llvm-6502/test/Transforms
Chandler Carruth ffc97a6ae1 [slp] Fix a nasty bug in the SLP vectorizer that Joerg pointed out.
Apparently some code finally started to tickle this after my
canonicalization changes to instcombine.

The bug stems from trying to form a vector type out of scalars that
aren't compatible at all. In this example, from x86_mmx values. The code
in the vectorizer that checks for reasonable types whas checking for
aggregates or vectors, but there are lots of other types that should
just never reach the vectorizer.

Debugging this was made more confusing by the lie in an assert in
VectorType::get() -- it isn't that the types are *primitive*. The types
must be integer, pointer, or floating point types. No other types are
allowed.

I've improved the assert and added a helper to the vectorizer to handle
the element type validity checks. It now re-uses the VectorType static
function and then further excludes weird target-specific types that we
probably shouldn't be touching here (x86_fp80 and ppc_fp128). Neither of
these are really reachable anyways (neither 80-bit nor 128-bit things
will get vectorized) but it seems better to just eagerly exclude such
nonesense.

I've added a test case, but while it definitely covers two of the paths
through this code there may be more paths that would benefit from test
coverage. I'm not familiar enough with the SLP vectorizer to synthesize
test cases for all of these, but was able to update the code itself by
inspection.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@228899 91177308-0d34-0410-b5e6-96231b3b80d8
2015-02-12 02:30:56 +00:00
..
ADCE
AddDiscriminators IR: Move MDLocation into place 2015-01-14 22:27:36 +00:00
AlignmentFromAssumptions [AlignmentFromAssumptions] Don't crash just because the target is 32-bit 2014-09-11 08:40:17 +00:00
ArgumentPromotion IR: Move MDLocation into place 2015-01-14 22:27:36 +00:00
AtomicExpand/ARM Use target-dependent emitLeading/TrailingFence instead of the target-independent insertLeading/TrailingFence (in AtomicExpandPass) 2014-09-03 21:01:03 +00:00
BBVectorize This patch teaches IndVarSimplify to add nuw and nsw to certain kinds 2015-01-06 19:02:56 +00:00
BranchFolding Delete -std-compile-opts. 2014-10-16 20:00:02 +00:00
CodeExtractor
CodeGenPrepare [GC] CodeGenPrep transform: simplify offsetable relocate 2015-01-14 23:27:07 +00:00
ConstantHoisting
ConstantMerge
ConstProp
CorrelatedValuePropagation LazyValueInfo: Actually re-visit partially solved block-values in solveBlockValue() 2014-11-25 17:23:05 +00:00
DeadArgElim DeadArgElim: aggregate Return assessment properly. 2015-02-11 23:13:11 +00:00
DeadStoreElimination IR: Make metadata typeless in assembly 2014-12-15 19:07:53 +00:00
EarlyCSE EarlyCSE: Add check lines for test added in r228760 2015-02-10 23:11:02 +00:00
FunctionAttrs Normally an 'optnone' function goes through fast-isel, which does not 2014-11-03 18:19:26 +00:00
GCOVProfiling IR: Move MDLocation into place 2015-01-14 22:27:36 +00:00
GlobalDCE Change the .ll syntax for comdats and add a syntactic sugar. 2015-01-06 22:55:16 +00:00
GlobalOpt IR: Move MDLocation into place 2015-01-14 22:27:36 +00:00
GVN Allow PRE to insert no-cost phi nodes 2015-02-03 20:37:08 +00:00
IndVarSimplify Fix PR22222 2015-01-15 01:46:09 +00:00
Inline Value soft float calls as more expensive in the inliner. 2015-02-05 02:09:33 +00:00
InstCombine Revert r228556: InstCombine: propagate nonNull through assume 2015-02-10 08:07:32 +00:00
