Fangrui Song [Sat, 20 Apr 2019 02:10:48 +0000 (02:10 +0000)]
[llvm-objdump] Simplify --{start,stop}-address
llvm-svn: 358803
Julian Lettner [Sat, 20 Apr 2019 00:18:44 +0000 (00:18 +0000)]
[TSan] Support fiber API on macOS
Committing on behalf of Yuri Per (yuri).
Reviewers: dvyukov, kubamracek, yln
Reviewed By: kubamracek
Authored By: yuri
Differential Revision: https://reviews.llvm.org/D58110
llvm-svn: 358802
Sam Clegg [Sat, 20 Apr 2019 00:11:46 +0000 (00:11 +0000)]
[WebAssembly] Object: Improve error messages on invalid section
Also add a test.
Differential Revision: https://reviews.llvm.org/D60836
llvm-svn: 358801
Amara Emerson [Fri, 19 Apr 2019 23:54:44 +0000 (23:54 +0000)]
Revert "Revert "[GlobalISel] Add legalization support for non-power-2 loads and stores""
We were shifting the wrong component of a split load when trying to combine them
back into a single value.
llvm-svn: 358800
Jessica Paquette [Fri, 19 Apr 2019 23:41:52 +0000 (23:41 +0000)]
[GlobalISel][AArch64] Legalize + select G_FRINT
Exactly the same as G_FCEIL, G_FABS, etc.
Add tests for the fp16/nofp16 behaviour, update arm64-vfloatintrinsics, etc.
Differential Revision: https://reviews.llvm.org/D60895
llvm-svn: 358799
Thomas Lively [Fri, 19 Apr 2019 23:40:36 +0000 (23:40 +0000)]
[WebAssembly] Emit the DataCount section when bulk memory is enabled
Summary:
The DataCount section is necessary for the bulk memory operations
memory.init and data.drop to validate, but it is not recognized by
engines that do not support bulk memory, so emit the section only if
bulk-memory is enabled.
Reviewers: aheejin, dschuff, sbc100
Subscribers: jgravelle-google, sunfish, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D60637
llvm-svn: 358798
Kristof Umann [Fri, 19 Apr 2019 23:33:50 +0000 (23:33 +0000)]
[analyzer] Move UninitializedObjectChecker out of alpha
Moved UninitializedObjectChecker from the 'alpha.cplusplus' to the
'optin.cplusplus' package.
Differential Revision: https://reviews.llvm.org/D58573
llvm-svn: 358797
David Blaikie [Fri, 19 Apr 2019 23:04:05 +0000 (23:04 +0000)]
Modules: Adopt template parameters for variable templates to set their decl context correctly
Exposed by a related bug about visibility of default arguments of nested
templates - without the correct decl context, default template
parameters of variable templates nested in classes would have incorrect
visibility computed.
llvm-svn: 358796
David Blaikie [Fri, 19 Apr 2019 23:02:30 +0000 (23:02 +0000)]
Modules: Search for a visible definition of the decl context when computing visibility of a default template parameter
The code is/was already correct for the case where a parameter is a
parameter of its enclosing lexical DeclContext (functions and classes).
But for other templates (alias and variable templates) they don't create
their own scope to be members of - in those cases, they parameter should
be considered visible if any definition of the lexical decl context is
visible.
[this should cleanup the failure on the libstdc++ modules buildbot]
[this doesn't actually fix the variable template case for a
secondary/compounding reason (its lexical decl context is incorrectly
considered to be the translation unit)]
Test covers all 4 kinds of templates with default args, including a
regression test for the still broken variable template case.
Reviewers: rsmith
Differential Revision: https://reviews.llvm.org/D60892
llvm-svn: 358795
Reid Kleckner [Fri, 19 Apr 2019 22:51:49 +0000 (22:51 +0000)]
[COFF] Pack Name in Symbol as is done in ELF
Summary:
This assumes all symbols are <4GB long, so we can store them as a 32-bit
integer. This reorders the fields so the length appears first, packing
with the other bitfield data in the base Symbol object.
This saved 70MB / 3.60% of heap allocations when linking
browser_tests.exe with no PDB. It's not much as a percentage, but worth
doing. I didn't do performance measurements, I don't think it will be
measurable in time.
Reviewers: ruiu, inglorion, amccarth, aganea
Subscribers: llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D60297
llvm-svn: 358794
Sam Clegg [Fri, 19 Apr 2019 22:43:32 +0000 (22:43 +0000)]
[WebAssembly] FastISel: Don't fallback to SelectionDAG after BuildMI in selectCall
My understanding is that once BuildMI has been called we can't fallback
to SelectionDAG.
This change moves the fallback for when getRegForValue() fails for
that target of an indirect call. This was failing in -fPIC mode when
the callee is GlobalValue.
