Although the previous code would construct a bundle and add the correct elements
to it, it would not finalise the bundle. This resulted in the InternalRead
markers not being added to the MachineOperands nor, more importantly, the
externally visible defs to the bundle itself. So, although the bundle was not
exposing the def, the generated code would be correct because there was no
optimisations being performed. When optimisations were enabled, the post
register allocator would kick in, and the hazard recognizer would reorder
operations around the load which would define the value being operated upon.
Rather than manually constructing the bundle, simply construct and finalise the
bundle via the finaliseBundle call after both MIs have been emitted. This
improves the code generation with optimisations where IMAGE_REL_ARM_MOV32T
relocations are emitted.
The changes to the other tests are the result of the bundle generation
preventing the scheduler from hoisting the moves across the loads. The net
effect of the generated code is equivalent, but, is much more identical to what
is actually being lowered.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209267 91177308-0d34-0410-b5e6-96231b3b80d8
Povray and dealII currently assert with "Overran sorted position" in
AssignTopologicalOrder. The problem is that performPostLD1Combine can
introduce cycles.
Consider:
(insert_vector_elt (INSERT_SUBREG undef,
(load (add %vreg0, Constant<8>), undef), <= A
TargetConstant<2>),
(load %vreg0, undef), <= B
Constant<1>)
This is turned into a LD1LANEpost node. However the address in A is not a
valid user of the post-incremented address of B in LD1LANEpost.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209242 91177308-0d34-0410-b5e6-96231b3b80d8
make the functions to set them non-static.
Move and rename the llvm specific backend options to avoid conflicting
with the clang option.
Paired with a backend commit to update.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209238 91177308-0d34-0410-b5e6-96231b3b80d8
The SplitIndexingFromLoad changes exposed a latent isel bug in the PowerPC64
backend. We matched an immediate offset with STWX8 even though it only
supports register offset.
The culprit is the complex-pattern predicate, SelectAddrIdx, which decides
that if the offset is not ISD::Constant it must be a register.
Many thanks to Bill Schmidt for testing this.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209219 91177308-0d34-0410-b5e6-96231b3b80d8
After discussion with Zoran, we have decided to temporarily revert this commit.
It's causing some difficult to resolve conflicts and we are under time pressure
to deliver an initial MIPS64r6 compiler.
We will re-apply an equivalent patch once the time pressure has passed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209211 91177308-0d34-0410-b5e6-96231b3b80d8
This allows the results of a ComplexPattern check to be distributed to separate
named Operands, instead of the current system where all results must apply (and
match perfectly) with a single Operand.
For example, if "some_addrmode" is a ComplexPattern producing two results, you
can write:
def : Pat<(load (some_addrmode GPR64:$base, imm:$offset)),
(INST GPR64:$base, imm:$offset)>;
This should allow neater instruction definitions in TableGen that don't put all
possible aspects of addressing into a single operand, but are still usable with
relatively simple C++ CodeGen idioms.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209206 91177308-0d34-0410-b5e6-96231b3b80d8
When multiple aliases overlap, the correct string to print can often be
determined purely by considering the InstAlias declarations in some particular
order. This allows the user to specify that order manually when desired,
without resorting to hacking around with the default lexicographical order on
Record instantiation, which is error-prone and ugly.
I was also mistaken about "add w2, w3, w4" being the same as "add w2, w3, w4,
uxtw". That's only true if Rn is the stack pointer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209199 91177308-0d34-0410-b5e6-96231b3b80d8
According to Intel Software Optimization Manual on Silvermont in some cases LEA
is better to be replaced with ADD instructions:
"The rule of thumb for ADDs and LEAs is that it is justified to use LEA
with a valid index and/or displacement for non-destructive destination purposes
(especially useful for stack offset cases), or to use a SCALE.
Otherwise, ADD(s) are preferable."
Differential Revision: http://reviews.llvm.org/D3826
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209198 91177308-0d34-0410-b5e6-96231b3b80d8
Currently the X86 backend doesn't support types larger than i128 very well. For
example an i192 multiply will assert in codegen when the 2nd argument is a constant and the constant got hoisted.
This fix changes the cost model to never hoist constants for types larger than
i128. Once the codegen issues have been resolved, the cost model can be updated
to allow also larger types.
This is related to <rdar://problem/16954938>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209162 91177308-0d34-0410-b5e6-96231b3b80d8
Instructions TZCNT (requires BMI1) and LZCNT (requires LZCNT), always
provide the operand size as output if the input operand is zero.
We can take advantage of this knowledge during instruction selection
stage in order to simplify a few corner case.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209159 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
When inserting an element that's coming from a vector load or a broadcast
of a vector (or scalar) load, combine the load into the insertps
instruction.
Added PerformINSERTPSCombine for the case where we need to fix the load
(load of a vector + insertps with a non-zero CountS).
Added patterns for the broadcasts.
Also added tests for SSE4.1, AVX, and AVX2.
Reviewers: delena, nadav, craig.topper
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D3581
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209156 91177308-0d34-0410-b5e6-96231b3b80d8
- On ARM/ARM64 we get a vrev because the shuffle matching code is really smart. We still unroll anything that's not v4i32 though.
