experimental pass that allocates locals relative to one another before
register allocation and then assigns them to actual stack slots as a block
later in PEI. This will eventually allow targets with limited index offset
range to allocate additional base registers (not just FP and SP) to
more efficiently reference locals, as well as handle situations where
locals cannot be referenced via SP or FP at all (dynamic stack realignment
together with variable sized objects, for example). It's currently
incomplete and almost certainly buggy. Work in progress.

Disabled by default and gated via the -enable-local-stack-alloc command
line option.

rdar://8277890

llvm-svn: 111059

had its address taken.

llvm-svn: 111058

This fixes another part of PR7792.

llvm-svn: 111057

The earliestStart argument is entirely specific to linear scan allocation, and
can be easily calculated by RegAllocLinearScan.

Replace std::vector with SmallVector.

llvm-svn: 111055

llvm-svn: 111050

instruction opcode.  This fixes part of PR7792.

llvm-svn: 111047

when they are the same loop. Don't compare two instructions'
loop depths when they are in the same block.

llvm-svn: 111045

When a live range is contained a single block, we can split it around
instruction clusters. The current approach is very primitive, splitting before
and after the largest gap between uses.

llvm-svn: 111043

ScalarEvolution::getAddExpr, which can be pretty expensive, when nothing
has changed, which is pretty common.

llvm-svn: 111042

llvm-svn: 111041

llvm-svn: 111040

rather than testing whether the loop contains the other's header.

llvm-svn: 111039

llvm-svn: 111038

the constant operand on the left, as that's where ScalarEvolution
will end up canonicalizing to.

llvm-svn: 111037

associated loop. This avoids potentially expensive traversals
of the add recurrence's operands.

llvm-svn: 111034

llvm-svn: 111033

Apparently, this is now fixed in Clang.

llvm-svn: 111032

since they can support trivial implementations. This avoids potentially
expensive traversals of the operands.

llvm-svn: 111031

same lines as the change I made for ARM saturate instructions.

llvm-svn: 111029

misanalysis and is undesirable.

llvm-svn: 111028

llvm-svn: 111022

llvm-svn: 111021

llvm-svn: 111019

llvm-svn: 111010

llvm-svn: 111008

llvm-svn: 111007

llvm-svn: 111004

llvm-svn: 111001

llvm-svn: 111000

llvm-svn: 110999

Tested on Linux and Darwin; please add platform-specific XFAILs/mail me a bug
report if this still fails.

llvm-svn: 110998

numbers match. The old check could accidentally leave holes in openli.

Also let useIntv add all ranges for the phi-def value inserted by
enterIntvAtEnd. This works as long at the value mapping is established in
enterIntvAtEnd.

llvm-svn: 110995

function to do it.

llvm-svn: 110994

llvm-svn: 110991

llvm-svn: 110987

Move some code from Verifier into SVI::isValidOperands.  This allows us to catch bad shufflevector operations when they are created, rather than waiting for someone to notice later on.

llvm-svn: 110986

llvm-svn: 110983

llvm-svn: 110982

llvm-svn: 110981

This can happen if the original interval has been broken into two disconnected
parts. Ideally, we should be able to detect when the graph is disconnected and
create separate intervals, but that code is not implemented yet.

Example:

Two basic blocks are both branching to a loop header. Our interval is defined in
both basic blocks, and live into the loop along both edges.

We decide to split the interval around the loop. The interval is split into an
inside part and an outside part. The outside part now has two disconnected
segments, one in each basic block.

If we later decide to split the outside interval into single blocks, we get one
interval per basic block and an empty dupli for the remainder.

llvm-svn: 110976