and the follow-up r254356: "Fix a bug in MachineBlockPlacement that may cause assertion failure during BranchProbability construction."
Asserts were firing in Chromium builds. See PR25687.
llvm-svn: 254366
The patch in http://reviews.llvm.org/D13745 is broken into four parts:
1. New interfaces without functional changes (http://reviews.llvm.org/D13908).
2. Use new interfaces in SelectionDAG, while in other passes treat probabilities
as weights (http://reviews.llvm.org/D14361).
3. Use new interfaces in all other passes.
4. Remove old interfaces.
This patch is 3+4 above. In this patch, MBB won't provide weight-based
interfaces any more, which are totally replaced by probability-based ones.
The interface addSuccessor() is redesigned so that the default probability is
unknown. We allow unknown probabilities but don't allow using it together
with known probabilities in successor list. That is to say, we either have a
list of successors with all known probabilities, or all unknown
probabilities. In the latter case, we assume each successor has 1/N
probability where N is the number of successors. An assertion checks if the
user is attempting to add a successor with the disallowed mixed use as stated
above. This can help us catch many misuses.
All uses of weight-based interfaces are now updated to use probability-based
ones.
Differential revision: http://reviews.llvm.org/D14973
llvm-svn: 254348
We currently output FMA instructions on targets which support both FMA4 + FMA (i.e. later Bulldozer CPUS bdver2/bdver3/bdver4).
This patch flips this so FMA4 is preferred; this is for several reasons:
1 - FMA4 is non-destructive reducing the need for mov instructions.
2 - Its more straighforward to commute and fold inputs (although the recent work on FMA has reduced this difference).
3 - All supported targets have FMA4 performance equal or better to FMA - Piledriver (bdver2) in particular has half the throughput when executing FMA instructions.
Its looks like no future AMD processor lines will support FMA4 after the Bulldozer series so we're not causing problems for later CPUs.
Differential Revision: http://reviews.llvm.org/D14997
llvm-svn: 254339
The MachineVerifier wants to check that the register operands of an
instruction belong to the instruction's register class. RIP-relative
control flow instructions violated this by referencing RIP. While this
was fixed for SysV, it was never fixed for Win64.
llvm-svn: 254315
We could already recognise shuffle(FSUB, FADD) -> ADDSUB, this allow us to recognise shuffle(FADD, FSUB) -> ADDSUB by commuting the shuffle mask prior to matching.
llvm-svn: 254259
Added FMADD/FMSUB/FNMADD/FNMSUB tests for all types
Added load folding tests for 512-bit vectors
NOTE: Many of the AVX512 FMA instructions don't yet commute/fold correctly
As discussed on D14909
llvm-svn: 254232
generated for _mm_losd_s{s,d}() intrinsics and used in scalar FMAs generated
for FMA intrinsics _mm_f{madd,msub,nmadd,nmsub}_s{s,d}().
Reviewer: David Kreitzer
Differential Revision: http://reviews.llvm.org/D14762
llvm-svn: 254140
to a simple type when lowering a truncating store of a vector type. In this
case for an EVT we'll return Expand as we should in all of the cases anyhow.
The testcase triggered at the one in VectorLegalizer::LegalizeOp, inspection
found the rest.
llvm-svn: 254061
It was wrong order of operands (from intrinsic to DAG node).
I added more strict type specification for instruction selection.
Differential Revision: http://reviews.llvm.org/D14942
llvm-svn: 254059
This caused PR25607 and also caused Chromium to crash on start-up.
(Also had to update test/CodeGen/X86/avx-splat.ll, which was committed
after shrink wrapping was enabled.)
llvm-svn: 254044
X86 needs to use its own FMA opcodes, preventing the standard FNEG(FMA) pattern table recognition method used by other platforms. This patch adds support for lowering FNEG(FMA(X,Y,Z)) into a single suitably negated FMA instruction.
Fix for PR24364
Differential Revision: http://reviews.llvm.org/D14906
llvm-svn: 254016
This patch fixes the following issues:
1. Fix the return type of X86psadbw: it should not be the same type of inputs.
For vNi8 inputs the output should be vMi64, where M = N/8.