- On X86 we get a pshufb with SSSE3. Required more cleverness in isShuffleMaskLegal.
- On PPC we get a vperm for v8i16 and v4i32. v2i64 is unrolled.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209123 91177308-0d34-0410-b5e6-96231b3b80d8
Rather than create a series of function calls to setup the library calls, create
a table with the information and just use the table to drive the configuration
of the library calls. This makes it easier to both inspect the list as well as
to modify it. NFC.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209089 91177308-0d34-0410-b5e6-96231b3b80d8
Windows on ARM uses R11 for the frame pointer even though the environment is a
pure Thumb-2, thumb-only environment. Replicate this behaviour to improve
Windows ABI compatibility. This register is used for fast stack walking, and
thus is part of the Windows ABI.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209085 91177308-0d34-0410-b5e6-96231b3b80d8
Use the ARMBaseRegisterInfo to query the frame register. The base register info
is aware of the frame register that is used for the frame pointer. Use that to
determine the frame register rather than duplicating the knowledge. Although,
the code path is slightly different in that it may return SP, that can only
occur if the frame pointer has been omitted in the machine function, which is
supposed to contain the desired value in that case.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209084 91177308-0d34-0410-b5e6-96231b3b80d8
This is mostly a mechanical change changing all the call sites to the newer
chained-function construction pattern. This removes the horrible 15-parameter
constructor for the CallLoweringInfo in favour of setting properties of the call
via chained functions. No functional change beyond the removal of the old
constructors are intended.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209082 91177308-0d34-0410-b5e6-96231b3b80d8
This is a preliminary step to help ease the construction of CallLoweringInfo.
Changing the construction to a chained function pattern requires that the
parameter be nullable. However, rather than copying the vector, save a pointer
rather than the reference to permit a late binding of the arguments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209080 91177308-0d34-0410-b5e6-96231b3b80d8
WoA uses COFF, not ELF. ARMISelLowering::createTLOF would previously return ELF
for any non-MachO platform. This was a missed site when the original change for
target format support for Windows on ARM was done.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209057 91177308-0d34-0410-b5e6-96231b3b80d8
were added in SSE2, no SSSE3. Found this while auditing all uses of
SSSE3 in the X86 target. I don't actually expect this to make
a significant difference on anything and I don't have any detailed test
cases but I updated the existing test cases that already covered some of
this code path.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209056 91177308-0d34-0410-b5e6-96231b3b80d8
vselects with constant masks, after legalization, will get turned into
specialized shuffle_vectors so they can be matched to blend+imm
instructions.
Fixed some tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209044 91177308-0d34-0410-b5e6-96231b3b80d8
LowerVSELECT will, if possible, generate a X86ISD::BLENDI DAG node if the
condition is constant and we can emit that instruction, given the
subtarget.
This is not enough for all cases. An additional SELECTCombine optimization
will be committed.
Fixed tests that were expecting variable blends but where a blend+imm can
be generated.
Added test where we can't emit blend+immediate.
Added avx2 blend+imm tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209043 91177308-0d34-0410-b5e6-96231b3b80d8
No functionality change intended. The types that previously were set to
lower as Expand or Legal are doing the same thing with this lowering
function.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209042 91177308-0d34-0410-b5e6-96231b3b80d8
This will allow us to use a single MachineInstr to represent
instructions which behave the same but have different encodings
on some subtargets.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209028 91177308-0d34-0410-b5e6-96231b3b80d8
This reverts commit r208934.
The patch depends on aliases to GEPs with non zero offsets. That is not
supported and fairly broken.
The good news is that GlobalAlias is being redesigned and will have support
for offsets, so this patch should be a nice match for it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208978 91177308-0d34-0410-b5e6-96231b3b80d8
TableGen has a fairly dubious heuristic to decide whether an alias should be
printed: does the alias have lest operands than the real instruction. This is
bad enough (particularly with no way to override it), but it should at least be
calculated consistently for both strings.
This patch implements that logic: first get the *correct* string for the
variant, in the same way as the Matcher, without guessing; then count the
number of whitespace chars.
There are basically 4 changes this brings about after the previous
commits; all of these appear to be good, so I have changed the tests:
+ ARM64: we print "neg X, Y" instead of "sub X, xzr, Y".
+ ARM64: we skip implicit "uxtx" and "uxtw" modifiers.
+ Sparc: we print "mov A, B" instead of "or %g0, A, B".
+ Sparc: we print "fcmpX A, B" instead of "fcmpX %fcc0, A, B"
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208969 91177308-0d34-0410-b5e6-96231b3b80d8
The canonical syntax is "fcmXY ..., #0.0".
This will be tested when the TableGen "should I print this Alias" heuristic is
fixed (very soon).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208968 91177308-0d34-0410-b5e6-96231b3b80d8
This alias appears not to have an appropriate PrintMethod. Normally, I'd look
into it, but since AArch64 is disappearing soon it's probably not worth it.
This will be tested when the TableGen "should I print this Alias" heuristic is
fixed (very soon).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208967 91177308-0d34-0410-b5e6-96231b3b80d8
These aliases are handled entirely in C++ and only having TableGen InstAliases
for some of them was confusing LLVM.
