This patch implements the outlining for offloading functions for code
annotated with the OpenMP target directive. It uses a temporary naming
of the outlined functions that will have to be updated later on once
target side codegen and registration of offloading libraries is
implemented - the naming needs to be made unique in the produced
library.
llvm-svn: 249148
Description.
If the simd clause is specified, the ordered regions encountered by any thread will use only a single SIMD lane to execute the ordered regions in the order of the loop iterations.
Restrictions.
An ordered construct with the simd clause is the only OpenMP construct that can appear in the simd region.
An ordered directive with ‘simd’ clause is generated as an outlined function and corresponding function call to prevent this part of code from vectorization later in backend.
llvm-svn: 248772
Patch improves codegen for OpenMP constructs. If the OpenMP region does not have internal 'cancel' construct, a call to 'void __kmpc_barrier()' runtime function is generated for all implicit/explicit barriers. If the region has inner 'cancel' directive, then
```
if (__kmpc_cancel_barrier())
exit from outer construct;
```
code is generated.
Also, the code for 'canellation point' directive is not generated if parent directive does not have 'cancel' directive.
llvm-svn: 247681
Current implementation may end up emitting an undefined reference for
an "inline __attribute__((always_inline))" function by generating an
"available_externally alwaysinline" IR function for it and then failing to
inline all the calls. This happens when a call to such function is in dead
code. As the inliner is an SCC pass, it does not process dead code.
Libc++ relies on the compiler never emitting such undefined reference.
With this patch, we emit a pair of
1. internal alwaysinline definition (called F.alwaysinline)
2a. A stub F() { musttail call F.alwaysinline }
-- or, depending on the linkage --
2b. A declaration of F.
The frontend ensures that F.inlinefunction is only used for direct
calls, and the stub is used for everything else (taking the address of
the function, really). Declaration (2b) is emitted in the case when
"inline" is meant for inlining only (like __gnu_inline__ and some
other cases).
This approach, among other nice properties, ensures that alwaysinline
functions are always internal, making it impossible for a direct call
to such function to produce an undefined symbol reference.
This patch is based on ideas by Chandler Carruth and Richard Smith.
llvm-svn: 247494
Current implementation may end up emitting an undefined reference for
an "inline __attribute__((always_inline))" function by generating an
"available_externally alwaysinline" IR function for it and then failing to
inline all the calls. This happens when a call to such function is in dead
code. As the inliner is an SCC pass, it does not process dead code.
Libc++ relies on the compiler never emitting such undefined reference.
With this patch, we emit a pair of
1. internal alwaysinline definition (called F.alwaysinline)
2a. A stub F() { musttail call F.alwaysinline }
-- or, depending on the linkage --
2b. A declaration of F.
The frontend ensures that F.inlinefunction is only used for direct
calls, and the stub is used for everything else (taking the address of
the function, really). Declaration (2b) is emitted in the case when
"inline" is meant for inlining only (like __gnu_inline__ and some
other cases).
This approach, among other nice properties, ensures that alwaysinline
functions are always internal, making it impossible for a direct call
to such function to produce an undefined symbol reference.
This patch is based on ideas by Chandler Carruth and Richard Smith.
llvm-svn: 247465
Currently all variables used in OpenMP regions are captured into a record and passed to outlined functions in this record. It may result in some poor performance because of too complex analysis later in optimization passes. Patch makes to emit outlined functions for parallel-based regions with a list of captured variables. It reduces code for 2*n GEPs, stores and loads at least.
Codegen for task-based regions remains unchanged because runtime requires that all captured variables are passed in captured record.
llvm-svn: 247251
Seems it broke the Polly build.