Add a test case that tickles this.
Differential Revision: https://reviews.llvm.org/D60908
llvm-svn: 358793
Jonas Devlieghere [Fri, 19 Apr 2019 22:37:55 +0000 (22:37 +0000)]
[Tests] Split float test into float and doubles
As I was waiting for the test suite to complete at 99% I noticed this
test taking quite a bit of time. Since it's easy to split I just went
ahead and did so.
llvm-svn: 358792
Vedant Kumar [Fri, 19 Apr 2019 22:36:40 +0000 (22:36 +0000)]
[GVN+LICM] Use line 0 locations for better crash attribution
This is a follow-up to r291037+r291258, which used null debug locations
to prevent jumpy line tables.
Using line 0 locations achieves the same effect, but works better for
crash attribution because it preserves the right inline scope.
Differential Revision: https://reviews.llvm.org/D60913
llvm-svn: 358791
Vitaly Buka [Fri, 19 Apr 2019 22:27:50 +0000 (22:27 +0000)]
Update GN files to build with r358103
llvm-svn: 358790
Eric Christopher [Fri, 19 Apr 2019 22:18:53 +0000 (22:18 +0000)]
Remove the EnableEarlyCSEMemSSA set of options from the legacy
and new pass managers. They were default to true and not being
used.
Differential Revision: https://reviews.llvm.org/D60747
llvm-svn: 358789
Eli Friedman [Fri, 19 Apr 2019 21:58:10 +0000 (21:58 +0000)]
[AArch64] Fix checks for AArch64MCExpr::VK_SABS flag.
VK_SABS is part of the SymLoc bitfield in the variant kind which should
be compared for equality, not by checking the VK_SABS bit.
As far as I know, the existing code happened to produce the correct
results in all cases, so this is just a cleanup.
Patch by Stephen Crane.
Differential Revision: https://reviews.llvm.org/D60596
llvm-svn: 358788
Jessica Paquette [Fri, 19 Apr 2019 21:46:12 +0000 (21:46 +0000)]
[GlobalISel] Add IRTranslator support for G_FRINT
Add it as a simple intrinsic, update arm64-irtranslator.ll.
Differential Revision: https://reviews.llvm.org/D60893
llvm-svn: 358787
Amy Huang [Fri, 19 Apr 2019 21:44:30 +0000 (21:44 +0000)]
Attempt to fix buildbot failure in commit
1bb57bac959ac163fd7d8a76d734ca3e0ecee6ab.
llvm-svn: 358786
Jessica Paquette [Fri, 19 Apr 2019 21:44:16 +0000 (21:44 +0000)]
[GlobalISel] Add a G_FRINT opcode
Equivalent to SelectionDAG's frint node.
Differential Revision: https://reviews.llvm.org/D60891
llvm-svn: 358785
Craig Topper [Fri, 19 Apr 2019 21:39:16 +0000 (21:39 +0000)]
[X86] Add test case for D60801. NFC
llvm-svn: 358784
Amy Huang [Fri, 19 Apr 2019 21:09:11 +0000 (21:09 +0000)]
[MS] Emit S_HEAPALLOCSITE debug info
Summary:
This emits labels around heapallocsite calls and S_HEAPALLOCSITE debug
info in codeview. Currently only changes FastISel, so emitting labels still
needs to be implemented in SelectionDAG.
Reviewers: hans, rnk
Subscribers: aprantl, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D60800
llvm-svn: 358783
Casey Carter [Fri, 19 Apr 2019 20:39:41 +0000 (20:39 +0000)]
[libc++] [test] Silence C++20 deprecation warnings in the MSVC STL
... when including msvc_stdlib_force_include.hpp.
llvm-svn: 358782
Artem Dergachev [Fri, 19 Apr 2019 20:23:29 +0000 (20:23 +0000)]
Reapply "[analyzer] Introduce a simplified API for adding custom path notes."
This reapplies commit r357323, fixing memory leak found by LSan.
Differential Revision: https://reviews.llvm.org/D58367
llvm-svn: 358781
Chris Bieneman [Fri, 19 Apr 2019 20:13:25 +0000 (20:13 +0000)]
[CMake] Add fuzzer as a component for runtime builds
Calling `add_compiler_rt_component` sets up the component connection between runtime builds and the parent CMake configuration. Adding this call allows specifying `fuzzer` as a `LLVM_RUNTIME_DISTRIBUTION_COMPONENT`.
llvm-svn: 358780
Chris Bieneman [Fri, 19 Apr 2019 20:08:55 +0000 (20:08 +0000)]
[CMake] Pass monorepo build settings in cross compile
This allows the cross compiled build targets to configure the LLVM tools and sub-projects that are part of the main build.
