Commit Graph

4328 Commits

Author SHA1 Message Date
Kazu Hirata 360c1111e3 Use llvm::is_contained (NFC) 2022-07-20 09:09:19 -07:00
Kazu Hirata 5cff5142a8 Use value instead of getValue (NFC) 2022-07-15 20:03:13 -07:00
Nikita Popov 2a721374ae [IR] Don't use blockaddresses as callbr arguments
Following some recent discussions, this changes the representation
of callbrs in IR. The current blockaddress arguments are replaced
with `!` label constraints that refer directly to callbr indirect
destinations:

    ; Before:
    %res = callbr i8* asm "", "=r,r,i"(i8* %x, i8* blockaddress(@test8, %foo))
    to label %asm.fallthrough [label %foo]
    ; After:
    %res = callbr i8* asm "", "=r,r,!i"(i8* %x)
    to label %asm.fallthrough [label %foo]

The benefit of this is that we can easily update the successors of
a callbr, without having to worry about also updating blockaddress
references. This should allow us to remove some limitations:

* Allow unrolling/peeling/rotation of callbr, or any other
  clone-based optimizations
  (https://github.com/llvm/llvm-project/issues/41834)
* Allow duplicate successors
  (https://github.com/llvm/llvm-project/issues/45248)

This is just the IR representation change though, I will follow up
with patches to remove limtations in various transformation passes
that are no longer needed.

Differential Revision: https://reviews.llvm.org/D129288
2022-07-15 10:18:17 +02:00
Kazu Hirata e5f568a49f Use has_value instead of hasValue (NFC) 2022-07-13 01:58:03 -07:00
Michael Kruse 6fa65f8a98 [Polly][MatMul] Abandon dependence analysis.
The copy statements inserted by the matrix-multiplication optimization
introduce new dependencies between the copy statements and other
statements. As a result, the DependenceInfo must be recomputed.

Not recomputing them caused IslAstInfo to deduce that some loops are
parallel but cause race conditions when accessing the packed arrays.
As a result, matrix-matrix multiplication currently cannot be
parallelized.

Also see discussion at https://reviews.llvm.org/D125202
2022-06-29 17:20:05 -05:00
Kazu Hirata 94460f5136 Don't use Optional::hasValue (NFC)
This patch replaces x.hasValue() with x where x is contextually
convertible to bool.
2022-06-26 19:54:41 -07:00
Kazu Hirata 3b7c3a654c Revert "Don't use Optional::hasValue (NFC)"
This reverts commit aa8feeefd3.
2022-06-25 11:56:50 -07:00
Kazu Hirata aa8feeefd3 Don't use Optional::hasValue (NFC) 2022-06-25 11:55:57 -07:00
Mingming Liu 67dc8021a1 [Support] Change TrackingStatistic and NoopStatistic to use uint64_t instead of unsigned.
Binary size of `clang` is trivial; namely, numerical value doesn't
change when measured in MiB, and `.data` section increases from 139Ki to
173 Ki.

Differential Revision: https://reviews.llvm.org/D128070
2022-06-22 10:11:40 -07:00
Arthur Eubanks c80b88ee29 [polly] #include <algorithm>
For the usage of std::max in the header.

Speculative fix for
https://ci.chromium.org/ui/p/fuchsia/builders/toolchain.ci/clang-windows-x64/b8810806780048763729/overview
reported in https://reviews.llvm.org/D125263.
2022-06-21 13:27:55 -07:00
Kazu Hirata ed8fceaa09 Don't use Optional::getValue (NFC) 2022-06-20 23:35:53 -07:00
Kazu Hirata 30c675878c Use value_or instead of getValueOr (NFC) 2022-06-19 10:34:41 -07:00
Guillaume Chatelet 4296f91323 [NFC][Alignment] Simplify code in JSONExporter 2022-06-13 13:36:36 +00:00
Guillaume Chatelet 6e930503f4 [NFC][polly] Removed dead code 2022-06-13 07:50:35 +00:00
jacquesguan bed7d707ac [NFC] Use predecessors to replace make_range.
Reviewed By: rriddle

Differential Revision: https://reviews.llvm.org/D127085
2022-06-07 02:22:35 +00:00
Fangrui Song 95a134254a Remove unneeded cl::ZeroOrMore for cl::opt/cl::list options 2022-06-05 01:07:51 -07:00
Fangrui Song d86a206f06 Remove unneeded cl::ZeroOrMore for cl::opt/cl::list options 2022-06-05 00:31:44 -07:00
Fangrui Song d0d1c416cb Remove unneeded cl::ZeroOrMore for cl::list options 2022-06-04 23:51:13 -07:00
Fangrui Song 36c7d79dc4 Remove unneeded cl::ZeroOrMore for cl::opt options
Similar to 557efc9a8b.
This commit handles options where cl::ZeroOrMore is more than one line below
cl::opt.
2022-06-04 00:10:42 -07:00
Nikita Popov 41d5033eb1 [IR] Enable opaque pointers by default
This enabled opaque pointers by default in LLVM. The effect of this
is twofold:

* If IR that contains *neither* explicit ptr nor %T* types is passed
  to tools, we will now use opaque pointer mode, unless
  -opaque-pointers=0 has been explicitly passed.
* Users of LLVM as a library will now default to opaque pointers.
  It is possible to opt-out by calling setOpaquePointers(false) on
  LLVMContext.

