This adds a llvm.copysign intrinsic; We already have Libfunc recognition for
copysign (which is turned into the FCOPYSIGN SDAG node). In order to
autovectorize calls to copysign in the loop vectorizer, we need a corresponding
intrinsic as well.

In addition to the expected changes to the language reference, the loop
vectorizer, BasicTTI, and the SDAG builder (the intrinsic is transformed into
an FCOPYSIGN node, just like the function call), this also adds FCOPYSIGN to a
few lists in LegalizeVector{Ops,Types} so that vector copysigns can be
expanded.

In TargetLoweringBase::initActions, I've made the default action for FCOPYSIGN
be Expand for vector types. This seems correct for all in-tree targets, and I
think is the right thing to do because, previously, there was no way to generate
vector-values FCOPYSIGN nodes (and most targets don't specify an action for
vector-typed FCOPYSIGN).

llvm-svn: 188728

copysign/copysignf never become function calls (because the SDAG expansion code
does not lower to the corresponding function call, but rather directly
implements the associated logic), but copysignl almost always is lowered into a
call to the requested libm functon (and, thus, might clobber CTR).

llvm-svn: 188727

llvm-svn: 188690

The Thumb2 add immediate is in fact defined for SP. The manual is misleading as it points to a different section for add immediate with SP, however the encoding is the same as for add immediate with register only with the SP operand hard coded. As such add immediate with SP and add immediate with register can safely be treated as the same instruction.

All the patch does is adjust a register constraint on an instruction alias.

llvm-svn: 188676

ADD, SUB, MUL integer and FP types. OR, AND, XOR.
Added embeded broadcast form for these instructions.

llvm-svn: 188673

For now this matches the equivalent of (neg (abs ...)), which did hit a few
times in projects/test-suite.  We should probably also match cases where
absolute-like selects are used with reversed arguments.

llvm-svn: 188671

llvm-svn: 188670

This first cut is pretty conservative.  The final argument register (R6)
is call-saved, so we would need to make sure that the R6 argument to a
sibling call is the same as the R6 argument to the calling function,
which seems worth keeping as a separate patch.

Saying that integer truncations are free means that we no longer
use the extending instructions LGF and LLGF for spills in int-conv-09.ll
and int-conv-10.ll.  Instead we treat the registers as 64 bits wide and
truncate them to 32-bits where necessary.  I think it's unlikely we'd
use LGF and LLGF for spills in other situations for the same reason,
so I'm removing the tests rather than replacing them.  The associated
code is generic and applies to many more instructions than just
LGF and LLGF, so there is no corresponding code removal.

llvm-svn: 188669

Modern PPC cores support a floating-point copysign instruction, and we can use
this to lower the FCOPYSIGN node (which is created from calls to the libm
copysign function). A couple of extra patterns are necessary because the
operand types of FCOPYSIGN need not agree.

llvm-svn: 188653

When patching inlineasm nodes to use GPRPair for 64-bit values, we
were dropping the information that two operands were tied, which
effectively broke the live-interval of vregs affected.

llvm-svn: 188643

llvm-svn: 188637

llvm-svn: 188629

llvm-svn: 188626

Spotted by Nick Lewycky!

llvm-svn: 188599

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
llvm-svn: 188598

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
llvm-svn: 188597

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
llvm-svn: 188596

Ongoing 'make the verifier happy' improvements to ARM fast-isel.

rdar://12594152

llvm-svn: 188595

Properly constrain the operand register class for instructions used
in [sz]ext expansion. Update more tests to use the verifier now that
we're getting the register classes correct.

rdar://12594152

llvm-svn: 188594

Teach the generic instruction selection helper functions to constrain
the register classes of their input operands. For non-physical register
references, the generic code needs to be careful not to mess that up
when replacing references to result registers. As the comment indicates
for MachineRegisterInfo::replaceRegWith(), it's important to call
constrainRegClass() first.

rdar://12594152

llvm-svn: 188593

Lots of machine verifier errors result from using a plain GPR regclass
for incoming argument copies. A more restrictive rGPR class is more
appropriate since it more accurately represents what's happening, plus
it lines up better with isel later on so the verifier is happier.
Reduces the number of ARM fast-isel tests not running with the verifier
enabled by over half.

rdar://12594152

llvm-svn: 188592

This regards how mips16 is viewed. It's not really a target type but
there has always been a target for it in the td files. It's more properly
-mcpu=mips32 -mattr=+mips16 . This is how clang treats it but we have
always had the -mcpu=mips16 which I probably should delete now but it will
require updating all the .ll test cases for mips16. In this case it changed
how we decide if we have a count bits instruction and whether instruction
lowering should then expand ctlz. Now that we have dual mode compilation,
-mattr=+mips16 really just indicates the inital processor mode that
we are compiling for. (It is also possible to have -mcpu=64 -mattr=+mips16
but as far as I know, nobody has even built such a processor, though there
is an architecture manual for this).

llvm-svn: 188586

safe on PPC32 SVR4 ABI

[Patch and following text by Mark Minich; committing on his behalf.]

