Commit Graph

116072 Commits

Author SHA1 Message Date
Chad Rosier 7a20ed7627 Improve RefreshCallGraph to remove invalid call graph edge.
With commit r219944, InstCombine can now turn a sqrtl into a llvm.fabs.f64.
The call graph edge originally representing the call to sqrtl becomes invalid.
This patch modifies CGPassManager::RefreshCallGraph() to remove the invalid
call graph edge, which can triggers an assert in
CallGraphNode::addCalledFunction().

Phabricator Review: http://reviews.llvm.org/D7705
Patch by Lawrence Hu <lawrence@codeaurora.org>.

llvm-svn: 234902
2015-04-14 15:52:57 +00:00
Alexander Kornienko fb37cfa346 Refactor: Simplify boolean expressions in ARM target
Simplify boolean expressions using `true` and `false` with `clang-tidy`

http://reviews.llvm.org/D8524

Patch by Richard Thomson!

llvm-svn: 234901
2015-04-14 15:32:58 +00:00
Krzysztof Parzyszek c49ce520d3 Change the testcase mtriple to x86_64-unknown-unknown
llvm-svn: 234900
2015-04-14 15:28:42 +00:00
Rafael Espindola 701d79e8fa Fix MSVC build.
llvm-svn: 234899
2015-04-14 15:25:14 +00:00
Daniel Jasper 8229ebb926 The code that originally made me discover this is:
if ((a & 0x1) == 0x1) {
    ..
  }

In this case we don't actually have any branch probability information and
should not assume to have any. LLVM transforms this into:

  %and = and i32 %a, 1
  %tobool = icmp eq i32 %and, 0

So, in this case, the result of a bitwise and is compared against 0,
but nevertheless, we should not assume to have probability
information.

llvm-svn: 234898
2015-04-14 15:20:37 +00:00
Bradley Smith b913653b91 [AArch64] Allow non-standard INS/DUP encodings
The ARMv8 ARMARM states that for these instructions in A64 state:

  "Unspecified bits in "imm5" are ignored but should be set to zero by an assembler.", (imm4 for INS).

Make the disassembler accept any encoding with these ignored bits set to 1.

llvm-svn: 234896
2015-04-14 15:07:26 +00:00
Rafael Espindola 37b7015916 Add raw_pwrite_stream type.
This is a raw_ostream that also supports pwrite.
I will be used in a sec.

llvm-svn: 234895
2015-04-14 15:00:34 +00:00
Tom Stellard d4a1950500 R600/SI: Fix verifier error caused by SIAnnotateControlFlow
This pass will always try to insert llvm.SI.ifbreak intrinsics
in the same block that its conditional value is computed in.  This is
a problem when conditions for breaks or continue are computed outside
of the loop, because the llvm.SI.ifbreak intrinsic ends up being inserted
outside of the loop.

This patch fixes this problem by inserting the llvm.SI.ifbreak
intrinsics in the loop header when the condition is computed outside
the loop.

llvm-svn: 234891
2015-04-14 14:36:45 +00:00
Aaron Ballman c20e6d755d Fixing a compile error with MSVC 2013 where there is no conversion from DISubprogram to DIScope directly.
llvm-svn: 234890
2015-04-14 14:19:09 +00:00
Filipe Cabecinhas 225542713b Error out of ParseBitcodeInto(Module*) if we haven't read a Module
Summary:
Without this check the following case failed:

Skip a SubBlock which is not a MODULE_BLOCK_ID nor a BLOCKINFO_BLOCK_ID
Got to end of file

TheModule would still be == nullptr, and we would subsequentially fail
when materializing the Module (assert at the start of
BitcodeReader::MaterializeModule).

Bug found with AFL.

Reviewers: dexonsmith, rafael

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D9014

llvm-svn: 234887
2015-04-14 14:07:15 +00:00
Aaron Ballman fc3530a0f6 Silencing a -Wreturn-type warning, as the switch is fully-covered; NFC.
llvm-svn: 234884
2015-04-14 13:26:11 +00:00
Petar Jovanovic 0380d0b88f Re-enable target-specific relocation table sorting and use it for Mips
Some targets (ie. Mips) have additional rules for ordering the relocation
table entries. Allow them to override generic sortRelocs(), which sorts
entries by Offset.
Then override this function for Mips, to emit HI16 and GOT16 relocations
against the local symbol in pair with the corresponding LO16 relocation.

Patch by Vladimir Stefanovic.

Differential Revision: http://reviews.llvm.org/D7414

llvm-svn: 234883
2015-04-14 13:23:34 +00:00
Daniel Jasper 274928fec6 [MachineLICM] Use newer model of register pressure sets.
TargetRegisterInfo::getRegPressureLimit has a note that it is an old
model that relies on manually entered classes. Using the newer model of
register pressure sets seems more appropriate. We might eventually even
switch to lib/CodeGen/RegisterPressure.cpp, but we should probably do
incremental changes here.

Using the newer model also makes it easier to take regmasks into account
which is necessary to fix llvm.org/PR23143. I am currently also
preparing a patch for that, but would like to do this switch
independently.

Review: http://reviews.llvm.org/D8986
llvm-svn: 234880
2015-04-14 11:56:25 +00:00
NAKAMURA Takumi 80ccca3702 Roll back llvm/test/ExecutionEngine/MCJIT/cross-module-sm-pic-a.ll, possibly wrong commit.
It reverts part of r234839, "[RuntimeDyldELF] Improve GOT support".

llvm-svn: 234879
2015-04-14 10:54:14 +00:00
Anders Waldenborg 1433fd4699 Fix crash in DebugInfoFinder when adding a module with forward declared composite type
The testcase that is included in the patch caused a crash when doing DebugInfoFinder::processModule
on the module due to DCT->getElements() returning nullptr in DebugInfoFinder::processType.

By doing "DCT->getElements()" instead of "DCT->getElements()->operands()" one gets a DIArray
instead of a raw MDTuple. The former has code to handle null as a 0-element array and
therefore avoids the crash.

Differential Revision: http://reviews.llvm.org/D9008

llvm-svn: 234875
2015-04-14 09:18:17 +00:00
Jingyue Wu 8cb6b2a292 Simplify n-ary adds by reassociation
Summary:
This transformation reassociates a n-ary add so that the add can partially reuse
existing instructions. For example, this pass can simplify

  void foo(int a, int b) {
    bar(a + b);
    bar((a + 2) + b);
  }

to

  void foo(int a, int b) {
    int t = a + b;
    bar(t);
    bar(t + 2);
  }

saving one add instruction.

Fixes PR22357 (https://llvm.org/bugs/show_bug.cgi?id=22357).

Test Plan: nary-add.ll

Reviewers: broune, dberlin, hfinkel, meheff, sanjoy, atrick

Reviewed By: sanjoy, atrick

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D8950

llvm-svn: 234855
2015-04-14 04:59:22 +00:00
Duncan P. N. Exon Smith acdee690c8 DebugInfo: Update signature of DICompileUnit::replace*()
Change `DICompileUnit::replaceSubprograms()` and
`DICompileUnit::replaceGlobalVariables()` to match the `MDCompileUnit`
equivalents that they're wrapping.

llvm-svn: 234852
2015-04-14 03:51:36 +00:00
Dmitri Gribenko d81be6d664 LinkAllPasses.h: also link in parts of libLLVMSupport
When a loadable (.so or .dylib) pass is built with assertions enabled and
loaded into the 'opt' tool, we need to ensure that the extra symbols that such
passes depend on are linked into the tool.

llvm-svn: 234851
2015-04-14 03:49:53 +00:00
Duncan P. N. Exon Smith 537b4a8159 DebugInfo: Gut DISubprogram and DILexicalBlock*
Gut the `DIDescriptor` wrappers around `MDLocalScope` subclasses.  Note
that `DILexicalBlock` wraps `MDLexicalBlockBase`, not `MDLexicalBlock`.

llvm-svn: 234850
2015-04-14 03:40:37 +00:00
Sanjoy Das b96610410f [SCEVExpander] Fix comments for functions. NFC.
Bring function documentation for ScalarEvolutionExpander up to code by
not repeating the function name in the comment documenting
functionality.  Reflow the edited comments where needed.

llvm-svn: 234847
2015-04-14 03:20:40 +00:00
Sanjoy Das e178f46965 [LoopUnrollRuntime] Avoid high-cost trip count computation.
Summary:
Runtime unrolling of loops needs to emit an expression to compute the
loop's runtime trip-count.  Avoid runtime unrolling if this computation
will be expensive.

