If the error is generated during analysis of implicitly or explicitly
mapped variables, it may cause compiler crash because of incorrect
analysis.
llvm-svn: 319774
Though it is incorrect from point of view of OpenMP standard to have
dependent iteration space in OpenMP loops, compiler should not crash.
Patch fixes this problem.
llvm-svn: 319700
Previously we emitted `__tgt_target_teams` only for standalone teams
directives. This patch allows emit this function for all teams-based
directives.
llvm-svn: 319585
distribute directives.
OpenMP standard does not allow to mark the variables as firstprivate and lastprivate at the same time in distribute-based directives. Patch fixes this problem.
llvm-svn: 319560
Clang asserts on undeclared variables on the to or link clause in the declare
target directive. The patch is to properly diagnose the error.
// foo1 and foo2 are not declared
#pragma omp declare target to(foo1)
#pragma omp declare target link(foo2)
Differential Revision: https://reviews.llvm.org/D40588
llvm-svn: 319458
teams region.
If the inner teams region is not correct, it may cause an assertion when
processing outer target region. Patch fixes this problem.
llvm-svn: 319450
directives.
According to the OpenMP standard, only loop control variables can be
used in linear clauses of distribute-based simd directives.
llvm-svn: 319362
In the future the compiler will analyze whether the OpenMP
runtime needs to be (fully) initialized and avoid that overhead
if possible. The functions already take an argument to transfer
that information to the runtime, so pass in the default value 1.
(This is needed for binary compatibility with libomptarget-nvptx
currently being upstreamed.)
Differential Revision: https://reviews.llvm.org/D40354
llvm-svn: 318836
This clang patch changes the __tgt_* API function signatures in preparation for the new map interface.
Changes are: Device IDs 32bits --> 64bits, Flags 32bits --> 64bits
Differential revision: https://reviews.llvm.org/D40281
llvm-svn: 318789
Summary:
This patch is part of the development effort to add support in the current OpenMP GPU offloading implementation for implicitly sharing variables between a target region executed by the team master thread and the worker threads within that team.
This patch is the first of three required for successfully performing the implicit sharing of master thread variables with the worker threads within a team. The remaining two patches are:
- Patch D38978 to the LLVM NVPTX backend which ensures the lowering of shared variables to an device memory which allows the sharing of references;
- Patch (coming soon) is a patch to libomptarget runtime library which ensures that a list of references to shared variables is properly maintained.
A simple code snippet which illustrates an implicit data sharing situation is as follows:
```
#pragma omp target
{
// master thread only
int v;
#pragma omp parallel
{
// worker threads
// use v
}
}
```
Variable v is implicitly shared from the team master thread which executes the code in between the target and parallel directives. The worker threads must operate on the latest version of v, including any updates performed by the master.
The code generated in this patch relies on the LLVM NVPTX patch (mentioned above) which prevents v from being lowered in the thread local memory of the master thread thus making the reference to this variable un-shareable with the workers. This ensures that the code generated by this patch is correct.
Since the parallel region is outlined the passing of arguments to the outlined regions must preserve the original order of arguments. The runtime therefore maintains a list of references to shared variables thus ensuring their passing in the correct order. The passing of arguments to the outlined parallel function is performed in a separate function which the data sharing infrastructure constructs in this patch. The function is inlined when optimizations are enabled.
Reviewers: hfinkel, carlo.bertolli, arpith-jacob, Hahnfeld, ABataev, caomhin
Reviewed By: ABataev
Subscribers: cfe-commits, jholewinski
Differential Revision: https://reviews.llvm.org/D38976
llvm-svn: 318773
https://reviews.llvm.org/D40187
This patch implements code gen for 'teams distribute parallel for' on the host, including all its clauses and related regression tests.
llvm-svn: 318692
Some target devices (e.g. Nvidia GPUs) don't support dynamic stack
allocation and hence no VLAs. Print errors with description instead
of failing in the backend or generating code that doesn't work.
This patch handles explicit uses of VLAs (local variable in target
or declare target region) or implicitly generated (private) VLAs
for reductions on VLAs or on array sections with non-constant size.
Differential Revision: https://reviews.llvm.org/D39505
llvm-svn: 318601
Summary:
[OpenMP] diagnose assign to firstprivate const
Clang does not diagnose assignments to const variables declared
firstprivate. Furthermore, codegen is broken such that, at run time,
such assignments simply have no effect. For example, the following
prints 0 not 1:
int main() {
const int i = 0;
#pragma omp parallel firstprivate(i)
{ i=1; printf("%d\n", i); }
return 0;
}
This commit makes these assignments a compile error, which is
consistent with other OpenMP compilers I've tried (pgcc 17.4-0, gcc
6.3.0).
Reviewers: ABataev
Reviewed By: ABataev
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D39859
llvm-svn: 317891
If the thread id is requested in windows mode within funclets, we may
generate incorrect function call that could lead to broken codegen.
llvm-svn: 317208
We can generate constant sized arrays whenever the array section has constant
length, even if the base expression itself is a VLA.
