Commit Graph

1691 Commits

Author SHA1 Message Date
Pete Cooper b4eede2c07 Rename NumOperands to make it clear its managed by the User. NFC.
This is to try make it very clear that subclasses shouldn't be changing
the value directly.  Now that OperandList for normal instructions is computed
using the NumOperands, its critical that the NumOperands is accurate or we
could compute the wrong offset to the first operand.

I looked over all places which update NumOperands and they are all safe.
Hung off use User's don't use NumOperands to compute the OperandList so they
are safe to continue to manipulate it.  The only other User which changed it
was GlobalVariable which has an optional init list but always allocated space
for a single Use.  It was correctly setting NumOperands to 1 before setting an
initializer, and setting it to 0 after clearing the init list, so the order was safe.

Added some comments to that code to make sure that this isn't changed in future
without being aware of this constraint.

Reviewed by Duncan Exon Smith.

llvm-svn: 239621
2015-06-12 17:48:10 +00:00
Pete Cooper 74510a409d Replace all accesses to User::OperandList with getter and setter methods. NFC.
We don't want anyone to access OperandList directly as its going to be removed
and computed instead.  This uses getter's and setter's instead in which we
can later change the underlying implementation of OperandList.

Reviewed by Duncan Exon Smith.

llvm-svn: 239620
2015-06-12 17:48:05 +00:00
Alexey Samsonov 770f65ca6a Set proper debug location for branch added in BasicBlock::splitBasicBlock().
This improves debug locations in passes that do a lot of basic block
transformations. Important case is LoopUnroll pass, the test for correct
debug locations accompanies this change.

Test Plan: regression test suite

Reviewers: dblaikie, sanjoy

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D10367

llvm-svn: 239551
2015-06-11 18:25:54 +00:00
Benjamin Kramer 2d221406fa Replace an instance of custom atomics with standard ones.
Eventually I want to get rid of them entirely, but Statistic.h is still blocked
on MSVC bugs. No functionality change.

llvm-svn: 239545
2015-06-11 17:30:34 +00:00
Reid Kleckner c35e7f52ba Revert "Move dllimport name mangling to IR mangler."
This reverts commit r239437.

This broke clang-cl self-hosts. We'd end up calling the __imp_ symbol
directly instead of using it to do an indirect function call.

llvm-svn: 239502
2015-06-11 01:31:48 +00:00
Pete Cooper 3fc3040860 Stop returning a Use* from allocHungOffUses.
This always just set the User::OperandList which is now set
in that method instead of being returned.

Reviewed by Duncan Exon Smith.

llvm-svn: 239493
2015-06-10 22:38:46 +00:00
Pete Cooper 93f9ff5781 Add User::growHungoffUses and use it to grow the hung off uses. NFC.
PhiNode, SwitchInst, LandingPad and IndirectBr all had virtually identical
logic for growing the hung off uses.
Move it to User so that they can all call a single shared implementation.

Their destructors were all empty after this change and were deleted.  They all
have virtual clone_impl methods which can be used as vtable anchors.

Reviewed by Duncan Exon Smith.

llvm-svn: 239492
2015-06-10 22:38:41 +00:00
Pete Cooper 178dcc2938 Delete User::dropHungOffUses and move it in to ~User which is the only caller. NFC.
Now that the subclasses which care about hung off uses let ~User clean it up,
there's no need for a separate method.  Just inline it to ~User and delete it.

Reviewed by Duncan Exon Smith.

llvm-svn: 239491
2015-06-10 22:38:38 +00:00
Pete Cooper c6c0439d2a Make User track whether a class has 'hung off uses' and delete them in its destructor.
Currently all of the logic for deleting hung off uses, which PHI/switch/etc use,
is in their classes.

This adds a bit to Value which tracks whether that user had hung off uses,
then User can be responsible for clearing them instead of the sub classes.

Note, the bit used here was taken from NumOperands which was 30-bits.
Given the reduction to 29 bits, and the average User being just over 100 bytes,
a single User with 29-bits of num operands would need 50GB of RAM for itself
so its reasonable to assume that 29-bits is enough for now.

This is a step towards hiding all the hung off uses logic in the User.

Reviewed by Duncan Exon Smith.

llvm-svn: 239490
2015-06-10 22:38:34 +00:00
Pete Cooper 87b925b064 Move the special Phi logic for hung off uses in to User::allocHungOffUses. NFC.
PhiNode's need to allocate space for an array of Use[N] and then BasicBlock*[N].

They had their own allocHungOffUses to handle all of this.  This moves the logic
in to User::allocHungOffUses and PhiNode passes in a bool to say to allocate
the BB* space too.

