The sqrt case is faster and we already do this for the case where
the exponent is 0.25. This adds the 0.75 case which is also not
sensitive to signed zeros.
Patch by Whitney Tsang (Whitney)
Differential revision: https://reviews.llvm.org/D57434
llvm-svn: 353557
Replace OR(SHL,SRL) pattern with ISD::FSHR (legalization expands this later if necessary) - this helps with the scale == 0 'undefined' drop-through case that was discussed on D55720.
llvm-svn: 353546
clampScalar doesn't do anything for non-power-of-2 in range.
There should probably be a combination rule to reduce the number
of matching rules.
llvm-svn: 353526
Make behavior of G_LOAD in widenScalar same as for G_ZEXTLOAD and
G_SEXTLOAD. That is perform widenScalarDst to size given by the target
and avoid additional checks in common code. Targets can reorder or add
additional rules in LegalizeRuleSet for the opcode to achieve desired
behavior.
Select extending load that does not have specified type of extension
into zero extending load.
Select truncating store that stores number of bytes indicated by size
in MachineMemoperand.
Differential Revision: https://reviews.llvm.org/D57454
llvm-svn: 353520
Differential revision: https://reviews.llvm.org/D55444
dpp move with uses and old reg initializer should be in the same BB.
bound_ctrl:0 is only considered when bank_mask and row_mask are fully enabled (0xF). Otherwise the old register value is checked for identity.
Added add, subrev, and, or instructions to the old folding function.
Kill flag is cleared for the src0 (DPP register) as it may be copied into more than one user.
The pass is still disabled by default.
llvm-svn: 353513
Check that when SimplifyCFG is flattening a 'br', all their debug intrinsic instructions are removed, including any dbg.label referencing a label associated with the basic blocks being removed.
Differential Revision: https://reviews.llvm.org/D57444
llvm-svn: 353511
Summary:
FPCW contains the rounding mode control which we manipulate to implement fp to integer conversion by changing the roudning mode, storing the value to the stack, and then changing the rounding mode back. Because we didn't model FPCW and its dependency chain, other instructions could be scheduled into the middle of the sequence.
This patch introduces the register and adds it as an implciit def of FLDCW and implicit use of the FP binary arithmetic instructions and store instructions. There are more instructions that need to be updated, but this is a good start. I believe this fixes at least the reduced test case from PR40529.
Reviewers: RKSimon, spatel, rnk, efriedma, andrew.w.kaylor
Subscribers: dim, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D57735
llvm-svn: 353489
AArch64 NEON has a bunch of instructions with a "2" suffix that extract
the top half of the source vectors, instead of the bottom half. We have
some DAGCombines to try to take advantage of that. However, they
assumed that any EXTRACT_VECTOR was extracting the high half of the
vector in question.
This issue has apparently existed since the AArch64 backend was merged.
Fixes https://bugs.llvm.org/show_bug.cgi?id=40632 .
Differential Revision: https://reviews.llvm.org/D57862
llvm-svn: 353486
When a landing pad is calculated in a program that is compiled for micromips
with -fPIC flag, it will point to an even address.
Such an error will cause a segmentation fault, as the instructions in
micromips are aligned on odd addresses. This patch sets the last bit of the
offset where a landing pad is, to 1, which will effectively be an odd
address and point to the instruction exactly.
r344591 fixed this issue for -static compilation.
Patch by Aleksandar Beserminji.
Differential Revision: https://reviews.llvm.org/D57677
llvm-svn: 353480
This is part of https://bugs.llvm.org/show_bug.cgi?id=40442.
Vector legalization is implemented for the add/sub overflow opcodes.
UMULO/SMULO are also handled as far as legalization is concerned, but
they don't support vector expansion yet (so no tests for them).
The vector result widening implementation is suboptimal, because it
could result in a legalization loop.
Differential Revision: https://reviews.llvm.org/D57639
llvm-svn: 353464
This is pretty much directly ported from SelectionDAG. Doesn't include
the shift by non-constant but known bits version, since there isn't a
globalisel version of computeKnownBits yet.
This shows a disadvantage of targets not specifically which type
should be used for the shift amount. If type 0 is legalized before
type 1, the operations on the shift amount type use the wider type
(which are also less likely to legalize). This can be avoided by
targets specifying legalization actions on type 1 earlier than for
type 0.
llvm-svn: 353455
I noticed that we are missing this canonicalization in IR:
rL352515
...and then realized that we don't get this right in SDAG either,
so this has to be fixed first regardless of what we choose to do in IR.
The existing fold was limited to scalars and using the wrong predicate
to guard the transform. We have a boolean contents TLI query that can
be used to decide which direction to fold.
This may eventually lead back to the problems/question in:
https://bugs.llvm.org/show_bug.cgi?id=40486
...but it makes no difference to that yet.
Differential Revision: https://reviews.llvm.org/D57401
llvm-svn: 353433
Introduce a new function which handles instructions with multiple type
indices, but have the same number of vector elements.
