binutils/readelf: handle AMDGPU relocation types

Make readelf recognize AMDGPU relocation types, as documented here: https://llvm.org/docs/AMDGPUUsage.html#amdgpu-relocation-records The user-visible change looks like: -000000000004 000400000001 unrecognized: 1 0000000000000000 SCRATCH_RSRC_DWORD0 -00000000000c 000500000001 unrecognized: 1 0000000000000000 SCRATCH_RSRC_DWORD1 -000000000014 000600000007 unrecognized: 7 0000000000000000 global_var0 -00000000001c 000700000008 unrecognized: 8 0000000000000000 global_var1 -000000000024 000800000009 unrecognized: 9 0000000000000000 global_var2 -00000000002c 00090000000a unrecognized: a 0000000000000000 global_var3 -000000000034 000a0000000b unrecognized: b 0000000000000000 global_var4 +000000000004 000400000001 R_AMDGPU_ABS32_LO 0000000000000000 SCRATCH_RSRC_DWORD0 +00000000000c 000500000001 R_AMDGPU_ABS32_LO 0000000000000000 SCRATCH_RSRC_DWORD1 +000000000014 000600000007 R_AMDGPU_GOTPCREL 0000000000000000 global_var0 +00000000001c 000700000008 R_AMDGPU_GOTPCREL 0000000000000000 global_var1 +000000000024 000800000009 R_AMDGPU_GOTPCREL 0000000000000000 global_var2 +00000000002c 00090000000a R_AMDGPU_REL32_LO 0000000000000000 global_var3 +000000000034 000a0000000b R_AMDGPU_REL32_HI 0000000000000000 global_var4 binutils/ChangeLog: * readelf.c (dump_relocations): Handle EM_AMDGPU. include/ChangeLog: * elf/amdgpu.h: Add relocation values. Change-Id: I2ed4589f4cd37ea11ad2e0cb38d4b682271e1334
binutils/readelf: build against msgpack, dump NT_AMDGPU_METADATA note contents
2022-03-15 15:09:55 -04:00 · 2022-03-15 15:09:55 -04:00 · 2022-03-15 15:09:55 -04:00 · 2022-03-15 15:09:55 -04:00 · 2022-03-15 15:09:55 -04:00 · 2022-03-15 15:09:55 -04:00
21 changed files with 852 additions and 19 deletions
--- a/bfd/Makefile.am
+++ b/bfd/Makefile.am
@@ -97,6 +97,7 @@ BFD64_LIBS_CFILES = archive64.c
 ALL_MACHINES = \
 	cpu-aarch64.lo \
 	cpu-alpha.lo \
+	cpu-amdgcn.lo \
 	cpu-arc.lo \
 	cpu-arm.lo \
 	cpu-avr.lo \
@@ -182,6 +183,7 @@ ALL_MACHINES = \
 ALL_MACHINES_CFILES = \
 	cpu-aarch64.c \
 	cpu-alpha.c \
+	cpu-amdgcn.c \
 	cpu-arc.c \
 	cpu-arm.c \
 	cpu-avr.c \
@@ -550,6 +552,7 @@ BFD64_BACKENDS = \
 	elf32-score.lo \
 	elf32-score7.lo \
 	elf64-alpha.lo \
+	elf64-amdgcn.lo \
 	elf64-gen.lo \
 	elf64-hppa.lo \
 	elf64-ia64.lo \
@@ -596,6 +599,7 @@ BFD64_BACKENDS_CFILES = \
 	elf32-score.c \
 	elf32-score7.c \
 	elf64-alpha.c \
+	elf64-amdgcn.c \
 	elf64-gen.c \
 	elf64-hppa.c \
 	elf64-ia64-vms.c \
--- a/bfd/Makefile.in
+++ b/bfd/Makefile.in
@@ -564,6 +564,7 @@ BFD64_LIBS_CFILES = archive64.c
 ALL_MACHINES = \
 	cpu-aarch64.lo \
 	cpu-alpha.lo \
+	cpu-amdgcn.lo \
 	cpu-arc.lo \
 	cpu-arm.lo \
 	cpu-avr.lo \
@@ -649,6 +650,7 @@ ALL_MACHINES = \
 ALL_MACHINES_CFILES = \
 	cpu-aarch64.c \
 	cpu-alpha.c \
+	cpu-amdgcn.c \
 	cpu-arc.c \
 	cpu-arm.c \
 	cpu-avr.c \
@@ -1019,6 +1021,7 @@ BFD64_BACKENDS = \
 	elf32-score.lo \
 	elf32-score7.lo \
 	elf64-alpha.lo \
+	elf64-amdgcn.lo \
 	elf64-gen.lo \
 	elf64-hppa.lo \
 	elf64-ia64.lo \
@@ -1065,6 +1068,7 @@ BFD64_BACKENDS_CFILES = \
 	elf32-score.c \
 	elf32-score7.c \
 	elf64-alpha.c \
+	elf64-amdgcn.c \
 	elf64-gen.c \
 	elf64-hppa.c \
 	elf64-ia64-vms.c \
@@ -1480,6 +1484,7 @@ distclean-compile:
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/corefile.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/cpu-aarch64.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/cpu-alpha.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/cpu-amdgcn.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/cpu-arc.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/cpu-arm.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/cpu-avr.Plo@am__quote@
@@ -1646,6 +1651,7 @@ distclean-compile:
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/elf32.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/elf64-aarch64.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/elf64-alpha.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/elf64-amdgcn.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/elf64-bpf.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/elf64-gen.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/elf64-hppa.Plo@am__quote@
--- a/bfd/archures.c
+++ b/bfd/archures.c
@@ -559,6 +559,19 @@ DESCRIPTION
 .  bfd_arch_loongarch,       {* LoongArch *}
 .#define bfd_mach_loongarch32	1
 .#define bfd_mach_loongarch64	2
+.  bfd_arch_amdgcn,     {* AMDGCN *}
+.#define bfd_mach_amdgcn_unknown 0x000
+.#define bfd_mach_amdgcn_gfx900  0x02c
+.#define bfd_mach_amdgcn_gfx904  0x02e
+.#define bfd_mach_amdgcn_gfx906  0x02f
+.#define bfd_mach_amdgcn_gfx908  0x030
+.#define bfd_mach_amdgcn_gfx90a  0x03f
+.#define bfd_mach_amdgcn_gfx1010 0x033
+.#define bfd_mach_amdgcn_gfx1011 0x034
+.#define bfd_mach_amdgcn_gfx1012 0x035
+.#define bfd_mach_amdgcn_gfx1030 0x036
+.#define bfd_mach_amdgcn_gfx1031 0x037
+.#define bfd_mach_amdgcn_gfx1032 0x038
 .  bfd_arch_last
 .  };
 */
@@ -614,6 +627,7 @@ DESCRIPTION

 extern const bfd_arch_info_type bfd_aarch64_arch;
 extern const bfd_arch_info_type bfd_alpha_arch;
+extern const bfd_arch_info_type bfd_amdgcn_arch;
 extern const bfd_arch_info_type bfd_arc_arch;
 extern const bfd_arch_info_type bfd_arm_arch;
 extern const bfd_arch_info_type bfd_avr_arch;
@@ -704,6 +718,7 @@ static const bfd_arch_info_type * const bfd_archures_list[] =
 #else
    &bfd_aarch64_arch,
    &bfd_alpha_arch,
+    &bfd_amdgcn_arch,
    &bfd_arc_arch,
    &bfd_arm_arch,
    &bfd_avr_arch,
--- a/bfd/bfd-in2.h
+++ b/bfd/bfd-in2.h
@@ -1938,6 +1938,19 @@ enum bfd_architecture
  bfd_arch_loongarch,       /* LoongArch */
 #define bfd_mach_loongarch32   1
 #define bfd_mach_loongarch64   2
+  bfd_arch_amdgcn,     /* AMDGCN */
+#define bfd_mach_amdgcn_unknown 0x000
+#define bfd_mach_amdgcn_gfx900  0x02c
+#define bfd_mach_amdgcn_gfx904  0x02e
+#define bfd_mach_amdgcn_gfx906  0x02f
+#define bfd_mach_amdgcn_gfx908  0x030
+#define bfd_mach_amdgcn_gfx90a  0x03f
+#define bfd_mach_amdgcn_gfx1010 0x033
+#define bfd_mach_amdgcn_gfx1011 0x034
+#define bfd_mach_amdgcn_gfx1012 0x035
+#define bfd_mach_amdgcn_gfx1030 0x036
+#define bfd_mach_amdgcn_gfx1031 0x037
+#define bfd_mach_amdgcn_gfx1032 0x038
  bfd_arch_last
  };