From http://lab.llvm.org:8011/builders/perf-x86_64-penryn-O3-polly-fast/builds/11687/steps/compile/logs/stdio:
In file included from /home/grosser/buildslave/perf-x86_64-penryn-O3-polly-fast/llvm.src/lib/TableGen/Record.cpp:14:0:
/home/grosser/buildslave/perf-x86_64-penryn-O3-polly-fast/llvm.src/include/llvm/TableGen/Record.h:369:3: error: looser throw specifier for 'virtual llvm::TypedInit::~TypedInit()'
/home/grosser/buildslave/perf-x86_64-penryn-O3-polly-fast/llvm.src/include/llvm/TableGen/Record.h:270:11: error: overriding 'virtual llvm::Init::~Init() noexcept (true)'
llvm-svn: 247222
Introduce an Address type to bundle a pointer value with an
alignment. Introduce APIs on CGBuilderTy to work with Address
values. Change core APIs on CGF/CGM to traffic in Address where
appropriate. Require alignments to be non-zero. Update a ton
of code to compute and propagate alignment information.
As part of this, I've promoted CGBuiltin's EmitPointerWithAlignment
helper function to CGF and made use of it in a number of places in
the expression emitter.
The end result is that we should now be significantly more correct
when performing operations on objects that are locally known to
be under-aligned. Since alignment is not reliably tracked in the
type system, there are inherent limits to this, but at least we
are no longer confused by standard operations like derived-to-base
conversions and array-to-pointer decay. I've also fixed a large
number of bugs where we were applying the complete-object alignment
to a pointer instead of the non-virtual alignment, although most of
these were hidden by the very conservative approach we took with
member alignment.
Also, because IRGen now reliably asserts on zero alignments, we
should no longer be subject to an absurd but frustrating recurring
bug where an incomplete type would report a zero alignment and then
we'd naively do a alignmentAtOffset on it and emit code using an
alignment equal to the largest power-of-two factor of the offset.
We should also now be emitting much more aggressive alignment
attributes in the presence of over-alignment. In particular,
field access now uses alignmentAtOffset instead of min.
Several times in this patch, I had to change the existing
code-generation pattern in order to more effectively use
the Address APIs. For the most part, this seems to be a strict
improvement, like doing pointer arithmetic with GEPs instead of
ptrtoint. That said, I've tried very hard to not change semantics,
but it is likely that I've failed in a few places, for which I
apologize.
ABIArgInfo now always carries the assumed alignment of indirect and
indirect byval arguments. In order to cut down on what was already
a dauntingly large patch, I changed the code to never set align
attributes in the IR on non-byval indirect arguments. That is,
we still generate code which assumes that indirect arguments have
the given alignment, but we don't express this information to the
backend except where it's semantically required (i.e. on byvals).
This is likely a minor regression for those targets that did provide
this information, but it'll be trivial to add it back in a later
patch.
I partially punted on applying this work to CGBuiltin. Please
do not add more uses of the CreateDefaultAligned{Load,Store}
APIs; they will be going away eventually.
llvm-svn: 246985
Added codegen for array section in 'depend' clause of 'task' directive. It emits to pointers, one for the begin of array section and another for the end of array section. Size of the section is calculated as (end + 1 - start) * sizeof(basic_element_type).
llvm-svn: 246422
Added codegen for array section in 'depend' clause of 'task' directive. It emits to pointers, one for the begin of array section and another for the end of array section. Size of the section is calculated as (end + 1 - start) * sizeof(basic_element_type).
llvm-svn: 246278
Summary:
float_cast_overflow is the only UBSan check without a source location attached.
This patch propagates SourceLocations where necessary to get them to the
EmitCheck() call.
Reviewers: rsmith, ABataev, rjmccall
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D11757
llvm-svn: 244568
The next code is generated for this construct:
```
if (__kmpc_cancellationpoint(ident_t *loc, kmp_int32 global_tid, kmp_int32 cncl_kind) != 0)
<exit from outer innermost construct>;
```
llvm-svn: 241239
If task directive has associated 'depend' clause then function kmp_int32 __kmpc_omp_task_with_deps ( ident_t *loc_ref, kmp_int32 gtid, kmp_task_t * new_task, kmp_int32 ndeps, kmp_depend_info_t *dep_list,kmp_int32 ndeps_noalias, kmp_depend_info_t *noalias_dep_list) must be called instead of __kmpc_omp_task().