Also legalize v2s16 shifts when applicable.
llvm-svn: 353432
This is intentionally a small step because it's hard to know exactly
where we might introduce a conflicting transform with the code that
tries to form wider shuffles. But I think this is safe - if we have
a wide shuffle with 2 operands, then we should do better with an
extract + narrow shuffle.
Differential Revision: https://reviews.llvm.org/D57867
llvm-svn: 353427
Modify GenerateConstantOffsetsImpl to create offsets that can be used
by indexed addressing modes. If formulae can be generated which
result in the constant offset being the same size as the recurrence,
we can generate a pre-indexed access. This allows the pointer to be
updated via the single pre-indexed access so that (hopefully) no
add/subs are required to update it for the next iteration. For small
cores, this can significantly improve performance DSP-like loops.
Differential Revision: https://reviews.llvm.org/D55373
llvm-svn: 353403
Summary:
Experimentally we found that promotion to scalars carries less benefits
than sinking and hoisting in LICM. When using MemorySSA, we build an
AliasSetTracker on demand in order to reuse the current infrastructure.
We only build it if less than AccessCapForMSSAPromotion exist in the
loop, a cap that is by default set to 250. This value ensures there are
no runtime regressions, and there are small compile time gains for
pathological cases. A much lower value (20) was found to yield a single
regression in the llvm-test-suite and much higher benefits for compile
times. Conservatively we set the current cap to a high value, but we will
explore lowering it when MemorySSA is enabled by default.
Reviewers: sanjoy, chandlerc
Subscribers: nemanjai, jlebar, Prazek, george.burgess.iv, jfb, jsji, llvm-commits
Differential Revision: https://reviews.llvm.org/D56625
llvm-svn: 353339
Since SystemZ supports counting of leading zeros with the FLOGR instruction,
isCheapToSpeculateCtlz() should return true, which it now does.
ISD::CTLZ_ZERO_UNDEF i32 is now handled the same way as ISD::CTLZ is, which
is needed since promotion to i64 is required and CTLZ_ZERO_UNDEF is only
expanded to CTLZ if it is Legal or Custom.
Review: Ulrich Weigand
https://reviews.llvm.org/D57710
llvm-svn: 353330
Don't lower BUILD_VECTORs to BYTE_MASK, but instead expose the BUILD_VECTORs
to the DAGCombiner and select them to VGBM in Select(). This allows the
DAGCombiner to understand the constant vector values.
For floating point, only all-zeros vectors are now generated with VGBM, as it
turned out to be somewhat complicated to handle any arbitrary constants,
while in practice this is very rare and hardly needed.
The SystemZ ISD opcodes z_byte_mask, z_vzero and z_vones have been removed.
Review: Ulrich Weigand
https://reviews.llvm.org/D57152
llvm-svn: 353325
ARMv8.1a CASP instructions need the first of the pair to be an even register
(otherwise the encoding is unallocated). We enforced this during assembly, but
not CodeGen before.
llvm-svn: 353308
The IPM sequence currently generated to compute the strcmp/memcmp
result will return INT_MIN for the "less than zero" case. While
this is in compliance with the standard, strictly speaking, it
turns out that common applications cannot handle this, e.g. because
they negate a comparison result in order to implement reverse
compares.
This patch changes code to use a different sequence that will result
in -2 for the "less than zero" case (same as GCC). However, this
requires that the two source operands of the compare instructions
are inverted, which breaks the optimization in removeIPMBasedCompare.
Therefore, I've removed this (and all of optimizeCompareInstr), and
replaced it with a mostly equivalent optimization in combineCCMask
at the DAGcombine level.
llvm-svn: 353304
The proposal in D56796 may cross the line because we're trying to avoid vectorization
transforms in generic DAG combining. So this is an alternate, later, x86-specific
translation of that patch.
There are several potential follow-ups to enhance this:
1. Allow extraction from non-zero element index.
2. Peek through extends of smaller width integers.
3. Support x86-specific conversion opcodes like X86ISD::CVTSI2P
Differential Revision: https://reviews.llvm.org/D56864
llvm-svn: 353302
We now print the implicit %st register on these instruction, but since they occur at the end of the line, FileCheck didn't see they were missing.
llvm-svn: 353222
Ensure the XOR in the waterfall loop for indirect addressing is considered a terminator.
Differential Revision: https://reviews.llvm.org/D57703
llvm-svn: 353207
The v2i64 argument is lowered to a bitcast of v4i32 build_vector.
This would then attempt to use the i32-element as the source of the
vector truncate. This really would need to collect 2 elements from the
build_vector to produce the intended truncate.
llvm-svn: 353202
rL352997 enabled ZERO_EXTEND from non-shuffle-able value types. I've disabled it for now to fix a regression identified by @asbirlea until I can fix this properly.
llvm-svn: 353198
