Simon Pilgrim [Sun, 8 Feb 2015 18:33:13 +0000 (18:33 +0000)]
[X86][AVX2] AVX2 broadcast + permute memory folding tests.
llvm-svn: 228528
Zachary Turner [Sun, 8 Feb 2015 18:08:51 +0000 (18:08 +0000)]
Make UTF8->UTF16 conversion null terminate output on empty input.
llvm-svn: 228527
Simon Pilgrim [Sun, 8 Feb 2015 17:13:54 +0000 (17:13 +0000)]
Moved AVX2 vbroadcast (reg) instruction foldings under the correct grouping. NFC.
llvm-svn: 228526
Bjorn Steinbrink [Sun, 8 Feb 2015 17:07:14 +0000 (17:07 +0000)]
Correctly combine alias.scope metadata by a union instead of intersecting
Summary:
The alias.scope metadata represents sets of things an instruction might
alias with. When generically combining the metadata from two
instructions the result must be the union of the original sets, because
the new instruction might alias with anything any of the original
instructions aliased with.
Reviewers: hfinkel
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D7490
llvm-svn: 228525
NAKAMURA Takumi [Sun, 8 Feb 2015 11:15:08 +0000 (11:15 +0000)]
[CMake] Kaleidoscope-Orc: Update libdeps.
llvm-svn: 228524
David Majnemer [Sun, 8 Feb 2015 10:55:14 +0000 (10:55 +0000)]
Sema: Don't give attribute alias vars with struct type an init expr
We'd give the VarDecl a CXXConstructExpr even though it is annotated
with an alias attribute. This would make us trip over sanity checking
asserts.
This fixes PR22493.
llvm-svn: 228523
Daniel Jasper [Sun, 8 Feb 2015 09:34:49 +0000 (09:34 +0000)]
clang-format: Correctly mark preprocessor lines in child blocks.
This prevents contracting:
auto lambda = []() {
int a = 2
#if A
+ 2
#endif
;
};
into:
auto lambda = []() { int a = 2
#if A + 2
#endif ; };
Which is obviously BAD.
This fixes llvm.org/PR22496.
llvm-svn: 228522
Elena Demikhovsky [Sun, 8 Feb 2015 08:27:19 +0000 (08:27 +0000)]
Masked Gather and Scatter Intrinsics.
Gather and Scatter are new introduced intrinsics, comming after recently implemented masked load and store.
This is the first patch for Gather and Scatter intrinsics. It includes only the syntax, parsing and verification.
Gather and Scatter intrinsics allow to perform multiple memory accesses (read/write) in one vector instruction.
The intrinsics are not target specific and will have the following syntax:
Gather:
declare <16 x i32> @llvm.masked.gather.v16i32(<16 x i32*> <vector of ptrs>, i32 <alignment>, <16 x i1> <mask>, <16 x i32> <passthru>)
declare <8 x float> @llvm.masked.gather.v8f32(<8 x float*><vector of ptrs>, i32 <alignment>, <8 x i1> <mask>, <8 x float><passthru>)
Scatter:
declare void @llvm.masked.scatter.v8i32(<8 x i32><vector value to be stored> , <8 x i32*><vector of ptrs> , i32 <alignment>, <8 x i1> <mask>)
declare void @llvm.masked.scatter.v16i32(<16 x i32> <vector value to be stored> , <16 x i32*> <vector of ptrs>, i32 <alignment>, <16 x i1><mask> )
Vector of ptrs - a set of source/destination addresses, to load/store the value.
Mask - switches on/off vector lanes to prevent memory access for switched-off lanes
vector of ptrs, value and mask should have the same vector width.
These are code examples where gather / scatter should be used and will allow function vectorization
;void foo1(int * restrict A, int * restrict B, int * restrict C) {
; for (int i=0; i<SIZE; i++) {
; A[i] = B[C[i]];
; }
;}
;void foo3(int * restrict A, int * restrict B) {
; for (int i=0; i<SIZE; i++) {
; A[B[i]] = i+5;
; }
;}
Tests will come in the following patches, with CodeGen and Vectorizer.
http://reviews.llvm.org/D7433
llvm-svn: 228521
David Blaikie [Sun, 8 Feb 2015 07:20:04 +0000 (07:20 +0000)]
Add CMake build support for Orc examples (& fix some clang -Werror build
breaks due to unused variables).
llvm-svn: 228520
Lang Hames [Sun, 8 Feb 2015 04:34:13 +0000 (04:34 +0000)]
[Orc][Kaleidoscope] Fix method-name think-o. NFC.
llvm-svn: 228519
Tim Northover [Sun, 8 Feb 2015 00:50:47 +0000 (00:50 +0000)]
ARM & AArch64: teach LowerVSETCC that output type size may differ from input.