This is needed for generating native non llvm *-tablegen tools when cross compiling clang in the monorepo build environment.
llvm-svn: 358779
Martin Storsjo [Fri, 19 Apr 2019 19:04:22 +0000 (19:04 +0000)]
[MSVC] If unable to find link.exe from a MSVC installation, look for link.exe next to cl.exe
Previously, if the MSVC installation isn't detected properly, clang
will later just fail to execute link.exe.
This improves using clang in msvc mode on linux, where one intentionally
might not want to point clang to the MSVC installation itself (which
isn't executable as such), but where a link.exe named wine wrapper is
available in the path next to a cl.exe named wine wrapper.
Differential Revision: https://reviews.llvm.org/D60094
llvm-svn: 358778
Jonas Devlieghere [Fri, 19 Apr 2019 19:03:58 +0000 (19:03 +0000)]
[Docs] Add more info about building the docs
Including the C++ and Python reference.
llvm-svn: 358777
Jim Ingham [Fri, 19 Apr 2019 18:46:56 +0000 (18:46 +0000)]
This test doesn't need to be run for all debug formats.
llvm-svn: 358776
Brad Smith [Fri, 19 Apr 2019 18:41:40 +0000 (18:41 +0000)]
Enable frame pointer elimination for OpenBSD on powerpc.
llvm-svn: 358775
Petr Hosek [Fri, 19 Apr 2019 18:29:17 +0000 (18:29 +0000)]
[gn] Support dots in CMake paths in the sync script
Some file paths use dots to pick up sources from parent directories.
Differential Revision: https://reviews.llvm.org/D60734
llvm-svn: 358774
Jonas Devlieghere [Fri, 19 Apr 2019 18:02:35 +0000 (18:02 +0000)]
[Docs] Make Doxygen functional
This fixes the doxygen configuration to be functional again. I removed
the customer header and footer, as well as the no-longer-existent style
sheet. I also widened the scope of the documentation, from just the
public API to include the private interfaces as well.
llvm-svn: 358773
Alina Sbirlea [Fri, 19 Apr 2019 17:46:50 +0000 (17:46 +0000)]
[LICM & MemorySSA] Make limit flags pass tuning options.
Summary:
Make the flags in LICM + MemorySSA tuning options in the old and new
pass managers.
Subscribers: mehdi_amini, jlebar, Prazek, george.burgess.iv, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D60490
llvm-svn: 358772
Amara Emerson [Fri, 19 Apr 2019 17:42:13 +0000 (17:42 +0000)]
Revert "[GlobalISel] Add legalization support for non-power-2 loads and stores"
This introduces some runtime failures which I'll need to investigate further.
llvm-svn: 358771
Mitch Phillips [Fri, 19 Apr 2019 17:36:58 +0000 (17:36 +0000)]
Removed regiser dump checks from FP unit test.
llvm-svn: 358770
Mitch Phillips [Fri, 19 Apr 2019 17:36:56 +0000 (17:36 +0000)]
[HWASan] Added no-FP unit test for register dump.
Summary: Unit test for D60798.
Reviewers: eugenis
Subscribers: kubamracek, #sanitizers, llvm-commits, pcc
Tags: #sanitizers, #llvm
Differential Revision: https://reviews.llvm.org/D60806
llvm-svn: 358769
Mitch Phillips [Fri, 19 Apr 2019 17:36:55 +0000 (17:36 +0000)]
Updated test case.
llvm-svn: 358768
Mitch Phillips [Fri, 19 Apr 2019 17:36:53 +0000 (17:36 +0000)]
Added no-FP unit test for register dump.
llvm-svn: 358767
Alexey Bataev [Fri, 19 Apr 2019 16:48:38 +0000 (16:48 +0000)]
[OPENMP][NVPTX] target [teams distribute] simd maybe run without
runtime.
target [teams distribute] simd costructs do not require full runtime for
the correct execution, we can run them without full runtime.
llvm-svn: 358766
Alina Sbirlea [Fri, 19 Apr 2019 16:32:08 +0000 (16:32 +0000)]
Update to use PipelineTuningOptions. Corresponds to llvm change: D59723.
llvm-svn: 358765
Jessica Paquette [Fri, 19 Apr 2019 16:28:08 +0000 (16:28 +0000)]
[GlobalISel][AArch64] Legalize vector G_FPOW
This instruction is legalized in the same way as G_FSIN, G_FCOS, G_FLOG10, etc.
Update legalize-pow.mir and arm64-vfloatintrinsics.ll to reflect the change.
Differential Revision: https://reviews.llvm.org/D60218
llvm-svn: 358764
Alina Sbirlea [Fri, 19 Apr 2019 16:11:59 +0000 (16:11 +0000)]
[NewPassManager] Adding pass tuning options: loop vectorize.
Summary:
Trying to add the plumbing necessary to add tuning options to the new pass manager.
Testing with the flags for loop vectorize.
