Eli Friedman b80f778bd3 Get rid of an optimization in SCCP which appears to have many issues. Specifically, it doesn't handle many cases involving undef correctly, and it is missing other checks which ...

lead to it trying to re-mark a value marked as a constant with a different value.  It also appears to trigger very rarely.

Fixes PR11357.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@144352 91177308-0d34-0410-b5e6-96231b3b80d8

2011-11-11 01:16:15 +00:00

..

2002-05-02-MissSecondInst.ll

…

2002-05-20-MissedIncomingValue.ll

…

2002-05-21-InvalidSimplify.ll

…

2002-08-30-GetElementPtrTest.ll

…

2003-06-24-OverdefinedPHIValue.ll

…

2003-08-26-InvokeHandling.ll

Update the tests to the new EH scheme.

2011-08-31 20:55:40 +00:00

2004-11-16-DeadInvoke.ll

Update the tests to the new EH scheme.

2011-08-31 20:55:40 +00:00

2004-12-10-UndefBranchBug.ll

…

2006-10-23-IPSCCP-Crash.ll

…

2006-12-04-PackedType.ll

…

2006-12-19-UndefBug.ll

…

2007-05-16-InvokeCrash.ll

Update the tests to the new EH scheme.

2011-08-31 20:55:40 +00:00

2008-01-27-UndefCorrelate.ll

…

2008-04-22-multiple-ret-sccp.ll

remove asmparser support for the old getresult instruction, which has been subsumed by extractvalue.

2011-06-17 06:57:15 +00:00

2008-05-23-UndefCallFold.ll

…

2009-01-14-IPSCCP-Invoke.ll

Update the tests to the new EH scheme.

2011-08-31 20:55:40 +00:00

2009-05-27-VectorOperandZero.ll

…

apint-array.ll

…

apint-basictest2.ll

…

apint-basictest3.ll

…

apint-basictest4.ll

…

apint-basictest.ll

…

apint-bigarray.ll

…

apint-bigint2.ll

…

apint-bigint.ll

…

apint-ipsccp1.ll

…

apint-ipsccp2.ll

…

apint-ipsccp3.ll

…

apint-ipsccp4.ll

…

apint-load.ll

…

apint-phi.ll

…

apint-select.ll

…

atomic-load-store.ll

Add missing newline.

2011-09-01 21:20:11 +00:00

calltest.ll

…

crash.ll

…

dg.exp

…

ipsccp-addr-taken.ll

…

ipsccp-basic.ll

Make sure IPSCCP never marks a tracked call as overdefined in SCCPSolver::ResolvedUndefsIn. If we do, we can end up in a situation where a function is resolved to return a constant, but the caller is marked overdefined, which confuses the code later.

2011-09-20 23:28:51 +00:00

loadtest.ll

…

logical-nuke.ll

…

retvalue-undef.ll

…

sccptest.ll

…

select.ll

…

switch.ll

Minor bug in SCCP found by inspection. (I don't think it's possible to hit this with a normal pass pipeline, but fixing for completeness.)

2011-08-16 21:12:35 +00:00

undef-resolve.ll

Silly mistake from r137777; restore significant isStructTy() checks. While here, be a bit more defensive

2011-08-17 18:10:43 +00:00