Commit Graph

9444 Commits

Author SHA1 Message Date
Duncan Sands
c5b969a019 Generalize the transform that boosts GEP indices to the size of a pointer to
also do it for vectors of pointers.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167354 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-03 11:44:17 +00:00
Alexey Samsonov
9ce84c1c95 Fix whitespaces
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167295 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-02 12:20:34 +00:00
Chandler Carruth
a864748284 Revert the switch of loop-idiom to use the new dependence analysis.
The new analysis is not yet ready for prime time. It has a *critical*
flawed assumption, and some troubling shortages of testing. Until it's
been hammered into better shape, let's stick with the working code. This
should be easy to revert itself when the analysis is ready.

Fixes PR14241, a miscompile of any memcpy-able loop which uses a pointer
as the induction mechanism. If you have been seeing miscompiles in this
revision range, you really want to test with this backed out. The
results of this miscompile are a bit subtle as they can lead to
downstream passes concluding things are impossible which are in fact
possible.

Thanks to David Blaikie for the majority of the reduction of this
miscompile. I'll be checking in the test case in a non-revert commit.

Revesions reverted here:

r167045: LoopIdiom: Fix a serious missed optimization: we only turned
         top-level loops into memmove.
r166877: LoopIdiom: Add checks to avoid turning memmove into an infinite
         loop.
r166875: LoopIdiom: Recognize memmove loops.
r166874: LoopIdiom: Replace custom dependence analysis with
         DependenceAnalysis.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167286 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-02 08:33:25 +00:00
Duncan Sands
d295c9729d Fix an obvious typo that causes an assertion failure when running
test/Transforms/GVN/rle.ll if the (currently disabled) check for a
pointer type in getIntPtrType is turned on.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167285 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-02 07:49:32 +00:00
Chandler Carruth
6e48f03077 Fix sign compare warning. Patch by Mahesha HS.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167282 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-02 05:24:00 +00:00
Hal Finkel
78fd353d5e BBVectorize: Use target costs for incoming and outgoing values instead of the depth heuristic.
When target cost information is available, compute explicit costs of inserting and
extracting values from vectors. At this point, all costs are estimated using the
target information, and the chain-depth heuristic is not needed. As a result, it is now, by
default, disabled when using target costs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167256 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-01 21:50:12 +00:00
Kostya Serebryany
a5f54f1443 [asan] don't instrument globals that we've created ourselves (reduces the binary size a bit)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167230 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-01 13:42:40 +00:00
Chandler Carruth
426c2bf5cd Revert the majority of the next patch in the address space series:
r165941: Resubmit the changes to llvm core to update the functions to
         support different pointer sizes on a per address space basis.

Despite this commit log, this change primarily changed stuff outside of
VMCore, and those changes do not carry any tests for correctness (or
even plausibility), and we have consistently found questionable or flat
out incorrect cases in these changes. Most of them are probably correct,
but we need to devise a system that makes it more clear when we have
handled the address space concerns correctly, and ideally each pass that
gets updated would receive an accompanying test case that exercises that
pass specificaly w.r.t. alternate address spaces.

However, from this commit, I have retained the new C API entry points.
Those were an orthogonal change that probably should have been split
apart, but they seem entirely good.

In several places the changes were very obvious cleanups with no actual
multiple address space code added; these I have not reverted when
I spotted them.

In a few other places there were merge conflicts due to a cleaner
solution being implemented later, often not using address spaces at all.
In those cases, I've preserved the new code which isn't address space
dependent.

This is part of my ongoing effort to clean out the partial address space
code which carries high risk and low test coverage, and not likely to be
finished before the 3.2 release looms closer. Duncan and I would both
like to see the above issues addressed before we return to these
changes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167222 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-01 09:14:31 +00:00
Chandler Carruth
ece6c6bb63 Revert the series of commits starting with r166578 which introduced the
getIntPtrType support for multiple address spaces via a pointer type,
and also introduced a crasher bug in the constant folder reported in
PR14233.