While various DAG combines try to guarantee that a vector SETCC
operation will have the same output size as input, there's nothing
intrinsic to either creation or LegalizeTypes that actually guarantees
it, so the function needs to be ready to handle a mismatch.
Fortunately this is easy enough, just extend or truncate the naturally
compared result.
I couldn't reproduce the failure in other backends that I know have
SIMD, so it's probably only an issue for these two due to shared
heritage.
Should fix PR21645.
llvm-svn: 228518
Zachary Turner [Sun, 8 Feb 2015 00:41:31 +0000 (00:41 +0000)]
Removed unused function mistakenly left in, triggering -Werror.
llvm-svn: 228517
Zachary Turner [Sun, 8 Feb 2015 00:29:29 +0000 (00:29 +0000)]
Some cleanup for libpdb.
This patch implements a few of the optional suggestions from the
initial patch comitting libpdb. In particular, it implements a
virtual function out of line for each of the concrete classes.
A few other minor cleanups exist as well, such as using override
instead of virtual, etc.
llvm-svn: 228516
Craig Topper [Sat, 7 Feb 2015 23:36:51 +0000 (23:36 +0000)]
[X86] Add register use/def for wrmsr and rdmsr.
llvm-svn: 228515
Craig Topper [Sat, 7 Feb 2015 23:36:36 +0000 (23:36 +0000)]
[X86] Add GETSEC instruction.
llvm-svn: 228514
Simon Pilgrim [Sat, 7 Feb 2015 23:28:16 +0000 (23:28 +0000)]
[X86][AVX2] AVX2 integer stack folding tests.
This adds tests for the remaining AVX2 instructions that currently support memory folding.
llvm-svn: 228513
David Blaikie [Sat, 7 Feb 2015 23:23:43 +0000 (23:23 +0000)]
Rename the 'Extending the Language: Debug Information' to 'Adding Debug Information' since this isn't actually modifying/extending the language.
llvm-svn: 228512
Yaron Keren [Sat, 7 Feb 2015 22:16:16 +0000 (22:16 +0000)]
Replace DeclContext::getNextContext with DeclContext::collectAllContexts
in the Clang CFE Internals Manual (done in r147729).
llvm-svn: 228510
Simon Pilgrim [Sat, 7 Feb 2015 21:44:06 +0000 (21:44 +0000)]
[X86][AVX] Added missing stack folding support + test for vptest ymm instruction
llvm-svn: 228509
Benjamin Kramer [Sat, 7 Feb 2015 21:37:08 +0000 (21:37 +0000)]
LoopIdiom: Use utility functions.
The only difference between deleteIfDeadInstruction and
RecursivelyDeleteTriviallyDeadInstructions is that the former also
manually invalidates SCEV. That's unnecessary because SCEV automatically
gets informed when an instruction is deleted via a ValueHandle. NFC.
llvm-svn: 228508
Joerg Sonnenberger [Sat, 7 Feb 2015 21:24:06 +0000 (21:24 +0000)]
Avoid integer overflows around realloc calls resulting in potential
heap. Problem identified by Guido Vranken. Changes differ from original
OpenBSD sources by not depending on non-portable reallocarray.
llvm-svn: 228507
Simon Pilgrim [Sat, 7 Feb 2015 21:20:11 +0000 (21:20 +0000)]
[X86][SSE] Added missing stack folding tests for (v)mpsadbw instruction
llvm-svn: 228506
Gabor Horvath [Sat, 7 Feb 2015 19:54:19 +0000 (19:54 +0000)]
[clang-tidy] Checker for inefficient use of algorithms on associative containers
Summary:
Associative containers implements some of the algorithms as methods which
should be preferred to the algorithms in the algorithm header. The methods
can take advantage of the order of the elements.
