Adhemerval Zanella [Mon, 5 Dec 2016 14:15:44 +0000 (14:15 +0000)]
ELF/AArch64: Fix R_AARCH64_LDST16_ABS_LO12_NC mask
The relocation R_AARCH64_LDST16_ABS_LO12_NC should set a ld/st
immediate value to bits [11:1] not [11:2]. This patches fixes it
and adds a testcase for regression.
With this fix all the faulty tests on test-suite (clavm, lencod,
and trimaran) pass.
llvm-svn: 288670
Adhemerval Zanella [Mon, 5 Dec 2016 14:15:03 +0000 (14:15 +0000)]
ELF/AArch64: Simplify R_AARCH64_ADD_ABS_LO12_NC relocation
This patch uses the updateAArch64Add on relocation apply and remove
the comment.
llvm-svn: 288669
Adhemerval Zanella [Mon, 5 Dec 2016 14:14:26 +0000 (14:14 +0000)]
ELF/AArch64: consolidate getAArch64Page implementation
This patch avoid getAArch64Page code duplication by removing the
implementation at InputSection.
llvm-svn: 288668
Nirav Dave [Mon, 5 Dec 2016 14:11:03 +0000 (14:11 +0000)]
[PPC] Slightly Improve Assembly Parsing errors and add EOL comment
parsing tests.
NFC intended.
llvm-svn: 288667
Simon Dardis [Mon, 5 Dec 2016 12:55:19 +0000 (12:55 +0000)]
[mips][ias] N32/N64 must not sort the relocation table.
Doing so changes the evaluation order for relocation composition.
Patch By: Daniel Sanders
Reviewers: vkalintiris, atanasyan
Differential Revision: https://reviews.llvm.org/D26401
llvm-svn: 288666
Martin Bohme [Mon, 5 Dec 2016 11:33:19 +0000 (11:33 +0000)]
CFGBuilder: Fix crash when visiting delete expression on dependent type
Summary:
CXXDeleteExpr::getDestroyedType() can return a null QualType if the destroyed
type is a dependent type. This patch protects against this.
Reviewers: klimek
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D27350
llvm-svn: 288665
Benjamin Kramer [Mon, 5 Dec 2016 11:30:24 +0000 (11:30 +0000)]
Clean out unused diagnostics. NFC.
llvm-svn: 288664
Simon Pilgrim [Mon, 5 Dec 2016 11:25:13 +0000 (11:25 +0000)]
[X86][SSE] Add support for combining target shuffles to UNPCKL/UNPCKH.
llvm-svn: 288663
Eric Liu [Mon, 5 Dec 2016 11:17:04 +0000 (11:17 +0000)]
[change-namespace] get changing namespace to global correct.
llvm-svn: 288662
Pavel Labath [Mon, 5 Dec 2016 11:15:36 +0000 (11:15 +0000)]
Clean up some use of __ANDROID_NDK__ in the cmake files
Rationale:
scripts/Python/modules: android is excluded at a higher level, so no point in
checking here
tools/lldb-mi: lldb-mi builds fine (with some cosmetic tweaks) on android, and
there is no reason it shouldn't.
tools/lldb-server: LLDB_DISABLE_LIBEDIT/CURSES already take the platform into
account, so there is no point in checking again.
I am reasonably confident this should not break the build on any platform, but
I'll keep an eye out on the bots.
llvm-svn: 288661
Roger Ferrer Ibanez [Mon, 5 Dec 2016 11:05:09 +0000 (11:05 +0000)]
Handle tests for noexcept that expect a false value
Under libcpp-no-exceptions, noexcept is trivially true. Some tests expect in
the usual setting to return false, so adjust them to expect true under
libcpp-no-exceptions.
Differential Revision: https://reviews.llvm.org/D27310
llvm-svn: 288660
Simon Pilgrim [Mon, 5 Dec 2016 11:00:25 +0000 (11:00 +0000)]
[X86][SSE] Add helper function to create UNPCKL/UNPCKH shuffle masks. NFCI.
llvm-svn: 288659
Diana Picus [Mon, 5 Dec 2016 10:40:33 +0000 (10:40 +0000)]
[GlobalISel] Extract handleAssignments out of AArch64CallLowering
This function seems target-independent so far: all the target-specific behaviour
is isolated in the CCAssignFn and the ValueHandler (which we're also extracting
into the generic CallLowering).