--- a/bfd/config.bfd
+++ b/bfd/config.bfd
@@ -184,6 +184,7 @@ alpha*)		 targ_archs=bfd_alpha_arch ;;
 am33_2.0*)	 targ_archs=bfd_mn10300_arch ;;
 arc*)		 targ_archs=bfd_arc_arch ;;
 arm*)		 targ_archs=bfd_arm_arch ;;
+amdgcn*)	 targ_archs=bfd_amdgcn_arch ;;
 bfin*)		 targ_archs=bfd_bfin_arch ;;
 c30*)		 targ_archs=bfd_tic30_arch ;;
 c4x*)		 targ_archs=bfd_tic4x_arch ;;
@@ -321,6 +322,10 @@ case "${targ}" in
    targ_defvec=alpha_ecoff_le_vec
    want64=true
    ;;
+  amdgcn-*-*)
+    targ_defvec=amdgcn_elf64_le_vec
+    want64=true
+    ;;
  ia64*-*-freebsd* | ia64*-*-netbsd* | ia64*-*-linux-* | ia64*-*-elf* | ia64*-*-kfreebsd*-gnu)
    targ_defvec=ia64_elf64_le_vec
    targ_selvecs="ia64_elf64_be_vec ia64_pei_vec"
--- a/bfd/configure
+++ b/bfd/configure
@@ -13364,6 +13364,7 @@ do
    alpha_vms_vec)		 tb="$tb vms-alpha.lo vms-misc.lo vms-lib.lo"; target_size=64 ;;
    alpha_vms_lib_txt_vec)	 tb="$tb vms-lib.lo vms-misc.lo" ;;
    am33_elf32_linux_vec)	 tb="$tb elf32-am33lin.lo elf32.lo $elf" ;;
+    amdgcn_elf64_le_vec)	 tb="$tb elf64-amdgcn.lo elf64.lo $elf"; target_size=64 ;;
    aout0_be_vec)		 tb="$tb aout0.lo aout32.lo" ;;
    aout64_vec)			 tb="$tb demo64.lo aout64.lo"; target_size=64 ;;
    aout_vec)			 tb="$tb host-aout.lo aout32.lo" ;;
--- a/bfd/configure.ac
+++ b/bfd/configure.ac
@@ -443,6 +443,7 @@ do
    alpha_vms_vec)		 tb="$tb vms-alpha.lo vms-misc.lo vms-lib.lo"; target_size=64 ;;
    alpha_vms_lib_txt_vec)	 tb="$tb vms-lib.lo vms-misc.lo" ;;
    am33_elf32_linux_vec)	 tb="$tb elf32-am33lin.lo elf32.lo $elf" ;;
+    amdgcn_elf64_le_vec)	 tb="$tb elf64-amdgcn.lo elf64.lo $elf"; target_size=64 ;;
    aout0_be_vec)		 tb="$tb aout0.lo aout32.lo" ;;
    aout64_vec)			 tb="$tb demo64.lo aout64.lo"; target_size=64 ;;
    aout_vec)			 tb="$tb host-aout.lo aout32.lo" ;;
--- a/bfd/cpu-amdgcn.c
+++ b/bfd/cpu-amdgcn.c
@@ -0,0 +1,59 @@
+/* BFD support for the AMDGCN GPU architecture.
+
+   Copyright (C) 2019-2022 Free Software Foundation, Inc.
+
+   This file is part of BFD, the Binary File Descriptor library.
+
+   This program 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 of the License, or
+   (at your option) any later version.
+
+   This program 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 this program.  If not, see <http://www.gnu.org/licenses/>.  */
+
+#include "sysdep.h"
+#include "bfd.h"
+#include "libbfd.h"
+
+#define N(MACHINE, PRINTABLE_NAME, DEFAULT, NEXT)       \
+  {                                                     \
+    32, /* 32 bits in a word */                         \
+    64, /* 64 bits in an address */                     \
+    8,  /* 8 bits in a byte */                          \
+    bfd_arch_amdgcn,                                    \
+    MACHINE,                                            \
+    "amdgcn",                                           \
+    PRINTABLE_NAME,                                     \
+    3, /* section align power */                        \
+    DEFAULT,                                            \
+    bfd_default_compatible,                             \
+    bfd_default_scan,                                   \
+    bfd_arch_default_fill,                              \
+    NEXT,                                               \
+    0                                                   \
+  }
+
+#define NN(index) (&arch_info_struct[index])
+
+static const bfd_arch_info_type arch_info_struct[] =
+{
+  N (bfd_mach_amdgcn_gfx904, "amdgcn:gfx904", false, NN (1)),
+  N (bfd_mach_amdgcn_gfx906, "amdgcn:gfx906", false, NN (2)),
+  N (bfd_mach_amdgcn_gfx908, "amdgcn:gfx908", false, NN (3)),
+  N (bfd_mach_amdgcn_gfx90a, "amdgcn:gfx90a", false, NN (4)),
+  N (bfd_mach_amdgcn_gfx1010, "amdgcn:gfx1010", false, NN (5)),
+  N (bfd_mach_amdgcn_gfx1011, "amdgcn:gfx1011", false, NN (6)),
+  N (bfd_mach_amdgcn_gfx1012, "amdgcn:gfx1012", false, NN (7)),
+  N (bfd_mach_amdgcn_gfx1030, "amdgcn:gfx1030", false, NN (8)),
+  N (bfd_mach_amdgcn_gfx1031, "amdgcn:gfx1031", false, NN (9)),
+  N (bfd_mach_amdgcn_gfx1032, "amdgcn:gfx1032", false, NULL)
+};
+
+const bfd_arch_info_type bfd_amdgcn_arch =
+  N (bfd_mach_amdgcn_gfx900, "amdgcn:gfx900", true, NN (0));
--- a/bfd/elf-bfd.h
+++ b/bfd/elf-bfd.h
@@ -499,6 +499,7 @@ enum elf_target_id
 {
  AARCH64_ELF_DATA = 1,
  ALPHA_ELF_DATA,
+  AMDGCN_ELF_DATA,
  ARC_ELF_DATA,
  ARM_ELF_DATA,
  AVR_ELF_DATA,
--- a/bfd/elf64-amdgcn.c
+++ b/bfd/elf64-amdgcn.c
@@ -0,0 +1,80 @@
+/* AMDGCN ELF support for BFD.
+
+   Copyright (C) 2019-2022 Free Software Foundation, Inc.
+
+   This file is part of BFD, the Binary File Descriptor library.
+
+   This program 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 of the License, or
+   (at your option) any later version.
+
+   This program 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 this program.  If not, see <http://www.gnu.org/licenses/>.  */
+
+/* This file handles ELF files that are of the AMDGCN architecture.  The
+   format is documented here:
+
+     https://llvm.org/docs/AMDGPUUsage.html#elf-code-object */
+
+#include "sysdep.h"
+#include "bfd.h"
+#include "libbfd.h"
+#include "elf-bfd.h"
+#include "elf/amdgpu.h"
+
+#include <string.h>
+
+static bool
+elf64_amdgcn_object_p (bfd *abfd)
+{
+  Elf_Internal_Ehdr *hdr = elf_elfheader (abfd);
+  unsigned int mach;
+  unsigned char osabi;
+  unsigned char osabi_version;
+
+  BFD_ASSERT (hdr->e_machine == EM_AMDGPU);
+
+  osabi = hdr->e_ident[EI_OSABI];
+  osabi_version = hdr->e_ident[EI_ABIVERSION];
+
+  /* Objects with OS ABI HSA version 2 encoded the GPU model differently (in a
+     note), but they are deprecated, so we don't need to support them.  Reject
+     them specifically.
+
+     At the time of writing, all AMDGCN objects encode the specific GPU
+     model in the EF_AMDGPU_MACH field of e_flags.  */
+  if (osabi == ELFOSABI_AMDGPU_HSA
+      && osabi_version < ELFABIVERSION_AMDGPU_HSA_V3)
+    return false;
+
+  mach = elf_elfheader (abfd)->e_flags & EF_AMDGPU_MACH;
+
+  /* Avoid matching non-AMDGCN AMDGPU objects (e.g. r600).  */
+  if (mach < EF_AMDGPU_MACH_AMDGCN_MIN)
+    return false;
+
+  bfd_default_set_arch_mach (abfd, bfd_arch_amdgcn, mach);
+  return true;
+}
+
+
+#define TARGET_LITTLE_SYM	amdgcn_elf64_le_vec
+#define TARGET_LITTLE_NAME	"elf64-amdgcn"
+#define ELF_ARCH		bfd_arch_amdgcn
+#define ELF_TARGET_ID		AMDGCN_ELF_DATA
+#define ELF_MACHINE_CODE	EM_AMDGPU
+#define ELF_MAXPAGESIZE		0x10000 /* 64KB */
+#define ELF_COMMONPAGESIZE	0x1000  /* 4KB */
+
+#define bfd_elf64_bfd_reloc_type_lookup bfd_default_reloc_type_lookup
+#define bfd_elf64_bfd_reloc_name_lookup _bfd_norelocs_bfd_reloc_name_lookup
+
+#define elf_backend_object_p elf64_amdgcn_object_p
+
+#include "elf64-target.h"
--- a/bfd/targets.c
+++ b/bfd/targets.c
@@ -686,6 +686,7 @@ extern const bfd_target alpha_elf64_fbsd_vec;
 extern const bfd_target alpha_vms_vec;
 extern const bfd_target alpha_vms_lib_txt_vec;
 extern const bfd_target am33_elf32_linux_vec;
+extern const bfd_target amdgcn_elf64_le_vec;
 extern const bfd_target aout_vec;
 extern const bfd_target arc_elf32_be_vec;
 extern const bfd_target arc_elf32_le_vec;
@@ -997,6 +998,10 @@ static const bfd_target * const _bfd_target_vector[] =
 	&aarch64_pei_vec,
 #endif

