Martin Storsjo [Tue, 20 Mar 2018 20:37:51 +0000 (20:37 +0000)]
[X86] Don't use the MSVC stack protector names on mingw
Mingw uses the same stack protector functions as GCC provides
on other platforms as well.
Patch by Valentin Churavy!
Differential Revision: https://reviews.llvm.org/D27296
llvm-svn: 328039
Abderrazek Zaafrani [Tue, 20 Mar 2018 20:37:31 +0000 (20:37 +0000)]
[AArch64] Add vmulxh_lane fp16 vector intrinsic
https://reviews.llvm.org/D44591
llvm-svn: 328038
Kevin Enderby [Tue, 20 Mar 2018 20:29:52 +0000 (20:29 +0000)]
For llvm-objdump and Mach-O files, fix the printing of module init and
term sections from .o files to look to see if the pointers have a relocation
entry and if so print the symbol name from the relocation entry. If not fall
back to the existing code and use the pointer value to look up that value
in the symbol table.
rdar://
38337506
llvm-svn: 328037
Rafael Espindola [Tue, 20 Mar 2018 20:27:30 +0000 (20:27 +0000)]
Set dso_local for runtime function.
This is another case where there is special logic for adding dllimport
and so we cannot use setGVProperties.
llvm-svn: 328036
Abderrazek Zaafrani [Tue, 20 Mar 2018 20:25:40 +0000 (20:25 +0000)]
[AArch64] Add vmulxh_lane fp16 vector intrinsic
https://reviews.llvm.org/D44591
llvm-svn: 328035
Craig Topper [Tue, 20 Mar 2018 20:24:16 +0000 (20:24 +0000)]
[TableGen] Use range based for loop. NFC
llvm-svn: 328034
Craig Topper [Tue, 20 Mar 2018 20:24:14 +0000 (20:24 +0000)]
[TableGen] Use vector::append instead of looping and calling push_back. NFC
Both vectors contain unsigned so we can just use append to do the copying. Not only is this shorter, but it should be able to predict the final size and only grow the vector once if needed.
llvm-svn: 328033
Craig Topper [Tue, 20 Mar 2018 20:24:12 +0000 (20:24 +0000)]
[TableGen] Use llvm::transform to simplify some loops. NFCI
llvm-svn: 328032
Craig Topper [Tue, 20 Mar 2018 20:24:10 +0000 (20:24 +0000)]
[TableGen] Pass result of std::unique to vector::erase instead of calculating a size and calling resize.
llvm-svn: 328031
Alexey Bataev [Tue, 20 Mar 2018 20:21:38 +0000 (20:21 +0000)]
[DEBUGINFO] Add -no-dwarf-debug-ranges option.
Summary:
Added option -no-dwarf-debug-ranges option to disable emission of
.debug_ranges section.
Reviewers: probinson, echristo
Subscribers: aprantl, JDevlieghere, llvm-commits
Differential Revision: https://reviews.llvm.org/D44384
llvm-svn: 328030
Andrea Di Biagio [Tue, 20 Mar 2018 20:18:36 +0000 (20:18 +0000)]
[llvm-mca] add keyword override to a couple of methods in BackendStatistics.
This should fix the buildbots after r328011.
llvm-svn: 328029
Derek Schuff [Tue, 20 Mar 2018 20:06:35 +0000 (20:06 +0000)]
[WebAssembly] Added initial AsmParser implementation.
It uses the MC framework and the tablegen matcher to do the
heavy lifting. Can handle both explicit and implicit locals
(-disable-wasm-explicit-locals). Comes with a small regression
test.
This is a first basic implementation that can parse most llvm .s
output and round-trips most instructions succesfully, but in order
to keep the commit small, does not address all issues.
There are a fair number of mismatches between what MC / assembly
matcher think a "CPU" should look like and what WASM provides,
some already have workarounds in this commit (e.g. the way it
deals with register operands) and some that require further work.
Some of that further work may involve changing what the
Disassembler outputs (and what s2wasm parses), so are probably
best left to followups.
Some known things missing:
- Many directives are ignored and not emitted.
- Vararg calls are parsed but extra args not emitted.
- Loop signatures are likely incorrect.
- $drop= is not emitted.
- Disassembler does not output SIMD types correctly, so assembler
can't test them.
Patch by Wouter van Oortmerssen
Differential Revision: https://reviews.llvm.org/D44329
llvm-svn: 328028
Evandro Menezes [Tue, 20 Mar 2018 20:00:29 +0000 (20:00 +0000)]
[AArch64] Adjust the cost model for Exynos M3
Fix typo in the number of integer dividers.
llvm-svn: 328027
Alexander Shaposhnikov [Tue, 20 Mar 2018 19:50:14 +0000 (19:50 +0000)]
[llvm-objcopy] Revert the tests from r328012
Temporarily revert the tests from r328012 as well.
llvm-svn: 328026
Davide Italiano [Tue, 20 Mar 2018 19:46:32 +0000 (19:46 +0000)]
[ExpressionParser] Re-implement r327356 in a less disruptive way.
Instead of applying the sledgehammer of refusing to insert any
C++ symbol in the ASTContext, try to validate the decl if what
we have is an operator. There was other code in lldb which was
responsible for this, just not really exposed (or used) in this
codepath. Also, add a better/more comprehensive test.
