Craig Topper [Thu, 29 Sep 2016 05:54:43 +0000 (05:54 +0000)]
[AVX-512] Replicate pattern from AVX to select VMOVDDUP for (v2f64 (X86VBroadcast f64:)). Add AVX512VL to command line of existing AVX2 test that hits this condition.
llvm-svn: 282688
Craig Topper [Thu, 29 Sep 2016 05:54:39 +0000 (05:54 +0000)]
[X86] Add EVEX encoded VBROADCASTSS/SD and VPBROADCASTD/Q to execution domain fixing table.
llvm-svn: 282687
Craig Topper [Thu, 29 Sep 2016 05:54:34 +0000 (05:54 +0000)]
[X86] Remove AddedComplexity adjustments that don't seem to be needed.
llvm-svn: 282686
Craig Topper [Thu, 29 Sep 2016 05:54:32 +0000 (05:54 +0000)]
[X86] Add 512-bit VPBROADCASTB and VPBROADCASTW tests.
llvm-svn: 282685
Craig Topper [Thu, 29 Sep 2016 05:54:28 +0000 (05:54 +0000)]
[X86] Add VBROADCASTF128/VBROADCASTI128 to execution domain fixing tables.
llvm-svn: 282684
Jason Molenda [Thu, 29 Sep 2016 04:01:43 +0000 (04:01 +0000)]
Add a unit test for an x86_64 assembly inspection of
a large stack frame with lots of spilled registers.
While writing the i386 version of this test, it looks
like I found a bug in the 32-bit instruction profiler
code. I may ned to fix the assembly inspection engine
before I can finish writing that test, so I'm only
committing the 64-bit one tonight.
<rdar://problem/
28509178>
llvm-svn: 282683
Vitaly Buka [Thu, 29 Sep 2016 03:53:20 +0000 (03:53 +0000)]
Move verbose clang binaries detection
llvm-svn: 282682
Shoaib Meenai [Thu, 29 Sep 2016 03:35:41 +0000 (03:35 +0000)]
[libc++] Add missing locale aliases
Add underscore aliases for strtof_l and strtod_l. _strtold_l exists in
VS 2013 and above, so fix that definition as a drive-by fix.
Differential Revision: https://reviews.llvm.org/D25059
llvm-svn: 282681
Justin Bogner [Thu, 29 Sep 2016 03:35:19 +0000 (03:35 +0000)]
IR: Rename the tablegen'd Attributes file to .gen
All of the other tablegen'd include files are named .gen, so it's best
to be consistent.
llvm-svn: 282680
Aditya Kumar [Thu, 29 Sep 2016 03:32:04 +0000 (03:32 +0000)]
[PR30341] Alias must point to a definition
Inlining the destructor caused the compiler to generate bad IR which failed the Verifier in the backend.
https://llvm.org/bugs/show_bug.cgi?id=30341
This patch disables alias to available_externally definitions.
Reviewers: eugenis, rsmith
Differential Revision: https://reviews.llvm.org/D24682
llvm-svn: 282679
Peter Collingbourne [Thu, 29 Sep 2016 03:29:28 +0000 (03:29 +0000)]
Add explanatory comment.
llvm-svn: 282678
Eric Christopher [Thu, 29 Sep 2016 02:03:52 +0000 (02:03 +0000)]
Remove an unnecessary duplicate initialization of TLOF from the Mips
AsmPrinter. This was reinitializing the Mangler after we moved the
Mangler down to TLOF and causing us to have two different unnamed
global values accessed with the same name.
This should fix the problems on the ubsan tests here:
http://lab.llvm.org:8011/builders/clang-cmake-mips/builds/15307
llvm-svn: 282675
Eric Christopher [Thu, 29 Sep 2016 02:03:50 +0000 (02:03 +0000)]
Remove the default constructor and count variable from the Mangler since
we can just use the size of the DenseMap as a unique counter.
llvm-svn: 282674
Eric Christopher [Thu, 29 Sep 2016 02:03:47 +0000 (02:03 +0000)]
Update comment about initializing TLOF with a pointer at the previous
line or the other commented out place.
llvm-svn: 282673
Eric Christopher [Thu, 29 Sep 2016 02:03:44 +0000 (02:03 +0000)]
Tidy spelling and grammar.
llvm-svn: 282672
Lei Liu [Thu, 29 Sep 2016 02:01:27 +0000 (02:01 +0000)]
[AArch64] Fix test case ELF/aarch64-tls-le.s
The add instruction should be generated with shift bit (lsl #12).