The intention is to use this in the ARM backend.
Differential Revision: https://reviews.llvm.org/D27045
llvm-svn: 288658
Sam Kolton [Mon, 5 Dec 2016 09:58:51 +0000 (09:58 +0000)]
[AMDGPU] Disassembler: fix s_buffer_store_dword instructions
Summary: s_buffer_store_dword instructions sdata operand was called sdst in encoding. This caused disassembler to fail.
Reviewers: tstellarAMD, vpykhtin, artem.tamazov
Subscribers: arsenm, nhaehnle, rampitec
Differential Revision: https://reviews.llvm.org/D27100
llvm-svn: 288657
Michal Gorny [Mon, 5 Dec 2016 09:15:05 +0000 (09:15 +0000)]
[cmake] Include component in Sphinx install rules
Include component in install rules for Sphinx targets. Based on
a similar suggestion for other doc targets in D24935.
Differential Revision: https://reviews.llvm.org/D24982
llvm-svn: 288656
Matthias Braun [Mon, 5 Dec 2016 08:15:57 +0000 (08:15 +0000)]
TableGen/AsmMatcherEmitter: Trust that stable_sort works
A debug build of AsmMatcherEmitter would use a quadratic algorithm to
check whether std::stable_sort() actually sorted. Let's hope the authors
of our C++ standard library did that testing for us. Removing the check
gives a 3x speedup in the X86 case.
llvm-svn: 288655
Richard Smith [Mon, 5 Dec 2016 07:49:14 +0000 (07:49 +0000)]
DR1213: element access on an array xvalue or prvalue produces an xvalue. In the
latter case, a temporary array object is materialized, and can be
lifetime-extended by binding a reference to the member access. Likewise, in an
array-to-pointer decay, an rvalue array is materialized before being converted
into a pointer.
This caused IR generation to stop treating file-scope array compound literals
as having static storage duration in some cases in C++; that has been rectified
by modeling such a compound literal as an lvalue. This also improves clang's
compatibility with GCC for those cases.
llvm-svn: 288654
Matthias Braun [Mon, 5 Dec 2016 07:35:13 +0000 (07:35 +0000)]
TableGen: Some more std::string->StringInit* replacements
llvm-svn: 288653
Matthias Braun [Mon, 5 Dec 2016 07:35:09 +0000 (07:35 +0000)]
TableGen/Record: Shortcut member access in hottest function
This may seem unusual, but makes most debug tblgen builds ~10% faster.
Usually we wouldn't care about speed that much in debug builds, but for
tblgen that also translates into build time.
llvm-svn: 288652
Matthias Braun [Mon, 5 Dec 2016 07:04:19 +0000 (07:04 +0000)]
TableGen: TableGenStringKey is no longer necessary as of r288642
llvm-svn: 288651
Matthias Braun [Mon, 5 Dec 2016 07:00:44 +0000 (07:00 +0000)]
TableGen: Use range based for; reserve vectors where possible
llvm-svn: 288650
Matthias Braun [Mon, 5 Dec 2016 06:41:54 +0000 (06:41 +0000)]
TableGen/TGParser: Prefer SmallVector/ArrayRef over std::vector
llvm-svn: 288649
Matthias Braun [Mon, 5 Dec 2016 06:41:51 +0000 (06:41 +0000)]
TableGen/Record: Replace std::vector with SmallVector/ArrayRef
llvm-svn: 288648
Matthias Braun [Mon, 5 Dec 2016 06:41:47 +0000 (06:41 +0000)]
ListInit::convertInitializerTo: avoid foldingset lookup if nothing changed
llvm-svn: 288647
Craig Topper [Mon, 5 Dec 2016 06:09:55 +0000 (06:09 +0000)]
[X86] Remove unnecessary explicit uses of .SimpleTy just to do an equality comparison. MVT's operator== already takes care of this. NFCI
llvm-svn: 288646
Matthias Braun [Mon, 5 Dec 2016 06:00:51 +0000 (06:00 +0000)]
Adapt to llvm/TableGen DagInit changes.