Reviewers: chandlerc
Subscribers: sanjoy, mehdi_amini, jlebar, steven_wu, dexonsmith, dang, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D59723
llvm-svn: 358763
Fangrui Song [Fri, 19 Apr 2019 15:45:25 +0000 (15:45 +0000)]
[dsymutil] DwarfLinker: delete unused parameter
llvm-svn: 358762
Sanjay Patel [Fri, 19 Apr 2019 15:31:11 +0000 (15:31 +0000)]
[SelectionDAG] soften splat mask assert/unreachable (PR41535)
These are general queries, so they should not die when given
a degenerate input like an all undef mask. Callers should be
able to deal with an op that will eventually be simplified away.
llvm-svn: 358761
Nico Weber [Fri, 19 Apr 2019 14:13:11 +0000 (14:13 +0000)]
llvm-undname: Attempt to fix leak-on-invalid found by oss-fuzz
llvm-svn: 358760
Martin Storsjo [Fri, 19 Apr 2019 13:50:43 +0000 (13:50 +0000)]
[MinGW] Add an --appcontainer flag, passed through to lld-link
GNU ld doesn't have such a flag though, so this is a lld specific
option.
Differential Revision: https://reviews.llvm.org/D60860
llvm-svn: 358759
Sylvestre Ledru [Fri, 19 Apr 2019 13:48:52 +0000 (13:48 +0000)]
Debian: Add two missing version code in sid
llvm-svn: 358758
Sylvestre Ledru [Fri, 19 Apr 2019 13:46:58 +0000 (13:46 +0000)]
Add support of the future Debian (Debian 11 - Bullseye)
https://wiki.debian.org/DebianBullseye
llvm-svn: 358757
Sylvestre Ledru [Fri, 19 Apr 2019 13:43:28 +0000 (13:43 +0000)]
Add support of the next Ubuntu (Ubuntu 19.10 - Eoan EANIMAL)
llvm-svn: 358756
Nico Weber [Fri, 19 Apr 2019 13:18:41 +0000 (13:18 +0000)]
gn build: Merge r358722
llvm-svn: 358755
Nico Weber [Fri, 19 Apr 2019 13:16:26 +0000 (13:16 +0000)]
gn build: Merge r358691
llvm-svn: 358754
Florian Hahn [Fri, 19 Apr 2019 12:36:41 +0000 (12:36 +0000)]
[LTO] Add plumbing to save stats during LTO on Darwin.
Gold and ld on Linux already support saving stats, but the
infrastructure is missing on Darwin. Unfortunately it seems like the
configuration from lib/LTO/LTO.cpp is not used.
This patch adds a new LTOStatsFile option and adds plumbing in Clang to
use it on Darwin, similar to the way remarks are handled.
Currnetly the handling of LTO flags seems quite spread out, with a bunch
of duplication. But I am not sure if there is an easy way to improve
that?
Reviewers: anemet, tejohnson, thegameg, steven_wu
Reviewed By: steven_wu
Differential Revision: https://reviews.llvm.org/D60516
llvm-svn: 358753
Kristof Umann [Fri, 19 Apr 2019 12:32:10 +0000 (12:32 +0000)]
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
Fangrui Song [Fri, 19 Apr 2019 12:28:36 +0000 (12:28 +0000)]
Change \r\n -> \n for llvm-symbolizer/help.test after rL358749
llvm-svn: 358751
Kristof Umann [Fri, 19 Apr 2019 11:01:35 +0000 (11:01 +0000)]
[analyzer] Fix an assertion failure if plugins added dependencies
Ideally, there is no reason behind not being able to depend on checkers that
come from a different plugin (or on builtin checkers) -- however, this is only
possible if all checkers are added to the registry before resolving checker
dependencies. Since I used a binary search in my addDependency method, this also
resulted in an assertion failure (due to CheckerRegistry::Checkers not being
sorted), since the function used by plugins to register their checkers
(clang_registerCheckers) calls addDependency.
This patch resolves this issue by only noting which dependencies have to
established when addDependency is called, and resolves them at a later stage
when no more checkers are added to the registry, by which point
CheckerRegistry::Checkers is already sorted.
Differential Revision: https://reviews.llvm.org/D59461
llvm-svn: 358750
Igor Kudrin [Fri, 19 Apr 2019 10:17:52 +0000 (10:17 +0000)]
[llvm-symbolizer] Add llvm-addr2line
This adds an alias for llvm-symbolizer with different defaults so that
it can be used as a drop-in replacement for GNU's addr2line.
If a substring "addr2line" is found in the tool's name:
* it defaults "-i", "-f" and "-C" to OFF;
* it uses "--output-style=GNU" by default.
