Commit Graph

4259 Commits

Author SHA1 Message Date
Kazu Hirata 8afcfbfb8f Use true/false instead of 1/0 (NFC)
Identified by modernize-use-bool-literals.
2022-01-09 12:21:06 -08:00
Kazu Hirata b12fd13812 Fix bugprone argument comments.
Identified by bugprone-argument-comment.
2022-01-09 12:21:02 -08:00
Kazu Hirata fb7cf90071 Use nullptr instead of 0 or NULL (NFC)
Identified with modernize-use-nullptr.
2022-01-07 10:17:29 -08:00
Kazu Hirata 42a4f5103b [Transform] Remove redundant declaration PollyAllowFullFunction (NFC)
The variable is declared in ScopDetection.h, which ScopInliner.cpp
includes.

Identified by readability-redundant-declaration.
2022-01-02 23:08:40 -08:00
Kazu Hirata e7774f499b Use static_assert instead of assert (NFC)
Identified with misc-static-assert.
2021-12-26 14:26:44 -08:00
Kazu Hirata 2d303e6781 Remove redundant return and continue statements (NFC)
Identified with readability-redundant-control-flow.
2021-12-24 23:17:54 -08:00
Kazu Hirata 76f0f1cc5c Use {DenseSet,SetVector,SmallPtrSet}::contains (NFC) 2021-12-24 21:43:06 -08:00
Kazu Hirata a8cbddc994 [CodeGen] Fix a memory leak 2021-12-24 19:51:10 -08:00
Michael Kruse 937b00ab2c [Polly][SchedOpt] Account for prevectorization of multiple statements.
A prevectorized loop may contain multiple statements, in which case
isl_schedule_node_band_sink will sink the vector band to multiple
leaves. Instead of statically assuming a specific tree structure after
sinking, add a SIMD marker to all inner bands.

Fixes llvm.org/PR52637
2021-12-23 14:06:41 -06:00
Stephan T. Lavavej 8bd106a891 [NFC] Fix typos in release notes.
Reviewed By: ldionne, Mordante, MaskRay

Differential Revision: https://reviews.llvm.org/D115685
2021-12-14 14:19:42 -08:00
Fangrui Song 0b39ec835d [Polly][Isl] Fix -Wsign-compare after D113101 2021-11-11 00:17:52 -08:00
Quinn Pham c3b15b71ce [NFC] Inclusive Language: change master to main for .chm files
[NFC] As part of using inclusive language within the llvm project,
this patch replaces master with main when referring to `.chm` files.

Reviewed By: teemperor

Differential Revision: https://reviews.llvm.org/D113299
2021-11-08 08:23:04 -06:00
Arthur Eubanks 7f62759697 [polly] Properly create and initialize new PM analysis managers
If we don't properly initialize all the analysis managers, we may be
missing analyses that other analyses depend on.

Fixes broken polly test, e.g.
https://lab.llvm.org/buildbot/#/builders/10/builds/7501.
2021-11-05 09:32:54 -07:00
Riccardo Mori 44596fe6a9 [Polly][Isl] Use the function unsignedFromIslSize to manage a isl::size object. NFCI
This is part of an effort to reduce the differences between the custom C++ bindings used right now by polly in lib/External/isl/include/isl/isl-noxceptions.h and the official isl C++ interface.
In the official interface the type `isl::size` cannot be casted to an unsigned without previously having checked if it contains a valid value with the function `isl::size::is_error()`.
For this reason two helping functions have been added:
 - `IslAssert`: assert that no errors are present in debug builds and just disables the mandatory error check in non-debug builds
 - `unisgnedFromIslSIze`: cast the `isl::size` object to `unsigned`

Changes made:
 - Add the functions `IslAssert` and `unsignedFromIslSize`
 - Add the utility function `rangeIslSize()`
 - Retype `MaxDisjunctsInDomain` from `int` to `unsigned`
 - Retype `RunTimeChecksMaxAccessDisjuncts` from `int` to `unsigned`
 - Retype `MaxDimensionsInAccessRange` from `int` to `unsigned`
 - Replaced some usages of `isl_size` to `unsigned` since we aim not to use `isl_size` anymore
 - `isl-noexceptions.h` has been generated by e704f73c88

No functional change intended.