llvm-svn: 288645
Matthias Braun [Mon, 5 Dec 2016 06:00:46 +0000 (06:00 +0000)]
TableGen: Use StringInit instead of std::string for DagInit arg names
llvm-svn: 288644
Matthias Braun [Mon, 5 Dec 2016 06:00:41 +0000 (06:00 +0000)]
TableGen: Use StringInit instead of std::string for DagInit name
llvm-svn: 288643
Matthias Braun [Mon, 5 Dec 2016 06:00:36 +0000 (06:00 +0000)]
TableGen: Use more StringInit instead of StringRef
This forces the code to call StringInit::get on the string early and
avoids storing duplicates in std::string and sometimes allows pointer
comparisons instead of string comparisons.
llvm-svn: 288642
Craig Topper [Mon, 5 Dec 2016 05:50:51 +0000 (05:50 +0000)]
[AVX-512] Teach fast isel to handle 512-bit vector bitcasts.
llvm-svn: 288641
Kuba Mracek [Mon, 5 Dec 2016 05:21:44 +0000 (05:21 +0000)]
Use Darwin libtool's -no_warning_for_no_symbols if available to silence the "has no symbols" link warning
Building compiler-rt on Darwin produces dozens of meaningless warnings about object files having no symbols during static archive creation. This is very intentional as compiler-rt uses #ifdefs to conditionally compile platform-specific code, and we even have a .cpp source file that only contains static asserts to make sure the environment is configured right. On Linux, this situation is fine and no warning is produced. This patch adds a libtool version detection and if it's new enough, we'll use the -no_warning_for_no_symbols flag that suppresses this warning. Build logs should be much cleaner now!
Differential Revision: https://reviews.llvm.org/D27119
llvm-svn: 288640
Matthias Braun [Mon, 5 Dec 2016 05:21:18 +0000 (05:21 +0000)]
TableGen: Factor out STRCONCAT constructor, add shortcut.
Introduce new constructor for STRCONCAT binop with a shortcut that
immediately concatenates if the two arguments are StringInits.
Makes the QualifyName code more readable and tablegen 2-3% faster.
llvm-svn: 288639
Matthias Braun [Mon, 5 Dec 2016 05:21:13 +0000 (05:21 +0000)]
TableGen/Record: Move PointerIntPair to less used field of RecordVal
llvm-svn: 288638
Colin LeMahieu [Mon, 5 Dec 2016 04:52:28 +0000 (04:52 +0000)]
[Hexagon] Adding additional tokenization characters in preparation for removing spacing from syntax.
llvm-svn: 288637
Craig Topper [Mon, 5 Dec 2016 04:51:31 +0000 (04:51 +0000)]
[AVX-512] Teach fast isel to use masked compare and movss for handling scalar cmp and select sequence when AVX-512 is enabled. This matches the behavior of normal isel.
llvm-svn: 288636
Craig Topper [Mon, 5 Dec 2016 04:51:28 +0000 (04:51 +0000)]
[AVX-512] Add avx512f command lines to fast isel SSE select test.
Currently the fast isel code emits an avx1 instruction sequence even with avx512. This is different than normal isel. A follow up commit will fix this.
llvm-svn: 288635
Colin LeMahieu [Mon, 5 Dec 2016 04:29:00 +0000 (04:29 +0000)]
[Hexagon] Changing from literal numeric value to argument since #-1 will not parse when '-' is converted to a token.
llvm-svn: 288634
Chris Bieneman [Mon, 5 Dec 2016 03:29:10 +0000 (03:29 +0000)]
[CMake] Use add_llvm_tool_symlink's OUTPUT_DIR option
This is updating to take r288632 into account.
llvm-svn: 288633
Chris Bieneman [Mon, 5 Dec 2016 03:28:03 +0000 (03:28 +0000)]
[CMake] Refactor add_llvm_tool_symlink for reuse
The old implementation of add_llvm_tool_symlink could fail in odd ways when building out of tree. This version solves that problem by not using the LLVM_* variables, and instead reaeding the target's properties.
llvm-svn: 288632
Rui Ueyama [Mon, 5 Dec 2016 02:07:29 +0000 (02:07 +0000)]
Run the last iteration of parallel_for_loop using a threadpool.