These commits also contained several problems that should really be
addressed before they are re-committed. I have avoided reverting various
cleanups to the DataLayout APIs that are reasonable to have moving
forward in order to reduce the amount of churn, and minimize the number
of commits that were reverted. I've also manually updated merge
conflicts and manually arranged for the getIntPtrType function to stay
in DataLayout and to be defined in a plausible way after this revert.

Thanks to Duncan for working through this exact strategy with me, and
Nick Lewycky for tracking down the really annoying crasher this
triggered. (Test case to follow in its own commit.)

After discussing with Duncan extensively, and based on a note from
Micah, I'm going to continue to back out some more of the more
problematic patches in this series in order to ensure we go into the
LLVM 3.2 branch with a reasonable story here. I'll send a note to
llvmdev explaining what's going on and why.

Summary of reverted revisions:

r166634: Fix a compiler warning with an unused variable.
r166607: Add some cleanup to the DataLayout changes requested by
         Chandler.
r166596: Revert "Back out r166591, not sure why this made it through
         since I cancelled the command. Bleh, sorry about this!
r166591: Delete a directory that wasn't supposed to be checked in yet.
r166578: Add in support for getIntPtrType to get the pointer type based
         on the address space.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167221 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-01 08:07:29 +00:00
Hal Finkel
86ccc55c82 BBVectorize: Account for internal shuffle costs
When target costs are available, use them to account for the costs of
shuffles on internal edges of the DAG of candidate pairs.

Because the shuffle costs here are currently for only the internal edges,
the current target cost model is trivial, and the chain depth requirement
is still in place, I don't yet have an easy test
case. Nevertheless, by looking at the debug output, it does seem to do the right
think to the effective "size" of each DAG of candidate pairs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167217 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-01 06:26:34 +00:00
Jakub Staszak
5801ff93e3 Don't insert and erase load instruction. Simply create (new) and delete it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167196 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-01 01:10:43 +00:00
Nadav Rotem
e57b2cbce6 LoopVectorize: Preserve NSW, NUW and IsExact flags.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167174 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 21:40:39 +00:00
Benjamin Kramer
ccd492c9d6 LCSSA: Try to recover compile time regressions due to SCEV updates.
- Use value handle tricks to communicate use replacements instead of forgetLoop, this is a lot faster.
- Move the "big hammer" out of the main loop so it's not called for every instruction.

This should recover most (if not all) compile time regressions introduced by this code.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167136 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 16:30:03 +00:00
Nadav Rotem
4c1b4b1fe7 Put the threshold magic number in a variable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167134 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 16:22:16 +00:00
Hans Wennborg
eb3e089fb6 Remove fixme about unreachable cases from SwitchToLookupTable
SimplifyCFG will have removed those cases for us.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167132 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 16:15:25 +00:00
Nadav Rotem
2a5401cc75 Remove enum values since they are not used anymore.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167131 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 16:14:06 +00:00
Hans Wennborg
daf4cfc8d8 Address Duncan's comments on r167121.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167130 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 15:31:09 +00:00
Hal Finkel
72465ea23d BBVectorize: Choose pair ordering to minimize shuffles
BBVectorize would, except for loads and stores, always fuse instructions
so that the first instruction (in the current source order) would always
represent the low part of the input vectors and the second instruction
would always represent the high part. This lead to too many shuffles
being produced because sometimes the opposite order produces fewer of them.

With this change, BBVectorize tracks the kind of pair connections that form
the DAG of candidate pairs, and uses that information to reorder the pairs to
avoid excess shuffles. Using this information, a future commit will be able
to add VTTI-based shuffle costs to the pair selection procedure. Importantly,
the number of remaining shuffles can now be estimated during pair selection.

There are some trivial instruction reorderings in the test cases, and one
simple additional test where we certainly want to do a reordering to
avoid an unnecessary shuffle.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167122 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 15:17:07 +00:00
Hans Wennborg
ef026f1b5e Address Duncan's comments on r167115
- Use 0 instead of NULL
 - Helper function for "dyn_cast, else lookup in the constant pool".

