This patch implement OpenMP 5.0 requirements of incrementing/decrementing
the reference count of a mapped structure at most once (across all elements)
on a construct.
This is implemented by pulling in libgomp/hashtab.h and using htab_t as a
pointer set. Structure element list siblings also have pointers-to-refcounts
linked together, to naturally achieve uniform increment/decrement without
repeating.
There are still some questions on whether using such a htab_t based set is
faster/slower than using a sorted pointer array based implementation. This
is to be researched on later.
libgomp/ChangeLog:
* hashtab.h (htab_clear): New function with initialization code
factored out from...
(htab_create): ...here, adjust to use htab_clear function.
* libgomp.h (REFCOUNT_SPECIAL): New symbol to denote range of
special refcount values, add comments.
(REFCOUNT_INFINITY): Adjust definition to use REFCOUNT_SPECIAL.
(REFCOUNT_LINK): Likewise.
(REFCOUNT_STRUCTELEM): New special refcount range for structure
element siblings.
(REFCOUNT_STRUCTELEM_P): Macro for testing for structure element
sibling maps.
(REFCOUNT_STRUCTELEM_FLAG_FIRST): Flag to indicate first sibling.
(REFCOUNT_STRUCTELEM_FLAG_LAST): Flag to indicate last sibling.
(REFCOUNT_STRUCTELEM_FIRST_P): Macro to test _FIRST flag.
(REFCOUNT_STRUCTELEM_LAST_P): Macro to test _LAST flag.
(struct splay_tree_key_s): Add structelem_refcount and
structelem_refcount_ptr fields into a union with dynamic_refcount.
Add comments.
(gomp_map_vars): Delete declaration.
(gomp_map_vars_async): Likewise.
(gomp_unmap_vars): Likewise.
(gomp_unmap_vars_async): Likewise.
(goacc_map_vars): New declaration.
(goacc_unmap_vars): Likewise.
* oacc-mem.c (acc_map_data): Adjust to use goacc_map_vars.
(goacc_enter_datum): Likewise.
(goacc_enter_data_internal): Likewise.
* oacc-parallel.c (GOACC_parallel_keyed): Adjust to use goacc_map_vars
and goacc_unmap_vars.
(GOACC_data_start): Adjust to use goacc_map_vars.
(GOACC_data_end): Adjust to use goacc_unmap_vars.
* target.c (hash_entry_type): New typedef.
(htab_alloc): New function hook for hashtab.h.
(htab_free): Likewise.
(htab_hash): Likewise.
(htab_eq): Likewise.
(hashtab.h): Add file include.
(gomp_increment_refcount): New function.
(gomp_decrement_refcount): Likewise.
(gomp_map_vars_existing): Add refcount_set parameter, adjust to use
gomp_increment_refcount.
(gomp_map_fields_existing): Add refcount_set parameter, adjust calls
to gomp_map_vars_existing.
(gomp_map_vars_internal): Add refcount_set parameter, add local openmp_p
variable to guard OpenMP specific paths, adjust calls to
gomp_map_vars_existing, add structure element sibling splay_tree_key
sequence creation code, adjust Fortran map case to avoid increment
under OpenMP.
(gomp_map_vars): Adjust to static, add refcount_set parameter, manage
local refcount_set if caller passed in NULL, adjust call to
gomp_map_vars_internal.
(gomp_map_vars_async): Adjust and rename into...
(goacc_map_vars): ...this new function, adjust call to
gomp_map_vars_internal.
(gomp_remove_splay_tree_key): New function with code factored out from
gomp_remove_var_internal.
(gomp_remove_var_internal): Add code to handle removing multiple
splay_tree_key sequence for structure elements, adjust code to use
gomp_remove_splay_tree_key for splay-tree key removal.
(gomp_unmap_vars_internal): Add refcount_set parameter, adjust to use
gomp_decrement_refcount.
(gomp_unmap_vars): Adjust to static, add refcount_set parameter, manage
local refcount_set if caller passed in NULL, adjust call to
gomp_unmap_vars_internal.
(gomp_unmap_vars_async): Adjust and rename into...
(goacc_unmap_vars): ...this new function, adjust call to
gomp_unmap_vars_internal.
(GOMP_target): Manage refcount_set and adjust calls to gomp_map_vars and
gomp_unmap_vars.
(GOMP_target_ext): Likewise.
(gomp_target_data_fallback): Adjust call to gomp_map_vars.
(GOMP_target_data): Likewise.
(GOMP_target_data_ext): Likewise.
(GOMP_target_end_data): Adjust call to gomp_unmap_vars.
(gomp_exit_data): Add refcount_set parameter, adjust to use
gomp_decrement_refcount, adjust to queue splay-tree keys for removal
after main loop.
(GOMP_target_enter_exit_data): Manage refcount_set and adjust calls to
gomp_map_vars and gomp_exit_data.
(gomp_target_task_fn): Likewise.
* testsuite/libgomp.c-c++-common/refcount-1.c: New testcase.
* testsuite/libgomp.c-c++-common/struct-elem-1.c: New testcase.
* testsuite/libgomp.c-c++-common/struct-elem-2.c: New testcase.
* testsuite/libgomp.c-c++-common/struct-elem-3.c: New testcase.
* testsuite/libgomp.c-c++-common/struct-elem-4.c: New testcase.
* testsuite/libgomp.c-c++-common/struct-elem-5.c: New testcase.
Move the OpenACC enter and exit data directives from using a single builtin to
having one each. For most purposes it was easy to tell which was which, from
the clauses given, but it's overhead we can easily avoid, and there may be
future uses where that isn't possible.
gcc/
* omp-builtins.def (BUILT_IN_GOACC_ENTER_EXIT_DATA): Split into...
(BUILT_IN_GOACC_ENTER_DATA, BUILT_IN_GOACC_EXIT_DATA): ... these.
* gimple.h (enum gf_mask): Split
'GF_OMP_TARGET_KIND_OACC_ENTER_EXIT_DATA' into
'GF_OMP_TARGET_KIND_OACC_ENTER_DATA' and
'GF_OMP_TARGET_KIND_OACC_EXIT_DATA'.
(is_gimple_omp_oacc): Update.
* gimple-pretty-print.c (dump_gimple_omp_target): Likewise.
* gimplify.c (gimplify_omp_target_update): Likewise.
* omp-expand.c (expand_omp_target, build_omp_regions_1)
(omp_make_gimple_edges): Likewise.
* omp-low.c (check_omp_nesting_restrictions, lower_omp_target):
Likewise.
gcc/testsuite/
* c-c++-common/goacc-gomp/nesting-fail-1.c: Adjust patterns.
* c-c++-common/goacc/finalize-1.c: Likewise.
* c-c++-common/goacc/mdc-1.c: Likewise.
* c-c++-common/goacc/nesting-fail-1.c: Likewise.
* c-c++-common/goacc/struct-enter-exit-data-1.c: Likewise.
* gfortran.dg/goacc/attach-descriptor.f90: Likewise.
* gfortran.dg/goacc/finalize-1.f: Likewise.
* gfortran.dg/goacc/mapping-tests-3.f90: Likewise.
libgomp/
* libgomp.map (GOACC_2.0.2): New symbol version.
* libgomp_g.h (GOACC_enter_data, GOACC_exit_data) New prototypes.
* oacc-mem.c (GOACC_enter_data, GOACC_exit_data) New functions.
Co-Authored-By: Thomas Schwinge <thomas@codesourcery.com>
libgomp/
* oacc-mem.c (goacc_enter_exit_data_internal): New function,
extracted from...
(GOACC_enter_exit_data): ... here.
(GOACC_declare): Use it.
Co-Authored-By: Andrew Stubbs <ams@codesourcery.com>
This deals with data management, after all.
Small fix-up for r230275 (commit 6e232ba424)
"[OpenACC] declare directive".
libgomp/
* oacc-parallel.c (GOACC_declare): Move...
* oacc-mem.c: ... here.
* libgomp_g.h: Adjust.
Given that we 'continue' for 'GOMP_MAP_POINTER', we cannot possibly encounter
it afterwards.
Small fix-up for r230275 (commit 6e232ba424)
"[OpenACC] declare directive".
libgomp/
* oacc-parallel.c (GOACC_declare): Clean up 'GOMP_MAP_POINTER'
handling.
Co-Authored-By: Thomas Schwinge <thomas@codesourcery.com>
The dsdotr and dsdoti variables uninitialized and the testcase fails e.g.
on i686-linux. Fixed by zero initialization.
2021-06-10 Jakub Jelinek <jakub@redhat.com>
PR tree-optimization/100981
* testsuite/libgomp.fortran/pr100981-2.f90 (cdcdot): Initialize
dsdotr and dsdoti to 0.
Don't add -march=i486 if atomic compare-and-swap is supported on 'int'.
This fixes libgomp tests with "-march=x86-64 -m32 -fcf-protection".
* testsuite/lib/libgomp.exp (libgomp_init): Don't add -march=i486
if atomic compare-and-swap is supported on 'int'.
The following fixes the SLP FMA patterns to preserve reduction
info and the reduction vectorization to consider internal function
call defs for the reduction stmt.
2021-06-09 Richard Biener <rguenther@suse.de>
PR tree-optimization/100981
gcc/
* tree-vect-loop.c (vect_create_epilog_for_reduction): Use
gimple_get_lhs to also handle calls.
* tree-vect-slp-patterns.c (complex_pattern::build): Transfer
reduction info.
gcc/testsuite/
* gfortran.dg/vect/pr100981-1.f90: New testcase.
libgomp/
* testsuite/libgomp.fortran/pr100981-2.f90: New testcase.
... which currently has *not* been forced to 'num_workers (1)'.
In addition to the testcases modified here, this also fixes:
FAIL: libgomp.oacc-c/../libgomp.oacc-c-c++-common/mode-transitions.c -DACC_DEVICE_TYPE_radeon=1 -DACC_MEM_SHARED=0 -foffload=amdgcn-amdhsa -O0 execution test
[Etc.]
mode-transitions.exe: [...]/libgomp.oacc-c-c++-common/mode-transitions.c:702: t17: Assertion `arr_b[i] == (i ^ 31) * 8' failed.
libgomp/
* plugin/plugin-gcn.c (gcn_exec): Force 'num_workers (1)'
unconditionally.
* testsuite/libgomp.oacc-c-c++-common/acc_prof-kernels-1.c:
Update.
* testsuite/libgomp.oacc-c-c++-common/parallel-dims.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/routine-wv-2.c: Likewise.
..., by simplifying 'libgomp.oacc-c-c++-common/parallel-dims.c', and updating
the former correspondingly. '__builtin_goacc_parlevel_id' does the right thing
for all 'acc_device_*'.
Follow-up to commit 09e0ad6253 "Update OpenACC
tests for amdgcn".
libgomp/
* testsuite/libgomp.oacc-c-c++-common/parallel-dims.c: Simplify.
* testsuite/libgomp.oacc-fortran/parallel-dims-aux.c: Update.
... on top of r279378 (commit 26b74ed022)
"Update OpenACC tests for amdgcn".
libgomp/
* testsuite/libgomp.oacc-c-c++-common/acc_prof-kernels-1.c: Fix
for 'acc_device_radeon'.
That is, re-enable it for host-fallback, and enable it for GCN offloading.
Fix-up for r279378 (commit 26b74ed022)
"Update OpenACC tests for amdgcn".
libgomp/
* testsuite/libgomp.oacc-c-c++-common/async_queue-1.c: Don't
require 'openacc_nvidia_accel_selected'. Fix up for
'ACC_DEVICE_TYPE_radeon'.
The GCN support that got added in r278935 (commit
83caa34e2a) "Enable OpenACC GCN testing" was
forked before my r269107 (commit ee332b4a9a)
"[libgomp] Clarify difference between offload target, offload plugin, and
OpenACC device type", and didn't later pick up these changes.
No functional change.
libgomp/
* testsuite/lib/libgomp.exp
(check_effective_target_openacc_radeon_accel_selected):
Streamline.
This problem has been fixed long ago, in r267934 (commit
d41d952c9b) "[nvptx] Handle assignment to
gang-level reduction variable".
libgomp/
* testsuite/libgomp.oacc-c-c++-common/parallel-dims.c: Revert
PR80547 workaround.
Small fix-up for r267889 (commit 2b9d9e3937)
"[nvptx] Enable large vectors":
> * testsuite/libgomp.oacc-c-c++-common/parallel-dims.c: Expect vector
> length 2097152 to be reduced to 1024 instead of 32.
libgomp/
* testsuite/libgomp.oacc-c-c++-common/parallel-dims.c
<acc_device_nvidia>: Update comment.
When gcc is configured for nvptx offloading with --without-cuda-driver
and full CUDA isn't installed, many libgomp.oacc-*/* tests fail,
some of them because cuda.h header can't be found, others because
the tests can't be linked against -lcuda, -lcudart or -lcublas.
I usually only have akmod-nvidia and xorg-x11-drv-nvidia-cuda rpms
installed, so libcuda.so.1 can be dlopened and the offloading works,
but linking against those libraries isn't possible nor are the
headers around (for the plugin itself there is the fallback
libgomp/plugin/cuda/cuda.h).
The following patch adds 3 new effective targets and uses them in tests that
needs those.
2021-05-27 Jakub Jelinek <jakub@redhat.com>
* testsuite/lib/libgomp.exp (check_effective_target_openacc_cuda,
check_effective_target_openacc_cublas,
check_effective_target_openacc_cudart): New.
* testsuite/libgomp.oacc-fortran/host_data-4.f90: Require effective
target openacc_cublas.
* testsuite/libgomp.oacc-fortran/host_data-2.f90: Likewise.