Remainders of tasks were ran in the main thread, so parallel_for_each
could theoretically take 2x time than the ideal.
llvm-svn: 288631
Rui Ueyama [Mon, 5 Dec 2016 01:31:39 +0000 (01:31 +0000)]
Simplify ICF alignment handling.
llvm-svn: 288630
Peter Collingbourne [Mon, 5 Dec 2016 00:02:18 +0000 (00:02 +0000)]
IRGen: Remove an unused overload of CreateAlignedLoad.
llvm-svn: 288629
Simon Pilgrim [Sun, 4 Dec 2016 22:55:57 +0000 (22:55 +0000)]
[X86][XOP] Add target shuffle tests showing missing UNPCKL combine.
llvm-svn: 288628
Simon Pilgrim [Sun, 4 Dec 2016 22:54:21 +0000 (22:54 +0000)]
[X86][AVX512] Add target shuffle tests showing missing UNPCK combines.
llvm-svn: 288627
Daniel Jasper [Sun, 4 Dec 2016 22:34:37 +0000 (22:34 +0000)]
Revert "Recover better from an incompatible .pcm file being provided by -fmodule-file=. We try to include the headers of the module textually in this case, still enforcing the modules semantic rules. In order to make that work, we need to still track that we're entering and leaving the module. Also, if the module was also marked as unavailable (perhaps because it was missing a file), we shouldn't mark the module unavailable -- we don't need the module to be complete if we're going to enter it textually."
This reverts commit r288449.
I believe that this is currently faulty wrt. modules being imported
inside namespaces. Adding these lines to the new test:
namespace n {
#include "foo.h"
}
Makes it break with
fatal error: import of module 'M' appears within namespace 'n'
However, I believe it should fail with
error: redundant #include of module 'M' appears within namespace 'n'
I have tracked this down to us now inserting a tok::annot_module_begin
instead of a tok::annot_module_include in
Preprocessor::HandleIncludeDirective() and then later in
Parser::parseMisplacedModuleImport(), we hit the code path for
tok::annot_module_begin, which doesn't set FromInclude of
checkModuleImportContext to true (thus leading to the "wrong"
diagnostic).
llvm-svn: 288626
Eric Fiselier [Sun, 4 Dec 2016 22:14:53 +0000 (22:14 +0000)]
Update status page for variant implementation
llvm-svn: 288625
Kuba Mracek [Sun, 4 Dec 2016 21:52:21 +0000 (21:52 +0000)]
[sanitizer] Make atos stdin a non-tty pipe to make sure it's not stuck waiting for user input
On macOS, we often symbolicate using atos (when llvm-symbolizer is not found). The current way we invoke atos involves creating a pseudo-terminal to make sure atos doesn't buffer its output. This however also makes atos think that it's stdin is interactive and in some error situations it will ask the user to enter some input instead of just printing out an error message. For example, when Developer Mode isn't enabled on a machine, atos cannot examine processes, and it will ask the user to enter an administrator's password, which will make the sanitized process get stuck. This patch only connects the pseudo-terminal to the stdout of atos, and uses a regular pipe as its stdin.
Differential Revision: https://reviews.llvm.org/D27239
llvm-svn: 288624
Eric Fiselier [Sun, 4 Dec 2016 21:37:37 +0000 (21:37 +0000)]
Choose better hash values for std::monostate and valueless variants.
Previously these hashes were 0 and -1 respectively. These seem like common
sentinel values and should be avoided to prevent needless collisions.
This patch changes those values to different arbitrary numbers, which should
hopefully cause less collisions. Because I couldn't help myself I choose the
fundamental constants for gravity and the speed of light.
llvm-svn: 288623
Craig Topper [Sun, 4 Dec 2016 21:13:05 +0000 (21:13 +0000)]
[X86] Mark 256-bit DPPS intrinsic as commutable to increase load matching opportunities.
llvm-svn: 288622
Craig Topper [Sun, 4 Dec 2016 21:13:01 +0000 (21:13 +0000)]
[X86] Add Commutative property to several MMX arithmetic and logic intrinsics.
