Commit Graph

347 Commits

Author SHA1 Message Date
Alexey Bataev 6e01dc1b84 [OPENMP][DEBUG] Fix for PR33676: Debug info for OpenMP region is broken.
After some changes in clang/LLVM debug info for task-based regions was
not generated at all. Patch fixes this problem.

llvm-svn: 310850
2017-08-14 16:03:47 +00:00
Alexey Bataev 3c595a6b2c [OPENMP] Generalization of calls of the outlined functions.
General improvement of the outlined functions calls.

llvm-svn: 310840
2017-08-14 15:01:03 +00:00
Alexey Bataev 3b8d5586ec [OPENMP][DEBUG] Set proper address space info if required by target.
Arguments, passed to the outlined function, must have correct address
space info for proper Debug info support. Patch sets global address
space for arguments that are mapped and passed by reference.

Also, cuda-gdb does not handle reference types correctly, so reference
arguments are represented as pointers.

llvm-svn: 310387
2017-08-08 18:04:06 +00:00
Alexey Bataev 2c7eee5b84 [OPENMP] Unify generation of outlined function calls.
llvm-svn: 310098
2017-08-04 19:10:54 +00:00
Alexey Bataev be5a8b42cd [OPENMP] Codegen for reduction clauses in 'taskloop' directives.
Adds codegen for taskloop-based directives.

llvm-svn: 308174
2017-07-17 13:30:36 +00:00
Alexey Bataev 5c40bec5eb [OPENMP] Generalization of codegen for reduction clauses.
Reworked codegen for reduction clauses for future support of reductions
in task-based directives.

llvm-svn: 307910
2017-07-13 13:36:14 +00:00
Alexey Bataev 3344603f7b [OPENMP] Emit implicit taskgroup block around taskloop directives.
If taskloop directive has no associated nogroup clause, it must emitted
inside implicit taskgroup block. Runtime supports it, but we need to
generate implicit taskgroup block explicitly to support future
reductions codegen.

llvm-svn: 307822
2017-07-12 18:09:32 +00:00
Alexey Bataev 1fdfdf7155 [OPENMP][DEBUG] Generate second function with correct arg types.
Currently, if the some of the parameters are captured by value, this
argument is converted to uintptr_t type and thus we loosing the debug
info about real type of the argument (captured variable):
```
void @.outlined_function.(uintptr %par);

...
%a = alloca i32
%a.casted = alloca uintptr
%cast = bitcast uintptr* %a.casted to i32*
%a.val = load i32, i32 *%a
store i32 %a.val, i32 *%cast
%a.casted.val = load uintptr, uintptr* %a.casted
call void @.outlined_function.(uintptr %a.casted.val)
...
```

To resolve this problem, in debug mode a speciall external wrapper
function is generated, that calls the outlined function with the correct
parameters types:
```
void @.wrapper.(uintptr %par) {
  %a = alloca i32
  %cast = bitcast i32* %a to uintptr*
  store uintptr %par, uintptr *%cast
  %a.val = load i32, i32* %a
  call void @.outlined_function.(i32 %a)
  ret void
}
void @.outlined_function.(i32 %par);

...
%a = alloca i32
%a.casted = alloca uintptr
%cast = bitcast uintptr* %a.casted to i32*
%a.val = load i32, i32 *%a
store i32 %a.val, i32 *%cast
%a.casted.val = load uintptr, uintptr* %a.casted
call void @.wrapper.(uintptr %a.casted.val)
...
```

llvm-svn: 306697
2017-06-29 16:43:05 +00:00
Alexey Bataev 5d1c3f6add [OPENMP] Use MapVector instead of DenseMap for stable codegen, NFC.
llvm-svn: 306419
2017-06-27 15:46:42 +00:00
Gheorghe-Teodor Bercea 47633db42e Add comma to comment.
llvm-svn: 305294
2017-06-13 15:35:27 +00:00
Alexey Bataev 56223237b0 [DebugInfo] Add kind of ImplicitParamDecl for emission of FlagObjectPointer.
Summary:
If the first parameter of the function is the ImplicitParamDecl, codegen
automatically marks it as an implicit argument with `this` or `self`
pointer. Added internal kind of the ImplicitParamDecl to separate
'this', 'self', 'vtt' and other implicit parameters from other kind of
parameters.

Reviewers: rjmccall, aaron.ballman

Subscribers: cfe-commits

Differential Revision: https://reviews.llvm.org/D33735

llvm-svn: 305075
2017-06-09 13:40:18 +00:00
Mehdi Amini 6aa9e9b41a IRGen: Add optnone attribute on function during O0
Amongst other, this will help LTO to correctly handle/honor files
compiled with O0, helping debugging failures.
It also seems in line with how we handle other options, like how
-fnoinline adds the appropriate attribute as well.

Differential Revision: https://reviews.llvm.org/D28404

llvm-svn: 304127
2017-05-29 05:38:20 +00:00
George Rokos 29d0f00340 [OpenMP] Create COMDAT group for OpenMP offload registration code to avoid multiple copies
Thanks to Sergey Dmitriev for submitting the patch.

Differential Revision: https://reviews.llvm.org/D33509

llvm-svn: 304056
2017-05-27 03:03:13 +00:00
Krzysztof Parzyszek 8f248234fa [CodeGen] Propagate LValueBaseInfo instead of AlignmentSource
The functions creating LValues propagated information about alignment
source. Extend the propagated data to also include information about
possible unrestricted aliasing. A new class LValueBaseInfo will
contain both AlignmentSource and MayAlias info.

This patch should not introduce any functional changes.

Differential Revision: https://reviews.llvm.org/D33284

llvm-svn: 303358
2017-05-18 17:07:11 +00:00
Serge Guelton 1d993270b3 Suppress all uses of LLVM_END_WITH_NULL. NFC.
Use variadic templates instead of relying on <cstdarg> + sentinel.

This enforces better type checking and makes code more readable.

Differential revision: https://reviews.llvm.org/D32550

llvm-svn: 302572
2017-05-09 19:31:30 +00:00
Carlo Bertolli b0ff0a69c3 Recommit of
[OpenMP] Initial implementation of code generation for pragma 'distribute parallel for' on host

https://reviews.llvm.org/D29508

This patch makes the following additions:

It abstracts away loop bound generation code from procedures associated with pragma 'for' and loops in general, in such a way that the same procedures can be used for 'distribute parallel for' without the need for a full re-implementation.
It implements code generation for 'distribute parallel for' and adds regression tests. It includes tests for clauses.
It is important to notice that most of the clauses are implemented as part of existing procedures. For instance, firstprivate is already implemented for 'distribute' and 'for' as separate pragmas. As the implementation of 'distribute parallel for' is based on the same procedures, then we automatically obtain implementation for such clauses without the need to add new code. However, this requires regression tests that verify correctness of produced code.

llvm-svn: 301340
2017-04-25 17:52:12 +00:00
Carlo Bertolli f09daae75d Revert r301223
llvm-svn: 301233
2017-04-24 19:50:35 +00:00
Carlo Bertolli 4287d65c10 [OpenMP] Initial implementation of code generation for pragma 'distribute parallel for' on host
https://reviews.llvm.org/D29508

This patch makes the following additions:

1. It abstracts away loop bound generation code from procedures associated with pragma 'for' and loops in general, in such a way that the same procedures can be used for 'distribute parallel for' without the need for a full re-implementation.
2. It implements code generation for 'distribute parallel for' and adds regression tests. It includes tests for clauses.

It is important to notice that most of the clauses are implemented as part of existing procedures. For instance, firstprivate is already implemented for 'distribute' and 'for' as separate pragmas. As the implementation of 'distribute parallel for' is based on the same procedures, then we automatically obtain implementation for such clauses without the need to add new code. However, this requires regression tests that verify correctness of produced code.

Looking forward to comments.

llvm-svn: 301223
2017-04-24 19:26:11 +00:00
Simon Pilgrim 2c51880a82 Spelling mistakes in comments. NFCI. (PR27635)
llvm-svn: 299083
2017-03-30 14:13:19 +00:00
Reid Kleckner e258c44002 Use arg_begin() instead of getArgumentList().begin(), the argument list is an implementation detail
llvm-svn: 297975
2017-03-16 18:55:46 +00:00
John McCall 5ad740756f Promote ConstantInitBuilder to be a public CodeGen API; it's
a generally useful utility for other frontends.  NFC.

llvm-svn: 296806
2017-03-02 20:04:19 +00:00
Jonas Hahnfeld b07931f01d [OpenMP] Fix cancellation point in task with no cancel
With tasks, the cancel may happen in another task. This has a different
region info which means that we can't find it here.

