This helps generate better code in functions with high register
pressure.

The previous version of compact region splitting caused regressions
because the regions were a bit too large. A stronger negative bias
applied in r136832 fixed this problem.

llvm-svn: 136836

[analyzer] rename all experimental checker packages to have 'experimental' be the common root package.

llvm-svn: 136835

Do not drop undef debug values. These are used as range termination marker by live debug variable pass.

llvm-svn: 136834

llvm-svn: 136833

Apply twice the negative bias on transparent blocks when computing the
compact regions. This excludes loop backedges from the region when only
one of the loop blocks uses the register.

Previously, we would include the backedge in the region if the loop
preheader and the loop latch both used the register, but the loop header
didn't.

When both the header and latch blocks use the register, we still keep it
live on the backedge.

llvm-svn: 136832

lib/CodeGen/RegAllocGreedy.cpp:1176:18: warning: unused variable 'B' [-Wunused-variable]
    if (unsigned B = Cand.getBundles(BundleCand, BestCand)) {
                 ^
lib/CodeGen/RegAllocGreedy.cpp:1188:18: warning: unused variable 'B' [-Wunused-variable]
    if (unsigned B = Cand.getBundles(BundleCand, 0)) {
                 ^

llvm-svn: 136831

the SBType implementation classes.

Fixed LLDB core and the test suite to not use deprecated SBValue APIs.

Added a few new APIs to SBValue:

    int64_t
    SBValue::GetValueAsSigned(int64_t fail_value=0);

    uint64_t
    SBValue::GetValueAsUnsigned(uint64_t fail_value=0)

 

llvm-svn: 136829

llvm-svn: 136828

llvm-svn: 136826

own, incorrectly, for releasing objects at the end of a full-expression.

llvm-svn: 136823

This is some of my original LLVM code. *wipes tear*

llvm-svn: 136821

Basic handling of Objective-C tagged pointers: return a custom ISA and typename when one is detected

llvm-svn: 136819

llvm-svn: 136818

reader, and fix up the one (!) place where we were improperly mapping
a local ID to a global ID. Tested via the usual "gaps" trick.

llvm-svn: 136817

llvm-svn: 136816

llvm-svn: 136815

llvm-svn: 136814

New approach to r136737: insert the necessary fences for atomic ops in platform-independent code, since a bunch of platforms (ARM, Mips, PPC, Alpha are the relevant targets here) need to do essentially the same thing.

I think this completes the basic CodeGen for atomicrmw and cmpxchg.

llvm-svn: 136813

llvm-svn: 136812

llvm-svn: 136811

llvm-svn: 136809

llvm-svn: 136808

Change TypoCorrection to store a set of NamedDecls instead of a single
NamedDecl. Also add initial support for performing function overload
resolution to Sema::DiagnoseEmptyLookup.

llvm-svn: 136807

llvm-svn: 136806

llvm-svn: 136805

[analyzer] Introduce MallocOverflowSecurityChecker, a simple flow-sensitive checker that may be useful for security auditing.  This checker is currently too noisy to be on by default.

llvm-svn: 136804

llvm-svn: 136803

llvm-svn: 136802

to make & delete directories in the test case.  Make a real copy of libd.dylib
in that directory so the two libraries are actually different.  Use (and remove)
the DYLD_LIBRARY_PATH to point to the new library.

llvm-svn: 136801

llvm-svn: 136800

Go back to the old definition. It's not clear that a 'resume' can be 'outlined' from the function with the landing pad for all platforms.

llvm-svn: 136799

rdar://problem/9890530

llvm-svn: 136798

to notify SCEV of a change. Add forgetLoop in a couple of those places.

llvm-svn: 136797

Skip the test case test_dyld_library_path(self) because it causes the debugserver to crash.

llvm-svn: 136796

llvm-svn: 136795

those declated in its protocols. First half or // rdar://6191214

llvm-svn: 136794

debugserver) but set it in the debugger settings (which will just get passed down to
the target).

llvm-svn: 136793

llvm-svn: 136792

llvm-svn: 136791

llvm-svn: 136790