+#ifdef BFD64
+	&amdgcn_elf64_le_vec,
+#endif
+
 #ifdef BFD64
 	&alpha_ecoff_le_vec,
 	&alpha_elf64_vec,
--- a/binutils/Makefile.am
+++ b/binutils/Makefile.am
@@ -256,7 +256,7 @@ objcopy_SOURCES = objcopy.c not-strip.c rename.c $(WRITE_DEBUG_SRCS) $(BULIBS)
 strings_SOURCES = strings.c $(BULIBS)

 readelf_SOURCES = readelf.c version.c unwind-ia64.c dwarf.c demanguse.c $(ELFLIBS)
-readelf_LDADD   = $(LIBCTF_NOBFD) $(LIBINTL) $(LIBIBERTY) $(ZLIB) $(DEBUGINFOD_LIBS)
+readelf_LDADD   = $(LIBCTF_NOBFD) $(LIBINTL) $(LIBIBERTY) $(ZLIB) $(DEBUGINFOD_LIBS) $(MSGPACK_LIBS)

 elfedit_SOURCES = elfedit.c version.c $(ELFLIBS)
 elfedit_LDADD = $(LIBINTL) $(LIBIBERTY)
@@ -381,6 +381,18 @@ endif
 	$(AM_V_CC)$(COMPILE) $(DEBUGINFOD_CFLAGS) -c -o $@ $(srcdir)/dwarf.c
 endif

+readelf.@OBJEXT@: readelf.c
+if am__fastdepCC
+	$(AM_V_CC)$(COMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo $(MSGPACK_CFLAGS) -c -o $@ $(srcdir)/readelf.c
+	$(AM_V_at)mv -f $(DEPDIR)/$*.Tpo $(DEPDIR)/$*.Po
+else
+if AMDEP
+	source='readelf.c' object='$@' libtool=no @AMDEPBACKSLASH@
+	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+endif
+	$(AM_V_CC)$(COMPILE) $(MSGPACK_CFLAGS) -c -o $@ $(srcdir)/readelf.c
+endif
+
 sysroff.@OBJEXT@: sysroff.c
 if am__fastdepCC
 	$(AM_V_CC)$(COMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ `test -f sysroff.c || echo $(srcdir)/`sysroff.c $(NO_WERROR)
--- a/binutils/Makefile.in
+++ b/binutils/Makefile.in
@@ -537,6 +537,8 @@ MKDIR_P = @MKDIR_P@
 MKINSTALLDIRS = @MKINSTALLDIRS@
 MSGFMT = @MSGFMT@
 MSGMERGE = @MSGMERGE@
+MSGPACK_CFLAGS = @MSGPACK_CFLAGS@
+MSGPACK_LIBS = @MSGPACK_LIBS@
 NM = @NM@
 NMEDIT = @NMEDIT@
 NO_WERROR = @NO_WERROR@
@@ -785,7 +787,7 @@ size_SOURCES = size.c $(BULIBS)
 objcopy_SOURCES = objcopy.c not-strip.c rename.c $(WRITE_DEBUG_SRCS) $(BULIBS)
 strings_SOURCES = strings.c $(BULIBS)
 readelf_SOURCES = readelf.c version.c unwind-ia64.c dwarf.c demanguse.c $(ELFLIBS)
-readelf_LDADD = $(LIBCTF_NOBFD) $(LIBINTL) $(LIBIBERTY) $(ZLIB) $(DEBUGINFOD_LIBS)
+readelf_LDADD = $(LIBCTF_NOBFD) $(LIBINTL) $(LIBIBERTY) $(ZLIB) $(DEBUGINFOD_LIBS) $(MSGPACK_LIBS)
 elfedit_SOURCES = elfedit.c version.c $(ELFLIBS)
 elfedit_LDADD = $(LIBINTL) $(LIBIBERTY)
 strip_new_SOURCES = objcopy.c is-strip.c rename.c $(WRITE_DEBUG_SRCS) $(BULIBS)
@@ -1919,6 +1921,13 @@ dwarf.@OBJEXT@: dwarf.c
@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
@am__fastdepCC_FALSE@	$(AM_V_CC)$(COMPILE) $(DEBUGINFOD_CFLAGS) -c -o $@ $(srcdir)/dwarf.c

+readelf.@OBJEXT@: readelf.c
+@am__fastdepCC_TRUE@	$(AM_V_CC)$(COMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo $(MSGPACK_CFLAGS) -c -o $@ $(srcdir)/readelf.c
+@am__fastdepCC_TRUE@	$(AM_V_at)mv -f $(DEPDIR)/$*.Tpo $(DEPDIR)/$*.Po
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='readelf.c' object='$@' libtool=no @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(AM_V_CC)$(COMPILE) $(MSGPACK_CFLAGS) -c -o $@ $(srcdir)/readelf.c
+
 sysroff.@OBJEXT@: sysroff.c
@am__fastdepCC_TRUE@	$(AM_V_CC)$(COMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ `test -f sysroff.c || echo $(srcdir)/`sysroff.c $(NO_WERROR)
@am__fastdepCC_TRUE@	$(AM_V_at)mv -f $(DEPDIR)/$*.Tpo $(DEPDIR)/$*.Po
--- a/binutils/config.in
+++ b/binutils/config.in
@@ -104,6 +104,9 @@
 /* Define to 1 if you have a working `mmap' system call. */
 #undef HAVE_MMAP

+/* Define to 1 if msgpack is available. */
+#undef HAVE_MSGPACK
+
 /* Define to 1 if you have the `sbrk' function. */
 #undef HAVE_SBRK

--- a/binutils/configure
+++ b/binutils/configure
@@ -648,6 +648,8 @@ BUILD_DLLTOOL
 BUILD_SRCONV
 LTLIBICONV
 LIBICONV
+MSGPACK_LIBS
+MSGPACK_CFLAGS
 zlibinc
 zlibdir
 DEMANGLER_NAME
@@ -830,6 +832,7 @@ enable_build_warnings
 enable_nls
 enable_maintainer_mode
 with_system_zlib
+with_msgpack
 enable_rpath
 with_libiconv_prefix
 with_libiconv_type
@@ -849,7 +852,9 @@ PKG_CONFIG_LIBDIR
 DEBUGINFOD_CFLAGS
 DEBUGINFOD_LIBS
 YACC
-YFLAGS'
+YFLAGS
+MSGPACK_CFLAGS
+MSGPACK_LIBS'


