Matt Arsenault [Thu, 28 Jun 2018 10:18:23 +0000 (10:18 +0000)]
AMDGPU: Fix AMDGPUCodeGenPrepare using uninitialized AMDGPUAS struct
Not sure how this wasn't noticed before.
llvm-svn: 335828
Matt Arsenault [Thu, 28 Jun 2018 10:18:11 +0000 (10:18 +0000)]
AMDGPU: Fix assert on aggregate type kernel arguments
Just fix the crash for now by not doing the optimization since
figuring out how to properly convert the bits for an arbitrary
struct is a pain.
Also fix a crash when there is only an empty struct argument.
llvm-svn: 335827
George Rimar [Thu, 28 Jun 2018 10:12:59 +0000 (10:12 +0000)]
[ELF] - Removed trailing whitespace. NFC.
llvm-svn: 335826
George Rimar [Thu, 28 Jun 2018 10:10:02 +0000 (10:10 +0000)]
[ELF] - Add test case to cover "corrupted CIE (failed to read string)"
It was uncovered by our test cases.
llvm-svn: 335824
Benjamin Kramer [Thu, 28 Jun 2018 10:03:45 +0000 (10:03 +0000)]
Unify sorted asserts to use the existing atomic pattern
These are all benign races and only visible in !NDEBUG. tsan complains
about it, but a simple atomic bool is sufficient to make it happy.
llvm-svn: 335823
Pavel Labath [Thu, 28 Jun 2018 10:03:42 +0000 (10:03 +0000)]
Retrieve a function PDB symbol correctly from nested blocks
Summary:
This patch fixes a problem with retrieving a function symbol by an
address in a nested block. In the current implementation of
ResolveSymbolContext function it retrieves a symbol with
PDB_SymType::None and then checks if found symbol's tag equals to
PDB_SymType::Function. So, if nested block's symbol was found,
ResolveSymbolContext does not resolve a function.
Reviewers: asmith, labath, zturner
Reviewed By: asmith, labath
Differential Revision: https://reviews.llvm.org/D47939
Patch by Aleksandr Urakov <aleksandr.urakov@jetbrains.com>
llvm-svn: 335822
Simon Pilgrim [Thu, 28 Jun 2018 09:54:28 +0000 (09:54 +0000)]
[DAGCombiner] Ensure we use the correct CC result type in visitSDIV
We could get away with it for constant folded cases, but not for rL335719.
Thanks to Krzysztof Parzyszek for noticing.
llvm-svn: 335821
Florian Hahn [Thu, 28 Jun 2018 09:53:38 +0000 (09:53 +0000)]
[SCCP] Mark CFG as preserved.
SCCP does not change the CFG, so we can mark it as preserved.
Reviewers: dberlin, efriedma, davide
Reviewed By: davide
Differential Revision: https://reviews.llvm.org/D47149
llvm-svn: 335820
George Rimar [Thu, 28 Jun 2018 09:34:59 +0000 (09:34 +0000)]
[ELF] - Add test case to cover "unexpected end of CIE" error.
It was uncovered by our test cases.
llvm-svn: 335819
George Rimar [Thu, 28 Jun 2018 09:29:49 +0000 (09:29 +0000)]
[ELF] - Get rid of precompiled elf object in test case.
We has precompiled object with unsupported FDE version (=2).
It is possible to use llvm-mc instead for this test.
Patch do this change.
llvm-svn: 335818
Simon Pilgrim [Thu, 28 Jun 2018 09:29:08 +0000 (09:29 +0000)]
[DAGCombiner] Remove unused variable. NFCI.
Noticed in D45806 review.
llvm-svn: 335817
Max Kazantsev [Thu, 28 Jun 2018 08:20:03 +0000 (08:20 +0000)]
[IndVarSimplify] Ignore unreachable users of truncs
If a trunc has a user in a block which is not reachable from entry,
we can safely perform trunc elimination as if this user didn't exist.
llvm-svn: 335816
Clement Courbet [Thu, 28 Jun 2018 07:41:16 +0000 (07:41 +0000)]
[llvm-exegesis] Add partial X87 support.
Summary:
This enables the X86-specific X86FloatingPointStackifierPass, and allow
llvm-exegesis to generate and measure X87 latency/uops for some FP ops.
Reviewers: gchatelet
Subscribers: tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D48592
llvm-svn: 335815
Adam Balogh [Thu, 28 Jun 2018 07:35:23 +0000 (07:35 +0000)]
[Analyzer] Constraint Manager Negates Difference
If range [m .. n] is stored for symbolic expression A - B, then we can deduce the range for B - A which is [-n .. -m]. This is only true for signed types, unless the range is [0 .. 0].
Differential Revision: https://reviews.llvm.org/D35110
llvm-svn: 335814
Tobias Grosser [Thu, 28 Jun 2018 07:11:48 +0000 (07:11 +0000)]
test: use regex matchers to make test-case robust against register renumberings
Suggested-by: Michael Kruse
llvm-svn: 335813
Petr Hosek [Thu, 28 Jun 2018 05:15:46 +0000 (05:15 +0000)]
[CMake] Use explicit targets for building Linux runtimes
Previously we were using default logic when building Linux runtimes
in Fuchsia toolchain, but that leads to various issues due to how
the CMake logic in compiler-rt for determining the platform support
is implemented. With this change, we will use explicit target for
every provided Linux sysroot.
Differential Revision: https://reviews.llvm.org/D48563
llvm-svn: 335812
Petar Jovanovic [Thu, 28 Jun 2018 04:50:40 +0000 (04:50 +0000)]
[DwarfDebug] Remove unused argument (NFC)
Remove unused ByteStreamer argument from function emitDebugLocValue.
Patch by Nikola Prica.
Differential Revision: https://reviews.llvm.org/D48590
llvm-svn: 335811
Petr Hosek [Thu, 28 Jun 2018 03:54:08 +0000 (03:54 +0000)]
Handle both Linux and Windows path separator for the resource dir
The resource dir path used for the multiarch runtimes support is
constructed in a platform independent way and therefore will use
native path separators on each platform. We need to make sure that
the per target runtime directory test handles both to not fail
when the test is being executed on Windows.
llvm-svn: 335810
Petr Hosek [Thu, 28 Jun 2018 03:11:52 +0000 (03:11 +0000)]
Support for multiarch runtimes layout
This change adds a support for multiarch style runtimes layout, so in
addition to the existing layout where runtimes get installed to:
lib/clang/$version/lib/$os
Clang now allows runtimes to be installed to:
lib/clang/$version/$target/lib
This also includes libc++, libc++abi and libunwind; today those are
assumed to be in Clang library directory built for host, with the
new layout it is possible to install libc++, libc++abi and libunwind
into the runtime directory built for different targets.