* testsuite/libgomp.oacc-fortran/host_data-3.f: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-91.c: Require effective
target openacc_cuda.
* testsuite/libgomp.oacc-c-c++-common/lib-70.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-90.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-75.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-69.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-74.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-81.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-72.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-85.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/pr87835.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-82.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-73.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-83.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-78.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-76.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-84.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-79.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/host_data-1.c: Require effective
targets openacc_cublas and openacc_cudart.
* testsuite/libgomp.oacc-c-c++-common/context-1.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/context-2.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/context-3.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/context-4.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/acc_get_property-nvptx.c:
Require effective target openacc_cudart.
* testsuite/libgomp.oacc-c-c++-common/asyncwait-1.c: Add -DUSE_CUDA_H
for effective target openacc_cuda and add && defined USE_CUDA_H to
preprocessor conditionals. Guard -lcuda also on openacc_cuda
effective target.
OpenMP Nesting of Regions restrictions say:
- If a target update, target data, target enter data, or target exit data
construct is encountered during execution of a target region, the behavior is unspecified.
- If a target construct is encountered during execution of a target region and a device
clause in which the ancestor device-modifier appears is not present on the construct, the
behavior is unspecified.
That wording is about the dynamic (runtime) behavior, not about lexical nesting,
so while it is UB if omp target * is encountered in the target region, we need to make
it compile and link (for lexical nesting of target * inside of target we actually
emit a warning).
To make this work, I had to do multiple changes.
One was to mark .omp_data_{sizes,kinds}.* variables when static as "omp declare target".
Another one was to add stub GOMP_target* entrypoints to nvptx and gcn libgomp.a.
The entrypoint functions shouldn't be called or passed in the offload regions,
otherwise
libgomp: cuLaunchKernel error: too many resources requested for launch
was reported; fixed by changing those arguments of calls to GOMP_target_ext
to NULL.
And we didn't mark the entrypoints "omp target entrypoint" when the caller
has been "omp declare target".
2021-05-26 Jakub Jelinek <jakub@redhat.com>
PR libgomp/100573
gcc/
* omp-low.c: Include omp-offload.h.
(create_omp_child_function): If current_function_decl has
"omp declare target" attribute and is_gimple_omp_offloaded,
remove that attribute from the copy of attribute list and
add "omp target entrypoint" attribute instead.
(lower_omp_target): Mark .omp_data_sizes.* and .omp_data_kinds.*
variables for offloading if in omp_maybe_offloaded_ctx.
* omp-offload.c (pass_omp_target_link::execute): Nullify second
argument to GOMP_target_data_ext in offloaded code.
libgomp/
* config/nvptx/target.c (GOMP_target_ext, GOMP_target_data_ext,
GOMP_target_end_data, GOMP_target_update_ext,
GOMP_target_enter_exit_data): New dummy entrypoints.
* config/gcn/target.c (GOMP_target_ext, GOMP_target_data_ext,
GOMP_target_end_data, GOMP_target_update_ext,
GOMP_target_enter_exit_data): Likewise.
* testsuite/libgomp.c-c++-common/for-3.c (DO_PRAGMA, OMPTEAMS,
OMPFROM, OMPTO): Define.
(main): Remove #pragma omp target teams around all the tests.
* testsuite/libgomp.c-c++-common/target-41.c: New test.
* testsuite/libgomp.c-c++-common/target-42.c: New test.
When a directive isn't combined with worksharing-loop, it takes much
simpler clause splitting path for reduction, and that one was missing
handling of teams when combined with simd.
2021-05-25 Jakub Jelinek <jakub@redhat.com>
PR middle-end/99928
gcc/c-family/
* c-omp.c (c_omp_split_clauses): Copy reduction to teams when teams is
combined with simd and not with taskloop or for.
gcc/testsuite/
* c-c++-common/gomp/pr99928-8.c: Remove xfails from omp teams r21 and
r28 checks.
* c-c++-common/gomp/pr99928-9.c: Likewise.
* c-c++-common/gomp/pr99928-10.c: Likewise.
libgomp/
* testsuite/libgomp.c-c++-common/reduction-17.c: New test.
This patch implements a method to track the "private-ness" of
OpenACC variables declared in offload regions in gang-partitioned,
worker-partitioned or vector-partitioned modes. Variables declared
implicitly in scoped blocks and those declared "private" on enclosing
directives (e.g. "acc parallel") are both handled. Variables that are
e.g. gang-private can then be adjusted so they reside in GPU shared
memory.
The reason for doing this is twofold: correct implementation of OpenACC
semantics, and optimisation, since shared memory might be faster than
the main memory on a GPU. Handling of private variables is intimately
tied to the execution model for gangs/workers/vectors implemented by
a particular target: for current targets, we use (or on mainline, will
soon use) a broadcasting/neutering scheme.
That is sufficient for code that e.g. sets a variable in worker-single
mode and expects to use the value in worker-partitioned mode. The
difficulty (semantics-wise) comes when the user wants to do something like
an atomic operation in worker-partitioned mode and expects a worker-single
(gang private) variable to be shared across each partitioned worker.
Forcing use of shared memory for such variables makes that work properly.
In terms of implementation, the parallelism level of a given loop is
not fixed until the oaccdevlow pass in the offload compiler, so the
patch delays fixing the parallelism level of variables declared on or
within such loops until the same point. This is done by adding a new
internal UNIQUE function (OACC_PRIVATE) that lists (the address of) each
private variable as an argument, and other arguments set so as to be able
to determine the correct parallelism level to use for the listed
variables. This new internal function fits into the existing scheme for
demarcating OpenACC loops, as described in comments in the patch.
Two new target hooks are introduced: TARGET_GOACC_ADJUST_PRIVATE_DECL and
TARGET_GOACC_EXPAND_VAR_DECL. The first can tweak a variable declaration
at oaccdevlow time, and the second at expand time. The first or both
of these target hooks can be used by a given offload target, depending
on its strategy for implementing private variables.
This patch updates the TARGET_GOACC_ADJUST_PRIVATE_DECL target hook in
the AMD GCN backend to the current name and prototype. (An earlier
version of the hook was already present, but dormant.)
gcc/
PR middle-end/90115
* doc/tm.texi.in (TARGET_GOACC_EXPAND_VAR_DECL)
(TARGET_GOACC_ADJUST_PRIVATE_DECL): Add documentation hooks.
* doc/tm.texi: Regenerate.
* expr.c (expand_expr_real_1): Expand decls using the
expand_var_decl OpenACC hook if defined.
* internal-fn.c (expand_UNIQUE): Handle IFN_UNIQUE_OACC_PRIVATE.
* internal-fn.h (IFN_UNIQUE_CODES): Add OACC_PRIVATE.
* omp-low.c (omp_context): Add oacc_privatization_candidates
field.
(lower_oacc_reductions): Add PRIVATE_MARKER parameter. Insert
before fork.
(lower_oacc_head_tail): Add PRIVATE_MARKER parameter. Modify
private marker's gimple call arguments, and pass it to
lower_oacc_reductions.
(oacc_privatization_scan_clause_chain)
(oacc_privatization_scan_decl_chain, lower_oacc_private_marker):
New functions.
(lower_omp_for, lower_omp_target, lower_omp_1): Use these.
* omp-offload.c (convert.h): Include.
(oacc_loop_xform_head_tail): Treat private-variable markers like
fork/join when transforming head/tail sequences.
(struct var_decl_rewrite_info): Add struct.
(oacc_rewrite_var_decl, is_sync_builtin_call): New functions.
(execute_oacc_device_lower): Support rewriting gang-private
variables using target hook, and fix up addr_expr and var_decl
nodes afterwards.
* target.def (adjust_private_decl, expand_var_decl): New hooks.
* config/gcn/gcn-protos.h (gcn_goacc_adjust_gangprivate_decl):
Rename to...
(gcn_goacc_adjust_private_decl): ...this.
* config/gcn/gcn-tree.c (gcn_goacc_adjust_gangprivate_decl):
Rename to...
(gcn_goacc_adjust_private_decl): ...this. Add LEVEL parameter.
* config/gcn/gcn.c (TARGET_GOACC_ADJUST_GANGPRIVATE_DECL): Rename
definition using gcn_goacc_adjust_gangprivate_decl...
(TARGET_GOACC_ADJUST_PRIVATE_DECL): ...to this, using
gcn_goacc_adjust_private_decl.
* config/nvptx/nvptx.c (tree-pretty-print.h): Include.
(gang_private_shared_size): New global variable.
(gang_private_shared_align): Likewise.
(gang_private_shared_sym): Likewise.
(gang_private_shared_hmap): Likewise.
(nvptx_option_override): Initialize these.
(nvptx_file_end): Output gang_private_shared_sym.
(nvptx_goacc_adjust_private_decl, nvptx_goacc_expand_var_decl):
New functions.
(nvptx_set_current_function): Clear gang_private_shared_hmap.
(TARGET_GOACC_ADJUST_PRIVATE_DECL): Define hook.
(TARGET_GOACC_EXPAND_VAR_DECL): Likewise.
libgomp/
PR middle-end/90115
* testsuite/libgomp.oacc-c-c++-common/private-atomic-1-gang.c: New
test.
* testsuite/libgomp.oacc-fortran/private-atomic-1-gang.f90:
Likewise.
* testsuite/libgomp.oacc-fortran/private-atomic-1-worker.f90:
Likewise.
Co-Authored-By: Chung-Lin Tang <cltang@codesourcery.com>
Co-Authored-By: Thomas Schwinge <thomas@codesourcery.com>
... to at least document/test/XFAIL nvptx offloading: PR83812 "operation not
supported on global/shared address space".
libgomp/
PR target/83812
* testsuite/libgomp.oacc-c-c++-common/private-atomic-1.c: New.
See local 'offload_targets' variable in
'libgomp/testsuite/lib/libgomp.exp:libgomp_check_effective_target_offload_target'
vs. global 'libgomp/testsuite/libgomp-test-support.exp.in:offload_targets'
variable.
libgomp/
* testsuite/lib/libgomp.exp
(check_effective_target_offload_target_nvptx): Don't shadow global
'offload_targets' variable.
Fix-up for recent commit 33b647956c
"OpenMP: Fix SIMT for complex/float reduction with && and ||"; see
commit d42088e453 "Avoid -latomic for amdgcn
offloading".
libgomp/
* testsuite/libgomp.c-c++-common/reduction-5.c: Restrict
'-latomic' to nvptx offloading compilation.
* testsuite/libgomp.c-c++-common/reduction-6.c: Likewise.
The team barrier should be notified of any new tasks that become runnable
as the result of a completing task, otherwise the barrier threads might
not resume processing available tasks, resulting in a hang.
2021-05-17 Kwok Cheung Yeung <kcy@codesourcery.com>
libgomp/
* task.c (omp_fulfill_event): Call gomp_team_barrier_set_task_pending
if new tasks generated.
* testsuite/libgomp.c-c++-common/task-detach-13.c: New.
gcc/ChangeLog:
* omp-low.c (finish_taskreg_scan): Use the proper detach decl.
libgomp/ChangeLog:
* testsuite/libgomp.c-c++-common/task-detach-12.c: New test.
* testsuite/libgomp.fortran/task-detach-12.f90: New test.
When a taskloop doesn't have any iterations, GOMP_taskloop* takes an early
return, doesn't create any tasks and more importantly, doesn't create
a taskgroup and doesn't register task reductions. But, the code emitted
in the callers assumes task reductions have been registered and performs
the reduction handling and task reduction unregistration. The pointer
to the task reduction private variables is reused, on input it is the alignment
and only on output it is the pointer, so in the case taskloop with no iterations
the caller attempts to dereference the alignment value as if it was a pointer
and crashes. We could in the early returns register the task reductions
only to have them looped over and unregistered in the caller, but I think
it is better to tell the caller there is nothing to task reduce and bypass
all that.
2021-05-11 Jakub Jelinek <jakub@redhat.com>
PR middle-end/100471
* omp-low.c (lower_omp_task_reductions): For OMP_TASKLOOP, if data
is 0, bypass the reduction loop including
GOMP_taskgroup_reduction_unregister call.
* taskloop.c (GOMP_taskloop): If GOMP_TASK_FLAG_REDUCTION and not
GOMP_TASK_FLAG_NOGROUP, when doing early return clear the task
reduction pointer.
* testsuite/libgomp.c/task-reduction-4.c: New test.
2021-05-07 Tobias Burnus <tobias@codesourcery.com>
Tom de Vries <tdevries@suse.de>
gcc/ChangeLog:
* omp-low.c (lower_rec_simd_input_clauses): Set max_vf = 1 if
a truth_value_p reduction variable is nonintegral.
libgomp/ChangeLog:
* testsuite/libgomp.c-c++-common/reduction-5.c: New test, testing
complex/floating-point || + && reduction with 'omp target'.
* testsuite/libgomp.c-c++-common/reduction-6.c: Likewise.
C/C++ permit logical AND and logical OR also with floating-point or complex
arguments by doing an unequal zero comparison; the result is an 'int' with
value one or zero. Hence, those are also permitted as reduction variable,
even though it is not the most sensible thing to do.
gcc/c/ChangeLog:
* c-typeck.c (c_finish_omp_clauses): Accept float + complex
for || and && reductions.
gcc/cp/ChangeLog:
* semantics.c (finish_omp_reduction_clause): Accept float + complex
for || and && reductions.
gcc/ChangeLog:
* omp-low.c (lower_rec_input_clauses, lower_reduction_clauses): Handle
&& and || with floating-point and complex arguments.
gcc/testsuite/ChangeLog:
* gcc.dg/gomp/clause-1.c: Use 'reduction(&:..)' instead of '...(&&:..)'.
libgomp/ChangeLog:
* testsuite/libgomp.c-c++-common/reduction-1.c: New test.
