use can't be dominated, saving one domtree lookup.

llvm-svn: 142066

I rewrote the algorithm a while back so it doesn't require map lookup,
but neglected to change the data structure. This was caught by
llvm-gcc self host, not because there's anything special about
llvm-gcc, but because it is the only test for nondeterminism we
currently have. Unit tests don't work well for everything; we should
always try to have a nondeterminism stress test running.

Fixes PR11133: llvm-gcc self host .o mismatch after enable-iv-rewrite=false

llvm-svn: 142036

Someone more familiar with LSR should double-check that the extra cast is actually doing the right thing in the overflow cases; I'm not completely confident that's that case. 

llvm-svn: 141916

Enhance the memdep interface so that users can tell the difference between a dependency which cannot be calculated and a path reaching the entry point of the function. This patch introduces isNonFuncLocal, which replaces isUnknown in some cases.

Patch by Xiaoyi Guo.

llvm-svn: 141896

Fix a couple hash functions so that they do not depend on undefined shifts.  Based on patch by Ahmed Charles.

llvm-svn: 141820

llvm-svn: 141750

would have never worked, since the element type of a vector type is never a
vector type. Also fix the conditional to be more direct in checking whether
EltTy is a vector type.

llvm-svn: 141713

lowering of NEON code. It provides little-to-no benefit now and only introduces
additional complexity.

llvm-svn: 141646

I'm not sure we will need it in the long run, but the option is
currently useful for checking if the output of LSR is "clean".

llvm-svn: 141634

IVs.

Indvars previously chose randomly between congruent IVs. Now it will
bias the decision toward IVs that SCEVExpander likes to create. This
was not done to fix any problem, it's just a welcome side effect of
factoring code.

llvm-svn: 141633

promoting allocas to preferred alignments that exceed the natural
alignment. This avoids some potentially expensive dynamic stack realignments.

The natural stack alignment is set in target data strings via the "S<size>"
option. Size is in bits and must be a multiple of 8. The natural stack alignment
defaults to "unspecified" (represented by a zero value), and the "unspecified"
value does not prevent any alignment promotions. Target maintainers that care
about avoiding promotions should explicitly add the "S<size>" option to their
target data strings.

llvm-svn: 141599

Fixes rdar://problem/5064068

llvm-svn: 141442

  switch (n) {
    case 27:
      do_something(x);
    ...
  }
the call do_something(x) will be replaced with do_something(27).  In
gcc-as-one-big-file this results in the removal of about 500 lines of
bitcode (about 0.02%), so has about 1/10 of the effect of propagating
branch conditions.

llvm-svn: 141360

llvm-svn: 141333

While I'm here, fix the related issue with strncmp, add some actual tests for strcmp and strncmp, and start using StringRef::compare for constant folding instead of using strcmp/strncmp so that the optimized IR isn't dependent on the host's implementation of strcmp.

llvm-svn: 141227

Just pull the instruction name, but don't change the order of anything
else. That keeps --debug happy and non-crashing, but doesn't change
how the worklist gets built.

llvm-svn: 141210

llvm-svn: 141209

When updating the worklist for InstCombine, the Add/AddUsersToWorklist
functions may access the instruction(s) being added, for debug output for
example. If the instructions aren't yet added to the basic block, this
can result in a crash. Finish the instruction transformation before
adjusting the worklist instead.

rdar://10238555

llvm-svn: 141203

branch "br i1 %x, label %if_true, label %if_false" then it replaces
"%x" with "true" in places only reachable via the %if_true arm, and
with "false" in places only reachable via the %if_false arm.  Except
that actually it doesn't: if value numbering shows that %y is equal
to %x then, yes, %y will be turned into true/false in this way, but
any occurrences of %x itself are not transformed.  Fix this.  What's
more, it's often the case that %x is an equality comparison such as
"%x = icmp eq %A, 0", in which case every occurrence of %A that is
only reachable via the %if_true arm can be replaced with 0.  Implement
this and a few other variations on this theme.  This reduces the number
of lines of LLVM IR in "GCC as one big file" by 0.2%.  It has a bigger
impact on Ada code, typically reducing the number of lines of bitcode
by around 0.4% by removing repeated compiler generated checks.  Passes
the LLVM nightly testsuite and the Ada ACATS testsuite.

llvm-svn: 141177

it's OK for the false/true destination to have multiple
predecessors as long as the extra ones are dominated by
the branch destination.

llvm-svn: 141176

This handles the case in which LSR rewrites an IV user that is a phi and
splits critical edges originating from a switch.
Fixes <rdar://problem/6453893> LSR is not splitting edges "nicely"

llvm-svn: 141059

llvm-svn: 141058

r140966.

llvm-svn: 140969

but not load instructions. Noticed by inspection.

llvm-svn: 140966

We want heuristics to be based on accurate data, but more importantly
we don't want llvm to behave randomly. A benign trunc inserted by an
upstream pass should not cause a wild swings in optimization
level. See PR11034. It's a general problem with threshold-based
heuristics, but we can make it less bad.

llvm-svn: 140919

llvm-svn: 140916

llvm-svn: 140875

InstCombine was incorrectly considering the conversion of the constant
zero to be unsafe.

We want to transform:
define float @bar(float %x) nounwind readnone optsize ssp {
  %conv = fpext float %x to double
  %cmp = fcmp olt double %conv, 0.000000e+00
  %conv1 = zext i1 %cmp to i32
  %conv2 = sitofp i32 %conv1 to float
  ret float %conv2
}

Into:
define float @bar(float %x) nounwind readnone optsize ssp {
  %cmp = fcmp olt float %x, 0.000000e+00   ; <---- This
  %conv1 = zext i1 %cmp to i32
  %conv2 = sitofp i32 %conv1 to float
  ret float %conv2
}


rdar://10215914

llvm-svn: 140869

llvm-svn: 140865

catch or repeated filter clauses.  Teach instcombine a bunch
of tricks for simplifying landingpad clauses.  Currently the
code only recognizes the GNU C++ and Ada personality functions,
but that doesn't stop it doing a bunch of "generic" transforms
which are hopefully fine for any real-world personality function.
If these "generic" transforms turn out not to be generic, they
can always be conditioned on the personality function.  Probably
someone should add the ObjC++ personality function.  I didn't as
I don't know anything about it.

llvm-svn: 140852

llvm-svn: 140821

handle the case where the retain is in a different basic block.
rdar://10210274.

llvm-svn: 140815

objc_retainBlock call is potentially responsible for copying
the block to the heap to extend its lifetime. rdar://10209613.

llvm-svn: 140814

Clean up uses of switch instructions so they are not dependent on the operand ordering.  Patch by Stepan Dyatkovskiy.

llvm-svn: 140803

llvm-svn: 140769

Rewriting the entire loop nest now requires -enable-lsr-nested.
See PR11035 for some performance data.
A few unit tests specifically test nested LSR, and are now under a flag.

llvm-svn: 140762

llvm-svn: 140670

Stop emitting instructions with the name "tmp" they eat up memory and have to be uniqued, without any benefit.

If someone prefers %tmp42 to %42, run instnamer.

llvm-svn: 140634

split landingpad instructions into a PHI node.
PR11016

llvm-svn: 140592

Disabling aggressive LSR saves compilation time, and with the new
indvars behavior usually improves performance.

llvm-svn: 140590