 # Initialize some variables set by options.
@@ -1512,6 +1517,7 @@ Optional Packages:
  --with-debuginfod       Enable debuginfo lookups with debuginfod
                          (auto/yes/no)
  --with-system-zlib      use installed libz
+  --with-msgpack          Enable msgpack support (auto/yes/no)
  --with-gnu-ld           assume the C compiler uses GNU ld default=no
  --with-libiconv-prefix[=DIR]  search for libiconv in DIR/include and DIR/lib
  --without-libiconv-prefix     don't search for libiconv in includedir and libdir
@@ -1541,6 +1547,10 @@ Some influential environment variables:
  YFLAGS      The list of arguments that will be passed by default to $YACC.
              This script will default YFLAGS to the empty string to avoid a
              default value of `-d' given by some make applications.
+  MSGPACK_CFLAGS
+              C compiler flags for MSGPACK, overriding pkg-config
+  MSGPACK_LIBS
+              linker flags for MSGPACK, overriding pkg-config

 Use these variables to override the choices made by `configure' or to help
 it to find libraries and programs with nonstandard names/locations.
@@ -10971,7 +10981,7 @@ else
  lt_dlunknown=0; lt_dlno_uscore=1; lt_dlneed_uscore=2
  lt_status=$lt_dlunknown
  cat > conftest.$ac_ext <<_LT_EOF
-#line 10974 "configure"
+#line 10984 "configure"
 #include "confdefs.h"

 #if HAVE_DLFCN_H
@@ -11077,7 +11087,7 @@ else
  lt_dlunknown=0; lt_dlno_uscore=1; lt_dlneed_uscore=2
  lt_status=$lt_dlunknown
  cat > conftest.$ac_ext <<_LT_EOF
-#line 11080 "configure"
+#line 11090 "configure"
 #include "confdefs.h"

 #if HAVE_DLFCN_H
@@ -13723,6 +13733,127 @@ $as_echo "#define USE_BINARY_FOPEN 1" >>confdefs.h
 ;;
 esac

+# Support for the msgpack C library.
+
+# Check whether --with-msgpack was given.
+if test "${with_msgpack+set}" = set; then :
+  withval=$with_msgpack;
+else
+  with_msgpack=auto
+fi
+
+
+
+if test "$with_msgpack" != no; then
+
+pkg_failed=no
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for msgpack" >&5
+$as_echo_n "checking for msgpack... " >&6; }
+
+if test -n "$MSGPACK_CFLAGS"; then
+    pkg_cv_MSGPACK_CFLAGS="$MSGPACK_CFLAGS"
+ elif test -n "$PKG_CONFIG"; then
+    if test -n "$PKG_CONFIG" && \
+    { { $as_echo "$as_me:${as_lineno-$LINENO}: \$PKG_CONFIG --exists --print-errors \"msgpack\""; } >&5
+  ($PKG_CONFIG --exists --print-errors "msgpack") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then
+  pkg_cv_MSGPACK_CFLAGS=`$PKG_CONFIG --cflags "msgpack" 2>/dev/null`
+		      test "x$?" != "x0" && pkg_failed=yes
+else
+  pkg_failed=yes
+fi
+ else
+    pkg_failed=untried
+fi
+if test -n "$MSGPACK_LIBS"; then
+    pkg_cv_MSGPACK_LIBS="$MSGPACK_LIBS"
+ elif test -n "$PKG_CONFIG"; then
+    if test -n "$PKG_CONFIG" && \
+    { { $as_echo "$as_me:${as_lineno-$LINENO}: \$PKG_CONFIG --exists --print-errors \"msgpack\""; } >&5
+  ($PKG_CONFIG --exists --print-errors "msgpack") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then
+  pkg_cv_MSGPACK_LIBS=`$PKG_CONFIG --libs "msgpack" 2>/dev/null`
+		      test "x$?" != "x0" && pkg_failed=yes
+else
+  pkg_failed=yes
+fi
+ else
+    pkg_failed=untried
+fi
+
+if test $pkg_failed = no; then
+  pkg_save_LDFLAGS="$LDFLAGS"
+  LDFLAGS="$LDFLAGS $pkg_cv_MSGPACK_LIBS"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+
+else
+  pkg_failed=yes
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+  LDFLAGS=$pkg_save_LDFLAGS
+fi
+
+
+
+if test $pkg_failed = yes; then
+        { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+
+if $PKG_CONFIG --atleast-pkgconfig-version 0.20; then
+        _pkg_short_errors_supported=yes
+else
+        _pkg_short_errors_supported=no
+fi
+        if test $_pkg_short_errors_supported = yes; then
+	        MSGPACK_PKG_ERRORS=`$PKG_CONFIG --short-errors --print-errors --cflags --libs "msgpack" 2>&1`
+        else
+	        MSGPACK_PKG_ERRORS=`$PKG_CONFIG --print-errors --cflags --libs "msgpack" 2>&1`
+        fi
+	# Put the nasty error message in config.log where it belongs
+	echo "$MSGPACK_PKG_ERRORS" >&5
+
+
+      if test "$with_msgpack" = yes; then
+	as_fn_error $? "--with-msgpack was given, but msgpack is missing or unusable." "$LINENO" 5
+      fi
+
+elif test $pkg_failed = untried; then
+        { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+
+      if test "$with_msgpack" = yes; then
+	as_fn_error $? "--with-msgpack was given, but msgpack is missing or unusable." "$LINENO" 5
+      fi
+
+else
+	MSGPACK_CFLAGS=$pkg_cv_MSGPACK_CFLAGS
+	MSGPACK_LIBS=$pkg_cv_MSGPACK_LIBS
+        { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+
+
+$as_echo "#define HAVE_MSGPACK 1" >>confdefs.h
+
+
+fi
+fi
+
 # target-specific stuff:

 # Canonicalize the secondary target names.
--- a/binutils/configure.ac
+++ b/binutils/configure.ac
@@ -275,6 +275,23 @@ AM_ZLIB

 BFD_BINARY_FOPEN

+# Support for the msgpack C library.
+AC_ARG_WITH([msgpack],
+	    AC_HELP_STRING([--with-msgpack], [Enable msgpack support (auto/yes/no)]),
+	    [],
+	    [with_msgpack=auto])
+
+
+if test "$with_msgpack" != no; then
+  PKG_CHECK_MODULES(MSGPACK, msgpack, [
+    AC_DEFINE([HAVE_MSGPACK], [1], [Define to 1 if msgpack is available.])
+  ], [
+      if test "$with_msgpack" = yes; then
+	AC_MSG_ERROR([--with-msgpack was given, but msgpack is missing or unusable.])
+      fi
+  ])
+fi
+
 # target-specific stuff:

 # Canonicalize the secondary target names.
--- a/binutils/readelf.c
+++ b/binutils/readelf.c
@@ -46,6 +46,10 @@
 #include <zlib.h>
 #include <wchar.h>

+#if defined HAVE_MSGPACK
+#include <msgpack.h>
+#endif
+
 #if __GNUC__ >= 2
 /* Define BFD64 here, even if our default architecture is 32 bit ELF
   as this will allow us to read in and parse 64bit and 32bit ELF files.
@@ -92,6 +96,7 @@

 #include "elf/aarch64.h"
 #include "elf/alpha.h"
+#include "elf/amdgpu.h"
 #include "elf/arc.h"
 #include "elf/arm.h"
 #include "elf/avr.h"
@@ -1925,6 +1930,9 @@ dump_relocations (Filedata *          filedata,
 	  rtype = elf_loongarch_reloc_type (type);
 	  break;

+	case EM_AMDGPU:
+	  rtype = elf_amdgpu_reloc_type (type);
+	  break;
 	}

      if (rtype == NULL)
@@ -3565,6 +3573,153 @@ decode_NDS32_machine_flags (unsigned e_flags, char buf[], size_t size)
    r += snprintf (buf + r, size -r, ", L2C");
 }

