Commit Graph

162 Commits

Author SHA1 Message Date
David Greene
465e2b950d Change errs() to dbgs().
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@92097 91177308-0d34-0410-b5e6-96231b3b80d8
2009-12-24 00:34:21 +00:00
Bill Wendling
fe586b3e38 Initialize uninitialized variables.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91477 91177308-0d34-0410-b5e6-96231b3b80d8
2009-12-16 00:01:27 +00:00
Bill Wendling
408e9d166a Initialize uninitialized variables.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91475 91177308-0d34-0410-b5e6-96231b3b80d8
2009-12-16 00:00:18 +00:00
Bill Wendling
43cf6c3939 Revert these. They may have been causing 483_xalancbmk to fail:
$ svn merge -c -91161 https://llvm.org/svn/llvm-project/llvm/trunk
--- Reverse-merging r91161 into '.':
U    lib/CodeGen/BranchFolding.cpp
U    lib/CodeGen/MachineBasicBlock.cpp
$ svn merge -c -91113 https://llvm.org/svn/llvm-project/llvm/trunk
--- Reverse-merging r91113 into '.':
G    lib/CodeGen/MachineBasicBlock.cpp
$ svn merge -c -91101 https://llvm.org/svn/llvm-project/llvm/trunk
--- Reverse-merging r91101 into '.':
U    include/llvm/CodeGen/MachineBasicBlock.h
G    lib/CodeGen/MachineBasicBlock.cpp
$ svn merge -c -91092 https://llvm.org/svn/llvm-project/llvm/trunk
--- Reverse-merging r91092 into '.':
G    include/llvm/CodeGen/MachineBasicBlock.h
G    lib/CodeGen/MachineBasicBlock.cpp



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91376 91177308-0d34-0410-b5e6-96231b3b80d8
2009-12-15 00:39:24 +00:00
Bill Wendling
c42a0b75a4 Don't try to move a MBB into the fall-through position if it's a landing pad or
branches only to a landing pad. Without this check, the compiler would go into
an infinite loop because the branch to a landing pad is an "abnormal" edge which
wasn't being taken into account.

This is the meat of that fix:

  if (!PrevBB.canFallThrough() && !MBB->BranchesToLandingPad(MBB)) {

The other stuff is simplification of the "branches to a landing pad" code.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91161 91177308-0d34-0410-b5e6-96231b3b80d8
2009-12-11 21:47:36 +00:00
Dan Gohman
864e2efce2 Remove the target hook TargetInstrInfo::BlockHasNoFallThrough in favor of
MachineBasicBlock::canFallThrough(), which is target-independent and more
thorough.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@90634 91177308-0d34-0410-b5e6-96231b3b80d8
2009-12-05 00:44:40 +00:00
Chris Lattner
7896c9f436 improve portability to avoid conflicting with std::next in c++'0x.
Patch by Howard Hinnant!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@90365 91177308-0d34-0410-b5e6-96231b3b80d8
2009-12-03 00:50:42 +00:00
Bob Wilson
15acadde5f Split tail duplication into a separate pass. This is needed to avoid
running tail duplication when doing branch folding for if-conversion, and
we also want to be able to run tail duplication earlier to fix some
reg alloc problems.  Move the CanFallThrough function from BranchFolding
to MachineBasicBlock so that it can be shared by TailDuplication.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89904 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-26 00:32:21 +00:00
Bob Wilson
f8c4cfb7cc Refactor target hook for tail duplication as requested by Chris.
Make tail duplication of indirect branches much more aggressive (for targets
that indicate that it is profitable), based on further experience with
this transformation.  I compiled 3 large applications with and without
this more aggressive tail duplication and measured minimal changes in code
size.  ("size" on Darwin seems to round the text size up to the nearest
page boundary, so I can only say that any code size increase was less than
one 4k page.) Radar 7421267.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89814 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-24 23:35:49 +00:00
Bob Wilson
fbb6b67df0 There should be no need to keep renumbering blocks during tail duplication.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89275 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-18 23:48:57 +00:00
Bob Wilson
60f34b9fb0 Tail duplication still needs to iterate. Duplicating new instructions onto
the tail of a block may make that block a new candidate for duplication.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89264 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-18 22:52:37 +00:00
Bob Wilson
79d621035e Add another statistic to measure code size due to tail duplication.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89254 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-18 22:12:31 +00:00
Bob Wilson
7cd5d3e05c Add statistics for tail duplication.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89225 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-18 19:29:37 +00:00
Bob Wilson
834b08af8d Add a target hook to allow changing the tail duplication limit based on the
contents of the block to be duplicated.  Use this for ARM Cortex A8/9 to
be more aggressive tail duplicating indirect branches, since it makes it
much more likely that they will be predicted in the branch target buffer.
Testcase coming soon.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89187 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-18 03:34:27 +00:00
Bob Wilson
aa60ddac39 Remove a special case for tail merging that seems to be both broken and
unnecessary.  It is broken because the "isIdenticalTo" check should be
negated.  If that is fixed, this code causes the CodeGen/X86/tail-opts.ll
test to fail, in the dont_merge_oddly function.  And, I confirmed that the
regression is real -- the generated code is worse.  As far as I can tell,
that tail-opts.ll test is checking for what this code is supposed to handle
and we're doing the right thing anyway.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89121 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-17 18:30:09 +00:00
Dan Gohman
37d60fd142 Set MadeChange instead of MadeChangeThisIteration.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89114 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-17 18:04:15 +00:00
Bob Wilson
465c8252f2 Update a comment, now that tail duplication happens after other branch
folding optimizations.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89109 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-17 17:40:31 +00:00
Bob Wilson
56ea69c3a2 Perform tail duplication only once, after tail merging is complete.
It was too difficult to keep the heuristics for merging and duplication
consistent.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89105 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-17 17:06:18 +00:00
Bob Wilson
4a504c6ff8 Fix a comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@88940 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-16 19:33:27 +00:00
Bob Wilson
c4a0c843e4 Fix some comments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@88932 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-16 18:54:08 +00:00
Bob Wilson
d34f5d91bc Whitespace: be consistent with pointer syntax.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@88929 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-16 18:08:46 +00:00
Bob Wilson
3cbc312087 Clean up whitespace.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@88927 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-16 17:56:13 +00:00
Dan Gohman
c4c550c758 When optimizing for size, don't tail-merge unless it's likely to be a
code-size win, and not when it's only likely to be code-size neutral,
such as when only a single instruction would be eliminated and a new
branch would be required.