Reviewers: alexfh
Reviewed By: alexfh
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D7246
llvm-svn: 228505
Benjamin Kramer [Sat, 7 Feb 2015 19:29:02 +0000 (19:29 +0000)]
ValueTracking: Make isBytewiseValue simpler and more powerful at the same time.
Turns out there is a simpler way of checking that all bytes in a word are equal
than binary decomposition.
llvm-svn: 228503
Bjorn Steinbrink [Sat, 7 Feb 2015 17:54:36 +0000 (17:54 +0000)]
Properly update AA metadata when performing call slot optimization
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D7482
llvm-svn: 228500
Jonathan Roelofs [Sat, 7 Feb 2015 17:18:26 +0000 (17:18 +0000)]
Fix docs typo regarding lit.local.cfg files
llvm-svn: 228499
Ahmed Bougacha [Sat, 7 Feb 2015 17:04:29 +0000 (17:04 +0000)]
[BasicAA] Try to disambiguate GEPs through arrays of structs into
different fields.
We can show that two GEPs off of the same (possibly multidimensional)
array of structs, into different fields, can't alias. Quoting:
For two GEPOperators GEP1 and GEP2, if we find that:
- both GEPs begin indexing from the exact same pointer;
- the last indices in both GEPs are constants, indexing into a struct;
- said indices are different, hence,the pointed-to fields are different;
- and both GEPs only index through arrays prior to that;
this lets us determine that the struct that GEP1 indexes into and the
struct that GEP2 indexes into must either precisely overlap or be
completely disjoint. Because they cannot partially overlap, indexing
into different non-overlapping fields of the struct will never alias.
The other BasicAA::aliasGEP rules worked in some cases, but not all
(for example, the i32x3 struct in the testcase).
We can add this simple ad-hoc rule to complement them.
rdar://
19717375
Differential Revision: http://reviews.llvm.org/D7453
llvm-svn: 228498
Benjamin Kramer [Sat, 7 Feb 2015 16:41:12 +0000 (16:41 +0000)]
SCEV: Compress disposition pairs.
Composing DenseMaps and SmallVectors is still somewhat suboptimal,
but this at least halves the size of the vector elements. NFC.
llvm-svn: 228497
Benjamin Kramer [Sat, 7 Feb 2015 16:41:02 +0000 (16:41 +0000)]
SmallVector: Move emplace_back to SmallVectorImpl.
This resolves the strange effect that emplace_back is only available
when the type contained in the vector is not trivially copyable.
llvm-svn: 228496
Simon Pilgrim [Sat, 7 Feb 2015 16:14:55 +0000 (16:14 +0000)]
[X86] Force fp stack folding tests to keep to specific domain.
General boolean instructions (AND, ANDN, OR, XOR) need to use a specific domain instruction (and not just the default).
llvm-svn: 228495
Simon Pilgrim [Sat, 7 Feb 2015 16:07:27 +0000 (16:07 +0000)]
[X86][AVX2] More AVX2 integer stack folding tests.
llvm-svn: 228494
Andrea Di Biagio [Sat, 7 Feb 2015 13:56:20 +0000 (13:56 +0000)]
Fix typos; NFC.
llvm-svn: 228493
Benjamin Kramer [Sat, 7 Feb 2015 13:15:54 +0000 (13:15 +0000)]
CodeGen: Move DebugLocs.
It's slightly cheaper than copying it, if the DebugLoc points to replaceable
metadata every copy is recorded in a DenseMap, moving reduces the peak size of
that map.
llvm-svn: 228492
Benjamin Kramer [Sat, 7 Feb 2015 12:28:15 +0000 (12:28 +0000)]
Move DebugLocs around instead of copying.
llvm-svn: 228491
David Majnemer [Sat, 7 Feb 2015 08:26:40 +0000 (08:26 +0000)]
MC: Emit COFF section flags in the "proper" order
COFF section flags are not idempotent:
'rd' will make a read-write section because 'd' implies write
'dr' will make a read-only section because 'r' disables write
llvm-svn: 228490
Chandler Carruth [Sat, 7 Feb 2015 08:12:22 +0000 (08:12 +0000)]
Consistently use override rather than virtual.