If this directive has associated 'if' clause then also before a call of kmpc_omp_task_begin_if0() a function void __kmpc_omp_wait_deps ( ident_t *loc_ref, kmp_int32 gtid, kmp_int32 ndeps, kmp_depend_info_t *dep_list, kmp_int32 ndeps_noalias, kmp_depend_info_t *noalias_dep_list) must be called.
Array sections are not supported yet.
llvm-svn: 240532
Added parsing, sema analysis and codegen for '#pragma omp taskgroup' directive (OpenMP 4.0).
The code for directive is generated the following way:
#pragma omp taskgroup
<body>
void __kmpc_taskgroup(<loc>, thread_id);
<body>
void __kmpc_end_taskgroup(<loc>, thread_id);
llvm-svn: 240011
The following code is generated for reduction clause within 'omp simd' loop construct:
#pragma omp simd reduction(op:var)
for (...)
<body>
alloca priv_var
priv_var = <initial reduction value>;
<loop_start>:
<body> // references to original 'var' are replaced by 'priv_var'
<loop_end>:
var op= priv_var;
llvm-svn: 239881
This modification generates proper copyin/initialization sequences for array variables/parameters. Before they were considered as pointers, not arrays.
llvm-svn: 237691
Internal task structure must be generated like
typedef struct kmp_task {
void * shareds;
kmp_routine_entry_t routine;
kmp_int32 part_id;
kmp_routine_entry_t destructors;
} kmp_task_t;
struct kmp_task_t_with_privates {
kmp_task_t task_data;
.kmp_private. privates;
};
to avoid possible additional alignment bytes in first fields (shareds, routine, part_id and destructors). Runtime library is not aware of such kind additional alignment bytes.
llvm-svn: 237561
Fixed codegen for reduction operations min, max, && and ||. Codegen for them is quite similar and I was confused by this similarity.
Also added a call to kmpc_end_reduce() in atomic part of reduction codegen (call to kmpc_end_reduce_nowait() is not required).
Differential Revision: http://reviews.llvm.org/D9513
llvm-svn: 236689
Destructors are never called for cleanups, so we can't use SmallVector as a member.
Differential Revision: http://reviews.llvm.org/D9399
llvm-svn: 236491
Destructors are never called for cleanups, so we can't use SmallVector as a member.
Differential Revision: http://reviews.llvm.org/D9399
llvm-svn: 236487
Destructors are never called for cleanups, so we can't use SmallVector as a member.
Differential Revision: http://reviews.llvm.org/D9399
llvm-svn: 236482
Destructors are never called for cleanups, so we can't use SmallVector as a member.
Differential Revision: http://reviews.llvm.org/D9399
llvm-svn: 236480
For tasks codegen for private/firstprivate variables are different rather than for other directives.
1. Build an internal structure of privates for each private variable:
struct .kmp_privates_t. {
Ty1 var1;
...
Tyn varn;
};
2. Add a new field to kmp_task_t type with list of privates.
struct kmp_task_t {
void * shareds;
kmp_routine_entry_t routine;
kmp_int32 part_id;
kmp_routine_entry_t destructors;
.kmp_privates_t. privates;
};
3. Create a function with destructors calls for all privates after end of task region.
kmp_int32 .omp_task_destructor.(kmp_int32 gtid, kmp_task_t *tt) {
~Destructor(&tt->privates.var1);
...
~Destructor(&tt->privates.varn);
return 0;
}
4. Perform initialization of all firstprivate fields (by simple copying for POD data, copy constructor calls for classes) + provide address of a destructor function after kmpc_omp_task_alloc() and before kmpc_omp_task() calls.
kmp_task_t *new_task = __kmpc_omp_task_alloc(ident_t *, kmp_int32 gtid, kmp_int32 flags, size_t sizeof_kmp_task_t, size_t sizeof_shareds, kmp_routine_entry_t *task_entry);
CopyConstructor(new_task->privates.var1, *new_task->shareds.var1_ref);
new_task->shareds.var1_ref = &new_task->privates.var1;
...