<rdar://problem/
35645893>
llvm-svn: 328025
Alexander Shaposhnikov [Tue, 20 Mar 2018 19:46:00 +0000 (19:46 +0000)]
[llvm-objcopy] Revert r328012
Temporarily revert r328012 (since it broke down the big-endian bots),
will resubmit an updated version later.
llvm-svn: 328024
Krzysztof Parzyszek [Tue, 20 Mar 2018 19:35:09 +0000 (19:35 +0000)]
[Hexagon] Add a few more lit tests, NFC
llvm-svn: 328023
Krzysztof Parzyszek [Tue, 20 Mar 2018 19:26:27 +0000 (19:26 +0000)]
[Hexagon] Add heuristic to exclude critical path cost for scheduling
Patch by Brendon Cahoon.
llvm-svn: 328022
Krzysztof Parzyszek [Tue, 20 Mar 2018 19:23:18 +0000 (19:23 +0000)]
[Hexagon] Fix fall-through warnings in HexagonMCDuplexInfo.cpp
llvm-svn: 328021
Jonas Devlieghere [Tue, 20 Mar 2018 19:18:11 +0000 (19:18 +0000)]
[lldb-dotest] Wrap arguments in single quotes
If we don't wrap arguments to the wrapper in single quotes, combined
arguments, for example for -E, don't reach dotest.py as a unit but as
separate arguments, causing the latter to fail.
llvm-svn: 328020
Nirav Dave [Tue, 20 Mar 2018 19:12:41 +0000 (19:12 +0000)]
[MC,X86] Cleanup some X86 parser functions to use MCParser helpers. NFCI.
llvm-svn: 328019
Andrea Di Biagio [Tue, 20 Mar 2018 19:06:34 +0000 (19:06 +0000)]
[llvm-mca] Remove const from a bunch of ArrayRef. NFC
llvm-svn: 328018
Craig Topper [Tue, 20 Mar 2018 18:49:28 +0000 (18:49 +0000)]
[PowerPC][LegalizeFloatTypes] Move the PPC hacks for (i32 fp_to_sint/fp_to_uint (ppcf128 X)) out of LegalizeFloatTypes and into PPC specific code
I'm not entirely sure these hacks are still needed. If you remove the hacks completely, the name of the library call that gets generated doesn't match the grep the test previously had. So the test wasn't really checking anything.
If the hack is still needed it belongs in PPC specific code. I believe the FP_TO_SINT code here is the only place in the tree where a FP_ROUND_INREG node is created today. And I don't think its even being used correctly because the legalization returned a BUILD_PAIR with the same value twice. That doesn't seem right to me. By moving the code entirely to PPC we can avoid creating the FP_ROUND_INREG at all.
I replaced the grep in the existing test with full checks generated by hacking update_llc_test_check.py to support ppc32 just long enough to generate it.
Differential Revision: https://reviews.llvm.org/D44061
llvm-svn: 328017
Krzysztof Parzyszek [Tue, 20 Mar 2018 18:46:55 +0000 (18:46 +0000)]
[X86] Add phony registers for high halves of regs with low halves
Registers E[A-D]X, E[SD]I, E[BS]P, and EIP have 16-bit subregisters
that cover the low halves of these registers. This change adds artificial
subregisters for the high halves in order to differentiate (in terms of
register units) between the 32- and the low 16-bit registers.
This patch contains parts that aim to preserve the calculated register
pressure. This is in order to preserve the current codegen (minimize the
impact of this patch). The approach of having artificial subregisters
could be used to fix PR23423, but the pressure calculation would need
to be changed.
Differential Revision: https://reviews.llvm.org/D43353
llvm-svn: 328016
Philip Reames [Tue, 20 Mar 2018 18:43:44 +0000 (18:43 +0000)]
[MustExecute] Use the annotation style printer
As suggested in the original review (https://reviews.llvm.org/D44524), use an annotation style printer instead.
Note: The switch from -analyze to -disable-output in tests was driven by the fact that seems to be the idiomatic style used in annoation passes. I tried to keep both working, but the old style pass API for printers really doesn't make this easy. It invokes (runOnFunction, print(Module)) repeatedly. I decided the extra state wasn't worth it given the old pass manager is going away soonish anyway.
llvm-svn: 328015
Zachary Turner [Tue, 20 Mar 2018 18:37:03 +0000 (18:37 +0000)]
Revert "Resubmit "Support embedding natvis files in PDBs.""
This is still failing on a different bot this time due to some
issue related to hashing absolute paths. Reverting until I can
figure it out.
llvm-svn: 328014
Davide Italiano [Tue, 20 Mar 2018 18:32:47 +0000 (18:32 +0000)]
[TestExpr] Fix a typo in a test, unbreaking the lldb Mac OS X bot.
llvm-svn: 328013
Alexander Shaposhnikov [Tue, 20 Mar 2018 18:20:42 +0000 (18:20 +0000)]
[llvm-objcopy] Implement support for section groups
This diff adds support for SHT_GROUP sections to llvm-objcopy.
Some sections are interrelated and comprise a group.
For example, a definition of an inline function might require,
in addition to the section containing its instructions,
a read-only data section containing literals referenced inside the function.
A section of the type SHT_GROUP contains the indices of the group members,
therefore, it needs to be updated whenever the indices change.
Similarly, the fields sh_link, sh_info should be recalculated as well.
Test plan: make check-all
Differential revision: https://reviews.llvm.org/D43996
llvm-svn: 328012
Andrea Di Biagio [Tue, 20 Mar 2018 18:20:39 +0000 (18:20 +0000)]
[llvm-mca] Move the logic that computes the scheduler's queue usage to the BackendStatistics view.
