Chris Lattner [Wed, 5 Jan 2011 01:03:32 +0000 (01:03 +0000)]
Fix PR8906: -fno-builtin should disable loop-idiom recognition.
It forms memset and memcpy's, and will someday form popcount and
other stuff. All of this is bad when compiling the implementation
of memset, memcpy, popcount, etc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122854
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Tue, 4 Jan 2011 22:31:50 +0000 (22:31 +0000)]
Remove TODO, these appear to be implemented.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122849
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Tue, 4 Jan 2011 22:15:21 +0000 (22:15 +0000)]
Don't bother value numbering instructions with void types in GVN. In theory this should allow us to insert
fewer things into the value numbering maps, but any speedup is beneath the noise threshold on my machine
on 403.gcc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122844
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Tue, 4 Jan 2011 21:10:11 +0000 (21:10 +0000)]
Use the EdgeBundles analysis in X86FloatingPoint instead of recomputing CFG
bundles in the pass.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122833
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Tue, 4 Jan 2011 21:10:05 +0000 (21:10 +0000)]
Turn the EdgeBundles class into a stand-alone machine CFG analysis pass.
The analysis will be needed by both the greedy register allocator and the
X86FloatingPoint pass. It only needs to be computed once when the CFG doesn't
change.
This pass is very fast, usually showing up as 0.0% wall time.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122832
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Tue, 4 Jan 2011 19:31:24 +0000 (19:31 +0000)]
Eliminate a warning compiling with llvm-gcc. (IMO the
warning is overzealous but gcc is what it is.)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122829
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Tue, 4 Jan 2011 19:29:46 +0000 (19:29 +0000)]
Complete the NumberTable --> LeaderTable rename.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122828
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Tue, 4 Jan 2011 19:25:18 +0000 (19:25 +0000)]
Fix typo in a comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122827
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Tue, 4 Jan 2011 19:24:57 +0000 (19:24 +0000)]
Prune #include's.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122826
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Tue, 4 Jan 2011 19:13:25 +0000 (19:13 +0000)]
Clarify terminology, settling on referring to what was the "number table" as the "leader table", and
rename methods to make it much more clear what they're doing.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122823
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Tue, 4 Jan 2011 19:10:54 +0000 (19:10 +0000)]
When removing a value from GVN's leaders list, don't drop the Next pointer in a corner case.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122822
91177308-0d34-0410-b5e6-
96231b3b80d8
Dale Johannesen [Tue, 4 Jan 2011 19:01:54 +0000 (19:01 +0000)]
Improve the accuracy of the inlining heuristic looking for the
case where a static caller is itself inlined everywhere else, and
thus may go away if it doesn't get too big due to inlining other
things into it. If there are references to the caller other than
calls, it will not be removed; account for this.
This results in same-day completion of the case in PR8853.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122821
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Tue, 4 Jan 2011 18:54:18 +0000 (18:54 +0000)]
Branch instructions don't produce values, so there's no need to generate a value number for them. This
avoids adding them to the various value numbering tables, resulting in a minor (~3%) speedup for GVN
on 40.gcc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122819
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Tue, 4 Jan 2011 18:22:08 +0000 (18:22 +0000)]
Remove commented out code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122817
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Tue, 4 Jan 2011 18:21:18 +0000 (18:21 +0000)]
Give MachineFunctionAnalysis a getPassName() implementation to make timing reports prettier.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122816
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Tue, 4 Jan 2011 18:19:19 +0000 (18:19 +0000)]
Switch to the new style of asterisk placement.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122815
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 4 Jan 2011 18:19:15 +0000 (18:19 +0000)]
fix an off-by-one bug that caused a crash analyzing
ashr's with huge shift amounts, PR8896
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122814
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael J. Spencer [Tue, 4 Jan 2011 17:00:18 +0000 (17:00 +0000)]
UnitTests/Path: Produce useful diagnostics on error.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122812
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Tue, 4 Jan 2011 16:24:51 +0000 (16:24 +0000)]
Switch to path halving from path compression for a small speedup. This also
makes getLeader() nonrecursive.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122811
91177308-0d34-0410-b5e6-
96231b3b80d8
Tobias Grosser [Tue, 4 Jan 2011 16:01:17 +0000 (16:01 +0000)]
Include llvm-gcc dir before llvm_tools_dir
This ensures that always the recently compiled tools are picked for testing.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122810
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Tue, 4 Jan 2011 12:52:29 +0000 (12:52 +0000)]
These methods should be "const"; make them so.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122809
91177308-0d34-0410-b5e6-
96231b3b80d8
Francois Pichet [Tue, 4 Jan 2011 10:23:42 +0000 (10:23 +0000)]
Disable r122754 on Windows: was causing all lit tests to fail.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122808
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 4 Jan 2011 07:46:33 +0000 (07:46 +0000)]
Teach loop-idiom to turn a loop containing a memset into a larger memset
when safe.
