diff mbox series

[5/8,og9] AMD GCN offloading support

Message ID dfe3cbfb88da3984ed8c791e941059ad514af0b2.1564776080.git.julian@codesourcery.com
State New
Headers show
Series AMD GCN offloading support | expand

Commit Message

Julian Brown Aug. 2, 2019, 8:12 p.m. UTC
This patch provides the basis of support for offloading (OpenMP/OpenACC)
to AMD GCN, in particular a version of "mkoffload" for that target,
and libgomp support.

Some parts of the latter can be shared with the nvptx backend, and have
been moved to an "accel" subdirectory to indicate their newly-generic
nature.

The patch also includes support for several routines useful for debugging
-- gomp_print_string, gomp_print_integer and gomp_print_double -- which
can be used from offload code without pulling in the whole of e.g. printf.

Several places in libgomp and several test cases have been fixed to
correct target-specific semantics that do not apply to GCN.

Julian

ChangeLog

2019-07-31  Julian Brown  <julian@codesourcery.com>
	    Andrew Stubbs  <ams@codesourcery.com>

	gcc/
	* config.gcc (amdgcn-*-*): Add default option for gfx906.
	* config/gcn/mkoffload.c: New.
	* config/gcn/offload.h: New.

	libgcc/
	* Makefile.in: Allow disabling of emutls.
	* config/gcn/gomp_print.c: New.
	* config/gcn/reduction.c: New.
	* config/gcn/t-amdgcn (LIB2ADD): Add gomp_print.c and reduction.c.
	Disable emutls.c.
	* config/gcn/t-gcn-hsa: New.

	libgomp/
	* Makefile.am (libgomp_la_SOURCES): Add gomp_print.c.
	* Makefile.in: Regenerate.
	* affinity-fmt.c: Rename calls to gomp_write_string from
	gomp_print_string.
	* config.h.in (PLUGIN_GCN): Add #undef.
	* config/nvptx/libgomp-plugin.c: Rename to...
	* config/accel/libgomp-plugin.c: ...this.
	* config/nvptx/lock.c: Rename to...
	* config/accel/lock.c: ...this.
	* config/nvptx/mutex.c: Rename to...
	* config/accel/mutex.c: ...this.
	* config/nvptx/mutex.h: Rename to...
	* config/accel/mutex.h: ...this.
	* config/nvptx/oacc-async.c: Rename to...
	* config/accel/oacc-async.c: ...this.
	* config/nvptx/oacc-cuda.c: Rename to...
	* config/accel/oacc-cuda.c: ...this.
	* config/nvptx/oacc-host.c: Rename to...
	* config/accel/oacc-host.c: ...this.
	* config/nvptx/oacc-init.c: Rename to...
	* config/accel/oacc-init.c: ...this.
	* config/nvptx/oacc-mem.c: Rename to...
	* config/accel/oacc-mem.c: ...this.
	* config/nvptx/oacc-plugin.c: Rename to...
	* config/accel/oacc-plugin.c: ...this.
	* config/nvptx/omp-lock.h: Rename to...
	* config/accel/omp-lock.h: ...this.
	* config/nvptx/openacc.f90: Rename to...
	* config/accel/openacc.f90: ...this.  Add acc_device_hsa and
	acc_device_gcn.
	* config/nvptx/pool.h: Rename to...
	* config/accel/pool.h: ...this.
	* config/nvptx/proc.c: Rename to...
	* config/accel/proc.c: ...this.  Add omp_get_num_procs alias.
	* config/nvptx/ptrlock.c: Rename to...
	* config/accel/ptrlock.c: ...this.
	* config/nvptx/ptrlock.h: Rename to...
	* config/accel/ptrlock.h: ...this.
	* config/nvptx/sem.c: Rename to...
	* config/accel/sem.c: ...this.
	* config/nvptx/sem.h: Rename to...
	* config/accel/sem.h: ...this.
	* config/nvptx/thread-stacksize.h: Rename to...
	* config/accel/thread-stacksize.h: ...this.
	* config/gcn/affinity-fmt.c: New.
	* config/gcn/bar.c: New.
	* config/gcn/bar.h: New.
	* config/gcn/doacross.h: New.
	* config/gcn/gomp_print.c: New.
	* config/gcn/icv-device.c: New.
	* config/gcn/simple-bar.h: New.
	* config/gcn/target.c: New.
	* config/gcn/task.c: New.
	* config/gcn/team.c: New.
	* config/gcn/time.c: New.
	* config/linux/gomp_print.c: New.
	* configure.ac (amdgcn*-*-*): Disable pthreads.
	* configure: Regenerated.
	* configure.tgt (nvptx*-*-*): Add 'accel' config_path.
	(amdgcn*-*-*): Set config_path.
	* fortran.c (omp_display_affinity_): Rename calls to gomp_write_string
	from gomp_print_string.
	* libgomp-plugin.h (enum offload_target_type): Add
	OFFLOAD_TARGET_TYPE_GCN.
	(GOMP_OFFLOAD_openacc_async_construct): Change parameter type to int.
	* libgomp.h (gcn_thrs, set_gcn_thrs, gomp_thread): Add for __AMDGCN__.
	(gomp_print_string): Rename to...
	(gomp_write_string): ...this.
	* libgomp.map (GOMP_4.5): Add gomp_rpint_string, gomp_print_integer,
	gomp_print_double.
	* oacc-async.c (lookup_goacc_asyncqueue): Pass target_id to async queue
	construct function.
	* oacc-host.c (host_openacc_async_construct): Add dummy device
	parameter.
	* oacc-init.c (name_of_acc_device_t): Add acc_device_gcn.
	* oacc-int.h (goacc_thread): Add dummy implementation for __AMDGCN__.
	* oacc-parallel.c (GOACC_enter_exit_data): Support acc_async_noval and
	zero-length array sections.
	* omp.h.in (gomp_print_string, gomp_print_integer, gomp_print_double):
	Add prototypes.
	* omp_lib.f90.in (gomp_print_string, gomp_print_integer,
	gomp_print_double): Add interfaces.
	* openacc.f90 (openacc_kinds): Add acc_device_gcn.  Bump
	acc_device_current code.
	* openacc.h (acc_device_t): Add acc_device_gcn, bump acc_device_current
	code.
	* openacc_lib.h (acc_device_hsa, acc_device_gcn): Add.
	* plugin/Makefrag.am (PLUGIN_GCN): Support building GCN plugin.
	* plugin/configfrag.am (PLUGIN_GCN, PLUGIN_GCN_CPPFLAGS,
	PLUGIN_GCN_LDFLAGS, PLUGIN_GCN_LIBS): Add.  Add suport for GCN plugin.
	* plugin/plugin-gcn.c: New.
	* target.c (stdio.h): Include unconditionally.
	(gomp_copy_host2dev): Add function comment.
	(copy_host2dev_immediate): New function.
	(gomp_map_pointer, gomp_map_vars_internal): Use
	copy_host2dev_immediate where appropriate.
	(offload_target_to_plugin_name): Support gcn.
	* team.c (gomp_free_pool_helper): Support gcn.
	* testsuite/Makefile.in: Regenerated.
	* testsuite/lib/libgomp.exp
	(check_effective_target_openacc_amdgcn_accel_present): New.
	(check_effective_target_openacc_amdgcn_accel_selected): New.
	* testsuite/libgomp.c/c.exp (generate_tests, test_lists,
	generated_tests): New.
	(tests): Add generated tests.
	* testsuite/libgomp.c/for-1.h: New.
	* testsuite/libgomp.c/for-2.h: New.
	* testsuite/libgomp.c/for-3.h: New.
	* testsuite/libgomp.c/for-3.list: New.
	* testsuite/libgomp.c/for-5.c: New.
	* testsuite/libgomp.c/for-5.list: New.
	* testsuite/libgomp.c/for-6.c: New.
	* testsuite/libgomp.c/for-6.list: New.
	* testsuite/libgomp.c/target-print-1.c: New.
	* testsuite/libgomp.fortran/target-print-1.f90: New.
	* testsuite/libgomp.oacc-c++/c++.exp (amdgcn*): Add support for AMD GCN.
	* testsuite/libgomp.oacc-c-c++-common/atomic_capture-2.c: Adjust for
	portability.
	* testsuite/libgomp.oacc-c-c++-common/loop-auto-1.c: Skip unsuitable
	test for AMD GCN.
	* testsuite/libgomp.oacc-c-c++-common/loop-gwv-1.c: Adjust for
	portability.
	* testsuite/libgomp.oacc-c-c++-common/loop-v-1.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/loop-w-1.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/loop-wv-1.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/loop-red-gwv-1.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/loop-red-v-1.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/loop-red-v-2.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/loop-red-w-1.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/loop-red-w-2.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/loop-red-wv-1.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/routine-gwv-1.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/routine-v-1.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/routine-w-1.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/routine-wv-1.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/routine-wv-2.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/parallel-dims.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/serial-dims.c: Likewise.
	* testsuite/libgomp.oacc-c-c++-common/private-variables-2.c: New.
	* testsuite/libgomp.oacc-c-c++-common/tile-1.c: Skip for AMD GCN.
	* testsuite/libgomp.oacc-c/c.exp (amdgcn*): Add support for AMD GCN.
	* testsuite/libgomp.oacc-c/offload-target-1.c: Add AMD GCN support.
	* testsuite/libgomp.oacc-c/print-1.c: New.
	* testsuite/libgomp.oacc-fortran/fortran.exp (amdgcn*): Add AMD GCN
	support.
	* testsuite/libgomp.oacc-fortran/atomic_capture-1.f90: Adjust for
	portability.
	* testsuite/libgomp.oacc-fortran/collapse-1.f90: Likewise.
	* testsuite/libgomp.oacc-fortran/collapse-2.f90: Likewise.
	* testsuite/libgomp.oacc-fortran/error_stop-1.f: Support AMD GCN.
	* testsuite/libgomp.oacc-fortran/error_stop-2.f: Support AMD GCN.
	* testsuite/libgomp.oacc-fortran/error_stop-3.f: Support AMD GCN.
	* testsuite/libgomp.oacc-fortran/print-1.f90: New.
---
 gcc/ChangeLog.openacc                         |    7 +
 gcc/config.gcc                                |    2 +-
 gcc/config/gcn/mkoffload.c                    |  702 ++++
 gcc/config/gcn/offload.h                      |   35 +
 include/gomp-constants.h                      |    4 +-
 libgcc/ChangeLog.openacc                      |   10 +
 libgcc/Makefile.in                            |    2 +
 libgcc/config/gcn/gomp_print.c                |  101 +
 libgcc/config/gcn/reduction.c                 |   30 +
 libgcc/config/gcn/t-amdgcn                    |   11 +-
 libgcc/config/gcn/t-gcn-hsa                   |   52 +
 libgomp/ChangeLog.openacc                     |  157 +
 libgomp/Makefile.am                           |    2 +-
 libgomp/Makefile.in                           |   63 +-
 libgomp/affinity-fmt.c                        |   10 +-
 libgomp/config.h.in                           |    3 +
 .../config/{nvptx => accel}/libgomp-plugin.c  |    0
 libgomp/config/{nvptx => accel}/lock.c        |    0
 libgomp/config/{nvptx => accel}/mutex.c       |    0
 libgomp/config/{nvptx => accel}/mutex.h       |    0
 libgomp/config/{nvptx => accel}/oacc-async.c  |    0
 libgomp/config/{nvptx => accel}/oacc-cuda.c   |    0
 libgomp/config/{nvptx => accel}/oacc-host.c   |    0
 libgomp/config/{nvptx => accel}/oacc-init.c   |    0
 libgomp/config/{nvptx => accel}/oacc-mem.c    |    0
 libgomp/config/{nvptx => accel}/oacc-plugin.c |    0
 libgomp/config/{nvptx => accel}/omp-lock.h    |    0
 libgomp/config/{nvptx => accel}/openacc.f90   |    2 +
 libgomp/config/{nvptx => accel}/pool.h        |    0
 libgomp/config/{nvptx => accel}/proc.c        |    1 +
 libgomp/config/{nvptx => accel}/ptrlock.c     |    0
 libgomp/config/{nvptx => accel}/ptrlock.h     |    0
 libgomp/config/{nvptx => accel}/sem.c         |    0
 libgomp/config/{nvptx => accel}/sem.h         |    0
 .../{nvptx => accel}/thread-stacksize.h       |    0
 libgomp/config/gcn/affinity-fmt.c             |   51 +
 libgomp/config/gcn/bar.c                      |  230 ++
 libgomp/config/gcn/bar.h                      |  168 +
 libgomp/config/gcn/doacross.h                 |   58 +
 libgomp/config/gcn/gomp_print.c               |    2 +
 libgomp/config/gcn/icv-device.c               |   72 +
 libgomp/config/gcn/simple-bar.h               |   61 +
 libgomp/config/gcn/target.c                   |   49 +
 libgomp/config/gcn/task.c                     |   39 +
 libgomp/config/gcn/team.c                     |  202 +
 libgomp/config/gcn/time.c                     |   52 +
 libgomp/config/linux/gomp_print.c             |   20 +
 libgomp/configure                             |   61 +-
 libgomp/configure.ac                          |    2 +-
 libgomp/configure.tgt                         |    6 +-
 libgomp/fortran.c                             |    4 +-
 libgomp/libgomp-plugin.h                      |    5 +-
 libgomp/libgomp.h                             |   20 +-
 libgomp/libgomp.map                           |    3 +
 libgomp/oacc-async.c                          |    3 +-
 libgomp/oacc-host.c                           |    2 +-
 libgomp/oacc-init.c                           |    3 +-
 libgomp/oacc-int.h                            |    9 +-
 libgomp/oacc-parallel.c                       |   22 +-
 libgomp/omp.h.in                              |   12 +
 libgomp/omp_lib.f90.in                        |   26 +
 libgomp/openacc.f90                           |    6 +-
 libgomp/openacc.h                             |    3 +-
 libgomp/openacc_lib.h                         |    2 +
 libgomp/plugin/Makefrag.am                    |   14 +
 libgomp/plugin/configfrag.ac                  |   35 +
 libgomp/plugin/plugin-gcn.c                   | 3482 +++++++++++++++++
 libgomp/target.c                              |   59 +-
 libgomp/team.c                                |    3 +
 libgomp/testsuite/Makefile.in                 |    5 +
 libgomp/testsuite/lib/libgomp.exp             |   25 +
 libgomp/testsuite/libgomp.c/c.exp             |   73 +-
 libgomp/testsuite/libgomp.c/for-1.h           |   25 +
 libgomp/testsuite/libgomp.c/for-2.h           |  313 ++
 libgomp/testsuite/libgomp.c/for-3.c           |  123 +
 libgomp/testsuite/libgomp.c/for-3.list        |   24 +
 libgomp/testsuite/libgomp.c/for-5.c           |  161 +
 libgomp/testsuite/libgomp.c/for-5.list        |   35 +
 libgomp/testsuite/libgomp.c/for-6.c           |  135 +
 libgomp/testsuite/libgomp.c/for-6.list        |   24 +
 libgomp/testsuite/libgomp.c/target-print-1.c  |   17 +
 .../libgomp.fortran/target-print-1.f90        |   15 +
 libgomp/testsuite/libgomp.oacc-c++/c++.exp    |    4 +
 .../atomic_capture-2.c                        |   92 +-
 .../libgomp.oacc-c-c++-common/loop-auto-1.c   |    3 +
 .../libgomp.oacc-c-c++-common/loop-gwv-1.c    |   17 +-
 .../loop-red-gwv-1.c                          |   20 +-
 .../libgomp.oacc-c-c++-common/loop-red-v-1.c  |    6 +-
 .../libgomp.oacc-c-c++-common/loop-red-v-2.c  |    6 +-
 .../libgomp.oacc-c-c++-common/loop-red-w-1.c  |    7 +-
 .../libgomp.oacc-c-c++-common/loop-red-w-2.c  |    7 +-
 .../libgomp.oacc-c-c++-common/loop-red-wv-1.c |   10 +-
 .../libgomp.oacc-c-c++-common/loop-v-1.c      |    7 +-
 .../libgomp.oacc-c-c++-common/loop-w-1.c      |    9 +-
 .../libgomp.oacc-c-c++-common/loop-wv-1.c     |   10 +-
 .../libgomp.oacc-c-c++-common/parallel-dims.c |   33 +-
 .../private-variables-2.c                     |  217 +
 .../libgomp.oacc-c-c++-common/routine-gwv-1.c |   15 +-
 .../libgomp.oacc-c-c++-common/routine-v-1.c   |    7 +-
 .../libgomp.oacc-c-c++-common/routine-w-1.c   |    7 +-
 .../libgomp.oacc-c-c++-common/routine-wv-1.c  |   10 +-
 .../libgomp.oacc-c-c++-common/routine-wv-2.c  |    9 +-
 .../libgomp.oacc-c-c++-common/serial-dims.c   |    9 +-
 .../libgomp.oacc-c-c++-common/tile-1.c        |    3 +
 libgomp/testsuite/libgomp.oacc-c/c.exp        |    4 +
 .../libgomp.oacc-c/offload-targets-1.c        |    3 +
 libgomp/testsuite/libgomp.oacc-c/print-1.c    |   17 +
 .../libgomp.oacc-fortran/atomic_capture-1.f90 |  155 +-
 .../libgomp.oacc-fortran/collapse-1.f90       |    4 +-
 .../libgomp.oacc-fortran/collapse-2.f90       |    4 +-
 .../libgomp.oacc-fortran/error_stop-1.f       |    2 +-
 .../libgomp.oacc-fortran/error_stop-2.f       |    2 +-
 .../libgomp.oacc-fortran/error_stop-3.f       |    2 +-
 .../libgomp.oacc-fortran/fortran.exp          |    4 +
 .../libgomp.oacc-fortran/print-1.f90          |   15 +
 115 files changed, 7461 insertions(+), 214 deletions(-)
 create mode 100644 gcc/config/gcn/mkoffload.c
 create mode 100644 gcc/config/gcn/offload.h
 create mode 100644 libgcc/config/gcn/gomp_print.c
 create mode 100644 libgcc/config/gcn/reduction.c
 create mode 100644 libgcc/config/gcn/t-gcn-hsa
 rename libgomp/config/{nvptx => accel}/libgomp-plugin.c (100%)
 rename libgomp/config/{nvptx => accel}/lock.c (100%)
 rename libgomp/config/{nvptx => accel}/mutex.c (100%)
 rename libgomp/config/{nvptx => accel}/mutex.h (100%)
 rename libgomp/config/{nvptx => accel}/oacc-async.c (100%)
 rename libgomp/config/{nvptx => accel}/oacc-cuda.c (100%)
 rename libgomp/config/{nvptx => accel}/oacc-host.c (100%)
 rename libgomp/config/{nvptx => accel}/oacc-init.c (100%)
 rename libgomp/config/{nvptx => accel}/oacc-mem.c (100%)
 rename libgomp/config/{nvptx => accel}/oacc-plugin.c (100%)
 rename libgomp/config/{nvptx => accel}/omp-lock.h (100%)
 rename libgomp/config/{nvptx => accel}/openacc.f90 (96%)
 rename libgomp/config/{nvptx => accel}/pool.h (100%)
 rename libgomp/config/{nvptx => accel}/proc.c (98%)
 rename libgomp/config/{nvptx => accel}/ptrlock.c (100%)
 rename libgomp/config/{nvptx => accel}/ptrlock.h (100%)
 rename libgomp/config/{nvptx => accel}/sem.c (100%)
 rename libgomp/config/{nvptx => accel}/sem.h (100%)
 rename libgomp/config/{nvptx => accel}/thread-stacksize.h (100%)
 create mode 100644 libgomp/config/gcn/affinity-fmt.c
 create mode 100644 libgomp/config/gcn/bar.c
 create mode 100644 libgomp/config/gcn/bar.h
 create mode 100644 libgomp/config/gcn/doacross.h
 create mode 100644 libgomp/config/gcn/gomp_print.c
 create mode 100644 libgomp/config/gcn/icv-device.c
 create mode 100644 libgomp/config/gcn/simple-bar.h
 create mode 100644 libgomp/config/gcn/target.c
 create mode 100644 libgomp/config/gcn/task.c
 create mode 100644 libgomp/config/gcn/team.c
 create mode 100644 libgomp/config/gcn/time.c
 create mode 100644 libgomp/config/linux/gomp_print.c
 create mode 100644 libgomp/plugin/plugin-gcn.c
 create mode 100644 libgomp/testsuite/libgomp.c/for-1.h
 create mode 100644 libgomp/testsuite/libgomp.c/for-2.h
 create mode 100644 libgomp/testsuite/libgomp.c/for-3.c
 create mode 100644 libgomp/testsuite/libgomp.c/for-3.list
 create mode 100644 libgomp/testsuite/libgomp.c/for-5.c
 create mode 100644 libgomp/testsuite/libgomp.c/for-5.list
 create mode 100644 libgomp/testsuite/libgomp.c/for-6.c
 create mode 100644 libgomp/testsuite/libgomp.c/for-6.list
 create mode 100644 libgomp/testsuite/libgomp.c/target-print-1.c
 create mode 100644 libgomp/testsuite/libgomp.fortran/target-print-1.f90
 create mode 100644 libgomp/testsuite/libgomp.oacc-c-c++-common/private-variables-2.c
 create mode 100644 libgomp/testsuite/libgomp.oacc-c/print-1.c
 create mode 100644 libgomp/testsuite/libgomp.oacc-fortran/print-1.f90
diff mbox series

Patch

diff --git a/gcc/ChangeLog.openacc b/gcc/ChangeLog.openacc
index 9e1e9315923..4a806549d50 100644
--- a/gcc/ChangeLog.openacc
+++ b/gcc/ChangeLog.openacc
@@ -1,3 +1,10 @@ 
+2019-07-31  Julian Brown  <julian@codesourcery.com>
+	    Andrew Stubbs  <ams@codesourcery.com>
+
+	* config.gcc (amdgcn-*-*): Add default option for gfx906.
+	* config/gcn/mkoffload.c: New.
+	* config/gcn/offload.h: New.
+
 2019-06-25  Kwok Cheung Yeung  <kcy@codesourcery.com>
             Andrew Stubbs  <ams@codesourcery.com>
 
diff --git a/gcc/config.gcc b/gcc/config.gcc
index aff3bfad3d1..38e60c9bee8 100644
--- a/gcc/config.gcc
+++ b/gcc/config.gcc
@@ -4126,7 +4126,7 @@  case "${target}" in
 		for which in arch tune; do
 			eval "val=\$with_$which"
 			case ${val} in
-			"" | carrizo | fiji | gfx900 )
+			"" | carrizo | fiji | gfx900 | gfx906 )
 				# OK
 				;;
 			*)
diff --git a/gcc/config/gcn/mkoffload.c b/gcc/config/gcn/mkoffload.c
new file mode 100644
index 00000000000..f26b9023969
--- /dev/null
+++ b/gcc/config/gcn/mkoffload.c
@@ -0,0 +1,702 @@ 
+/* Offload image generation tool for AMD GCN.
+
+   Copyright (C) 2014-2019 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/>.  */
+
+/* Munges GCN assembly into a C source file defining the GCN code as a
+   string.
+
+   This is not a complete assembler.  We presume the source is well
+   formed from the compiler and can die horribly if it is not.  */
+
+#include "config.h"
+#include "system.h"
+#include "coretypes.h"
+#include "obstack.h"
+#include "diagnostic.h"
+#include "intl.h"
+#include <libgen.h>
+#include "collect-utils.h"
+#include "gomp-constants.h"
+
+const char tool_name[] = "gcn mkoffload";
+
+#define COMMENT_PREFIX "#"
+
+struct id_map
+{
+  id_map *next;
+  char *gcn_name;
+};
+
+static id_map *func_ids, **funcs_tail = &func_ids;
+static id_map *var_ids, **vars_tail = &var_ids;
+
+/* Files to unlink.  */
+static const char *gcn_s1_name;
+static const char *gcn_s2_name;
+static const char *gcn_o_name;
+static const char *gcn_cfile_name;
+
+enum offload_abi offload_abi = OFFLOAD_ABI_UNSET;
+
+/* Delete tempfiles.  */
+
+void
+tool_cleanup (bool from_signal ATTRIBUTE_UNUSED)
+{
+  if (gcn_cfile_name)
+    maybe_unlink (gcn_cfile_name);
+  if (gcn_s1_name)
+    maybe_unlink (gcn_s1_name);
+  if (gcn_s2_name)
+    maybe_unlink (gcn_s2_name);
+  if (gcn_o_name)
+    maybe_unlink (gcn_o_name);
+}
+
+static void
+mkoffload_cleanup (void)
+{
+  tool_cleanup (false);
+}
+
+/* Unlink FILE unless requested otherwise.  */
+
+void
+maybe_unlink (const char *file)
+{
+  if (!save_temps)
+    {
+      if (unlink_if_ordinary (file) && errno != ENOENT)
+	fatal_error (input_location, "deleting file %s: %m", file);
+    }
+  else if (verbose)
+    fprintf (stderr, "[Leaving %s]\n", file);
+}
+
+/* Add or change the value of an environment variable, outputting the
+   change to standard error if in verbose mode.  */
+
+static void
+xputenv (const char *string)
+{
+  if (verbose)
+    fprintf (stderr, "%s\n", string);
+  putenv (CONST_CAST (char *, string));
+}
+
+/* Read the whole input file.  It will be NUL terminated (but
+   remember, there could be a NUL in the file itself.  */
+
+static const char *
+read_file (FILE *stream, size_t *plen)
+{
+  size_t alloc = 16384;
+  size_t base = 0;
+  char *buffer;
+
+  if (!fseek (stream, 0, SEEK_END))
+    {
+      /* Get the file size.  */
+      long s = ftell (stream);
+      if (s >= 0)
+	alloc = s + 100;
+      fseek (stream, 0, SEEK_SET);
+    }
+  buffer = XNEWVEC (char, alloc);
+
+  for (;;)
+    {
+      size_t n = fread (buffer + base, 1, alloc - base - 1, stream);
+
+      if (!n)
+	break;
+      base += n;
+      if (base + 1 == alloc)
+	{
+	  alloc *= 2;
+	  buffer = XRESIZEVEC (char, buffer, alloc);
+	}
+    }
+  buffer[base] = 0;
+  *plen = base;
+  return buffer;
+}
+
+/* Parse STR, saving found tokens into PVALUES and return their number.
+   Tokens are assumed to be delimited by ':'.  */
+
+static unsigned
+parse_env_var (const char *str, char ***pvalues)
+{
+  const char *curval, *nextval;
+  char **values;
+  unsigned num = 1, i;
+
+  curval = strchr (str, ':');
+  while (curval)
+    {
+      num++;
+      curval = strchr (curval + 1, ':');
+    }
+
+  values = (char **) xmalloc (num * sizeof (char *));
+  curval = str;
+  nextval = strchr (curval, ':');
+  if (nextval == NULL)
+    nextval = strchr (curval, '\0');
+
+  for (i = 0; i < num; i++)
+    {
+      int l = nextval - curval;
+      values[i] = (char *) xmalloc (l + 1);
+      memcpy (values[i], curval, l);
+      values[i][l] = 0;
+      curval = nextval + 1;
+      nextval = strchr (curval, ':');
+      if (nextval == NULL)
+	nextval = strchr (curval, '\0');
+    }
+  *pvalues = values;
+  return num;
+}
+
+/* Auxiliary function that frees elements of PTR and PTR itself.
+   N is number of elements to be freed.  If PTR is NULL, nothing is freed.
+   If an element is NULL, subsequent elements are not freed.  */
+
+static void
+free_array_of_ptrs (void **ptr, unsigned n)
+{
+  unsigned i;
+  if (!ptr)
+    return;
+  for (i = 0; i < n; i++)
+    {
+      if (!ptr[i])
+	break;
+      free (ptr[i]);
+    }
+  free (ptr);
+  return;
+}
+
+/* Check whether NAME can be accessed in MODE.  This is like access,
+   except that it never considers directories to be executable.  */
+
+static int
+access_check (const char *name, int mode)
+{
+  if (mode == X_OK)
+    {
+      struct stat st;
+
+      if (stat (name, &st) < 0 || S_ISDIR (st.st_mode))
+	return -1;
+    }
+
+  return access (name, mode);
+}
+
+/* Parse an input assembler file, extract the offload tables etc.,
+   and output (1) the assembler code, minus the tables (which can contain
+   problematic relocations), and (2) a C file with the offload tables
+   encoded as structured data.  */
+
+static void
+process_asm (FILE *in, FILE *out, FILE *cfile)
+{
+  int fn_count = 0, var_count = 0, dims_count = 0;
+  struct obstack fns_os, vars_os, varsizes_os, dims_os;
+  obstack_init (&fns_os);
+  obstack_init (&vars_os);
+  obstack_init (&varsizes_os);
+  obstack_init (&dims_os);
+
+  struct oaccdims
+  {
+    int d[3];
+    char *name;
+  } dim;
+
+  /* Always add _init_array and _fini_array as kernels.  */
+  obstack_ptr_grow (&fns_os, xstrdup ("_init_array"));
+  obstack_ptr_grow (&fns_os, xstrdup ("_fini_array"));
+  fn_count += 2;
+
+  char buf[1000];
+  enum { IN_CODE, IN_VARS, IN_FUNCS } state = IN_CODE;
+  while (fgets (buf, sizeof (buf), in))
+    {
+      switch (state)
+	{
+	case IN_CODE:
+	  {
+	    if (sscanf (buf, " ;; OPENACC-DIMS: %d, %d, %d : %ms\n",
+			&dim.d[0], &dim.d[1], &dim.d[2], &dim.name) == 4)
+	      {
+		obstack_grow (&dims_os, &dim, sizeof (dim));
+		dims_count++;
+	      }
+	    break;
+	  }
+	case IN_VARS:
+	  {
+	    char *varname;
+	    unsigned varsize;
+	    if (sscanf (buf, " .8byte %ms\n", &varname))
+	      {
+		obstack_ptr_grow (&vars_os, varname);
+		fgets (buf, sizeof (buf), in);
+		if (!sscanf (buf, " .8byte %u\n", &varsize))
+		  abort ();
+		obstack_int_grow (&varsizes_os, varsize);
+		var_count++;
+
+		/* The HSA Runtime cannot locate the symbol if it is not
+		   exported from the kernel.  */
+		fprintf (out, "\t.global %s\n", varname);
+	      }
+	    break;
+	  }
+	case IN_FUNCS:
+	  {
+	    char *funcname;
+	    if (sscanf (buf, "\t.8byte\t%ms\n", &funcname))
+	      {
+		obstack_ptr_grow (&fns_os, funcname);
+		fn_count++;
+		continue;
+	      }
+	    break;
+	  }
+	}
+
+      char dummy;
+      if (sscanf (buf, " .section .gnu.offload_vars%c", &dummy) > 0)
+	state = IN_VARS;
+      else if (sscanf (buf, " .section .gnu.offload_funcs%c", &dummy) > 0)
+	state = IN_FUNCS;
+      else if (sscanf (buf, " .section %c", &dummy) > 0
+	       || sscanf (buf, " .text%c", &dummy) > 0
+	       || sscanf (buf, " .bss%c", &dummy) > 0
+	       || sscanf (buf, " .data%c", &dummy) > 0
+	       || sscanf (buf, " .ident %c", &dummy) > 0)
+	state = IN_CODE;
+
+      if (state == IN_CODE)
+	fputs (buf, out);
+    }
+
+  char **fns = XOBFINISH (&fns_os, char **);
+  struct oaccdims *dims = XOBFINISH (&dims_os, struct oaccdims *);
+
+  fprintf (cfile, "#include <stdlib.h>\n");
+  fprintf (cfile, "#include <stdbool.h>\n\n");
+
+  char **vars = XOBFINISH (&vars_os, char **);
+  unsigned *varsizes = XOBFINISH (&varsizes_os, unsigned *);
+  fprintf (cfile,
+	   "static const struct global_var_info {\n"
+	   "  const char *name;\n"
+	   "  void *address;\n"
+	   "} vars[] = {\n");
+  int i;
+  for (i = 0; i < var_count; ++i)
+    {
+      const char *sep = i < var_count - 1 ? "," : " ";
+      fprintf (cfile, "  { \"%s\", NULL }%s /* size: %u */\n", vars[i], sep,
+	       varsizes[i]);
+    }
+  fprintf (cfile, "};\n\n");
+
+  obstack_free (&vars_os, NULL);
+  obstack_free (&varsizes_os, NULL);
+
+  /* Dump out function idents.  */
+  fprintf (cfile, "static const struct hsa_kernel_description {\n"
+	   "  const char *name;\n"
+	   "  unsigned omp_data_size;\n"
+	   "  bool gridified_kernel_p;\n"
+	   "  unsigned kernel_dependencies_count;\n"
+	   "  const char **kernel_dependencies;\n"
+	   "  int oacc_dims[3];\n"
+	   "} gcn_kernels[] = {\n  ");
+  dim.d[0] = dim.d[1] = dim.d[2] = 0;
+  const char *comma;
+  for (comma = "", i = 0; i < fn_count; comma = ",\n  ", i++)
+    {
+      /* Find if we recorded dimensions for this function.  */
+      int *d = dim.d;		/* Previously zeroed.  */
+      for (int j = 0; j < dims_count; j++)
+	if (strcmp (fns[i], dims[j].name) == 0)
+	  {
+	    d = dims[j].d;
+	    break;
+	  }
+
+      fprintf (cfile, "%s{\"%s\", 0, 0, 0, NULL, {%d, %d, %d}}", comma,
+	       fns[i], d[0], d[1], d[2]);
+
+      free (fns[i]);
+    }
+  fprintf (cfile, "\n};\n\n");
+
+  obstack_free (&fns_os, NULL);
+  for (i = 0; i < dims_count; i++)
+    free (dims[i].name);
+  obstack_free (&dims_os, NULL);
+}
+
+/* Embed an object file into a C source file.  */
+
+static void
+process_obj (FILE *in, FILE *cfile)
+{
+  size_t len = 0;
+  const char *input = read_file (in, &len);
+  id_map const *id;
+  unsigned ix;
+
+  /* Dump out an array containing the binary.
+     FIXME: do this with objcopy.  */
+  fprintf (cfile, "static unsigned char gcn_code[] = {");
+  for (size_t i = 0; i < len; i += 17)
+    {
+      fprintf (cfile, "\n\t");
+      for (size_t j = i; j < i + 17 && j < len; j++)
+	fprintf (cfile, "%3u,", (unsigned char) input[j]);
+    }
+  fprintf (cfile, "\n};\n\n");
+
+  fprintf (cfile,
+	   "static const struct gcn_image {\n"
+	   "  char magic[4];\n"
+	   "  size_t size;\n"
+	   "  void *image;\n"
+	   "} gcn_image = {\n"
+	   "  \"GCN\",\n"
+	   "  %zu,\n"
+	   "  gcn_code\n"
+	   "};\n\n",
+	   len);
+
+  fprintf (cfile,
+	   "static const struct gcn_image_desc {\n"
+	   "  const struct gcn_image *gcn_image;\n"
+	   "  unsigned kernel_count;\n"
+	   "  const struct hsa_kernel_description *kernel_infos;\n"
+	   "  unsigned global_variable_count;\n"
+	   "  const struct global_var_info *global_variables;\n"
+	   "} target_data = {\n"
+	   "  &gcn_image,\n"
+	   "  sizeof (gcn_kernels) / sizeof (gcn_kernels[0]),\n"
+	   "  gcn_kernels,\n"
+	   "  sizeof (vars) / sizeof (vars[0]),\n"
+	   "  vars\n"
+	   "};\n\n");
+
+  fprintf (cfile,
+	   "#ifdef __cplusplus\n"
+	   "extern \"C\" {\n"
+	   "#endif\n"
+	   "extern void GOMP_offload_register_ver"
+	   " (unsigned, const void *, int, const void *);\n"
+	   "extern void GOMP_offload_unregister_ver"
+	   " (unsigned, const void *, int, const void *);\n"
+	   "#ifdef __cplusplus\n"
+	   "}\n"
+	   "#endif\n\n");
+
+  fprintf (cfile, "extern const void *const __OFFLOAD_TABLE__[];\n\n");
+
+  fprintf (cfile, "static __attribute__((constructor)) void init (void)\n"
+	   "{\n"
+	   "  GOMP_offload_register_ver (%#x, __OFFLOAD_TABLE__,"
+	   " %d/*GCN*/, &target_data);\n"
+	   "};\n",
+	   GOMP_VERSION_PACK (GOMP_VERSION, GOMP_VERSION_GCN),
+	   GOMP_DEVICE_GCN);
+
+  fprintf (cfile, "static __attribute__((destructor)) void fini (void)\n"
+	   "{\n"
+	   "  GOMP_offload_unregister_ver (%#x, __OFFLOAD_TABLE__,"
+	   " %d/*GCN*/, &target_data);\n"
+	   "};\n",
+	   GOMP_VERSION_PACK (GOMP_VERSION, GOMP_VERSION_GCN),
+	   GOMP_DEVICE_GCN);
+}
+
+/* Compile a C file using the host compiler.  */
+
+static void
+compile_native (const char *infile, const char *outfile, const char *compiler)
+{
+  const char *collect_gcc_options = getenv ("COLLECT_GCC_OPTIONS");
+  if (!collect_gcc_options)
+    fatal_error (input_location,
+		 "environment variable COLLECT_GCC_OPTIONS must be set");
+
+  struct obstack argv_obstack;
+  obstack_init (&argv_obstack);
+  obstack_ptr_grow (&argv_obstack, compiler);
+  if (save_temps)
+    obstack_ptr_grow (&argv_obstack, "-save-temps");
+  if (verbose)
+    obstack_ptr_grow (&argv_obstack, "-v");
+  switch (offload_abi)
+    {
+    case OFFLOAD_ABI_LP64:
+      obstack_ptr_grow (&argv_obstack, "-m64");
+      break;
+    case OFFLOAD_ABI_ILP32:
+      obstack_ptr_grow (&argv_obstack, "-m32");
+      break;
+    default:
+      gcc_unreachable ();
+    }
+  obstack_ptr_grow (&argv_obstack, infile);
+  obstack_ptr_grow (&argv_obstack, "-c");
+  obstack_ptr_grow (&argv_obstack, "-o");
+  obstack_ptr_grow (&argv_obstack, outfile);
+  obstack_ptr_grow (&argv_obstack, NULL);
+
+  const char **new_argv = XOBFINISH (&argv_obstack, const char **);
+  fork_execute (new_argv[0], CONST_CAST (char **, new_argv), true);
+  obstack_free (&argv_obstack, NULL);
+}
+
+int
+main (int argc, char **argv)
+{
+  FILE *in = stdin;
+  FILE *out = stdout;
+  FILE *cfile = stdout;
+  const char *outname = 0, *offloadsrc = 0;
+
+  progname = "mkoffload";
+  diagnostic_initialize (global_dc, 0);
+
+  if (atexit (mkoffload_cleanup) != 0)
+    fatal_error (input_location, "atexit failed");
+
+  char *collect_gcc = getenv ("COLLECT_GCC");
+  if (collect_gcc == NULL)
+    fatal_error (input_location, "COLLECT_GCC must be set.");
+  const char *gcc_path = dirname (ASTRDUP (collect_gcc));
+  const char *gcc_exec = basename (ASTRDUP (collect_gcc));
+
+  size_t len = (strlen (gcc_path) + 1 + strlen (GCC_INSTALL_NAME) + 1);
+  char *driver = XALLOCAVEC (char, len);
+
+  if (strcmp (gcc_exec, collect_gcc) == 0)
+    /* collect_gcc has no path, so it was found in PATH.  Make sure we also
+       find accel-gcc in PATH.  */
+    gcc_path = NULL;
+
+  int driver_used = 0;
+  if (gcc_path != NULL)
+    driver_used = sprintf (driver, "%s/", gcc_path);
+  sprintf (driver + driver_used, "%s", GCC_INSTALL_NAME);
+
+  bool found = false;
+  if (gcc_path == NULL)
+    found = true;
+  else if (access_check (driver, X_OK) == 0)
+    found = true;
+  else
+    {
+      /* Don't use alloca pointer with XRESIZEVEC.  */
+      driver = NULL;
+      /* Look in all COMPILER_PATHs for GCC_INSTALL_NAME.  */
+      char **paths = NULL;
+      unsigned n_paths;
+      n_paths = parse_env_var (getenv ("COMPILER_PATH"), &paths);
+      for (unsigned i = 0; i < n_paths; i++)
+	{
+	  len = strlen (paths[i]) + 1 + strlen (GCC_INSTALL_NAME) + 1;
+	  driver = XRESIZEVEC (char, driver, len);
+	  sprintf (driver, "%s/%s", paths[i], GCC_INSTALL_NAME);
+	  if (access_check (driver, X_OK) == 0)
+	    {
+	      found = true;
+	      break;
+	    }
+	}
+      free_array_of_ptrs ((void **) paths, n_paths);
+    }
+
+  if (!found)
+    fatal_error (input_location,
+		 "offload compiler %s not found", GCC_INSTALL_NAME);
+
+  /* We may be called with all the arguments stored in some file and
+     passed with @file.  Expand them into argv before processing.  */
+  expandargv (&argc, &argv);
+
+  /* Scan the argument vector.  */
+  bool fopenmp = false;
+  bool fopenacc = false;
+  for (int i = 1; i < argc; i++)
+    {
+#define STR "-foffload-abi="
+      if (strncmp (argv[i], STR, strlen (STR)) == 0)
+	{
+	  if (strcmp (argv[i] + strlen (STR), "lp64") == 0)
+	    offload_abi = OFFLOAD_ABI_LP64;
+	  else if (strcmp (argv[i] + strlen (STR), "ilp32") == 0)
+	    offload_abi = OFFLOAD_ABI_ILP32;
+	  else
+	    fatal_error (input_location,
+			 "unrecognizable argument of option " STR);
+	}
+#undef STR
+      else if (strcmp (argv[i], "-fopenmp") == 0)
+	fopenmp = true;
+      else if (strcmp (argv[i], "-fopenacc") == 0)
+	fopenacc = true;
+      else if (strcmp (argv[i], "-save-temps") == 0)
+	save_temps = true;
+      else if (strcmp (argv[i], "-v") == 0)
+	verbose = true;
+    }
+  if (!(fopenacc ^ fopenmp))
+    fatal_error (input_location, "either -fopenacc or -fopenmp must be set");
+
+  const char *abi;
+  switch (offload_abi)
+    {
+    case OFFLOAD_ABI_LP64:
+      abi = "-m64";
+      break;
+    case OFFLOAD_ABI_ILP32:
+      abi = "-m32";
+      break;
+    default:
+      gcc_unreachable ();
+    }
+
+  gcn_s1_name = make_temp_file (".mkoffload.1.s");
+  gcn_s2_name = make_temp_file (".mkoffload.2.s");
+  gcn_o_name = make_temp_file (".mkoffload.hsaco");
+  gcn_cfile_name = make_temp_file (".c");
+
+  /* Build arguments for compiler pass.  */
+  struct obstack cc_argv_obstack;
+  obstack_init (&cc_argv_obstack);
+  obstack_ptr_grow (&cc_argv_obstack, driver);
+  obstack_ptr_grow (&cc_argv_obstack, "-S");
+
+  if (save_temps)
+    obstack_ptr_grow (&cc_argv_obstack, "-save-temps");
+  if (verbose)
+    obstack_ptr_grow (&cc_argv_obstack, "-v");
+  obstack_ptr_grow (&cc_argv_obstack, abi);
+  obstack_ptr_grow (&cc_argv_obstack, "-xlto");
+  if (fopenmp)
+    obstack_ptr_grow (&cc_argv_obstack, "-mgomp");
+
+  for (int ix = 1; ix != argc; ix++)
+    {
+      if (!strcmp (argv[ix], "-o") && ix + 1 != argc)
+	outname = argv[++ix];
+      else
+	{
+	  obstack_ptr_grow (&cc_argv_obstack, argv[ix]);
+
+	  if (argv[ix][0] != '-')
+	    offloadsrc = argv[ix];
+	}
+    }
+
+  obstack_ptr_grow (&cc_argv_obstack, "-o");
+  obstack_ptr_grow (&cc_argv_obstack, gcn_s1_name);
+  obstack_ptr_grow (&cc_argv_obstack,
+		    concat ("-mlocal-symbol-id=", offloadsrc, NULL));
+  obstack_ptr_grow (&cc_argv_obstack, NULL);
+  const char **cc_argv = XOBFINISH (&cc_argv_obstack, const char **);
+
+  /* Build arguments for assemble/link pass.  */
+  struct obstack ld_argv_obstack;
+  obstack_init (&ld_argv_obstack);
+  obstack_ptr_grow (&ld_argv_obstack, driver);
+  obstack_ptr_grow (&ld_argv_obstack, gcn_s2_name);
+  obstack_ptr_grow (&ld_argv_obstack, "-lgomp");
+
+  for (int i = 1; i < argc; i++)
+    if (strncmp (argv[i], "-l", 2) == 0
+	|| strncmp (argv[i], "-Wl", 3) == 0
+	|| strncmp (argv[i], "-march", 6) == 0)
+      obstack_ptr_grow (&ld_argv_obstack, argv[i]);
+
+  obstack_ptr_grow (&ld_argv_obstack, "-o");
+  obstack_ptr_grow (&ld_argv_obstack, gcn_o_name);
+  obstack_ptr_grow (&ld_argv_obstack, NULL);
+  const char **ld_argv = XOBFINISH (&ld_argv_obstack, const char **);
+
+  /* Clean up unhelpful environment variables.  */
+  char *execpath = getenv ("GCC_EXEC_PREFIX");
+  char *cpath = getenv ("COMPILER_PATH");
+  char *lpath = getenv ("LIBRARY_PATH");
+  unsetenv ("GCC_EXEC_PREFIX");
+  unsetenv ("COMPILER_PATH");
+  unsetenv ("LIBRARY_PATH");
+
+  /* Run the compiler pass.  */
+  fork_execute (cc_argv[0], CONST_CAST (char **, cc_argv), true);
+  obstack_free (&cc_argv_obstack, NULL);
+
+  in = fopen (gcn_s1_name, "r");
+  if (!in)
+    fatal_error (input_location, "cannot open intermediate gcn asm file");
+
+  out = fopen (gcn_s2_name, "w");
+  if (!out)
+    fatal_error (input_location, "cannot open '%s'", gcn_s2_name);
+
+  cfile = fopen (gcn_cfile_name, "w");
+  if (!cfile)
+    fatal_error (input_location, "cannot open '%s'", gcn_cfile_name);
+
+  process_asm (in, out, cfile);
+
+  fclose (in);
+  fclose (out);
+
+  /* Run the assemble/link pass.  */
+  fork_execute (ld_argv[0], CONST_CAST (char **, ld_argv), true);
+  obstack_free (&ld_argv_obstack, NULL);
+
+  in = fopen (gcn_o_name, "r");
+  if (!in)
+    fatal_error (input_location, "cannot open intermediate gcn obj file");
+
+  process_obj (in, cfile);
+
+  fclose (in);
+  fclose (cfile);
+
+  xputenv (concat ("GCC_EXEC_PREFIX=", execpath, NULL));
+  xputenv (concat ("COMPILER_PATH=", cpath, NULL));
+  xputenv (concat ("LIBRARY_PATH=", lpath, NULL));
+
+  compile_native (gcn_cfile_name, outname, collect_gcc);
+
+  return 0;
+}
diff --git a/gcc/config/gcn/offload.h b/gcc/config/gcn/offload.h
new file mode 100644
index 00000000000..795ee3f0e0e
--- /dev/null
+++ b/gcc/config/gcn/offload.h
@@ -0,0 +1,35 @@ 
+/* Support for AMD GCN offloading.
+
+   Copyright (C) 2014-2019 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.
+
+   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/>.  */
+
+#ifndef GCC_GCN_OFFLOAD_H
+#define GCC_GCN_OFFLOAD_H
+
+/* Support for OpenACC acc_on_device.  */
+
+#include "gomp-constants.h"
+
+#define ACCEL_COMPILER_acc_device GOMP_DEVICE_GCN
+
+#endif
diff --git a/include/gomp-constants.h b/include/gomp-constants.h
index 22f9520524d..0e15cfb303e 100644
--- a/include/gomp-constants.h
+++ b/include/gomp-constants.h
@@ -219,7 +219,8 @@  enum gomp_map_kind
 #define GOMP_DEVICE_NVIDIA_PTX		5
 #define GOMP_DEVICE_INTEL_MIC		6
 #define GOMP_DEVICE_HSA			7
-#define GOMP_DEVICE_CURRENT		8
+#define GOMP_DEVICE_GCN			8
+#define GOMP_DEVICE_CURRENT		9
 