This patch introduces two new callbacks in the event listener interface to
handle the "buffered resource reserved" event and the "buffered resource
released" event. Every time a buffered resource is used, an event is generated.
Before this patch, the Scheduler (with the help of the ResourceManager) was
responsible for tracking the scheduler's queue usage. However, that design
forced the Scheduler to 'publish' scheduler's queue pressure information through
the Backend interface.
The goal of this patch is to break the dependency between the BackendStatistics
view, and the Backend. Now the Scheduler knows how to notify "buffer
reserved/released" events. The scheduler's queue usage analysis has been moved
to the BackendStatistics.
Differential Revision: https://reviews.llvm.org/D44686
llvm-svn: 328011
Eric Christopher [Tue, 20 Mar 2018 18:10:30 +0000 (18:10 +0000)]
Fix consitent -> consistent.
llvm-svn: 328010
Fangrui Song [Tue, 20 Mar 2018 18:07:25 +0000 (18:07 +0000)]
[ELF] Simplify compCtors and writeInt.
Reviewers: ruiu
Subscribers: emaste, arichardson, llvm-commits
Differential Revision: https://reviews.llvm.org/D44690
llvm-svn: 328009
Saleem Abdulrasool [Tue, 20 Mar 2018 17:33:26 +0000 (17:33 +0000)]
Basic: support PreserveMost and PreserveAll on Windows ARM
Do not ignore these calling conventions on Windows ARM. They are used
by the swift runtime for certain calls.
llvm-svn: 328007
Artem Belevich [Tue, 20 Mar 2018 17:18:59 +0000 (17:18 +0000)]
[NVPTX] Make tensor load/store intrinsics overloaded.
This way we can support address-space specific variants without explicitly
encoding the space in the name of the intrinsic. Less intrinsics to deal with ->
less boilerplate.
Added a bit of tablegen magic to match/replace an intrinsics with a pointer
argument in particular address space with the space-specific instruction
variant.
Updated tests to use non-default address spaces.
Differential Revision: https://reviews.llvm.org/D43268
llvm-svn: 328006
Tobias Grosser [Tue, 20 Mar 2018 17:16:32 +0000 (17:16 +0000)]
Adjust to clang-format changes
llvm-svn: 328005
Philip Reames [Tue, 20 Mar 2018 17:09:21 +0000 (17:09 +0000)]
Add an analysis printer for must execute reasoning
Many of our loop passes make use of so called "must execute" or "guaranteed to execute" facts to prove the legality of code motion. The basic notion is that we know (by assumption) an instruction didn't fault at it's original location, so if the location we move it to is strictly post dominated by the original, then we can't have introduced a new fault.
At the moment, the testing for this logic is somewhat adhoc and done mostly through LICM. Since I'm working on that code, I want to improve the testing. This patch is the first step in that direction. It doesn't actually test the variant used by the loop passes - I need to move that to the Analysis library first - but instead exercises an alternate implementation used by SCEV. (I plan on merging both implementations.)
Note: I'll be replacing the printing logic within this with an annotation based version in the near future. Anna suggested this in review, and it seems like a strictly better format.
Differential Revision: https://reviews.llvm.org/D44524
llvm-svn: 328004
Rafael Espindola [Tue, 20 Mar 2018 17:07:25 +0000 (17:07 +0000)]
MC: fix layering violation introduced in r325139.
Move AsmToken from MC/MCParser/MCAsmLexer.h into MC/MCAsmMacro.h in
order to avoid MCAsmMacro definition depending on MCParser.
Patch by Roger Pau Monné <royger@freebsd.org>.
llvm-svn: 328003
Zachary Turner [Tue, 20 Mar 2018 17:06:39 +0000 (17:06 +0000)]
Resubmit "Support embedding natvis files in PDBs."
The issue causing this to fail in certain configurations
should be fixed.
It was due to the fact that DIA apparently expects there to be
a null string at ID 1 in the string table. I'm not sure why this
is important but it seems to make a difference, so set it.
llvm-svn: 328002
Sanjay Patel [Tue, 20 Mar 2018 17:05:19 +0000 (17:05 +0000)]
[LangRef] fix link formatting
llvm-svn: 328001
Krzysztof Parzyszek [Tue, 20 Mar 2018 17:03:27 +0000 (17:03 +0000)]
[Hexagon] Correct the computation of TopReadyCycle and BotReadyCycle of SU
TopReadyCycle and BotReadyCycle were off by one cycle when an SU is either
the first instruction or the last instruction in a packet.
Patch by Ikhlas Ajbar.
llvm-svn: 328000
Michael Zolotukhin [Tue, 20 Mar 2018 17:02:29 +0000 (17:02 +0000)]
[XRay] Lazily compute MachineLoopInfo instead of requiring it.
Summary:
Currently X-Ray Instrumentation pass has a dependency on MachineLoopInfo
(and thus on MachineDominatorTree as well) and we have to compute them
even if X-Ray is not used. This patch changes it to a lazy computation
to save compile time by avoiding these redundant computations.
Reviewers: dberris, kubamracek
Subscribers: llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D44666
llvm-svn: 327999
Sanjay Patel [Tue, 20 Mar 2018 16:38:22 +0000 (16:38 +0000)]
[LangRef] describe the default FP environment
Follow-up for D44216: add a section and examples to describe the FP env.
