Reid Kleckner [Wed, 31 May 2017 19:23:09 +0000 (19:23 +0000)]
[IR] Add additional addParamAttr/removeParamAttr to AttributeList API
Summary:
Fairly straightforward patch to fill in some of the holes in the
attributes API with respect to accessing parameter/argument attributes.
The patch aims to step further towards encapsulating the
idx+FirstArgIndex pattern to access these attributes to within the
AttributeList.
Craig Topper [Wed, 31 May 2017 19:01:11 +0000 (19:01 +0000)]
[TableGen] Make Record::getValueAsString and getValueAsListOfStrings return StringRefs instead of std::string
Internally both these methods just return the result of getValue on either a StringInit or a CodeInit object. In both cases this returns a StringRef pointing to a string allocated in the BumpPtrAllocator so its not going anywhere. So we can just pass that StringRef along.
This is a fairly naive patch that targets just the build failures caused by this change. There's additional work that can be done to avoid creating std::string at call sites that still think getValueAsString returns a std::string. I'll try to clean those up in future patches.
Teresa Johnson [Wed, 31 May 2017 18:58:11 +0000 (18:58 +0000)]
[ThinLTO] Reduce unnecessary map lookups during combined summary write
Summary:
Don't assign values to undefined references, simply don't emit those
reference edges as they are not useful (we were already not emitting
call edges to undefined refs).
Also, streamline the later lookup of value ids when writing the
summaries, by combining the check for value id existence with the access
of that value id.
Nirav Dave [Wed, 31 May 2017 18:43:17 +0000 (18:43 +0000)]
[ScheduleDAG] Deal with already scheduled loads in ScheduleDAG.
Summary:
If we attempt to unfold an SUnit in ScheduleDAG that results in
finding an already scheduled load, we must should abort the
unfold as it will not improve scheduling.
This adds a callback to the LLVMTargetMachine that lets target indicate
that they do not pass the machine verifier checks in all cases yet.
This is intended to be a temporary measure while the targets are fixed
allowing us to enable the machine verifier by default with
EXPENSIVE_CHECKS enabled!
Zaara Syeda [Wed, 31 May 2017 17:12:38 +0000 (17:12 +0000)]
[PPC] Inline expansion of memcmp
This patch does an inline expansion of memcmp.
It changes the memcmp library call into an inline expansion when the size is
known at compile time and is under a target specified threshold.
This expansion is implemented in CodeGenPrepare and expands into straight line
code. The target specifies a maximum load size and the expansion works by using
this size to load the two sources, compare, and exit early if a difference is
found. It also has a special case when the memcmp result is used in a compare
to zero equality.
Mark Searles [Wed, 31 May 2017 16:44:23 +0000 (16:44 +0000)]
[AMDGPU] Fix bugs in new waitcnt pass. Add test.
- new waitcnt pass remains off by default; -enable-si-insert-waitcnts=1 to enable it
- fix handling of PERMUTE ops
- fix insertion of waitcnt instrs at function begin/end ( port of analogous code that was added to old waitcnt pass )
- add new test
Summary:
Expanding the loop idiom test for memcpy to also recognize unordered atomic memcpy.
The only difference for recognizing
an unordered atomic memcpy and instead of a normal memcpy is
that the loads and/or stores involved are unordered atomic operations.
Background: http://lists.llvm.org/pipermail/llvm-dev/2017-May/112779.html
Tony Jiang [Wed, 31 May 2017 13:09:57 +0000 (13:09 +0000)]
[PowerPC] Fix a performance bug for PPC::XXPERMDI.
There are some VectorShuffle Nodes in SDAG which can be selected to XXPERMDI
Instruction, this patch recognizes them and does the selection to improve
the PPC performance.
This patch builds upon https://reviews.llvm.org/rL302810 to add
handling for bitwise logical operations in general purpose registers.
The idea is to keep the values in GPRs as long as possible - only
extracting them to a condition register bit when no further operations
are to be done.
Zachary Turner [Wed, 31 May 2017 04:17:13 +0000 (04:17 +0000)]
[ObjectYAML] Split CodeViewYAML into 3 pieces.
The code was a mess and disorganized due to the sheer amount
of it being in one file. So I'm splitting this into three files.
One for CodeView types, one for CodeView symbols, and one for
CodeView debug subsections. NFC.
[CFLAA] Add missing break; note things are broken.
Thanks to Galina Kistanova for finding the missing break!
When trying to make a test for this, I realized our logic for handling
extractvalue/insertvalue/... is somewhat broken. This makes constructing
a test-case for this missing break nontrivial.
Matthias Braun [Wed, 31 May 2017 02:11:10 +0000 (02:11 +0000)]
X86FrameLowering: No need to mark FP as live-in everywhere
The frame pointer (when used as frame pointer) is a reserved register.
We do not track liveness of reserved registers and hence do not need to
add them to the basic block livein lists.
Vedant Kumar [Wed, 31 May 2017 01:42:55 +0000 (01:42 +0000)]
Mark a test as requiring a default triple
This test assumes that llc can infer a default triple. I'm not sure why
exactly, but the Verify MachineInstrs bot requires tests to be explicit
about this dependency.