* testsuite/libgomp.c-c++-common/reduction-2.c: New test.
* testsuite/libgomp.c-c++-common/reduction-3.c: New test.
The test-case included in this patch contains this target region:
...
for (int i0 = 0 ; i0 < N0 ; i0++ )
counter_N0.i += 1;
...
When running with nvptx accelerator, the counter variable is expected to
be N0 after the region, but instead is N0 / 32. The problem is that rather
than getting the result for all warp lanes, we get it for just one lane.
This is caused by the implementation of SIMT being incomplete. It handles
regular reductions, but appearantly not user-defined reductions.
For now, handle this by disabling SIMT in this case, specifically by setting
sctx->max_vf to 1.
Tested libgomp on x86_64-linux with nvptx accelerator.
gcc/ChangeLog:
2021-05-03 Tom de Vries <tdevries@suse.de>
PR target/100321
* omp-low.c (lower_rec_input_clauses): Disable SIMT for user-defined
reduction.
libgomp/ChangeLog:
2021-05-03 Tom de Vries <tdevries@suse.de>
PR target/100321
* testsuite/libgomp.c/target-44.c: New test.
PR84878 fix adds an assertion which can fail, e.g. when stack pointer
is adjusted inside the loop. We have to prevent it and search earlier
for any 'strange' instruction. The solution is to skip the whole loop
if using 'note_stores' we found that one of hard registers is in
'df->regular_block_artificial_uses' set.
Also patch properly prohibit not single-set instruction in loop body.
gcc/ChangeLog:
PR rtl-optimization/100225
PR rtl-optimization/84878
* modulo-sched.c (sms_schedule): Use note_stores to skip loops
where we have an instruction which touches (writes) any hard
register from df->regular_block_artificial_uses set.
Allow not-single-set instruction only right before basic block
tail.
gcc/testsuite/ChangeLog:
PR rtl-optimization/100225
PR rtl-optimization/84878
* gcc.dg/pr100225.c: New test.
libgomp/ChangeLog:
* testsuite/libgomp.oacc-c-c++-common/atomic_capture-3.c: New test.
Consider the test-case libgomp.c/pr81778.c added in this commit, with
this core loop (note: CANARY_SIZE set to 0 for simplicity):
...
int s = 1;
#pragma omp target simd
for (int i = N - 1; i > -1; i -= s)
a[i] = 1;
...
which, given that N is 32, sets a[0..31] to 1.
After omp-expand, this looks like:
...
<bb 5> :
simduid.7 = .GOMP_SIMT_ENTER (simduid.7);
.omp_simt.8 = .GOMP_SIMT_ENTER_ALLOC (simduid.7);
D.3193 = -s;
s.9 = s;
D.3204 = .GOMP_SIMT_LANE ();
D.3205 = -s.9;
D.3206 = (int) D.3204;
D.3207 = D.3205 * D.3206;
i = D.3207 + 31;
D.3209 = 0;
D.3210 = -s.9;
D.3211 = D.3210 - i;
D.3210 = -s.9;
D.3212 = D.3211 / D.3210;
D.3213 = (unsigned int) D.3212;
D.3213 = i >= 0 ? D.3213 : 0;
<bb 19> :
if (D.3209 < D.3213)
goto <bb 6>; [87.50%]
else
goto <bb 7>; [12.50%]
<bb 6> :
a[i] = 1;
D.3215 = -s.9;
D.3219 = .GOMP_SIMT_VF ();
D.3216 = (int) D.3219;
D.3220 = D.3215 * D.3216;
i = D.3220 + i;
D.3209 = D.3209 + 1;
goto <bb 19>; [100.00%]
...
On nvptx, the first time bb6 is executed, i is in the 0..31 range (depending
on the lane that is executing) at bb entry.
So we have the following sequence:
- a[0..31] is set to 1
- i is updated to -32..-1
- D.3209 is updated to 1 (being 0 initially)
- bb19 is executed, and if condition (D.3209 < D.3213) == (1 < 32) evaluates
to true
- bb6 is once more executed, which should not happen because all the elements
that needed to be handled were already handled.
- consequently, elements that should not be written are written
- with CANARY_SIZE == 0, we may run into a libgomp error:
...
libgomp: cuCtxSynchronize error: an illegal memory access was encountered
...
and with CANARY_SIZE unmodified, we run into:
...
Expected 0, got 1 at base[-961]
Aborted (core dumped)
...
The cause of this is as follows:
- because the step s is a variable rather than a constant, an alternative
IV (D.3209 in our example) is generated in expand_omp_simd, and the
loop condition is tested in terms of the alternative IV rather than
the original IV (i in our example).
- the SIMT code in expand_omp_simd works by modifying step and initial value.
- The initial value fd->loop.n1 is loaded into a variable n1, which is
modified by the SIMT code and then used there-after.
- The step fd->loop.step is loaded into a variable step, which is modified
by the SIMT code, but afterwards there are uses of both step and
fd->loop.step.
- There are uses of fd->loop.step in the alternative IV handling code,
which should use step instead.
Fix this by introducing an additional variable orig_step, which is not
modified by the SIMT code and replacing all remaining uses of fd->loop.step
by either step or orig_step.
Build on x86_64-linux with nvptx accelerator, tested libgomp.
This fixes for-5.c and for-6.c FAILs I'm currently seeing on a quadro m1200
with driver 450.66.
gcc/ChangeLog:
2020-10-02 Tom de Vries <tdevries@suse.de>
* omp-expand.c (expand_omp_simd): Add step_orig, and replace uses of
fd->loop.step by either step or orig_step.
libgomp/ChangeLog:
2020-10-02 Tom de Vries <tdevries@suse.de>
* testsuite/libgomp.c/pr81778.c: New test.
When running the test-case included in this patch using an
nvptx accelerator, it fails in execution.
The problem is that the expansion of GOMP_SIMT_XCHG_BFLY is optimized away
during pass_jump as "trivially dead insns".
This is caused by this code in expand_GOMP_SIMT_XCHG_BFLY:
...
class expand_operand ops[3];
create_output_operand (&ops[0], target, mode);
...
expand_insn (targetm.code_for_omp_simt_xchg_bfly, 3, ops);
...
which doesn't guarantee that target is assigned to by the expanded insn.
F.i., if target is:
...
(gdb) call debug_rtx ( target )
(subreg/s/u:QI (reg:SI 40 [ _61 ]) 0)
...
then after expand_insn, we have:
...
(gdb) call debug_rtx ( ops[0].value )
(reg:QI 57)
...
See commit 3af3bec2e4 "internal-fn: Avoid dropping the lhs of some
calls [PR94941]" for a similar problem.
Fix this in the same way, by adding:
...
if (!rtx_equal_p (target, ops[0].value))
emit_move_insn (target, ops[0].value);
...
where applicable in the expand_GOMP_SIMT_* functions.
Tested libgomp on x86_64 with nvptx accelerator.
gcc/ChangeLog:
2021-04-28 Tom de Vries <tdevries@suse.de>
PR target/100232
* internal-fn.c (expand_GOMP_SIMT_ENTER_ALLOC)
(expand_GOMP_SIMT_LAST_LANE, expand_GOMP_SIMT_ORDERED_PRED)
(expand_GOMP_SIMT_VOTE_ANY, expand_GOMP_SIMT_XCHG_BFLY)
(expand_GOMP_SIMT_XCHG_IDX): Ensure target is assigned to.
If configured with --enable-offload-defaulted, configured but not installed
offload compilers and libgomp plugins are silently ignored. Useful for
distribution compilers where those are in separate optional packages.
2021-04-28 Jakub Jelinek <jakub@redhat.com>
Tobias Burnus <tobias@codesourcery.com>
ChangeLog:
* configure.ac (--enable-offload-defaulted): New.
* configure: Regenerate.
gcc/ChangeLog:
* configure.ac (OFFLOAD_DEFAULTED): AC_DEFINE if offload-defaulted.
* gcc.c (process_command): New variable.
(driver::maybe_putenv_OFFLOAD_TARGETS): If OFFLOAD_DEFAULTED,
set it if -foffload is defaulted.
* lto-wrapper.c (OFFLOAD_TARGET_DEFAULT_ENV): Define.
(compile_offload_image): If OFFLOAD_DEFAULTED and
OFFLOAD_TARGET_DEFAULT is in the environment, don't fail
if corresponding mkoffload can't be found.
(compile_images_for_offload_targets): Likewise. Free and clear
offload_names if no valid offload is found.
* config.in: Regenerate.
* configure: Regenerate.
libgomp/ChangeLog:
* configure.ac (OFFLOAD_DEFAULTED): AC_DEFINE if offload-defaulted.
* target.c (gomp_load_plugin_for_device): If set and if a plugin
can't be dlopened, silently assume it has no devices.
* Makefile.in: Regenerate.
* config.h.in: Regenerate.
* configure: Regenerate.
It turned out that a compiler built without offloading support
and one with can produce slightly different diagnostic.
Offloading support implies ENABLE_OFFLOAD which implies that
g->have_offload is set when offloading is actually needed.
In cgraphunit.c, the latter causes flag_generate_offload = 1,
which in turn affects tree.c's free_lang_data.
The result is that the front-end specific diagnostic gets reset
('tree_diagnostics_defaults (global_dc)'), which affects in this
case 'Warning' vs. 'warning' via the Fortran frontend.
Result: 'Warning:' vs. 'warning:'.
Side note: Other FE also override the diagnostic, leading to
similar differences, e.g. the C++ FE outputs mangled function
names differently, cf. patch thread.
libgomp/ChangeLog:
* testsuite/libgomp.oacc-fortran/par-reduction-2-1.f:
Use [Ww]arning in dg-bogus as FE diagnostic and default
diagnostic differ and the result depends on ENABLE_OFFLOAD.
* testsuite/libgomp.oacc-fortran/par-reduction-2-2.f: Likewise.
* testsuite/libgomp.oacc-fortran/parallel-dims.f90: Likewise.
* testsuite/libgomp.oacc-fortran/parallel-reduction.f90: Likewise.
gcc/testsuite/ChangeLog:
* gfortran.dg/goacc/classify-serial.f95:
Use [Ww]arning in dg-bogus as FE diagnostic and default
diagnostic differ and the result depends on ENABLE_OFFLOAD.
* gfortran.dg/goacc/kernels-decompose-2.f95: Likewise.
* gfortran.dg/goacc/routine-module-mod-1.f90: Likewise.
libatomic isn't built for amdgcn but reduction-16.c adds it
via -foffload=-latomic when offloading for nvptx is enabled.
The following avoids linker errors when offloading to amdgcn is enabled
as well.
2021-04-21 Richard Biener <rguenther@suse.de>
libgomp/
* testsuite/libgomp.c-c++-common/reduction-16.c: Use -latomic
only on nvptx-none.
For the tests modified below, the effective target line has to be effective
when compiling for an offload target, except that variable-not-offloaded.c
would compile with unified-share memory and pr86416-*.c if long double/float128
is supported.
The previous check used a run-time device ability check. This new variant
now enables those dg- lines when _compiling_ for nvptx or gcn.
libgomp/ChangeLog:
* testsuite/lib/libgomp.exp (offload_target_to_openacc_device_type):
New, based on check_effective_target_offload_target_nvptx.
(check_effective_target_offload_target_nvptx): Call it.
(check_effective_target_offload_target_amdgcn): New.
* testsuite/libgomp.c-c++-common/function-not-offloaded.c:
Require target offload_target_nvptx || offload_target_amdgcn.
* testsuite/libgomp.c-c++-common/variable-not-offloaded.c: Likewise.
* testsuite/libgomp.c/pr86416-1.c: Likewise.
* testsuite/libgomp.c/pr86416-2.c: Likewise.
As can be seen under valgrind, the testcase didn't bind in the last part
the fortran pointers properly to the c pointers.
2021-04-14 Jakub Jelinek <jakub@redhat.com>
PR testsuite/100071
* testsuite/libgomp.fortran/alloc-1.F90: Call c_f_pointer after last
cp = omp_alloc with cp, p arguments instead of cq, q and call
c_f_pointer after last cq = omp_alloc with cq, q.
We have seen an ICE both on trunk and devel/omp/gcc-10 branches which can
be reprodued with this simple testcase. It occurs if an OpenACC loop has
a collapse clause and any of the loop being collapsed uses GT or GE
condition. This issue is specific to OpenACC.
int main (void)
{
int ix, iy;
int dim_x = 16, dim_y = 16;
{
for (iy = dim_y - 1; iy > 0; --iy)
for (ix = dim_x - 1; ix > 0; --ix)
;
}
}
The problem is caused by a failing assertion in expand_oacc_collapse_init.
It checks that cond_code for fd->loop should be same as cond_code for all
the loops that are being collapsed. As the cond_code for fd->loop is
LT_EXPR with collapse clause (set at the end of omp_extract_for_data),
this assertion forces that all the loop in collapse clause should use
< operator.
There does not seem to be anything in the code which demands this
condition as loop with > condition works ok otherwise. I digged old
mailing list a bit but could not find any discussion on this change.
Looking at the code, expand_oacc_for checks that fd->loop->cond_code is
either LT_EXPR or GT_EXPR. I guess the original intention was to have
similar checks on the loop which are being collapsed. But the way check
was written does not acheive that.
I have fixed it by modifying the check in the assertion to be same as
check on fd->loop->cond_code.
I tested goacc and libgomp (with nvptx offloading) and did not see any
regression. I have added new tests to check collapse with GT/GE condition.
PR middle-end/98088
gcc/
* omp-expand.c (expand_oacc_collapse_init): Update condition in
a gcc_assert.
gcc/testsuite/
* c-c++-common/goacc/collapse-2.c: New.
libgomp/
* testsuite/libgomp.oacc-c-c++-common/collapse-2.c: Add check
for loop with GT/GE condition.
