Dan Gohman 2199dbc5ff Don't (unconditionally) use getSCEVAtScope to simplify the step ...

expression in IVUsers, because in the case of a use of a non-linear
addrec outside of a loop, this causes the addrec to be evaluated as
a linear addrec.

llvm-svn: 73774

2009-06-19 17:33:15 +00:00

..

2005-08-15-AddRecIV.ll

…

2005-08-17-OutOfLoopVariant.ll

…

2005-09-12-UsesOutOutsideOfLoop.ll

…

2007-04-23-UseIterator.ll

…

2008-08-06-CmpStride.ll

PR2535, not PR2355.

2008-08-06 18:06:48 +00:00

2008-08-13-CmpStride.ll

Check sign to detect overflow before changing compare stride.

2008-08-13 02:05:14 +00:00

2008-08-14-ShadowIV.ll

If IV is used in a int-to-float cast inside the loop then try to eliminate the cast operation.

2008-08-26 17:57:54 +00:00

2008-09-09-Overflow.ll

fix overflow check.

2008-09-09 20:54:34 +00:00

2009-01-13-nonconstant-stride-outside-loop.ll

Fix the time regression I introduced in 464.h264ref with

2009-01-14 02:35:31 +00:00

2009-02-09-ivs-different-sizes.ll

Fix PR 3471, and some cleanups.

2009-02-09 22:14:15 +00:00

2009-04-28-no-reduce-mul.ll

Teach SCEVExpander to expand arithmetic involving pointers into GEP

2009-05-19 02:15:55 +00:00

change-compare-stride-trickiness-0.ll

Refine the change in r52258 for avoiding use-before-def conditions

2008-06-16 22:34:15 +00:00

change-compare-stride-trickiness-1.ll

Teach getZeroExtendExpr and getSignExtendExpr to use trip-count

2009-04-27 20:16:15 +00:00

change-compare-stride-trickiness-2.ll

Revert 72493 and replace it with a more conservative fix, for now: don't

2009-05-27 21:10:47 +00:00

dead-phi.ll

Improve LSR's dead-phi detection to handle use-def cycles

2008-06-22 20:44:02 +00:00

dg.exp

sabre brings to my attention that the 'tr' suffix is also obsolete

2008-05-20 21:00:03 +00:00

different-type-ivs.ll

…

dont_insert_redundant_ops.ll

…

dont_reduce_bytes.ll

…

dont_reverse.ll

Suppress the IV reversal transformation in the case that the RHS

2009-05-20 00:34:08 +00:00

dont-hoist-simple-loop-constants.ll

Update old-style syntax in some "not grep" tests.

2008-05-01 23:50:07 +00:00

exit_compare_live_range.ll

…

invariant_value_first_arg.ll

…

invariant_value_first.ll

…

nested-reduce.ll

…

ops_after_indvar.ll

…

phi_node_update_multiple_preds.ll

…

pr2537.ll

Testcase for PR2537.

2009-04-23 04:26:42 +00:00

pr2570.ll

Previously, RecursivelyDeleteDeadInstructions provided an option

2009-05-02 18:29:22 +00:00

pr3086.ll

Don't try to mix integers and pointers in an icmp instruction

2009-04-30 16:40:30 +00:00

pr3399.ll

Strengthen the "non-constant stride must dominate loop preheader" check.

2009-02-17 00:13:06 +00:00

pr3571.ll

Fix pr3571: If stride is a value defined by an instruction, make sure it dominates the loop preheader. When IV users are strength reduced, the stride is inserted into the preheader. It could create a use before def situation.

2009-02-15 06:06:15 +00:00

quadradic-exit-value.ll

Don't (unconditionally) use getSCEVAtScope to simplify the step

2009-06-19 17:33:15 +00:00

related_indvars.ll

…

remove_indvar.ll

…

share_code_in_preheader.ll

…

share_ivs.ll

…

use_postinc_value_outside_loop.ll

Implement "superhero" strength reduction, or full strength

2009-02-20 04:17:46 +00:00

var_stride_used_by_compare.ll

…

variable_stride.ll

…