Reviewed by Duncan Exon Smith.

llvm-svn: 239489
2015-06-10 22:38:30 +00:00
Peter Collingbourne 115fe37621 ArgumentPromotion: Drop sret attribute on functions that are only called directly.
If the first argument to a function is a 'this' argument and the second
has the sret attribute, the ArgumentPromotion pass may promote the 'this'
argument to more than one argument, violating the IR constraint that 'sret'
may only be applied to the first or second argument.

Although this IR constraint is arguably unnecessary, it highlighted the fact
that ArgPromotion does not need to preserve this attribute. Dropping the
attribute reduces register pressure in the backend by avoiding the register
copy required by sret. Because sret implies noalias, we also replace the
former with the latter.

Differential Revision: http://reviews.llvm.org/D10353

llvm-svn: 239488
2015-06-10 21:14:34 +00:00
Peter Collingbourne 9fe51fdf18 Move dllimport name mangling to IR mangler.
This ensures that LTO clients see the correct external symbol name.

Differential Revision: http://reviews.llvm.org/D10318

llvm-svn: 239437
2015-06-09 22:09:53 +00:00
Benjamin Kramer f1cfc4244c Prefer copy init over direct init. NFC.
llvm-svn: 239327
2015-06-08 18:58:57 +00:00
Benjamin Kramer bbd05a2470 [AsmWriter] Rewrite module asm printing using StringRef::split.
No change in output intended.

llvm-svn: 239251
2015-06-07 13:59:33 +00:00
Yaron Keren 4849aa35b7 Rangify more for loops in LegacyPassManager.cpp.
llvm-svn: 239166
2015-06-05 17:48:47 +00:00
Yaron Keren 3b1e24be56 Rangify for loops in LegacyPassManager.cpp.
llvm-svn: 239155
2015-06-05 14:15:07 +00:00
Peter Zotov c164a3f4e6 [C API] Add LLVMStructGetTypeAtIndex.
Patch by deadalnix (Amaury SECHET).

llvm-svn: 239029
2015-06-04 09:09:53 +00:00
David Majnemer 38eb9f46db [ConstantFold] Don't skip the first gep index when folding geps
We neglected to check if the first index made the GEP ineligible for
'inbounds'.

This fixes PR23753.

llvm-svn: 239015
2015-06-04 07:01:56 +00:00
Nick Lewycky 2f41757634 Int128 is also a built-in preconstructed type.
llvm-svn: 238889
2015-06-02 23:56:53 +00:00
Filipe Cabecinhas 62431b1d71 [IR/AsmWriter] Output escape sequences if the first character isdigit()
If the first character in a metadata attachment's name is a digit, it has
to be output using an escape sequence, otherwise it's not valid text IR.

Removed an over-zealous assert from LLVMContext which didn't allow this.
The rule should only apply to text IR. Actual names can have any sequence
of non-NUL bytes.

Also added some documentation on accepted names.

Bug found with AFL fuzz.

llvm-svn: 238867
2015-06-02 21:25:08 +00:00
Filipe Cabecinhas 1ac0d2de3a clang-format a few functions. NFC
llvm-svn: 238865
2015-06-02 21:25:00 +00:00
Owen Anderson 7349ab9385 Move the name pointer out of Value into a map that lives on the
LLVMContext.  Production builds of clang do not set names on most
Value's, so this is wasted space on almost all subclasses of Value.
This reduces the size of all Value subclasses by 8 bytes on 64 bit
hosts.

The one tricky part of this change is averting compile time regression
by keeping Value::hasName() fast.  This required stealing bits out of
NumOperands.

With this change, peak memory usage on verify-uselistorder-nodbg.lto.bc
is decreased by approximately 2.3% (~3MB absolute on my machine).

llvm-svn: 238791
2015-06-01 22:24:01 +00:00
Artur Pilipenko a82f8db0b3 Add isConstant argument to MDBuilder::createTBAAStructTagNode
According to the TBAA description struct-path tag node can have an optional IsConstant field. Add corresponding argument to MDBuilder::createTBAAStructTagNode.