 #define GOMP_DEVICE_ICV			-1
 #define GOMP_DEVICE_HOST_FALLBACK	-2
@@ -276,6 +277,7 @@  enum gomp_map_kind
 #define GOMP_VERSION_NVIDIA_PTX 1
 #define GOMP_VERSION_INTEL_MIC 0
 #define GOMP_VERSION_HSA 0
+#define GOMP_VERSION_GCN 0
 
 #define GOMP_VERSION_PACK(LIB, DEV) (((LIB) << 16) | (DEV))
 #define GOMP_VERSION_LIB(PACK) (((PACK) >> 16) & 0xffff)
diff --git a/libgcc/ChangeLog.openacc b/libgcc/ChangeLog.openacc
index 45fc6aaf530..a237a700837 100644
--- a/libgcc/ChangeLog.openacc
+++ b/libgcc/ChangeLog.openacc
@@ -1,3 +1,13 @@ 
+2019-07-31  Julian Brown  <julian@codesourcery.com>
+	    Andrew Stubbs  <ams@codesourcery.com>
+
+	* Makefile.in: Allow disabling of emutls.
+	* config/gcn/gomp_print.c: New.
+	* config/gcn/reduction.c: New.
+	* config/gcn/t-amdgcn (LIB2ADD): Add gomp_print.c and reduction.c.
+	Disable emutls.c.
+	* config/gcn/t-gcn-hsa: New.
+
 2019-06-25  Andrew Stubbs  <ams@codesourcery.com>
 
 	Backport from mainline:
diff --git a/libgcc/Makefile.in b/libgcc/Makefile.in
index ea390a5bbea..3261ab05cd9 100644
--- a/libgcc/Makefile.in
+++ b/libgcc/Makefile.in
@@ -430,9 +430,11 @@  LIB2ADD += enable-execute-stack.c
 # While emutls.c has nothing to do with EH, it is in LIB2ADDEH*
 # instead of LIB2ADD because that's the way to be sure on some targets
 # (e.g. *-*-darwin*) only one copy of it is linked.
+ifneq ($(enable_emutls),no)
 LIB2ADDEH += $(srcdir)/emutls.c
 LIB2ADDEHSTATIC += $(srcdir)/emutls.c
 LIB2ADDEHSHARED += $(srcdir)/emutls.c
+endif
 
 # Library members defined in libgcc2.c.
 lib2funcs = _muldi3 _negdi2 _lshrdi3 _ashldi3 _ashrdi3 _cmpdi2 _ucmpdi2	   \
diff --git a/libgcc/config/gcn/gomp_print.c b/libgcc/config/gcn/gomp_print.c
new file mode 100644
index 00000000000..5d1a3fc811d
--- /dev/null
+++ b/libgcc/config/gcn/gomp_print.c
@@ -0,0 +1,101 @@ 
+/* Newlib may not have been built yet.  */
+typedef long int64_t;
+typedef long size_t;
+extern char *strncpy (char *dst, const char *src, size_t length);
+extern void exit(int);
+
+void gomp_print_string (const char *msg, const char *value);
+void gomp_print_integer (const char *msg, int64_t value);
+void gomp_print_double (const char *msg, double value);
+
+/* This struct must match the one used by gcn-run and libgomp.
+   It holds all the data output from a kernel (besides mapping data).
+
+   The base address pointer can be found at kernargs+16.
+
+   The next_output counter must be atomically incremented for each
+   print output.  Only when the print data is fully written can the
+   "written" flag be set.  */
+struct output {
+  int return_value;
+  unsigned int next_output;
+  struct printf_data {
+    int written;
+    char msg[128];
+    int type;
+    union {
+      int64_t ivalue;
+      double dvalue;
+      char text[128];
+    };
+  } queue[1024];
+  unsigned int consumed;
+};
+
+static struct printf_data *
+reserve_print_slot (void) {
+  /* The kernargs pointer is in s[8:9].
+     This will break if the enable_sgpr_* flags are ever changed.  */
+  char *kernargs;
+  asm ("s_mov_b64 %0, s[8:9]" : "=Sg"(kernargs));
+
+  /* The output data is at kernargs[2].  */
+  struct output *data = *(struct output **)(kernargs + 16);
+
+  /* Reserve the slot.  */
+  unsigned int index = __atomic_fetch_add (&data->next_output, 1,
+					   __ATOMIC_ACQUIRE);
+
+  /* Spinlock while the host catches up.  */
+  if (index >= 1024)
+    while (__atomic_load_n (&data->consumed, __ATOMIC_ACQUIRE)
+	   <= (index - 1024))
+      asm ("s_sleep 64");
+
+  if ((unsigned int)(index + 1) < data->consumed)
+    {
+      /* Overflow.  */
+      exit (1);
+    }
+  return &(data->queue[index%1024]);
+}
+
+void
+gomp_print_string (const char *msg, const char *value)
+{
+  struct printf_data *output = reserve_print_slot ();
+  output->type = 2; /* String.  */
+
+  strncpy (output->msg, msg, 127);
+  output->msg[127] = '\0';
+  strncpy (output->text, value, 127);
+  output->text[127] = '\0';
+
+  __atomic_store_n (&output->written, 1, __ATOMIC_RELEASE);
+}
+
+void
+gomp_print_integer (const char *msg, int64_t value)
+{
+  struct printf_data *output = reserve_print_slot ();
+  output->type = 0; /* Integer.  */
+
+  strncpy (output->msg, msg, 127);
+  output->msg[127] = '\0';
+  output->ivalue = value;
+
+  __atomic_store_n (&output->written, 1, __ATOMIC_RELEASE);
+}
+
+void
+gomp_print_double (const char *msg, double value)
+{
+  struct printf_data *output = reserve_print_slot ();
+  output->type = 1; /* Double.  */
+
+  strncpy (output->msg, msg, 127);
+  output->msg[127] = '\0';
+  output->dvalue = value;
+
+  __atomic_store_n (&output->written, 1, __ATOMIC_RELEASE);
+}
diff --git a/libgcc/config/gcn/reduction.c b/libgcc/config/gcn/reduction.c
new file mode 100644
index 00000000000..eafcee8ca8a
--- /dev/null
+++ b/libgcc/config/gcn/reduction.c
@@ -0,0 +1,30 @@ 
+/* Oversized reductions lock variable
+   Copyright (C) 2017-2019 Free Software Foundation, Inc.
+   Contributed by Mentor Graphics.
+
+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.
+
+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/>.  */
+
+/* We use a global lock variable for reductions on objects larger than
+   64 bits.  Until and unless proven that lock contention for
+   different reductions is a problem, a single lock will suffice.  */
+
+unsigned volatile __reduction_lock = 0;
diff --git a/libgcc/config/gcn/t-amdgcn b/libgcc/config/gcn/t-amdgcn
index adbd866a1d9..c04de3ce4bc 100644
--- a/libgcc/config/gcn/t-amdgcn
+++ b/libgcc/config/gcn/t-amdgcn
@@ -1,7 +1,11 @@ 
+LIB2ADD += $(srcdir)/config/gcn/gomp_print.c
+
 LIB2ADD += $(srcdir)/config/gcn/lib2-divmod.c \
 	   $(srcdir)/config/gcn/lib2-divmod-hi.c \
 	   $(srcdir)/config/gcn/unwind-gcn.c
 
+LIB2ADD += $(srcdir)/config/gcn/reduction.c
+
 LIB2ADDEH=
 LIB2FUNCS_EXCLUDE=__main
 
@@ -13,5 +17,10 @@  LIBGCC2_DEBUG_CFLAGS = -g0
 crt0.o: $(srcdir)/config/gcn/crt0.c
 	$(crt_compile) -c $<
 
-# Prevent building "advanced" stuff (for example, gcov support).
+# Prevent building "advanced" stuff (for example, gcov support).  We don't
+# support it, and it may cause the build to fail, because of alloca usage, for
+# example.
 INHIBIT_LIBC_CFLAGS = -Dinhibit_libc
+
+# Disable emutls.c (temporarily?)
+enable_emutls = no
diff --git a/libgcc/config/gcn/t-gcn-hsa b/libgcc/config/gcn/t-gcn-hsa
new file mode 100644
index 00000000000..1600a586ac4
--- /dev/null
+++ b/libgcc/config/gcn/t-gcn-hsa
@@ -0,0 +1,52 @@ 
+#  Copyright (C) 2016-2019 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 of the License, 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.
+#
+#  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/>.
+
+GTM_H += $(HASH_TABLE_H)
+
+driver-gcn.o: $(srcdir)/config/gcn/driver-gcn.c
+	$(COMPILE) $<
+	$(POSTCOMPILE)
+
+CFLAGS-mkoffload.o += $(DRIVER_DEFINES) \
+	-DGCC_INSTALL_NAME=\"$(GCC_INSTALL_NAME)\"
+mkoffload.o: $(srcdir)/config/gcn/mkoffload.c
+	$(COMPILE) $<
+	$(POSTCOMPILE)
+ALL_HOST_OBJS += mkoffload.o
+
+mkoffload$(exeext): mkoffload.o collect-utils.o libcommon-target.a \
+		      $(LIBIBERTY) $(LIBDEPS)
+	+$(LINKER) $(ALL_LINKERFLAGS) $(LDFLAGS) -o $@ \
+	  mkoffload.o collect-utils.o libcommon-target.a $(LIBIBERTY) $(LIBS)
+
+CFLAGS-gcn-run.o += -DVERSION_STRING=$(PKGVERSION_s)
+COMPILE-gcn-run.o = $(filter-out -fno-rtti,$(COMPILE))
+gcn-run.o: $(srcdir)/config/gcn/gcn-run.c
+	$(COMPILE-gcn-run.o) -x c -std=gnu11 -Wno-error=pedantic $<
+	$(POSTCOMPILE)
+ALL_HOST_OBJS += gcn-run.o
+
+gcn-run$(exeext): gcn-run.o
+	+$(LINKER) $(ALL_LINKERFLAGS) $(LDFLAGS) -o $@ $< -ldl
+
+MULTILIB_OPTIONS = march=gfx900 march=gfx906
+MULTILIB_DIRNAMES = gfx900 gfx906
+
+PASSES_EXTRA += $(srcdir)/config/gcn/gcn-passes.def
+gcn-tree.o: $(srcdir)/config/gcn/gcn-tree.c
+	$(COMPILE) $<
+	$(POSTCOMPILE)
+ALL_HOST_OBJS += gcn-tree.o
diff --git a/libgomp/ChangeLog.openacc b/libgomp/ChangeLog.openacc
index caf5f6e2fb1..e6c81d4f43f 100644
--- a/libgomp/ChangeLog.openacc
+++ b/libgomp/ChangeLog.openacc
@@ -1,3 +1,160 @@ 
+2019-07-31  Julian Brown  <julian@codesourcery.com>
+	    Andrew Stubbs  <ams@codesourcery.com>
+
+	* Makefile.am (libgomp_la_SOURCES): Add gomp_print.c.
+	* Makefile.in: Regenerate.
+	* affinity-fmt.c: Rename calls to gomp_write_string from
+	gomp_print_string.
+	* config.h.in (PLUGIN_GCN): Add #undef.
+	* config/nvptx/libgomp-plugin.c: Rename to...
+	* config/accel/libgomp-plugin.c: ...this.
+	* config/nvptx/lock.c: Rename to...
+	* config/accel/lock.c: ...this.
+	* config/nvptx/mutex.c: Rename to...
+	* config/accel/mutex.c: ...this.
+	* config/nvptx/mutex.h: Rename to...
+	* config/accel/mutex.h: ...this.
+	* config/nvptx/oacc-async.c: Rename to...
+	* config/accel/oacc-async.c: ...this.
+	* config/nvptx/oacc-cuda.c: Rename to...
+	* config/accel/oacc-cuda.c: ...this.
+	* config/nvptx/oacc-host.c: Rename to...
+	* config/accel/oacc-host.c: ...this.
+	* config/nvptx/oacc-init.c: Rename to...
+	* config/accel/oacc-init.c: ...this.
+	* config/nvptx/oacc-mem.c: Rename to...
+	* config/accel/oacc-mem.c: ...this.
+	* config/nvptx/oacc-plugin.c: Rename to...
+	* config/accel/oacc-plugin.c: ...this.
+	* config/nvptx/omp-lock.h: Rename to...
+	* config/accel/omp-lock.h: ...this.
+	* config/nvptx/openacc.f90: Rename to...
+	* config/accel/openacc.f90: ...this.  Add acc_device_hsa and
+	acc_device_gcn.
+	* config/nvptx/pool.h: Rename to...
+	* config/accel/pool.h: ...this.
+	* config/nvptx/proc.c: Rename to...
+	* config/accel/proc.c: ...this.  Add omp_get_num_procs alias.
+	* config/nvptx/ptrlock.c: Rename to...
+	* config/accel/ptrlock.c: ...this.
+	* config/nvptx/ptrlock.h: Rename to...
+	* config/accel/ptrlock.h: ...this.
+	* config/nvptx/sem.c: Rename to...
+	* config/accel/sem.c: ...this.
+	* config/nvptx/sem.h: Rename to...
+	* config/accel/sem.h: ...this.
+	* config/nvptx/thread-stacksize.h: Rename to...
+	* config/accel/thread-stacksize.h: ...this.
+	* config/gcn/affinity-fmt.c: New.
+	* config/gcn/bar.c: New.
+	* config/gcn/bar.h: New.
+	* config/gcn/doacross.h: New.
+	* config/gcn/gomp_print.c: New.
+	* config/gcn/icv-device.c: New.
+	* config/gcn/simple-bar.h: New.
+	* config/gcn/target.c: New.
+	* config/gcn/task.c: New.
+	* config/gcn/team.c: New.
+	* config/gcn/time.c: New.
+	* config/linux/gomp_print.c: New.
+	* configure.ac (amdgcn*-*-*): Disable pthreads.
+	* configure: Regenerated.
+	* configure.tgt (nvptx*-*-*): Add 'accel' config_path.
+	(amdgcn*-*-*): Set config_path.
+	* fortran.c (omp_display_affinity_): Rename calls to gomp_write_string
+	from gomp_print_string.
+	* libgomp-plugin.h (enum offload_target_type): Add
+	OFFLOAD_TARGET_TYPE_GCN.
+	(GOMP_OFFLOAD_openacc_async_construct): Change parameter type to int.
+	* libgomp.h (gcn_thrs, set_gcn_thrs, gomp_thread): Add for __AMDGCN__.
+	(gomp_print_string): Rename to...
+	(gomp_write_string): ...this.
+	* libgomp.map (GOMP_4.5): Add gomp_rpint_string, gomp_print_integer,
+	gomp_print_double.
+	* oacc-async.c (lookup_goacc_asyncqueue): Pass target_id to async queue
+	construct function.
+	* oacc-host.c (host_openacc_async_construct): Add dummy device
+	parameter.
+	* oacc-init.c (name_of_acc_device_t): Add acc_device_gcn.
+	* oacc-int.h (goacc_thread): Add dummy implementation for __AMDGCN__.
+	* oacc-parallel.c (GOACC_enter_exit_data): Support acc_async_noval and
+	zero-length array sections.
+	* omp.h.in (gomp_print_string, gomp_print_integer, gomp_print_double):
+	Add prototypes.
+	* omp_lib.f90.in (gomp_print_string, gomp_print_integer,
+	gomp_print_double): Add interfaces.
+	* openacc.f90 (openacc_kinds): Add acc_device_gcn.  Bump
+	acc_device_current code.
+	* openacc.h (acc_device_t): Add acc_device_gcn, bump acc_device_current
+	code.
+	* openacc_lib.h (acc_device_hsa, acc_device_gcn): Add.
+	* plugin/Makefrag.am (PLUGIN_GCN): Support building GCN plugin.
+	* plugin/configfrag.am (PLUGIN_GCN, PLUGIN_GCN_CPPFLAGS,
+	PLUGIN_GCN_LDFLAGS, PLUGIN_GCN_LIBS): Add.  Add suport for GCN plugin.
+	* plugin/plugin-gcn.c: New.
+	* target.c (stdio.h): Include unconditionally.
+	(gomp_copy_host2dev): Add function comment.
+	(copy_host2dev_immediate): New function.
+	(gomp_map_pointer, gomp_map_vars_internal): Use
+	copy_host2dev_immediate where appropriate.
+	(offload_target_to_plugin_name): Support gcn.
+	* team.c (gomp_free_pool_helper): Support gcn.
+	* testsuite/Makefile.in: Regenerated.
+	* testsuite/lib/libgomp.exp
+	(check_effective_target_openacc_amdgcn_accel_present): New.
+	(check_effective_target_openacc_amdgcn_accel_selected): New.
+	* testsuite/libgomp.c/c.exp (generate_tests, test_lists,
+	generated_tests): New.
+	(tests): Add generated tests.
+	* testsuite/libgomp.c/for-1.h: New.
+	* testsuite/libgomp.c/for-2.h: New.
+	* testsuite/libgomp.c/for-3.h: New.
+	* testsuite/libgomp.c/for-3.list: New.
+	* testsuite/libgomp.c/for-5.c: New.
+	* testsuite/libgomp.c/for-5.list: New.
+	* testsuite/libgomp.c/for-6.c: New.
+	* testsuite/libgomp.c/for-6.list: New.
+	* testsuite/libgomp.c/target-print-1.c: New.
+	* testsuite/libgomp.fortran/target-print-1.f90: New.
+	* testsuite/libgomp.oacc-c++/c++.exp (amdgcn*): Add support for AMD GCN.
+	* testsuite/libgomp.oacc-c-c++-common/atomic_capture-2.c: Adjust for
+	portability.
+	* testsuite/libgomp.oacc-c-c++-common/loop-auto-1.c: Skip unsuitable
+	test for AMD GCN.
+	* testsuite/libgomp.oacc-c-c++-common/loop-gwv-1.c: Adjust for
+	portability.
+	* testsuite/libgomp.oacc-c-c++-common/loop-v-1.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/loop-w-1.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/loop-wv-1.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/loop-red-gwv-1.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/loop-red-v-1.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/loop-red-v-2.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/loop-red-w-1.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/loop-red-w-2.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/loop-red-wv-1.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/routine-gwv-1.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/routine-v-1.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/routine-w-1.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/routine-wv-1.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/routine-wv-2.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/parallel-dims.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/serial-dims.c: Likewise.
+	* testsuite/libgomp.oacc-c-c++-common/private-variables-2.c: New.
+	* testsuite/libgomp.oacc-c-c++-common/tile-1.c: Skip for AMD GCN.
+	* testsuite/libgomp.oacc-c/c.exp (amdgcn*): Add support for AMD GCN.
+	* testsuite/libgomp.oacc-c/offload-target-1.c: Add AMD GCN support.
+	* testsuite/libgomp.oacc-c/print-1.c: New.
+	* testsuite/libgomp.oacc-fortran/fortran.exp (amdgcn*): Add AMD GCN
+	support.
+	* testsuite/libgomp.oacc-fortran/atomic_capture-1.f90: Adjust for
+	portability.
+	* testsuite/libgomp.oacc-fortran/collapse-1.f90: Likewise.
+	* testsuite/libgomp.oacc-fortran/collapse-2.f90: Likewise.
+	* testsuite/libgomp.oacc-fortran/error_stop-1.f: Support AMD GCN.
+	* testsuite/libgomp.oacc-fortran/error_stop-2.f: Support AMD GCN.
+	* testsuite/libgomp.oacc-fortran/error_stop-3.f: Support AMD GCN.
+	* testsuite/libgomp.oacc-fortran/print-1.f90: New.
+
 2019-01-23  Thomas Schwinge <thomas@codesourcery.com>
 
 	* testsuite/libgomp.oacc-c-c++-common/acc_prof-kernels-1.c: Update.
diff --git a/libgomp/Makefile.am b/libgomp/Makefile.am
index 00848cdc792..ba9b56c4cfa 100644
--- a/libgomp/Makefile.am
+++ b/libgomp/Makefile.am
@@ -72,7 +72,7 @@  libgomp_la_SOURCES = alloc.c atomic.c barrier.c critical.c env.c error.c \
 	proc.c sem.c bar.c ptrlock.c time.c fortran.c affinity.c target.c \
 	splay-tree.c libgomp-plugin.c oacc-parallel.c oacc-host.c oacc-init.c \
 	oacc-mem.c oacc-async.c oacc-plugin.c oacc-cuda.c priority_queue.c \
-	affinity-fmt.c teams.c oacc-profiling.c \
+	affinity-fmt.c teams.c gomp_print.c oacc-profiling.c \
 	oacc-profiling-acc_register_library.c
 
 include $(top_srcdir)/plugin/Makefrag.am
diff --git a/libgomp/Makefile.in b/libgomp/Makefile.in
index 63f919cb7ca..650437f2557 100644
--- a/libgomp/Makefile.in
+++ b/libgomp/Makefile.in
@@ -120,7 +120,8 @@  host_triplet = @host@
 target_triplet = @target@
 @PLUGIN_NVPTX_TRUE@am__append_1 = libgomp-plugin-nvptx.la
 @PLUGIN_HSA_TRUE@am__append_2 = libgomp-plugin-hsa.la
-@USE_FORTRAN_TRUE@am__append_3 = openacc.f90
+@PLUGIN_GCN_TRUE@am__append_3 = libgomp-plugin-gcn.la
+@USE_FORTRAN_TRUE@am__append_4 = openacc.f90
 subdir = .
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/../config/acx.m4 \
@@ -180,15 +181,26 @@  am__installdirs = "$(DESTDIR)$(toolexeclibdir)" "$(DESTDIR)$(infodir)" \
 	"$(DESTDIR)$(toolexeclibdir)"
 LTLIBRARIES = $(toolexeclib_LTLIBRARIES)
 am__DEPENDENCIES_1 =
+@PLUGIN_GCN_TRUE@libgomp_plugin_gcn_la_DEPENDENCIES = libgomp.la \
+@PLUGIN_GCN_TRUE@	$(am__DEPENDENCIES_1)
+@PLUGIN_GCN_TRUE@am_libgomp_plugin_gcn_la_OBJECTS =  \
+@PLUGIN_GCN_TRUE@	libgomp_plugin_gcn_la-plugin-gcn.lo
+libgomp_plugin_gcn_la_OBJECTS = $(am_libgomp_plugin_gcn_la_OBJECTS)
+AM_V_lt = $(am__v_lt_@AM_V@)
+am__v_lt_ = $(am__v_lt_@AM_DEFAULT_V@)
+am__v_lt_0 = --silent
+am__v_lt_1 = 
+libgomp_plugin_gcn_la_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CC \
+	$(libgomp_plugin_gcn_la_LIBTOOLFLAGS) $(LIBTOOLFLAGS) \
+	--mode=link $(CCLD) $(AM_CFLAGS) $(CFLAGS) \
+	$(libgomp_plugin_gcn_la_LDFLAGS) $(LDFLAGS) -o $@
+@PLUGIN_GCN_TRUE@am_libgomp_plugin_gcn_la_rpath = -rpath \
+@PLUGIN_GCN_TRUE@	$(toolexeclibdir)
 @PLUGIN_HSA_TRUE@libgomp_plugin_hsa_la_DEPENDENCIES = libgomp.la \
 @PLUGIN_HSA_TRUE@	$(am__DEPENDENCIES_1)
 @PLUGIN_HSA_TRUE@am_libgomp_plugin_hsa_la_OBJECTS =  \
 @PLUGIN_HSA_TRUE@	libgomp_plugin_hsa_la-plugin-hsa.lo
 libgomp_plugin_hsa_la_OBJECTS = $(am_libgomp_plugin_hsa_la_OBJECTS)
-AM_V_lt = $(am__v_lt_@AM_V@)
-am__v_lt_ = $(am__v_lt_@AM_DEFAULT_V@)
-am__v_lt_0 = --silent
-am__v_lt_1 = 
 libgomp_plugin_hsa_la_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CC \
 	$(libgomp_plugin_hsa_la_LIBTOOLFLAGS) $(LIBTOOLFLAGS) \
 	--mode=link $(CCLD) $(AM_CFLAGS) $(CFLAGS) \
@@ -216,7 +228,7 @@  am_libgomp_la_OBJECTS = alloc.lo atomic.lo barrier.lo critical.lo \
 	target.lo splay-tree.lo libgomp-plugin.lo oacc-parallel.lo \
 	oacc-host.lo oacc-init.lo oacc-mem.lo oacc-async.lo \
 	oacc-plugin.lo oacc-cuda.lo priority_queue.lo affinity-fmt.lo \
-	teams.lo oacc-profiling.lo \
+	teams.lo gomp_print.lo oacc-profiling.lo \
 	oacc-profiling-acc_register_library.lo $(am__objects_1)
 libgomp_la_OBJECTS = $(am_libgomp_la_OBJECTS)
 AM_V_P = $(am__v_P_@AM_V@)
@@ -265,7 +277,8 @@  AM_V_FCLD = $(am__v_FCLD_@AM_V@)
 am__v_FCLD_ = $(am__v_FCLD_@AM_DEFAULT_V@)
 am__v_FCLD_0 = @echo "  FCLD    " $@;
 am__v_FCLD_1 = 
-SOURCES = $(libgomp_plugin_hsa_la_SOURCES) \
+SOURCES = $(libgomp_plugin_gcn_la_SOURCES) \
+	$(libgomp_plugin_hsa_la_SOURCES) \
 	$(libgomp_plugin_nvptx_la_SOURCES) $(libgomp_la_SOURCES)
 AM_V_DVIPS = $(am__v_DVIPS_@AM_V@)
 am__v_DVIPS_ = $(am__v_DVIPS_@AM_DEFAULT_V@)
@@ -429,6 +442,10 @@  PACKAGE_URL = @PACKAGE_URL@
 PACKAGE_VERSION = @PACKAGE_VERSION@
 PATH_SEPARATOR = @PATH_SEPARATOR@
 PERL = @PERL@
+PLUGIN_GCN = @PLUGIN_GCN@
+PLUGIN_GCN_CPPFLAGS = @PLUGIN_GCN_CPPFLAGS@
+PLUGIN_GCN_LDFLAGS = @PLUGIN_GCN_LDFLAGS@
+PLUGIN_GCN_LIBS = @PLUGIN_GCN_LIBS@
 PLUGIN_HSA = @PLUGIN_HSA@
 PLUGIN_HSA_CPPFLAGS = @PLUGIN_HSA_CPPFLAGS@
 PLUGIN_HSA_LDFLAGS = @PLUGIN_HSA_LDFLAGS@
@@ -502,6 +519,7 @@  pdfdir = @pdfdir@
 prefix = @prefix@
 program_transform_name = @program_transform_name@
 psdir = @psdir@
+runstatedir = @runstatedir@
 sbindir = @sbindir@
 sharedstatedir = @sharedstatedir@
 srcdir = @srcdir@
@@ -529,7 +547,8 @@  libsubincludedir = $(libdir)/gcc/$(target_alias)/$(gcc_version)/include
 AM_CPPFLAGS = $(addprefix -I, $(search_path)) $(LIBFFIINCS)
 AM_CFLAGS = $(XCFLAGS)
 AM_LDFLAGS = $(XLDFLAGS) $(SECTION_LDFLAGS) $(OPT_LDFLAGS)
-toolexeclib_LTLIBRARIES = libgomp.la $(am__append_1) $(am__append_2)
+toolexeclib_LTLIBRARIES = libgomp.la $(am__append_1) $(am__append_2) \
+	$(am__append_3)
 nodist_toolexeclib_HEADERS = libgomp.spec
 
 # -Wc is only a libtool option.
@@ -554,8 +573,8 @@  libgomp_la_SOURCES = alloc.c atomic.c barrier.c critical.c env.c \
 	affinity.c target.c splay-tree.c libgomp-plugin.c \
 	oacc-parallel.c oacc-host.c oacc-init.c oacc-mem.c \
 	oacc-async.c oacc-plugin.c oacc-cuda.c priority_queue.c \
-	affinity-fmt.c teams.c oacc-profiling.c \
-	oacc-profiling-acc_register_library.c $(am__append_3)
+	affinity-fmt.c teams.c gomp_print.c oacc-profiling.c \
+	oacc-profiling-acc_register_library.c $(am__append_4)
 
 # Nvidia PTX OpenACC plugin.
 @PLUGIN_NVPTX_TRUE@libgomp_plugin_nvptx_version_info = -version-info $(libtool_VERSION)
@@ -578,6 +597,18 @@  libgomp_la_SOURCES = alloc.c atomic.c barrier.c critical.c env.c \
 @PLUGIN_HSA_TRUE@	$(lt_host_flags) $(PLUGIN_HSA_LDFLAGS)
 @PLUGIN_HSA_TRUE@libgomp_plugin_hsa_la_LIBADD = libgomp.la $(PLUGIN_HSA_LIBS)
 @PLUGIN_HSA_TRUE@libgomp_plugin_hsa_la_LIBTOOLFLAGS = --tag=disable-static
+
+# AMD GCN plugin
+@PLUGIN_GCN_TRUE@libgomp_plugin_gcn_version_info = -version-info $(libtool_VERSION)
+@PLUGIN_GCN_TRUE@libgomp_plugin_gcn_la_SOURCES = plugin/plugin-gcn.c
+@PLUGIN_GCN_TRUE@libgomp_plugin_gcn_la_CPPFLAGS = $(AM_CPPFLAGS) $(PLUGIN_GCN_CPPFLAGS) \
+@PLUGIN_GCN_TRUE@	-D_GNU_SOURCE
+
+@PLUGIN_GCN_TRUE@libgomp_plugin_gcn_la_LDFLAGS =  \
+@PLUGIN_GCN_TRUE@	$(libgomp_plugin_gcn_version_info) \
+@PLUGIN_GCN_TRUE@	$(lt_host_flags) $(PLUGIN_GCN_LDFLAGS)
+@PLUGIN_GCN_TRUE@libgomp_plugin_gcn_la_LIBADD = libgomp.la $(PLUGIN_GCN_LIBS)
+@PLUGIN_GCN_TRUE@libgomp_plugin_gcn_la_LIBTOOLFLAGS = --tag=disable-static
 nodist_noinst_HEADERS = libgomp_f.h
 nodist_libsubinclude_HEADERS = omp.h openacc.h acc_prof.h
 @USE_FORTRAN_TRUE@nodist_finclude_HEADERS = omp_lib.h omp_lib.f90 omp_lib.mod omp_lib_kinds.mod \
@@ -714,6 +745,9 @@  clean-toolexeclibLTLIBRARIES:
 	  rm -f $${locs}; \
 	}
 
+libgomp-plugin-gcn.la: $(libgomp_plugin_gcn_la_OBJECTS) $(libgomp_plugin_gcn_la_DEPENDENCIES) $(EXTRA_libgomp_plugin_gcn_la_DEPENDENCIES) 
+	$(AM_V_CCLD)$(libgomp_plugin_gcn_la_LINK) $(am_libgomp_plugin_gcn_la_rpath) $(libgomp_plugin_gcn_la_OBJECTS) $(libgomp_plugin_gcn_la_LIBADD) $(LIBS)
+
 libgomp-plugin-hsa.la: $(libgomp_plugin_hsa_la_OBJECTS) $(libgomp_plugin_hsa_la_DEPENDENCIES) $(EXTRA_libgomp_plugin_hsa_la_DEPENDENCIES) 
 	$(AM_V_CCLD)$(libgomp_plugin_hsa_la_LINK) $(am_libgomp_plugin_hsa_la_rpath) $(libgomp_plugin_hsa_la_OBJECTS) $(libgomp_plugin_hsa_la_LIBADD) $(LIBS)
 
@@ -739,11 +773,13 @@  distclean-compile:
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/env.Plo@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/error.Plo@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/fortran.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/gomp_print.Plo@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/icv-device.Plo@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/icv.Plo@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/iter.Plo@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/iter_ull.Plo@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libgomp-plugin.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libgomp_plugin_gcn_la-plugin-gcn.Plo@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libgomp_plugin_hsa_la-plugin-hsa.Plo@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libgomp_plugin_nvptx_la-plugin-nvptx.Plo@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/lock.Plo@am__quote@
@@ -796,6 +832,13 @@  distclean-compile:
 @AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
 @am__fastdepCC_FALSE@	$(AM_V_CC@am__nodep@)$(LTCOMPILE) -c -o $@ $<
 
+libgomp_plugin_gcn_la-plugin-gcn.lo: plugin/plugin-gcn.c
+@am__fastdepCC_TRUE@	$(AM_V_CC)$(LIBTOOL) $(AM_V_lt) --tag=CC $(libgomp_plugin_gcn_la_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libgomp_plugin_gcn_la_CPPFLAGS) $(CPPFLAGS) $(AM_CFLAGS) $(CFLAGS) -MT libgomp_plugin_gcn_la-plugin-gcn.lo -MD -MP -MF $(DEPDIR)/libgomp_plugin_gcn_la-plugin-gcn.Tpo -c -o libgomp_plugin_gcn_la-plugin-gcn.lo `test -f 'plugin/plugin-gcn.c' || echo '$(srcdir)/'`plugin/plugin-gcn.c
+@am__fastdepCC_TRUE@	$(AM_V_at)$(am__mv) $(DEPDIR)/libgomp_plugin_gcn_la-plugin-gcn.Tpo $(DEPDIR)/libgomp_plugin_gcn_la-plugin-gcn.Plo
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	$(AM_V_CC)source='plugin/plugin-gcn.c' object='libgomp_plugin_gcn_la-plugin-gcn.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(AM_V_CC@am__nodep@)$(LIBTOOL) $(AM_V_lt) --tag=CC $(libgomp_plugin_gcn_la_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libgomp_plugin_gcn_la_CPPFLAGS) $(CPPFLAGS) $(AM_CFLAGS) $(CFLAGS) -c -o libgomp_plugin_gcn_la-plugin-gcn.lo `test -f 'plugin/plugin-gcn.c' || echo '$(srcdir)/'`plugin/plugin-gcn.c
+
 libgomp_plugin_hsa_la-plugin-hsa.lo: plugin/plugin-hsa.c
 @am__fastdepCC_TRUE@	$(AM_V_CC)$(LIBTOOL) $(AM_V_lt) --tag=CC $(libgomp_plugin_hsa_la_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(libgomp_plugin_hsa_la_CPPFLAGS) $(CPPFLAGS) $(AM_CFLAGS) $(CFLAGS) -MT libgomp_plugin_hsa_la-plugin-hsa.lo -MD -MP -MF $(DEPDIR)/libgomp_plugin_hsa_la-plugin-hsa.Tpo -c -o libgomp_plugin_hsa_la-plugin-hsa.lo `test -f 'plugin/plugin-hsa.c' || echo '$(srcdir)/'`plugin/plugin-hsa.c
 @am__fastdepCC_TRUE@	$(AM_V_at)$(am__mv) $(DEPDIR)/libgomp_plugin_hsa_la-plugin-hsa.Tpo $(DEPDIR)/libgomp_plugin_hsa_la-plugin-hsa.Plo
diff --git a/libgomp/affinity-fmt.c b/libgomp/affinity-fmt.c
index d9c6e181233..cd5783f4658 100644
--- a/libgomp/affinity-fmt.c
+++ b/libgomp/affinity-fmt.c
@@ -38,7 +38,7 @@ 
 #endif
 
 void
-gomp_print_string (const char *str, size_t len)
+gomp_write_string (const char *str, size_t len)
 {
   fwrite (str, 1, len, stderr);
 }
@@ -462,13 +462,13 @@  omp_display_affinity (const char *format)
   if (ret < sizeof buf)
     {
       buf[ret] = '\n';
-      gomp_print_string (buf, ret + 1);
+      gomp_write_string (buf, ret + 1);
       return;
     }
   b = gomp_malloc (ret + 1);
   ialias_call (omp_capture_affinity) (b, ret + 1, format);
   b[ret] = '\n';
-  gomp_print_string (b, ret + 1);
+  gomp_write_string (b, ret + 1);
   free (b);
 }
 
@@ -483,13 +483,13 @@  gomp_display_affinity_thread (gomp_thread_handle handle,
   if (ret < sizeof buf)
     {
       buf[ret] = '\n';
-      gomp_print_string (buf, ret + 1);
+      gomp_write_string (buf, ret + 1);
       return;
     }
   b = gomp_malloc (ret + 1);
   gomp_display_affinity (b, ret + 1, gomp_affinity_format_var,
   			 handle, ts, place);
   b[ret] = '\n';
-  gomp_print_string (b, ret + 1);
+  gomp_write_string (b, ret + 1);
   free (b);
 }
diff --git a/libgomp/config.h.in b/libgomp/config.h.in
index 5ad5f532af1..778a59eb1b7 100644
--- a/libgomp/config.h.in
+++ b/libgomp/config.h.in
@@ -170,6 +170,9 @@ 
 /* Define to the version of this package. */
 #undef PACKAGE_VERSION
 
+/* Define to 1 if the GCN plugin is built, 0 if not. */
+#undef PLUGIN_GCN
+
 /* Define to 1 if the HSA plugin is built, 0 if not. */
 #undef PLUGIN_HSA
 
diff --git a/libgomp/config/nvptx/libgomp-plugin.c b/libgomp/config/accel/libgomp-plugin.c
similarity index 100%
rename from libgomp/config/nvptx/libgomp-plugin.c
rename to libgomp/config/accel/libgomp-plugin.c
diff --git a/libgomp/config/nvptx/lock.c b/libgomp/config/accel/lock.c
similarity index 100%
rename from libgomp/config/nvptx/lock.c
rename to libgomp/config/accel/lock.c
diff --git a/libgomp/config/nvptx/mutex.c b/libgomp/config/accel/mutex.c
similarity index 100%
rename from libgomp/config/nvptx/mutex.c
rename to libgomp/config/accel/mutex.c
diff --git a/libgomp/config/nvptx/mutex.h b/libgomp/config/accel/mutex.h
similarity index 100%
rename from libgomp/config/nvptx/mutex.h
rename to libgomp/config/accel/mutex.h
diff --git a/libgomp/config/nvptx/oacc-async.c b/libgomp/config/accel/oacc-async.c
similarity index 100%
rename from libgomp/config/nvptx/oacc-async.c
rename to libgomp/config/accel/oacc-async.c
diff --git a/libgomp/config/nvptx/oacc-cuda.c b/libgomp/config/accel/oacc-cuda.c
similarity index 100%
rename from libgomp/config/nvptx/oacc-cuda.c
rename to libgomp/config/accel/oacc-cuda.c
diff --git a/libgomp/config/nvptx/oacc-host.c b/libgomp/config/accel/oacc-host.c
similarity index 100%
rename from libgomp/config/nvptx/oacc-host.c
rename to libgomp/config/accel/oacc-host.c
diff --git a/libgomp/config/nvptx/oacc-init.c b/libgomp/config/accel/oacc-init.c
similarity index 100%
rename from libgomp/config/nvptx/oacc-init.c
rename to libgomp/config/accel/oacc-init.c
diff --git a/libgomp/config/nvptx/oacc-mem.c b/libgomp/config/accel/oacc-mem.c
similarity index 100%
rename from libgomp/config/nvptx/oacc-mem.c
rename to libgomp/config/accel/oacc-mem.c
diff --git a/libgomp/config/nvptx/oacc-plugin.c b/libgomp/config/accel/oacc-plugin.c
similarity index 100%
rename from libgomp/config/nvptx/oacc-plugin.c
rename to libgomp/config/accel/oacc-plugin.c
diff --git a/libgomp/config/nvptx/omp-lock.h b/libgomp/config/accel/omp-lock.h
similarity index 100%
rename from libgomp/config/nvptx/omp-lock.h
rename to libgomp/config/accel/omp-lock.h
diff --git a/libgomp/config/nvptx/openacc.f90 b/libgomp/config/accel/openacc.f90
similarity index 96%
rename from libgomp/config/nvptx/openacc.f90
rename to libgomp/config/accel/openacc.f90
index a7f690e1572..c2331d893ba 100644
--- a/libgomp/config/nvptx/openacc.f90
+++ b/libgomp/config/accel/openacc.f90
@@ -51,6 +51,8 @@  module openacc_kinds
   ! integer (acc_device_kind), parameter :: acc_device_host_nonshm = 3 removed.
   integer (acc_device_kind), parameter :: acc_device_not_host = 4
   integer (acc_device_kind), parameter :: acc_device_nvidia = 5
+  integer (acc_device_kind), parameter :: acc_device_hsa = 6
+  integer (acc_device_kind), parameter :: acc_device_gcn = 7
 