Reviewed By: Meinersbur

Differential Revision: https://reviews.llvm.org/D113101
2021-11-05 11:15:22 +01:00
Kazu Hirata 8568ca789e Use llvm::erase_if (NFC) 2021-10-18 09:33:42 -07:00
Michael Kruse da2e1f622d [Polly][docs] Fix Sphinx warning.
ReStructured Text is not Markdown.
2021-10-15 21:13:43 -05:00
Max Fan 5bf32a0e48 [Polly] Remove checkIslAstExprInt and use RAII instead of manually freeing Expr. NFC.
Polly is trying to move towards using isl::ast_expr / isl-noexceptions.h
(which implements RAII) where possible instead of manually managing memory.
checkIslAstExprInt manually frees Expr, so it has been removed to be
more idiomatic and consistent.

Reviewed By: Meinersbur

Differential Revision: https://reviews.llvm.org/D111769
2021-10-15 09:17:00 -05:00
Michael Kruse 19db33c06e [Polly] Remove support for code generated by gfortran+DragonEgg.
DragonEgg is not maintained anymore, hence there is no need for this
functionality.

Fixes llvm.org/PR52173
2021-10-14 14:12:06 -05:00
Michael Kruse a5e52ce3f2 [Polly][docs] Fix itemize list for release notes.
Make the changes top-level items, instead of subitems of the
"Changes..." placeholder.
2021-10-14 13:50:18 -05:00
Michael Kruse 5f668bba55 [Polly] Clean up Polly's getting started docs.
This patch removes the broken bash scipt (polly.sh) and fixes the broken setup
instructions in get_started.html. It also adds instructions for using Ninja and
links to the LLVM getting started page.

Reviewed By: Meinersbur, InnovativeInventor

Differential Revision: https://reviews.llvm.org/D111685
2021-10-14 12:26:57 -05:00
Michael Kruse fa6c5107c3 [Polly] Fix unused variable in non-assert builds. NFC. 2021-10-13 00:20:22 -05:00
Michael Kruse ec2029f986 [Polly] Do not inline dumpIslObj methods. NFC.
Instead of being inline and having a neverCalled() workaround to make it
work in the debugger, define it as a regular exported function.

Also add overloads for the C API types isl_* so it works with managed as
well as unmanaged ISL objects.
2021-10-12 23:52:36 -05:00
Michael Kruse 203c7fab73 [Polly] Fix test case fixing the colon.
Commit 573531fb1f fixed the colon at the
end of a CHECK line (was a semicolon by mistake). With the check
enabled, it turned out that it was failing. Check for the correct
content.

Also add the missing colon to the next CHECK line.
2021-10-08 22:46:55 -05:00
Qiu Chaofan 573531fb1f Fix typo of colon to semicolon in lit tests 2021-10-09 10:03:50 +08:00
Michael Kruse 64489255be [Polly] Add greedy fusion algorithm.
When the option -polly-loopfusion-greedy is set, the ScheduleOptimizer
tries to aggressively fuse any band it can and does not violate any
dependences.

As part if the implementation, the functionalty for copying a band
into an new schedule was extracted out of the ScheduleTreeRewriter.
2021-10-08 20:33:30 -05:00
Reid Kleckner 89b57061f7 Move TargetRegistry.(h|cpp) from Support to MC
This moves the registry higher in the LLVM library dependency stack.
Every client of the target registry needs to link against MC anyway to
actually use the target, so we might as well move this out of Support.

This allows us to ensure that Support doesn't have includes from MC/*.