The use of new layout is enabled by setting the
LLVM_ENABLE_RUNTIME_TARGET_DIR CMake variable and is supported by both
projects and runtimes layouts. The runtimes CMake build has been further
modified to use the new layout when building runtimes for multiple
targets.
Differential Revision: https://reviews.llvm.org/D45604
llvm-svn: 335809
Artem Dergachev [Thu, 28 Jun 2018 02:33:16 +0000 (02:33 +0000)]
[analyzer] Remove redundant ';'.
Fixes a compiler warning. No functionan change intended.
llvm-svn: 335808
Richard Smith [Thu, 28 Jun 2018 01:57:04 +0000 (01:57 +0000)]
[modules] Do not serialize / deserialize pending new/delete mismatch
checks across module boundaries. This was causing us to load constructor
definitions for all consumers of a module with a pending check.
(In one case we saw ~7% of total frontend time spent loading
constructors for this check.)
llvm-svn: 335807
Craig Topper [Thu, 28 Jun 2018 01:45:44 +0000 (01:45 +0000)]
[X86] Use PatFrag with hardcoded numbers for FROUND_NO_EXC/FROUND_CURRENT instead of ImmLeafs with predicates where one of the two numbers was hardcoded.
This more efficient for the isel table generator since we can use CheckChildInteger instead of MoveChild, CheckPredicate, MoveParent. This reduced the table size by 1-2K.
I wish there was a way to share the values with X86BaseInfo.h and still use a PatFrag like this. These numbers are fixed by the X86 intrinsic spec going back many years and we should never need to change them. So we shouldn't waste table bytes to support sharing.
llvm-svn: 335806
Richard Smith [Thu, 28 Jun 2018 01:07:28 +0000 (01:07 +0000)]
[modules] Ensure that an in-class function definition is attached to the
declaration of the function that ends up in the primary definition of
the class.
... at least for class templates. This is necessary for us to be able to
track when an inline friend function has a definition that needs to be
(lazily) instantiated.
llvm-svn: 335805
Craig Topper [Thu, 28 Jun 2018 00:47:41 +0000 (00:47 +0000)]
[X86] Change how we prefer shift by immediate over folding a load into a shift.
BMI2 added new shift by register instructions that have the ability to fold a load.
Normally without doing anything special isel would prefer folding a load over folding an immediate because the load folding pattern has higher "complexity". This would require an instruction to move the immediate into a register. We would rather fold the immediate instead and have a separate instruction for the load.
We used to enforce this priority by artificially lowering the complexity of the load pattern.
This patch changes this to instead reject the load fold in isProfitableToFoldLoad if there is an immediate. This is more consistent with other binops and feels less hacky.
llvm-svn: 335804
Artem Dergachev [Thu, 28 Jun 2018 00:42:11 +0000 (00:42 +0000)]
[analyzer] Use sufficiently large types for index bounds calculation.
The ProgramState::assumeInBound() API is used by checkers to make an assumption
that a certain array index is within the array's bounds (i.e. is greater than or
equal to 0 and is less than the length of the array). When the type of the
index was unspecified by the caller, it assumed that the type is 'int', which
caused some indices and sizes to truncate during calculations.
Use ArrayIndexTy by default instead, which is used by the analyzer to represent
index types and is currently hardcoded to long long.
Patch by Bevin Hansson!
Differential Revision: https://reviews.llvm.org/D46944
llvm-svn: 335803
George Karpenkov [Thu, 28 Jun 2018 00:39:50 +0000 (00:39 +0000)]
[libFuzzer] [Tests] [NFC] Change seed for reduce_inputs.test
Differential Revision: https://reviews.llvm.org/D48686
llvm-svn: 335802
Alex Lorenz [Thu, 28 Jun 2018 00:39:09 +0000 (00:39 +0000)]
[cmake][xcode-toolchain] add support for major Xcode version >= 10
The regex that extracts the Xcode version should support major versions with two
digits.
rdar://
41465184
llvm-svn: 335801
Artem Dergachev [Thu, 28 Jun 2018 00:30:18 +0000 (00:30 +0000)]
[analyzer] Add support for pre-C++17 copy elision.
r335795 adds copy elision information to CFG. This commit allows static analyzer
to elide elidable copy constructors by constructing the objects that were
previously subject to elidable copy directly in the target region of the copy.
The chain of elided constructors may potentially be indefinitely long. This
only happens when the object is being returned from a function which in turn is
returned from another function, etc.
NRVO is not supported yet.
Differential Revision: https://reviews.llvm.org/D47671
llvm-svn: 335800
Douglas Yung [Thu, 28 Jun 2018 00:19:12 +0000 (00:19 +0000)]
Fixup test to compile with -frtti when trying to use typeid() as the PS4 does not have it on by default and it was failing on the PS4 linux bot because of this.
llvm-svn: 335799
Artem Dergachev [Thu, 28 Jun 2018 00:18:52 +0000 (00:18 +0000)]
[CFG] [analyzer] Simplify lifetime-extended temporary construction contexts.
When a temporary object is materialized and through that obtain lifetime that
is longer than the duration of the full-expression, it does not require a
temporary object destructor; it will be destroyed in a different manner.
Therefore it's not necessary to include CXXBindTemporaryExpr into the
construction context for such temporary in the CFG only to make clients
throw it away.
Differential Revision: https://reviews.llvm.org/D47667
llvm-svn: 335798
Michael J. Spencer [Thu, 28 Jun 2018 00:12:04 +0000 (00:12 +0000)]
[CGProfile] Fix unused variable warning.
llvm-svn: 335797
Artem Dergachev [Thu, 28 Jun 2018 00:11:42 +0000 (00:11 +0000)]
[analyzer] Re-enable lifetime extension for temporaries without destructors.
