
   tree label;
+#ifdef EMIT_COLLECT2_LABELS_FOR_EH_FRAMES
+  bool emit_collect2_labels = EMIT_COLLECT2_LABELS_FOR_EH_FRAMES;
+#else
+  bool emit_collect2_labels = false;
+#endif

 #ifdef EH_FRAME_SECTION_NAME
   if (eh_frame_section == 0)
@@ -464,7 +469,11 @@ switch_to_eh_frame_section (bool back)
       /* We have no special eh_frame section.  Put the information in
         the data section and emit special labels to guide collect2.  */
       switch_to_section (data_section);
+      emit_collect2_labels = true;
+    }

+  if (emit_collect2_labels)
+    {
       if (!back)
        {
          label = get_file_function_name ("F");

Re: Move unwind info to read-only section on AIX

[David Edelsohn]

On Tue, Jul 29, 2014 at 5:33 PM, Andrew Dixie <andrewdixie50@gmail.com> wrote:
> Hi,
>
> The following patch moves dwarf unwind information from the data
> section to the read-only/text section on AIX.  This means the memory
> for the unwind information can be shared across multiple processes.
>
> The frame tables are still registered through collect2 scanning for
> "_GLOBAL_F" symbol names, however these symbols now exist in the
> eh_frame section.
>
> Previously all the unwind pointers were absolute (requiring
> relocations), these have been changed to PC-relative and data-relative
> offsets.
>
> The DW_EH_PE_datarel encoding doesn't appear to have previously been
> used in the compiler, but it is already implemented in the runtime
> unwind code.
> The data relative encoding is needed as a PC relative offset cannot
> point from the text section to the data section on AIX.
> For the base for the data relative offset, I couldn't find any way to
> get a reference to the start of the data segment, so this patch uses
> the symbol __dso_handle as an arbitrary base for the data relative
> offsets.
> I'm considering changing the base to be the TOC.
> I'm Interested in thoughts or ideas on a better base for the data
> relative offsets?

Hi, Andrew

Thanks for working on this.

As the comment in the code states DWARF2 unwind info was placed in the
data section to prevent the AIX linker from garbage collecting it.
How are you avoiding that problem?

Also, the AIX linker uses "$" for the current location, so the kludge
of creating PCREL symbols should not be necessary.

Using __dso_handle as the base address is underwhelming.

Thanks, David

Re: Move unwind info to read-only section on AIX

[Andrew Dixie]

Hi David,

On Thu, Jul 31, 2014 at 2:22 PM, David Edelsohn <dje.gcc@gmail.com> wrote:
> As the comment in the code states DWARF2 unwind info was placed in the
> data section to prevent the AIX linker from garbage collecting it.
> How are you avoiding that problem?

That comment is very old.  I'm not sure what the original problem was.
The code generated by collect2 which calls register_frame_info_table
has explicit references to the DWARF unwind info.  The linker won't
garbage collect the unwind info as the linker will see it as used
through these references.

> Also, the AIX linker uses "$" for the current location, so the kludge
> of creating PCREL symbols should not be necessary.

I will swap it over to this.

> Using __dso_handle as the base address is underwhelming.

Does the using the TOC sound like a reasonable solution?
Or do you know how to get a reference to the start of the data section
from the AIX assembler/linker?
The ld manpage lists _data, but this symbol doesn't work for shared libraries.

Thanks,
Andrew

Re: Move unwind info to read-only section on AIX

[Andrew Dixie]

Hi David,

As discussed, I have updated the to use SYMBOL-$ syntax for
PC-relative addressing and to use a new symbol, __gcc_unwind_dbase,
rather than overloading __dso_handle.

Regards,
Andrew

2014-09-09  Andrew Dixie  <andrewd@gentrack.com>

       Move exception tables to read-only memory on AIX.
       * collect2.c (write_c_file_stat): Provide dbase on AIX.
       (scan_prog_file): Don't output __dso_handle nor __gcc_unwind_dbase.
       * config/rs6000/aix.h (ASM_PREFERRED_EH_DATA_FORMAT): define.
       (EH_TABLES_CAN_BE_READ_ONLY): define.
       (ASM_OUTPUT_DWARF_PCREL): define.
       (ASM_OUTPUT_DWARF_DATAREL): define.
       (EH_FRAME_IN_DATA_SECTION): undefine.
       * config/rs6000/rs6000-aix.c: new file.
       (rs6000_aix_asm_output_dwarf_pcrel): new function.
       (rs6000_aix_asm_output_dwarf_datarel): new function.
       * config/rs6000/rs6000.c (rs6000_xcoff_asm_init_sections): remove
       assignment of exception_section.
       * dwarf2asm.c (dw2_asm_output_encoded_addr_rtx): Add call to
       ASM_OUTPUT_DWARF_DATAREL.
       * dwarf2out.c (switch_to_eh_frame_section): Add call to
       EMIT_COLLECT2_LABELS_FOR_EH_FRAMES.

