| Commit message (Collapse) | Author | Age | Files | Lines | 
| ... |  | 
| | 
| 
| 
| 
| 
| 
| 
|  | 
Reviewed By: MatzeB
Differential Revision: https://reviews.llvm.org/D51495
llvm-svn: 342457
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
We were mapping an instruction every time we saw something we couldn't map
before this. Since each illegal mapping is unique, we only have to do this once.
This makes it so that we don't map illegal instructions when the previous
mapped instruction was illegal.
In CTMark (AArch64), this results in 240 fewer instruction mappings on
average over 619 files in total. The largest improvement is 12576 fewer
mappings in one file, and the smallest is 0. The median improvement is 101
fewer mappings.
llvm-svn: 342405
 | 
| | 
| 
| 
| 
| 
| 
| 
|  | 
extract_subvector with invalid index.""
Fixed the assertion failure.
llvm-svn: 342397
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
std::vector::iterator type may be a pointer, then
iterator::value_type fails to compile since iterator is not a class,
namespace, or enumeration.
Patch by orivej (Orivej Desh)
Differential Revision: https://reviews.llvm.org/D52142
llvm-svn: 342354
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
This is a follow-up suggested in D51630 and originally proposed as an IR transform in D49040.
Copying the motivational statement by @evandro from that patch:
"This transformation helps some benchmarks in SPEC CPU2000 and CPU2006, such as 188.ammp, 
447.dealII, 453.povray, and especially 300.twolf, as well as some proprietary benchmarks. 
Otherwise, no regressions on x86-64 or A64."
I'm proposing to add only the minimum support for a DAG node here. Since we don't have an 
LLVM IR intrinsic for cbrt, and there are no other DAG ways to create a FCBRT node yet, I 
don't think we need to worry about DAG builder, legalization, a strict variant, etc. We 
should be able to expand as needed when adding more functionality/transforms. For reference, 
these are transform suggestions currently listed in SimplifyLibCalls.cpp:
//   * cbrt(expN(X))  -> expN(x/3)
//   * cbrt(sqrt(x))  -> pow(x,1/6)
//   * cbrt(cbrt(x))  -> pow(x,1/9)
Also, given that we bail out on long double for now, there should not be any logical 
differences between platforms (unless there's some platform out there that has pow()
but not cbrt()).
Differential Revision: https://reviews.llvm.org/D51753
llvm-svn: 342348
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
|  | 
CodeGenPrepare has a transform that sinks {lshr, trunc} pairs to make it
easier for the backend to emit fancy extract-bits instructions (e.g UBFX).
Teach it to preserve debug locations and salvage debug values.
llvm-svn: 342319
 | 
| | 
| 
| 
|  | 
llvm-svn: 342293
 | 
| | 
| 
| 
|  | 
llvm-svn: 342285
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
|  | 
extract_subvector with invalid index."
Causes 'isVector() && "Invalid vector type!"' assertion when building
Skia in Chrome.
llvm-svn: 342265
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
This patch fixes the debug info handling for SelectionDAG legalization
of DAG nodes with two results. When an replaced SDNode has more than
one result, transferDbgValues was always copying the SDDbgValue from
the first result and attaching them to all members. In reality
SelectionDAG::ReplaceAllUsesWith() is given an array of SDNodes
(though the type signature doesn't make this obvious (cf. the call
site code in ReplaceNode()).
rdar://problem/44162227
Differential Revision: https://reviews.llvm.org/D52112
llvm-svn: 342264
 | 
| | 
| 
| 
|  | 
llvm-svn: 342247
 | 
| | 
| 
| 
|  | 
llvm-svn: 342245
 | 
| | 
| 
| 
|  | 
llvm-svn: 342241
 | 
| | 
| 
| 
| 
| 
| 
| 
|  | 
invalid index.
Differential Revision: https://reviews.llvm.org/D51831
llvm-svn: 342183
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
operands from inline assembly instructions if they have an associated flag operand.
INLINEASM instructions use extra operands to carry flags. If a register operand is removed without removing the flag operand, then the flags will no longer make sense.
This patch fixes this by preventing the removal when a flag operand is present.
The included test case was generated by MS inline assembly. Longer term maybe we should fix the inline assembly parsing to not generate redundant operands.
