80-col violation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107361 91177308-0d34-0410-b5e6-96231b3b80d8

Trailing whitespace.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107360 91177308-0d34-0410-b5e6-96231b3b80d8

Add memory operand folding support to InlineSpiller.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107355 91177308-0d34-0410-b5e6-96231b3b80d8

Add support for rematerialization to InlineSpiller.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107351 91177308-0d34-0410-b5e6-96231b3b80d8

Use the catch-all selectors we already found when converting them to use the
correct catch-all value. This saves having to iterate through all of the
selectors in the program again.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107345 91177308-0d34-0410-b5e6-96231b3b80d8

lowerinvoke needs to handle aggregate function args like sjlj eh does.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107335 91177308-0d34-0410-b5e6-96231b3b80d8

build: Update install-clang target.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107334 91177308-0d34-0410-b5e6-96231b3b80d8

Handle array and vector typed parameters in sjljehprepare like we do
structs. rdar://8145832

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107332 91177308-0d34-0410-b5e6-96231b3b80d8

Remove all debug info related named mdnodes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107323 91177308-0d34-0410-b5e6-96231b3b80d8

grammar tweak in comment.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107321 91177308-0d34-0410-b5e6-96231b3b80d8

In ScalarEvolution::forgetValue, eliminate any SCEVUnknown
entries associated with the value being erased in the
folding set map.  These entries used to be harmless, because
a SCEVUnknown doesn't store any information about its Value*,
so having a new Value allocated at the old Value's address
wasn't a problem. But now that ScalarEvolution is storing more
information about values, this is no longer safe.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107316 91177308-0d34-0410-b5e6-96231b3b80d8

Move MOVD/MODQ code around, creating sections for each of them

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107308 91177308-0d34-0410-b5e6-96231b3b80d8

Some fool committed without testing (or even building) first.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107307 91177308-0d34-0410-b5e6-96231b3b80d8

Add AVX SSE2 mask creation and conditional store instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107306 91177308-0d34-0410-b5e6-96231b3b80d8

Remember to track spill slot uses in VirtRegMap when inserting loads and stores.
LocalRewriter::runOnMachineFunction uses this information to mark dead spill
slots.

This means that InlineSpiller now also works for functions that spill.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107302 91177308-0d34-0410-b5e6-96231b3b80d8

Fix a bug introduced in r107211 where instructions with memory operands are declared as commutable

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107300 91177308-0d34-0410-b5e6-96231b3b80d8

Revert the part of r107257 which introduced new logic for using
nsw and nuw flags from IR Instructions. On further consideration,
this isn't valid.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107298 91177308-0d34-0410-b5e6-96231b3b80d8

Rename NextPowerOfTwo to RoundUpToPowerOfTwo.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107297 91177308-0d34-0410-b5e6-96231b3b80d8

Remove an unused variable.  The call to getRoot has side-effects, so
this could break something (but doesn't seem to).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107295 91177308-0d34-0410-b5e6-96231b3b80d8

Add AVX SSE2 packed integer extract/insert instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107293 91177308-0d34-0410-b5e6-96231b3b80d8

Clarify that the NextPowerOfTwo template is idempotent.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107286 91177308-0d34-0410-b5e6-96231b3b80d8

Rather than giving SmallPtrSetImpl a member field SmallArray which is magically
replaced by a bigger array in SmallPtrSet (by overridding it), instead just use a
pointer to the start of the storage, and have SmallPtrSet pass in the value to use.
This has the disadvantage that SmallPtrSet becomes bigger by one pointer.  It has
the advantage that it no longer uses tricky C++ rules, and is clearly correct while
I'm not sure the previous version was.  This was inspired by g++-4.6 pointing out
that SmallPtrSetImpl was writing off the end of SmallArray, which it was.  Since
SmallArray is replaced with a bigger array in SmallPtrSet, the write was still to
valid memory.  But it was writing off the end of the declared array type - sounds
kind of dubious to me, like it sounded dubious to g++-4.6.  Maybe g++-4.6 is wrong
and this construct is perfectly valid and correctly compiled by all compilers, but
I think it is better to avoid the whole can of worms by avoiding this construct.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107285 91177308-0d34-0410-b5e6-96231b3b80d8