This will be tested when the TableGen "should I print this Alias" heuristic is
fixed (very soon).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208966 91177308-0d34-0410-b5e6-96231b3b80d8
Certainly not without having a custom PrintMethod to invert the immediate
beforehand. But probably not at all.
This will be tested when the TableGen "should I print this Alias" heuristic is
fixed (very soon).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208964 91177308-0d34-0410-b5e6-96231b3b80d8
In AT&T syntax, we should probably print the full "movl" or "movw". TableGen
used to ignore these aliases because it was miscounting the number of operands.
This fixes the issue.
This will be tested when the TableGen "should I print this Alias"
heuristic is fixed (very soon).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208963 91177308-0d34-0410-b5e6-96231b3b80d8
Actually, MOV sometimes is canonical, but for now this is a better
approximation than what's there.
This will be tested when the TableGen "should I print this Alias" heuristic is
fixed (very soon).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208962 91177308-0d34-0410-b5e6-96231b3b80d8
You can perform (say) an fcmle operation by swapping the operands on an fcmge,
but it shouldn't be printed like that.
This will be tested when the TableGen "should I print this Alias" heuristic is
fixed (very soon).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208961 91177308-0d34-0410-b5e6-96231b3b80d8
We accept "ldr w3, [x1, #-1]" as a convenience, but we should still print the
canonical "ldur" form.
This will be tested when the TableGen "should I print this Alias" heuristic is
fixed (very soon).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208960 91177308-0d34-0410-b5e6-96231b3b80d8
If an ANDS instruction has Rd == ZR it should be printed as TST since
its only effect is on the flags register NZCV.
This will be tested when the TableGen "should I print this Alias"
heuristic is fixed (very soon).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208959 91177308-0d34-0410-b5e6-96231b3b80d8
MOV is almost always the right thing to print if possile. People understand it.
This will be tested when the TableGen "should I print this Alias" heuristic is
fixed (very soon).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208958 91177308-0d34-0410-b5e6-96231b3b80d8
For example, the full instruction "sub w0, wzr, w1, uxtw" could print as either
"neg w0, w1" or "sub w0, wzr, w1". The former is better.
This will be tested when the TableGen "should I print this Alias" heuristic is
fixed (very soon).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208957 91177308-0d34-0410-b5e6-96231b3b80d8
You can write "lslv w0, w1, w2" (probably for legacy reasons), but it should be
printed as simply "lsl".
This will be tested when the TableGen "should I print this Alias" heuristic is
fixed (very soon).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208956 91177308-0d34-0410-b5e6-96231b3b80d8
Add some Windows on ARM specific library calls. These are provided by msvcrt,
and can be used to perform integer to floating-point conversions (and
vice-versa) mirroring similar functions in the RTABI.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208949 91177308-0d34-0410-b5e6-96231b3b80d8
This commit implements two command line switches -global-merge-on-external
and -global-merge-aligned, and both of them are false by default, so this
optimization is disabled by default for all targets.
For ARM64, some back-end behaviors need to be tuned to get this optimization
further enabled.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208934 91177308-0d34-0410-b5e6-96231b3b80d8
Added target specific combine rules to fold blend intrinsics according
to the following rules:
1) fold(blend A, A, Mask) -> A;
2) fold(blend A, B, <allZeros>) -> A;
3) fold(blend A, B, <allOnes>) -> B.
Added two new tests to verify that the new folding rules work for all
the optimized blend intrinsics.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208895 91177308-0d34-0410-b5e6-96231b3b80d8
We now use SReg_* for integer types and VReg_* for floating-point types.
This should help simplify the SIFixSGPRCopies pass and no longer causes
ISel to insert a COPY after termiator instuctions that output a value.
This change is covered by exisitng tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208888 91177308-0d34-0410-b5e6-96231b3b80d8
Previously, TableGen assumed that every aliased operand consumed precisely 1
MachineInstr slot (this was reasonable because until a couple of days ago,
nothing more complicated was eligible for printing).
This allows a couple more ARM64 aliases to print so we can remove the special
code.
On the X86 side, I've gone for explicit AT&T size specifiers as the default, so
turned off a few of the aliases that would have just started printing.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208880 91177308-0d34-0410-b5e6-96231b3b80d8
In all cases, if a "mov" alias exists, it is the canonical form of the
instruction. Now that TableGen can support aliases containing syntax variants,
we can enable them and improve the quality of the asm output.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208874 91177308-0d34-0410-b5e6-96231b3b80d8
To get at least one use of the change (and some actual tests) in with its
commit, I've enabled the AArch64 & ARM64 NEON mov aliases.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208867 91177308-0d34-0410-b5e6-96231b3b80d8
Previously, we ignored the difference between V64 and V128 when parsing
assembly: they both got mapped to registers in the FPR128 class. This is
basically harmless at the moment because they both print and encode the same
way. However, it will affect the printing of aliases.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208866 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
No support for symbols in place of the immediate yet since it requires new
relocations.
