Commit Graph

42 Commits

Author SHA1 Message Date
Dan Gohman
e521ccb5bd Add a cast to void to show that the return value is being
intentionally ignored.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100984 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-11 19:30:19 +00:00
Dan Gohman
87527c594f Delete a dead check.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100983 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-11 19:29:41 +00:00
Dan Gohman
86eeeaf877 Delete dead code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100981 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-11 19:28:47 +00:00
Dan Gohman
b3cdb0ec31 Fix a bug in IVUsers which was permitting non-affine addrecs to
be sent to LSR, which it isn't prepared to handle.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100839 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-09 01:22:56 +00:00
Dan Gohman
448db1cdef Generalize IVUsers to track arbitrary expressions rather than expressions
explicitly split into stride-and-offset pairs. Also, add the
ability to track multiple post-increment loops on the same expression.

This refines the concept of "normalizing" SCEV expressions used for
to post-increment uses, and introduces a dedicated utility routine for
normalizing and denormalizing expressions.

This fixes the expansion of expressions which are post-increment users
of more than one loop at a time. More broadly, this takes LSR another
step closer to being able to reason about more than one loop at a time.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100699 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-07 22:27:08 +00:00
Dan Gohman
3d32386992 Add a const qualifier.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100515 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-06 01:31:12 +00:00
Dan Gohman
3f46a3abee Spelling fixes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97453 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-01 17:49:51 +00:00
Dan Gohman
454d26dc43 Remove unused variables and parameters.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96780 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-22 04:11:59 +00:00
Dan Gohman
0402577f38 Simplify this code; no need for a custom subclass if it doesn't need
to override anything from the parent class.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96150 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-14 02:48:58 +00:00
Dan Gohman
572645cf84 Reapply the new LoopStrengthReduction code, with compile time and
bug fixes, and with improved heuristics for analyzing foreign-loop
addrecs.

This change also flattens IVUsers, eliminating the stride-oriented
groupings, which makes it easier to work with.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95975 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-12 10:34:29 +00:00
Dan Gohman
4207d6af5e Use an AssemblyAnnotatorWriter to clean up IVUsers' debug output.
The "uses=" comments are just clutter in this context.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95799 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-10 20:42:37 +00:00
Bill Wendling
f7d848349e Add "dump" method to IVUsersOneStride.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95022 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-01 22:51:23 +00:00
Dan Gohman
a10756ee65 Re-implement the main strength-reduction portion of LoopStrengthReduction.
This new version is much more aggressive about doing "full" reduction in
cases where it reduces register pressure, and also more aggressive about
rewriting induction variables to count down (or up) to zero when doing so
reduces register pressure.

It currently uses fairly simplistic algorithms for finding reuse
opportunities, but it introduces a new framework allows it to combine
multiple strategies at once to form hybrid solutions, instead of doing
all full-reduction or all base+index.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@94061 91177308-0d34-0410-b5e6-96231b3b80d8
2010-01-21 02:09:26 +00:00
Dan Gohman
bafbbdde38 Add a new helper function to IVUsers for returning the "canonical"
form of an expression. This is the expression without the
post-increment adjustment made, which is useful in determining
which registers will be used by the expansion.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93921 91177308-0d34-0410-b5e6-96231b3b80d8
2010-01-19 21:55:32 +00:00
Dan Gohman
3073329c53 Use WriteAsOperand instead of getName() to print loop header names,
so that unnamed blocks are handled.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@93059 91177308-0d34-0410-b5e6-96231b3b80d8
2010-01-09 18:17:45 +00:00
David Greene
63c4560526 Convert debug messages to use dbgs(). Generally this means
s/errs/dbgs/g except for certain special cases.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@92037 91177308-0d34-0410-b5e6-96231b3b80d8
2009-12-23 20:20:46 +00:00
Dan Gohman
92329c7fbe Add Loop contains utility methods for testing whether a loop
contains another loop, or an instruction. The loop form is
substantially more efficient on large loops than the typical
code it replaces.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91654 91177308-0d34-0410-b5e6-96231b3b80d8
2009-12-18 01:24:09 +00:00
Dan Gohman
6bec5bb344 Reapply LoopStrengthReduce and IVUsers cleanups, excluding the part
of 91296 that caused trouble -- the Processed list needs to be
preserved for the livetime of the pass, as AddUsersIfInteresting
is called from other passes.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91641 91177308-0d34-0410-b5e6-96231b3b80d8
2009-12-18 00:06:20 +00:00
Evan Cheng
04149f7ffd Revert 91280-91283, 91286-91289, 91291, 91293, 91295-91296. It apparently introduced a non-deterministic behavior in the optimizer somewhere.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91598 91177308-0d34-0410-b5e6-96231b3b80d8
2009-12-17 09:39:49 +00:00
Dan Gohman
e2f65ee4b1 Clear the Processed set when it is no longer used, and clear the
IVUses list in releaseMemory().


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91296 91177308-0d34-0410-b5e6-96231b3b80d8
2009-12-14 17:35:17 +00:00
Dan Gohman
30844c3950 Fix a thinko; isNotAlreadyContainedIn had a built-in negative, so the
condition was inverted when the code was converted to contains().


