Sam Parker [Wed, 27 Sep 2017 08:30:45 +0000 (08:30 +0000)]
[ARM] isTruncateFree fix
I implemented isTruncateFree in rL313533, this patch fixes the logic
to match my comment, as the previous logic was too general. Now the
only truncates that are free are i64 -> i32.
Martin Storsjo [Wed, 27 Sep 2017 06:08:16 +0000 (06:08 +0000)]
[X86] Remove erroneous callsite offsetting in SJLJ landing pads
The callsite value is already stored indexed from 0 in
the _Unwind_Context struct. When accessed via the functions
_Unwind_GetIP and _Unwind_SetIP, the value is indexed from 1,
but those functions handle the offseting. When reading directly
from the struct here, we shouldn't subtract 1.
This matches the code generated by the ARM target, where SJLJ
exception handling is used by default on iOS.
This makes clang-built object files for 32 bit x86 mingw work when
linked with libgcc/libstdc++.
Martin Pelikan [Wed, 27 Sep 2017 04:48:03 +0000 (04:48 +0000)]
[XRay] convert FDR arg1 log entries
Summary:
A new FDR metadata record will support logging a function call argument;
appending multiple metadata records will represent a sequence of arguments
meaning that "holes" are not representable by the buffer format. Each
call argument is currently a 64-bit value (useful for "this" pointers and
synchronization objects).
If present, we put this argument to the function call "entry" record it
belongs to, and alter its type to notify the user of its presence.
Sanjoy Das [Tue, 26 Sep 2017 21:54:27 +0000 (21:54 +0000)]
[BypassSlowDivision] Improve our handling of divisions by constants
Summary:
Don't bail out on constant divisors for divisions that can be narrowed without
introducing control flow . This gives us a 32 bit multiply instead of an
emulated 64 bit multiply in the generated PTX assembly.
[AArch64][Falkor] Fix bug in falkor prefetcher fix pass.
Summary:
In rare cases, loads that don't get prefetched that were marked as
strided loads could cause a crash if they occurred in a loop with other
colliding loads.
[AArch64][Falkor] Fix correctness bug in falkor prefetcher fix pass and correct some opcode tag computations.
Summary:
This addresses a correctness bug for LD[1234]*_POST opcodes that have
the prefetcher fix applied to them: the base register was not being
written back from the temp after being incremented, so it would appear
to never be incremented.
Also, fix some opcode tag computations based on some updated HW details
to get better tag avoidance and thus better prefetcher performance.
[X86] Don't emit X86::MOV8rr_NOREX from X86InstrInfo::copyPhysReg.
This hook is called after register allocation with two physical registers. We don't need a separate instruction at that time to force register class constraints. I left in the assert though. We also have a fatal error in X86MCCodeEmitter if we ever encode an H-reg and a REX prefix.
Sam Clegg [Tue, 26 Sep 2017 21:10:09 +0000 (21:10 +0000)]
[WebAssembly] Model weakly defined symbols as wasm exports
Previously these were being included as both imports and
exports, with the import being satisfied by the export
(or some strong symbol) at runtime. However proved
unnecessary and actually complicated linking as it meant
there was not a 1-to-1 mapping between a wasm function
/global index and a linker symbol.
[PowerPC] Reverting sequence of patches for elimination of comparison instructions
In the past while, I've committed a number of patches in the PowerPC back end
aimed at eliminating comparison instructions. However, this causes some failures
in proprietary source and these issues are not observed in SPEC or any open
source packages I've been able to run.
As a result, I'm pulling the entire series and will refactor it to:
- Have a single entry point for easy control
- Have fine-grained control over which patterns we transform
A side-effect of this is that test cases for these patches (and modified by
them) are XFAIL-ed. This is a temporary measure as it is counter-productive
to remove/modify these test cases and then have to modify them again when
the refactored patch is recommitted.
The failure will be investigated in parallel to the refactoring effort and
the recommit will either have a fix for it or will leave this transformation
off by default until the problem is resolved.