Depends on D3671
Reviewers: jkolek, zoran.jovanovic, vmedic
Reviewed By: vmedic
Differential Revision: http://reviews.llvm.org/D3689
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208858 91177308-0d34-0410-b5e6-96231b3b80d8
For example
tzcntl %edi, %ebx
testl %edi, %edi
je .label
can be rewritten into
tzcntl %edi, %ebx
jb .label
A minor complication is that tzcnt sets CF instead of ZF when the input
is zero, we have to rewrite users of the flags from ZF to CF. Currently
we recognize patterns using lzcnt, tzcnt and popcnt.
Differential Revision: http://reviews.llvm.org/D3454
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208788 91177308-0d34-0410-b5e6-96231b3b80d8
The UDF instruction is a reserved undefined instruction space. The assembler
mnemonic was introduced with ARM ARM rev C.a. The instruction is not predicated
and the immediate constant is ignored by the CPU. Add support for the three
encodings for this instruction.
The changes to the invalid instruction test is due to the fact that the invalid
instructions actually overlap with the undefined instruction. Introduction of
the new instruction results in a partial decode as an undefined sequence. Drop
the tests as they are invalid instruction patterns anyways.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208751 91177308-0d34-0410-b5e6-96231b3b80d8
member variable and sink the initialization of crbits into the
subtarget feature reset code.
No functional change, but this refactor will be used in a future
commit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208726 91177308-0d34-0410-b5e6-96231b3b80d8
This commit was already commited as revision rL208689 and discussd in
phabricator revision D3704.
But the test file was crashing on OS X and windows.
I fixed the test file in the same way as in rL208340.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208711 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
This required a new instruction group representing the 32-bit subset of
MIPS-3 that was available in MIPS32R2.
To limit the number of tests required, only one 32-bit and one 64-bit ISA
prior to MIPS32/MIPS64 are tested.
rdhwr has been deliberately left without an ISA annotation for now. This is
because the assembler and CodeGen disagree on when the instruction is
available. Strictly speaking, it is only available in MIPS32r2 and
MIPS64r2. However, it is emulated by a kernel trap on earlier ISA's and is
necessary for TLS so CodeGen should emit it on older ISA's too.
Depends on D3696
Reviewers: vmedic
Reviewed By: vmedic
Differential Revision: http://reviews.llvm.org/D3697
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208690 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
We are currently very close to the 32-bit limit of the current assembler
implementation. This is because there is no way to represent an instruction
that is available in, for example, Mips3 or Mips32. We have to define a
feature bit that represents this.
This patch cleans up a pair of redundant feature bits and slightly postpones the
point we will reach the limit.
Reviewers: zoran.jovanovic, jkolek, vmedic
Reviewed By: vmedic
Differential Revision: http://reviews.llvm.org/D3703
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208685 91177308-0d34-0410-b5e6-96231b3b80d8
Normally, patterns like (add x, (setcc cc ...)) will be folded into
(csel x, x+1, not cc). However, if there is a ZEXT after SETCC, they
won't be folded. This patch recognizes the ZEXT and allows the
generation of CSINC.
This patch fixes bug 19680.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208660 91177308-0d34-0410-b5e6-96231b3b80d8
r208453 added support for having sret on the second parameter. In that
change, the code for copying sret into a virtual register was hoisted
into the loop that lowers formal parameters. This caused a "Wrong
topological sorting" assertion failure during scheduling when a
parameter is passed in memory. This change undoes that by creating a
second loop that deals with sret.
I'm worried that this fix is incomplete. I don't fully understand the
dependence issues. However, with this change we produce the same DAGs
we used to produce, so if they are broken, they are just as broken as
they have always been.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208637 91177308-0d34-0410-b5e6-96231b3b80d8
The current patterns for REV16 misses mostn __builtin_bswap16() due to
legalization promoting the operands to from load/stores toi32s and then
truncing/extending them. This patch adds new patterns that catch the resultant
DAGs and codegens them to rev16 instructions. Tests included.
rdar://15353652
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208620 91177308-0d34-0410-b5e6-96231b3b80d8
This is a slightly different approach to AArch64 (the base instruction
definitions aren't quite right for that to work), but achieves the
same thing and reduces C++ hackery in AsmParser.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208605 91177308-0d34-0410-b5e6-96231b3b80d8
Summary: The 'mul' line of the test is temporarily commented out because it currently matches the MIPS32 mul instead of the MIPS32r6 mul. This line will be uncommented when we disable the MIPS32 mul on MIPS32r6.
Reviewers: jkolek, zoran.jovanovic, vmedic
Reviewed By: vmedic
Differential Revision: http://reviews.llvm.org/D3668
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208576 91177308-0d34-0410-b5e6-96231b3b80d8
If this code triggers, any immediate has already been validated so it can't
possibly trigger a diagnostic.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208564 91177308-0d34-0410-b5e6-96231b3b80d8
In terms of assembly, these have too much overlap to be neatly modelled as
disjoint classes: in many cases "lsl" is an acceptable alternative to either
"uxtw" or "uxtx".
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208563 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
This required a new instruction group representing the 32-bit subset of
MIPS-V that was available in MIPS32R2
Most of these instructions are correctly rejected but with the wrong error
message. These have been placed in a separate test for now. It happens
because many of the MIPS V instructions have not been implemented.
