Hans Wennborg [Tue, 9 Jun 2015 00:39:03 +0000 (00:39 +0000)]
Enable propagation of dll attributes to previously instantiated base class templates in some cases
It is safe to add a dll attribute if the base class template previously only had
an explicit instantiation declaration, or was implicitly instantiated.
I both those cases, the members would not have been codegenned yet. In the case
of explicit instantiation declaration this is natural, and for implicit
instantiations, codegen is deferred (see r225570).
This is work towards fixing PR23770.
llvm-svn: 239373
Hans Wennborg [Tue, 9 Jun 2015 00:38:56 +0000 (00:38 +0000)]
Narrow the -Wunsupported-dll-base-class-template warning.
Don't warn about not being able to propagate dll attribute to a base class template
when that base already has a different attribute.
MSVC doesn't actually try to do this; the first attribute that was propagated
takes precedence, so Clang is already doing the right thing and there's no
need to warn.
(This is a step towards fixing PR21718.)
llvm-svn: 239372
Richard Smith [Tue, 9 Jun 2015 00:35:49 +0000 (00:35 +0000)]
[modules] Fix some visibility issues with default template arguments.
There are still problems here, but this is a better starting point.
The main part of the change is: when doing a lookup that would accept visible
or hidden declarations, prefer to produce the latest visible declaration if
there are any visible declarations, rather than always producing the latest
declaration.
Thus, when we inherit default arguments (and other properties) from a previous
declaration, we inherit them from the previous visible declaration; if the
previous declaration is hidden, we already suppress inheritance of default
arguments.
There are a couple of other changes here that fix latent bugs exposed by this
change.
llvm-svn: 239371
Matt Arsenault [Tue, 9 Jun 2015 00:31:39 +0000 (00:31 +0000)]
MC: Add target hook to control symbol quoting
llvm-svn: 239370
Arnold Schwaighofer [Tue, 9 Jun 2015 00:17:40 +0000 (00:17 +0000)]
Fix unused variable warning
llvm-svn: 239369
Jingyue Wu [Tue, 9 Jun 2015 00:05:56 +0000 (00:05 +0000)]
[NVPTX] run SROA after NVPTXFavorNonGenericAddrSpaces
Summary:
This cleans up most allocas NVPTXLowerKernelArgs emits for byval
parameters.
Test Plan: makes bug21465.ll more stronger to verify no redundant local load/store.
Reviewers: eliben, jholewinski
Reviewed By: eliben, jholewinski
Subscribers: jholewinski, llvm-commits
Differential Revision: http://reviews.llvm.org/D10322
llvm-svn: 239368
Arnold Schwaighofer [Tue, 9 Jun 2015 00:03:29 +0000 (00:03 +0000)]
MergeFunctions: Impose a total order on the replacement of functions
We don't want to replace function A by Function B in one module and Function B
by Function A in another module.
If these functions are marked with linkonce_odr we would end up with a function
stub calling B in one module and a function stub calling A in another module. If
the linker decides to pick these two we will have two stubs calling each other.
rdar://
21265586
llvm-svn: 239367
Pavel Labath [Mon, 8 Jun 2015 23:38:06 +0000 (23:38 +0000)]
Revert "Introduce a TypeSystem interface to support adding non-clang languages."
This seems to break expression evaluation on the linux build.
llvm-svn: 239366
Tyler Nowicki [Mon, 8 Jun 2015 23:27:35 +0000 (23:27 +0000)]
Eliminate unnecessary namespace to prevent conflicts.
llvm-svn: 239365
Anna Zaks [Mon, 8 Jun 2015 23:24:55 +0000 (23:24 +0000)]
[asan] Fixup to r239134. This test does pass on darwin.
This should restore the darwin buildbot.
llvm-svn: 239364
Tyler Nowicki [Mon, 8 Jun 2015 23:13:43 +0000 (23:13 +0000)]
Correct Loop Hint Diagnostic Message
When pragma clang loop unroll() is specified without an argument the diagnostic message should inform that user that 'full' and 'disable' are valid arguments (not 'enable').
llvm-svn: 239363
Tyler Nowicki [Mon, 8 Jun 2015 22:53:36 +0000 (22:53 +0000)]
Moved CPP CodeGen tests into CodeGenCXX.
llvm-svn: 239362
Reid Kleckner [Mon, 8 Jun 2015 22:43:32 +0000 (22:43 +0000)]
[WinEH] Cache declarations of frame intrinsics
llvm-svn: 239361
Pavel Labath [Mon, 8 Jun 2015 22:27:10 +0000 (22:27 +0000)]
Introduce a TypeSystem interface to support adding non-clang languages.