Reviewed By: hfinkel

Differential Revision: http://reviews.llvm.org/D10160

llvm-svn: 238749
2015-06-01 14:53:55 +00:00
David Blaikie f5147ef0b9 [opaque pointer type] Explicitly store the pointee type of the result of a GEP
Alternatively, this type could be derived on-demand whenever
getResultElementType is called - if someone thinks that's the better
choice (simple time/space tradeoff), I'm happy to give it a go.

llvm-svn: 238716
2015-06-01 03:09:34 +00:00
Ahmed Bougacha 0ea9d1e753 [IR] fptrunc-of-fptrunc isn't an EliminableCastPair.
Double and single rounding can produce different results.
This is the IR counterpart to r228911.

llvm-svn: 238531
2015-05-29 00:04:30 +00:00
Owen Anderson 85fa7d5037 Add initial support for the convergent attribute.
llvm-svn: 238264
2015-05-26 23:48:40 +00:00
Philip Reames 38840245e4 [PlaceSafepoints] Cleanup InsertSafepointPoll function
While working on another change, I noticed that the naming in this function was mildly deceptive.  While fixing that, I took the oppurtunity to modernize some of the code.  NFC intended.

llvm-svn: 238252
2015-05-26 21:03:23 +00:00
Kit Barton 6646033e6e This patch adds support for the vector quadword add/sub instructions introduced
in POWER8:

vadduqm
vaddeuqm
vaddcuq
vaddecuq
vsubuqm
vsubeuqm
vsubcuq
vsubecuq
In addition to adding the instructions themselves, it also adds support for the
v1i128 type for intrinsics (Intrinsics.td, Function.cpp, and
IntrinsicEmitter.cpp).

http://reviews.llvm.org/D9081

llvm-svn: 238144
2015-05-25 15:49:26 +00:00
Akira Hatanaka bd881834c5 Simplify and rename function overrideFunctionAttributes. NFC.
This is in preparation to making changes needed to stop resetting
NoFramePointerElim in resetTargetOptions.

llvm-svn: 238079
2015-05-23 01:12:26 +00:00
Pete Cooper eb31b68cef Use Op<0> accessor instead of OperandList for Instructions. NFC
This is consistent with other uses of the operand list.  I'm planning a future commit where this will actually matter.

llvm-svn: 237967
2015-05-21 22:48:54 +00:00
David Blaikie 6f0d522475 Put GEPOperator member function definition in the appropriate .cpp file
Last commit put it in Constants.cpp instead of Operator.cpp

llvm-svn: 237960
2015-05-21 21:17:12 +00:00
David Blaikie 457343dcaa [opaque pointer type] Allow gep_type_iterator to work with the pointee type from the GEP instruction
The raw non-instruction/constant form of this is still relying on being
able to access the pointee type from a pointer type - those will be
cleaned up later. For now, just focus on the cases where the pointee
type is easily accessible.

llvm-svn: 237958
2015-05-21 21:12:43 +00:00
Adrian Prantl 1f599f9f65 IR / debug info: Add a DWOId field to DICompileUnit,
so DWARF skeleton CUs can be expression in IR. A skeleton CU is a
(typically empty) DW_TAG_compile_unit that has a DW_AT_(GNU)_dwo_name and
a DW_AT_(GNU)_dwo_id attribute. It is used to refer to external debug info.

This is a prerequisite for clang module debugging as discussed in
http://lists.cs.uiuc.edu/pipermail/cfe-dev/2014-November/040076.html.
In order to refer to external types stored in split DWARF (dwo) objects,
such as clang modules, we need to emit skeleton CUs, which identify the
dwarf object (i.e., the clang module) by filename (the SplitDebugFilename)
and a hash value, the dwo_id.

This patch only contains the IR changes. The idea is that a CUs with a
non-zero dwo_id field will be emitted together with a DW_AT_GNU_dwo_name
and DW_AT_GNU_dwo_id attribute.

http://reviews.llvm.org/D9488
rdar://problem/20091852

llvm-svn: 237949
2015-05-21 20:37:30 +00:00
David Blaikie f87cc6d3ab [opaque pointer type] Pass explicit pointee type in another case of GEP constant folding
llvm-svn: 237860
2015-05-21 00:06:38 +00:00
David Blaikie 0c28fd7fda [opaque pointer type] Pass explicit type to Load instruction creation in AutoUpgrade
llvm-svn: 237838
2015-05-20 21:46:30 +00:00
David Blaikie 79009f88b4 [opaque pointer type] LoadInst: assert that the explicit type matches the implicit one
llvm-svn: 237830
2015-05-20 20:22:31 +00:00
Pete Cooper 9e1d335697 Change Function::getIntrinsicID() to return an Intrinsic::ID. NFC.
Now that Intrinsic::ID is a typed enum, we can forward declare it and so return it from this method.

This updates all users which were either using an unsigned to store it, or had a now unnecessary cast.

llvm-svn: 237810
2015-05-20 17:16:39 +00:00
Sanjoy Das f999547d11 Dereferenceable, dereferenceable_or_null metadata for loads
Summary:
Introduce dereferenceable, dereferenceable_or_null metadata for loads
with the same semantic as corresponding attributes.