diff -rupN orig/gcc-4.10-20140706/gcc/collect2.c
gcc-4.10-20140706/gcc/collect2.c
--- orig/gcc-4.10-20140706/gcc/collect2.c       2014-06-26
21:16:28.000000000 +1200
+++ gcc-4.10-20140706/gcc/collect2.c    2014-09-09 19:37:49.383271225 +1200
@@ -2101,12 +2110,23 @@ write_c_file_stat (FILE *stream, const c
       fprintf (stream, "  struct object *next;\n");
       fprintf (stream, "};\n");

+      fprintf (stream, "extern void __register_frame_info_table_bases
(void *, struct object *, void *tbase, void *dbase);\n");
       fprintf (stream, "extern void __register_frame_info_table (void
*, struct object *);\n");
       fprintf (stream, "extern void *__deregister_frame_info (void *);\n");
+#ifdef TARGET_AIX_VERSION
+      fprintf (stream, "extern void *__gcc_unwind_dbase;\n");
+#endif

       fprintf (stream, "static void reg_frame () {\n");
       fprintf (stream, "\tstatic struct object ob;\n");
+#ifdef TARGET_AIX_VERSION
+      /* As per config/rs6000/rs6000-aix.c, we use __gcc_unwind_dbase
as the dbase on AIX.
+        This might not be the start of the segment, we assume signed offsets.
+       */
+      fprintf (stream, "\t__register_frame_info_table_bases
(frame_table, &ob, (void *)0, &__gcc_unwind_dbase);\n");
+#else
       fprintf (stream, "\t__register_frame_info_table (frame_table, &ob);\n");
+#endif
       fprintf (stream, "\t}\n");

       fprintf (stream, "static void dereg_frame () {\n");
@@ -2868,7 +2888,16 @@ scan_prog_file (const char *prog_name, s
                             provides an explicit export list.  */
                          if (shared_obj && !is_shared
                              && which_pass == PASS_OBJ && !export_flag)
-                           add_to_list (&exports, name);
+                           {
+                                   /* Do not auto-export __dso_handle
or __gcc_unwind_dbase.
+                                      They are required to be local
to each module.
+                                    */
+                              if (strcmp(name, "__dso_handle") != 0
+                                              && strcmp(name,
"__gcc_unwind_dbase") != 0)
+                                {
+                                  add_to_list (&exports, name);
+                                }
+                           }
 #endif
                          continue;
                        }
diff -rupN orig/gcc-4.10-20140706/gcc/config/rs6000/aix61.h
gcc-4.10-20140706/gcc/config/rs6000/aix61.h
--- orig/gcc-4.10-20140706/gcc/config/rs6000/aix61.h    2014-01-03
11:23:26.000000000 +1300
+++ gcc-4.10-20140706/gcc/config/rs6000/aix61.h 2014-09-10
17:52:02.125993022 +1200
@@ -167,7 +167,7 @@ do {
                         \
    %{!maix64:\
      %{pthread:%{pg:gcrt0_r%O%s}%{!pg:%{p:mcrt0_r%O%s}%{!p:crt0_r%O%s}}}\
      %{!pthread:%{pg:gcrt0%O%s}%{!pg:%{p:mcrt0%O%s}%{!p:crt0%O%s}}}}}\
-   %{shared:crtcxa_s%O%s;:crtcxa%O%s}"
+   %{shared:crtcxa_s%O%s;:crtcxa%O%s} crtdbase%O%s"