Differential Revision: https://reviews.llvm.org/D51829
llvm-svn: 342176
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
|  | 
This was trying to scalarizing a scalar FP type,
resulting in an assert.
Fixes unaligned f64 stack stores for AMDGPU.
llvm-svn: 342132
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
The Technical Reference Manuals for these two CPUs state that branching
to an unaligned 32-bit instruction incurs an extra pipeline reload
penalty. That's bad.
This also enables the optimization at -Os since it costs on average one
byte per loop in return for 1 cycle per iteration, which is pretty good
going.
llvm-svn: 342127
 | 
| | 
| 
| 
|  | 
llvm-svn: 342095
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
The output of splitLargeGEPOffsets does not appear to be deterministic because
of the way that we iterate over a DenseMap. I've changed it to a MapVector for
consistent output.
The test here isn't particularly great, only showing a consmetic difference in
output. The original reproducer is much larger but show a diffierence in
instruction ordering, leading to different codegen.
Differential Revision: https://reviews.llvm.org/D51851
llvm-svn: 342043
 | 
| | 
| 
| 
| 
| 
| 
| 
|  | 
UpdateNodeOperands returning an existing node instead of updating.
I suspect this became unecessary when the CSE of mgather was fixed in r338080. It may still be possible to hit this if we widen the element type of a gather outside of type legalization and the promote the mask of a separate gather node so they become the same. But that seems pretty unlikely.
llvm-svn: 342022
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
Since the outliner is a module pass, it doesn't get codegen size remarks like
the other codegen passes do. This adds size remarks *to* the outliner.
This is kind of a workaround, so it's peppered with FIXMEs; size remarks
really ought to not ever be handled by the pass itself. However, since the
outliner is the only "MachineModulePass", this works for now. Since the
entire purpose of the MachineOutliner is to produce code size savings, it
really ought to be included in codgen size remarks.
If we ever go ahead and make a MachineModulePass (say, something similar to
MachineFunctionPass), then all of this ought to be moved there.
llvm-svn: 342009
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
Summary: Initial support for nsw, nuw and exact flags in MI
Reviewers: spatel, hfinkel, wristow
Reviewed By: spatel
Subscribers: nlopes
Differential Revision: https://reviews.llvm.org/D51738
llvm-svn: 341996
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
Summary:
D31439 changed the semantics of dbg.declare to take the address of a
variable as the first argument, making it indirect.  It specifically
updated FastISel for this change here:
https://reviews.llvm.org/D31439#change-WVArzi177jPl
GlobalISel needs to follow suit, or else it will be missing a level of
indirection in the generated debuginfo.  This problem was seen in a Rust
debuginfo test on aarch64, since GlobalISel is used at -O0 for aarch64.
https://github.com/rust-lang/rust/issues/49807
https://bugzilla.redhat.com/show_bug.cgi?id=1611597
https://bugzilla.redhat.com/show_bug.cgi?id=1625768
Reviewers: dblaikie, aprantl, t.p.northover, javed.absar, rnk
Reviewed By: rnk
Subscribers: #debug-info, rovka, kristof.beyls, JDevlieghere, llvm-commits, tstellar
Differential Revision: https://reviews.llvm.org/D51749
llvm-svn: 341969
 | 
| | 
| 
| 
| 
| 
| 
|  | 
Just some tidy-up. Pull the mapper stuff into `populateMapper`. This makes it
a bit easier to read what's going on in `runOnModule`.
llvm-svn: 341959
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
This adds per-function size remarks to codegen, similar to what we have in the
IR layer as of r341588. This only impacts MachineFunctionPasses.
This does the same thing, but for `MachineInstr`s instead of just
`Instructions`. After this, when a `MachineFunctionPass` modifies the number of
`MachineInstr`s in the function it ran on, you'll get a remark.
To enable this, use the size-info analysis remark as before.
llvm-svn: 341876
 | 
| | 
| 
| 
| 
| 
|  | 
Loop's getBlocks returns an ArrayRef.
llvm-svn: 341821
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
This already worked if only one register piece was used,
but didn't if a type was split into multiple, unequal
sized pieces.
Fixes not splitting 3i16/v3f16 into two registers for
AMDGPU.
This will also allow fixing the ABI for 16-bit vectors
in a future commit so that it's the same for all subtargets.
llvm-svn: 341801
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
condition operand
This is the DAG equivalent of D51433.