use ArgOperand API

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107282 91177308-0d34-0410-b5e6-96231b3b80d8

use ArgOperand API

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107280 91177308-0d34-0410-b5e6-96231b3b80d8

use ArgOperand API

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107279 91177308-0d34-0410-b5e6-96231b3b80d8

use ArgOperand API

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107278 91177308-0d34-0410-b5e6-96231b3b80d8

use ArgOperand API

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107277 91177308-0d34-0410-b5e6-96231b3b80d8

use CallSite::arg_end instead of CallInst::op_end

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107276 91177308-0d34-0410-b5e6-96231b3b80d8

use getArgOperand (corrected by CallInst::ArgOffset) instead of getOperand

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107275 91177308-0d34-0410-b5e6-96231b3b80d8

use getArgOperand (corrected by CallInst::ArgOffset) instead of getOperand

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107273 91177308-0d34-0410-b5e6-96231b3b80d8

use getNumArgOperands instead of getNumOperands

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107272 91177308-0d34-0410-b5e6-96231b3b80d8

use getArgOperand instead of getOperand

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107271 91177308-0d34-0410-b5e6-96231b3b80d8

use getArgOperand instead of getOperand

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107270 91177308-0d34-0410-b5e6-96231b3b80d8

use getArgOperand instead of getOperand

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107269 91177308-0d34-0410-b5e6-96231b3b80d8

Add a testcase for scev-aa's new capability.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107258 91177308-0d34-0410-b5e6-96231b3b80d8

Improve ScalarEvolution's nsw and nuw preservation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107257 91177308-0d34-0410-b5e6-96231b3b80d8

When computing a new ConservativeResult, intersect it with
the old one instead of replacing it, to be more precise.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107256 91177308-0d34-0410-b5e6-96231b3b80d8

Rework scev-aa's basic computation so that it doesn't depend
on ScalarEvolution successfully folding and preserving
range information for both A-B and B-A. Now, if it gets
either one, it's sufficient.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107249 91177308-0d34-0410-b5e6-96231b3b80d8

Simplify.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107248 91177308-0d34-0410-b5e6-96231b3b80d8

Add AVX SSE2 integer unpack instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107246 91177308-0d34-0410-b5e6-96231b3b80d8

Add AVX SSE2 packed integer shuffle instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107245 91177308-0d34-0410-b5e6-96231b3b80d8

Remove trailing whitespace, no functionality changes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107244 91177308-0d34-0410-b5e6-96231b3b80d8

Small refactoring of SSE2 packed integer shuffle instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107243 91177308-0d34-0410-b5e6-96231b3b80d8

Add AVX SSE2 pack with saturation integer instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107241 91177308-0d34-0410-b5e6-96231b3b80d8

Add AVX SSE2 integer packed compare instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107240 91177308-0d34-0410-b5e6-96231b3b80d8

- Add AVX form of all SSE2 logical instructions
- Add VEX encoding bits to x86 MRM0r-MRM7r

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107238 91177308-0d34-0410-b5e6-96231b3b80d8

Do not construct DIE for already processed MDNode.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107237 91177308-0d34-0410-b5e6-96231b3b80d8

Use skipInstruction() as a simpler way of iterating over instructions using SrcReg

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107234 91177308-0d34-0410-b5e6-96231b3b80d8

Use clEnumValN macro to work around keyword clash

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107233 91177308-0d34-0410-b5e6-96231b3b80d8

Add variables into a scope before constructing scope DIE otherwise variables won't be included DIE tree.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107228 91177308-0d34-0410-b5e6-96231b3b80d8

Begin implementation of an inline spiller.

InlineSpiller inserts loads and spills immediately instead of deferring to
VirtRegMap. This is possible now because SlotIndexes allows instructions to be
inserted and renumbered.