There are FIXME's in PowerPC/PPCFrameLowering.cpp, method
PPCFrameLowering::emitPrologue() related to "negative offsets of R1"
on PPC32 SVR4. They're true, but the real issue is that on PPC32 SVR4
(and any ABI without a Red Zone), no spills may be made until after
the stackframe is claimed, which also includes the LR spill which is
at a positive offset. The same problem exists in emitEpilogue(),
though there's no FIXME for it. I intend to fix this issue, making
LLVM-compiled code finally safe for use on SVR4/EABI/e500 32-bit
platforms (including in particular, OS-free embedded systems & kernel
code, where interrupts may share the same stack as user code).

In preparation for making these changes, to make the diffs for the
functional changes less cluttered, I am providing the non-functional
refactorings in two stages:

Stage 1 does some minor fluffy refactorings to pull multiple method
calls up into a single bool, creating named bools for repeated uses of
obscure logic, moving some code up earlier because either stage 2 or
my final version will require it earlier, and rewording/adding some
comments. My stage 1 changes can be characterized as primarily fluffy
cleanup, the purpose of which may be unclear until the stage 2 or
final changes are made.

My stage 2 refactorings combine the separate PPC32 & PPC64 logic,
which is currently performed by largely duplicate code, into a single
flow, with the differences handled by a group of constants initialized
early in the methods.

This submission is for my stage 1 changes. There should be no
functional changes whatsoever; this is a pure refactoring.

llvm-svn: 188573

Fixes two recent piglit regressions with radeonsi.

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 188559

The logic in SIInsertWaits::getHwCounts() only really made sense for SMRD
instructions, and trying to shoehorn it into handling DS_WRITE_B32 caused
it to corrupt the encoding of that by clobbering the first operand with
the second one.

Undo that damage and only apply the SMRD logic to that.

Fixes some derivates related piglit regressions with radeonsi.

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 188558

llvm-svn: 188557

llvm-svn: 188556

llvm-svn: 188555

This unbreaks PIC with fast isel on ELF targets (PR16717). The output matches
what GCC and SDag do for PIC but may not cover all of the many flavors of PIC
that exist.

llvm-svn: 188551

Thumb2 literal loads use an offset encoding which allows for 
negative zero. This fixes parsing and encoding so that #-0 
is correctly processed. The parser represents #-0 as INT32_MIN.

llvm-svn: 188549

There are many Thumb instructions which take 12-bit immediates encoded in a special
8-byte value + 4-byte rotator form. Not all numbers are represented, and it's legal
to transform an assembly instruction to be able to encode the immediate.

For example: AND and BIC are complementary instructions; one can switch the AND
to a BIC as long as the immediate is complemented. 

The intent is to switch one instruction into its complementary one when the immediate
cannot be encoded in the form requested in the original assembly and when the 
complementary immediate is encodable.

The patch addresses two issues:
1. definition of t2SOImmNot immediate - it has to check that the orignal value is
not encoded naturally
2. t2AND and t2BIC instruction aliases which should use the Thumb2 SOImm operand 
rather than the ARM one.

llvm-svn: 188548

It would also make sense to use it for memchr; I'm working on that now.

llvm-svn: 188547

llvm-svn: 188546

llvm-svn: 188544

Generalize r188163 to cope with return types other than MVT::i32, just
as the existing visitMemCmpCall code did.  I've split this out into a
subroutine so that it can be used for other upcoming patches.

I also noticed that I'd used the wrong API to record the out chain.
It's a load that uses DAG.getRoot() rather than getRoot(), so the out
chain should go on PendingLoads.  I don't have a testcase for that because
we don't do any interesting scheduling on z yet.

llvm-svn: 188540

r188163 used CLC to implement memcmp.  Code that compares the result
directly against zero can test the CC value produced by CLC, but code
that needs an integer result must use IPM.  The sequence I'd used was:

   ipm <reg>
   sll <reg>, 2
   sra <reg>, 30

but I'd forgotten that this inverts the order, so that CC==1 ("less")
becomes an integer greater than zero, and CC==2 ("greater") becomes
an integer less than zero.  This sequence should only be used if the
CLC arguments are reversed to compensate.  The problem then is that
the branch condition must also be reversed when testing the CLC
result directly.

Rather than do that, I went for a different sequence that works with
the natural CLC order:

   ipm <reg>
   srl <reg>, 28
   rll <reg>, <reg>, 31

One advantage of this is that it doesn't clobber CC.  A disadvantage
is that any sign extension to 64 bits must be done separately,
rather than being folded into the shifts.

llvm-svn: 188538

llvm-svn: 188537

llvm-svn: 188534

This reverts commit a6a39ced095c2f453624ce62c4aead25db41a18f.
This is the wrong version of this fix.

llvm-svn: 188523

The SIInsertWaits pass was overwriting the first operand (gds bit) of
DS_WRITE_B32 with the second operand (value to write).  This meant that
any time the value to write was stored in an odd number VGPR, the gds
bit would be set causing the instruction to write to GDS instead of LDS.

llvm-svn: 188522