[X86][LLVM]Expanding Supports lowerInterleavedStore() in X86InterleavedAccess (VF{8|16|32} stride 3)
This patch expands the support of lowerInterleavedStore to {8|16|32}x8i stride 3.
LLVM creates suboptimal shuffle code-gen for AVX2. In overall, this patch is a specific fix for the pattern (Strid=3 VF={8|16|32}) .
This patch is part two of two patches and it covers the store (interlevaed) side.
The patch goal is to optimize the following sequence:
a0 a1 a2 a3 a4 a5 a6 a7
b0 b1 b2 b3 b4 b5 b6 b7
c0 c1 c2 c3 c4 c5 c6 c7
[InstCombine] Remove one use restriction on the shift for calls to foldICmpAndShift.
If this transformation succeeds, we're going to remove our dependency on the shift by rewriting the and. So it doesn't matter how many uses the shift has.
This distributes the one use check to other transforms in foldICmpAndConstConst that do need it.
Sam Clegg [Tue, 26 Sep 2017 18:21:12 +0000 (18:21 +0000)]
[WebAssembly] Use function/global index space in WasmSymbol
It is useful for the symbol to contain the index of the
function of global it represents in the function/global
index space.
For imports we also store the import index so that the
linker can find, for example, the signature of the
corresponding function, which is defined by the import
In the long run we need to decide whether this API
surface should be closer to binary (where imported
functions are seperate) or the wasm spec (where the
function index space is unified).
When dsymutil generates the companion file, its strips all unnecessary
sections by omitting their body and setting the offset in their
corresponding load command to zero.
One such section is the .eh_frame section, as it contains runtime
information rather than debug information and is part of the __TEXT
segment. When reading this section, we would just read the number of
bytes specified in the load command, starting from offset 0 (i.e. the
beginning of the file).
Rather than trying to parse this obviously invalid section, dwarfdump
now skips this.
Simon Pilgrim [Tue, 26 Sep 2017 14:12:50 +0000 (14:12 +0000)]
[X86][XOP] Merge rotation opcodes with AVX512 equivalents. NFCI.
The XOP rotations act as ROTL with +ve values and ROTR with -ve values, which means that we can treat them all as ROTL with unsigned modulo. We already check that we're only trying to lower as ROTL for XOP rotations.
[DSE] Merge stores when the later store only writes to memory locations the early store also wrote to (2nd try)
This is a 2nd attempt at:
https://reviews.llvm.org/rL310055
...which was reverted at rL310123 because of PR34074:
https://bugs.llvm.org/show_bug.cgi?id=34074
In this version, we break out of the inner loop after we successfully merge and kill a pair of stores. In the
earlier rev, we were continuing instead, which meant we could process the invalid info from a now dead store.
Original commit message (authored by Filipe Cabecinhas):
This fixes PR31777.
If both stores' values are ConstantInt, we merge the two stores
(shifting the smaller store appropriately) and replace the earlier (and
larger) store with an updated constant.
In the future we should also support vectors of integers. And maybe
float/double if we can.
This patch adds logic to follow a symbol's aliases when the symbol name
cannot be found in the current object file. It checks the main binary
for the symbol's address and queries the current object for its aliases
(symbols with the same address) before printing out a warning.
[X86] Finishing broadcastf32x2 and broadcasti32x2 intrinsics lowering to IR. llvm side.
Removing X86 broadcast(f/i)32x2 intrinsics from llvm.
Adding autoUpgrade support.
Moving matching tests from avx512dq-intrinsics.ll to avx512dq-intrinsics-upgrade.ll and from avx512dqvl-intrinsics.ll to avx512dqvl-intrinsics-upgrade.ll.
Usually the frontend communicates the size of wchar_t via metadata and
we can optimize wcslen (and possibly other calls in the future). In
cases without the wchar_size metadata we would previously try to guess
the correct size based on the target triple; however this is fragile to
keep up to date and may miss users manually changing the size via flags.