* testsuite/libgomp.oacc-c-c++-common/collapse-3.c: Likewise.
pthread_setspecific second argument is const void *, so that one can
call it even with pointers to const, but the function only stores the
pointer and does nothing else, so the new assumption of -Wmaybe-uninitialized
that functions taking such pointers will read from what those pointers
will point to is wrong. Maybe it would be useful to have some whitelist
of functions that surely don't do that.
Anyway, in this case it is easy to workaround the warning by moving the
pthread_setspecific call after the initialization without slowing anything
down.
2021-04-09 Jakub Jelinek <jakub@redhat.com>
PR libgomp/99984
* team.c (gomp_thread_start): Call pthread_setspecific for
!(defined HAVE_TLS || defined USE_EMUTLS) only after local_thr
has been initialized to avoid false positive warning.
For unknown reasons, this had gotten added for the libgomp HSA plugin in commit
b8d89b03db (r242749) "Remove build dependence on
HSA run-time", and later propagated into the GCN plugin.
libgomp/
* plugin/plugin-gcn.c (init_environment_variables): Don't prepend
the 'HSA_RUNTIME_LIB' path to 'libhsa-runtime64.so'.
* plugin/configfrag.ac (HSA_RUNTIME_LIB): Clean up.
* config.h.in: Regenerate.
* configure: Likewise.
Fixup for recent commit d28f3da11d "openacc: Fix
lowering for derived-type mappings through array elements". With nvptx
offloading we see the usual:
[...]/libgomp.oacc-fortran/derivedtypes-arrays-1.f90: In function 'MAIN__._omp_fn.0':
[...]/libgomp.oacc-fortran/derivedtypes-arrays-1.f90:90:40: warning: using vector_length (32), ignoring 1
libgomp/
* testsuite/libgomp.oacc-fortran/derivedtypes-arrays-1.f90:
OpenACC 'serial' construct diagnostic for nvptx offloading.
For variables with 'declare target' attribute,
varpool_node::get_create marks variables as offload; however,
if the node already exists, it is not updated. C/C++ may tag
decl with 'declare target implicit', which may only be after
varpool creation turned into 'declare target' or 'declare target link';
in this case, the tagging has to happen in the FE.
gcc/c/ChangeLog:
PR c++/99509
* c-decl.c (finish_decl): For 'omp declare target implicit' vars,
ensure that the varpool node is marked as offloadable.
gcc/cp/ChangeLog:
PR c++/99509
* decl.c (cp_finish_decl): For 'omp declare target implicit' vars,
ensure that the varpool node is marked as offloadable.
libgomp/ChangeLog:
PR c++/99509
* testsuite/libgomp.c-c++-common/declare_target-1.c: New test.
Some gcc configurations default to -m32 but support -m64 too. This patch
just makes the ILP32 tests more reliable by following what e.g. libsanitizer
configury does.
2021-03-04 Jakub Jelinek <jakub@redhat.com>
* configure.ac: Add AC_CHECK_SIZEOF([void *]).
* plugin/configfrag.ac: Check $ac_cv_sizeof_void_p value instead of
checking of -m32 or -mx32 options on the command line.
* config.h.in: Regenerated.
* configure: Regenerated.
This fails everywhere on Darwin, which does not have support for
symbol aliases. Add a dg-require-alias to UNSUPPORT it.
libgomp/ChangeLog:
* testsuite/libgomp.c-c++-common/pr96390.c: Require alias
support from the target.
This adds support for the task detach clause to taskwait and taskgroup, and
simplifies the handling of the detach clause by moving most of the extra
handling required for detach tasks to omp_fulfill_event.
2021-02-25 Kwok Cheung Yeung <kcy@codesourcery.com>
Jakub Jelinek <jakub@redhat.com>
libgomp/
PR libgomp/98738
* libgomp.h (enum gomp_task_kind): Add GOMP_TASK_DETACHED.
(struct gomp_task): Replace detach and completion_sem fields with
union containing completion_sem and detach_team. Add deferred_p
field.
(struct gomp_team): Remove task_detach_queue.
* task.c: Include assert.h.
(gomp_init_task): Initialize deferred_p and completion_sem fields.
Rearrange initialization order of fields.
(task_fulfilled_p): Delete.
(GOMP_task): Use address of task as the event handle. Remove
initialization of detach field. Initialize deferred_p field.
Use automatic local for completion_sem. Initialize detach_team field
for deferred tasks.
(gomp_barrier_handle_tasks): Remove handling of task_detach_queue.
Set kind of suspended detach task to GOMP_TASK_DETACHED and
decrement task_running_count. Move finish_cancelled block out of
else branch. Relocate call to gomp_team_barrier_done.
(GOMP_taskwait): Handle tasks with completion events that have not
been fulfilled.
(GOMP_taskgroup_end): Likewise.
(omp_fulfill_event): Use address of task as event handle. Post to
completion_sem for undeferred tasks. Clear detach_team if task
has not finished. For finished tasks, handle post-execution tasks,
call gomp_team_barrier_wake if necessary, and free task.
* team.c (gomp_new_team): Remove initialization of task_detach_queue.
(free_team): Remove free of task_detach_queue.
* testsuite/libgomp.c-c++-common/task-detach-1.c: Fix formatting.
* testsuite/libgomp.c-c++-common/task-detach-2.c: Fix formatting.
* testsuite/libgomp.c-c++-common/task-detach-3.c: Fix formatting.
* testsuite/libgomp.c-c++-common/task-detach-4.c: Fix formatting.
* testsuite/libgomp.c-c++-common/task-detach-5.c: Fix formatting.
Change data-sharing of detach events on enclosing parallel to private.
* testsuite/libgomp.c-c++-common/task-detach-6.c: Likewise. Remove
taskwait directive.
* testsuite/libgomp.c-c++-common/task-detach-7.c: New.
* testsuite/libgomp.c-c++-common/task-detach-8.c: New.
* testsuite/libgomp.c-c++-common/task-detach-9.c: New.
* testsuite/libgomp.c-c++-common/task-detach-10.c: New.
* testsuite/libgomp.c-c++-common/task-detach-11.c: New.
* testsuite/libgomp.fortran/task-detach-1.f90: Fix formatting.
* testsuite/libgomp.fortran/task-detach-2.f90: Fix formatting.
* testsuite/libgomp.fortran/task-detach-3.f90: Fix formatting.
* testsuite/libgomp.fortran/task-detach-4.f90: Fix formatting.
* testsuite/libgomp.fortran/task-detach-5.f90: Fix formatting.
Change data-sharing of detach events on enclosing parallel to private.
* testsuite/libgomp.fortran/task-detach-6.f90: Likewise. Remove
taskwait directive.
* testsuite/libgomp.fortran/task-detach-7.f90: New.
* testsuite/libgomp.fortran/task-detach-8.f90: New.
* testsuite/libgomp.fortran/task-detach-9.f90: New.
* testsuite/libgomp.fortran/task-detach-10.f90: New.
* testsuite/libgomp.fortran/task-detach-11.f90: New.
gcc/fortran/ChangeLog:
PR fortran/99171
* trans-openmp.c (gfc_omp_is_optional_argument): Regard optional
dummy procs as nonoptional as no special treatment is needed.
libgomp/ChangeLog:
PR fortran/99171
* testsuite/libgomp.fortran/dummy-procs-1.f90: New test.
This patch disallows selecting components of array sections in update
directives for OpenACC, as specified in OpenACC 3.0, "2.14.4. Update
Directive":
In Fortran, members of variables of derived type may appear, including
a subarray of a member. Members of subarrays of derived type may
not appear.
The diagnostic for attempting to use the same construct on other
directives has also been improved.
gcc/fortran/
* openmp.c (resolve_omp_clauses): Disallow selecting components
of arrays of derived type.
gcc/testsuite/
* gfortran.dg/goacc/array-with-dt-2.f90: Remove expected errors.
* gfortran.dg/goacc/array-with-dt-6.f90: New test.
* gfortran.dg/goacc/mapping-tests-2.f90: Update expected error.
* gfortran.dg/goacc/ref_inquiry.f90: Update expected errors.
* gfortran.dg/gomp/ref_inquiry.f90: Likewise.
libgomp/
* testsuite/libgomp.oacc-fortran/array-stride-dt-1.f90: Remove
expected errors.
This patch fixes lowering of derived-type mappings which select elements
of arrays of derived types, and similar. These would previously lead
to ICEs.
With this change, OpenACC directives can pass through constructs that
are no longer recognized by the gimplifier, hence alterations are needed
there also.
gcc/fortran/
* trans-openmp.c (gfc_trans_omp_clauses): Handle element selection
for arrays of derived types.
gcc/
* gimplify.c (gimplify_scan_omp_clauses): Handle ATTACH_DETACH
for non-decls.
gcc/testsuite/
* gfortran.dg/goacc/array-with-dt-1.f90: New test.
* gfortran.dg/goacc/array-with-dt-3.f90: Likewise.
* gfortran.dg/goacc/array-with-dt-4.f90: Likewise.
* gfortran.dg/goacc/array-with-dt-5.f90: Likewise.
* gfortran.dg/goacc/derived-chartypes-1.f90: Re-enable test.
* gfortran.dg/goacc/derived-chartypes-2.f90: Likewise.
* gfortran.dg/goacc/derived-classtypes-1.f95: Uncomment
previously-broken directives.
libgomp/
* testsuite/libgomp.oacc-fortran/derivedtypes-arrays-1.f90: New test.
* testsuite/libgomp.oacc-fortran/update-dt-array.f90: Likewise.
Linux man-pages 5.07 wrongly declares syscall output type as int. This error
was fixed in release 5.10, so this patch reverts my recent change.
2021-02-11 Uroš Bizjak <ubizjak@gmail.com>
libgomp/
* config/linux/x86/futex.h (__futex_wait):
Revert output type back to long.
(__futex_wake): Ditto.
(futex_wait): Update for revert.
(futex_wake): Ditto.
Move syscall asms to static inline wrapper functions to improve #ifdeffery.
Also correct output type to int and timeout type to void *.
2021-02-11 Uroš Bizjak <ubizjak@gmail.com>
libgomp/
* config/linux/x86/futex.h (__futex_wait): New static inline
wrapper function. Correct output type to int and
timeout type to void *.
(__futex_wake): New static inline wrapper function.
Correct output type to int.
(futex_wait): Use __futex_wait.
(futex_wake): Use __futex_wake.
This patch adds some XFAILs for PR98979 until the patch to fix them has
been approved. See:
https://gcc.gnu.org/pipermail/gcc-patches/2021-February/564711.html
gcc/testsuite/
PR fortran/98979
* gfortran.dg/goacc/array-with-dt-2.f90: Add expected errors.
* gfortran.dg/goacc/derived-chartypes-1.f90: Skip ICEing test.
* gfortran.dg/goacc/derived-chartypes-2.f90: Likewise.
libgomp/
PR fortran/98979
* testsuite/libgomp.oacc-fortran/array-stride-dt-1.f90: Add expected
errors.
OpenACC 3.0 ("2.14.4. Update Directive") states:
Noncontiguous subarrays may appear. It is implementation-specific
whether noncontiguous regions are updated by using one transfer for
each contiguous subregion, or whether the non-contiguous data is
packed, transferred once, and unpacked, or whether one or more larger
subarrays (no larger than the smallest contiguous region that contains
the specified subarray) are updated.
This patch relaxes some conditions in the Fortran front-end so that
strided accesses are permitted for update directives.
gcc/fortran/
* openmp.c (resolve_omp_clauses): Omit OpenACC update in
contiguity check and stride-specified error.
gcc/testsuite/
* gfortran.dg/goacc/array-with-dt-2.f90: New test.
libgomp/
* testsuite/libgomp.oacc-fortran/array-stride-dt-1.f90: New test.
On Wed, Jan 20, 2021 at 05:04:39PM +0100, Florian Weimer wrote:
> Sorry, this appears to cause OpenMP task state corruption in RPM. We
> have only seen this on s390x.
Haven't actually verified it, but my suspection is that this is a caller
stack corruption.
We play with fire with the GOMP_task API/ABI extensions, the GOMP_task
function used to be:
void
GOMP_task (void (*fn) (void *), void *data, void (*cpyfn) (void *, void *),
long arg_size, long arg_align, bool if_clause, unsigned flags);
and later:
void
GOMP_task (void (*fn) (void *), void *data, void (*cpyfn) (void *, void *),
long arg_size, long arg_align, bool if_clause, unsigned flags,
void **depend);
and later:
void
GOMP_task (void (*fn) (void *), void *data, void (*cpyfn) (void *, void *),
long arg_size, long arg_align, bool if_clause, unsigned flags,
void **depend, int priority);
and now:
void
GOMP_task (void (*fn) (void *), void *data, void (*cpyfn) (void *, void *),
long arg_size, long arg_align, bool if_clause, unsigned flags,
void **depend, int priority, void *detach)
and which of those depend, priority and detach argument is present depends
on the bits in flags.
I'm afraid the compiler just decided to spill the detach = NULL store in
if ((flags & GOMP_TASK_FLAG_DETACH) == 0)
detach = NULL;
on s390x into the argument stack slot. Not a problem if the caller passes
all those 10 arguments, but if not, can clobber random stack location.
This hack should fix it up. Priority doesn't need changing, but I've
changed it anyway just to be safe. With the patch none of the 3 arguments
are ever modified, so I'd hope gcc doesn't decide to spill something
unrelated there.