Reviewers: clayborg
Reviewed By: clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D8712
Original Author: Ryan Brown <ribrdb@google.com>
llvm-svn: 239360
Chaoren Lin [Mon, 8 Jun 2015 22:13:28 +0000 (22:13 +0000)]
Skip TestInferiorChanged if host platform is windows.
Summary: Opened files on Windows cannot be modified, so this test doesn't make sense.
Reviewers: ovyalov, zturner, flackr, clayborg
Reviewed By: clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D10295
llvm-svn: 239359
Chaoren Lin [Mon, 8 Jun 2015 22:12:58 +0000 (22:12 +0000)]
Fix TestSymbolContext for remote Windows to Android.
Summary: Denormalize path returned by SBFileSpec::GetDirectory().
Reviewers: zturner, clayborg
Reviewed By: clayborg
Subscribers: tberghammer, lldb-commits
Differential Revision: http://reviews.llvm.org/D10298
llvm-svn: 239358
Reid Kleckner [Mon, 8 Jun 2015 22:12:44 +0000 (22:12 +0000)]
[MC] Use unsigned for the Kind bitfield in MCSymbol
Fixes most of the test suite on Windows with clang-cl.
I'm not sure why the test suite was passing with MSVC 2013. Maybe they
changed their behavior and we are emulating their old sign extension
behavior. I think this deserves more investigation, but I want to green
the bot first.
llvm-svn: 239357
Hubert Tong [Mon, 8 Jun 2015 21:59:59 +0000 (21:59 +0000)]
Consider unsigned long for non-u/U decimal literals (C90/C++03)
Summary:
This modifies Clang to reflect that under pre-C99 ISO C, decimal
constants may have type `unsigned long` even if they do not contain `u`
or `U` in their suffix (C90 subclause 6.1.3.2 paragraph 5). The same is
done for C++ without C++11 which--because of undefined behaviour--allows
for behaviour compatible with ISO C90 in the case of an unsuffixed
decimal literal and is otherwise identical to C90 in its treatment of
integer literals (C++03 subclause 2.13.1 [lex.icon] paragraph 2).
Messages are added to the `c99-compat` and `c++11-compat` groups to warn
on such literals, since they behave differently under the newer
standards.
Fixes PR 16678.
Test Plan:
A new test file is added to exercise both pre-C99/C++11 and C99/C++11-up
on decimal literals with no suffix or suffixes `l`/`L` for both 32-bit
and 64-bit `long`.
In the file, 2^31 (being `INT_MAX+1`) is tested for the expected type
using `__typeof__` and multiple declarations of the same entity. 2^63
is similarly tested when it is within the range of `unsigned long`.
Preprocessor arithmetic tests are added to ensure consistency given
that Clang (like GCC) uses greater than 32 bits for preprocessor
arithmetic even when `long` and `unsigned long` is 32 bits and a
pre-C99/C++11 mode is in effect.
Tests added:
test/Sema/PR16678.c
Reviewers: fraggamuffin, rsmith
Reviewed By: rsmith
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D9794
llvm-svn: 239356
Reid Kleckner [Mon, 8 Jun 2015 21:57:57 +0000 (21:57 +0000)]
Fix clang-cl self-host -Wc++11-narrowing bug
Use unsigned as the underlying storage type of the AMDGPU address space
enum.
llvm-svn: 239355
Greg Clayton [Mon, 8 Jun 2015 21:53:11 +0000 (21:53 +0000)]
If we have a MachO file loaded from memory, make sure we can always get the symbol table even if we don't have the __LINKEDIT load address set in the target.
<rdar://problem/
21208168>
llvm-svn: 239354
Ranjeet Singh [Mon, 8 Jun 2015 21:32:16 +0000 (21:32 +0000)]
[AArch64] AsmParser should be case insensitive about accepting vector register names.
Differential Revision: http://reviews.llvm.org/D10320
llvm-svn: 239353
Evgeniy Stepanov [Mon, 8 Jun 2015 21:07:24 +0000 (21:07 +0000)]
[asan] Disable asan for aarch64.
lib/Target/AArch64/AArch64FrameLowering.cpp:370: virtual void llvm::AArch64FrameLowering::emitPrologue(llvm::MachineFunction&, llvm::MachineBasicBlock&) const: Assertion `!(NeedsRealignment && NumBytes==0) && "NumBytes should never be 0 when realignment is needed"' failed.
llvm-svn: 239352
Peter Collingbourne [Mon, 8 Jun 2015 20:21:28 +0000 (20:21 +0000)]
COFF: Skip internal symbols in bitcode files.