2. Fix the return type of int_x86_avx512_psad_bw_512 accordingly.
3. Fix the definiton of PSADBW, VPSADBW, and VPSADBWY accordingly.
4. Adjust the return type when building a DAG node of X86ISD::PSADBW type.
5. Update related tests.
Differential revision: http://reviews.llvm.org/D14897
llvm-svn: 254010
We had duplicated definitions for the same hardware '[v]movq' instructions. For example with SSE:
def MOVZQI2PQIrr : RS2I<0x6E, MRMSrcReg, (outs VR128:$dst), (ins GR64:$src),
"mov{d|q}\t{$src, $dst|$dst, $src}", // X86-64 only
[(set VR128:$dst, (v2i64 (X86vzmovl (v2i64 (scalar_to_vector GR64:$src)))))],
IIC_SSE_MOVDQ>;
def MOV64toPQIrr : RS2I<0x6E, MRMSrcReg, (outs VR128:$dst), (ins GR64:$src),
"mov{d|q}\t{$src, $dst|$dst, $src}",
[(set VR128:$dst, (v2i64 (scalar_to_vector GR64:$src)))],
IIC_SSE_MOVDQ>, Sched<[WriteMove]>;
As shown in the test case and PR25554:
https://llvm.org/bugs/show_bug.cgi?id=25554
This causes us to miss reusing an operand because later passes don't know these 'movq' are the same instruction.
This patch deletes one pair of these defs.
Sadly, this won't fix the original test case in the bug report. Something else is still broken.
Differential Revision: http://reviews.llvm.org/D14941
llvm-svn: 253988
The patch in http://reviews.llvm.org/D13745 is broken into four parts:
1. New interfaces without functional changes.
2. Use new interfaces in SelectionDAG, while in other passes treat probabilities
as weights.
3. Use new interfaces in all other passes.
4. Remove old interfaces.
This the second patch above. In this patch SelectionDAG starts to use
probability-based interfaces in MBB to add successors but other MC passes are
still using weight-based interfaces. Therefore, we need to maintain correct
weight list in MBB even when probability-based interfaces are used. This is
done by updating weight list in probability-based interfaces by treating the
numerator of probabilities as weights. This change affects many test cases
that check successor weight values. I will update those test cases once this
patch looks good to you.
Differential revision: http://reviews.llvm.org/D14361
llvm-svn: 253965
This patch detects the AVG pattern in vectorized code, which is simply
c = (a + b + 1) / 2, where a, b, and c have the same type which are vectors of
either unsigned i8 or unsigned i16. In the IR, i8/i16 will be promoted to
i32 before any arithmetic operations. The following IR shows such an example:
%1 = zext <N x i8> %a to <N x i32>
%2 = zext <N x i8> %b to <N x i32>
%3 = add nuw nsw <N x i32> %1, <i32 1 x N>
%4 = add nuw nsw <N x i32> %3, %2
%5 = lshr <N x i32> %N, <i32 1 x N>
%6 = trunc <N x i32> %5 to <N x i8>
and with this patch it will be converted to a X86ISD::AVG instruction.
The pattern recognition is done when combining instructions just before type
legalization during instruction selection. We do it here because after type
legalization, it is much more difficult to do pattern recognition based
on many instructions that are doing type conversions. Therefore, for
target-specific instructions (like X86ISD::AVG), we need to take care of type
legalization by ourselves. However, as X86ISD::AVG behaves similarly to
ISD::ADD, I am wondering if there is a way to legalize operands and result
types of X86ISD::AVG together with ISD::ADD. It seems that the current design
doesn't support this idea.
Tests are added for SSE2, AVX2, and AVX512BW and both i8 and i16 types of
variant vector sizes.
Differential revision: http://reviews.llvm.org/D14761
llvm-svn: 253952
Caller saved regs differ between SysV and Win64. Use the tail call available set to scavenge from.
Refactor register info to create new helper to get at tail call GPRs. Added a new test case for windows. Fixed up a number of X64 tests since now RCX is preferred over RDX on SysV.
Differential Revision: http://reviews.llvm.org/D14878
llvm-svn: 253927
autogenerated.
Also update existing test cases which appear to be generated by it and
weren't modified (other than addition of the header) by rerunning it.
llvm-svn: 253917
ISERT_SUBVECTOR for i1 vectors may be done with shifts, when we insert into the lower part, or into the upper part, on into all-zero vector.
CONCAT_VECTORS uses ISERT_SUBVECTOR.
Differential Revision: http://reviews.llvm.org/D14815
llvm-svn: 253819
MachineInstrBuilder::addDisp can already add an immediate or global address MO with an adjusted offset, this patch adds support for constant pool indices as well.
All remaining MO types still assert - there are a number of other types that could support adjusted offsets but I have no test cases at this time.
Required to fix a regression in D13988 found by Mikael Holmén during stress testing (test case attached).
Differential Revision: http://reviews.llvm.org/D14867
llvm-svn: 253795
It turns out we have a number of places that just grab the first type attached to a register class for various reasons. This is fine unless for some reason that type isn't legal on the current target, such as for SSE1 which doesn't support v16i8/v8i16/v4i32/v2i64 - all of which were included before 4f32 in the class.