2021-01-20 Jakub Jelinek <jakub@redhat.com>
* task.c (GOMP_task): Rename priority argument to priority_arg,
add priority automatic variable and modify that variable. Instead of
clearing detach argument when GOMP_TASK_FLAG_DETACH bit is not set,
check flags for that bit.
This patch introduces gomp_sem_getcount wrapper, which uses sem_getvalue
for POSIX and atomic loads for linux futex and accel. rtems for now
remains broken.
2021-01-18 Jakub Jelinek <jakub@redhat.com>
* config/linux/sem.h (gomp_sem_getcount): New function.
* config/posix/sem.h (gomp_sem_getcount): New function.
* config/posix/sem.c (gomp_sem_getcount): New function.
* config/accel/sem.h (gomp_sem_getcount): New function.
* task.c (task_fulfilled_p): Use gomp_sem_getcount.
(omp_fulfill_event): Likewise.
The recent changes to error on mixing -march=i386 and -fcf-protection broke
bootstrap. This patch changes lib{atomic,gomp,itm} configury, so that it
only adds -march=i486 to flags if really needed (i.e. when 486 or later isn't
on by default already). Similarly, it will not use ifuncs if -mcx16
(or -march=i686 for 32-bit) is on by default.
2021-01-15 Jakub Jelinek <jakub@redhat.com>
PR target/70454
libatomic/
* configure.tgt: For i?86 and x86_64 determine if -march=i486 needs to
be added through preprocessor check on
__GCC_HAVE_SYNC_COMPARE_AND_SWAP_4. Determine if try_ifunc is needed
based on preprocessor check on __GCC_HAVE_SYNC_COMPARE_AND_SWAP_16
or __GCC_HAVE_SYNC_COMPARE_AND_SWAP_8.
libgomp/
* configure.tgt: For i?86 and x86_64 determine if -march=i486 needs to
be added through preprocessor check on
__GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
libitm/
* configure.tgt: For i?86 and x86_64 determine if -march=i486 needs to
be added through preprocessor check on
__GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
As recently again discussed in <https://gcc.gnu.org/PR97436> "[nvptx] -m32
support", nvptx offloading other than for 64-bit host has never been
implemented, tested, supported. So we simply should buildn't the nvptx libgomp
plugin in this case.
This avoids build problems if, for example, in a (standard) bi-arch
x86_64-pc-linux-gnu '-m64'/'-m32' build, libcuda is available only in a 64-bit
variant but not in a 32-bit one, which, for example, is the case if you build
GCC against the CUDA toolkit's 'stubs/libcuda.so' (see
<https://stackoverflow.com/a/52784819>).
This amends PR65099 commit a92defdab7 (r225560)
"[nvptx offloading] Only 64-bit configurations are currently supported" to
match the way we're doing this for the HSA/GCN plugins.
libgomp/
PR libgomp/65099
* plugin/configfrag.ac (PLUGIN_NVPTX): Restrict to supported
configurations.
* configure: Regenerate.
* plugin/plugin-nvptx.c (nvptx_get_num_devices): Remove 64-bit
check.
The libgomp texinfo docs lead to an invalid "up" link on the Top node,
which we can avoid similarly to the Top link in the main GCC manual.
2020-12-28 Sandra Loosemore <sandra@codesourcery.com>
libgomp/
* libgomp.texi (Top): Avoid bad "up" link.
The attached testcase is miscompiled, because we optimize shared clauses
to firstprivate when task body can't modify the variable even when the
task has depend clause. That is wrong, because firstprivate means the
variable will be copied immediately when the task is created, while with
depend clause some other task might change it later before the dependencies
are satisfied and the task should observe the value only after the change.
2020-12-18 Jakub Jelinek <jakub@redhat.com>
* gimplify.c (struct gimplify_omp_ctx): Add has_depend member.
(gimplify_scan_omp_clauses): Set it to true if OMP_CLAUSE_DEPEND
appears on OMP_TASK.
(gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Force
GOVD_WRITTEN on shared variables if task construct has depend clause.
* testsuite/libgomp.c/task-6.c: New test.
These are the same header files that exist in the Radeon Open Compute Runtime
project (as of October 2020), but they have been specially relicensed by AMD
for use in GCC.
The header files retain AMD copyright.
include/ChangeLog:
* hsa.h: Replace whole file.
* hsa_ext_amd.h: New file.
* hsa_ext_image.h: New file.
libgomp/ChangeLog:
* plugin/plugin-gcn.c: Include hsa_ext_amd.h.
(HSA_AMD_AGENT_INFO_COMPUTE_UNIT_COUNT): Delete redundant definition.
The change in major version (and the increment from Darwin19 to 20)
caused libtool tests to fail which resulted in incorrect build settings
for shared libraries.
We take this opportunity to sort out the shared undefined symbols state
rather than propagating the current unsound behaviour into a new rev.
This change means that we default to the case that missing symbols are
considered an error, and if one wants to allow this intentionally, the
confiuration for that case should be set appropriately.
Three existing cases need undefined dynamic lookup:
libitm, where there is already a configuration mechanism to add the
flags.
libcc1, where we add simple configuration to add the flags for Darwin.
libsanitizer, where we can add to the existing extra flags.
libcc1/ChangeLog:
PR target/97865
* Makefile.am: Add dynamic_lookup to LD flags for Darwin.
* configure.ac: Test for Darwin host and set a flag.
* Makefile.in: Regenerate.
* configure: Regenerate.
libitm/ChangeLog:
PR target/97865
* configure.tgt: Add dynamic_lookup to XLDFLAGS for Darwin.
* configure: Regenerate.
libsanitizer/ChangeLog:
PR target/97865
* configure.tgt: Add dynamic_lookup to EXTRA_CXXFLAGS for
Darwin.
* configure: Regenerate.
ChangeLog:
PR target/97865
* libtool.m4: Update handling of Darwin platform link flags
for Darwin20.
gcc/ChangeLog:
PR target/97865
* configure: Regenerate.
libatomic/ChangeLog:
PR target/97865
* configure: Regenerate.
libbacktrace/ChangeLog:
PR target/97865
* configure: Regenerate.
libffi/ChangeLog:
PR target/97865
* configure: Regenerate.
libgfortran/ChangeLog:
PR target/97865
* configure: Regenerate.
libgomp/ChangeLog:
PR target/97865
* configure: Regenerate.
libhsail-rt/ChangeLog:
PR target/97865
* configure: Regenerate.
libobjc/ChangeLog:
PR target/97865
* configure: Regenerate.
libphobos/ChangeLog:
PR target/97865
* configure: Regenerate.
libquadmath/ChangeLog:
PR target/97865
* configure: Regenerate.
libssp/ChangeLog:
PR target/97865
* configure: Regenerate.
libstdc++-v3/ChangeLog:
PR target/97865
* configure: Regenerate.
libvtv/ChangeLog:
PR target/97865
* configure: Regenerate.
zlib/ChangeLog:
PR target/97865
* configure: Regenerate.
The testcase had invalid assumptions about which loop iterations would run
first and last.
libgomp/ChangeLog
* testsuite/libgomp.oacc-fortran/atomic_capture-1.f90 (main): Adjust
expected results.
Ensure the code will continue to compile when elf.h gets these definitions.
libgomp/ChangeLog:
* plugin/plugin-gcn.c: Don't redefine relocations if elf.h has them.
(reserved): Delete unused define.
This removes the nest-var ICV, expressing nesting in terms of the
max-active-levels-var ICV instead. The max-active-levels-var ICV
is now per data environment rather than per device.
2020-11-18 Kwok Cheung Yeung <kcy@codesourcery.com>
libgomp/
* env.c (gomp_global_icv): Remove nest_var field. Add
max_active_levels_var field.
(gomp_max_active_levels_var): Remove.
(parse_boolean): Return true on success.
(handle_omp_display_env): Express OMP_NESTED in terms of
max_active_levels_var. Change format specifier for
max_active_levels_var.
(initialize_env): Set max_active_levels_var from
OMP_MAX_ACTIVE_LEVELS, OMP_NESTED, OMP_NUM_THREADS and
OMP_PROC_BIND.
* icv.c (omp_set_nested): Express in terms of
max_active_levels_var.
(omp_get_nested): Likewise.
(omp_set_max_active_levels): Use max_active_levels_var field instead
of gomp_max_active_levels_var.
(omp_get_max_active_levels): Likewise.
* libgomp.h (struct gomp_task_icv): Remove nest_var field. Add
max_active_levels_var field.
(gomp_supported_active_levels): Set to UCHAR_MAX.
(gomp_max_active_levels_var): Delete.
* libgomp.texi (omp_get_nested): Update documentation.
(omp_set_nested): Likewise.
(OMP_MAX_ACTIVE_LEVELS): Likewise.
(OMP_NESTED): Likewise.
(OMP_NUM_THREADS): Likewise.
(OMP_PROC_BIND): Likewise.
* parallel.c (gomp_resolve_num_threads): Replace reference
to nest_var with max_active_levels_var. Use max_active_levels_var
field instead of gomp_max_active_levels_var.
As typically configured, newlib's libc.a does not build 'posix' and,
hence, usleep is not available. Thus, use the same fallback as for nvptx.
libgomp/
* testsuite/libgomp.c/usleep.h (fallback_usleep): Renamed from
nvptx_usleep; use also for device={arch(gcn)}.
This patch adds support for custom allocators on private/firstprivate
clauses for task (and taskloop) constructs. Private didn't need anything
special, but firstprivate if it is passed by reference needs the GOMP_alloc
calls in the copyfn and GOMP_free in the task body.
2020-11-14 Jakub Jelinek <jakub@redhat.com>
* gimplify.c (gimplify_omp_for): Add OMP_CLAUSE_ALLOCATE_ALLOCATOR
decls as firstprivate on task clauses even when allocate clause
decl is not lastprivate.
* omp-low.c (install_var_field): Don't dereference omp_is_reference
types if mask is 33 rather than 1.
(scan_sharing_clauses): Populate allocate_map even for task
constructs. For now remove it back for variables mentioned in
reduction and in_reduction clauses on task/taskloop constructs
or on VLA task firstprivates. For firstprivate on task construct,
install the var field into field_map with by_ref and 33 instead
of false and 1 if mentioned in allocate clause.
(lower_private_allocate): Set TREE_THIS_NOTRAP on the created
MEM_REF.
(lower_rec_input_clauses): Handle allocate for task firstprivatized
non-VLA variables.
(create_task_copyfn): Likewise.
* testsuite/libgomp.c-c++-common/allocate-1.c (struct S): New type.
(foo): Add tests for non-VLA private and firstprivate clauses on
omp task.
(bar): Likewise. Remove taking of address from private/firstprivate
variables.
* testsuite/libgomp.c++/allocate-1.C (struct S): New type.
(foo): Add p, q, px and s arguments. Add tests for array reductions
and for non-VLA private and firstprivate clauses on omp task.
(bar): Removed.
(main): Adjust foo caller. Don't call bar.
Document status quo re PR94358 "[OMP] Privatize internal array variables
introduced by the Fortran FE".
libgomp/
PR fortran/94358
* testsuite/libgomp.oacc-fortran/pr94358-1.f90: New.
Co-authored-by: Thomas Schwinge <thomas@codesourcery.com>
This adds allocate clause support for array section reductions.
Furthermore, it fixes one bug that would cause inscan reductions with
allocate to be rejected by C, and for now just ignores allocate for
inscan/task reductions, that will need slightly more work.
2020-11-13 Jakub Jelinek <jakub@redhat.com>
gcc/
* omp-low.c (scan_sharing_clauses): For now remove for reduction
clauses with inscan or task modifiers decl from allocate_map.
(lower_private_allocate): Handle TYPE_P (new_var).
(lower_rec_input_clauses): Handle allocate clause for C/C++ array
reductions.
gcc/c/
* c-typeck.c (c_finish_omp_clauses): Don't clear
OMP_CLAUSE_REDUCTION_INSCAN unless reduction_seen == -2.
libgomp/
* testsuite/libgomp.c-c++-common/allocate-1.c (foo): Add tests
for array reductions.
(main): Adjust foo callers.
For now, task/taskloop constructs aren't handled and C/C++ array reductions
and reductions with task or inscan modifiers need further work.
Instead of calling omp_alloc/omp_free (where the former doesn't have
alignment argument and omp_aligned_alloc is 5.1 only feature), this calls
GOMP_alloc/GOMP_free, so that the library can fail if it would fall back
into NULL (exception is zero length allocations).
2020-11-12 Jakub Jelinek <jakub@redhat.com>
gcc/
* builtin-types.def (BT_FN_PTR_SIZE_SIZE_PTRMODE): New function type.
* omp-builtins.def (BUILT_IN_GOACC_DECLARE): Move earlier.
(BUILT_IN_GOMP_ALLOC, BUILT_IN_GOMP_FREE): New builtins.
* gimplify.c (gimplify_scan_omp_clauses): Force allocator into a
decl if it is not NULL, INTEGER_CST or decl.
(gimplify_adjust_omp_clauses): Clear GOVD_EXPLICIT on explicit clauses
which are being removed. Remove allocate clauses for variables not seen
if they are private, firstprivate or linear too. Call
omp_notice_variable on the allocator otherwise.
(gimplify_omp_for): Handle iterator vars mentioned in allocate clauses
similarly to non-is_gimple_reg iterators.
* omp-low.c (struct omp_context): Add allocate_map field.
(delete_omp_context): Delete it.
(scan_sharing_clauses): Fill it from allocate clauses. Remove it
if mentioned also in shared clause.
(lower_private_allocate): New function.
(lower_rec_input_clauses): Handle allocate clause for privatized
variables, except for task/taskloop, C/C++ array reductions for now
and task/inscan variables.
(lower_send_shared_vars): Don't consider variables in allocate_map
as shared.