Also, add pointers from the FP instructions to this new section to reduce
bloat.
Differential Revision: https://reviews.llvm.org/D44318
llvm-svn: 327998
Krzysztof Parzyszek [Tue, 20 Mar 2018 16:22:06 +0000 (16:22 +0000)]
[Hexagon] Check weak dependences when only 1 instruction is available
Patch by Brendon Cahoon.
llvm-svn: 327997
Pavel Labath [Tue, 20 Mar 2018 16:14:00 +0000 (16:14 +0000)]
Move StringExtractorGDBRemote.h to the include folder
While trying to use this header I noticed that it is not in the include
folder. Move it to there and update all #includes to reference that file
correctly.
llvm-svn: 327996
Sanjay Patel [Tue, 20 Mar 2018 16:08:47 +0000 (16:08 +0000)]
[AArch64] add fabs tests for PR36600; NFC
llvm-svn: 327995
Alexey Bataev [Tue, 20 Mar 2018 16:04:40 +0000 (16:04 +0000)]
[DEBUGINFO] Add flag -no-dwarf-pub-sections to disable pub sections.
Summary:
Added a flag -no-dwarf-pub-sections, which allows to disable
emission of DWARF public sections.
Reviewers: probinson, echristo
Subscribers: aprantl, JDevlieghere, llvm-commits
Differential Revision: https://reviews.llvm.org/D44385
llvm-svn: 327994
Rafael Espindola [Tue, 20 Mar 2018 15:48:00 +0000 (15:48 +0000)]
Set dso_local for CFConstantStringClassReference.
This one cannot use setGVProperties since it has special logic for
when it is dllimport or not.
llvm-svn: 327993
Simon Pilgrim [Tue, 20 Mar 2018 15:44:47 +0000 (15:44 +0000)]
[X86][Haswell][Znver1] Fix typo in fldl instregexs
Missing comma was casing 2 instregex entries to be concatenated together by mistake.
Found while investigating PR35548
llvm-svn: 327992
Rafael Espindola [Tue, 20 Mar 2018 15:42:58 +0000 (15:42 +0000)]
Set dso_local for guid decls.
llvm-svn: 327991
Alexey Bataev [Tue, 20 Mar 2018 15:41:05 +0000 (15:41 +0000)]
[OPENMP, NVPTX] Codegen for target distribute parallel combined
constructs in generic mode.
Fixed codegen for distribute parallel combined constructs. We have to
pass and read the shared lower and upper bound from the distribute
region in the inner parallel region. Patch is for generic mode.
llvm-svn: 327990
Sjoerd Meijer [Tue, 20 Mar 2018 15:37:15 +0000 (15:37 +0000)]
[Kaleidoscope] doc fix
This fixes the compile command of the example in Chapter 2.
Differential Revision: https://reviews.llvm.org/D43196
llvm-svn: 327989
Krzysztof Parzyszek [Tue, 20 Mar 2018 14:54:01 +0000 (14:54 +0000)]
[Hexagon] Improve scheduling heuristic for large basic blocks
This patch changes the isLatencyBound heuristic to look at the
path length based upon the number of packets needed to schedule
a basic block. For small basic blocks, the heuristic uses a small
threshold for isLatencyBound. For large basic blocks, the
heuristic uses a large threshold.
The goal is to increase the priority of an instruction in a small
basic block that has a large height or depth relative to the code
size. For large functions, the height and depth are ignored
because it increases the live range of a register and causes more
spills. That is, for large functions, it is more important to
schedule instructions when available, and attempt to keep the defs
and uses closer together.
Patch by Brendon Cahoon.
llvm-svn: 327987
Ben Hamilton [Tue, 20 Mar 2018 14:53:25 +0000 (14:53 +0000)]
[clang-format] Fix ObjC selectors with multiple params passed to macro
Summary:
Objective-C selectors with arguments take the form of:
foo:
foo:bar:
foo:bar:baz:
These can be passed to a macro, like NS_SWIFT_NAME():
https://developer.apple.com/library/content/documentation/Swift/Conceptual/BuildingCocoaApps/MixandMatch.html
and must never have spaces inserted around the colons.
Previously, there was logic in TokenAnnotator's tok::colon parser to
handle the single-argument case, but it failed for the
multiple-argument cases.
This diff fixes the bug and adds more tests.
Test Plan: New tests added. Ran tests with:
% make -j12 FormatTests && ./tools/clang/unittests/Format/FormatTests
Reviewers: jolesiak, djasper, Wizard
Reviewed By: jolesiak, Wizard
Subscribers: klimek, cfe-commits
Differential Revision: https://reviews.llvm.org/D44638
llvm-svn: 327986
Alexey Bataev [Tue, 20 Mar 2018 14:45:59 +0000 (14:45 +0000)]
[OPENMP, NVPTX] Globalization of the private redeclarations.
If the generic codegen is enabled and private copy of the original
variable escapes the declaration context, this private copy should be
globalized just like it was the original variable.
llvm-svn: 327985
Ilya Biryukov [Tue, 20 Mar 2018 14:39:12 +0000 (14:39 +0000)]
Backport changes from llvm/.clang_tidy to clang/.clang_tidy configs
Summary: LLVM .clang_tidy seems to be more up-to-date.