+static void
+decode_AMDGPU_machine_flags (Filedata *filedata, unsigned int e_flags,
+			     char *buf)
+{
+  unsigned char *e_ident = filedata->file_header.e_ident;
+  unsigned char osabi = e_ident[EI_OSABI];
+  unsigned char abiversion = e_ident[EI_ABIVERSION];
+  unsigned int mach;
+
+  /* HSA OS ABI v2 used a different encoding, but we don't need to support it,
+     it has been deprecated for a while.
+
+     The PAL, MESA3D and NONE OS ABIs are not properly versioned, at the time
+     of writing, they use the same flags as HSA v3, so the code below uses that
+     assumption.  */
+  if (osabi == ELFOSABI_AMDGPU_HSA && abiversion < ELFABIVERSION_AMDGPU_HSA_V3)
+    return;
+
+  mach = e_flags & EF_AMDGPU_MACH;
+  switch (mach)
+    {
+#define AMDGPU_CASE(code, string) \
+  case code: strcat (buf, ", " string); break;
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX600, "gfx600")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX601, "gfx601")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX700, "gfx700")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX701, "gfx701")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX702, "gfx702")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX703, "gfx703")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX704, "gfx704")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX801, "gfx801")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX802, "gfx802")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX803, "gfx803")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX810, "gfx810")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX900, "gfx900")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX902, "gfx902")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX904, "gfx904")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX906, "gfx906")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX908, "gfx908")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX909, "gfx909")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX90C, "gfx90c")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX1010, "gfx1010")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX1011, "gfx1011")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX1012, "gfx1012")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX1030, "gfx1030")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX1031, "gfx1031")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX1032, "gfx1032")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX1033, "gfx1033")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX602, "gfx602")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX705, "gfx705")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX805, "gfx805")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX1035, "gfx1035")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX1034, "gfx1034")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX90A, "gfx90a")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX940, "gfx940")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX1013, "gfx1013")
+    AMDGPU_CASE (EF_AMDGPU_MACH_AMDGCN_GFX1036, "gfx1036")
+    default:
+      sprintf (buf, _(", <unknown AMDGPU GPU type: %#x>"), mach);
+      break;
+#undef AMDGPU_CASE
+    }
+
+  buf += strlen (buf);
+  e_flags &= ~EF_AMDGPU_MACH;
+
+  if ((osabi == ELFOSABI_AMDGPU_HSA
+       && abiversion == ELFABIVERSION_AMDGPU_HSA_V3)
+      || osabi != ELFOSABI_AMDGPU_HSA)
+    {
+      /* For HSA v3 and other OS ABIs.  */
+      if (e_flags & EF_AMDGPU_FEATURE_XNACK_V3)
+	{
+	  strcat (buf, ", xnack on");
+	  buf += strlen (buf);
+	  e_flags &= ~EF_AMDGPU_FEATURE_XNACK_V3;
+	}
+
+      if (e_flags & EF_AMDGPU_FEATURE_SRAMECC_V3)
+	{
+	  strcat (buf, ", sramecc on");
+	  buf += strlen (buf);
+	  e_flags &= ~EF_AMDGPU_FEATURE_SRAMECC_V3;
+	}
+    }
+  else
+    {
+      /* For HSA v4+.  */
+      int xnack, sramecc;
+
+      xnack = e_flags & EF_AMDGPU_FEATURE_XNACK_V4;
+      switch (xnack)
+	{
+	case EF_AMDGPU_FEATURE_XNACK_UNSUPPORTED_V4:
+	  break;
+
+	case EF_AMDGPU_FEATURE_XNACK_ANY_V4:
+	  strcat (buf, ", xnack any");
+	  break;
+
+	case EF_AMDGPU_FEATURE_XNACK_OFF_V4:
+	  strcat (buf, ", xnack off");
+	  break;
+
+	case EF_AMDGPU_FEATURE_XNACK_ON_V4:
+	  strcat (buf, ", xnack on");
+	  break;
+
+	default:
+	  sprintf (buf, _(", <unknown xnack value: %#x>"), xnack);
+	  break;
+	}
+
+      buf += strlen (buf);
+      e_flags &= ~EF_AMDGPU_FEATURE_XNACK_V4;
+
+      sramecc = e_flags & EF_AMDGPU_FEATURE_SRAMECC_V4;
+      switch (sramecc)
+	{
+	case EF_AMDGPU_FEATURE_SRAMECC_UNSUPPORTED_V4:
+	  break;
+
+	case EF_AMDGPU_FEATURE_SRAMECC_ANY_V4:
+	  strcat (buf, ", sramecc any");
+	  break;
+
+	case EF_AMDGPU_FEATURE_SRAMECC_OFF_V4:
+	  strcat (buf, ", sramecc off");
+	  break;
+
+	case EF_AMDGPU_FEATURE_SRAMECC_ON_V4:
+	  strcat (buf, ", sramecc on");
+	  break;
+
+	default:
+	  sprintf (buf, _(", <unknown sramecc value: %#x>"), sramecc);
+	  break;
+	}
+
+      buf += strlen (buf);
+      e_flags &= ~EF_AMDGPU_FEATURE_SRAMECC_V4;
+    }
+
+  if (e_flags != 0)
+    sprintf (buf, _(", unknown flags bits: %#x"), e_flags);
+}
+
 static char *
 get_machine_flags (Filedata * filedata, unsigned e_flags, unsigned e_machine)
 {
@@ -3717,6 +3872,10 @@ get_machine_flags (Filedata * filedata, unsigned e_flags, unsigned e_machine)
 	    }
 	  break;

+	case EM_AMDGPU:
+	  decode_AMDGPU_machine_flags (filedata, e_flags, buf);
+	  break;
+
 	case EM_CYGNUS_MEP:
 	  switch (e_flags & EF_MEP_CPU_MASK)
 	    {
@@ -4284,6 +4443,17 @@ get_osabi_name (Filedata * filedata, unsigned int osabi)
      if (osabi >= 64)
 	switch (filedata->file_header.e_machine)
 	  {
+	  case EM_AMDGPU:
+	    switch (osabi)
+	      {
+	      case ELFOSABI_AMDGPU_HSA:    return "AMD HSA";
+	      case ELFOSABI_AMDGPU_PAL:    return "AMD PAL";
+	      case ELFOSABI_AMDGPU_MESA3D: return "AMD Mesa3D";
+	      default:
+		break;
+	      }
+	    break;
+
 	  case EM_ARM:
 	    switch (osabi)
 	      {
@@ -19561,6 +19731,22 @@ decode_x86_compat_2_isa (unsigned int bitmask)
    }
 }

+static const char *
+get_amdgpu_elf_note_type (unsigned int e_type)
+{
+  switch (e_type)
+    {
+    case NT_AMDGPU_METADATA:
+      return _("NT_AMDGPU_METADATA (code object metadata)");
+    default:
+      {
+	static char buf[64];
+	snprintf (buf, sizeof (buf), _("Unknown note type: (0x%08x)"), e_type);
+	return buf;
+      }
+    }
+}
+
 static void
 decode_x86_isa (unsigned int bitmask)
 {
@@ -21128,6 +21314,177 @@ print_gnu_build_attribute_name (Elf_Internal_Note * pnote)
  return true;
 }