* omp-expand.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
expand_omp_for_static_chunk): Use expand_omp_build_assign instead of
gimple_build_assign + gsi_insert_after.
* builtins.c (builtin_fnspec): Handle BUILTIN_GOMP_ALLOC and
BUILTIN_GOMP_FREE.
* tree-ssa-ccp.c (evaluate_stmt): Handle BUILTIN_GOMP_ALLOC.
* tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Handle
BUILTIN_GOMP_ALLOC.
(mark_all_reaching_defs_necessary_1): Handle BUILTIN_GOMP_ALLOC
and BUILTIN_GOMP_FREE.
(propagate_necessity): Likewise.
gcc/fortran/
* f95-lang.c (ATTR_ALLOC_WARN_UNUSED_RESULT_SIZE_2_NOTHROW_LIST):
Define.
(gfc_init_builtin_functions): Add alloc_size and warn_unused_result
attributes to __builtin_GOMP_alloc.
* types.def (BT_PTRMODE): New primitive type.
(BT_FN_VOID_PTR_PTRMODE, BT_FN_PTR_SIZE_SIZE_PTRMODE): New function
types.
libgomp/
* libgomp.map (GOMP_alloc, GOMP_free): Export at GOMP_5.0.1.
* omp.h.in (omp_alloc): Add malloc and alloc_size attributes.
* libgomp_g.h (GOMP_alloc, GOMP_free): Declare.
* allocator.c (omp_aligned_alloc): New for now static function,
add alignment argument and handle it.
(omp_alloc): Reimplement using omp_aligned_alloc.
(GOMP_alloc, GOMP_free): New functions.
(omp_free): Add ialias.
* testsuite/libgomp.c-c++-common/allocate-1.c: New test.
* testsuite/libgomp.c++/allocate-1.C: New test.
This patch implements some parts of the target variable mapping changes
specified in OpenMP 5.0, including base-pointer attachment/detachment
behavior for array section list-items in map clauses, and ordering of
map clauses according to map kind.
2020-11-10 Chung-Lin Tang <cltang@codesourcery.com>
gcc/c-family/ChangeLog:
* c-common.h (c_omp_adjust_map_clauses): New declaration.
* c-omp.c (struct map_clause): Helper type for c_omp_adjust_map_clauses.
(c_omp_adjust_map_clauses): New function.
gcc/c/ChangeLog:
* c-parser.c (c_parser_omp_target_data): Add use of
new c_omp_adjust_map_clauses function. Add GOMP_MAP_ATTACH_DETACH as
handled map clause kind.
(c_parser_omp_target_enter_data): Likewise.
(c_parser_omp_target_exit_data): Likewise.
(c_parser_omp_target): Likewise.
* c-typeck.c (handle_omp_array_sections): Adjust COMPONENT_REF case to
use GOMP_MAP_ATTACH_DETACH map kind for C_ORT_OMP region type.
(c_finish_omp_clauses): Adjust bitmap checks to allow struct decl and
same struct field access to co-exist on OpenMP construct.
gcc/cp/ChangeLog:
* parser.c (cp_parser_omp_target_data): Add use of
new c_omp_adjust_map_clauses function. Add GOMP_MAP_ATTACH_DETACH as
handled map clause kind.
(cp_parser_omp_target_enter_data): Likewise.
(cp_parser_omp_target_exit_data): Likewise.
(cp_parser_omp_target): Likewise.
* semantics.c (handle_omp_array_sections): Adjust COMPONENT_REF case to
use GOMP_MAP_ATTACH_DETACH map kind for C_ORT_OMP region type. Fix
interaction between reference case and attach/detach.
(finish_omp_clauses): Adjust bitmap checks to allow struct decl and
same struct field access to co-exist on OpenMP construct.
gcc/ChangeLog:
* gimplify.c (is_or_contains_p): New static helper function.
(omp_target_reorder_clauses): New function.
(gimplify_scan_omp_clauses): Add use of omp_target_reorder_clauses to
reorder clause list according to OpenMP 5.0 rules. Add handling of
GOMP_MAP_ATTACH_DETACH for OpenMP cases.
* omp-low.c (is_omp_target): New static helper function.
(scan_sharing_clauses): Add scan phase handling of GOMP_MAP_ATTACH/DETACH
for OpenMP cases.
(lower_omp_target): Add lowering handling of GOMP_MAP_ATTACH/DETACH for
OpenMP cases.
gcc/testsuite/ChangeLog:
* c-c++-common/gomp/clauses-2.c: Remove dg-error cases now valid.
* gfortran.dg/gomp/map-2.f90: Likewise.
* c-c++-common/gomp/map-5.c: New testcase.
libgomp/ChangeLog:
* libgomp.h (enum gomp_map_vars_kind): Adjust enum values to be bit-flag
usable.
* oacc-mem.c (acc_map_data): Adjust gomp_map_vars argument flags to
'GOMP_MAP_VARS_OPENACC | GOMP_MAP_VARS_ENTER_DATA'.
(goacc_enter_datum): Likewise for call to gomp_map_vars_async.
(goacc_enter_data_internal): Likewise.
* target.c (gomp_map_vars_internal):
Change checks of GOMP_MAP_VARS_ENTER_DATA to use bit-and (&). Adjust use
of gomp_attach_pointer for OpenMP cases.
(gomp_exit_data): Add handling of GOMP_MAP_DETACH.
(GOMP_target_enter_exit_data): Add handling of GOMP_MAP_ATTACH.
* testsuite/libgomp.c-c++-common/ptr-attach-1.c: New testcase.
Avoid code duplication, and better test what we expect to happen.
libgomp/
PR target/85486
* testsuite/libgomp.oacc-c-c++-common/pr85486-2.c: Simplify and enhance.
* testsuite/libgomp.oacc-c-c++-common/pr85486-3.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/pr85486.c: Likewise.
This changes makes 'dg-warning', 'dg-error', 'dg-bogus', 'dg-message' behave as
expected, and also enables use of relative line numbers as well as 'dg-line'.
libgomp/
PR testsuite/80219
PR testsuite/85303
* testsuite/lib/libgomp.exp (libgomp_init): Set
'gcc_warning_prefix', 'gcc_error_prefix'.
This marks all variants of declare variant also declare target if the base
functions are called directly in target regions or declare target functions.
2020-10-28 Jakub Jelinek <jakub@redhat.com>
gcc/
* omp-offload.c (omp_declare_target_tgt_fn_r): Handle direct calls to
declare variant base functions.
libgomp/
* testsuite/libgomp.c/target-42.c: New test.
With the patch I've posted today to fix up declare variant LTO handling,
Tobias reported the patch still doesn't work, and there are two
reasons for that.
One is that when the base function is marked implicitly as declare target,
we don't mark also implicitly the variants. I'll need to ask on omp-lang
about details for that, but generally the compiler should do it some way.
The other one is that the way base_delay is written, it will always
call the usleep function, which is undesirable for nvptx. While the
compiler will replace all direct calls to base_delay to nvptx_delay,
the base_delay definition which calls usleep stays.
2020-10-28 Jakub Jelinek <jakub@redhat.com>
Tom de Vries <tdevries@suse.de>
PR testsuite/81690
* testsuite/libgomp.c/usleep.h: New file.
* testsuite/libgomp.c/target-32.c: Include usleep.h.
(main): Use tgt_usleep instead of usleep.
* testsuite/libgomp.c/thread-limit-2.c: Include usleep.h.
(main): Use tgt_usleep instead of usleep.
> I've tried to add the saving/restoring next to ipa refs saving/restoring, as
> the declare variant alt stuff is kind of extension of those, unfortunately
> following doesn't compile, because I need to also write or read a tree there
> (ctx is a portion of DECL_ATTRIBUTES of the base function), but the ipa refs
> write/read back functions don't have arguments that can be used for that.
This patch adds the streaming out and in of those omp_declare_variant_alt
hash table on the side data for the declare_variant_alt cgraph_nodes and
treats for LTO purposes the declare_variant_alt nodes (which have no body)
as if they contained a body that calls all the possible variants.
After IPA all the calls to these magic declare_variant_alt calls are
replaced with call to one of the variant depending on which one has the
highest score in the context.
2020-10-28 Jakub Jelinek <jakub@redhat.com>
PR lto/96680
gcc/
* lto-streamer.h (omp_lto_output_declare_variant_alt,
omp_lto_input_declare_variant_alt): Declare variant.
* symtab.c (symtab_node::get_partitioning_class): Return
SYMBOL_DUPLICATE for declare_variant_alt nodes.
* passes.c (ipa_write_summaries): Add declare_variant_alt to
partition.
* lto-cgraph.c (output_refs): Call omp_lto_output_declare_variant_alt
on declare_variant_alt nodes.
(input_refs): Call omp_lto_input_declare_variant_alt on
declare_variant_alt nodes.
* lto-streamer-out.c (output_function): Don't call
collect_block_tree_leafs if DECL_INITIAL is error_mark_node.
(lto_output): Call output_function even for declare_variant_alt
nodes.
* omp-general.c (omp_lto_output_declare_variant_alt,
omp_lto_input_declare_variant_alt): New functions.
gcc/lto/
* lto-common.c (lto_fixup_prevailing_decls): Don't use
LTO_NO_PREVAIL on TREE_LIST's TREE_PURPOSE.
* lto-partition.c (lto_balanced_map): Treat declare_variant_alt
nodes like definitions.
libgomp/
* testsuite/libgomp.c/declare-variant-1.c: New test.
> Therefore, I think until omp_get_initial_device () value is changed, we
The following so far untested patch implements that change.
OpenMP 4.5 said for omp_get_initial_device:
The value of the device number is implementation defined. If it is between 0 and one less than
omp_get_num_devices() then it is valid for use with all device constructs and routines; if it is
outside that range, then it is only valid for use with the device memory routines and not in the
device clause.
and OpenMP 5.0 similarly, but OpenMP 5.1 says:
The value of the device number is the value returned by the omp_get_num_devices routine.
As the new value is compatible with what has been required earlier, I think
we can change it already now.
2020-10-22 Jakub Jelinek <jakub@redhat.com>
* icv.c (omp_get_initial_device): Remove including corresponding
ialias.
* icv-device.c (omp_get_initial_device): New function. Return
gomp_get_num_devices (). Add ialias.
* target.c (resolve_device): Don't fail with
OMP_TARGET_OFFLOAD=mandatory if device_id is equal to
gomp_get_num_devices ().
(omp_target_alloc, omp_target_free, omp_target_is_present,
omp_target_memcpy, omp_target_memcpy_rect, omp_target_associate_ptr,
omp_target_disassociate_ptr, omp_pause_resource): Use
gomp_get_num_devices () instead of GOMP_DEVICE_HOST_FALLBACK on the
first use in the functions, in uses dominated by the
gomp_get_num_devices call use num_devices_openmp instead.
* libgomp.texi (omp_get_initial_device): Document.
* config/gcn/icv-device.c (omp_get_initial_device): New function.
Add ialias.
* config/nvptx/icv-device.c (omp_get_initial_device): Likewise.
* testsuite/libgomp.c/target-40.c: New test.
> the patch also breaks bootstrap on both i386-pc-solaris2.11 and
> sparc-sun-solaris2.11:
>
> /vol/gcc/src/hg/master/local/libgomp/env.c: In function 'initialize_env':
> /vol/gcc/src/hg/master/local/libgomp/env.c:414:16: error: 'new_offload' may be used uninitialized in this function [-Werror=maybe-uninitialized]
> 414 | *offload = new_offload;
> | ~~~~~~~~~^~~~~~~~~~~~~
> /vol/gcc/src/hg/master/local/libgomp/env.c:384:30: note: 'new_offload' was declared here
> 384 | enum gomp_target_offload_t new_offload;
> | ^~~~~~~~~~~
I can't reproduce that, but I fail to see why we need two separate
variables, one with actual value and one tracking if the value is valid.
So, I'm going with:
2020-10-21 Jakub Jelinek <jakub@redhat.com>
* env.c (parse_target_offload): Change new_offload var type to int,
preinitialize to -1, remove found var and test new_offload != -1
instead of found.
> On 10/20/20 2:11 PM, Tobias Burnus wrote:
>
> > Unfortunately, the committed patch
> > (r11-4121-g1bfc07d150790fae93184a79a7cce897655cb37b)
> > causes build errors.
> >
> > The error seems to be provoked by function cloning – as the code
> > itself looks fine:
> > ...
> > struct gomp_device_descr *devices_s
> > = malloc (num_devices * sizeof (struct gomp_device_descr));
> > ...
> > for (i = 0; i < num_devices; i++)
> > if (!(devices[i].capabilities & GOMP_OFFLOAD_CAP_OPENMP_400))
> > devices_s[num_devices_after_openmp++] = devices[i];
>
> gomp_target_init.part.0 ()
> {
> ...
> <bb 2>
> devices_s_1 = malloc (0);
> ...
> num_devices.16_67 = num_devices;
> ...
> if (num_devices.16_67 > 0)
> goto <bb 3>; [89.00%]
> else
> goto <bb 18>; [11.00%]
>
> Which seems to have an ordering problem.
This patch fixes the warning that breaks the bootstrap.
2020-10-20 Jakub Jelinek <jakub@redhat.com>
* target.c (gomp_target_init): Inside of the function, use automatic
variables corresponding to num_devices, num_devices_openmp and devices
global variables and update the globals only at the end of the
function.
This implements support for the OMP_TARGET_OFFLOAD environment variable
introduced in the OpenMP 5.0 standard, which controls how offloading
is handled. It may be set to MANDATORY (abort if offloading cannot be
performed), DISABLED (no offloading to devices) or DEFAULT (offload to
device if possible, fall back to host if not).