If we know we're not using all vector lanes, use that knowledge to potentially simplify a vselect condition.
The reduction/horizontal tests show that we are eliminating AVX1 operations on the upper half of 256-bit 
vectors because we don't need those anyway.
I'm not sure what the pr34592 test is showing. That's run with -O0; is SimplifyDemandedVectorElts supposed 
to be running there?
Differential Revision: https://reviews.llvm.org/D51696
llvm-svn: 341762
 | 
| | 
| 
| 
|  | 
llvm-svn: 341740
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
This patch removes addBlockByrefAddress(), it is dead code as far as
clang is concerned: Every byref block capture is emitted with a
complex expression that is equivalent to what this function does.
rdar://problem/31629055
Differential Revision: https://reviews.llvm.org/D51763
llvm-svn: 341737
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
Summary:
MSVC and LLD sort sections ASCII-betically, so we need to use section
names that sort between .CRT$XCA (the start) and .CRT$XCU (the default
priority).
In the general case, use .CRT$XCT12345 as the section name, and let the
linker sort the zero-padded digits.
Users with low priorities typically want to initialize as early as
possible, so use .CRT$XCA00199 for prioties less than 200. This number
is arbitrary.
Implements PR38552.
Reviewers: majnemer, mstorsjo
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D51820
llvm-svn: 341727
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
Summary:
Extend LDV so that stack slot offsets for spilled sub-registers
are added to the emitted debug locations. This is accomplished
by querying InstrInfo::getStackSlotRange().
With this change, LDV will add a DW_OP_plus_uconst operation to
the expression if a sub-register is spilled. Later on, PEI will
add an offset operation for the stack slot, meaning that we will
get expressions of the forms:
 * {DW_OP_constu #fp-offset, DW_OP_minus,
    DW_OP_plus_uconst #subreg-offset}
 * {DW_OP_plus_const #fp-offset,
    DW_OP_minus, DW_OP_plus_uconst #subreg-offset}
The two offset operations should ideally be merged.
Reviewers: rnk, aprantl, stoklund
Reviewed By: aprantl
Subscribers: dblaikie, bjope, nemanjai, JDevlieghere, llvm-commits
Tags: #debug-info
Differential Revision: https://reviews.llvm.org/D51612
llvm-svn: 341659
 | 
| | 
| 
| 
| 
| 
| 
| 
|  | 
Add support for bitcasts from float type to an integer type of the same element bitwidth.
There maybe cases where we need to support different widths (e.g. as SSE __m128i is treated as v2i64) - but I haven't seen cases of this in the wild yet.
llvm-svn: 341652
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
The MCID::Flag enumeration now has more than 32 items, this means that
the hasPropertyBundle argument 'Mask' can overflow.
This patch changes the argument to be 64 bits instead.
Patch by Mikael Nilsson.
Differential Revision: https://reviews.llvm.org/D51596
llvm-svn: 341536
 | 
| | 
| 
| 
| 
| 
| 
| 
|  | 
In DwarfDebug::collectEntityInfo(), if the label entity is processed in
DbgLabels list, it means the label is not optimized out. There is no
need to generate debug info for it with null position.
llvm-svn: 341513
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
This was proposed as an IR transform in D49306, but it was not clearly justifiable as a canonicalization. 
Here, we only do the transform when the target tells us that sqrt can be lowered with inline code.
This is the basic case. Some potential enhancements are in the TODO comments:
1. Generalize the transform for other exponents (allow more than 2 sqrt calcs if that's really cheaper).
2. If we have less fast-math-flags, generate code to avoid -0.0 and/or INF.
3. Allow the transform when optimizing/minimizing size (might require a target hook to get that right).
Note that by default, x86 converts single-precision sqrt calcs into sqrt reciprocal estimate with 
refinement. That codegen is controlled by CPU attributes and can be manually overridden. We have plenty 
of test coverage for that already, so I didn't bother to include extra testing for that here. AArch uses 
its full-precision ops in all cases (not sure if that's the intended behavior or not, but that should 
also be covered by existing tests).
Differential Revision: https://reviews.llvm.org/D51630 
llvm-svn: 341481
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
Normalize common kinds of DWARF sub-expressions to make debug info
encoding a bit more compact:
  DW_OP_constu [X < 32] -> DW_OP_litX
  DW_OP_constu [all ones] -> DW_OP_lit0, DW_OP_not (64-bit only)
Differential revision: https://reviews.llvm.org/D51640
llvm-svn: 341457
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
This removes the FrameAccess struct that was added to the interface
in D51537, since the PseudoValue from the MachineMemoryOperand
can be safely casted to a FixedStackPseudoSourceValue.