+/* Print the contents of PNOTE as hex.  */
+
+static void
+print_note_contents_hex (Elf_Internal_Note *pnote)
+{
+  if (pnote->descsz)
+    {
+      unsigned long i;
+
+      printf (_("   description data: "));
+      for (i = 0; i < pnote->descsz; i++)
+	printf ("%02x ", pnote->descdata[i] & 0xff);
+      if (!do_wide)
+	printf ("\n");
+    }
+
+  if (do_wide)
+    printf ("\n");
+}
+
+#if defined HAVE_MSGPACK
+
+static void
+print_indents (int n)
+{
+  printf ("    ");
+
+  for (int i = 0; i < n; i++)
+    printf ("  ");
+}
+
+/* Print OBJ in human-readable form.  */
+
+static void
+dump_msgpack_obj (const msgpack_object *obj, int indent)
+{
+  switch (obj->type)
+    {
+    case MSGPACK_OBJECT_NIL:
+      printf ("(nil)");
+      break;
+
+    case MSGPACK_OBJECT_BOOLEAN:
+      printf ("%s", obj->via.boolean ? "true" : "false");
+      break;
+
+    case MSGPACK_OBJECT_POSITIVE_INTEGER:
+      printf ("%" PRIu64, obj->via.u64);
+      break;
+
+    case MSGPACK_OBJECT_NEGATIVE_INTEGER:
+      printf ("%" PRIi64, obj->via.i64);
+      break;
+
+    case MSGPACK_OBJECT_FLOAT32:
+    case MSGPACK_OBJECT_FLOAT64:
+      printf ("%f", obj->via.f64);
+      break;
+
+    case MSGPACK_OBJECT_STR:
+      printf ("\"%.*s\"", obj->via.str.size, obj->via.str.ptr);
+      break;
+
+    case MSGPACK_OBJECT_ARRAY:
+      {
+	const msgpack_object_array *array = &obj->via.array;
+
+	printf ("[\n");
+	++indent;
+
+	for (uint32_t i = 0; i < array->size; ++i)
+	  {
+	    const msgpack_object *item = &array->ptr[i];
+
+	    print_indents (indent);
+	    dump_msgpack_obj (item, indent);
+	    printf (",\n");
+	  }
+
+	--indent;
+	print_indents (indent);
+	printf ("]");
+	break;
+      }
+      break;
+
+    case MSGPACK_OBJECT_MAP:
+      {
+	const msgpack_object_map *map = &obj->via.map;
+
+	printf ("{\n");
+	++indent;
+
+	for (uint32_t i = 0; i < map->size; ++i)
+	  {
+	    const msgpack_object_kv *kv = &map->ptr[i];
+	    const msgpack_object *key = &kv->key;
+	    const msgpack_object *val = &kv->val;
+
+	    print_indents (indent);
+	    dump_msgpack_obj (key, indent);
+	    printf (": ");
+	    dump_msgpack_obj (val, indent);
+
+	    printf (",\n");
+	  }
+
+	--indent;
+	print_indents (indent);
+	printf ("}");
+
+	break;
+      }
+
+    case MSGPACK_OBJECT_BIN:
+      printf ("(bin)");
+      break;
+
+    case MSGPACK_OBJECT_EXT:
+      printf ("(ext)");
+      break;
+    }
+}
+
+static void
+dump_msgpack (const msgpack_unpacked *msg)
+{
+  print_indents (0);
+  dump_msgpack_obj (&msg->data, 0);
+  printf ("\n");
+}
+
+#endif /* defined HAVE_MSGPACK */
+
+static bool
+print_amdgpu_note (Elf_Internal_Note *pnote)
+{
+#if defined HAVE_MSGPACK
+  /* If msgpack is available, decode and dump the note's content.  */
+  bool ret;
+  msgpack_unpacked msg;
+  msgpack_unpack_return msgpack_ret;
+
+  assert (pnote->type == NT_AMDGPU_METADATA);
+
+  msgpack_unpacked_init (&msg);
+  msgpack_ret = msgpack_unpack_next (&msg, pnote->descdata, pnote->descsz,
+				     NULL);
+
+  switch (msgpack_ret)
+    {
+    case MSGPACK_UNPACK_SUCCESS:
+      dump_msgpack (&msg);
+      ret = true;
+      break;
+
+    default:
+      error (_("failed to unpack msgpack contents in NT_AMDGPU_METADATA note"));
+      ret = false;
+      break;
+    }
+
+  msgpack_unpacked_destroy (&msg);
+  return ret;
+#else
+  /* msgpack is not available, dump contents as hex.  */
+  print_note_contents_hex (pnote);
+  return true;
+#endif
+}
+
 /* Note that by the ELF standard, the name field is already null byte
   terminated, and namesz includes the terminating null byte.
   I.E. the value of namesz for the name "FSF" is 4.
@@ -21150,6 +21507,10 @@ process_note (Elf_Internal_Note *  pnote,
    /* GNU-specific object file notes.  */
    nt = get_gnu_elf_note_type (pnote->type);

+  else if (startswith (pnote->namedata, "AMDGPU"))
+    /* AMDGPU-specific object file notes.  */
+    nt = get_amdgpu_elf_note_type (pnote->type);
+
  else if (startswith (pnote->namedata, "FreeBSD"))
    /* FreeBSD-specific core file notes.  */
    nt = get_freebsd_elfcore_note_type (filedata, pnote->type);
@@ -21221,21 +21582,11 @@ process_note (Elf_Internal_Note *  pnote,
 	   && (pnote->type == NT_GNU_BUILD_ATTRIBUTE_OPEN
 	       || pnote->type == NT_GNU_BUILD_ATTRIBUTE_FUNC))
    return print_gnu_build_attribute_description (pnote, filedata);
+  else if (startswith (pnote->namedata, "AMDGPU")
+	   && pnote->type == NT_AMDGPU_METADATA)
+    return print_amdgpu_note (pnote);

-  if (pnote->descsz)
-    {
-      unsigned long i;
-
-      printf (_("   description data: "));
-      for (i = 0; i < pnote->descsz; i++)
-	printf ("%02x ", pnote->descdata[i] & 0xff);
-      if (!do_wide)
-	printf ("\n");
-    }
-
-  if (do_wide)
-    printf ("\n");
-
+  print_note_contents_hex (pnote);
  return true;
 }