Differential Revision: https://reviews.llvm.org/D39504
llvm-svn: 317207
In some cases the compiler can deduce the length of an array section
as constants. With this information, VLAs can be avoided in place of
a constant sized array or even a scalar value if the length is 1.
Example:
int a[4], b[2];
pragma omp parallel reduction(+: a[1:2], b[1:1])
{ }
For chained array sections, this optimization is restricted to cases
where all array sections except the last have a constant length 1.
This trivially guarantees that there are no holes in the memory region
that needs to be privatized.
Example:
int c[3][4];
pragma omp parallel reduction(+: c[1:1][1:2])
{ }
This relands commit r316229 that I reverted in r316235 because it
failed on some bots. During investigation I found that this was because
Clang and GCC evaluate the two arguments to emplace_back() in
ReductionCodeGen::emitSharedLValue() in a different order, hence
leading to a different order of generated instructions in the final
LLVM IR. Fix this by passing in the arguments from temporary variables
that are evaluated in a defined order.
Differential Revision: https://reviews.llvm.org/D39136
llvm-svn: 316362
In some cases the compiler can deduce the length of an array section
as constants. With this information, VLAs can be avoided in place of
a constant sized array or even a scalar value if the length is 1.
Example:
int a[4], b[2];
pragma omp parallel reduction(+: a[1:2], b[1:1])
{ }
For chained array sections, this optimization is restricted to cases
where all array sections except the last have a constant length 1.
This trivially guarantees that there are no holes in the memory region
that needs to be privatized.
Example:
int c[3][4];
pragma omp parallel reduction(+: c[1:1][1:2])
{ }
Differential Revision: https://reviews.llvm.org/D39136
llvm-svn: 316229
If the variables is boolean and we generating inner function with real
types, the codegen may crash because of not loading boolean value from
memory.
llvm-svn: 316011
reduction.
If the reduction is an array or an array section and reduction operation
is declare reduction without initializer, it may lead to crash.
llvm-svn: 315611
in C.
If we try to get the lvalue for thread_id variables in inlined regions,
we did not use the correct version of function. Fixed this bug by adding
overrided version of the function getThreadIDVariableLValue for inlined
regions.
llvm-svn: 315578
If both taskloop and task directives are used at the same time in one
program, we may ran into the situation when the particular type for task
directive is reused for taskloop directives. Patch fixes this problem.
llvm-svn: 315464
Previously we may erroneously try to capture locally declared static
variables, which will lead to crash for target-based constructs.
Patch fixes this problem.
llvm-svn: 315076
In C++11 variable to global variables are considered as constant
expressions and these variables are not captured in the outlined
regions. Patch allows capturing of such variables in the OpenMP regions.
llvm-svn: 315074
If the `defaultmap(tofrom:scalar)` clause is specified, the scalars must
be mapped with 'tofrom' modifiers, otherwise they must be captured as
firstprivates.
llvm-svn: 314995
https://reviews.llvm.org/D38371
This patch implements codegen for the combined 'teams distribute" OpenMP pragma and adds regression tests for all its clauses.
llvm-svn: 314905
declaration.
Patch allows using of the `#pragma omp declare target`| `#pragma omp end
declare target` directives inside the structures if we need to mark as
declare target only some static members.
llvm-svn: 314833
directives.
The argument of the `device` clause in target-based executable
directives must be captured to support codegen for the `target`
directives with the `depend` clauses.
llvm-svn: 314686
Simplified and generalized codegen for non-offloading part that works if
offloading is failed or condition of the `if` clause is `false`.
llvm-svn: 314670
Summary: Test for checking if the mapping is performed correctly. This is a test initially included in Patch https://reviews.llvm.org/D29905
Reviewers: Hahnfeld, carlo.bertolli, caomhin, ABataev
Reviewed By: Hahnfeld
Subscribers: tra, cfe-commits
Differential Revision: https://reviews.llvm.org/D38040
llvm-svn: 314303
Summary: Test for checking if the mapping is performed correctly. This is a test initially included in Patch https://reviews.llvm.org/D29905
Reviewers: Hahnfeld, carlo.bertolli, caomhin
Reviewed By: Hahnfeld
Subscribers: tra, cfe-commits
Differential Revision: https://reviews.llvm.org/D38040
llvm-svn: 314228
Summary: Test for checking if the mapping is performed correctly. This is a test initially included in Patch https://reviews.llvm.org/D29905
Reviewers: Hahnfeld, carlo.bertolli, caomhin
Reviewed By: Hahnfeld
Subscribers: tra, cfe-commits
Differential Revision: https://reviews.llvm.org/D38040
llvm-svn: 314210
directives.
If the variable is used in the target-based region but is not found in
any private|mapping clause, then generate implicit firstprivate|map
clauses for these implicitly mapped variables.
llvm-svn: 314205
If the captured variable has re-declaration we may end up with the
situation where the captured variable is the re-declaration while the
referenced variable is the canonical declaration (or vice versa). In
this case we may generate wrong code. Patch fixes this situation.
llvm-svn: 313995
This is to fix PR31620. MaxAtomicInlineWidth is set to 128 for x86_64. However
for target without cx16 support, 128 atomic operation will generate __sync_*
libcalls. The patch set MaxAtomicInlineWidth to 64 if the target doesn't support
cx16.
