1. Better document what is going on here.
2. Only hack on one branch per iteration, making the results less conservative.
3. Handle the problematic case by marking edges executable instead of by
playing with value lattice states. This is far less pessimistic, and fixes
SCCP/ipsccp-gvar.ll.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@31106 91177308-0d34-0410-b5e6-96231b3b80d8
Make IPSCCP strip off dead constant exprs that are using functions, making
them appear as though their address is taken. This allows us to propagate
some more pool descriptors, lowering the overhead of pool alloc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@21363 91177308-0d34-0410-b5e6-96231b3b80d8
using Function::arg_{iterator|begin|end}. Likewise Module::g* -> Module::global_*.
This patch is contributed by Gabor Greif, thanks!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@20597 91177308-0d34-0410-b5e6-96231b3b80d8
if the other side is overdefined.
This allows us to fold conditions like: if (X < Y || Y > Z) in some cases.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@18807 91177308-0d34-0410-b5e6-96231b3b80d8
1. Actually increment the Statistic for the GV elim optzn
2. When resolving undef branches, only resolve branches in executable blocks,
avoiding marking a bunch of completely dead blocks live. This has a big
impact on the quality of the generated code.
With this patch, we positively rip up vortex, compiling Ut_MoveBytes to a
single memcpy call. In vortex we get this:
12 ipsccp - Number of globals found to be constant
986 ipsccp - Number of arguments constant propagated
1378 ipsccp - Number of basic blocks unreachable
8919 ipsccp - Number of instructions removed
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@18796 91177308-0d34-0410-b5e6-96231b3b80d8
In functions where we fully constant prop the return value, replace all
ret instructions with 'ret undef'.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@18786 91177308-0d34-0410-b5e6-96231b3b80d8
This implements SCCP/ipsccp-basic.ll, rips apart Olden/mst (as described in
PR415), and does other nice things.
There is still more to come with this, but it's a start.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@18752 91177308-0d34-0410-b5e6-96231b3b80d8
1. Speedup getValueState by having it not consider Arguments. It's better
to just add them before we start SCCP'ing.
2. SCCP can delete the contents of dead blocks. No really, it's ok! This
reduces the size of the IR for subsequent passes, even though
simplifycfg would do the same job. In practice, simplifycfg does not
run until much later than sccp in gccas
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@17820 91177308-0d34-0410-b5e6-96231b3b80d8
class. The only changes are minor:
* Do not try to SCCP instructions that return void in the rewrite loop.
This is silly and fool hardy, wasting a map lookup and adding an entry
to the map which is never used.
* If we decide something has an undefined value, rewrite it to undef,
potentially leading to further simplications.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@17816 91177308-0d34-0410-b5e6-96231b3b80d8
Move include/Config and include/Support into include/llvm/Config,
include/llvm/ADT and include/llvm/Support. From here on out, all LLVM
public header files must be under include/llvm/.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@16137 91177308-0d34-0410-b5e6-96231b3b80d8
Speed up SCCP substantially by processing overdefined values quickly. This
patch speeds up SCCP by about 30-40% on large testcases.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@14861 91177308-0d34-0410-b5e6-96231b3b80d8
time from 615s to 1.49s on a large testcase that has a gigantic switch statement
that all of the blocks in the function go to (an intepreter).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@12442 91177308-0d34-0410-b5e6-96231b3b80d8
constants as being "true" when evaluating branches. This was introduced
because we now create constantexprs for the constants instead of failing the
fold.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10778 91177308-0d34-0410-b5e6-96231b3b80d8
* Implement SCCP of load instructions, implementing Transforms/SCCP/loadtest.ll
This allows us to fold expressions like "foo"[2], even if the pointer is only
a conditional constant.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10767 91177308-0d34-0410-b5e6-96231b3b80d8
to avoid reprocessing PHI nodes needlessly. This speeds up the big bad PHI
testcase 43%: from 104.9826 to 73.5157s
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@8964 91177308-0d34-0410-b5e6-96231b3b80d8
of a test that Bill Wendling sent me from 228.5s to 105s. Obviously there is
more improvement to be had, but this is a nice speedup which should be "felt"
by many programs.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@8962 91177308-0d34-0410-b5e6-96231b3b80d8
* Spell propagate right
* Improve performance of phi node handling
* Delete using directive
* Other minor changes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@5920 91177308-0d34-0410-b5e6-96231b3b80d8
* Renamed StatisticReporter.h/cpp to Statistic.h/cpp
* Broke constructor to take two const char * arguments instead of one, so
that indendation can be taken care of automatically.