Given that this is such a rare situation I've just re-ordered the types and placed the float types first.
Fix for PR16133
Differential Revision: http://reviews.llvm.org/D14787
llvm-svn: 253773
Now that the register allocator knows about the barriers on funclet
entry and exit, testing has shown that this is unnecessary.
We still demote PHIs on unsplittable blocks due to the differences
between the IR CFG and the Machine CFG.
llvm-svn: 253619
This is another step towards allowing SimplifyCFG to speculate harder, but then have
CGP clean things up if the target doesn't like it.
Previous patches in this series:
http://reviews.llvm.org/D12882http://reviews.llvm.org/D13297
D13297 should catch most expensive ops, but speculation of cttz/ctlz requires special
handling because of weirdness in the intrinsic definition for handling a zero input
(that definition can probably be blamed on x86).
For example, if we have the usual speculated-by-select expensive op pattern like this:
%tobool = icmp eq i64 %A, 0
%0 = tail call i64 @llvm.cttz.i64(i64 %A, i1 true) ; is_zero_undef == true
%cond = select i1 %tobool, i64 64, i64 %0
ret i64 %cond
There's an instcombine that will turn it into:
%0 = tail call i64 @llvm.cttz.i64(i64 %A, i1 false) ; is_zero_undef == false
This CGP patch is looking for that case and despeculating it back into:
entry:
%tobool = icmp eq i64 %A, 0
br i1 %tobool, label %cond.end, label %cond.true
cond.true:
%0 = tail call i64 @llvm.cttz.i64(i64 %A, i1 true) ; is_zero_undef == true
br label %cond.end
cond.end:
%cond = phi i64 [ %0, %cond.true ], [ 64, %entry ]
ret i64 %cond
This unfortunately may lead to poorer codegen (see the changes in the existing x86 test),
but if we increase speculation in SimplifyCFG (the next step in this patch series), then
we should avoid those kinds of cases in the first place.
The need for this patch was originally mentioned here:
http://reviews.llvm.org/D7506
with follow-up here:
http://reviews.llvm.org/D7554
Differential Revision: http://reviews.llvm.org/D14630
llvm-svn: 253573
Copying one mask register to another under BW should be done with kmovq instruction, otherwise we can loose some bits.
Copying 8 bits under DQ may be done with kmovb.
Differential Revision: http://reviews.llvm.org/D14812
llvm-svn: 253563
The lowering patterns for X86ISD::VZEXT_MOVL for 128-bit to 256-bit vectors were just copying the lower xmm instead of actually masking off the first scalar using a blend.
Fix for PR25320.
Differential Revision: http://reviews.llvm.org/D14151
llvm-svn: 253561
The masked intrinsics support all integer and floating point data types. I added the pointer type to this list.
Added tests for CodeGen and for Loop Vectorizer.
Updated the Language Reference.
Differential Revision: http://reviews.llvm.org/D14150
llvm-svn: 253544
Note, this was reviewed (and more details are in) http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20151109/312083.html
These intrinsics currently have an explicit alignment argument which is
required to be a constant integer. It represents the alignment of the
source and dest, and so must be the minimum of those.
This change allows source and dest to each have their own alignments
by using the alignment attribute on their arguments. The alignment
argument itself is removed.
There are a few places in the code for which the code needs to be
checked by an expert as to whether using only src/dest alignment is
safe. For those places, they currently take the minimum of src/dest
alignments which matches the current behaviour.
For example, code which used to read:
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 500, i32 8, i1 false)
will now read:
call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 8 %dest, i8* align 8 %src, i32 500, i1 false)
For out of tree owners, I was able to strip alignment from calls using sed by replacing:
(call.*llvm\.memset.*)i32\ [0-9]*\,\ i1 false\)
with:
$1i1 false)
and similarly for memmove and memcpy.
I then added back in alignment to test cases which needed it.
A similar commit will be made to clang which actually has many differences in alignment as now
IRBuilder can generate different source/dest alignments on calls.
In IRBuilder itself, a new argument was added. Instead of calling:
CreateMemCpy(Dst, Src, getInt64(Size), DstAlign, /* isVolatile */ false)
you now call
CreateMemCpy(Dst, Src, getInt64(Size), DstAlign, SrcAlign, /* isVolatile */ false)
There is a temporary class (IntegerAlignment) which takes the source alignment and rejects
implicit conversion from bool. This is to prevent isVolatile here from passing its default
parameter to the source alignment.
Note, changes in future can now be made to codegen. I didn't change anything here, but this
change should enable better memcpy code sequences.