Differential Revision: https://reviews.llvm.org/D111454
2021-10-08 14:51:48 -07:00
Reid Kleckner 46ef2e0bf9 Update polly for removal of CallInst::arg_operands/getNumArgOperands
Fixes polly build
2021-10-08 10:46:05 -07:00
Michael Kruse cb879d00d8 [Polly] Completely remove -polly-opt-fusion.
This was missing from 07e7cb9433.
The switch did nothing since then.
2021-10-08 02:10:34 -05:00
Simon Pilgrim f1be391bed [polly] Replace report_fatal_error(std::string) uses with report_fatal_error(Twine)
As described on D111049, we're trying to remove the <string> dependency from error handling and replace uses of report_fatal_error(const std::string&) with the Twine() variant which can be forward declared.
2021-10-06 13:32:57 +01:00
Christopher Tetreault 67acc772d0 [NFC] Fix build failure in ScopDetection
In some build environments, the C++ compiler is unable to infer the
correct type for the DenseMap::insert in isErrorBlock. Typing out
std::make_pair helps.
2021-10-04 09:19:27 -07:00
Philip Reames d02db32644 [SCEV] Use full logic when infering flags on add and gep
This is a followon to D109845. With that landed, we will have fixed all known instances of pr51817, and can thus start inferring flags more aggressively with greatly reduced risk of miscompiles. This patch simply applies the same inference logic used in that patch to our other major flag inference path.

We can still do much better here (on both paths), but this is our first step.

Differential Revision: https://reviews.llvm.org/D111003
2021-10-03 15:32:15 -07:00
Philip Reames 2ca8a3f213 [SCEV] Stop blindly propagating flags from inbound geps to SCEV nodes
This fixes a violation of the wrap flag rules introduced in c4048d8f. This was also noted in the (very old) PR23527.

The issue being fixed is that we assume the inbound flag on any GEP assumes that all users of *any* gep (or add) which happens to map to that SCEV would also be UB if the (other) gep overflowed. That's simply not true.

In terms of the test diffs, I don't see anything seriously problematic. The lost flags are expected (given the semantic restriction on when its legal to tag the SCEV), and there are several cases where the previously inferred flags are unsound per the new semantics.

The only common trend I noticed when looking at the deltas is that by not considering branch on poison as immediate UB in ValueTracking, we do miss a few cases we could reclaim. We may be able to claw some of these back with the follow ideas mentioned in PR51817.

It's worth noting that most of the changes are analysis result only changes. The two transform changes are pretty minimal. In one case, we miss the opportunity to infer a nuw (correctly). In the other, we fail to fold an exit and produce a loop invariant form instead. This one is probably over-reduced as the program appears to be undefined in practice, and neither before or after exploits that.

Differential Revision: https://reviews.llvm.org/D109789
2021-10-01 16:30:44 -07:00
Roman Gareev 113fa82c3c [Polly] Check the properties of accesses to operands of a matrix-matrix
multiplication

The following code modifies elements of the array D.

    for (i = 0; i < _PB_NI; i++)
      for (j = 0; j < _PB_NJ; j++)
      {
        for (k = 0; k < _PB_NK; k++)
        {
          double Mul = A[i][k] * B[k][j];
          D[i][j][k] += Mul;
          C[i][j] += Mul;
        }
      }

Nevertheless, the code is recognised as a matrix-matrix multiplication, since
the second and third dimensions of D are accessed with non-zero strides.

This fixes the typo, which was made during the translation to C++ bindings
(https://reviews.llvm.org/D35845).

Reviewed By: Michael Kruse <llvm@meinersbur.de>

Differential Revision: https://reviews.llvm.org/D110491
2021-09-28 22:58:57 +05:00
Michael Kruse 027c036663 [Polly] Reject regions entered by an indirectbr/callbr.
SplitBlockPredecessors is unable to insert an additional BasicBlock
between an indirectbr/callbr terminator and the successor blocks.
This is needed by Polly to normalize the control flow before emitting
its optimzed code.