--- a/include/elf/amdgpu.h
+++ b/include/elf/amdgpu.h
@@ -0,0 +1,115 @@
+/* AMDGPU ELF support for BFD.
+
+   Copyright (C) 2019-2021 Free Software Foundation, Inc.
+
+   This file is part of BFD, the Binary File Descriptor library.
+
+   This program 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 of the License, or
+   (at your option) any later version.
+
+   This program 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 this program.  If not, see <http://www.gnu.org/licenses/>.  */
+
+#ifndef _ELF_AMDGPU_H
+#define _ELF_AMDGPU_H
+
+#include "elf/reloc-macros.h"
+
+/* e_ident[EI_ABIVERSION] values, when e_ident[EI_OSABI] is
+   ELFOSABI_AMDGPU_HSA.  */
+
+#define ELFABIVERSION_AMDGPU_HSA_V2 0
+#define ELFABIVERSION_AMDGPU_HSA_V3 1
+#define ELFABIVERSION_AMDGPU_HSA_V4 2
+#define ELFABIVERSION_AMDGPU_HSA_V5 3
+
+/* Processor selection mask for EF_AMDGPU_MACH_* values.  */
+
+#define EF_AMDGPU_MACH 0x0ff
+#define EF_AMDGPU_MACH_AMDGCN_MIN 0x020
+
+#define EF_AMDGPU_MACH_AMDGCN_GFX600  0x020
+#define EF_AMDGPU_MACH_AMDGCN_GFX601  0x021
+#define EF_AMDGPU_MACH_AMDGCN_GFX700  0x022
+#define EF_AMDGPU_MACH_AMDGCN_GFX701  0x023
+#define EF_AMDGPU_MACH_AMDGCN_GFX702  0x024
+#define EF_AMDGPU_MACH_AMDGCN_GFX703  0x025
+#define EF_AMDGPU_MACH_AMDGCN_GFX704  0x026
+#define EF_AMDGPU_MACH_AMDGCN_GFX801  0x028
+#define EF_AMDGPU_MACH_AMDGCN_GFX802  0x029
+#define EF_AMDGPU_MACH_AMDGCN_GFX803  0x02a
+#define EF_AMDGPU_MACH_AMDGCN_GFX810  0x02b
+#define EF_AMDGPU_MACH_AMDGCN_GFX900  0x02c
+#define EF_AMDGPU_MACH_AMDGCN_GFX902  0x02d
+#define EF_AMDGPU_MACH_AMDGCN_GFX904  0x02e
+#define EF_AMDGPU_MACH_AMDGCN_GFX906  0x02f
+#define EF_AMDGPU_MACH_AMDGCN_GFX908  0x030
+#define EF_AMDGPU_MACH_AMDGCN_GFX909  0x031
+#define EF_AMDGPU_MACH_AMDGCN_GFX90C  0x032
+#define EF_AMDGPU_MACH_AMDGCN_GFX1010 0x033
+#define EF_AMDGPU_MACH_AMDGCN_GFX1011 0x034
+#define EF_AMDGPU_MACH_AMDGCN_GFX1012 0x035
+#define EF_AMDGPU_MACH_AMDGCN_GFX1030 0x036
+#define EF_AMDGPU_MACH_AMDGCN_GFX1031 0x037
+#define EF_AMDGPU_MACH_AMDGCN_GFX1032 0x038
+#define EF_AMDGPU_MACH_AMDGCN_GFX1033 0x039
+#define EF_AMDGPU_MACH_AMDGCN_GFX602  0x03a
+#define EF_AMDGPU_MACH_AMDGCN_GFX705  0x03b
+#define EF_AMDGPU_MACH_AMDGCN_GFX805  0x03c
+#define EF_AMDGPU_MACH_AMDGCN_GFX1035 0x03d
+#define EF_AMDGPU_MACH_AMDGCN_GFX1034 0x03e
+#define EF_AMDGPU_MACH_AMDGCN_GFX90A  0x03f
+#define EF_AMDGPU_MACH_AMDGCN_GFX940  0x040
+#define EF_AMDGPU_MACH_AMDGCN_GFX1013 0x042
+#define EF_AMDGPU_MACH_AMDGCN_GFX1036 0x045
+
+/* Code object v3 machine flags.  */
+
+#define EF_AMDGPU_FEATURE_XNACK_V3   0x100
+#define EF_AMDGPU_FEATURE_SRAMECC_V3 0x200
+
+/* Code object v4 (and later) machine flags.  */
+
+#define EF_AMDGPU_FEATURE_XNACK_V4             0x300
+#define EF_AMDGPU_FEATURE_XNACK_UNSUPPORTED_V4 0x000
+#define EF_AMDGPU_FEATURE_XNACK_ANY_V4         0x100
+#define EF_AMDGPU_FEATURE_XNACK_OFF_V4         0x200
+#define EF_AMDGPU_FEATURE_XNACK_ON_V4          0x300
+
+#define EF_AMDGPU_FEATURE_SRAMECC_V4             0xc00
+#define EF_AMDGPU_FEATURE_SRAMECC_UNSUPPORTED_V4 0x000
+#define EF_AMDGPU_FEATURE_SRAMECC_ANY_V4         0x400
+#define EF_AMDGPU_FEATURE_SRAMECC_OFF_V4         0x800
+#define EF_AMDGPU_FEATURE_SRAMECC_ON_V4          0xc00
+
+/* Notes. */
+
+#define NT_AMDGPU_METADATA                32
+
+/* Relocations.  */
+
+START_RELOC_NUMBERS (elf_amdgpu_reloc_type)
+ RELOC_NUMBER (R_AMDGPU_NONE,           0)
+ RELOC_NUMBER (R_AMDGPU_ABS32_LO,       1)
+ RELOC_NUMBER (R_AMDGPU_ABS32_HI,       2)
+ RELOC_NUMBER (R_AMDGPU_ABS64,          3)
+ RELOC_NUMBER (R_AMDGPU_REL32,          4)
+ RELOC_NUMBER (R_AMDGPU_REL64,          5)
+ RELOC_NUMBER (R_AMDGPU_ABS32,          6)
+ RELOC_NUMBER (R_AMDGPU_GOTPCREL,       7)
+ RELOC_NUMBER (R_AMDGPU_GOTPCREL32_LO,  8)
+ RELOC_NUMBER (R_AMDGPU_GOTPCREL32_HI,  9)
+ RELOC_NUMBER (R_AMDGPU_REL32_LO,      10)
+ RELOC_NUMBER (R_AMDGPU_REL32_HI,      11)
+ RELOC_NUMBER (R_AMDGPU_RELATIVE64,    13)
+ RELOC_NUMBER (R_AMDGPU_REL16,         16)
+END_RELOC_NUMBERS (R_AMDGPU_max)
+
+#endif /* _ELF_AMDGPU_H */
--- a/include/elf/common.h
+++ b/include/elf/common.h
@@ -77,8 +77,11 @@
 #define ELFOSABI_OPENVOS     18 /* Stratus Technologies OpenVOS */

 #define ELFOSABI_C6000_ELFABI 64 /* Bare-metal TMS320C6000 */
+#define ELFOSABI_AMDGPU_HSA  64 /* AMD HSA Runtime */
 #define ELFOSABI_C6000_LINUX 65 /* Linux TMS320C6000 */
+#define ELFOSABI_AMDGPU_PAL  65 /* AMD PAL Runtime */
 #define ELFOSABI_ARM_FDPIC   65 /* ARM FDPIC */
+#define ELFOSABI_AMDGPU_MESA3D 66 /* AMD Mesa3D Runtime */
 #define ELFOSABI_ARM	     97	/* ARM */
 #define ELFOSABI_STANDALONE 255	/* Standalone (embedded) application */

