difficult on current ARM implementations for a few reasons.
1. Even though a single vmla has latency that is one cycle shorter than a pair
of vmul + vadd, a RAW hazard during the first (4? on Cortex-a8) can cause
additional pipeline stall. So it's frequently better to single codegen
vmul + vadd.
2. A vmla folowed by a vmul, vmadd, or vsub causes the second fp instruction to
stall for 4 cycles. We need to schedule them apart.
3. A vmla followed vmla is a special case. Obvious issuing back to back RAW
vmla + vmla is very bad. But this isn't ideal either:
vmul
vadd
vmla
Instead, we want to expand the second vmla:
vmla
vmul
vadd
Even with the 4 cycle vmul stall, the second sequence is still 2 cycles
faster.
Up to now, isel simply avoid codegen'ing fp vmla / vmls. This works well enough
but it isn't the optimial solution. This patch attempts to make it possible to
use vmla / vmls in cases where it is profitable.
A. Add missing isel predicates which cause vmla to be codegen'ed.
B. Make sure the fmul in (fadd (fmul)) has a single use. We don't want to
compute a fmul and a fmla.
C. Add additional isel checks for vmla, avoid cases where vmla is feeding into
fp instructions (except for the #3 exceptional case).
D. Add ARM hazard recognizer to model the vmla / vmls hazards.
E. Add a special pre-regalloc case to expand vmla / vmls when it's likely the
vmla / vmls will trigger one of the special hazards.
Work in progress, only A+B are enabled.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120960 91177308-0d34-0410-b5e6-96231b3b80d8
The only reasonable way I could find to do this is to provide an alternate
version of the addrmode6 operand with a different encoding function. Use it
for all the VLD-dup instructions for the sake of consistency.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120358 91177308-0d34-0410-b5e6-96231b3b80d8
The op11_8 field is the same for all of them so put it in the instruction
classes instead of specifying it separately for each instruction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120234 91177308-0d34-0410-b5e6-96231b3b80d8
'db', 'ib', 'da') instead of having that mode as a separate field in the
instruction. It's more convenient for the asm parser and much more readable for
humans.
<rdar://problem/8654088>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@119310 91177308-0d34-0410-b5e6-96231b3b80d8
We could be more aggressive about making this work for a larger range of constants,
but this seems like a good start.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@118201 91177308-0d34-0410-b5e6-96231b3b80d8
all of the different element sizes are pseudo instructions that map down to vext.8 underneath, with
the immediate shifted left to reflect the increased element size.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@118183 91177308-0d34-0410-b5e6-96231b3b80d8
half of the Q register), rather than with just regno. This allows us to unify the encodings for
a lot of different NEON instrucitons that differ only in whether they have Q or D register operands.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@117056 91177308-0d34-0410-b5e6-96231b3b80d8
1. Cortex-A8 load / store multiplies can only issue on ALU0.
2. Eliminate A8_Issue, A8_LSPipe will correctly limit the load / store issues.
3. Correctly model all vld1 and vld2 variants.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@116134 91177308-0d34-0410-b5e6-96231b3b80d8
"The register specified for a dregpair is the corresponding Q register, so to
get the pair, we need to look up the sub-regs based on the qreg. Create a
lookup function since we don't have access to TargetRegisterInfo here to
be able to use getSubReg(ARM::dsub_[01])."
Additionaly, fix the NEON VLD1* and VST1* instruction patterns not to use
the dregpair modifier for the 2xdreg versions. Explicitly specifying the two
registers as operands is more correct and more consistent with the other
instruction patterns. This enables further cleanup of special case code in the
disassembler as a nice side-effect.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113903 91177308-0d34-0410-b5e6-96231b3b80d8
an argument, so that we can distinguish instructions with the same register
classes but different numbers of registers (e.g., vld3 and vld4). Fix some
of the non-pseudo NEON ld/st instruction itineraries to reflect the number
of registers loaded or stored, not just the opcode name.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113854 91177308-0d34-0410-b5e6-96231b3b80d8
pseudo-instruction approach. Change ARMExpandPseudoInsts to use a table
to record all the NEON load/store information.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113812 91177308-0d34-0410-b5e6-96231b3b80d8
the VST pseudos. The VLD/VST scheduling still needs work (see pr6722), but
at least we shouldn't confuse the loads with the stores.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113473 91177308-0d34-0410-b5e6-96231b3b80d8
instructions prior to regalloc. Since it's getting a little close to
the 2.8 branch deadline, I'll have to leave the rest of the instructions
handled by the NEONPreAllocPass for now, but I didn't want to leave half
of the VLD instructions converted and the other half not.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112983 91177308-0d34-0410-b5e6-96231b3b80d8
vabd intrinsic and add and/or zext operations. In the case of vaba, this
also avoids the need for a DAG combine pattern to combine vabd with add.