Update the test case.
llvm-svn: 282671
Rui Ueyama [Thu, 29 Sep 2016 01:59:03 +0000 (01:59 +0000)]
Attempt to fix Windows buildbots.
llvm-svn: 282670
Matthias Braun [Thu, 29 Sep 2016 01:47:42 +0000 (01:47 +0000)]
MachineFunction: Add missing newline in debug print()
Should not be a functional but an aesthetic change.
llvm-svn: 282669
Rui Ueyama [Thu, 29 Sep 2016 01:45:22 +0000 (01:45 +0000)]
Remove unused #includes.
llvm-svn: 282668
Matt Arsenault [Thu, 29 Sep 2016 01:44:16 +0000 (01:44 +0000)]
AMDGPU: Partially fix control flow at -O0
Fixes to allow spilling all registers at the end of the block
work with exec modifications. Don't emit s_and_saveexec_b64 for
if lowering, and instead emit copies. Mark control flow mask
instructions as terminators to get correct spill code placement
with fast regalloc, and then have a separate optimization pass
form the saveexec.
This should work if SGPRs are spilled to VGPRs, but
will likely fail in the case that an SGPR spills to memory
and no workitem takes a divergent branch.
llvm-svn: 282667
Matthias Braun [Thu, 29 Sep 2016 01:32:31 +0000 (01:32 +0000)]
ScheduleDAGInstrs: There is no need to set OrigNode for MI SUnits; NFC
llvm-svn: 282666
Peter Collingbourne [Thu, 29 Sep 2016 01:28:36 +0000 (01:28 +0000)]
LTO: Fix use-after-scope error.
llvm-svn: 282665
Rui Ueyama [Thu, 29 Sep 2016 01:26:48 +0000 (01:26 +0000)]
Add comments.
llvm-svn: 282664
Rafael Espindola [Thu, 29 Sep 2016 01:20:40 +0000 (01:20 +0000)]
Read addend from the correct location.
llvm-svn: 282663
Davide Italiano [Thu, 29 Sep 2016 01:13:32 +0000 (01:13 +0000)]
[LTO] Fully qualify make_unique to make MSVC happy.
llvm-svn: 282662
Lei Liu [Thu, 29 Sep 2016 01:05:48 +0000 (01:05 +0000)]
AArch64: Set shift bit of TLSLE HI12 add instruction
Summary: AArch64 LLVM assembler emits add instruction without shift bit to calculate the higher 12-bit address of TLS variables in local exec model. This generates wrong code sequence to access TLS variables with thread offset larger than 0x1000.
Reviewers: t.p.northover, peter.smith, rovka
Subscribers: salim.nasser, aemerson, llvm-commits, rengolin
Differential Revision: https://reviews.llvm.org/D24702
llvm-svn: 282661
Eric Fiselier [Thu, 29 Sep 2016 01:01:26 +0000 (01:01 +0000)]
Partially revert overflow checking in last_write_time
llvm-svn: 282660
Jason Molenda [Thu, 29 Sep 2016 01:00:16 +0000 (01:00 +0000)]
Re-commit the changes from r282565 that I had to back out because of
a linux bot test failure. That one is fixed; hopefully there won't
be any others turned up this time.
The eh_frame augmentation code wasn't working right after the
reorg/rewrite of the classes. It works correctly now for the one
test that was failing - but we'll see what the test bots come up
with.
<rdar://problem/
28509178>
llvm-svn: 282659
Rui Ueyama [Thu, 29 Sep 2016 00:58:10 +0000 (00:58 +0000)]
Use InputFile::Symbol instead of auto for consistency.
llvm-svn: 282658
Sean Callanan [Thu, 29 Sep 2016 00:45:33 +0000 (00:45 +0000)]
Fixed TestObjCStructArgument/i386; expressions can now call ObjC class methods.
<rdar://problem/
28502241>
llvm-svn: 282657
Davide Italiano [Thu, 29 Sep 2016 00:40:08 +0000 (00:40 +0000)]
[ELF/LTO] Switch to the new resolution-based API.
Differential Revision: https://reviews.llvm.org/D24492
llvm-svn: 282656
Davide Italiano [Thu, 29 Sep 2016 00:31:45 +0000 (00:31 +0000)]
[LTO] Add a FIXME, we shouldn't expose getComdat().
Thanks to Peter for the suggestion.
llvm-svn: 282655
Davide Italiano [Thu, 29 Sep 2016 00:29:33 +0000 (00:29 +0000)]
[LTO] Expose getComdatSymbolTable() to linkers.
llvm-svn: 282654
Sean Callanan [Thu, 29 Sep 2016 00:16:37 +0000 (00:16 +0000)]
Introduced a null check to avoid a crash in a test on i386.
llvm-svn: 282653
Richard Smith [Thu, 29 Sep 2016 00:08:05 +0000 (00:08 +0000)]
Mark P0127R3 as done, and replace its __has_feature check with the corresponding SD-6 macro.
llvm-svn: 282652
Richard Smith [Wed, 28 Sep 2016 23:55:27 +0000 (23:55 +0000)]
P0127R2: Support type deduction for types of non-type template parameters in
C++1z.