Depends on D8993.

Reviewers: atrick

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D8994

llvm-svn: 234846
2015-04-14 03:20:38 +00:00
Sanjoy Das a9f1e27a04 [SCEV] Strengthen SCEVExpander::isHighCostExpansion.
Summary:

Teach `isHighCostExpansion` to consider divisions by power-of-two
constants as cheap and add a test case.  This change is needed for a new
user of `isHighCostExpansion` that will be added in a subsequent change.

Depends on D8995.

Reviewers: atrick

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D8993

llvm-svn: 234845
2015-04-14 03:20:32 +00:00
Sanjoy Das 2e6bb3b947 [SCEV] Refactor out isHighCostExpansion. NFCI.
Summary:
Move isHighCostExpansion from IndVarSimplify to SCEVExpander.  This
exposed function will be used in a subsequent change.

Reviewers: bogner, atrick

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D8995

llvm-svn: 234844
2015-04-14 03:20:28 +00:00
Duncan P. N. Exon Smith 20caafbfd6 DebugInfo: Gut DINamespace and DITemplate*Parameter
Continue gutting `DIDescriptor` subclasses, turning them into
as-bare-as-possible pointer wrappers.

llvm-svn: 234843
2015-04-14 03:01:27 +00:00
Duncan P. N. Exon Smith 5f88ba1d5d DebugInfo: Add MDLexicalBlockBase::getLine(), etc.
Add a few functions from `DILexicalBlock` to `MDLexicalBlockBase`,
leaving `DILexicalBlock` a simple wrapper.

IMO, the new functions (`getLine()` and `getColumn()`) don't really
belong in the base class, but to simplify transitioning old code it
seems like the right incremental step.  I've explicitly deleted them in
`MDLexicalBlockFile`, and eventually the callers should be updated to
downcast to `MDLexicalBlock` directly and the forwarding functions
removed.

llvm-svn: 234842
2015-04-14 02:50:07 +00:00
Duncan P. N. Exon Smith 7348ddaa74 DebugInfo: Gut DIVariable and DIGlobalVariable
Gut all the non-pointer API from the variable wrappers, except an
implicit conversion from `DIGlobalVariable` to `DIDescriptor`.  Note
that if you're updating out-of-tree code, `DIVariable` wraps
`MDLocalVariable` (`MDVariable` is a common base class shared with
`MDGlobalVariable`).

llvm-svn: 234840
2015-04-14 02:22:36 +00:00
Keno Fischer 02628def32 [RuntimeDyldELF] Improve GOT support
Summary:
This is the first in a series of patches to eventually add support for TLS relocations to RuntimeDyld. This patch resolves an issue in the current GOT handling, where GOT entries would be reused between object files, which leads to the same situation that necessitates the GOT in the first place, i.e. that the 32-bit offset can not cover all of the address space. Thus this patch makes the GOT object-file-local.
Unfortunately, this still isn't quite enough, because the MemoryManager does not yet guarantee that sections are allocated sufficiently close to each other, even if they belong to the same object file. To address this concern, this patch also adds a small API abstraction on top of the GOT allocation mechanism that will allow (temporarily, until the MemoryManager is improved) using the stub mechanism instead of allocating a different section. The actual switch from separate section to stub mechanism will be part of a follow-on commit, so that it can be easily reverted independently at the appropriate time.

Test Plan: Includes a test case where the GOT of two object files is artificially forced to be apart by several GB.

Reviewers: lhames

Reviewed By: lhames

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D8813

llvm-svn: 234839
2015-04-14 02:10:35 +00:00
Duncan P. N. Exon Smith 32e7f2860b DebugInfo: Move DIVariable::printExtendedName() to its only caller
Move the local function `printDebugLoc()` along with it.

llvm-svn: 234838
2015-04-14 02:09:32 +00:00
Duncan P. N. Exon Smith 571baebacb DebugInfo: Inline DIVariable::isBlockByrefVariable() into its callers
I don't think this API is helping much.  Change the callers to call
`MDType::isBlockByrefStruct()` directly.

llvm-svn: 234837
2015-04-14 01:59:58 +00:00
Duncan P. N. Exon Smith de8e427394 DebugInfo: Gut DIObjCProperty and DIImportedEntity
Gut a couple more classes in the DIDescriptor hierarchy.  Leave behind
an implicit conversion to `DIDescriptor`, the old base class.

llvm-svn: 234836
2015-04-14 01:46:44 +00:00
Duncan P. N. Exon Smith b7e221ba55 DebugInfo: Gut DILocation
This is along the same lines as r234832, but for `DILocation`.  Clean
out all accessors from `DILocation`.  Any callers should be using
`MDLocation` directly (e.g., via `operator->()`).

llvm-svn: 234835
2015-04-14 01:35:55 +00:00
Adam Nemet 26da8e9800 [LoopAccesses] Properly print whether memchecks are needed
Fix oversight in -analyze output.  PtrRtCheck contains the pointers that
need to be checked against each other and not whether memchecks are
necessary.

For instance in the testcase PtrRtCheck has four elements but all
no-alias so no checking is necessary.

llvm-svn: 234833
2015-04-14 01:12:55 +00:00
Duncan P. N. Exon Smith 6a0320a991 DebugInfo: Gut DIExpression
Completely gut `DIExpression`, turning it into a simple wrapper around
`MDExpression *`.  There are two bits of magic left:

  - It's constructed from `const MDExpression*` but convertible to
    `MDExpression*`.
  - It's default-constructed to `nullptr`.

Otherwise, it should behave quite like a raw pointer.  Once I've done
the same to the rest of the `DIDescriptor` subclasses, I'll come back to
delete them entirely (and update call sites as necessary to deal with
the missing magic).

llvm-svn: 234832
2015-04-14 01:12:42 +00:00
Duncan P. N. Exon Smith e7515ebf66 DebugInfo: Move an assertion into MDCompositeTypeBase
In the name of gutting the `DIDescriptor` hierarchy.

llvm-svn: 234829
2015-04-14 00:57:50 +00:00
Philip Reames ba1984958d [RewriteStatepointsForGC] Delete dead code [NFC]
Before we had real liveness, we needed to track every value that base pointer
insertion code created because these now might be live.  We now just rerun 
the data flow liveness algorithm (which is actually faster!) and no longer 
need the associated code.

llvm-svn: 234827
2015-04-14 00:41:34 +00:00
Duncan P. N. Exon Smith 843237f573 DebugInfo: Move DILocation::computeNewDiscriminators()
As documented in PR23200 (and the FIXMEs I've added to the code here),
this logic is fairly broken: it modifies the `LLVMContext` in a way that
affects other modules and cannot be serialized to assembly/bitcode.  For
now, move it over to `MDLocation::computeNewDiscriminators()` anyway.

llvm-svn: 234825
2015-04-14 00:35:42 +00:00
Duncan P. N. Exon Smith 4fd839b0da AddDiscriminators: Create new MDLocation directly
I don't see a reason to add the `copyWithNewScope()` API over to
`MDLocation` -- it seems to be a holdover from when creating locations
required knowing details of operand layout -- so change
`AddDiscriminators` to call `MDLocation::get()` directly.  Should be no
functionality change here.

llvm-svn: 234824
2015-04-14 00:34:30 +00:00
Lang Hames 47260c23ca [Orc] Revert 234815. Still haven't quite got this test figured out apparently.
llvm-svn: 234822
2015-04-14 00:27:47 +00:00
Duncan P. N. Exon Smith 2c8291e3b4 DebugInfo: Move DILocation::getDiscriminator() to MDLocation
llvm-svn: 234819
2015-04-14 00:05:13 +00:00
Duncan P. N. Exon Smith 26e1cda1f4 DebugInfo: Remove DIObjCProperty attribute accessors, NFC
There's only one user of the various `DIObjCProperty::is*Property()`
accessors -- `DwarfUnit::constructTypeDIE()` -- and it's just using the
reverse logic to reconstruct the bitfield.  Drop this API and simplify
the only caller.

llvm-svn: 234818
2015-04-13 23:55:01 +00:00
Duncan P. N. Exon Smith 0bf8d4c6c1 DebugInfo: Move DIDerivedType accessors to MDDerivedType, NFC
Add accessors in `MDDerivedType` to downcast `getExtraData()`, matching
those in `DIDerivedType`.

llvm-svn: 234816
2015-04-13 23:36:36 +00:00
Lang Hames 2bde68c2e6 [Orc] Make the OrcLazy hello.ll regression test output via stderr.
This keeps the program and JIT output in sync, enabling FileCheck to test the
order of target program and JIT events.