 /* AIX V5 typedefs ptrdiff_t as "long" while earlier releases used "int".  */

diff -rupN orig/gcc-4.10-20140706/gcc/config/rs6000/rs6000-aix.c
gcc-4.10-20140706/gcc/config/rs6000/rs6000-aix.c
--- orig/gcc-4.10-20140706/gcc/config/rs6000/rs6000-aix.c
1970-01-01 12:00:00.000000000 +1200
+++ gcc-4.10-20140706/gcc/config/rs6000/rs6000-aix.c    2014-09-09
19:39:34.770296080 +1200
@@ -0,0 +1,48 @@
+/* Functions for AIX on PowerPC.
+   Copyright (C) 2014 Free Software Foundation, Inc.
+
+This file is part of GCC.
+
+GCC is free software; you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation; either version 3, or (at your option)
+any later version.
+
+GCC is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with GCC; see the file COPYING3.  If not see
+<http://www.gnu.org/licenses/>.  */
+
+#include "config.h"
+#include "system.h"
+#include "coretypes.h"
+#include "tm.h"
+#include "rtl.h"
+#include "tm_p.h"
+#include "output.h"
+
+/* Overide the default 'SYMBOL-.' syntax with AIX compatible 'SYMBOL-$'. */
+void rs6000_asm_output_dwarf_pcrel(FILE *file, int size, const char *label)
+{
+       fputs (integer_asm_op (size, FALSE), file);
+       assemble_name (file, label);
+       fputs ("-$", file);
+}
+
+/* Output a symbol offset relative to the dbase for the current object.
+   We use __gcc_unwind_dbase as an arbitrary base for dbase and
assume signed offsets.
+
+   __gcc_unwind_dbase is embedded in all executables/libraries through
+   libgcc/config/rs6000/crtdbase.S
+ */
+void rs6000_asm_output_dwarf_datarel(FILE *file, int size, const char *label)
+{
+       fputs (integer_asm_op (size, FALSE), file);
+       assemble_name (file, label);
+       fputs("-__gcc_unwind_dbase", file);
+}
+
diff -rupN orig/gcc-4.10-20140706/gcc/config/rs6000/rs6000.c
gcc-4.10-20140706/gcc/config/rs6000/rs6000.c
--- orig/gcc-4.10-20140706/gcc/config/rs6000/rs6000.c   2014-07-03
18:13:48.000000000 +1200
+++ gcc-4.10-20140706/gcc/config/rs6000/rs6000.c        2014-07-25
13:43:54.151118199 +1200
@@ -29251,7 +29251,6 @@ rs6000_xcoff_asm_init_sections (void)
     = get_unnamed_section (0, rs6000_xcoff_output_toc_section_asm_op, NULL);

   readonly_data_section = read_only_data_section;
-  exception_section = data_section;
 }

 static int
diff -rupN orig/gcc-4.10-20140706/gcc/config/rs6000/rs6000-protos.h
gcc-4.10-20140706/gcc/config/rs6000/rs6000-protos.h
--- orig/gcc-4.10-20140706/gcc/config/rs6000/rs6000-protos.h
2014-06-26 02:46:08.000000000 +1200
+++ gcc-4.10-20140706/gcc/config/rs6000/rs6000-protos.h 2014-07-28
12:41:54.852926208 +1200
@@ -195,6 +195,8 @@ extern void rs6000_aix_asm_output_dwarf_
 extern void get_ppc476_thunk_name (char name[32]);
 extern bool rs6000_overloaded_builtin_p (enum rs6000_builtins);
 extern HOST_WIDE_INT rs6000_builtin_mask_calculate (void);
+extern void rs6000_asm_output_dwarf_pcrel(FILE *file, int size, const
char *label);
+extern void rs6000_asm_output_dwarf_datarel(FILE *file, int size,
const char *label);

 /* Declare functions in rs6000-c.c */

diff -rupN orig/gcc-4.10-20140706/gcc/config/rs6000/t-aix52
gcc-4.10-20140706/gcc/config/rs6000/t-aix52
--- orig/gcc-4.10-20140706/gcc/config/rs6000/t-aix52    2014-01-03
11:23:26.000000000 +1300
+++ gcc-4.10-20140706/gcc/config/rs6000/t-aix52 2014-07-25
13:43:54.151118199 +1200
@@ -24,3 +24,7 @@ MULTILIB_OPTIONS      = pthread maix64
 MULTILIB_DIRNAMES      = pthread ppc64