Patch by James Touton! Some bugfixes and rebasing by me.
llvm-svn: 282651
Evgeny Stupachenko [Wed, 28 Sep 2016 23:39:39 +0000 (23:39 +0000)]
Wisely choose sext or zext when widening IV.
Summary:
The patch fixes regression caused by two earlier patches D18777 and D18867.
Reviewers: reames, sanjoy
Differential Revision: http://reviews.llvm.org/D24280
From: Li Huang
llvm-svn: 282650
Kevin Enderby [Wed, 28 Sep 2016 23:16:01 +0000 (23:16 +0000)]
Next set of additional error checks for invalid Mach-O files for the
load command that uses the Mach::rpath_command type
but not used in llvm libObject code but used in llvm tool code.
This includes just the LC_RPATH load command.
llvm-svn: 282649
Enrico Granata [Wed, 28 Sep 2016 22:53:16 +0000 (22:53 +0000)]
Fix an issue where libc++ changed the type information we get for std::map::iterator, rendering LLDB unable to display elements vended by an iterator
Fixes <rdar://problem/
28237521>
llvm-svn: 282648
Justin Lebar [Wed, 28 Sep 2016 22:45:58 +0000 (22:45 +0000)]
[CUDA] Disallow variable-length arrays in CUDA device code.
Reviewers: tra
Subscribers: cfe-commits, jhen
Differential Revision: https://reviews.llvm.org/D25050
llvm-svn: 282647
Justin Lebar [Wed, 28 Sep 2016 22:45:54 +0000 (22:45 +0000)]
[CUDA] Disallow exceptions in device code.
Reviewers: tra
Subscribers: cfe-commits, jhen
Differential Revision: https://reviews.llvm.org/D25036
llvm-svn: 282646
Gor Nishanov [Wed, 28 Sep 2016 22:37:17 +0000 (22:37 +0000)]
[Coroutines] Add proper mangling for operator co_await for MicrosoftABI
Reviewers: rnk, rsmith
Subscribers: mehdi_amini, cfe-commits
Differential Revision: https://reviews.llvm.org/D25045
llvm-svn: 282645
Shoaib Meenai [Wed, 28 Sep 2016 22:28:51 +0000 (22:28 +0000)]
[libc++] Clarify _LIBCPP_NEW_DELETE_VIS for Windows
Replace a stale reference to cxx_EXPORTS with _LIBCPP_BUILDING_LIBRARY,
and clarify why the operator new and delete family of functions are
marked dllexport when building but *not* dllimport when including the
header externally.
The new code is identical to the intent of the old code (and would be
functionally equivalent were cxx_EXPORTS still defined when building
libc++). The overall behavior is not ideal, since Microsoft's operator
new and delete functions will get called instead of libc++'s, but I
think consistently calling msvcrt's functions is better than either
calling msvcrt's or libc++'s functions depending on header inclusion.
Differential Revision: https://reviews.llvm.org/D25042
llvm-svn: 282644
Quentin Colombet [Wed, 28 Sep 2016 22:20:49 +0000 (22:20 +0000)]
[RegisterBankInfo] Uniquely generate OperandsMapping.
This is a step toward statically allocate InstructionMapping. Like the
previous few commits, the goal is to move toward a TableGen'ed like
structure with no dynamic allocation at all.
This should already improve compile time by getting rid of a bunch of
memmove of SmallVectors.
llvm-svn: 282643
Quentin Colombet [Wed, 28 Sep 2016 22:20:24 +0000 (22:20 +0000)]
[RegisterBankInfo] Rework the APIs of ValueMapping.
This is a preparatory commit for more TableGen-like structure.
NFC
llvm-svn: 282642
Richard Smith [Wed, 28 Sep 2016 22:08:38 +0000 (22:08 +0000)]
Fix bug where template argument deduction of a non-type template parameter used
as a template argument in a template-id, from a null non-type template
argument, failed.
Extracted from a patch by James Touton!
llvm-svn: 282641
Eric Fiselier [Wed, 28 Sep 2016 22:08:13 +0000 (22:08 +0000)]
Mark libc++ internal globals with _LIBCPP_SAFE_STATIC.
This patch applies the _LIBCPP_SAFE_STATIC attribute to internal globals,
most of which are locking primitives, in order to ensure that they can
safely be used during program startup.
This patch also fixes an unsafe static init issue with the global locks
used to implement atomic operations on shared pointers. Previously the
locks were initialized using a dynamically initialized pointer, so it was
possible that the pointer was uninitialized.
llvm-svn: 282640
Mike Aizatsky [Wed, 28 Sep 2016 21:39:28 +0000 (21:39 +0000)]
[sancov] introducing symbolized coverage files (.symcov)
Summary:
Answering any meaningful questions about .sancov files requires
accessing symbol information from the corresponding binary.