In particular we can now test that main is not compiled until after the global
constructor has run.

llvm-svn: 234815
2015-04-13 23:28:46 +00:00
Daniel Berlin 8de312d2a9 Add new getModRefInfo API to determine whether an Instruction and a call modify the same memory
llvm-svn: 234814
2015-04-13 23:25:41 +00:00
Daniel Berlin b8a4d41327 Common some code from MemoryDependenceAnalysis that will be used in MemorySSA
llvm-svn: 234813
2015-04-13 23:20:13 +00:00
Duncan P. N. Exon Smith 56b1d00484 DebugInfo: Make DIDerivedType accessors more strict
These accessors in `DIDerivedType` should only be called when `DbgNode`
really is a `MDDerivedType`, not just a `MDDerivedTypeBase`.  Assume
that it is.

llvm-svn: 234812
2015-04-13 23:13:18 +00:00
Daniel Berlin b2d227693f Make getModRefInfo with a default location not crash.
Add getModRefInfo that works without location.
Add unit tests.

llvm-svn: 234811
2015-04-13 23:05:45 +00:00
Lang Hames cf0ed3a836 [Orc] Back out r234805 for hello.ll until I can figure out how to sync up the
output.

llvm-svn: 234810
2015-04-13 22:58:39 +00:00
Ahmed Bougacha 8ebcdb3bc3 [CodeGen] Combine concat_vectors of scalars into build_vector.
Combine something like:
  (v8i8 concat_vectors (v2i8 bitcast (i16)) x4)
into:
  (v8i8 (bitcast (v4i16 BUILD_VECTOR (i16) x4)))

If any of the scalars are floating point, use that throughout.

Differential Revision: http://reviews.llvm.org/D8948

llvm-svn: 234809
2015-04-13 22:57:21 +00:00
Daniel Berlin d8d6046e69 Allow printing functions with an optional annotationwriter
llvm-svn: 234807
2015-04-13 22:36:38 +00:00
Lang Hames 271f4d1c1b [Orc] Disambiguate call to make_unique. This should fix the builders broken by
r234805.

llvm-svn: 234806
2015-04-13 22:33:05 +00:00
Lang Hames cd3fd83c3a [Orc] Add an Orc layer for applying arbitrary transforms to IR, use it to add
debugging output to the LLI orc-lazy JIT, and update the orc-lazy "hello.ll"
test to actually test for lazy compilation.

llvm-svn: 234805
2015-04-13 22:12:54 +00:00
Duncan P. N. Exon Smith d11d9c265f DebugInfo: Simplify a few more wrappers
This is almost NFC, but I'm removing some assertions against `nullptr`.
The assertions aren't worth all that much since we'll typically get
segfaults at the same site (and I imagine ASan catches this sort of
thing), and besides: the whole idea is to replace the `DIDescriptor`
hierarchy with raw pointers to the new one.

llvm-svn: 234802
2015-04-13 21:58:26 +00:00
Krzysztof Parzyszek 9c43a08615 Settle on a specific triple for the aarch64 testcase
llvm-svn: 234801
2015-04-13 21:55:21 +00:00
Duncan P. N. Exon Smith 745a5db444 SelectionDAG: Stop using DIVariable::isInlinedFnArgument()
Instead of calling the somewhat confusingly-named
`DIVariable::isInlinedFnArgument()`, do the check directly here.
There's possibly a small functionality change here: instead of
`dyn_cast<>`'ing `DV->getScope()` to `MDSubprogram`, I'm looking up the
scope chain for the actual subprogram.  I suspect that this is a no-op
for function arguments so in practise there isn't a real difference.

I've also added a `FIXME` to check the `inlinedAt:` chain instead, since
I wonder if that would be more reliable than the
`MDSubprogram::describes()` function.

Since this was the only user of `DIVariable::isInlinedFnArgument()`,
delete it.

llvm-svn: 234799
2015-04-13 21:38:48 +00:00
Chris Bieneman 168ddf4dc1 [CMake] Updates to llvm-shlib to support overriding exports list and added an option to export all symbols.
llvm-svn: 234798
2015-04-13 21:29:46 +00:00
Krzysztof Parzyszek ac1b966604 Also add mtriple to the aarch64 testcase
llvm-svn: 234797
2015-04-13 20:49:08 +00:00
Duncan P. N. Exon Smith cca5f68e53 DebugInfo: Remove DIGlobalVariable::getGlobal()
`DIGlobalVariable::getGlobal()` isn't really helpful, it just does a
`dyn_cast_or_null<>`.  Simplify its only user by doing the cast directly
and delete the code.

llvm-svn: 234796
2015-04-13 20:39:25 +00:00
Krzysztof Parzyszek 2c4487de6d Expand ADDO/SUBO on Hexagon
llvm-svn: 234795
2015-04-13 20:37:01 +00:00
Krzysztof Parzyszek cab9ce9163 Add mtriple to test case to avoid problems with different naming schemes
llvm-svn: 234793
2015-04-13 20:24:40 +00:00
Duncan P. N. Exon Smith 7fa6629d7d StripSymbols: Use DIGlobalVariable::getConstant() instead of getGlobal()
The only difference between the two is a `dyn_cast<>` to
`GlobalVariable`.  If optimizations have left anything behind when a
global gets replaced, then it doesn't seem like the debug info is dead.

I can't seem to find an optimization that would leave behind a
non-`GlobalVariable` without nulling the reference entirely, so I
haven't added a testcase (but I'll be deleting `getGlobal()` in a future
commit).

llvm-svn: 234792
2015-04-13 20:13:30 +00:00
Nick Lewycky d6f241d53b GCC complains thusly: "attributes at the beginning of statement are ignored [-Werror=attributes]". Very well then! NFC
llvm-svn: 234788
2015-04-13 20:03:08 +00:00
Philip Reames f209a153f1 [RwriteStatepointsForGC] Minor indentation and naming [NFC]
Use early-return style that's preferred in LLVM and updating the naming in places I touched with other changes in the last few days.  Hopefully, NFC.

llvm-svn: 234785
2015-04-13 20:00:30 +00:00
Duncan P. N. Exon Smith 7f38919e05 DebugInfo: Remove a few unnecessary wrappers
llvm-svn: 234783
2015-04-13 19:44:31 +00:00
Duncan P. N. Exon Smith f24e76c36c DebugInfo: Assume valid pointer in DISubprogram::replaceFunction()
Other accessors assume this already; not sure how `replaceFunction()`
got left behind.

llvm-svn: 234782
2015-04-13 19:41:30 +00:00
Nick Lewycky abe2cc17da Subtraction is not commutative. Fixes PR23212!
llvm-svn: 234780
2015-04-13 19:17:37 +00:00
Philip Reames 2114275263 [RewriteStatepointsForGC] Avoid inserting empty holder
We use dummy calls to adjust the liveness of values over statepoints in the midst of the insertion.  If there are no values which need held live, there's no point in actually inserting the holder.  

llvm-svn: 234779
2015-04-13 19:07:47 +00:00
Duncan P. N. Exon Smith 3c2d70434b DebugInfo: Migrate DISubprogram::describes() to new hierarchy, NFC
I don't really like this function at all -- I think it should be as
simple as `return getFunction() == F` -- but for now this seems like the
best we can do.

llvm-svn: 234778
2015-04-13 19:07:27 +00:00
Krzysztof Parzyszek 3ff5657329 Remove this test until I figure out why it fails
llvm-svn: 234777
2015-04-13 18:57:50 +00:00
Duncan P. N. Exon Smith c82570b68b Reapply "Verifier: Check for incompatible bit piece expressions"
This reverts commit r234717, reapplying r234698 (in spirit).

As described in r234717, the original `Verifier` check had a
use-after-free.  Instead of storing pointers to "interesting" debug info
intrinsics whose bit piece expressions should be verified once we have
typerefs, do a second traversal.  I've added a testcase to catch the
`llc` crasher.

Original commit message:

    Verifier: Check for incompatible bit piece expressions

    Convert an assertion into a `Verifier` check.  Bit piece expressions
    must fit inside the variable, and mustn't be the entire variable.
    Catching this in the verifier will help us find bugs sooner, and makes
    `DIVariable::getSizeInBits()` dead code.

llvm-svn: 234776
2015-04-13 18:53:11 +00:00
Matthias Braun d0d5cbe4dd Use FileCheck for test
llvm-svn: 234774
2015-04-13 18:47:19 +00:00
Akira Hatanaka f99e1913ae [inliner] Don't inline a function if it doesn't have exactly the same
target-cpu and target-features attribute strings as the caller.

