The key idea is to lower COPY nodes populating EFLAGS by scanning the
uses of EFLAGS and introducing dedicated code to preserve the necessary
state in a GPR. In the vast majority of cases, these uses are cmovCC and
jCC instructions. For such cases, we can very easily save and restore
the necessary information by simply inserting a setCC into a GPR where
the original flags are live, and then testing that GPR directly to feed
the cmov or conditional branch.
However, things are a bit more tricky if arithmetic is using the flags.
This patch handles the vast majority of cases that seem to come up in
practice: adc, adcx, adox, rcl, and rcr; all without taking advantage of
partially preserved EFLAGS as LLVM doesn't currently model that at all.
There are a large number of operations that techinaclly observe EFLAGS
currently but shouldn't in this case -- they typically are using DF.
Currently, they will not be handled by this approach. However, I have
never seen this issue come up in practice. It is already pretty rare to
have these patterns come up in practical code with LLVM. I had to resort
to writing MIR tests to cover most of the logic in this pass already.
I suspect even with its current amount of coverage of arithmetic users
of EFLAGS it will be a significant improvement over the current use of
pushf/popf. It will also produce substantially faster code in most of
the common patterns.
This patch also removes all of the old lowering for EFLAGS copies, and
the hack that forced us to use a frame pointer when EFLAGS copies were
found anywhere in a function so that the dynamic stack adjustment wasn't
a problem. None of this is needed as we now lower all of these copies
directly in MI and without require stack adjustments.
Lots of thanks to Reid who came up with several aspects of this
approach, and Craig who helped me work out a couple of things tripping
me up while working on this.
Differential Revision: https://reviews.llvm.org/D45146
llvm-svn: 329657
Summary:
Another clean up, following D43208.
Interleaved memory access analysis/optimization has nothing to do with vectorization legality. It doesn't really belong there. On the other hand, cost model certainly has to know about it.
In principle, vectorization should proceed like Legality ==> Optimization ==> CostModel ==> CodeGen, and this change just does that,
by moving the interleaved access analysis/decision out of Legal, and run it just before CostModel object is created.
After this, I can move LoopVectorizationLegality and Hints/Requirements classes into it's own header file, making it shareable within Transform tree. I have the patch already but I don't want to mix with this change. Eventual goal is to move to Analysis tree, but I first need to move RecurrenceDescriptor/InductionDescriptor from Transform/Util/LoopUtil.* to Analysis.
Reviewers: rengolin, hfinkel, mkuper, dcaballe, sguggill, fhahn, aemerson
Reviewed By: rengolin
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D45072
llvm-svn: 329645
Summary:
SSAUpdater is a bottleneck in JumpThreading, and this patch improves the
situation by using SSAUpdaterBulk instead.
Compile time impact: no noticable changes on CTMark, a big improvement
on the test from PR16756.
Reviewers: dberlin, davide, MatzeB
Subscribers: llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D44282
llvm-svn: 329644
Summary:
SSAUpdater is a bottleneck in a number of passes, and one of the reasons
is that it performs a lot of unnecessary computations (DT/IDF) over and
over again. This patch adds a new SSAUpdaterBulk that uses existing DT
and avoids recomputing IDF when possible.
Reviewers: dberlin, davide, MatzeB
Subscribers: llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D44282
llvm-svn: 329643
The caching walker used to hold its own caches, which made its `reset()`
function meaningful. Since caching has been moved out of it, there's no
reason to continue to have these cache-related methods.
Similarly, the EXPENSIVE_CHECKS block that's getting removed used to
rerun the query with caching disabled. Since that's how we always do
queries now, it's redundant.
llvm-svn: 329638
Lower is slightly odd. It often doesn't change the type but the lowerings
do use the new type to decide what code to create. Treat it like a mutation
but provide convenience functions that re-use the existing type.