Differential Revision: http://reviews.llvm.org/D10319
llvm-svn: 239338
Jonathan Peyton [Mon, 8 Jun 2015 20:18:10 +0000 (20:18 +0000)]
Remove unused variables '__kmp_build_check_*'
As an ongoing effort to sanitize the openmp code, this one word change
eliminates creating 1 byte arrays named __kmp_build_check_* and instead
creates one byte array types. The KMP_BUILD_ASSERT macro still offers the same
functionality; array types with negative number of elements is illegal
and will cause a compiler failure.
llvm-svn: 239337
Keno Fischer [Mon, 8 Jun 2015 20:09:58 +0000 (20:09 +0000)]
[InstrInfo] Refactor foldOperandImpl to thread through InsertPt. NFC
Summary:
This was a longstanding FIXME and is a necessary precursor to cases
where foldOperandImpl may have to create more than one instruction
(e.g. to constrain a register class). This is the split out NFC changes from
D6262.
Reviewers: pete, ributzka, uweigand, mcrosier
Reviewed By: mcrosier
Subscribers: mcrosier, ted, llvm-commits
Differential Revision: http://reviews.llvm.org/D10174
llvm-svn: 239336
Rafael Espindola [Mon, 8 Jun 2015 20:08:55 +0000 (20:08 +0000)]
Fix a regression in .pop_section.
It was calling ChangeSection with the wrong current section, eventually leading
to a crash.
llvm-svn: 239335
Jonathan Peyton [Mon, 8 Jun 2015 20:01:14 +0000 (20:01 +0000)]
Remove unused variable warnings by deletion.
As an ongoing effort to sanitize the openmp code, these changes delete
variables that aren't used at all.
http://lists.cs.uiuc.edu/pipermail/openmp-dev/2015-June/000701.html
Patch by Jack Howarth
llvm-svn: 239334
Simon Pilgrim [Mon, 8 Jun 2015 19:58:43 +0000 (19:58 +0000)]
[X86][SSE] Added lzcnt vector tests.
llvm-svn: 239333
Rui Ueyama [Mon, 8 Jun 2015 19:43:59 +0000 (19:43 +0000)]
COFF: Read symbol names lazily.
This change seems to make the linker about 10% faster.
Reading symbol name is not very cheap because it needs strlen()
on the string table. We were wasting time on reading non-external
symbol names that would never be used by the linker.
llvm-svn: 239332
Jonathan Peyton [Mon, 8 Jun 2015 19:39:07 +0000 (19:39 +0000)]
Remove unused variable warnings by moving variables.
As an ongoing effort to sanitize the openmp code, these changes move
variables under already existing macro guards.
Patch by Jack Howarth
llvm-svn: 239331
Jonathan Peyton [Mon, 8 Jun 2015 19:25:25 +0000 (19:25 +0000)]
Remove unused variable warnings by adding proper macro guards.
As an ongoing effort to sanitize the openmp code, these changes remove unused variables
by adding proper macros around both variables and functions.
Patch by Jack Howarth
llvm-svn: 239330
Akira Hatanaka [Mon, 8 Jun 2015 19:03:32 +0000 (19:03 +0000)]
Include header file <functional>.
This is a follow-up to r239325.
llvm-svn: 239329
Benjamin Kramer [Mon, 8 Jun 2015 18:58:57 +0000 (18:58 +0000)]
Prefer copy init over direct init. NFC.
llvm-svn: 239327
Jonathan Peyton [Mon, 8 Jun 2015 18:56:33 +0000 (18:56 +0000)]
Remove unused variable warnings by fooling compiler.
Some variables are convenient to keep around even if they aren't
really used in a release build. This is often seen in DEBUG guarded code where the variable
is only used in a DEBUG build.
Patch by Jack Howarth
llvm-svn: 239326
Akira Hatanaka [Mon, 8 Jun 2015 18:50:43 +0000 (18:50 +0000)]
[ARM] Pass a callback to FunctionPass constructors to enable skipping execution
on a per-function basis.