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167121 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 15:14:39 +00:00
Meador Inge
e0f1dca1c8 instcombine: Migrate strto* optimizations
This patch migrates the strto* optimizations from the simplify-libcalls
pass into the instcombine library call simplifier.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167119 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 14:58:26 +00:00
Hans Wennborg
2857866138 Fix false -> NULL conversion from r167115 spotted by Benjamin Kramer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167117 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 14:36:48 +00:00
Benjamin Kramer
b4c9d9c51f Replace some instances of UniqueVector with SetVector, which is slightly cheaper.
No functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167116 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 13:45:49 +00:00
Hans Wennborg
e03d9e4ec7 Do simple constant propagation in lookup table formation for switches
By propagating the value for the switch condition, LLVM can now build
lookup tables for code such as:

  switch (x) {
    case 1: return 5;
    case 2: return 42;
    case 3: case 4: case 5:
      return x - 123;
    default:
      return 123;
  }

Given that x is known for each case, "x - 123" becomes a constant for
cases 3, 4, and 5.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167115 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 13:42:45 +00:00
Benjamin Kramer
4ad3d981b9 LCSSA: Add a workaround for another nasty SCEV cache invalidation issue.
I'm not entirely happy with this solution, but I don't see a smarter way currently.
Fixes PR14214.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167112 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 10:01:29 +00:00
Meador Inge
08684d1f06 instcombine: Migrate strpbrk optimizations
This patch migrates the strpbrk optimizations from the simplify-libcalls
pass into the instcombine library call simplifier.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167105 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 04:29:58 +00:00
Meador Inge
57cfd71f88 instcombine: Migrate strlen optimizations
This patch migrates the strlen optimizations from the simplify-libcalls
pass into the instcombine library call simplifier.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167103 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 03:33:06 +00:00
Meador Inge
a0885fb882 instcombine: Migrate strncpy optimizations
This patch migrates the strncpy optimizations from the simplify-libcalls
pass into the instcombine library call simplifier.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167102 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 03:33:00 +00:00
Nadav Rotem
5b2c4dc5f8 LoopVectorize: Do not vectorize loops with tiny constant trip counts.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167101 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 03:31:07 +00:00
Nadav Rotem
462d1ca428 Add support for loops that don't start with Zero.
This is important for loops in the LAPACK test-suite.
These loops start at 1 because they are auto-converted from fortran.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167084 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 00:45:26 +00:00
Meador Inge
e6d781fd3c instcombine: Migrate stpcpy optimizations
This patch migrates the stpcpy optimizations from the simplify-libcalls
pass into the instcombine library call simplifier.  Note that the
__stpcpy_chk simplifications were migrated in a previous commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167083 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 00:20:56 +00:00
Meador Inge
fa9d1372c9 instcombine: Split out the __stpcpy_chk simplifications from StrCpyChkOpt
r166198 migrated the strcpy optimization to instcombine.  The strcpy
simplifier that was migrated from Transforms/Scalar/SimplifyLibCalls.cpp
was also doing some __strcpy_chk simplifications.  Those fortified
simplifications were migrated as well, but introduced a bug in the
__stpcpy_chk simplifier in the process.  This happened because the
__strcpy_chk and __stpcpy_chk simplifiers were both mapped to StrCpyChkOpt
which was updated with simplifications that worked for __strcpy_chk, but
not __stpcpy_chk.

This patch fixes the problem by adding proper test coverage and creating a
new simplifier for __stpcpy_chk (instead of sharing one with __strcpy_chk).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167082 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-31 00:20:51 +00:00
Nadav Rotem
a368b89f28 Add documentation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167055 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-30 22:06:26 +00:00
Chandler Carruth
70dace3052 Fix PR14212: For some strange reason I treated vectors differently from
integers in that the code to handle split alloca-wide integer loads or
stores doesn't come first. It should, for the same reasons as with
integers, and the PR attests to that. Also had to fix a busted assert in
that this test case also covers.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167051 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-30 20:52:40 +00:00
Hal Finkel
a9779bfbc9 BBVectorize: Cache fixed-order pairs instead of recomputing pointer info.
Instead of recomputing relative pointer information just prior to fusing,
cache this information (which also needs to be computed during the
candidate-pair selection process). This cuts down on the total number of
SE queries made, and also is a necessary intermediate step on the road toward
including shuffle costs in the pair selection procedure.