Re-uses the existing tests:
test/CodeGen/AArch64/GlobalISel/legalize-rem.mir
test/CodeGen/AArch64/GlobalISel//legalize-mul.mir
test/CodeGen/AArch64/GlobalISel//legalize-cmpxchg-with-success.mir
llvm-svn: 329623
Fix PR36484, as suggested:
<quote>
during moves, mark the direct users of the erased things that were phis as "not to be optimized"
<quote>
llvm-svn: 329621
1. Remove max_scratch_backing_memory_byte_size from kernel header
2. Make it a reserved field
3. Ignore it while parsing assembly for backwards compatibility
4. Bump up minor version of kernel header
Differential Revision: https://reviews.llvm.org/D45452
llvm-svn: 329620
LowerIntUnary as its name says has an assert for integer types. But for the bitcast case one side might be an FP type.
Rather than making sure the function really works for fp types and renaming it. Just do really basic splitting directly. The LowerIntUnary has the advantage that it can peek through BUILD_VECTOR because every other call is during Lowering. But these calls are during legalization and will be followed by a DAG combine round.
Revert some change to LowerVectorIntUnary that were originally made just to make these two calls work even in pure integer cases.
This was found purely by compiling the avx512f-builtins.c test from clang so I've copied over the offending function from that.
llvm-svn: 329616
This is a code size win in code that takes offseted addresses
frequently, such as C++ constructors that typically need to compute
an offseted address of a vtable. It reduces the size of Chromium for
Android's .text section by 46KB, or 56KB with ThinLTO (which exposes
more opportunities to use a direct access rather than a GOT access).
Because the addend range is limited in COFF and Mach-O, this is
enabled for ELF only.
Differential Revision: https://reviews.llvm.org/D45199
llvm-svn: 329611
Summary:
r327219 added wrappers to std::sort which randomly shuffle the container before sorting.
This will help in uncovering non-determinism caused due to undefined sorting
order of objects having the same key.
To make use of that infrastructure we need to invoke llvm::sort instead of std::sort.
Note: This patch is one of a series of patches to replace *all* std::sort to llvm::sort.
Refer the comments section in D44363 for a list of all the required patches.
Reviewers: sunfish, RKSimon
Reviewed By: sunfish
Subscribers: jfb, dschuff, sbc100, jgravelle-google, aheejin, llvm-commits
Differential Revision: https://reviews.llvm.org/D44873
llvm-svn: 329607
In somes cases fast-isel fails to remove the and/shifts and uses blends or conditional moves.
But once masking gets involved, fast-isel aborts on the mask portion and we DAG combine more thorougly.
llvm-svn: 329604
This allows MachineMemOperand::getSize()'s result to be fed directly into
MachineMemOperand::MachineMemOperand() without a narrowing type conversion
warning.
llvm-svn: 329602
building.
https://reviews.llvm.org/D45067
This change attempts to do two things:
1) It separates out the state that is stored in the
MachineIRBuilder(InsertionPt, MF, MRI, InsertFunction etc) into a
separate object called MachineIRBuilderState.
2) Add the ability to constant fold operations while building instructions
(optionally). MachineIRBuilder is now refactored into a MachineIRBuilderBase
which contains lots of non foldable build methods and their implementation.
Instructions which can be constant folded/transformed are now in a class
called FoldableInstructionBuilder which uses CRTP to use the implementation
of the derived class for buildBinaryOps. Additionally buildInstr in the derived
class can be used to implement other kinds of transformations.
Also because of separation of state, given a MachineIRBuilder in an API,
if one wishes to use another MachineIRBuilder, a new one can be
constructed from the state locally. For eg,
void doFoo(MachineIRBuilder &B) {
MyCustomBuilder CustomB(B.getState());
// Use CustomB for building.
}
reviewed by : aemerson
llvm-svn: 329596
While it appears to be correct information based on Intel's optimization manual and Agner's data, it causes perf regressions on a couple of the benchmarks in our internal list.
llvm-svn: 329593
Author: Samuel Pitoiset
ds_read_b128 and ds_write_b128 have been recently enabled
under the amdgpu-ds128 option because the performance benefit
is unclear.