Depends on D3694
Reviewers: vmedic
Reviewed By: vmedic
Differential Revision: http://reviews.llvm.org/D3695
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208546 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
DCL[ZO] are now correctly marked as being MIPS64 instructions. This has no
effect on the CodeGen tests since expansion of i64 prevented their use
anyway.
The check for MIPS16 to prevent the use of CLZ no longer prevents DCLZ as
well. This is not a functional change since DCLZ is still prohibited by
being a MIPS64 instruction (MIPS16 is only compatible with MIPS32).
No functional change
Reviewers: vmedic
Reviewed By: vmedic
Differential Revision: http://reviews.llvm.org/D3694
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208544 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
dsbh and dshd are not available on Mips32r2. No codegen test changes
required since expansion of i64 prevented the use of these instructions
anyway.
Depends on D3690
Reviewers: vmedic
Reviewed By: vmedic
Differential Revision: http://reviews.llvm.org/D3692
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208542 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
No functional change.
The minor change to the MIPS16 code is in preparation for a patch that will handle 32-bit FPIdx instructions separately to 64-bit (because they were added in different revisions)
Depends on D3677
Reviewers: rkotler, vmedic
Reviewed By: vmedic
Differential Revision: http://reviews.llvm.org/D3690
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208541 91177308-0d34-0410-b5e6-96231b3b80d8
Unfortunately, since ARM64 models all these instructions as aliases,
the checks need to be done at the time the alias is seen rather than
during instruction validation as AArch64 does it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208529 91177308-0d34-0410-b5e6-96231b3b80d8
1) Changed gather and scatter intrinsics. Now they are aligned with GCC built-ins. There is no more non-masked form. Masked intrinsic receives -1 if all lanes are executed.
2) I changed the function that works with intrinsics inside X86ISelLowering.cpp. I put all intrinsics in one table. I did it for INTRINSICS_W_CHAIN and plan to put all intrinsics from WO_CHAIN set to the same table in order to avoid the long-long "switch". (I wanted to use static map initialization that allowed by C++11 but I wasn't able to compile it on VS2012).
3) I added gather/scatter prefetch intrinsics.
4) I fixed MRMm encoding for masked instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208522 91177308-0d34-0410-b5e6-96231b3b80d8
Support for the intrinsics that read from and write to global named registers
is added for r1, r2 and r13 (depending on the subtarget).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208509 91177308-0d34-0410-b5e6-96231b3b80d8
We must validate the value type in TLI::getRegisterByName, because if we
don't and the wrong type was used with the IR intrinsic, then we'll assert
(because we won't be able to find a valid register class with which to
construct the requested copy operation). For PPC64, additionally, the type
information is necessary to decide between the 64-bit register and the 32-bit
subregister.
No functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208508 91177308-0d34-0410-b5e6-96231b3b80d8
The counter-loops formation pass needs to know what operations might be
function calls (because they can't appear in counter-based loops). On PPC32,
128-bit shifts might be runtime calls (even though you can't use __int128 on
PPC32, it seems that SROA might form them).
Fixes PR19709.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208501 91177308-0d34-0410-b5e6-96231b3b80d8
When lowering build_vector to an insertps, we would still lower it, even
if the source vectors weren't v4x32. This would break on avx if the source
was a v8x32. We now check the type of the source vectors.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208487 91177308-0d34-0410-b5e6-96231b3b80d8
We were swapping the true & false results while testing for FMAX/FMIN,
but not putting them back to the original state if the later checks
failed.
Should fix PR19700.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208469 91177308-0d34-0410-b5e6-96231b3b80d8
This reverts commit r200561.
This calling convention was an attempt to match the MSVC C++ ABI for
methods that return structures by value. This solution didn't scale,
because it would have required splitting every CC available on Windows
into two: one for methods and one for free functions.
Now that we can put sret on the second arg (r208453), and Clang does
that (r208458), revert this hack.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208459 91177308-0d34-0410-b5e6-96231b3b80d8
MSVC always places the implicit sret parameter after the implicit this
parameter of instance methods. We used to handle this for
x86_thiscallcc by allocating the sret parameter on the stack and leaving
the this pointer in ecx, but that doesn't handle alternative calling
conventions like cdecl, stdcall, fastcall, or the win64 convention.
Instead, change the verifier to allow sret on the second parameter.
This also requires changing the Mips and X86 backends to return the
argument with the sret parameter, instead of assuming that the sret
parameter comes first.
The Sparc backend also returns sret parameters in a register, but I
wasn't able to update it to handle secondary sret parameters. It
currently calls report_fatal_error if you feed it an sret in the second
parameter.
Reviewers: rafael.espindola, majnemer
Differential Revision: http://reviews.llvm.org/D3617
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208453 91177308-0d34-0410-b5e6-96231b3b80d8
This patch adds support to ARM for custom lowering of the
llvm.{u|s}add.with.overflow.i32 intrinsics for i32/i64. This is particularly useful
for handling idiomatic saturating math functions as generated by
InstCombineCompare.