Previously some of the passes were conditionally added to ARM's pass pipeline
based on the target machine's subtarget. This patch makes changes to add those
passes unconditionally and execute them conditonally based on the predicate
functor passed to the pass constructors. This enables running different sets of
passes for different functions in the module.
rdar://problem/
20542263
Differential Revision: http://reviews.llvm.org/D8717
llvm-svn: 239325
Pete Cooper [Mon, 8 Jun 2015 18:41:57 +0000 (18:41 +0000)]
Use a PointerUnion in MCSymbol for Section and Fragment. NFC.
The Fragment and Section, and a bool for HasFragment were all used to create
a PointerUnion. Just use a pointer union instead.
llvm-svn: 239324
Jonathan Peyton [Mon, 8 Jun 2015 18:38:23 +0000 (18:38 +0000)]
Removed unused functions.
As an ongoing effort to sanitize the openmp code, these changes remove unused functions.
The unused functions are:
__kmp_fini_allocator_thread(), __kmp_env_isDefined(), __kmp_strip_quotes(),
__kmp_convert_to_seconds(), and __kmp_convert_to_nanoseconds().
Patch by Jack Howarth
llvm-svn: 239323
Evgeniy Stepanov [Mon, 8 Jun 2015 18:19:26 +0000 (18:19 +0000)]
[asan] Enable AArch64 build of ASan runtime and tests.
llvm-svn: 239322
Evgeniy Stepanov [Mon, 8 Jun 2015 18:15:52 +0000 (18:15 +0000)]
[asan] Fix Android build for API >= 21.
/code/llvm/projects/compiler-rt/lib/sanitizer_common/sanitizer_linux.cc:971:8: error: address of function 'dl_iterate_phdr' will always evaluate to 'true' [-Werror,-Wpointer-bool-conversion]
if (!dl_iterate_phdr)
~^~~~~~~~~~~~~~~
/code/llvm/projects/compiler-rt/lib/sanitizer_common/sanitizer_linux.cc:971:8: note: prefix with the address-of operator to silence this warning
if (!dl_iterate_phdr)
^
&
llvm-svn: 239321
Evgeniy Stepanov [Mon, 8 Jun 2015 18:03:33 +0000 (18:03 +0000)]
[asan] Fix build of AsanNoinstTest on Android.
Some of the asan-ubsan build changes were not replicated in the android branch in CMakeLists.
llvm-svn: 239320
Pete Cooper [Mon, 8 Jun 2015 17:25:57 +0000 (17:25 +0000)]
Remove includes of MCMachOSymbolFlags.h after it was deleted
llvm-svn: 239318
Pete Cooper [Mon, 8 Jun 2015 17:17:30 +0000 (17:17 +0000)]
Make flags get/set method protected so that all flags have to be managed by subclasses.
All of ELF, COFF and MachO now manipulate the flags in helpers so we don't need
anyone to read the flags directly, but instead via those helpers.
Reviewed by Rafael Espíndola.
llvm-svn: 239317
Pete Cooper [Mon, 8 Jun 2015 17:17:28 +0000 (17:17 +0000)]
Move all flags logic to MCSymbolMachO.
Also delete the now unused MCMachOSymbolFlags.h header as the only enum in there was moved to MCSymbolMachO.
Similarly to ELF and COFF, manipulating the flags is now done via helpers instead of spread
throughout the codebase.
Reviewed by Rafael Espíndola.
llvm-svn: 239316
Pete Cooper [Mon, 8 Jun 2015 17:17:23 +0000 (17:17 +0000)]
Add MCSymbolMachO which will be used to hide the MCSymbolMachO flags.
Reviewed by Rafael Espíndola.
llvm-svn: 239315
Pete Cooper [Mon, 8 Jun 2015 17:17:19 +0000 (17:17 +0000)]
Move all of the MCSymbol COFF flags logic in to MCSymbolCOFF.
All flags setting/getting is now done in the class with helper methods instead
of users having to get the bits in the correct order.
Reviewed by Rafael Espíndola.
llvm-svn: 239314
Pete Cooper [Mon, 8 Jun 2015 17:17:16 +0000 (17:17 +0000)]
Move COFF Type in to the MCSymbolCOFF class.
The flags field in MCSymbol only needs to be 16-bits on ELF and MachO.
This moves the 16-bit Type out of there so that it can be reduced in size in a future commit.
Reviewed by Rafael Espíndola.
llvm-svn: 239313
Pete Cooper [Mon, 8 Jun 2015 17:17:12 +0000 (17:17 +0000)]
Add MCSymbolCOFF class and use it to get and set the COFF type field.
