Patch by Louis Zhuang!

llvm-svn: 119394

llvm-svn: 119393

llvm-svn: 119391

llvm-svn: 119390

llvm-svn: 119389

increased.  The following program is running 49% faster:

#include <iostream>
#include <memory>
#include <chrono>
#include <vector>
#include "chrono_io"

int main()
{
    typedef std::chrono::high_resolution_clock Clock;
    Clock::time_point t0 = Clock::now();
    {
        std::shared_ptr<int> p(new int (1));
        std::vector<std::shared_ptr<int> > v(1000000, p);
        v.insert(v.begin(), p);
        v.insert(v.begin(), p);
        v.insert(v.begin(), p);
        v.insert(v.begin(), p);
    }
    Clock::time_point t1 = Clock::now();
    std::cout << (t1-t0) << '\n';
}

llvm-svn: 119388

llvm-svn: 119387

llvm-svn: 119386

llvm-svn: 119385

Next: Add support for the !HasDotLocAndDotFile case to the MCAsmStreamer
and then switch codegen to use it.

llvm-svn: 119384

llvm-svn: 119383

easier to debug, and to avoid complications when the CFG changes
in the middle of the instruction selection process.

llvm-svn: 119382

llvm-svn: 119381

llvm-svn: 119380

where we failed to free this buffer along one of the paths, and
detangles the code a little.

llvm-svn: 119379

turned pointers into pointers to const in function parameters in all functions/builtins accepting pointers to a const-qualified type according to PIM and "Language Extensions for CBEA"

llvm-svn: 119376

Always spill the full representative register at any point where any subregister
is live.

This fixes PR8620 which caused the old logic to get confused and not spill
anything at all.

The fundamental problem here is that the coalescer is too aggressive about
physical register coalescing. It sometimes makes it impossible to allocate
registers without these emergency spills.

llvm-svn: 119375

llvm-svn: 119374

result including the session logs of test failures/errors as a MIME message.

llvm-svn: 119371

llvm-svn: 119370

Stop defining types with "__neon_" prefixes and then using typedefs without
the prefix; there's no reason to do that anymore.  Remove types that combine
multiple Neon vectors and treat them as a single long vector; they are not
used.

llvm-svn: 119369

issue with runtime which I am discussing it with Blaine.
This is wip (so no test yet).

llvm-svn: 119368

I've temporarily disabled the failing clang test.

llvm-svn: 119367

I'll reenable it soon when I'm done reworking <arm_neon.h>.

llvm-svn: 119366

llvm-svn: 119365

allocation sizes of 0 bytes.

Fixes PR 2899.

llvm-svn: 119364

It's breaking buildbots.

llvm-svn: 119363

llvm-svn: 119362

The system API's will be shifted over to returning an error_code, and returning
other return values as out parameters to the function.

Code that needs to check error conditions will use the errc enum values which
are the same as the posix_errno defines (EBADF, E2BIG, etc...), and are
compatable with the error codes in WinError.h due to some magic in system_error.

An example would be:

if (error_code ec = KillEvil("Java")) { // error_code can be converted to bool.
  handle_error(ec);
}

llvm-svn: 119360

Fixes rdar://8365458

llvm-svn: 119359

Thanks to Nate Begeman for an earlier version of this patch.

llvm-svn: 119358

llvm-svn: 119355

llvm-svn: 119354

llvm-svn: 119353

rather than calling hasConstantValue.  No intended functionality change.

llvm-svn: 119352

for Arnaud Allard de Grandmaison for preparing a patch.

llvm-svn: 119351

llvm-svn: 119349

llvm-svn: 119348

over a phi node by applying it to each operand may be wrong if the
operation and the phi node are mutually interdependent (the testcase
has a simple example of this).  So only do this transform if it would
be correct to perform the operation in each predecessor of the block
containing the phi, i.e. if the other operands all dominate the phi.
This should fix the FFMPEG snow.c regression reported by İsmail Dönmez.

llvm-svn: 119347

conversion.

llvm-svn: 119346