When an object's class provides no destructor, it's less important to
materialize that object properly because we don't have to model the destructor
correctly, so previously we skipped the support for these syntax patterns.
Additionally, fix support for construction contexts of "static temporaries"
(temporaries that are lifetime-extended by static references) because
it turned out that we only had tests for them without destructors, which caused
us to regress when we re-introduced the construction context for such
temporaries.
Differential Revision: https://reviews.llvm.org/D47658
llvm-svn: 335796
Artem Dergachev [Thu, 28 Jun 2018 00:04:54 +0000 (00:04 +0000)]
[CFG] [analyzer] Add construction contexts that explain pre-C++17 copy elision.
Before C++17 copy elision was optional, even if the elidable copy/move
constructor had arbitrary side effects. The elidable constructor is present
in the AST, but marked as elidable.
In these cases CFG now contains additional information that allows its clients
to figure out if a temporary object is only being constructed so that to pass
it to an elidable constructor. If so, it includes a reference to the elidable
constructor's construction context, so that the client could elide the
elidable constructor and construct the object directly at its final destination.
Differential Revision: https://reviews.llvm.org/D47616
llvm-svn: 335795
Michael J. Spencer [Wed, 27 Jun 2018 23:58:08 +0000 (23:58 +0000)]
Add support for generating a call graph profile from Branch Frequency Info.
=== Generating the CG Profile ===
The CGProfile module pass simply gets the block profile count for each BB and scans for call instructions. For each call instruction it adds an edge from the current function to the called function with the current BB block profile count as the weight.
After scanning all the functions, it generates an appending module flag containing the data. The format looks like:
```
!llvm.module.flags = !{!0}
!0 = !{i32 5, !"CG Profile", !1}
!1 = !{!2, !3, !4} ; List of edges
!2 = !{void ()* @a, void ()* @b, i64 32} ; Edge from a to b with a weight of 32
!3 = !{void (i1)* @freq, void ()* @a, i64 11}
!4 = !{void (i1)* @freq, void ()* @b, i64 20}
```
Differential Revision: https://reviews.llvm.org/D48105
llvm-svn: 335794
George Karpenkov [Wed, 27 Jun 2018 23:11:24 +0000 (23:11 +0000)]
[libFuzzer] [NFC] [Tests] Mark signal handling tests as UNSUPPORTED on Darwin
Under load, these tests tend to fail sporadically on our bots.
In my understanding, the signal handling is not guaranteed to happen
within 2 seconds, and the test is inherently flaky.
llvm-svn: 335792
Heejin Ahn [Wed, 27 Jun 2018 22:05:09 +0000 (22:05 +0000)]
[analyzer] Add clangFrontend to target_link_libraries
Without this, builds with `-DSHARED_LIB=ON` fail.
llvm-svn: 335791
Petr Hosek [Wed, 27 Jun 2018 21:25:21 +0000 (21:25 +0000)]
[sanitizer] zx_cprng_draw no longer returns any value
Remove the return value check.
Differential Revision: https://reviews.llvm.org/D48671
llvm-svn: 335790
Zachary Turner [Wed, 27 Jun 2018 21:18:15 +0000 (21:18 +0000)]
Move some code from PDBFileBuilder to MSFBuilder.
The code to emit the pieces of the MSF file were actually in
PDBFileBuilder. Move this to MSFBuilder so that we can
theoretically emit an MSF without having a PDB file.
llvm-svn: 335789
Benjamin Kramer [Wed, 27 Jun 2018 21:01:53 +0000 (21:01 +0000)]
[X86] Make folding table checking threadsafe
This is a benign race, but tsan likes to complain about it. Just make it
happy.
llvm-svn: 335788
Craig Topper [Wed, 27 Jun 2018 20:58:46 +0000 (20:58 +0000)]
[X86] In X86DAGToDAGISel::PreprocessISelDAG, make sure we don't access N after we delete it.
If we turn X86ISD::AND into ISD::AND, we delete N. But we were continuing onto the next block of code even though N no longer existed.
Just happened to notice it. I assume asan didn't notice it because we explicitly unpoison deleted nodes and give them a DELETE_NODE opcode.
llvm-svn: 335787
Sameer AbuAsal [Wed, 27 Jun 2018 20:51:42 +0000 (20:51 +0000)]
[RISCV] Add machine function pass to merge base + offset
Summary:
In r333455 we added a peephole to fix the corner cases that result
from separating base + offset lowering of global address.The
peephole didn't handle some of the cases because it only has a basic
block view instead of a function level view.
This patch replaces that logic with a machine function pass. In
addition to handling the original cases it handles uses of the global
address across blocks in function and folding an offset from LW\SW
instruction. This pass won't run for OptNone compilation, so there
will be a negative impact overall vs the old approach at O0.
Reviewers: asb, apazos, mgrang
Reviewed By: asb
Subscribers: MartinMosbeck, brucehoult, the_o, rogfer01, mgorny, rbar, johnrusso, simoncook, niosHD, kito-cheng, shiva0217, zzheng, llvm-commits, edward-jones
Differential Revision: https://reviews.llvm.org/D47857
llvm-svn: 335786
Fangrui Song [Wed, 27 Jun 2018 20:45:11 +0000 (20:45 +0000)]
[llvm-objdump] Add -x --all-headers options
Reviewers: paulsemel, echristo
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D48622
llvm-svn: 335785
Fangrui Song [Wed, 27 Jun 2018 20:43:23 +0000 (20:43 +0000)]
[ELF] Use %eiz after disassembler change in r335768
llvm-svn: 335784
Eli Friedman [Wed, 27 Jun 2018 20:35:02 +0000 (20:35 +0000)]
[ScopHelper] Cache ScopExpander results.
The number of SCEV expressions is usually linear in the number of IR
instructions being modeled. However, a naive SCEV visitor is not. For
an expression like x*x, "x" will be visited twice. If x is itself an
expression like x*x, that will be visited twice, etc, and the overall
runtime is O(2^N) in the number of SCEV expressions.
To prevent this from happening, add a cache, so we only visit each SCEV
expression once.
Not sure this is the best solution. Maybe we can instead check whether
the SCEV is scop-invariant (in which case we never need to map the
value). But we don't have a utility for that at the moment.
