The value map is currently not used, all values are 'complex mapped' and
LiveIntervalMap::mapValue is used to dig them out.

This is the first step in a series changes leading to the removal of
LiveIntervalMap. Its data structures can be shared among all the live intervals
created by a split, so it is wasteful to create a copy for each.

llvm-svn: 126800

Local live range splitting is better driven by interference. This code was just
guessing.

llvm-svn: 126799

This is a waste of time since we already know how to evict all interferences
which is a better approach anyway.

llvm-svn: 126798

llvm-svn: 126797

Patch by Erik Olofsson!

llvm-svn: 126796

This fixes few blocks.exp regressions.

llvm-svn: 126795

llvm-svn: 126794

Today, the language front ends produces llvm.dbg.* intrinsics, used to encode arguments' debug info, in order any way, most of the times. However, if a front end mix-n-matches llvm.dbg.declare and llvm.dbg.value intrinsics to encode debug info for arguments then code generator needs a way to find argument order.

Use 8 bits from line number field to keep track of argument ordering while encoding debug info for an argument. That leaves 24 bit for line no, DebugLoc also allocates 24 bit for line numbers. If a function has more than 255 arguments then rest of the arguments will be ordered by llvm.dbg.* intrinsics' ordering in IR.

llvm-svn: 126793

among other SBProcess APIs, to write (int)256 into a memory location of a global variable
(int)my_int and reads/checks the variable afterwards.

llvm-svn: 126792

compare it with getDriver().Dir.c_str(), since that is a pointer
comparison, not a "are these strings equal" comparison.

Instead, just compare with getDriver().Dir directly, so both sides will
get promoted to std::string, and the regular std::string comparison
operator applies.

Patch by Dimitry Andric!

llvm-svn: 126791

On the first cmake run before the caches has been updated with the
default options, options defined after HandleLLVMOptions are always
treated as off inside HandleLLVMOptions.

Patch by Erik Olofsson!

llvm-svn: 126790

Add code to emulate LDRSB (literal) Arm instruction.

llvm-svn: 126789

llvm-svn: 126788

which constructs a diagnostic with no line to show.

llvm-svn: 126787

llvm-svn: 126786

llvm-svn: 126785

a StringRef, for the benefit of clients that want the result as a
nul-terminated string. Clients that expect a StringRef will get one via
the implicit conversion.

llvm-svn: 126784

Add code to emulate LDRSB (immediate) Arm instruction.

llvm-svn: 126783

addressing code. On 403.gcc this almost halves CodeGenPrepare time and reduces
total llc time by 9.5%. Unfortunately, getNumUses() is still the hottest function
in llc.

llvm-svn: 126782

This effectively disables the 'turbo' functionality of the greedy register
allocator where all new live ranges created by splitting would be reconsidered
as if they were originals.

There are two reasons for doing this, 1. It guarantees that the algorithm
terminates. Early versions were prone to infinite looping in certain corner
cases. 2. It is a 2x speedup. We can skip a lot of unnecessary interference
checks that won't lead to good splitting anyway.

The problem is that region splitting only gets one shot, so it should probably
be changed to target multiple physical registers at once.

Local live range splitting is still 'turbo' enabled. It only accounts for a
small fraction of compile time, so it is probably not necessary to do anything
about that.

llvm-svn: 126781

llvm-svn: 126780

Don't warn about unused values in ternary ?: expressions unless both the LHS and RHS are "unused" (side-effect free).

Patch by Justin Bogner!  Fixes PR 8282.

llvm-svn: 126779

template specialization types. There are still a few rough edges to
clean up with some of the parser actions dropping
nested-name-specifiers too early.

llvm-svn: 126776

llvm-svn: 126775

llvm-svn: 126774

llvm-svn: 126773

llvm-svn: 126772

llvm-svn: 126770

This makes the number of total tests equal to 201.

llvm-svn: 126769

llvm-svn: 126768

llvm-svn: 126767

llvm-svn: 126766

nested-name-speciciers within elaborated type names, e.g.,
 
  enum clang::NestedNameSpecifier::SpecifierKind

Fixes in this iteration include:

  (1) Compute the type-source range properly for a dependent template
  specialization type that starts with "template template-id ::", as
  in a member access expression

    dep->template f<T>::f()

  This is a latent bug I triggered with this change (because now we're
  checking the computed source ranges for dependent template
  specialization types). But the real problem was...

  (2) Make sure to set the qualifier range on a dependent template
  specialization type appropriately. This will go away once we push
  nested-name-specifier locations into dependent template
  specialization types, but it was the source of the
  valgrind errors on the buildbots.
  

llvm-svn: 126765

Patch by Dimitry Andric!

llvm-svn: 126763

llvm-svn: 126762

path.  This avoids accidentally including the wrong headers.

Patch by Dimitry Andric!

llvm-svn: 126761

ToolChain's FilePaths.  If clang is installed as a port in /usr/local,
it is *not* supposed to use /usr/local/lib by default, for example.

Additionally, there are no clang-related executables in either
/usr/libexec, or getDriver().Dir + "/../libexec", anymore, so remove
that from the ToolChain's ProgramPaths.

Patch by Dimitry Andric!

llvm-svn: 126760

llvm-svn: 126759

Add code to emulate LDRH (register) Arm instruction.

llvm-svn: 126758

retrieve the library paths from the ToolChain object instead.

Copy the relevant code from linuxtools::Link::ConstructJob(), and
replace the std::string stuff with llvm::StringRef, while we're here.

Patch by Dimitry Andric!

llvm-svn: 126757