No functionality change is intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167049 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-30 20:17:37 +00:00
Benjamin Kramer
415f869cf3 LoopIdiom: Fix a serious missed optimization: we only turned top-level loops into memmove.
Thanks to Preston Briggs for catching this!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167045 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-30 19:49:39 +00:00
Hal Finkel
8a69a95f91 BBVectorize: Fix a small bug introduced in r167042.
We need to make sure that we take the correct load/store alignment
when the inputs are flipped.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167044 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-30 19:47:37 +00:00
Hal Finkel
202d1cb8a5 BBVectorize: Simplify how input swapping is handled.
Stop propagating the FlipMemInputs variable into the routines that
create the replacement instructions. Instead, just flip the arguments
of those routines. This allows for some associated cleanup (not all
of which is done here). No functionality change is intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167042 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-30 19:35:29 +00:00
Hal Finkel
93f6f45761 BBVectorize: Don't make calls to SE when the result is unused.
SE was being called during the instruction-fusion process (when the result
is unreliable, and thus ignored). No functionality change is intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167037 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-30 18:55:49 +00:00
Nadav Rotem
9210cc3a17 80-col
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167036 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-30 18:37:43 +00:00
Nadav Rotem
e709f5b600 LoopVectorize: Add support for write-only loops when the write destination is a single pointer.
Speedup SciMark by 1%



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167035 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-30 18:36:45 +00:00
Nadav Rotem
cc03331caa LoopVectorize: Fix a bug in the initialization of reduction variables. AND needs to start at all-one
while XOR, and OR need to start at zero.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167032 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-30 18:12:36 +00:00
Duncan Sands
446cf94cdb Fix isEliminableCastPair to work correctly in the presence of pointers
with different sizes.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167018 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-30 16:03:32 +00:00
Ulrich Weigand
3467b9fd5d Enable some additional constant folding for PPCDoubleDouble.
This fixes Clang :: CodeGen/complex-builtints.c on PowerPC.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167013 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-30 12:33:18 +00:00
Hans Wennborg
04d7d13d30 Use TargetTransformInfo to control switch-to-lookup table transformation
When the switch-to-lookup tables transform landed in SimplifyCFG, it
was pointed out that this could be inappropriate for some targets.
Since there was no way at the time for the pass to know anything about
the target, an awkward reverse-transform was added in CodeGenPrepare
that turned lookup tables back into switches for some targets.

This patch uses the new TargetTransformInfo to determine if a
switch should be transformed, and removes
CodeGenPrepare::ConvertLoadToSwitch.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167011 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-30 11:23:25 +00:00
Nadav Rotem
803ea1c40a LoopVectorizer: change debug prints: Print the module identifier when deciding to vectorize. When deciding not to vectorize do not print the called function name because it can be null.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@166989 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-30 00:40:39 +00:00
Nadav Rotem
369ff7b740 LoopVectorize: Update and preserve the dominator tree info.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@166970 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-29 21:52:38 +00:00
Ulrich Weigand
e669c930a6 In various places throughout the code generator, there were special
checks to avoid performing compile-time arithmetic on PPCDoubleDouble.

Now that APFloat supports arithmetic on PPCDoubleDouble, those checks
are no longer needed, and we can treat the type like any other.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@166958 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-29 18:35:49 +00:00
Nadav Rotem
2459bf2427 Rename the BB-vectorize flag to match the dragonegg name
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@166948 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-29 18:01:14 +00:00
Duncan Sands
7ed4f94c13 Remove a wrapper around getIntPtrType added to GVN by Hal in commit 166624 (the
wrapper returns a vector of integers when passed a vector of pointers) by having
getIntPtrType itself return a vector of integers in this case.  Outside of this
wrapper, I didn't find anywhere in the codebase that was relying on the old
behaviour for vectors of pointers, so give this a whirl through the buildbots.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@166939 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-29 17:31:46 +00:00
Nadav Rotem
d233b78760 Change the PassManagerBuilder (used by -O3) loop vectorizer flag from -vectorize to -vectorize-loops because we dont want to share the same flag as the bb-vectorizer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@166937 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-29 16:36:25 +00:00