2020-10-20 Kwok Cheung Yeung <kcy@codesourcery.com>
libgomp/
* env.c (gomp_target_offload_var): New.
(parse_target_offload): New.
(handle_omp_display_env): Print value of OMP_TARGET_OFFLOAD.
(initialize_env): Parse OMP_TARGET_OFFLOAD.
* libgomp.h (gomp_target_offload_t): New.
(gomp_target_offload_var): New.
* libgomp.texi (OMP_TARGET_OFFLOAD): New section.
* target.c (resolve_device): Generate error if device not found and
offloading is mandatory.
(gomp_target_fallback): Generate error if offloading is mandatory.
(GOMP_target): Add argument in call to gomp_target_fallback.
(GOMP_target_ext): Likewise.
(gomp_target_data_fallback): Generate error if offloading is mandatory.
(GOMP_target_data): Add argument in call to gomp_target_data_fallback.
(GOMP_target_data_ext): Likewise.
(gomp_target_task_fn): Add argument in call to gomp_target_fallback.
(gomp_target_init): Return early if offloading is disabled.
This patch implements the omp_get_supported_active_levels runtime routine
from the OpenMP 5.0 specification, which returns the maximum number of
active nested parallel regions supported by this implementation. The
current maximum (set using the omp_set_max_active_levels routine or the
OMP_MAX_ACTIVE_LEVELS environment variable) cannot exceed this number.
2020-10-13 Kwok Cheung Yeung <kcy@codesourcery.com>
libgomp/
* env.c (gomp_max_active_levels_var): Initialize to
gomp_supported_active_levels.
(initialize_env): Limit gomp_max_active_levels_var to be at most
equal to gomp_supported_active_levels.
* fortran.c (omp_get_supported_active_levels): Add ialias_redirect.
(omp_get_supported_active_levels_): New.
* icv.c (omp_set_max_active_levels): Limit gomp_max_active_levels_var
to at most equal to gomp_supported_active_levels.
(omp_get_supported_active_levels): New.
* libgomp.h (gomp_supported_active_levels): New.
* libgomp.map (OMP_5.0.1): Add omp_get_supported_active_levels and
omp_get_supported_active_levels_.
* libgomp.texi (omp_get_supported_active_levels): New.
(omp_set_max_active_levels): Update. Add reference to
omp_get_supported_active_levels.
* omp.h.in (omp_get_supported_active_levels): New.
* omp_lib.f90.in (omp_get_supported_active_levels): New.
* omp_lib.h.in (omp_get_supported_active_levels): New.
* testsuite/libgomp.c/lib-2.c (main): Check omp_get_max_active_levels
against omp_get_supported_active_levels.
* testsuite/libgomp.fortran/lib4.f90 (lib4): Likewise.
AIX caches shared objects in archives with read-other permission.
libgomp and libatomic might be in use during the build or testing, which
may cause archiver operations on them to fail. This patch adjusts the
Makefile fragments to delete the library archives before creating fresh
archives containing both the 32 bit and 64 bit shared objects.
libatomic/ChangeLog:
2020-10-11 Clement Chigot <clement.chigot@atos.net>
* config/t-aix: Delete and recreate libatomic before creating
FAT library.
libgomp/ChangeLog:
2020-10-11 Clement Chigot <clement.chigot@atos.net>
* config/t-aix: Delete and recreate libgomp before creating
FAT library.
Using this patch, when using GOMP_DEBUG=1 and launching a kernel in
GOMP_OFFLOAD_run (used by the omp implementation), we see the kernel launch
dimensions:
...
GOMP_OFFLOAD_run: kernel main$_omp_fn$0: \
launch [(teams: 1), 1, 1] [(lanes: 32), (threads: 1), 1]
...
Build on x86_64-linux with nvptx accelerator, tested libgomp.
libgomp/ChangeLog:
2020-10-08 Tom de Vries <tdevries@suse.de>
PR libgomp/81802
* plugin/plugin-nvptx.c (GOMP_OFFLOAD_run): Report launch
dimensions.
Consider test-case test.c, with VLA A:
...
int main (void) {
int N = 1000;
int A[N];
#pragma acc declare copy(A)
return 0;
}
...
compiled using:
...
$ gcc test.c -fopenacc -S -fdump-tree-all
...
At original, we have:
...
#pragma acc declare map(tofrom:A);
...
but at gimple, we have a map (to:A.1), but not a map (from:A.1):
...
int[0:D.2074] * A.1;
{
int A[0:D.2074] [value-expr: *A.1];
saved_stack.2 = __builtin_stack_save ();
try
{
A.1 = __builtin_alloca_with_align (D.2078, 32);
#pragma omp target oacc_declare map(to:(*A.1) [len: D.2076])
}
finally
{
__builtin_stack_restore (saved_stack.2);
}
}
...
This is caused by the following incompatibility. When storing the desired
from clause in oacc_declare_returns, we use 'A.1' as the key:
...
10898 oacc_declare_returns->put (decl, c);
(gdb) call debug_generic_expr (decl)
A.1
(gdb) call debug_generic_expr (c)
map(from:(*A.1))
...
but when looking it up, we use 'A' as the key:
...
(gdb)
1471 tree *c = oacc_declare_returns->get (t);
(gdb) call debug_generic_expr (t)
A
...
Fix this by extracing the 'A.1' lookup key from 'A' using the decl-expr.
In addition, unshare the looked up value, to fix avoid running into
an "incorrect sharing of tree nodes" error.
Using these two fixes, we get our desired:
...
finally
{
+ #pragma omp target oacc_declare map(from:(*A.1))
__builtin_stack_restore (saved_stack.2);
}
...
Build on x86_64-linux with nvptx accelerator, tested libgomp.
gcc/ChangeLog:
2020-10-06 Tom de Vries <tdevries@suse.de>
PR middle-end/90861
* gimplify.c (gimplify_bind_expr): Handle lookup in
oacc_declare_returns using key with decl-expr.
libgomp/ChangeLog:
2020-10-06 Tom de Vries <tdevries@suse.de>
PR middle-end/90861
* testsuite/libgomp.oacc-c-c++-common/declare-vla.c: Remove xfail.
When running the libgomp testsuite on x86_64-linux with nvptx accelerator on
the test-case included in this patch, we run into:
...
FAIL: libgomp.fortran/pr95654.f90 -O3 -fomit-frame-pointer -funroll-loops \
-fpeel-loops -ftracer -finline-functions execution test
...
The test-case is a minimal version of this FAIL:
...
FAIL: libgomp.fortran/pr66199-5.f90 -O3 -fomit-frame-pointer -funroll-loops \
-fpeel-loops -ftracer -finline-functions execution test
...
but that one has stopped failing at commit c2ebf4f10d "openmp: Add support
for non-rect simd and improve collapsed simd support".
The problem is that ftracer duplicates a block containing GOMP_SIMT_VOTE_ANY.
That is, before ftracer we have (dropping the GOMP_SIMT_ prefix):
...
bb4(ENTER_ALLOC)
*----------+
| \
| \
| v
| *
v bb8
*<------------*
bb5(VOTE_ANY)
*-------------+
| |
| |
| |
| |
| v
| *
v bb7(XCHG_IDX)
*<------------*
bb6(EXIT)
...
The XCHG_IDX internal-fn does inter-SIMT-lane communication, which for nvptx
maps onto shfl, an operator which has the requirement that the warp executing
the operator is convergent. The warp diverges at bb4, and
reconverges at bb5, and does not diverge by going to bb7, so the shfl is
indeed executed by a convergent warp.
After ftracer, we have:
...
bb4(ENTER_ALLOC)
*----------+
| \
| \
| \
| \
v v
* *
bb5(VOTE_ANY) bb8(VOTE_ANY)
* *
|\ /|
| \ +--------+ |
| \/ |
| /\ |
| / +----------v
|/ *
v bb7(XCHG_IDX)
*<--------------*
bb6(EXIT)
...
The warp diverges again at bb5, but does not reconverge again before bb6, so
the shfl is executed by a divergent warp, which causes the FAIL.
Fix this by making ftracer ignore blocks containing ENTER_ALLOC, VOTE_ANY and
EXIT, effectively treating the SIMT region conservatively.
An argument can be made that the test needs to be added in a more
generic place, like gimple_can_duplicate_bb_p or some such, and that ftracer
then needs to use the generic test. But that's a discussion with a much
broader scope, so I'm leaving that for another patch.
Bootstrapped and reg-tested on x86_64-linux.
Build on x86_64-linux with nvptx accelerator, tested with libgomp.
gcc/ChangeLog:
PR fortran/95654
* tracer.c (ignore_bb_p): Ignore GOMP_SIMT_ENTER_ALLOC,
GOMP_SIMT_VOTE_ANY and GOMP_SIMT_EXIT.
libgomp/ChangeLog:
2020-10-05 Tom de Vries <tdevries@suse.de>
PR fortran/95654
* testsuite/libgomp.fortran/pr95654.f90: New test.
Accelerators with fixed thread-counts will break if nested teams are expected
to have multiple threads each.
libgomp/ChangeLog:
2020-09-29 Andrew Stubbs <ams@codesourcery.com>
* parallel.c (gomp_resolve_num_threads): Ignore nest_var on nvptx
and amdgcn targets.
Both GCN and NVPTX allow nested parallel regions, but the barrier
implementation did not allow the nested teams to run independently of each
other (due to hardware limitations). This patch fixes that, under the
assumption that each thread will create a new subteam of one thread, by
simply not using barriers when there's no other thread to synchronise.
libgomp/ChangeLog:
* config/gcn/bar.c (gomp_barrier_wait_end): Skip the barrier if the
total number of threads is one.
(gomp_team_barrier_wake): Likewise.
(gomp_team_barrier_wait_end): Likewise.
(gomp_team_barrier_wait_cancel_end): Likewise.
* config/nvptx/bar.c (gomp_barrier_wait_end): Likewise.
(gomp_team_barrier_wake): Likewise.
(gomp_team_barrier_wait_end): Likewise.
(gomp_team_barrier_wait_cancel_end): Likewise.
* testsuite/libgomp.c-c++-common/nested-parallel-unbalanced.c: New test.
AIX FAT libraries should be built with the version of AR chosen by configure.
The GNU Make $(AR) variable includes the AIX -X32_64 option needed
by the default Makefile rules to accept both 32 bit and 64 bit object files.
The -X32_64 option conflicts with ar archiving objects of the same name
used to build FAT libraries.
This patch changes the Makefile fragments for AIX FAT libraries to use $(AR),
but strips the -X32_64 option from the Make variable.
libgcc/ChangeLog:
2020-09-27 Clement Chigot <clement.chigot@atos.net>
* config/rs6000/t-slibgcc-aix: Use $(AR) without -X32_64.
libatomic/ChangeLog:
2020-09-27 Clement Chigot <clement.chigot@atos.net>
* config/t-aix: Use $(AR) without -X32_64.
libgomp/ChangeLog:
2020-09-27 Clement Chigot <clement.chigot@atos.net>
* config/t-aix: Use $(AR) without -X32_64.
libstdc++-v3/ChangeLog:
2020-09-27 Clement Chigot <clement.chigot@atos.net>
* config/os/aix/t-aix: Use $(AR) without -X32_64.
libgfortran/ChangeLog:
2020-09-27 Clement Chigot <clement.chigot@atos.net>
* config/t-aix: Use $(AR) without -X32_64.
The following change adds support for non-rectangular simd loops.
While working on that, I've noticed we actually don't vectorize collapsed
simd loops at all, because the code that I thought would be vectorizable
actually is not vectorized. While in theory for the constant lower/upper
bounds and constant step of all but the outermost loop we could in theory
vectorize by computing the seprate iterators using vectorized division
and modulo for each of them from the single iterator that increments
by 1 from 0 to total iteration count in the loop nest, I think that would
be fairly expensive and the chances of the loop body being vectorizable
would be low e.g. because of array indices unlikely to be linear and would
need scatters/gathers.
This patch changes the generated code to vectorize only the innermost
loop which has higher chance of being vectorized. Below is the list of
tests and function names in which the patch resulted in vectorizing something
that hasn't been vectorized before (ok, the first line is a new test).
I've also found that the vectorizer will not vectorize loops with non-constant
steps, I plan to do something about those incrementally on the omp-expand.c
side (basically, compute number of iterations before the loop and use a 0 to
number_of_iterations step 1 IV as the main one).
I have problem with the composite simd vectorization though.
The point is that each thread (or task etc.) is given only a range of
consecutive iterations, so somewhere earlier it computes total number of iterations
and splits the work between the workers and then the intent is to try to vectorize it.
So, each thread is then given a begin ... end-1 range that it would handle.
This means that from the single begin value I need to compute the individual iteration
vars I should start at and then goto into the loop nest to begin iterating there
(and actually compute how many iterations the innermost loop should do each time
so that it stops before end).
Very roughly the IL I emit is something like:
int t[100][100][100];
void
foo (int a, int b, int c, int d, int e, int f, int g, int h, int u, int v, int w, int x)
{
int i, j, k;
int cnt;
if (x)
{
i = u; j = v; k = w; goto doit;
}
for (i = a; i < b; i += c)
for (j = d; j < e; j += f)
{
k = g;
doit:
for (; k < h; k++)
t[i][j][k] += i + j + k;
}
}
Unfortunately, some pass then turns the innermost loop to have more than 2 basic blocks
and it isn't vectorized because of that.
Also, I have disabled (for now) SIMTization of collapsed simd loops, because for SIMT
it would be using a single thread anyway and I didn't want to bother with checking
SIMT on all places I've been changing. If SIMT support is added for some or all
collapsed loops, that omp-low.c change needs to be reverted.