Differential Revision: https://reviews.llvm.org/D47087
llvm-svn: 335783
Richard Smith [Wed, 27 Jun 2018 20:30:36 +0000 (20:30 +0000)]
[www] Update cxx_dr_status for recent DR fixes.
llvm-svn: 335782
Richard Smith [Wed, 27 Jun 2018 20:30:34 +0000 (20:30 +0000)]
DR1687: When overload resolution selects a built-in operator, implicit
conversions are only applied to operands of class type, and the second
standard conversion sequence is not applied.
When diagnosing an invalid builtin binary operator, talk about the
original types rather than the converted types. If these differ by a
user-defined conversion, tell the user what happened.
llvm-svn: 335781
Bruno Cardoso Lopes [Wed, 27 Jun 2018 20:29:36 +0000 (20:29 +0000)]
[Modules][ObjC] Warn on the use of '@import' in framework headers
Using @import in framework headers inhibit the use of such headers
when not using modules, this is specially bad for headers that end
up in the SDK (or any other system framework). Add a warning to give
users some indication that this is discouraged.
rdar://problem/
39192894
llvm-svn: 335780
Richard Smith [Wed, 27 Jun 2018 20:29:32 +0000 (20:29 +0000)]
DR1213: Ignore implicit conversions when determining if an operand of an
array subscript expression is an array prvalue.
Also apply DR1213 to vector prvalues for consistency.
llvm-svn: 335779
Sanjay Patel [Wed, 27 Jun 2018 20:23:47 +0000 (20:23 +0000)]
[InstCombine] add tests for vector-select-of-binops with 2 variables; NFC
llvm-svn: 335778
Vlad Tsyrklevich [Wed, 27 Jun 2018 20:00:55 +0000 (20:00 +0000)]
Another shot at fixing android r335644 failure
The android buildbot moves the build outputs to a different directory
and rewrites the executable path, the DSO passed as an argument does not
get re-written. Use rpaths to load the DSO the same way the
test/cfi/cross-dso/ tests do and test the DSO name differently.
llvm-svn: 335777
Jonas Devlieghere [Wed, 27 Jun 2018 19:58:39 +0000 (19:58 +0000)]
[SymbolFile] Implement GetCompleteObjCClass for .debug_names
When running the test suite with .debug_names a bunch of tests were
failing because GetCompleteObjCClass was not yet implemented for
DebugNamesDWARFIndex. This patch adds the required logic.
We use the .debug_names to find the Objective-C class and then rely on
DW_AT_APPLE_objc_complete_type to find the complete type. If we can't
find it or the attribute is not supported, we return a list of potential
complete types.
Differential revision: https://reviews.llvm.org/D48596
llvm-svn: 335776
Paul Robinson [Wed, 27 Jun 2018 19:58:28 +0000 (19:58 +0000)]
Document the git config for Windows to do line-endings correctly.
Differential Revision: https://reviews.llvm.org/D48494
llvm-svn: 335775
Aaron Enye Shi [Wed, 27 Jun 2018 19:51:42 +0000 (19:51 +0000)]
[HIP] Fix ordering of device-libs linking
Summary:
HIP should link the bitcodes with caller functions before callee functions. Also added lit test to check the ordering of the linked bitcodes is matches.
Reviewers: yaxunl, b-sumner
Reviewed By: yaxunl, b-sumner
Subscribers: cfe-commits, yaxunl, b-sumner, scchan
Differential Revision: https://reviews.llvm.org/D48667
llvm-svn: 335774
Nirav Dave [Wed, 27 Jun 2018 19:41:25 +0000 (19:41 +0000)]
[DAGCombine] Disable TokenFactor simplifications when optnone.
llvm-svn: 335773
Michael Kruse [Wed, 27 Jun 2018 19:39:03 +0000 (19:39 +0000)]
[ADT] drop_begin: use adl_begin/adl_end. NFC.
Summary:
The instantiation of the drop_begin function template usually fails because the functions begin() and end() do not exist. Only when using on a container from the std namespace (or `llvm::iterator_range`s of something derived from `std::iterator`), they are matched to std::begin() and std::end() due to Koenig-lookup.
Explicitly use llvm::adl_begin and llvm::adl_end to make drop_begin applicable to anything iterable (including C-style arrays).
A solution for general `llvm::iterator_range`s was already tried in r244620, but got reverted in r244621 due to MSVC not liking it.
Reviewers: dblaikie, grosbach, aaron.ballman, ruiu
Reviewed By: dblaikie, aaron.ballman
Subscribers: aaron.ballman, llvm-commits
Differential Revision: https://reviews.llvm.org/D48598
llvm-svn: 335772
Fangrui Song [Wed, 27 Jun 2018 19:35:50 +0000 (19:35 +0000)]
[WebAssembly] Try fixing test/CodeGen/WebAssembly/vector_sdiv.ll
llvm-svn: 335771
Ben Hamilton [Wed, 27 Jun 2018 19:13:09 +0000 (19:13 +0000)]
[clang-tidy/ObjC] Add hashing algorithm acronyms to objc-property-declaration
Summary:
This PR adds a few acronyms related to hashing algorithms to the standard
list in `objc-property-declaration`.
Reviewers: Wizard
Reviewed By: Wizard
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D48652
llvm-svn: 335770
Fangrui Song [Wed, 27 Jun 2018 19:12:07 +0000 (19:12 +0000)]
[X86] Fix unmatched parenthesis in r335768
llvm-svn: 335769
Craig Topper [Wed, 27 Jun 2018 19:03:36 +0000 (19:03 +0000)]
[X86] Teach the disassembler to use %eiz/%riz instead of NoRegister when the SIB byte is present, but doesn't encode an index register and there was another shorter encoding that would achieve the same result.
The %eiz/%riz are dummy registers that force the encoder to emit a SIB byte when it normally wouldn't. By emitting them in the disassembly output we ensure that assembling the disassembler output would also produce a SIB byte.
This should match the behavior of objdump from binutils.
llvm-svn: 335768
Daniel Sanders [Wed, 27 Jun 2018 19:03:21 +0000 (19:03 +0000)]
[globalisel][legalizer] Add AtomicOrdering to LegalityQuery and use it in AArch64
Now that we have the ability to legalize based on MMO's. Add support for
legalizing based on AtomicOrdering and use it to correct the legalization
of the atomic instructions.