 MULTILIB_MATCHES       =
+
+rs6000-aix.o: $(srcdir)/config/rs6000/rs6000-aix.c
+       $(COMPILE) $<
+       $(POSTCOMPILE)
diff -rupN orig/gcc-4.10-20140706/gcc/config/rs6000/xcoff.h
gcc-4.10-20140706/gcc/config/rs6000/xcoff.h
--- orig/gcc-4.10-20140706/gcc/config/rs6000/xcoff.h    2014-06-26
02:46:08.000000000 +1200
+++ gcc-4.10-20140706/gcc/config/rs6000/xcoff.h 2014-07-28
12:48:43.361071782 +1200
@@ -299,8 +299,25 @@
 #define DATA_SECTION_ASM_OP \
   "\t.csect .data[RW]," XCOFF_CSECT_DEFAULT_ALIGNMENT_STR

+/* The eh_frames are put in the read-only text segment.
+   Local code labels/function will also be in the local text segment so use
+   PC relative addressing.
+   Global symbols must be in the data segment to allow loader relocations.
+   So use DW_EH_PE_indirect to allocate a slot in the local data segment.
+   There is no constant offset to this data segment from the text segment,
+   so use addressing relative to the data segment.
+ */
+#define ASM_PREFERRED_EH_DATA_FORMAT(CODE,GLOBAL) \
+       ((GLOBAL) ? (DW_EH_PE_indirect | DW_EH_PE_datarel |
DW_EH_PE_sdata4) : (DW_EH_PE_pcrel | DW_EH_PE_sdata4))

-/* Define to prevent DWARF2 unwind info in the data section rather
-   than in the .eh_frame section.  We do this because the AIX linker
-   would otherwise garbage collect these sections.  */
-#define EH_FRAME_IN_DATA_SECTION 1
+#define EMIT_COLLECT2_LABELS_FOR_EH_FRAMES 1
+
+#define TARGET_TERMINATE_DW2_EH_FRAME_INFO true
+
+#define EH_TABLES_CAN_BE_READ_ONLY 1
+
+#define ASM_OUTPUT_DWARF_PCREL(FILE, SIZE, LABEL)       \
+       rs6000_asm_output_dwarf_pcrel(FILE, SIZE, LABEL);
+
+#define ASM_OUTPUT_DWARF_DATAREL(FILE, SIZE, LABEL)       \
+       rs6000_asm_output_dwarf_datarel(FILE, SIZE, LABEL);
diff -rupN orig/gcc-4.10-20140706/gcc/config.gcc
gcc-4.10-20140706/gcc/config.gcc
--- orig/gcc-4.10-20140706/gcc/config.gcc       2014-07-04
19:18:19.000000000 +1200
+++ gcc-4.10-20140706/gcc/config.gcc    2014-07-29 12:46:11.459162754 +1200
@@ -2369,6 +2369,7 @@ rs6000-ibm-aix5.1.* | powerpc-ibm-aix5.1
 rs6000-ibm-aix5.2.* | powerpc-ibm-aix5.2.*)
        tm_file="${tm_file} rs6000/aix.h rs6000/aix52.h rs6000/xcoff.h
rs6000/aix-stdint.h"
        tmake_file="rs6000/t-aix52 t-slibgcc"
+       extra_objs="$extra_objs rs6000-aix.o"
        extra_options="${extra_options} rs6000/aix64.opt"
        use_collect2=yes
        thread_file='aix'
@@ -2378,6 +2379,7 @@ rs6000-ibm-aix5.2.* | powerpc-ibm-aix5.2
 rs6000-ibm-aix5.3.* | powerpc-ibm-aix5.3.*)
        tm_file="${tm_file} rs6000/aix.h rs6000/aix53.h rs6000/xcoff.h
rs6000/aix-stdint.h"
        tmake_file="rs6000/t-aix52 t-slibgcc"
+       extra_objs="$extra_objs rs6000-aix.o"
        extra_options="${extra_options} rs6000/aix64.opt"
        use_collect2=yes
        thread_file='aix'
@@ -2387,6 +2389,7 @@ rs6000-ibm-aix5.3.* | powerpc-ibm-aix5.3
 rs6000-ibm-aix[6789].* | powerpc-ibm-aix[6789].*)
        tm_file="${tm_file} rs6000/aix.h rs6000/aix61.h rs6000/xcoff.h
rs6000/aix-stdint.h"
        tmake_file="rs6000/t-aix52 t-slibgcc"
+       extra_objs="$extra_objs rs6000-aix.o"
        extra_options="${extra_options} rs6000/aix64.opt"
        use_collect2=yes
        thread_file='aix'
diff -rupN orig/gcc-4.10-20140706/gcc/doc/tm.texi.in
gcc-4.10-20140706/gcc/doc/tm.texi.in
--- orig/gcc-4.10-20140706/gcc/doc/tm.texi.in   2014-06-06
13:04:22.000000000 +1200
+++ gcc-4.10-20140706/gcc/doc/tm.texi.in        2014-07-28
12:41:54.852926208 +1200
@@ -6525,6 +6525,15 @@ Do not define this macro unless @code{TA
 also defined.
 @end defmac