Differential Revision: https://reviews.llvm.org/D60067
llvm-svn: 358749
Igor Kudrin [Fri, 19 Apr 2019 10:14:18 +0000 (10:14 +0000)]
[llvm-symbolizer] Unhide and document the "-output-style" option
With the latest changes, the option gets useful for users of
llvm-symbolizer, not only for the upcoming llvm-addr2line.
Differential Revision: https://reviews.llvm.org/D60816
llvm-svn: 358748
Igor Kudrin [Fri, 19 Apr 2019 10:12:56 +0000 (10:12 +0000)]
[llvm-symbolizer] Make the output with -output-style=GNU closer to addr2line's
This patch addresses two differences in the output of llvm-symbolizer
and GNU's addr2line:
* llvm-symbolizer prints an empty line after the report for an address.
* With "-f -i=0", llvm-symbolizer replaces the name of an inlined
function with the name from the symbol table, i. e., the top caller
function in the inlining chain. addr2line preserves the name of the
inlined function.
Differential Revision: https://reviews.llvm.org/D60770
llvm-svn: 358747
Simon Pilgrim [Fri, 19 Apr 2019 09:49:04 +0000 (09:49 +0000)]
[AMDGPU] Regenerate extractelt->truncate test.
Prep work for D60462
llvm-svn: 358746
Bjorn Pettersson [Fri, 19 Apr 2019 09:10:42 +0000 (09:10 +0000)]
[LibTooling] Fix -Wsign-compare after r358697
llvm-svn: 358745
Bjorn Pettersson [Fri, 19 Apr 2019 09:08:38 +0000 (09:08 +0000)]
[CodeGen] Add "const" to MachineInstr::mayAlias
Summary:
The basic idea here is to make it possible to use
MachineInstr::mayAlias also when the MachineInstr
is const (or the "Other" MachineInstr is const).
The addition of const in MachineInstr::mayAlias
then rippled down to the need for adding const
in several other places, such as
TargetTransformInfo::getMemOperandWithOffset.
Reviewers: hfinkel
Reviewed By: hfinkel
Subscribers: hfinkel, MatzeB, arsenm, jvesely, nhaehnle, hiraditya, javed.absar, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D60856
llvm-svn: 358744
James Molloy [Fri, 19 Apr 2019 09:00:55 +0000 (09:00 +0000)]
[PATCH] [MachineScheduler] Check pending instructions when an instruction is scheduled
Pending instructions that may have been blocked from being available by the HazardRecognizer may no longer may not be blocked any more when an instruction is scheduled; pending instructions should be re-checked in this case.
This is primarily aimed at VLIW targets with large parallelism and esoteric constraints.
No testcase as no in-tree targets have this behavior.
Differential revision: https://reviews.llvm.org/D60861
llvm-svn: 358743
Fangrui Song [Fri, 19 Apr 2019 08:03:20 +0000 (08:03 +0000)]
[MergeFunc] Delete unused FunctionNode::release()
llvm-svn: 358742
Fangrui Song [Fri, 19 Apr 2019 07:57:51 +0000 (07:57 +0000)]
[MergeFunc] removeUsers: call remove() only on direct users
removeUsers uses a work list to collect indirect users and call remove()
on those functions. However it has a bug (`if (!Visited.insert(UU).second)`).
Actually, we don't have to collect indirect users.
After the merge of F and G, G's callers will be considered (added to
Deferred). If G's callers can be merged, G's callers' callers will be
considered.
Update the test unnamed-addr-reprocessing.ll to make it clear we can
still merge indirect callers.
llvm-svn: 358741
Piotr Sobczak [Fri, 19 Apr 2019 06:19:14 +0000 (06:19 +0000)]
[AMDGPU] Ignore non-SUnits edges
Summary:
Ignore edges to non-SUnits (e.g. ExitSU) when checking
for low latency instructions.
When calling the function isLowLatencyInstruction(),
an ExitSU could be on the list of successors, not necessarily
a regular SU. In other places in the code there is a check
"Succ->NodeNum >= DAGSize" to prevent further processing of
ExitSU as "Succ->getInstr()" is NULL in such a case.
Also, 8 out of 9 cases of "SUnit *Succ = SuccDep.getSUnit())"
has the guard, so it is clearly an omission here.
Change-Id: Ica86f0327c7b2e6bcb56958e804ea6c71084663b
Reviewers: nhaehnle
Reviewed By: nhaehnle
Subscribers: MatzeB, arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, javed.absar, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D60864
llvm-svn: 358740
Chandler Carruth [Fri, 19 Apr 2019 05:59:42 +0000 (05:59 +0000)]
[CallSite removal] Move the legacy PM, call graph, and some inliner
code to `CallBase`.
This patch focuses on the legacy PM, call graph, and some of inliner and legacy
passes interacting with those APIs from `CallSite` to the new `CallBase` class.
No interesting changes.