This is work in progress, and is mostly a copy of TrivialSpiller so far. It
works very well for functions that don't require spilling.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107227 91177308-0d34-0410-b5e6-96231b3b80d8

Add *several* AVX integer packed binop instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107225 91177308-0d34-0410-b5e6-96231b3b80d8

Fix ScalarEvolution's tripcount computation for chains of loops
where each loop's induction variable's start value is the exit
value of a preceding loop.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107224 91177308-0d34-0410-b5e6-96231b3b80d8

Revert r107205 and r107207.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107215 91177308-0d34-0410-b5e6-96231b3b80d8

Print InlinedAt location.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107214 91177308-0d34-0410-b5e6-96231b3b80d8

Add another bswap idiom that isn't matched.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107213 91177308-0d34-0410-b5e6-96231b3b80d8

Use -l option to remove symbols from i386.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107212 91177308-0d34-0410-b5e6-96231b3b80d8

Move SSE2 Packed Integer instructions around, and create specific sections for each of them

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107211 91177308-0d34-0410-b5e6-96231b3b80d8

Print InlinedAt location.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107208 91177308-0d34-0410-b5e6-96231b3b80d8

Improve explanation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107207 91177308-0d34-0410-b5e6-96231b3b80d8

Add AVX Move Aligned/Unaligned packed integers

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107206 91177308-0d34-0410-b5e6-96231b3b80d8

Introducing the "linker_weak" linkage type. This will be used for Objective-C
metadata types which should be marked as "weak", but which the linker will
remove upon final linkage. For example, the "objc_msgSend_fixup_alloc" symbol is
defined like this:

       .globl l_objc_msgSend_fixup_alloc
       .weak_definition l_objc_msgSend_fixup_alloc
       .section __DATA, __objc_msgrefs, coalesced
       .align 3
l_objc_msgSend_fixup_alloc:
        .quad   _objc_msgSend_fixup
        .quad   L_OBJC_METH_VAR_NAME_1

This is different from the "linker_private" linkage type, because it can't have
the metadata defined with ".weak_definition".

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107205 91177308-0d34-0410-b5e6-96231b3b80d8

Add AVX ld/st XCSR register.
Add VEX encoding bits for MRMXm x86 form

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107204 91177308-0d34-0410-b5e6-96231b3b80d8

Do not hardcode DW_AT_stmt_list value.
Inspired by Artur Pietrek.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107202 91177308-0d34-0410-b5e6-96231b3b80d8

Add support for encoding VDUP (ARM core register) instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107201 91177308-0d34-0410-b5e6-96231b3b80d8

Fix a buffer overflow noticed by gcc-4.6: zero is written into
SmallArray[SmallSize] in the SmallPtrSetIteratorImpl, and this is
one off the end of the array.  For those who care, right now gcc
warns about writing off the end because it is confused about the
declaration of SmallArray as having length 1 in the parent class
SmallPtrSetIteratorImpl.  However if you tweak code to unconfuse
it, then it still warns about writing off the end of the array,
because of this buffer overflow.  In short, even with this fix
gcc-4.6 will warn about writing off the end of the array, but now
that is only because it is confused.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107200 91177308-0d34-0410-b5e6-96231b3b80d8

NextPowerOfTwo was already used to ensure that SmallSizePowTwo is a power
of two, no need to do it a second time (NextPowerOfTwo is idempotent).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107199 91177308-0d34-0410-b5e6-96231b3b80d8

give PATypeHolder an explicit copy ctor which initializes the type pointer,
and make PATypeHolder work with null pointers.

The implicitly generated one didn't work on numerous levels, but was still
accepted, allowing all sorts of bugs with default constructed pa type holders.

Previously, they "sort of" worked if they were default constructed and then
destructed.  Now they really work, and you can even default construct one,
then assign to it, amazing.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107195 91177308-0d34-0410-b5e6-96231b3b80d8

Fix the handling of partial redefines in the fast register allocator.

A partial redefine needs to be treated like a tied operand, and the register
must be reloaded while processing use operands.