A cmake option to toggle this default will not be provided. Frontends
or other tools that want to (temporarily) keep using typed pointers
should disable opaque pointers via LLVMContext.

Differential Revision: https://reviews.llvm.org/D126689
2022-06-02 09:40:56 +02:00
Fangrui Song 8d3dda7624 [Polly] Fix -Wreorder-ctor. NFC 2022-06-01 17:33:14 -07:00
Yang Keao 02f640672e [Polly] Migrate -polly-mse to the new pass manager.
This patch implements the `MaximalStaticExpansion` and its printer in NPM.

Reviewed By: Meinersbur

Differential Revision: https://reviews.llvm.org/D125870
2022-06-01 13:37:58 -05:00
Nikita Popov 03aceab08b [ValueTracking] Enable -branch-on-poison-as-ub by default
Now that SimpleLoopUnswitch and other transforms no longer introduce
branch on poison, enable the -branch-on-poison-as-ub option by
default. The practical impact of this is mostly better flag
preservation in SCEV, and some freeze instructions no longer being
necessary.

Differential Revision: https://reviews.llvm.org/D125299
2022-06-01 10:46:06 +02:00
Michael Kruse cc871cf6b5 [Polly][Test] Fix race condition while printing dot files.
The tests dot-scops.ll and dot-scops-npm.ll both wrote to the same file
scops.func.dot. If they are executed in parallel they will race for
the file. Fix by renaming func to func_npm in dot-scops-npm.ll so this
test writes dot scops.func_npm.dot.

Long-term, we will probably pass a file name (prefix) to the
printer pass such that we can use the guaranteed-unique LIT %t
placeholder in tests.
2022-05-26 15:58:53 -05:00
Ivan Kosarev 8894c05b0d [FileCheck] GetCheckTypeAbbreviation() to handle the misspelled case.
Also fix directives not covered by D125604.
2022-05-26 12:20:15 +01:00
Ivan Kosarev ad1d60c3be [FileCheck] Catch missspelled directives.
Reviewed By: MaskRay

Differential Revision: https://reviews.llvm.org/D125604
2022-05-26 11:37:19 +01:00
Vitaly Buka d33c36235d [lit] Fix setup of sanitizer environment
Not all options were propageted into tests.

Reviewed By: ychen

Differential Revision: https://reviews.llvm.org/D122869
2022-05-19 19:24:16 -07:00
Jay Foad 6bec3e9303 [APInt] Remove all uses of zextOrSelf, sextOrSelf and truncOrSelf
Most clients only used these methods because they wanted to be able to
extend or truncate to the same bit width (which is a no-op). Now that
the standard zext, sext and trunc allow this, there is no reason to use
the OrSelf versions.

The OrSelf versions additionally have the strange behaviour of allowing
extending to a *smaller* width, or truncating to a *larger* width, which
are also treated as no-ops. A small amount of client code relied on this
(ConstantRange::castOp and MicrosoftCXXNameMangler::mangleNumber) and
needed rewriting.

Differential Revision: https://reviews.llvm.org/D125557
2022-05-19 11:23:13 +01:00
Michael Kruse bd93df937a [Polly] Mark classes as final by default. NFC.
This make is obivious that a class was not intended to be derived from.

NPM analysis pass can unfortunately not marked as final because they are
derived from a llvm::Checker<T> template internally by the NPM.

Also normalize the use of classes/structs
 * NPM passes are structs
 * Legacy passes are classes
 * structs that have methods and are not a visitor pattern are classes
 * structs have public inheritance by default, remove "public" keyword
 * Use typedef'ed type instead of inline forward declaration
2022-05-17 12:05:39 -05:00
Simon Pilgrim 3cc2c7deed [polly] Remove 'using namespace llvm/polly' from ScopGraphPrinter.h header.
As mentioned on D123678 this appears to be causing namespace resolution issues on some versions of gcc.
2022-05-16 16:19:03 +01:00
Michael Kruse b554c643c5 [polly] Fix type in function name. NFC. 2022-05-09 18:19:38 -05:00
Michael Kruse e61baceedb [polly] Load NPM pass plugin for NPM test.
This fixes the polly-*-plugin buildbots.
2022-05-09 16:10:01 -05:00
Michael Kruse 6b3b87376b [polly] migrate -polly-show to the new pass manager
Reviewed By: Meinersbur

Differential Revision: https://reviews.llvm.org/D123678
2022-05-09 14:04:29 -05:00
Michael Kruse a6b399ad79 [PassManager] Implement DOTGraphTraitsViewer under NPM
Rename the legacy `DOTGraphTraits{Module,}{Viewer,Printer}` to the corresponding `DOTGraphTraits...WrapperPass`, and implement a new `DOTGraphTraitsViewer` with new pass manager.

