llvm-svn: 131480

llvm-svn: 131427

Radar 9422775.

llvm-svn: 131422

corrupted when setjmp returns again.

llvm-svn: 131399

llvm-svn: 131345

rdar://problem/9267970

Patch by Julien Lerouge!

llvm-svn: 131339

set.

rdar://9423996

llvm-svn: 131283

the users.

llvm-svn: 131277

Interleave IV simplifications. Currently involves EliminateComparison
and EliminateRemainder. Next I'll add EliminateExtend.

llvm-svn: 131210

llvm-svn: 131112

return the pointer being dereferenced, it returns the pointee, but a call
might return the pointer itself.

llvm-svn: 130979

llvm-svn: 130967

PR9838: Fix transform introduced in r127064 to not trigger when only one side of the icmp is an exact shift.

llvm-svn: 130954

crash.

llvm-svn: 130911

llvm-svn: 130903

filename.

llvm-svn: 130897

llvm-svn: 130895

llvm-svn: 130894

llvm-svn: 130880

llvm-svn: 130876

llvm-svn: 130869

frontend, if applicable.

llvm-svn: 130835

This adds functionality to remove size/zero extension during indvars
without generating a canonical IV and rewriting all IV users. It's
disabled by default so should have no effect on codegen. Work in progress.

llvm-svn: 130829

Only create a canonical IV for backedge taken count if it will
actually be used by LinearFunctionTestReplace. And some related
cleanup, preparing to reduce dependence on canonical IVs.
No significant effect on x86 or arm in the test-suite.

llvm-svn: 130799

llvm-svn: 130755

model constants which can be added to base registers via add-immediate
instructions which don't require an additional register to materialize
the immediate.

llvm-svn: 130743

Scanning entire basic block may be too expensive in terms of compile time. Instead, just use whatever location info first non-phi instruction has.

llvm-svn: 130729

llvm-svn: 130705

This automagically provides a transform noticed by my super-optimizer
as occurring quite often: "rem x, (select cond, x, 1)" -> 0.

llvm-svn: 130694

problem reported on cfe-dev.

llvm-svn: 130661

This obviously helps a lot if the division would be turned into a libcall
(think i64 udiv on i386), but div is also one of the few remaining instructions
on modern CPUs that become more expensive when the bitwidth gets bigger.

This also helps register pressure on i386 when dividing chars, divb needs
two 8-bit parts of a 16 bit register as input where divl uses two registers.

int foo(unsigned char a) { return a/10; }
int bar(unsigned char a, unsigned char b) { return a/b; }

compiles into (x86_64)
_foo:
  imull $205, %edi, %eax
  shrl  $11, %eax
  ret
_bar:
  movzbl        %dil, %eax
  divb  %sil, %al
  movzbl        %al, %eax
  ret

llvm-svn: 130615

This folds away silly stuff like (a&255)/1000 -> 0.

llvm-svn: 130614

llvm-svn: 130551

llvm-svn: 130536

llvm-svn: 130528

llvm-svn: 130527

llvm-svn: 130526

llvm-svn: 130489

Fixes PR9809.

llvm-svn: 130485

llvm-svn: 130450