llvm-svn: 94616

After running a batch of measurements, it is clear that the inliner metrics
need some adjustments:

Own argument bonus:       20 -> 5
Outgoing argument penalty: 0 -> 5
Alloca bonus:             10 -> 5
Constant instr bonus:      7 -> 5
Dead successor bonus:     40 -> 5*(avg instrs/block)

The new cost metrics are generaly 25 points higher than before, so we may need
to move thresholds.

With this change, InlineConstants::CallPenalty becomes a political correction:

if (!isa<IntrinsicInst>(II) && !callIsSmall(CS.getCalledFunction()))
  NumInsts += InlineConstants::CallPenalty + CS.arg_size();

The code size is accurately modelled by CS.arg_size(). CallPenalty is added
because calls tend to take a long time, so it may not be worth it to inline a
function with lots of calls.

All of the political corrections are in the InlineConstants namespace:
IndirectCallBonus, CallPenalty, LastCallToStaticBonus, ColdccPenalty,
NoreturnPenalty.

llvm-svn: 94615

logic up from X86 into the common code.  The other targets will 
hopefully start using this soon.

llvm-svn: 94614

is constified.

llvm-svn: 94613

llvm-svn: 94612

llvm-svn: 94611

llvm-svn: 94610

llvm-svn: 94601

Before existing NamedMDNode entry in the symbol table, remove any existing entry with the same name.

llvm-svn: 94600

assignments.

.set x, a-b

is the same as:

x = a-b

llvm-svn: 94596

llvm-as: t.ll:1:25: error: invalid cast opcode for cast from '[4 x i8]' to '[1 x i32]'
@x = constant [1 x i32] bitcast ([4 x i8] c"abcd" to [1 x i32])
                        ^

llvm-svn: 94595

llvm-svn: 94594

llvm-svn: 94593

A GEP with all constant indices is already considered free by
analyzeBasicBlock(), so don't give it an extra bonus in
CountCodeReductionForAlloca().

This patch should remove a small positive bias toward inlining functions with
variable-index GEPs, and remove a smaller negative bias from functions with
all-constant index GEPs.

llvm-svn: 94591

Functions containing indirectbr are marked NeverInline by analyzeBasicBlock(),
so there is no point in giving indirectbr special treatment in
CountCodeReductionForConstant. It is never called.

No functional change intended.

llvm-svn: 94590

Save a few bytes by allocating the correct size vector.

No functional change intended.

llvm-svn: 94589

llvm-svn: 94587

Add extra element to composite type. This new element will be used to record c++ class that holds current class's vtable.

llvm-svn: 94586

Default HasSetDirective to true, since most targets have it.

The targets that claim to not have it probably do, or it is
spelled differently. These include Blackfin, Mips, Alpha, and
PIC16.  All of these except pic16 are normal ELF targets, so
they almost certainly have it.

llvm-svn: 94585

llvm-svn: 94583

Original patch by Sandeep Patel and updated by me.

llvm-svn: 94582

that has it.

llvm-svn: 94581

llvm-svn: 94580

llvm-svn: 94578

llvm-svn: 94574

and clean up some loose ends.

llvm-svn: 94572

llvm-svn: 94570

llvm-svn: 94569

llvm-svn: 94567

llvm-svn: 94562

llvm-svn: 94560

llvm-svn: 94558

than an iteration count.

llvm-svn: 94549

in the case of empty and full ranges.

llvm-svn: 94548

wrapping that Duncan pointed out.

llvm-svn: 94547

llvm-svn: 94546

llvm-svn: 94545

llvm-svn: 94544

llvm-svn: 94543

to append a newline when using setModuleInlineAsm in case later
someone calls appendModuleInlineAsm - make newline handling fully
automatic.  In case anyone is wondering there is only one user of
appendModuleInlineAsm: the dragonegg plugin.

llvm-svn: 94539