Though, using 128-bit loads/stores for the local address space
appears to introduce regressions in tessellation shaders. Not
sure what is broken, but as ds_read_b128/ds_write_b128 are not
enabled by default, just introduce a global option and enable
128-bit only if requested (until it's fixed/used correctly).
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105464
llvm-svn: 329591
This patch teaches llvm-mca how to parse code comments in search for special
"markers" used to select regions of code.
Example:
# LLVM-MCA-BEGIN My Code Region
....
# LLVM-MCA-END
The MCAsmLexer now delegates to an object of class MCACommentParser (i.e. an
AsmCommentConsumer) the parsing of code comments to search for begin/end code
region markers.
A comment starting with substring "LLVM-MCA-BEGIN" marks the beginning of a new
region of code. A comment starting with substring "LLVM-MCA-END" marks the end
of the last region.
This implementation doesn't allow regions to overlap. Each region can have a
optional description; internally, each region is identified by a range of source
code locations (SMLoc).
MCInst objects are added to a region R only if the source location for the
MCInst is in the range of locations specified by R.
By default, the tool allocates an implicit "Default" code region which contains
every source location. See new tests llvm-mca-marker-*.s for a few examples.
A new Backend object is created for every region. So, the analysis is conducted
on every parsed code region. The final report is the union of the reports
generated for every code region. Note that empty regions are skipped.
Special "[#] Code Region - ..." strings are used in the report to mark the
portion which is specific to a code region only. For example, see
llvm-mca-markers-5.s.
Differential Revision: https://reviews.llvm.org/D45433
llvm-svn: 329590
Summary:
This fixes AMDGPU GlobalISel test failures when enabling the AMDGPU
target without any other targets that use GlobalISel.
Reviewers: arsenm
Subscribers: kzhuravl, wdng, nhaehnle, yaxunl, rovka, kristof.beyls, dstuttard, tpr, llvm-commits, t-tye
Differential Revision: https://reviews.llvm.org/D45353
llvm-svn: 329588
Without the fast math flags, the llvm.experimental.vector.reduce.fadd/fmul intrinsic expansions must be expanded in order.
This patch scalarizes the reduction, applying the accumulator at the start of the sequence: ((((Acc + Scl[0]) + Scl[1]) + Scl[2]) + ) ... + Scl[NumElts-1]
Differential Revision: https://reviews.llvm.org/D45366
llvm-svn: 329585
Summary:
The option is helpful for large projects where it's not feasible to specify sources which
user would like to see in the report. Instead, it allows to black-list specific sources via
regular expressions (e.g. now it's possible to skip all files that have "test" in its name).
This also partially fixes https://bugs.llvm.org/show_bug.cgi?id=34277
Reviewers: vsk, morehouse, liaoyuke
Reviewed By: vsk
Subscribers: kcc, mgorny, llvm-commits
Differential Revision: https://reviews.llvm.org/D43907
llvm-svn: 329581
This patch fixes an issue exposed on the SystemZ build bots when committing
https://reviews.llvm.org/rL327856. The hoisting was temporarily disabled with
an option. This patch now re-enables hoisting and checks that we only hoist a
store instruction when all its operands are either constant caller preserved
registers or immediates.
Differential Revision: https://reviews.llvm.org/D45286
llvm-svn: 329577
Summary:
If an input DICompileUnit is completely empty (e.g., the result of
running "clang -g" on an empty file), we don't bother emitting an empty
DWARF CU. When we do that, we must make sure we don't also emit a DWARF v5
name index, as DWARF specifies that each index must reference at least
one compilation unit.
Reviewers: JDevlieghere, aprantl, dblaikie
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D45435
llvm-svn: 329575
Summary: The bit widths are wrong.
Reviewers: bkramer, lhames, hans
Reviewed By: hans
Subscribers: hans, nemanjai, kbarton, llvm-commits
Differential Revision: https://reviews.llvm.org/D45361
llvm-svn: 329573
Summary:
We do not try to move the instructions and split the block till we
know the blocks can be split, i.e. BCE-cmp-insts can be separated from
non-BCE-cmp-insts.
Reviewers: davide, courbet
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D44443
llvm-svn: 329564