Reviewers: alexfh, simark
Reviewed By: alexfh
Subscribers: simark, cfe-commits
Differential Revision: https://reviews.llvm.org/D44628
llvm-svn: 327984
George Rimar [Tue, 20 Mar 2018 14:22:10 +0000 (14:22 +0000)]
[ELF] - Make __start_/__stop_<section_name> symbols STV_PROTECTED
There are no reasons for them to be STV_DEFAULT,
recently bfd did the same change.
Differential revision: https://reviews.llvm.org/D44566
llvm-svn: 327983
Geoff Berry [Tue, 20 Mar 2018 13:46:35 +0000 (13:46 +0000)]
[AArch64][Falkor] Correct load/store increment scheduling details
llvm-svn: 327982
Pavel Labath [Tue, 20 Mar 2018 13:46:05 +0000 (13:46 +0000)]
[LLDB] Fix TestTargetXMLArch's expected arch
Summary:
When running on an architecture other than x86_64, the
target.ConnectRemote() part of the test may add platform information to
the target triple.
It was observed that this happens at Process::CompleteAttach() method,
after the platform_sp->IsCompatibleArchitecture() check fails.
This method then calls platform_sp->GetPlatformForArchitecture(), that
on a Linux machine ends up returning a generic Linux platform, that then
ends up getting added to the original target architecture.
Reviewers: clayborg, labath, jasonmolenda
Reviewed By: labath
Subscribers: alexandreyy, lbianc
Differential Revision: https://reviews.llvm.org/D44022
Patch by Leandro Lupori <leandro.lupori@gmail.com>.
llvm-svn: 327981
Krzysztof Parzyszek [Tue, 20 Mar 2018 13:28:46 +0000 (13:28 +0000)]
[Hexagon] Fix division by zero in machine scheduler
llvm-svn: 327980
Alex Bradbury [Tue, 20 Mar 2018 13:26:12 +0000 (13:26 +0000)]
[RISCV] Add codegen for RV32F floating point load/store
As part of this, add support for load/store from the constant pool. This is
used to materialise f32 constants.
llvm-svn: 327979
Andrea Di Biagio [Tue, 20 Mar 2018 12:58:34 +0000 (12:58 +0000)]
[llvm-mca] Use llvm::make_unique in a few places. NFC
Also, clang-format a couple of DEBUG functions.
llvm-svn: 327978
Pavel Labath [Tue, 20 Mar 2018 12:46:33 +0000 (12:46 +0000)]
Fix some tests for PPC64le architecture
Summary:
- Fix test jump for powerpc64le
Jumping directly to the return line on power architecture dos not means
returning the value that is seen on the code. The last test fails, because
it needs the execution of some assembly in the beginning of the function.
Avoiding this test for this architecture.
- Avoid evaluate environ variable name on Linux
On Linux the Symbol environ conflicts with another variable, then in
order to avoid it, this test was moved into a specific test, which is not
supported if the OS is Linux.
- Added PPC64le as MIPS behavior
Checking the disassembler output, on PPC64le machines behaves as MPIS.
Added method to identify PPC64le architecture and checking it when
disassembling instructions in the test case.
Reviewers: labath
Reviewed By: labath
Subscribers: clayborg, labath, luporl, alexandreyy, sdardis, ki.stfu, arichardson
Differential Revision: https://reviews.llvm.org/D44101
Patch by Leonardo Bianconi <leonardo.bianconi@eldorado.org.br>.
llvm-svn: 327977
Alex Bradbury [Tue, 20 Mar 2018 12:45:35 +0000 (12:45 +0000)]
[RISCV] Add codegen for RV32F arithmetic and conversion operations
Currently, only a soft floating point ABI is supported.
llvm-svn: 327976
Krzysztof Parzyszek [Tue, 20 Mar 2018 12:28:43 +0000 (12:28 +0000)]
[Hexagon] Improve scheduling based on register pressure
Patch by Brendon Cahoon.
llvm-svn: 327975
Simon Pilgrim [Tue, 20 Mar 2018 12:26:55 +0000 (12:26 +0000)]
[X86][SandyBridge] Merge multiple InstrRW entries that map to the same SchedWriteRes group (NFCI) (PR35955)
I've also merged some VEX/non-VEX instregex strings with a (V?) prefix - there are still a lot more of these to do.
llvm-svn: 327974
Andrea Di Biagio [Tue, 20 Mar 2018 12:25:54 +0000 (12:25 +0000)]
[llvm-mca] Move the routine that computes processor resource masks to its own file.
Function computeProcResourceMasks is used by the ResourceManager (owned by the
Scheduler) to compute resource masks for processor resources. Before this
refactoring, there was an implicit dependency between the Scheduler and the
InstrBuilder. That is because InstrBuilder has to know about resource masks when
computing the set of processor resources consumed by a new instruction.
With this patch, the functionality that computes resource masks has been
extracted from the ResourceManager, and moved to a separate file (Support.h).
This helps removing the dependency between the Scheduler and the InstrBuilder.
No functional change intended.
llvm-svn: 327973
Xin Tong [Tue, 20 Mar 2018 12:03:25 +0000 (12:03 +0000)]
[MergeICmps] Break eargerly out of loop
llvm-svn: 327972
Xin Tong [Tue, 20 Mar 2018 11:57:54 +0000 (11:57 +0000)]
[MergeICmp] Fix a bug in entry block shuffled to middle of the chain
Summary: Fix a bug in entry block shuffled to middle of the chain.