Also extend all() to be a variadic template as this ruleset now requires
3 and 4 argument versions.
llvm-svn: 335767
Teresa Johnson [Wed, 27 Jun 2018 19:00:35 +0000 (19:00 +0000)]
[ThinLTO] Fix test
Fix test changes added in r335760. Even though we are invoking llvm-lto2
in single threaded mode, the order of processing the modules in the
backend is apparently not deterministic. Handle the expected debug
messages in any order. (The determinism would be good to fix, but not
related to this change.)
This also undoes the change I made in r335764 to help debug this.
llvm-svn: 335766
Aaron Enye Shi [Wed, 27 Jun 2018 18:58:55 +0000 (18:58 +0000)]
[HIP] Support flush denormals bitcode
Summary:
Use oclc_daz_opt_on.amdgcn.bc bitcode when option fcuda-flush-denormal-to-zero is enabled, otherwise use oclc_daz_opt_off.amdgcn.bc bitcode. Added lit tests to verify that the correct bitcode is linked when -fcuda-flush-denormal-to-zero option is enabled or disabled.
Reviewers: yaxunl, scchan, b-sumner
Reviewed By: yaxunl, scchan, b-sumner
Subscribers: cfe-commits, yaxunl
Differential Revision: https://reviews.llvm.org/D48493
llvm-svn: 335765
Teresa Johnson [Wed, 27 Jun 2018 18:36:53 +0000 (18:36 +0000)]
[ThinLTO] Modify test to help diagnose bot failures
I am getting bot failures from r335760 that are difficult to diagnose
since the stderr is getting redirected to FileCheck. Save and dump the
debug output to stderr to help debug the issue.
llvm-svn: 335764
Artem Belevich [Wed, 27 Jun 2018 18:32:51 +0000 (18:32 +0000)]
[CUDA] Use atexit() to call module destructor.
This matches the way NVCC does it. Doing module cleanup at global
destructor phase used to work, but is, apparently, too late for
the CUDA runtime in CUDA-9.2, which ends up crashing with double-free.
Differential Revision: https://reviews.llvm.org/D48613
llvm-svn: 335763
Matt Morehouse [Wed, 27 Jun 2018 18:24:46 +0000 (18:24 +0000)]
[UBSan] Add silence_unsigned_overflow flag.
Summary:
Setting UBSAN_OPTIONS=silence_unsigned_overflow=1 will silence all UIO
reports. This feature, combined with
-fsanitize-recover=unsigned-integer-overflow, is useful for providing
fuzzing signal without the excessive log output.
Helps with https://github.com/google/oss-fuzz/issues/910.
Reviewers: kcc, vsk
Reviewed By: vsk
Subscribers: vsk, kubamracek, Dor1s, llvm-commits
Differential Revision: https://reviews.llvm.org/D48660
llvm-svn: 335762
Sanjay Patel [Wed, 27 Jun 2018 18:16:40 +0000 (18:16 +0000)]
[DAGCombiner] restrict (float)((int) f) --> ftrunc with no-signed-zeros
As noted in the D44909 review, the transform from (fptosi+sitofp) to ftrunc
can produce -0.0 where the original code does not:
#include <stdio.h>
int main(int argc) {
float x;
x = -0.8 * argc;
printf("%f\n", (float)((int)x));
return 0;
}
$ clang -O0 -mavx fp.c ; ./a.out
0.000000
$ clang -O1 -mavx fp.c ; ./a.out
-0.000000
Ideally, we'd use IR/node flags to predicate the transform, but the IR parser
doesn't currently allow fast-math-flags on the cast instructions. So for now,
just use the function attribute that corresponds to clang's "-fno-signed-zeros"
option.
Differential Revision: https://reviews.llvm.org/D48085
llvm-svn: 335761
Teresa Johnson [Wed, 27 Jun 2018 18:03:39 +0000 (18:03 +0000)]
[ThinLTO] Print names in function import debug messages when available
Summary:
Rather than just print the GUID, when it is available in the index,
print the global name as well in the function import thin link debug
messages. Names will be available when the combined index is being
built by the same process, e.g. a linker or "llvm-lto2 run".
Reviewers: davidxl
Subscribers: mehdi_amini, inglorion, eraman, steven_wu, llvm-commits
Differential Revision: https://reviews.llvm.org/D48612
llvm-svn: 335760
Justin Bogner [Wed, 27 Jun 2018 17:58:32 +0000 (17:58 +0000)]
[Object] Allow iterating over an IRObjectFile's modules
If you've already loaded an IRObjectFile and need access to the
Modules themselves you shouldn't have to reparse a byte stream to do
it. Adds an accessor for the modules in IRObjectFile.
llvm-svn: 335759
Jessica Paquette [Wed, 27 Jun 2018 17:43:27 +0000 (17:43 +0000)]
[MachineOutliner] Don't outline sequences where x16/x17/nzcv are live across
It isn't safe to outline sequences of instructions where x16/x17/nzcv live
across the sequence.
This teaches the outliner to check whether or not a specific canidate has
x16/x17/nzcv live across it and discard the candidate in the case that that is
true.
https://bugs.llvm.org/show_bug.cgi?id=37573
https://reviews.llvm.org/D47655
llvm-svn: 335758
Jonas Devlieghere [Wed, 27 Jun 2018 17:31:59 +0000 (17:31 +0000)]
[DebugInfo] Emit ObjC methods as part of interface
As brought up during the discussion of the DWARF5 accelerator tables,
there is currently no way to associate Objective-C methods with the
interface they belong to, other than the .apple_objc accelerator table.
After due consideration we came to the conclusion that it makes more
sense to follow Pavel's suggestion of just emitting this information in
the .debug_info section. One concern was that categories were
emitted in the .apple_names as well, but it turns out that LLDB doesn't
rely on the accelerator tables for this information.
This patch changes the codegen behavior to emit subprograms for
structure types, like we do for C++. This will result in the
DW_TAG_subprogram being nested as a child under its
DW_TAG_structure_type. This behavior is only enabled for DWARF5 and
later, so we can have a unique code path in LLDB with regards to
obtaining the class methods.