Reviewed by Hal Finkel.
llvm-svn: 253511
This patch adds support for vector constant folding of integer/float comparisons.
This requires FoldConstantVectorArithmetic to support scalar constant operands (in this case ISD::CONDCASE). In future we should be able to support other scalar constant types as necessary (and possibly start calling FoldConstantVectorArithmetic for all node creations)
Differential Revision: http://reviews.llvm.org/D14683
llvm-svn: 253504
If a section is rw, it is irrelevant if the dynamic linker will write to
it or not.
It looks like llvm implemented this because gcc was doing it. It looks
like gcc implemented this in the hope that it would put all the
relocated items close together and speed up the dynamic linker.
There are two problem with this:
* It doesn't work. Both bfd and gold will map .data.rel to .data and
concatenate the input sections in the order they are seen.
* If we want a feature like that, it can be implemented directly in the
linker since it knowns where the dynamic relocations are.
llvm-svn: 253436
When looking for the best successor from the outer loop for a block
belonging to an inner loop, the edge probability computation can be
improved so that edges in the inner loop are ignored. For example,
suppose we are building chains for the non-loop part of the following
code, and looking for B1's best successor. Assume the true body is very
hot, then B3 should be the best candidate. However, because of the
existence of the back edge from B1 to B0, the probability from B1 to B3
can be very small, preventing B3 to be its successor. In this patch, when
computing the probability of the edge from B1 to B3, the weight on the
back edge B1->B0 is ignored, so that B1->B3 will have 100% probability.
if (...)
do {
B0;
... // some branches
B1;
} while(...);
else
B2;
B3;
Differential revision: http://reviews.llvm.org/D10825
llvm-svn: 253414
Summary:
Now that there is a one-to-one mapping from MachineFunction to
WinEHFuncInfo, we don't need to use a DenseMap to select the right
WinEHFuncInfo for the current funclet.
The main challenge here is that X86WinEHStatePass is an IR pass that
doesn't have access to the MachineFunction. I gave it its own
WinEHFuncInfo object that it uses to calculate state numbers, which it
then throws away. As long as nobody creates or removes EH pads between
this pass and SDAG construction, we will get the same state numbers.
The other thing X86WinEHStatePass does is to mark the EH registration
node. Instead of communicating which alloca was the registration through
WinEHFuncInfo, I added the llvm.x86.seh.ehregnode intrinsic. This
intrinsic generates no code and simply marks the alloca in use.
Reviewers: JCTremoulet
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D14668
llvm-svn: 253378
Statepoint lowering currently expects that the target method of a
statepoint only defines a single value. This precludes using
statepoints with ABIs that return values in multiple registers
(e.g. the SysV AMD64 ABI). This change adds support for lowering
statepoints with mutli-def targets.
llvm-svn: 253339
Several places in AsmPrinter.cpp print comments describing MachineOperand
registers using MCRegisterInfo, which uses MCOperand-oriented names. This
doesn't work for targets that use virtual registers exclusively, as
WebAssembly does, since virtual registers are represented and printed
differently.
This patch preserves what seems to be the spirit of r229978, avoiding the
use of TM.getSubtargetImpl(), while still using MachineOperand-oriented
printing for MachineOperands.
Differential Revision: http://reviews.llvm.org/D14709
llvm-svn: 253338
The way prelink used to work was
* The compiler decides if a given section only has relocations that
are know to point to the same DSO. If so, it names it
.data.rel.ro.local<something>.
* The static linker puts all of these together.
* The prelinker program assigns addresses to each library and resolves
the local relocations.
There are many problems with this:
* It is incompatible with address space randomization.
* The information passed by the compiler is redundant. The linker
knows if a given relocation is in the same DSO or not. If could sort
by that if so desired.
* There are newer ways of speeding up DSO (gnu hash for example).
* Even if we want to implement this again in the compiler, the previous
implementation is pretty broken. It talks about relocations that are
"resolved by the static linker". If they are resolved, there are none
left for the prelinker. What one needs to track is if an expression
will require only dynamic relocations that point to the same DSO.
At this point it looks like the prelinker is an historical curiosity.
For example, fedora has retired it because it failed to build for two
releases
(http://pkgs.fedoraproject.org/cgit/prelink.git/commit/?id=eb43100a8331d91c801ee3dcdb0a0bb9babfdc1f)
This patch removes support for it. That is, it stops printing the
".local" sections.
llvm-svn: 253280
On top of that, don't bother allocating and initializing UnwindHelp if
we don't have any funclets. Currently we always use RBP as our frame
pointer when funclets are present, so this change makes it impossible to
come here without any fixed stack objects.
Fixes PR25533.
llvm-svn: 253245
attribute.