Reviewers: MatzeB, thegameg, javed.absar
Reviewed By: thegameg
Differential Revision: https://reviews.llvm.org/D51617
llvm-svn: 341454
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
In lib/CodeGen/LiveDebugVariables.cpp, it uses std::prev(MBBI) to
get DebugValue's SlotIndex. However, the previous instruction may be
also a debug instruction. It could not use a debug instruction to query
SlotIndex in mi2iMap.
Scan all debug instructions and use the first debug instruction to query
SlotIndex for following debug instructions. Only handle DBG_VALUE in
handleDebugValue().
Differential Revision: https://reviews.llvm.org/D50621
llvm-svn: 341446
 | 
| | 
| 
| 
|  | 
llvm-svn: 341392
 | 
| | 
| 
| 
| 
| 
| 
| 
|  | 
Fix remaining cases not committed in https://reviews.llvm.org/D49574
Differential Revision: https://reviews.llvm.org/D50659
llvm-svn: 341380
 | 
| | 
| 
| 
|  | 
llvm-svn: 341317
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
|  | 
A condition in isSpillInstruction() updates a small vector rather
than the 'FI' by-ref parameter, which was used in a subsequent
call to 'isSpillSlotObjectIndex()'. This patch fixes the condition
to check the FIs in the vector instead.
llvm-svn: 341305
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
For instructions that spill/fill to and from multiple frame-indices
in a single instruction, hasStoreToStackSlot and hasLoadFromStackSlot
should return an array of accesses, rather than just the first encounter
of such an access.
This better describes FI accesses for AArch64 (paired) LDP/STP
instructions.
Reviewers: t.p.northover, gberry, thegameg, rengolin, javed.absar, MatzeB
Reviewed By: MatzeB
Differential Revision: https://reviews.llvm.org/D51537
llvm-svn: 341301
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
|  | 
This reverts commit 8f548ff2a1819e1bc051e8218584f1a3d2cf178a.
buildbot failure in LLVM on clang-ppc64be-linux
http://lab.llvm.org:8011/builders/clang-ppc64le-linux/builds/19765
llvm-svn: 341290
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
In lib/CodeGen/LiveDebugVariables.cpp, it uses std::prev(MBBI) to
get DebugValue's SlotIndex. However, the previous instruction may be
also a debug instruction. It could not use a debug instruction to query
SlotIndex in mi2iMap.
Scan all debug instructions and use the first debug instruction to query
SlotIndex for following debug instructions. Only handle DBG_VALUE in
handleDebugValue().
Differential Revision: https://reviews.llvm.org/D50621
llvm-svn: 341289
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
Summary:
A follow-up for D49266 / rL337166 + D49497 / rL338044.
This is still the same pattern to check for the [lack of]
signed truncation, but in this case the constants and the predicate
are negated.
https://rise4fun.com/Alive/BDV
https://rise4fun.com/Alive/n7Z
Reviewers: spatel, craig.topper, RKSimon, javed.absar, efriedma, dmgreen
Reviewed By: spatel
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D51532
llvm-svn: 341287
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
Summary:
Currently, the SafeStack analysis disallows out-of-bounds writes but not
out-of-bounds reads for mem intrinsics like llvm.memcpy. This could
cause leaks of pointers to the safe stack by leaking spilled registers/
frame pointers. Check for allocas used as source or destination pointers
to mem intrinsics.
Reviewers: eugenis
Reviewed By: eugenis
Subscribers: pcc, llvm-commits, kcc
Differential Revision: https://reviews.llvm.org/D51334
llvm-svn: 341116
 | 
| | 
| 
| 
|  | 
llvm-svn: 341103
 | 
| | 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
| 
|  | 
With r295105, some 'noreturn' blocks (those that don't return and have no
successors) may be merged.
If such blocks' predecessors have different outgoing offset or register, don't
report an error in CFIInstrInserter verify().
Thanks to Vlad Tsyrklevich for reporting the issue.
Differential Revision: https://reviews.llvm.org/D51161
llvm-svn: 341087
 |