This change introduces a separate intermediate data structure and
format: symbolized coverage. It contains all symbol information that
is required to answer common queries:
- merging
- coverd/uncovered files and functions
- line status.
Also removing the html report functionality from sancov: generated
HTML files are too huge, and a different approach is required.
Maintaining this half-working approach in the C++ is painful.
Differential Revision: https://reviews.llvm.org/D24947
llvm-svn: 282639
Adrian Prantl [Wed, 28 Sep 2016 21:34:23 +0000 (21:34 +0000)]
Remove dead code from LiveDebugVariables.cpp (NFC)
LiveDebugVariables doesn't propagate DBG_VALUEs accross basic block
boundaries any more; this functionality was split into LiveDebugValues.
We can thus drop the now dead references to LexicalScopes from LiveDebugVariables.
llvm-svn: 282638
Mike Aizatsky [Wed, 28 Sep 2016 21:27:58 +0000 (21:27 +0000)]
[sancov] a simple .symcov coverage report server
Coverage reports for gigabyte-sized binaries are huge. There's no
practical reason to generate them statically.
Implementing an experiment http coverage report server. The server
loads .symcov file and serves interactive coverage pages.
llvm-svn: 282637
Kevin Enderby [Wed, 28 Sep 2016 21:20:45 +0000 (21:20 +0000)]
Next set of additional error checks for invalid Mach-O files for the
other load commands that use the Mach::version_min_command type
but not used in llvm libObject code but used in llvm tool code.
This includes LC_VERSION_MIN_MACOSX, LC_VERSION_MIN_IPHONEOS,
LC_VERSION_MIN_TVOS and LC_VERSION_MIN_WATCHOS load commands.
llvm-svn: 282635
Eric Fiselier [Wed, 28 Sep 2016 21:16:58 +0000 (21:16 +0000)]
Improve 'last_write_time(...)' accuracy and detect overflow errors.
The ::stat struct on Linux, FreeBSD, and OS X provides the access and
modification times as an instance of 'timespec', which has a nanosecond
resolution. The 'st_mtime' and 'st_atime' members simply reference the 'tv_sec'
value of the timespec struct. This patch changes 'last_write_time(...)' so that
it extracts both the seconds and nanoseconds values of the last modification
time, providing a more accurate implementation of 'last_write_time(...)'.
Additionally this patch fixes a possible signed integer overflow bug. The
'file_time_type' type cannot represent all possible values returned by
the filesystem. Attempting to construct a 'file_time_type' from one of these
values is undefined behavior. This patch avoids that UB by detecting possible
overflows before the conversion.
llvm-svn: 282634
Rui Ueyama [Wed, 28 Sep 2016 21:10:54 +0000 (21:10 +0000)]
Warn on empty archive files.
Differential Revision: https://reviews.llvm.org/D25044
llvm-svn: 282633
Greg Clayton [Wed, 28 Sep 2016 21:07:34 +0000 (21:07 +0000)]
Add the ability for the task port to change when a process execs.
<rdar://problem/
28476369>
llvm-svn: 282632
Richard Smith [Wed, 28 Sep 2016 21:01:37 +0000 (21:01 +0000)]
cxx_status: use HTML5 details/summary elements to hide implementation status
tables for fully-implemented language modes by default. Also add some missing
elements to TS support table.
llvm-svn: 282631
Dehao Chen [Wed, 28 Sep 2016 21:00:58 +0000 (21:00 +0000)]
Refactor the ProfileSummaryInfo to use doInitialization and doFinalization to handle Module update.
Summary: This refactors the change in r282616
Reviewers: davidxl, eraman, mehdi_amini
Subscribers: mehdi_amini, davide, llvm-commits
Differential Revision: https://reviews.llvm.org/D25041
llvm-svn: 282630
Richard Smith [Wed, 28 Sep 2016 20:42:56 +0000 (20:42 +0000)]
Add a couple more tentative names for upcoming SD-6 feature checks. These might
not reflect the final chosen names, but supporting them now seems to have
little downside.
llvm-svn: 282629
Todd Fiala [Wed, 28 Sep 2016 20:39:50 +0000 (20:39 +0000)]
use assertEquals in TestSBTypeClassMembers
This change replaces the self.assertTrue() calls with
self.assertEquals() so that test failures get more context on failure
values.
llvm-svn: 282628
Richard Smith [Wed, 28 Sep 2016 20:26:06 +0000 (20:26 +0000)]
P0095R3: Implement the latest published revision of SD-6 (C++ feature test macros).
llvm-svn: 282627
Krzysztof Parzyszek [Wed, 28 Sep 2016 20:07:41 +0000 (20:07 +0000)]
IfConversion: Add implicit uses for redefined regs with live subregisters
Normally, if conversion would add implicit uses for redefined registers,
e.g. R0<def> = add_if ..., R0<imp-use>. However, if only subregisters of
R0 are known to be live but not R0 itself, such implicit uses will not be
added, causing prior definitions of such subregisters and R0 itself to
become dead.