I think these intrinsics were added after the Commutative was added to most of the rest of the intrinsics and it must have been forgotten.
llvm-svn: 288621
Rui Ueyama [Sun, 4 Dec 2016 16:33:13 +0000 (16:33 +0000)]
Re-implement the optimization that I removed in r288527.
I removed a wrong optimization for ICF in r288527. Sean Silva suggested
in a post commit review that the correct algorithm can be implemented
easily. So is this patch.
llvm-svn: 288620
Michael Gottesman [Sun, 4 Dec 2016 10:26:53 +0000 (10:26 +0000)]
[stl-extras] Provide an adaptor of std::count for ranges.
llvm-svn: 288619
Dylan McKay [Sun, 4 Dec 2016 09:50:42 +0000 (09:50 +0000)]
[AVR] Remove 'XFAIL' from a CodeGen test
This seems to be fixed as of r288052.
llvm-svn: 288618
Rafael Espindola [Sun, 4 Dec 2016 08:34:17 +0000 (08:34 +0000)]
Don't discard .L symbol with -r.
They might be used by relocations.
Fixes pr31252.
llvm-svn: 288617
Rafael Espindola [Sun, 4 Dec 2016 07:27:02 +0000 (07:27 +0000)]
Always use / as the path separator.
It is not clear if it is worth the complexity to use \ on
windows. This should fix the bots.
llvm-svn: 288616
Rafael Espindola [Sun, 4 Dec 2016 06:52:30 +0000 (06:52 +0000)]
Prefix path when displaying thin archives.
Patch by Mark Santaniello.
llvm-svn: 288615
Matthias Braun [Sun, 4 Dec 2016 05:55:09 +0000 (05:55 +0000)]
TableGen: Adapt to llvm r288612
llvm-svn: 288614
Matthias Braun [Sun, 4 Dec 2016 05:48:20 +0000 (05:48 +0000)]
TableGen: Store Records on a BumpPtrAllocator
All these records are internalized and will live until exit. This makes
them perfect candidates for a fast BumpPtrAllocator.
llvm-svn: 288613
Matthias Braun [Sun, 4 Dec 2016 05:48:16 +0000 (05:48 +0000)]
TableGen: Use StringRef instead of const std::string& in return vals.
This will allow to switch to a different string storage in an upcoming
commit.
llvm-svn: 288612
Matthias Braun [Sun, 4 Dec 2016 05:48:06 +0000 (05:48 +0000)]
TableGen: Optimize common string concatenation with SmallString
llvm-svn: 288611
Matthias Braun [Sun, 4 Dec 2016 05:48:03 +0000 (05:48 +0000)]
TableGen: Use StringRef instead of const std::string& for parameters
This avoid an extra construction of a std::string (and a heap
allocation) when the caller only has a StringRef but no std::string at
hand.
llvm-svn: 288610
Rui Ueyama [Sun, 4 Dec 2016 02:34:29 +0000 (02:34 +0000)]
Update comment to clarify the machine spec.
llvm-svn: 288609
Lang Hames [Sun, 4 Dec 2016 01:56:10 +0000 (01:56 +0000)]
[Object][MachO] Reference-ify some helper function arguments. NFC.
Changes all static helper functions in MachOObjectFile.cpp that expect a
non-null MachOObjectFile pointer to take a reference instead.
llvm-svn: 288608
Dan Gohman [Sat, 3 Dec 2016 23:55:57 +0000 (23:55 +0000)]
[MC] Generalize MCContext's SectionSymbols field.
Change SectionSymbols so that it doesn't hard-code ELF types, so that
it can be used for non-ELF targets.
llvm-svn: 288607
Rui Ueyama [Sat, 3 Dec 2016 23:35:22 +0000 (23:35 +0000)]
Add comments about the use of threads in LLD.
llvm-svn: 288606
Sylvestre Ledru [Sat, 3 Dec 2016 23:22:45 +0000 (23:22 +0000)]
Add the --no-color option to the git call in the doc when using clang-format-diff
llvm-svn: 288605
Dan Gohman [Sat, 3 Dec 2016 23:03:52 +0000 (23:03 +0000)]
[WebAssembly] Revert r288447.
