Commit Graph

123 Commits

Author SHA1 Message Date
Chris Lattner
76c1b97e40 Merge DenseMapKeyInfo & DenseMapValueInfo into DenseMapInfo
Add a new DenseMapInfo::isEqual method to allow clients to redefine
the equality predicate used when probing the hash table.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@42042 91177308-0d34-0410-b5e6-96231b3b80d8
2007-09-17 18:34:04 +00:00
Anton Korobeynikov
9f528e6280 Don't promote volatile loads/stores. This is needed (for example) to handle setjmp/longjmp properly.
This fixes PR1520.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@41461 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-26 21:43:30 +00:00
Chris Lattner
f12f8def39 rewrite the code used to construct pruned SSA form with the IDF method.
In the old way, we computed and inserted phi nodes for the whole IDF of 
the definitions of the alloca, then computed which ones were dead and
removed them.

In the new method, we first compute the region where the value is live,
and use that information to only insert phi nodes that are live.  This
eliminates the need to compute liveness later, and stops the algorithm
from inserting a bunch of phis which it then later removes.

This speeds up the testcase in PR1432 from 2.00s to 0.15s (14x) in a
release build and 6.84s->0.50s (14x) in a debug build.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40825 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 22:50:14 +00:00
Chris Lattner
0ec8df3d5f Factor out a whole bunch of code into it's own method.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40824 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 21:14:29 +00:00
Chris Lattner
127ed3c929 Use getNumPreds(BB) instead of computing them manually. This is a very small but
measurable speedup.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40823 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 21:06:15 +00:00
Chris Lattner
1e76af3011 Change the rename pass to be "tail recursive", only adding N-1 successors
to the worklist, and handling the last one with a 'tail call'.  This speeds
up PR1432 from 2.0578s to 2.0012s (2.8%)


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40822 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 20:40:27 +00:00
Chris Lattner
e7b653dabe cache computation of #preds for a BB. This speeds up
mem2reg from 2.0742->2.0522s on PR1432.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40821 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 20:24:50 +00:00
Chris Lattner
384c7e0436 reserve operand space for phi nodes when we insert them.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40820 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 20:14:34 +00:00
Chris Lattner
6c81213b4c use continue to avoid nesting, no functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40819 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 20:07:06 +00:00
Chris Lattner
fb312c7e44 Promoting allocas with the 'single store' fastpath is
faster than with the 'local to a block' fastpath.  This speeds
up PR1432 from 2.1232 to 2.0686s (2.6%)



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40818 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 20:03:23 +00:00
Chris Lattner
7a5745b38c When PromoteLocallyUsedAllocas promoted allocas, it didn't remember
to increment NumLocalPromoted, and didn't actually delete the
dead alloca, leading to an extra iteration of mem2reg.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40817 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 20:01:43 +00:00
Chris Lattner
59a2837715 std::map -> DenseMap
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40816 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 19:52:20 +00:00
Chris Lattner
c69e491575 fix a logic bug where we wouldn't promote single store allocas if the
stored value was a non-instruction value.  Doh.

This increase the # single store allocas from 8982 to 9026, and
speeds up mem2reg on the testcase in PR1432 from 2.17 to 2.13s.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40813 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 02:45:02 +00:00
Chris Lattner
4f63e76cda When we do the single-store optimization, delete both the store
and the alloca so they don't get reprocessed.

This speeds up PR1432 from 2.20s to 2.17s.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40812 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 02:38:38 +00:00
Chris Lattner
d0458e5d4f Three improvements:
1. Check for revisiting a block before checking domination, which is faster.
  2. If the stored value isn't an instruction, we don't have to check for domination.
  3. If we have a value used in the same block more than once, make sure to remove the
     block from the UsingBlocks vector.  Not doing so forces us to go through the slow
     path for the alloca.