llvm-svn: 282626
Matthias Gehre [Wed, 28 Sep 2016 20:06:18 +0000 (20:06 +0000)]
[clang-tidy] fix false-positive for cppcoreguidelines-pro-type-member-init with in-class initializers
Summary:
This fixes https://llvm.org/bugs/show_bug.cgi?id=30487 where
```
warning: uninitialized record type: 's' [cppcoreguidelines-pro-type-member-init]
```
is emitted on
```
struct MyStruct
{
int a = 5;
int b = 7;
};
int main()
{
MyStruct s;
}
```
Reviewers: alexfh, aaron.ballman
Subscribers: nemanjai, cfe-commits
Differential Revision: https://reviews.llvm.org/D24848
llvm-svn: 282625
Konstantin Zhuravlyov [Wed, 28 Sep 2016 20:05:39 +0000 (20:05 +0000)]
[AMDGPU] Promote uniform i16 ops to i32 ops for targets that have 16 bit instructions
Differential Revision: https://reviews.llvm.org/D24125
llvm-svn: 282624
Richard Smith [Wed, 28 Sep 2016 19:44:50 +0000 (19:44 +0000)]
P0096R2: Implement more recent revision of SD-6 (C++ feature test macros).
llvm-svn: 282622
Richard Smith [Wed, 28 Sep 2016 19:22:36 +0000 (19:22 +0000)]
Long-overdue update to cxx_status: C++14 is no longer "upcoming".
llvm-svn: 282621
Sanjay Patel [Wed, 28 Sep 2016 19:10:16 +0000 (19:10 +0000)]
[InstCombine] update to use FileCheck
Also, remove unnecessary function attributes, parameters, and comments.
It looks like at least some of these tests are not minimal though...
llvm-svn: 282620
Richard Smith [Wed, 28 Sep 2016 19:09:10 +0000 (19:09 +0000)]
Re-commit r282556, reverted in r282564, with a fix to CallArgList::addFrom to
function correctly when targeting MS ABIs (this appears to have never mattered
prior to this change).
Update test case to always cover both 32-bit and 64-bit Windows ABIs, since
they behave somewhat differently from each other here.
Update test case to also cover operators , && and ||, which it appears are also
affected by P0145R3 (they're not explicitly called out by the design document,
but this is the emergent behavior of the existing wording).
Original commit message:
P0145R3 (C++17 evaluation order tweaks): evaluate the right-hand side of
assignment and compound-assignment operators before the left-hand side. (Even
if it's an overloaded operator.)
This completes the implementation of P0145R3 + P0400R0 for all targets except
Windows, where the evaluation order guarantees for <<, >>, and ->* are
unimplementable as the ABI requires the function arguments are evaluated from
right to left (because parameter destructors are run from left to right in the
callee).
llvm-svn: 282619
Dehao Chen [Wed, 28 Sep 2016 18:54:36 +0000 (18:54 +0000)]
Fix the bug introduced in r282616.
llvm-svn: 282618
Vitaly Buka [Wed, 28 Sep 2016 18:46:52 +0000 (18:46 +0000)]
Add build script for symbolizer which can be linked into instrumented process.
Reviewers: eugenis
Subscribers: llvm-commits, kubabrecka
Differential Revision: https://reviews.llvm.org/D24974
llvm-svn: 282617
Dehao Chen [Wed, 28 Sep 2016 18:41:14 +0000 (18:41 +0000)]
Fix the bug when -compile-twice is specified, the PSI will be invalidated.
Summary:
When using llc with -compile-twice, module is generated twice, but getAnalysis<ProfileSummaryInfoWrapperPass>().getPSI will still get the old PSI with the original (invalidated) Module. This patch checks if the module has changed when calling getPSI, if yes, update the module and invalidate the Summary.
The bug does not show up in the current llc because PSI is not used in CodeGen yet. But with https://reviews.llvm.org/D24989, the bug will be exposed by test/CodeGen/PowerPC/pr26378.ll
Reviewers: eraman, davidxl
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D24993
llvm-svn: 282616
Etienne Bergeron [Wed, 28 Sep 2016 18:04:07 +0000 (18:04 +0000)]
[compiler-rt] Fix interception of multiple defined symbols.
Summary:
The MSVC compiler is generating multiple instance of the exception handler
when compiling on win64 with /MD.
see: https://msdn.microsoft.com/en-us/library/2kzt1wy3.aspx
Two tests were failing when running:
```
ninja check-asan-dynamic.
```
The tests were failing because only the first occurence of the function was patched.
The function `__C_specific_handler` is defined in `ntdll` and `vcruntime140`.