Differential Revision: https://reviews.llvm.org/D30091

llvm-svn: 295474
2017-02-17 18:32:58 +00:00
Jonas Hahnfeld 20fce72f1b [OpenMP] Remove barriers at cancel and cancellation point
This resolves a deadlock with the cancel directive when there is no explicit
cancellation point. In that case, the implicit barrier acts as cancellation
point. After removing the barrier after cancel, the now unmatched barrier for
the explicit cancellation point has to go as well.

This has probably worked before rL255992: With the calls for the explicit
barrier, it was sure that all threads passed a barrier before exiting.

Reported by Simon Convent and Joachim Protze!

Differential Revision: https://reviews.llvm.org/D30088

llvm-svn: 295473
2017-02-17 18:32:51 +00:00
Arpith Chacko Jacob 101e8fb1f3 [OpenMP] Parallel reduction on the NVPTX device.
This patch implements codegen for the reduction clause on
any parallel construct for elementary data types.  An efficient
implementation requires hierarchical reduction within a
warp and a threadblock.  It is complicated by the fact that
variables declared in the stack of a CUDA thread cannot be
shared with other threads.

The patch creates a struct to hold reduction variables and
a number of helper functions.  The OpenMP runtime on the GPU
implements reduction algorithms that uses these helper
functions to perform reductions within a team.  Variables are
shared between CUDA threads using shuffle intrinsics.

An implementation of reductions on the NVPTX device is
substantially different to that of CPUs.  However, this patch
is written so that there are minimal changes to the rest of
OpenMP codegen.

The implemented design allows the compiler and runtime to be
decoupled, i.e., the runtime does not need to know of the
reduction operation(s), the type of the reduction variable(s),
or the number of reductions.  The design also allows reuse of
host codegen, with appropriate specialization for the NVPTX
device.

While the patch does introduce a number of abstractions, the
expected use case calls for inlining of the GPU OpenMP runtime.
After inlining and optimizations in LLVM, these abstractions
are unwound and performance of OpenMP reductions is comparable
to CUDA-canonical code.

Patch by Tian Jin in collaboration with Arpith Jacob

Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D29758

llvm-svn: 295333
2017-02-16 16:20:16 +00:00
Arpith Chacko Jacob bd6344c0be Revert r295319 while investigating buildbot failure.
llvm-svn: 295323
2017-02-16 14:25:35 +00:00
Arpith Chacko Jacob 8e170fc857 [OpenMP] Parallel reduction on the NVPTX device.
This patch implements codegen for the reduction clause on
any parallel construct for elementary data types.  An efficient
implementation requires hierarchical reduction within a
warp and a threadblock.  It is complicated by the fact that
variables declared in the stack of a CUDA thread cannot be
shared with other threads.

The patch creates a struct to hold reduction variables and
a number of helper functions.  The OpenMP runtime on the GPU
implements reduction algorithms that uses these helper
functions to perform reductions within a team.  Variables are
shared between CUDA threads using shuffle intrinsics.

An implementation of reductions on the NVPTX device is
substantially different to that of CPUs.  However, this patch
is written so that there are minimal changes to the rest of
OpenMP codegen.

The implemented design allows the compiler and runtime to be
decoupled, i.e., the runtime does not need to know of the
reduction operation(s), the type of the reduction variable(s),
or the number of reductions.  The design also allows reuse of
host codegen, with appropriate specialization for the NVPTX
device.

While the patch does introduce a number of abstractions, the
expected use case calls for inlining of the GPU OpenMP runtime.
After inlining and optimizations in LLVM, these abstractions
are unwound and performance of OpenMP reductions is comparable
to CUDA-canonical code.

Patch by Tian Jin in collaboration with Arpith Jacob

Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D29758

llvm-svn: 295319
2017-02-16 14:03:36 +00:00
Arpith Chacko Jacob 99a1e0eba5 [OpenMP] Codegen support for 'target teams' on the host.
This patch adds support for codegen of 'target teams' on the host.
This combined directive has two captured statements, one for the
'teams' region, and the other for the 'parallel'.

This target teams region is offloaded using the __tgt_target_teams()
call. The patch sets the number of teams as an argument to
this call.

Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D29084

llvm-svn: 293005
2017-01-25 02:18:43 +00:00
Arpith Chacko Jacob 86f9e46365 Reverting commit because an NVPTX patch sneaked in. Break up into two
patches.

llvm-svn: 293003
2017-01-25 01:45:59 +00:00
Arpith Chacko Jacob 4dbf368e14 [OpenMP] Codegen support for 'target teams' on the host.
This patch adds support for codegen of 'target teams' on the host.
This combined directive has two captured statements, one for the
'teams' region, and the other for the 'parallel'.

This target teams region is offloaded using the __tgt_target_teams()
call. The patch sets the number of teams as an argument to
this call.

Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D29084

llvm-svn: 293001
2017-01-25 01:38:33 +00:00
Arpith Chacko Jacob 33c849a007 [OpenMP] Support for the num_threads-clause on 'target parallel'.
The num_threads-clause on the combined directive applies to the
'parallel' region of this construct. We modify the NumThreadsClause
class to capture the clause expression within the 'target' region.

The offload runtime call for 'target parallel' is changed to
__tgt_target_teams() with 1 team and the number of threads set by
this clause or a default if none.

Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D29082

llvm-svn: 292997
2017-01-25 00:57:16 +00:00
Arpith Chacko Jacob 19b911cb75 [OpenMP] Codegen support for 'target parallel' on the host.
This patch adds support for codegen of 'target parallel' on the host.
It is also the first combined directive that requires two or more
captured statements.  Support for this functionality is included in
the patch.

A combined directive such as 'target parallel' has two captured
statements, one for the 'target' and the other for the 'parallel'
region.  Two captured statements are required because each has
different implicit parameters (see SemaOpenMP.cpp).  For example,
the 'parallel' has 'global_tid' and 'bound_tid' while the 'target'
does not.  The patch adds support for handling multiple captured
statements based on the combined directive.

When codegen'ing the 'target parallel' directive, the 'target'
outlined function is created using the outer captured statement
and the 'parallel' outlined function is created using the inner
captured statement.

Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D28753

llvm-svn: 292419
2017-01-18 18:18:53 +00:00
Arpith Chacko Jacob 42793e000a Revert r292374 to debug Windows buildbot failure.
llvm-svn: 292400
2017-01-18 15:36:05 +00:00
Arpith Chacko Jacob 68019578a3 [OpenMP] Codegen support for 'target parallel' on the host.
This patch adds support for codegen of 'target parallel' on the host.
It is also the first combined directive that requires two or more
captured statements.  Support for this functionality is included in
the patch.

A combined directive such as 'target parallel' has two captured
statements, one for the 'target' and the other for the 'parallel'
region.  Two captured statements are required because each has
different implicit parameters (see SemaOpenMP.cpp).  For example,
the 'parallel' has 'global_tid' and 'bound_tid' while the 'target'
does not.  The patch adds support for handling multiple captured
statements based on the combined directive.

When codegen'ing the 'target parallel' directive, the 'target'
outlined function is created using the outer captured statement
and the 'parallel' outlined function is created using the inner
captured statement.

Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D28753

llvm-svn: 292374
2017-01-18 15:14:52 +00:00
Arpith Chacko Jacob 43a8b7bc8c [OpenMP] Refactor code that calls codegen for target regions on the device.
This patch refactors code that calls codegen for target regions.  Currently
the codebase only supports the 'target' directive.  The patch pulls out
common target processing code into a static function that can be called
by codegen for any target directive.

Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D28752

llvm-svn: 292134
2017-01-16 15:26:02 +00:00
Malcolm Parsons c6e4583dbb Remove unused lambda captures. NFC
llvm-svn: 291939
2017-01-13 18:55:32 +00:00
Arpith Chacko Jacob bb36fe8dba [OpenMP] Basic support for a parallel directive in a target region on an NVPTX device
Summary:

This patch introduces support for the execution of parallel constructs in a target
region on the NVPTX device.  Parallel regions must be in the lexical scope of the
target directive.