This fixes a bug where partially redefined registers were processed as normal
defs with a reload added. The reload could clobber another use operand if it was
a kill that allowed register reuse.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107193 91177308-0d34-0410-b5e6-96231b3b80d8

Fix a register scavenger crash when dealing with undefined subregs.
The LowerSubregs pass needs to preserve implicit def operands attached to
EXTRACT_SUBREG instructions when it replaces those instructions with copies.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107189 91177308-0d34-0410-b5e6-96231b3b80d8

Add AVX non-temporal stores

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107178 91177308-0d34-0410-b5e6-96231b3b80d8

Add a few more interesting testcases.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107177 91177308-0d34-0410-b5e6-96231b3b80d8

Fix whitespace style.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107175 91177308-0d34-0410-b5e6-96231b3b80d8

Move non-temporal movs to their own section

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107168 91177308-0d34-0410-b5e6-96231b3b80d8

Add support for encoding NEON VMOV (from core register to scalar) instructions.
The encoding is the same as VMOV (from scalar to core register) except that
the operands are in different places.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107167 91177308-0d34-0410-b5e6-96231b3b80d8

Add sqrt, rsqrt and rcp AVX instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107166 91177308-0d34-0410-b5e6-96231b3b80d8

skip dbg_value instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107154 91177308-0d34-0410-b5e6-96231b3b80d8

minor cosmetics

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107148 91177308-0d34-0410-b5e6-96231b3b80d8

The t2MOVi16 and t2MOVTi16 instructions do not set CPSR.  Trying to add
a CPSR operand to them causes an assertion failure, so apparently these
instructions haven't been getting a lot of use.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107147 91177308-0d34-0410-b5e6-96231b3b80d8

use ArgOffset constant to prepare for operand rotation

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107146 91177308-0d34-0410-b5e6-96231b3b80d8

use ArgOperand API

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107145 91177308-0d34-0410-b5e6-96231b3b80d8

use ArgOperand API

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107144 91177308-0d34-0410-b5e6-96231b3b80d8

Return Changed.  This required setting Changed if dbg metadata
is stripped off.  Currently set unconditionally, since the API
does not provide a way of working out if anything was actually
stripped off.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107142 91177308-0d34-0410-b5e6-96231b3b80d8

It seems clear that this should return Changed.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107141 91177308-0d34-0410-b5e6-96231b3b80d8

Add a VT argument to getMinimalPhysRegClass and replace the copy related uses
of getPhysicalRegisterRegClass with it.

If we want to make a copy (or estimate its cost), it is better to use the
smallest class as more efficient operations might be possible.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107140 91177308-0d34-0410-b5e6-96231b3b80d8

getMachineBasicBlockAddress returns a uintptr_t - don't truncate
to unsigned only to extend back to a pointer sized value on the
next line.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107139 91177308-0d34-0410-b5e6-96231b3b80d8

The variable ValueSize is set to 1 on both code paths, and then
ignored!  Remove it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107138 91177308-0d34-0410-b5e6-96231b3b80d8

The variable "Value" is carefully set to Layout.getSymbolAddress,
but then not actually used - maybe a bug?  Remove the variable.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107137 91177308-0d34-0410-b5e6-96231b3b80d8

Remove unused calls to Lexer.getLoc and the pointless variable HasFillExpr.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107136 91177308-0d34-0410-b5e6-96231b3b80d8

Remove pointless variable LastDef.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107135 91177308-0d34-0410-b5e6-96231b3b80d8

Looks like this test is missing an XFAIL line.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107134 91177308-0d34-0410-b5e6-96231b3b80d8

Remove unused variable Loc and pointless variables unified_syntax
and thumb_mode.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107133 91177308-0d34-0410-b5e6-96231b3b80d8

use ArgOperand APIs

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107132 91177308-0d34-0410-b5e6-96231b3b80d8

Remove an unused and a pointless variable.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107131 91177308-0d34-0410-b5e6-96231b3b80d8

Remove pointless and unused variables.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107130 91177308-0d34-0410-b5e6-96231b3b80d8

encode operand initializations (at fixed index)
in terms of Op<> and ArgOffset. This works for
values of {0, 1} for ArgOffset.
Please note that ArgOffset will become 0 soon and
will go away eventually.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107129 91177308-0d34-0410-b5e6-96231b3b80d8

Remove a pointless variable.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107128 91177308-0d34-0410-b5e6-96231b3b80d8

Remove initialized but otherwise unused variables.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107127 91177308-0d34-0410-b5e6-96231b3b80d8

Remove variables that are written by not read.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107126 91177308-0d34-0410-b5e6-96231b3b80d8

Use a more obvious way to avoid compiling functions which are only used when XDEBUG is enabled.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107125 91177308-0d34-0410-b5e6-96231b3b80d8