The patch that introduced the weak_undefined effective-target keyword
and corresponding dg-add-options support
commit 378ec7b87a
Author: Alexandre Oliva <oliva@adacore.com>
Date: Thu Mar 23 00:45:05 2023 -0300
[testsuite] test for weak_undefined support and add options
badly broke the affected tests on macOS like so:
ERROR: gcc.dg/addr_equal-1.c: unknown dg option: 89 for " dg-add-options 5 weak_undefined "
ERROR: gcc.dg/addr_equal-1.c: unknown dg option: 89 for " dg-add-options 5 weak_undefined "
add_options_for_weak_undefined tries to call an non-existant proc "89".
Even after fixing this by escaping the brackets, two tests still failed to
link since they lacked the corresponding calls do dg-add-options
weak_undefined.
Tested on x86_64-apple-darwin20.6.0 and i386-pc-solaris2.11.
2023-03-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
gcc/testsuite:
* lib/target-supports.exp (add_options_for_weak_undefined): Escape
brackets.
* gcc.dg/visibility-22.c: Add weak_undefined options.
libgomp:
* testsuite/libgomp.oacc-c-c++-common/routine-nohost-2.c: Add
weak_undefined options.
Thereby considerably simplify the device plugins' 'GOMP_OFFLOAD_openacc_exec',
'GOMP_OFFLOAD_openacc_async_exec' functions: in terms of lines of code, but in
particular conceptually: no more device memory allocation, host to device data
copying, device memory deallocation -- 'GOMP_MAP_VARS_TARGET' does all that for
us.
This depends on commit 2b2340e236
"Allow libgomp 'cbuf' buffering with OpenACC 'async' for 'ephemeral' data",
where I said that "a use will emerge later", which is this one here.
PR libgomp/90596
libgomp/
* target.c (gomp_map_vars_internal): Allow for
'param_kind == GOMP_MAP_VARS_OPENACC | GOMP_MAP_VARS_TARGET'.
* oacc-parallel.c (GOACC_parallel_keyed): Pass
'GOMP_MAP_VARS_TARGET' to 'goacc_map_vars'.
* plugin/plugin-gcn.c (alloc_by_agent, gcn_exec)
(GOMP_OFFLOAD_openacc_exec, GOMP_OFFLOAD_openacc_async_exec):
Adjust, simplify.
(gomp_offload_free): Remove.
* plugin/plugin-nvptx.c (nvptx_exec, GOMP_OFFLOAD_openacc_exec)
(GOMP_OFFLOAD_openacc_async_exec): Adjust, simplify.
(cuda_free_argmem): Remove.
* testsuite/libgomp.oacc-c-c++-common/acc_prof-parallel-1.c:
Adjust.
This does *allow*, but under no circumstances is this currently going to be
used: all potentially applicable data is non-'ephemeral', and thus not
considered for 'gomp_coalesce_buf_add' for OpenACC 'async'. (But a use will
emerge later.)
Follow-up to commit r12-2530-gd88a6951586c7229b25708f4486eaaf4bf4b5bbe
"Don't use libgomp 'cbuf' buffering with OpenACC 'async'", addressing this
TODO comment:
TODO ... but we could allow CBUF usage for EPHEMERAL data? (Open question:
is it more performant to use libgomp CBUF buffering or individual device
asyncronous copying?)
Ephemeral data is small, and therefore individual device asyncronous copying
does seem dubious -- in particular given that for all those, we'd individually
have to allocate and queue for deallocation a temporary buffer to capture the
ephemeral data. Instead, just let the 'cbuf' *be* the temporary buffer.
libgomp/
* target.c (gomp_copy_host2dev, gomp_map_vars_internal): Allow
libgomp 'cbuf' buffering with OpenACC 'async' for 'ephemeral'
data.
For 'OFFSET_INLINED', 'gomp_map_val' does the right thing, and we may then
simplify the device plugins accordingly.
This is a follow-up to
Subversion r279551 (Git commit a6163563f2)
"Add OpenACC 2.6's no_create",
Subversion r279622 (Git commit 5bcd470bf0)
"Use gomp_map_val for OpenACC host-to-device address translation".
libgomp/
* target.c (gomp_map_vars_internal): Use 'OFFSET_INLINED' for
'GOMP_MAP_IF_PRESENT'.
* plugin/plugin-gcn.c (gcn_exec, GOMP_OFFLOAD_openacc_exec)
(GOMP_OFFLOAD_openacc_async_exec): Adjust.