 end module
 
diff --git a/libgomp/config/nvptx/pool.h b/libgomp/config/accel/pool.h
similarity index 100%
rename from libgomp/config/nvptx/pool.h
rename to libgomp/config/accel/pool.h
diff --git a/libgomp/config/nvptx/proc.c b/libgomp/config/accel/proc.c
similarity index 98%
rename from libgomp/config/nvptx/proc.c
rename to libgomp/config/accel/proc.c
index 8ca0b0a1ee1..fb2d4830df7 100644
--- a/libgomp/config/nvptx/proc.c
+++ b/libgomp/config/accel/proc.c
@@ -39,3 +39,4 @@  omp_get_num_procs (void)
 {
   return gomp_icv (false)->nthreads_var;
 }
+ialias (omp_get_num_procs)
diff --git a/libgomp/config/nvptx/ptrlock.c b/libgomp/config/accel/ptrlock.c
similarity index 100%
rename from libgomp/config/nvptx/ptrlock.c
rename to libgomp/config/accel/ptrlock.c
diff --git a/libgomp/config/nvptx/ptrlock.h b/libgomp/config/accel/ptrlock.h
similarity index 100%
rename from libgomp/config/nvptx/ptrlock.h
rename to libgomp/config/accel/ptrlock.h
diff --git a/libgomp/config/nvptx/sem.c b/libgomp/config/accel/sem.c
similarity index 100%
rename from libgomp/config/nvptx/sem.c
rename to libgomp/config/accel/sem.c
diff --git a/libgomp/config/nvptx/sem.h b/libgomp/config/accel/sem.h
similarity index 100%
rename from libgomp/config/nvptx/sem.h
rename to libgomp/config/accel/sem.h
diff --git a/libgomp/config/nvptx/thread-stacksize.h b/libgomp/config/accel/thread-stacksize.h
similarity index 100%
rename from libgomp/config/nvptx/thread-stacksize.h
rename to libgomp/config/accel/thread-stacksize.h
diff --git a/libgomp/config/gcn/affinity-fmt.c b/libgomp/config/gcn/affinity-fmt.c
new file mode 100644
index 00000000000..3585f414460
--- /dev/null
+++ b/libgomp/config/gcn/affinity-fmt.c
@@ -0,0 +1,51 @@ 
+/* Copyright (C) 2018-2019 Free Software Foundation, Inc.
+
+   This file is part of the GNU Offloading and Multi Processing Library
+   (libgomp).
+
+   Libgomp 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.
+
+   Libgomp 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/>.  */
+
+#include "libgomp.h"
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
+#ifdef HAVE_UNISTD_H
+#include <unistd.h>
+#endif
+#ifdef HAVE_INTTYPES_H
+# include <inttypes.h>  /* For PRIx64.  */
+#endif
+#ifdef HAVE_UNAME
+#include <sys/utsname.h>
+#endif
+
+/* The HAVE_GETPID and HAVE_GETHOSTNAME configure tests are passing for nvptx,
+   while the nvptx newlib implementation does not support those functions.
+   Override the configure test results here.  */
+#undef HAVE_GETPID
+#undef HAVE_GETHOSTNAME
+
+/* The GCN newlib implementation does not support fwrite, but it does support
+   write.  Map fwrite to write.  */
+#undef fwrite
+#define fwrite(ptr, size, nmemb, stream) write (1, (ptr), (nmemb) * (size))
+
+#include "../../affinity-fmt.c"
+
diff --git a/libgomp/config/gcn/bar.c b/libgomp/config/gcn/bar.c
new file mode 100644
index 00000000000..592dacd26f2
--- /dev/null
+++ b/libgomp/config/gcn/bar.c
@@ -0,0 +1,230 @@ 
+/* Copyright (C) 2015-2019 Free Software Foundation, Inc.
+   Contributed by Mentor Embedded.
+
+   This file is part of the GNU Offloading and Multi Processing Library
+   (libgomp).
+
+   Libgomp 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.
+
+   Libgomp 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/>.  */
+
+/* This is an AMD GCN specific implementation of a barrier synchronization
+   mechanism for libgomp.  This type is private to the library.  This
+   implementation uses atomic instructions and s_barrier instruction.  It
+   uses MEMMODEL_RELAXED here because barriers are within workgroups and
+   therefore don't need to flush caches.  */
+
+#include <limits.h>
+#include "libgomp.h"
+
+
+void
+gomp_barrier_wait_end (gomp_barrier_t *bar, gomp_barrier_state_t state)
+{
+  if (__builtin_expect (state & BAR_WAS_LAST, 0))
+    {
+      /* Next time we'll be awaiting TOTAL threads again.  */
+      bar->awaited = bar->total;
+      __atomic_store_n (&bar->generation, bar->generation + BAR_INCR,
+			MEMMODEL_RELAXED);
+    }
+  asm ("s_barrier" ::: "memory");
+}
+
+void
+gomp_barrier_wait (gomp_barrier_t *bar)
+{
+  gomp_barrier_wait_end (bar, gomp_barrier_wait_start (bar));
+}
+
+/* Like gomp_barrier_wait, except that if the encountering thread
+   is not the last one to hit the barrier, it returns immediately.
+   The intended usage is that a thread which intends to gomp_barrier_destroy
+   this barrier calls gomp_barrier_wait, while all other threads
+   call gomp_barrier_wait_last.  When gomp_barrier_wait returns,
+   the barrier can be safely destroyed.  */
+
+void
+gomp_barrier_wait_last (gomp_barrier_t *bar)
+{
+  /* Deferring to gomp_barrier_wait does not use the optimization opportunity
+     allowed by the interface contract for all-but-last participants.  The
+     original implementation in config/linux/bar.c handles this better.  */
+  gomp_barrier_wait (bar);
+}
+
+void
+gomp_team_barrier_wake (gomp_barrier_t *bar, int count)
+{
+  asm ("s_barrier" ::: "memory");
+}
+
+void
+gomp_team_barrier_wait_end (gomp_barrier_t *bar, gomp_barrier_state_t state)
+{
+  unsigned int generation, gen;
+
+  if (__builtin_expect (state & BAR_WAS_LAST, 0))
+    {
+      /* Next time we'll be awaiting TOTAL threads again.  */
+      struct gomp_thread *thr = gomp_thread ();
+      struct gomp_team *team = thr->ts.team;
+
+      bar->awaited = bar->total;
+      team->work_share_cancelled = 0;
+      if (__builtin_expect (team->task_count, 0))
+	{
+	  gomp_barrier_handle_tasks (state);
+	  state &= ~BAR_WAS_LAST;
+	}
+      else
+	{
+	  state &= ~BAR_CANCELLED;
+	  state += BAR_INCR - BAR_WAS_LAST;
+	  __atomic_store_n (&bar->generation, state, MEMMODEL_RELAXED);
+	  asm ("s_barrier" ::: "memory");
+	  return;
+	}
+    }
+
+  generation = state;
+  state &= ~BAR_CANCELLED;
+  int retry = 100;
+  do
+    {
+      if (retry-- == 0)
+	{
+	  /* It really shouldn't happen that barriers get out of sync, but
+	     if they do then this will loop until they realign, so we need
+	     to avoid an infinite loop where the thread just isn't there.  */
+	  gomp_print_string ("Barrier sync failed (another thread died?);",
+			     " aborting.");
+	  abort();
+	}
+
+      asm ("s_barrier" ::: "memory");
+      gen = __atomic_load_n (&bar->generation, MEMMODEL_ACQUIRE);
+      if (__builtin_expect (gen & BAR_TASK_PENDING, 0))
+	{
+	  gomp_barrier_handle_tasks (state);
+	  gen = __atomic_load_n (&bar->generation, MEMMODEL_ACQUIRE);
+	}
+      generation |= gen & BAR_WAITING_FOR_TASK;
+    }
+  while (gen != state + BAR_INCR);
+}
+
+void
+gomp_team_barrier_wait (gomp_barrier_t *bar)
+{
+  gomp_team_barrier_wait_end (bar, gomp_barrier_wait_start (bar));
+}
+
+void
+gomp_team_barrier_wait_final (gomp_barrier_t *bar)
+{
+  gomp_barrier_state_t state = gomp_barrier_wait_final_start (bar);
+  if (__builtin_expect (state & BAR_WAS_LAST, 0))
+    bar->awaited_final = bar->total;
+  gomp_team_barrier_wait_end (bar, state);
+}
+
+bool
+gomp_team_barrier_wait_cancel_end (gomp_barrier_t *bar,
+				   gomp_barrier_state_t state)
+{
+  unsigned int generation, gen;
+
+  if (__builtin_expect (state & BAR_WAS_LAST, 0))
+    {
+      /* Next time we'll be awaiting TOTAL threads again.  */
+      /* BAR_CANCELLED should never be set in state here, because
+	 cancellation means that at least one of the threads has been
+	 cancelled, thus on a cancellable barrier we should never see
+	 all threads to arrive.  */
+      struct gomp_thread *thr = gomp_thread ();
+      struct gomp_team *team = thr->ts.team;
+
+      bar->awaited = bar->total;
+      team->work_share_cancelled = 0;
+      if (__builtin_expect (team->task_count, 0))
+	{
+	  gomp_barrier_handle_tasks (state);
+	  state &= ~BAR_WAS_LAST;
+	}
+      else
+	{
+	  state += BAR_INCR - BAR_WAS_LAST;
+	  __atomic_store_n (&bar->generation, state, MEMMODEL_RELAXED);
+	  asm ("s_barrier" ::: "memory");
+	  return false;
+	}
+    }
+
+  if (__builtin_expect (state & BAR_CANCELLED, 0))
+    return true;
+
+  generation = state;
+  int retry = 100;
+  do
+    {
+      if (retry-- == 0)
+	{
+	  /* It really shouldn't happen that barriers get out of sync, but
+	     if they do then this will loop until they realign, so we need
+	     to avoid an infinite loop where the thread just isn't there.  */
+	  gomp_print_string ("Barrier sync failed (another thread died?);",
+			     " aborting.");
+	  abort();
+	}
+
+      asm ("s_barrier" ::: "memory");
+      gen = __atomic_load_n (&bar->generation, MEMMODEL_RELAXED);
+      if (__builtin_expect (gen & BAR_CANCELLED, 0))
+	return true;
+      if (__builtin_expect (gen & BAR_TASK_PENDING, 0))
+	{
+	  gomp_barrier_handle_tasks (state);
+	  gen = __atomic_load_n (&bar->generation, MEMMODEL_RELAXED);
+	}
+      generation |= gen & BAR_WAITING_FOR_TASK;
+    }
+  while (gen != state + BAR_INCR);
+
+  return false;
+}
+
+bool
+gomp_team_barrier_wait_cancel (gomp_barrier_t *bar)
+{
+  return gomp_team_barrier_wait_cancel_end (bar, gomp_barrier_wait_start (bar));
+}
+
+void
+gomp_team_barrier_cancel (struct gomp_team *team)
+{
+  gomp_mutex_lock (&team->task_lock);
+  if (team->barrier.generation & BAR_CANCELLED)
+    {
+      gomp_mutex_unlock (&team->task_lock);
+      return;
+    }
+  team->barrier.generation |= BAR_CANCELLED;
+  gomp_mutex_unlock (&team->task_lock);
+  gomp_team_barrier_wake (&team->barrier, INT_MAX);
+}
diff --git a/libgomp/config/gcn/bar.h b/libgomp/config/gcn/bar.h
new file mode 100644
index 00000000000..ec8851ba078
--- /dev/null
+++ b/libgomp/config/gcn/bar.h
@@ -0,0 +1,168 @@ 
+/* Copyright (C) 2015-2019 Free Software Foundation, Inc.
+   Contributed by Mentor Embedded.
+
+   This file is part of the GNU Offloading and Multi Processing Library
+   (libgomp).
+
+   Libgomp 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.
+
+   Libgomp 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/>.  */
+
+/* This is an AMD GCN specific implementation of a barrier synchronization
+   mechanism for libgomp.  This type is private to the library.  This
+   implementation uses atomic instructions and s_barrier instruction.  It
+   uses MEMMODEL_RELAXED here because barriers are within workgroups and
+   therefore don't need to flush caches.  */
+
+#ifndef GOMP_BARRIER_H
+#define GOMP_BARRIER_H 1
+
+#include "mutex.h"
+
+typedef struct
+{
+  unsigned total;
+  unsigned generation;
+  unsigned awaited;
+  unsigned awaited_final;
+} gomp_barrier_t;
+
+typedef unsigned int gomp_barrier_state_t;
+
+/* The generation field contains a counter in the high bits, with a few
+   low bits dedicated to flags.  Note that TASK_PENDING and WAS_LAST can
+   share space because WAS_LAST is never stored back to generation.  */
+#define BAR_TASK_PENDING	1
+#define BAR_WAS_LAST		1
+#define BAR_WAITING_FOR_TASK	2
+#define BAR_CANCELLED		4
+#define BAR_INCR		8
+
+static inline void gomp_barrier_init (gomp_barrier_t *bar, unsigned count)
+{
+  bar->total = count;
+  bar->awaited = count;
+  bar->awaited_final = count;
+  bar->generation = 0;
+}
+
+static inline void gomp_barrier_reinit (gomp_barrier_t *bar, unsigned count)
+{
+  __atomic_add_fetch (&bar->awaited, count - bar->total, MEMMODEL_RELAXED);
+  bar->total = count;
+}
+
+static inline void gomp_barrier_destroy (gomp_barrier_t *bar)
+{
+}
+
+extern void gomp_barrier_wait (gomp_barrier_t *);
+extern void gomp_barrier_wait_last (gomp_barrier_t *);
+extern void gomp_barrier_wait_end (gomp_barrier_t *, gomp_barrier_state_t);
+extern void gomp_team_barrier_wait (gomp_barrier_t *);
+extern void gomp_team_barrier_wait_final (gomp_barrier_t *);
+extern void gomp_team_barrier_wait_end (gomp_barrier_t *,
+					gomp_barrier_state_t);
+extern bool gomp_team_barrier_wait_cancel (gomp_barrier_t *);
+extern bool gomp_team_barrier_wait_cancel_end (gomp_barrier_t *,
+					       gomp_barrier_state_t);
+extern void gomp_team_barrier_wake (gomp_barrier_t *, int);
+struct gomp_team;
+extern void gomp_team_barrier_cancel (struct gomp_team *);
+
+static inline gomp_barrier_state_t
+gomp_barrier_wait_start (gomp_barrier_t *bar)
+{
+  unsigned int ret = __atomic_load_n (&bar->generation, MEMMODEL_RELAXED);
+  ret &= -BAR_INCR | BAR_CANCELLED;
+  /* A memory barrier is needed before exiting from the various forms
+     of gomp_barrier_wait, to satisfy OpenMP API version 3.1 section
+     2.8.6 flush Construct, which says there is an implicit flush during
+     a barrier region.  This is a convenient place to add the barrier,
+     so we use MEMMODEL_ACQ_REL here rather than MEMMODEL_ACQUIRE.  */
+  if (__atomic_add_fetch (&bar->awaited, -1, MEMMODEL_RELAXED) == 0)
+    ret |= BAR_WAS_LAST;
+  return ret;
+}
+
+static inline gomp_barrier_state_t
+gomp_barrier_wait_cancel_start (gomp_barrier_t *bar)
+{
+  return gomp_barrier_wait_start (bar);
+}
+
+/* This is like gomp_barrier_wait_start, except it decrements
+   bar->awaited_final rather than bar->awaited and should be used
+   for the gomp_team_end barrier only.  */
+static inline gomp_barrier_state_t
+gomp_barrier_wait_final_start (gomp_barrier_t *bar)
+{
+  unsigned int ret = __atomic_load_n (&bar->generation, MEMMODEL_RELAXED);
+  ret &= -BAR_INCR | BAR_CANCELLED;
+  /* See above gomp_barrier_wait_start comment.  */
+  if (__atomic_add_fetch (&bar->awaited_final, -1, MEMMODEL_RELAXED) == 0)
+    ret |= BAR_WAS_LAST;
+  return ret;
+}
+
+static inline bool
+gomp_barrier_last_thread (gomp_barrier_state_t state)
+{
+  return state & BAR_WAS_LAST;
+}
+
+/* All the inlines below must be called with team->task_lock
+   held.  */
+
+static inline void
+gomp_team_barrier_set_task_pending (gomp_barrier_t *bar)
+{
+  bar->generation |= BAR_TASK_PENDING;
+}
+
+static inline void
+gomp_team_barrier_clear_task_pending (gomp_barrier_t *bar)
+{
+  bar->generation &= ~BAR_TASK_PENDING;
+}
+
+static inline void
+gomp_team_barrier_set_waiting_for_tasks (gomp_barrier_t *bar)
+{
+  bar->generation |= BAR_WAITING_FOR_TASK;
+}
+
+static inline bool
+gomp_team_barrier_waiting_for_tasks (gomp_barrier_t *bar)
+{
+  return (bar->generation & BAR_WAITING_FOR_TASK) != 0;
+}
+
+static inline bool
+gomp_team_barrier_cancelled (gomp_barrier_t *bar)
+{
+  return __builtin_expect ((bar->generation & BAR_CANCELLED) != 0, 0);
+}
+
+static inline void
+gomp_team_barrier_done (gomp_barrier_t *bar, gomp_barrier_state_t state)
+{
+  bar->generation = (state & -BAR_INCR) + BAR_INCR;
+}
+
+#endif /* GOMP_BARRIER_H */
diff --git a/libgomp/config/gcn/doacross.h b/libgomp/config/gcn/doacross.h
new file mode 100644
index 00000000000..2bff18ae1a8
--- /dev/null
+++ b/libgomp/config/gcn/doacross.h
@@ -0,0 +1,58 @@ 
+/* Copyright (C) 2015-2019 Free Software Foundation, Inc.
+   Contributed by Mentor Embedded.
+
+   This file is part of the GNU Offloading and Multi Processing Library
+   (libgomp).
+
+   Libgomp 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.
+
+   Libgomp 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/>.  */
+
+/* This is the AMD GCN implementation of doacross spinning.  */
+
+#ifndef GOMP_DOACROSS_H
+#define GOMP_DOACROSS_H 1
+
+#include "libgomp.h"
+
+static inline int
+cpu_relax (void)
+{
+  /* This can be implemented as just a memory barrier, but a sleep seems
+     like it should allow the wavefront to yield (maybe?)
+     Use the shortest possible sleep time of 1*64 cycles.  */
+  asm volatile ("s_sleep\t1" ::: "memory");
+  return 0;
+}
+
+static inline void doacross_spin (unsigned long *addr, unsigned long expected,
+				  unsigned long cur)
+{
+  /* Prevent compiler from optimizing based on bounds of containing object.  */
+  asm ("" : "+r" (addr));
+  do
+    {
+       /* An alternative implementation might use s_setprio to lower the
+	  priority temporarily, and then restore it after.  */
+      int i = cpu_relax ();
+      cur = addr[i];
+    }
+  while (cur <= expected);
+}
+
+#endif /* GOMP_DOACROSS_H */
diff --git a/libgomp/config/gcn/gomp_print.c b/libgomp/config/gcn/gomp_print.c
new file mode 100644
index 00000000000..1c755730f08
--- /dev/null
+++ b/libgomp/config/gcn/gomp_print.c
@@ -0,0 +1,2 @@ 
+/* The GCN gomp_print routines live in libgcc where they are available
+   to stand-alone toolchains configured without libgomp.  */
diff --git a/libgomp/config/gcn/icv-device.c b/libgomp/config/gcn/icv-device.c
new file mode 100644
index 00000000000..cbb9dfa1133
--- /dev/null
+++ b/libgomp/config/gcn/icv-device.c
@@ -0,0 +1,72 @@ 
+/* Copyright (C) 2015-2019 Free Software Foundation, Inc.
+   Contributed by Mentor Embedded.
+
+   This file is part of the GNU Offloading and Multi Processing Library
+   (libgomp).
+
+   Libgomp 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.
+
+   Libgomp 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/>.  */
+
+/* This file defines OpenMP API entry points that accelerator targets are
+   expected to replace.  */
+
+#include "libgomp.h"
+
+void
+omp_set_default_device (int device_num __attribute__((unused)))
+{
+}
+
+int
+omp_get_default_device (void)
+{
+  return 0;
+}
+
+int
+omp_get_num_devices (void)
+{
+  return 0;
+}
+
+int
+omp_get_num_teams (void)
+{
+  return gomp_num_teams_var + 1;
+}
+
+int __attribute__ ((__optimize__ ("O2")))
+omp_get_team_num (void)
+{
+  return __builtin_gcn_dim_pos (0);
+}
+
+int
+omp_is_initial_device (void)
+{
+  /* AMD GCN is an accelerator-only target.  */
+  return 0;
+}
+
+ialias (omp_set_default_device)
+ialias (omp_get_default_device)
+ialias (omp_get_num_devices)
+ialias (omp_get_num_teams)
+ialias (omp_get_team_num)
+ialias (omp_is_initial_device)
diff --git a/libgomp/config/gcn/simple-bar.h b/libgomp/config/gcn/simple-bar.h
new file mode 100644
index 00000000000..802e0f5c301
--- /dev/null
+++ b/libgomp/config/gcn/simple-bar.h
@@ -0,0 +1,61 @@ 
+/* Copyright (C) 2015-2019 Free Software Foundation, Inc.
+   Contributed by Mentor Embedded.
+
+   This file is part of the GNU Offloading and Multi Processing Library
+   (libgomp).
+
+   Libgomp 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.
+
+   Libgomp 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/>.  */
+
+/* This is a simplified barrier that is suitable for thread pool
+   synchronizaton.  Only a subset of full barrier API (bar.h) is exposed.
+   Here in the AMD GCN-specific implementation, we expect that thread pool
+   corresponds to the wavefronts within a work group.  */
+
+#ifndef GOMP_SIMPLE_BARRIER_H
+#define GOMP_SIMPLE_BARRIER_H 1
+
+/* AMD GCN has no use for this type.  */
+typedef int gomp_simple_barrier_t;
+
+/* GCN barriers block all wavefronts, so the count is not interesting.  */
+static inline void
+gomp_simple_barrier_init (gomp_simple_barrier_t *bar, unsigned count)
+{
+}
+
+static inline void
+gomp_simple_barrier_destroy (gomp_simple_barrier_t *bar)
+{
+}
+
+static inline void
+gomp_simple_barrier_wait (gomp_simple_barrier_t *bar)
+{
+  asm volatile ("s_barrier" ::: "memory");
+}
+
+static inline void
+gomp_simple_barrier_wait_last (gomp_simple_barrier_t *bar)
+{
+  /* GCN has no way to signal a barrier without waiting.  */
+  asm volatile ("s_barrier" ::: "memory");
+}
+
+#endif /* GOMP_SIMPLE_BARRIER_H */
diff --git a/libgomp/config/gcn/target.c b/libgomp/config/gcn/target.c
new file mode 100644
index 00000000000..5ec57778772
--- /dev/null
+++ b/libgomp/config/gcn/target.c
@@ -0,0 +1,49 @@ 
+/* Copyright (C) 2017-2019 Free Software Foundation, Inc.
+   Contributed by Mentor Embedded.
+
+   This file is part of the GNU Offloading and Multi Processing Library
+   (libgomp).
+
+   Libgomp 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.
+
+   Libgomp 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/>.  */
+
+#include "libgomp.h"
+#include <limits.h>
+
+void
+GOMP_teams (unsigned int num_teams, unsigned int thread_limit)
+{
+  if (thread_limit)
+    {
+      struct gomp_task_icv *icv = gomp_icv (true);
+      icv->thread_limit_var
+	= thread_limit > INT_MAX ? UINT_MAX : thread_limit;
+    }
+  unsigned int num_workgroups, workgroup_id;
+  num_workgroups = __builtin_gcn_dim_size (0);
+  workgroup_id = __builtin_gcn_dim_pos (0);
+  if (!num_teams || num_teams >= num_workgroups)
+    num_teams = num_workgroups;
+  else if (workgroup_id >= num_teams)
+    {
+      gomp_free_thread (gcn_thrs ());
+      exit (0);
+    }
+  gomp_num_teams_var = num_teams - 1;
+}
diff --git a/libgomp/config/gcn/task.c b/libgomp/config/gcn/task.c
new file mode 100644
index 00000000000..a13565034b6
--- /dev/null
+++ b/libgomp/config/gcn/task.c
@@ -0,0 +1,39 @@ 
+/* Copyright (C) 2017-2019 Free Software Foundation, Inc.
+   Contributed by Mentor Embedded.
+
+   This file is part of the GNU Offloading and Multi Processing Library
+   (libgomp).
+
+   Libgomp 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.
+
+   Libgomp 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/>.  */
+
+/* This file handles the maintainence of tasks in response to task
+   creation and termination.  */
+
+#include "libgomp.h"
+
+/* AMD GCN is an accelerator-only target, so this should never be called.  */
+
+bool
+gomp_target_task_fn (void *data)
+{
+  __builtin_unreachable ();
+}
+
+#include "../../task.c"
diff --git a/libgomp/config/gcn/team.c b/libgomp/config/gcn/team.c
new file mode 100644
index 00000000000..79aec65a24e
--- /dev/null
+++ b/libgomp/config/gcn/team.c
@@ -0,0 +1,202 @@ 
+/* Copyright (C) 2017-2019 Free Software Foundation, Inc.
+   Contributed by Mentor Embedded.
+
+   This file is part of the GNU Offloading and Multi Processing Library
+   (libgomp).
+
+   Libgomp 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.
+
+   Libgomp 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/>.  */
+
+/* This file handles maintainance of threads on AMD GCN.  */
+
+#include "libgomp.h"
+#include <stdlib.h>
+#include <string.h>
+
+static void gomp_thread_start (struct gomp_thread_pool *);
+void gomp_print_string (const char *msg, const char *val);
+
+/* This externally visible function handles target region entry.  It
+   sets up a per-team thread pool and transfers control by returning to
+   the kernel in the master thread or gomp_thread_start in other threads.
+
+   The name of this function is part of the interface with the compiler: for
+   each OpenMP kernel the compiler configures the stack, then calls here.
+
+   Likewise, gomp_gcn_exit_kernel is called during the kernel epilogue.  */
+
+void
+gomp_gcn_enter_kernel (void)
+{
+  int tid, ntids;
+  tid = __builtin_gcn_dim_pos (1);
+  ntids = __builtin_gcn_dim_size (1);
+  if (tid == 0)
+    {
+      gomp_global_icv.nthreads_var = ntids;
+      /* Starting additional threads is not supported.  */
+      gomp_global_icv.dyn_var = true;
+
+      set_gcn_thrs (calloc (ntids, sizeof (struct gomp_thread)));
+      if (gcn_thrs () == NULL)
+	goto oom;
+
+      struct gomp_thread_pool *pool = malloc (sizeof (*pool));
+      if (pool == NULL)
+	goto oom;
+
+      pool->threads = malloc (ntids * sizeof (*pool->threads));
+      if (pool->threads == NULL)
+	goto oom;
+
+      for (tid = 0; tid < ntids; tid++)
+	pool->threads[tid] = gcn_thrs () + tid;
+      pool->threads_size = ntids;
+      pool->threads_used = ntids;
+      pool->threads_busy = 1;
+      pool->last_team = NULL;
+      gomp_simple_barrier_init (&pool->threads_dock, ntids);
+
+      gcn_thrs ()[0].thread_pool = pool;
+      asm ("s_barrier" ::: "memory");
+      return;  /* Return to kernel.  */
+    }
+  else
+    {
+      asm ("s_barrier" ::: "memory");
+      gomp_thread_start (gcn_thrs ()[0].thread_pool);
+      /* gomp_thread_start does not return.  */
+    }
+
+oom:
+  gomp_print_string ("GCN heap exhausted; try setting GCN_HEAP_SIZE.", "");
+  abort();
+}
+
+void
+gomp_gcn_exit_kernel (void)
+{
+  gomp_free_thread (gcn_thrs ());
+}
+
+/* This function contains the idle loop in which a thread waits
+   to be called up to become part of a team.  */
+
+static void
+gomp_thread_start (struct gomp_thread_pool *pool)
+{
+  struct gomp_thread *thr = gomp_thread ();
+
+  gomp_sem_init (&thr->release, 0);
+  thr->thread_pool = pool;
+
+  /* The loop exits only when "fn" is assigned "gomp_free_pool_helper",
+     which contains "s_endpgm", or an infinite no-op loop is
+     suspected (this happens when the thread master crashes).  */
+  int nul_limit = 99;
+  do
+    {
+      gomp_simple_barrier_wait (&pool->threads_dock);
+      if (!thr->fn)
+	{
+	  if (nul_limit-- > 0)
+	    continue;
+	  else
+	    {
+	      gomp_print_string ("team master not responding;",
+				 " slave thread aborting");
+	      abort();
+	    }
+	}
+      thr->fn (thr->data);
+      thr->fn = NULL;
+
+      struct gomp_task *task = thr->task;
+      gomp_team_barrier_wait_final (&thr->ts.team->barrier);
+      gomp_finish_task (task);
+    }
+  while (1);
+}
+
+/* Launch a team.  */
+
+void
+gomp_team_start (void (*fn) (void *), void *data, unsigned nthreads,
+		 unsigned flags, struct gomp_team *team,
+		 struct gomp_taskgroup *taskgroup)
+{
+  struct gomp_thread *thr, *nthr;
+  struct gomp_task *task;
+  struct gomp_task_icv *icv;
+  struct gomp_thread_pool *pool;
+  unsigned long nthreads_var;
+
+  thr = gomp_thread ();
+  pool = thr->thread_pool;
+  task = thr->task;
+  icv = task ? &task->icv : &gomp_global_icv;
+
+  /* Always save the previous state, even if this isn't a nested team.
+     In particular, we should save any work share state from an outer
+     orphaned work share construct.  */
+  team->prev_ts = thr->ts;
+
+  thr->ts.team = team;
+  thr->ts.team_id = 0;
+  ++thr->ts.level;
+  if (nthreads > 1)
+    ++thr->ts.active_level;
+  thr->ts.work_share = &team->work_shares[0];
+  thr->ts.last_work_share = NULL;
+  thr->ts.single_count = 0;
+  thr->ts.static_trip = 0;
+  thr->task = &team->implicit_task[0];
+  nthreads_var = icv->nthreads_var;
+  gomp_init_task (thr->task, task, icv);
+  team->implicit_task[0].icv.nthreads_var = nthreads_var;
+  team->implicit_task[0].taskgroup = taskgroup;
+
+  if (nthreads == 1)
+    return;
+
+  /* Release existing idle threads.  */
+  for (unsigned i = 1; i < nthreads; ++i)
+    {
+      nthr = pool->threads[i];
+      nthr->ts.team = team;
+      nthr->ts.work_share = &team->work_shares[0];
+      nthr->ts.last_work_share = NULL;
+      nthr->ts.team_id = i;
+      nthr->ts.level = team->prev_ts.level + 1;
+      nthr->ts.active_level = thr->ts.active_level;
+      nthr->ts.single_count = 0;
+      nthr->ts.static_trip = 0;
+      nthr->task = &team->implicit_task[i];
+      gomp_init_task (nthr->task, task, icv);
+      team->implicit_task[i].icv.nthreads_var = nthreads_var;
+      team->implicit_task[i].taskgroup = taskgroup;
+      nthr->fn = fn;
+      nthr->data = data;
+      team->ordered_release[i] = &nthr->release;
+    }
+
+  gomp_simple_barrier_wait (&pool->threads_dock);
+}
+
+#include "../../team.c"
diff --git a/libgomp/config/gcn/time.c b/libgomp/config/gcn/time.c
new file mode 100644
index 00000000000..f189e55889c
--- /dev/null
+++ b/libgomp/config/gcn/time.c
@@ -0,0 +1,52 @@ 
+/* Copyright (C) 2015-2019 Free Software Foundation, Inc.
+   Contributed by Mentor Embedded.
+
+   This file is part of the GNU Offloading and Multi Processing Library
+   (libgomp).
+
+   Libgomp 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.
+
+   Libgomp 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/>.  */
+
+/* This file implements timer routines for AMD GCN.  */
+
+#include "libgomp.h"
+
+/* According to AMD:
+    dGPU RTC is 27MHz
+    AGPU RTC is 100MHz
+   FIXME: DTRT on an APU.  */
+#define RTC_TICKS (1.0 / 27000000.0) /* 27MHz */
+
+double
+omp_get_wtime (void)
+{
+  uint64_t clock;
+  asm ("s_memrealtime %0\n\t"
+       "s_waitcnt 0" : "=r" (clock));
+  return clock * RTC_TICKS;
+}
+
+double
+omp_get_wtick (void)
+{
+  return RTC_TICKS;
+}
+
+ialias (omp_get_wtime)
+ialias (omp_get_wtick)
diff --git a/libgomp/config/linux/gomp_print.c b/libgomp/config/linux/gomp_print.c
new file mode 100644
index 00000000000..811bdd6e9a9
--- /dev/null
+++ b/libgomp/config/linux/gomp_print.c
@@ -0,0 +1,20 @@ 
+#include <stdio.h>
+#include <stdint.h>
+
+void
+gomp_print_string (const char *msg, const char *value)
+{
+  printf ("%s%s\n", msg, value);
+}
+
+void
+gomp_print_integer (const char *msg, int64_t value)
+{
+  printf ("%s%ld\n", msg, value);
+}
+
+void
+gomp_print_double (const char *msg, double value)
+{
+  printf ("%s%f\n", msg, value);
+}
diff --git a/libgomp/configure b/libgomp/configure
index afceea5cf88..39da8af4546 100755
--- a/libgomp/configure
+++ b/libgomp/configure
@@ -661,6 +661,8 @@  LIBGOMP_BUILD_VERSIONED_SHLIB_FALSE
 LIBGOMP_BUILD_VERSIONED_SHLIB_TRUE
 OPT_LDFLAGS
 SECTION_LDFLAGS
+PLUGIN_GCN_FALSE
+PLUGIN_GCN_TRUE
 PLUGIN_HSA_FALSE
 PLUGIN_HSA_TRUE
 PLUGIN_NVPTX_FALSE
@@ -669,6 +671,10 @@  offload_additional_lib_paths
 offload_additional_options
 offload_targets
 offload_plugins
+PLUGIN_GCN_LIBS
+PLUGIN_GCN_LDFLAGS
+PLUGIN_GCN_CPPFLAGS
+PLUGIN_GCN
 PLUGIN_HSA_LIBS
 PLUGIN_HSA_LDFLAGS
 PLUGIN_HSA_CPPFLAGS
@@ -11396,7 +11402,7 @@  else
   lt_dlunknown=0; lt_dlno_uscore=1; lt_dlneed_uscore=2
   lt_status=$lt_dlunknown
   cat > conftest.$ac_ext <<_LT_EOF
-#line 11399 "configure"
+#line 11405 "configure"
 #include "confdefs.h"
 
 #if HAVE_DLFCN_H
@@ -11502,7 +11508,7 @@  else
   lt_dlunknown=0; lt_dlno_uscore=1; lt_dlneed_uscore=2
   lt_status=$lt_dlunknown
   cat > conftest.$ac_ext <<_LT_EOF
-#line 11505 "configure"
+#line 11511 "configure"
 #include "confdefs.h"
 
 #if HAVE_DLFCN_H
@@ -15315,7 +15321,7 @@  case "$host" in
   *-*-rtems*)
     # RTEMS supports Pthreads, but the library is not available at GCC build time.
     ;;
-  nvptx*-*-*)
+  nvptx*-*-* | amdgcn*-*-*)
     # NVPTX does not support Pthreads, has its own code replacement.
     libgomp_use_pthreads=no
     # NVPTX is an accelerator-only target
@@ -15645,6 +15651,15 @@  PLUGIN_HSA_LIBS=
 
 
 
+PLUGIN_GCN=0
+PLUGIN_GCN_CPPFLAGS=
+PLUGIN_GCN_LDFLAGS=
+PLUGIN_GCN_LIBS=
+
+
+
+
+
 # Parse '--enable-offload-targets', figure out the corresponding libgomp
 # plugins, and configure to find the corresponding offload compilers.
 # 'offload_plugins' and 'offload_targets' will be populated in the same order.
@@ -15756,6 +15771,29 @@  rm -f core conftest.err conftest.$ac_objext \
             ;;
         esac
         ;;
+
+      amdgcn*)
+	case "${target}" in
+	  x86_64-*-*)
+	    case " ${CC} ${CFLAGS} " in
+	      *" -m32 "*)
+		PLUGIN_GCN=0
+		;;
+	      *)
+		tgt_name=gcn
+		PLUGIN_GCN=$tgt
+		PLUGIN_GCN_CPPFLAGS=$HSA_RUNTIME_CPPFLAGS
+		PLUGIN_GCN_LDFLAGS="$HSA_RUNTIME_LDFLAGS"
+		PLUGIN_GCN_LIBS="-ldl"
+		PLUGIN_GCN=1
+		;;
+	      esac
+	    ;;
+	  *-*-*)
+	    PLUGIN_GCN=0
+	     ;;
+	esac
+	;;
       *)
 	as_fn_error $? "unknown offload target specified" "$LINENO" 5
 	;;
@@ -15820,6 +15858,19 @@  cat >>confdefs.h <<_ACEOF
 #define PLUGIN_HSA $PLUGIN_HSA
 _ACEOF
 
+ if test $PLUGIN_GCN = 1; then
+  PLUGIN_GCN_TRUE=
+  PLUGIN_GCN_FALSE='#'
+else
+  PLUGIN_GCN_TRUE='#'
+  PLUGIN_GCN_FALSE=
+fi
+
+
+cat >>confdefs.h <<_ACEOF
+#define PLUGIN_GCN $PLUGIN_GCN
+_ACEOF
+
 
 if test "$HSA_RUNTIME_LIB" != ""; then
   HSA_RUNTIME_LIB="$HSA_RUNTIME_LIB/"
@@ -17445,6 +17496,10 @@  if test -z "${PLUGIN_HSA_TRUE}" && test -z "${PLUGIN_HSA_FALSE}"; then
   as_fn_error $? "conditional \"PLUGIN_HSA\" was never defined.
 Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
+if test -z "${PLUGIN_GCN_TRUE}" && test -z "${PLUGIN_GCN_FALSE}"; then
+  as_fn_error $? "conditional \"PLUGIN_GCN\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
+fi
 if test -z "${LIBGOMP_BUILD_VERSIONED_SHLIB_TRUE}" && test -z "${LIBGOMP_BUILD_VERSIONED_SHLIB_FALSE}"; then
   as_fn_error $? "conditional \"LIBGOMP_BUILD_VERSIONED_SHLIB\" was never defined.
 Usually this means the macro was only invoked conditionally." "$LINENO" 5
diff --git a/libgomp/configure.ac b/libgomp/configure.ac
index 2350b78f310..33e183909b3 100644
--- a/libgomp/configure.ac
+++ b/libgomp/configure.ac
@@ -177,7 +177,7 @@  case "$host" in
   *-*-rtems*)
     # RTEMS supports Pthreads, but the library is not available at GCC build time.
     ;;
-  nvptx*-*-*)
+  nvptx*-*-* | amdgcn*-*-*)
     # NVPTX does not support Pthreads, has its own code replacement.
     libgomp_use_pthreads=no
     # NVPTX is an accelerator-only target
diff --git a/libgomp/configure.tgt b/libgomp/configure.tgt
index b88bf72fe3d..06ee115ece9 100644
--- a/libgomp/configure.tgt
+++ b/libgomp/configure.tgt
@@ -154,7 +154,7 @@  case "${target}" in
 	;;
 
   nvptx*-*-*)
-	config_path="nvptx"
+	config_path="nvptx accel"
 	;;
 
   *-*-rtems*)
@@ -164,6 +164,10 @@  case "${target}" in
 	fi
 	;;
 
+  amdgcn*-*-*)
+	config_path="gcn accel"
+	;;
+
   *)
 	;;
 
diff --git a/libgomp/fortran.c b/libgomp/fortran.c
index 4d544be1c99..fdc324d129d 100644
--- a/libgomp/fortran.c
+++ b/libgomp/fortran.c
@@ -626,7 +626,7 @@  omp_display_affinity_ (const char *format, size_t format_len)
   if (ret < sizeof buf)
     {
       buf[ret] = '\n';
-      gomp_print_string (buf, ret + 1);
+      gomp_write_string (buf, ret + 1);
     }
   else
     {
@@ -635,7 +635,7 @@  omp_display_affinity_ (const char *format, size_t format_len)
 			     format_len ? fmt : gomp_affinity_format_var,
 			     gomp_thread_self (), &thr->ts, thr->place);
       b[ret] = '\n';
-      gomp_print_string (b, ret + 1);
+      gomp_write_string (b, ret + 1);
       free (b);
     }
   if (fmt && fmt != fmt_buf)
diff --git a/libgomp/libgomp-plugin.h b/libgomp/libgomp-plugin.h
index da8ce260564..bd63c422b0c 100644
--- a/libgomp/libgomp-plugin.h
+++ b/libgomp/libgomp-plugin.h
@@ -50,7 +50,8 @@  enum offload_target_type
   /* OFFLOAD_TARGET_TYPE_HOST_NONSHM = 3 removed.  */
   OFFLOAD_TARGET_TYPE_NVIDIA_PTX = 5,
   OFFLOAD_TARGET_TYPE_INTEL_MIC = 6,
-  OFFLOAD_TARGET_TYPE_HSA = 7
+  OFFLOAD_TARGET_TYPE_HSA = 7,
+  OFFLOAD_TARGET_TYPE_GCN = 8
 };
 
 /* Container type for passing device properties.  */
@@ -120,7 +121,7 @@  extern void GOMP_OFFLOAD_openacc_exec (void (*) (void *), size_t, void **,
 				       void **, unsigned *, void *);
 extern void *GOMP_OFFLOAD_openacc_create_thread_data (int);
 extern void GOMP_OFFLOAD_openacc_destroy_thread_data (void *);
-extern struct goacc_asyncqueue *GOMP_OFFLOAD_openacc_async_construct (void);
+extern struct goacc_asyncqueue *GOMP_OFFLOAD_openacc_async_construct (int);
 extern bool GOMP_OFFLOAD_openacc_async_destruct (struct goacc_asyncqueue *);
 extern int GOMP_OFFLOAD_openacc_async_test (struct goacc_asyncqueue *);
 extern bool GOMP_OFFLOAD_openacc_async_synchronize (struct goacc_asyncqueue *);
diff --git a/libgomp/libgomp.h b/libgomp/libgomp.h
index 31403ba67a9..803f72db922 100644
--- a/libgomp/libgomp.h
+++ b/libgomp/libgomp.h
@@ -692,6 +692,24 @@  static inline struct gomp_thread *gomp_thread (void)
   asm ("mov.u32 %0, %%tid.y;" : "=r" (tid));
   return nvptx_thrs + tid;
 }
+#elif defined __AMDGCN__
+static inline struct gomp_thread *gcn_thrs (void)
+{
+  /* The value is at the bottom of LDS.  */
+  struct gomp_thread * __lds *thrs = (struct gomp_thread * __lds *)4;
+  return *thrs;
+}
+static inline void set_gcn_thrs (struct gomp_thread *val)
+{
+  /* The value is at the bottom of LDS.  */
+  struct gomp_thread * __lds *thrs = (struct gomp_thread * __lds *)4;
+  *thrs = val;
+}
+static inline struct gomp_thread *gomp_thread (void)
+{
+  int tid = __builtin_gcn_dim_pos(1);
+  return gcn_thrs () + tid;
+}
 #elif defined HAVE_TLS || defined USE_EMUTLS
 extern __thread struct gomp_thread gomp_tls_data;
 static inline struct gomp_thread *gomp_thread (void)
@@ -751,7 +769,7 @@  extern void gomp_display_affinity_place (char *, size_t, size_t *, int);
 
 /* affinity-fmt.c */
 
-extern void gomp_print_string (const char *str, size_t len);
+extern void gomp_write_string (const char *str, size_t len);
 extern void gomp_set_affinity_format (const char *, size_t);
 extern void gomp_display_string (char *, size_t, size_t *, const char *,
 				 size_t);
diff --git a/libgomp/libgomp.map b/libgomp/libgomp.map
index 3be4a0cd37e..02596b0e265 100644
--- a/libgomp/libgomp.map
+++ b/libgomp/libgomp.map
@@ -312,6 +312,9 @@  GOMP_4.5 {
 	GOMP_loop_ull_nonmonotonic_guided_start;
 	GOMP_parallel_loop_nonmonotonic_dynamic;
 	GOMP_parallel_loop_nonmonotonic_guided;
+	gomp_print_string;
+	gomp_print_integer;
+	gomp_print_double;
 } GOMP_4.0.1;
 
 GOMP_5.0 {
diff --git a/libgomp/oacc-async.c b/libgomp/oacc-async.c
index 1760e8c90c6..2b24ae7adc2 100644
--- a/libgomp/oacc-async.c
+++ b/libgomp/oacc-async.c
@@ -100,7 +100,8 @@  lookup_goacc_asyncqueue (struct goacc_thread *thr, bool create, int async)
 
   if (!dev->openacc.async.asyncqueue[async])
     {
-      dev->openacc.async.asyncqueue[async] = dev->openacc.async.construct_func ();
+      dev->openacc.async.asyncqueue[async]
+	= dev->openacc.async.construct_func (dev->target_id);
 
       if (!dev->openacc.async.asyncqueue[async])
 	{
diff --git a/libgomp/oacc-host.c b/libgomp/oacc-host.c
index beeca287e15..21f73302f03 100644
--- a/libgomp/oacc-host.c
+++ b/libgomp/oacc-host.c
@@ -257,7 +257,7 @@  host_openacc_async_queue_callback (struct goacc_asyncqueue *aq
 }
 
 static struct goacc_asyncqueue *
-host_openacc_async_construct (void)
+host_openacc_async_construct (int device __attribute__((unused)))
 {
   /* Non-NULL 0xffff... value as opaque dummy.  */
   return (struct goacc_asyncqueue *) -1;
diff --git a/libgomp/oacc-init.c b/libgomp/oacc-init.c
index 766f59f3963..c9896e7ca24 100644
--- a/libgomp/oacc-init.c
+++ b/libgomp/oacc-init.c
@@ -110,8 +110,9 @@  name_of_acc_device_t (enum acc_device_t type)
     case acc_device_host: return "host";
     case acc_device_not_host: return "not_host";
     case acc_device_nvidia: return "nvidia";
-    case /* not supported */ _acc_device_intel_mic:
+    case acc_device_gcn: return "gcn";
     case /* not supported */ _acc_device_hsa:
+    case /* not supported */ _acc_device_intel_mic:
     default: gomp_fatal ("unknown device type %u", (unsigned) type);
     }
 }
diff --git a/libgomp/oacc-int.h b/libgomp/oacc-int.h
index 5ca9944601e..9dc6c8a5713 100644
--- a/libgomp/oacc-int.h
+++ b/libgomp/oacc-int.h
@@ -82,7 +82,14 @@  struct goacc_thread
   void *target_tls;
 };
 
-#if defined HAVE_TLS || defined USE_EMUTLS
+#ifdef __AMDGCN__
+static inline struct goacc_thread *
+goacc_thread (void)
+{
+  /* Unused in the offload libgomp for OpenACC: return a dummy value.  */
+  return 0;
+}
+#elif defined HAVE_TLS || defined USE_EMUTLS
 extern __thread struct goacc_thread *goacc_tls_data;
 static inline struct goacc_thread *
 goacc_thread (void)
diff --git a/libgomp/oacc-parallel.c b/libgomp/oacc-parallel.c
index a374f91f0d7..6b089763e9a 100644
--- a/libgomp/oacc-parallel.c
+++ b/libgomp/oacc-parallel.c
@@ -732,6 +732,8 @@  GOACC_enter_exit_data (int flags_m, size_t mapnum,
 	  || kind == GOMP_MAP_FORCE_FROM)
 	finalize = true;
     }
+  else if (num_waits == acc_async_noval)
+    acc_wait_all_async (async);
 
   /* Determine if this is an "acc enter data".  */
   for (i = 0; i < mapnum; ++i)
@@ -749,7 +751,8 @@  GOACC_enter_exit_data (int flags_m, size_t mapnum,
 	  || kind == GOMP_MAP_FORCE_TO
 	  || kind == GOMP_MAP_TO
 	  || kind == GOMP_MAP_ALLOC
-	  || kind == GOMP_MAP_DECLARE_ALLOCATE)
+	  || kind == GOMP_MAP_DECLARE_ALLOCATE
+	  || kind == GOMP_MAP_ZERO_LEN_ARRAY_SECTION)
 	{
 	  data_enter = true;
 	  break;
@@ -761,7 +764,8 @@  GOACC_enter_exit_data (int flags_m, size_t mapnum,
 	  || kind == GOMP_MAP_FORCE_DETACH
 	  || kind == GOMP_MAP_FROM
 	  || kind == GOMP_MAP_FORCE_FROM
-	  || kind == GOMP_MAP_DECLARE_DEALLOCATE)
+	  || kind == GOMP_MAP_DECLARE_DEALLOCATE
+	  || kind == GOMP_MAP_DELETE_ZERO_LEN_ARRAY_SECTION)
 	break;
 
       gomp_fatal (">>>> GOACC_enter_exit_data UNHANDLED kind 0x%.2x",
@@ -866,6 +870,10 @@  GOACC_enter_exit_data (int flags_m, size_t mapnum,
 		case GOMP_MAP_FORCE_ALLOC:
 		  acc_create_async (hostaddrs[i], sizes[i], async);
 		  break;
+		case GOMP_MAP_ZERO_LEN_ARRAY_SECTION:
+		  if (hostaddrs[i] != NULL)
+		    acc_create_async (hostaddrs[i], 1, async);
+		  break;
 		case GOMP_MAP_TO:
 		case GOMP_MAP_FORCE_TO:
 		  if (hostaddrs[i])
@@ -986,6 +994,15 @@  GOACC_enter_exit_data (int flags_m, size_t mapnum,
 			acc_delete_async (hostaddrs[i], sizes[i], async);
 		    }
 		  break;
+		case GOMP_MAP_DELETE_ZERO_LEN_ARRAY_SECTION:
+		  if (acc_is_present (hostaddrs[i], 1))
+		    {
+		      if (finalize)
+			acc_delete_finalize_async (hostaddrs[i], 1, async);
+		      else
+			acc_delete_async (hostaddrs[i], 1, async);
+		    }
+		  break;
 		case GOMP_MAP_DETACH:
 		case GOMP_MAP_FORCE_DETACH:
 		case GOMP_MAP_FORCE_PRESENT:
@@ -1196,6 +1213,7 @@  GOACC_update (int flags_m, size_t mapnum,
 	{
 	case GOMP_MAP_POINTER:
 	case GOMP_MAP_TO_PSET:
+	case GOMP_MAP_ZERO_LEN_ARRAY_SECTION:
 	  break;
 
 	case GOMP_MAP_ALWAYS_POINTER:
diff --git a/libgomp/omp.h.in b/libgomp/omp.h.in
index d7ac71400ad..0d8f4763b55 100644
--- a/libgomp/omp.h.in
+++ b/libgomp/omp.h.in
@@ -188,6 +188,18 @@  extern __SIZE_TYPE__ omp_capture_affinity (char *, __SIZE_TYPE__, const char *)
 extern int omp_pause_resource (omp_pause_resource_t, int) __GOMP_NOTHROW;
 extern int omp_pause_resource_all (omp_pause_resource_t) __GOMP_NOTHROW;
 