Differential Revision: http://reviews.llvm.org/D8984

llvm-svn: 234773
2015-04-13 18:43:38 +00:00
Krzysztof Parzyszek 13b4b013b5 Make the ARM testcase from r234764 also pass on Thumb
llvm-svn: 234772
2015-04-13 18:39:52 +00:00
Philip Reames 69e51cae33 [RewriteStatepointsForGC] Fix a latent bug in normalization for invoke statepoint [NFC]
Since we're restructuring the CFG, we also need to make sure to update the analsis passes. While I'm touching the code, I dedicided to restructure it a bit.  The code involved here was very confusing.  This change moves the normalization to essentially being a pre-pass before the main insertion work and updates a few comments to actually say what is happening and *why*.

The restructuring should be covered by existing tests.  I couldn't easily see how to create a test for the invalidation bug.  Suggestions welcome.

llvm-svn: 234769
2015-04-13 18:07:21 +00:00
Jan Vesely ffcd968647 Revert revisions r234755, r234759, r234760
Revert "Remove default in fully-covered switch (to fix Clang -Werror -Wcovered-switch-default)"
Revert "R600: Add carry and borrow instructions. Use them to implement UADDO/USUBO"
Revert "LegalizeDAG: Try to use Overflow operations when expanding ADD/SUB"

Using overflow operations fails CodeGen/Generic/2011-07-07-ScheduleDAGCrash.ll
on hexagon, nvptx, and r600. Revert while I investigate.

llvm-svn: 234768
2015-04-13 17:47:15 +00:00
Philip Reames 9a2e01d908 [RewriteStatepointsForGC] Strengthen assertions around liveness
This is related to the issues addressed in 234651.  These assertions check the properties ensured by that change at the place of use.  Note that a similiar property is checked in checkBasicSSA, but without the reachability constraint.  Technically, the liveness would be correct to include unreachable values, but this would be problematic for actual relocation.

llvm-svn: 234766
2015-04-13 17:35:55 +00:00
Krzysztof Parzyszek a46c36b8f4 Allow memory intrinsics to be tail calls
llvm-svn: 234764
2015-04-13 17:16:45 +00:00
Matthias Braun a283cb3265 DAGCombiner: Fix crash in select(select) opt.
In case of different types used for the condition of the selects the
select(select) -> select(and) normalisation cannot be performed.

See also: http://reviews.llvm.org/D7622

llvm-svn: 234763
2015-04-13 17:16:33 +00:00
Philip Reames e73300b925 [RewriteStatepointsForGC] Move an expensive debugging check to XDEBUG
The check in question is attempting to help find cases where we haven't relocated a pointer at a safepoint we should have.  It does this by coercing the value to null at any safepoint which doesn't relocate it.  

Unfortunately, this turns out to be rather expensive in terms of memory usage and time.  The number of stores inserted can grow with O(number of values x number of statepoints).  On at least one example I looked at, over half of peak memory usage was coming from this check.  

With this change, the check is no longer enabled by default in Asserts builds.  It is enabled for expensive asserts builds and has a command line option to enable it in both Asserts and non-Asserts builds.  

llvm-svn: 234761
2015-04-13 16:41:32 +00:00
David Blaikie 155f38e0d8 Remove default in fully-covered switch (to fix Clang -Werror -Wcovered-switch-default)
llvm-svn: 234760
2015-04-13 16:37:50 +00:00
Jan Vesely d579048464 R600: Add carry and borrow instructions. Use them to implement UADDO/USUBO
v2: tighten the sub64 tests
v3: rename to CARRY/BORROW
v4: fixup test cmdline
    add known bits computation
    use sign extend instead of sub 0,x
    better add test
v5: remove redundant break
    move lowering to separate functions
    fix comments

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewers: arsenm
llvm-svn: 234759
2015-04-13 16:26:00 +00:00
David Blaikie 29ac43c361 llvm-readobj: teach it to handle MachO Universal Archive correctly
Patch by Chilledheart (rwindz0@gmail.com).

Reviewed By: rafael

Differential Revision: http://reviews.llvm.org/D8773

llvm-svn: 234758
2015-04-13 16:05:49 +00:00
David Blaikie 8d7554ceff [docs] Update outdated ExtendingLLVM.rst
Summary:
The document is still incomplete in some degrees, but updated to reflect the
latest changes. Anyway we can detail it if any one think it is not enough. For
the sake of it, some useful examples are listed below:

Refer to r113618 "Add X86 MMX type to bitcode and Type" for how to add a new
type.

> One notable change from then is only one thing that ``lib/VMCore`` is renamed
to ``lib/IR``.

Refer to r194760 "Add addrspacecast instruction" for how to add a new
instruction.

Patch by Chilledheart (rwindz0@gmail.com).

Reviewed By: echristo

Differential Revision: http://reviews.llvm.org/D8897

llvm-svn: 234757
2015-04-13 16:04:17 +00:00
Jan Vesely a835555e40 LegalizeDAG: Try to use Overflow operations when expanding ADD/SUB
v2: consider BooleanContents when processing overflow

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewers: resistor, jholewinsky (nvidia parts)
Differential Revision: http://reviews.llvm.org/D6340

llvm-svn: 234755
2015-04-13 15:32:01 +00:00
Rafael Espindola a9b84abba8 Fix SupportsSeeking detection on windows.
Will be tested by existing tests once used (soon).

llvm-svn: 234737
2015-04-13 11:09:48 +00:00
John Brawn e8fd6c8563 [ARM] Align global variables passed to memory intrinsics
Fill in the TODO in CodeGenPrepare::OptimizeCallInst so that global
variables that are passed to memory intrinsics are aligned in the same
way that allocas are.

Differential Revision: http://reviews.llvm.org/D8421

llvm-svn: 234735
2015-04-13 10:47:39 +00:00
Rafael Espindola 79be4cc634 Add r234615 back, but make sure outs() is binary.
Original message.

Have one raw_fd_ostream constructor forward to the other.

This fixes some odd behaviour differences between the two. In particular,
the version that takes a FD no longer unconditionally sets stdout to binary.

llvm-svn: 234734
2015-04-13 10:28:56 +00:00
NAKAMURA Takumi 379ed3942e Revert r234615, "Have one raw_fd_ostream constructor forward to the other."
It broke MSVCRT hosts:

  LLVM :: Object/check_binary_output.ll
  LLVM :: Object/extract.ll

llvm-svn: 234721
2015-04-13 04:54:19 +00:00
NAKAMURA Takumi 57184325e1 [CMake] Uniquify LLVM_ABI_BREAKING_CHECKS. It reverts r233784, "[CMake] Cache in LLVMConfig.cmake."
llvm-svn: 234720
2015-04-13 04:54:12 +00:00
NAKAMURA Takumi fa88e0d2b9 llvm/test/CodeGen/R600/fminnum.ll: Relax an expression for NaN on MSVCRT like r204118.
<stdin>:202:2: note: possible intended match here
   2143289344(1.#QNAN0e+00), 2(2.802597e-45)

llvm-svn: 234719
2015-04-13 04:54:06 +00:00
Duncan P. N. Exon Smith 5cd454161c Revert "Verifier: Check for incompatible bit piece expressions"
This reverts commit r234698.

This caused a use-after-free: `QueuedBitPieceExpressions` holds onto
references to `DbgInfoIntrinsic`s and references them past where they're
deleted (this is because the verifier is run as a function pass, and
then `verifyTypeRefs()` is called during `doFinalization()`).

I'll include a reduced crasher for `llc` when I recommit the check.

llvm-svn: 234717
2015-04-13 00:06:28 +00:00
Jan Vesely 452b036697 R600: Make FMIN/MAXNUM legal on all asics
v2: Add tests

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
reviewer: arsenm
llvm-svn: 234716
2015-04-12 23:45:05 +00:00
Jan Vesely 811ef52db7 R600: remove manual BFE optimization
Fixed since r233079

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
reviewer: arsenm
llvm-svn: 234715
2015-04-12 23:45:01 +00:00
Petr Hosek 9e0c890f3e [MC] Write padding into fragments when -mc-relax-all flag is used
Summary:
When instruction bundling is enabled and the -mc-relax-all flag is
set, we can write bundle padding directly into fragments and avoid
creating large number of fragments significantly reducing LLVM MC
memory usage.