Revert r288447 which introduced -mdirect. It turns out we don't need a
custom flag for this, as the information we need is in the target triple.
llvm-svn: 288604
Matt Arsenault [Sat, 3 Dec 2016 23:03:26 +0000 (23:03 +0000)]
DAG: Fold out out of bounds insert_vector_elt
getNode already prevents formation of out of bounds constant
extract_vector_elts. Do the same for insert_vector_elt.
llvm-svn: 288603
Dan Gohman [Sat, 3 Dec 2016 23:00:12 +0000 (23:00 +0000)]
[WebAssembly] Eliminate an ad-hoc command-line argument.
Use the target triple to determine whether to run the explicit-locals
pass, rather than using a separate command-line argument.
llvm-svn: 288602
Saleem Abdulrasool [Sat, 3 Dec 2016 22:25:21 +0000 (22:25 +0000)]
AMDGPU: remove a couple of unused variables
lib/Target/AMDGPU/SIRegisterInfo.cpp: In member function 'void llvm::SIRegisterInfo::spillSGPR(llvm::MachineBasicBlock::iterator, int, llvm::RegScavenger*) const':
lib/Target/AMDGPU/SIRegisterInfo.cpp:572:30: warning: variable 'SubRC' set but not used [-Wunused-but-set-variable]
const TargetRegisterClass *SubRC = nullptr;
^
lib/Target/AMDGPU/SIRegisterInfo.cpp: In member function 'void llvm::SIRegisterInfo::restoreSGPR(llvm::MachineBasicBlock::iterator, int, llvm::RegScavenger*) const':
lib/Target/AMDGPU/SIRegisterInfo.cpp:723:30: warning: variable 'SubRC' set but not used [-Wunused-but-set-variable]
const TargetRegisterClass *SubRC = nullptr;
^
The variable was assigned to, but never used. The functions called did not
mutate state. Simplify the logic and remove the variable. Identified by gcc
5.4.0.
llvm-svn: 288601
Saleem Abdulrasool [Sat, 3 Dec 2016 22:03:24 +0000 (22:03 +0000)]
build: allow specifying the component to `llvm_install_symlink`
Add an optional parameter to `llvm_install_symlink` which allows the symlink
installation to be placed into a specific component rather than the default
value.
llvm-svn: 288600
Rui Ueyama [Sat, 3 Dec 2016 21:24:51 +0000 (21:24 +0000)]
Factor out common code to a header.
llvm-svn: 288599
Justin Lebar [Sat, 3 Dec 2016 19:49:35 +0000 (19:49 +0000)]
[PM] Rename lookupPass to lookUpPass.
Summary:
"Lookup" is a noun ("lookup table"), "look up" is a verb ("look up
'table' in the dictionary").
Reviewers: chandlerc
Subscribers: silvas, llvm-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D27374
llvm-svn: 288598
Justin Lebar [Sat, 3 Dec 2016 19:49:31 +0000 (19:49 +0000)]
[PM] Get rid of an unused variable in AnalysisManager::clear(IRUnitT&).
Reviewers: chandlerc
Subscribers: silvas, llvm-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D27373
llvm-svn: 288597
Justin Lebar [Sat, 3 Dec 2016 19:49:27 +0000 (19:49 +0000)]
[PM] Consistently use curly braces rather than std::make_pair in AnalysisResults.find().
Reviewers: chandlerc
Subscribers: silvas, llvm-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D27372
llvm-svn: 288596
Justin Lebar [Sat, 3 Dec 2016 19:49:23 +0000 (19:49 +0000)]
[PM] Don't walk the AM's ResultsList if nothing was invalidated.
Summary:
Previously in AnalysisManager::invalidate(), we would walk the full
ResultsList even if we knew that nothing was invalidated.
Reviewers: chandlerc
Subscribers: silvas, llvm-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D27371
llvm-svn: 288595
Justin Lebar [Sat, 3 Dec 2016 19:49:19 +0000 (19:49 +0000)]
[PM] Make AnalysisManager::registerPass take its parameter by universal reference.