Differential Revision: https://reviews.llvm.org/D38046
llvm-svn: 313992
If the captured variable has some redeclarations we may run into the
situation where the redeclaration is used instead of the canonical
declaration and we may consider this variable as one not captured
before.
llvm-svn: 313880
When the value specified for n in ordered(n) is larger than the number of loops a segmentation fault can occur in one of two ways when attempting to print out a diagnostic for an associated depend(sink : vec):
1) The iteration vector vec contains less than n items
2) The iteration vector vec contains a variable that is not a loop control variable
This patch addresses both of these issues.
Differential Revision: https://reviews.llvm.org/D38049
llvm-svn: 313675
__kmpc_for_static_fini().
Added special flags for calls of __kmpc_for_static_fini(), like previous
ly for __kmpc_for_static_init(). Added flag OMP_IDENT_WORK_DISTRIBUTE
for distribute cnstruct, OMP_IDENT_WORK_SECTIONS for sections-based
constructs and OMP_IDENT_WORK_LOOP for loop-based constructs in
location flags.
llvm-svn: 312642
move constructor.
Previously user-defined reduction initializer was considered as an
assignment expression, not as initializer. Fixed this by treating the
initializer expression as an initializer.
llvm-svn: 312638
step>1.
If the loop is a loot with random access iterators and the iteration
construct is represented it += n, then the compiler crashed because of
reusing of the same MaterializedTemporaryExpr around N. Patch fixes it
by using the expression as written, without any special kind of
wrappings.
llvm-svn: 312292
Capturing of the global variables occurs only in target regions. Patch
fixes it and allows capturing of globals in all target executable
directives.
llvm-svn: 312024
Summary:
Most DIExpressions are empty or very simple. When they are complex, they
tend to be unique, so checking them inline is reasonable.
This also avoids the need for CodeGen passes to append to the
llvm.dbg.mir named md node.
See also PR22780, for making DIExpression not be an MDNode.
Reviewers: aprantl, dexonsmith, dblaikie
Subscribers: qcolombet, javed.absar, eraman, hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D37075
llvm-svn: 311594
of class fails to map class static variable.
If the global variable is captured and it has several redeclarations,
sometimes it may lead to a compiler crash. Patch fixes this by working
only with canonical declarations.
llvm-svn: 311479
If worksharing construct has at least one linear item, an implicit
synchronization point must be emitted to avoid possible conflict with
the loading/storing values to the original variables. Added implicit
barrier if the linear item is found before actual start of the
worksharing construct.
llvm-svn: 311013
If exceptions are enabled, there may be a problem with the codegen of
the finalization functions from OpenMP runtime. It happens because of
the problem with the getting of thread identifier value. Patch tries to
fix it by using the result of the call of function
__kmpc_global_thread_num() rather than loading of value of outlined
function parameter.
llvm-svn: 311007
When translating arguments for NVPTX target it is not taken into account
that function may have variable number of arguments. Patch fixes this
problem.
llvm-svn: 310920
__kmpc_for_static_init().
OpenMP 5.0 will include OpenMP Tools interface that requires distinguishing different worksharing constructs.
Since the same entry point (__kmp_for_static_init(ident_t *loc,
kmp_int32 global_tid,........)) is called in case static
loop/sections/distribute it is suggested using 'flags' field of the
ident_t structure to pass the type of the construct.
llvm-svn: 310865
name.
If the host code is compiled with the debug info, while the target
without, there is a problem that the compiler is unable to find the
debug wrapper. Patch fixes this problem by emitting special name for the
debug version of the code.
llvm-svn: 310511
Converting a _Complex type to a real one simply discards the imaginary part.
This can easily lead to loss of information so for safety (and GCC
compatibility) this patch disallows that when the conversion would be implicit.
The one exception is bool, which actually compares both real and imaginary
parts and so is safe.
llvm-svn: 310427
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
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: 310377
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: 310360
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: 310104
'#pragma pack (pop)' and suspicious uses of '#pragma pack' in included files
The second recommit (r309106) was reverted because the "non-default #pragma
pack value chages the alignment of struct or union members in the included file"
warning proved to be too aggressive for external projects like Chromium
(https://bugs.chromium.org/p/chromium/issues/detail?id=749197). This recommit
makes the problematic warning a non-default one, and gives it the
-Wpragma-pack-suspicious-include warning option.
The first recommit (r308441) caused a "non-default #pragma pack value might
change the alignment of struct or union members in the included file" warning
in LLVM itself. This recommit tweaks the added warning to avoid warnings for
#includes that don't have any records that are affected by the non-default
alignment. This tweak avoids the previously emitted warning in LLVM.
Original message:
This commit adds a new -Wpragma-pack warning. It warns in the following cases:
- When a translation unit is missing terminating #pragma pack (pop) directives.
- When entering an included file if the current alignment value as determined
by '#pragma pack' directives is different from the default alignment value.
- When leaving an included file that changed the state of the current alignment
value.
rdar://10184173
Differential Revision: https://reviews.llvm.org/D35484
llvm-svn: 309386
Alexey Bataev