Reviewed by Rafael Espíndola.
llvm-svn: 239312
Pete Cooper [Mon, 8 Jun 2015 17:17:09 +0000 (17:17 +0000)]
Change MCSymbol IsELF to an enum to support future MCSymbolCOFF and MCSymbolMachO.
Reviewed by Rafael Espíndola.
llvm-svn: 239311
Oleksiy Vyalov [Mon, 8 Jun 2015 17:10:27 +0000 (17:10 +0000)]
Switch from setPrintImmHex to setPrintHexStyle to follow changes from r239263.
llvm-svn: 239310
Matthias Braun [Mon, 8 Jun 2015 16:56:23 +0000 (16:56 +0000)]
X86: Reject register operands with obvious type mismatches.
While we have some code to transform specification like {ax} into
{eax}/{rax} if the operand type isn't 16bit, we should reject cases
where there is no sane way to do this, like the i128 type in the
example.
Related to rdar://
21042280
Differential Revision: http://reviews.llvm.org/D10260
llvm-svn: 239309
Oliver Stannard [Mon, 8 Jun 2015 16:55:31 +0000 (16:55 +0000)]
Fix assertion failure in global-merge with unused ConstantExpr
The global-merge pass was crashing because it assumes that all ConstantExprs
(reached via the global variables that they use) have at least one user.
I haven't worked out a way to test this, as an unused ConstantExpr cannot be
represented by serialised IR, and global-merge can only be run in llc, which
does not run any passes which can make a ConstantExpr dead.
This (reduced to the point of silliness) C code triggers this bug when compiled
for arm-none-eabi at -O1:
static a = 7;
static volatile b[10] = {&a};
c;
main() {
c = 0;
for (; c < 10;)
printf(b[c]);
}
Differential Revision: http://reviews.llvm.org/D10314
llvm-svn: 239308
Colin LeMahieu [Mon, 8 Jun 2015 16:34:47 +0000 (16:34 +0000)]
[Hexagon] Adding functionality for searching for compound instruction pairs. Compound instructions reduce slot resource requirements freeing those packet slots up for more instructions.
llvm-svn: 239307
Tobias Grosser [Mon, 8 Jun 2015 16:24:49 +0000 (16:24 +0000)]
Revert "Add NVIDIA vprintf printing to RuntimeDebugBuilder"
This reverts commit 239219 which requires some LLVM changes I forgot to commit.
Reported-by: Marshall Clow
llvm-svn: 239306
Simon Pilgrim [Mon, 8 Jun 2015 16:19:00 +0000 (16:19 +0000)]
[DAGCombiner] Added CTLZ vector constant folding support.
llvm-svn: 239305
Sanjay Patel [Mon, 8 Jun 2015 15:21:38 +0000 (15:21 +0000)]
fix typos; NFC
llvm-svn: 239303
Javed Absar [Mon, 8 Jun 2015 15:01:11 +0000 (15:01 +0000)]
ARM]: Add support for MMFR4_EL1 in assembler
This patch adds support for system register MMFR4_EL1 (memory model feature register) in the assembler.
This register provides information about the implemented memory model and memory management support.
llvm-svn: 239302
Petar Jovanovic [Mon, 8 Jun 2015 14:10:23 +0000 (14:10 +0000)]
[Mips64][mcjit] Add R_MIPS_PC32 relocation
This patch adds R_MIPS_PC32 relocation for Mips64.
Patch by Vladimir Radosavljevic.
Differential Revision: http://reviews.llvm.org/D10235
llvm-svn: 239301
Igor Breger [Mon, 8 Jun 2015 14:03:17 +0000 (14:03 +0000)]
AVX-512: Implemented 256/128bit VALIGND/Q instructions for SKX and KNL
Implemented DAG lowering for all these forms.
Added tests for DAG lowering and encoding.
Differential Revision: http://reviews.llvm.org/D10310
llvm-svn: 239300
Artur Pilipenko [Mon, 8 Jun 2015 11:58:13 +0000 (11:58 +0000)]
Minor refactoring of GEP handling in isDereferenceablePointer
For GEP instructions isDereferenceablePointer checks that all indices are constant and within bounds. Replace this index calculation logic to a call to accumulateConstantOffset. Separated from the http://reviews.llvm.org/D9791
Reviewed By: sanjoy
Differential Revision: http://reviews.llvm.org/D9874
llvm-svn: 239299
Leny Kholodov [Mon, 8 Jun 2015 11:39:16 +0000 (11:39 +0000)]
Update test with target flag and amended results
llvm-svn: 239298
Bruce Mitchener [Mon, 8 Jun 2015 11:15:09 +0000 (11:15 +0000)]
[LLDB-MI] Properly detect missing mandatory arguments to MI commands
Summary:
Previously if an MI command had **X** mandatory and **Y** optional arguments you could provide **X** or more optional arguments without providing any of the mandatory arguments, and the argument validation code wouldn't complain.