Differential Revision: https://reviews.llvm.org/D60412
llvm-svn: 358739
Fangrui Song [Fri, 19 Apr 2019 05:49:29 +0000 (05:49 +0000)]
[MergeFunc] Use less_first() as the comparator of Schwartzian transform
llvm-svn: 358738
Craig Topper [Fri, 19 Apr 2019 05:48:13 +0000 (05:48 +0000)]
[X86] Turn (and (shl X, C1), C2) into (shl (and X, (C1 >> C2), C2) if the AND could match a movzx.
Could get further improvements by recognizing (i64 and (anyext (i32 shl))).
llvm-svn: 358737
Craig Topper [Fri, 19 Apr 2019 05:48:09 +0000 (05:48 +0000)]
[X86] Add test cases for turning (and (shl X, C1), C2) into (shl (and X, (C1 >> C2), C2) when the AND could match to a movzx.
We already reorder when C1 >> C2 would allow a smaller immediate encoding.
llvm-svn: 358736
Craig Topper [Fri, 19 Apr 2019 04:52:21 +0000 (04:52 +0000)]
[X86] Make sure we copy the HandleSDNode back to N before executing the default code after the switch in matchAddressRecursively
Summary:
There are two places where we create a HandleSDNode in address matching in order to handle the case where N is changed by CSE. But if we end up not matching, we fall back to code at the bottom of the switch that really would like N to point to something that wasn't CSEd away. So we should make sure we copy the handle back to N on any paths that can reach that code.
This appears to be the true reason we needed to check DELETED_NODE in the negation matching. In pr32329.ll we had two subtracts back to back. We recursed through the first subtract, and onto the second subtract. The second subtract called matchAddressRecursively on its LHS which caused that subtract to CSE. We ultimately failed the match and ended up in the default code. But N was pointing at the old node that had been deleted, but the default code didn't know that and took it as the base register. Then we unwound back to the first subtract and tried to access this bogus base reg requiring the check for deleted node. With this patch we now use the CSE result as the base reg instead.
matchAdd has been broken since sometime in 2015 when it was pulled out of the switch into a helper function. The assignment to N at the end was still there, but N was passed by value and not by reference so the update didn't go anywhere.
Reviewers: niravd, spatel, RKSimon, bkramer
Reviewed By: niravd
Subscribers: llvm-commits, hiraditya
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D60843
llvm-svn: 358735
Fangrui Song [Fri, 19 Apr 2019 03:34:28 +0000 (03:34 +0000)]
[DWARF] Use hasFileAtIndex to properly verify DWARF 5 after rL358732
llvm-svn: 358734
Richard Smith [Fri, 19 Apr 2019 02:46:50 +0000 (02:46 +0000)]
Fix overly-long line after r358731.
llvm-svn: 358733
Ali Tamur [Fri, 19 Apr 2019 02:26:56 +0000 (02:26 +0000)]
[llvm] Prevent duplicate files in debug line header in dwarf 5: another attempt
Another attempt to land the changes in debug line header to prevent duplicate
files in Dwarf 5. I rolled back my previous commit because of a mistake in
generating the object file in a test. Meanwhile, I addressed some offline
comments and changed the implementation; the largest difference is that
MCDwarfLineTableHeader does not keep DwarfVersion but gets it as a parameter. I
also merged the patch to fix two lld tests that will strt to fail into this
patch.
Original Commit:
https://reviews.llvm.org/D59515
Original Message:
Motivation: In previous dwarf versions, file name indexes started from 1, and
the primary source file was not explicit. Dwarf 5 standard (6.2.4) prescribes
the primary source file to be explicitly given an entry with an index number 0.
The current implementation honors the specification by just duplicating the
main source file, once with index number 0, and later maybe with another
index number. While this is compliant with the letter of the standard, the
duplication causes problems for consumers of this information such as lldb.
(Some files are duplicated, where only some of them have a line table although
all refer to the same file)
With this change, dwarf 5 debug line section files always start from 0, and
the zeroth entry is not duplicated whenever possible. This requires different
handling of dwarf 4 and dwarf 5 during generation (e.g. when a function returns
an index zero for a file name, it signals an error in dwarf 4, but not in dwarf
5) However, I think the minor complication is worth it, because it enables all
consumers (lldb, gdb, dwarfdump, objdump, and so on) to treat all files in the
file name list homogenously.
llvm-svn: 358732
Ali Tamur [Fri, 19 Apr 2019 02:15:57 +0000 (02:15 +0000)]
Fix typo in function name [NFC]
llvm-svn: 358731
Fangrui Song [Fri, 19 Apr 2019 02:06:06 +0000 (02:06 +0000)]
[APInt] Optimize umul_ov
Change two costly udiv() calls to lshr(1)*RHS + left-shift + plus
On one 64-bit umul_ov benchmark, I measured an obvious improvement: 12.8129s -> 3.6257s
Note, there may be some value to special case 64-bit (the most common
case) with __builtin_umulll_overflow().
