llvm-svn: 165599

llvm-svn: 165597

Have 'addFnAttr' take the attribute enum value. Then have it build the attribute object and add it appropriately. No functionality change.

llvm-svn: 165596

Have 'addFnAttr' take the attribute enum value. Then have it build the attribute object and add it appropriately. No functionality change.

llvm-svn: 165595

llvm-svn: 165594

and modules are not enabled.

llvm-svn: 165593

an invalid location if the location points to the synthetic buffer
for the module input.

llvm-svn: 165592

This means the main file for modules will always be a virtual one.

llvm-svn: 165591

llvm-svn: 165590

Patch committed on behalf of Kirill Uhanov

llvm-svn: 165589

llvm-svn: 165588

API (to get the length of x86 instructions) to using the LLVM-MC
disassembler.
<rdar://problem/12411000>

llvm-svn: 165587

This patch adds new functions to the SectionRef and ObjectFile interfaces to determine whether or not a section is meant to be read-only.  These functions will be used by the MCJIT RuntimeDyld to give hints to the memory manager during the object loading process in a future patch.

Patch by Ashok Thirumurthi.

llvm-svn: 165586

Patch by Sasa Stankovic. 

llvm-svn: 165585

llvm-svn: 165584

llvm-svn: 165583

llvm-svn: 165582

llvm-svn: 165581

llvm-svn: 165580

over to simply update its cache and then look up
the descriptor in the cache.  This is fine because
the cache now builds much faster (since descriptors
are minimal).

Metaclasses aren't in the cache, so I switched
the Describe method for class descriptors from using
GetClassDescriptor to manually creating an automatic
ClassDescriptorV2.

llvm-svn: 165579

This reverts commit 165428 in an attempt to get our buildbots going.

llvm-svn: 165574

This reverts commit 165429 in an attempt to get our buildbots going.

llvm-svn: 165573

-fcatch-undefined-behavior: store the type name directly at the end of a type descriptor. 5% binary size reduction due to fewer relocations.

llvm-svn: 165572

-fcatch-undefined-behavior: store the type name directly at the end of a type descriptor. 5% binary size reduction due to fewer relocations.

llvm-svn: 165571

llvm-svn: 165570

llvm-svn: 165569

llvm-svn: 165568

The following are now derived lazily:

- The name of the class (cached);
- the instance size of the class (not cached);

The following have been removed entirely:

- Whether the class is realized.  This is an
  implementation detail.
- The contents of the objc_class object.  That
  object can be read as needed.
- Whether the class is valid.  The fact that
  we vended a class to begin with means it's
  valid.  We will only give up looking parts
  of it up if they are not in the format we
  expect.

llvm-svn: 165567

llvm-svn: 165566

llvm-svn: 165565

misched: Allow flags to disable hasInstrSchedModel/hasInstrItineraries for external users of TargetSchedule.

llvm-svn: 165564

This wasn't contributing anything significant to postRA heuristics except compile time (by my measurements) and will be replaced by a more general heuristic for cross-region dependencies within the scheduler itself.

llvm-svn: 165563

Use the attribute builder to add attributes to call/invoke instruction. No functionality change intended.

llvm-svn: 165562

This patch provides initial implementation of load address 
macro instruction for Mips. We have implemented two kinds 
of expansions with their variations depending on the size 
of immediate operand:

 1) load address with immediate value directly:
    * la d,j => addiu d,$zero,j   (for -32768 <= j <= 65535)
    * la d,j => lui d,hi16(j)
                ori d,d,lo16(j)   (for any other 32 bit value of j)

 2) load load address with register offset value
    * la d,j(s) => addiu d,s,j     (for -32768 <= j <= 65535)
    * la d,j(s) => lui d,hi16(j)   (for any other 32 bit value of j)
                   ori d,d,lo16(j)
                   addu d,d,s

This patch does not cover the case when the address is loaded 
from the value of the label or function.

Contributer: Vladimir Medic
llvm-svn: 165561

MacroInfo*. Instead of simply dumping an offset into the current file,
give each macro definition a proper ID with all of the standard
modules-remapping facilities. Additionally, when a macro is modified
in a subsequent AST file (e.g., #undef'ing a macro loaded from another
module or from a precompiled header), provide a macro update record
rather than rewriting the entire macro definition. This gives us
greater consistency with the way we handle declarations, and ties
together macro definitions much more cleanly.

Note that we're still not actually deserializing macro history (we
never were), but it's far easy to do properly now.

llvm-svn: 165560

llvm-svn: 165559

I think our general framework for parser pragmas needs a bit more work,
but I'm not planning on working on it at the moment.

llvm-svn: 165558

llvm-svn: 165557

llvm-svn: 165554

llvm-svn: 165551