After this patch, there is still two remaining tests failing.
```
********************
Testing: 0 .. 10.. 20.. 30.. 40.. 50.. 60.. 70.. 80.. 90..
Testing Time: 87.81s
********************
Failing Tests (2):
AddressSanitizer-x86_64-windows-dynamic :: TestCases/Windows/dll_intercept_memchr.cc
AddressSanitizer-x86_64-windows-dynamic :: TestCases/Windows/dll_intercept_memcpy_indirect.cc
Expected Passes : 342
Passes With Retry : 2
Expected Failures : 16
Unsupported Tests : 152
Unexpected Failures: 2
```
Reviewers: rnk, vitalybuka
Subscribers: vitalybuka, llvm-commits, chrisha, dberris
Differential Revision: https://reviews.llvm.org/D24983
llvm-svn: 282614
Simon Pilgrim [Wed, 28 Sep 2016 17:59:30 +0000 (17:59 +0000)]
[X86][AVX] Add test showing that VBROADCAST loads don't correctly respect dependencies
llvm-svn: 282613
Artur Pilipenko [Wed, 28 Sep 2016 17:57:16 +0000 (17:57 +0000)]
Don't look through addrspacecast in GetPointerBaseWithConstantOffset
Pointers in different addrspaces can have different sizes, so it's not valid to look through addrspace cast calculating base and offset for a value.
This is similar to D13008.
Reviewed By: reames
Differential Revision: https://reviews.llvm.org/D24729
llvm-svn: 282612
Adrian Prantl [Wed, 28 Sep 2016 17:51:14 +0000 (17:51 +0000)]
Teach LiveDebugValues about lexical scopes.
This addresses PR26055 LiveDebugValues is very slow.
Contrary to the old LiveDebugVariables pass LiveDebugValues currently
doesn't look at the lexical scopes before inserting a DBG_VALUE
intrinsic. This means that we often propagate DBG_VALUEs much further
down than necessary. This is especially noticeable in large C++
functions with many inlined method calls that all use the same
"this"-pointer.
For example, in the following code it makes no sense to propagate the
inlined variable a from the first inlined call to f() into any of the
subsequent basic blocks, because the variable will always be out of
scope:
void sink(int a);
void __attribute((always_inline)) f(int a) { sink(a); }
void foo(int i) {
f(i);
if (i)
f(i);
f(i);
}
This patch reuses the LexicalScopes infrastructure we have for
LiveDebugVariables to take this into account.
The effect on compile time and memory consumption is quite noticeable:
I tested a benchmark that is a large C++ source with an enormous
amount of inlined "this"-pointers that would previously eat >24GiB
(most of them for DBG_VALUE intrinsics) and whose compile time was
dominated by LiveDebugValues. With this patch applied the memory
consumption is 1GiB and 1.7% of the time is spent in LiveDebugValues.
https://reviews.llvm.org/D24994
Thanks to Daniel Berlin and Keith Walker for reviewing!
llvm-svn: 282611
Artem Belevich [Wed, 28 Sep 2016 17:47:40 +0000 (17:47 +0000)]
[CUDA] Added support for CUDA-8
Differential Revision: https://reviews.llvm.org/D24946
llvm-svn: 282610
Artem Belevich [Wed, 28 Sep 2016 17:47:35 +0000 (17:47 +0000)]
[CUDA] added __nvvm_atom_{sys|cta}_* builtins.
These builtins are available on sm_60+ GPU only.
Differential Revision: https://reviews.llvm.org/D24944
llvm-svn: 282609
Adrian Prantl [Wed, 28 Sep 2016 17:31:17 +0000 (17:31 +0000)]
Rewrite loops to use range-based for. (NFC)
llvm-svn: 282608
Artem Belevich [Wed, 28 Sep 2016 17:25:38 +0000 (17:25 +0000)]
[NVPTX] Added intrinsics for atom.gen.{sys|cta}.* instructions.
These are only available on sm_60+ GPUs.
Differential Revision: https://reviews.llvm.org/D24943
llvm-svn: 282607
Sanjoy Das [Wed, 28 Sep 2016 17:14:58 +0000 (17:14 +0000)]
[SCEV] Use a SmallPtrSet as a temporary union predicate; NFC
Summary:
Instead of creating and destroying SCEVUnionPredicate instances (which
internally creates and destroys a DenseMap), use temporary SmallPtrSet
instances of remember the set of predicates that will get reified into a
SCEVUnionPredicate.
Reviewers: silviu.baranga, sbaranga
Subscribers: sanjoy, mcrosier, llvm-commits, mzolotukhin
Differential Revision: https://reviews.llvm.org/D25000
llvm-svn: 282606
Todd Fiala [Wed, 28 Sep 2016 16:43:47 +0000 (16:43 +0000)]
zorg Xcode python test suite target arch update
This changes the Xcode target used by the Green Dragon Xcode CI.