Even if the target supports shrink-wrapping, the prologue and epilogue
must not move because a crash can happen anywhere and sanitizers need
to be able to unwind from the PC of the crash.
llvm-svn: 253116
The C++ EH personality automatically restores ESP from the C++ EH
registration node after a catchret. I mistakenly thought it was like
SEH, which does not restore ESP.
It makes sense for C++ EH to differ from SEH here because SEH does not
use funclets for catches, and does not allow catching inside of finally.
C++ EH may need to unwind through multiple catch funclets and eventually
catchret to some outer funclet. Therefore, the runtime has to keep track
of which ESP to use with catchret, rather than having the compiler
reload it manually.
llvm-svn: 253084
This patch is enabling combining UNPCKL with vector_shuffle that moves the upper
half of a vector into the lower half, into a UNPCKH instruction. For example:
t2: v16i8 = vector_shuffle<8,9,10,11,12,13,14,15,u,u,u,u,u,u,u,u> t1, undef:v16i8
t3: v16i8 = X86ISD::UNPCKL undef:v16i8, t2
will be combined to:
t3: v16i8 = X86ISD::UNPCKH undef:v16i8, t1
Differential revision: http://reviews.llvm.org/D14399
llvm-svn: 253067
It made it possible to apply the memory folding optimization for the 2nd
operand of FMA*_Int instructions.
Reviewer: Quentin Colombet
Differential Revision: http://reviews.llvm.org/D14550
llvm-svn: 252973
ShrinkWrapping does not understand exception handling constraints for now, so
make sure we do not mess with them by aborting on functions that use EH
funclets.
llvm-svn: 252917
Several backends have instructions to reverse the order of bits in an integer. Conceptually matching such patterns is similar to @llvm.bswap, and it was mentioned in http://reviews.llvm.org/D14234 that it would be best if these patterns were matched in InstCombine instead of reimplemented in every different target.
This patch introduces an intrinsic @llvm.bitreverse.i* that operates similarly to @llvm.bswap. For plumbing purposes there is also a new ISD node ISD::BITREVERSE, with simple expansion and promotion support.
The intention is that InstCombine's BSWAP detection logic will be extended to support BITREVERSE too, and @llvm.bitreverse intrinsics emitted (if the backend supports lowering it efficiently).
llvm-svn: 252878
- Factor out code to query and modify the sign bit of a floatingpoint
value as an integer. This also works if none of the targets integer
types is big enough to hold all bits of the floatingpoint value.
- Legalize FABS(x) as FCOPYSIGN(x, 0.0) if FCOPYSIGN is available,
otherwise perform bit manipulation on the sign bit. The previous code
used "x >u 0 ? x : -x" which is incorrect for x being -0.0! It also
takes 34 instructions on ARM Cortex-M4. With this patch we only
require 5:
vldr d0, LCPI0_0
vmov r2, r3, d0
lsrs r2, r3, #31
bfi r1, r2, #31, #1
bx lr
(This could be further improved if the compiler would recognize that
r2, r3 is zero).
- Only lower FCOPYSIGN(x, y) = sign(x) ? -FABS(x) : FABS(x) if FABS is
available otherwise perform bit manipulation on the sign bit.
- Perform the sign(x) test by masking out the sign bit and comparing
with 0 rather than shifting the sign bit to the highest position and
testing for "<s 0". For x86 copysignl (on 80bit values) this gets us:
testl $32768, %eax
rather than:
shlq $48, %rax
sets %al
testb %al, %al
Differential Revision: http://reviews.llvm.org/D11172
llvm-svn: 252839
Summary: Other personalities don't use this special frame slot.
Reviewers: majnemer, andrew.w.kaylor, rnk
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D14580
llvm-svn: 252778
If possible and profitable, replace lea %reg, 1(%reg) and lea %reg, -1(%reg) with inc %reg and dec %reg respectively.
Patch by: anton.nadolsky@intel.com
Differential Revision: http://reviews.llvm.org/D14059
llvm-svn: 252722
This is one of the problems noted in PR25016:
https://llvm.org/bugs/show_bug.cgi?id=25016
and:
http://lists.llvm.org/pipermail/llvm-dev/2015-October/090998.html
The spilling problem is independent and not addressed by this patch.
The MachineCombiner was doing reassociations that don't improve or even worsen the critical path.
This is caused by inclusion of the "slack" factor when calculating the critical path of the original
code sequence. If we don't add that, then we have a more conservative cost comparison of the old code
sequence vs. a new sequence. The more liberal calculation must be preserved, however, for the AArch64
MULADD patterns because benchmark regressions were observed without that.