Better be safe and stop guessing and optimizing if the frontend didn't
communicate the size.
[llvm-cov] Warn if -show-functions is used without query files
llvm-cov's report mode does not print any output when -show-functions is
specified and no source files are specified. This can be surprising, so
the tool should at least print out an error message when this happens.
Summary:
Sanitizer blacklist entries currently apply to all sanitizers--there
is no way to specify that an entry should only apply to a specific
sanitizer. This is important for Control Flow Integrity since there are
several different CFI modes that can be enabled at once. For maximum
security, CFI blacklist entries should be scoped to only the specific
CFI mode(s) that entry applies to.
Adding section headers to SpecialCaseLists allows users to specify more
information about list entries, like sanitizer names or other metadata,
like so:
The section headers are regular expressions. For backwards compatbility,
blacklist entries entered before a section header are put into the '[*]'
section so that blacklists without sections retain the same behavior.
SpecialCaseList has been modified to also accept a section name when
matching against the blacklist. It has also been modified so the
follow-up change to clang can define a derived class that allows
matching sections by SectionMask instead of by string.
It leads to some improvements, but also a regression for the simple
case, so it's not clearly a good idea.
test/CodeGen/ARM/vcvt.ll now has test coverage to show the difference.
Ultimately, the right solution is probably to custom-lower fp-to-int
conversions, to something like ARMISD::VCVT_F32_S32 plus a bitcast.
It's hard to do the right thing when the implicit bitcast isn't visible
to DAG transforms.
R12 is used for the SwiftError parameter. It is no longer a CSR as it
is used for transfer the SwiftError, and the caller must preserve it if
they need to.
[InstCombine] Move an optimization from foldICmpAndConstConst to foldICmpUsingKnownBits
All this optimization cares about is knowing how many low bits of LHS is known to be zero and whether that means that the result is 0 or greater than the RHS constant. It doesn't matter where the zeros in the low bits came from. So we don't need to specifically look for an AND. Instead we can use known bits.
[X86] Don't select anyext GR32->GR64 to SUBREG_TO_REG. Use INSERT_SUBREG instead.
As far as I know SUBREG_TO_REG is stating that the upper bits are 0. But if we are just converting the GR32 with no checks, then we have no reason to say the upper bits are 0.
I don't really know how to test this today since I can't find anything that looks that closely at SUBREG_TO_REG. The test changes here seems to be some perturbance of register allocation.
[llvm-objcopy] Refactor code to include initialize method
This change refactors some of the code to allow for some code
deduplication in later diffs as well as just to make adding a new
section type more self contained to the class itself. The idea for this
was first mentioned by James in D 37915 and will be used in that change
as recommended.
This change follows changes for dynamic sections but precedes support
for dynamic relocations.
The 1st attempt at this:
https://reviews.llvm.org/rL314117
was reverted at:
https://reviews.llvm.org/rL314118
because of bot fails for clang tests that were checking optimized IR. That should be fixed with:
https://reviews.llvm.org/rL314144
...so try again.
Original commit message:
The transform to convert an extract-of-a-select-of-vectors was added at:
https://reviews.llvm.org/rL194013
And a question about the validity of this transform was raised in the review:
https://reviews.llvm.org/D1539:
...but not answered AFAICT>
Most of the motivating cases in that patch are now handled by other combines. These are the tests that were added with
the original commit, but they are not regressing even after we remove the transform in this patch.
The diffs we see after removing this transform cause us to avoid increasing the instruction count, so we don't want to do
those transforms as canonicalizations.
The motivation for not turning a vector-select-of-vectors into a scalar operation is shown in PR33301:
https://bugs.llvm.org/show_bug.cgi?id=33301
...in those cases, we'll get vector ops with this patch rather than the vector/scalar mix that we currently see.
[SelectionDAG] Teach simplifyDemandedBits to handle shifts by constant splat vectors
This teach simplifyDemandedBits to handle constant splat vector shifts.