Reviewers: davide, courbet
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D44642
llvm-svn: 327971
Pavel Labath [Tue, 20 Mar 2018 11:56:24 +0000 (11:56 +0000)]
Re-land: [lldb] Use vFlash commands when writing to target's flash memory regions
The difference between this and the previous patch is that now we use
ELF physical addresses only for loading objects into the target (and the
rest of the module load address logic still uses virtual addresses).
Summary:
When writing an object file over gdb-remote, use the vFlashErase, vFlashWrite, and vFlashDone commands if the write address is in a flash memory region. A bare metal target may have this kind of setup.
- Update ObjectFileELF to set load addresses using physical addresses. A typical case may be a data section with a physical address in ROM and a virtual address in RAM, which should be loaded to the ROM address.
- Add support for querying the target's qXfer:memory-map, which contains information about flash memory regions, leveraging MemoryRegionInfo data structures with minor modifications
- Update ProcessGDBRemote to use vFlash commands in DoWriteMemory when the target address is in a flash region
Original discussion at http://lists.llvm.org/pipermail/lldb-dev/2018-January/013093.html
Reviewers: clayborg, labath
Reviewed By: labath
Subscribers: llvm-commits, arichardson, emaste, mgorny, lldb-commits
Differential Revision: https://reviews.llvm.org/D42145
Patch by Owen Shaw <llvm@owenpshaw.net>.
llvm-svn: 327970
Igor Laevsky [Tue, 20 Mar 2018 11:32:13 +0000 (11:32 +0000)]
[llvm-opt-fuzzer] Add irce to the fuzzing options
llvm-svn: 327969
Jonas Devlieghere [Tue, 20 Mar 2018 11:26:50 +0000 (11:26 +0000)]
Revert "Fix layering between llvm-c and Support by factoring out some typedefs into Support"
This reverts r327940 because it fails on several GreenDragon bots.
http://green.lab.llvm.org/green/job/clang-stage2-configure-Rlto/
http://green.lab.llvm.org/green/job/clang-stage2-coverage-R/
llvm-svn: 327968
Dylan McKay [Tue, 20 Mar 2018 11:23:03 +0000 (11:23 +0000)]
[AVR] Add a regression test for struct return lowering
The test is taken from
https://github.com/avr-rust/rust/issues/57
The originally implementation of struct return lowering was made in
r325474.
Patch by Peter Nimmervoll
llvm-svn: 327967
Hans Wennborg [Tue, 20 Mar 2018 10:43:02 +0000 (10:43 +0000)]
Revert r327964 "lit: Make config.llvm_libxml2_enabled an int, not a string"
It seems @LLVM_LIBXML2_ENABLED@ doesn't always get set to 0 when not available,
but to nothing, which broke parsing of lit.site.cfg.py.
> @LLVM_LIBXML2_ENABLED@ will be replaced with 0 or 1. Putting quotes
> around that is unnecessary and just makes it harder to use the value.
> This matches what have_zlib does below.
>
> This also puts the flag together with the feature-related ones instead
> of the path-related flags.
llvm-svn: 327966
Andrea Di Biagio [Tue, 20 Mar 2018 10:25:36 +0000 (10:25 +0000)]
[Release Notes] Add release note for llvm-mca.
Differential Revision: https://reviews.llvm.org/D44636
llvm-svn: 327965
Hans Wennborg [Tue, 20 Mar 2018 10:14:31 +0000 (10:14 +0000)]
lit: Make config.llvm_libxml2_enabled an int, not a string
@LLVM_LIBXML2_ENABLED@ will be replaced with 0 or 1. Putting quotes
around that is unnecessary and just makes it harder to use the value.
This matches what have_zlib does below.
This also puts the flag together with the feature-related ones instead
of the path-related flags.
llvm-svn: 327964
George Rimar [Tue, 20 Mar 2018 09:49:03 +0000 (09:49 +0000)]
[ELF] - Add test case for "unsupported relocation reference" error. NFCI.
We do not have test showing we explicitly reject objects
where relocation section goes before the target, i.e
.rel[a].text is listed before .text, for example.
The patch adds it.
llvm-svn: 327963
Henry Wong [Tue, 20 Mar 2018 09:27:02 +0000 (09:27 +0000)]
[analyzer] Fix the crash in IteratorChecker.cpp when 'SymbolConjured' has a null Stmt.
When the loop has a null terminator statement and sets 'widen-loops=true', 'invalidateRegions' will constructs the 'SymbolConjured' with null 'Stmt'. And this will lead to a crash in 'IteratorChecker.cpp'. This patch use 'dyn_cast_or_null<>' instead of 'dyn_cast<>' in IteratorChecker.cpp.
Differential Revision: https://reviews.llvm.org/D44606
llvm-svn: 327962
Bjorn Pettersson [Tue, 20 Mar 2018 09:06:37 +0000 (09:06 +0000)]
[CGP] Avoid segmentation fault when doing PHI node simplifications
Summary:
Made PHI node simplifiations more robust in several ways:
- Minor refactoring to let the SimplificationTracker own the
sets with new PHI/Select nodes that are introduced. This is
maybe not mapping to the original intention with the
SimplificationTracker, but IMHO it encapsulates the logic behind
those sets a little bit better.
- MatchPhiNode can sometimes populate the Matched set with
several entries, where it maps one PHI node to different candidates
for replacement. The Matched set is changed into a SmallSetVector
to make sure we get a deterministic iteration when doing
the replacements.
- As described above we may get several different replacements
for a single PHI node. The loop in MatchPhiSet that is doing
the replacements could end up calling eraseFromParent several
times for the same PHI node, resulting in segmentation faults.