Reviewed By: Meinersbur

Differential Revision: https://reviews.llvm.org/D123677
2022-05-09 14:04:28 -05:00
Michael Kruse 809ca66eac [Polly] Fix test after D119669. 2022-05-01 13:32:42 -05:00
Florian Hahn fb4113ef0c
[Passes] Remove legacy LoopUnswitch pass.
The legacy LoopUnswitch pass is only used in the legacy pass manager
pipeline, which is deprecated.

The NewPM replacement is SimpleLoopUnswitch and I think it is time to
remove the legacy LoopUnswitch code.

Fixes #31000.

Reviewed By: aeubanks, Meinersbur, asbirlea

Differential Revision: https://reviews.llvm.org/D124376
2022-04-29 10:30:49 +01:00
Nikita Popov e1616dc59e [ScopBuilder] Avoid pointer element type access
Rather than checking the bitcast pointer element types, compare
the element type of the access and the GEP result type.

The entire code is dubious due to the inspection of GEP structure,
but this at least preserves the spirit of the existing code.
2022-04-20 11:52:36 +02:00
Nikita Popov dbe6d85b8b [PPCGCodeGeneration] Look for function instead of function pointer type
What this code is actually interested in are references to functions.
Use of a function pointer type is being used as an imprecise proxy
for that.
2022-04-19 17:59:34 +02:00
Nikita Popov 880014b593 [PPCGCodeGeneration] Avoid another pointer element type access
Use an API that returns both the address and the element type,
and use that for the load type.
2022-04-19 17:26:33 +02:00
Nikita Popov ee6bd28f23 [PPCGCodeGeneration] Avoid pointer element type access
Pass through the ArrayTy instead.
2022-04-19 17:09:34 +02:00
Nikita Popov 76174459ac [RuntimeDebugBuilder] Remove pointer element type accesses 2022-03-30 14:02:41 +02:00
Arthur Eubanks caf6af2ed7 [polly] Remove last instances of -analyze
As mentioned in D120782, the loop block order can be different depending
on if LoopInfo is incrementally updated or freshly computed.

Reviewed By: Meinersbur

Differential Revision: https://reviews.llvm.org/D122195
2022-03-24 09:47:43 -07:00
Philip Reames 93102505aa Rename mayBeMemoryDependent in polly to fix build bot
This case was missed in ee7324b8.
2022-03-21 10:11:31 -07:00
Michael Kruse 12ac339e9e [polly] Fix NPM unittests after D121566. 2022-03-18 14:25:44 -05:00
Wael Yehia c80198b3d3 Reland "Load pass plugins during option processing, so that plugin options are registered and live."
Fix Polly failures.

Reviewed By: mehdi_amini, Meinersbur

Differential Revision: https://reviews.llvm.org/D121566
2022-03-18 03:27:53 +00:00
Sam McCall 75acad41bc Use lit_config.substitute instead of foo % lit_config.params everywhere
This mechanically applies the same changes from D121427 everywhere.

Differential Revision: https://reviews.llvm.org/D121746
2022-03-16 09:57:41 +01:00
Michael Kruse 5c02808131 [polly] Introduce -polly-print-* passes to replace -analyze.
The `opt -analyze` option only works with the legacy pass manager and might be removed in the future, as explained in llvm.org/PR53733. This patch introduced -polly-print-* passes that print what the pass would print with the `-analyze` option and replaces all uses of `-analyze` in the regression tests.

There are two exceptions: `CodeGen\single_loop_param_less_equal.ll` and `CodeGen\loop_with_condition_nested.ll` use `-analyze on the `-loops` pass which is not part of Polly.

Reviewed By: aeubanks

Differential Revision: https://reviews.llvm.org/D120782
2022-03-14 10:27:15 -05:00
Petr Hosek 0c0f6cfb7b [CMake] Rename TARGET_TRIPLE to LLVM_TARGET_TRIPLE
This clarifies that this is an LLVM specific variable and avoids
potential conflicts with other projects.

Differential Revision: https://reviews.llvm.org/D119918
2022-03-11 15:43:01 -08:00
Arthur Eubanks 30f1cef86b Revert "[polly] Fix regression test after D110620."
This reverts commit 2aa624a94f.

D110620 was reverted.
2022-03-04 20:37:15 -08:00
Michael Kruse d7851685a3 [polly] Remove trailing whitespace from tests. NFC. 2022-02-22 15:41:13 -06:00