This was tested on the LLDB side and doesn't lead to a regression.
There's already code in place to deal with member functions in C++,
which deals with this transparently.
For more background please refer to the discussion on the mailing list:
http://lists.llvm.org/pipermail/llvm-dev/2018-June/123986.html
Differential revision: https://reviews.llvm.org/D48241
llvm-svn: 335757
Sanjay Patel [Wed, 27 Jun 2018 17:21:57 +0000 (17:21 +0000)]
[InstCombine] add more tests for shuffle with different binops; NFC
llvm-svn: 335756
Petr Hosek [Wed, 27 Jun 2018 16:49:37 +0000 (16:49 +0000)]
[sanitizer] zx_cprng_draw no longer takes the output argument
The zx_cprng_draw system call no longer takes the output argument.
Differential Revision: https://reviews.llvm.org/D48657
llvm-svn: 335755
Craig Topper [Wed, 27 Jun 2018 16:47:39 +0000 (16:47 +0000)]
[X86] Use bts/btr/btc for single bit set/clear/complement of a variable bit position
If we are just modifying a single bit at a variable bit position we can use the BT* instructions to make the change instead of shifting a 1(or rotating a -1) and doing a binop. These instruction also ignore the upper bits of their index input so we can also remove an and if one is present on the index.
Fixes PR37938.
llvm-svn: 335754
Craig Topper [Wed, 27 Jun 2018 16:47:36 +0000 (16:47 +0000)]
[X86] Add test cases for D48606.
llvm-svn: 335753
Mikhail R. Gadelha [Wed, 27 Jun 2018 16:45:58 +0000 (16:45 +0000)]
Revert "[Analyzer] Moved RangeConstraintManager to header. NFC."
This broke a number of bots.
This reverts commit
5e1a89912d37a21c3b49ccf30600d7f498dffa9c.
llvm-svn: 335752
Jakub Kuderski [Wed, 27 Jun 2018 16:34:30 +0000 (16:34 +0000)]
[AliasSet] Fix UnknownInstructions printing
Summary:
AliasSet::print uses `I->printAsOperand` to print UnknownInstructions. The problem is that not all UnknownInstructions have names (e.g. call instructions). When such instructions are printed, they appear as `<badref>` in AliasSets, which is very confusing, as the values are perfectly valid.
This patch fixes that by printing UnknownInstructions without a name using `print` instead of `printAsOperand`.
Reviewers: asbirlea, chandlerc, sanjoy, grosser
Reviewed By: asbirlea
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D48609
llvm-svn: 335751
Francis Visoiu Mistrih [Wed, 27 Jun 2018 16:17:32 +0000 (16:17 +0000)]
[NEON] Remove empty test file from r335734
Fails on Green Dragon:
http://green.lab.llvm.org/green/job/clang-stage1-cmake-RA-incremental/50174/consoleFull
UNRESOLVED: Clang :: CodeGen/vld_dup.c (5546 of 38947)
******************** TEST 'Clang :: CodeGen/vld_dup.c' FAILED ********************
Test has no run line!
llvm-svn: 335750
Jonas Devlieghere [Wed, 27 Jun 2018 16:13:40 +0000 (16:13 +0000)]
[dsymutil] Move abstractions into separate files (NFC)
This patch splits off some abstractions used by dsymutil's dwarf linker
and moves them into separate header and implementation files. This
almost halves the number of LOC in DwarfLinker.cpp and makes it a lot
easier to understand what functionality lives where.
Differential revision: https://reviews.llvm.org/D48647
llvm-svn: 335749
Matt Davis [Wed, 27 Jun 2018 16:09:33 +0000 (16:09 +0000)]
[llvm-mca] Register listeners with stages; remove Pipeline dependency from Stage.
Summary:
This patch removes a few callbacks from Pipeline. It comes at the cost of
registering Listeners with all Stages. Not all stages need listeners or issue
callbacks, this registration is a bit redundant. However, as we build-out the
API, this redundancy can disappear.
The main purpose here is to move callback code from the Pipeline and into the
stages that actually issue those callbacks. This removes the back-pointer to
the Pipeline that was put into a few Stage subclasses.
Reviewers: andreadb, courbet, RKSimon
Reviewed By: andreadb, courbet
Subscribers: tschuett, gbedwell, llvm-commits
Differential Revision: https://reviews.llvm.org/D48576
llvm-svn: 335748
Vedant Kumar [Wed, 27 Jun 2018 16:06:15 +0000 (16:06 +0000)]
[asan] Relax the null_deref.cc test even further
On Darwin/x86_64, asan may report the crashing line of NullDeref as line
19 (i.e the closing brace of the function), whereas on other targets we
see line 15 ("ptr[10]++").
The optimized debug info here isn't reliable enough to check.
rdar://problem/
41526369
llvm-svn: 335747
Simon Pilgrim [Wed, 27 Jun 2018 16:00:53 +0000 (16:00 +0000)]
[X86][SSE] Add missing AVX512 rotation tests
Increase coverage to make sure we're not doing anything stupid without AVX512BW
llvm-svn: 335746
Craig Topper [Wed, 27 Jun 2018 15:57:57 +0000 (15:57 +0000)]
[X86] Rename llvm.x86.avx512.mask.fpclass.p* to exclude 'mask.' from the name to match llvm.
llvm-svn: 335745
Craig Topper [Wed, 27 Jun 2018 15:57:53 +0000 (15:57 +0000)]
[X86] Rename the autoupgraded of packed fp compare and fpclass intrinsics that don't take a mask as input to exclude '.mask.' from their name.
I think the intrinsics named 'avx512.mask.' should refer to the previous behavior of taking a mask argument in the intrinsic instead of using a 'select' or 'and' instruction in IR to accomplish the masking. This is more consistent with the goal that eventually we will have no intrinsics that have masking builtin. When we reach that goal, we should have no intrinsics named "avx512.mask".
llvm-svn: 335744
Fangrui Song [Wed, 27 Jun 2018 15:56:32 +0000 (15:56 +0000)]
[ELF] Rename RF_PROGBITS_NOT_EXEC_OR_WRITE to RF_RODATA
Post commit review at D48406
llvm-svn: 335743
Stanislav Mekhanoshin [Wed, 27 Jun 2018 15:33:33 +0000 (15:33 +0000)]
[AMDGPU] Convert rcp to rcp_iflag
If a source of rcp instruction is a result of any conversion from
an integer convert it into rcp_iflag instruction. No FP exception
can ever happen except division by zero if a single precision rcp
argument is a representation of an integral number.