+/************************************************************************/
+/* Libgomp extensions.                                                  */
+
+#include <stdint.h>
+
+/* Print a message, and value, possibly from a GPU-offloaded function.
+   Primarily intended for debug messages.
+   Maximum message & value length is 128 bytes.  */
+void gomp_print_string (const char *msg, const char *value);
+void gomp_print_integer (const char *msg, int64_t value);
+void gomp_print_double (const char *msg, double value);
+
 #ifdef __cplusplus
 }
 #endif
diff --git a/libgomp/omp_lib.f90.in b/libgomp/omp_lib.f90.in
index c4e7c0dff4a..65aec54da6b 100644
--- a/libgomp/omp_lib.f90.in
+++ b/libgomp/omp_lib.f90.in
@@ -484,4 +484,30 @@ 
           end function
         end interface
 
+        ! Libgomp extensions.
+
+        interface
+          subroutine gomp_print_string (msg, str) bind(C)
+            use iso_c_binding, only: c_char
+            character (kind=c_char) :: msg(*)
+            character (kind=c_char) :: str(*)
+          end subroutine gomp_print_string
+        end interface
+
+        interface
+          subroutine gomp_print_integer (msg, i) bind(C)
+            use iso_c_binding, only: c_char, c_int64_t
+            character (kind=c_char) :: msg(*)
+            integer (kind=c_int64_t), value :: i
+          end subroutine gomp_print_integer
+        end interface
+
+        interface
+          subroutine gomp_print_double (msg, d) bind(C)
+            use iso_c_binding, only: c_char, c_double
+            character (kind=c_char) :: msg(*)
+            real (kind=c_double), value :: d
+          end subroutine gomp_print_double
+        end interface
+
       end module omp_lib
diff --git a/libgomp/openacc.f90 b/libgomp/openacc.f90
index 8b514c54e39..e675843a411 100644
--- a/libgomp/openacc.f90
+++ b/libgomp/openacc.f90
@@ -37,7 +37,7 @@  module openacc_kinds
   integer, parameter :: acc_device_kind = int32
 
   public :: acc_device_none, acc_device_default, acc_device_host
-  public :: acc_device_not_host, acc_device_nvidia
+  public :: acc_device_not_host, acc_device_nvidia, acc_device_gcn
 
   ! Keep in sync with include/gomp-constants.h.
   integer (acc_device_kind), parameter :: acc_device_none = 0
@@ -46,7 +46,9 @@  module openacc_kinds
   ! integer (acc_device_kind), parameter :: acc_device_host_nonshm = 3 removed.
   integer (acc_device_kind), parameter :: acc_device_not_host = 4
   integer (acc_device_kind), parameter :: acc_device_nvidia = 5
-  integer (acc_device_kind), parameter :: acc_device_current = 8
+  integer (acc_device_kind), parameter :: acc_device_hsa = 7
+  integer (acc_device_kind), parameter :: acc_device_gcn = 8
+  integer (acc_device_kind), parameter :: acc_device_current = 9
 
   public :: acc_device_property
 
diff --git a/libgomp/openacc.h b/libgomp/openacc.h
index 26084dc5ddd..83e7c46a70d 100644
--- a/libgomp/openacc.h
+++ b/libgomp/openacc.h
@@ -57,7 +57,8 @@  typedef enum acc_device_t {
   acc_device_nvidia = 5,
   /* not supported */ _acc_device_intel_mic = 6,
   /* not supported */ _acc_device_hsa = 7,
-  acc_device_current = 8,
+  acc_device_gcn = 8,
+  acc_device_current = 9,
   _ACC_device_hwm,
   /* Ensure enumeration is layout compatible with int.  */
   _ACC_highest = __INT_MAX__,
diff --git a/libgomp/openacc_lib.h b/libgomp/openacc_lib.h
index 9c26fecff41..3284b961631 100644
--- a/libgomp/openacc_lib.h
+++ b/libgomp/openacc_lib.h
@@ -42,6 +42,8 @@ 
 !     removed.
       integer (acc_device_kind), parameter :: acc_device_not_host = 4
       integer (acc_device_kind), parameter :: acc_device_nvidia = 5
+      integer (acc_device_kind), parameter :: acc_device_hsa = 7
+      integer (acc_device_kind), parameter :: acc_device_gcn = 8
 
       integer, parameter :: acc_handle_kind = 4
 
diff --git a/libgomp/plugin/Makefrag.am b/libgomp/plugin/Makefrag.am
index 168ef59de41..45ed043e333 100644
--- a/libgomp/plugin/Makefrag.am
+++ b/libgomp/plugin/Makefrag.am
@@ -52,3 +52,17 @@  libgomp_plugin_hsa_la_LDFLAGS += $(PLUGIN_HSA_LDFLAGS)
 libgomp_plugin_hsa_la_LIBADD = libgomp.la $(PLUGIN_HSA_LIBS)
 libgomp_plugin_hsa_la_LIBTOOLFLAGS = --tag=disable-static
 endif
+
+if PLUGIN_GCN
+# AMD GCN plugin
+libgomp_plugin_gcn_version_info = -version-info $(libtool_VERSION)
+toolexeclib_LTLIBRARIES += libgomp-plugin-gcn.la
+libgomp_plugin_gcn_la_SOURCES = plugin/plugin-gcn.c
+libgomp_plugin_gcn_la_CPPFLAGS = $(AM_CPPFLAGS) $(PLUGIN_GCN_CPPFLAGS) \
+	-D_GNU_SOURCE
+libgomp_plugin_gcn_la_LDFLAGS = $(libgomp_plugin_gcn_version_info) \
+	$(lt_host_flags)
+libgomp_plugin_gcn_la_LDFLAGS += $(PLUGIN_GCN_LDFLAGS)
+libgomp_plugin_gcn_la_LIBADD = libgomp.la $(PLUGIN_GCN_LIBS)
+libgomp_plugin_gcn_la_LIBTOOLFLAGS = --tag=disable-static
+endif
diff --git a/libgomp/plugin/configfrag.ac b/libgomp/plugin/configfrag.ac
index 13ca26f47d9..6fedd28eccc 100644
--- a/libgomp/plugin/configfrag.ac
+++ b/libgomp/plugin/configfrag.ac
@@ -137,6 +137,15 @@  AC_SUBST(PLUGIN_HSA_CPPFLAGS)
 AC_SUBST(PLUGIN_HSA_LDFLAGS)
 AC_SUBST(PLUGIN_HSA_LIBS)
 
+PLUGIN_GCN=0
+PLUGIN_GCN_CPPFLAGS=
+PLUGIN_GCN_LDFLAGS=
+PLUGIN_GCN_LIBS=
+AC_SUBST(PLUGIN_GCN)
+AC_SUBST(PLUGIN_GCN_CPPFLAGS)
+AC_SUBST(PLUGIN_GCN_LDFLAGS)
+AC_SUBST(PLUGIN_GCN_LIBS)
+
 # Parse '--enable-offload-targets', figure out the corresponding libgomp
 # plugins, and configure to find the corresponding offload compilers.
 # 'offload_plugins' and 'offload_targets' will be populated in the same order.
@@ -237,6 +246,29 @@  if test x"$enable_offload_targets" != x; then
             ;;
         esac
         ;;
+
+      amdgcn*)
+	case "${target}" in
+	  x86_64-*-*)
+	    case " ${CC} ${CFLAGS} " in
+	      *" -m32 "*)
+		PLUGIN_GCN=0
+		;;
+	      *)
+		tgt_name=gcn
+		PLUGIN_GCN=$tgt
+		PLUGIN_GCN_CPPFLAGS=$HSA_RUNTIME_CPPFLAGS
+		PLUGIN_GCN_LDFLAGS="$HSA_RUNTIME_LDFLAGS"
+		PLUGIN_GCN_LIBS="-ldl"
+		PLUGIN_GCN=1
+		;;
+	      esac
+	    ;;
+	  *-*-*)
+	    PLUGIN_GCN=0
+	     ;;
+	esac
+	;;
       *)
 	AC_MSG_ERROR([unknown offload target specified])
 	;;
@@ -275,6 +307,9 @@  AC_DEFINE_UNQUOTED([PLUGIN_NVPTX_DYNAMIC], [$PLUGIN_NVPTX_DYNAMIC],
 AM_CONDITIONAL([PLUGIN_HSA], [test $PLUGIN_HSA = 1])
 AC_DEFINE_UNQUOTED([PLUGIN_HSA], [$PLUGIN_HSA],
   [Define to 1 if the HSA plugin is built, 0 if not.])
+AM_CONDITIONAL([PLUGIN_GCN], [test $PLUGIN_GCN = 1])
+AC_DEFINE_UNQUOTED([PLUGIN_GCN], [$PLUGIN_GCN],
+  [Define to 1 if the GCN plugin is built, 0 if not.])
 