The master thread in the master warp signals parallel work for worker threads in worker
warps on encountering a parallel region.

Note: The patch does not yet support capture of arguments in a parallel region so
the test cases are simple.

Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D28145

llvm-svn: 291565
2017-01-10 15:42:51 +00:00
Samuel Antao f83efdb77a [OpenMP] Add fields for flags in the offload entry descriptor.
Summary:
This patch adds two fields to the offload entry descriptor. One field is meant to signal Ctors/Dtors and `link` global variables, and the other is reserved for runtime library use. 

 Currently, these fields are only filled with zeros in the current code generation, but that will change when `declare target` is added. 

The reason, we are adding these fields now is to make the code generation consistent with the runtime library proposal under review in https://reviews.llvm.org/D14031.

Reviewers: ABataev, hfinkel, carlo.bertolli, kkwli0, arpith-jacob, Hahnfeld

Subscribers: cfe-commits, caomhin, jholewinski

Differential Revision: https://reviews.llvm.org/D28298

llvm-svn: 291124
2017-01-05 16:02:49 +00:00
Chandler Carruth fcd33149b4 Cleanup the handling of noinline function attributes, -fno-inline,
-fno-inline-functions, -O0, and optnone.

These were really, really tangled together:
- We used the noinline LLVM attribute for -fno-inline
  - But not for -fno-inline-functions (breaking LTO)
  - But we did use it for -finline-hint-functions (yay, LTO is happy!)
  - But we didn't for -O0 (LTO is sad yet again...)
- We had weird structuring of CodeGenOpts with both an inlining
  enumeration and a boolean. They interacted in weird ways and
  needlessly.
- A *lot* of set smashing went on with setting these, and then got worse
  when we considered optnone and other inlining-effecting attributes.
- A bunch of inline affecting attributes were managed in a completely
  different place from -fno-inline.
- Even with -fno-inline we failed to put the LLVM noinline attribute
  onto many generated function definitions because they didn't show up
  as AST-level functions.
- If you passed -O0 but -finline-functions we would run the normal
  inliner pass in LLVM despite it being in the O0 pipeline, which really
  doesn't make much sense.
- Lastly, we used things like '-fno-inline' to manipulate the pass
  pipeline which forced the pass pipeline to be much more
  parameterizable than it really needs to be. Instead we can *just* use
  the optimization level to select a pipeline and control the rest via
  attributes.

Sadly, this causes a bunch of churn in tests because we don't run the
optimizer in the tests and check the contents of attribute sets. It
would be awesome if attribute sets were a bit more FileCheck friendly,
but oh well.

I think this is a significant improvement and should remove the semantic
need to change what inliner pass we run in order to comply with the
requested inlining semantics by relying completely on attributes. It
also cleans up tho optnone and related handling a bit.

One unfortunate aspect of this is that for generating alwaysinline
routines like those in OpenMP we end up removing noinline and then
adding alwaysinline. I tried a bunch of other approaches, but because we
recompute function attributes from scratch and don't have a declaration
here I couldn't find anything substantially cleaner than this.

Differential Revision: https://reviews.llvm.org/D28053

llvm-svn: 290398
2016-12-23 01:24:49 +00:00
Samuel Antao 4b75b8726d Fix typo and remove unnecessary statement.
llvm-svn: 289458
2016-12-12 19:26:31 +00:00
Samuel Antao 4c8035bca4 Fix format and a few typos in comments.
llvm-svn: 289450
2016-12-12 18:00:20 +00:00
John McCall f1788639c5 Hide the result of building a constant initializer. NFC.
llvm-svn: 288080
2016-11-28 22:18:30 +00:00
John McCall 23c9dc6585 ConstantBuilder -> ConstantInitBuilder for clarity, and
move the member classes up to top level to allow forward
declarations to name them.  NFC.

llvm-svn: 288079
2016-11-28 22:18:27 +00:00
Benjamin Kramer 81cb4b7103 [CodeGen] Pass objects that are expensive to copy by const ref.
No functionality change. Found by clang-tidy's
performance-unnecessary-value-param.

llvm-svn: 287894
2016-11-24 16:01:20 +00:00
John McCall 6c9f1fdb5c Introduce a helper class for building complex constant initializers. NFC.
I've adopted this in most of the places it makes sense, but v-tables
and CGObjCMac will need a second pass.

llvm-svn: 287437
2016-11-19 08:17:24 +00:00
Peter Collingbourne d9445c49ad Bitcode: Change module reader functions to return an llvm::Expected.
Differential Revision: https://reviews.llvm.org/D26562

llvm-svn: 286752
2016-11-13 07:00:17 +00:00
Teresa Johnson ffc4e2420f Mirror the llvm changes that split Bitcode/ReaderWriter.h
The change in D26502 splits ReaderWriter.h, which contains the APIs
into both the BitReader and BitWriter libraries, into BitcodeReader.h
and BitcodeWriter.h.

Change clang uses to the appropriate split header(s).

llvm-svn: 286567
2016-11-11 05:35:12 +00:00
Paul Robinson 78fb132af0 Add FIXMEs for MSVC 2013 hacks in r277211. NFC.
llvm-svn: 277396
2016-08-01 22:12:46 +00:00
Hans Wennborg bc1b58d086 Fix VS2013 build of CGOpenMPRuntime.cpp
It seems the compiler was getting confused by the in-class initializers
in local struct MapInfo, so moving those to a default constructor
instead.

llvm-svn: 277256
2016-07-30 00:41:37 +00:00
Paul Robinson 15c840052e Fix CGOpenMPRuntime.cpp for VS2013. NFC.
I don't know why these changes work but they do.

llvm-svn: 277211
2016-07-29 20:46:16 +00:00
Samuel Antao 44bcdb3731 [OpenMP] Change name of variable in mappble expression.
This attempts to fix a failure in Windows bots pottentially related with a reserved keyword.

llvm-svn: 276988
2016-07-28 15:31:29 +00:00
Samuel Antao cf3f83e46b [OpenMP] Do not use default argument in lambda from mappable expressions handlers.
Windows bots were complaining about that.

llvm-svn: 276981
2016-07-28 14:47:35 +00:00
Samuel Antao 6890b09634 [OpenMP] Code generation for the is_device_ptr clause
Summary: This patch adds support for the is_device_ptr clause. It expands SEMA to use the mappable expression logic that can only be tested with code generation in place and check conflicts with other data sharing related clauses using the mappable expressions infrastructure.

Reviewers: hfinkel, carlo.bertolli, arpith-jacob, kkwli0, ABataev

Subscribers: caomhin, cfe-commits

Differential Revision: https://reviews.llvm.org/D22788

llvm-svn: 276978
2016-07-28 14:25:09 +00:00
Samuel Antao cc10b85789 [OpenMP] Codegen for use_device_ptr clause.
Summary: This patch adds support for the use_device_ptr clause. It includes changes in SEMA that could not be tested without codegen, namely, the use of the first private logic and mappable expressions support.

Reviewers: hfinkel, carlo.bertolli, arpith-jacob, kkwli0, ABataev

Subscribers: caomhin, cfe-commits

Differential Revision: https://reviews.llvm.org/D22691

llvm-svn: 276977
2016-07-28 14:23:26 +00:00
Samuel Antao 03a3cec480 [OpenMP] Add support to map member expressions with references to pointers.
Summary: This patch add support to map pointers through references in class members. Although a reference does not have storage that a user can access, it still has to be mapped in order to get the deep copy right and the dereferencing code work properly.

Reviewers: hfinkel, carlo.bertolli, arpith-jacob, kkwli0, ABataev

Subscribers: caomhin, cfe-commits

Differential Revision: https://reviews.llvm.org/D22787

llvm-svn: 276934
2016-07-27 22:52:16 +00:00
Samuel Antao 403ffd409f [OpenMP] Add support for mapping array sections through pointer references.
Summary:
This patch fixes a bug in the map of array sections whose base is a reference to a pointer. The existing mapping support was not prepared to deal with it, causing the compiler to crash.

Mapping a reference to a pointer enjoys the same characteristics of a regular pointer, i.e., it is passed by value. Therefore, the reference has to be materialized in the target region.