When calling xcodebuild with LLDB_PYTHON_TESTSUITE_ARCH set, the
arch's xUnit XML output is now set to an arch-specific filename:
$(BUILD_DIR)/test-results-$(LLDB_PYTHON_TESTSUITE_ARCH).xml.
The change also ensures that the Python testsuite sees the Xcode
build settings passed in through environment variables.
llvm-svn: 282605
Nirav Dave [Wed, 28 Sep 2016 16:37:50 +0000 (16:37 +0000)]
Revert "In visitSTORE, always use FindBetterChain, rather than only when UseAA is enabled."
This reverts commit r282600 due to test failues with MCJIT
llvm-svn: 282604
Rafael Espindola [Wed, 28 Sep 2016 16:31:05 +0000 (16:31 +0000)]
Sort headers. NFC.
llvm-svn: 282603
Dylan McKay [Wed, 28 Sep 2016 16:04:40 +0000 (16:04 +0000)]
[AVR] Rename the builtin calling convention names
'BUILTIN' is clearer than 'RT' in this context.
llvm-svn: 282602
Marina Yatsina [Wed, 28 Sep 2016 15:52:56 +0000 (15:52 +0000)]
[x86] Accept 'retn' as an alias to 'ret[lqw]'\'ret' (At&t\Intel)
Implement 'retn' simply by aliasing it to the relevant 'ret' instruction
Commit on behalf of coby
Differential Revision: https://reviews.llvm.org/D24346
llvm-svn: 282601
Nirav Dave [Wed, 28 Sep 2016 15:50:43 +0000 (15:50 +0000)]
In visitSTORE, always use FindBetterChain, rather than only when UseAA is enabled.
Simplify Consecutive Merge Store Candidate Search
Now that address aliasing is much less conservative, push through
simplified store merging search which only checks for parallel stores
through the chain subgraph. This is cleaner as the separation of
non-interfering loads/stores from the store-merging logic.
Whem merging stores, search up the chain through a single load, and
finds all possible stores by looking down from through a load and a
TokenFactor to all stores visited. This improves the quality of the
output SelectionDAG and generally the output CodeGen (with some
exceptions).
Additional Minor Changes:
1. Finishes removing unused AliasLoad code
2. Unifies the the chain aggregation in the merged stores across
code paths
3. Re-add the Store node to the worklist after calling
SimplifyDemandedBits.
4. Increase GatherAllAliasesMaxDepth from 6 to 18. That number is
arbitrary, but seemed sufficient to not cause regressions in
tests.
This finishes the change Matt Arsenault started in r246307 and
jyknight's original patch.
Many tests required some changes as memory operations are now
reorderable. Some tests relying on the order were changed to use
volatile memory operations
Noteworthy tests:
CodeGen/AArch64/argument-blocks.ll -
It's not entirely clear what the test_varargs_stackalign test is
supposed to be asserting, but the new code looks right.
CodeGen/AArch64/arm64-memset-inline.lli -
CodeGen/AArch64/arm64-stur.ll -
CodeGen/ARM/memset-inline.ll -
The backend now generates *worse* code due to store merging
succeeding, as we do do a 16-byte constant-zero store efficiently.
CodeGen/AArch64/merge-store.ll -
Improved, but there still seems to be an extraneous vector insert
from an element to itself?
CodeGen/PowerPC/ppc64-align-long-double.ll -
Worse code emitted in this case, due to the improved store->load
forwarding.
CodeGen/X86/dag-merge-fast-accesses.ll -
CodeGen/X86/MergeConsecutiveStores.ll -
CodeGen/X86/stores-merging.ll -
CodeGen/Mips/load-store-left-right.ll -
Restored correct merging of non-aligned stores
CodeGen/AMDGPU/promote-alloca-stored-pointer-value.ll -
Improved. Correctly merges buffer_store_dword calls
CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll -
Improved. Sidesteps loading a stored value and merges two stores
CodeGen/X86/pr18023.ll -
This test has been removed, as it was asserting incorrect
behavior. Non-volatile stores *CAN* be moved past volatile loads,
and now are.
CodeGen/X86/vector-idiv.ll -
CodeGen/X86/vector-lzcnt-128.ll -
It's basically impossible to tell what these tests are actually
testing. But, looks like the code got better due to the memory
operations being recognized as non-aliasing.
CodeGen/X86/win32-eh.ll -
Both loads of the securitycookie are now merged.
CodeGen/AMDGPU/vgpr-spill-emergency-stack-slot-compute.ll -
This test appears to work but no longer exhibits the spill
behavior.