 if test "$HSA_RUNTIME_LIB" != ""; then
   HSA_RUNTIME_LIB="$HSA_RUNTIME_LIB/"
diff --git a/libgomp/plugin/plugin-gcn.c b/libgomp/plugin/plugin-gcn.c
new file mode 100644
index 00000000000..b059348c7bf
--- /dev/null
+++ b/libgomp/plugin/plugin-gcn.c
@@ -0,0 +1,3482 @@ 
+/* Plugin for AMD GCN execution.
+
+   Copyright (C) 2013-2019 Free Software Foundation, Inc.
+
+   Contributed by Mentor Embedded
+
+   This file is part of the GNU Offloading and Multi Processing Library
+   (libgomp).
+
+   Libgomp 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.
+
+   Libgomp 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/>.  */
+
+#include "config.h"
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <pthread.h>
+#include <inttypes.h>
+#include <stdbool.h>
+#include <limits.h>
+#include <hsa.h>
+#include <dlfcn.h>
+#include <signal.h>
+#include "libgomp-plugin.h"
+#include "gomp-constants.h"
+#include <elf.h>
+#include "oacc-plugin.h"
+#include "oacc-int.h"
+#include <assert.h>
+
+#define obstack_chunk_alloc GOMP_PLUGIN_malloc
+#define obstack_chunk_free free
+#include "obstack.h"
+
+/* These probably won't be in elf.h for a while.  */
+#define R_AMDGPU_NONE		0
+#define R_AMDGPU_ABS32_LO	1	/* (S + A) & 0xFFFFFFFF  */
+#define R_AMDGPU_ABS32_HI	2	/* (S + A) >> 32  */
+#define R_AMDGPU_ABS64		3	/* S + A  */
+#define R_AMDGPU_REL32		4	/* S + A - P  */
+#define R_AMDGPU_REL64		5	/* S + A - P  */
+#define R_AMDGPU_ABS32		6	/* S + A  */
+#define R_AMDGPU_GOTPCREL	7	/* G + GOT + A - P  */
+#define R_AMDGPU_GOTPCREL32_LO	8	/* (G + GOT + A - P) & 0xFFFFFFFF  */
+#define R_AMDGPU_GOTPCREL32_HI	9	/* (G + GOT + A - P) >> 32  */
+#define R_AMDGPU_REL32_LO	10	/* (S + A - P) & 0xFFFFFFFF  */
+#define R_AMDGPU_REL32_HI	11	/* (S + A - P) >> 32  */
+#define reserved		12
+#define R_AMDGPU_RELATIVE64	13	/* B + A  */
+
+/* Secure getenv() which returns NULL if running as SUID/SGID.  */
+#ifndef HAVE_SECURE_GETENV
+#ifdef HAVE___SECURE_GETENV
+#define secure_getenv __secure_getenv
+#elif defined (HAVE_UNISTD_H) && defined(HAVE_GETUID) && defined(HAVE_GETEUID) \
+  && defined(HAVE_GETGID) && defined(HAVE_GETEGID)
+
+#include <unistd.h>
+
+/* Implementation of secure_getenv() for targets where it is not provided but
+   we have at least means to test real and effective IDs. */
+
+static char *
+secure_getenv (const char *name)
+{
+  if ((getuid () == geteuid ()) && (getgid () == getegid ()))
+    return getenv (name);
+  else
+    return NULL;
+}
+
+#else
+#define secure_getenv getenv
+#endif
+#endif
+
+struct gcn_thread
+{
+  int async;
+};
+
+static inline struct gcn_thread *
+gcn_thread (void)
+{
+  return (struct gcn_thread *) GOMP_PLUGIN_acc_thread ();
+}
+
+/* As an HSA runtime is dlopened, following structure defines function
+   pointers utilized by the HSA plug-in.  */
+
+struct hsa_runtime_fn_info
+{
+  /* HSA runtime.  */
+  hsa_status_t (*hsa_status_string_fn) (hsa_status_t status,
+					const char **status_string);
+  hsa_status_t (*hsa_system_get_info_fn) (hsa_system_info_t attribute,
+					  void *value);
+  hsa_status_t (*hsa_agent_get_info_fn) (hsa_agent_t agent,
+					 hsa_agent_info_t attribute,
+					 void *value);
+  hsa_status_t (*hsa_isa_get_info_fn)(hsa_isa_t isa,
+				      hsa_isa_info_t attribute,
+				      uint32_t index,
+				      void *value);
+  hsa_status_t (*hsa_init_fn) (void);
+  hsa_status_t (*hsa_iterate_agents_fn)
+    (hsa_status_t (*callback)(hsa_agent_t agent, void *data), void *data);
+  hsa_status_t (*hsa_region_get_info_fn) (hsa_region_t region,
+					  hsa_region_info_t attribute,
+					  void *value);
+  hsa_status_t (*hsa_queue_create_fn)
+    (hsa_agent_t agent, uint32_t size, hsa_queue_type_t type,
+     void (*callback)(hsa_status_t status, hsa_queue_t *source, void *data),
+     void *data, uint32_t private_segment_size,
+     uint32_t group_segment_size, hsa_queue_t **queue);
+  hsa_status_t (*hsa_agent_iterate_regions_fn)
+    (hsa_agent_t agent,
+     hsa_status_t (*callback)(hsa_region_t region, void *data), void *data);
+  hsa_status_t (*hsa_executable_destroy_fn) (hsa_executable_t executable);
+  hsa_status_t (*hsa_executable_create_fn)
+    (hsa_profile_t profile, hsa_executable_state_t executable_state,
+     const char *options, hsa_executable_t *executable);
+  hsa_status_t (*hsa_executable_global_variable_define_fn)
+    (hsa_executable_t executable, const char *variable_name, void *address);
+  hsa_status_t (*hsa_executable_load_code_object_fn)
+    (hsa_executable_t executable, hsa_agent_t agent,
+     hsa_code_object_t code_object, const char *options);
+  hsa_status_t (*hsa_executable_freeze_fn)(hsa_executable_t executable,
+					   const char *options);
+  hsa_status_t (*hsa_signal_create_fn) (hsa_signal_value_t initial_value,
+					uint32_t num_consumers,
+					const hsa_agent_t *consumers,
+					hsa_signal_t *signal);
+  hsa_status_t (*hsa_memory_allocate_fn) (hsa_region_t region, size_t size,
+					  void **ptr);
+  hsa_status_t (*hsa_memory_copy_fn)(void *dst, const void *src, size_t size);
+  hsa_status_t (*hsa_memory_free_fn) (void *ptr);
+  hsa_status_t (*hsa_signal_destroy_fn) (hsa_signal_t signal);
+  hsa_status_t (*hsa_executable_get_symbol_fn)
+    (hsa_executable_t executable, const char *module_name,
+     const char *symbol_name, hsa_agent_t agent, int32_t call_convention,
+     hsa_executable_symbol_t *symbol);
+  hsa_status_t (*hsa_executable_symbol_get_info_fn)
+    (hsa_executable_symbol_t executable_symbol,
+     hsa_executable_symbol_info_t attribute, void *value);
+  hsa_status_t (*hsa_executable_iterate_symbols_fn)
+    (hsa_executable_t executable,
+     hsa_status_t (*callback)(hsa_executable_t executable,
+			      hsa_executable_symbol_t symbol, void *data),
+     void *data);
+  uint64_t (*hsa_queue_add_write_index_release_fn) (const hsa_queue_t *queue,
+						    uint64_t value);
+  uint64_t (*hsa_queue_load_read_index_acquire_fn) (const hsa_queue_t *queue);
+  void (*hsa_signal_store_relaxed_fn) (hsa_signal_t signal,
+				       hsa_signal_value_t value);
+  void (*hsa_signal_store_release_fn) (hsa_signal_t signal,
+				       hsa_signal_value_t value);
+  hsa_signal_value_t (*hsa_signal_wait_acquire_fn)
+    (hsa_signal_t signal, hsa_signal_condition_t condition,
+     hsa_signal_value_t compare_value, uint64_t timeout_hint,
+     hsa_wait_state_t wait_state_hint);
+  hsa_signal_value_t (*hsa_signal_load_acquire_fn) (hsa_signal_t signal);
+  hsa_status_t (*hsa_queue_destroy_fn) (hsa_queue_t *queue);
+
+  hsa_status_t (*hsa_code_object_deserialize_fn)
+    (void *serialized_code_object, size_t serialized_code_object_size,
+     const char *options, hsa_code_object_t *code_object);
+};
+
+/* HSA runtime functions that are initialized in init_hsa_context.  */
+
+static struct hsa_runtime_fn_info hsa_fns;
+
+/* Keep the following GOMP prefixed structures in sync with respective parts of
+   the compiler.  */
+
+/* Structure describing the run-time and grid properties of an HSA kernel
+   lauch.  */
+
+struct GOMP_kernel_launch_attributes
+{
+  /* Number of dimensions the workload has.  Maximum number is 3.  */
+  uint32_t ndim;
+  /* Size of the grid in the three respective dimensions.  */
+  uint32_t gdims[3];
+  /* Size of work-groups in the respective dimensions.  */
+  uint32_t wdims[3];
+};
+
+/* Collection of information needed for a dispatch of a kernel from a
+   kernel.  */
+
+struct GOMP_hsa_kernel_dispatch
+{
+  /* Pointer to a command queue associated with a kernel dispatch agent.  */
+  void *queue;
+  /* Pointer to reserved memory for OMP data struct copying.  */
+  void *omp_data_memory;
+  /* Pointer to a memory space used for kernel arguments passing.  */
+  void *kernarg_address;
+  /* Kernel object.  */
+  uint64_t object;
+  /* Synchronization signal used for dispatch synchronization.  */
+  uint64_t signal;
+  /* Private segment size.  */
+  uint32_t private_segment_size;
+  /* Group segment size.  */
+  uint32_t group_segment_size;
+  /* Number of children kernel dispatches.  */
+  uint64_t kernel_dispatch_count;
+  /* Debug purpose argument.  */
+  uint64_t debug;
+  /* Levels-var ICV.  */
+  uint64_t omp_level;
+  /* Kernel dispatch structures created for children kernel dispatches.  */
+  struct GOMP_hsa_kernel_dispatch **children_dispatches;
+  /* Number of threads.  */
+  uint32_t omp_num_threads;
+};
+
+/* Structure of the default kernargs segment, supporting gomp_print_*.
+   This will only be used if the requested space is less than 9 bytes.  */
+
+struct kernargs {
+  /* Leave space for the real kernel arguments.
+     OpenACC and OpenMP only use one pointer.  */
+  int64_t dummy1;
+  int64_t dummy2;
+
+  /* A pointer to struct output, below, for console output data.  */
+  int64_t out_ptr;
+
+  /* A pointer to struct heap, below.  */
+  int64_t heap_ptr;
+
+  /* Output data.  */
+  struct output {
+    int return_value;
+    unsigned int next_output;
+    struct printf_data {
+      int written;
+      char msg[128];
+      int type;
+      union {
+	int64_t ivalue;
+	double dvalue;
+	char text[128];
+      };
+    } queue[1024];
+    unsigned int consumed;
+  } output_data;
+};
+
+/* Heap space, allocated target-side, provided for use of newlib malloc.
+   Each module should have it's own heap allocated.
+   Beware that heap usage increases with OpenMP teams.  */
+static size_t gcn_kernel_heap_size = 100*1024*1024;  /* 100MB.  */
+struct heap {
+  int64_t size;
+  char data[0];
+};
+
+/* GCN specific definition of asynchronous queues.  */
+
+#define ASYNC_QUEUE_SIZE 64
+#define DRAIN_QUEUE_SYNCHRONOUS_P false
+#define DEBUG_QUEUES 0
+#define DEBUG_THREAD_SLEEP 0
+#define DEBUG_THREAD_SIGNAL 0
+
+struct kernel_launch
+{
+  struct kernel_info *kernel;
+  void *vars;
+  struct GOMP_kernel_launch_attributes kla;
+};
+
+struct callback
+{
+  void (*fn)(void *);
+  void *data;
+};
+
+struct queue_entry
+{
+  int type;
+  union {
+    struct kernel_launch launch;
+    struct callback callback;
+  } u;
+};
+
+struct goacc_asyncqueue
+{
+  struct agent_info *agent;
+  hsa_queue_t *hsa_queue;
+
+  pthread_t thread_drain_queue;
+  pthread_mutex_t mutex;
+  pthread_cond_t queue_cond_in;
+  pthread_cond_t queue_cond_out;
+  struct queue_entry queue[ASYNC_QUEUE_SIZE];
+  int queue_first;
+  int queue_n;
+  int drain_queue_stop;
+
+  int id;
+  struct goacc_asyncqueue *prev;
+  struct goacc_asyncqueue *next;
+};
+
+/* Part of the libgomp plugin interface.  Return the name of the accelerator,
+   which is "gcn".  */
+
+const char *
+GOMP_OFFLOAD_get_name (void)
+{
+  return "gcn";
+}
+
+/* Part of the libgomp plugin interface.  Return the specific capabilities the
+   HSA accelerator have.  */
+
+unsigned int
+GOMP_OFFLOAD_get_caps (void)
+{
+  /* FIXME: Enable shared memory for APU, but not discrete GPU.  */
+  return /*GOMP_OFFLOAD_CAP_SHARED_MEM |*/ GOMP_OFFLOAD_CAP_OPENMP_400
+	    | GOMP_OFFLOAD_CAP_OPENACC_200;
+}
+
+/* Part of the libgomp plugin interface.  Identify as HSA accelerator.  */
+
+int
+GOMP_OFFLOAD_get_type (void)
+{
+  return OFFLOAD_TARGET_TYPE_GCN;
+}
+
+/* Return the libgomp version number we're compatible with.  There is
+   no requirement for cross-version compatibility.  */
+
+unsigned
+GOMP_OFFLOAD_version (void)
+{
+  return GOMP_VERSION;
+}
+
+/* Flag to decide whether print to stderr information about what is going on.
+   Set in init_debug depending on environment variables.  */
+
+static bool debug;
+
+/* Flag to decide if the runtime should suppress a possible fallback to host
+   execution.  */
+
+static bool suppress_host_fallback;
+
+/* Flag to locate HSA runtime shared library that is dlopened
+   by this plug-in.  */
+
+static const char *hsa_runtime_lib;
+
+/* Flag to decide if the runtime should support also CPU devices (can be
+   a simulator).  */
+
+static bool support_cpu_devices;
+
+/* Runtime dimension overrides.  Zero indicates default.  */
+
+static int override_x_dim = 0;
+static int override_z_dim = 0;
+
+/* Initialize debug and suppress_host_fallback according to the environment.  */
+
+static void
+init_environment_variables (void)
+{
+  if (secure_getenv ("GCN_DEBUG"))
+    debug = true;
+  else
+    debug = false;
+
+  if (secure_getenv ("GCN_SUPPRESS_HOST_FALLBACK"))
+    suppress_host_fallback = true;
+  else
+    suppress_host_fallback = false;
+
+  hsa_runtime_lib = secure_getenv ("HSA_RUNTIME_LIB");
+  if (hsa_runtime_lib == NULL)
+    hsa_runtime_lib = HSA_RUNTIME_LIB "libhsa-runtime64.so";
+
+  support_cpu_devices = secure_getenv ("GCN_SUPPORT_CPU_DEVICES");
+
+  const char *x = secure_getenv ("GCN_NUM_TEAMS");
+  if (!x)
+    x = secure_getenv ("GCN_NUM_GANGS");
+  if (x)
+    override_x_dim = atoi (x);
+
+  const char *z = secure_getenv ("GCN_NUM_THREADS");
+  if (!z)
+    z = secure_getenv ("GCN_NUM_WORKERS");
+  if (z)
+    override_z_dim = atoi (z);
+
+  const char *heap = secure_getenv ("GCN_HEAP_SIZE");
+  if (heap)
+    {
+      size_t tmp = atol (heap);
+      if (tmp)
+	gcn_kernel_heap_size = tmp;
+    }
+}
+
+/* Print a message to stderr if HSA_DEBUG value is set to true.  */
+
+#define HSA_DPRINT(...) \
+  do \
+  { \
+    if (debug) \
+      { \
+	fprintf (stderr, __VA_ARGS__); \
+      } \
+  } \
+  while (false);
+
+/* Flush stderr if GCN_DEBUG value is set to true.  */
+
+#define HSA_FLUSH()				\
+  do {						\
+    if (debug)					\
+      fflush (stderr);				\
+  } while (0)
+
+/* Print a logging message with PREFIX to stderr if HSA_DEBUG value
+   is set to true.  */
+
+#define HSA_LOG(prefix, ...)			\
+  do						\
+    {						\
+      HSA_DPRINT (prefix);			\
+      HSA_DPRINT (__VA_ARGS__);			\
+      HSA_FLUSH ();				\
+    } while (false)
+
+/* Print a debugging message to stderr.  */
+
+#define HSA_DEBUG(...) HSA_LOG ("GCN debug: ", __VA_ARGS__)
+
+/* Print a warning message to stderr.  */
+
+#define HSA_WARNING(...) HSA_LOG ("GCN warning: ", __VA_ARGS__)
+
+/* Print HSA warning STR with an HSA STATUS code.  */
+
+static void
+hsa_warn (const char *str, hsa_status_t status)
+{
+  if (!debug)
+    return;
+
+  const char *hsa_error_msg;
+  hsa_fns.hsa_status_string_fn (status, &hsa_error_msg);
+
+  fprintf (stderr, "GCN warning: %s\nRuntime message: %s\n", str,
+	   hsa_error_msg);
+}
+
+/* Report a fatal error STR together with the HSA error corresponding to STATUS
+   and terminate execution of the current process.  */
+
+static void
+hsa_fatal (const char *str, hsa_status_t status)
+{
+  const char *hsa_error_msg;
+  hsa_fns.hsa_status_string_fn (status, &hsa_error_msg);
+  GOMP_PLUGIN_fatal ("GCN fatal error: %s\nRuntime message: %s\n", str,
+		     hsa_error_msg);
+}
+
+/* Like hsa_fatal, except only report error message, and return FALSE
+   for propagating error processing to outside of plugin.  */
+
+static bool
+hsa_error (const char *str, hsa_status_t status)
+{
+  const char *hsa_error_msg;
+  hsa_fns.hsa_status_string_fn (status, &hsa_error_msg);
+  GOMP_PLUGIN_error ("GCN fatal error: %s\nRuntime message: %s\n", str,
+		     hsa_error_msg);
+  return false;
+}
+
+struct hsa_kernel_description
+{
+  const char *name;
+  unsigned omp_data_size;
+  bool gridified_kernel_p;
+  unsigned kernel_dependencies_count;
+  const char **kernel_dependencies;
+  int oacc_dims[3];  /* Only present for GCN kernels.  */
+};
+
+struct global_var_info
+{
+  const char *name;
+  void *address;
+};
+
+/* Data passed by the static initializer of a compilation unit containing GCN
+   object code to GOMP_offload_register.  */
+
+struct gcn_image_desc
+{
+  union {
+    struct gcn_image {
+      char magic[4];  /* Will be "GCN" for GCN code objects.  */
+      size_t size;
+      void *image;
+    } *gcn_image;
+  };
+  const unsigned kernel_count;
+  struct hsa_kernel_description *kernel_infos;
+  const unsigned global_variable_count;
+  struct global_var_info *global_variables;
+};
+
+struct agent_info;
+
+/* Information required to identify, finalize and run any given kernel.  */
+
+struct kernel_info
+{
+  /* Name of the kernel, required to locate it within the GCN object-code
+     module.  */
+  const char *name;
+  /* Size of memory space for OMP data.  */
+  unsigned omp_data_size;
+  /* The specific agent the kernel has been or will be finalized for and run
+     on.  */
+  struct agent_info *agent;
+  /* The specific module where the kernel takes place.  */
+  struct module_info *module;
+  /* Mutex enforcing that at most once thread ever initializes a kernel for
+     use.  A thread should have locked agent->module_rwlock for reading before
+     acquiring it.  */
+  pthread_mutex_t init_mutex;
+  /* Flag indicating whether the kernel has been initialized and all fields
+     below it contain valid data.  */
+  bool initialized;
+  /* Flag indicating that the kernel has a problem that blocks an execution.  */
+  bool initialization_failed;
+  /* The object to be put into the dispatch queue.  */
+  uint64_t object;
+  /* Required size of kernel arguments.  */
+  uint32_t kernarg_segment_size;
+  /* Required size of group segment.  */
+  uint32_t group_segment_size;
+  /* Required size of private segment.  */
+  uint32_t private_segment_size;
+  /* List of all kernel dependencies.  */
+  const char **dependencies;
+  /* Number of dependencies.  */
+  unsigned dependencies_count;
+  /* Maximum OMP data size necessary for kernel from kernel dispatches.  */
+  unsigned max_omp_data_size;
+  /* True if the kernel is gridified.  */
+  bool gridified_kernel_p;
+};
+
+/* Information about a particular GCN module, its image and kernels.  */
+
+struct module_info
+{
+  /* The description with which the program has registered the image.  */
+  struct gcn_image_desc *image_desc;
+  /* GCN heap allocation.  */
+  struct heap *heap;
+  /* Physical boundaries of the loaded module.  */
+  Elf64_Addr phys_address_start;
+  Elf64_Addr phys_address_end;
+
+  bool constructors_run_p;
+  struct kernel_info *init_array_func, *fini_array_func;
+
+  /* Number of kernels in this module.  */
+  int kernel_count;
+  /* An array of kernel_info structures describing each kernel in this
+     module.  */
+  struct kernel_info kernels[];
+};
+
+/* Description of an HSA GPU agent and the program associated with it.  */
+
+struct agent_info
+{
+  /* The HSA ID of the agent.  Assigned when hsa_context is initialized.  */
+  hsa_agent_t id;
+  /* The user-visible device number.  */
+  int device_id;
+  /* Whether the agent has been initialized.  The fields below are usable only
+     if it has been.  */
+  bool initialized;
+  /* Precomuted check for problem architectures.  */
+  bool gfx900_p;
+
+  /* Command queues of the agent.  */
+  hsa_queue_t *sync_queue;
+  struct goacc_asyncqueue *async_queues, *omp_async_queue;
+  pthread_mutex_t async_queues_mutex;
+
+  /* The HSA memory region from which to allocate kernel arguments.  */
+  hsa_region_t kernarg_region;
+
+  /* Read-write lock that protects kernels which are running or about to be run
+     from interference with loading and unloading of images.  Needs to be
+     locked for reading while a kernel is being run, and for writing if the
+     list of modules is manipulated (and thus the HSA program invalidated).  */
+  pthread_rwlock_t module_rwlock;
+
+  /* The module associated with this kernel.  */
+  struct module_info *module;
+
+  /* Mutex enforcing that only one thread will finalize the HSA program.  A
+     thread should have locked agent->module_rwlock for reading before
+     acquiring it.  */
+  pthread_mutex_t prog_mutex;
+  /* Flag whether the HSA program that consists of all the modules has been
+     finalized.  */
+  bool prog_finalized;
+  /* HSA executable - the finalized program that is used to locate kernels.  */
+  hsa_executable_t executable;
+};
+
+static bool create_and_finalize_hsa_program (struct agent_info *);
+
+/* Information about the whole HSA environment and all of its agents.  */
+
+struct hsa_context_info
+{
+  /* Whether the structure has been initialized.  */
+  bool initialized;
+  /* Number of usable GPU HSA agents in the system.  */
+  int agent_count;
+  /* Array of agent_info structures describing the individual HSA agents.  */
+  struct agent_info *agents;
+};
+
+/* Information about the whole HSA environment and all of its agents.  */
+
+static struct hsa_context_info hsa_context;
+
+static bool
+init_hsa_runtime_functions (void)
+{
+#define DLSYM_FN(function) \
+  hsa_fns.function##_fn = dlsym (handle, #function); \
+  if (hsa_fns.function##_fn == NULL) \
+    return false;
+  void *handle = dlopen (hsa_runtime_lib, RTLD_LAZY);
+  if (handle == NULL)
+    return false;
+
+  DLSYM_FN (hsa_status_string)
+  DLSYM_FN (hsa_system_get_info)
+  DLSYM_FN (hsa_agent_get_info)
+  DLSYM_FN (hsa_init)
+  DLSYM_FN (hsa_iterate_agents)
+  DLSYM_FN (hsa_region_get_info)
+  DLSYM_FN (hsa_queue_create)
+  DLSYM_FN (hsa_agent_iterate_regions)
+  DLSYM_FN (hsa_executable_destroy)
+  DLSYM_FN (hsa_executable_create)
+  DLSYM_FN (hsa_executable_global_variable_define)
+  DLSYM_FN (hsa_executable_load_code_object)
+  DLSYM_FN (hsa_executable_freeze)
+  DLSYM_FN (hsa_signal_create)
+  DLSYM_FN (hsa_memory_allocate)
+  DLSYM_FN (hsa_memory_copy)
+  DLSYM_FN (hsa_memory_free)
+  DLSYM_FN (hsa_signal_destroy)
+  DLSYM_FN (hsa_executable_get_symbol)
+  DLSYM_FN (hsa_executable_symbol_get_info)
+  DLSYM_FN (hsa_executable_iterate_symbols)
+  DLSYM_FN (hsa_queue_add_write_index_release)
+  DLSYM_FN (hsa_queue_load_read_index_acquire)
+  DLSYM_FN (hsa_signal_wait_acquire)
+  DLSYM_FN (hsa_signal_store_relaxed)
+  DLSYM_FN (hsa_signal_store_release)
+  DLSYM_FN (hsa_signal_load_acquire)
+  DLSYM_FN (hsa_queue_destroy)
+  DLSYM_FN (hsa_code_object_deserialize)
+  return true;
+#undef DLSYM_FN
+}
+
+static void
+dump_hsa_system_info (void)
+{
+  hsa_status_t status;
+
+  hsa_endianness_t endianness;
+  status = hsa_fns.hsa_system_get_info_fn (HSA_SYSTEM_INFO_ENDIANNESS,
+					   &endianness);
+  if (status == HSA_STATUS_SUCCESS)
+    switch (endianness)
+      {
+      case HSA_ENDIANNESS_LITTLE:
+	HSA_DEBUG ("HSA_SYSTEM_INFO_ENDIANNESS: LITTLE\n");
+	break;
+      case HSA_ENDIANNESS_BIG:
+	HSA_DEBUG ("HSA_SYSTEM_INFO_ENDIANNESS: BIG\n");
+	break;
+      default:
+	HSA_DEBUG ("HSA_SYSTEM_INFO_ENDIANNESS: UNKNOWN\n");
+      }
+  else
+    HSA_DEBUG ("HSA_SYSTEM_INFO_ENDIANNESS: FAILED\n");
+
+  uint8_t extensions[128];
+  status = hsa_fns.hsa_system_get_info_fn (HSA_SYSTEM_INFO_EXTENSIONS,
+					   &extensions);
+  if (status == HSA_STATUS_SUCCESS)
+    {
+      if (extensions[0] & (1 << HSA_EXTENSION_IMAGES))
+	HSA_DEBUG ("HSA_SYSTEM_INFO_EXTENSIONS: IMAGES\n");
+    }
+  else
+    HSA_DEBUG ("HSA_SYSTEM_INFO_EXTENSIONS: FAILED\n");
+}
+
+static void
+dump_machine_model (hsa_machine_model_t machine_model, const char *s)
+{
+  switch (machine_model)
+    {
+    case HSA_MACHINE_MODEL_SMALL:
+      HSA_DEBUG ("%s: SMALL\n", s);
+      break;
+    case HSA_MACHINE_MODEL_LARGE:
+      HSA_DEBUG ("%s: LARGE\n", s);
+      break;
+    default:
+      HSA_DEBUG ("%s: UNKNOWN\n", s);
+      break;
+    }
+}
+
+static void
+dump_profile (hsa_profile_t profile, const char *s)
+{
+  switch (profile)
+    {
+    case HSA_PROFILE_FULL:
+      HSA_DEBUG ("%s: FULL\n", s);
+      break;
+    case HSA_PROFILE_BASE:
+      HSA_DEBUG ("%s: BASE\n", s);
+      break;
+    default:
+      HSA_DEBUG ("%s: UNKNOWN\n", s);
+      break;
+    }
+}
+
+static void dump_hsa_regions (hsa_agent_t agent);
+
+static hsa_status_t
+dump_hsa_agent_info (hsa_agent_t agent, void *data __attribute__((unused)))
+{
+  hsa_status_t status;
+
+  char buf[64];
+  status = hsa_fns.hsa_agent_get_info_fn (agent, HSA_AGENT_INFO_NAME,
+					  &buf);
+  if (status == HSA_STATUS_SUCCESS)
+    HSA_DEBUG ("HSA_AGENT_INFO_NAME: %s\n", buf);
+  else
+    HSA_DEBUG ("HSA_AGENT_INFO_NAME: FAILED\n");
+
+  status = hsa_fns.hsa_agent_get_info_fn (agent, HSA_AGENT_INFO_VENDOR_NAME,
+					  &buf);
+  if (status == HSA_STATUS_SUCCESS)
+    HSA_DEBUG ("HSA_AGENT_INFO_VENDOR_NAME: %s\n", buf);
+  else
+    HSA_DEBUG ("HSA_AGENT_INFO_VENDOR_NAME: FAILED\n");
+
+  hsa_machine_model_t machine_model;
+  status
+    = hsa_fns.hsa_agent_get_info_fn (agent, HSA_AGENT_INFO_MACHINE_MODEL,
+				     &machine_model);
+  if (status == HSA_STATUS_SUCCESS)
+    dump_machine_model (machine_model, "HSA_AGENT_INFO_MACHINE_MODEL");
+  else
+    HSA_DEBUG ("HSA_AGENT_INFO_MACHINE_MODEL: FAILED\n");
+
+  hsa_profile_t profile;
+  status = hsa_fns.hsa_agent_get_info_fn (agent, HSA_AGENT_INFO_PROFILE,
+					  &profile);
+  if (status == HSA_STATUS_SUCCESS)
+    dump_profile (profile, "HSA_AGENT_INFO_PROFILE");
+  else
+    HSA_DEBUG ("HSA_AGENT_INFO_PROFILE: FAILED\n");
+
+  hsa_device_type_t device_type;
+  status = hsa_fns.hsa_agent_get_info_fn (agent, HSA_AGENT_INFO_DEVICE,
+					  &device_type);
+  if (status == HSA_STATUS_SUCCESS)
+    {
+      switch (device_type)
+	{
+	case HSA_DEVICE_TYPE_CPU:
+	  HSA_DEBUG ("HSA_AGENT_INFO_DEVICE: CPU\n");
+	  break;
+	case HSA_DEVICE_TYPE_GPU:
+	  HSA_DEBUG ("HSA_AGENT_INFO_DEVICE: GPU\n");
+	  break;
+	case HSA_DEVICE_TYPE_DSP:
+	  HSA_DEBUG ("HSA_AGENT_INFO_DEVICE: DSP\n");
+	  break;
+	default:
+	  HSA_DEBUG ("HSA_AGENT_INFO_DEVICE: UNKNOWN\n");
+	  break;
+	}
+    }
+  else
+    HSA_DEBUG ("HSA_AGENT_INFO_DEVICE: FAILED\n");
+
+  uint32_t size;
+  status = hsa_fns.hsa_agent_get_info_fn (agent, HSA_AGENT_INFO_WAVEFRONT_SIZE,
+					  &size);
+  if (status == HSA_STATUS_SUCCESS)
+    HSA_DEBUG ("HSA_AGENT_INFO_WAVEFRONT_SIZE: %u\n", size);
+  else
+    HSA_DEBUG ("HSA_AGENT_INFO_WAVEFRONT_SIZE: FAILED\n");
+
+  uint32_t max_dim;
+  status = hsa_fns.hsa_agent_get_info_fn (agent,
+					  HSA_AGENT_INFO_WORKGROUP_MAX_DIM,
+					  &max_dim);
+  if (status == HSA_STATUS_SUCCESS)
+    HSA_DEBUG ("HSA_AGENT_INFO_WORKGROUP_MAX_DIM: %u\n", max_dim);
+  else
+    HSA_DEBUG ("HSA_AGENT_INFO_WORKGROUP_MAX_DIM: FAILED\n");
+
+  uint32_t max_size;
+  status = hsa_fns.hsa_agent_get_info_fn (agent,
+					  HSA_AGENT_INFO_WORKGROUP_MAX_SIZE,
+					  &max_size);
+  if (status == HSA_STATUS_SUCCESS)
+    HSA_DEBUG ("HSA_AGENT_INFO_WORKGROUP_MAX_SIZE: %u\n", max_size);
+  else
+    HSA_DEBUG ("HSA_AGENT_INFO_WORKGROUP_MAX_SIZE: FAILED\n");
+
+  uint32_t grid_max_dim;
+  status = hsa_fns.hsa_agent_get_info_fn (agent, HSA_AGENT_INFO_GRID_MAX_DIM,
+					  &grid_max_dim);
+  if (status == HSA_STATUS_SUCCESS)
+    HSA_DEBUG ("HSA_AGENT_INFO_GRID_MAX_DIM: %u\n", grid_max_dim);
+  else
+    HSA_DEBUG ("HSA_AGENT_INFO_GRID_MAX_DIM: FAILED\n");
+
+  uint32_t grid_max_size;
+  status = hsa_fns.hsa_agent_get_info_fn (agent, HSA_AGENT_INFO_GRID_MAX_SIZE,
+					  &grid_max_size);
+  if (status == HSA_STATUS_SUCCESS)
+    HSA_DEBUG ("HSA_AGENT_INFO_GRID_MAX_SIZE: %u\n", grid_max_size);
+  else
+    HSA_DEBUG ("HSA_AGENT_INFO_GRID_MAX_SIZE: FAILED\n");
+
+  dump_hsa_regions (agent);
+
+  return HSA_STATUS_SUCCESS;
+}
+
+/* Return true if the agent is a GPU and acceptable of concurrent submissions
+   from different threads.  */
+
+static bool
+suitable_hsa_agent_p (hsa_agent_t agent)
+{
+  hsa_device_type_t device_type;
+  hsa_status_t status
+    = hsa_fns.hsa_agent_get_info_fn (agent, HSA_AGENT_INFO_DEVICE,
+				     &device_type);
+  if (status != HSA_STATUS_SUCCESS)
+    return false;
+
+  switch (device_type)
+    {
+    case HSA_DEVICE_TYPE_GPU:
+      break;
+    case HSA_DEVICE_TYPE_CPU:
+      if (!support_cpu_devices)
+	return false;
+      break;
+    default:
+      return false;
+    }
+
+  uint32_t features = 0;
+  status = hsa_fns.hsa_agent_get_info_fn (agent, HSA_AGENT_INFO_FEATURE,
+					  &features);
+  if (status != HSA_STATUS_SUCCESS
+      || !(features & HSA_AGENT_FEATURE_KERNEL_DISPATCH))
+    return false;
+  hsa_queue_type_t queue_type;
+  status = hsa_fns.hsa_agent_get_info_fn (agent, HSA_AGENT_INFO_QUEUE_TYPE,
+					  &queue_type);
+  if (status != HSA_STATUS_SUCCESS
+      || (queue_type != HSA_QUEUE_TYPE_MULTI))
+    return false;
+
+  return true;
+}
+
+/* Callback of hsa_iterate_agents, if AGENT is a GPU device, increment
+   agent_count in hsa_context.  */
+
+static hsa_status_t
+count_gpu_agents (hsa_agent_t agent, void *data __attribute__ ((unused)))
+{
+  if (suitable_hsa_agent_p (agent))
+    hsa_context.agent_count++;
+  return HSA_STATUS_SUCCESS;
+}
+
+/* Callback of hsa_iterate_agents, if AGENT is a GPU device, assign the agent
+   id to the describing structure in the hsa context.  The index of the
+   structure is pointed to by DATA, increment it afterwards.  */
+
+static hsa_status_t
+assign_agent_ids (hsa_agent_t agent, void *data)
+{
+  if (suitable_hsa_agent_p (agent))
+    {
+      int *agent_index = (int *) data;
+      hsa_context.agents[*agent_index].id = agent;
+      ++*agent_index;
+    }
+  return HSA_STATUS_SUCCESS;
+}
+
+static void
+finalize_async_thread (struct goacc_asyncqueue *aq)
+{
+  pthread_mutex_lock (&aq->mutex);
+  if (aq->drain_queue_stop == 2)
+    {
+      pthread_mutex_unlock (&aq->mutex);
+      return;
+    }
+
+  aq->drain_queue_stop = 1;
+
+  if (DEBUG_THREAD_SIGNAL)
+    HSA_DEBUG ("Signalling async thread %d:%d: cond_in\n",
+	       aq->agent->device_id, aq->id);
+  pthread_cond_signal (&aq->queue_cond_in);
+
+  while (aq->drain_queue_stop != 2)
+    {
+      if (DEBUG_THREAD_SLEEP)
+	HSA_DEBUG ("Waiting for async thread %d:%d to finish, putting thread"
+		   " to sleep\n", aq->agent->device_id, aq->id);
+      pthread_cond_wait (&aq->queue_cond_out, &aq->mutex);
+      if (DEBUG_THREAD_SLEEP)
+	HSA_DEBUG ("Waiting, woke up thread %d:%d.  Rechecking\n",
+		   aq->agent->device_id, aq->id);
+    }
+
+  HSA_DEBUG ("Done waiting for async thread %d:%d\n", aq->agent->device_id,
+	     aq->id);
+  pthread_mutex_unlock (&aq->mutex);
+
+  int err = pthread_join (aq->thread_drain_queue, NULL);
+  if (err != 0)
+    GOMP_PLUGIN_fatal ("Join async thread %d:%d: failed: %s",
+		       aq->agent->device_id, aq->id, strerror (err));
+  HSA_DEBUG ("Joined with async thread %d:%d\n", aq->agent->device_id, aq->id);
+}
+
+/* Initialize hsa_context if it has not already been done.
+   Return TRUE on success.  */
+
+static bool
+init_hsa_context (void)
+{
+  hsa_status_t status;
+  int agent_index = 0;
+
+  if (hsa_context.initialized)
+    return true;
+  init_environment_variables ();
+  if (!init_hsa_runtime_functions ())
+    {
+      HSA_DEBUG ("Run-time could not be dynamically opened\n");
+      if (suppress_host_fallback)
+	GOMP_PLUGIN_fatal ("GCN host fallback has been suppressed");
+      return false;
+    }
+  status = hsa_fns.hsa_init_fn ();
+  if (status != HSA_STATUS_SUCCESS)
+    return hsa_error ("Run-time could not be initialized", status);
+  HSA_DEBUG ("HSA run-time initialized for GCN\n");
+
+  if (debug)
+    dump_hsa_system_info ();
+
+  status = hsa_fns.hsa_iterate_agents_fn (count_gpu_agents, NULL);
+  if (status != HSA_STATUS_SUCCESS)
+    return hsa_error ("GCN GPU devices could not be enumerated", status);
+  HSA_DEBUG ("There are %i GCN GPU devices.\n", hsa_context.agent_count);
+
+  hsa_context.agents
+    = GOMP_PLUGIN_malloc_cleared (hsa_context.agent_count
+				  * sizeof (struct agent_info));
+  status = hsa_fns.hsa_iterate_agents_fn (assign_agent_ids, &agent_index);
+  if (agent_index != hsa_context.agent_count)
+    {
+      GOMP_PLUGIN_error ("Failed to assign IDs to all GCN agents");
+      return false;
+    }
+
+  if (debug)
+    {
+      status = hsa_fns.hsa_iterate_agents_fn (dump_hsa_agent_info, NULL);
+      if (status != HSA_STATUS_SUCCESS)
+	GOMP_PLUGIN_error ("Failed to list all HSA runtime agents");
+    }
+
+  hsa_context.initialized = true;
+  return true;
+}
+
+/* Verify that hsa_context has already been initialized and return the
+   agent_info structure describing device number N.  Return NULL on error.  */
+
+static struct agent_info *
+get_agent_info (int n)
+{
+  if (!hsa_context.initialized)
+    {
+      GOMP_PLUGIN_error ("Attempt to use uninitialized GCN context.");
+      return NULL;
+    }
+  if (n >= hsa_context.agent_count)
+    {
+      GOMP_PLUGIN_error ("Request to operate on non-existent GCN device %i", n);
+      return NULL;
+    }
+  if (!hsa_context.agents[n].initialized)
+    {
+      GOMP_PLUGIN_error ("Attempt to use an uninitialized GCN agent.");
+      return NULL;
+    }
+  return &hsa_context.agents[n];
+}
+
+/* Callback of dispatch queues to report errors.  */
+
+static void
+queue_callback (hsa_status_t status,
+		hsa_queue_t *queue __attribute__ ((unused)),
+		void *data __attribute__ ((unused)))
+{
+  hsa_fatal ("Asynchronous queue error", status);
+}
+
+static hsa_status_t
+dump_hsa_region (hsa_region_t region, void *data __attribute__((unused)))
+{
+  hsa_status_t status;
+
+  hsa_region_segment_t segment;
+  status = hsa_fns.hsa_region_get_info_fn (region, HSA_REGION_INFO_SEGMENT,
+					   &segment);
+  if (status == HSA_STATUS_SUCCESS)
+    {
+      if (segment == HSA_REGION_SEGMENT_GLOBAL)
+	HSA_DEBUG ("HSA_REGION_INFO_SEGMENT: GLOBAL\n");
+      else if (segment == HSA_REGION_SEGMENT_READONLY)
+	HSA_DEBUG ("HSA_REGION_INFO_SEGMENT: READONLY\n");
+      else if (segment == HSA_REGION_SEGMENT_PRIVATE)
+	HSA_DEBUG ("HSA_REGION_INFO_SEGMENT: PRIVATE\n");
+      else if (segment == HSA_REGION_SEGMENT_GROUP)
+	HSA_DEBUG ("HSA_REGION_INFO_SEGMENT: GROUP\n");
+      else
+	HSA_DEBUG ("HSA_REGION_INFO_SEGMENT: UNKNOWN\n");
+    }
+  else
+    HSA_DEBUG ("HSA_REGION_INFO_SEGMENT: FAILED\n");
+
+  if (segment == HSA_REGION_SEGMENT_GLOBAL)
+    {
+      uint32_t flags;
+      status
+	= hsa_fns.hsa_region_get_info_fn (region, HSA_REGION_INFO_GLOBAL_FLAGS,
+					  &flags);
+      if (status == HSA_STATUS_SUCCESS)
+	{
+	  if (flags & HSA_REGION_GLOBAL_FLAG_KERNARG)
+	    HSA_DEBUG ("HSA_REGION_INFO_GLOBAL_FLAGS: KERNARG\n");
+	  if (flags & HSA_REGION_GLOBAL_FLAG_FINE_GRAINED)
+	    HSA_DEBUG ("HSA_REGION_INFO_GLOBAL_FLAGS: FINE_GRAINED\n");
+	  if (flags & HSA_REGION_GLOBAL_FLAG_COARSE_GRAINED)
+	    HSA_DEBUG ("HSA_REGION_INFO_GLOBAL_FLAGS: COARSE_GRAINED\n");
+	}
+      else
+	HSA_DEBUG ("HSA_REGION_INFO_GLOBAL_FLAGS: FAILED\n");
+    }
+
+  size_t size;
+  status = hsa_fns.hsa_region_get_info_fn (region, HSA_REGION_INFO_SIZE, &size);
+  if (status == HSA_STATUS_SUCCESS)
+    HSA_DEBUG ("HSA_REGION_INFO_SIZE: %zu\n", size);
+  else
+    HSA_DEBUG ("HSA_REGION_INFO_SIZE: FAILED\n");
+
+  status
+    = hsa_fns.hsa_region_get_info_fn (region, HSA_REGION_INFO_ALLOC_MAX_SIZE,
+				      &size);
+  if (status == HSA_STATUS_SUCCESS)
+    HSA_DEBUG ("HSA_REGION_INFO_ALLOC_MAX_SIZE: %zu\n", size);
+  else
+    HSA_DEBUG ("HSA_REGION_INFO_ALLOC_MAX_SIZE: FAILED\n");
+
+  bool alloc_allowed;
+  status
+    = hsa_fns.hsa_region_get_info_fn (region,
+				      HSA_REGION_INFO_RUNTIME_ALLOC_ALLOWED,
+				      &alloc_allowed);
+  if (status == HSA_STATUS_SUCCESS)
+    HSA_DEBUG ("HSA_REGION_INFO_RUNTIME_ALLOC_ALLOWED: %u\n", alloc_allowed);
+  else
+    HSA_DEBUG ("HSA_REGION_INFO_RUNTIME_ALLOC_ALLOWED: FAILED\n");
+
+  if (status != HSA_STATUS_SUCCESS || !alloc_allowed)
+    return HSA_STATUS_SUCCESS;
+
+  status
+    = hsa_fns.hsa_region_get_info_fn (region,
+				      HSA_REGION_INFO_RUNTIME_ALLOC_GRANULE,
+				      &size);
+  if (status == HSA_STATUS_SUCCESS)
+    HSA_DEBUG ("HSA_REGION_INFO_RUNTIME_ALLOC_GRANULE: %zu\n", size);
+  else
+    HSA_DEBUG ("HSA_REGION_INFO_RUNTIME_ALLOC_GRANULE: FAILED\n");
+
+  size_t align;
+  status
+    = hsa_fns.hsa_region_get_info_fn (region,
+				      HSA_REGION_INFO_RUNTIME_ALLOC_ALIGNMENT,
+				      &align);
+  if (status == HSA_STATUS_SUCCESS)
+    HSA_DEBUG ("HSA_REGION_INFO_RUNTIME_ALLOC_ALIGNMENT: %zu\n", align);
+  else
+    HSA_DEBUG ("HSA_REGION_INFO_RUNTIME_ALLOC_ALIGNMENT: FAILED\n");
+
+  return HSA_STATUS_SUCCESS;
+}
+
+static void
+dump_hsa_regions (hsa_agent_t agent)
+{
+  hsa_status_t status;
+  status = hsa_fns.hsa_agent_iterate_regions_fn (agent,
+						 dump_hsa_region,
+						 NULL);
+  if (status != HSA_STATUS_SUCCESS)
+    hsa_error ("Dumping hsa regions failed", status);
+}
+
+/* Return malloc'd string with name of SYMBOL.  */
+
+static char *
+get_executable_symbol_name (hsa_executable_symbol_t symbol)
+{
+  hsa_status_t status;
+  char *res;
+  uint32_t len;
+  const hsa_executable_symbol_info_t info_name_length
+    = HSA_EXECUTABLE_SYMBOL_INFO_NAME_LENGTH;
+
+  status = hsa_fns.hsa_executable_symbol_get_info_fn (symbol, info_name_length,
+						      &len);
+  if (status != HSA_STATUS_SUCCESS)
+    {
+      hsa_error ("Could not get length of symbol name", status);
+      return NULL;
+    }
+
+  res = GOMP_PLUGIN_malloc (len + 1);
+
+  const hsa_executable_symbol_info_t info_name
+    = HSA_EXECUTABLE_SYMBOL_INFO_NAME;
+
+  status = hsa_fns.hsa_executable_symbol_get_info_fn (symbol, info_name, res);
+
+  if (status != HSA_STATUS_SUCCESS)
+    {
+      hsa_error ("Could not get symbol name", status);
+      free (res);
+      return NULL;
+    }
+
+  res[len] = '\0';
+
+  return res;
+}
+
+/* Helper function for dump_executable_symbols.  */
+
+static hsa_status_t
+dump_executable_symbol (hsa_executable_t executable,
+			hsa_executable_symbol_t symbol,
+			void *data __attribute__((unused)))
+{
+  char *name = get_executable_symbol_name (symbol);
+
+  if (name)
+    {
+      HSA_DEBUG ("executable symbol: %s\n", name);
+      free (name);
+    }
+
+  return HSA_STATUS_SUCCESS;
+}
+
+/* Dump all global symbol in executable.  */
+
+static void
+dump_executable_symbols (hsa_executable_t executable)
+{
+  hsa_status_t status;
+  status
+    = hsa_fns.hsa_executable_iterate_symbols_fn (executable,
+						 dump_executable_symbol,
+						 NULL);
+  if (status != HSA_STATUS_SUCCESS)
+    hsa_fatal ("Could not dump HSA executable symbols", status);
+}
+
+/* Helper function for find_executable_symbol.  */
+
+static hsa_status_t
+find_executable_symbol_1 (hsa_executable_t executable,
+			  hsa_executable_symbol_t symbol,
+			  void *data)
+{
+  hsa_executable_symbol_t *res = (hsa_executable_symbol_t *)data;
+  *res = symbol;
+  return HSA_STATUS_INFO_BREAK;
+}
+
+/* Find a global symbol in EXECUTABLE, save to *SYMBOL and return true.  If not
+   found, return false.  */
+
+static bool
+find_executable_symbol (hsa_executable_t executable,
+			hsa_executable_symbol_t *symbol)
+{
+  hsa_status_t status;
+
+  status
+    = hsa_fns.hsa_executable_iterate_symbols_fn (executable,
+						 find_executable_symbol_1,
+						 symbol);
+  if (status != HSA_STATUS_INFO_BREAK)
+    {
+      hsa_error ("Could not find executable symbol", status);
+      return false;
+    }
+
+  return true;
+}
+
+/* Callback of hsa_agent_iterate_regions.  Determine if a memory REGION can be
+   used for kernarg allocations and if so write it to the memory pointed to by
+   DATA and break the query.  */
+
+static hsa_status_t
+get_kernarg_memory_region (hsa_region_t region, void *data)
+{
+  hsa_status_t status;
+  hsa_region_segment_t segment;
+
+  status = hsa_fns.hsa_region_get_info_fn (region, HSA_REGION_INFO_SEGMENT,
+					   &segment);
+  if (status != HSA_STATUS_SUCCESS)
+    return status;
+  if (segment != HSA_REGION_SEGMENT_GLOBAL)
+    return HSA_STATUS_SUCCESS;
+
+  uint32_t flags;
+  status = hsa_fns.hsa_region_get_info_fn (region, HSA_REGION_INFO_GLOBAL_FLAGS,
+					   &flags);
+  if (status != HSA_STATUS_SUCCESS)
+    return status;
+  if (flags & HSA_REGION_GLOBAL_FLAG_KERNARG)
+    {
+      hsa_region_t *ret = (hsa_region_t *) data;
+      *ret = region;
+      return HSA_STATUS_INFO_BREAK;
+    }
+  return HSA_STATUS_SUCCESS;
+}
+
+/* Part of the libgomp plugin interface.  Return the number of HSA devices on
+   the system.  */
+
+int
+GOMP_OFFLOAD_get_num_devices (void)
+{
+  if (!init_hsa_context ())
+    return 0;
+  return hsa_context.agent_count;
+}
+
+union gomp_device_property_value
+GOMP_OFFLOAD_get_property (int device, int prop)
+{
+  struct agent_info *agent = get_agent_info (device);
+  hsa_region_t region = agent->kernarg_region;
+
+  union gomp_device_property_value propval = { .val = 0 };
+
+  static char buf[64];
+  buf[0] = '\0';
+  size_t size;
+  hsa_status_t status;
+
+  switch (prop)
+    {
+    case GOMP_DEVICE_PROPERTY_FREE_MEMORY:
+      /* Not known: fall through.  */
+    case GOMP_DEVICE_PROPERTY_MEMORY:
+      status = hsa_fns.hsa_region_get_info_fn (region, HSA_REGION_INFO_SIZE,
+					       &size);
+      propval.val = size;
+      break;
+      break;
+    case GOMP_DEVICE_PROPERTY_NAME:
+      status = hsa_fns.hsa_agent_get_info_fn (agent->id, HSA_AGENT_INFO_NAME,
+					      buf);
+      if (status == HSA_STATUS_SUCCESS)
+	propval.ptr = buf;
+      break;
+    case GOMP_DEVICE_PROPERTY_VENDOR:
+      status = hsa_fns.hsa_agent_get_info_fn (agent->id,
+					      HSA_AGENT_INFO_VENDOR_NAME,
+					      buf);
+      if (status == HSA_STATUS_SUCCESS)
+	propval.ptr = buf;
+      break;
+    case GOMP_DEVICE_PROPERTY_DRIVER:
+      propval.ptr = "HSA Runtime";
+      break;
+    }
+
+  return propval;
+}
+
+static void
+queue_push_launch (struct goacc_asyncqueue *aq, struct kernel_info *kernel,
+		   void *vars, struct GOMP_kernel_launch_attributes *kla)
+{
+  assert (aq->agent == kernel->agent);
+
+  if (aq->queue_n == ASYNC_QUEUE_SIZE)
+    GOMP_PLUGIN_fatal ("ran out of async queue in thread %d:%d",
+		       aq->agent->device_id, aq->id);
+
+  pthread_mutex_lock (&aq->mutex);
+
+  int queue_last = ((aq->queue_first + aq->queue_n)
+		    % ASYNC_QUEUE_SIZE);
+  if (DEBUG_QUEUES)
+    HSA_DEBUG ("queue_push_launch %d:%d: at %i\n", aq->agent->device_id,
+	       aq->id, queue_last);
+
+  aq->queue[queue_last].type = 0;
+  aq->queue[queue_last].u.launch.kernel = kernel;
+  aq->queue[queue_last].u.launch.vars = vars;
+  aq->queue[queue_last].u.launch.kla = *kla;
+
+  aq->queue_n++;
+
+  if (DEBUG_THREAD_SIGNAL)
+    HSA_DEBUG ("signalling async thread %d:%d: cond_in\n",
+	       aq->agent->device_id, aq->id);
+  pthread_cond_signal (&aq->queue_cond_in);
+
+  pthread_mutex_unlock (&aq->mutex);
+}
+
+static void
+queue_push_callback (struct goacc_asyncqueue *aq, void (*fn)(void *),
+		     void *data)
+{
+  if (aq->queue_n == ASYNC_QUEUE_SIZE)
+    GOMP_PLUGIN_fatal ("Async thread %d:%d: error: queue overflowed",
+		       aq->agent->device_id, aq->id);
+
+  pthread_mutex_lock (&aq->mutex);
+
+  int queue_last = ((aq->queue_first + aq->queue_n)
+		    % ASYNC_QUEUE_SIZE);
+  if (DEBUG_QUEUES)
+    HSA_DEBUG ("queue_push_callback %d:%d: at %i\n", aq->agent->device_id,
+	       aq->id, queue_last);
+
+  aq->queue[queue_last].type = 1;
+  aq->queue[queue_last].u.callback.fn = fn;
+  aq->queue[queue_last].u.callback.data = data;
+
+  aq->queue_n++;
+
+  if (DEBUG_THREAD_SIGNAL)
+    HSA_DEBUG ("signalling async thread %d:%d: cond_in\n",
+	       aq->agent->device_id, aq->id);
+  pthread_cond_signal (&aq->queue_cond_in);
+
+  pthread_mutex_unlock (&aq->mutex);
+}
+
+static void run_kernel (struct kernel_info *kernel, void *vars,
+			struct GOMP_kernel_launch_attributes *kla,
+			struct goacc_asyncqueue *aq, bool module_locked);
+
+static void
+execute_queue_entry (struct goacc_asyncqueue *aq, int index)
+{
+  struct queue_entry *entry = &aq->queue[index];
+  if (entry->type == 0)
+    {
+      if (DEBUG_QUEUES)
+	HSA_DEBUG ("Async thread %d:%d: Executing launch entry (%d)\n",
+		   aq->agent->device_id, aq->id, index);
+      run_kernel (entry->u.launch.kernel,
+		  entry->u.launch.vars,
+		  &entry->u.launch.kla, aq, false);
+      if (DEBUG_QUEUES)
+	HSA_DEBUG ("Async thread %d:%d: Executing launch entry (%d) done\n",
+		   aq->agent->device_id, aq->id, index);
+    }
+  else if (entry->type == 1)
+    {
+      if (DEBUG_QUEUES)
+	HSA_DEBUG ("Async thread %d:%d: Executing callback entry (%d)\n",
+		   aq->agent->device_id, aq->id, index);
+      entry->u.callback.fn (entry->u.callback.data);
+      if (DEBUG_QUEUES)
+	HSA_DEBUG ("Async thread %d:%d: Executing callback entry (%d) done\n",
+		   aq->agent->device_id, aq->id, index);
+    }
+  else
+    GOMP_PLUGIN_fatal ("Unknown queue element");
+}
+
+static void *
+drain_queue (void *thread_arg)
+{
+  struct goacc_asyncqueue *aq = thread_arg;
+
+  if (DRAIN_QUEUE_SYNCHRONOUS_P)
+    {
+      aq->drain_queue_stop = 2;
+      return NULL;
+    }
+
+  pthread_mutex_lock (&aq->mutex);
+
+  while (true)
+    {
+      if (aq->drain_queue_stop)
+	break;
+
+      if (aq->queue_n > 0)
+	{
+	  pthread_mutex_unlock (&aq->mutex);
+	  execute_queue_entry (aq, aq->queue_first);
+
+	  pthread_mutex_lock (&aq->mutex);
+	  aq->queue_first = ((aq->queue_first + 1)
+			     % ASYNC_QUEUE_SIZE);
+	  aq->queue_n--;
+
+	  if (DEBUG_THREAD_SIGNAL)
+	    HSA_DEBUG ("Async thread %d:%d: broadcasting queue out update\n",
+		       aq->agent->device_id, aq->id);
+	  pthread_cond_broadcast (&aq->queue_cond_out);
+	  pthread_mutex_unlock (&aq->mutex);
+
+	  if (DEBUG_QUEUES)
+	    HSA_DEBUG ("Async thread %d:%d: continue\n", aq->agent->device_id,
+		       aq->id);
+	  pthread_mutex_lock (&aq->mutex);
+	}
+      else
+	{
+	  if (DEBUG_THREAD_SLEEP)
+	    HSA_DEBUG ("Async thread %d:%d: going to sleep\n",
+		       aq->agent->device_id, aq->id);
+	  pthread_cond_wait (&aq->queue_cond_in, &aq->mutex);
+	  if (DEBUG_THREAD_SLEEP)
+	    HSA_DEBUG ("Async thread %d:%d: woke up, rechecking\n",
+		       aq->agent->device_id, aq->id);
+	}
+    }
+
+  aq->drain_queue_stop = 2;
+  if (DEBUG_THREAD_SIGNAL)
+    HSA_DEBUG ("Async thread %d:%d: broadcasting last queue out update\n",
+	       aq->agent->device_id, aq->id);
+  pthread_cond_broadcast (&aq->queue_cond_out);
+  pthread_mutex_unlock (&aq->mutex);
+
+  HSA_DEBUG ("Async thread %d:%d: returning\n", aq->agent->device_id, aq->id);
+  return NULL;
+}
+
+static void
+drain_queue_synchronous (struct goacc_asyncqueue *aq)
+{
+  pthread_mutex_lock (&aq->mutex);
+
+  while (aq->queue_n > 0)
+    {
+      execute_queue_entry (aq, aq->queue_first);
+
+      aq->queue_first = ((aq->queue_first + 1)
+			 % ASYNC_QUEUE_SIZE);
+      aq->queue_n--;
+    }
+
+  pthread_mutex_unlock (&aq->mutex);
+}
+
+/* Part of the libgomp plugin interface.  Initialize agent number N so that it
+   can be used for computation.  Return TRUE on success.  */
+
+bool
+GOMP_OFFLOAD_init_device (int n)
+{
+  if (!init_hsa_context ())
+    return false;
+  if (n >= hsa_context.agent_count)
+    {
+      GOMP_PLUGIN_error ("Request to initialize non-existent GCN device %i", n);
+      return false;
+    }
+  struct agent_info *agent = &hsa_context.agents[n];
+
+  if (agent->initialized)
+    return true;
+
+  agent->device_id = n;
+
+  if (pthread_rwlock_init (&agent->module_rwlock, NULL))
+    {
+      GOMP_PLUGIN_error ("Failed to initialize a GCN agent rwlock");
+      return false;
+    }
+  if (pthread_mutex_init (&agent->prog_mutex, NULL))
+    {
+      GOMP_PLUGIN_error ("Failed to initialize a GCN agent program mutex");
+      return false;
+    }
+  if (pthread_mutex_init (&agent->async_queues_mutex, NULL))
+    {
+      GOMP_PLUGIN_error ("Failed to initialize a GCN agent queue mutex");
+      return false;
+    }
+  agent->async_queues = NULL;
+  agent->omp_async_queue = NULL;
+
+  uint32_t queue_size;
+  hsa_status_t status;
+  status = hsa_fns.hsa_agent_get_info_fn (agent->id,
+					  HSA_AGENT_INFO_QUEUE_MAX_SIZE,
+					  &queue_size);
+  if (status != HSA_STATUS_SUCCESS)
+    return hsa_error ("Error requesting maximum queue size of the GCN agent",
+		      status);
+
+  char buf[64];
+  status = hsa_fns.hsa_agent_get_info_fn (agent->id, HSA_AGENT_INFO_NAME,
+					  &buf);
+  if (status != HSA_STATUS_SUCCESS)
+    return hsa_error ("Error querying the name of the agent", status);
+  agent->gfx900_p = (strncmp (buf, "gfx900", 6) == 0);
+
+  status = hsa_fns.hsa_queue_create_fn (agent->id, queue_size,
+					HSA_QUEUE_TYPE_MULTI, queue_callback,
+					NULL, UINT32_MAX, UINT32_MAX,
+					&agent->sync_queue);
+  if (status != HSA_STATUS_SUCCESS)
+    return hsa_error ("Error creating command queue", status);
+
+  agent->kernarg_region.handle = (uint64_t) -1;
+  status = hsa_fns.hsa_agent_iterate_regions_fn (agent->id,
+						 get_kernarg_memory_region,
+						 &agent->kernarg_region);
+  if (agent->kernarg_region.handle == (uint64_t) -1)
+    {
+      GOMP_PLUGIN_error ("Could not find suitable memory region for kernel "
+			 "arguments");
+      return false;
+    }
+  HSA_DEBUG ("Selected kernel arguments memory region:\n");
+  dump_hsa_region (agent->kernarg_region, NULL);
+
+  HSA_DEBUG ("GCN agent %d initialized\n", n);
+
+  agent->initialized = true;
+  return true;
+}
+
+/* Free the HSA program in agent and everything associated with it and set
+   agent->prog_finalized and the initialized flags of all kernels to false.
+   Return TRUE on success.  */
+
+static bool
+destroy_hsa_program (struct agent_info *agent)
+{
+  if (!agent->prog_finalized)
+    return true;
+
+  hsa_status_t status;
+
+  HSA_DEBUG ("Destroying the current GCN program.\n");
+
+  status = hsa_fns.hsa_executable_destroy_fn (agent->executable);
+  if (status != HSA_STATUS_SUCCESS)
+    return hsa_error ("Could not destroy GCN executable", status);
+
+  if (agent->module)
+    {
+      int i;
+      for (i = 0; i < agent->module->kernel_count; i++)
+	agent->module->kernels[i].initialized = false;
+
+      if (agent->module->heap)
+	{
+	  hsa_fns.hsa_memory_free_fn (agent->module->heap);
+	  agent->module->heap = NULL;
+	}
+    }
+  agent->prog_finalized = false;
+  return true;
+}
+
+/* Initialize KERNEL from D and other parameters.  Return true on success. */
+
+static bool
+init_basic_kernel_info (struct kernel_info *kernel,
+			struct hsa_kernel_description *d,
+			struct agent_info *agent,
+			struct module_info *module)
+{
+  kernel->agent = agent;
+  kernel->module = module;
+  kernel->name = d->name;
+  kernel->omp_data_size = d->omp_data_size;
+  kernel->gridified_kernel_p = d->gridified_kernel_p;
+  kernel->dependencies_count = d->kernel_dependencies_count;
+  kernel->dependencies = d->kernel_dependencies;
+  if (pthread_mutex_init (&kernel->init_mutex, NULL))
+    {
+      GOMP_PLUGIN_error ("Failed to initialize a GCN kernel mutex");
+      return false;
+    }
+  return true;
+}
+
+static void init_kernel (struct kernel_info *kernel);
+
+/* Part of the libgomp plugin interface.  Load GCN object-code module
+   described by struct gcn_image_desc in TARGET_DATA and return references to
+   kernel descriptors in TARGET_TABLE.  */
+
+int
+GOMP_OFFLOAD_load_image (int ord, unsigned version, const void *target_data,
+			 struct addr_pair **target_table)
+{
+  if (GOMP_VERSION_DEV (version) > GOMP_VERSION_GCN)
+    {
+      GOMP_PLUGIN_error ("Offload data incompatible with GCN plugin"
+			 " (expected %u, received %u)",
+			 GOMP_VERSION_GCN, GOMP_VERSION_DEV (version));
+      return -1;
+    }
+
+  struct gcn_image_desc *image_desc = (struct gcn_image_desc *) target_data;
+  struct agent_info *agent;
+  struct addr_pair *pair;
+  struct module_info *module;
+  struct kernel_info *kernel;
+  int kernel_count = image_desc->kernel_count;
+  unsigned var_count = image_desc->global_variable_count;
+
+  agent = get_agent_info (ord);
+  if (!agent)
+    return -1;
+
+  if (pthread_rwlock_wrlock (&agent->module_rwlock))
+    {
+      GOMP_PLUGIN_error ("Unable to write-lock a GCN agent rwlock");
+      return -1;
+    }
+  if (agent->prog_finalized
+      && !destroy_hsa_program (agent))
+    return -1;
+
+  HSA_DEBUG ("Encountered %d kernels in an image\n", kernel_count);
+  HSA_DEBUG ("Encountered %u global variables in an image\n", var_count);
+  pair = GOMP_PLUGIN_malloc ((kernel_count + var_count - 2)
+			     * sizeof (struct addr_pair));
+  *target_table = pair;
+  module = (struct module_info *)
+    GOMP_PLUGIN_malloc_cleared (sizeof (struct module_info)
+				+ kernel_count * sizeof (struct kernel_info));
+  module->image_desc = image_desc;
+  module->kernel_count = kernel_count;
+  module->heap = NULL;
+  module->constructors_run_p = false;
+
+  kernel = &module->kernels[0];
+
+  /* We have the magic code for a native GCN ELF kernel, not something
+     else.  */
+  if (strcmp (image_desc->gcn_image->magic, "GCN") != 0)
+    return -1;
+
+  /* Allocate memory for kernel dependencies.  */
+  for (unsigned i = 0; i < kernel_count; i++)
+    {
+      struct hsa_kernel_description *d = &image_desc->kernel_infos[i];
+      if (!init_basic_kernel_info (kernel, d, agent, module))
+	return -1;
+      if (strcmp (d->name, "_init_array") == 0)
+	module->init_array_func = kernel;
+      else if (strcmp (d->name, "_fini_array") == 0)
+        module->fini_array_func = kernel;
+      else