Test cases included.
rdar://14853450
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208435 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
This required a new instruction group representing the 32-bit subset of
MIPS-IV that was available in MIPS32
A small number of instructions are correctly rejected but with the wrong error
message. These have been placed in a separate test for now.
Depends on D3676
Reviewers: vmedic
Reviewed By: vmedic
Differential Revision: http://reviews.llvm.org/D3677
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208414 91177308-0d34-0410-b5e6-96231b3b80d8
When using the ARM AAPCS, HFAs (Homogeneous Floating-point Aggregates) must
be passed in a block of consecutive floating-point registers, or on the stack.
This means that unused floating-point registers cannot be back-filled with
part of an HFA, however this can currently happen. This patch, along with the
corresponding clang patch (http://reviews.llvm.org/D3083) prevents this.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208413 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
This required a new instruction group representing the 32-bit subset of
MIPS-III that was available in MIPS32
A small number of instructions are correctly rejected but with the wrong error
message. These have been placed in a separate test for now.
There's some obvious InstAlias's that ought to be marked MIPS-III but arent.
This is because they are not currently tested. I intend to catch these with
a final pass through the tablegen records to find tablegen records without
ISA annotations.
Depends on D3674
Reviewers: vmedic
Reviewed By: vmedic
Differential Revision: http://reviews.llvm.org/D3675
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208408 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
Adds MIPS32r6/MIPS64r6 and checks the compatibility requirements for these
processors.
I've also included comments to describe removed and re-encoded instructions,
along with placeholder def's for the new instructions but there are no
functional changes to codegen at this point.
Reviewers: jkolek, vmedic
Reviewed By: vmedic
Differential Revision: http://reviews.llvm.org/D3622
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208399 91177308-0d34-0410-b5e6-96231b3b80d8
Handle lowering of global addresses for PIC mode compilation on Windows. Always
use the movw/movt load to load the address as Windows on ARM requires ARMv7+ and
is a pure Thumb environment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208385 91177308-0d34-0410-b5e6-96231b3b80d8
This patch doesn't introduce any functionality change. Test cases will be
added later when v5 support is added.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208349 91177308-0d34-0410-b5e6-96231b3b80d8
This patch teaches the backend how to combine packed SSE2/AVX2 arithmetic shift
intrinsics.
The rules are:
- Always fold a packed arithmetic shift by zero to its first operand;
- Convert a packed arithmetic shift intrinsic dag node into a ISD::SRA only if
the shift count is known to be smaller than the vector element size.
This patch also teaches to function 'getTargetVShiftByConstNode' how fold
target specific vector shifts by zero.
Added two new tests to verify that the DAGCombiner is able to fold
sequences of SSE2/AVX2 packed arithmetic shift calls.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208342 91177308-0d34-0410-b5e6-96231b3b80d8
The parsing of ADD/SUB shifted immediates needs to be done explicitly so
that better diagnostics can be emitted, as a side effect this also
removes some of the hacks in the current method of handling this operand
type.
Additionally remove manual CMP aliasing to ADD/SUB and use InstAlias
instead.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208329 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
These instructions were added in MIPS-I, and MIPS-II but were removed in
MIPS-III. Interestingly, GAS continues to accept them when assembling for
MIPS-III.
For the moment, these instructions will follow GAS and accept them for
MIPS-III and newer but this will be tightened up when the invalid-*.s
tests are added.
Depends on D3647
Reviewers: vmedic
Reviewed By: vmedic
Differential Revision: http://reviews.llvm.org/D3648
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208311 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
A small number of instructions are rejected with the wrong error message.
These have been placed in a separate test for now. There seems to be some
parsing quirk that triggers when these instructions are disabled.
Depends on D3571
Reviewers: vmedic
Reviewed By: vmedic
Differential Revision: http://reviews.llvm.org/D3647
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208305 91177308-0d34-0410-b5e6-96231b3b80d8
The old method used by X86TTI to determine partial-unrolling thresholds was
messy (because it worked by testing target features), and also would not
correctly identify the target CPU if certain target features were disabled.
After some discussions on IRC with Chandler et al., it was decided that the
processor scheduling models were the right containers for this information
(because it is often tied to special uop dispatch-buffer sizes).
This does represent a small functionality change:
- For generic x86-64 (which uses the SB model and, thus, will get some
unrolling).
- For AMD cores (because they still currently use the SB scheduling model)
- For Haswell (based on benchmarking by Louis Gerbarg, it was decided to bump
the default threshold to 50; we're working on a test case for this).
Otherwise, nothing has changed for any other targets. The logic, however, has
been moved into BasicTTI, so other targets may now also opt-in to this
functionality simply by setting LoopMicroOpBufferSize in their processor
model definitions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208289 91177308-0d34-0410-b5e6-96231b3b80d8
This adds FK_SecRel_2 relocation support to ARM. This enables the building of
object files for armv7-windows-msvc which enables CodeView line tables for
debugging as opposed to armv7-windows-itanium which currently uses DWARF.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208273 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
Vectors built with zeros and elements in the same order as another
(source) vector are optimized to be built using a single insertps
instruction.
Also optimize when we move one element in a vector to a different place
in that vector while zeroing out some of the other elements.
Further optimizations are possible, described in TODO comments.