Here is that list of what hasn't been vectorized before and is now:
gcc/testsuite/gcc.dg/vect/vect-simd-17.c doit
gcc/testsuite/gfortran.dg/gomp/openmp-simd-6.f90 bar
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-10.c f28_taskloop_simd_normal._omp_fn.0
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-10.c _Z24f28_taskloop_simd_normalv._omp_fn.0
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-11.c f25_t_simd_normal._omp_fn.0
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-11.c f26_t_simd_normal._omp_fn.0
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-11.c f27_t_simd_normal._omp_fn.0
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-11.c f28_tpf_simd_guided32._omp_fn.1
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-11.c f28_tpf_simd_runtime._omp_fn.1
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-11.c _Z17f25_t_simd_normaliiiiiii._omp_fn.0
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-11.c _Z17f26_t_simd_normaliiiixxi._omp_fn.0
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-11.c _Z17f27_t_simd_normalv._omp_fn.0
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-11.c _Z20f28_tpf_simd_runtimev._omp_fn.1
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-11.c _Z21f28_tpf_simd_guided32v._omp_fn.1
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-2.c f7_simd_normal
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-2.c f7_simd_normal
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-2.c f8_f_simd_guided32
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-2.c f8_f_simd_guided32
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-2.c f8_f_simd_runtime
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-2.c f8_f_simd_runtime
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-2.c f8_pf_simd_guided32._omp_fn.0
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-2.c f8_pf_simd_runtime._omp_fn.0
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-2.c _Z18f8_pf_simd_runtimev._omp_fn.0
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-2.c _Z19f8_pf_simd_guided32v._omp_fn.0
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-4.c f8_taskloop_simd_normal._omp_fn.0
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-4.c _Z23f8_taskloop_simd_normalv._omp_fn.0
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-5.c f7_t_simd_normal._omp_fn.0
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-5.c f8_tpf_simd_guided32._omp_fn.1
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-5.c f8_tpf_simd_runtime._omp_fn.1
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-5.c _Z16f7_t_simd_normalv._omp_fn.0
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-5.c _Z19f8_tpf_simd_runtimev._omp_fn.1
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-5.c _Z20f8_tpf_simd_guided32v._omp_fn.1
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-8.c f25_simd_normal
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-8.c f25_simd_normal
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-8.c f26_simd_normal
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-8.c f26_simd_normal
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-8.c f27_simd_normal
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-8.c f27_simd_normal
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-8.c f28_f_simd_guided32
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-8.c f28_f_simd_guided32
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-8.c f28_f_simd_runtime
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-8.c f28_f_simd_runtime
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-8.c f28_pf_simd_guided32._omp_fn.0
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/for-8.c f28_pf_simd_runtime._omp_fn.0
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-8.c _Z19f28_pf_simd_runtimev._omp_fn.0
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/for-8.c _Z20f28_pf_simd_guided32v._omp_fn.0
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/master-combined-1.c main._omp_fn.9
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/master-combined-1.c main._omp_fn.9
libgomp/testsuite/libgomp.c++/../libgomp.c-c++-common/simd-1.c f2
libgomp/testsuite/libgomp.c/../libgomp.c-c++-common/simd-1.c f2
libgomp/testsuite/libgomp.c/pr70680-2.c f1._omp_fn.0
libgomp/testsuite/libgomp.c/pr70680-2.c f2._omp_fn.0
libgomp/testsuite/libgomp.c/pr70680-2.c f3._omp_fn.0
libgomp/testsuite/libgomp.c/pr70680-2.c f4._omp_fn.0
libgomp/testsuite/libgomp.c/simd-8.c foo
libgomp/testsuite/libgomp.c/simd-9.c bar
libgomp/testsuite/libgomp.c/simd-9.c foo
2020-09-25 Jakub Jelinek <jakub@redhat.com>
gcc/
* omp-low.c (scan_omp_1_stmt): Don't call scan_omp_simd for
collapse > 1 loops as simt doesn't support collapsed loops yet.
* omp-expand.c (expand_omp_for_init_counts, expand_omp_for_init_vars):
Small tweaks to function comment.
(expand_omp_simd): Rewritten collapse > 1 support to only attempt
to vectorize the innermost loop and emit set of outer loops around it.
For non-composite simd with collapse > 1 without broken loop don't
even try to compute number of iterations first. Add support for
non-rectangular simd loops.
(expand_omp_for): Don't sorry_at on non-rectangular simd loops.
gcc/testsuite/
* gcc.dg/vect/vect-simd-17.c: New test.
libgomp/
* testsuite/libgomp.c/loop-25.c: New test.
By running libgomp test-case libgomp.c/target-28.c with GOMP_NVPTX_PTXRW=w
(using a maintenance patch that adds support for this env var), we dump the
ptx in target-28.exe to file. By editing one ptx file to rename
gomp_nvptx_main to gomp_nvptx_main2 in both declaration and call, and
running with GOMP_NVPTX_PTXRW=r, we trigger a link error:
...
$ GOMP_NVPTX_PTXRW=r ./target-28.exe
libgomp: cuLinkComplete error: unknown error
...
The error is somewhat uninformative.
Fix this by dumping the error log returned by the failing cuda call, such
that we have instead:
...
$ GOMP_NVPTX_PTXRW=r ./target-28.exe
libgomp: Link error log error : \
Undefined reference to 'gomp_nvptx_main2' in ''
libgomp: cuLinkComplete error: unknown error
...
Build on x86_64 with nvptx accelerator, tested libgomp.
libgomp/ChangeLog:
* plugin/plugin-nvptx.c (link_ptx): Print elog if cuLinkComplete call
fails.
This corrects the earlier problems with removing the template header
from local omp reductions. And it uncovered a latent bug. When we
tsubst such a decl, we immediately tsubst its body.
cp_check_omp_declare_reduction gets a success return value to gate
that instantiation.
udr-2.C got a further error, as the omp checking machinery doesn't
appear to turn the reduction into an error mark when failing. I
didn't dig into that further. udr-3.C appears to have been invalid
and accidentally worked.
gcc/cp/
* cp-tree.h (cp_check_omp_declare_reduction): Return bool.
* semantics.c (cp_check_omp_declare_reduction): Return true on for
success.
* pt.c (push_template_decl_real): OMP reductions do not get a
template header.
(tsubst_function_decl): Remove special casing for local decl omp
reductions.
(tsubst_expr): Call instantiate_body for a local omp reduction.
(instantiate_body): Add nested_p parm, and deal with such
instantiations.
(instantiate_decl): Reject FUNCTION_SCOPE entities, adjust
instantiate_body call.
gcc/testsuite/
* g++.dg/gomp/udr-2.C: Add additional expected error.
libgomp/
* testsuite/libgomp.c++/udr-3.C: Add missing ctor.
As reported here
( https://gcc.gnu.org/pipermail/gcc-patches/2020-September/553070.html ),
when running test-case libgomp.c-c++-common/reduction-16.c for powerpc host
with nvptx accelerator, we run into:
...
unresolved symbol __sync_val_compare_and_swap_16
...
I can reproduce the problem on x86_64 with a trigger patch that:
- initializes ix86_isa_flags2 to TARGET_ISA2_CX16
- enables define_expand "atomic_load<mode>" in gcc/config/i386/sync.md
for TImode
The problem is that omp-expand.c generates atomic builtin calls based on
checks whether those are supported on the host, which forces the target to
support these, even though those checks fail for the accelerator target.
Fix this by:
- adding a __sync_val_compare_and_swap_16 in libgomp for nvptx,
which falls back onto libatomic's __atomic_compare_and_swap_16
- adding -foffload=-latomic in the test-case
Tested libgomp on x86_64-linux with nvptx accelerator.
Tested libgomp with trigger patch on x86_64-linux with nvptx accelerator.
libgomp/ChangeLog:
* config/nvptx/atomic.c: New file. Add
__sync_val_compare_and_swap_16.
* testsuite/libgomp.c-c++-common/reduction-16.c: Add -latomic for
target offload_target_nvptx.
The test case was written with assumptions about loop iteration ordering
that are not guaranteed by OpenACC and do not apply on all targets,
in particular AMD GCN. This patch removes those assumptions.
2020-09-08 Julian Brown <julian@codesourcery.com>
libgomp/
* testsuite/libgomp.oacc-c-c++-common/atomic_capture-2.c: Remove
iteration-ordering assumptions.
The gangs participating in a gang-partitioned loop are not all guaranteed
to complete before some given gang continues to execute beyond that loop.
This means that two existing test cases contain a race condition,
because a loop that may be gang-partitioned is followed immediately by
another loop. The fix is to place the loops in separate parallel regions.
2020-09-08 Julian Brown <julian@codesourcery.com>
libgomp/
* testsuite/libgomp.oacc-fortran/collapse-1.f90: Fix race condition.
* testsuite/libgomp.oacc-fortran/collapse-2.f90: Likewise.
Change test for CUDA callback context in nvptx_free() from using
GOMP_PLUGIN_acc_thread () into checking for CUDA_ERROR_NOT_PERMITTED,
for the former only works for OpenACC, but not OpenMP offloading.
2020-08-20 Chung-Lin Tang <cltang@codesourcery.com>
libgomp/
* plugin/plugin-nvptx.c (nvptx_free):
Change "GOMP_PLUGIN_acc_thread () == NULL" test into check of
CUDA_ERROR_NOT_PERMITTED status for cuMemGetAddressRange. Adjust
comments.
This adds support for __sync_val_compare_and_swap and
__sync_bool_compare_and_swap for 1-byte and 2-byte long
values, which are not natively supported on nvptx.
Build and reg-tested on nvptx.
Build and reg-tested libgomp on x86_64 with nvptx accelerator.
2020-07-16 Kwok Cheung Yeung <kcy@codesourcery.com>
libgcc/
* config/nvptx/atomic.c: New.
* config/nvptx/t-nvptx (LIB2ADD): Add atomic.c.
gcc/testsuite/
* gcc.target/nvptx/ia64-sync-5.c: New.
libgomp/
* testsuite/libgomp.c-c++-common/reduction-16.c: New.
2020-08-13 Jakub Jelinek <jakub@redhat.com>
* gimplify.c (gimplify_omp_taskloop_expr): New function.
(gimplify_omp_for): Use it. For OMP_FOR_NON_RECTANGULAR
loops adjust in outer taskloop the var-outer decls.
* omp-expand.c (expand_omp_taskloop_for_inner): Handle non-rectangular
loops.
(expand_omp_for): Don't reject non-rectangular taskloop.
* omp-general.c (omp_extract_for_data): Don't assert that
non-rectangular loops have static schedule, instead treat loop->m1
or loop->m2 as if loop->n1 or loop->n2 is non-constant.
* testsuite/libgomp.c/loop-22.c (main): Add some further tests.
* testsuite/libgomp.c/loop-23.c (main): Likewise.
* testsuite/libgomp.c/loop-24.c: New test.
If the walk_body on the various sequences of reduction, lastprivate and/or linear
clauses needs to create a temporary variable, we should declare that variable
in that sequence rather than outside, where it would need to be privatized inside of
the construct.
2020-08-08 Jakub Jelinek <jakub@redhat.com>
PR fortran/93553
* tree-nested.c (convert_nonlocal_omp_clauses): For
OMP_CLAUSE_REDUCTION, OMP_CLAUSE_LASTPRIVATE and OMP_CLAUSE_LINEAR
save info->new_local_var_chain around walks of the clause gimple
sequences and declare_vars if needed into the sequence.
2020-08-08 Tobias Burnus <tobias@codesourcery.com>
PR fortran/93553
* testsuite/libgomp.fortran/pr93553.f90: New test.
The number of loops computation and logical iteration -> actual iterator values
computations can now be done separately even on composite constructs (though
for triangular loops it would still be more efficient to propagate a few values
through, will handle that incrementally).
simd and taskloop are still unhandled.
2020-08-05 Jakub Jelinek <jakub@redhat.com>
* omp-expand.c (expand_omp_for): Don't disallow combined non-rectangular
loops.
* testsuite/libgomp.c/loop-22.c: New test.
* testsuite/libgomp.c/loop-23.c: New test.
As the new testcase shows, we weren't actually performing reductions on
host teams construct. And fixing that revealed a flaw in the for-14.c testcase.
The problem is that the tests perform also initialization and checking around the
calls to the functions with the OpenMP constructs. In that testcase, all the
tests have been spawned from a teams construct but only the tested loops were
distribute, which means the initialization and checking has been performed
redundantly and racily in each team. Fixed by performing the initialization
and checking outside of host teams and only do the calls to functions with
the tested constructs inside of host teams.
2020-08-05 Jakub Jelinek <jakub@redhat.com>
PR middle-end/96459
* omp-low.c (lower_omp_taskreg): Call lower_reduction_clauses even in
for host teams.
* testsuite/libgomp.c/teams-3.c: New test.
* testsuite/libgomp.c-c++-common/for-2.h (OMPTEAMS): Define to nothing
if not defined yet.
(N(test)): Use it before all N(f*) calls.
* testsuite/libgomp.c-c++-common/for-14.c (DO_PRAGMA, OMPTEAMS): Define.
(main): Don't call all test_* functions from within
#pragma omp teams reduction(|:err), call them directly.
Chung-Lin Tang
GCC Administrator
Tobias Burnus
Andrew Stubbs
Thomas Schwinge
Jakub Jelinek
H.J. Lu
Richard Biener
Julian Brown
Kwok Cheung Yeung
Martin Liska
Tom de Vries
Roman Zhuykov
Hafiz Abid Qadeer
Iain Sandoe
Uros Bizjak
Andreas Schwab
Sebastian Huber
Samuel Thibault
Gerald Pfeifer
John David Anglin
Gergö Barany
Clément Chigot
Nathan Sidwell