This fixes -Winconsistent-missing-override warnings.
llvm-svn: 228489
Hal Finkel [Sat, 7 Feb 2015 07:32:58 +0000 (07:32 +0000)]
[PowerPC] Handle loop predecessor invokes
If a loop predecessor has an invoke as its terminator, and the return value
from that invoke is used to determine the loop iteration space, then we can't
insert a computation based on that value in the loop predecessor prior to the
terminator (oops). If there's such an invoke, or just no predecessor for that
matter, insert a new loop preheader.
llvm-svn: 228488
Bruce Mitchener [Sat, 7 Feb 2015 06:35:30 +0000 (06:35 +0000)]
Add more DWARF 5 language constants.
Differential Revision: http://reviews.llvm.org/D7430
llvm-svn: 228487
Jason Molenda [Sat, 7 Feb 2015 06:03:49 +0000 (06:03 +0000)]
When creating a disassembler for one of the arm variants that can
only execute thumb instructions, force the arch triple string to
be "thumbv..." instead of "armv..." so we do the right thing by
default when disassembling arbitrary chunks of code.
<rdar://problem/
15126397>
llvm-svn: 228486
Richard Smith [Sat, 7 Feb 2015 03:11:11 +0000 (03:11 +0000)]
[modules] Treat friend declarations that are lexically within a dependent
context as anonymous for merging purposes. They can't be found by their names,
so we merge them based on their position within the surrounding context.
llvm-svn: 228485
Zachary Turner [Sat, 7 Feb 2015 02:02:23 +0000 (02:02 +0000)]
Change RHS-style decltype to LHS-style decltype<declval()>.
Seems some compilers don't like the RHS-style decltype specifier.
This should fix the buildbots.
llvm-svn: 228484
Nico Weber [Sat, 7 Feb 2015 01:57:32 +0000 (01:57 +0000)]
clang-format: Format Objective-C try blocks like all the other try blocks.
Before:
@try {
// ...
}
@finally {
// ...
}
Now:
@try {
// ...
} @finally {
// ...
}
This is consistent with how we format C++ try blocks and SEH try blocks.
clang-format not doing this before was an implementation oversight.
This is dependent on BraceBreakingStyle. The snippet above is with the
Attach style. Style Stroustrip for example still results in the "Before:"
snippet, which makes sense since other blocks (try, else) break after '}' too.
llvm-svn: 228483
Zachary Turner [Sat, 7 Feb 2015 01:47:14 +0000 (01:47 +0000)]
Resubmit unittests for DebugInfoPDB.
These were originally submitted as part of r228428, but this part
caused a build breakage in LLVMConfig. The library portion was
resubmitted independently since it was not causing breakage.
There were two reasons this was causing the build to fail. The
first is that there were no Makefiles added for the PDB tests. And
the second is that the DebugInfoPDB library was only being built by
CMake behind an "if (MSVC)" check. This is wrong since this the
library hides platform specific details, and it was causing
LLVM-Config to not find the library when trying to build unittests.
llvm-svn: 228482
Filipe Cabecinhas [Sat, 7 Feb 2015 01:37:09 +0000 (01:37 +0000)]
Make the byte-shift SSE intrinsics emit vector shuffles which we know the backend can handle.
Also removed unused builtins.
Original patch by Andrea Di Biagio!
Reviewers: craig.topper, nadav
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D7199
llvm-svn: 228481
Sean Silva [Sat, 7 Feb 2015 01:16:26 +0000 (01:16 +0000)]
[docs][LLVM-style RTTI] Add a mention of multiple inheritance.
llvm-svn: 228479
Duncan P. N. Exon Smith [Sat, 7 Feb 2015 01:07:30 +0000 (01:07 +0000)]
Support: Fix tests for VirtualityString
Since these `dwarf` functions return `const char *`, the tests need to
use `StringRef` for checks. Should fix, e.g., hexagon [1].
[1]: http://lab.llvm.org:8011/builders/clang-hexagon-elf/builds/22435
llvm-svn: 228478
Rui Ueyama [Sat, 7 Feb 2015 00:54:46 +0000 (00:54 +0000)]
Remove unused parameter.
llvm-svn: 228476
Richard Smith [Sat, 7 Feb 2015 00:45:52 +0000 (00:45 +0000)]
[modules] Don't accidentally trigger deserialization from DeclContext::noload_lookup.
llvm-svn: 228475
Duncan P. N. Exon Smith [Sat, 7 Feb 2015 00:37:15 +0000 (00:37 +0000)]
Support: Add dwarf::getVirtuality()
llvm-svn: 228474
Duncan P. N. Exon Smith [Sat, 7 Feb 2015 00:36:23 +0000 (00:36 +0000)]
Support: Use Dwarf.def for DW_VIRTUALITY, NFC
Use definition file for `DW_VIRTUALITY_*`. Add a `DW_VIRTUALITY_max`
both for ease of testing and for future use by the `LLParser`.
llvm-svn: 228473
Zachary Turner [Sat, 7 Feb 2015 00:14:55 +0000 (00:14 +0000)]
Dont' use close_fds = True on Windows.