I will be implementing at least some of them in the near future.
Added some tests for different cases where this optimization triggers.
Reviewers: nadav, delena, craig.topper
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D3521
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208271 91177308-0d34-0410-b5e6-96231b3b80d8
The loop stream detector (LSD) on modern Intel cores, which optimizes the
execution of small loops, has limits on the number of taken branches in
addition to uop-count limits (modern AMD cores have similar limits).
Unfortunately, at the IR level, estimating the number of branches that will be
taken is difficult. For one thing, it strongly depends on later passes (block
placement, etc.). The original implementation took a conservative approach and
limited the maximal BB DFS depth of the loop. However, fairly-extensive
benchmarking by several of us has revealed that this is the wrong approach. In
fact, there are zero known cases where the branch limit prevents a detrimental
unrolling (but plenty of cases where it does prevent beneficial unrolling).
While we could improve the current branch counting logic by incorporating
branch probabilities, this further complication seems unjustified without a
motivating regression. Instead, unless and until a regression appears, the
branch counting will be removed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208255 91177308-0d34-0410-b5e6-96231b3b80d8
Given a FMA family (e.g., 213, 231), not all the variants (i.e., register or
memory) are commutable.
E.g., for the 213 family (with the syntax src1, src2, src3):
fmaXXX213 A, B, reg3/mem3 == fmaXXX213 B, A, reg3/mem3
Now consider the 231 family:
fmaXXX231 A, B, reg3 == fmaXXX231 A, reg3, B
But
fmaXXX231 A, B, mem3 != fmaXXX231 A, mem3, B
Indeed, mem3 cannot be the second argument of the memory variant of fmaXXX231.
Working on a reduced test case!
<rdar://problem/16800495>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208252 91177308-0d34-0410-b5e6-96231b3b80d8
default architecture for reasonable modern x86 processors, actually be
modern. This processor model should essentially be "tuned" for modern
x86 chips as much as possible without undue penalties on any specific
architecture. Previously we weren't even using the nice scheduling
models. There are a few other tweaks needed here, but this change at
least I have benchmarked across a decent swatch of chips (intel's
clovertown, westmere, and sandybridge; amd's istanbul) and seen no
significant regressions.
If anyone has suggested ways to test this, just let me know. Somewhat
alarmingly, no existing tests failed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208230 91177308-0d34-0410-b5e6-96231b3b80d8
this patch disables the dead register elimination pass and the load/store pair
optimization pass at -O0. The ILP optimizations don't require the optimization
level to be checked because the call to addILPOpts is predicated with the
necessary check. The AdvSIMDScalar pass is disabled by default at all
optimization levels. This patch leaves that pass disabled by default.
Also, move command-line options into ARM64TargetMachine.cpp and add a few
additional flags to aid in debugging. This fixes an issue with the
-debug-pass=Structure flag where passes were printed, but not actually run
(i.e., AdvSIMDScalar pass).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208223 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
These processors will only be available for the integrated assembler at
first (CodeGen will emit a fatal error saying they are not implemented).
The intention is to work through the existing instructions and correctly
annotate the ISA they were added in so that we have a sufficiently good
base to start MIPS64r6 development. MIPS64r6 removes/re-encodes certain
instructions and I believe it is best to define ISA's using set-union's
as far as possible rather than using set-subtraction.
Reviewers: vmedic
Subscribers: emaste, llvm-commits
Differential Revision: http://reviews.llvm.org/D3569
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208221 91177308-0d34-0410-b5e6-96231b3b80d8
When performing a scalar comparison that feeds into a vector select,
it's actually better to do the comparison on the vector side: the
scalar route would be "CMP -> CSEL -> DUP", the vector is "CM -> DUP"
since the vector comparisons are all mask based.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208210 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
One small functional change. The recently added PAUSE instruction now has
the HasStdEnc predicate which was accidentally removed by a Requires<>.
Depends on D3640
Reviewers: vmedic
Reviewed By: vmedic
Differential Revision: http://reviews.llvm.org/D3641
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208209 91177308-0d34-0410-b5e6-96231b3b80d8
The AAPCS states that values passed in registers must have a value as though
they had been loaded with "LDR". LDR is equivalent to "LD1.64 vX.1D" - that is,
loading scalars to vector registers and loading 1-element vectors is equivalent.
The logic implemented here is to ensure that at all call boundaries and during
formal argument lowering all vectors are treated as their bitwidth-based floating
point scalar counterpart, which is always one of f64 or f128 (v2i32 -> f64,
v4i32 -> f128 etc). A BITCAST is inserted so that the appropriate REV will be
generated during code generation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208198 91177308-0d34-0410-b5e6-96231b3b80d8
Because we've canonicalised on using LD1/ST1, every time we do a bitcast
between vector types we must do an equivalent lane reversal.
Consider a simple memory load followed by a bitconvert then a store.
v0 = load v2i32
v1 = BITCAST v2i32 v0 to v4i16
store v4i16 v2
In big endian mode every memory access has an implicit byte swap. LDR and
STR do a 64-bit byte swap, whereas LD1/ST1 do a byte swap per lane - that
is, they treat the vector as a sequence of elements to be byte-swapped.
