llvm-svn: 59975

There might be other, similar bugs lurking there.

llvm-svn: 59974

llvm-svn: 59973

with an undef.

llvm-svn: 59972

- Mark "add with overflow" as having a custom lowering for X86. Give it a null
  lowering representation for now.

llvm-svn: 59971

(a) Slight rethink on i64 zero/sign/any extend code - use a shuffle to
    directly zero-extend i32 to i64, but use rotates and shifts for
    sign extension. Also ensure unified register consistency.
(b) Add new test harness for i64 operations: i64ops.ll

llvm-svn: 59970

to removePred because an SUnit can both data-depend and anti-depend
on the same SUnit.

llvm-svn: 59969

llvm-svn: 59968

obscure tail-merging opportunities.

llvm-svn: 59967

llvm-svn: 59966

(a) Improve the extract element code: there's no need to do gymnastics with
    rotates into the preferred slot if a shuffle will do the same thing.
(b) Rename a couple of SPUISD pseudo-instructions for readability and better
    semantic correspondence.
(c) Fix i64 sign/any/zero extension lowering.

llvm-svn: 59965

llvm-svn: 59964

llvm-svn: 59963

llvm-svn: 59962

llvm-svn: 59961

(this doesn't happen that often, since most code
does not use illegal types) then follow it by a
DAG combiner run that is allowed to generate
illegal operations but not illegal types.  I didn't
modify the target combiner code to distinguish like
this between illegal operations and illegal types,
so it will not produce illegal operations as well
as not producing illegal types.

llvm-svn: 59960

llvm-svn: 59958

One design problem that is emerging is the signed-ness problem during static
analysis. Many unsigned value have to be converted into signed value because
it partipates in operations with signed values. 

On the other hand, we cannot blindly make all values occuring in static analysis
signed, because we do have cases where unsignedness is required, for example,
integer overflow detection.

llvm-svn: 59957

llvm-svn: 59956

noalias attribute parameters/return values.

llvm-svn: 59955

llvm-svn: 59954

Move target independent td files from lib/Target/ to include/llvm/Target so they can be distributed along with the header files.

llvm-svn: 59953

llvm-svn: 59952

llvm-svn: 59951

instead of converting them to strings first.  This also fixes a
bunch of minor inconsistencies in the diagnostics emitted by clang
and adds a bunch of FIXME's to DiagnosticKinds.def.

llvm-svn: 59948

uses of getName() with uses of getDeclName().  This upgrades a bunch of
diags to take DeclNames instead of std::strings.

This also tweaks a couple of diagnostics to be cleaner and changes
CheckInitializerTypes/PerformInitializationByConstructor to pass
around DeclarationNames instead of std::strings.

llvm-svn: 59947

llvm-svn: 59946

llvm-svn: 59945

llvm-svn: 59944

llvm-svn: 59943

llvm-svn: 59942

llvm-svn: 59941

value. It must now be as if the pointer were allocated and has not escaped to
the caller. Thanks to Dan Gohman for pointing out the error in the original
and helping devise this definition.

llvm-svn: 59940

llvm-svn: 59939

llvm-svn: 59938

NamedDecl::getNameAsString() to make it more explicit.

llvm-svn: 59937

assert if the name is not an identifier.  Update callers to do the right
thing and avoid this method in unsafe cases.  This also fixes an objc
warning that was missing a space, and migrates a couple more to taking
IdentifierInfo and QualTypes instead of std::strings.

llvm-svn: 59936

llvm-svn: 59935

indicate functions that allocate, such as operator new, or list::insert. The
actual definition is slightly less strict (for now).

No changes to the bitcode reader/writer, asm printer or verifier were needed.

llvm-svn: 59934

a new NamedDecl::getAsString() method.

Change uses of Selector::getName() to just pass in a Selector 
where possible (e.g. to diagnostics) instead of going through
an std::string.

This also adds new formatters for objcinstance and objcclass
as described in the dox.

llvm-svn: 59933