The two failing test cases now have identical asm that does what we want:
a + b + c + d ---> (a + b) + (c + d)
Differential Revision: http://reviews.llvm.org/D13417
llvm-svn: 252616
For CoreCLR on Windows, stack probes must be emitted as inline sequences that probe successive stack pages
between the current stack limit and the desired new stack pointer location. This implements support for
the inline expansion on x64.
For in-body alloca probes, expansion is done during instruction lowering. For prolog probes, a stub call
is initially emitted during prolog creation, and expanded after epilog generation, to avoid complications
that arise when introducing new machine basic blocks during prolog and epilog creation.
Added a new test case, modified an existing one to exclude non-x64 coreclr (for now).
Add test case
Fix tests
llvm-svn: 252578
The motivation for this patch starts with the epic fail example in PR18007:
https://llvm.org/bugs/show_bug.cgi?id=18007
...unfortunately, this patch makes no difference for that case, but it solves some
simpler cases. We'll get there some day. :)
The current 'or' matching code was using computeKnownBits() via
isBaseWithConstantOffset() -> MaskedValueIsZero(), but that's an unnecessarily limited use.
We can do more by copying the logic in ValueTracking's haveNoCommonBitsSet(), so we can
treat the 'or' as if it was an 'add'.
There's a TODO comment here because we should lift the bit-checking logic into a helper
function, so it's not duplicated in DAGCombiner.
An example of the better LEA matching:
leal (%rdi,%rdi), %eax
andl $1, %esi
orl %esi, %eax
Becomes:
andl $1, %esi
leal (%rsi,%rdi,2), %eax
Differential Revision: http://reviews.llvm.org/D13956
llvm-svn: 252515
For some reason we'd never run MachineVerifier on WinEH code, and you
explicitly have to ask for it with llc. I added it to a few test cases
to get some coverage.
Fixes PR25461.
llvm-svn: 252512
The TailDuplication machine pass ran across a malformed CFG: a PHI node
referred it's predecessor's predecessor instead of it's predecessor.
This occurred because we split the edge in X86ISelLowering when we
processed the CATCHRET but forgot to do something about the PHI nodes.
This fixes PR25444.
llvm-svn: 252413
When matching non-LSB-extracting truncating broadcasts, we now insert
the necessary SRL. If the scalar resulted from a load, the SRL will be
folded into it, creating a narrower, offset, load.
However, i16 loads aren't Desirable, so we get i16->i32 zextloads.
We already catch i16 aextloads; catch these as well.
llvm-svn: 252363
Now that we recognize this, we can support it instead of bailing out.
That is, we can fold:
(v8i16 (shufflevector
(v8i16 (bitcast (v4i32 (build_vector X, Y, ...)))),
<1,1,...,1>))
into:
(v8i16 (vbroadcast (i16 (trunc (srl Y, 16)))))
llvm-svn: 252362
We used to incorrectly assume that the offset we're extracting from
was a multiple of the element size. So, we'd fold:
(v8i16 (shufflevector
(v8i16 (bitcast (v4i32 (build_vector X, Y, ...)))),
<1,1,...,1>))
into:
(v8i16 (vbroadcast (i16 (trunc Y))))
whereas we should have extracted the higher bits from X.
Instead, bail out if the assumption doesn't hold.
llvm-svn: 252361
All 3 operands of FMA3 instructions are commutable now.
Patch by Slava Klochkov
Reviewers: Quentin Colombet(qcolombet), Ahmed Bougacha(ab).
Differential Revision: http://reviews.llvm.org/D13269
llvm-svn: 252335
Summary:
In this implementation, LiveIntervalAnalysis invents a few register
masks on basic block boundaries that preserve no registers. The nice
thing about this is that it prevents the prologue inserter from thinking
it needs to spill all XMM CSRs, because it doesn't see any explicit
physreg defs in the MI.
Reviewers: MatzeB, qcolombet, JosephTremoulet, majnemer
Subscribers: MatzeB, llvm-commits
Differential Revision: http://reviews.llvm.org/D14407
llvm-svn: 252318
We now create the .eh_frame section early, just like every other special
section.
This means that the special flags are visible in code that explicitly
asks for ".eh_frame".
llvm-svn: 252313
Previously, subprograms contained a metadata reference to the function they
described. Because most clients need to get or set a subprogram for a given
function rather than the other way around, this created unneeded inefficiency.
For example, many passes needed to call the function llvm::makeSubprogramMap()
to build a mapping from functions to subprograms, and the IR linker needed to
fix up function references in a way that caused quadratic complexity in the IR
linking phase of LTO.
This change reverses the direction of the edge by storing the subprogram as
function-level metadata and removing DISubprogram's function field.