The two pairs of instructions are fundamentally incompatible. We've decided
to use LD1/ST1 only to simplify compiler implementation.
LD1/ST1 perform the equivalent of a sequence of LDR/STR + REV. This makes
the original code sequence: v0 = load v2i32
v1 = REV v2i32 (implicit)
v2 = BITCAST v2i32 v1 to v4i16
v3 = REV v4i16 v2 (implicit)
store v4i16 v3
But this is now broken - the value stored is different to the value loaded
due to lane reordering. To fix this, on every BITCAST we must perform two
other REVs:
v0 = load v2i32
v1 = REV v2i32 (implicit)
v2 = REV v2i32
v3 = BITCAST v2i32 v2 to v4i16
v4 = REV v4i16
v5 = REV v4i16 v4 (implicit)
store v4i16 v5
This means an extra two instructions, but actually in most cases the two REV
instructions can be combined into one. For example:
(REV64_2s (REV64_4h X)) === (REV32_4h X)
There is also no 128-bit REV instruction. This must be synthesized with an
EXT instruction.
Most bitconverts require some sort of conversion. The only exceptions are:
a) Identity conversions - vNfX <-> vNiX
b) Single-lane-to-scalar - v1fX <-> fX or v1iX <-> iX
Even though there are hundreds of changed lines, I have a fairly high confidence
that they are somewhat correct. The changes to add two REV instructions per
bitcast were pretty mechanical, and once I'd done that I threw the resulting
.td at a script I wrote which combined the two REVs together (and added
an EXT instruction, for f128) based on an instruction description I gave it.
This was much less prone to error than doing it all manually, plus my brain
would not just have melted but would have vapourised.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208194 91177308-0d34-0410-b5e6-96231b3b80d8
This completes the port of r204814 (cpirker "AArch64_BE function argument
passing for ARM ABI") from AArch64 to ARM64, and fixes a bunch of issues
found during later development along the way. The biggest of these was
that the alignment fixup logic wasn't replicated into all the places it
should have been.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208192 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
The overall idea is to chop the Predicates list into subsets that are
usually overridden independently. This allows subclasses to partially
override the predicates of their superclasses without having to re-add all
the existing predicates.
This patch starts the process by moving HasStdEnc into a new
EncodingPredicates list and almost everything else into
AdditionalPredicates.
It has revealed a couple likely bugs where 'let Predicates' has removed
the HasStdEnc predicate.
No functional change (confirmed by diffing tablegen-erated files).
Depends on D3549, D3506
Reviewers: vmedic
Differential Revision: http://reviews.llvm.org/D3550
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208184 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
This will make it easier to prove that a more complicated change in the
following commit is non-functional.
No functional change.
Depends on D3506
Reviewers: vmedic
Reviewed By: vmedic
Differential Revision: http://reviews.llvm.org/D3549
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208179 91177308-0d34-0410-b5e6-96231b3b80d8
Mark up additional instructions which are part of the function prologue as
MachineFrameSetup. These instructions are part of the function prologue,
emitted by the PEI pass to setup the stack for use in the activating frame.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208153 91177308-0d34-0410-b5e6-96231b3b80d8
The ARM::BLX instruction is an ARM mode instruction. The Windows on ARM target
is limited to Thumb instructions. Correctly use the thumb mode tBLXr
instruction. This would manifest as an errant write into the object file as the
instruction is 4-bytes in length rather than 2. The result would be a corrupted
object file that would eventually result in an executable that would crash at
runtime.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208152 91177308-0d34-0410-b5e6-96231b3b80d8
remove it from the list of unspilled registers. Otherwise the following
attempt to keep the stack aligned by picking an extra GPR register to
spill will not work as it picks up r11.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208129 91177308-0d34-0410-b5e6-96231b3b80d8
Before this patch, the backend always emitted a store+load sequence to
bitconvert from f64 to i64 the input operand of a ISD::BITCAST dag node that
performed a bitconvert from type MVT::f64 to type MVT::v2i32. The resulting
i64 node was then used to build a v2i32 vector.
With this patch, the backend now produces a cheaper SCALAR_TO_VECTOR from
MVT::f64 to MVT::v2f64. That SCALAR_TO_VECTOR is then followed by a "free"
bitcast to type MVT::v4i32. The elements of the resulting
v4i32 are then extracted to build a v2i32 vector (which is illegal and
therefore promoted to MVT::v2i64).
This is in general cheaper than emitting a stack store+load sequence
to bitconvert the operand from type f64 to type i64.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208107 91177308-0d34-0410-b5e6-96231b3b80d8
This patch implements the infrastructure to use named register constructs in
programs that need access to specific registers (bare metal, kernels, etc).
So far, only the stack pointer is supported as a technology preview, but as it
is, the intrinsic can already support all non-allocatable registers from any
architecture.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208104 91177308-0d34-0410-b5e6-96231b3b80d8
The Win64 docs are very clear that anything larger than 8 bytes is
passed by reference, and GCC MinGW64 honors that for __modti3 and
friends.
Patch by Jameson Nash!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208029 91177308-0d34-0410-b5e6-96231b3b80d8