* plugin/plugin-nvptx.c (nvptx_exec, GOMP_OFFLOAD_openacc_exec)
(GOMP_OFFLOAD_openacc_async_exec): Likewise.
* testsuite/libgomp.oacc-c-c++-common/no_create-1.c: Add 'async'
testing.
* testsuite/libgomp.oacc-c-c++-common/no_create-2.c: Likewise.
We're not taking such a shortcut anywhere else, and (with future changes) it
has potential to confuse things if synchronization in a libgomp plugin happens
to have side effects even if an async queue currently is empty.
libgomp/
* oacc-async.c (goacc_wait): Remove 'acc_async_test' -> skip
shortcut.
... that I almost broke with later implementation changes.
libgomp/
* testsuite/libgomp.oacc-c-c++-common/data-3.c: Document/verify
another aspect of OpenACC 'async' semantics.
When OMP_WAIT_POLICY is not specified, current implementation will cause
icv flag GOMP_ICV_WAIT_POLICY unset, so global variable wait_policy
will remain its uninitialized value. Initialize it to -1 to make
GOMP_SPINCOUNT behavior consistent with its description.
libgomp/ChangeLog:
PR libgomp/109062
* env.c (wait_policy): Initialize to -1.
(initialize_icvs): Initialize icvs->wait_policy to -1.
* testsuite/libgomp.c-c++-common/pr109062.c: New test.
Calls to vectorized versions of routines in the math library will now
be inserted when vectorizing code containing supported math functions.
2023-03-02 Kwok Cheung Yeung <kcy@codesourcery.com>
Paul-Antoine Arras <pa@codesourcery.com>
gcc/
* builtins.cc (mathfn_built_in_explicit): New.
* config/gcn/gcn.cc: Include case-cfn-macros.h.
(mathfn_built_in_explicit): Add prototype.
(gcn_vectorize_builtin_vectorized_function): New.
(gcn_libc_has_function): New.
(TARGET_LIBC_HAS_FUNCTION): Define.
(TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Define.
gcc/testsuite/
* gcc.target/gcn/simd-math-1.c: New testcase.
* gcc.target/gcn/simd-math-2.c: New testcase.
libgomp/
* testsuite/libgomp.c/simd-math-1.c: New testcase.
For is_device_ptr, optional checks should only be done before calling
libgomp, afterwards they are NULL either because of absent or, by
chance, because it is unallocated or unassociated (for pointers/allocatables).
Additionally, it fixes an issue with explicit mapping for 'type(c_ptr)'.
PR middle-end/108546
gcc/fortran/ChangeLog:
* trans-openmp.cc (gfc_trans_omp_clauses): Fix mapping of
type(C_ptr) variables.
gcc/ChangeLog:
* omp-low.cc (lower_omp_target): Remove optional handling
on the receiver side, i.e. inside target (data), for
use_device_ptr.
libgomp/ChangeLog:
* testsuite/libgomp.fortran/is_device_ptr-3.f90: New test.
* testsuite/libgomp.fortran/use_device_ptr-optional-4.f90: New test.
I noticed that GCC/Rust recently lost all LTO variants in torture testing:
PASS: rust/compile/torture/all_doc_comment_line_blocks.rs -O0 (test for excess errors)
PASS: rust/compile/torture/all_doc_comment_line_blocks.rs -O1 (test for excess errors)
PASS: rust/compile/torture/all_doc_comment_line_blocks.rs -O2 (test for excess errors)
-PASS: rust/compile/torture/all_doc_comment_line_blocks.rs -O2 -flto -fno-use-linker-plugin -flto-partition=none (test for excess errors)
-PASS: rust/compile/torture/all_doc_comment_line_blocks.rs -O2 -flto -fuse-linker-plugin -fno-fat-lto-objects (test for excess errors)
PASS: rust/compile/torture/all_doc_comment_line_blocks.rs -O3 -g (test for excess errors)
PASS: rust/compile/torture/all_doc_comment_line_blocks.rs -Os (test for excess errors)
Etc.