Differential Revision: https://reviews.llvm.org/D48569
llvm-svn: 335742
Vedant Kumar [Wed, 27 Jun 2018 14:57:59 +0000 (14:57 +0000)]
[asan] Relax a test which depends on optimized debug info
On some ARM platforms this test depends on debug locations being
present on constant materialization code, which was eliminated in
r335497.
Relax the test to allow two outcomes: the backtrace either contains the
right line numbers, or no line numbers.
llvm-svn: 335741
Alexander Kornienko [Wed, 27 Jun 2018 14:56:12 +0000 (14:56 +0000)]
[analyzer] Allow registering custom statically-linked analyzer checkers
Summary:
Add an extension point to allow registration of statically-linked Clang Static
Analyzer checkers that are not a part of the Clang tree. This extension point
employs the mechanism used when checkers are registered from dynamically loaded
plugins.
Reviewers: george.karpenkov, NoQ, xazax.hun, dcoughlin
Reviewed By: george.karpenkov
Subscribers: mgorny, mikhail.ramalho, rnkovacs, xazax.hun, szepet, a.sidorin, cfe-commits
Differential Revision: https://reviews.llvm.org/D45718
llvm-svn: 335740
Mikhail R. Gadelha [Wed, 27 Jun 2018 14:39:41 +0000 (14:39 +0000)]
[analyzer] Fix string not being formatted with extra arguments
Signed-off-by: Mikhail Ramalho <mikhail.ramalho@gmail.com>
llvm-svn: 335739
George Rimar [Wed, 27 Jun 2018 14:37:02 +0000 (14:37 +0000)]
[ELF] - ICF: Add test case to check many sections case.
Currently, ICF does not enable threading if we have less than 1024
sections in each equivalence class.
And the following code is uncovered by our test cases:
https://github.com/llvm-mirror/lld/blob/master/ELF/ICF.cpp#L404
This patch adds a test case that triggers the mentioned code to execute.
llvm-svn: 335738
Luke Geeson [Wed, 27 Jun 2018 14:34:40 +0000 (14:34 +0000)]
[AArch64] Reverting FP16 vcvth_n_s64_f16 to fix
llvm-svn: 335737
Alexander Kornienko [Wed, 27 Jun 2018 14:30:55 +0000 (14:30 +0000)]
[clang-tidy] Add ExprMutationAnalyzer, that analyzes whether an expression is mutated within a statement.
Summary:
(Originally started as a clang-tidy check but there's already D45444 so shifted to just adding ExprMutationAnalyzer)
`ExprMutationAnalyzer` is a generally useful helper that can be used in different clang-tidy checks for checking whether a given expression is (potentially) mutated within a statement (typically the enclosing compound statement.)
This is a more general and more powerful/accurate version of isOnlyUsedAsConst, which is used in ForRangeCopyCheck, UnnecessaryCopyInitialization.
It should also be possible to construct checks like D45444 (suggest adding const to variable declaration) or https://bugs.llvm.org/show_bug.cgi?id=21981 (suggest adding const to member function) using this helper function.
This function is tested by itself and is intended to stay generally useful instead of tied to any particular check.
Reviewers: hokein, alexfh, aaron.ballman, ilya-biryukov, george.karpenkov
Reviewed By: aaron.ballman
Subscribers: lebedev.ri, shuaiwang, rnkovacs, hokein, alexfh, aaron.ballman, a.sidorin, Eugene.Zelenko, xazax.hun, JonasToth, klimek, mgorny, cfe-commits
Tags: #clang-tools-extra
Patch by Shuai Wang.
Differential Revision: https://reviews.llvm.org/D45679
llvm-svn: 335736
Adhemerval Zanella [Wed, 27 Jun 2018 13:58:46 +0000 (13:58 +0000)]
[AArch64] Add custom lowering for v4i8 trunc store
This patch adds a custom trunc store lowering for v4i8 vector types.
Since there is not v.4b register, the v4i8 is promoted to v4i16 (v.4h)
and default action for v4i8 is to extract each element and issue 4
byte stores.
A better strategy would be to extended the promoted v4i16 to v8i16
(with undef elements) and extract and store the word lane which
represents the v4i8 subvectores. The construction:
define void @foo(<4 x i16> %x, i8* nocapture %p) {
%0 = trunc <4 x i16> %x to <4 x i8>
%1 = bitcast i8* %p to <4 x i8>*
store <4 x i8> %0, <4 x i8>* %1, align 4, !tbaa !2
ret void
}
Can be optimized from:
umov w8, v0.h[3]
umov w9, v0.h[2]
umov w10, v0.h[1]
umov w11, v0.h[0]
strb w8, [x0, #3]
strb w9, [x0, #2]
strb w10, [x0, #1]
strb w11, [x0]
ret
To:
xtn v0.8b, v0.8h
str s0, [x0]
ret
The patch also adjust the memory cost for autovectorization, so the C
code:
void foo (const int *src, int width, unsigned char *dst)
{
for (int i = 0; i < width; i++)
*dst++ = *src++;
}
can be vectorized to:
.LBB0_4: // %vector.body
// =>This Inner Loop Header: Depth=1
ldr q0, [x0], #16
subs x12, x12, #4 // =4
xtn v0.4h, v0.4s
xtn v0.8b, v0.8h
st1 { v0.s }[0], [x2], #4
b.ne .LBB0_4
Instead of byte operations.
llvm-svn: 335735
Ivan A. Kosarev [Wed, 27 Jun 2018 13:58:43 +0000 (13:58 +0000)]
[NEON] Support vldNq intrinsics in AArch32 (Clang part)
This patch reworks the support for dup NEON intrinsics as
described in D48439.
Differential Revision: https://reviews.llvm.org/D48440
llvm-svn: 335734
Ivan A. Kosarev [Wed, 27 Jun 2018 13:57:52 +0000 (13:57 +0000)]
[NEON] Support vldNq intrinsics in AArch32 (LLVM part)
This patch adds support for the q versions of the dup
(load-to-all-lanes) NEON intrinsics, such as vld2q_dup_f16() for
example.