Update tests. Auto-upgrade the old intrinsics.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112941 91177308-0d34-0410-b5e6-96231b3b80d8
add, and subtract operations with zero-extended or sign-extended vectors.
Update tests. Add auto-upgrade support for the old intrinsics.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112773 91177308-0d34-0410-b5e6-96231b3b80d8
IR add/sub operations with one or both operands sign- or zero-extended.
Auto-upgrade the old intrinsics.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112416 91177308-0d34-0410-b5e6-96231b3b80d8
all the other LDM/STM instructions. This fixes asm printer crashes when
compiling with -O0. I've changed one of the NEON tests (vst3.ll) to run
with -O0 to check this in the future.
Prior to this change VLDM/VSTM used addressing mode #5, but not really.
The offset field was used to hold a count of the number of registers being
loaded or stored, and the AM5 opcode field was expanded to specify the IA
or DB mode, instead of the standard ADD/SUB specifier. Much of the backend
was not aware of these special cases. The crashes occured when rewriting
a frameindex caused the AM5 offset field to be changed so that it did not
have a valid submode. I don't know exactly what changed to expose this now.
Maybe we've never done much with -O0 and NEON. Regardless, there's no longer
any reason to keep a count of the VLDM/VSTM registers, so we can use
addressing mode #4 and clean things up in a lot of places.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112322 91177308-0d34-0410-b5e6-96231b3b80d8
with the VST4 instructions. Until after register allocation, we want to
represent sets of adjacent registers by a single super-register. These
VST4 pseudo instructions have a single QQ or QQQQ source register operand.
They get expanded to the real VST4 instructions with 4 separate D register
operands. Once this conversion is complete, we'll be able to remove the
NEONPreAllocPass and avoid some fragile and hacky code elsewhere.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@112108 91177308-0d34-0410-b5e6-96231b3b80d8
immediate" operands. These functions have so far only been used for VMOV
but they also apply to other NEON instructions with modified immediate
operands. No functional changes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@105969 91177308-0d34-0410-b5e6-96231b3b80d8
the machine instruction representation of the immediate value to be encoded
into an integer with similar fields as the actual VMOV instruction. This makes
things easier for the disassembler, since it can just stuff the bits into the
immediate operand, but harder for the asm printer since it has to decode the
value to be printed. Testcase for the encoding will follow later when MC has
more support for ARM.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@105836 91177308-0d34-0410-b5e6-96231b3b80d8
- change isShuffleMaskLegal to show that all shuffles with 32-bit and 64-bit
elements are legal
- the Neon shuffle instructions do not support 64-bit elements, but we were
not checking for that before lowering shuffles to use them
- remove some 64-bit element vduplane patterns that are no longer needed
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@105586 91177308-0d34-0410-b5e6-96231b3b80d8
Reverse-merging r103156 into '.':
U lib/Target/ARM/ARMInstrNEON.td
U lib/Target/ARM/ARMRegisterInfo.h
U lib/Target/ARM/ARMBaseRegisterInfo.cpp
U lib/Target/ARM/ARMBaseInstrInfo.cpp
U lib/Target/ARM/ARMRegisterInfo.td
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103159 91177308-0d34-0410-b5e6-96231b3b80d8
These instructions use byte index in a control vector (M:Vm) to lookup byte
values in a table and generate a new vector (D:Vd). The table is specified via
a list of vectors, which can be:
{Dn}
{Dn D<n+1>}
{Dn D<n+1> D<n+2>}
{Dn D<n+1> D<n+2> D<n+3>}
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99789 91177308-0d34-0410-b5e6-96231b3b80d8
input to be v8i8 or v16i8, which buildvectors get canonicalized to.
This allows the patterns that were previously using a bare 'vnot' to
match, before they couldn't.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99754 91177308-0d34-0410-b5e6-96231b3b80d8
it as the format for the appropriate N3V*SL*<> classes. These instructions
require special handling of the M:Vm field which encodes the restricted Dm and
the lane index within Dm.
Examples are A8.6.325 VMLA, VMLAL, VMLS, VMLSL (by scalar):
vmlal.s32 q3, d2, d10[0]
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99690 91177308-0d34-0410-b5e6-96231b3b80d8
to now take a format argument. N3VDInt<> and N3VQInt<> are modified to take a
format argument as well.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99676 91177308-0d34-0410-b5e6-96231b3b80d8
to encode the byte location of the extracted result in the concatenation of the
operands, from the least significant end.
