out aliases when allocating. Clean up allocVirtReg().

Use calcSpillCost() to allow more aggressive hinting. Now the hint is always
taken unless blocked by a reserved register. This leads to more coalescing,
lower register pressure, and less spilling.

llvm-svn: 103939

Pick the correct personality function based on the language.  This prevents link failures when C/ObjC code uses __attribute__((cleanup())) (previously this was inserting references to two libstc++ symbols; the personality function and the __terminate() function).  

This is still probably wrong for Objective-C++ and adds a couple of lines in CGException that should probably be in the CGObjCRuntime subclass.  The personality function is now only looked up in one place in CGException though, so this should be easier to fix in the future.

llvm-svn: 103938

llvm-svn: 103937

llvm-svn: 103936

This makes allocation independent on the ordering of use-def chains.

llvm-svn: 103935

llvm-svn: 103934

This is safe to do because the physreg has been marked UsedInInstr and the kill flag will be set on the last operand using the virtreg if there are more then one.

llvm-svn: 103933

definitions.
 

llvm-svn: 103932

llvm-svn: 103931

through the very long list of call-clobbered registers. We just assume all
registers are clobbered.

llvm-svn: 103930

llvm-svn: 103929

symbol to the file as we have it.  Simplifies out tbss handling.

llvm-svn: 103928

Debug code doesn't use callee saved registers anyway, and the code is simpler this way. Now spillVirtReg always kills, and the isKill parameter is not needed.

llvm-svn: 103927

llvm-svn: 103926

llvm-svn: 103925

Yes, if the redef is a copy, update the old val# with the copy. But make sure to clear the copy field if the redef is not a copy.

llvm-svn: 103922

llvm-svn: 103917

llvm-svn: 103916

llvm-svn: 103915

<1xi64> -> i64 to work in MMX registers on hosts where -no-sse
is the default (not mine).  The right thing is
to accept this and make i64->f64 conversions go through memory,
but I don't have time right now.

llvm-svn: 103914

a missing */ since we truncated the file.

This fixes rdar://7948776

llvm-svn: 103913

This aligns with how gcc compiler does things.

llvm-svn: 103912

(This worked as of about 6 months ago and I didn't track down
exactly what broke it; I think this fix is appropriate.)

llvm-svn: 103911

Beefed up the tests for all of the distributions to include checks against the expected skewness and kurtosis

llvm-svn: 103910

Correctly diagnose array 'new' with initialization arguments when the new type is a typedef to an array type.

llvm-svn: 103909

of an array type, use the outermost array bound as the number of
elements to allocate. Fixes PR7147.

llvm-svn: 103908

llvm-svn: 103907

llvm-svn: 103906

compute a CFG for a function.

llvm-svn: 103905

initializer, don't fold paramters. Their initializers are just default
arguments which can be overridden. This fixes some spectacular regressions due
to more things making it into the constant folding.

llvm-svn: 103904

llvm-svn: 103903

Patch by Charles Davis and Steven Watanabe!

llvm-svn: 103902

llvm-svn: 103901

llvm-svn: 103900

passed to va_start, it doesn't actually pass it.

llvm-svn: 103899

llvm-svn: 103898

LLVMgold.so both in both the build and install directories.

llvm-svn: 103897

llvm-svn: 103896

entirely sure what this does, to be honest.

llvm-svn: 103895

ObjCObjectTypeLoc.  This should clear up some valgrind errors we've
been seeing.

llvm-svn: 103894