+@defmac EMIT_COLLECT2_LABELS_FOR_EH_FRAMES
+If defined, gcc will output specially named labels for DWARF 2 frame unwind
+information, and collect2 will detect these labels and emit code to register
+the frames at runtime.
+
+If not defined, the target must register frame unwind tables
+through another mechanism.
+@end defmac
+
 @defmac EH_TABLES_CAN_BE_READ_ONLY
 Define this macro to 1 if your target is such that no frame unwind
 information encoding used with non-PIC code will ever require a
@@ -7067,6 +7076,11 @@ A C statement to issue assembly directiv
 reference to the given @var{label}, using an integer of the given @var{size}.
 @end defmac

+@defmac ASM_OUTPUT_DWARF_DATAREL (@var{stream}, @var{size}, @var{label})
+A C statement to issue assembly directives that create a reference to the
+given @var{label} relative to the dbase, using an integer of the
given @var{size}.
+@end defmac
+
 @defmac ASM_OUTPUT_DWARF_TABLE_REF (@var{label})
 A C statement to issue assembly directives that create a reference to
 the DWARF table identifier @var{label} from the current section.  This
diff -rupN orig/gcc-4.10-20140706/gcc/dwarf2asm.c
gcc-4.10-20140706/gcc/dwarf2asm.c
--- orig/gcc-4.10-20140706/gcc/dwarf2asm.c      2014-01-03
11:23:26.000000000 +1300
+++ gcc-4.10-20140706/gcc/dwarf2asm.c   2014-07-25 14:07:57.050230529 +1200
@@ -992,6 +992,13 @@ dw2_asm_output_encoded_addr_rtx (int enc
          dw2_assemble_integer (size, addr);
          break;

+#ifdef ASM_OUTPUT_DWARF_DATAREL
+       case DW_EH_PE_datarel:
+         gcc_assert (GET_CODE (addr) == SYMBOL_REF);
+         ASM_OUTPUT_DWARF_DATAREL (asm_out_file, size, XSTR (addr, 0));
+         break;
+#endif
+
        case DW_EH_PE_pcrel:
          gcc_assert (GET_CODE (addr) == SYMBOL_REF);
 #ifdef ASM_OUTPUT_DWARF_PCREL
diff -rupN orig/gcc-4.10-20140706/gcc/dwarf2out.c
gcc-4.10-20140706/gcc/dwarf2out.c
--- orig/gcc-4.10-20140706/gcc/dwarf2out.c      2014-07-02
09:35:41.000000000 +1200
+++ gcc-4.10-20140706/gcc/dwarf2out.c   2014-07-28 12:48:12.925359014 +1200
@@ -424,6 +424,11 @@ static void
 switch_to_eh_frame_section (bool back)
 {
   tree label;
+#ifdef EMIT_COLLECT2_LABELS_FOR_EH_FRAMES
+  bool emit_collect2_labels = EMIT_COLLECT2_LABELS_FOR_EH_FRAMES;
+#else
+  bool emit_collect2_labels = true;
+#endif

 #ifdef EH_FRAME_SECTION_NAME
   if (eh_frame_section == 0)
@@ -464,7 +469,11 @@ switch_to_eh_frame_section (bool back)
       /* We have no special eh_frame section.  Put the information in
         the data section and emit special labels to guide collect2.  */
       switch_to_section (data_section);
+      emit_collect2_labels = true;
+    }

+  if (emit_collect2_labels)
+    {
       if (!back)
        {
          label = get_file_function_name ("F");
@@ -929,6 +938,9 @@ output_call_frame_info (int for_eh)
                    augmentation, any_lsda_needed, lsda_encoding);
     }