InstMerge Added 5 more tests related to sink store revision 224247 2014-12-17 08:12:59 +00:00
InstSimplify Fold fcmp in cases where value is provably non-negative. By Arch Robison. 2015-01-28 08:03:58 +00:00
Internalize
IPConstantProp
IRCE Teach IRCE to look at branch weights when recognizing range checks 2015-01-27 21:38:12 +00:00
JumpThreading ValueTracking: ComputeNumSignBits should tolerate misshapen phi nodes 2015-01-04 07:06:53 +00:00
LCSSA [LCSSA] Handle PHI insertion in disjoint loops 2014-12-22 22:35:46 +00:00
LICM Teach isDereferenceablePointer() to look through bitcast constant expressions. 2015-02-05 09:15:37 +00:00
LoadCombine Use AA in LoadCombine 2014-11-03 23:19:16 +00:00
LoopDeletion
LoopIdiom IR: Move MDLocation into place 2015-01-14 22:27:36 +00:00
LoopReroll [LoopReroll] Introduce the concept of DAGRootSets. 2015-02-11 09:19:47 +00:00
LoopRotate LoopRotate: Don't walk the uses of a Constant 2015-01-27 06:21:43 +00:00
LoopSimplify Teach SplitBlockPredecessors how to handle landingpad blocks. 2015-01-28 23:06:47 +00:00
LoopStrengthReduce Reverting VLD1/VST1 base-updating/post-incrementing combining 2015-02-04 10:11:59 +00:00
LoopUnroll Add a test case for new unrolling heuristics. 2015-02-10 17:54:54 +00:00
LoopUnswitch
LoopVectorize Move the target specific test case arbitrary-induction-step.ll to test/Transforms/LoopVectorize/AArch64 folder. 2015-01-30 07:33:31 +00:00
LowerAtomic
LowerExpectIntrinsic [PM] Port LowerExpectIntrinsic to the new pass manager. 2015-01-24 11:13:02 +00:00
LowerInvoke
LowerSwitch LowerSwitch: replace unreachable default with popular case destination 2015-01-23 20:43:51 +00:00
Mem2Reg IR: Move MDLocation into place 2015-01-14 22:27:36 +00:00
MemCpyOpt ValueTracking: Make isBytewiseValue simpler and more powerful at the same time. 2015-02-07 19:29:02 +00:00
MergeFunc IR: Make metadata typeless in assembly 2014-12-15 19:07:53 +00:00
MetaRenamer
ObjCARC IR: Move MDLocation into place 2015-01-14 22:27:36 +00:00
PartiallyInlineLibCalls
PhaseOrdering
PlaceSafepoints PlaceSafepoints: modernize gc.result.* -> gc.result 2015-02-09 23:00:40 +00:00
PruneEH Fix invalid LLVM IR in PruneEH tests 2015-02-11 02:06:47 +00:00
Reassociate Reassociate: cannot negate a INT_MIN value 2015-02-11 19:54:44 +00:00
Reg2Mem
SampleProfile IR: Move MDLocation into place 2015-01-14 22:27:36 +00:00
Scalarizer IR: Move MDLocation into place 2015-01-14 22:27:36 +00:00
ScalarRepl IR: Move MDLocation into place 2015-01-14 22:27:36 +00:00
SCCP SCCP: overdefined calls cannot become constant 2014-11-07 08:54:19 +00:00
SeparateConstOffsetFromGEP/NVPTX [SeparateConstOffsetFromGEP] Fixed a bug related to unsigned modulo 2014-10-25 18:34:03 +00:00
SimplifyCFG [TTI] Improved cost heuristic for cttz/ctlz calls. 2015-02-11 14:22:18 +00:00
Sink
SLPVectorizer [slp] Fix a nasty bug in the SLP vectorizer that Joerg pointed out. 2015-02-12 02:30:56 +00:00
SROA [SROA] Apply a somewhat heavy and unpleasant hammer to fix PR22093, an 2015-01-05 04:17:53 +00:00
StraightLineStrengthReduce Add straight-line strength reduction to LLVM 2015-02-03 19:37:06 +00:00
StripSymbols IR: Move MDLocation into place 2015-01-14 22:27:36 +00:00
StructurizeCFG StructurizeCFG: Use a reverse post-order traversal 2015-02-04 20:49:44 +00:00
TailCallElim Fix tail recursion elimination 2014-11-19 13:32:51 +00:00
TailDup
Util Correctly combine alias.scope metadata by a union instead of intersecting 2015-02-08 17:07:14 +00:00