Summary:
Previously, we were forcing a copy if you passed an lvalue argument; now
we'll take it by reference.
Reviewers: chandlerc
Subscribers: mehdi_amini, silvas, llvm-commits
Differential Revision: https://reviews.llvm.org/D27370
llvm-svn: 288594
Justin Lebar [Sat, 3 Dec 2016 19:49:15 +0000 (19:49 +0000)]
[PM] Make PassManager's constructor explicit.
Reviewers: chandlerc
Subscribers: silvas, llvm-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D27369
llvm-svn: 288593
Justin Lebar [Sat, 3 Dec 2016 19:49:12 +0000 (19:49 +0000)]
[PM] Make PreservedAnalyses::preserved take its parameter by const ref.
Summary: Previously we were unnecessarily copying the argument.
Reviewers: chandlerc
Subscribers: mehdi_amini, silvas, llvm-commits
Differential Revision: https://reviews.llvm.org/D27368
llvm-svn: 288592
Craig Topper [Sat, 3 Dec 2016 19:37:39 +0000 (19:37 +0000)]
[AVX-512] Add many of the VPERM instructions to the load folding table. Move VPERMPDZri to the correct table.
llvm-svn: 288591
Matt Arsenault [Sat, 3 Dec 2016 18:22:49 +0000 (18:22 +0000)]
AMDGPU: Clean up struct initializers
llvm-svn: 288590
Sanjay Patel [Sat, 3 Dec 2016 18:03:53 +0000 (18:03 +0000)]
[InstSimplify] add more helper functions for SimplifyICmpInst; NFCI
llvm-svn: 288589
Sanjay Patel [Sat, 3 Dec 2016 17:30:22 +0000 (17:30 +0000)]
[InstSimplify] add helper functions for SimplifyICmpInst; NFCI
llvm-svn: 288588
Craig Topper [Sat, 3 Dec 2016 17:19:15 +0000 (17:19 +0000)]
[AVX-512] Add EVEX VPMADDUBSW and VPMADDWD to the load folding tables.
llvm-svn: 288587
Eric Liu [Sat, 3 Dec 2016 15:28:03 +0000 (15:28 +0000)]
[clang-move] don't miss ',' in json output when there are duplicate elements.
llvm-svn: 288586
Rafael Espindola [Sat, 3 Dec 2016 15:26:18 +0000 (15:26 +0000)]
Ignone SHF_INFO_LINK.
Some elf producers (dtrace) put this flag in relocation sections and
some (MC) don't. If we don't ignore the flag we end up with multiple
relocation sections poiting to the same section, which we don't
support.
llvm-svn: 288585
Sanjay Patel [Sat, 3 Dec 2016 15:25:16 +0000 (15:25 +0000)]
[InstCombine] change select type to eliminate bitcasts
This solves a secondary problem seen in PR6137:
https://llvm.org/bugs/show_bug.cgi?id=6137#c6
This is similar to the bitwise logic op fold added with:
https://reviews.llvm.org/rL287707
And like that patch, I'm artificially restricting the
transform from vector <-> scalar types until we're sure
that the backend can handle that.
llvm-svn: 288584
Sergey Kalinichev [Sat, 3 Dec 2016 12:53:06 +0000 (12:53 +0000)]
[libclang] Fix python tests
It was broken in r286421
llvm-svn: 288582
George Rimar [Sat, 3 Dec 2016 07:30:30 +0000 (07:30 +0000)]
[ELF] - Implemented R_386_16 and R_386PC16 relocations
A program or object file using R_386_8, R_386_16, R_386_PC16 or R_386_PC8
relocations is not conformant to latest ABI. The R_386_16, and R_386_8
relocations truncate the computed value to 16 - bits and 8 - bits
respectively. R_386_PC16 and R_386_16 are used by some
applications, for example by FreeBSD loaders.
Previously we did not take addend in account for these relocation,
counting it as 0, what is wrong and was a reason of hangs.
This patch needed for example for FreeBSD pmbr (protective mbr).