This fixes rdar://7392894.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@88692 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-13 21:02:15 +00:00
Dan Gohman
72b2990a74 Make the BranchFolderPass class local to BranchFolding.cpp.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86928 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-12 01:59:26 +00:00
Dan Gohman
8520149db1 Minor code cleanups.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86926 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-12 01:51:28 +00:00
Dan Gohman
ad6af45dc1 Tail merge at any size when there are two potentials blocks and one
can be made to fall through into the other.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86909 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-12 00:39:10 +00:00
Dan Gohman
ffe644ebf4 Promote MergePotentialsElt and SameTailElt to be regular classes
instead of typedefs for std::pair. This simplifies the type of
SameTails, which previously was std::vector<std::pair<std::vector<std::pair<unsigned, MachineBasicBlock *> >::iterator, MachineBasicBlock::iterator>


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86885 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-11 21:57:02 +00:00
Dan Gohman
7cc253e3b8 Revert this line of 86871.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86875 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-11 19:56:05 +00:00
Dan Gohman
2210c0bea8 Add support for tail duplication to BranchFolding, and extend
tail merging support to handle more cases.
 - Recognize several cases where tail merging is beneficial even when
   the tail size is smaller than the generic threshold.
 - Make use of MachineInstrDesc::isBarrier to help detect
   non-fallthrough blocks.
 - Check for and avoid disrupting fall-through edges in more cases.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86871 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-11 19:48:59 +00:00
Dan Gohman
cdc06ba5df Fix indentation level.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86856 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-11 18:42:28 +00:00
Dan Gohman
4e3f125e18 Whitespace cleanups.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86855 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-11 18:38:14 +00:00
Dan Gohman
a127edceae Prefix MBB numbers with "BB#" in debug output to make it clear what
the numbers mean.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86854 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-11 18:23:17 +00:00
Dan Gohman
d194498d26 Minor code simplification.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86853 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-11 18:18:34 +00:00
Dan Gohman
7071528e4e Remove an unused variable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86642 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-10 01:37:57 +00:00
Dan Gohman
b2e40990c4 Minor code simplification.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86641 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-10 01:36:20 +00:00
Bob Wilson
888acc35a3 Fix branch folding bug for indirect branches: for a block containing only
an unconditional branch (possibly from tail merging), this code is
trying to redirect all of its predecessors to go directly to the branch
target, but that isn't feasible for indirect branches.  The other
predecessors (that don't end with indirect branches) could theoretically
still be handled, but that is not easily done right now.