Modify VEXTd and VEXTq classes to use the format.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99659 91177308-0d34-0410-b5e6-96231b3b80d8
follow the N3RegFrm's operand order of D:Vd N:Vn M:Vm. The operand order of
N3RegVShFrm is D:Vd M:Vm N:Vn (notice that M:Vm is the first src operand).
Add a parent class N3Vf which requires passing a Format argument and which the
N3V class is modified to inherit from. N3V class represents the "normal"
3-Register NEON Instructions with N3RegFrm.
Also add a multiclass N3VSh_QHSD to represent clusters of NEON 3-Register Shift
Instructions and replace 8 invocations with it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99655 91177308-0d34-0410-b5e6-96231b3b80d8
dispatch to the appropriate routines to handle the different interpretations of
the shift amount encoded in the imm6 field. The Vd, Vm fields are interpreted
the same between the two, though.
See, for example, A8.6.367 VQSHL, VQSHLU (immediate) for N2RegVShLFrm format and
A8.6.368 VQSHRN, VQSHRUN for N2RegVShRFrm format.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99590 91177308-0d34-0410-b5e6-96231b3b80d8
These instructions are only needed for codegen, so I've removed all the
explicit encoding bits for now; they should be set in the same way as the for
VLDMD and VSTMD whenever we add encodings for VFP. The use of addrmode5
requires that the instructions be custom-selected so that the number of
registers can be set in the AM5Opc value.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99309 91177308-0d34-0410-b5e6-96231b3b80d8
of D registers. Add a separate VST1q instruction with a Q register
source operand for use by storeRegToStackSlot.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99265 91177308-0d34-0410-b5e6-96231b3b80d8
of D registers. Add a separate VLD1q instruction with a Q register
destination operand for use by loadRegFromStackSlot.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99261 91177308-0d34-0410-b5e6-96231b3b80d8
with changes to add a separate optional register update argument. Change all
the NEON instructions with address register writeback to use it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99095 91177308-0d34-0410-b5e6-96231b3b80d8
writeback, and refactor the existing double-spaced VST2 instructions.
These are only for the disassembler since codegen doesn't use them, at
least for now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99090 91177308-0d34-0410-b5e6-96231b3b80d8
load/stores with address register writeback, and use "odd" suffix to distinguish
instructions to access odd numbered registers (instead of "a" and "b").
No functional changes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99066 91177308-0d34-0410-b5e6-96231b3b80d8
writeback, and refactor the existing double-spaced VLD2 instructions.
These are only for the disassembler since codegen doesn't use them, at
least for now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99065 91177308-0d34-0410-b5e6-96231b3b80d8
--- Reverse-merging r98889 into '.':
U lib/Target/ARM/ARMInstrNEON.td
U lib/Target/ARM/ARMISelLowering.h
U lib/Target/ARM/ARMInstrInfo.td
U lib/Target/ARM/ARMInstrVFP.td
U lib/Target/ARM/ARMISelLowering.cpp
U lib/Target/ARM/ARMInstrFormats.td
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99010 91177308-0d34-0410-b5e6-96231b3b80d8
instructions to help disassembly.
We also changed the output of the addressing modes to omit the '+' from the
assembler syntax #+/-<imm> or +/-<Rm>. See, for example, A8.6.57/58/60.
And modified test cases to not expect '+' in +reg or #+num. For example,
; CHECK: ldr.w r9, [r7, #28]
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98745 91177308-0d34-0410-b5e6-96231b3b80d8
optional register update argument, which is currently unused -- when we add
support for that, it can just be a separate operand.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98683 91177308-0d34-0410-b5e6-96231b3b80d8
U test/CodeGen/ARM/tls2.ll
U test/CodeGen/ARM/arm-negative-stride.ll
U test/CodeGen/ARM/2009-10-30.ll
U test/CodeGen/ARM/globals.ll
U test/CodeGen/ARM/str_pre-2.ll
U test/CodeGen/ARM/ldrd.ll
U test/CodeGen/ARM/2009-10-27-double-align.ll
U test/CodeGen/Thumb2/thumb2-strb.ll
U test/CodeGen/Thumb2/ldr-str-imm12.ll
U test/CodeGen/Thumb2/thumb2-strh.ll
U test/CodeGen/Thumb2/thumb2-ldr.ll
U test/CodeGen/Thumb2/thumb2-str_pre.ll
U test/CodeGen/Thumb2/thumb2-str.ll
U test/CodeGen/Thumb2/thumb2-ldrh.ll
U utils/TableGen/TableGen.cpp
U utils/TableGen/DisassemblerEmitter.cpp
D utils/TableGen/RISCDisassemblerEmitter.h
D utils/TableGen/RISCDisassemblerEmitter.cpp
U Makefile.rules
U lib/Target/ARM/ARMInstrNEON.td
U lib/Target/ARM/Makefile
U lib/Target/ARM/AsmPrinter/ARMInstPrinter.cpp
U lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp
U lib/Target/ARM/AsmPrinter/ARMInstPrinter.h
D lib/Target/ARM/Disassembler
U lib/Target/ARM/ARMInstrFormats.td
U lib/Target/ARM/ARMAddressingModes.h
U lib/Target/ARM/Thumb2ITBlockPass.cpp
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98640 91177308-0d34-0410-b5e6-96231b3b80d8
(RISCDisassemblerEmitter) which emits the decoder functions for ARM and Thumb,
and the disassembler core which invokes the decoder function and builds up the
MCInst based on the decoded Opcode.