This patches rejects regions entered by an indirectbr/callbr to not fail
later at code generation.

This fixes llvm.org/PR51964

Recommit with "REQUIRES: asserts" in test that uses statistics.
2021-09-27 18:49:11 -05:00
Haowei Wu 283ed7de32 Revert "[Polly] Reject reject regions entered by an indirectbr/callbr."
This reverts commit 91f46bb77e which
causes test failures when assertions are off.
2021-09-27 16:05:33 -07:00
Michael Kruse 91f46bb77e [Polly] Reject reject regions entered by an indirectbr/callbr.
SplitBlockPredecessors is unable to insert an additional BasicBlock
between an indirectbr/callbr terminator and the successor blocks.
This is needed by Polly to normalize the control flow before emitting
its optimzed code.

This patches rejects regions entered by an indirectbr/callbr to not fail
later at code generation.

This fixes llvm.org/PR51964
2021-09-26 21:21:50 -05:00
Michael Kruse 9820dd970c [Polly] Support for InlineAsm.
Inline assembly was not handled at all and treated like a llvm::Value.
In particular, it tried to create a pointer it which is not allowed.

Fix by handling like a llvm::Constant such that it is just reused when
required, instead of trying to marshall it in memory.

Fixes llvm.org/PR51960
2021-09-26 03:26:43 -05:00
Michael Kruse d5c87162db [Polly] Use VirtualUse to determine references.
VirtualUse ensures consistency over different source of values with
Polly. In particular, this enables its use of instructions moved between
Statement. Before the patch, the code wrongly assumed that the BB's
instructions are also the ScopStmt's instructions. Reference are
determined for OpenMP outlining and GPGPU kernel extraction.

GPGPU CodeGen had some problems. For one, it generated GPU kernel
parameters for constants. Second, it emitted GPU-side invariant loads
which have already been loaded by the host. This has been partially
fixed, it still generates a store for the invariant load result, but
using the value that the host has already written.

WARNING: I did not test the generated PollyACC code on an actual GPU.

The improved consistency will be made use of in the next patch.
2021-09-26 03:26:43 -05:00
Michael Kruse 1cea25eec9 [Polly] Remove isConstCall.
The function was intended to catch OpenMP functions such as
get_thread_id(). If matched, the call would be considered synthesizable.

There were a few problems with this:

 * get_thread_id() is not 'const' in the sense of have the gcc manual
   defines it: "do not examine any values except their arguments".
   get_thread_id() reads OpenCL runtime libreary global state.
   What was inteded was probably 'speculable'.

 * isConstCall was implemented using mayReadOrWriteMemory(). 'const' is
   stricter than that, mayReadOrWriteMemory is e.g. true for malloc(),
   since it may only read/write addresses that are considered
   inaccessible fro the application. However, malloc is certainly not
   speculable.

 * Values that are isConstCall were not handled consistently throughout
   Polly. In particular, it was not considered for referenced values
   (OpenMP outlining and PollyACC).

Fix by removing special handling for isConstCall entirely.
2021-09-26 03:26:43 -05:00
Michael Kruse a5d47b3fa0 [Polly] Fix wrong redirect in test case. 2021-09-24 14:53:00 -05:00
Michael Kruse e470f9268a [Polly] Implement user-directed loop distribution/fission.
This is a simple version without the possibility to define distribute
points or followup-transformations. However, it is the first
transformation that has to check whether the transformation is correct.

It interprets the same metadata as the LoopDistribute pass.