If you do, the test runner will fail immediately with the error:
close_fds is not supported on Windows platforms if you redirect
stdin/stdout/stderr.
llvm-svn: 228472
Duncan P. N. Exon Smith [Fri, 6 Feb 2015 23:46:49 +0000 (23:46 +0000)]
Support: Add dwarf::getAttributeEncoding()
llvm-svn: 228470
Duncan P. N. Exon Smith [Fri, 6 Feb 2015 23:45:37 +0000 (23:45 +0000)]
Support: Rewrite AttributeEncodingString(), NFC
llvm-svn: 228469
Duncan P. N. Exon Smith [Fri, 6 Feb 2015 23:44:24 +0000 (23:44 +0000)]
Support: Stop stringifying DW_ATE_{lo,hi}_user
llvm-svn: 228468
Hal Finkel [Fri, 6 Feb 2015 23:30:06 +0000 (23:30 +0000)]
[PowerPC] Fixup incomplete revert of test/CodeGen/PowerPC/tls-pic.ll
llvm-svn: 228467
Lang Hames [Fri, 6 Feb 2015 23:26:33 +0000 (23:26 +0000)]
[Orc] Add a Kaleidoscope/Orc tutorial demonstrating lazy-irgen.
This tutorial builds on the lazy_codegen kaleidoscope/orc tutorial by making
a small set of changes (~75 lines diff) to defer ir-generation for function
definitions until functions are actually referenced.
llvm-svn: 228466
Kevin Enderby [Fri, 6 Feb 2015 23:25:38 +0000 (23:25 +0000)]
Add code to llvm-objdump so the -section option with -macho will dump literal
sections with the Mach-O S_{4,8,16}BYTE_LITERALS section types.
llvm-svn: 228465
Richard Smith [Fri, 6 Feb 2015 23:20:21 +0000 (23:20 +0000)]
PR22405: don't lose implicit-deleted-ness across AST write / read.
llvm-svn: 228464
Ahmed Bougacha [Fri, 6 Feb 2015 23:15:39 +0000 (23:15 +0000)]
[AArch64] Use the source location of the IR branch when creating Bcc
from a conditional branch fed by an add/sub/mul-with-overflow node.
We previously used the SDLoc of the overflow node, for no good reason.
In some cases, this led to the Bcc and B terminators having different
source orders, and DBG_VALUEs being inserted between them.
The real issue is with the code that can't handle DBG_VALUEs between
terminators: the few places affected by this will be fixed soon.
In the meantime, fixing the SDLoc is a positive change no matter what.
No tests, as I have no idea how to get .loc emitted for branches?
rdar://
19347133
llvm-svn: 228463
Simon Pilgrim [Fri, 6 Feb 2015 23:12:15 +0000 (23:12 +0000)]
[X86][AVX2] Begun adding AVX2 integer stack folding tests.
llvm-svn: 228462
Rui Ueyama [Fri, 6 Feb 2015 23:09:13 +0000 (23:09 +0000)]
PECOFF: Move error check for invalid command line combination to validateImpl.
llvm-svn: 228461
Hal Finkel [Fri, 6 Feb 2015 23:07:40 +0000 (23:07 +0000)]
Revert "r227976 - [PowerPC] Yet another approach to __tls_get_addr" and related fixups
Unfortunately, even with the workaround of disabling the linker TLS
optimizations in Clang restored (which has already been done), this still
breaks self-hosting on my P7 machine (-O3 -DNDEBUG -mcpu=native).
Bill is currently working on an alternate implementation to address the TLS
issue in a way that also fully elides the linker bug (which, unfortunately,
this approach did not fully), so I'm reverting this now.
llvm-svn: 228460
Lang Hames [Fri, 6 Feb 2015 23:04:53 +0000 (23:04 +0000)]
[Orc] Add a Kaleidoscope/Orc tutorial demonstrating lazy-codegen.