Differential Revision: https://reviews.llvm.org/D60669
llvm-svn: 358730
Fangrui Song [Fri, 19 Apr 2019 01:54:36 +0000 (01:54 +0000)]
[analyzer] Fix -Wunused-local-typedef after rC358695
llvm-svn: 358729
Saleem Abdulrasool [Fri, 19 Apr 2019 01:48:36 +0000 (01:48 +0000)]
MergeFunc: preserve COMDAT information when creating a thunk
We would previously drop the COMDAT on the thunk we generated when replacing a
function body with the forwarding thunk. This would result in a function that
may have been multiply emitted and multiply merged to be emitted with the same
name without the COMDAT. This is a hard error with PE/COFF where the COMDAT is
used for the deduplication of Value Witness functions for Swift.
llvm-svn: 358728
Kostya Serebryany [Fri, 19 Apr 2019 01:40:16 +0000 (01:40 +0000)]
[libFuzzer] revert part of r358726 committed by mistake (oops)
llvm-svn: 358727
Kostya Serebryany [Fri, 19 Apr 2019 01:39:14 +0000 (01:39 +0000)]
[libFuzzer] in fork mode, make sure the child stats are always collected. This should fix the flakiness in fork-ubsan.test
llvm-svn: 358726
Thomas Anderson [Fri, 19 Apr 2019 00:52:54 +0000 (00:52 +0000)]
[libc++] Make __debug_less::operator() constexpr
This is a followup to [1] which added a new `__debug_less::operator()` overload.
[2] added `_LIBCPP_CONSTEXPR_AFTER_CXX17` to the original
`__debug_less::operator()` between the time of writing [1] and landing it. This
change adds `_LIBCPP_CONSTEXPR_AFTER_CXX17` to the new overload too.
[1] https://reviews.llvm.org/rL358423
[2] https://reviews.llvm.org/rL358252
Differential Revision: https://reviews.llvm.org/D60724
llvm-svn: 358725
Ali Tamur [Fri, 19 Apr 2019 00:42:54 +0000 (00:42 +0000)]
Add header guard to Reusables.h [NFC]
llvm-svn: 358724
Alina Sbirlea [Thu, 18 Apr 2019 23:43:49 +0000 (23:43 +0000)]
[LoopUnroll] Move list of params into a struct [NFCI].
Summary: Cleanup suggested in review of r358304.
Reviewers: sanjoy, efriedma
Subscribers: jlebar, zzheng, dmgreen, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D60638
llvm-svn: 358723
Artem Dergachev [Thu, 18 Apr 2019 23:35:56 +0000 (23:35 +0000)]
[analyzer] Make default bindings to variables actually work.
Default RegionStore bindings represent values that can be obtained by loading
from anywhere within the region, not just the specific offset within the region
that they are said to be bound to. For example, default-binding a character \0
to an int (eg., via memset()) means that the whole int is 0, not just
that its lower byte is 0.
Even though memset and bzero were modeled this way, it didn't work correctly
when applied to simple variables. Eg., in
int x;
memset(x, 0, sizeof(x));
we did produce a default binding, but were unable to read it later, and 'x'
was perceived as an uninitialized variable even after memset.
At the same time, if we replace 'x' with a variable of a structure or array
type, accessing fields or elements of such variable was working correctly,
which was enough for most cases. So this was only a problem for variables of
simple integer/enumeration/floating-point/pointer types.
Fix loading default bindings from RegionStore for regions of simple variables.
Add a unit test to document the API contract as well.
Differential Revision: https://reviews.llvm.org/D60742
llvm-svn: 358722
Davide Italiano [Thu, 18 Apr 2019 23:24:54 +0000 (23:24 +0000)]
[Python] Simplify the code. NFCI.
llvm-svn: 358721
Artem Dergachev [Thu, 18 Apr 2019 23:24:50 +0000 (23:24 +0000)]
[analyzer] NFC: Make reusable unittest mocks reusable.
Put them in a header for other Analyzer unittests to include.
Differential Revision: https://reviews.llvm.org/D60739
llvm-svn: 358720
Artem Dergachev [Thu, 18 Apr 2019 23:17:58 +0000 (23:17 +0000)]
[analyzer] NFC: MoveChecker: Refactor tests to use -verify=prefix.
This -verify=prefix feature is quite underrated.
Differential Revision: https://reviews.llvm.org/D60732
llvm-svn: 358719
Sanjay Patel [Thu, 18 Apr 2019 21:48:46 +0000 (21:48 +0000)]
[AArch64] add tests for mul-by-element; NFC
llvm-svn: 358718
Davide Italiano [Thu, 18 Apr 2019 21:32:36 +0000 (21:32 +0000)]
[crashlog] Strip trailing `\n` from check_output return.