Test Plan: Regression test attached

Reviewers: eliben

Subscribers: jfb, mseaborn

Differential Revision: http://reviews.llvm.org/D8072

llvm-svn: 234714
2015-04-12 23:42:25 +00:00
Lang Hames 44398e4961 [Orc] Remove duplicate "then" in comment.
llvm-svn: 234710
2015-04-12 22:23:57 +00:00
Lang Hames 65a0109565 [Orc] Fix out-of-date comment for the IRCompileLayer.
llvm-svn: 234709
2015-04-12 22:21:48 +00:00
Lang Hames c6de458f7c [Orc] During module partitioning, rename anonymous and asm-private globals.
If they're not (re)named, these globals will fail to resolve when the
partitioned modules are linked.

llvm-svn: 234707
2015-04-12 20:05:51 +00:00
Mark Lacey 274f48b5a8 Fix typo.
llvm-svn: 234706
2015-04-12 18:18:51 +00:00
Hal Finkel 5551f2528c [PowerPC] Really iterate over all loops in PPCLoopDataPrefetch/PPCLoopPreIncPrep
When I fixed these a couple of days ago to iterate over all loops, not just
depth == 1 loops, I inadvertently made it such that we'd only look at the first
top-level loop. Make sure that we really look at all of them.

llvm-svn: 234705
2015-04-12 17:18:56 +00:00
Sanjoy Das 71190feca5 [LoopUnrollRuntime] Clean up a predicate.
Clean up a predicate I added in r229731, fix the relevant comment and
add a test case.  The earlier version is confusing to read and was also
buggy (probably not a coincidence) till Alexey fixed it in r233881.

llvm-svn: 234701
2015-04-12 01:24:01 +00:00
Duncan P. N. Exon Smith 7ad0bd54d3 DebugInfo: Make MDSubprogram::getFunction() return Constant
Change `MDSubprogram::getFunction()` and
`MDGlobalVariable::getConstant()` to return a `Constant`.  Previously,
both returned `ConstantAsMetadata`.

llvm-svn: 234699
2015-04-11 20:27:40 +00:00
Duncan P. N. Exon Smith 5ad6ff76dc Verifier: Check for incompatible bit piece expressions
Convert an assertion into a `Verifier` check.  Bit piece expressions
must fit inside the variable, and mustn't be the entire variable.
Catching this in the verifier will help us find bugs sooner, and makes
`DIVariable::getSizeInBits()` dead code.

llvm-svn: 234698
2015-04-11 19:58:35 +00:00
Duncan P. N. Exon Smith 127ea4b616 DebugInfo: Remove dead DIDescriptor::replaceAllUsesWith()
r234696 replaced the only use of `DIDescriptor::replaceAllUsesWith()`
with `DIBuilder::replaceTemporary()` (added in r234695).  Delete the
dead code.

llvm-svn: 234697
2015-04-11 19:29:09 +00:00
Duncan P. N. Exon Smith c89d664f72 DebugInfo: Introduce DIBuilder::replaceTemporary()
Add `DIBuilder::replaceTemporary()` as a replacement for
`DIDescriptor::replaceAllUsesWith()`.  I'll update clang to use the new
method, and then come back to delete the original.

This method dispatches to `replaceAllUsesWith()` or
`replaceWithUniqued()`, depending on whether the replacement is actually
a different node from the original.

llvm-svn: 234695
2015-04-11 19:04:09 +00:00
Benjamin Kramer 79de6e6d89 Mark empty default constructors as =default if it makes the type POD
NFC

llvm-svn: 234694
2015-04-11 18:57:14 +00:00
Duncan P. N. Exon Smith 19f87726d5 DebugInfo: Assume a valid pointer for DISubprogram::getFunction()
llvm-svn: 234693
2015-04-11 18:15:48 +00:00
Benjamin Kramer 30bfab3cb4 Make OnDiskChainedHashTableGenerator's Bucket type a POD type and reorder fields to optimize padding
For the common case of offset_type=unsigned this brings Bucket from 24 to 16
bytes.

llvm-svn: 234692
2015-04-11 18:00:28 +00:00
Duncan P. N. Exon Smith f0d81a50bf DebugInfo: Move DIScope::getName() and getContext() to MDScope
Continue gutting the `DIDescriptor` hierarchy.  In this case, move the
guts of `DIScope::getName()` and `DIScope::getContext()` to
`MDScope::getName()` and `MDScope::getScope()`.

llvm-svn: 234691
2015-04-11 17:37:23 +00:00
Benjamin Kramer dd0ff85701 Remove empty non-virtual destructors or mark them =default when non-public
These add no value but can make a class non-trivially copyable. NFC.

llvm-svn: 234688
2015-04-11 15:32:26 +00:00
Hal Finkel 58f5f9c393 [PowerPC] Disable part-word atomics on the P7
As it turns out, even though these are part of ISA 2.06, the P7 does not
support them (or, at least, not any P7s we're tested so far).

llvm-svn: 234686
2015-04-11 13:40:36 +00:00
Nemanja Ivanovic c38b5311cb Add direct moves to/from VSR and exploit them for FP/INT conversions
This patch corresponds to review:
http://reviews.llvm.org/D8928

It adds direct move instructions to/from VSX registers to GPR's. These are
exploited for FP <-> INT conversions.

llvm-svn: 234682
2015-04-11 10:40:42 +00:00
Alexander Kornienko f817c1cb9a Use 'override/final' instead of 'virtual' for overridden methods
The patch is generated using clang-tidy misc-use-override check.

This command was used:

  tools/clang/tools/extra/clang-tidy/tool/run-clang-tidy.py \
    -checks='-*,misc-use-override' -header-filter='llvm|clang' \
    -j=32 -fix -format

http://reviews.llvm.org/D8925

llvm-svn: 234679
2015-04-11 02:11:45 +00:00
Duncan P. N. Exon Smith 63ffa21d90 DebugInfo: Rewrite atSameLineAs() as MDLocation::canDiscriminate()
Rewrite `DILocation::atSameLineAs()` as `MDLocation::canDiscriminate()`
with a doxygen comment explaining its purpose.  I've added a few FIXMEs
where I think this check is too weak; fixing that is tracked by PR23199.

llvm-svn: 234674
2015-04-11 01:00:47 +00:00
Lang Hames 709a0dcfef [Orc] Update kaleidoscope tutorial for API change in r234669.
llvm-svn: 234673
2015-04-11 00:56:21 +00:00
Duncan P. N. Exon Smith 5ba747b9b1 DebugInfo: Remove dead DIRef friends
llvm-svn: 234672
2015-04-11 00:45:24 +00:00
Duncan P. N. Exon Smith 50c065bc7d DebugInfo: Add forwarding getFilename() accessor to new hierarchy
Add forwarding `getFilename()` and `getDirectory()` accessors to nodes
in the new hierarchy that define a `getFile()`.  Use that to
re-implement existing functionality in the `DIDescriptor` hierarchy.

llvm-svn: 234671
2015-04-11 00:39:43 +00:00
Hal Finkel ffb460fdf0 [PowerPC] Fix PPCLoopPreIncPrep for depth > 1 loops
This pass had the same problem as the data-prefetching pass: it was only
checking for depth == 1 loops in practice. Fix that, add some debugging
statements, and make sure that, when we grab an AddRec, it is for the loop we
expect.

llvm-svn: 234670
2015-04-11 00:33:08 +00:00
Lang Hames eb9bdb5d16 [Orc] Tidy up IndirectionUtils API a little, add some comments. NFC.
llvm-svn: 234669
2015-04-11 00:23:49 +00:00
Philip Reames 9638ff9b37 [Statepoints] Fix a release only build failure
A function which is used only in Asserts builds needs to be defined only in Asserts builds.

llvm-svn: 234667
2015-04-11 00:06:47 +00:00
Ahmed Bougacha b96444efd1 [CodeGen] Split -enable-global-merge into ARM and AArch64 options.
Currently, there's a single flag, checked by the pass itself.
It can't force-enable the pass (and is on by default), because it
might not even have been created, as that's the targets decision.
Instead, have separate explicit flags, so that the decision is
consistently made in the target.