This commit follows the lead from r248452 and adds in 'REQUIRES:
default_triple' to omit-empty.ll.
Matthias Braun [Wed, 31 May 2017 01:21:35 +0000 (01:21 +0000)]
ARM: Fix cmpxchg O0 expansion
This is the equivalent of r304048 for ARM:
- Rewrite livein calculation to use the computeLiveIns() helper
function. This is slightly less efficient but easier to reason about
and doesn't unnecessarily add pristine and reserved registers[1]
- Zero the status register at the beginning of the loop to make sure it
has a defined value.
- Remove kill flags of values that need to stay alive throughout the loop.
[1] An upcoming commit of mine will tighten the MachineVerifier to catch
these.
Zachary Turner [Wed, 31 May 2017 01:08:36 +0000 (01:08 +0000)]
[ObjectYAML] Clean up the CodeView headers a bit.
CodeViewYAML.h attempts to hide the details of many of the
CodeView yaml structures and types, but at the same time it
exposes the mapping traits for them to external users of the
header.
This patch just hides these in the implementation files so that
the interface is kept as simple as possible.
Zachary Turner [Tue, 30 May 2017 23:50:44 +0000 (23:50 +0000)]
[CodeView] Move CodeView symbol yaml logic to ObjectYAML.
This continues the effort to get the CodeView YAML parsing logic
into ObjectYAML. After this patch, the only missing piece will
be the CodeView debug symbol subsections.
Tim Shen [Tue, 30 May 2017 22:26:52 +0000 (22:26 +0000)]
[AntiDepBreaker] Revert r299124 and add a test.
Summary:
AntiDepBreaker intends to add all live-outs, including the implicit
CSRs, in StartBlock. r299124 was done without understanding that
intention.
Now with the live-ins propagated correctly (D32464), we can revert this change.
Zachary Turner [Tue, 30 May 2017 21:53:05 +0000 (21:53 +0000)]
[CodeView] Move CodeView YAML code to ObjectYAML.
This is the beginning of an effort to move the codeview yaml
reader / writer into ObjectYAML so that it can be shared.
Currently the only consumer / producer of CodeView YAML is
llvm-pdbdump, but CodeView can exist outside of PDB files, and
indeed is put into object files and passed to the linker to
produce PDB files. Furthermore, there are subtle differences
in the types of records that show up in object file CodeView
vs PDB file CodeView, but they are otherwise 99% the same.
By having this code in ObjectYAML, we can have llvm-pdbdump
reuse this code, while teaching obj2yaml and yaml2obj to use
this syntax for dealing with object files that can contain
CodeView.
This patch only adds support for CodeView type information
to ObjectYAML. Subsequent patches will add support for
CodeView symbol information.
Matthias Braun [Tue, 30 May 2017 21:36:41 +0000 (21:36 +0000)]
TargetPassConfig: Keep a reference to an LLVMTargetMachine; NFC
TargetPassConfig is not useful for targets that do not use the CodeGen
library, so we may just as well store a pointer to an
LLVMTargetMachine instead of just to a TargetMachine.
While at it, also change the constructor to take a reference instead of a
pointer as the TM must not be nullptr.
Quentin Colombet [Tue, 30 May 2017 20:53:06 +0000 (20:53 +0000)]
[Localizer] Don't trick to be smart for the insertion point
There is no guarantee that the first use of a constant that is traversed
is actually the first in the related basic block. Thus, if we use that
as the insertion point we may end up with definitions that don't
dominate there use.
Matthew Simpson [Tue, 30 May 2017 19:55:57 +0000 (19:55 +0000)]
[LV] Reapply r303763 with fix for PR33193
r303763 caused build failures in some out-of-tree tests due to an assertion in
TTI. The original patch updated cost estimates for induction variable update
instructions marked for scalarization. However, it didn't consider that the
incoming value of an induction variable phi node could be a cast instruction.
This caused queries for cast instruction costs with a mix of vector and scalar
types. This patch includes a fix for cast instructions and the test case from
PR33193.
The fix was suggested by Jonas Paulsson <paulsson@linux.vnet.ibm.com>.
Reference: https://bugs.llvm.org/show_bug.cgi?id=33193
Original Differential Revision: https://reviews.llvm.org/D33457
I reproduced the failure locally (without ThinLTO), reverted the commit, rebuilt the stage1 clang, rebuilt the stage2 llvm-tblgen tool, and found that the crash disappears when the commit is reverted. Here is the stack trace:
Keno Fischer [Tue, 30 May 2017 18:28:30 +0000 (18:28 +0000)]
[Cloning] Take another pass at properly cloning debug info
Summary:
In rL302576, DISubprograms gained the constraint that a !dbg attachments to functions must
have a 1:1 mapping to DISubprograms. As part of that change, the function cloning support
was adjusted to attempt to enforce this invariant during cloning. However, there
were several problems with the implementation. Part of these were fixed in rL304079.