Reviewers: hfinkel, carlo.bertolli, kkwli0, ABataev

Subscribers: caomhin, cfe-commits

Differential Revision: https://reviews.llvm.org/D22690

llvm-svn: 276933
2016-07-27 22:49:49 +00:00
David Majnemer 59f7792136 Use more ArrayRefs
No functional change is intended, just a small refactoring.

llvm-svn: 273647
2016-06-24 04:05:48 +00:00
Samuel Antao 6d0042642a Re-apply r272900 - [OpenMP] Cast captures by copy when passed to fork call so that they are compatible to what the runtime library expects.
An issue in one of the regression tests was fixed for 32-bit hosts.

llvm-svn: 272931
2016-06-16 18:39:34 +00:00
Samuel Antao b1f9501242 Revert r272900 - [OpenMP] Cast captures by copy when passed to fork call so that they are compatible to what the runtime library expects.
Was causing trouble in one of the regression tests for a 32-bit address space.

llvm-svn: 272908
2016-06-16 16:06:22 +00:00
Samuel Antao 4951617980 [OpenMP] Cast captures by copy when passed to fork call so that they are compatible to what the runtime library expects.
Summary:
This patch fixes an issue detected when firstprivate variables are passed to an OpenMP outlined function vararg list. Currently they are not compatible with what the runtime library expects causing malfunction in some targets.

This patch fixes the issue by moving the casting logic already in place for offloading to the common code that creates the outline function and arguments and updates the regression tests accordingly.

Reviewers: hfinkel, arpith-jacob, carlo.bertolli, kkwli0, ABataev

Subscribers: cfe-commits, caomhin

Differential Revision: http://reviews.llvm.org/D21150

llvm-svn: 272900
2016-06-16 15:09:31 +00:00
Peter Collingbourne bcf909d737 Update clang for D20348
Differential Revision: http://reviews.llvm.org/D20339

llvm-svn: 272710
2016-06-14 21:02:05 +00:00
Nico Weber a691689ae8 Remove a few gendered pronouns.
llvm-svn: 272415
2016-06-10 18:53:04 +00:00
Alexey Bataev 6cff62484a [OPENMP 4.5] Additional codegen for statically scheduled loops with
'simd' modifier.

Runtime library defines new schedule constant kmp_sch_static_balanced_chunked = 45 for static loop-based directives  static with chunk adjustment (e.g., simd). Added codegen for this kind of schedule.

llvm-svn: 271204
2016-05-30 13:05:14 +00:00
Alexey Bataev ad537bb8a0 [OPENMP 4.5] Fixed codegen for 'priority' and destructors in task-based
directives.

'kmp_task_t' record type added a new field for 'priority' clause and
changed the representation of pointer to destructors for privates used
within loop-based directives.
Old representation:

typedef struct kmp_task {                   /* GEH: Shouldn't this be
aligned somehow? */
  void *shareds;                            /**< pointer to block of
    pointers to shared vars   */
  kmp_routine_entry_t routine;              /**< pointer to routine
    to call for executing task */
  kmp_int32 part_id;                        /**< part id for the
    task                          */
  kmp_routine_entry_t destructors;        /* pointer to function to
  invoke deconstructors of firstprivate C++ objects */
  /*  private vars  */
} kmp_task_t;

New representation:

typedef struct kmp_task {                   /* GEH: Shouldn't this be
aligned somehow? */
  void *shareds;                            /**< pointer to block of
    pointers to shared vars   */
  kmp_routine_entry_t routine;              /**< pointer to routine
    to call for executing task */
  kmp_int32 part_id;                        /**< part id for the
    task                          */
  kmp_cmplrdata_t data1; /* Two known
optional additions: destructors and priority */
  kmp_cmplrdata_t data2; /* Process
destructors first, priority second */
/* future data */
  /*  private vars  */
} kmp_task_t;

Also excessive initialization of 'destructors' fields to 'null' was
removed from codegen if it is known that no destructors shal be used.
Currently a special bit is used in 'kmp_tasking_flags_t' bitfields
('destructors_thunk' bitfield).

llvm-svn: 271201
2016-05-30 09:06:50 +00:00
Samuel Antao 8d2d730f2a [OpenMP] Codegen for target update directive.
Summary: This patch implements the code generation for the `target update` directive. The implemntation relies on the logic already in place for target data standalone directives, i.e. target enter/exit data.

Reviewers: hfinkel, carlo.bertolli, arpith-jacob, kkwli0, ABataev

Subscribers: caomhin, cfe-commits

Differential Revision: http://reviews.llvm.org/D20650

llvm-svn: 270886
2016-05-26 18:30:22 +00:00
Samuel Antao d486f84c57 [OpenMP] Add support for the 'private pointer' flag to signal variables captured in target regions and used in first-private clauses.
Summary: If a variable is implicitly mapped (doesn't show in a map clause), the runtime library has to be informed if the corresponding capture shows up in first-private clause, so that the storage previously allocated in the device is used. This patch adds the support for that.

Reviewers: hfinkel, carlo.bertolli, arpith-jacob, kkwli0, ABataev

Subscribers: caomhin, cfe-commits

Differential Revision: http://reviews.llvm.org/D20112

llvm-svn: 270870
2016-05-26 16:53:38 +00:00
Samuel Antao 6782e944d2 [OpenMP] Adjust map type bits according to latest spec and use zero size array sections for pointers.
Summary: This patch changes the bits used to specify the map types according to the latest version of the libomptarget document and add the support for zero size array section when pointers are being implicitly mapped. This completes the missing new 4.5 map semantics.

Reviewers: hfinkel, carlo.bertolli, arpith-jacob, kkwli0, ABataev

Subscribers: caomhin, cfe-commits

Differential Revision: http://reviews.llvm.org/D20111

llvm-svn: 270868
2016-05-26 16:48:10 +00:00
Alexey Bataev 8b42706a6e [OPENMP 4.5] Codegen for dacross loop synchronization constructs.
OpenMP 4.5 adds support for doacross loop synchronization. Patch
implements codegen for this construct.

llvm-svn: 270690
2016-05-25 12:36:08 +00:00
Alexey Bataev 7ace49dff1 [OPENMP] Pass scalar firstprivate vars by value.
For better performance and to unify code with offloading part we pass
scalar firstprivate values by value, instead of by reference. It will
remove some extra copying operations.

llvm-svn: 269751
2016-05-17 08:55:33 +00:00
Alexey Bataev 1e1e286a6b [OPENMP 4.5] Initial codegen for 'priority' clause in task-based
directives.

OpenMP 4.5 supports clause 'priority' in task-based directives. Patch
adds initial codegen support for this clause in codegen.

llvm-svn: 269050
2016-05-10 12:21:02 +00:00
Alexey Bataev 8a83159731 [OPENMP 4.0] Fixed codegen for destructors in task-based directives.
If private variables require destructors call at the deletion of the
task, additional flag in task flags must be set. Patch fixes this
problem.

llvm-svn: 269039
2016-05-10 10:36:51 +00:00
Alexey Bataev 9ebd742748 [OPENMP 4.5] Add codegen support in runtime for '[non]monotonic'
schedule modifiers.

Runtime library expects some additional data in schedule argument for
loop-based directives, that have additional schedule modifiers
'monotonic|nonmonotonic'.

llvm-svn: 269035
2016-05-10 09:57:36 +00:00
Samuel Antao e49645cf12 [OpenMP] Check for associated statements with hasAssociatedStmt() when scanning for device code.
Summary: `getAssociatedStmt()` contains an assertion that assumes the statement always exists. In device code scanning, we need to look into the associated statement therefore we check its existence. This patch replaces  `getAssociatedStmt` by `hasAssociatedStmt` so that we do not trigger the assertion for directives that happen not to have an associated statement (e.g target enter/exit data).