Keep the flag as a last-resort "force-disable GlobalMerge" for now,
for backwards compatibility.

llvm-svn: 234666
2015-04-11 00:06:36 +00:00
Duncan P. N. Exon Smith 241982c380 DebugInfo: Remove dead DIDescriptor::getDescriptorField()
llvm-svn: 234665
2015-04-10 23:53:44 +00:00
Quentin Colombet fd7475b5e8 [AArch64] Strengthen the code for the prologue insertion.
The spilled registers are pristine and thus, correctly handled by
the register scavenger and so on, but the liveness information is
strictly speaking wrong at this point.
Fix that.

llvm-svn: 234664
2015-04-10 23:14:34 +00:00
Reid Kleckner 9405ef0e1f [WinEH] Recognize SEH finally block inserted by the frontend
This allows winehprepare to build sensible llvm.eh.actions calls for SEH
finally blocks.  The pattern matching in this change is brittle and
should be replaced with something more robust soon.  In the meantime,
this will let us write the code that produces __C_specific_handler xdata
tables, which we need regardless of how we decide to get finally blocks
through EH preparation.

llvm-svn: 234663
2015-04-10 23:12:29 +00:00
Philip Reames 4d80ede538 [RewriteStatepointsForGC] Use a SetVector for a worklist [NFC]
Using a SetVector to replace equivelent but more verbose functionality.

llvm-svn: 234662
2015-04-10 23:11:26 +00:00
Tim Northover 1290047a7d Generic: Make isMask_N and isShiftedMask_N consistent over 0
Previously, isMask_N returned false for 0 but isShiftedMask_N returned true.
Almost all uses are for pattern matching bitfield operations in the backends,
and expect false (this was discovered because of AArch64's copy of this logic).

Unfortunately, I couldn't put together a small non-fragile test for this. The
nature of the bitfield operations means that this edge case is only really
triggered for nodes like "(and N, 0)", which the DAG combiner is usually very
good at folding away before they get to this stage.

rdar://20501377

llvm-svn: 234659
2015-04-10 22:58:48 +00:00
Philip Reames 8106a264d3 [RewriteStatepointsForGC] test case missing from 234657
llvm-svn: 234658
2015-04-10 22:58:39 +00:00
Philip Reames df1ef08c0c [RewriteStatepointsForGC] Use an actual liveness algorithm
When rewriting statepoints to make relocations explicit, we need to have a conservative but consistent notion of where a particular pointer is live at a particular site. The old code just used dominance, which is correct, but decidedly more conservative then it needed to be. This patch implements a simple dataflow algorithm that's run one per function (well, twice counting fixup after base pointer insertion). There's still lots of room to make this faster, but it's fast enough for all practical purposes today.

Differential Revision: http://reviews.llvm.org/D8674

llvm-svn: 234657
2015-04-10 22:53:14 +00:00
Philip Reames 704e78b149 [RewriteStatepointsForGC] clang-format file
Format the entire file to reduce diff of change to follow.

llvm-svn: 234656
2015-04-10 22:34:56 +00:00
Benjamin Kramer b4bf14ceaa [CodeGenPrepare] Report all changes made during instruction sinking
r234638 chained another transform below which was tripping over the
deleted instruction. Use after free found by asan in many regression
tests.

llvm-svn: 234654
2015-04-10 22:25:36 +00:00
Philip Reames f66d73708b [RewriteStatepointsForGC] Missed review comment from 234651 & build fix
After submitting 234651, I noticed I hadn't responded to a review comment by mjacob.  This patch addresses that comment and fixes a Release only build problem due to an unused variable.  

llvm-svn: 234653
2015-04-10 22:16:58 +00:00
Philip Reames 85b36a8157 [RewriteStatepointsForGC] Preprocess the IR to remove unreachable blocks and single entry phis
Two related small changes:

    Various dominance based queries about liveness can get confused if we're talking about unreachable blocks. To avoid reasoning about such cases, just remove them before rewriting statepoints.
    Remove single entry phis (likely left behind by LCSSA) to reduce the number of live values.

Both of these are motivated by http://reviews.llvm.org/D8674 which will be submitted shortly.

Differential Revision: http://reviews.llvm.org/D8675

llvm-svn: 234651
2015-04-10 22:07:04 +00:00
Philip Reames 8531d8c491 [RewriteStatepointsForGC] Limited support for vectors of pointers
This patch adds limited support for inserting explicit relocations when there's a vector of pointers live over the statepoint. This doesn't handle the case where the vector contains a mix of base and non-base pointers; that's future work.

The current implementation just scalarizes the vector over the gc.statepoint before doing the explicit rewrite. An alternate approach would be to plumb the vector all the way though the backend lowering, but doing that appears challenging. In particular, the size of the indirect spill slot is currently assumed to be sizeof(pointer) throughout the backend.

In practice, this is enough to allow running the SLP and Loop vectorizers before RewriteStatepointsForGC.

Differential Revision: http://reviews.llvm.org/D8671

llvm-svn: 234647
2015-04-10 21:48:25 +00:00
Sanjoy Das b6c5914308 [InstCombine][CodeGenPrep] Create llvm.uadd.with.overflow in CGP.
Summary:
This change moves creating calls to `llvm.uadd.with.overflow` from
InstCombine to CodeGenPrep.  Combining overflow check patterns into
calls to the said intrinsic in InstCombine inhibits optimization because
it introduces an intrinsic call that not all other transforms and
analyses understand.

Depends on D8888.

Reviewers: majnemer, atrick

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D8889

llvm-svn: 234638
2015-04-10 21:07:09 +00:00
Reid Kleckner aa7a5a3a0f Avoid spewing binary to stdout in some filetype=obj tests
llvm-svn: 234627
2015-04-10 19:36:55 +00:00
Sanjay Patel 89e28f67cc use update_llc_test_checks.py to tighten checking
test features, not CPUs

remove unnecessary cruft

llvm-svn: 234622
2015-04-10 18:31:42 +00:00
Rafael Espindola 74d2f617f2 Remember if lseek works in this FD.
It will be used in clang in a sec.

llvm-svn: 234619
2015-04-10 18:15:51 +00:00
Duncan P. N. Exon Smith 457bfc779a DebugInfo: Stop leaking temporaries in DIBuilder::createCompileUnit()
Stop leaking temporary nodes from `DIBuilder::createCompileUnit()`.
`replaceAllUsesWith()` doesn't delete the nodes, so we need to delete
them "manually" (well, `TempMDTuple` does that for us).

Similarly, stop leaking the temporary nodes used for variables of
subprograms.

llvm-svn: 234617
2015-04-10 18:01:58 +00:00
Rafael Espindola a4800720b0 Have one raw_fd_ostream constructor forward to the other.
This fixes some odd behavior differences between the two. In particular,
the version that takes a FD no longer unconditionally sets stdout to binary.

llvm-svn: 234615
2015-04-10 17:52:22 +00:00
Reid Kleckner d03f9f4016 [FS] Report errors from llvm::sys::fs::rename on Windows
Previously we would always report success, which is pretty bogus.

I'm too lazy to write a test where rename will portably fail on all
platforms. I'm just trying to fix breakage introduced by r234597, which
happened to tickle this.

llvm-svn: 234611
2015-04-10 17:20:45 +00:00
Reid Kleckner 6e48a826e8 [WinEH] Try to make outlining invokes work a little better
WinEH currently turns invokes into calls. Long term, we will reconsider
this, but for now, make sure we remap the operands and clone the
successors of the new terminator.

llvm-svn: 234608
2015-04-10 16:26:42 +00:00
Benjamin Kramer b3bcf604d1 [CallSite] Use the injected base class name. NFC.
llvm-svn: 234606
2015-04-10 16:02:34 +00:00
Hal Finkel a9fceb803d [PowerPC] Prefetching should also consider depth > 1 loops
Iterating over loops from the LoopInfo instance only provides top-level loops.
We need to search the whole tree of loops to find the inner ones.

llvm-svn: 234603
2015-04-10 15:05:02 +00:00
Benjamin Kramer 3a09ef64ee [CallSite] Make construction from Value* (or Instruction*) explicit.
CallSite roughly behaves as a common base CallInst and InvokeInst. Bring
the behavior closer to that model by making upcasts explicit. Downcasts
remain implicit and work as before.