This required changing some uses of getZExtValue to getLimitedValue since we can't rely on legalization using getShiftAmountTy for the shift amount.
I believe there may have been a bug in the ((X << C1) >>u ShAmt) handling where we didn't check if the inner shift was too large. I've fixed that here.
I had to add new patterns to ARM because the zext/sext the patterns were trying to look for got turned into an any_extend with this patch. Happy to split that out too, but not sure how to test without this change.
[Hexagon] Better determination of register classes in bit tracker
Add two callbacks to MachineEvaluator, so that specific implementations
can specify more details about register classes:
- composeWithSubRegIndex(RC,Idx), to provide the register class for a
register from RC used in conjunction with a subregister index Idx.
- getPhysRegBitWidth(Reg), to provide the size in bits of the given
physical register.
[AVX-512] Replace large number of explicit patterns that check for insert_subvector with zero after masked compares with fewer patterns with predicate
This replaces the large number of patterns that handle every possible case of zeroing after a masked compare with a few simpler patterns that use a predicate to check for a masked compare producer.
This is similar to what we do for detecting free GR32->GR64 zero extends and free xmm->ymm/zmm zero extends.
This shrinks the isel table from ~590k to ~531k. This is a roughly 10% reduction in size.
The transform to convert an extract-of-a-select-of-vectors was added at:
rL194013
And a question about the validity of this transform was raised in the review:
https://reviews.llvm.org/D1539:
...but not answered AFAICT>
Most of the motivating cases in that patch are now handled by other combines. These are the tests that were added with
the original commit, but they are not regressing even after we remove the transform in this patch.
The diffs we see after removing this transform cause us to avoid increasing the instruction count, so we don't want to do
those transforms as canonicalizations.
The motivation for not turning a vector-select-of-vectors into a scalar operation is shown in PR33301:
https://bugs.llvm.org/show_bug.cgi?id=33301
...in those cases, we'll get vector ops with this patch rather than the vector/scalar mix that we currently see.
[DebugInfo] Sort the SDDbgValue list before assuming it is in IR order
Summary:
This code iterates the 'Orders' vector in parallel with the DbgValue
list, emitting all DBG_VALUEs that occurred between the last IR order
insertion point and the next insertion point. This assumes the
SDDbgValue list is sorted in IR order, which it usually is. However, it
is not sorted when a node with a debug value is replaced with another
one. When this happens, TransferDbgValues is called, and the new value
is added to the end of the list.
The problem can be solved by stably sorting the list by IR order.
[X86][LLVM]Expanding Supports lowerInterleavedStore() in X86InterleavedAccess (VF8 stride 4):
This patch expands the support of lowerInterleavedStore to 8x8i stride 4.
LLVM creates suboptimal shuffle code-gen for AVX2.
In overall, this patch is a specific fix for the pattern (Strid=4 VF=8) and we plan to include more patterns in the future.
The patch goal is to optimize the following sequence:
At the end of the computation, we have xmm2, xmm0, xmm12 and xmm3 holding
each 8 chars:
[PowerPC] Eliminate compares - add i64 sext/zext handling for SETLT/SETGT
As mentioned in https://reviews.llvm.org/D33718, this simply adds another
pattern to the compare elimination sequence and is committed without a
differential review.
Summary:
SLP vectorizer supports horizontal reductions for Add/FAdd binary operations. Patch adds support for horizontal min/max reductions.
Function getReductionCost() is split to getArithmeticReductionCost() for binary operation reductions and getMinMaxReductionCost() for min/max reductions.
Patch fixes PR26956.
Summary:
Right now there are two functions with the same name, one does the work
and the other one returns true if expansion is needed. Rename
TargetTransformInfo::expandMemCmp to make it more consistent with other
members of TargetTransformInfo.
[X86] Make IFMA instructions during isel so we can fold broadcast loads.
This required changing the ISD opcode for these instructions to have the commutable operands first and the addend last. This way tablegen can autogenerate the additional patterns for us.
projects / llvm / log