The combination of these improvements increases the number of allocas on the fastpath
from 8935 to 8982 on PR1432.  This speeds it up from 2.90s to 2.20s (31%)



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40811 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 02:32:22 +00:00
Chris Lattner
c3f6ea8330 switch from using a std::set to using a SmallPtrSet. This speeds up the
testcase in PR1432 from 6.33s to 2.90s (2.22x)


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40810 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 02:21:22 +00:00
Chris Lattner
b776a335b1 In mem2reg, when handling the single-store case, make sure to remove
a using block from the list if we handle it.  Not doing this caused us
to not be able to promote (with the fast path) allocas which have uses (whoops).

This increases the # allocas hitting this fastpath from 4042 to 8935 on the
testcase in PR1432, speeding up mem2reg by 2.6x



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40809 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 02:15:24 +00:00
Chris Lattner
5dd75b4ca7 split rewriting of single-store allocas into its own
method.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40806 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 01:47:41 +00:00
Chris Lattner
bbe104002f refactor some code to shrink PromoteMem2Reg::run a bit
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40805 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 01:41:18 +00:00
Chris Lattner
483ce14bf4 add a typedef, no other change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40804 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 01:19:38 +00:00
Chris Lattner
63cdcaa3d6 avoid an unneeded vector copy. This speeds up mem2reg on the testcase
in PR1432 by 6%


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40803 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 01:07:49 +00:00
Chris Lattner
ac4aa4be9b make RenamePassWorkList a local var instead of an ivar.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40802 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-04 01:04:40 +00:00
Devang Patel
326821ef12 Use DominatorTree instead of ETForest.
This allows faster immediate domiantor walk.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37500 91177308-0d34-0410-b5e6-96231b3b80d8
2007-06-07 21:57:03 +00:00
Devang Patel
68c01b3cf3 Mem2Reg does not need TargetData.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36444 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-25 18:32:35 +00:00
Devang Patel
41968df51e Remove unused function argument.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36441 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-25 17:15:20 +00:00
Owen Anderson
cdacc498e4 Fix a comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36299 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-21 07:12:44 +00:00
Owen Anderson
df07335b46 Move more passes to using ETForest instead of DominatorTree.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36271 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-20 06:27:13 +00:00
Devang Patel
d64d3a1d28 Reduce malloc/free traffic.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35370 91177308-0d34-0410-b5e6-96231b3b80d8
2007-03-26 23:19:29 +00:00
Devang Patel
f7543fe35a Remove dead comments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35053 91177308-0d34-0410-b5e6-96231b3b80d8
2007-03-09 23:41:03 +00:00
Devang Patel
a5b7dc5ef8 Avoid recursion. Use iterative algorithm for RenamePass().
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35052 91177308-0d34-0410-b5e6-96231b3b80d8
2007-03-09 23:39:14 +00:00
Chris Lattner
dfb22c3521 redesign the primary datastructure used by mem2reg to eliminate an
std::map of std::vector's (ouch!).  This speeds up mem2reg by 10% on 176.gcc.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@33974 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-07 01:15:04 +00:00
Chris Lattner
d3874049a5 With the last change, we no longer need both directions of mapping from
BBNumbers.  Instead of using a bi-directional mapping, just use a single
densemap.  This speeds up mem2reg on 176.gcc by 8%, from  1.3489 to
1.2485s.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@33940 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-05 23:37:20 +00:00
Chris Lattner
a0d1548d0d Simplify use of DFBlocks, this makes no noticable performance difference,
but paves the way to eliminate BBNumbers.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@33938 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-05 23:31:26 +00:00
Chris Lattner
6a1a28dedd Switch InsertedPHINodes back to SmallPtrSet now that the SmallPtrSet::erase
bug is fixed.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@33932 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-05 23:11:37 +00:00
Chris Lattner
fc7d13d898 switch a SmallPtrSet back to an std::set for now, this caused problems.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@33930 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-05 22:28:52 +00:00
Chris Lattner
c670f3da72 switch an std::set over to a SmallPtrSet, speeding up mem2reg 6% on 176.gcc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@33929 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-05 22:15:21 +00:00
Chris Lattner
c837615cf0 switch an std::set over to SmallPtrSet, speeding up mem2reg 3.4% on 176.gcc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@33928 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-05 22:13:11 +00:00
Chris Lattner
40b6555561 eliminate some malloc traffic, this speeds up mem2reg by 3.4%.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@33927 91177308-0d34-0410-b5e6-96231b3b80d8
2007-02-05 21:58:48 +00:00
Chris Lattner
a4f0b3a084 s|llvm/Support/Visibility.h|llvm/Support/Compiler.h|
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29911 91177308-0d34-0410-b5e6-96231b3b80d8
2006-08-27 12:54:02 +00:00
Chris Lattner
9525528a7d Use hidden visibility to make symbols in an anonymous namespace get
dropped.  This shrinks libllvmgcc.dylib another 67K