+  if (!targetm.terminate_dw2_eh_frame_info) {
+    abort();
+  }
   if (for_eh && targetm.terminate_dw2_eh_frame_info)
     dw2_asm_output_data (4, 0, "End of Table");

diff -rupN orig/gcc-4.10-20140706/libgcc/config/rs6000/crtdbase.c
gcc-4.10-20140706/libgcc/config/rs6000/crtdbase.c
--- orig/gcc-4.10-20140706/libgcc/config/rs6000/crtdbase.c
1970-01-01 12:00:00.000000000 +1200
+++ gcc-4.10-20140706/libgcc/config/rs6000/crtdbase.c   2014-09-10
17:28:48.189945402 +1200
@@ -0,0 +1,26 @@
+/* Defines __gcc_unwind_dbase
+
+   Copyright (C) 2014 Free Software Foundation, Inc.
+
+   This file is free software; you can redistribute it and/or modify it
+   under the terms of the GNU General Public License as published by the
+   Free Software Foundation; either version 3, or (at your option) any
+   later version.
+
+   This file is distributed in the hope that it will be useful, but
+   WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   General Public License for more details.
+
+   Under Section 7 of GPL version 3, you are granted additional
+   permissions described in the GCC Runtime Library Exception, version
+   3.1, as published by the Free Software Foundation.
+
+   You should have received a copy of the GNU General Public License and
+   a copy of the GCC Runtime Library Exception along with this program;
+   see the files COPYING3 and COPYING.RUNTIME respectively.  If not, see
+   <http://www.gnu.org/licenses/>.  */
+
+__asm__(".csect data[RW]\n__gcc_unwind_dbase:\n.globl __gcc_unwind_dbase");
+/* Zero length symbol used as an arbitrary base for offsets inside the data
+ * segment for unwind information. */
diff -rupN orig/gcc-4.10-20140706/libgcc/config/rs6000/t-aix-cxa
gcc-4.10-20140706/libgcc/config/rs6000/t-aix-cxa
--- orig/gcc-4.10-20140706/libgcc/config/rs6000/t-aix-cxa
2013-11-24 04:38:07.000000000 +1300
+++ gcc-4.10-20140706/libgcc/config/rs6000/t-aix-cxa    2014-09-10
16:36:16.850791121 +1200
@@ -5,6 +5,9 @@ LIB2ADD_ST += $(srcdir)/config/rs6000/ai

 SHLIB_MAPFILES += $(srcdir)/config/rs6000/libgcc-aix-cxa.ver

+crtdbase.o: $(srcdir)/config/rs6000/crtdbase.c
+       $(crt_compile) -c $<
+
 crtcxa.o: $(srcdir)/config/rs6000/crtcxa.c
        $(crt_compile) -c $<

diff -rupN orig/gcc-4.10-20140706/libgcc/config.host
gcc-4.10-20140706/libgcc/config.host
--- orig/gcc-4.10-20140706/libgcc/config.host   2014-05-21
23:08:58.000000000 +1200
+++ gcc-4.10-20140706/libgcc/config.host        2014-09-09
19:33:58.321410780 +1200
@@ -1027,7 +1027,7 @@ rs6000-ibm-aix5.1.* | powerpc-ibm-aix5.1
 rs6000-ibm-aix[56789].* | powerpc-ibm-aix[56789].*)
        md_unwind_header=rs6000/aix-unwind.h
        tmake_file="t-fdpbit rs6000/t-ppc64-fp rs6000/t-slibgcc-aix
rs6000/t-ibm-ldouble rs6000/t-aix-cxa"
-       extra_parts="crtcxa.o crtcxa_s.o"
+       extra_parts="crtcxa.o crtcxa_s.o crtdbase.o"
        ;;
 rl78-*-elf)
        tmake_file="$tm_file t-fdpbit rl78/t-rl78"

Re: Move unwind info to read-only section on AIX

[David Edelsohn]