+	{
+	  pair->start = (uintptr_t) kernel;
+	  pair->end = (uintptr_t) (kernel + 1);
+	  pair++;
+	}
+      kernel++;
+    }
+
+  agent->module = module;
+  if (pthread_rwlock_unlock (&agent->module_rwlock))
+    {
+      GOMP_PLUGIN_error ("Unable to unlock a GCN agent rwlock");
+      return -1;
+    }
+
+  if (!create_and_finalize_hsa_program (agent))
+    return -1;
+
+  for (unsigned i = 0; i < var_count; i++)
+    {
+      struct global_var_info *v = &image_desc->global_variables[i];
+      HSA_DEBUG ("Looking for variable %s\n", v->name);
+
+      hsa_status_t status;
+      hsa_executable_symbol_t var_symbol;
+      status = hsa_fns.hsa_executable_get_symbol_fn (agent->executable, NULL,
+						     v->name, agent->id,
+						     0, &var_symbol);
+
+      if (status != HSA_STATUS_SUCCESS)
+	hsa_fatal ("Could not find symbol for variable in the code object",
+		   status);
+
+      uint64_t var_addr;
+      uint32_t var_size;
+      status = hsa_fns.hsa_executable_symbol_get_info_fn
+	(var_symbol, HSA_EXECUTABLE_SYMBOL_INFO_VARIABLE_ADDRESS, &var_addr);
+      if (status != HSA_STATUS_SUCCESS)
+	hsa_fatal ("Could not extract a variable from its symbol", status);
+      status = hsa_fns.hsa_executable_symbol_get_info_fn
+	(var_symbol, HSA_EXECUTABLE_SYMBOL_INFO_VARIABLE_SIZE, &var_size);
+      if (status != HSA_STATUS_SUCCESS)
+	hsa_fatal ("Could not extract a variable size from its symbol", status);
+
+      pair->start = var_addr;
+      pair->end = var_addr + var_size;
+      HSA_DEBUG ("Found variable %s at %p with size %u\n", v->name,
+		 (void *)var_addr, var_size);
+      pair++;
+    }
+
+  /* Ensure that constructors are run first.  */
+  struct GOMP_kernel_launch_attributes kla =
+    { 3,
+      /* Grid size.  */
+      { 1, 64, 1 },
+      /* Work-group size.  */
+      { 1, 64, 1 }
+    };
+
+  if (module->init_array_func)
+    {
+      init_kernel (module->init_array_func);
+      run_kernel (module->init_array_func, NULL, &kla, NULL, false);
+    }
+  module->constructors_run_p = true;
+
+  return kernel_count + var_count;
+}
+
+/* Find the load_offset for MODULE, savte to *LOAD_OFFSET, and return true.  If
+   not found, return false.  */
+
+static bool
+find_load_offset (Elf64_Addr *load_offset, struct agent_info *agent,
+		  struct module_info *module, Elf64_Ehdr *image,
+		  Elf64_Shdr *sections)
+{
+  bool res = false;
+
+  hsa_status_t status;
+
+  hsa_executable_symbol_t symbol;
+  if (!find_executable_symbol (agent->executable, &symbol))
+    return false;
+
+  status = hsa_fns.hsa_executable_symbol_get_info_fn
+    (symbol, HSA_EXECUTABLE_SYMBOL_INFO_VARIABLE_ADDRESS, load_offset);
+  if (status != HSA_STATUS_SUCCESS)
+    {
+      hsa_error ("Could not extract symbol address", status);
+      return false;
+    }
+
+  char *symbol_name = get_executable_symbol_name (symbol);
+  if (symbol_name == NULL)
+    return false;
+
+  /* Find the kernel function in ELF, and calculate actual load offset.  */
+  for (int i = 0; i < image->e_shnum; i++)
+    if (sections[i].sh_type == SHT_SYMTAB)
+      {
+	Elf64_Shdr *strtab = &sections[sections[i].sh_link];
+	char *strings = (char *)image + strtab->sh_offset;
+
+	for (size_t offset = 0;
+	     offset < sections[i].sh_size;
+	     offset += sections[i].sh_entsize)
+	  {
+	    Elf64_Sym *sym = (Elf64_Sym*)((char*)image
+					  + sections[i].sh_offset
+					  + offset);
+	    if (strcmp (symbol_name, strings + sym->st_name) == 0)
+	      {
+		*load_offset -= sym->st_value;
+		res = true;
+		break;
+	      }
+	  }
+      }
+
+  free (symbol_name);
+  return res;
+}
+
+/* Create and finalize the program consisting of all loaded modules.  */
+
+static bool
+create_and_finalize_hsa_program (struct agent_info *agent)
+{
+  hsa_status_t status;
+  int reloc_count = 0;
+  bool res = true;
+  if (pthread_mutex_lock (&agent->prog_mutex))
+    {
+      GOMP_PLUGIN_error ("Could not lock a GCN agent program mutex");
+      return false;
+    }
+  if (agent->prog_finalized)
+    goto final;
+
+  status
+    = hsa_fns.hsa_executable_create_fn (HSA_PROFILE_FULL,
+					HSA_EXECUTABLE_STATE_UNFROZEN,
+					"", &agent->executable);
+  if (status != HSA_STATUS_SUCCESS)
+    {
+      hsa_error ("Could not create GCN executable", status);
+      goto fail;
+    }
+
+  struct obstack unmodified_sections_os;
+  obstack_init (&unmodified_sections_os);
+
+  /* Load any GCN modules.  */
+  struct module_info *module = agent->module;
+  if (module)
+    {
+      Elf64_Ehdr *image = (Elf64_Ehdr *)module->image_desc->gcn_image->image;
+
+      /* Hide relocations from the HSA runtime loader.
+	 Keep a copy of the unmodified section headers to use later.  */
+      Elf64_Shdr *image_sections = (Elf64_Shdr *)((char *)image
+						  + image->e_shoff);
+      Elf64_Shdr *sections = malloc (sizeof (Elf64_Shdr) * image->e_shnum);
+      memcpy (sections, image_sections, sizeof (Elf64_Shdr) * image->e_shnum);
+      for (int i = image->e_shnum - 1; i >= 0; i--)
+	{
+	  if (image_sections[i].sh_type == SHT_RELA
+	      || image_sections[i].sh_type == SHT_REL)
+	    /* Change section type to something harmless.  */
+	    image_sections[i].sh_type = SHT_NOTE;
+	}
+      obstack_ptr_grow (&unmodified_sections_os, sections);
+
+      hsa_code_object_t co = { 0 };
+      status = hsa_fns.hsa_code_object_deserialize_fn
+	(module->image_desc->gcn_image->image,
+	 module->image_desc->gcn_image->size,
+	 NULL, &co);
+      if (status != HSA_STATUS_SUCCESS)
+	{
+	  hsa_error ("Could not deserialize GCN code object", status);
+	  goto fail;
+	}
+
+      status = hsa_fns.hsa_executable_load_code_object_fn
+	(agent->executable, agent->id, co, "");
+      if (status != HSA_STATUS_SUCCESS)
+	{
+	  hsa_error ("Could not load GCN code object", status);
+	  goto fail;
+	}
+
+      if (!module->heap)
+	{
+	  status = hsa_fns.hsa_memory_allocate_fn (agent->kernarg_region,
+						   gcn_kernel_heap_size,
+						   (void**)&module->heap);
+	  if (status != HSA_STATUS_SUCCESS)
+	    {
+	      hsa_error ("Could not allocate memory for GCN heap", status);
+	      goto fail;
+	    }
+
+	  module->heap->size = gcn_kernel_heap_size;
+	}
+
+    }
+  Elf64_Shdr **unmodified_sections = obstack_finish (&unmodified_sections_os);
+
+  if (debug)
+    dump_executable_symbols (agent->executable);
+
+  status = hsa_fns.hsa_executable_freeze_fn (agent->executable, "");
+  if (status != HSA_STATUS_SUCCESS)
+    {
+      hsa_error ("Could not freeze the GCN executable", status);
+      goto fail;
+    }
+
+  int s = 0;
+  if (agent->module)
+    {
+      struct module_info *module = agent->module;
+      Elf64_Ehdr *image = (Elf64_Ehdr *)module->image_desc->gcn_image->image;
+      Elf64_Shdr *sections = unmodified_sections[s++];
+
+      Elf64_Addr load_offset;
+      if (!find_load_offset (&load_offset, agent, module, image, sections))
+	goto fail;
+
+      /* Record the physical load address range.
+	 We need this for data copies later.  */
+      Elf64_Phdr *segments = (Elf64_Phdr *)((char*)image + image->e_phoff);
+      Elf64_Addr low = ~0, high = 0;
+      for (int i = 0; i < image->e_phnum; i++)
+	if (segments[i].p_memsz > 0)
+	  {
+	    if (segments[i].p_paddr < low)
+	      low = segments[i].p_paddr;
+	    if (segments[i].p_paddr > high)
+	      high = segments[i].p_paddr + segments[i].p_memsz - 1;
+	  }
+      module->phys_address_start = low + load_offset;
+      module->phys_address_end = high + load_offset;
+
+      // Find dynamic symbol table
+      Elf64_Shdr *dynsym = NULL;
+      for (int i = 0; i < image->e_shnum; i++)
+	if (sections[i].sh_type == SHT_DYNSYM)
+	  {
+	    dynsym = &sections[i];
+	    break;
+	  }
+
+      /* Fix up relocations.  */
+      for (int i = 0; i < image->e_shnum; i++)
+	{
+	  if (sections[i].sh_type == SHT_RELA)
+	    for (size_t offset = 0;
+		 offset < sections[i].sh_size;
+		 offset += sections[i].sh_entsize)
+	      {
+		Elf64_Rela *reloc = (Elf64_Rela*)((char*)image
+						  + sections[i].sh_offset
+						  + offset);
+		Elf64_Sym *sym =
+		  (dynsym
+		   ? (Elf64_Sym*)((char*)image
+				  + dynsym->sh_offset
+				  + (dynsym->sh_entsize
+				     * ELF64_R_SYM (reloc->r_info)))
+		   : NULL);
+
+		int64_t S = (sym ? sym->st_value : 0);
+		int64_t P = reloc->r_offset + load_offset;
+		int64_t A = reloc->r_addend;
+		int64_t B = load_offset;
+		int64_t V, size;
+		switch (ELF64_R_TYPE (reloc->r_info))
+		  {
+		  case R_AMDGPU_ABS32_LO:
+		    V = (S + A) & 0xFFFFFFFF;
+		    size = 4;
+		    break;
+		  case R_AMDGPU_ABS32_HI:
+		    V = (S + A) >> 32;
+		    size = 4;
+		    break;
+		  case R_AMDGPU_ABS64:
+		    V = S + A;
+		    size = 8;
+		    break;
+		  case R_AMDGPU_REL32:
+		    V = S + A - P;
+		    size = 4;
+		    break;
+		  case R_AMDGPU_REL64:
+		    /* FIXME
+		       LLD seems to emit REL64 where the the assembler has
+		       ABS64.  This is clearly wrong because it's not what the
+		       compiler is expecting.  Let's assume, for now, that
+		       it's a bug.  In any case, GCN kernels are always self
+		       contained and therefore relative relocations will have
+		       been resolved already, so this should be a safe
+		       workaround.  */
+		    V = S + A/* - P*/;
+		    size = 8;
+		    break;
+		  case R_AMDGPU_ABS32:
+		    V = S + A;
+		    size = 4;
+		    break;
+		    /* TODO R_AMDGPU_GOTPCREL */
+		    /* TODO R_AMDGPU_GOTPCREL32_LO */
+		    /* TODO R_AMDGPU_GOTPCREL32_HI */
+		  case R_AMDGPU_REL32_LO:
+		    V = (S + A - P) & 0xFFFFFFFF;
+		    size = 4;
+		    break;
+		  case R_AMDGPU_REL32_HI:
+		    V = (S + A - P) >> 32;
+		    size = 4;
+		    break;
+		  case R_AMDGPU_RELATIVE64:
+		    V = B + A;
+		    size = 8;
+		    break;
+		  default:
+		    fprintf (stderr, "Error: unsupported relocation type.\n");
+		    exit (1);
+		  }
+		status = hsa_fns.hsa_memory_copy_fn ((void*)P, &V, size);
+		if (status != HSA_STATUS_SUCCESS)
+		  {
+		    hsa_error ("Failed to fix up relocation", status);
+		    goto fail;
+		  }
+		reloc_count++;
+	      }
+	}
+
+      free (sections);
+    }
+  obstack_free (&unmodified_sections_os, NULL);
+
+  HSA_DEBUG ("Loaded GCN kernels to device %d (%d relocations)\n",
+	     agent->device_id, reloc_count);
+
+final:
+  agent->prog_finalized = true;
+
+  if (pthread_mutex_unlock (&agent->prog_mutex))
+    {
+      GOMP_PLUGIN_error ("Could not unlock a GCN agent program mutex");
+      res = false;
+    }
+
+  return res;
+
+fail:
+  res = false;
+  goto final;
+}
+
+/* Create kernel dispatch data structure for given KERNEL.  */
+
+static struct GOMP_hsa_kernel_dispatch *
+create_single_kernel_dispatch (struct kernel_info *kernel,
+			       unsigned omp_data_size)
+{
+  struct agent_info *agent = kernel->agent;
+  struct GOMP_hsa_kernel_dispatch *shadow
+    = GOMP_PLUGIN_malloc_cleared (sizeof (struct GOMP_hsa_kernel_dispatch));
+
+  shadow->omp_data_memory
+    = omp_data_size > 0 ? GOMP_PLUGIN_malloc (omp_data_size) : NULL;
+  unsigned dispatch_count = kernel->dependencies_count;
+  if (dispatch_count != 0)
+    GOMP_PLUGIN_fatal ("kernel->dependencies_count != 0");
+  shadow->kernel_dispatch_count = 0;
+
+  shadow->object = kernel->object;
+
+  hsa_signal_t sync_signal;
+  hsa_status_t status = hsa_fns.hsa_signal_create_fn (1, 0, NULL, &sync_signal);
+  if (status != HSA_STATUS_SUCCESS)
+    hsa_fatal ("Error creating the GCN sync signal", status);
+
+  shadow->signal = sync_signal.handle;
+  shadow->private_segment_size = kernel->private_segment_size;
+  shadow->group_segment_size = kernel->group_segment_size;
+
+  /* Ensure that there is space for the gomp_print data.
+     See also gcn-run.c, in GCC.  */
+  size_t kss = kernel->kernarg_segment_size;
+  bool use_gomp_print = false;
+  if (kss <= 8)
+    {
+      kss = sizeof (struct kernargs);
+      use_gomp_print = true;
+    }
+
+  status
+    = hsa_fns.hsa_memory_allocate_fn (agent->kernarg_region,
+				      kss,
+				      &shadow->kernarg_address);
+  if (status != HSA_STATUS_SUCCESS)
+    hsa_fatal ("Could not allocate memory for GCN kernel arguments", status);
+
+  struct kernargs *kernargs = shadow->kernarg_address;
+  if (use_gomp_print)
+    {
+      /* Zero-initialize the output_data (minimum needed).  */
+      kernargs->out_ptr = (int64_t)&kernargs->output_data;
+      kernargs->output_data.next_output = 0;
+      for (unsigned i = 0;
+	   i < (sizeof (kernargs->output_data.queue)
+		/ sizeof (kernargs->output_data.queue[0]));
+	   i++)
+	kernargs->output_data.queue[i].written = 0;
+      kernargs->output_data.consumed = 0;
+
+      /* Pass in the heap location.  */
+      kernargs->heap_ptr = (int64_t)kernel->module->heap;
+    }
+
+  kernargs->output_data.return_value = 0xcafe0000;
+
+  return shadow;
+}
+
+/* Output any data written by gomp_print_*.
+   Only enabled when the requested kernarg_segment_size would not
+   overwrite the gomp_print data.
+   We print all entries from print_index to the next entry without a "written"
+   flag.  Subsequent calls should use the returned print_index value to resume
+   from the same point.  */
+static void
+gomp_print_output (struct kernel_info *kernel, struct kernargs *kernargs,
+		   bool final)
+{
+  if (kernel->kernarg_segment_size <= 8)
+    {
+      unsigned int limit = (sizeof (kernargs->output_data.queue)
+			    / sizeof (kernargs->output_data.queue[0]));
+
+      unsigned int from = __atomic_load_n (&kernargs->output_data.consumed,
+					   __ATOMIC_ACQUIRE);
+      unsigned int to = kernargs->output_data.next_output;
+
+      if (from > to)
+	{
+	  /* Overflow.  */
+	  if (final)
+	    printf ("GCN print buffer overflowed.\n");
+	  return;
+	}
+
+      unsigned int i;
+      for (i = from; i < to; i++)
+	{
+	  struct printf_data *data = &kernargs->output_data.queue[i%limit];
+
+	  if (!data->written && !final)
+	    break;
+
+	  switch (data->type)
+	    {
+	    case 0: printf ("%.128s%ld\n", data->msg, data->ivalue); break;
+	    case 1: printf ("%.128s%f\n", data->msg, data->dvalue); break;
+	    case 2: printf ("%.128s%.128s\n", data->msg, data->text); break;
+	    case 3: printf ("%.128s%.128s", data->msg, data->text); break;
+	    default: printf ("GCN print buffer error!\n"); break;
+	    }
+	  data->written = 0;
+	  __atomic_store_n (&kernargs->output_data.consumed, i+1,
+			    __ATOMIC_RELEASE);
+	}
+      fflush (stdout);
+    }
+}
+
+/* Release data structure created for a kernel dispatch in SHADOW argument.  */
+
+static void
+release_kernel_dispatch (struct GOMP_hsa_kernel_dispatch *shadow)
+{
+  HSA_DEBUG ("Released kernel dispatch: %p has value: %lu (%p)\n", shadow,
+	     shadow->debug, (void *) shadow->debug);
+
+  hsa_fns.hsa_memory_free_fn (shadow->kernarg_address);
+
+  hsa_signal_t s;
+  s.handle = shadow->signal;
+  hsa_fns.hsa_signal_destroy_fn (s);
+
+  free (shadow->omp_data_memory);
+
+  free (shadow);
+}
+
+/* Initialize a KERNEL without its dependencies.  MAX_OMP_DATA_SIZE is used
+   to calculate maximum necessary memory for OMP data allocation.  */
+
+static void
+init_single_kernel (struct kernel_info *kernel, unsigned *max_omp_data_size)
+{
+  hsa_status_t status;
+  struct agent_info *agent = kernel->agent;
+  hsa_executable_symbol_t kernel_symbol;
+  status = hsa_fns.hsa_executable_get_symbol_fn (agent->executable, NULL,
+						 kernel->name, agent->id,
+						 0, &kernel_symbol);
+  if (status != HSA_STATUS_SUCCESS)
+    {
+      hsa_warn ("Could not find symbol for kernel in the code object", status);
+      fprintf (stderr, "not found name: '%s'\n", kernel->name);
+      dump_executable_symbols (agent->executable);
+      goto failure;
+    }
+  HSA_DEBUG ("Located kernel %s\n", kernel->name);
+  status = hsa_fns.hsa_executable_symbol_get_info_fn
+    (kernel_symbol, HSA_EXECUTABLE_SYMBOL_INFO_KERNEL_OBJECT, &kernel->object);
+  if (status != HSA_STATUS_SUCCESS)
+    hsa_fatal ("Could not extract a kernel object from its symbol", status);
+  status = hsa_fns.hsa_executable_symbol_get_info_fn
+    (kernel_symbol, HSA_EXECUTABLE_SYMBOL_INFO_KERNEL_KERNARG_SEGMENT_SIZE,
+     &kernel->kernarg_segment_size);
+  if (status != HSA_STATUS_SUCCESS)
+    hsa_fatal ("Could not get info about kernel argument size", status);
+  status = hsa_fns.hsa_executable_symbol_get_info_fn
+    (kernel_symbol, HSA_EXECUTABLE_SYMBOL_INFO_KERNEL_GROUP_SEGMENT_SIZE,
+     &kernel->group_segment_size);
+  if (status != HSA_STATUS_SUCCESS)
+    hsa_fatal ("Could not get info about kernel group segment size", status);
+  status = hsa_fns.hsa_executable_symbol_get_info_fn
+    (kernel_symbol, HSA_EXECUTABLE_SYMBOL_INFO_KERNEL_PRIVATE_SEGMENT_SIZE,
+     &kernel->private_segment_size);
+  if (status != HSA_STATUS_SUCCESS)
+    hsa_fatal ("Could not get info about kernel private segment size",
+	       status);
+
+  HSA_DEBUG ("Kernel structure for %s fully initialized with "
+	     "following segment sizes: \n", kernel->name);
+  HSA_DEBUG ("  group_segment_size: %u\n",
+	     (unsigned) kernel->group_segment_size);
+  HSA_DEBUG ("  private_segment_size: %u\n",
+	     (unsigned) kernel->private_segment_size);
+  HSA_DEBUG ("  kernarg_segment_size: %u\n",
+	     (unsigned) kernel->kernarg_segment_size);
+  HSA_DEBUG ("  omp_data_size: %u\n", kernel->omp_data_size);
+  HSA_DEBUG ("  gridified_kernel_p: %u\n", kernel->gridified_kernel_p);
+
+  if (kernel->omp_data_size > *max_omp_data_size)
+    *max_omp_data_size = kernel->omp_data_size;
+
+  return;
+
+failure:
+  kernel->initialization_failed = true;
+}
+
+/* Indent stream F by INDENT spaces.  */
+
+static void
+indent_stream (FILE *f, unsigned indent)
+{
+  fprintf (f, "%*s", indent, "");
+}
+
+/* Dump kernel DISPATCH data structure and indent it by INDENT spaces.  */
+
+static void
+print_kernel_dispatch (struct GOMP_hsa_kernel_dispatch *dispatch,
+		       unsigned indent)
+{
+  indent_stream (stderr, indent);
+  fprintf (stderr, "this: %p\n", dispatch);
+  indent_stream (stderr, indent);
+  fprintf (stderr, "queue: %p\n", dispatch->queue);
+  indent_stream (stderr, indent);
+  fprintf (stderr, "omp_data_memory: %p\n", dispatch->omp_data_memory);
+  indent_stream (stderr, indent);
+  fprintf (stderr, "kernarg_address: %p\n", dispatch->kernarg_address);
+  indent_stream (stderr, indent);
+  fprintf (stderr, "object: %lu\n", dispatch->object);
+  indent_stream (stderr, indent);
+  fprintf (stderr, "signal: %lu\n", dispatch->signal);
+  indent_stream (stderr, indent);
+  fprintf (stderr, "private_segment_size: %u\n",
+	   dispatch->private_segment_size);
+  indent_stream (stderr, indent);
+  fprintf (stderr, "group_segment_size: %u\n",
+	   dispatch->group_segment_size);
+  indent_stream (stderr, indent);
+  fprintf (stderr, "children dispatches: %lu\n",
+	   dispatch->kernel_dispatch_count);
+  indent_stream (stderr, indent);
+  fprintf (stderr, "omp_num_threads: %u\n",
+	   dispatch->omp_num_threads);
+  fprintf (stderr, "\n");
+}
+
+/* Create kernel dispatch data structure for a KERNEL and all its
+   dependencies.  */
+
+static struct GOMP_hsa_kernel_dispatch *
+create_kernel_dispatch (struct kernel_info *kernel, unsigned omp_data_size)
+{
+  struct GOMP_hsa_kernel_dispatch *shadow
+    = create_single_kernel_dispatch (kernel, omp_data_size);
+  shadow->omp_num_threads = 64;
+  shadow->debug = 0;
+  shadow->omp_level = kernel->gridified_kernel_p ? 1 : 0;
+
+  return shadow;
+}
+
+/* Do all the work that is necessary before running KERNEL for the first time.
+   The function assumes the program has been created, finalized and frozen by
+   create_and_finalize_hsa_program.  */
+
+static void
+init_kernel (struct kernel_info *kernel)
+{
+  if (pthread_mutex_lock (&kernel->init_mutex))
+    GOMP_PLUGIN_fatal ("Could not lock a GCN kernel initialization mutex");
+  if (kernel->initialized)
+    {
+      if (pthread_mutex_unlock (&kernel->init_mutex))
+	GOMP_PLUGIN_fatal ("Could not unlock a GCN kernel initialization "
+			   "mutex");
+
+      return;
+    }
+
+  /* Precomputed maximum size of OMP data necessary for a kernel from kernel
+     dispatch operation.  */
+  init_single_kernel (kernel, &kernel->max_omp_data_size);
+
+  if (!kernel->initialization_failed)
+    {
+      HSA_DEBUG ("\n");
+
+      kernel->initialized = true;
+    }
+  if (pthread_mutex_unlock (&kernel->init_mutex))
+    GOMP_PLUGIN_fatal ("Could not unlock a GCN kernel initialization "
+		       "mutex");
+}
+
+/* Calculate the maximum grid size for OMP threads / OACC workers.
+   This depends on the kernel's resource usage levels.  */
+
+static int
+limit_worker_threads (int threads)
+{
+  /* FIXME Do something more inteligent here.
+     GCN can always run 4 threads within a Compute Unit, but
+     more than that depends on register usage.  */
+  if (threads > 16)
+    threads = 16;
+  return threads;
+}
+
+/* Parse the target attributes INPUT provided by the compiler and return true
+   if we should run anything all.  If INPUT is NULL, fill DEF with default
+   values, then store INPUT or DEF into *RESULT.  */
+
+static bool
+parse_target_attributes (void **input,
+			 struct GOMP_kernel_launch_attributes *def,
+			 struct GOMP_kernel_launch_attributes **result,
+			 struct agent_info *agent)
+{
+  if (!input)
+    GOMP_PLUGIN_fatal ("No target arguments provided");
+
+  bool grid_attrs_found = false;
+  bool gcn_dims_found = false;
+  int gcn_teams = 0;
+  int gcn_threads = 0;
+  while (*input)
+    {
+      intptr_t id = (intptr_t) *input++, val;
+
+      if (id & GOMP_TARGET_ARG_SUBSEQUENT_PARAM)
+	val = (intptr_t) *input++;
+      else
+	val = id >> GOMP_TARGET_ARG_VALUE_SHIFT;
+
+      val = (val > INT_MAX) ? INT_MAX : val;
+
+      if ((id & GOMP_TARGET_ARG_DEVICE_MASK) == GOMP_DEVICE_GCN
+	  && ((id & GOMP_TARGET_ARG_ID_MASK)
+	      == GOMP_TARGET_ARG_HSA_KERNEL_ATTRIBUTES))
+	{
+	  grid_attrs_found = true;
+	  break;
+	}
+      else if ((id & GOMP_TARGET_ARG_DEVICE_ALL) == GOMP_TARGET_ARG_DEVICE_ALL)
+	{
+	  gcn_dims_found = true;
+	  switch (id & GOMP_TARGET_ARG_ID_MASK)
+	    {
+	    case GOMP_TARGET_ARG_NUM_TEAMS:
+	      gcn_teams = val;
+	      break;
+	    case GOMP_TARGET_ARG_THREAD_LIMIT:
+	      gcn_threads = limit_worker_threads (val);
+	      break;
+	    default:
+	      ;
+	    }
+	}
+    }
+
+  if (gcn_dims_found)
+    {
+      if (agent->gfx900_p && gcn_threads == 0 && override_z_dim == 0)
+	{
+	  gcn_threads = 4;
+	  HSA_DEBUG ("VEGA BUG WORKAROUND: reducing default number of "
+		     "threads to 4 per team.\n");
+	  HSA_DEBUG (" - If this is not a Vega 10 device, please use "
+		     "GCN_NUM_THREADS=16\n");
+	}
+
+      def->ndim = 3;
+      /* Fiji has 64 CUs.  */
+      def->gdims[0] = (gcn_teams > 0) ? gcn_teams : 64;
+      /* Each thread is 64 work items wide.  */
+      def->gdims[1] = 64;
+      /* A work group can have 16 wavefronts.  */
+      def->gdims[2] = (gcn_threads > 0) ? gcn_threads : 16;
+      def->wdims[0] = 1; /* Single team per work-group.  */
+      def->wdims[1] = 64;
+      def->wdims[2] = 16;
+      *result = def;
+      return true;
+    }
+  else if (!grid_attrs_found)
+    {
+      def->ndim = 1;
+      def->gdims[0] = 1;
+      def->gdims[1] = 1;
+      def->gdims[2] = 1;
+      def->wdims[0] = 1;
+      def->wdims[1] = 1;
+      def->wdims[2] = 1;
+      *result = def;
+      HSA_DEBUG ("GOMP_OFFLOAD_run called with no launch attributes\n");
+      return true;
+    }
+
+  struct GOMP_kernel_launch_attributes *kla;
+  kla = (struct GOMP_kernel_launch_attributes *) *input;
+  *result = kla;
+  if (kla->ndim == 0 || kla->ndim > 3)
+    GOMP_PLUGIN_fatal ("Invalid number of dimensions (%u)", kla->ndim);
+
+  HSA_DEBUG ("GOMP_OFFLOAD_run called with %u dimensions:\n", kla->ndim);
+  unsigned i;
+  for (i = 0; i < kla->ndim; i++)
+    {
+      HSA_DEBUG ("  Dimension %u: grid size %u and group size %u\n", i,
+		 kla->gdims[i], kla->wdims[i]);
+      if (kla->gdims[i] == 0)
+	return false;
+    }
+  return true;
+}
+
+/* Return the group size given the requested GROUP size, GRID size and number
+   of grid dimensions NDIM.  */
+
+static uint32_t
+get_group_size (uint32_t ndim, uint32_t grid, uint32_t group)
+{
+  if (group == 0)
+    {
+      /* TODO: Provide a default via environment or device characteristics.  */
+      if (ndim == 1)
+	group = 64;
+      else if (ndim == 2)
+	group = 8;
+      else
+	group = 4;
+    }
+
+  if (group > grid)
+    group = grid;
+  return group;
+}
+
+/* Return true if the HSA runtime can run function FN_PTR.  */
+
+bool
+GOMP_OFFLOAD_can_run (void *fn_ptr)
+{
+  struct kernel_info *kernel = (struct kernel_info *) fn_ptr;
+
+  init_kernel (kernel);
+  if (kernel->initialization_failed)
+    goto failure;
+
+  return true;
+
+failure:
+  if (suppress_host_fallback)
+    GOMP_PLUGIN_fatal ("GCN host fallback has been suppressed");
+  HSA_DEBUG ("GCN target cannot be launched, doing a host fallback\n");
+  return false;
+}
+
+/* Atomically store pair of uint16_t values (HEADER and REST) to a PACKET.  */
+
+void
+packet_store_release (uint32_t* packet, uint16_t header, uint16_t rest)
+{
+  __atomic_store_n (packet, header | (rest << 16), __ATOMIC_RELEASE);
+}
+
+/* Run KERNEL on its agent, pass VARS to it as arguments and take
+   launchattributes from KLA.  MODULE_LOCKED indicates that the caller
+   already holds the lock and run_kernel need not lock it again.
+   If AQ is NULL then agent->sync_queue will be used.  */
+
+static void
+run_kernel (struct kernel_info *kernel, void *vars,
+	    struct GOMP_kernel_launch_attributes *kla,
+	    struct goacc_asyncqueue *aq, bool module_locked)
+{
+  HSA_DEBUG ("GCN launch on queue: %d:%d\n", kernel->agent->device_id,
+	     (aq ? aq->id : 0));
+  HSA_DEBUG ("GCN launch attribs: gdims:[");
+  int i;
+  for (i = 0; i < kla->ndim; ++i)
+    {
+      if (i)
+	HSA_DPRINT (", ");
+      HSA_DPRINT ("%u", kla->gdims[i]);
+    }
+  HSA_DPRINT ("], normalized gdims:[");
+  for (i = 0; i < kla->ndim; ++i)
+    {
+      if (i)
+	HSA_DPRINT (", ");
+      HSA_DPRINT ("%u", kla->gdims[i] / kla->wdims[i]);
+    }
+  HSA_DPRINT ("], wdims:[");
+  for (i = 0; i < kla->ndim; ++i)
+    {
+      if (i)
+	HSA_DPRINT (", ");
+      HSA_DPRINT ("%u", kla->wdims[i]);
+    }
+  HSA_DPRINT ("]\n");
+  HSA_FLUSH ();
+
+  struct agent_info *agent = kernel->agent;
+  if (!module_locked && pthread_rwlock_rdlock (&agent->module_rwlock))
+    GOMP_PLUGIN_fatal ("Unable to read-lock a GCN agent rwlock");
+
+  if (!agent->initialized)
+    GOMP_PLUGIN_fatal ("Agent must be initialized");
+
+  if (!kernel->initialized)
+    GOMP_PLUGIN_fatal ("Called kernel must be initialized");
+
+  struct GOMP_hsa_kernel_dispatch *shadow
+    = create_kernel_dispatch (kernel, kernel->max_omp_data_size);
+
+  hsa_queue_t *command_q = (aq ? aq->hsa_queue : kernel->agent->sync_queue);
+  shadow->queue = command_q;
+
+  if (debug)
+    {
+      fprintf (stderr, "\nKernel has following dependencies:\n");
+      print_kernel_dispatch (shadow, 2);
+    }
+
+  uint64_t index
+    = hsa_fns.hsa_queue_add_write_index_release_fn (command_q, 1);
+  HSA_DEBUG ("Got AQL index %llu\n", (long long int) index);
+
+  /* Wait until the queue is not full before writing the packet.   */
+  while (index - hsa_fns.hsa_queue_load_read_index_acquire_fn (command_q)
+	 >= command_q->size)
+    ;
+
+  /* Do not allow the dimensions to be overridden when running
+     constructors or destructors.  */
+  struct module_info *module = kernel->module;
+  bool init_fini_p = kernel == module->init_array_func
+		     || kernel == module->fini_array_func;
+  int override_x = init_fini_p ? 0 : override_x_dim;
+  int override_z = init_fini_p ? 0 : override_z_dim;
+
+  hsa_kernel_dispatch_packet_t *packet;
+  packet = ((hsa_kernel_dispatch_packet_t *) command_q->base_address)
+	   + index % command_q->size;
+
+  memset (((uint8_t *) packet) + 4, 0, sizeof (*packet) - 4);
+  packet->grid_size_x = override_x ? : kla->gdims[0];
+  packet->workgroup_size_x = get_group_size (kla->ndim,
+					     packet->grid_size_x,
+					     kla->wdims[0]);
+
+  if (kla->ndim >= 2)
+    {
+      packet->grid_size_y = kla->gdims[1];
+      packet->workgroup_size_y = get_group_size (kla->ndim, kla->gdims[1],
+						 kla->wdims[1]);
+    }
+  else
+    {
+      packet->grid_size_y = 1;
+      packet->workgroup_size_y = 1;
+    }
+
+  if (kla->ndim == 3)
+    {
+      packet->grid_size_z = limit_worker_threads (override_z
+						  ? : kla->gdims[2]);
+      packet->workgroup_size_z = get_group_size (kla->ndim,
+						 packet->grid_size_z,
+						 kla->wdims[2]);
+    }
+  else
+    {
+      packet->grid_size_z = 1;
+      packet->workgroup_size_z = 1;
+    }
+
+  HSA_DEBUG ("GCN launch actuals: grid:[%u, %u, %u],"
+	     " normalized grid:[%u, %u, %u], workgroup:[%u, %u, %u]\n",
+	     packet->grid_size_x, packet->grid_size_y, packet->grid_size_z,
+	     packet->grid_size_x / packet->workgroup_size_x,
+	     packet->grid_size_y / packet->workgroup_size_y,
+	     packet->grid_size_z / packet->workgroup_size_z,
+	     packet->workgroup_size_x, packet->workgroup_size_y,
+	     packet->workgroup_size_z);
+
+  packet->private_segment_size = kernel->private_segment_size;
+  packet->group_segment_size = kernel->group_segment_size;
+  packet->kernel_object = kernel->object;
+  packet->kernarg_address = shadow->kernarg_address;
+  hsa_signal_t s;
+  s.handle = shadow->signal;
+  packet->completion_signal = s;
+  hsa_fns.hsa_signal_store_relaxed_fn (s, 1);
+  memcpy (shadow->kernarg_address, &vars, sizeof (vars));
+
+  /* PR hsa/70337.  */
+  size_t vars_size = sizeof (vars);
+  if (kernel->kernarg_segment_size > vars_size)
+    {
+      if (kernel->kernarg_segment_size != vars_size
+	  + sizeof (struct hsa_kernel_runtime *))
+	GOMP_PLUGIN_fatal ("Kernel segment size has an unexpected value");
+      memcpy (packet->kernarg_address + vars_size, &shadow,
+	      sizeof (struct hsa_kernel_runtime *));
+    }
+
+  HSA_DEBUG ("Copying kernel runtime pointer to kernarg_address\n");
+
+  uint16_t header;
+  header = HSA_PACKET_TYPE_KERNEL_DISPATCH << HSA_PACKET_HEADER_TYPE;
+  header |= HSA_FENCE_SCOPE_SYSTEM << HSA_PACKET_HEADER_ACQUIRE_FENCE_SCOPE;
+  header |= HSA_FENCE_SCOPE_SYSTEM << HSA_PACKET_HEADER_RELEASE_FENCE_SCOPE;
+
+  HSA_DEBUG ("Going to dispatch kernel %s on device %d\n", kernel->name,
+	     agent->device_id);
+
+  packet_store_release ((uint32_t *) packet, header,
+			(uint16_t) kla->ndim
+			<< HSA_KERNEL_DISPATCH_PACKET_SETUP_DIMENSIONS);
+
+  hsa_fns.hsa_signal_store_release_fn (command_q->doorbell_signal,
+				       index);
+
+  HSA_DEBUG ("Kernel dispatched, waiting for completion\n");
+
+  /* Root signal waits with 1ms timeout.  */
+  while (hsa_fns.hsa_signal_wait_acquire_fn (s, HSA_SIGNAL_CONDITION_LT, 1,
+					     1000 * 1000,
+					     HSA_WAIT_STATE_BLOCKED) != 0)
+    {
+      gomp_print_output (kernel, shadow->kernarg_address, false);
+    }
+  gomp_print_output (kernel, shadow->kernarg_address, true);
+
+  struct kernargs *kernargs = shadow->kernarg_address;
+  unsigned int return_value = (unsigned int)kernargs->output_data.return_value;
+
+  release_kernel_dispatch (shadow);
+
+  if (!module_locked && pthread_rwlock_unlock (&agent->module_rwlock))
+    GOMP_PLUGIN_fatal ("Unable to unlock a GCN agent rwlock");
+
+  unsigned int upper = (return_value & ~0xffff) >> 16;
+  if (upper == 0xcafe)
+    ; // exit not called, normal termination.
+  else if (upper == 0xffff)
+    ; // exit called.
+  else
+    {
+      GOMP_PLUGIN_error ("Possible kernel exit value corruption, 2 most"
+			 " significant bytes aren't 0xffff or 0xcafe: 0x%x\n",
+			 return_value);
+      abort ();
+    }
+
+  if (upper == 0xffff)
+    {
+      unsigned int signal = (return_value >> 8) & 0xff;
+
+      if (signal == SIGABRT)
+	{
+	  HSA_DEBUG ("GCN Kernel aborted\n");
+	  abort ();
+	}
+      else if (signal != 0)
+	{
+	  HSA_DEBUG ("GCN Kernel received unknown signal\n");
+	  abort ();
+	}
+
+      HSA_DEBUG ("GCN Kernel exited with value: %d\n", return_value & 0xff);
+      exit (return_value & 0xff);
+    }
+}
+
+/* Part of the libgomp plugin interface.  Run a kernel on device N (the number
+   is actually ignored, we assume the FN_PTR has been mapped using the correct
+   device) and pass it an array of pointers in VARS as a parameter.  The kernel
+   is identified by FN_PTR which must point to a kernel_info structure.  */
+
+void
+GOMP_OFFLOAD_run (int device, void *fn_ptr, void *vars, void **args)
+{
+  struct agent_info *agent = get_agent_info (device);
+  struct kernel_info *kernel = (struct kernel_info *) fn_ptr;
+  struct GOMP_kernel_launch_attributes def;
+  struct GOMP_kernel_launch_attributes *kla;
+  assert (agent == kernel->agent);
+
+  if (!parse_target_attributes (args, &def, &kla, agent))
+    {
+      HSA_DEBUG ("Will not run GCN kernel because the grid size is zero\n");
+      return;
+    }
+  run_kernel (kernel, vars, kla, NULL, false);
+}
+
+/* Set up an async queue for OpenMP.  There will be only one.
+   FIXME: is this thread-safe if two threads call this function?  */
+static void
+maybe_init_omp_async (struct agent_info *agent)
+{
+  if (!agent->omp_async_queue)
+    agent->omp_async_queue
+      = GOMP_OFFLOAD_openacc_async_construct (agent->device_id);
+}
+
+/* Part of the libgomp plugin interface.  Run a kernel like GOMP_OFFLOAD_run
+   does, but asynchronously and call GOMP_PLUGIN_target_task_completion when it
+   has finished.  */
+
+void
+GOMP_OFFLOAD_async_run (int device, void *tgt_fn, void *tgt_vars,
+			void **args, void *async_data)
+{
+  HSA_DEBUG ("GOMP_OFFLOAD_async_run invoked\n");
+  struct agent_info *agent = get_agent_info (device);
+  struct kernel_info *kernel = (struct kernel_info *) tgt_fn;
+  struct GOMP_kernel_launch_attributes def;
+  struct GOMP_kernel_launch_attributes *kla;
+  assert (agent == kernel->agent);
+
+  if (!parse_target_attributes (args, &def, &kla, agent))
+    {
+      HSA_DEBUG ("Will not run GCN kernel because the grid size is zero\n");
+      return;
+    }
+
+  maybe_init_omp_async (agent);
+  queue_push_launch (agent->omp_async_queue, kernel, tgt_vars, kla);
+  queue_push_callback (agent->omp_async_queue,
+		       GOMP_PLUGIN_target_task_completion, async_data);
+}
+
+/* Deinitialize all information associated with MODULE and kernels within
+   it.  Return TRUE on success.  */
+
+static bool
+destroy_module (struct module_info *module, bool locked)
+{
+  /* Run destructors before destroying module.  */
+  struct GOMP_kernel_launch_attributes kla =
+    { 3,
+      /* Grid size.  */
+      { 1, 64, 1 },
+      /* Work-group size.  */
+      { 1, 64, 1 }
+    };
+
+  if (module->fini_array_func)
+    {
+      init_kernel (module->fini_array_func);
+      run_kernel (module->fini_array_func, NULL, &kla, NULL, locked);
+    }
+  module->constructors_run_p = false;
+
+  int i;
+  for (i = 0; i < module->kernel_count; i++)
+    if (pthread_mutex_destroy (&module->kernels[i].init_mutex))
+      {
+	GOMP_PLUGIN_error ("Failed to destroy a GCN kernel initialization "
+			   "mutex");
+	return false;
+      }
+
+  return true;
+}
+
+/* Part of the libgomp plugin interface.  Unload GCN object-code module
+   described by struct gcn_image_desc in TARGET_DATA from agent number N.
+   Return TRUE on success.  */
+
+bool
+GOMP_OFFLOAD_unload_image (int n, unsigned version, const void *target_data)
+{
+  if (GOMP_VERSION_DEV (version) > GOMP_VERSION_HSA)
+    {
+      GOMP_PLUGIN_error ("Offload data incompatible with GCN plugin"
+			 " (expected %u, received %u)",
+			 GOMP_VERSION_GCN, GOMP_VERSION_DEV (version));
+      return false;
+    }
+
+  struct agent_info *agent;
+  agent = get_agent_info (n);
+  if (!agent)
+    return false;
+
+  if (pthread_rwlock_wrlock (&agent->module_rwlock))
+    {
+      GOMP_PLUGIN_error ("Unable to write-lock a GCN agent rwlock");
+      return false;
+    }
+
+  if (!agent->module || agent->module->image_desc != target_data)
+    {
+      GOMP_PLUGIN_error ("Attempt to unload an image that has never been "
+			 "loaded before");
+      return false;
+    }
+
+  if (!destroy_module (agent->module, true))
+    return false;
+  free (agent->module);
+  agent->module = NULL;
+  if (!destroy_hsa_program (agent))
+    return false;
+  if (pthread_rwlock_unlock (&agent->module_rwlock))
+    {
+      GOMP_PLUGIN_error ("Unable to unlock a GCN agent rwlock");
+      return false;
+    }
+  return true;
+}
+
+/* Part of the libgomp plugin interface.  Deinitialize all information and
+   status associated with agent number N.  We do not attempt any
+   synchronization, assuming the user and libgomp will not attempt
+   deinitialization of a device that is in any way being used at the same
+   time.  Return TRUE on success.  */
+
+bool
+GOMP_OFFLOAD_fini_device (int n)
+{
+  struct agent_info *agent = get_agent_info (n);
+  if (!agent)
+    return false;
+
+  if (!agent->initialized)
+    return true;
+
+  if (agent->omp_async_queue)
+    {
+      GOMP_OFFLOAD_openacc_async_destruct (agent->omp_async_queue);
+      agent->omp_async_queue = NULL;
+    }
+
+  if (agent->module)
+    {
+      if (!destroy_module (agent->module, false))
+        return false;
+      free (agent->module);
+      agent->module = NULL;
+    }
+
+  if (!destroy_hsa_program (agent))
+    return false;
+
+  /*release_agent_shared_libraries (agent);*/
+
+  hsa_status_t status = hsa_fns.hsa_queue_destroy_fn (agent->sync_queue);
+  if (status != HSA_STATUS_SUCCESS)
+    return hsa_error ("Error destroying command queue", status);
+
+  if (pthread_mutex_destroy (&agent->prog_mutex))
+    {
+      GOMP_PLUGIN_error ("Failed to destroy a GCN agent program mutex");
+      return false;
+    }
+  if (pthread_rwlock_destroy (&agent->module_rwlock))
+    {
+      GOMP_PLUGIN_error ("Failed to destroy a GCN agent rwlock");
+      return false;
+    }
+
+  if (pthread_mutex_destroy (&agent->async_queues_mutex))
+    {
+      GOMP_PLUGIN_error ("Failed to destroy a GCN agent queue mutex");
+      return false;
+    }
+  agent->initialized = false;
+  return true;
+}
+
+static void *
+GOMP_OFFLOAD_alloc_by_agent (struct agent_info *agent, size_t size)
+{
+  HSA_DEBUG ("Allocating %zu bytes on device %d\n", size, agent->device_id);
+
+  /* Zero-size allocations are invalid, so in order to return a valid pointer
+     we need to pass a valid size.  One source of zero-size allocations is
+     kernargs for kernels that have no inputs or outputs (the kernel may
+     only use gomp_print, for example).  */
+  if (size == 0)
+    size = 4;
+
+  void *ptr;
+  hsa_status_t status = hsa_fns.hsa_memory_allocate_fn (agent->kernarg_region,
+							size, &ptr);
+  if (status != HSA_STATUS_SUCCESS)
+    {
+      hsa_error ("Could not allocate device memory", status);
+      return NULL;
+    }
+
+  return ptr;
+}
+
+void *
+GOMP_OFFLOAD_alloc (int n, size_t size)
+{
+  struct agent_info *agent = get_agent_info (n);
+  return GOMP_OFFLOAD_alloc_by_agent (agent, size);
+}
+
+bool
+GOMP_OFFLOAD_free (int device, void *ptr)
+{
+  HSA_DEBUG ("Freeing memory on device %d\n", device);
+
+  hsa_status_t status = hsa_fns.hsa_memory_free_fn (ptr);
+  if (status != HSA_STATUS_SUCCESS)
+    {
+      hsa_error ("Could not free device memory", status);
+      return false;
+    }
+
+  return true;
+}
+
+/* Returns true if PTR falls within the bounds of any loaded kernel image.  */
+
+static bool
+image_address_p (struct agent_info *agent, const void *ptr)
+{
+  Elf64_Addr addr = (Elf64_Addr)ptr;
+  if (agent->module)
+    {
+      if (addr >= agent->module->phys_address_start
+	  && addr <= agent->module->phys_address_end)
+	return true;
+    }
+  return false;
+}
+
+struct copy_data
+{
+  void *dst;
+  const void *src;
+  size_t len;
+  bool use_hsa_memory_copy;
+  struct goacc_asyncqueue *aq;
+};
+
+static void
+copy_data (void *data_)
+{
+  struct copy_data *data = (struct copy_data *)data_;
+  HSA_DEBUG ("Async thread %d:%d: Copying %zu bytes from (%p) to (%p)\n",
+	     data->aq->agent->device_id, data->aq->id, data->len, data->src,
+	     data->dst);
+  if (data->use_hsa_memory_copy)
+    hsa_fns.hsa_memory_copy_fn (data->dst, data->src, data->len);
+  else
+    memcpy (data->dst, data->src, data->len);
+  free (data);
+}
+
+static void
+queue_push_copy (struct goacc_asyncqueue *aq, void *dst, const void *src,
+		 size_t len, bool use_hsa_memory_copy)
+{
+  if (DEBUG_QUEUES)
+    HSA_DEBUG ("queue_push_copy %d:%d: %zu bytes from (%p) to (%p)\n",
+	       aq->agent->device_id, aq->id, len, src, dst);
+  struct copy_data *data
+    = (struct copy_data *)GOMP_PLUGIN_malloc (sizeof (struct copy_data));
+  data->dst = dst;
+  data->src = src;
+  data->len = len;
+  data->use_hsa_memory_copy = use_hsa_memory_copy;
+  data->aq = aq;
+  queue_push_callback (aq, copy_data, data);
+}
+
+bool
+GOMP_OFFLOAD_dev2host (int device, void *dst, const void *src, size_t n)
+{
+  HSA_DEBUG ("Copying %zu bytes from device %d (%p) to host (%p)\n", n, device,
+	     src, dst);
+
+  /* memcpy only works for addresses allocated with hsa_memory_allocate,
+     but hsa_memory_copy seems unable to read from .rodata variables.  */
+  if (image_address_p (get_agent_info (device), src))
+    hsa_fns.hsa_memory_copy_fn (dst, src, n);
+  else
+    memcpy (dst, src, n);
+  return true;
+}
+
+bool
+GOMP_OFFLOAD_host2dev (int device, void *dst, const void *src, size_t n)
+{
+  HSA_DEBUG ("Copying %zu bytes from host (%p) to device %d (%p)\n", n, src,
+	     device, dst);
+  /* memcpy only works for addresses allocated with hsa_memory_allocate,
+     but hsa_memory_copy seems unable to read from .rodata variables.  */
+  if (image_address_p (get_agent_info (device), dst))
+    hsa_fns.hsa_memory_copy_fn (dst, src, n);
+  else
+    memcpy (dst, src, n);
+  return true;
+}
+
+/* Part of the libgomp plugin interface.  */
+
+bool
+GOMP_OFFLOAD_dev2dev (int device, void *dst, const void *src, size_t n)
+{
+  struct gcn_thread *thread_data = gcn_thread ();
+
+  if (thread_data && !async_synchronous_p (thread_data->async))
+    {
+      struct agent_info *agent = get_agent_info (device);
+      maybe_init_omp_async (agent);
+      queue_push_copy (agent->omp_async_queue, dst, src, n, false);
+      return true;
+    }
+
+  HSA_DEBUG ("Copying %zu bytes from device %d (%p) to device %d (%p)\n", n,
+	     device, src, device, dst);
+  /* We can assume that dev2dev moves are always within allocated memory.  */
+  memcpy (dst, src, n);
+  return true;
+}
+
+static int
+queue_empty (struct goacc_asyncqueue *aq)
+{
+  pthread_mutex_lock (&aq->mutex);
+  int res = aq->queue_n == 0 ? 1 : 0;
+  pthread_mutex_unlock (&aq->mutex);
+
+  return res;
+}
+
+static void
+wait_queue (struct goacc_asyncqueue *aq)
+{
+  if (DRAIN_QUEUE_SYNCHRONOUS_P)
+    {
+      drain_queue_synchronous (aq);
+      return;
+    }
+
+  pthread_mutex_lock (&aq->mutex);
+
+  while (aq->queue_n > 0)
+    {
+      if (DEBUG_THREAD_SLEEP)
+	HSA_DEBUG ("waiting for thread %d:%d, putting thread to sleep\n",
+		   aq->agent->device_id, aq->id);
+      pthread_cond_wait (&aq->queue_cond_out, &aq->mutex);
+      if (DEBUG_THREAD_SLEEP)
+	HSA_DEBUG ("thread %d:%d woke up.  Rechecking\n", aq->agent->device_id,
+		   aq->id);
+    }
+
+  pthread_mutex_unlock (&aq->mutex);
+  HSA_DEBUG ("waiting for thread %d:%d, done\n", aq->agent->device_id, aq->id);
+}
+
+static void
+gomp_offload_free (void *ptr)
+{
+  HSA_DEBUG ("Async thread ?:?: Freeing %p\n", ptr);
+  GOMP_OFFLOAD_free (0, ptr);
+}
+
+static void
+gcn_exec (struct kernel_info *kernel, size_t mapnum, void **hostaddrs,
+	  void **devaddrs, unsigned *dims, void *targ_mem_desc, bool async,
+	  struct goacc_asyncqueue *aq)
+{
+  if (!GOMP_OFFLOAD_can_run (kernel))
+    GOMP_PLUGIN_fatal ("OpenACC host fallback unimplemented.");
+
+  // For some reason, devaddrs must be double-indirect on the target
+  void **ind_da = GOMP_OFFLOAD_alloc_by_agent (kernel->agent,
+					       sizeof (void*) * mapnum);
+  for (size_t i = 0; i < mapnum; i++)
+    ind_da[i] = devaddrs[i] ? devaddrs[i] : hostaddrs[i];
+
+  struct hsa_kernel_description *hsa_kernel_desc = NULL;
+  for (unsigned i = 0; i < kernel->module->image_desc->kernel_count; i++)
+    {
+      struct hsa_kernel_description *d
+	= &kernel->module->image_desc->kernel_infos[i];
+      if (d->name == kernel->name)
+	{
+	  hsa_kernel_desc = d;
+	  break;
+	}
+    }
+
+  /* We may have statically-determined dimensions in
+     hsa_kernel_desc->oacc_dims[] or dimensions passed to this offload kernel
+     invocation at runtime in dims[].  We allow static dimensions to take
+     priority over dynamic dimensions when present (non-zero).  */
+  if (hsa_kernel_desc->oacc_dims[0] > 0)
+    dims[0] = hsa_kernel_desc->oacc_dims[0];
+  if (hsa_kernel_desc->oacc_dims[1] > 0)
+    dims[1] = hsa_kernel_desc->oacc_dims[1];
+  if (hsa_kernel_desc->oacc_dims[2] > 0)
+    dims[2] = hsa_kernel_desc->oacc_dims[2];
+
+  /* If any of the OpenACC dimensions remain 0 then we get to pick a number.
+     There isn't really a correct answer for this without a clue about the
+     problem size, so let's do a reasonable number of single-worker gangs.
+     64 gangs matches a typical Fiji device.  */
+
+  if (dims[0] == 0) dims[0] = 64; /* Gangs.  */
+  if (dims[1] == 0) dims[1] = 16; /* Workers.  */
+
+  /* The incoming dimensions are expressed in terms of gangs, workers, and
+     vectors.  The HSA dimensions are expressed in terms of "work-items",
+     which means multiples of vector lanes.
+
+     The "grid size" specifies the size of the problem space, and the
+     "work-group size" specifies how much of that we want a single compute
+     unit to chew on at once.
+
+     The three dimensions do not really correspond to hardware, but the
+     important thing is that the HSA runtime will launch as many
+     work-groups as it takes to process the entire grid, and each
+     work-group will contain as many wave-fronts as it takes to process
+     the work-items in that group.
+
+     Essentially, as long as we set the Y dimension to 64 (the number of
+     vector lanes in hardware), and the Z group size to the maximum (16),
+     then we will get the gangs (X) and workers (Z) launched as we expect.
+
+     The reason for the apparent reversal of vector and worker dimension
+     order is to do with the way the run-time distributes work-items across
+     v1 and v2.  */
+  struct GOMP_kernel_launch_attributes kla =
+    {3,
+     /* Grid size.  */
+     {dims[0], 64, dims[1]},
+     /* Work-group size.  */
+     {1,       64, 16}
+    };
+
+  if (!async)
+    {
+      run_kernel (kernel, ind_da, &kla, NULL, false);
+      gomp_offload_free (ind_da);
+    }
+  else
+    {
+      queue_push_launch (aq, kernel, ind_da, &kla);
+      if (DEBUG_QUEUES)
+	HSA_DEBUG ("queue_push_callback %d:%d gomp_offload_free, %p\n",
+		   aq->agent->device_id, aq->id, ind_da);
+      queue_push_callback (aq, gomp_offload_free, ind_da);
+    }
+}
+
+void
+GOMP_OFFLOAD_openacc_exec (void (*fn_ptr) (void *), size_t mapnum,
+			   void **hostaddrs, void **devaddrs, unsigned *dims,
+			   void *targ_mem_desc)
+{
+  struct kernel_info *kernel = (struct kernel_info *) fn_ptr;
+
+  gcn_exec (kernel, mapnum, hostaddrs, devaddrs, dims, targ_mem_desc, false,
+	    NULL);
+}
+
+void
+GOMP_OFFLOAD_openacc_async_exec (void (*fn_ptr) (void *), size_t mapnum,
+				 void **hostaddrs, void **devaddrs,
+				 unsigned *dims, void *targ_mem_desc,
+				 struct goacc_asyncqueue *aq)
+{
+  struct kernel_info *kernel = (struct kernel_info *) fn_ptr;
+
+  gcn_exec (kernel, mapnum, hostaddrs, devaddrs, dims, targ_mem_desc, true,
+	    aq);
+}
+
+struct goacc_asyncqueue *
+GOMP_OFFLOAD_openacc_async_construct (int device)
+{
+  struct agent_info *agent = get_agent_info (device);
+
+  pthread_mutex_lock (&agent->async_queues_mutex);
+
+  struct goacc_asyncqueue *aq = GOMP_PLUGIN_malloc (sizeof (*aq));
+  aq->agent = get_agent_info (device);
+  aq->prev = NULL;
+  aq->next = agent->async_queues;
+  if (aq->next)
+    {
+      aq->next->prev = aq;
+      aq->id = aq->next->id + 1;
+    }
+  else
+    aq->id = 1;
+  agent->async_queues = aq;
+
+  aq->queue_first = 0;
+  aq->queue_n = 0;
+  aq->drain_queue_stop = 0;
+
+  if (pthread_mutex_init (&aq->mutex, NULL))
+    {
+      GOMP_PLUGIN_error ("Failed to initialize a GCN agent queue mutex");
+      return false;
+    }
+  if (pthread_cond_init (&aq->queue_cond_in, NULL))
+    {
+      GOMP_PLUGIN_error ("Failed to initialize a GCN agent queue cond");
+      return false;
+    }
+  if (pthread_cond_init (&aq->queue_cond_out, NULL))
+    {
+      GOMP_PLUGIN_error ("Failed to initialize a GCN agent queue cond");
+      return false;
+    }
+
+  hsa_status_t status = hsa_fns.hsa_queue_create_fn (agent->id,
+						     ASYNC_QUEUE_SIZE,
+						     HSA_QUEUE_TYPE_MULTI,
+						     queue_callback, NULL,
+						     UINT32_MAX, UINT32_MAX,
+						     &aq->hsa_queue);
+  if (status != HSA_STATUS_SUCCESS)
+    hsa_fatal ("Error creating command queue", status);
+
+  int err = pthread_create (&aq->thread_drain_queue, NULL, &drain_queue, aq);
+  if (err != 0)
+    GOMP_PLUGIN_fatal ("GCN asynchronous thread creation failed: %s",
+		       strerror (err));
+  HSA_DEBUG ("Async thread %d:%d: created\n", aq->agent->device_id,
+	     aq->id);
+
+  pthread_mutex_unlock (&agent->async_queues_mutex);
+
+  return aq;
+}
+
+bool
+GOMP_OFFLOAD_openacc_async_destruct (struct goacc_asyncqueue *aq)
+{
+  struct agent_info *agent = aq->agent;
+
+  finalize_async_thread (aq);
+
+  pthread_mutex_lock (&agent->async_queues_mutex);
+
+  int err;
+  if ((err = pthread_mutex_destroy (&aq->mutex)))
+    {
+      GOMP_PLUGIN_error ("Failed to destroy a GCN async queue mutex: %d", err);
+      goto fail;
+    }
+  if (pthread_cond_destroy (&aq->queue_cond_in))
+    {
+      GOMP_PLUGIN_error ("Failed to destroy a GCN async queue cond");
+      goto fail;
+    }
+  if (pthread_cond_destroy (&aq->queue_cond_out))
+    {
+      GOMP_PLUGIN_error ("Failed to destroy a GCN async queue cond");
+      goto fail;
+    }
+  hsa_status_t status = hsa_fns.hsa_queue_destroy_fn (aq->hsa_queue);
+  if (status != HSA_STATUS_SUCCESS)
+    {
+      hsa_error ("Error destroying command queue", status);
+      goto fail;
+    }
+
+  if (aq->prev)
+    aq->prev->next = aq->next;
+  if (aq->next)
+    aq->next->prev = aq->prev;
+  if (agent->async_queues == aq)
+    agent->async_queues = aq->next;
+
+  HSA_DEBUG ("Async thread %d:%d: destroyed\n", agent->device_id, aq->id);
+
+  free (aq);
+  pthread_mutex_unlock (&agent->async_queues_mutex);
+  return true;
+
+fail:
+  pthread_mutex_unlock (&agent->async_queues_mutex);
+  return false;
+}
+
+int
+GOMP_OFFLOAD_openacc_async_test (struct goacc_asyncqueue *aq)
+{
+  return queue_empty (aq);
+}
+
+bool
+GOMP_OFFLOAD_openacc_async_synchronize (struct goacc_asyncqueue *aq)
+{
+  wait_queue (aq);
+  return true;
+}
+
+bool
+GOMP_OFFLOAD_openacc_async_serialize (struct goacc_asyncqueue *aq1,
+				      struct goacc_asyncqueue *aq2)
+{
+  /* FIXME: what should happen here????  */
+  wait_queue (aq1);
+  wait_queue (aq2);
+  return true;
+}
+
+void
+GOMP_OFFLOAD_openacc_async_queue_callback (struct goacc_asyncqueue *aq,
+					   void (*fn) (void *), void *data)
+{
+  queue_push_callback (aq, fn, data);
+}
+
+bool
+GOMP_OFFLOAD_openacc_async_host2dev (int device, void *dst, const void *src,
+				     size_t n, struct goacc_asyncqueue *aq)
+{
+  struct agent_info *agent = get_agent_info (device);
+  assert (agent == aq->agent);
+  queue_push_copy (aq, dst, src, n, image_address_p (agent, dst));
+  return true;
+}
+
+bool
+GOMP_OFFLOAD_openacc_async_dev2host (int device, void *dst, const void *src,
+				     size_t n, struct goacc_asyncqueue *aq)
+{
+  struct agent_info *agent = get_agent_info (device);
+  assert (agent == aq->agent);
+  queue_push_copy (aq, dst, src, n, image_address_p (agent, src));
+  return true;
+}
+
+void *
+GOMP_OFFLOAD_openacc_create_thread_data (int ord __attribute__((unused)))
+{
+  struct gcn_thread *thread_data
+    = GOMP_PLUGIN_malloc (sizeof (struct gcn_thread));
+
+  thread_data->async = GOMP_ASYNC_SYNC;
+
+  return (void *) thread_data;
+}
+
+void
+GOMP_OFFLOAD_openacc_destroy_thread_data (void *data)
+{
+  free (data);
+}
diff --git a/libgomp/target.c b/libgomp/target.c
index c81e5ababb7..4645894f869 100644
--- a/libgomp/target.c
+++ b/libgomp/target.c
@@ -39,7 +39,7 @@ 
 #include <string.h>
 #include <assert.h>
 #include <errno.h>
-#ifdef RC_CHECKING
+#if defined(RC_CHECKING)
 #include <stdio.h>
 #endif
 