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91295 91177308-0d34-0410-b5e6-96231b3b80d8
2009-12-14 17:31:01 +00:00
Dan Gohman
440b40d865 Drop Loop::isNotAlreadyContainedIn in favor of Loop::contains. The
former was just exposing a LoopInfoBase implementation detail.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91286 91177308-0d34-0410-b5e6-96231b3b80d8
2009-12-14 17:06:50 +00:00
Jim Grosbach
dbc3577ea1 enable iv-users simplification by default
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89713 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-23 23:25:54 +00:00
Jim Grosbach
97200e4dd4 Teach IVUsers to keep things simpler and track loop-invariant strides only
for uses inside the loop. This works better with LSR. Disabled behind
-simplify-iv-users while benchmarking.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89299 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-19 02:05:44 +00:00
Evan Cheng
586f69a118 - Teach LSR to avoid changing cmp iv stride if it will create an immediate that
cannot be folded into target cmp instruction.
- Avoid a phase ordering issue where early cmp optimization would prevent the
  later count-to-zero optimization.
- Add missing checks which could cause LSR to reuse stride that does not have
  users.
- Fix a bug in count-to-zero optimization code which failed to find the pre-inc
  iv's phi node.
- Remove, tighten, loosen some incorrect checks disable valid transformations.
- Quite a bit of code clean up.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86969 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-12 07:35:05 +00:00
Dan Gohman
a6a4aaefee Fix IVUsers to avoid assuming that the loop has a unique backedge.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86161 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-05 19:41:37 +00:00
Dan Gohman
00cb67359f Instead of testing whether an instruction dominates the loop preheader,
test whether it properly dominates the loop header. This is equivalent
when the loop has a preheader, and has the advantage of working when
the loop doesn't have a preheader. Since IVUsers doesn't Require
LoopSimplify, the loop isn't guaranteed to have a preheader.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82899 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-27 15:30:00 +00:00
Dan Gohman
59faf04a5d Remove a redundant #include.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82897 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-27 15:21:52 +00:00
Chris Lattner
9661c13c37 remove a few dead insertion methods.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79882 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-24 02:39:26 +00:00
Chris Lattner
bbbfa99d3d remove a few DOUTs here and there.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79832 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-23 06:35:02 +00:00
Chris Lattner
bdff548e4d eliminate the "Value" printing methods that print to a std::ostream.
This required converting a bunch of stuff off DOUT and other cleanups.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79819 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-23 04:37:46 +00:00
Dan Gohman
094ae454e8 Fix more missing newlines.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79244 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-17 15:29:31 +00:00
Daniel Dunbar
ce63ffb52f More migration to raw_ostream, the water has dried up around the iostream hole.
- Some clients which used DOUT have moved to DEBUG. We are deprecating the
   "magic" DOUT behavior which avoided calling printing functions when the
   statement was disabled. In addition to being unnecessary magic, it had the
   downside of leaving code in -Asserts builds, and of hiding potentially
   unnecessary computations.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@77019 91177308-0d34-0410-b5e6-96231b3b80d8
2009-07-25 00:23:56 +00:00
Dan Gohman
d9ef1a860e Add a newline, now that Value's operator<< doesn't print one.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@75568 91177308-0d34-0410-b5e6-96231b3b80d8
2009-07-14 00:32:49 +00:00
Dan Gohman
c8d76d5afb Make Loop and MachineLoop be subclasses of LoopBase, rather than typedefs,
using the Curiously Recurring Template Pattern with LoopBase.

This will help further refactoring, and future functionality for
Loop. Also, Headers can now foward-declare Loop, instead of pulling
in LoopInfo.h or doing tricks.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@75519 91177308-0d34-0410-b5e6-96231b3b80d8
2009-07-13 21:51:15 +00:00
Dan Gohman
0bba49cebc Change all SCEV* to SCEV *.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@74918 91177308-0d34-0410-b5e6-96231b3b80d8
2009-07-07 17:06:11 +00:00
Owen Anderson
372b46cad9 SCEVHandle is no more!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73906 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-22 21:39:50 +00:00
Dan Gohman
958e292c7a 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.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73774 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-19 17:33:15 +00:00
Dan Gohman
4e8a98519e Remove the code from IVUsers that attempted to handle
casted induction variables in cases where the cast
isn't foldable. It ended up being a pessimization in
many cases. This could be fixed, but it would require
a bunch of complicated code in IVUsers' clients. The
advantages of this approach aren't visible enough to
justify it at this time.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73706 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-18 16:54:06 +00:00
Dan Gohman
743e41eae1 Code cleanups. getSCEVAtScope no longer uses SCEVCouldNotCompute.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73401 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-15 18:38:59 +00:00
Duncan Sands
18395d2c3d Avoid getting a compiler warning
IVUsers.cpp: In member function ‘bool llvm::IVUsers::AddUsersIfInteresting(llvm::Instruction*)’:
 IVUsers.cpp:221: warning: ‘isSigned’ may be used uninitialized in this function
with gcc-4.3.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@71654 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-13 12:52:44 +00:00
Dan Gohman
81db61a2e6 Factor the code for collecting IV users out of LSR into an IVUsers class,
and generalize it so that it can be used by IndVarSimplify. Implement the
base IndVarSimplify transformation code using IVUsers. This removes
TestOrigIVForWrap and associated code, as ScalarEvolution now has enough
builtin overflow detection and folding logic to handle all the same cases,
and more. Run "opt -iv-users -analyze -disable-output" on your favorite
loop for an example of what IVUsers does.

This lets IndVarSimplify eliminate IV casts and compute trip counts in
more cases. Also, this happens to finally fix the remaining testcases
in PR1301.

Now that IndVarSimplify is being more aggressive, it occasionally runs
into the problem where ScalarEvolutionExpander's code for avoiding
duplicate expansions makes it difficult to ensure that all expanded
instructions dominate all the instructions that will use them. As a
temporary measure, IndVarSimplify now uses a FixUsesBeforeDefs function
to fix up instructions inserted by SCEVExpander. Fortunately, this code
is contained, and can be easily removed once a more comprehensive
solution is available.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@71535 91177308-0d34-0410-b5e6-96231b3b80d8
2009-05-12 02:17:14 +00:00