Currently, non-q versions of the dup intrinsics are implemented
in clang by generating IR that first loads the elements of the
structure into the first lane with the lane (to-single-lane)
intrinsics, and then propagating it other lanes. There are at
least two problems with this approach. First, there are no
double-spaced to-single-lane byte-element instructions. For
example, there is no such instruction as 'vld2.8 { d0[0], d2[0]
}, [r0]'. That means we cannot rely on the to-single-lane
intrinsics and instructions to implement the q versions of the
dup intrinsics. Note that to-all-lanes instructions do support
all sizes of data items, including bytes.
The second problem with the current approach is that we need a
separate vdup instruction to propagate the structure to each
lane. So for vld4q_dup_f16() we would need four vdup instructions
in addition to the initial vld instruction.
This patch introduces dup LLVM intrinsics and reworks handling of
the currently supported (non-q) NEON dup intrinsics to expand
them into those LLVM intrinsics, thus eliminating the need for
using to-single-lane intrinsics and instructions.
Additionally, this patch adds support for u64 and s64 dup NEON
intrinsics. These are marked as Arch64-only in the ARM NEON
Reference, but it seems there are no reasons to not support them
in AArch32 mode. Please correct, if that is wrong.
That's what we generate with this patch applied:
vld2q_dup_f16:
vld2.16 {d0[], d2[]}, [r0]
vld2.16 {d1[], d3[]}, [r0]
vld3q_dup_f16:
vld3.16 {d0[], d2[], d4[]}, [r0]
vld3.16 {d1[], d3[], d5[]}, [r0]
vld4q_dup_f16:
vld4.16 {d0[], d2[], d4[], d6[]}, [r0]
vld4.16 {d1[], d3[], d5[], d7[]}, [r0]
Differential Revision: https://reviews.llvm.org/D48439
llvm-svn: 335733
Zaara Syeda [Wed, 27 Jun 2018 13:55:41 +0000 (13:55 +0000)]
[PPC64] Add support for R_PPC64_GOT_DTPREL16* relocations
The local dynamic TLS access on PPC64 ELF v2 ABI uses R_PPC64_GOT_DTPREL16*
relocations when a TLS variables falls outside 2 GB of the thread storage
block. This patch adds support for these relocations by adding a new RelExpr
called R_TLSLD_GOT_OFF which emits a got entry for the TLS variable relative
to the dynamic thread pointer using the relocation R_PPC64_DTPREL64. It then
evaluates the R_PPC64_GOT_DTPREL16* relocations as the got offset for the
R_PPC64_DTPREL64 got entries.
Differential Revision: https://reviews.llvm.org/D48484
llvm-svn: 335732
Gabor Marton [Wed, 27 Jun 2018 13:32:50 +0000 (13:32 +0000)]
Re-apply: [ASTImporter] Import the whole redecl chain of functions
Summary:
With this patch when any `FunctionDecl` of a redeclaration chain is imported
then we bring in the whole declaration chain. This involves functions and
function template specializations. Also friend functions are affected. The
chain is imported as it is in the "from" tu, the order of the redeclarations
are kept. I also changed the lookup logic in order to find friends, but first
making them visible in their declaration context. We may have long
redeclaration chains if all TU contains the same prototype, but our
measurements shows no degradation in time of CTU analysis (Tmux, Xerces,
Bitcoin, Protobuf). Also, as further work we could squash redundant
prototypes, but first ensure that functionality is working properly; then
should we optimize.
This may seem like a huge patch, sorry about that. But, most of the changes are
new tests, changes in the production code is not that much. I also tried to
create a smaller patch which does not affect specializations, but that patch
failed to pass some of the `clang-import-test`s because there we import
function specializations. Also very importantly, we can't just change the
import of `FunctionDecl`s without changing the import of function template
specializations because they are handled as `FunctionDecl`s.
Reviewers: a.sidorin, r.stahl, xazax.hun, balazske, a_sidorin
Reviewed By: a_sidorin
Subscribers: labath, aprantl, a_sidorin, rnkovacs, dkrupp, cfe-commits
Differential Revision: https://reviews.llvm.org/D47532
Re-apply commit rC335480
llvm-svn: 335731
Zaara Syeda [Wed, 27 Jun 2018 13:27:29 +0000 (13:27 +0000)]
[PPC64] Add TLS global dynamic to local exec relaxation
This patch adds the target call back relaxTlsGdToLe to support TLS relaxation
from global dynamic to local exec model.
The relaxation performs the following transformation:
addis r3, r2, x@got@tlsgd@ha --> nop
addi r3, r3, x@got@tlsgd@l --> addis r3, r13, x@tprel@ha
bl __tls_get_addr(x@tlsgd) --> nop
nop --> addi r3, r3, x@tprel@l
Differential Revision: https://reviews.llvm.org/D48082
llvm-svn: 335730
Florian Hahn [Wed, 27 Jun 2018 12:57:51 +0000 (12:57 +0000)]
[ValueLattice] Return false if value range did not change in mergeIn.
llvm-svn: 335729
Dan Liew [Wed, 27 Jun 2018 12:56:34 +0000 (12:56 +0000)]
[CMake] Tidy up the organisation of compiler-rt when configured as a standalone
build with an IDE (e.g. Xcode) as the generator.
Previously the global `USE_FOLDERS` property wasn't set in standalone
builds leading to existing settings of FOLDER not being respected.
In addition to this there were several targets that appeared at the top
level that were not interesting and clustered up the view. These have
been changed to be displayed in "Compiler-RT Misc".
Now when an Xcode project is generated from a standalone compiler-rt
build the project navigator is much less cluttered. The interesting
libraries should appear in "Compiler-RT Libraries" in the IDE.
Differential Revision: https://reviews.llvm.org/D48378
llvm-svn: 335728
Simon Pilgrim [Wed, 27 Jun 2018 12:45:31 +0000 (12:45 +0000)]
[DAGCombiner] visitSDIV - add special case handling for (sdiv X, 1) -> X in pow2 expansion
For divisor = 1, perform a select of X - reduces scalarisation of simple SDIVs
llvm-svn: 335727