Re-apply after revert in c7bcd72a38 with
fix: Take isBand out of #ifndef NDEBUG since it now is used
unconditionally.
2021-09-23 21:11:01 -05:00
Petr Hosek c7bcd72a38 Revert "[Polly] Implement user-directed loop distribution/fission."
This reverts commit 52c30adc7d which
breaks the build when NDEBUG is defined.
2021-09-23 14:04:25 -07:00
Michael Kruse 07e7cb9433 [Polly] Remove -polly-opt-fusion option.
The name of the option is misleading and has been renamed by isl to
"serialize-sccs". Instead of also renaming the option, remove it.
The option is still accessible using

    -polly-isl-arg=--no-schedule-serialize-sccs
2021-09-23 15:43:08 -05:00
Michael Kruse 35f7020098 [Polly] Dissolve Isl test directory. NFC.
All tests use ISL, integrate its subfolder into the components they
belong to.
2021-09-22 17:45:07 -05:00
Michael Kruse 52c30adc7d [Polly] Implement user-directed loop distribution/fission.
This is a simple version without the possibility to define distribute
points or followup-transformations. However, it is the first
transformation that has to check whether the transformation is correct.

It interprets the same metadata as the LoopDistribute pass.
2021-09-22 17:28:25 -05:00
Michael Kruse ced20c6672 [Polly] Add -polly-reschedule and -polly-postopts options.
This command line options allow to off parts of the schedule tree optimization pipeline.
2021-09-22 00:18:19 -05:00
Michael Kruse cad9f98a2a [Polly] Don't generate inter-iteration noalias metadata.
This metadata was intended to mark all accesses within an iteration to be pairwise non-aliasing, in this case because every memory of a base pointer is touched (read or write) at most once. This is typical for 'sweeps' over all data. The stated motivation from D30606 is to ensure that unrolled iterations are considered non-aliasing.

Rhe implemention had multiple issues:

 * The structure of the noalias metadata was malformed. D110026 added check in the verifier for this metadata, and the tests were failing since then.

 * This is not true for the outer loops of the BLIS matrix multiplication, where it was being inserted. Each element of A, B, C is accessed multiple times, as often as the loop not used as an index is iterating.

 * Scopes were added to SecondLevelOtherAliasScopeList (used for the !noalias scop list) on-the-fly when another SCEV was seen. This meant that previously visited instructions would not be updated with alias scopes that are only seen later, missing out those SCEVs they should not be aliasing with.

 * Since the !noalias scope list would ideally consists of all other SCEV for this base pointer, we might run quickly into scalability issues. Especially after unrolling there would probably at least once SCEV per instruction and unroll instance.

 * The inter-iteration noalias base pointer was not removed after leaving the loop marked with it, effectively marking everything after it to noalias as well.

A solution I considered was to mark each instruction as non-aliasing with its own scope. The instruction itself would obviously alias itself, but such construction might also be considered invalid. Duplicating the instruction (e.g. due to speculation) would mark the instruction non-aliasing with its clone. I don't want to go into this territory, especially since the original motivation of determining unrolled instances as noalias based on SCEV is the what scev-aa does as well.

This effectively reverts D30606 and D35761.
2021-09-20 22:20:17 -05:00
Nikita Popov 53720f74e4 [Polly] Partially fix scoped alias metadata
This partially addresses the verifier failures caused by D110026.
In particular, it does not fix the "second level" alias metadata.
2021-09-20 22:51:31 +02:00
Nikita Popov 0fc624f029 [IR] Return AAMDNodes from Instruction::getMetadata() (NFC)
getMetadata() currently uses a weird API where it populates a
structure passed to it, and optionally merges into it. Instead,
we can return the AAMDNodes and provide a separate merge() API.
This makes usages more compact.

Differential Revision: https://reviews.llvm.org/D109852
2021-09-16 21:06:57 +02:00
Michael Kruse 658eb9e142 [Polly] Remove autotools build systems from Externals. NFC.
Building a source distribution using autotools adds GPL-licenced
files into the the sources. Although redistribution of theses files is
explicitly allowed with an exception, these are not used by Polly
which uses a CMake replacement. Use the direct source checkout
instead (replacing the output of 'make dist').

Some m4 scripts with the same licence are also included in isl/ppcg
repository. Removing them renders the autotools-based build scipts
inoperable, so remove the autotools build system altogether.
2021-09-15 17:11:15 -05:00