Duncan Sands d462ba8539 Teach LegalizeTypes how to expand and promote CTLZ, ... CTTZ and CTPOP. The expansion code differs from that in LegalizeDAG in that it chooses to take the CTLZ/CTTZ count from the Hi/Lo part depending on whether the Hi/Lo value is zero, not on whether CTLZ/CTTZ of Hi/Lo returned 32 (or whatever the width of the type is) for it. I made this change because the optimizers may well know that Hi/Lo is zero and exploit it. The promotion code for CTTZ also differs from that in LegalizeDAG: it uses an "or" to get the right result when the original value is zero, rather than using a compare and select. This also means the value doesn't need to be zero extended. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47075 91177308-0d34-0410-b5e6-96231b3b80d8		2008-02-13 18:01:53 +00:00
..
GC	Setting GlobalDirective in TargetAsmInfo by default rather than	2008-01-07 02:31:11 +00:00
2002-04-14-UnexpectedUnsignedType.ll
2002-04-16-StackFrameSizeAlignment.ll
2003-05-27-phifcmpd.ll
2003-05-27-useboolinotherbb.ll
2003-05-27-usefsubasbool.ll
2003-05-28-ManyArgs.ll
2003-05-30-BadFoldGEP.ll
2003-05-30-BadPreselectPhi.ll
2003-07-06-BadIntCmp.ll
2003-07-07-BadLongConst.ll
2003-07-08-BadCastToBool.ll
2003-07-29-BadConstSbyte.ll
2004-02-08-UnwindSupport.llx
2004-05-09-LiveVarPartialRegister.llx
2005-01-18-SetUO-InfLoop.ll
2005-04-09-GlobalInPHI.ll
2005-07-12-memcpy-i64-length.ll
2005-10-18-ZeroSizeStackObject.ll
2005-10-21-longlonggtu.ll
2005-12-01-Crash.ll
2005-12-12-ExpandSextInreg.ll
2006-01-12-BadSetCCFold.ll	avoid ambiguity in target selection	2007-07-05 22:16:19 +00:00
2006-01-18-InvalidBranchOpcodeAssert.ll
2006-02-12-InsertLibcall.ll
2006-03-01-dagcombineinfloop.ll
2006-03-27-DebugInfoNULLDeclare.ll
2006-04-11-vecload.ll
2006-04-26-SetCCAnd.ll
2006-04-28-Sign-extend-bool.ll
2006-05-06-GEP-Cast-Sink-Crash.ll
2006-06-12-LowerSwitchCrash.ll
2006-06-13-ComputeMaskedBitsCrash.ll
2006-06-28-SimplifySetCCCrash.ll
2006-07-03-schedulers.ll	-pre-RA-sched=none, simple, simple-noitin are gone.	2007-10-01 22:17:20 +00:00
2006-08-30-CoallescerCrash.ll
2006-09-02-LocalAllocCrash.ll
2006-09-06-SwitchLowering.ll
2006-10-27-CondFolding.ll
2006-10-29-Crash.ll
2006-11-06-MemIntrinsicExpand.ll
2006-11-20-DAGCombineCrash.ll
2006-12-16-InlineAsmCrash.ll
2007-01-15-LoadSelectCycle.ll
2007-02-16-BranchFold.ll	Test assumes tail merging is off; this must now be explicit.	2007-05-22 17:18:05 +00:00
2007-02-23-DAGCombine-Miscompile.ll	Change the x86 assembly output to use tab characters to separate the	2007-07-31 20:11:57 +00:00
2007-02-25-invoke.ll
2007-04-08-MultipleFrameIndices.ll	Don't use * in XFAIL line unless you want it to XFAIL on everything. I don't believe you can use true regular expressions here. It will just look for those keywords. Please test this on sparc and see if they still are xfailed.	2007-07-17 18:08:32 +00:00
2007-04-13-SwitchLowerBadPhi.ll
2007-04-14-BitTestsBadMask.ll
2007-04-14-EHSelectorCrash.ll	Workaround for PR1508.	2007-06-15 19:04:19 +00:00
2007-04-17-lsr-crash.ll
2007-04-27-BitTestsBadMask.ll	Convert tests using "grep -c ... | grep ..." to use the count script.	2007-08-15 13:49:33 +00:00
2007-04-27-InlineAsm-X-Dest.ll	Don't use * in XFAIL line unless you want it to XFAIL on everything. I don't believe you can use true regular expressions here. It will just look for those keywords. Please test this on sparc and see if they still are xfailed.	2007-07-17 18:08:32 +00:00
2007-04-27-LargeMemObject.ll	Don't use * in XFAIL line unless you want it to XFAIL on everything. I don't believe you can use true regular expressions here. It will just look for those keywords. Please test this on sparc and see if they still are xfailed.	2007-07-17 18:08:32 +00:00
2007-04-30-LandingPadBranchFolding.ll
2007-05-03-EHTypeInfo.ll	Split eh.select / eh.typeid.for intrinsics into i32/i64 versions. This is needed, because they just "mark" register	2007-09-07 11:39:35 +00:00