Following dyn_cast as a mental model checking whether a Value *V isa
CallSite now looks like this: 
  if (auto CS = CallSite(V)) // think dyn_cast
instead of:
  if (CallSite CS = V)

This is an extra token but I think it is slightly clearer. Making the
ctor explicit has the advantage of not accidentally creating nullptr
CallSites, e.g. when you pass a Value * to a function taking a CallSite
argument.

llvm-svn: 234601
2015-04-10 14:50:08 +00:00
Toma Tabacu ae47f93b74 [mips] [IAS] Improve comments in MipsAsmParser::expandLoadImm. NFC.
llvm-svn: 234595
2015-04-10 13:28:16 +00:00
Chad Rosier 518659d9b4 [AArch64] Changes some SchedAlias to WriteRes for Cortex-A57.
Using SchedAliases is convenient and works well for latency and resource
lookup for instructions.  However, this creates an entry in
AArch64WriteLatencyTable with a WriteResourceID of 0, breaking any
SchedReadAdvance since the lookup will fail.

http://reviews.llvm.org/D8043
Patch by Dave Estes <cestes@codeaurora.org>!

llvm-svn: 234594
2015-04-10 13:19:27 +00:00
Chad Rosier a82c876045 [AArch64] Adjusts Cortex-A57 machine model to handle zero shift.
http://reviews.llvm.org/D8043
Patch by Dave Estes <cestes@codeaurora.org>!

llvm-svn: 234593
2015-04-10 13:19:21 +00:00
Benjamin Kramer 9b88fed323 Microoptimize DenseMap::clear.
Cache NumEntries locally, it's only used in an assert and using the member
variable prevents the compiler from eliminating the tombstone check for types
with trivial destructors. No functionality change intended.

llvm-svn: 234589
2015-04-10 12:46:44 +00:00
Benjamin Kramer 619c4e57ba Reduce dyn_cast<> to isa<> or cast<> where possible.
No functional change intended.

llvm-svn: 234586
2015-04-10 11:24:51 +00:00
Toma Tabacu 038a330ede [mips] [IAS] Make the mips-expansions-bad.s test more readable. NFC.
Move the check lines below the code lines and change the indentation from 8
spaces to 2 spaces.

llvm-svn: 234584
2015-04-10 10:46:59 +00:00
Kostya Serebryany cd073d5475 [lib/Fuzzer] Section: How good is my fuzzer?
llvm-svn: 234571
2015-04-10 06:32:29 +00:00
Kostya Serebryany 6bd016be8e [lib/Fuzzer] explain compatibility with AFL
llvm-svn: 234570
2015-04-10 05:44:43 +00:00
Jingyue Wu 5da831cc31 Divergence analysis for GPU programs
Summary:
Some optimizations such as jump threading and loop unswitching can negatively
affect performance when applied to divergent branches. The divergence analysis
added in this patch conservatively estimates which branches in a GPU program
can diverge. This information can then help LLVM to run certain optimizations
selectively.

Test Plan: test/Analysis/DivergenceAnalysis/NVPTX/diverge.ll

Reviewers: resistor, hfinkel, eliben, meheff, jholewinski

Subscribers: broune, bjarke.roune, madhur13490, tstellarAMD, dberlin, echristo, jholewinski, llvm-commits

Differential Revision: http://reviews.llvm.org/D8576

llvm-svn: 234567
2015-04-10 05:03:50 +00:00
David Majnemer 5c65f58f64 [WinEHPrepare] Don't rely on the order of IR
The IPToState table must be emitted after we have generated labels for
all functions in the table.  Don't rely on the order of the list of
globals.  Instead, utilize WinEHFuncInfo to tell us how many catch
handlers we expect to outline.  Once we know we've visited all the catch
handlers, emit the cppxdata.

llvm-svn: 234566
2015-04-10 04:56:17 +00:00
Hal Finkel 93138503ae [PowerPC] Don't crash on PPC32 i64 fp_to_uint on modern cores
When we have an instruction for this (and, thus, don't generate a runtime
call), we need to custom type legalize this (in a trivial way, just as we do
for fp_to_sint).

Fixes PR23173.

llvm-svn: 234561
2015-04-10 03:39:00 +00:00
Ahmed Bougacha 1ffe7c7d36 [AArch64] Promote f16 operations to f32.
For the most common ones (such as fadd), we already did the promotion.
Do the same thing for all the others.

Currently, we'll just crash/assert on all these operations, as
there's no hardware or libcall support whatsoever.

f16 (half) is specified as an interchange - not arithmetic - format,
and is expected to be promoted to single-precision for arithmetic
operations.

While there, teach the legalizer about promoting some of the (mostly
floating-point) operations that we never needed before.

Differential Revision: http://reviews.llvm.org/D8648
See related discussion on the thread for: http://reviews.llvm.org/D8755

llvm-svn: 234550
2015-04-10 00:08:48 +00:00
Nemanja Ivanovic c09047916a Add LLVM support for remaining integer divide and permute instructions from ISA 2.06
This is the patch corresponding to review:
http://reviews.llvm.org/D8406

It adds some missing instructions from ISA 2.06 to the PPC back end.

llvm-svn: 234546
2015-04-09 23:54:37 +00:00
Rafael Espindola 5682ce2ceb Simplify use of formatted_raw_ostream.
formatted_raw_ostream is a wrapper over another stream to add column and line
number tracking.

It is used only for asm printing.

This patch moves the its creation down to where we know we are printing
assembly. This has the following advantages:

* Simpler lifetime management: std::unique_ptr
* We don't compute column and line number of object files :-)

llvm-svn: 234535
2015-04-09 21:06:08 +00:00
Ahmed Bougacha df43737782 [CodeGen] Combine concat_vector of trunc'd scalar to scalar_to_vector.
We already do:
  concat_vectors(scalar, undef) -> scalar_to_vector(scalar)
When the scalar is legal.
When it's not, but is a truncated legal scalar, we can also do:
  concat_vectors(trunc(scalar), undef) -> scalar_to_vector(scalar)
Which is equivalent, since the upper lanes are undef anyway.
While there, teach the combine to look at more than 2 operands.

Differential Revision: http://reviews.llvm.org/D8883

llvm-svn: 234530
2015-04-09 20:04:47 +00:00
Juergen Ributzka bd0c7eb4dc [AArch64][FastISel] Fix integer extend optimization.
The integer extend optimization tries to fold the extend into the load
instruction. This requires us to identify if the extend has already been
emitted or not and act accordingly on it.

The check that was originally performed for this was not sufficient. Besides
checking the ValueMap for a mapped register we also need to check if the
virtual register has already an associated machine instruction that defines it.

This fixes rdar://problem/20470788.

llvm-svn: 234529
2015-04-09 20:00:46 +00:00
Eric Christopher fbe80f5f63 Remove duplicated code and consolidate initializers.
llvm-svn: 234525
2015-04-09 19:20:37 +00:00
Rafael Espindola 49286e9f4a clang-format bits of code to make a followup patch easy to read.
llvm-svn: 234519
2015-04-09 18:32:58 +00:00
Rafael Espindola 1c84271694 Revert "Refactoring and enhancement to FMA combine."
This reverts commit r234513. It was failing on the bots.

llvm-svn: 234518
2015-04-09 18:29:32 +00:00
Rafael Espindola 37099d9afd Define a function with "... llvm::func...".
Using this instead of
namespace llvm {
  func...
}

Has the advantage that the build fails with a compiler error if it gets out
of sync with the .h file.

llvm-svn: 234515
2015-04-09 18:08:15 +00:00
Olivier Sallenave 53703d0862 Refactoring and enhancement to FMA combine.
llvm-svn: 234513
2015-04-09 17:55:26 +00:00
Duncan P. N. Exon Smith 628d6057d0 IR: Preserve use-list order by default in bitcode
Pull the `-preserve-*-use-list-order` flags out of "experimental" mode,
and preserve use-list order by default when serializing to bitcode.

llvm-svn: 234510
2015-04-09 17:41:20 +00:00
Rafael Espindola f546d0f9c5 Use a raw_svector_ostream instead of a raw_string_ostream.
It saves a bit of copying.

llvm-svn: 234507
2015-04-09 17:16:25 +00:00
Rafael Espindola df7305a438 Don't repeat name in comment. NFC.
llvm-svn: 234506
2015-04-09 17:10:57 +00:00
Jingyue Wu b89f42254b [NFC] add more comments for SLSR
llvm-svn: 234505
2015-04-09 17:04:28 +00:00
Rafael Espindola 62e6ec066e Misc cleanup. NFC.
These were lost when I reverted the raw_ostream changes.

llvm-svn: 234504
2015-04-09 16:59:07 +00:00
Rafael Espindola 2e089d8d95 clang-format. NFC.
llvm-svn: 234502
2015-04-09 16:43:22 +00:00
Rafael Espindola 3a2b3f4129 clang-format this constructor.
llvm-svn: 234501
2015-04-09 16:37:11 +00:00
Rafael Espindola 504a5dccc4 Don't repeat names in comments.
llvm-svn: 234498
2015-04-09 16:06:26 +00:00
Rafael Espindola 7a051fa27a Use implicit calls to parent constructor. NFC.
llvm-svn: 234497
2015-04-09 16:00:24 +00:00
Rafael Espindola ee0dd4d289 This reverts commit r234460 and r234461.
Revert "Add classof implementations to the raw_ostream classes."
Revert "Use the cast machinery to remove dummy uses of formatted_raw_ostream."