Since this is an IR change, a bitcode upgrade has been provided.
Fixes PR23367. An upgrade script for textual IR for out-of-tree clients is
attached to the PR.
Differential Revision: http://reviews.llvm.org/D14265
llvm-svn: 252219
We already had a test for this for 32-bit SEH catchpads, but those don't
actually create funclets. We had a bug that only appeared in funclet
prologues, where we would establish EBP and ESI as our FP and BP, and
then downstream prologue code would overwrite them.
While I was at it, I fixed Win64+funclets+stackrealign. This issue
doesn't come up as often there due to the ABI requring 16 byte stack
alignment, but now we can rest easy that AVX and WinEH will work well
together =P.
llvm-svn: 252210
This fixes the issue of wrong CFA calculation in the following case:
0x08048400 <+0>: push %ebx
0x08048401 <+1>: sub $0x8,%esp
0x08048404 <+4>: **call 0x8048409 <test+9>**
0x08048409 <+9>: **pop %eax**
0x0804840a <+10>: add $0x1bf7,%eax
0x08048410 <+16>: mov %eax,%ebx
0x08048412 <+18>: call 0x80483f0 <bar>
0x08048417 <+23>: add $0x8,%esp
0x0804841a <+26>: pop %ebx
0x0804841b <+27>: ret
The highlighted instructions are a product of movpc instruction. The call
instruction changes the stack pointer, and pop instruction restores its
value. However, the rule for computing CFA is not updated and is wrong on
the pop instruction. So, e.g. backtrace in gdb does not work when on the pop
instruction. This adds cfi instructions for both call and pop instructions.
cfi_adjust_cfa_offset** instruction is used with the appropriate offset for
setting the rules to calculate CFA correctly.
Patch by Violeta Vukobrat.
Differential Revision: http://reviews.llvm.org/D14021
llvm-svn: 252176
Win64 has some strict requirements for the epilogue. As a result, we disable
shrink-wrapping for Win64 unless the block that gets the epilogue is already an
exit block.
Fixes PR24193.
llvm-svn: 252088
This patch improves the memory folding of the inserted float element for the (V)INSERTPS instruction.
The existing implementation occurs in the DAGCombiner and relies on the narrowing of a whole vector load into a scalar load (and then converted into a vector) to (hopefully) allow folding to occur later on. Not only has this proven problematic for debug builds, it also prevents other memory folds (notably stack reloads) from happening.
This patch removes the old implementation and moves the folding code to the X86 foldMemoryOperand handler. A new private 'special case' function - foldMemoryOperandCustom - has been added to deal with memory folding of instructions that can't just use the lookup tables - (V)INSERTPS is the first of several that could be done.
It also tweaks the memory operand folding code with an additional pointer offset that allows existing memory addresses to be modified, in this case to convert the vector address to the explicit address of the scalar element that will be inserted.
Unlike the previous implementation we now set the insertion source index to zero, although this is ignored for the (V)INSERTPSrm version, anything that relied on shuffle decodes (such as unfolding of insertps loads) was incorrectly calculating the source address - I've added a test for this at insertps-unfold-load-bug.ll
Differential Revision: http://reviews.llvm.org/D13988
llvm-svn: 252074
Patch by Slava Klochkov
The key difference between FMA* and FMA*_Int opcodes is that FMA*_Int opcodes are handled more conservatively. It is illegal to commute the 1st operand of FMA*_Int instructions as the upper bits of scalar FMA intrinsic result must be taken from the 1st operand, but such commute transformation would change those upper bits and invalidate the intrinsic's result.
Reviewers: Quentin Colombet, Elena Demikhovsky
Differential Revision: http://reviews.llvm.org/D13710
llvm-svn: 252060
The x86 "sitofp i64 to double" dag combine, in 32-bit mode, lowers sitofp
directly to X86ISD::FILD (or FILD_FLAG). This should not be done in soft-float mode.
llvm-svn: 252042
There is no point in having invoke safepoints handled differently than the
call safepoints. All relevant decisions could be made by looking at whether
or not gc.result and gc.relocate lay in a same basic block. This change will
allow to lower call safepoints with relocates and results in a different
basic blocks. See test case for example.
Differential Revision: http://reviews.llvm.org/D14158
llvm-svn: 252028
XOP has the VPCMOV instruction that performs the common vector bit select operation OR( AND( SRC1, SRC3 ), AND( SRC2, ~SRC3 ) )
This patch adds tablegen pattern matching for this instruction.
Differential Revision: http://reviews.llvm.org/D8841
llvm-svn: 251975
When push instructions are being used to pass function arguments on
the stack, and either EH or debugging are enabled, we need to generate
.cfi_adjust_cfa_offset directives appropriately. For (synch) EH, it is
enough for the CFA offset to be correct at every call site, while
for debugging we want to be correct after every push.