The AnalyzeBranch interface doesn't currently let us distinguish jump table
branches from indirect branches, and this code is currently handling
jump tables.  To avoid punting on address-taken blocks, we would have to give
up handling jump tables.  That seems like a bad tradeoff.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85975 91177308-0d34-0410-b5e6-96231b3b80d8
2009-11-03 23:44:31 +00:00
Dan Gohman
ab918103d2 Don't delete blocks which have their address taken.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85572 91177308-0d34-0410-b5e6-96231b3b80d8
2009-10-30 02:13:27 +00:00
Bob Wilson
7b888b8ad0 Refactor complicated predicate into a separate function.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85519 91177308-0d34-0410-b5e6-96231b3b80d8
2009-10-29 18:40:06 +00:00
Bob Wilson
2c04dae715 Reimplement BranchFolding change to avoid tail merging for a 1 instruction
common tail, except when the OptimizeForSize function attribute is present.
Radar 7338114.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85441 91177308-0d34-0410-b5e6-96231b3b80d8
2009-10-28 22:10:20 +00:00
Bob Wilson
a597103c32 Revert r85346 change to control tail merging by CodeGenOpt::Level.
I'm going to redo this using the OptimizeForSize function attribute.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85426 91177308-0d34-0410-b5e6-96231b3b80d8
2009-10-28 20:46:46 +00:00
Bob Wilson
cd4f04d6bc Record CodeGen optimization level in the BranchFolding pass so that we can
use it to control tail merging when there is a tradeoff between performance
and code size.  When there is only 1 instruction in the common tail, we have
been merging.  That can be good for code size but is a definite loss for
performance.  Now we will avoid tail merging in that case when the
optimization level is "Aggressive", i.e., "-O3".  Radar 7338114.

Since the IfConversion pass invokes BranchFolding, it too needs to know
the optimization level.  Note that I removed the RegisterPass instantiation
for IfConversion because it required a default constructor.  If someone
wants to keep that for some reason, we can add a default constructor with
a hard-wired optimization level.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85346 91177308-0d34-0410-b5e6-96231b3b80d8
2009-10-27 23:49:38 +00:00
Dan Gohman
6d31268a7d Revert the main portion of r31856. It was causing BranchFolding
to break up CFG diamonds by banishing one of the blocks to the end of
the function, which is bad for code density and branch size.

This does pessimize MultiSource/Benchmarks/Ptrdist/yacr2, the
benchmark cited as the reason for the change, however I've examined
the code and it looks more like a case of gaming a particular
branch than of being generally applicable.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84803 91177308-0d34-0410-b5e6-96231b3b80d8
2009-10-22 00:03:58 +00:00
Evan Cheng
030a0a0cdb Run branch folding if if-converter make some transformations.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80994 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-04 07:47:40 +00:00
Evan Cheng
b3c2742896 Funky indentation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80971 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-03 23:54:22 +00:00
Bill Wendling
3403bcd8f9 Convert DOUT to DEBUG(errs()...).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79747 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-22 20:03:00 +00:00
Dan Gohman
da65822cfc Make tail merging handle blocks with repeated predecessors correctly, and
remove RemoveDuplicateSuccessor, as it is no longer necessary, and because
it breaks assumptions made in
MachineBasicBlock::isOnlyReachableByFallthrough.

Convert test/CodeGen/X86/omit-label.ll to FileCheck and add a testcase
for PR4732.

test/CodeGen/Thumb2/thumb2-ifcvt2.ll sees a diff with this commit due to
it being bugpoint-reduced to the point where it doesn't matter what the
condition for the branch is.

Add some more interesting code to
test/CodeGen/X86/2009-08-06-branchfolder-crash.ll, which is the testcase
that originally motivated the RemoveDuplicateSuccessor code, to help
verify that the original problem isn't being re-broken.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79338 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-18 15:18:18 +00:00
Dale Johannesen
1c5a28706d Rewrite previous patch to follow Chris' stylistic
preference; no functional change.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78391 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-07 17:41:29 +00:00
Dale Johannesen
14e545d18e Fix PR 4626, a crash in branch folding after OptimizeBlock
produced a CFG it wasn't prepared for.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78351 91177308-0d34-0410-b5e6-96231b3b80d8
2009-08-06 22:56:40 +00:00
Torok Edwin
c23197a26f llvm_unreachable->llvm_unreachable(0), LLVM_UNREACHABLE->llvm_unreachable.
This adds location info for all llvm_unreachable calls (which is a macro now) in
!NDEBUG builds.
In NDEBUG builds location info and the message is off (it only prints
"UREACHABLE executed").


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@75640 91177308-0d34-0410-b5e6-96231b3b80d8
2009-07-14 16:55:14 +00:00
Torok Edwin
c25e7581b9 assert(0) -> LLVM_UNREACHABLE.
Make llvm_unreachable take an optional string, thus moving the cerr<< out of
line.
LLVM_UNREACHABLE is now a simple wrapper that makes the message go away for
NDEBUG builds.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@75379 91177308-0d34-0410-b5e6-96231b3b80d8
2009-07-11 20:10:48 +00:00