The reason is that when probing for availability of LTO, we run into:
spawn [...]/build-gcc/gcc/testsuite/rust/../../gccrs -B[...]/build-gcc/gcc/testsuite/rust/../../ -fdiagnostics-plain-output -frust-incomplete-and-experimental-compiler-do-not-use -flto -c -o lto8274.o lto8274.c
cc1: warning: command-line option '-frust-incomplete-and-experimental-compiler-do-not-use' is valid for Rust but not for C
For GCC/Rust testing, this flag is (as of recently) defaulted in
'gcc/testsuite/lib/rust.exp:rust_init':
lappend ALWAYS_RUSTFLAGS "additional_flags=-frust-incomplete-and-experimental-compiler-do-not-use"
A few more "command-line option [...] is valid for [...] but not for [...]"
instances were found in the test suite logs, when more than one language is
involved.
With '-Wno-complain-wrong-lang' used in
'gcc/testsuite/lib/target-supports.exp:check_compile', we get back:
PASS: rust/compile/torture/all_doc_comment_line_blocks.rs -O0 (test for excess errors)
PASS: rust/compile/torture/all_doc_comment_line_blocks.rs -O1 (test for excess errors)
PASS: rust/compile/torture/all_doc_comment_line_blocks.rs -O2 (test for excess errors)
+PASS: rust/compile/torture/all_doc_comment_line_blocks.rs -O2 -flto -fno-use-linker-plugin -flto-partition=none (test for excess errors)
+PASS: rust/compile/torture/all_doc_comment_line_blocks.rs -O2 -flto -fuse-linker-plugin -fno-fat-lto-objects (test for excess errors)
PASS: rust/compile/torture/all_doc_comment_line_blocks.rs -O3 -g (test for excess errors)
PASS: rust/compile/torture/all_doc_comment_line_blocks.rs -Os (test for excess errors)
Etc., and in total:
=== rust Summary for unix ===
# of expected passes [-4990-]{+6718+}
# of expected failures [-39-]{+51+}
Anything that 'gcc/opts-global.cc:complain_wrong_lang' might do is cut
short by '-Wno-complain-wrong-lang', not just the one 'warning'
diagnostic. This corresponds to what already exists via
'lang_hooks.complain_wrong_lang_p'.
The 'gcc/opts-common.cc:prune_options' changes follow the same rationale
as PR67640 "driver passes -fdiagnostics-color= always last": we need to
process '-Wno-complain-wrong-lang' early, so that it properly affects
other options appearing before it on the command line.
gcc/
* common.opt (-Wcomplain-wrong-lang): New.
* doc/invoke.texi (-Wno-complain-wrong-lang): Document it.
* opts-common.cc (prune_options): Handle it.
* opts-global.cc (complain_wrong_lang): Use it.
gcc/testsuite/
* gcc.dg/Wcomplain-wrong-lang-1.c: New.
* gcc.dg/Wcomplain-wrong-lang-2.c: Likewise.
* gcc.dg/Wcomplain-wrong-lang-3.c: Likewise.
* gcc.dg/Wcomplain-wrong-lang-4.c: Likewise.
* gcc.dg/Wcomplain-wrong-lang-5.c: Likewise.
* lib/target-supports.exp (check_compile): Use
'-Wno-complain-wrong-lang'.
* g++.dg/abi/empty12.C: Likewise.
* g++.dg/abi/empty13.C: Likewise.
* g++.dg/abi/empty14.C: Likewise.
* g++.dg/abi/empty15.C: Likewise.
* g++.dg/abi/empty16.C: Likewise.
* g++.dg/abi/empty17.C: Likewise.
* g++.dg/abi/empty18.C: Likewise.
* g++.dg/abi/empty19.C: Likewise.
* g++.dg/abi/empty22.C: Likewise.
* g++.dg/abi/empty25.C: Likewise.
* g++.dg/abi/empty26.C: Likewise.
* gfortran.dg/bind-c-contiguous-1.f90: Likewise.
* gfortran.dg/bind-c-contiguous-4.f90: Likewise.
* gfortran.dg/bind-c-contiguous-5.f90: Likewise.
libgomp/
* testsuite/libgomp.fortran/alloc-10.f90: Use
'-Wno-complain-wrong-lang'.
* testsuite/libgomp.fortran/alloc-11.f90: Likewise.
* testsuite/libgomp.fortran/alloc-7.f90: Likewise.
* testsuite/libgomp.fortran/alloc-9.f90: Likewise.
* testsuite/libgomp.fortran/allocate-1.f90: Likewise.
* testsuite/libgomp.fortran/depend-4.f90: Likewise.
* testsuite/libgomp.fortran/depend-5.f90: Likewise.
* testsuite/libgomp.fortran/depend-6.f90: Likewise.
* testsuite/libgomp.fortran/depend-7.f90: Likewise.