This problem was supposed to be fixed in rL327250, but due to
the non-determinism(?) it only appeared to be fixed (I still
got crashes sometime when turning on/off -print-after-all etc
to get different iteration order in the DenseSets).
With this patch we follow the deterministic ordering in the
Matched set when replacing the PHI nodes. If we find a new
replacement for an already replaced PHI node we replace the
new replacement by the old replacement instead. This is quite
similar to what happened in the rl327250 patch, but here we
also recursively verify that the old replacement hasn't been
replaced already.
- It was really hard to track down the fault described above
(segementation fault due to doing eraseFromParent multiple
times for the same instruction). The fault was intermittent and
small changes in the code, or simply turning on -print-after-all
etc could make the problem go away. This was basically due to
the iteration over PhiNodesToMatch in MatchPhiSet no being
deterministic. Therefore I've changed the data structure for
the SimplificationTracker::AllPhiNodes into an SmallSetVector.
This gives a deterministic behavior.
Reviewers: skatkov, john.brawn
Reviewed By: skatkov
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D44571
llvm-svn: 327961
Andrei Elovikov [Tue, 20 Mar 2018 09:04:39 +0000 (09:04 +0000)]
[LV] Let recordVectorLoopValueForInductionCast to check if IV was created from the cast.
Summary:
It turned out to be error-prone to expect the callers to handle that - better to
leave the decision to this routine and make the required data to be explicitly
passed to the function.
This handles the case that was missed in the r322473 and fixes the assert
mentioned in PR36524.
Reviewers: dorit, mssimpso, Ayal, dcaballe
Reviewed By: dcaballe
Subscribers: Ka-Ka, hiraditya, dneilson, hsaito, llvm-commits
Differential Revision: https://reviews.llvm.org/D43812
llvm-svn: 327960
Hans Wennborg [Tue, 20 Mar 2018 08:53:11 +0000 (08:53 +0000)]
[ms] Parse #pragma optimize and ignore it behind its own flag
This allows users to turn off warnings about this pragma specifically,
while still receiving warnings about other ignored pragmas.
Differential Revision: https://reviews.llvm.org/D44630
llvm-svn: 327959
Jonas Paulsson [Tue, 20 Mar 2018 08:29:19 +0000 (08:29 +0000)]
[SystemZ] Add "REQUIRES: asserts" to test case to fix build bots.
llvm-svn: 327958
Martin Storsjo [Tue, 20 Mar 2018 06:19:38 +0000 (06:19 +0000)]
[X86] Properly implement the calling convention for f80 for mingw/x86_64
In these cases, both parameters and return values are passed
as a pointer to a stack allocation.
MSVC doesn't use the f80 data type at all, while it is used
for long doubles on mingw.
Normally, this part of the calling convention is handled
within clang, but for intrinsics that are lowered to libcalls,
it may need to be handled within llvm as well.
Differential Revision: https://reviews.llvm.org/D44592
llvm-svn: 327957
Lang Hames [Tue, 20 Mar 2018 05:56:58 +0000 (05:56 +0000)]
[ORC] Don't fully qualify explicit destructor call -- it confuses some compilers.
This should fix the builder failure at
http://lab.llvm.org:8011/builders/lld-x86_64-darwin13/builds/19224
llvm-svn: 327955
Craig Topper [Tue, 20 Mar 2018 05:00:20 +0000 (05:00 +0000)]
[X86] Rename MOVSX32_NOREXrr8 to MOVSX32rr8_NOREX so that the scheduler model regular expressions will pick it up with the regular version.
Do the same for MOVSX32_NOREXrm8, MOVZX32_NOREXrr8, and MOVZX32_NOREXrm8
llvm-svn: 327948
Craig Topper [Tue, 20 Mar 2018 03:55:17 +0000 (03:55 +0000)]
[X86] Fix the SchedRW for memory forms of CMP and TEST.
They were incorrectly marked as RMW operations. Some of the CMP instrucions worked, but the ones that use a similar encoding as RMW form of ADD ended up marked as RMW.
TEST used the same tablegen class as some of the CMPs.
llvm-svn: 327947
Lang Hames [Tue, 20 Mar 2018 03:49:29 +0000 (03:49 +0000)]
[ORC] Rename SymbolSource to MaterializationUnit, and make the materialization
operation all-or-nothing, rather than allowing materialization on a per-symbol
basis.
This addresses a shortcoming of per-symbol materialization: If a
MaterializationUnit (/SymbolSource) wants to materialize more symbols than
requested (which is likely: most materializers will want to materialize whole
modules) then it needs a way to notify the symbol table about the extra symbols
being materialized. This process (checking what has been requested against what
is being provided and notifying the symbol table about the difference) has to
be repeated at every level of the JIT stack. Making materialization
all-or-nothing eliminates this issue, simplifying both materializer
implementations and the symbol table (VSO class) API. The cost is that
per-symbol materialization (e.g. for individual symbols in a module) now
requires multiple MaterializationUnits.
llvm-svn: 327946
George Burgess IV [Tue, 20 Mar 2018 03:27:44 +0000 (03:27 +0000)]
Properly construct `inline` members without initializers
Digging through commit logs, it appears the checks in this block predate
`inline` class variables. With them, we fail to emit dynamic
initializers for members that don't have an explicit initializer, and we
won't go out of our way to instantiate the class denoted by
`Var->getType()`.