For example this would pass argument validation even though the mandatory **address** and **count** arguments are missing:
-data-read-memory-bytes --thread 1 --frame 0
Part of the problem was that an empty string was considered a valid value for a mandatory argument, which didn't make much sense.
Patch by Vadim Macagon. Thanks!
Test Plan:
./dotest.py -A x86_64 -C clang --executable $BUILDDIR/bin/lldb tools/lldb-mi/
No unexpected failures on my Ubuntu 14.10 64bit Virtualbox VM.
Reviewers: domipheus, ki.stfu, abidh
Reviewed By: ki.stfu, abidh
Subscribers: brucem, lldb-commits
Differential Revision: http://reviews.llvm.org/D10299
llvm-svn: 239297
Leny Kholodov [Mon, 8 Jun 2015 11:06:59 +0000 (11:06 +0000)]
Fix for temporary variable names in stack reuse tests in revision 239294
llvm-svn: 239296
Silviu Baranga [Mon, 8 Jun 2015 10:27:06 +0000 (10:27 +0000)]
[LAA] Fix estimation of number of memchecks
Summary:
We need to add a runtime memcheck for pair of accesses (x,y) where at least one of x and y
are writes.
Assuming we have w writes and r reads, currently this number is estimated as being
w* (w+r-1). This estimation will count (write,write) pairs twice and will overestimate
the number of checks required.
This change adds a getNumberOfChecks method to RuntimePointerCheck, which
will count the number of runtime checks needed (similar in implementation to
needsAnyChecking) and uses it to produce the correct number of runtime checks.
Test Plan:
llvm test suite
spec2k
spec2k6
Performance results: no changes observed (not surprising since the formula for 1 writer is basically the same, which would covers most cases - at least with the current check limit).
Reviewers: anemet
Reviewed By: anemet
Subscribers: mzolotukhin, llvm-commits
Differential Revision: http://reviews.llvm.org/D10217
llvm-svn: 239295
Leny Kholodov [Mon, 8 Jun 2015 10:23:49 +0000 (10:23 +0000)]
[CodeGen] Reuse stack space from unused function results (with more accurate unused result detection)
This patch fixes issues with unused result detection which were found in patch http://reviews.llvm.org/D9743.
Differential Revision: http://reviews.llvm.org/D10042
llvm-svn: 239294
Simon Pilgrim [Mon, 8 Jun 2015 09:57:09 +0000 (09:57 +0000)]
[DAGCombiner] Added CTTZ vector constant folding support.
llvm-svn: 239293
Rui Ueyama [Mon, 8 Jun 2015 08:26:28 +0000 (08:26 +0000)]
COFF: Avoid callign stable_sort.
MSVC profiler reported that this stable_sort takes 7% time
when self-linking. As a result, createSection was taking 10% time.
Now createSection takes 3%. This small change actually makes
the linker a bit but perceptibly faster.
llvm-svn: 239292
Hao Liu [Mon, 8 Jun 2015 06:39:56 +0000 (06:39 +0000)]
[LoopVectorize] Teach Loop Vectorizor about interleaved memory accesses.
Interleaved memory accesses are grouped and vectorized into vector load/store and shufflevector.
E.g. for (i = 0; i < N; i+=2) {
a = A[i]; // load of even element
b = A[i+1]; // load of odd element
... // operations on a, b, c, d
A[i] = c; // store of even element
A[i+1] = d; // store of odd element
}
The loads of even and odd elements are identified as an interleave load group, which will be transfered into vectorized IRs like:
%wide.vec = load <8 x i32>, <8 x i32>* %ptr
%vec.even = shufflevector <8 x i32> %wide.vec, <8 x i32> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
%vec.odd = shufflevector <8 x i32> %wide.vec, <8 x i32> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
The stores of even and odd elements are identified as an interleave store group, which will be transfered into vectorized IRs like:
%interleaved.vec = shufflevector <4 x i32> %vec.even, %vec.odd, <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
store <8 x i32> %interleaved.vec, <8 x i32>* %ptr
This optimization is currently disabled by defaut. To try it by adding '-enable-interleaved-mem-accesses=true'.
llvm-svn: 239291
Rui Ueyama [Mon, 8 Jun 2015 06:13:12 +0000 (06:13 +0000)]
COFF: Use the empty string as the current directory instead of ".".