CopyConstructor(new_task->privates.varn, *new_task->shareds.varn_ref);
new_task->shareds.varn_ref = &new_task->privates.varn;
new_task->destructors = .omp_task_destructor.;
kmp_int32 __kmpc_omp_task(ident_t *, kmp_int32 gtid, kmp_task_t *new_task)
Differential Revision: http://reviews.llvm.org/D9370
llvm-svn: 236479
For tasks codegen for private/firstprivate variables are different rather than for other directives.
1. Build an internal structure of privates for each private variable:
struct .kmp_privates_t. {
Ty1 var1;
...
Tyn varn;
};
2. Add a new field to kmp_task_t type with list of privates.
struct kmp_task_t {
void * shareds;
kmp_routine_entry_t routine;
kmp_int32 part_id;
kmp_routine_entry_t destructors;
.kmp_privates_t. privates;
};
3. Create a function with destructors calls for all privates after end of task region.
kmp_int32 .omp_task_destructor.(kmp_int32 gtid, kmp_task_t *tt) {
~Destructor(&tt->privates.var1);
...
~Destructor(&tt->privates.varn);
return 0;
}
4. Perform default initialization of all private fields (no initialization for POD data, default constructor calls for classes) + provide address of a destructor function after kmpc_omp_task_alloc() and before kmpc_omp_task() calls.
kmp_task_t *new_task = __kmpc_omp_task_alloc(ident_t *, kmp_int32 gtid, kmp_int32 flags, size_t sizeof_kmp_task_t, size_t sizeof_shareds, kmp_routine_entry_t *task_entry);
DefaultConstructor(new_task->privates.var1);
new_task->shareds.var1_ref = &new_task->privates.var1;
...
DefaultConstructor(new_task->privates.varn);
new_task->shareds.varn_ref = &new_task->privates.varn;
new_task->destructors = .omp_task_destructor.;
kmp_int32 __kmpc_omp_task(ident_t *, kmp_int32 gtid, kmp_task_t *new_task)
Differential Revision: http://reviews.llvm.org/D9322
llvm-svn: 236207
Fixed initialization of 'single' region completion + changed type of the third argument of __kmpc_copyprivate() runtime function to size_t.
llvm-svn: 236198
Emit the following code for 'taskwait' directive within tied task:
call i32 @__kmpc_omp_taskwait(<loc>, i32 <thread_id>);
Differential Revision: http://reviews.llvm.org/D9245
llvm-svn: 235836
If condition evaluates to true, the code executes task by calling @__kmpc_omp_task() runtime function.
If condition evaluates to false, the code executes serial version of the code by executing the following code:
call void @__kmpc_omp_task_begin_if0(<loc>, <threadid>, <task_t_ptr, returned by @__kmpc_omp_task_alloc()>);
proxy_task_entry(<gtid>, <task_t_ptr, returned by @__kmpc_omp_task_alloc()>);
call void @__kmpc_omp_task_complete_if0(<loc>, <threadid>, <task_t_ptr, returned by @__kmpc_omp_task_alloc()>);
Also it checks if the condition is constant and if it is constant it evaluates its value and then generates either parallel version of the code (if the condition evaluates to true), or the serial version of the code (if the condition evaluates to false).
Differential Revision: http://reviews.llvm.org/D9143
llvm-svn: 235507
Add codegen for 'ordered' directive:
__kmpc_ordered(ident_t *, gtid);
<associated statement>;
__kmpc_end_ordered(ident_t *, gtid);
Also for 'for' directives with the dynamic scheduling and an 'ordered' clause added a call to '__kmpc_dispatch_fini_(4|8)[u]()' function after increment expression for loop control variable:
while(__kmpc_dispatch_next(&LB, &UB)) {
idx = LB;
while (idx <= UB) { BODY; ++idx;
__kmpc_dispatch_fini_(4|8)[u](); // For ordered loops only.
} // inner loop
}
Differential Revision: http://reviews.llvm.org/D9070
llvm-svn: 235496
Samuel Antao