The testcase is basically this nested loop:
void foo(char *X) {
for (int i = 0; i != 100; ++i)
for (int j = 0; j != 100; ++j)
X[j+i*100] = 0;
}
which gets turned into a single memset now. clang -O3 doesn't optimize
this yet though due to a phase ordering issue I haven't analyzed yet.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122806
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 4 Jan 2011 07:27:30 +0000 (07:27 +0000)]
restructure this a bit. Initialize the WeakVH with "I", the
instruction *after* the store. The store will always be deleted
if the transformation kicks in, so we'd do an N^2 scan of every
loop block. Whoops.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122805
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Tue, 4 Jan 2011 06:42:27 +0000 (06:42 +0000)]
Eliminate repeated allocation of a per-BB DenseMap for a 4.6% reduction of time
spent in StrongPHIElimination on 403.gcc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122803
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Tue, 4 Jan 2011 04:43:31 +0000 (04:43 +0000)]
Avoid finding loop back edges when we are not splitting critical edges in
CodeGenPrepare (which is the default behavior).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122801
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Tue, 4 Jan 2011 01:05:30 +0000 (01:05 +0000)]
Don't pattern match "/clang" so we don't mangle directory names. Some
tests use absolute paths to clang.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122796
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Tue, 4 Jan 2011 00:55:21 +0000 (00:55 +0000)]
Clean up a funky pass registration that got passed over when I got rid of static constructors.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122795
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Tue, 4 Jan 2011 00:32:57 +0000 (00:32 +0000)]
Fix the ARM IIC_iCMPsi itinerary and add an important assert.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122794
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Tue, 4 Jan 2011 00:12:46 +0000 (00:12 +0000)]
Address most of Duncan's review comments. Also, make LoopInstSimplify a simple
FunctionPass. It probably doesn't have a reason to be a LoopPass, as it will
probably drop the simple fixed point and either use RPO iteration or Duncan's
approach in instsimplify of only revisiting instructions that have changed.
The next step is to preserve LoopSimplify. This looks like it won't be too hard,
although the pass manager doesn't actually seem to respect when non-loop passes
claim to preserve LCSSA or LoopSimplify. This will have to be fixed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122791
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Tue, 4 Jan 2011 00:06:55 +0000 (00:06 +0000)]
use the very-handy getTruncateOrZeroExtend helper function, and
stop setting NSW: signed overflow is possible. Thanks to Dan
for pointing these out.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122790
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Mon, 3 Jan 2011 23:59:05 +0000 (23:59 +0000)]
Formatting changes. No functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122789
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Mon, 3 Jan 2011 23:51:56 +0000 (23:51 +0000)]
Fix comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122788
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Mon, 3 Jan 2011 23:51:43 +0000 (23:51 +0000)]
Use the new addEscapingValue callback to update GlobalsModRef when GVN adds PHIs of GEPs. For the moment,
have GlobalsModRef handle this conservatively by simply removing the value from its maps.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122787
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 3 Jan 2011 23:47:14 +0000 (23:47 +0000)]
Convert MC tests to .s so codegen changes won't break them.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122786
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 23:38:13 +0000 (23:38 +0000)]
Duncan deftly points out that readnone functions aren't
invalidated by stores, so they can be handled as 'simple'
operations.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122785
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 3 Jan 2011 22:53:22 +0000 (22:53 +0000)]
Use pushq / popq instead of subq $8, %rsp / addq $8, %rsp to adjust stack in
prologue and epilogue if the adjustment is 8. Similarly, use pushl / popl if
the adjustment is 4 in 32-bit mode.