This tutorial builds on the initial kaleidoscope/orc tutorial by adding a
LazyEmittingLayer to the custom stack. This extra layer defers compilation
of modules in the JIT until they are statically referenced.
llvm-svn: 228459
Duncan P. N. Exon Smith [Fri, 6 Feb 2015 22:55:13 +0000 (22:55 +0000)]
Support: Add dwarf::getLanguage()
llvm-svn: 228458
Duncan P. N. Exon Smith [Fri, 6 Feb 2015 22:53:19 +0000 (22:53 +0000)]
Support: Rewrite dwarf::LanguageString(), NFC
llvm-svn: 228457
Lang Hames [Fri, 6 Feb 2015 22:52:04 +0000 (22:52 +0000)]
[Orc] Add a Kaleidoscope tutorial for Orc demonstrating eager compilation.
This tutorial demonstrates a very basic custom Orc JIT stack that performs eager
compilation: All modules are CodeGen'd immediately upon being added to the JIT.
llvm-svn: 228456
Duncan P. N. Exon Smith [Fri, 6 Feb 2015 22:50:13 +0000 (22:50 +0000)]
IR: Allow 32-bits for lines in debug location
Remove unnecessary restriction of 24-bits for line numbers in
`MDLocation`.
The rest of the debug info schema (with the exception of local
variables) uses 32-bits for line numbers. As I introduce the
specialized nodes, it makes sense to canonicalize on one size or the
other.
llvm-svn: 228455
Lang Hames [Fri, 6 Feb 2015 22:48:43 +0000 (22:48 +0000)]
[Orc] Add more missing headers.
llvm-svn: 228454
Rui Ueyama [Fri, 6 Feb 2015 22:44:16 +0000 (22:44 +0000)]
Use make_unique.
llvm-svn: 228453
Sanjay Patel [Fri, 6 Feb 2015 22:43:52 +0000 (22:43 +0000)]
use local variables; NFC
llvm-svn: 228452
Duncan P. N. Exon Smith [Fri, 6 Feb 2015 22:34:48 +0000 (22:34 +0000)]
Support: Stop stringifying DW_LANG_{lo,hi}_user
llvm-svn: 228451
Rui Ueyama [Fri, 6 Feb 2015 22:34:29 +0000 (22:34 +0000)]
Remove unused #include.
llvm-svn: 228450
Rui Ueyama [Fri, 6 Feb 2015 22:32:09 +0000 (22:32 +0000)]
ELF: Split OrderPass::compareInitFini for readability. NFC.
llvm-svn: 228449
Duncan P. N. Exon Smith [Fri, 6 Feb 2015 22:29:35 +0000 (22:29 +0000)]
AsmParser: Use DW_TAG_hi_user instead of magic constant, NFC
llvm-svn: 228448
Duncan P. N. Exon Smith [Fri, 6 Feb 2015 22:28:05 +0000 (22:28 +0000)]
AsmWriter: Extract writeTag(), NFC
llvm-svn: 228447
Duncan P. N. Exon Smith [Fri, 6 Feb 2015 22:27:22 +0000 (22:27 +0000)]
AsmWriter: Extract writeMetadataAsOperand(), NFC
llvm-svn: 228446
Evgeniy Stepanov [Fri, 6 Feb 2015 21:47:39 +0000 (21:47 +0000)]
[msan] Fix "missing origin" in atomic store.
An atomic store always make the target location fully initialized (in the
current implementation). It should not store origin. Initialized memory can't
have meaningful origin, and, due to origin granularity (4 bytes) there is a
chance that this extra store would overwrite meaningfull origin for an adjacent
location.
llvm-svn: 228444
Johannes Doerfert [Fri, 6 Feb 2015 21:39:31 +0000 (21:39 +0000)]
[Refactor] Use only one BlockGenerator for a SCoP
This change has two main purposes:
1) We do not use a static interface to hide an object we create and
destroy for every basic block we copy.
2) We allow the BlockGenerator to store information between calls to
the copyBB method. This will ease scalar/phi code generation
later on.
While a lot of method signatures were changed this should not cause
any real behaviour change.