Darwin does not support this well, so don't use pushes whenever it
would be required.
Differential Revision: http://reviews.llvm.org/D13767
llvm-svn: 251904
In the current BB placement algorithm, a loop chain always contains all loop blocks. This has a drawback that cold blocks in the loop may be inserted on a hot function path, hence increasing branch cost and also reducing icache locality.
Consider a simple example shown below:
A
|
B⇆C
|
D
When B->C is quite cold, the best BB-layout should be A,B,D,C. But the current implementation produces A,C,B,D.
This patch filters those cold blocks off from the loop chain by comparing the ratio:
LoopBBFreq / LoopFreq
to 20%: if it is less than 20%, we don't include this BB to the loop chain. Here LoopFreq is the frequency of the loop when we reduce the loop into a single node. In general we have more cold blocks when the loop has few iterations. And vice versa.
Differential revision: http://reviews.llvm.org/D11662
llvm-svn: 251833
1) PR25154. This is basically a repeat of PR18102, which was fixed in
r200201, and broken again by r234430. The latter changed which of the
store nodes was merged into from the first to the last. Thus, we now
also need to prefer merging a later store at a given address into the
target node, instead of an earlier one.
2) While investigating that, I also realized I'd introduced a bug in
r236850. There, I removed a check for alignment -- not realizing that
nothing except the alignment check was ensuring that none of the stores
were overlapping! This is a really bogus way to ensure there's no
aliased stores.
A better solution to both of these issues is likely to always use the
code added in the 'if (UseAA)' branches which rearrange the chain based
on a more principled analysis. I'll look into whether that can be used
always, but in the interest of getting things back to working, I think a
minimal change makes sense.
llvm-svn: 251816
Optimized <8 x i32> to <8 x i16>
<4 x i64> to < 4 x i32>
<16 x i16> to <16 x i8>
All these oprtrations use now AVX512F set (KNL). Before this change it was implemented with AVX2 set.
Differential Revision: http://reviews.llvm.org/D14108
llvm-svn: 251764
This patch generalizes the zeroing of vector elements with the BLEND instructions. Currently a zero vector will only blend if the shuffled elements are correctly inline, this patch recognises when a vector input is zero (or zeroable) and modifies a local copy of the shuffle mask to support a blend. As a zeroable vector input may not be all zeroes, the zeroable vector is regenerated if necessary.
Differential Revision: http://reviews.llvm.org/D14050
llvm-svn: 251659
This is a usage of the IR-level fast-math-flags now that they are propagated to SDNodes.
This was originally part of D8900.
Removing the global 'enable-unsafe-fp-math' checks will require auto-upgrade and
possibly other changes.
Differential Revision: http://reviews.llvm.org/D9708
llvm-svn: 251450
convert float to half with mask/maskz for the reg to reg version and mask for the reg to mem version (there is no maskz version for reg to mem).
Differential Revision: http://reviews.llvm.org/D14113
llvm-svn: 251409
GNU tools require elfiamcu to take up the entire OS field, so, e.g.
i?86-*-linux-elfiamcu is not considered a legal triple.
Make us compatible.
Differential Revision: http://reviews.llvm.org/D14081
llvm-svn: 251390
When taking the remainder of a value divided by a constant, visitREM()
attempts to convert the REM to a longer but faster sequence of instructions.
This conversion calls combine() on a speculative DIV instruction. Commit
rL250825 may cause this combine() to return a DIVREM, corrupting nearby nodes.
Flow eventually hits unreachable().
This patch adds a test case and a check to prevent visitREM() from trying
to convert the REM instruction in cases where a DIVREM is possible.
See http://reviews.llvm.org/D14035
llvm-svn: 251373
When the target does not support these intrinsics they should be converted to a chain of scalar load or store operations.
If the mask is not constant, the scalarizer will build a chain of conditional basic blocks.
I added isLegalMaskedGather() isLegalMaskedScatter() APIs.
Differential Revision: http://reviews.llvm.org/D13722
llvm-svn: 251237
When using the MCU psABI, compiler-generated library calls should pass
some parameters in-register. However, since inreg marking for x86 is currently
done by the front end, it will not be applied to backend-generated calls.
This is a workaround for PR3997, which describes a similar issue for -mregparm.
Differential Revision: http://reviews.llvm.org/D13977
llvm-svn: 251223
We don't need a mask of a rotation result to be a constant splat - any constant scalar/vector can be usefully folded.
Followup to D13851.
llvm-svn: 251197
Hans Wennborg