--- a/opcodes/configure
+++ b/opcodes/configure
@@ -12491,6 +12491,7 @@ if test x${all_targets} = xfalse ; then
 	case "$arch" in
 	bfd_aarch64_arch)	ta="$ta aarch64-asm.lo aarch64-dis.lo aarch64-opc.lo aarch64-asm-2.lo aarch64-dis-2.lo aarch64-opc-2.lo" ;;
 	bfd_alpha_arch)		ta="$ta alpha-dis.lo alpha-opc.lo" ;;
+	bfd_amdgcn_arch)	;;
 	bfd_arc_arch)		ta="$ta arc-dis.lo arc-opc.lo arc-ext.lo" ;;
 	bfd_arm_arch)		ta="$ta arm-dis.lo" ;;
 	bfd_avr_arch)		ta="$ta avr-dis.lo" ;;
--- a/opcodes/configure.ac
+++ b/opcodes/configure.ac
@@ -263,6 +263,7 @@ if test x${all_targets} = xfalse ; then
 	case "$arch" in
 	bfd_aarch64_arch)	ta="$ta aarch64-asm.lo aarch64-dis.lo aarch64-opc.lo aarch64-asm-2.lo aarch64-dis-2.lo aarch64-opc-2.lo" ;;
 	bfd_alpha_arch)		ta="$ta alpha-dis.lo alpha-opc.lo" ;;
+	bfd_amdgcn_arch)	;;
 	bfd_arc_arch)		ta="$ta arc-dis.lo arc-opc.lo arc-ext.lo" ;;
 	bfd_arm_arch)		ta="$ta arm-dis.lo" ;;
 	bfd_avr_arch)		ta="$ta avr-dis.lo" ;;
Author	SHA1	Message	Date
Simon Marchi	0618b08844	binutils/readelf: handle AMDGPU relocation types Make readelf recognize AMDGPU relocation types, as documented here: https://llvm.org/docs/AMDGPUUsage.html#amdgpu-relocation-records The user-visible change looks like: -000000000004 000400000001 unrecognized: 1 0000000000000000 SCRATCH_RSRC_DWORD0 -00000000000c 000500000001 unrecognized: 1 0000000000000000 SCRATCH_RSRC_DWORD1 -000000000014 000600000007 unrecognized: 7 0000000000000000 global_var0 -00000000001c 000700000008 unrecognized: 8 0000000000000000 global_var1 -000000000024 000800000009 unrecognized: 9 0000000000000000 global_var2 -00000000002c 00090000000a unrecognized: a 0000000000000000 global_var3 -000000000034 000a0000000b unrecognized: b 0000000000000000 global_var4 +000000000004 000400000001 R_AMDGPU_ABS32_LO 0000000000000000 SCRATCH_RSRC_DWORD0 +00000000000c 000500000001 R_AMDGPU_ABS32_LO 0000000000000000 SCRATCH_RSRC_DWORD1 +000000000014 000600000007 R_AMDGPU_GOTPCREL 0000000000000000 global_var0 +00000000001c 000700000008 R_AMDGPU_GOTPCREL 0000000000000000 global_var1 +000000000024 000800000009 R_AMDGPU_GOTPCREL 0000000000000000 global_var2 +00000000002c 00090000000a R_AMDGPU_REL32_LO 0000000000000000 global_var3 +000000000034 000a0000000b R_AMDGPU_REL32_HI 0000000000000000 global_var4 binutils/ChangeLog: * readelf.c (dump_relocations): Handle EM_AMDGPU. include/ChangeLog: * elf/amdgpu.h: Add relocation values. Change-Id: I2ed4589f4cd37ea11ad2e0cb38d4b682271e1334	2022-03-15 15:09:55 -04:00
Simon Marchi	39306f8455	binutils/readelf: build against msgpack, dump NT_AMDGPU_METADATA note contents [This patch might be a bit more controversial, given its adds an optional dependency on a third party library. If this is really a problems, we can drop it from the series. The consequence being that there will be no human readable output for the AMDGPU ELF notes.] The AMDGPU HSA OS ABI (code object v3 and above) defines the NT_AMDGPU_METADATA ELF note [1]. The content is a msgpack object describing, among other things, the kernels present in the code object and how to call them. I think it would be useful for readelf to be able to display the content of those notes. msgpack is a structured format, a bit like JSON, except not text-based. It is therefore possible to dump the contents in human-readable form without knowledge of the specific layout of the note. Add configury to binutils to optionally check for the msgpack C library [2]. Add There is a new --with{,out}-msgpack configure flag, and the actual library lookup is done using pkg-config. If msgpack support is enabled, dumping a NT_AMDGPU_METADATA note looks like: $ readelf --notes amdgpu-code-object Displaying notes found in: .note Owner Data size Description AMDGPU 0x0000040d NT_AMDGPU_METADATA (code object metadata) { "amdhsa.kernels": [ { ".args": [ { ".address_space": "global", ".name": "out.coerce", ".offset": 0, ".size": 8, ".value_kind": "global_buffer", }, <snip> If msgpack support is disabled, dump the contents as hex, as is done with notes that are not handled in a special way. This allows one to decode the contents manually (maybe using a command-line msgpack decoder) if really needed. [1] https://llvm.org/docs/AMDGPUUsage.html#code-object-metadata [2] https://github.com/msgpack/msgpack-c/tree/c_master binutils/ChangeLog: * Makefile.am (readelf_CFLAGS): New. (readelf_LDADD): Add MSGPACK_LIBS. * Makefile.in: Re-generate. * config.in: Re-generate. * configure: Re-generate. * configure.ac: Add --with-msgpack flag and check for msgpack using pkg-config. * readelf.c: Include msgpack.h if HAVE_MSGPACK. (print_note_contents_hex): New. (print_indents): New. (dump_msgpack_obj): New. (dump_msgpack): New. (print_amdgpu_note): New. (process_note): Handle NT_AMDGPU_METADATA note contents. Use print_note_contents_hex. Change-Id: Ia60a654e620bc32dfdb1bccd845594e2af328b84	2022-03-15 15:09:55 -04:00
Simon Marchi	86b9105a36	binutils/readelf: handle NT_AMDGPU_METADATA note name Handle the NT_AMDGPU_METADATA note, which is described here: https://llvm.org/docs/AMDGPUUsage.html#code-object-v3-note-records As of this patch, just print out the name, not the contents, which is in the msgpack format. binutils/ChangeLog: * readelf.c (get_amdgpu_elf_note_type): New. (process_note): Handle "AMDGPU" notes. include/ChangeLog: * elf/amdgcn.h (NT_AMDGPU_METADATA): New. Change-Id: Id2dba2e2aeaa55ef7464fb35aee9c7d5f96ddb23	2022-03-15 15:09:55 -04:00
Simon Marchi	9452483507	binutils/readelf: decode AMDGPU-specific e_flags Decode and print the AMDGPU-specific fields of e_flags, as documented here: https://llvm.org/docs/AMDGPUUsage.html#header That is: - The specific GPU model - Whether the xnack and sramecc features are enabled The result looks like: - Flags: 0x52f + Flags: 0x52f, gfx906, xnack any, sramecc any The flags for the "HSA" OS ABI are properly versioned and documented on that page. But the NONE, PAL and MESA3D OS ABIs are not well documented nor versioned. Taking a peek at the LLVM source code, we see that they encode their flags the same way as HSA v3. For example, for PAL: `c8b614cd74/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUTargetStreamer.cpp (L601)` So for those other OS ABIs, we read them the same as HSA v3. binutils/ChangeLog: * readelf.c: Include elf/amdgcn.h. (decode_AMDGPU_machine_flags): New. (get_machine_flags): Handle flags for EM_AMDGPU machine type. include/ChangeLog: * elf/amdgcn.h: Add EF_AMDGPU_MACH_AMDGCN_* and EF_AMDGPU_FEATURE_* defines. Change-Id: Ib5b94df7cae0719a22cf4e4fd0629330e9485c12	2022-03-15 15:09:55 -04:00
Simon Marchi	9f8890e0d1	binutils/readelf: handle AMDGPU OS ABIs When the machine is EM_AMDGPU, handle the various OS ABIs described here: https://llvm.org/docs/AMDGPUUsage.html#header For a binary with the HSA OS ABI, the change looks like: - OS/ABI: <unknown: 40> + OS/ABI: AMD HSA binutils/ChangeLog: * readelf.c (get_osabi_name): Handle EM_AMDGPU OS ABIs. include/ChangeLog: * elf/common.h (ELFOSABI_AMDGPU_PAL, ELFOSABI_AMDGPU_MESA3D): New. Change-Id: I383590c390f7dc2fe0f902f50038735626d71863	2022-03-15 15:09:55 -04:00
Simon Marchi	780a00cbad	opcodes: handle bfd_amdgcn_arch in configure script There isn't an actual opcodes implementation for the AMDGCN arch (yet), this is just the bare minimum to get $ ./configure --target=amdgcn-hsa-amdhsa --disable-gas $ make all-binutils working later in this series. opcodes/ChangeLog: * configure.ac: Handle bfd_amdgcn_arch. * configure: Re-generate. Change-Id: Ib7d7c5533a803ed8b2a293e9275f667ed781ce79	2022-03-15 15:09:55 -04:00
Simon Marchi	daab8d1df7	bfd: add AMDGCN architecture Add support for the AMDGCN architecture to BFD. This is the bare minimum to get The specific AMDGCN models added here are a bit arbitrary, based on what we intend to initially support in GDB. This list will need to be updated in the future anyway. The complete up-to-date list of existing AMDGPU models can be found here: https://llvm.org/docs/AMDGPUUsage.html#processors The ELF format for this architecture is documented here: https://llvm.org/docs/AMDGPUUsage.html#elf-code-object The flags for the "HSA" OS ABI are properly versioned and documented on that page. But the NONE, PAL and MESA3D OS ABIs are not well documented nor versioned. Taking a peek at the LLVM source code, we see that they encode their flags the same way as HSA v3. For example, for PAL: `c8b614cd74/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUTargetStreamer.cpp (L601)` So at least, we know that all AMDGPU objects (of which AMDGCN objects are a subset of) at the time of writing encode the specific GPU model in the EF_AMDGPU_MACH field of e_flags. bfd/ChangeLog: * Makefile.am (ALL_MACHINES, ALL_MACHINES_CFILES): Add cpu-amdgcn.c. (BFD64_BACKENDS): Add elf64-amdgcn.lo. (BFD64_BACKENDS_CFILES): Add elf64-amdgcn.c. * Makefile.in: Re-generate. * cpu-amdgcn.c: New. * elf64-amdgcn.c: New. * archures.c (bfd_architecture): Add bfd_arch_amdgcn and related mach defines. (bfd_amdgcn_arch): New. (bfd_archures_list): Add bfd_amdgcn_arch. * bfd-in2.h: Re-generate. * config.bfd: Handle amdgcn* target. * configure.ac: Handle amdgcn_elf64_le_vec. * configure: Re-generate. * elf-bfd.h (elf_target_id): Add AMDGCN_ELF_DATA. * targets.c (amdgcn_elf64_le_vec): New. (_bfd_target_vector): Add amdgcn_elf64_le_vec. include/ChangeLog: * elf/amdgpu.h: New. * elf/common.h (ELFOSABI_AMDGPU_HSA): Add. Change-Id: I969f7b14960797e88891c308749a6e341eece5b2	2022-03-15 15:09:55 -04:00