Differential Revision: http://reviews.llvm.org/D7467
llvm-svn: 228443
Cameron Esfahani [Fri, 6 Feb 2015 21:33:08 +0000 (21:33 +0000)]
Test commit to see if it triggers an email to llvm-commits. No change.
llvm-svn: 228442
Johannes Doerfert [Fri, 6 Feb 2015 21:26:45 +0000 (21:26 +0000)]
Do not run independent blocks when we model all scalar dependences
llvm-svn: 228441
Shankar Easwaran [Fri, 6 Feb 2015 21:23:50 +0000 (21:23 +0000)]
[Revert] [ELF] Determine default search directories from Context
It looks like the Driver manages search path for each Target lld would support
on the Gnu flavor.
llvm-svn: 228440
Johannes Doerfert [Fri, 6 Feb 2015 21:20:14 +0000 (21:20 +0000)]
[Fix] Broken test case
llvm-svn: 228439
Eric Christopher [Fri, 6 Feb 2015 20:53:40 +0000 (20:53 +0000)]
Make this test a little less specific by removing the argument that
could change.
llvm-svn: 228438
Zachary Turner [Fri, 6 Feb 2015 20:42:03 +0000 (20:42 +0000)]
Try to fix Makefile build for LLVMDebugInfoPDB.
llvm-svn: 228437
Saleem Abdulrasool [Fri, 6 Feb 2015 20:34:17 +0000 (20:34 +0000)]
Unwind: hoist placement delete into base class
Move the placement delete into the base class. This permits the proper emission
of the virtual destructor in UnwindCursor by using the class specific placement
delete instead of the normal single element ::operator delete. With this patch,
we can finally build libunwind as a DSO without a runtime dependency on
libc++/libc++abi.
llvm-svn: 228436
Zachary Turner [Fri, 6 Feb 2015 20:30:52 +0000 (20:30 +0000)]
Resubmit "Create lib/DebugInfo/PDB" (r228428)
This change resubmits the patch that broke the build, this time
without unittests. The unittests will be submitted separately
after the problem has been addressed:
--Original Commit Message--
Create lib/DebugInfo/PDB.
This patch creates a platform-independent interface to a PDB reader.
There is currently no implementation of this interface, which will
be provided in future patches. This defines the basic object model
which any implementation must conform to.
Reviewed by: David Blaikie
Differential Revision: http://reviews.llvm.org/D7356
llvm-svn: 228435
Michael Zolotukhin [Fri, 6 Feb 2015 20:20:40 +0000 (20:20 +0000)]
Use estimated number of optimized insns in unroll-threshold computation.
If complete-unroll could help us to optimize away N% of instructions, we
might want to do this even if the final size would exceed loop-unroll
threshold. However, we don't want to unroll huge loop, and we are add
AbsoluteThreshold to avoid that - this threshold will never be crossed,
even if we expect to optimize 99% instructions after that.
llvm-svn: 228434
Johannes Doerfert [Fri, 6 Feb 2015 20:13:15 +0000 (20:13 +0000)]
Model PHI nodes without demoting them
This allows us to model PHI nodes in the polyhedral description
without demoting them. The modeling however will result in the
same accesses as the demotion would have introduced.
Differential Revision: http://reviews.llvm.org/D7415
llvm-svn: 228433
Michael Zolotukhin [Fri, 6 Feb 2015 20:02:51 +0000 (20:02 +0000)]
[InstSimplify] Add SimplifyFPBinOp function.
It is a variation of SimplifyBinOp, but it takes into account
FastMathFlags.
It is needed in inliner and loop-unroller to accurately predict the
transformation's outcome (previously we dropped the flags and were too
conservative in some cases).
Example:
float foo(float *a, float b) {
float r;
if (a[1] * b)
r = /* a lot of expensive computations */;
else
r = 1;
return r;
}
float boo(float *a) {
return foo(a, 0.0);
}
Without this patch, we don't inline 'foo' into 'boo'.
llvm-svn: 228432
Zachary Turner [Fri, 6 Feb 2015 20:00:18 +0000 (20:00 +0000)]
Revert "Create lib/DebugInfo/PDB."