2007-05-05-Personality.ll	Split eh.select / eh.typeid.for intrinsics into i32/i64 versions. This is needed, because they just "mark" register	2007-09-07 11:39:35 +00:00
2007-05-15-InfiniteRecursion.ll	Fix an infinite recursion in GetNegatedExpression.	2007-05-15 17:05:43 +00:00
2007-06-06-CriticalEdgeLandingPad.ll	Split eh.select / eh.typeid.for intrinsics into i32/i64 versions. This is needed, because they just "mark" register	2007-09-07 11:39:35 +00:00
2007-11-21-UndeadIllegalNode.ll	Fix a bug in which node A is replaced by node B, but later	2007-11-21 16:43:19 +00:00
2007-12-17-InvokeAsm.ll	Make invokes of inline asm legal. Teach codegen	2007-12-17 18:08:19 +00:00
2007-12-31-UnusedSelector.ll	Fix PR1833 - eh.exception and eh.selector return two	2007-12-31 18:35:50 +00:00
2008-01-25-dag-combine-mul.ll	Fix some bugs in SimplifyNodeWithTwoResults where it would call deletenode to	2008-01-26 01:09:19 +00:00
2008-01-30-LoadCrash.ll	remove target triple to make this test more "generic"	2008-02-04 18:02:37 +00:00
2008-02-04-Ctlz.ll	Teach LegalizeTypes how to expand and promote CTLZ,	2008-02-13 18:01:53 +00:00
2008-02-04-ExtractSubvector.ll	Crashes LegalizeTypes with "Do not know how to split	2008-02-04 18:05:42 +00:00
addc-fold2.ll
asm-large-immediate.ll	Allow target constants to be illegal types. The target should	2007-08-25 01:00:22 +00:00
badarg6.ll
badCallArgLRLLVM.ll	Remove the trailing semicolon from function declarations in LLVM,	2007-07-16 13:37:30 +00:00
badFoldGEP.ll
badlive.ll
BasicInstrs.llx
bit-intrinsics.ll	This is the patch to provide clean intrinsic function overloading support in LLVM. It cleans up the intrinsic definitions and generally smooths the process for more complicated intrinsic writing. It will be used by the upcoming atomic intrinsics as well as vector and float intrinsics in the future.	2007-08-04 01:51:18 +00:00
bool-to-double.ll
bool-vector.ll	new testcase for PR1845	2008-01-10 00:30:38 +00:00
BurgBadRegAlloc.ll
call2-ret0.ll
call-ret0.ll
call-ret42.ll
call-void.ll
cast-fp.ll
ConstantExprLowering.llx
constindices.ll	Change all floating constants that are not exactly	2007-09-05 17:50:36 +00:00
debug-info.ll
dg.exp
div-neg-power-2.ll
fneg-fabs.ll
fp_to_int.ll	upgrade this test	2007-11-28 18:22:12 +00:00
fpowi-promote.ll
fwdtwice.ll	Fix a typo in a comment.	2007-09-24 15:50:11 +00:00
global-ret0.ll
hello.ll
i128-addsub.ll	add testcase for 128-bit add/sub	2007-05-17 18:22:37 +00:00
i128-arith.ll
intrinsics.ll
invalid-memcpy.ll	Fix a crash on invalid code due to memcpy lowering.	2007-11-27 22:14:42 +00:00
isunord.ll	the sparc v8 backend handles this correctly	2007-05-05 22:12:57 +00:00
llvm-ct-intrinsics.ll
Makefile
negintconst.ll
nested-select.ll
phi-immediate-factoring.ll	Convert tests using "| wc -l | grep ..." to use the count script.	2007-08-15 13:36:28 +00:00
print-add.ll
print-arith-fp.ll
print-arith-int.ll
print-int.ll
print-mul-exp.ll
print-mul.ll
print-shift.ll
ret0.ll
ret42.ll
sched.ll	Change all floating constants that are not exactly	2007-09-05 17:50:36 +00:00
select.ll	Change all floating constants that are not exactly	2007-09-05 17:50:36 +00:00
shift-int64.ll
spillccr.ll
stacksave-restore.ll
storetrunc-fp.ll	new testcase.	2008-01-17 19:47:23 +00:00
switch-crit-edge-constant.ll	Convert tests using "| wc -l | grep ..." to use the count script.	2007-08-15 13:36:28 +00:00
switch-lower-feature-2.ll	Convert tests using "| wc -l | grep ..." to use the count script.	2007-08-15 13:36:28 +00:00
switch-lower-feature.ll	Convert tests using "| wc -l | grep ..." to use the count script.	2007-08-15 13:36:28 +00:00
switch-lower.ll
SwitchLowering.ll	Convert tests using "| wc -l | grep ..." to use the count script.	2007-08-15 13:36:28 +00:00
trap.ll	new testcase for llvm.trap.	2008-01-15 22:17:26 +00:00
v-split.ll	add testcase that has been sitting in my tree for awhile.	2008-01-17 06:54:09 +00:00
vector-constantexpr.ll
vector-identity-shuffle.ll
vector.ll	Change all floating constants that are not exactly	2007-09-05 17:50:36 +00:00