llvm-svn: 139705

THe LRE_DidCloneVirtReg callback may be called with vitual registers
that RAGreedy doesn't even know about yet.  In that case, there are no
data structures to update.

llvm-svn: 139702

llvm-dwarfdump: Make the "is debug info section" heuristic stricter so it doesn't accidentaly picks up the wrong section.

Also add some validation code to the aranges section parser.

Fixes PR10926.

llvm-svn: 139701

llvm-svn: 139699

When a back-copy is hoisted to the nearest common dominator, keep
looking up the dominator tree for a less loopy dominator, and place the
back-copy there instead.

Don't do this when a single existing back-copy dominates all the others.
Assume the client knows what he is doing, and keep the dominating
back-copy.

This prevents us from hoisting back-copies into loops in most cases.  If
a value is defined in a loop with multiple exits, we may still hoist
back-copies into that loop.  That is the speed/size tradeoff.

llvm-svn: 139698

llvm-svn: 139697

llvm-svn: 139695

llvm-svn: 139692

Fix mem type for VEX.128 form of VROUNDP*. Remove filter preventing VROUND from being recognized by disassembler.

llvm-svn: 139691

llvm-svn: 139690

- Add TSFlags for the instruction formats. The idea here is to use
  as much encoding as possible from getBinaryCodeForInstr, and having
  TSFLags formats for that would make it easier to encode most part
  of the instructions (since Mips encodings are pretty straightforward)
- Improve the mips mechanism for compilation callback
- Add Mips specific code for invalidating the instruction cache
- Next patch will address wrong tablegen encoding

Commit msg added by my own but the patch is from Sasa Stankovic.

llvm-svn: 139688

alignment check for 256-bit classes more strict. There're no testcases
but we catch more folding cases for AVX while running single and multi
sources in the llvm testsuite.

Since some 128-bit AVX instructions have different number of operands
than their SSE counterparts, they are placed in different tables.

256-bit AVX instructions should also be added in the table soon. And
there a few more 128-bit versions to handled, which should come in
the following commits.

llvm-svn: 139687

llvm-svn: 139686

llvm-svn: 139684

- Add enum SymbolType and function getSymbolType()
- Add function isGlobal() - it's returns true for symbols that can be used in another objects, such as library functions.
- Rename function getAddress() to getOffset() and add new function getAddress(), because currently getAddress() returns section offset of symbol first byte. new getAddress() return symbol address.
- Change usage SymbolRef::getAddress() to getOffset() in tools/llvm-nm and tools/llvm-objdump.

Patch by Danil Malyshev!

llvm-svn: 139683

Remove unimplemented function prototypes from PathV2. They can be readded when someone cares enough.

Patch by Aaron Ballman!

llvm-svn: 139682

This is only one half of it, the part that caches address ranges from the DIEs when .debug_aranges is
not available will be ported soon.

llvm-svn: 139680

llvm-svn: 139676

llvm-svn: 139671

#line directives with the needed support in the lexer.  Next will be to build
a simple file/line# table mapping source SMLoc's for later use by diagnostics.
And the last step will be to get the diagnostics to use the mapping for file
and line numbers.

llvm-svn: 139669

When a ParentVNI maps to multiple defs in a new interval, its live range
may still be derived directly from RegAssign by transferValues().

On the other hand, when instructions have been rematerialized or
hoisted, it may be necessary to completely recompute live ranges using
LiveRangeCalc::extend() to all uses.

Use a bit in the value map to indicate that a live range must be
recomputed.  Rename markComplexMapped() to forceRecompute().

This fixes some live range verification errors when
-split-spill-mode=size hoists back-copies by recomputing source ranges
when RegAssign kills can't be moved.

llvm-svn: 139660

llvm-svn: 139656

Whenever the complement interval is defined by multiple copies of the
same value, hoist those back-copies to the nearest common dominator.

This ensures that at most one copy is inserted per value in the
complement inteval, and no phi-defs are needed.

llvm-svn: 139651

llvm-svn: 139649

llvm-svn: 139646

llvm-svn: 139642

llvm-svn: 139641

llvm-svn: 139639

llvm-svn: 139637

llvm-svn: 139636

llvm-svn: 139635

tries to match a dead MipsLo node (explanation in the link below).

http://article.gmane.org/gmane.comp.compilers.llvm.devel/42757/match=dagcombiner+dead

llvm-svn: 139634

llvm-svn: 139633

llvm-svn: 139630

llvm-svn: 139628

This introduces a new library to LLVM: libDebugInfo. It will provide debug information
parsing to LLVM. Much of the design and some of the code is taken from the LLDB project.

It also contains an llvm-dwarfdump tool that can dump the abbrevs and DIEs from an
object file. It can be used to write tests for DWARF input and output easily.

llvm-svn: 139627

It is an endian-aware helper that can read data from a StringRef. It will
come in handy for DWARF parsing. This class is inspired by LLDB's
DataExtractor, but is stripped down to the bare minimum needed for DWARF.

Comes with unit tests!

llvm-svn: 139626

more strict about the alignment checking. This was found by inspection
and I don't have any testcases so far, although the llvm testsuite runs
without any problem.

llvm-svn: 139625

to be in the VSELECT operands order, so I'll leave the fix for Nadav.

llvm-svn: 139624

xor/and/or (For example SSE2).

llvm-svn: 139623