Generally having spurious `\n` doesn't matter, but here the
returning string is a command which is executed, so we want
to strip it. Pointed out by Jason.
llvm-svn: 358717
Adrian Prantl [Thu, 18 Apr 2019 21:22:50 +0000 (21:22 +0000)]
Implement sys::fs::copy_file using the macOS copyfile(3) API
to support APFS clones.
This patch adds a Darwin-specific implementation of
llvm::sys::fs::copy_file() that uses the macOS copyfile(3) API to
support APFS copy-on-write clones, which should be faster and much
more space efficient.
https://developer.apple.com/library/archive/documentation/FileManagement/Conceptual/APFS_Guide/ToolsandAPIs/ToolsandAPIs.html
Differential Revision: https://reviews.llvm.org/D60802
This reapplies 358628 with an additional bugfix handling the case
where the destination file already exists. (Caught by the clang testsuite).
llvm-svn: 358716
Jessica Paquette [Thu, 18 Apr 2019 21:15:48 +0000 (21:15 +0000)]
[GlobalISel][AArch64] Legalize/select G_(S/Z/ANY)_EXT for v8s8s
This adds legalization for G_SEXT, G_ZEXT, and G_ANYEXT for v8s8s.
We were falling back on G_ZEXT in arm64-vabs.ll before, preventing us from
selecting the @llvm.aarch64.neon.sabd.v8i8 intrinsic.
This adds legalizer support for those 3, which gives us selection via the
importer. Update the relevant tests (legalize-ext.mir, select-int-ext.mir) and
add a GISel line to arm64-vabs.ll.
Differential Revision: https://reviews.llvm.org/D60881
llvm-svn: 358715
Jessica Paquette [Thu, 18 Apr 2019 21:13:58 +0000 (21:13 +0000)]
[GlobalISel][AArch64] Legalize v8s8 loads
Add legalizer support for loads of v8s8 and update legalize-load-store.mir.
Differential Revision: https://reviews.llvm.org/D60877
llvm-svn: 358714
Richard Smith [Thu, 18 Apr 2019 21:12:54 +0000 (21:12 +0000)]
[c++2a] Add semantic support for private module fragments.
llvm-svn: 358713
Louis Dionne [Thu, 18 Apr 2019 20:59:49 +0000 (20:59 +0000)]
[libc++] Link against libc++abi in the libc++abi tests
PR27405
llvm-svn: 358712
Gheorghe-Teodor Bercea [Thu, 18 Apr 2019 20:34:43 +0000 (20:34 +0000)]
[OpenMP][NFC] Fix requires target test.
Summary:
Fix requires target test.
Reviewers: ABataev
Subscribers: guansong, jdoerfert, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D60886
llvm-svn: 358711
Owen Pan [Thu, 18 Apr 2019 20:17:08 +0000 (20:17 +0000)]
[clang-format] Fix incorrect formatting of keyword macro definition
See PR39719
Differential Revision: https://reviews.llvm.org/D60853
llvm-svn: 358710
Gheorghe-Teodor Bercea [Thu, 18 Apr 2019 19:53:43 +0000 (19:53 +0000)]
[OpenMP] Add checks for requires and target directives.
Summary: The requires directive containing target related clauses must appear before any target region in the compilation unit.
Reviewers: ABataev, AlexEichenberger, caomhin
Reviewed By: ABataev
Subscribers: guansong, jfb, jdoerfert, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D60875
llvm-svn: 358709
Nico Weber [Thu, 18 Apr 2019 19:52:32 +0000 (19:52 +0000)]
llvm-undname: Fix two more asserts-on-invalid, found by oss-fuzz
llvm-svn: 358708
Nico Weber [Thu, 18 Apr 2019 19:30:21 +0000 (19:30 +0000)]
llvm-undname: Fix two asserts-on-invalid
llvm-svn: 358707
Yi Kong [Thu, 18 Apr 2019 19:29:03 +0000 (19:29 +0000)]
[builtins] Build x86_64 with GENERIC_TF_SOURCES
llvm-svn: 358706
Yitzhak Mandelbaum [Thu, 18 Apr 2019 19:19:01 +0000 (19:19 +0000)]
[LibTooling] Fix build breakage from commit
7b7ce6683ee.
On configurations with -Werror,-Wmissing-field-initializers, the commit does not compile. This commit fixes the offending line.
Original Differential Revision: https://reviews.llvm.org/D60408
llvm-svn: 358705
Philip Reames [Thu, 18 Apr 2019 19:17:14 +0000 (19:17 +0000)]
[GuardWidening] Wire up a NPM version of the LoopGuardWidening pass
llvm-svn: 358704