* testsuite/libgomp.fortran/depend-inoutset-1.f90: Likewise.
* testsuite/libgomp.fortran/examples-4/declare_target-1.f90:
Likewise.
* testsuite/libgomp.fortran/examples-4/declare_target-2.f90:
Likewise.
* testsuite/libgomp.fortran/order-reproducible-1.f90: Likewise.
* testsuite/libgomp.fortran/order-reproducible-2.f90: Likewise.
* testsuite/libgomp.oacc-fortran/parallel-dims.f90: Likewise.
* testsuite/libgomp.fortran/task-detach-6.f90: Remove left-over
'dg-prune-output'.
I decided to check for repeated the the in libgomp and noticed
there are several occurrences of a typo theads rather than threads
in libgomp.texi.
2023-02-16 Jakub Jelinek <jakub@redhat.com>
* libgomp.texi: Fix typos - theads -> threads.
I saw
FAIL: libgomp.fortran/target-nowait-array-section.f90 -O execution test
in my last x86_64-linux bootstrap. From quick skimming, it might be just
unreliable test, which assumes that asynchronous execution wouldn't produce
ordered sequence, but can't it happen even with asynchronous execution?
That said, while skimming the test, I've noticed a comment typo and
this patch fixes that up.
2023-02-16 Jakub Jelinek <jakub@redhat.com>
* testsuite/libgomp.fortran/target-nowait-array-section.f90: Fix
comment typo and improve its wording.
libgomp/
* target.c (gomp_target_rev): Dereference ptr
to get device address.
* testsuite/libgomp.fortran/reverse-offload-5.f90: Add test
for unallocated allocatable.
As GOMP_MAP_ALWAYS_POINTER operates on the previous map item, ensure that
with 'target enter data' both are passed together to gomp_map_vars_internal.
libgomp/ChangeLog:
* target.c (gomp_map_vars_internal): Add 'i > 0' before doing a
kind check.
(GOMP_target_enter_exit_data): If the next map item is
GOMP_MAP_ALWAYS_POINTER map it together with the current item.
* testsuite/libgomp.fortran/target-enter-data-3.f90: New test.
This patch ensures that loop bounds depending on outer loop vars use the
proper TREE_VEC format. It additionally gives a sorry if such an outer
var has a non-one/non-minus-one increment as currently a count variable
is used in this case (see PR).
Finally, it avoids 'count' and just uses a local loop variable if the
step increment is +/-1.
PR fortran/107424
gcc/fortran/ChangeLog:
* trans-openmp.cc (struct dovar_init_d): Add 'sym' and
'non_unit_incr' members.
(gfc_nonrect_loop_expr): New.
(gfc_trans_omp_do): Call it; use normal loop bounds
for unit stride - and only create local loop var.
libgomp/ChangeLog:
* testsuite/libgomp.fortran/non-rectangular-loop-1.f90: New test.
* testsuite/libgomp.fortran/non-rectangular-loop-1a.f90: New test.
* testsuite/libgomp.fortran/non-rectangular-loop-2.f90: New test.
* testsuite/libgomp.fortran/non-rectangular-loop-3.f90: New test.
* testsuite/libgomp.fortran/non-rectangular-loop-4.f90: New test.
* testsuite/libgomp.fortran/non-rectangular-loop-5.f90: New test.
gcc/testsuite/ChangeLog:
* gfortran.dg/goacc/privatization-1-compute-loop.f90: Update dg-note.
* gfortran.dg/goacc/privatization-1-routine_gang-loop.f90: Likewise.
If there is nothing to map, skip the mapping and avoid attempting to
copy 0 bytes from addrs, sizes and kinds.
Additionally, it could happen that a non-allocated address was deallocated,
such as a pointer set, leading to a free for the actual data.
libgomp/
* target.c (gomp_target_rev): Handle mapnum == 0 and avoid
freeing not allocated memory.
* testsuite/libgomp.fortran/reverse-offload-6.f90: New test.
Switch from using stacks in the "private segment" to using a memory block
allocated on the host side. The primary reason is to permit the reverse
offload implementation to access values located on the device stack, but
there may also be performance benefits, especially with repeated kernel
invocations.
This implementation unifies the stacks with the "team arena" optimization
feature, and now allows both to have run-time configurable sizes.
A new ABI is needed, so all libraries must be rebuilt, and newlib must be
version 4.3.0.20230120 or newer.
gcc/ChangeLog:
* config/gcn/gcn-run.cc: Include libgomp-gcn.h.