Reviewers: hfinkel, carlo.bertolli, arpith-jacob, kkwli0, ABataev

Subscribers: cfe-commits, caomhin

Differential Revision: http://reviews.llvm.org/D19812

llvm-svn: 268870
2016-05-08 06:43:56 +00:00
Alexey Bataev c7a82b41a7 [OPENMP 4.0] Codegen for 'declare simd' directive.
OpenMP 4.0 adds support for elemental functions using declarative
directive '#pragma omp declare simd'. Patch adds mangling for simd
functions in accordance with
https://sourceware.org/glibc/wiki/libmvec?action=AttachFile&do=view&target=VectorABI.txt

llvm-svn: 268721
2016-05-06 09:40:08 +00:00
Alexey Bataev f93095a003 [OPENMP 4.5] Codegen for 'lastprivate' clauses in 'taskloop' directives.
OpenMP 4.5 adds taskloop/taskloop simd directives. These directives
allow to use lastprivate clause. Patch adds codegen for this clause.

llvm-svn: 268618
2016-05-05 08:46:22 +00:00
Carlo Bertolli 6eee9061ac [OPENMP] Enable correct generation of runtime call when target directive is separated from teams directive by multiple curly brackets
http://reviews.llvm.org/D18474

This patch fixes a bug in code generation of the correct OpenMP runtime library call in presence of target and teams, when target is separated by teams with multiple curly brackets. The current implementation will not be able to see the teams directive inside target and issue a call to tgt_target instead of the correct one tgt_target_teams.

llvm-svn: 267972
2016-04-29 01:37:30 +00:00
Alexey Bataev 24b5baed27 [OPENMP] Simplified interface for codegen of tasks, NFC.
Reduced number of arguments in member functions of runtime support
library for task-based directives.

llvm-svn: 267863
2016-04-28 09:23:51 +00:00
Alexey Bataev 2b19a6fe53 [OPENMP 4.5] Codegen for 'grainsize/num_tasks' clauses of 'taskloop'
directive.

OpenMP 4.5 defines 'taskloop' directive and 2 additional clauses
'grainsize' and 'num_tasks' for this directive. Patch adds codegen for
these clauses.
These clauses are generated as arguments of the '__kmpc_taskloop'
libcall and are encoded the following way:

void __kmpc_taskloop(ident_t *loc, int gtid, kmp_task_t *task, int if_val, kmp_uint64 *lb, kmp_uint64 *ub, kmp_int64 st, int nogroup,  int sched, kmp_uint64 grainsize, void *task_dup);

If 'grainsize' is specified, 'sched' argument must be set to '1' and
'grainsize' argument must be set to the value of the 'grainsize' clause.
If 'num_tasks' is specified, 'sched' argument must be set to '2' and
'grainsize' argument must be set to the value of the 'num_tasks' clause.
It is possible because these 2 clauses are mutually exclusive and can't
be used at the same time on the same directive.
If none of these clauses is specified, 'sched' argument must be set to
'0'.

llvm-svn: 267862
2016-04-28 09:15:06 +00:00
Samuel Antao 8dd6628743 [OpenMP] Code generation for target exit data directive
Summary:
This patch adds support for the target exit data directive code generation.

Given that, apart from the employed runtime call, target exit data requires the same code generation pattern as target enter data, the OpenMP codegen entry point was renamed and reused for both.

Reviewers: hfinkel, carlo.bertolli, arpith-jacob, kkwli0, ABataev

Subscribers: cfe-commits, fraggamuffin, caomhin

Differential Revision: http://reviews.llvm.org/D17369

llvm-svn: 267814
2016-04-27 23:14:30 +00:00
Samuel Antao bd0ae2e14c [OpenMP] Code generation for target enter data directive
Summary: This patch adds support for the target enter data directive code generation.

Reviewers: hfinkel, carlo.bertolli, arpith-jacob, kkwli0, ABataev

Subscribers: cfe-commits, fraggamuffin, caomhin

Differential Revision: http://reviews.llvm.org/D17368

llvm-svn: 267812
2016-04-27 23:07:29 +00:00
Samuel Antao df158d5567 [OpenMP] Code generation for target data directive
Summary:
This patch adds support for the target data directive code generation.

Part of the already existent functionality related with data maps is moved to a new function so that it could be reused.

Reviewers: hfinkel, carlo.bertolli, arpith-jacob, kkwli0, ABataev

Subscribers: cfe-commits, fraggamuffin, caomhin

Differential Revision: http://reviews.llvm.org/D17367

llvm-svn: 267811
2016-04-27 22:58:19 +00:00
Samuel Antao 86ace55d53 [OpenMP] Map clause codegeneration.
Summary:
Implement codegen for the map clause. All the new list items in 4.5 specification are supported. 

Fix bug in the generation of array sections that was exposed by some of the map clause tests: for pointer types the offsets have to be calculated from the pointee not the pointer.

Reviewers: hfinkel, kkwli0, carlo.bertolli, arpith-jacob, ABataev

Subscribers: ABataev, cfe-commits, caomhin, fraggamuffin

Differential Revision: http://reviews.llvm.org/D16749

llvm-svn: 267808
2016-04-27 22:40:57 +00:00
Alexey Bataev 4ba78a46ff [OPENMP] Fix for codegen of captured variables in inlined directives.
Currently there is a problem with codegen of inlined directives inside
lambdas, it may cause a crash during codegen because of incorrect
capturing of variables. Patch fixes this problem.

llvm-svn: 267677
2016-04-27 07:56:03 +00:00
Alexey Bataev 7292c29bb5 [OPENMP 4.5] Codegen for 'taskloop' directive.
The taskloop construct specifies that the iterations of one or more associated loops will be executed in parallel using OpenMP tasks. The iterations are distributed across tasks created by the construct and scheduled to be executed.
The next code will be generated for the taskloop directive:
    #pragma omp taskloop num_tasks(N) lastprivate(j)
        for( i=0; i<N*GRAIN*STRIDE-1; i+=STRIDE ) {
          int th = omp_get_thread_num();
          #pragma omp atomic
            counter++;
          #pragma omp atomic
            th_counter[th]++;
          j = i;
    }

Generated code:
task = __kmpc_omp_task_alloc(NULL,gtid,1,sizeof(struct
task),sizeof(struct shar),&task_entry);
psh = task->shareds;
psh->pth_counter = &th_counter;
psh->pcounter = &counter;
psh->pj = &j;
task->lb = 0;
task->ub = N*GRAIN*STRIDE-2;
task->st = STRIDE;
__kmpc_taskloop(
NULL,             // location
gtid,             // gtid
task,             // task structure
1,                // if clause value
&task->lb,        // lower bound
&task->ub,        // upper bound
STRIDE,           // loop increment
0,                // 1 if nogroup specified
2,                // schedule type: 0-none, 1-grainsize, 2-num_tasks
N,                // schedule value (ignored for type 0)
(void*)&__task_dup_entry // tasks duplication routine
);

llvm-svn: 267395
2016-04-25 12:22:29 +00:00
Adrian Prantl 1858c664de Debug info: Apply an empty debug location for global OpenMP destructors.
LLVM really wants a debug location on every inlinable call in a function
with debug info, because it otherwise cannot set up inlining debug info.

This change applies an artificial line 0 debug location (which is how
DWARF marks automatically generated code that has no corresponding
source code) to the .__kmpc_global_dtor_. functions to avoid the
LLVM Verifier complaining.

llvm-svn: 267369
2016-04-24 22:22:29 +00:00
Alexey Bataev 48591dd98c [OPENMP] Codegen for untied tasks.
If the untied clause is present on a task construct, any thread in the
team can resume the task region after a suspension. Patch adds proper
codegen for untied tasks.

llvm-svn: 266853
2016-04-20 04:01:36 +00:00
Alexey Bataev 995e861ba6 Revert "[OPENMP] Codegen for untied tasks."
This reverts commit r266754.

llvm-svn: 266755
2016-04-19 16:36:01 +00:00
Alexey Bataev 823acfacdf [OPENMP] Codegen for untied tasks.
If the untied clause is present on a task construct, any thread in the
team can resume the task region after a suspension. Patch adds proper
codegen for untied tasks.

llvm-svn: 266754
2016-04-19 16:27:55 +00:00
Alexey Bataev bec9572213 Revert "[OPENMP] Codegen for untied tasks."
This reverts commit 266722.

llvm-svn: 266724
2016-04-19 09:27:38 +00:00
Alexey Bataev 26b2577f6b [OPENMP] Codegen for untied tasks.
If the untied clause is present on a task construct, any thread in the team can resume the task region after a suspension. Patch adds proper codegen for untied tasks.

llvm-svn: 266722
2016-04-19 09:10:27 +00:00
JF Bastien 92f4ef1017 NFC: make AtomicOrdering an enum class
Summary: See LLVM change D18775 for details, this change depends on it.