This is NFC but makes log message a bit nicer because it doesn't
append .\ (or ./ on Unix) to files in the current directory.
llvm-svn: 239290
Rui Ueyama [Mon, 8 Jun 2015 06:00:10 +0000 (06:00 +0000)]
COFF: Add more log messages.
llvm-svn: 239289
Rui Ueyama [Mon, 8 Jun 2015 05:43:50 +0000 (05:43 +0000)]
COFF: Print out log messages to stdout.
llvm-svn: 239288
Rui Ueyama [Mon, 8 Jun 2015 05:00:42 +0000 (05:00 +0000)]
COFF: Check for auxiliary symbol's type.
We forgot to check for auxiliary symbol's type. So we sometimes read
garbage as associative section definitions.
Associative sections are considered as not live themselves by the
garbage collector because they are live only when associaited sections
are live.
By reading more data (or garbage) as associative section definitions,
we treated more sections as non-GC-roots, that caused the linker to
discard too many sections by mistake. That caused another mysterious
bug (such as some global constructors don't run at all for some reason.)
llvm-svn: 239287
Davide Italiano [Mon, 8 Jun 2015 04:55:29 +0000 (04:55 +0000)]
[llvm-readobj] Use the new symbol API.
llvm-svn: 239286
Hao Liu [Mon, 8 Jun 2015 04:48:37 +0000 (04:48 +0000)]
[LoopAccessAnalysis] Teach LAA to check the memory dependence between strided accesses.
Differential Revision: http://reviews.llvm.org/D9368
llvm-svn: 239285
Chaoren Lin [Mon, 8 Jun 2015 04:39:24 +0000 (04:39 +0000)]
Silence warning on Windows.
llvm-svn: 239284
Peter Collingbourne [Mon, 8 Jun 2015 04:04:13 +0000 (04:04 +0000)]
Do not build LLD from tools/Makefile, as LLD no longer has a Makefile build.
llvm-svn: 239283
Michael Zolotukhin [Mon, 8 Jun 2015 03:28:06 +0000 (03:28 +0000)]
Remove SCEVCache and FindConstantPointers from complete loop unrolling heuristic.
Summary:
Using some SCEV functionality helped to entirely remove SCEVCache class and FindConstantPointers SCEV visitor.
Also, this makes the code more universal - I'll take advandate of it in next patches where I start handling additional types of instructions.
Test Plan: Tests would be submitted in subsequent patches.
Reviewers: atrick, chandlerc
Reviewed By: atrick, chandlerc
Subscribers: atrick, llvm-commits
Differential Revision: http://reviews.llvm.org/D10205
llvm-svn: 239282
Rui Ueyama [Mon, 8 Jun 2015 03:27:57 +0000 (03:27 +0000)]
COFF: Simplify InputFile class.
Now that all InputFile subclasses have MemoryBufferRefs and
provides the same set of functions. Implement that in the base class.
llvm-svn: 239281
Rui Ueyama [Mon, 8 Jun 2015 03:17:07 +0000 (03:17 +0000)]
COFF: Set non-1 alignment to common chunks.
I don't know what the right thing to do here, but at least 1 does
not seem like a correct value. If we do not align common chunks at
all, a small program which calls puts() from global dtors crashes
mysteriously in a kernel32's function.
I believe the crash was caused by symbols overlapping each other,
and my guess is that alignment has something to do with that, but
I am not 100% sure. Needs investigating.
llvm-svn: 239280
Peter Collingbourne [Mon, 8 Jun 2015 02:43:32 +0000 (02:43 +0000)]
Fix Windows build.
llvm-svn: 239279
Peter Collingbourne [Mon, 8 Jun 2015 02:32:01 +0000 (02:32 +0000)]
llvm-ar: Move archive writer to Object.
No functional change intended, other than some minor changes to certain
diagnostics.