The underlying issue can be fixed without classof.

llvm-svn: 234495
2015-04-09 15:54:59 +00:00
Javed Absar 5c5e3c5e36 [ARM] support for Cortex-R4/R4F
Currently, llvm (backend) doesn't know cortex-r4, even though it is the
default target for armv7r. Using "--target=armv7r-arm-none-eabi" provokes
'cortex-r4' is not a recognized processor for this target' by llvm.
This patch adds support for cortex-r4 and, very closely related, r4f.

llvm-svn: 234486
2015-04-09 14:07:28 +00:00
Rafael Espindola 3e608b0f0c Nothing inherits from the asm streamer.
Make that explicit and remove protected:

llvm-svn: 234484
2015-04-09 13:04:20 +00:00
Toma Tabacu be218927f8 [mips] Refactor saved-registers bitmask creation in MipsAsmPrinter::printSavedRegsBitmask. NFC.
Summary:
Make the code more readable by fusing the for-loops together and explicitly checking for each register class.

Also, this version is more straightforward because it doesn't assume that FPU registers always come before CPU registers in the CalleeSavedInfo vector.

Reviewers: dsanders

Reviewed By: dsanders

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D8033

llvm-svn: 234475
2015-04-09 10:54:16 +00:00
Kristof Beyls 17cb8982f4 [AArch64] Add support for dynamic stack alignment
Differential Revision: http://reviews.llvm.org/D8876

llvm-svn: 234471
2015-04-09 08:49:47 +00:00
Lang Hames 522bf13b83 [AArch64] Remove redundant -march option. Also fix a think-o from r234462.
llvm-svn: 234467
2015-04-09 05:34:57 +00:00
Nick Lewycky d65d3d7b8b Not all triples put _ before function names. Specify a triple to make this test pass on Linux.
llvm-svn: 234466
2015-04-09 05:31:32 +00:00
Craig Topper 8b88cc284f Use SmallVector instead of std::vector for uniquing X86 disassembler operand sets. The number of operands is a small fixed size.
llvm-svn: 234465
2015-04-09 04:08:48 +00:00
Craig Topper 34136874fd Simplify some printing code by combining new lines onto previous strings. Don't work so hard not to print a comma on the last entry of an array.
llvm-svn: 234464
2015-04-09 04:08:46 +00:00
Craig Topper d434e39557 Don't convert enum to strings just to put them in the uniquing map. Use the enum directly. Only convert to a string for printing.
llvm-svn: 234463
2015-04-09 04:08:42 +00:00
Lang Hames 903338511b [AArch64] Teach AArch64TargetLowering::getOptimalMemOpType to consider alignment
restrictions when choosing a type for small-memcpy inlining in
SelectionDAGBuilder.

This ensures that the loads and stores output for the memcpy won't be further
expanded during legalization, which would cause the total number of instructions
for the memcpy to exceed (often significantly) the inlining thresholds.

<rdar://problem/17829180>

llvm-svn: 234462
2015-04-09 03:40:33 +00:00
Rafael Espindola 132381f981 Use the cast machinery to remove dummy uses of formatted_raw_ostream.
If we know we are producing an object, we don't need to wrap the stream
in a formatted_raw_ostream anymore.

llvm-svn: 234461
2015-04-09 02:28:12 +00:00
Rafael Espindola 0a261a3dda Add classof implementations to the raw_ostream classes.
More uses to follow in a another patch.

llvm-svn: 234460
2015-04-09 02:10:28 +00:00
Rafael Espindola 622be5793d Delete unused constructor.
llvm-svn: 234459
2015-04-09 01:11:26 +00:00
Eric Christopher bc91e908f4 Update comment to refer to software floating point rather than
a local variable.

llvm-svn: 234457
2015-04-09 00:14:49 +00:00
Akira Hatanaka 4cb3d064ce Use option -march instead of -mtriple to avoid overconditionalizing the test.
This fixes r234439, which was committed to fix the test failures caused by
r234430.

llvm-svn: 234451
2015-04-08 23:02:45 +00:00
Manman Ren ed6b5fc4b0 [LTO] do not run internalize pass from compileOptimized.
The input to compileOptimized is already optimized and internalized, so remove
internalize pass from compileOptimized.

rdar://20227235

llvm-svn: 234446
2015-04-08 22:02:11 +00:00
Akira Hatanaka bda7aa5938 Pass -mtriple to llc to appease buildbot.
This fixes the test case I committed in r234430.

llvm-svn: 234439
2015-04-08 21:30:48 +00:00
Andrew Kaylor e104d89c8f Formmatting correction
llvm-svn: 234438
2015-04-08 21:22:46 +00:00
Andrew Kaylor 67d3c0359d [WinEH] Minor bug fixes.
Fixed insert point for allocas created for demoted values.
Clear the nested landing pad list after it has been processed.

llvm-svn: 234433
2015-04-08 20:57:22 +00:00
Akira Hatanaka c6fab80536 [DAGCombine] Fix a bug in MergeConsecutiveStores.
The bug manifests when there are two loads and two stores chained as follows in
a DAG,

(ld v3f32) -> (st f32) -> (ld v3f32) -> (st f32)

and the stores' values are extracted from the preceding vector loads.

MergeConsecutiveStores would replace the first store in the chain with the
merged vector store, which would create a cycle between the merged store node
and the last load node that appears in the chain.

This commits fixes the bug by replacing the last store in the chain instead.

rdar://problem/20275084

Differential Revision: http://reviews.llvm.org/D8849

llvm-svn: 234430
2015-04-08 20:34:53 +00:00
Peter Collingbourne 0b9c7bb9fc Go bindings: make various DIBuilder arguments optional.
r234262 changed some code in DIBuilderBindings.cpp to use the unwrap function
to unwrap debug metadata. The problem with this is that unwrap asserts that
its argument is non-null, which is not what we want in a number of places
in DIBuilder where the argument is optional. This change makes certain
arguments optional by adding null checks in places where it is required,
fixing the llgo build.

llvm-svn: 234428
2015-04-08 20:18:57 +00:00
Rafael Espindola e70edce22b Don't repeat names in comments.
llvm-svn: 234427
2015-04-08 20:16:23 +00:00
Rafael Espindola 9a07495d6d Remove unused variable.
llvm-svn: 234426
2015-04-08 20:04:20 +00:00
Cameron Zwarich b282ef0111 Eliminate O(n^2) worst-case behavior in SSA construction
The code uses a priority queue and a worklist, which share the same
visited set, but the visited set is only updated when inserting into
the priority queue. Instead, switch to using separate visited sets
for the priority queue and worklist.

llvm-svn: 234425
2015-04-08 18:26:20 +00:00
Adam Nemet ce48250f11 [LoopAccesses] Allow analysis to complete in the presence of uniform stores
(Re-apply r234361 with a fix and a testcase for PR23157)

Both run-time pointer checking and the dependence analysis are capable
of dealing with uniform addresses. I.e. it's really just an orthogonal
property of the loop that the analysis computes.

Run-time pointer checking will only try to reason about SCEVAddRec
pointers or else gives up. If the uniform pointer turns out the be a
SCEVAddRec in an outer loop, the run-time checks generated will be
correct (start and end bounds would be equal).

In case of the dependence analysis, we work again with SCEVs. When
compared against a loop-dependent address of the same underlying object,
the difference of the two SCEVs won't be constant. This will result in
returning an Unknown dependence for the pair.

When compared against another uniform access, the difference would be
constant and we should return the right type of dependence
(forward/backward/etc).

The changes also adds support to query this property of the loop and
modify the vectorizer to use this.

Patch by Ashutosh Nema!

llvm-svn: 234424
2015-04-08 17:48:40 +00:00
Scott Douglass 7ad7792088 [ARM] make vminnm/vmaxnm work with ?le, ?ge and no-nans-fp-math
Because -menable-no-nans causes fcmp conditions to be rewritten
without 'o' or 'u' the recognition code in needs to cope. Also
extended it to handle 'le' and 'ge.

Differential Revision: http://reviews.llvm.org/D8725

llvm-svn: 234421
2015-04-08 17:18:28 +00:00
Sanjay Patel 0f712d04f2 fixed to test features, not CPU models
llvm-svn: 234413
2015-04-08 16:51:42 +00:00