However, there was a more fundamental problem with these changes, namely that it
bypasses the matadata value map, causing the cloned metadata to be a mix of metadata
pointing to the new suprogram (where manual code was added to fix those up) and the
old suprogram (where this was not the case). This mismatch could cause a number of
different assertion failures in the DWARF emitter. Some of these are given at
https://github.com/JuliaLang/julia/issues/22069, but some others have been observed
as well. Attempt to rectify this by partially reverting the manual DI metadata fixup,
and instead using the standard value map approach. To retain the desired semantics
of not duplicating the compilation unit and inlined subprograms, explicitly freeze
these in the value map.
[Hexagon] Improve code generation for 32x32-bit multiplication
For multiplications of 64-bit values (giving 64-bit result), detect
cases where the arguments are sign-extended 32-bit values, on a per-
operand basis. This will allow few patterns to match a wider variety
of combinations in which extensions can occur.
Zachary Turner [Tue, 30 May 2017 17:13:33 +0000 (17:13 +0000)]
[CodeView] Add more DebugSubsection implementations.
This adds implementations for Symbols and FrameData, and renames
the existing codeview::StringTable class to conform to the
DebugSectionStringTable convention.
Craig Topper [Tue, 30 May 2017 17:12:18 +0000 (17:12 +0000)]
[SelectionDAG] Remove special case for ISD::FPOWI from the strict FP intrinsic handling.
This code was compensating for FPOWI defaulting to Legal and many targets not changing it to Expand. This was fixed in r304215 to default to Expand so this special handling should no longer be necessary.
[AMDGPU] Allow SDWA in instructions with immediates and SGPRs
An encoding does not allow to use SDWA in an instruction with
scalar operands, either literals or SGPRs. That is however possible
to copy these operands into a VGPR first.
Several copies of the value are produced if multiple SDWA conversions
were done. To cleanup MachineLICM (to hoist copies out of loops),
MachineCSE (to remove duplicate copies) and SIFoldOperands (to replace
SGPR to VGPR copy with immediate copy right to the VGPR) runs are added
after the SDWA pass.
Craig Topper [Tue, 30 May 2017 15:27:55 +0000 (15:27 +0000)]
[SelectionDAG] Set ISD::FPOWI to Expand by default
Summary:
Currently FPOWI defaults to Legal and LegalizeDAG.cpp turns Legal into Expand for this opcode because Legal is a "lie".
This patch changes the default for this opcode to Expand and removes the hack from LegalizeDAG.cpp. It also removes all the code in the targets that set this opcode to Expand themselves since they can just rely on the default.
Ulrich Weigand [Tue, 30 May 2017 10:15:16 +0000 (10:15 +0000)]
[SystemZ] Add decimal floating-point instructions
This adds assembler / disassembler support for the decimal
floating-point instructions. Since LLVM does not yet have
support for decimal float types, these cannot be used for
codegen at this point.
This adds assembler / disassembler support for the hexadecimal
floating-point instructions. Since the Linux ABI does not use
any hex float data types, these are not useful for codegen.
Oliver Stannard [Tue, 30 May 2017 09:37:11 +0000 (09:37 +0000)]
[MC] Fix constant pools with DenseMap sentinel values
The MC ConstantPool class uses a DenseMap to track generated constants, with
the int64_t value of the constant as the key. This fails when values of
0x7fffffffffffffff or 0x7ffffffffffffffe are inserted into the constant pool, as
these are sentinel values for DenseMap.
The fix is to use std::map instead, which doesn't use sentinel values.
Zoran Jovanovic [Tue, 30 May 2017 09:33:43 +0000 (09:33 +0000)]
[mips] Expansion of LI.S and LI.D
Author: smaksimovic
Reviewers: dsanders sdardis
Introduces LI.S and LI.D pseudo instructions with floating point operands.
Differential Revision: https://reviews.llvm.org/D14390
Craig Topper [Mon, 29 May 2017 21:49:37 +0000 (21:49 +0000)]
[TableGen] Use StringMap instead of DenseMap<StringRef> to unique CodeInit and StringInit objects. Override the allocator to keep using the BumpPtrAllocator. NFCI
StringMap is better suited to mapping strings than a DenseMap.
Craig Topper [Mon, 29 May 2017 21:49:34 +0000 (21:49 +0000)]
[TableGen] Introduce DagInit::getArgs that returns an ArrayRef. Use it to fix 80 column violations in arg_begin/arg_end. Remove DagInit::args and use getArgs instead. NFC
Benjamin Kramer [Mon, 29 May 2017 16:25:20 +0000 (16:25 +0000)]
[X86] Don't fold away the memory operand of an xchg.
xchg with a mem operand has different locking semantics. If we unfold it
into a xchg r,r we will loose the implicit lock. Likewise we never want
to fold a register xchg into a memory one as it would be a lot slower.
Ayal Zaks [Mon, 29 May 2017 15:36:23 +0000 (15:36 +0000)]
[Docs] Add VectorizationPlan to docs/Proposals.
Following the request made in https://reviews.llvm.org/D32871, the
general documentation of the Vectorization Plan is hereby placed
under docs/Proposals.