Added sub-formats to the NeonI/NeonXI instructions to further refine the NEONFrm
instructions to help disassembly.
We also changed the output of the addressing modes to omit the '+' from the
assembler syntax #+/-<imm> or +/-<Rm>. See, for example, A8.6.57/58/60.
And modified test cases to not expect '+' in +reg or #+num. For example,
; CHECK: ldr.w r9, [r7, #28]
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98637 91177308-0d34-0410-b5e6-96231b3b80d8
example, this:
(set DPR:$dst, (fsub (fneg (fmul DPR:$a, DPR:$b)), DPR:$dstin))
is ambiguous because DPR contains both f64 and v2f32. tblgen
currently accidentally picks f64 because it's first in the
regclass.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97955 91177308-0d34-0410-b5e6-96231b3b80d8
memory from three or four registers and VST2 (multiple two-element structures)
which stores to memory from two double-spaced registers.
A8.6.391 & A8.6.393
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97018 91177308-0d34-0410-b5e6-96231b3b80d8
three or four registers and VLD2 (multiple two-element structures) which loads
memory into two double-spaced registers.
A8.6.307 & A8.6.310
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96980 91177308-0d34-0410-b5e6-96231b3b80d8
since it has no pattern, there's not much point in distinguishing an "N2VS"
class for intrinsics anyway.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96525 91177308-0d34-0410-b5e6-96231b3b80d8
* Use "S" abbreviation for scalar single FP registers in class and pattern
names, instead of keeping the "D" (for "double") abbreviation and tacking on
an "s" elsewhere in the name.
* Move the scalar single FP register classes and patterns to be more
consistent with other definitions in the file.
* Rename "VNEGf32d" definition to "VNEGfd" for consistency.
* Deleted the N2VDIntsPat pattern; N2VSPat is good enough.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96521 91177308-0d34-0410-b5e6-96231b3b80d8
with the rest of the assembly output, is easier to read, and matches the
expected output for gcc's Neon tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93703 91177308-0d34-0410-b5e6-96231b3b80d8
adding an "i" to the suffix, indicating that the elements are integers, is
accepted but not part of the standard syntax. This helps us pass a few more
of the Neon tests from gcc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93677 91177308-0d34-0410-b5e6-96231b3b80d8
For VMOVv*i[16,32], op bit is don't care, and some cmode bits vary depending on
the immediate values.
Ref: Table A7-15 Modified immediate values for Advanced SIMD instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@90173 91177308-0d34-0410-b5e6-96231b3b80d8
This fixes the NEON asm printing so the "predicate" field is printed between the opcode and the data type suffix.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89706 91177308-0d34-0410-b5e6-96231b3b80d8
VDUPLND and VDUPLNQ to derive from N2V instead of N2VDup. VDUPLND and VDUPLNQ
now expect op19_18 and op17_16 as the first two args.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89699 91177308-0d34-0410-b5e6-96231b3b80d8
fully specified at this level. Subclasses of NLdStLN can specify selective
bit(s) for Inst{7-4}, as is done for VLD[234]LN* and VST[234]LN* inside
ARMInstrNEON.td.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89377 91177308-0d34-0410-b5e6-96231b3b80d8
will not accept negative values for these. LLVM's default operand printing
sign extends values, so that valid unsigned values appear as negative
immediates. Print all VMOV immediate operands as hex values to resolve this.
Radar 7372576.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86301 91177308-0d34-0410-b5e6-96231b3b80d8
a default value of zero. This is important for decoding the instructions.
Patch by Johnny Chen, with some changes from me, too.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84730 91177308-0d34-0410-b5e6-96231b3b80d8
Leave Inst{11-8}, which represents the starting byte index of the extracted
result in the concatenation of the operands and is left unspecified.
Patch by Johnny Chen.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84572 91177308-0d34-0410-b5e6-96231b3b80d8