* Sort the list by pass name when printing
* Make sure to print all statistics as a group, instead of randomly when
the statistics dtors are called.
* Updated ProgrammersManual with new semantics.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@4002 91177308-0d34-0410-b5e6-96231b3b80d8
- Implement SCCP of globals into ConstantExprs hopefully opening new
opportunities for more SCCP.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@3555 91177308-0d34-0410-b5e6-96231b3b80d8
* Add new RegisterOpt/RegisterAnalysis templates for registering passes that
are to show up in opt or analyze
* Register Analyses now
* Change optimizations to use RegisterOpt instead of RegisterPass
* Add support for different "PassType's"
* Add new RegisterOpt/RegisterAnalysis templates for registering passes that
are to show up in opt or analyze
* Register Analyses now
* Change optimizations to use RegisterOpt instead of RegisterPass
* Remove getPassName implementations from various subclasses
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@3113 91177308-0d34-0410-b5e6-96231b3b80d8
test/Regression/Transforms/SCCP/2002-05-02-EdgeFailure.ll
* SCCP now preserves the CFG: It leaves conditional branches the way they
are in the program, not simplifying them. A seperate pass should eliminate
the potentially dead basic blocks and edges in the CFG.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@2446 91177308-0d34-0410-b5e6-96231b3b80d8
* Change worklist to a set so that duplicates cannot happen
* Add support for the invoke instruction
* Avoid marking store and free instructions as overdefined, since they
cannot produce a value anyway.
* Inline the OperandChangedState implementation
* Add isEdgeFeasible in preparation to fix a bug. Right now it behaves
exactly as before.
* Remove obsolete comment about constant pools
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@2445 91177308-0d34-0410-b5e6-96231b3b80d8
* Do not skip the instruction immediately after a folded instruction. This
was causing the testcase failure:
test/Regression/Transforms/SCCP/2002-05-02-MissSecondInst.ll
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@2443 91177308-0d34-0410-b5e6-96231b3b80d8
- Rename runOnMethod to runOnFunction
* Transform getAnalysisUsageInfo into getAnalysisUsage
- Method is now const
- It now takes one AnalysisUsage object to fill in instead of 3 vectors
to fill in
- Pass's now specify which other passes they _preserve_ not which ones
they modify (be conservative!)
- A pass can specify that it preserves all analyses (because it never
modifies the underlying program)
* s/Method/Function/g in other random places as well
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@2333 91177308-0d34-0410-b5e6-96231b3b80d8
Method::inst_* is now in llvm/Support/InstIterator.h
GraphTraits specializations for BasicBlock and Methods are now in llvm/Support/CFG.h
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@1746 91177308-0d34-0410-b5e6-96231b3b80d8
* Renamed getOpcode to getOpcodeName
* Changed getOpcodeName to return a const char * instead of string
* Added a getOpcode method to replace getInstType
* Changed code to use getOpcode instead of getInstType
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@152 91177308-0d34-0410-b5e6-96231b3b80d8
Operands are maintained as a vector<Use> in the User class, and operator
iterators are provided as before. Getting an operand no longer requires
a virtual function call.
WARNING: getOperand(x) where x >= getNumOperands() will now assert instead
of returning null!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@149 91177308-0d34-0410-b5e6-96231b3b80d8
Optimizations now live in the 'opt' namespace
include/llvm/Opt was renamed include/llvm/Optimizations
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113 91177308-0d34-0410-b5e6-96231b3b80d8