Fixes PR35599.
llvm-svn: 327945
Craig Topper [Tue, 20 Mar 2018 03:02:03 +0000 (03:02 +0000)]
[X86] Add TEST16mi/TEST32mi/TEST64mi32 to the Sandybridge/Haswell/Broadwell/Skylake scheduler models.
Move it from a load+store group on SNB to a load only group, the same group as CMP.
llvm-svn: 327944
Craig Topper [Tue, 20 Mar 2018 03:01:59 +0000 (03:01 +0000)]
[X86] Add ROR/ROL/SHR/SAR by 1 instructions to the Sandy Bridge scheduler model.
I assume these match the generic immediate version like they do in the other models.
llvm-svn: 327943
Quentin Colombet [Tue, 20 Mar 2018 02:44:40 +0000 (02:44 +0000)]
[ShrinkWrap] Take into account landing pad
When scanning the function for CSRs uses and defs, also check if
the basic block are landing pads.
Consider that landing pads needs the CSRs to be properly set.
That way we force the prologue/epilogue to always be pushed out
of the problematic "throw" region. The "throw" region is
problematic because the jumps are not properly modeled.
Fixes PR36513
llvm-svn: 327942
Jim Ingham [Tue, 20 Mar 2018 02:15:23 +0000 (02:15 +0000)]
Add a suggestion to convert dotest tests to use run_to_source_breakpoint.
llvm-svn: 327941
David Blaikie [Tue, 20 Mar 2018 02:14:50 +0000 (02:14 +0000)]
Fix layering between llvm-c and Support by factoring out some typedefs into Support
llvm-c depends on Support, but Support (TargetRegistry) uses some of
llvm-c's typedefs. Move those into a Support header to be used from both
llvm-c and Support.
llvm-svn: 327940
Akira Hatanaka [Tue, 20 Mar 2018 01:47:58 +0000 (01:47 +0000)]
[CodeGen] Ignore OpaqueValueExprs that are unique references to their
source expressions when iterating over a PseudoObjectExpr's semantic
subexpression list.
Previously the loop in emitPseudoObjectExpr would emit the IR for each
OpaqueValueExpr that was in a PseudoObjectExpr's semantic-form
expression list and use the result when the OpaqueValueExpr later
appeared in other expressions. This caused an assertion failure when
AggExprEmitter tried to copy the result of an OpaqueValueExpr and the
copied type didn't have trivial copy/move constructors or assignment
operators.
This patch adds flag IsUnique to OpaqueValueExpr which indicates it is a
unique reference to its source expression (it is not used in multiple
places). The loop in emitPseudoObjectExpr ignores OpaqueValueExprs that
are unique and CodeGen visitors simply traverse the source expressions
of such OpaqueValueExprs.
rdar://problem/
34363596
Differential Revision: https://reviews.llvm.org/D39562
llvm-svn: 327939
Shiva Chen [Tue, 20 Mar 2018 01:39:17 +0000 (01:39 +0000)]
[RISCV] Preserve stack space for outgoing arguments when the function contain variable size objects
E.g.
bar (int x)
{
char p[x];
push outgoing variables for foo.
call foo
}
We need to generate stack adjustment instructions for outgoing arguments by
eliminateCallFramePseudoInstr when the function contains variable size
objects to avoid outgoing variables corrupt the variable size object.
Default hasReservedCallFrame will return !hasFP().
We don't want to generate extra sp adjustment instructions when hasFP()
return true, So We override hasReservedCallFrame as !hasVarSizedObjects().
Differential Revision: https://reviews.llvm.org/D43752
llvm-svn: 327938
Craig Topper [Tue, 20 Mar 2018 01:30:21 +0000 (01:30 +0000)]
[X86] Fix the SNB scheduler for BLENDVB.
PBLENDVBrr0 was with the memory version of VBLENDVB and PBLENDVBrm0 was missing.
llvm-svn: 327937
Mandeep Singh Grang [Tue, 20 Mar 2018 01:17:18 +0000 (01:17 +0000)]
Revert "[compiler-rt] Change std::sort to llvm::sort in response to r327219"
This reverts commit
2ee210e1963e03aacc0f71c50e4994bb5c66586e.
llvm-svn: 327936
George Karpenkov [Tue, 20 Mar 2018 01:16:46 +0000 (01:16 +0000)]
[analyzer] Improve performance of NoStoreFuncVisitor
Compute modifying frames lazily on demand.
Differential Revision: https://reviews.llvm.org/D44503
llvm-svn: 327935
Rafael Espindola [Tue, 20 Mar 2018 01:06:29 +0000 (01:06 +0000)]
Run dos2unix on a test. NFC.
llvm-svn: 327934
Aaron Smith [Tue, 20 Mar 2018 01:04:21 +0000 (01:04 +0000)]
[DebugInfoPDB] Add backward support flags to PDB_NameSearchFlags
For NS_CaseInFileNameExt support.
llvm-svn: 327933
Petr Hosek [Tue, 20 Mar 2018 00:56:08 +0000 (00:56 +0000)]
Allow builtins for Cross-DSO CFI on Android
This is needed to avoid the test failure in case when compiler-rt
is set as the default runtime library for Clang.
Differential Revision: https://reviews.llvm.org/D44655
llvm-svn: 327932
Fangrui Song [Tue, 20 Mar 2018 00:55:39 +0000 (00:55 +0000)]
[ELF] Make SectionHeaderOffset less sensitive.
llvm-svn: 327931