Differential revision: https://reviews.llvm.org/D27303
llvm-svn: 288581
George Rimar [Sat, 3 Dec 2016 07:23:30 +0000 (07:23 +0000)]
[ELF] - Change the way how we compute offsets for binary output.
Binary output feature is a bit confuzing. bfd and gold output differs a lot sometimes,
though it is important for FreeBSD mbr loaders.
Patch change the way how we compute file offsets for binary output.
This fixes PR31196.
Previously offsets were calculated basing on offsets and addresses of sections
from the same loads:
if (Sec == First)
return alignTo(Off, Target->MaxPageSize, Sec->Addr);
return First->Offset + Sec->Addr - First->Addr;
bfd assigns offsets for each section to VA - MinVA:
https://github.com/redox-os/binutils-gdb/blob/master/bfd/binary.c#L27
https://github.com/redox-os/binutils-gdb/blob/master/bfd/binary.c#L255
(LMA == VA usually)
This patch for now just stops creating phdrs for binary output.
An effect from this that no any additional calculation for offset is performed:
uintX_t getFileAlignment(uintX_t Off, OutputSectionBase *Sec) {
OutputSectionBase *First = Sec->FirstInPtLoad;
// If the section is not in a PT_LOAD, we have no other constraint.
if (!First)
return Off; //**First is always null, condition always happens**
That is enough now with combination of another patch to generate output
that is similar to what bfd produce for mbr loader.
Differential revision: https://reviews.llvm.org/D27341
llvm-svn: 288580
George Rimar [Sat, 3 Dec 2016 07:09:28 +0000 (07:09 +0000)]
[ELF] - Disable relro when -omagic specified.
--omagic is an option to create old-fashioned executables in which
.text segments are writable. Today, the option is still in use to
create special-purpose programs such as boot loaders. It doesn't
make sense to create PT_GNU_RELRO for such executables.
DIfferential revision: https://reviews.llvm.org/D27297
llvm-svn: 288579
Craig Topper [Sat, 3 Dec 2016 05:35:44 +0000 (05:35 +0000)]
[X86] Fix VEX encoded VPMADDUBSW to not be marked commutable.
This was accidentallly broken in r285515 when we started lowering the intrinsic to an ISD node. Should fix PR31241.
llvm-svn: 288578
Craig Topper [Sat, 3 Dec 2016 05:35:38 +0000 (05:35 +0000)]
[X86] Add test cases demonstrating where we incorrectly commute VEX VPMADDUSBW due to a bug introduced in r285515.
I believe this is the cause of PR31241.
llvm-svn: 288577
Eric Fiselier [Sat, 3 Dec 2016 03:29:45 +0000 (03:29 +0000)]
Turn off testsuite warnings by default with GCC
llvm-svn: 288576
Eric Fiselier [Sat, 3 Dec 2016 03:22:11 +0000 (03:22 +0000)]
Make make_exception_ptr abort with -fno-exceptions
llvm-svn: 288575
Eric Fiselier [Sat, 3 Dec 2016 02:47:40 +0000 (02:47 +0000)]
Mark various <variant> items as complete
llvm-svn: 288574
Eric Fiselier [Sat, 3 Dec 2016 02:26:28 +0000 (02:26 +0000)]
Work around more -Wshadow warnings
llvm-svn: 288573
Michael Kuperstein [Sat, 3 Dec 2016 01:59:13 +0000 (01:59 +0000)]
Remove stale comment. NFC.
llvm-svn: 288572
Eric Fiselier [Sat, 3 Dec 2016 01:58:07 +0000 (01:58 +0000)]
Fix <variant> w/o exception support
llvm-svn: 288571
Saleem Abdulrasool [Sat, 3 Dec 2016 01:57:47 +0000 (01:57 +0000)]
Sema: delay the DLL exported member referencing
An explicit template specialization can cause the implicit template
specialization of a type which inherits the attributes. In such a case, we
would end up with a delayed template specialization for a dll exported type
which we would fail to reference. This would trigger an assertion.
We now propagate the dll storage attributes through the inheritance
chain. Only after having done so do we reference the delayed template
specializations. This allows any implicit specializations which inherit dll
storage to also be referenced.
llvm-svn: 288570