(struct kernargs): Replace the common content with kernargs_abi.
(struct heap): Delete.
(main): Read GCN_STACK_SIZE envvar.
Allocate space for the device stacks.
Write the new kernargs fields.
* config/gcn/gcn.cc (gcn_option_override): Remove stack_size_opt.
(default_requested_args): Remove PRIVATE_SEGMENT_BUFFER_ARG and
PRIVATE_SEGMENT_WAVE_OFFSET_ARG.
(gcn_addr_space_convert): Mask the QUEUE_PTR_ARG content.
(gcn_expand_prologue): Move the TARGET_PACKED_WORK_ITEMS to the top.
Set up the stacks from the values in the kernargs, not private.
(gcn_expand_builtin_1): Match the stack configuration in the prologue.
(gcn_hsa_declare_function_name): Turn off the private segment.
(gcn_conditional_register_usage): Ensure QUEUE_PTR is fixed.
* config/gcn/gcn.h (FIXED_REGISTERS): Fix the QUEUE_PTR register.
* config/gcn/gcn.opt (mstack-size): Change the description.
include/ChangeLog:
* gomp-constants.h (GOMP_VERSION_GCN): Bump.
libgomp/ChangeLog:
* config/gcn/libgomp-gcn.h (DEFAULT_GCN_STACK_SIZE): New define.
(DEFAULT_TEAM_ARENA_SIZE): New define.
(struct heap): Move to this file.
(struct kernargs_abi): Likewise.
* config/gcn/team.c (gomp_gcn_enter_kernel): Use team arena size from
the kernargs.
* libgomp.h: Include libgomp-gcn.h.
(TEAM_ARENA_SIZE): Remove.
(team_malloc): Update the error message.
* plugin/plugin-gcn.c (struct kernargs): Move common content to
struct kernargs_abi.
(struct agent_info): Rename team arenas to ephemeral memories.
(struct team_arena_list): Rename ....
(struct ephemeral_memories_list): to this.
(struct heap): Delete.
(team_arena_size): New variable.
(stack_size): New variable.
(print_kernel_dispatch): Update debug messages.
(init_environment_variables): Read GCN_TEAM_ARENA_SIZE.
Read GCN_STACK_SIZE.
(get_team_arena): Rename ...
(configure_ephemeral_memories): ... to this, and set up stacks.
(release_team_arena): Rename ...
(release_ephemeral_memories): ... to this.
(destroy_team_arenas): Rename ...
(destroy_ephemeral_memories): ... to this.
(create_kernel_dispatch): Add num_threads parameter.
Adjust for kernargs_abi refactor and ephemeral memories.
(release_kernel_dispatch): Adjust for ephemeral memories.
(run_kernel): Pass thread-count to create_kernel_dispatch.
(GOMP_OFFLOAD_init_device): Adjust for ephemeral memories.
(GOMP_OFFLOAD_fini_device): Adjust for ephemeral memories.
gcc/testsuite/ChangeLog:
* gcc.c-torture/execute/pr47237.c: Xfail on amdgcn.
* gcc.dg/builtin-apply3.c: Xfail for amdgcn.
* gcc.dg/builtin-apply4.c: Xfail for amdgcn.
* gcc.dg/torture/stackalign/builtin-apply-3.c: Xfail for amdgcn.
* gcc.dg/torture/stackalign/builtin-apply-4.c: Xfail for amdgcn.
Fix the 'strict' modifier status: it is already listed (as 'Y') for OpenMP
5.1 for num_task and grainsize; only strict on num_threads is new with TR11.
libgomp/
* libgomp.texi (OpenMP TR11): Fix item for 'strict' modifier.
expand_omp_for_init_counts was using for the case where collapse(2)
inner loop has init expression dependent on non-constant multiple of
the outer iterator and the condition upper bound expression doesn't
depend on the outer iterator fold_unary (NEGATE_EXPR, ...). This
will just return NULL if it can't be folded, we need fold_build1
instead.
2023-01-19 Jakub Jelinek <jakub@redhat.com>
PR middle-end/108459
* omp-expand.cc (expand_omp_for_init_counts): Use fold_build1 rather
than fold_unary for NEGATE_EXPR.
* testsuite/libgomp.c/pr108459.c: New test.
GCC Administrator
Rainer Orth
Tobias Burnus
Thomas Schwinge
Hongyu Wang
Kwok Cheung Yeung
Jakub Jelinek
Andrew Stubbs
Martin Liska