Differential Revision: http://reviews.llvm.org/D10296
llvm-svn: 239278
Craig Topper [Mon, 8 Jun 2015 01:35:45 +0000 (01:35 +0000)]
[TableGen] Remove unnecessary default constructed SMLoc that just existed to return. Instead just call the default constructor in the return.
llvm-svn: 239277
Craig Topper [Mon, 8 Jun 2015 01:35:43 +0000 (01:35 +0000)]
[TableGen] Use range-based for loop. NFC
llvm-svn: 239276
Craig Topper [Mon, 8 Jun 2015 01:35:42 +0000 (01:35 +0000)]
[TableGen] Remove unnecessary extern declaration that's already in an included header file.
llvm-svn: 239275
Craig Topper [Mon, 8 Jun 2015 01:35:40 +0000 (01:35 +0000)]
[TableGen] Use the SMLoc header file instead of SourceMgr header file in a couple places. NFC
llvm-svn: 239274
David Majnemer [Mon, 8 Jun 2015 00:22:46 +0000 (00:22 +0000)]
[Driver] Inject the MSVC compatibility version into the triple
Encoding the version into the triple will allow us to communicate to
LLVM what functions it can expect to depend upon in the implementation.
llvm-svn: 239273
Rui Ueyama [Mon, 8 Jun 2015 00:09:25 +0000 (00:09 +0000)]
COFF: Fix memory leak.
llvm-svn: 239272
Rui Ueyama [Sun, 7 Jun 2015 23:10:50 +0000 (23:10 +0000)]
COFF: Fix typo.
This change doesn't change its functionality since the value
passed here is converted to uint16_t immediately.
llvm-svn: 239271
Rui Ueyama [Sun, 7 Jun 2015 23:02:50 +0000 (23:02 +0000)]
COFF: Simplify. NFC.
llvm-svn: 239270
Rui Ueyama [Sun, 7 Jun 2015 23:00:29 +0000 (23:00 +0000)]
COFF: Use llvm::cl::ExpandReponseFiles.
llvm-svn: 239269
Rui Ueyama [Sun, 7 Jun 2015 22:42:52 +0000 (22:42 +0000)]
COFF: Add a glossary to README.
llvm-svn: 239268
Rui Ueyama [Sun, 7 Jun 2015 22:00:28 +0000 (22:00 +0000)]
COFF: Use named constants instead of sizeof().
llvm-svn: 239267
Rui Ueyama [Sun, 7 Jun 2015 21:58:34 +0000 (21:58 +0000)]
COFF: Remove dead code.
/include'ed symbols are already added to Config->GCRoots.
Marking symbols twice doesn't have any effect.
llvm-svn: 239266
Colin LeMahieu [Sun, 7 Jun 2015 21:07:17 +0000 (21:07 +0000)]
[objdump] Moving PrintImmHex out of MachODump and in to llvm-objdump and setting instprinter appropriately.
llvm-svn: 239265
Simon Pilgrim [Sun, 7 Jun 2015 21:01:34 +0000 (21:01 +0000)]
[X86] Added tzcnt vector tests.
llvm-svn: 239264
Colin LeMahieu [Sun, 7 Jun 2015 20:29:37 +0000 (20:29 +0000)]
[MC] Function naming NFC.
llvm-svn: 239263
Matt Arsenault [Sun, 7 Jun 2015 20:17:44 +0000 (20:17 +0000)]
SeparateConstOffsetFromGEP: Pass address space to isLegalAddressingMode
llvm-svn: 239262
Matt Arsenault [Sun, 7 Jun 2015 20:17:42 +0000 (20:17 +0000)]
Make NaryReassociate pass the address space to isLegalAddressingMode
No test since the kinds of transforms this prevents seem to not really
be relevant for SI's different addressing modes.
llvm-svn: 239261
Matt Arsenault [Sun, 7 Jun 2015 20:12:03 +0000 (20:12 +0000)]
Add isLegalAddressingMode address space argument to TTI
Update to match the TLI version, and remove the TLI version's
default argument.
llvm-svn: 239260
Peter Zotov [Sun, 7 Jun 2015 19:22:22 +0000 (19:22 +0000)]
[cmake] [OCaml] Make ocamldoc targets depend on output files
OCaml doc builds fail without .cmi files, and .cmi files are collected
in ocaml_outputs. Therefore, make doc targets depend on ocaml_outputs as
well.
Fixes: https://llvm.org/bugs/show_bug.cgi?id=23777
Patch by Michał Górny <mgorny@gentoo.org>
llvm-svn: 239259
Chaoren Lin [Sun, 7 Jun 2015 18:50:40 +0000 (18:50 +0000)]
Update dosep to display dotest command invoked on failure.
llvm-svn: 239258