On Thu, Sep 11, 2014 at 8:11 PM, Andrew Dixie <andrewdixie50@gmail.com> wrote:
> Hi David,
>
> As discussed, I have updated the to use SYMBOL-$ syntax for
> PC-relative addressing and to use a new symbol, __gcc_unwind_dbase,
> rather than overloading __dso_handle.
>
> Regards,
> Andrew
>
> 2014-09-09  Andrew Dixie  <andrewd@gentrack.com>
>
>        Move exception tables to read-only memory on AIX.
>        * collect2.c (write_c_file_stat): Provide dbase on AIX.
>        (scan_prog_file): Don't output __dso_handle nor __gcc_unwind_dbase.
>        * config/rs6000/aix.h (ASM_PREFERRED_EH_DATA_FORMAT): define.
>        (EH_TABLES_CAN_BE_READ_ONLY): define.
>        (ASM_OUTPUT_DWARF_PCREL): define.
>        (ASM_OUTPUT_DWARF_DATAREL): define.
>        (EH_FRAME_IN_DATA_SECTION): undefine.
>        * config/rs6000/rs6000-aix.c: new file.
>        (rs6000_aix_asm_output_dwarf_pcrel): new function.
>        (rs6000_aix_asm_output_dwarf_datarel): new function.
>        * config/rs6000/rs6000.c (rs6000_xcoff_asm_init_sections): remove
>        assignment of exception_section.
>        * dwarf2asm.c (dw2_asm_output_encoded_addr_rtx): Add call to
>        ASM_OUTPUT_DWARF_DATAREL.
>        * dwarf2out.c (switch_to_eh_frame_section): Add call to
>        EMIT_COLLECT2_LABELS_FOR_EH_FRAMES.

Hi, Andrew

Thanks for this work. The patch itself looks good. I cannot approve
the changes to dwarf2asm.c and dwarf2out.c.

My concern is this really is an ABI change.  One cannot mix object
files compiled with the old EH format and with the new EH format. One
can mix shared libraries of different styles, but not object files.
I'm not certain how much of a problem it is to require recompiling
everything from scratch.

Thanks, David

Re: Move unwind info to read-only section on AIX

[Andrew Dixie]

Hi David,

On AIX, mixing objects from different GCC versions should work as long
as the newest GCC is used for linking.  I tested a library with some
objects compiled by GCC-4.4 and some objects compiled by a patched
GCC-5.0.  Exceptions passed through the mixed objects without issue.
Did you have a specific example or reason to think it wouldn't work?

If one attempts to mix C++ objects from GCC-3.3 and GCC-4.4 on Linux
or Solaris there are link time errors.  If you're worried about the
risk of runtime problems, would you rather have a deliberate link time
error on AIX?

Regards,
Andrew

On Wed, Sep 17, 2014 at 6:37 AM, David Edelsohn <dje.gcc@gmail.com> wrote:
> My concern is this really is an ABI change.  One cannot mix object
> files compiled with the old EH format and with the new EH format. One
> can mix shared libraries of different styles, but not object files.
> I'm not certain how much of a problem it is to require recompiling
> everything from scratch.

Re: Move unwind info to read-only section on AIX

[David Edelsohn]

Hi, Andrew

I can understand the requirement of linking with the newer version GCC
to provide the dbase symbol. I'm just trying to understand how EH will
work when some objects have EH with absolute references in the data
section and some have relative references in the text section. I guess
the information is completely local to the object and only has to be
self-consistent within the object.

I want to make sure that there are no additional problems of a throw()
passing through objects that change the location of the EH
information.

Thanks, David

On Tue, Sep 16, 2014 at 5:04 PM, Andrew Dixie <andrewdixie50@gmail.com> wrote:
> Hi David,
>
> On AIX, mixing objects from different GCC versions should work as long
> as the newest GCC is used for linking.  I tested a library with some
> objects compiled by GCC-4.4 and some objects compiled by a patched
> GCC-5.0.  Exceptions passed through the mixed objects without issue.
> Did you have a specific example or reason to think it wouldn't work?
>
> If one attempts to mix C++ objects from GCC-3.3 and GCC-4.4 on Linux
> or Solaris there are link time errors.  If you're worried about the
> risk of runtime problems, would you rather have a deliberate link time
> error on AIX?
>
> Regards,
> Andrew
>
> On Wed, Sep 17, 2014 at 6:37 AM, David Edelsohn <dje.gcc@gmail.com> wrote:
>> My concern is this really is an ABI change.  One cannot mix object
>> files compiled with the old EH format and with the new EH format. One
>> can mix shared libraries of different styles, but not object files.
>> I'm not certain how much of a problem it is to require recompiling
>> everything from scratch.