This reverts commit 21028, as it is causing failures in LLVMConfig.
llvm-svn: 228431
Oleksiy Vyalov [Fri, 6 Feb 2015 19:56:33 +0000 (19:56 +0000)]
Make lldb-platform to clear m_process_launch_info when hanlding qProcessInfo request - otherwise subsequent process launches will reuse data from previous launch.
llvm-svn: 228430
Kostya Serebryany [Fri, 6 Feb 2015 19:52:07 +0000 (19:52 +0000)]
[fuzzer] move default sanitizer options to a separate file
llvm-svn: 228429
Zachary Turner [Fri, 6 Feb 2015 19:44:09 +0000 (19:44 +0000)]
Create lib/DebugInfo/PDB.
This patch creates a platform-independent interface to a PDB reader.
There is currently no implementation of this interface, which will
be provided in future patches. This defines the basic object model
which any implementation must conform to.
Reviewed by: David Blaikie
Differential Revision: http://reviews.llvm.org/D7356
llvm-svn: 228428
Lang Hames [Fri, 6 Feb 2015 19:36:40 +0000 (19:36 +0000)]
[Orc] Move SectionMemoryManager's implementation from MCJIT to ExecutionEngine.
This is a more sensible home for SectionMemoryManager, and allows the implementation
to be shared between Orc and MCJIT.
llvm-svn: 228427
Lang Hames [Fri, 6 Feb 2015 19:34:40 +0000 (19:34 +0000)]
[Orc] Add some missing headers.
llvm-svn: 228426
Lang Hames [Fri, 6 Feb 2015 19:34:04 +0000 (19:34 +0000)]
[Orc] Fix syntax error in LazyEmittingLayer::removeModuleSet.
This was a trivial think-o, but it's in a method of a templated class
and doesn't have any callers yet, so the compiler let it pass. I hope
to add a unit test to cover this soon.
llvm-svn: 228425
Benjamin Kramer [Fri, 6 Feb 2015 18:58:04 +0000 (18:58 +0000)]
ASTUnit: Fix a name clash in GCC builds.
Should fix the build. Looks like GCC 4.9 is using different scoping rules for
range-based for loops.
llvm-svn: 228424
Saleem Abdulrasool [Fri, 6 Feb 2015 18:48:07 +0000 (18:48 +0000)]
Unwind: replace pure virtual functions with aborts
Convert all pure virtual functions in the UnwindCursor with implementations that
abort. This is effectively manually replicating the current behaviour, whilst
removing the compiler generated calls to __cxa_pure_virtual, which will abort at
runtime with a message indicating that a pure virtual call was made.
The whitespace changes are the result of executing clang-format over the changed
region.
llvm-svn: 228423
Eric Christopher [Fri, 6 Feb 2015 18:44:18 +0000 (18:44 +0000)]
Inline asm IR input register constraints don't have early clobber
modifiers on them. If we have a matching output constraint with
an early clobber make sure we don't propagate that to the input
constraint.
llvm-svn: 228422
Quentin Colombet [Fri, 6 Feb 2015 18:42:41 +0000 (18:42 +0000)]
[LiveIntervalAnalysis] Speed up creation of live ranges for physical registers
by using a segment set.
The patch addresses a compile-time performance regression in the LiveIntervals
analysis pass (see http://llvm.org/bugs/show_bug.cgi?id=18580). This regression
is especially critical when compiling long functions. Our analysis had shown
that the most of time is taken for generation of live intervals for physical
registers. Insertions in the middle of the array of live ranges cause quadratic
algorithmic complexity, which is apparently the main reason for the slow-down.
Overview of changes:
- The patch introduces an additional std::set<Segment>* member in LiveRange for
storing segments in the phase of initial creation. The set is used if this
member is not NULL, otherwise everything works the old way.
- The set of operations on LiveRange used during initial creation (i.e. used by
createDeadDefs and extendToUses) have been reimplemented to use the segment
set if it is available.
- After a live range is created the contents of the set are flushed to the
segment vector, because the set is not as efficient as the vector for the
later uses of the live range. After the flushing, the set is deleted and
cannot be used again.
- The set is only for live ranges computed in
LiveIntervalAnalysis::computeLiveInRegUnits() and getRegUnit() but not in
computeVirtRegs(), because I did not bring any performance benefits to
computeVirtRegs() and for some examples even brought a slow down.
Patch by Vaidas Gasiunas <vaidas.gasiunas@sap.com>
Differential Revision: http://reviews.llvm.org/D6013
llvm-svn: 228421