@@ -302,6 +302,12 @@  gomp_to_device_kind_p (int kind)
     }
 }
 
+/* Copy host memory to an offload device.  In asynchronous mode (if AQ is
+   non-NULL), this is only safe when the source memory is a global or heap
+   location (otherwise a copy may take place from a dangling pointer to an
+   expired stack frame).  Use copy_host2dev_immediate for copies from stack
+   locations.  */
+
 attribute_hidden void
 gomp_copy_host2dev (struct gomp_device_descr *devicep,
 		    struct goacc_asyncqueue *aq,
@@ -340,6 +346,17 @@  gomp_copy_host2dev (struct gomp_device_descr *devicep,
     gomp_device_copy (devicep, devicep->host2dev_func, "dev", d, "host", h, sz);
 }
 
+/* Use this variant for host-to-device copies from stack locations that may not
+   be live at the time an asynchronous copy operation takes place.  */
+
+static void
+copy_host2dev_immediate (struct gomp_device_descr *devicep, void *d,
+			 const void *h, size_t sz,
+			 struct gomp_coalesce_buf *cbuf)
+{
+  gomp_copy_host2dev (devicep, NULL, d, h, sz, cbuf);
+}
+
 attribute_hidden void
 gomp_copy_dev2host (struct gomp_device_descr *devicep,
 		    struct goacc_asyncqueue *aq,
@@ -600,10 +617,10 @@  gomp_map_pointer (struct target_mem_desc *tgt, struct goacc_asyncqueue *aq,
   if (cur_node.host_start == (uintptr_t) NULL)
     {
       cur_node.tgt_offset = (uintptr_t) NULL;
-      gomp_copy_host2dev (devicep, aq,
-			  (void *) (tgt->tgt_start + target_offset),
-			  (void *) &cur_node.tgt_offset,
-			  sizeof (void *), cbuf);
+      copy_host2dev_immediate (devicep,
+			       (void *) (tgt->tgt_start + target_offset),
+			       (void *) &cur_node.tgt_offset,
+			       sizeof (void *), cbuf);
       return;
     }
   /* Add bias to the pointer value.  */
@@ -622,8 +639,9 @@  gomp_map_pointer (struct target_mem_desc *tgt, struct goacc_asyncqueue *aq,
      array section.  Now subtract bias to get what we want
      to initialize the pointer with.  */
   cur_node.tgt_offset -= bias;
-  gomp_copy_host2dev (devicep, aq, (void *) (tgt->tgt_start + target_offset),
-		      (void *) &cur_node.tgt_offset, sizeof (void *), cbuf);
+  copy_host2dev_immediate (devicep, (void *) (tgt->tgt_start + target_offset),
+			   (void *) &cur_node.tgt_offset, sizeof (void *),
+			   cbuf);
 }
 
 static void
@@ -1442,13 +1460,13 @@  gomp_map_vars_internal (struct gomp_device_descr *devicep,
 		  cur_node.tgt_offset = gomp_map_val (tgt, hostaddrs, i - 1);
 		if (cur_node.tgt_offset)
 		  cur_node.tgt_offset -= sizes[i];
-		gomp_copy_host2dev (devicep, aq,
-				    (void *) (n->tgt->tgt_start
-					      + n->tgt_offset
-					      + cur_node.host_start
-					      - n->host_start),
-				    (void *) &cur_node.tgt_offset,
-				    sizeof (void *), cbufp);
+		copy_host2dev_immediate (devicep,
+					 (void *) (n->tgt->tgt_start
+						   + n->tgt_offset
+						   + cur_node.host_start
+						   - n->host_start),
+					 (void *) &cur_node.tgt_offset,
+					 sizeof (void *), cbufp);
 		cur_node.tgt_offset = n->tgt->tgt_start + n->tgt_offset
 				      + cur_node.host_start - n->host_start;
 		continue;
@@ -1687,8 +1705,8 @@  gomp_map_vars_internal (struct gomp_device_descr *devicep,
 		    void *tgt_addr = (void *) (tgt->tgt_start + k->tgt_offset);
 		    /* We intentionally do not use coalescing here, as it's not
 		       data allocated by the current call to this function.  */
-		    gomp_copy_host2dev (devicep, aq, (void *) n->tgt_offset,
-					&tgt_addr, sizeof (void *), NULL);
+		    copy_host2dev_immediate (devicep, (void *) n->tgt_offset,
+					     &tgt_addr, sizeof (void *), NULL);
 		  }
 		array++;
 	      }
@@ -1810,10 +1828,9 @@  gomp_map_vars_internal (struct gomp_device_descr *devicep,
       for (i = 0; i < mapnum; i++)
 	{
 	  cur_node.tgt_offset = gomp_map_val (tgt, hostaddrs, i);
-	  gomp_copy_host2dev (devicep, aq,
-			      (void *) (tgt->tgt_start + i * sizeof (void *)),
-			      (void *) &cur_node.tgt_offset, sizeof (void *),
-			      cbufp);
+	  copy_host2dev_immediate (devicep,
+	    (void *) (tgt->tgt_start + i * sizeof (void *)),
+	    (void *) &cur_node.tgt_offset, sizeof (void *), cbufp);
 	}
     }
 
@@ -3725,6 +3742,8 @@  offload_target_to_plugin_name (const char *offload_target)
     return "nvptx";
   else if (strncmp (offload_target, "hsa", 3) == 0)
     return "hsa";
+  else if (strstr (offload_target, "gcn") != NULL)
+    return "gcn";
   else
     gomp_fatal ("Unknown offload target: %s", offload_target);
 }
diff --git a/libgomp/team.c b/libgomp/team.c
index c422da3701d..b26caaaaec6 100644
--- a/libgomp/team.c
+++ b/libgomp/team.c
@@ -239,6 +239,9 @@  gomp_free_pool_helper (void *thread_pool)
   pthread_exit (NULL);
 #elif defined(__nvptx__)
   asm ("exit;");
+#elif defined(__AMDGCN__)
+  asm ("s_dcache_wb\n\t"
+       "s_endpgm");
 #else
 #error gomp_free_pool_helper must terminate the thread
 #endif
diff --git a/libgomp/testsuite/Makefile.in b/libgomp/testsuite/Makefile.in
index db794f35d6d..9c4d7b37c6e 100644
--- a/libgomp/testsuite/Makefile.in
+++ b/libgomp/testsuite/Makefile.in
@@ -207,6 +207,10 @@  PACKAGE_URL = @PACKAGE_URL@
 PACKAGE_VERSION = @PACKAGE_VERSION@
 PATH_SEPARATOR = @PATH_SEPARATOR@
 PERL = @PERL@
+PLUGIN_GCN = @PLUGIN_GCN@
+PLUGIN_GCN_CPPFLAGS = @PLUGIN_GCN_CPPFLAGS@
+PLUGIN_GCN_LDFLAGS = @PLUGIN_GCN_LDFLAGS@
+PLUGIN_GCN_LIBS = @PLUGIN_GCN_LIBS@
 PLUGIN_HSA = @PLUGIN_HSA@
 PLUGIN_HSA_CPPFLAGS = @PLUGIN_HSA_CPPFLAGS@
 PLUGIN_HSA_LDFLAGS = @PLUGIN_HSA_LDFLAGS@
@@ -280,6 +284,7 @@  pdfdir = @pdfdir@
 prefix = @prefix@
 program_transform_name = @program_transform_name@
 psdir = @psdir@
+runstatedir = @runstatedir@
 sbindir = @sbindir@
 sharedstatedir = @sharedstatedir@
 srcdir = @srcdir@
diff --git a/libgomp/testsuite/lib/libgomp.exp b/libgomp/testsuite/lib/libgomp.exp
index a0fe4071cf9..19bee806fb0 100644
--- a/libgomp/testsuite/lib/libgomp.exp
+++ b/libgomp/testsuite/lib/libgomp.exp
@@ -445,3 +445,28 @@  proc check_effective_target_hsa_offloading_selected {} {
 	check_effective_target_hsa_offloading_selected_nocache
     }]
 }
+# Return 1 if at least one AMD GCN board is present.
+
+proc check_effective_target_openacc_amdgcn_accel_present { } {
+    return [check_runtime openacc_amdgcn_accel_present {
+	#include <openacc.h>
+	int main () {
+	return !(acc_get_num_devices (acc_device_gcn) > 0);
+	}
+    } "" ]
+}
+
+# Return 1 if at least one AMD GCN board is present, and the AMD GCN device
+# type is selected by default.
+
+proc check_effective_target_openacc_amdgcn_accel_selected { } {
+    if { ![check_effective_target_openacc_amdgcn_accel_present] } {
+	return 0;
+    }
+    global offload_target_openacc
+    if { [string match "amdgcn*" $offload_target_openacc] } {
+        return 1;
+    }
+    return 0;
+}
+
diff --git a/libgomp/testsuite/libgomp.c/c.exp b/libgomp/testsuite/libgomp.c/c.exp
index 31bdd5795dc..463d953f53d 100644
--- a/libgomp/testsuite/libgomp.c/c.exp
+++ b/libgomp/testsuite/libgomp.c/c.exp
@@ -23,10 +23,81 @@  dg-init
 # Turn on OpenMP.
 lappend ALWAYS_CFLAGS "additional_flags=-fopenmp"
 
+# Generate new tests for each DO_TEST entry in TEST_LIST.
+proc generate_tests { test_list } {
+    global srcdir
+    global subdir
+
+    # Get corresponding source file.
+    set base_file [regsub "\.list" $test_list ""]
+    set base_file [regsub "$srcdir/$subdir/" $base_file ""]
+    set c_file $base_file.c
+
+    # Get dg directives from c file.
+    set dg_directives ""
+    set fp [open "$srcdir/$subdir/$c_file" r]
+    while {[gets $fp line] >= 0} {
+	if {[regexp -line -- "^/\\* \{ dg-" $line]} {
+	    if { "$dg_directives" == "" } {
+		set sep ""
+	    } else {
+		set sep "\n"
+	    }
+	    set dg_directives "$dg_directives$sep$line"
+	}
+    }
+    close $fp
+
+    # Get list of tests.
+    set fp [open "$test_list" r]
+    set file_data [read $fp]
+    close $fp
+    set file_data [regsub -all "DO_TEST" $file_data ""]
+    set file_data [regsub -all "\\(" $file_data ""]
+    set file_data [regsub -all "\\)" $file_data ""]
+    set file_data [regsub -all \[\n\] $file_data ""]
+    set file_data [string trimleft $file_data " "]
+    set tests [split $file_data]
+
+    # Create directory to generate files.
+    set test_dir [pwd]
+    set generated_dir $test_dir/generated/libgomp.c
+    file mkdir $generated_dir
+
+    # Generate tests.
+    set new_files []
+    set i 1
+    foreach test $tests {
+	set new_file "$generated_dir/$base_file-$test.c"
+
+	set fp [open "$new_file" w]
+	puts $fp "$dg_directives"
+	puts $fp "#define ONE_TEST $test"
+	puts $fp "#define TEST_NR $i"
+	puts $fp "#include \"$srcdir/$subdir/$c_file\""
+	close $fp
+
+	set i [expr $i + 1]
+	lappend new_files $new_file
+    }
+
+    return $new_files
+}
+
+# Generate tests for each .list file
+set test_lists [find $srcdir/$subdir *.list]
+set generated_tests []
+foreach test_list $test_lists {
+    set generated_tests [concat \
+			     $generated_tests \
+			     [generate_tests $test_list]]
+}
+
 # Gather a list of all tests.
 set tests [lsort [concat \
 		      [find $srcdir/$subdir *.c] \
-		      [find $srcdir/$subdir/../libgomp.c-c++-common *.c]]]
+		      [find $srcdir/$subdir/../libgomp.c-c++-common *.c] \
+		      $generated_tests]]
 
 set ld_library_path $always_ld_library_path
 append ld_library_path [gcc-set-multilib-library-path $GCC_UNDER_TEST]
diff --git a/libgomp/testsuite/libgomp.c/for-1.h b/libgomp/testsuite/libgomp.c/for-1.h
new file mode 100644
index 00000000000..fa82c5b20d7
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/for-1.h
@@ -0,0 +1,25 @@ 
+#define S
+#define N(x) M(x, G, static)
+#include "for-2.h"
+#undef S
+#undef N
+#define S schedule(static, 32)
+#define N(x) M(x, G, static32)
+#include "for-2.h"
+#undef S
+#undef N
+#define S schedule(auto)
+#define N(x) M(x, G, auto)
+#include "for-2.h"
+#undef S
+#undef N
+#define S schedule(guided, 32)
+#define N(x) M(x, G, guided32)
+#include "for-2.h"
+#undef S
+#undef N
+#define S schedule(runtime)
+#define N(x) M(x, G, runtime)
+#include "for-2.h"
+#undef S
+#undef N
diff --git a/libgomp/testsuite/libgomp.c/for-2.h b/libgomp/testsuite/libgomp.c/for-2.h
new file mode 100644
index 00000000000..6d8e34974fd
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/for-2.h
@@ -0,0 +1,313 @@ 
+#ifndef VARS
+#define VARS
+int a[1500];
+float b[10][15][10];
+__attribute__((noreturn)) void
+noreturn (void)
+{
+  for (;;);
+}
+#endif
+#ifndef SC
+#define SC
+#endif
+#ifndef OMPTGT
+#define OMPTGT
+#endif
+#ifndef OMPTO
+#define OMPTO(v) do {} while (0)
+#endif
+#ifndef OMPFROM
+#define OMPFROM(v) do {} while (0)
+#endif
+
+__attribute__((noinline, noclone)) void
+N(f0) (void)
+{
+  int i;
+  OMPTGT
+#pragma omp F S
+  for (i = 0; i < 1500; i++)
+    a[i] += 2;
+}
+
+__attribute__((noinline, noclone)) void
+N(f1) (void)
+{
+  OMPTGT
+#pragma omp F S
+  for (unsigned int i = __INT_MAX__; i < 3000U + __INT_MAX__; i += 2)
+    a[(i - __INT_MAX__) >> 1] -= 2;
+}
+
+__attribute__((noinline, noclone)) void
+N(f2) (void)
+{
+  unsigned long long i;
+  OMPTGT
+#pragma omp F S
+  for (i = __LONG_LONG_MAX__ + 4500ULL - 27;
+       i > __LONG_LONG_MAX__ - 27ULL; i -= 3)
+    a[(i + 26LL - __LONG_LONG_MAX__) / 3] -= 4;
+}
+
+__attribute__((noinline, noclone)) void
+N(f3) (long long n1, long long n2, long long s3)
+{
+  OMPTGT
+#pragma omp F S
+  for (long long i = n1 + 23; i > n2 - 25; i -= s3)
+    a[i + 48] += 7;
+}
+
+__attribute__((noinline, noclone)) void
+N(f4) (void)
+{
+  unsigned int i;
+  OMPTGT
+#pragma omp F S
+  for (i = 30; i < 20; i += 2)
+    a[i] += 10;
+}
+
+__attribute__((noinline, noclone)) void
+N(f5) (int n11, int n12, int n21, int n22, int n31, int n32,
+       int s1, int s2, int s3)
+{
+  SC int v1, v2, v3;
+  OMPTGT
+#pragma omp F S collapse(3)
+  for (v1 = n11; v1 < n12; v1 += s1)
+    for (v2 = n21; v2 < n22; v2 += s2)
+      for (v3 = n31; v3 < n32; v3 += s3)
+	b[v1][v2][v3] += 2.5;
+}
+
+__attribute__((noinline, noclone)) void
+N(f6) (int n11, int n12, int n21, int n22, long long n31, long long n32,
+       int s1, int s2, long long int s3)
+{
+  SC int v1, v2;
+  SC long long v3;
+  OMPTGT
+#pragma omp F S collapse(3)
+  for (v1 = n11; v1 > n12; v1 += s1)
+    for (v2 = n21; v2 > n22; v2 += s2)
+      for (v3 = n31; v3 > n32; v3 += s3)
+	b[v1][v2 / 2][v3] -= 4.5;
+}
+
+__attribute__((noinline, noclone)) void
+N(f7) (void)
+{
+  SC unsigned int v1, v3;
+  SC unsigned long long v2;
+  OMPTGT
+#pragma omp F S collapse(3)
+  for (v1 = 0; v1 < 20; v1 += 2)
+    for (v2 = __LONG_LONG_MAX__ + 16ULL;
+	 v2 > __LONG_LONG_MAX__ - 29ULL; v2 -= 3)
+      for (v3 = 10; v3 > 0; v3--)
+	b[v1 >> 1][(v2 - __LONG_LONG_MAX__ + 64) / 3 - 12][v3 - 1] += 5.5;
+}
+
+__attribute__((noinline, noclone)) void
+N(f8) (void)
+{
+  SC long long v1, v2, v3;
+  OMPTGT
+#pragma omp F S collapse(3)
+  for (v1 = 0; v1 < 20; v1 += 2)
+    for (v2 = 30; v2 < 20; v2++)
+      for (v3 = 10; v3 < 0; v3--)
+	b[v1][v2][v3] += 5.5;
+}
+
+__attribute__((noinline, noclone)) void
+N(f9) (void)
+{
+  int i;
+  OMPTGT
+#pragma omp F S
+  for (i = 20; i < 10; i++)
+    {
+      a[i] += 2;
+      noreturn ();
+      a[i] -= 4;
+    }
+}
+
+__attribute__((noinline, noclone)) void
+N(f10) (void)
+{
+  SC int i;
+  OMPTGT
+#pragma omp F S collapse(3)
+  for (i = 0; i < 10; i++)
+    for (int j = 10; j < 8; j++)
+      for (long k = -10; k < 10; k++)
+	{
+	  b[i][j][k] += 4;
+	  noreturn ();
+	  b[i][j][k] -= 8;
+	}
+}
+
+__attribute__((noinline, noclone)) void
+N(f11) (int n)
+{
+  int i;
+  OMPTGT
+#pragma omp F S
+  for (i = 20; i < n; i++)
+    {
+      a[i] += 8;
+      noreturn ();
+      a[i] -= 16;
+    }
+}
+
+__attribute__((noinline, noclone)) void
+N(f12) (int n)
+{
+  SC int i;
+  OMPTGT
+#pragma omp F S collapse(3)
+  for (i = 0; i < 10; i++)
+    for (int j = n; j < 8; j++)
+      for (long k = -10; k < 10; k++)
+	{
+	  b[i][j][k] += 16;
+	  noreturn ();
+	  b[i][j][k] -= 32;
+	}
+}
+
+__attribute__((noinline, noclone)) void
+N(f13) (void)
+{
+  int *i;
+  OMPTGT
+#pragma omp F S
+  for (i = a; i < &a[1500]; i++)
+    i[0] += 2;
+}
+
+__attribute__((noinline, noclone)) void
+N(f14) (void)
+{
+  SC float *i;
+  OMPTGT
+#pragma omp F S collapse(3)
+  for (i = &b[0][0][0]; i < &b[0][0][10]; i++)
+    for (float *j = &b[0][15][0]; j > &b[0][0][0]; j -= 10)
+      for (float *k = &b[0][0][10]; k > &b[0][0][0]; --k)
+	b[i - &b[0][0][0]][(j - &b[0][0][0]) / 10 - 1][(k - &b[0][0][0]) - 1]
+	  -= 3.5;
+}
+
+__attribute__((noinline, noclone)) int
+N(test) (void)
+{
+  int i, j, k;
+  for (i = 0; i < 1500; i++)
+    a[i] = i - 25;
+  OMPTO (a);
+  N(f0) ();
+  OMPFROM (a);
+  for (i = 0; i < 1500; i++)
+    if (a[i] != i - 23)
+      return 1;
+  N(f1) ();
+  OMPFROM (a);
+  for (i = 0; i < 1500; i++)
+    if (a[i] != i - 25)
+      return 1;
+  N(f2) ();
+  OMPFROM (a);
+  for (i = 0; i < 1500; i++)
+    if (a[i] != i - 29)
+      return 1;
+  N(f3) (1500LL - 1 - 23 - 48, -1LL + 25 - 48, 1LL);
+  OMPFROM (a);
+  for (i = 0; i < 1500; i++)
+    if (a[i] != i - 22)
+      return 1;
+  N(f3) (1500LL - 1 - 23 - 48, 1500LL - 1, 7LL);
+  OMPFROM (a);
+  for (i = 0; i < 1500; i++)
+    if (a[i] != i - 22)
+      return 1;
+  N(f4) ();
+  OMPFROM (a);
+  for (i = 0; i < 1500; i++)
+    if (a[i] != i - 22)
+      return 1;
+  for (i = 0; i < 10; i++)
+    for (j = 0; j < 15; j++)
+      for (k = 0; k < 10; k++)
+	b[i][j][k] = i - 2.5 + 1.5 * j - 1.5 * k;
+  OMPTO (b);
+  N(f5) (0, 10, 0, 15, 0, 10, 1, 1, 1);
+  OMPFROM (b);
+  for (i = 0; i < 10; i++)
+    for (j = 0; j < 15; j++)
+      for (k = 0; k < 10; k++)
+	if (b[i][j][k] != i + 1.5 * j - 1.5 * k)
+	  return 1;
+  N(f5) (0, 10, 30, 15, 0, 10, 4, 5, 6);
+  OMPFROM (b);
+  for (i = 0; i < 10; i++)
+    for (j = 0; j < 15; j++)
+      for (k = 0; k < 10; k++)
+	if (b[i][j][k] != i + 1.5 * j - 1.5 * k)
+	  return 1;
+  N(f6) (9, -1, 29, 0, 9, -1, -1, -2, -1);
+  OMPFROM (b);
+  for (i = 0; i < 10; i++)
+    for (j = 0; j < 15; j++)
+      for (k = 0; k < 10; k++)
+	if (b[i][j][k] != i - 4.5 + 1.5 * j - 1.5 * k)
+	  return 1;
+  N(f7) ();
+  OMPFROM (b);
+  for (i = 0; i < 10; i++)
+    for (j = 0; j < 15; j++)
+      for (k = 0; k < 10; k++)
+	if (b[i][j][k] != i + 1.0 + 1.5 * j - 1.5 * k)
+	  return 1;
+  N(f8) ();
+  OMPFROM (b);
+  for (i = 0; i < 10; i++)
+    for (j = 0; j < 15; j++)
+      for (k = 0; k < 10; k++)
+	if (b[i][j][k] != i + 1.0 + 1.5 * j - 1.5 * k)
+	  return 1;
+  N(f9) ();
+  N(f10) ();
+  N(f11) (10);
+  N(f12) (12);
+  OMPFROM (a);
+  OMPFROM (b);
+  for (i = 0; i < 1500; i++)
+    if (a[i] != i - 22)
+      return 1;
+  for (i = 0; i < 10; i++)
+    for (j = 0; j < 15; j++)
+      for (k = 0; k < 10; k++)
+	if (b[i][j][k] != i + 1.0 + 1.5 * j - 1.5 * k)
+	  return 1;
+  N(f13) ();
+  N(f14) ();
+  OMPFROM (a);
+  OMPFROM (b);
+  for (i = 0; i < 1500; i++)
+    if (a[i] != i - 20)
+      return 1;
+  for (i = 0; i < 10; i++)
+    for (j = 0; j < 15; j++)
+      for (k = 0; k < 10; k++)
+	if (b[i][j][k] != i - 2.5 + 1.5 * j - 1.5 * k)
+	  return 1;
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/for-3.c b/libgomp/testsuite/libgomp.c/for-3.c
new file mode 100644
index 00000000000..d040f11b8e6
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/for-3.c
@@ -0,0 +1,123 @@ 
+/* { dg-additional-options "-std=gnu99" } */
+
+extern void abort ();
+
+#define M(x, y, z) O(x, y, z)
+#define O(x, y, z) x ## _ ## y ## _ ## z
+
+#ifndef ONE_TEST
+#define TEST_ALL 1
+#else
+#define TEST_ALL 0
+#endif
+
+#pragma omp declare target
+
+#if TEST_ALL || TEST_NR == 1
+#define F distribute
+#define G d
+#define S
+#define N(x) M(x, G, normal)
+#include "for-2.h"
+#undef S
+#undef N
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || TEST_NR == 2
+#define F distribute
+#define G d_ds128
+#define S dist_schedule(static, 128)
+#define N(x) M(x, G, normal)
+#include "for-2.h"
+#undef S
+#undef N
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || TEST_NR == 3
+#define F distribute simd
+#define G ds
+#define S
+#define N(x) M(x, G, normal)
+#include "for-2.h"
+#undef S
+#undef N
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || TEST_NR == 4
+#define F distribute simd
+#define G ds_ds128
+#define S dist_schedule(static, 128)
+#define N(x) M(x, G, normal)
+#include "for-2.h"
+#undef S
+#undef N
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || (5 <= TEST_NR && TEST_NR <= 9)
+#define F distribute parallel for
+#define G dpf
+#include "for-1.h"
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || (10 <= TEST_NR && TEST_NR <= 14)
+#define F distribute parallel for dist_schedule(static, 128)
+#define G dpf_ds128
+#include "for-1.h"
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || (15 <= TEST_NR && TEST_NR <= 19)
+#define F distribute parallel for simd
+#define G dpfs
+#include "for-1.h"
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || (20 <= TEST_NR && TEST_NR <= 24)
+#define F distribute parallel for simd dist_schedule(static, 128)
+#define G dpfs_ds128
+#include "for-1.h"
+#undef F
+#undef G
+#endif
+
+#pragma omp end declare target
+
+int
+main ()
+{
+  int err = 0;
+
+  #pragma omp target teams reduction(|:err)
+  {
+#define DO_TEST_1(test) \
+    do {	      \
+      err |= test (); \
+    } while (0)
+
+#ifdef ONE_TEST
+  DO_TEST_1 (ONE_TEST);
+#else
+#define DO_TEST(test) DO_TEST_1(test);
+#include "for-3.list"
+#undef DO_TEST
+#endif
+#undef DO_TEST_1
+  }
+
+  if (err)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/for-3.list b/libgomp/testsuite/libgomp.c/for-3.list
new file mode 100644
index 00000000000..6fb25a581dc
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/for-3.list
@@ -0,0 +1,24 @@ 
+DO_TEST (test_d_normal)
+DO_TEST (test_d_ds128_normal)
+DO_TEST (test_ds_normal)
+DO_TEST (test_ds_ds128_normal)
+DO_TEST (test_dpf_static)
+DO_TEST (test_dpf_static32)
+DO_TEST (test_dpf_auto)
+DO_TEST (test_dpf_guided32)
+DO_TEST (test_dpf_runtime)
+DO_TEST (test_dpf_ds128_static)
+DO_TEST (test_dpf_ds128_static32)
+DO_TEST (test_dpf_ds128_auto)
+DO_TEST (test_dpf_ds128_guided32)
+DO_TEST (test_dpf_ds128_runtime)
+DO_TEST (test_dpfs_static)
+DO_TEST (test_dpfs_static32)
+DO_TEST (test_dpfs_auto)
+DO_TEST (test_dpfs_guided32)
+DO_TEST (test_dpfs_runtime)
+DO_TEST (test_dpfs_ds128_static)
+DO_TEST (test_dpfs_ds128_static32)
+DO_TEST (test_dpfs_ds128_auto)
+DO_TEST (test_dpfs_ds128_guided32)
+DO_TEST (test_dpfs_ds128_runtime)
diff --git a/libgomp/testsuite/libgomp.c/for-5.c b/libgomp/testsuite/libgomp.c/for-5.c
new file mode 100644
index 00000000000..8f125792731
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/for-5.c
@@ -0,0 +1,161 @@ 
+/* { dg-additional-options "-std=gnu99" } */
+
+extern void abort ();
+
+#define M(x, y, z) O(x, y, z)
+#define O(x, y, z) x ## _ ## y ## _ ## z
+
+#ifndef ONE_TEST
+#define TEST_ALL 1
+#else
+#define TEST_ALL 0
+#endif
+
+#pragma omp declare target
+
+#define F for
+#define G f
+#define S
+#define N(x) M(x, G, normal)
+#include "for-2.h"
+#undef S
+#undef N
+#undef F
+#undef G
+
+#pragma omp end declare target
+
+#undef OMPFROM
+#undef OMPTO
+#define DO_PRAGMA(x) _Pragma (#x)
+#define OMPFROM(v) DO_PRAGMA (omp target update from(v))
+#define OMPTO(v) DO_PRAGMA (omp target update to(v))
+
+#if TEST_ALL || (1 <= TEST_NR && TEST_NR <= 5)
+#define F target parallel for
+#define G tpf
+#include "for-1.h"
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || TEST_NR == 6
+#define F target simd
+#define G t_simd
+#define S
+#define N(x) M(x, G, normal)
+#include "for-2.h"
+#undef S
+#undef N
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || (7 <= TEST_NR && TEST_NR <= 11)
+#define F target parallel for simd
+#define G tpf_simd
+#include "for-1.h"
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || TEST_NR == 12
+#define F target teams distribute
+#define G ttd
+#define S
+#define N(x) M(x, G, normal)
+#include "for-2.h"
+#undef S
+#undef N
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || TEST_NR == 13
+#define F target teams distribute
+#define G ttd_ds128
+#define S dist_schedule(static, 128)
+#define N(x) M(x, G, normal)
+#include "for-2.h"
+#undef S
+#undef N
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || TEST_NR == 14
+#define F target teams distribute simd
+#define G ttds
+#define S
+#define N(x) M(x, G, normal)
+#include "for-2.h"
+#undef S
+#undef N
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || TEST_NR == 15
+#define F target teams distribute simd
+#define G ttds_ds128
+#define S dist_schedule(static, 128)
+#define N(x) M(x, G, normal)
+#include "for-2.h"
+#undef S
+#undef N
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || (16 <= TEST_NR && TEST_NR <= 20)
+#define F target teams distribute parallel for
+#define G ttdpf
+#include "for-1.h"
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || (21 <= TEST_NR && TEST_NR <= 25)
+#define F target teams distribute parallel for dist_schedule(static, 128)
+#define G ttdpf_ds128
+#include "for-1.h"
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || (26 <= TEST_NR && TEST_NR <= 30)
+#define F target teams distribute parallel for simd
+#define G ttdpfs
+#include "for-1.h"
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || (31 <= TEST_NR && TEST_NR <= 35)
+#define F target teams distribute parallel for simd dist_schedule(static, 128)
+#define G ttdpfs_ds128
+#include "for-1.h"
+#undef F
+#undef G
+#endif
+
+int
+main ()
+{
+#define DO_TEST_1(test)				\
+  do {						\
+    if (test ())				\
+      abort ();					\
+  } while (0)
+
+#ifdef ONE_TEST
+  DO_TEST_1 (ONE_TEST);
+#else
+#define DO_TEST(test) DO_TEST_1 (test);
+#include "for-5.list"
+#undef DO_TEST
+#endif
+#undef DO_TEST_1
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/for-5.list b/libgomp/testsuite/libgomp.c/for-5.list
new file mode 100644
index 00000000000..48d0c3aa7f6
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/for-5.list
@@ -0,0 +1,35 @@ 
+DO_TEST (test_tpf_static)
+DO_TEST (test_tpf_static32)
+DO_TEST (test_tpf_auto)
+DO_TEST (test_tpf_guided32)
+DO_TEST (test_tpf_runtime)
+DO_TEST (test_t_simd_normal)
+DO_TEST (test_tpf_simd_static)
+DO_TEST (test_tpf_simd_static32)
+DO_TEST (test_tpf_simd_auto)
+DO_TEST (test_tpf_simd_guided32)
+DO_TEST (test_tpf_simd_runtime)
+DO_TEST (test_ttd_normal)
+DO_TEST (test_ttd_ds128_normal)
+DO_TEST (test_ttds_normal)
+DO_TEST (test_ttds_ds128_normal)
+DO_TEST (test_ttdpf_static)
+DO_TEST (test_ttdpf_static32)
+DO_TEST (test_ttdpf_auto)
+DO_TEST (test_ttdpf_guided32)
+DO_TEST (test_ttdpf_runtime)
+DO_TEST (test_ttdpf_ds128_static)
+DO_TEST (test_ttdpf_ds128_static32)
+DO_TEST (test_ttdpf_ds128_auto)
+DO_TEST (test_ttdpf_ds128_guided32)
+DO_TEST (test_ttdpf_ds128_runtime)
+DO_TEST (test_ttdpfs_static)
+DO_TEST (test_ttdpfs_static32)
+DO_TEST (test_ttdpfs_auto)
+DO_TEST (test_ttdpfs_guided32)
+DO_TEST (test_ttdpfs_runtime)
+DO_TEST (test_ttdpfs_ds128_static)
+DO_TEST (test_ttdpfs_ds128_static32)
+DO_TEST (test_ttdpfs_ds128_auto)
+DO_TEST (test_ttdpfs_ds128_guided32)
+DO_TEST (test_ttdpfs_ds128_runtime)
diff --git a/libgomp/testsuite/libgomp.c/for-6.c b/libgomp/testsuite/libgomp.c/for-6.c
new file mode 100644
index 00000000000..50a866a3cf7
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/for-6.c
@@ -0,0 +1,135 @@ 
+/* { dg-additional-options "-std=gnu99" } */
+
+extern void abort ();
+
+#define M(x, y, z) O(x, y, z)
+#define O(x, y, z) x ## _ ## y ## _ ## z
+
+#ifndef ONE_TEST
+#define TEST_ALL 1
+#else
+#define TEST_ALL 0
+#endif
+
+#pragma omp declare target
+
+#define F for
+#define G f
+#define S
+#define N(x) M(x, G, normal)
+#include "for-2.h"
+#undef S
+#undef N
+#undef F
+#undef G
+
+#pragma omp end declare target
+
+#undef OMPTGT
+#undef OMPFROM
+#undef OMPTO
+#define DO_PRAGMA(x) _Pragma (#x)
+#define OMPTGT DO_PRAGMA (omp target)
+#define OMPFROM(v) DO_PRAGMA (omp target update from(v))
+#define OMPTO(v) DO_PRAGMA (omp target update to(v))
+
+#if TEST_ALL || TEST_NR == 1
+#define F teams distribute
+#define G td
+#define S
+#define N(x) M(x, G, normal)
+#include "for-2.h"
+#undef S
+#undef N
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || TEST_NR == 2
+#define F teams distribute
+#define G td_ds128
+#define S dist_schedule(static, 128)
+#define N(x) M(x, G, normal)
+#include "for-2.h"
+#undef S
+#undef N
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || TEST_NR == 3
+#define F teams distribute simd
+#define G tds
+#define S
+#define N(x) M(x, G, normal)
+#include "for-2.h"
+#undef S
+#undef N
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || TEST_NR == 4
+#define F teams distribute simd
+#define G tds_ds128
+#define S dist_schedule(static, 128)
+#define N(x) M(x, G, normal)
+#include "for-2.h"
+#undef S
+#undef N
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || (5 <= TEST_NR && TEST_NR <= 9)
+#define F teams distribute parallel for
+#define G tdpf
+#include "for-1.h"
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || (10 <= TEST_NR && TEST_NR <= 14)
+#define F teams distribute parallel for dist_schedule(static, 128)
+#define G tdpf_ds128
+#include "for-1.h"
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || (15 <= TEST_NR && TEST_NR <= 19)
+#define F teams distribute parallel for simd
+#define G tdpfs
+#include "for-1.h"
+#undef F
+#undef G
+#endif
+
+#if TEST_ALL || (20 <= TEST_NR && TEST_NR <= 24)
+#define F teams distribute parallel for simd dist_schedule(static, 128)
+#define G tdpfs_ds128
+#include "for-1.h"
+#undef F
+#undef G
+#endif
+
+int
+main ()
+{
+#define DO_TEST_1(test)				\
+  do {						\
+    if (test ())				\
+      abort ();					\
+  } while (0)
+
+#ifdef ONE_TEST
+  DO_TEST_1 (ONE_TEST);
+#else
+#define DO_TEST(test) DO_TEST_1 (test);
+#include "for-6.list"
+#undef DO_TEST
+#endif
+#undef DO_TEST_1
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/for-6.list b/libgomp/testsuite/libgomp.c/for-6.list
new file mode 100644
index 00000000000..438ecff0a2f
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/for-6.list
@@ -0,0 +1,24 @@ 
+DO_TEST (test_td_normal)
+DO_TEST (test_td_ds128_normal)
+DO_TEST (test_tds_normal)
+DO_TEST (test_tds_ds128_normal)
+DO_TEST (test_tdpf_static)
+DO_TEST (test_tdpf_static32)
+DO_TEST (test_tdpf_auto)
+DO_TEST (test_tdpf_guided32)
+DO_TEST (test_tdpf_runtime)
+DO_TEST (test_tdpf_ds128_static)
+DO_TEST (test_tdpf_ds128_static32)
+DO_TEST (test_tdpf_ds128_auto)
+DO_TEST (test_tdpf_ds128_guided32)
+DO_TEST (test_tdpf_ds128_runtime)
+DO_TEST (test_tdpfs_static)
+DO_TEST (test_tdpfs_static32)
+DO_TEST (test_tdpfs_auto)
+DO_TEST (test_tdpfs_guided32)
+DO_TEST (test_tdpfs_runtime)
+DO_TEST (test_tdpfs_ds128_static)
+DO_TEST (test_tdpfs_ds128_static32)
+DO_TEST (test_tdpfs_ds128_auto)
+DO_TEST (test_tdpfs_ds128_guided32)
+DO_TEST (test_tdpfs_ds128_runtime)
diff --git a/libgomp/testsuite/libgomp.c/target-print-1.c b/libgomp/testsuite/libgomp.c/target-print-1.c
new file mode 100644
index 00000000000..5857b875ced
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/target-print-1.c
@@ -0,0 +1,17 @@ 
+/* Ensure that printf on the offload device works.  */
+
+/* { dg-do run } */
+/* { dg-output "The answer is 42(\n|\r\n|\r)+" } */
+
+#include <stdio.h>
+
+int var = 42;
+
+int
+main ()
+{
+#pragma omp target
+    {
+      printf ("The answer is %d\n", var);
+    }
+}
diff --git a/libgomp/testsuite/libgomp.fortran/target-print-1.f90 b/libgomp/testsuite/libgomp.fortran/target-print-1.f90
new file mode 100644
index 00000000000..f4f00e23373
--- /dev/null
+++ b/libgomp/testsuite/libgomp.fortran/target-print-1.f90
@@ -0,0 +1,15 @@ 
+! Ensure that printf on the offload device works.
+
+! { dg-do run }
+! { dg-output "The answer is 42(\n|\r\n|\r)+" }
+! { dg-xfail-if "no write for nvidia" { openacc_nvidia_accel_selected } }
+
+program main
+  implicit none
+  integer :: var = 42
+
+!$omp target
+  write (0, '("The answer is ", I2)') var
+!$omp end target
+
+end program main
diff --git a/libgomp/testsuite/libgomp.oacc-c++/c++.exp b/libgomp/testsuite/libgomp.oacc-c++/c++.exp
index b8b44518b86..1285a6a1c6d 100644
--- a/libgomp/testsuite/libgomp.oacc-c++/c++.exp
+++ b/libgomp/testsuite/libgomp.oacc-c++/c++.exp
@@ -111,6 +111,10 @@  if { $lang_test_file_found } {
 
 		set acc_mem_shared 0
 	    }
+	    amdgcn* {
+		set acc_mem_shared 0
+		set tagopt "-DACC_DEVICE_TYPE_gcn=\"$offload_target_openacc\""
+	    }
 	    default {
 		error "Unknown OpenACC device type: $openacc_device_type (offload target: $offload_target)"
 	    }
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/atomic_capture-2.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/atomic_capture-2.c
index 842f2de4722..6a24a2dde6a 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/atomic_capture-2.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/atomic_capture-2.c
@@ -37,11 +37,9 @@  main(int argc, char **argv)
       imin = idata[i] < imin ? idata[i] : imin;
     }
 
-  if (imax != 1234 || imin != 0)
+  if (imax != 1234 || imin < 0 || imin > 1)
     abort ();
 
-  return 0;
-
   igot = 0;
   iexp = 32;
 
@@ -443,17 +441,16 @@  main(int argc, char **argv)
     }
   }
 
+  int ones = 0, zeros = 0;
+
   for (i = 0; i < N; i++)
-    if (i % 2 == 0)
-      {
-	if (idata[i] != 1)
-	  abort ();
-      }
-    else
-      {
-	if (idata[i] != 0)
-	  abort ();
-      }
+    if (idata[i] == 1)
+      ones++;
+    else if (idata[i] == 0)
+      zeros++;
+
+  if (ones != N / 2 || zeros != N / 2)
+    abort ();
 
   if (iexp != igot)
     abort ();
@@ -491,17 +488,16 @@  main(int argc, char **argv)
       }
   }
 
+  ones = zeros = 0;
+
   for (i = 0; i < N; i++)
-    if (i % 2 == 0)
-      {
-	if (idata[i] != 0)
-	  abort ();
-      }
-    else
-      {
-	if (idata[i] != 1)
-	  abort ();
-      }
+    if (idata[i] == 1)
+      ones++;
+    else if (idata[i] == 0)
+      zeros++;
+
+  if (ones != N / 2 || zeros != N / 2)
+    abort ();
 
   if (iexp != igot)
     abort ();
@@ -579,7 +575,7 @@  main(int argc, char **argv)
   if (lexp != lgot)
     abort ();
 
-  lgot = 2LL;
+  lgot = 2LL << N;
   lexp = 2LL;
 
 #pragma acc data copy (lgot, ldata[0:N])
@@ -587,7 +583,7 @@  main(int argc, char **argv)
 #pragma acc parallel loop
     for (i = 0; i < N; i++)
       {
-        long long expr = 1LL << N;
+        long long expr = 2LL;
 
 #pragma acc atomic capture
         { lgot = lgot / expr; ldata[i] = lgot; }
@@ -1450,17 +1446,16 @@  main(int argc, char **argv)
       }
   }
 
+  ones = zeros = 0;
+
   for (i = 0; i < N; i++)
-    if (i % 2 == 0)
-      {
-	if (fdata[i] != 1.0)
-	  abort ();
-      }
-    else
-      {
-	if (fdata[i] != 0.0)
-	  abort ();
-      }
+    if (fdata[i] == 1.0)
+      ones++;
+    else if (fdata[i] == 0.0)
+      zeros++;
+
+  if (ones != N / 2 || zeros != N / 2)
+    abort ();
 
   if (fexp != fgot)
     abort ();
@@ -1498,17 +1493,16 @@  main(int argc, char **argv)
       }
   }
 
+  ones = zeros = 0;
+
   for (i = 0; i < N; i++)
-    if (i % 2 == 0)
-      {
-	if (fdata[i] != 0.0)
-	  abort ();
-      }
-    else
-      {
-	if (fdata[i] != 1.0)
-	  abort ();
-      }
+    if (fdata[i] == 1.0)
+      ones++;
+    else if (fdata[i] == 0.0)
+      zeros++;
+
+  if (ones != N / 2 || zeros != N / 2)
+    abort ();
 
   if (fexp != fgot)
     abort ();
@@ -1569,7 +1563,7 @@  main(int argc, char **argv)
     abort ();
 
   fgot = 8192.0*8192.0*64.0;
-  fexp = 1.0;
+  fexp = fgot;
 
 #pragma acc data copy (fgot, fdata[0:N])
   {
@@ -1586,15 +1580,15 @@  main(int argc, char **argv)
   if (fexp != fgot)
     abort ();
 
-  fgot = 4.0;
-  fexp = 4.0;
+  fgot = 2.0 * (1LL << N);
+  fexp = 2.0;
 
 #pragma acc data copy (fgot, fdata[0:N])
   {
 #pragma acc parallel loop
     for (i = 0; i < N; i++)
       {
-        long long expr = 1LL << N;
+        long long expr = 2LL;
 
 #pragma acc atomic capture
         { fgot = fgot / expr; fdata[i] = fgot; }
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-auto-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-auto-1.c
index 9642b391942..05946c900f1 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-auto-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-auto-1.c
@@ -1,3 +1,6 @@ 
+/* AMD GCN does not use 32-lane vectors.
+   { dg-skip-if "unsuitable dimensions" { openacc_amdgcn_accel_selected } { "*" } { "" } } */
+
 /* { dg-additional-options "-fopenacc-dim=32" } */
 
 #include <stdio.h>
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-gwv-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-gwv-1.c
index 766e5782b46..5c843012061 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-gwv-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-gwv-1.c
@@ -9,11 +9,13 @@  int main ()
   int ix;
   int exit = 0;
   int ondev = 0;
+  int gangsize, workersize, vectorsize;
 
   for (ix = 0; ix < N;ix++)
     ary[ix] = -1;
   
-#pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) copy(ary) copy(ondev)
+#pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+	    copy(ary) copy(ondev) copyout(gangsize, workersize, vectorsize)
   {
 #pragma acc loop gang worker vector
     for (unsigned ix = 0; ix < N; ix++)
@@ -32,6 +34,10 @@  int main ()
 	else
 	  ary[ix] = ix;
       }
+
+    gangsize = __builtin_goacc_parlevel_size (GOMP_DIM_GANG);
+    workersize = __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
+    vectorsize = __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
   }
 
   for (ix = 0; ix < N; ix++)
@@ -39,11 +45,12 @@  int main ()
       int expected = ix;
       if(ondev)
 	{
-	  int chunk_size = (N + 32*32*32 - 1) / (32*32*32);
+	  int chunk_size = (N + gangsize * workersize * vectorsize - 1)
+			   / (gangsize * workersize * vectorsize);
 	  
-	  int g = ix / (chunk_size * 32 * 32);
-	  int w = ix / 32 % 32;
-	  int v = ix % 32;
+	  int g = ix / (chunk_size * workersize * vectorsize);
+	  int w = (ix / vectorsize) % workersize;
+	  int v = ix % vectorsize;
 
 	  expected = (g << 16) | (w << 8) | v;
 	}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-gwv-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-gwv-1.c
index 0bec6e19510..9c4a85f7b16 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-gwv-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-gwv-1.c
@@ -8,8 +8,10 @@  int main ()
   int ix;
   int ondev = 0;
   int t = 0, h = 0;
-  
-#pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) copy(ondev)
+  int gangsize, workersize, vectorsize;
+
+#pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) \
+	copy(ondev) copyout(gangsize, workersize, vectorsize)
   {
 #pragma acc loop gang worker vector reduction(+:t)
     for (unsigned ix = 0; ix < N; ix++)
@@ -28,18 +30,22 @@  int main ()
 	  }
 	t += val;
       }
+    gangsize = __builtin_goacc_parlevel_size (GOMP_DIM_GANG);
+    workersize = __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
+    vectorsize = __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
   }
 
   for (ix = 0; ix < N; ix++)
     {
       int val = ix;
-      if(ondev)
+      if (ondev)
 	{
-	  int chunk_size = (N + 32*32*32 - 1) / (32*32*32);
+	  int chunk_size = (N + gangsize * workersize * vectorsize - 1)
+			   / (gangsize * workersize * vectorsize);
 	  