Reviewers: jyknight, reames

Subscribers: cfe-commits

Differential Revision: http://reviews.llvm.org/D18776

llvm-svn: 265569
2016-04-06 17:26:42 +00:00
Carlo Bertolli c687225b43 [OPENMP] Codegen for teams directive for NVPTX
This patch implements the teams directive for the NVPTX backend. It is different from the host code generation path as it:

Does not call kmpc_fork_teams. All necessary teams and threads are started upon touching the target region, when launching a CUDA kernel, and their execution is coordinated through sequential and parallel regions within the target region.
Does not call kmpc_push_num_teams even if a num_teams of thread_limit clause is present. Setting the number of teams and the thread limit is implemented by the nvptx-related runtime.
Please note that I am now passing a Clang Expr * to emitPushNumTeams instead of the originally chosen llvm::Value * type. The reason for that is that I want to avoid emitting expressions for num_teams and thread_limit if they are not needed in the target region.

http://reviews.llvm.org/D17963

llvm-svn: 265304
2016-04-04 15:55:02 +00:00
Alexey Bataev 14fa1c6b60 [OPENMP] Allow runtime insert its own code inside OpenMP regions.
Solution unifies interface of RegionCodeGenTy type to allow insert
runtime-specific code before/after main codegen action defined in
CGStmtOpenMP.cpp file. Runtime should not define its own RegionCodeGenTy
for general OpenMP directives, but must be allowed to insert its own
 (required) code to support target specific codegen.

llvm-svn: 264700
2016-03-29 05:34:15 +00:00
Alexey Bataev f539faa733 Revert "[OPENMP] Allow runtime insert its own code inside OpenMP regions."
Reverting because of failed tests.

llvm-svn: 264577
2016-03-28 12:58:34 +00:00
Alexey Bataev 424be92831 [OPENMP] Allow runtime insert its own code inside OpenMP regions.
Solution unifies interface of RegionCodeGenTy type to allow insert
runtime-specific code before/after main codegen action defined in
CGStmtOpenMP.cpp file. Runtime should not define its own RegionCodeGenTy
for general OpenMP directives, but must be allowed to insert its own
 (required) code to support target specific codegen.

llvm-svn: 264576
2016-03-28 12:52:58 +00:00
Alexey Bataev f662b5943c Revert "[OPENMP] Allow runtime insert its own code inside OpenMP regions."
This reverts commit 3ee791165100607178073f14531a0dc90c622b36.

llvm-svn: 264570
2016-03-28 10:12:03 +00:00
Alexey Bataev b8c425c4f7 [OPENMP] Allow runtime insert its own code inside OpenMP regions.
Solution unifies interface of RegionCodeGenTy type to allow insert
runtime-specific code before/after main codegen action defined in
CGStmtOpenMP.cpp file. Runtime should not define its own RegionCodeGenTy
for general OpenMP directives, but must be allowed to insert its own
  (required) code to support target specific codegen.

llvm-svn: 264569
2016-03-28 09:53:43 +00:00
Arpith Chacko Jacob 5c309e475d [OpenMP] Base support for target directive codegen on NVPTX device.
Summary:
This patch adds base support for codegen of the target directive on the NVPTX device.

Reviewers: ABataev

Differential Revision: http://reviews.llvm.org/D17877

Reworked test case after buildbot failure on windows.
Updated patch to integrate r263837 and test case nvptx_target_firstprivate_codegen.cpp.

llvm-svn: 264018
2016-03-22 01:48:56 +00:00
Carlo Bertolli b74bfc80a4 [OPENMP] Implementation of codegen for firstprivate clause of target directive
This patch implements the following aspects:

It extends sema to check that a variable is not reference in both a map clause and firstprivate or private. This is needed to ensure correct functioning at codegen level, apart from being useful for the user.
It implements firstprivate for target in codegen. The implementation applies to both host and nvptx devices.
It adds regression tests for codegen of firstprivate, host and device side when using the host as device, and nvptx side.
Please note that the regression test for nvptx codegen is missing VLAs. This is because VLAs currently require saving and restoring the stack which appears not to be a supported operation by nvptx backend.

It adds a check in sema regression tests for target map, firstprivate, and private clauses.

http://reviews.llvm.org/D18203

llvm-svn: 263837
2016-03-18 21:43:32 +00:00
Arpith Chacko Jacob 129fa9a048 Revert r263783 as buildbot failure is being investigated.
llvm-svn: 263784
2016-03-18 12:39:40 +00:00
Arpith Chacko Jacob ac563708ab [OpenMP] Base support for target directive codegen on NVPTX device.
Summary:
Reworked test case after buildbot failure on windows.

This patch adds base support for codegen of the target directive on the NVPTX device.

Reviewers: ABataev

Differential Revision: http://reviews.llvm.org/D17877

llvm-svn: 263783
2016-03-18 11:47:43 +00:00
Alexey Bataev a839dddf92 [OPENMP 4.0] Use 'declare reduction' constructs in 'reduction' clauses.
OpenMP 4.0 allows to define custom reduction operations using '#pragma
omp declare reduction' construct. Patch allows to use this custom
defined reduction operations in 'reduction' clauses.

llvm-svn: 263701
2016-03-17 10:19:46 +00:00
Carlo Bertolli a03acfa359 [OPENMP] Support for codegen of private clause of target, host side
This patch adds support for codegen of private clause of target and a regression test for host code generation, when the host is used as target device. I believe that code generation for nvptx backend would not require anything additional or different to what is done for the host.

http://reviews.llvm.org/D18105

llvm-svn: 263654
2016-03-16 19:04:22 +00:00
Arpith Chacko Jacob 9cb61faa61 Revert commit http://reviews.llvm.org/D17877 to fix tests on x86.
llvm-svn: 263589
2016-03-15 21:26:34 +00:00
Arpith Chacko Jacob 5e1493b560 [OpenMP] Base support for target directive codegen on NVPTX device.
Summary:
This patch adds base support for codegen of the target directive on the NVPTX device.

Reviewers: ABataev

Differential Revision: http://reviews.llvm.org/D17877

llvm-svn: 263587
2016-03-15 21:04:57 +00:00
Arpith Chacko Jacob fc46c25d74 Reverted http://reviews.llvm.org/D17877 to fix tests.
llvm-svn: 263555
2016-03-15 16:19:13 +00:00
Arpith Chacko Jacob c61744c26b [OpenMP] Base support for target directive codegen on NVPTX device.
Summary:
This patch adds base support for codegen of the target directive on the NVPTX device.

Reviewers: ABataev

Differential Revision: http://reviews.llvm.org/D17877

llvm-svn: 263552
2016-03-15 15:24:52 +00:00
John McCall c56a8b3284 Preserve ExtParameterInfos into CGFunctionInfo.
As part of this, make the function-arrangement interfaces
a little simpler and more semantic.

NFC.

llvm-svn: 263191
2016-03-11 04:30:31 +00:00
Carlo Bertolli fc35ad2bbc Reapply r262741 [OPENMP] Codegen for distribute directive
This patch provide basic implementation of codegen for teams directive, excluding all clauses except dist_schedule. It also fixes parts of AST reader/writer to enable correct pre-compiled header handling.

http://reviews.llvm.org/D17170

llvm-svn: 262832
2016-03-07 16:04:49 +00:00
Samuel Antao bf4d18d3d2 Revert r262741 - [OPENMP] Codegen for distribute directive
Was causing a failure in one of the buildbot slaves.

llvm-svn: 262744
2016-03-04 21:02:14 +00:00
Carlo Bertolli 4a56e3831d [OPENMP] Codegen for distribute directive
This patch provide basic implementation of codegen for teams directive, excluding all clauses except dist_schedule. It also fixes parts of AST reader/writer to enable correct pre-compiled header handling.

http://reviews.llvm.org/D17170

llvm-svn: 262741
2016-03-04 20:24:58 +00:00
Alexey Bataev c5b1d320b8 [OPENMP 4.0] Codegen for 'declare reduction' construct.
Emit function for 'combiner' part of 'declare reduction' construct and
'initialilzer' part, if any.

llvm-svn: 262699
2016-03-04 09:22:22 +00:00
Carlo Bertolli 430d8ecc55 Add code generation for teams directive inside target region
llvm-svn: 262652
2016-03-03 20:34:23 +00:00
Samuel Antao b68e2db8f9 [OpenMP] Code generation for teams - kernel launching
Summary:
This patch implements the launching of a target region in the presence of a nested teams region, i.e calls tgt_target_teams with the required arguments gathered from the enclosed teams directive.