Reviewers: arsenm, hfinkel, tstellarAMD, nhaehnle, jyknight
Subscribers: wdng, nhaehnle, nemanjai, arsenm, weimingz, niravd, RKSimon, aemerson, qcolombet, resistor, tstellarAMD, t.p.northover, spatel
Differential Revision: https://reviews.llvm.org/D14834
llvm-svn: 282600
Ed Maste [Wed, 28 Sep 2016 15:37:21 +0000 (15:37 +0000)]
libunwind: Add OpenBSD case for _Unwind_Ptr typedef
Patch by Mark Kettenis
llvm-svn: 282599
Dylan McKay [Wed, 28 Sep 2016 15:35:26 +0000 (15:35 +0000)]
[AVR] Import the LLVM namespace inside AVRMCTargetDesc.cpp
llvm-svn: 282598
Dylan McKay [Wed, 28 Sep 2016 15:31:12 +0000 (15:31 +0000)]
[AVR] Add AVRMCTargetDesc.cpp
Summary:
This adds the AVRMCTargetDesc file in tree. It allows creation of the
core classes used in the backend.
Reviewers: arsenm, kparzysz
Subscribers: wdng, beanz, mgorny
Differential Revision: https://reviews.llvm.org/D25023
llvm-svn: 282597
Petr Hosek [Wed, 28 Sep 2016 15:20:47 +0000 (15:20 +0000)]
[ELF] Support -z max-page-size option
This options issupported by both BFD ld and gold and allows
overriding the max page size whose default values are defined by
the target.
https://llvm.org/bugs/show_bug.cgi?id=30541
Differential Revision: https://reviews.llvm.org/D24891
llvm-svn: 282596
George Rimar [Wed, 28 Sep 2016 15:11:58 +0000 (15:11 +0000)]
[ELF] - Created new "Invalid" subfolder for testcases.
This subfolder just like "linkerscript" subfolder keeps
testcases with invalid input. According to PR30540 it seems
we might have many new ones soon, so it is seems reasonable to
separate them from regular testcases.
Differential revision: https://reviews.llvm.org/D25010
llvm-svn: 282595
Dylan McKay [Wed, 28 Sep 2016 14:35:07 +0000 (14:35 +0000)]
[AVR] Update the signature of createAVRAsmBackend
It has been recently changed to also take a MCTargetOptions structure.
llvm-svn: 282594
Dylan McKay [Wed, 28 Sep 2016 14:34:42 +0000 (14:34 +0000)]
[AVR] Enable the assembly parser
We very recently landed the code. This commit enables the parser.
It also adds a missing include to AVRAsmParser.cpp
llvm-svn: 282593
Sanjay Patel [Wed, 28 Sep 2016 14:27:21 +0000 (14:27 +0000)]
[InstSimplify] allow or-of-icmps folds with vector splat constants
llvm-svn: 282592
Sanjay Patel [Wed, 28 Sep 2016 14:17:35 +0000 (14:17 +0000)]
[InstSimplify] add vector splat tests for or-of-icmps
llvm-svn: 282591
Sanjay Patel [Wed, 28 Sep 2016 13:53:13 +0000 (13:53 +0000)]
[InstSimplify] allow and-of-icmps folds with vector splat constants
llvm-svn: 282590
Ed Schouten [Wed, 28 Sep 2016 13:51:23 +0000 (13:51 +0000)]
Also use the proper register numbers on CloudABI.
Without this change applied, unw_step() fails to obtain the next frame
properly.
llvm-svn: 282589
Dylan McKay [Wed, 28 Sep 2016 13:44:02 +0000 (13:44 +0000)]
[AVR] Merge most recent changes to AVRInstrInfo.td
This adds two new things:
- Operand types per fixup
- Atomic pseudo operations
llvm-svn: 282588
Dylan McKay [Wed, 28 Sep 2016 13:29:10 +0000 (13:29 +0000)]
[AVR] Update the data layout
The previous data layout caused issues when dealing with atomics.
Foe example, it is illegal to load a 16-bit value with less than 16-bits
of alignment.
This changes the data layout so that all types are aligned by at least
their own width.
Interestingly, this also _slightly_ decreased register pressure in some
cases.
llvm-svn: 282587
Dylan McKay [Wed, 28 Sep 2016 13:23:42 +0000 (13:23 +0000)]
[AVR] Handle AVR relocations when handling ELF files
llvm-svn: 282586
Dylan McKay [Wed, 28 Sep 2016 13:15:17 +0000 (13:15 +0000)]
[AVR] Allow llvm-objdump to handle AVR ELF files
llvm-svn: 282585
Dylan McKay [Wed, 28 Sep 2016 13:02:57 +0000 (13:02 +0000)]
[AVR] Add assembly parser
Summary: This patch adds the AVRAsmParser library.
Reviewers: arsenm, kparzysz
Subscribers: wdng, beanz, mgorny, kparzysz, simoncook, jtbandes, llvm-commits
Differential Revision: https://reviews.llvm.org/D20046
llvm-svn: 282584