This patch depends on http://reviews.llvm.org/D9253

Patch by Artur Pilipenko!

Reviewers: hfinkel, sanjoy, reames

Reviewed By: sanjoy, reames

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D9365

llvm-svn: 237720
2015-05-19 20:10:19 +00:00
Yaron Keren 004756d366 As r237678 was reverted, this is no longer needed.
llvm-svn: 237687
2015-05-19 11:46:27 +00:00
Yaron Keren d90d6fb53a Fix Visual C++ errors C2784, C2780, C2782 after r237678.
It does not like std::min(unsigned, uint32_t).

llvm-svn: 237683
2015-05-19 11:18:10 +00:00
Pete Cooper a7c0c18c4d Store intrinsic ID by value in Function instead of a string lookup. NFC.
On 64-bit targets, Function has 4-bytes of padding in its struct layout.

This uses the space for the intrinsic ID. It is set and recalculated whenever the function name is set.  This is similar to the current behavior which clears the function from the intrinsic ID cache when its renamed.

The intrinsic cache itself is removed as the only purpose was to speedup calls to getIntrinsicID() which now just reading the new field in the struct.

Reviewed by Duncan.  http://reviews.llvm.org/D9836

llvm-svn: 237642
2015-05-19 00:24:26 +00:00
Pete Cooper 2124113569 Move Function::lookupIntrinsicID to a static method. NFC
llvm-svn: 237641
2015-05-19 00:02:25 +00:00
David Blaikie ff6409d096 Simplify IRBuilder::CreateCall* by using ArrayRef+initializer_list/braced init only
llvm-svn: 237624
2015-05-18 22:13:54 +00:00
Eric Christopher 7a4d1090bc Fix some odd whitespace and formatting errors while making
changes in ConstantsContext.h.

llvm-svn: 237620
2015-05-18 21:49:02 +00:00
Filipe Cabecinhas 11bb8495f6 Extract the load/store type verification to a separate function.
Summary:
Added isLoadableOrStorableType to PointerType.

We were doing some checks in some places, occasionally assert()ing instead
of telling the caller. With this patch, I'm putting all type checking in
the same place for load/store type instructions, and verifying the same
thing every time.

I also added a check for load/store of a function type.

Applied extracted check to Load, Store, and Cmpxcg.

I don't have exhaustive tests for all of these, but all Error() calls in
TypeCheckLoadStoreInst are being tested (in invalid.test).

Reviewers: dblaikie, rafael

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D9785

llvm-svn: 237619
2015-05-18 21:48:55 +00:00
Chen Li 6d8635a743 [Verifier] Assert gc_relocate always return a pointer type
Summary: Add an assertion in verifier.cpp to make sure gc_relocate relocate a gc pointer, and its return type has the same address space with the relocated pointer.

Reviewers: reames, AndyAyers, sanjoy, pgavlin

Reviewed By: pgavlin

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D9695

llvm-svn: 237605
2015-05-18 19:50:14 +00:00
Eric Christopher 97cb56572a While in GlobalValue fix the function(s) that don't follow the
naming convention and update users.

llvm-svn: 237461
2015-05-15 18:20:14 +00:00
Jingyue Wu 154eb5aa1d Add a speculative execution pass
Summary:
This is a pass for speculative execution of instructions for simple if-then (triangle) control flow. It's aimed at GPUs, but could perhaps be used in other contexts. Enabling this pass gives us a 1.0% geomean improvement on Google benchmark suites, with one benchmark improving 33%.

Credit goes to Jingyue Wu for writing an earlier version of this pass.

Patched by Bjarke Roune. 

Test Plan:
This patch adds a set of tests in test/Transforms/SpeculativeExecution/spec.ll
The pass is controlled by a flag which defaults to having the pass not run.

Reviewers: eliben, dberlin, meheff, jingyue, hfinkel

Reviewed By: jingyue, hfinkel

Subscribers: majnemer, jholewinski, llvm-commits

Differential Revision: http://reviews.llvm.org/D9360

llvm-svn: 237459
2015-05-15 17:54:48 +00:00
David Blaikie e7107060f8 [opaque pointer type] Use GlobalVariable::getValueType rather than accessing it through the GV's pointee type
llvm-svn: 237311
2015-05-13 22:54:54 +00:00
Sanjoy Das 9af34eb795 [Safepoints][Verifier] Fix a tautological Assert.
llvm-svn: 237287
2015-05-13 20:11:59 +00:00