The actual codegen of the region enclosed by the teams construct will be contributed in a separate patch.

Reviewers: hfinkel, arpith-jacob, kkwli0, carlo.bertolli, ABataev

Subscribers: cfe-commits, caomhin, fraggamuffin

Differential Revision: http://reviews.llvm.org/D17019

llvm-svn: 262625
2016-03-03 16:20:23 +00:00
Alexey Bataev 50b3c95992 [OPENMP] Improved layout of CGOpenMPRuntime class, NFC.
llvm-svn: 261315
2016-02-19 10:38:26 +00:00
Richard Trieu cc3949d99a Remove use of builtin comma operator.
Cleanup for upcoming Clang warning -Wcomma.  No functionality change intended.

llvm-svn: 261271
2016-02-18 22:34:54 +00:00
Samuel Antao 2de62b0c89 [OpenMP] Rename the offload entry points.
Summary:
Unlike other outlined regions in OpenMP, offloading entry points have to have be visible (external linkage) for the device side. Using dots in the names of the entries can be therefore problematic for some toolchains, e.g. NVPTX.

Also the patch drops the column information in the unique name of the entry points. The parsing of directives ignore unknown tokens, preventing several target  regions to be implemented in the same line. Therefore, the line information is sufficient for the name to be unique. Also, the preprocessor printer does not preserve the column information, causing offloading-entry detection issues if the host uses an integrated preprocessor and the target doesn't (or vice versa).

Reviewers: hfinkel, arpith-jacob, carlo.bertolli, kkwli0, ABataev

Subscribers: cfe-commits, fraggamuffin, caomhin

Differential Revision: http://reviews.llvm.org/D17179

llvm-svn: 260837
2016-02-13 23:35:10 +00:00
Eugene Zelenko 0a4f3f4373 Fix some Clang-tidy readability-redundant-control-flow warnings; other minor fixes.
Differential revision: http://reviews.llvm.org/D17060

llvm-svn: 260414
2016-02-10 19:11:58 +00:00
Alexey Bataev 31300ed0a5 [OPENMP 4.0] Fixed support of array sections/array subscripts.
Codegen for array sections/array subscripts worked only for expressions with arrays as base. Patch fixes codegen for bases with pointer/reference types.

llvm-svn: 259776
2016-02-04 11:27:03 +00:00
Benjamin Kramer 8c30592e18 Move DebugInfoKind into its own header to cut the cyclic dependency edge from Driver to Frontend.
llvm-svn: 259489
2016-02-02 11:06:51 +00:00
Eugene Zelenko 1660a5d298 Fix Clang-tidy modernize-use-nullptr warnings; other minor fixes.
Differential revision: http://reviews.llvm.org/D16567

llvm-svn: 258836
2016-01-26 19:01:06 +00:00
Alexey Bataev 1189bd0205 [OPENMP 4.5] Allow arrays in 'reduction' clause.
OpenMP 4.5, alogn with array sections, allows to use variables of array type in reductions.

llvm-svn: 258804
2016-01-26 12:20:39 +00:00
Alexey Bataev 3015bcc62a [OPENMP] Generalize codegen for 'sections'-based directive.
If 'sections' directive has only one sub-section, the code for 'single'-based directive was emitted. Removed this codegen, because it causes crashes in different cases.

llvm-svn: 258495
2016-01-22 08:56:50 +00:00
Alexey Bataev 8524d15954 [OPENMP] Fix crash on reduction for complex variables.
reworked codegen for reduction operation for complex types to avoid crash

llvm-svn: 258394
2016-01-21 12:35:58 +00:00
Alexey Bataev 9619f04c0e [OPENMP 4.0] Fix for codegen of 'cancel' directive within 'sections' directive.
Allow to emit code for 'cancel' directive within 'sections' directive with single sub-section.

llvm-svn: 258307
2016-01-20 12:29:47 +00:00
Hans Wennborg 45c7439d11 Don't store CGOpenMPRegionInfo::CodeGen as a reference (PR26078)
The referenced llvm::function_ref<void(CodeGenFunction &)> object can go
away before CodeGen is used, resulting in a crash.

llvm-svn: 257516
2016-01-12 20:54:36 +00:00
Samuel Antao ee8fb302f5 [OpenMP] Reapply rL256842: [OpenMP] Offloading descriptor registration and device codegen.
This patch attempts to fix the regressions identified when the patch was committed initially. 

Thanks to Michael Liao for identifying the fix in the offloading metadata generation 
related with side effects in evaluation of function arguments. 
 

llvm-svn: 256933
2016-01-06 13:42:12 +00:00
Samuel Antao 7d5de9a1ee [OpenMP] Revert rL256842: [OpenMP] Offloading descriptor registration and device codegen.
It was causing two regression, so I'm reverting until the cause is found.

llvm-svn: 256858
2016-01-05 19:16:13 +00:00
Samuel Antao 4d5f0bbea1 [OpenMP] Offloading descriptor registration and device codegen.
Summary:
In order to offloading work properly two things need to be in place:
- a descriptor with all the offloading information (device entry functions, and global variable) has to be created by the host and registered in the OpenMP offloading runtime library.
- all the device functions need to be emitted for the device and a convention has to be in place so that the runtime library can easily map the host ID of an entry point with the actual function in the device.

This patch adds support for these two things. However, only entry functions are being registered given that 'declare target' directive is not yet implemented.

About offloading descriptor:

The details of the descriptor are explained with more detail in http://goo.gl/L1rnKJ. Basically the descriptor will have fields that specify the number of devices, the pointers to where the device images begin and end (that will be defined by the linker), and also pointers to a the begin and end of table whose entries contain information about a specific entry point. Each entry has the type:
```
struct __tgt_offload_entry{
 void *addr;
 char *name;
 int64_t size;
};
```  
and will be implemented in a pre determined (ELF) section `.omp_offloading.entries` with 1-byte alignment, so that when all the objects are linked, the table is in that section with no padding in between entries (will be like a C array). The code generation ensures that all `__tgt_offload_entry` entries are emitted in the same order for both host and device so that the runtime can have the corresponding entries in both host and device in same index of the table, and efficiently implement the mapping.

The resulting descriptor is registered/unregistered with the runtime library using the calls `__tgt_register_lib` and `__tgt_unregister_lib`. The registration is implemented in a high priority global initializer so that the registration happens always before any initializer (that can potentially include target regions) is run.

The driver flag -omptargets= was created to specify a comma separated list of devices the user wants to support so that the new functionality can be exercised. Each device is specified with its triple.


About target codegen:

The target codegen is pretty much straightforward as it reuses completely the logic of the host version for the same target region. The tricky part is to identify the meaningful target regions in the device side. Unlike other programming models, like CUDA, there are no already outlined functions with attributes that mark what should be emitted or not. So, the information on what to emit is passed in the form of metadata in host bc file. This requires a new option to pass the host bc to the device frontend. Then everything is similar to what happens in CUDA: the global declarations emission is intercepted to check to see if it is an "interesting" declaration. The difference is that instead of checking an attribute, the metadata information in checked. Right now, there is only a form of metadata to pass information about the device entry points (target regions). A class `OffloadEntriesInfoManagerTy` was created to manage all the information and queries related with the metadata. The metadata looks like this:
```
!omp_offload.info = !{!0, !1, !2, !3, !4, !5, !6}

!0 = !{i32 0, i32 52, i32 77426347, !"_ZN2S12r1Ei", i32 479, i32 13, i32 4}
!1 = !{i32 0, i32 52, i32 77426347, !"_ZL7fstatici", i32 461, i32 11, i32 5}
!2 = !{i32 0, i32 52, i32 77426347, !"_Z9ftemplateIiET_i", i32 444, i32 11, i32 6}
!3 = !{i32 0, i32 52, i32 77426347, !"_Z3fooi", i32 99, i32 11, i32 0}
!4 = !{i32 0, i32 52, i32 77426347, !"_Z3fooi", i32 272, i32 11, i32 3}
!5 = !{i32 0, i32 52, i32 77426347, !"_Z3fooi", i32 127, i32 11, i32 1}
!6 = !{i32 0, i32 52, i32 77426347, !"_Z3fooi", i32 159, i32 11, i32 2}
```
The fields in each metadata entry are (in sequence):
Entry 1) an ID of the type of metadata - right now only zero is used meaning "OpenMP target region".
Entry 2) a unique ID of the device where the input source file that contain the target region lives. 
Entry 3) a unique ID of the file where the input source file that contain the target region lives. 
Entry 4) a mangled name of the function that encloses the target region.
Entries 5) and 6) line and column number where the target region was found.
Entry 7) is the order the entry was emitted.