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28975 91177308-0d34-0410-b5e6-96231b3b80d8
2006-06-28 23:17:24 +00:00
Chris Lattner
634c76c08c Fix some nondeterminstic behavior in the mem2reg pass that (in addition to
nondeterminism being bad) could cause some trivial missed optimizations (dead
phi nodes being left around for later passes to clean up).

With this, llvm-gcc4 now bootstraps and correctly compares.  I don't know
why I never tried to do it before... :)


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27984 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-27 01:14:43 +00:00
Chris Lattner
36ba5006df Implement a refinement to the mem2reg algorithm for cases where an alloca
has a single def.  In this case, look for uses that are dominated by the def
and attempt to rewrite them to directly use the stored value.

This speeds up mem2reg on these values and reduces the number of phi nodes
inserted.  This should address PR665.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24411 91177308-0d34-0410-b5e6-96231b3b80d8
2005-11-18 07:31:42 +00:00
Chris Lattner
fed40df846 This needs proper dominance
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24410 91177308-0d34-0410-b5e6-96231b3b80d8
2005-11-18 07:29:44 +00:00
Chris Lattner
5e1b231921 This code can handle non-dominating instructions
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22667 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-05 00:57:45 +00:00
Nate Begeman
a83ba0f5c9 Fix a fixme in CondPropagate.cpp by moving a PhiNode optimization into
BasicBlock's removePredecessor routine.  This requires shuffling around
the definition and implementation of hasContantValue from Utils.h,cpp into
Instructions.h,cpp


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22664 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-04 23:24:19 +00:00
Jeff Cohen
00b16889ab Eliminate all remaining tabs and trailing spaces.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22523 91177308-0d34-0410-b5e6-96231b3b80d8
2005-07-27 06:12:32 +00:00
Chris Lattner
6cfd1ebcd3 Fix PR590 and Transforms/Mem2Reg/2005-06-30-ReadBeforeWrite.ll.
The optimization for locally used allocas was not safe for allocas that
were read before they were written.  This change disables that optimization
in that case.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22318 91177308-0d34-0410-b5e6-96231b3b80d8
2005-06-30 07:29:44 +00:00
Misha Brukman
fd93908ae8 Remove trailing whitespace
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21427 91177308-0d34-0410-b5e6-96231b3b80d8
2005-04-21 23:48:37 +00:00
Chris Lattner
5a85d9c664 Fix test/Regression/Transforms/LICM/2004-09-14-AliasAnalysisInvalidate.llx
This only fails on darwin or on X86 under valgrind.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@18377 91177308-0d34-0410-b5e6-96231b3b80d8
2004-11-30 07:01:15 +00:00
Reid Spencer
17e6e44298 Correction to allow compilation with Visual C++.
Patch contributed by Morten Ofstad. Thanks Morten!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@17123 91177308-0d34-0410-b5e6-96231b3b80d8
2004-10-18 14:38:48 +00:00