-	  int g = ix / (chunk_size * 32 * 32);
-	  int w = ix / 32 % 32;
-	  int v = ix % 32;
+	  int g = ix / (chunk_size * vectorsize * workersize);
+	  int w = ix / vectorsize % workersize;
+	  int v = ix % vectorsize;
 
 	  val = (g << 16) | (w << 8) | v;
 	}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-1.c
index da4921d15f9..1173c1f57bb 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-1.c
@@ -9,8 +9,9 @@  int main ()
   int ix;
   int ondev = 0;
   int t = 0,  h = 0;
+  int vectorsize;
 
-#pragma acc parallel vector_length(32) copy(ondev)
+#pragma acc parallel vector_length(32) copy(ondev) copyout(vectorsize)
   {
 #pragma acc loop vector reduction (+:t)
     for (unsigned ix = 0; ix < N; ix++)
@@ -29,6 +30,7 @@  int main ()
 	  }
 	t += val;
       }
+    vectorsize = __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
   }
 
   for (ix = 0; ix < N; ix++)
@@ -38,7 +40,7 @@  int main ()
 	{
 	  int g = 0;
 	  int w = 0;
-	  int v = ix % 32;
+	  int v = ix % vectorsize;
 
 	  val = (g << 16) | (w << 8) | v;
 	}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-2.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-2.c
index 15e2bc2f83b..84c2296a7b1 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-2.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-v-2.c
@@ -9,8 +9,9 @@  int main ()
   int ix;
   int ondev = 0;
   int q = 0,  h = 0;
+  int vectorsize;
 
-#pragma acc parallel vector_length(32) copy(q) copy(ondev)
+#pragma acc parallel vector_length(32) copy(q) copy(ondev) copyout(vectorsize)
   {
     int t = q;
     
@@ -32,6 +33,7 @@  int main ()
 	t += val;
       }
     q = t;
+    vectorsize = __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
   }
 
   for (ix = 0; ix < N; ix++)
@@ -41,7 +43,7 @@  int main ()
 	{
 	  int g = 0;
 	  int w = 0;
-	  int v = ix % 32;
+	  int v = ix % vectorsize;
 
 	  val = (g << 16) | (w << 8) | v;
 	}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-1.c
index a1bb845987d..7344fa8bf00 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-1.c
@@ -9,8 +9,10 @@  int main ()
   int ix;
   int ondev = 0;
   int t = 0,  h = 0;
+  int workersize;
 
-#pragma acc parallel num_workers(32) vector_length(32) copy(ondev)
+#pragma acc parallel num_workers(32) vector_length(32) copy(ondev) \
+	    copyout(workersize)
   {
 #pragma acc loop worker reduction(+:t)
     for (unsigned ix = 0; ix < N; ix++)
@@ -29,6 +31,7 @@  int main ()
 	  }
 	t += val;
       }
+    workersize = __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
   }
 
   for (ix = 0; ix < N; ix++)
@@ -37,7 +40,7 @@  int main ()
       if(ondev)
 	{
 	  int g = 0;
-	  int w = ix % 32;
+	  int w = ix % workersize;
 	  int v = 0;
 
 	  val = (g << 16) | (w << 8) | v;
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-2.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-2.c
index ae43bb47d8f..d99877ab8a9 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-2.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-w-2.c
@@ -9,8 +9,10 @@  int main ()
   int ix;
   int ondev = 0;
   int q = 0,  h = 0;
+  int workersize;
 
-#pragma acc parallel num_workers(32) vector_length(32) copy(q) copy(ondev)
+#pragma acc parallel num_workers(32) vector_length(32) copy(q) copy(ondev) \
+	    copyout(workersize)
   {
     int t = q;
     
@@ -32,6 +34,7 @@  int main ()
 	t += val;
       }
     q = t;
+    workersize = __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
   }
 
   for (ix = 0; ix < N; ix++)
@@ -40,7 +43,7 @@  int main ()
       if(ondev)
 	{
 	  int g = 0;
-	  int w = ix % 32;
+	  int w = ix % workersize;
 	  int v = 0;
 
 	  val = (g << 16) | (w << 8) | v;
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-wv-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-wv-1.c
index 71d3969f7b6..c360ad11e7c 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-wv-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-red-wv-1.c
@@ -8,8 +8,10 @@  int main ()
   int ix;
   int ondev = 0;
   int t = 0, h = 0;
+  int workersize, vectorsize;
   
-#pragma acc parallel num_workers(32) vector_length(32) copy(ondev)
+#pragma acc parallel num_workers(32) vector_length(32) copy(ondev) \
+	    copyout(workersize, vectorsize)
   {
 #pragma acc loop worker vector reduction (+:t)
     for (unsigned ix = 0; ix < N; ix++)
@@ -28,6 +30,8 @@  int main ()
 	  }
 	t += val;
       }
+    workersize = __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
+    vectorsize = __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
   }
 
   for (ix = 0; ix < N; ix++)
@@ -36,8 +40,8 @@  int main ()
       if(ondev)
 	{
 	  int g = 0;
-	  int w = (ix / 32) % 32;
-	  int v = ix % 32;
+	  int w = (ix / vectorsize) % workersize;
+	  int v = ix % vectorsize;
 
 	  val = (g << 16) | (w << 8) | v;
 	}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-v-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-v-1.c
index 6010cd2498a..8c858f30563 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-v-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-v-1.c
@@ -9,11 +9,13 @@  int main ()
   int ix;
   int exit = 0;
   int ondev = 0;
+  int vectorsize;
 
   for (ix = 0; ix < N;ix++)
     ary[ix] = -1;
   
-#pragma acc parallel vector_length(32) copy(ary) copy(ondev)
+#pragma acc parallel vector_length(32) copy(ary) copy(ondev) \
+	    copyout(vectorsize)
   {
 #pragma acc loop vector
     for (unsigned ix = 0; ix < N; ix++)
@@ -31,6 +33,7 @@  int main ()
 	else
 	  ary[ix] = ix;
       }
+    vectorsize = __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
   }
 
   for (ix = 0; ix < N; ix++)
@@ -40,7 +43,7 @@  int main ()
 	{
 	  int g = 0;
 	  int w = 0;
-	  int v = ix % 32;
+	  int v = ix % vectorsize;
 
 	  expected = (g << 16) | (w << 8) | v;
 	}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-w-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-w-1.c
index 10b80f197de..8731c805b79 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-w-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-w-1.c
@@ -9,12 +9,14 @@  int main ()
   int ix;
   int exit = 0;
   int ondev = 0;
+  int workersize;
 
   for (ix = 0; ix < N;ix++)
     ary[ix] = -1;
   
-#pragma acc parallel num_workers(32) vector_length(32) copy(ary) copy(ondev)
-  /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 16 } */
+#pragma acc parallel num_workers(32) vector_length(32) copy(ary) copy(ondev) \
+	    copyout(workersize)
+  /* { dg-warning "region is vector partitioned but does not contain vector partitioned code" "vector" { target *-*-* } 17 } */
   {
 #pragma acc loop worker
     for (unsigned ix = 0; ix < N; ix++)
@@ -32,6 +34,7 @@  int main ()
 	else
 	  ary[ix] = ix;
       }
+    workersize = __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
   }
 
   for (ix = 0; ix < N; ix++)
@@ -40,7 +43,7 @@  int main ()
       if(ondev)
 	{
 	  int g = 0;
-	  int w = ix % 32;
+	  int w = ix % workersize;
 	  int v = 0;
 
 	  expected = (g << 16) | (w << 8) | v;
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-wv-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-wv-1.c
index cd4cc994b82..fd4e4cf5ea9 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-wv-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/loop-wv-1.c
@@ -9,11 +9,13 @@  int main ()
   int ix;
   int exit = 0;
   int ondev = 0;
+  int workersize, vectorsize;
 
   for (ix = 0; ix < N;ix++)
     ary[ix] = -1;
   
-#pragma acc parallel num_workers(32) vector_length(32) copy(ary) copy(ondev)
+#pragma acc parallel num_workers(32) vector_length(32) copy(ary) copy(ondev) \
+	    copyout(workersize, vectorsize)
   {
 #pragma acc loop worker vector
     for (unsigned ix = 0; ix < N; ix++)
@@ -31,6 +33,8 @@  int main ()
 	else
 	  ary[ix] = ix;
       }
+    workersize = __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
+    vectorsize = __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
   }
 
   for (ix = 0; ix < N; ix++)
@@ -39,8 +43,8 @@  int main ()
       if(ondev)
 	{
 	  int g = 0;
-	  int w = (ix / 32) % 32;
-	  int v = ix % 32;
+	  int w = (ix / vectorsize) % workersize;
+	  int v = ix % vectorsize;
 
 	  expected = (g << 16) | (w << 8) | v;
 	}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/parallel-dims.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/parallel-dims.c
index c4430cecf71..ec63e3fe2c9 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/parallel-dims.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/parallel-dims.c
@@ -12,7 +12,8 @@  static unsigned int __attribute__ ((optimize ("O2"))) acc_gang ()
 {
   if (acc_on_device ((int) acc_device_host))
     return 0;
-  else if (acc_on_device ((int) acc_device_nvidia))
+  else if (acc_on_device ((int) acc_device_nvidia)
+	   || acc_on_device ((int) acc_device_gcn))
     return __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
   else
     __builtin_abort ();
@@ -23,7 +24,8 @@  static unsigned int __attribute__ ((optimize ("O2"))) acc_worker ()
 {
   if (acc_on_device ((int) acc_device_host))
     return 0;
-  else if (acc_on_device ((int) acc_device_nvidia))
+  else if (acc_on_device ((int) acc_device_nvidia)
+	   || acc_on_device ((int) acc_device_gcn))
     return __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
   else
     __builtin_abort ();
@@ -34,7 +36,8 @@  static unsigned int __attribute__ ((optimize ("O2"))) acc_vector ()
 {
   if (acc_on_device ((int) acc_device_host))
     return 0;
-  else if (acc_on_device ((int) acc_device_nvidia))
+  else if (acc_on_device ((int) acc_device_nvidia)
+	   || acc_on_device ((int) acc_device_gcn))
     return __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
   else
     __builtin_abort ();
@@ -177,9 +180,8 @@  int main ()
 	if (vectors_actual != 32)
 	  __builtin_abort ();
       }
-    else
-      if (vectors_actual != 1)
-	__builtin_abort ();
+    else if (vectors_actual != 1)
+      __builtin_abort ();
     if (gangs_min != 0 || gangs_max != 0
 	|| workers_min != 0 || workers_max != 0
 	|| vectors_min != 0 || vectors_max != vectors_actual - 1)
@@ -325,6 +327,10 @@  int main ()
 	  /* We're actually executing with num_workers (32).  */
 	  /* workers_actual = 32; */
 	}
+      else if (acc_on_device (acc_device_gcn))
+	{
+	  workers_actual = 4;
+	}
       else
 	__builtin_abort ();
 #pragma acc loop worker reduction (min: gangs_min, workers_min, vectors_min) reduction (max: gangs_max, workers_max, vectors_max)
@@ -404,6 +410,13 @@  int main ()
 	  /* The GCC nvptx back end enforces vector_length (32).  */
 	  vectors_actual = 32;
 	}
+      else if (acc_on_device (acc_device_gcn))
+	{
+	  /* Because of the way vectors are implemented for GCN, a vector loop
+	     containing a seq routine call will not vectorize calls to that
+	     routine.  Hence, we'll only get one "vector".  */
+	  vectors_actual = 1;
+	}
       else
 	__builtin_abort ();
 #pragma acc loop vector reduction (min: gangs_min, workers_min, vectors_min) reduction (max: gangs_max, workers_max, vectors_max)
@@ -430,6 +443,9 @@  int main ()
        in the following case.  So, limit ourselves here.  */
     if (acc_get_device_type () == acc_device_nvidia)
       gangs = 3;
+    /* Similar appears to be true for GCN.  */
+    if (acc_get_device_type () == acc_device_gcn)
+      gangs = 3;
     int gangs_actual = gangs;
 #define WORKERS 3
     int workers_actual = WORKERS;
@@ -456,6 +472,11 @@  int main ()
 	  /* The GCC nvptx back end enforces vector_length (32).  */
 	  vectors_actual = 32;
 	}
+      else if (acc_on_device (acc_device_gcn))
+	{
+	  /* See above comments about GCN vectors_actual.  */
+	  vectors_actual = 1;
+	}
       else
 	__builtin_abort ();
 #pragma acc loop gang reduction (min: gangs_min, workers_min, vectors_min) reduction (max: gangs_max, workers_max, vectors_max)
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/private-variables-2.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/private-variables-2.c
new file mode 100644
index 00000000000..51fb394a0b1
--- /dev/null
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/private-variables-2.c
@@ -0,0 +1,217 @@ 
+#include <assert.h>
+
+/* Worker propagation: plain scalar variables.  */
+
+void
+worker_bcast_1 (void)
+{
+  int i, arr[32 * 32];
+
+  for (i = 0; i < 32 * 32; i++)
+    arr[i] = i;
+
+  #pragma acc parallel copy(arr) num_gangs(32) num_workers(32)
+  {
+    #pragma acc loop gang
+    for (i = 0; i < 32; i++)
+      {
+	int j;
+	int x = (i ^ 3) * 3;
+
+	#pragma acc loop worker
+	for (j = 0; j < 32; j++)
+	  arr[i * 32 + j] += x * j;
+
+	x = (i | 5) * 5;
+
+	#pragma acc loop worker
+	for (j = 0; j < 32; j++)
+	  arr[i * 32 + j] += x * j;
+      }
+  }
+
+  for (i = 0; i < 32; i++)
+    for (int j = 0; j < 32; j++)
+      {
+	int idx = i * 32 + j;
+	assert (arr[idx] == idx + (i ^ 3) * 3 * j + (i | 5) * 5 * j);
+      }
+}
+
+#pragma acc routine seq
+__attribute__((noinline)) static int
+select_var (int s, int x, int y)
+{
+  if (s)
+    return x;
+  else
+    return y;
+}
+
+/* Worker propagation: scalars through function calls.  */
+
+void
+worker_bcast_2 (void)
+{
+  int i, arr[32 * 32];
+
+  for (i = 0; i < 32 * 32; i++)
+    arr[i] = i;
+
+  #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
+  {
+    int j;
+
+    #pragma acc loop gang
+    for (i = 0; i < 32; i++)
+      {
+	int x, y, z;
+
+	x = i * 5;
+	y = i * 7;
+
+	#pragma acc loop worker
+	for (j = 0; j < 32; j++)
+	  arr[i * 32 + j] += select_var (j & 1, x, y) * j;
+
+	#pragma acc loop worker vector
+	for (j = 0; j < 32; j++)
+	  arr[i * 32 + j] += select_var (j & 1, y, x) * j;
+      }
+  }
+
+  for (i = 0; i < 32; i++)
+    for (int j = 0; j < 32; j++)
+      {
+	int idx = i * 32 + j;
+	int answer = idx + ((j & 1) ? i * 5 : i * 7) * j
+		     + ((j & 1) ? i * 7 : i * 5) * j;
+        assert (arr[idx] == answer);
+      }
+}
+
+#pragma acc routine seq
+__attribute__((noinline)) static int
+select_addr (int s, int *x, int *y)
+{
+  if (s)
+    return *x;
+  else
+    return *y;
+}
+
+/* Worker propagation: addresses of locals through function calls.  */
+
+void
+worker_bcast_3 (void)
+{
+  int i, arr[32 * 32];
+
+  for (i = 0; i < 32 * 32; i++)
+    arr[i] = i;
+
+  #pragma acc parallel copy(arr) num_gangs(32) num_workers(32) vector_length(32)
+  {
+    int j;
+
+    #pragma acc loop gang
+    for (i = 0; i < 32; i++)
+      {
+        int x, y, z;
+
+	x = i * 5;
+	y = i * 7;
+
+	#pragma acc loop worker
+	for (j = 0; j < 32; j++)
+	  arr[i * 32 + j] += select_addr (j & 1, &x, &y) * j;
+
+	#pragma acc loop worker vector
+	for (j = 0; j < 32; j++)
+	  arr[i * 32 + j] += select_addr (j & 1, &y, &x) * j;
+      }
+  }
+
+  for (i = 0; i < 32; i++)
+    for (int j = 0; j < 32; j++)
+      {
+	int idx = i * 32 + j;
+	int answer = idx + ((j & 1) ? i * 5 : i * 7) * j
+		     + ((j & 1) ? i * 7 : i * 5) * j;
+	assert (arr[idx] == answer);
+      }
+}
+
+#pragma acc routine seq
+__attribute__((noinline)) static int *
+select_ptr (int s, int *x, int *y)
+{
+  if (s)
+    return x;
+  else
+    return y;
+}
+
+/* Worker propagation: writes through pointers.  */
+
+void
+worker_bcast_4 (void)
+{
+  int i, arr[32 * 32];
+
+  for (i = 0; i < 32 * 32; i++)
+    arr[i] = i;
+
+  #pragma acc parallel copy(arr) num_gangs(32) num_workers(32)
+  {
+    int j;
+
+    #pragma acc loop gang
+    for (i = 0; i < 32; i++)
+      {
+	int x, y, z;
+	int *p, *q, *r;
+
+	p = &x;
+	q = &y;
+
+	x = i * 5;
+	y = i * 7;
+	r = select_ptr (i & 1, p, q);
+
+	#pragma acc loop worker
+	for (j = 0; j < 32; j++)
+	  arr[i * 32 + j] += (x + y + 2 * (*p) + (*q)) * j;
+
+	/* This write can affect either x or y: both should be broadcast into
+	   the next loop.  */
+	(*r) += 20;
+
+	#pragma acc loop worker
+	for (j = 0; j < 32; j++)
+	  arr[i * 32 + j] += (x + y + 2 * (*p) + (*q)) * j;
+      }
+  }
+
+  for (i = 0; i < 32; i++)
+    for (int j = 0; j < 32; j++)
+      {
+	int idx = i * 32 + j;
+	int x = i * 5, y = i * 7;
+	int answer = idx + (3 * x + 2 * y) * j
+		     + ((i & 1) ? (3 * (x + 20) + 2 * y)
+				: (3 * x + 2 * (y + 20))) * j;
+	assert (arr[idx] == answer);
+      }
+}
+
+
+int main ()
+{
+  worker_bcast_1 ();
+  worker_bcast_2 ();
+  worker_bcast_3 ();
+  worker_bcast_4 ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-gwv-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-gwv-1.c
index a97e046b687..da13d84908a 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-gwv-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-gwv-1.c
@@ -30,14 +30,18 @@  int main ()
   int ix;
   int exit = 0;
   int ondev = 0;
+  int gangsize, workersize, vectorsize;
 
   for (ix = 0; ix < N;ix++)
     ary[ix] = -1;
   
-#pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) copy(ary) copy(ondev)
+#pragma acc parallel num_gangs(32) num_workers(32) vector_length(32) copy(ary) copy(ondev) copyout(gangsize, workersize, vectorsize)
   {
     ondev = acc_on_device (acc_device_not_host);
     gang (ary);
+    gangsize = __builtin_goacc_parlevel_size (GOMP_DIM_GANG);
+    workersize = __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
+    vectorsize = __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
   }
 
   for (ix = 0; ix < N; ix++)
@@ -45,11 +49,12 @@  int main ()
       int expected = ix;
       if(ondev)
 	{
-	  int chunk_size = (N + 32*32*32 - 1) / (32*32*32);
+	  int chunk_size = (N + gangsize * workersize * vectorsize - 1)
+			   / (gangsize * workersize * vectorsize);
 	  
-	  int g = ix / (chunk_size * 32 * 32);
-	  int w = ix / 32 % 32;
-	  int v = ix % 32;
+	  int g = ix / (chunk_size * vectorsize * workersize);
+	  int w = (ix / vectorsize) % workersize;
+	  int v = ix % vectorsize;
 
 	  expected = (g << 16) | (w << 8) | v;
 	}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-v-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-v-1.c
index b1e3e3a596a..dd7bb6cdcd1 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-v-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-v-1.c
@@ -30,14 +30,17 @@  int main ()
   int ix;
   int exit = 0;
   int ondev = 0;
+  int vectorsize;
 
   for (ix = 0; ix < N;ix++)
     ary[ix] = -1;
   
-#pragma acc parallel vector_length(32) copy(ary) copy(ondev)
+#pragma acc parallel vector_length(32) copy(ary) copy(ondev) \
+	    copyout(vectorsize)
   {
     ondev = acc_on_device (acc_device_not_host);
     vector (ary);
+    vectorsize = __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
   }
 
   for (ix = 0; ix < N; ix++)
@@ -47,7 +50,7 @@  int main ()
 	{
 	  int g = 0;
 	  int w = 0;
-	  int v = ix % 32;
+	  int v = ix % vectorsize;
 
 	  expected = (g << 16) | (w << 8) | v;
 	}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-w-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-w-1.c
index e14947cdf94..bf9228105c5 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-w-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-w-1.c
@@ -31,14 +31,17 @@  int main ()
   int ix;
   int exit = 0;
   int ondev = 0;
+  int workersize;
 
   for (ix = 0; ix < N;ix++)
     ary[ix] = -1;
   
-#pragma acc parallel num_workers(32) vector_length(32) copy(ary) copy(ondev)
+#pragma acc parallel num_workers(32) vector_length(32) copy(ary) copy(ondev) \
+	    copyout(workersize)
   {
     ondev = acc_on_device (acc_device_not_host);
     worker (ary);
+    workersize = __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
   }
 
   for (ix = 0; ix < N; ix++)
@@ -47,7 +50,7 @@  int main ()
       if(ondev)
 	{
 	  int g = 0;
-	  int w = ix % 32;
+	  int w = ix % workersize;
 	  int v = 0;
 
 	  expected = (g << 16) | (w << 8) | v;
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-1.c
index 23dbc1ae401..73696e4e59a 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-1.c
@@ -30,14 +30,18 @@  int main ()
   int ix;
   int exit = 0;
   int ondev = 0;
+  int workersize, vectorsize;
 
   for (ix = 0; ix < N;ix++)
     ary[ix] = -1;
   
-#pragma acc parallel num_workers(32) vector_length(32) copy(ary) copy(ondev)
+#pragma acc parallel num_workers(32) vector_length(32) copy(ary) copy(ondev) \
+	    copyout(workersize, vectorsize)
   {
     ondev = acc_on_device (acc_device_not_host);
     worker (ary);
+    workersize = __builtin_goacc_parlevel_size (GOMP_DIM_WORKER);
+    vectorsize = __builtin_goacc_parlevel_size (GOMP_DIM_VECTOR);
   }
 
   for (ix = 0; ix < N; ix++)
@@ -46,8 +50,8 @@  int main ()
       if(ondev)
 	{
 	  int g = 0;
-	  int w = (ix / 32) % 32;
-	  int v = ix % 32;
+	  int w = (ix / vectorsize) % workersize;
+	  int v = ix % vectorsize;
 
 	  expected = (g << 16) | (w << 8) | v;
 	}
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-2.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-2.c
index 886214843f1..b469eedd1bc 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-2.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/routine-wv-2.c
@@ -2,8 +2,14 @@ 
 #include <openacc.h>
 #include <gomp-constants.h>
 
+#ifdef ACC_DEVICE_TYPE_gcn
+/* FIXME: Max. number of workers may increase for GCN in the future.  */
+#define NUM_WORKERS 4
+#define NUM_VECTORS 1
+#else
 #define NUM_WORKERS 16
 #define NUM_VECTORS 32
+#endif
 #define WIDTH 64
 #define HEIGHT 32
 
@@ -37,7 +43,8 @@  int DoWorkVec (int nw)
       ary[ix][jx] = 0xdeadbeef;
 
   printf ("spawning %d ...", nw); fflush (stdout);
-  
+
+/* { dg-warning "region contains vector partitioned code but is not vector partitioned" "vector" { target openacc_amdgcn_accel_selected } 48 } */
 #pragma acc parallel num_workers(nw) vector_length (NUM_VECTORS) copy (ary)
   {
     WorkVec ((int *)ary, WIDTH, HEIGHT, nw, NUM_VECTORS);
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/serial-dims.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/serial-dims.c
index dc6e1eac938..d4692091b84 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/serial-dims.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/serial-dims.c
@@ -11,7 +11,8 @@  static unsigned int __attribute__ ((optimize ("O2"))) acc_gang ()
 {
   if (acc_on_device ((int) acc_device_host))
     return 0;
-  else if (acc_on_device ((int) acc_device_nvidia))
+  else if (acc_on_device ((int) acc_device_nvidia)
+	   || acc_on_device ((int) acc_device_gcn))
     return __builtin_goacc_parlevel_id (GOMP_DIM_GANG);
   else
     __builtin_abort ();
@@ -22,7 +23,8 @@  static unsigned int __attribute__ ((optimize ("O2"))) acc_worker ()
 {
   if (acc_on_device ((int) acc_device_host))
     return 0;
-  else if (acc_on_device ((int) acc_device_nvidia))
+  else if (acc_on_device ((int) acc_device_nvidia)
+	   || acc_on_device ((int) acc_device_gcn))
     return __builtin_goacc_parlevel_id (GOMP_DIM_WORKER);
   else
     __builtin_abort ();
@@ -33,7 +35,8 @@  static unsigned int __attribute__ ((optimize ("O2"))) acc_vector ()
 {
   if (acc_on_device ((int) acc_device_host))
     return 0;
-  else if (acc_on_device ((int) acc_device_nvidia))
+  else if (acc_on_device ((int) acc_device_nvidia)
+	   || acc_on_device ((int) acc_device_gcn))
     return __builtin_goacc_parlevel_id (GOMP_DIM_VECTOR);
   else
     __builtin_abort ();
diff --git a/libgomp/testsuite/libgomp.oacc-c-c++-common/tile-1.c b/libgomp/testsuite/libgomp.oacc-c-c++-common/tile-1.c
index 5130591dd81..076e3cd75fe 100644
--- a/libgomp/testsuite/libgomp.oacc-c-c++-common/tile-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/tile-1.c
@@ -1,3 +1,6 @@ 
+/* AMD GCN does not use 32-lane vectors.
+   { dg-skip-if "unsuitable dimensions" { openacc_amdgcn_accel_selected } { "*" } { "" } } */
+
 /* { dg-additional-options "-fopenacc-dim=32" } */
 
 #include <stdio.h>
diff --git a/libgomp/testsuite/libgomp.oacc-c/c.exp b/libgomp/testsuite/libgomp.oacc-c/c.exp
index aba6665b5cb..f7005ebba48 100644
--- a/libgomp/testsuite/libgomp.oacc-c/c.exp
+++ b/libgomp/testsuite/libgomp.oacc-c/c.exp
@@ -72,6 +72,10 @@  foreach offload_target [concat [split $offload_targets ":"] "disable"] {
 
 	    set acc_mem_shared 0
 	}
+	amdgcn* {
+	    set acc_mem_shared 0
+	    set tagopt "-DACC_DEVICE_TYPE_gcn=\"$offload_target_openacc\""
+	}
 	default {
 	    error "Unknown OpenACC device type: $openacc_device_type (offload target: $offload_target)"
 	}
diff --git a/libgomp/testsuite/libgomp.oacc-c/offload-targets-1.c b/libgomp/testsuite/libgomp.oacc-c/offload-targets-1.c
index b62a587ec08..7265f48deb9 100644
--- a/libgomp/testsuite/libgomp.oacc-c/offload-targets-1.c
+++ b/libgomp/testsuite/libgomp.oacc-c/offload-targets-1.c
@@ -26,6 +26,9 @@  int main ()
 #if defined ACC_DEVICE_TYPE_nvidia
   offload_target_requested = ACC_DEVICE_TYPE_nvidia;
   acc_device_type_requested = acc_device_nvidia;
+#elif defined ACC_DEVICE_TYPE_gcn
+  offload_target_requested = ACC_DEVICE_TYPE_gcn;
+  acc_device_type_requested = acc_device_gcn;
 #elif defined ACC_DEVICE_TYPE_host
   offload_target_requested = ACC_DEVICE_TYPE_host;
   acc_device_type_requested = acc_device_host;
diff --git a/libgomp/testsuite/libgomp.oacc-c/print-1.c b/libgomp/testsuite/libgomp.oacc-c/print-1.c
new file mode 100644
index 00000000000..593885b5c2c
--- /dev/null
+++ b/libgomp/testsuite/libgomp.oacc-c/print-1.c
@@ -0,0 +1,17 @@ 
+/* Ensure that printf on the offload device works.  */
+
+/* { dg-do run } */
+/* { dg-output "The answer is 42(\n|\r\n|\r)+" } */
+
+#include <stdio.h>
+
+int var = 42;
+
+int
+main ()
+{
+#pragma acc parallel
+    {
+      printf ("The answer is %d\n", var);
+    }
+}
diff --git a/libgomp/testsuite/libgomp.oacc-fortran/atomic_capture-1.f90 b/libgomp/testsuite/libgomp.oacc-fortran/atomic_capture-1.f90
index 5a4a1e03f64..9dd2339509c 100644
--- a/libgomp/testsuite/libgomp.oacc-fortran/atomic_capture-1.f90
+++ b/libgomp/testsuite/libgomp.oacc-fortran/atomic_capture-1.f90
@@ -275,7 +275,7 @@  program main
   if (ltmp .neqv. .not. lexp) STOP 33
   if (lgot .neqv. lexp) STOP 34
 
-  igot = 1
+  igot = 0
   iexp = N
 
   !$acc parallel loop copy (igot, itmp)
@@ -287,12 +287,16 @@  program main
     end do
   !$acc end parallel loop
 
+  itmp = 0
+  do i = 1, N
+     if (iarr(i) == 0 .and. itmp == 0) itmp = i
+  end do
   do i = 1, N
-     if (.not. (1 <= iarr(i) .and. iarr(i) < iexp)) STOP 35
+     if (iarr(i) == 0 .and. i /= itmp) STOP 35
   end do
   if (igot /= iexp) STOP 36
 
-  igot = N
+  igot = N + 1
   iexp = 1
 
   !$acc parallel loop copy (igot, itmp)
@@ -304,8 +308,12 @@  program main
     end do
   !$acc end parallel loop
 
+  itmp = 0
   do i = 1, N
-     if (.not. (iarr(i) == 1 .or. iarr(i) == N)) STOP 37
+     if (iarr(i) == N + 1 .and. itmp == 0) itmp = i
+  end do
+  do i = 1, N
+      if (iarr(i) == N + 1 .and. i /= itmp) STOP 37
   end do
   if (igot /= iexp) STOP 38
 
@@ -314,7 +322,7 @@  program main
 
   !$acc parallel loop copy (igot, itmp)
     do i = 0, N - 1
-      iexpr = ibclr (-2, i)
+      iexpr = ibclr (-1, i)
   !$acc atomic capture
       iarr(i) = igot
       igot = iand (igot, iexpr)
@@ -322,9 +330,13 @@  program main
     end do
   !$acc end parallel loop
 
-  do i = 1, N
-     if (.not. (iarr(i - 1) < 0)) STOP 39
+  itmp = 0
+  do i = 0, N - 1
+     do j = 0, N - 1
+        if (btest (iarr(i), j)) itmp = itmp + 1
+     end do
   end do
+  if (itmp /= 528) STOP 39
   if (igot /= iexp) STOP 40
 
   igot = 0
@@ -340,10 +352,14 @@  program main
     end do
   !$acc end parallel loop
 
-  do i = 1, N
-     if (.not. (iarr(i - 1) >= 0)) STOP 41
+  itmp = 0
+  do i = 0, N - 1
+     do j = 0, N - 1
+        if (btest (iarr(i), j)) itmp = itmp + 1
+     end do
   end do
-  if (igot /= iexp) STOP 42
+  if (igot /= iexp) STOP 41
+  if (itmp /= 496) STOP 42
 
   igot = -1
   iexp = 0 
@@ -358,12 +374,16 @@  program main
     end do
   !$acc end parallel loop
 
-  do i = 1, N
-     if (.not. (iarr(i - 1) < 0)) STOP 43
+  itmp = 0
+  do i = 0, N - 1
+     do j = 0, N - 1
+        if (btest (iarr(i), j)) itmp = itmp + 1
+     end do
   end do
-  if (igot /= iexp) STOP 44
+  if (igot /= iexp) STOP 43
+  if (itmp /= 528) STOP 44
 
-  igot = 1
+  igot = 0
   iexp = N
 
   !$acc parallel loop copy (igot, itmp)
@@ -375,12 +395,16 @@  program main
     end do
   !$acc end parallel loop
 
+  itmp = 0
+  do i = 1, N
+     if (iarr(i) == 0 .and. itmp == 0) itmp = i
+  end do
   do i = 1, N
-     if (.not. (1 <= iarr(i) .and. iarr(i) < iexp)) STOP 45
+     if (iarr(i) == 0 .and. itmp /= i) STOP 45
   end do
   if (igot /= iexp) STOP 46
 
-  igot = N
+  igot = N + 1
   iexp = 1
 
   !$acc parallel loop copy (igot, itmp)
@@ -392,8 +416,12 @@  program main
     end do
   !$acc end parallel loop
 
+  itmp = 0
   do i = 1, N
-     if (.not. (iarr(i) == 1 .or. iarr(i) == N)) STOP 47
+     if (iarr(i) == N + 1 .and. itmp == 0) itmp = i
+  end do
+  do i = 1, N
+      if (iarr(i) == N + 1 .and. i /= itmp) STOP 47
   end do
   if (igot /= iexp) STOP 48
 
@@ -402,7 +430,7 @@  program main
 
   !$acc parallel loop copy (igot, itmp)
     do i = 0, N - 1
-      iexpr = ibclr (-2, i)
+      iexpr = ibclr (-1, i)
   !$acc atomic capture
       iarr(i) = igot
       igot = iand (iexpr, igot)
@@ -410,14 +438,18 @@  program main
     end do
   !$acc end parallel loop
 
-  do i = 1, N
-     if (.not. (iarr(i - 1) < 0)) STOP 49
+  itmp = 0
+  do i = 0, N - 1
+     do j = 0, N - 1
+        if (btest (iarr(i), j)) itmp = itmp + 1
+     end do
   end do
+  if (itmp /= 528) STOP 49
   if (igot /= iexp) STOP 50
 
   igot = 0
   iexp = -1 
-	!!
+
   !$acc parallel loop copy (igot, itmp)
     do i = 0, N - 1
       iexpr = lshift (1, i)
@@ -428,10 +460,14 @@  program main
     end do
   !$acc end parallel loop
 
-  do i = 1, N
-     if (.not. (iarr(i - 1) >= 0)) STOP 51
+  itmp = 0
+  do i = 0, N - 1
+     do j = 0, N - 1
+        if (btest (iarr(i), j)) itmp = itmp + 1
+     end do
   end do
-  if (igot /= iexp) STOP 52
+  if (igot /= iexp) STOP 51
+  if (itmp /= 496) STOP 52
 
   igot = -1
   iexp = 0 
@@ -446,10 +482,14 @@  program main
     end do
   !$acc end parallel loop
 
-  do i = 1, N
-     if (.not. (iarr(i - 1) < 0)) STOP 53
+  itmp = 0
+  do i = 0, N - 1
+     do j = 0, N - 1
+        if (btest (iarr(i), j)) itmp = itmp + 1
+     end do
   end do
-  if (igot /= iexp) STOP 54
+  if (igot /= iexp) STOP 53
+  if (itmp /= 528) STOP 54
 
   fgot = 1234.0
   fexp = 1266.0
@@ -720,7 +760,7 @@  program main
   end do
   if (igot /= iexp) STOP 88
 
-  igot = N
+  igot = N + 1
   iexp = 1
 
   !$acc parallel loop copy (igot, itmp)
@@ -733,7 +773,7 @@  program main
   !$acc end parallel loop
 
   do i = 1, N
-     if (.not. (iarr(i) == iexp)) STOP 89
+     if (iarr(i) .lt. 1 .or. iarr(i) .gt. N) STOP 89
   end do
   if (igot /= iexp) STOP 90
 
@@ -742,7 +782,7 @@  program main
 
   !$acc parallel loop copy (igot, itmp)
     do i = 0, N - 1
-      iexpr = ibclr (-2, i)
+      iexpr = ibclr (-1, i)
   !$acc atomic capture
       igot = iand (igot, iexpr)
       iarr(i) = igot
@@ -750,9 +790,13 @@  program main
     end do
   !$acc end parallel loop
 
-  do i = 1, N
-     if (.not. (iarr(i - 1) <= 0)) STOP 91
+  itmp = 0
+  do i = 0, N - 1
+     do j = 0, N - 1
+        if (btest (iarr(i), j)) itmp = itmp + 1
+     end do
   end do
+  if (itmp /= 496) STOP 91
   if (igot /= iexp) STOP 92
 
   igot = 0
@@ -768,9 +812,13 @@  program main
     end do
   !$acc end parallel loop
 
-  do i = 1, N
-     if (.not. (iarr(i - 1) >= -1)) STOP 93
+  itmp = 0
+  do i = 0, N - 1
+     do j = 0, N - 1
+        if (btest (iarr(i), j)) itmp = itmp + 1
+     end do
   end do
+  if (itmp /= 528) STOP 93
   if (igot /= iexp) STOP 94
 
   igot = -1
@@ -786,9 +834,13 @@  program main
     end do
   !$acc end parallel loop
 
-  do i = 1, N
-     if (.not. (iarr(i - 1) <= 0)) STOP 95
+  itmp = 0
+  do i = 0, N - 1
+     do j = 0, N - 1
+        if (btest (iarr(i), j)) itmp = itmp + 1
+     end do
   end do
+  if (itmp /= 496) STOP 95
   if (igot /= iexp) STOP 96
 
   igot = 1
@@ -808,7 +860,7 @@  program main
   end do
   if (igot /= iexp) STOP 98
 
-  igot = N
+  igot = N + 1
   iexp = 1
 
   !$acc parallel loop copy (igot, itmp)
@@ -821,7 +873,7 @@  program main
   !$acc end parallel loop
 
   do i = 1, N
-     if (.not. (iarr(i) == iexp )) STOP 99
+     if (iarr(i) .lt. 1 .or. iarr(i) .gt. N) STOP 99
   end do
   if (igot /= iexp) STOP 100
 
@@ -830,7 +882,7 @@  program main
 
   !$acc parallel loop copy (igot, itmp)
     do i = 0, N - 1
-      iexpr = ibclr (-2, i)
+      iexpr = ibclr (-1, i)
   !$acc atomic capture
       igot = iand (iexpr, igot)
       iarr(i) = igot
@@ -838,9 +890,13 @@  program main
     end do
   !$acc end parallel loop
 
-  do i = 1, N
-     if (.not. (iarr(i - 1) <= 0)) STOP 101
+  itmp = 0
+  do i = 0, N - 1
+     do j = 0, N - 1
+        if (btest (iarr(i), j)) itmp = itmp + 1
+     end do
   end do
+  if (itmp /= 496) STOP 101
   if (igot /= iexp) STOP 102
 
   igot = 0
@@ -856,9 +912,13 @@  program main
     end do
   !$acc end parallel loop
 
-  do i = 1, N
-     if (.not. (iarr(i - 1) >= iexp)) STOP 103
+  itmp = 0
+  do i = 0, N - 1
+     do j = 0, N - 1
+        if (btest (iarr(i), j)) itmp = itmp + 1
+     end do
   end do
+  if (itmp /= 528) STOP 103
   if (igot /= iexp) STOP 104
 
   igot = -1
@@ -874,9 +934,12 @@  program main
     end do
   !$acc end parallel loop
 
-  do i = 1, N
-     if (.not. (iarr(i - 1) <= iexp)) STOP 105
+  itmp = 0
+  do i = 0, N - 1
+     do j = 0, N - 1
+        if (btest (iarr(i), j)) itmp = itmp + 1
+     end do
   end do
+  if (itmp /= 496) STOP 105
   if (igot /= iexp) STOP 106
-
 end program
diff --git a/libgomp/testsuite/libgomp.oacc-fortran/collapse-1.f90 b/libgomp/testsuite/libgomp.oacc-fortran/collapse-1.f90
index 918c5d0d5b1..659e5db5d0b 100644
--- a/libgomp/testsuite/libgomp.oacc-fortran/collapse-1.f90
+++ b/libgomp/testsuite/libgomp.oacc-fortran/collapse-1.f90
@@ -6,7 +6,7 @@  program collapse1
   l = .false.
   a(:, :, :) = 0
   !$acc parallel
-  !$acc loop collapse(4 - 1)
+  !$acc loop collapse(4 - 1) gang(static:*)
     do i = 1, 3
       do j = 4, 6
         do k = 5, 7
@@ -14,7 +14,7 @@  program collapse1
         end do
       end do
     end do
-  !$acc loop collapse(2) reduction(.or.:l)
+  !$acc loop collapse(3) gang(static:*) reduction(.or.:l)
     do i = 1, 3
       do j = 4, 6
         do k = 5, 7
diff --git a/libgomp/testsuite/libgomp.oacc-fortran/collapse-2.f90 b/libgomp/testsuite/libgomp.oacc-fortran/collapse-2.f90
index 98b6987750e..409d9794e00 100644
--- a/libgomp/testsuite/libgomp.oacc-fortran/collapse-2.f90
+++ b/libgomp/testsuite/libgomp.oacc-fortran/collapse-2.f90
@@ -7,13 +7,13 @@  program collapse2
   l = .false.
   a(:, :, :) = 0
   !$acc parallel
-  !$acc loop collapse(4 - 1)
+  !$acc loop collapse(4 - 1) gang(static:*)
     do 164 i = 1, 3
       do 164 j = 4, 6
         do 164 k = 5, 7
           a(i, j, k) = i + j + k
 164      end do
-  !$acc loop collapse(2) reduction(.or.:l)
+  !$acc loop collapse(3) gang(static:*) reduction(.or.:l)
 firstdo: do i = 1, 3
       do j = 4, 6
         do k = 5, 7
diff --git a/libgomp/testsuite/libgomp.oacc-fortran/error_stop-1.f b/libgomp/testsuite/libgomp.oacc-fortran/error_stop-1.f
index 4965e674c27..95810a6ae93 100644
--- a/libgomp/testsuite/libgomp.oacc-fortran/error_stop-1.f
+++ b/libgomp/testsuite/libgomp.oacc-fortran/error_stop-1.f
@@ -15,6 +15,6 @@ 
 ! { dg-output "ERROR STOP (\n|\r\n|\r)+" }
 ! PR85463.  The "minimal" libgfortran implementation used with nvptx
 ! offloading is a little bit different.
-! { dg-output "Error termination.*" { target { ! openacc_nvidia_accel_selected } } }
+! { dg-output "Error termination.*" { target { { ! openacc_nvidia_accel_selected } && { ! openacc_amdgcn_accel_selected } } } }
 ! { dg-output "libgomp: cuStreamSynchronize error.*" { target openacc_nvidia_accel_selected } }
 ! { dg-shouldfail "" }
diff --git a/libgomp/testsuite/libgomp.oacc-fortran/error_stop-2.f b/libgomp/testsuite/libgomp.oacc-fortran/error_stop-2.f
index 7103fdb5d8e..ce59bbda3c3 100644
--- a/libgomp/testsuite/libgomp.oacc-fortran/error_stop-2.f
+++ b/libgomp/testsuite/libgomp.oacc-fortran/error_stop-2.f
@@ -15,6 +15,6 @@ 
 ! { dg-output "ERROR STOP 35(\n|\r\n|\r)+" }
 ! PR85463.  The "minimal" libgfortran implementation used with nvptx
 ! offloading is a little bit different.
-! { dg-output "Error termination.*" { target { ! openacc_nvidia_accel_selected } } }
+! { dg-output "Error termination.*" { target { { ! openacc_nvidia_accel_selected } && { ! openacc_amdgcn_accel_selected } } } }
 ! { dg-output "libgomp: cuStreamSynchronize error.*" { target openacc_nvidia_accel_selected } }
 ! { dg-shouldfail "" }
diff --git a/libgomp/testsuite/libgomp.oacc-fortran/error_stop-3.f b/libgomp/testsuite/libgomp.oacc-fortran/error_stop-3.f
index 9c217f14ea1..9b606c83ad9 100644
--- a/libgomp/testsuite/libgomp.oacc-fortran/error_stop-3.f
+++ b/libgomp/testsuite/libgomp.oacc-fortran/error_stop-3.f
@@ -15,6 +15,6 @@ 
 ! { dg-output "ERROR STOP SiGN(\n|\r\n|\r)+" }
 ! PR85463.  The "minimal" libgfortran implementation used with nvptx
 ! offloading is a little bit different.
-! { dg-output "Error termination.*" { target { ! openacc_nvidia_accel_selected } } }
+! { dg-output "Error termination.*" { target { { ! openacc_nvidia_accel_selected } && { ! openacc_amdgcn_accel_selected } } } }
 ! { dg-output "libgomp: cuStreamSynchronize error.*" { target openacc_nvidia_accel_selected } }
 ! { dg-shouldfail "" }
diff --git a/libgomp/testsuite/libgomp.oacc-fortran/fortran.exp b/libgomp/testsuite/libgomp.oacc-fortran/fortran.exp
index 56694a6eef5..0f8d1db4895 100644
--- a/libgomp/testsuite/libgomp.oacc-fortran/fortran.exp
+++ b/libgomp/testsuite/libgomp.oacc-fortran/fortran.exp
@@ -90,6 +90,10 @@  if { $lang_test_file_found } {
 
 		set acc_mem_shared 0
 	    }
+	    amdgcn* {
+		set acc_mem_shared 0
+		set tagopt "-DACC_DEVICE_TYPE_gcn=\"$offload_target_openacc\""
+	    }
 	    default {
 		error "Unknown OpenACC device type: $openacc_device_type (offload target: $offload_target)"
 	    }
diff --git a/libgomp/testsuite/libgomp.oacc-fortran/print-1.f90 b/libgomp/testsuite/libgomp.oacc-fortran/print-1.f90
new file mode 100644
index 00000000000..9a94195b6f1
--- /dev/null
+++ b/libgomp/testsuite/libgomp.oacc-fortran/print-1.f90
@@ -0,0 +1,15 @@ 
+! Ensure that printf on the offload device works.
+
+! { dg-do run }
+! { dg-output "The answer is 42(\n|\r\n|\r)+" }
+! { dg-xfail-if "no write for nvidia" { openacc_nvidia_accel_selected } }
+
+program main
+  implicit none
+  integer :: var = 42
+
+!$acc parallel
+  write (0, '("The answer is ", I2)') var
+!$acc end parallel
+
+end program main