Entry 2) and 3) are required to distinguish files that have the same function name.
Entry 4) is required to distinguish different instances of the same declaration (usually templated ones)
Entries 5) and 6) are required to distinguish the particular target region in body of the function (it is possible that a given target region is not an entry point - if clause can evaluate always to zero - and therefore we need to identify the "interesting" target regions. )

This patch replaces http://reviews.llvm.org/D12306.

Reviewers: ABataev, hfinkel, tra, rjmccall, sfantao

Subscribers: FBrygidyn, piotr.rak, Hahnfeld, cfe-commits

Differential Revision: http://reviews.llvm.org/D12614

llvm-svn: 256842
2016-01-05 16:23:04 +00:00
Alexey Bataev a636c7f9b9 [OPENMP 4.5] Parsing/sema for 'depend(sink:vec)' clause in 'ordered' directive.
OpenMP 4.5 adds 'depend(sink:vec)' in 'ordered' directive for doacross loop synchronization. Patch adds parsing and semantic analysis for this clause.

llvm-svn: 256330
2015-12-23 10:27:45 +00:00
Alexey Bataev 29c9209357 [OPENMP] Revert r256238 to fix the problem with tests on Linux.
llvm-svn: 256239
2015-12-22 12:44:46 +00:00
Alexey Bataev ef4c5584d5 [OPENMP 4.5] Parsing/sema for 'depend(sink:vec)' clause in 'ordered' directive.
OpenMP 4.5 adds 'depend(sink:vec)' in 'ordered' directive for doacross loop synchronization. Patch adds parsing and semantic analysis for this clause.

llvm-svn: 256238
2015-12-22 12:21:47 +00:00
Alexey Bataev 8ef3141127 [OPENMP] Fix for http://llvm.org/PR25878: Error compiling an OpenMP program
OpenMP codegen tried to emit the code for its constructs even if it was detected as a dead-code. Added checks to ensure that the code is emitted if the code is not dead.

llvm-svn: 255990
2015-12-18 07:58:25 +00:00
Alexey Bataev eb48235033 [OPENMP 4.5] Parsing/sema analysis for 'depend(source)' clause in 'ordered' directive.
OpenMP 4.5 adds 'depend(source)' clause for 'ordered' directive to support cross-iteration dependence. Patch adds parsing and semantic analysis for this construct.

llvm-svn: 255986
2015-12-18 05:05:56 +00:00
Alexey Bataev fc57d1601d [OPENMP 4.5] Codegen for 'hint' clause of 'critical' directive
OpenMP 4.5 defines 'hint' clause for 'critical' directive. Patch adds codegen for this clause.

llvm-svn: 255639
2015-12-15 10:55:09 +00:00
Samuel Antao 4af1b7b693 [OpenMP] Update target directive codegen to use 4.5 implicit data mappings.
Summary:
This patch implements the 4.5 specification for the implicit data maps. OpenMP 4.5 specification changes the default way data is captured into a target region. All the non-aggregate kinds are passed by value by default. This required activating the capturing by value during SEMA for the target region. All the non-aggregate values that can be encoded in the size of a pointer are properly casted and forwarded to the runtime library. On top of fixing the previous weird behavior for mapping pointers in nested data regions (an explicit map was always required), this also improves performance as the number of allocations/transactions to the device per non-aggregate map are reduced from two to only one - instead of passing a reference and the value, only the value passed.

Explicit maps will be added later on once firstprivate, private, and map clauses' SEMA and parsing are available.

Reviewers: hfinkel, rjmccall, ABataev

Subscribers: cfe-commits, carlo.bertolli

Differential Revision: http://reviews.llvm.org/D14940

llvm-svn: 254521
2015-12-02 17:44:43 +00:00
Alexey Bataev 40e36f1f64 [OPENMP] Fix crash on codegen for 'task' directive with no shared variables.
If 'task' region does not have shared variables codegen could crash on calculation of size of list of shared variables.

llvm-svn: 253977
2015-11-24 13:01:44 +00:00
Alexey Bataev 92e82f9cce [OPENMP] 'out' dependency for 'task' directives must be the same as 'inout'.
Runtime library requires, that codegen for 'depend' clause for 'out' dependency kind must be the same as codegen for 'depend' clause with 'inout' dependency.

llvm-svn: 253866
2015-11-23 13:33:42 +00:00
Akira Hatanaka 7791f1a4a9 [CodeGen] Call SetInternalFunctionAttributes to attach function
attributes to internal functions.

This patch fixes CodeGenModule::CreateGlobalInitOrDestructFunction to
use SetInternalFunctionAttributes instead of SetLLVMFunctionAttributes
to attach function attributes to internal functions.

Also, make sure the correct CGFunctionInfo is passed instead of always
passing what arrangeNullaryFunction returns.

rdar://problem/20828324

Differential Revision: http://reviews.llvm.org/D13610

llvm-svn: 251734
2015-10-31 01:28:07 +00:00
Benjamin Kramer e003ca2a03 Put global classes into the appropriate namespace.
Most of the cases belong into an anonymous namespace. No functionality
change intended.

llvm-svn: 251514
2015-10-28 13:54:16 +00:00
Akira Hatanaka 44a59f8976 [CodeGen] Attach function attributes to Objective-C and OpenMP
functions.

This commit fixes a bug in CGOpenMPRuntime.cpp and CGObjC.cpp where
some of the function attributes are not attached to newly created
functions.

rdar://problem/20828324

Differential Revision: http://reviews.llvm.org/D13928

llvm-svn: 251476
2015-10-28 02:30:47 +00:00
Alexey Bataev f24e7b1f60 [OPENMP 4.1] Codegen for array sections/subscripts in 'reduction' clause.
OpenMP 4.1 adds support for array sections/subscripts in 'reduction' clause. Patch adds codegen for this feature.

llvm-svn: 249672
2015-10-08 09:10:53 +00:00
Samuel Antao bed3c46632 [OpenMP] Target directive host codegen.
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
2015-10-02 16:14:20 +00:00
Craig Topper 8674c5cf70 Remove 'const' from some ArrayRef arguments since they're passed by value anyway. NFC
llvm-svn: 248774
2015-09-29 04:30:07 +00:00
Alexey Bataev 5f600d6a49 [OPENMP 4.1] Codegen for ‘simd’ clause in ‘ordered’ directive.
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
2015-09-29 03:48:57 +00:00
Alexey Bataev 87933c7ced [OPENMP 4.0] Add 'if' clause for 'cancel' directive.
Add parsing, sema analysis and codegen for 'if' clause in 'cancel' directive.

llvm-svn: 247976
2015-09-18 08:07:34 +00:00
Alexey Bataev 25e5b44654 [OPENMP] Emit __kmpc_cancel_barrier() and code for 'cancellation point' only if 'cancel' is found.
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
2015-09-15 12:52:43 +00:00
Evgeniy Stepanov 6b2a61d3a5 Revert "Always_inline codegen rewrite" and 2 follow-ups.
Revert "Update cxx-irgen.cpp test to allow signext in alwaysinline functions."
Revert "[CodeGen] Remove wrapper-free always_inline functions from COMDATs"
Revert "Always_inline codegen rewrite."

Reason for revert: PR24793.

llvm-svn: 247620
2015-09-14 21:35:16 +00:00
Evgeniy Stepanov 93db40a147 Always_inline codegen rewrite.
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
2015-09-12 01:07:37 +00:00
Evgeniy Stepanov 67037ee21e Revert "Specify target triple in alwaysinline tests."
Revert "Always_inline codegen rewrite."

Breaks gdb & lldb tests.
Breaks on Fedora 22 x86_64.

llvm-svn: 247491
2015-09-11 23:48:37 +00:00
Evgeniy Stepanov 072e83500e Always_inline codegen rewrite.
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
2015-09-11 20:29:07 +00:00