In the epilogue, takes care to pop to a caller-saved register that's not live
at the exit (either return or tailcall instruction).
rdar://
8771137
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122783
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Mon, 3 Jan 2011 21:55:08 +0000 (21:55 +0000)]
Don't pattern match "clang-" as it may be part of a tool name with a
triple suffix.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122779
91177308-0d34-0410-b5e6-
96231b3b80d8
Wesley Peck [Mon, 3 Jan 2011 21:40:26 +0000 (21:40 +0000)]
Fix more stack layout issues in the MBlaze backend.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122778
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Mon, 3 Jan 2011 21:38:41 +0000 (21:38 +0000)]
Stub out a new updating interface to AliasAnalysis, allowing stateful analyses to be informed when
a pointer value has potentially become escaping. Implementations can choose to either fall back to
conservative responses for that value, or may recompute their analysis to accomodate the change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122777
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Mon, 3 Jan 2011 21:08:22 +0000 (21:08 +0000)]
We can count properly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122773
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 21:03:33 +0000 (21:03 +0000)]
fix rdar://
8813415 - a miscompilation of 164.gzip that loop-idiom
exposed. It turns out to be a latent bug in basicaa, scary.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122772
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 21:01:26 +0000 (21:01 +0000)]
filecheckize
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122771
91177308-0d34-0410-b5e6-
96231b3b80d8
Oscar Fuentes [Mon, 3 Jan 2011 20:01:32 +0000 (20:01 +0000)]
TableGen.cmake: sometimes the .td file is not in the current directory
(clang/include/clang/Basic/StmtNodes.td, for instance, is tablegenned
from clang/include/clang/AST/CMakeLists.txt) so it is not contained on
the list of all .td files on the current source directory which is
used as the DEPENDS of the custom command. We must add the .td file to
the DEPENDS list of the custom command. Otherwise some .inc files are
not regenerated when the corresponding .td file changes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122768
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Mon, 3 Jan 2011 19:00:11 +0000 (19:00 +0000)]
Simplify GVN's value expression structure, allowing the elimination of a lot of
almost-but-not-quite-identical code. No intended functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122760
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 18:43:03 +0000 (18:43 +0000)]
stength reduce my previous patch a bit. The only instructions
that are allowed to have metadata operands are intrinsic calls,
and the only ones that take metadata currently return void.
Just reject all void instructions, which should not be value
numbered anyway. To future proof things, add an assert to the
getHashValue impl for calls to check that metadata operands
aren't present.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122759
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 18:28:15 +0000 (18:28 +0000)]
fix PR8895: metadata operands don't have a strong use of their
nested values, so they can change and drop to null, which can
change the hash and cause havok.
It turns out that it isn't a good idea to value number stuff
with metadata operands anyway, so... don't.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122758
91177308-0d34-0410-b5e6-
96231b3b80d8
David Greene [Mon, 3 Jan 2011 17:30:25 +0000 (17:30 +0000)]
Reapply 122341 to fix PR8199 now that clang changes are in.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122754
91177308-0d34-0410-b5e6-
96231b3b80d8
Oscar Fuentes [Mon, 3 Jan 2011 16:59:52 +0000 (16:59 +0000)]
LLVMProcessSources: add .def files along with .h files to targets for
the benefit of project-based generators (VS, XCode, etc).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122749
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Mon, 3 Jan 2011 10:50:04 +0000 (10:50 +0000)]
Speed up instsimplify by about 10-15% by not bothering to retry
InstructionSimplify on instructions that didn't change since the
last time round the loop.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122745
91177308-0d34-0410-b5e6-
96231b3b80d8
Evan Cheng [Mon, 3 Jan 2011 07:53:18 +0000 (07:53 +0000)]
Undo what looks like accidental removal of an instcombine pass in r122740.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122743
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Mon, 3 Jan 2011 06:33:01 +0000 (06:33 +0000)]
Switch a worklist in CodeGenPrepare to SmallVector and increase the inline
capacity on the Visited SmallPtrSet. On 403.gcc, this is about a 4.5% speedup of
CodeGenPrepare time (which itself is 10% of time spent in the backend).
This is progress towards PR8889.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122741
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 06:19:09 +0000 (06:19 +0000)]
Turn on earlycse by default. This seems to be a small performance
improvement in the generated code, and speeds up 'opt -std-compile-opts'
compile time on 176.gcc from 24.84s to 23.2s (about 7%).
This also resolves a specific code quality issue in rdar://
7352081 which
was generating poor code for:
int t(int a, int b) {
if (a & b & 1)
return a & b;
return 3;
}
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122740
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Mon, 3 Jan 2011 06:16:07 +0000 (06:16 +0000)]
Further expand what a call graph pass may do.
The rationale is that after analyzing a function in the SCC, we may want to
modify it in a way that requires us to update its uses (f.e. to replace the
call with a constant) or its users (f.e. to call it with fewer arguments).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122739
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 04:17:24 +0000 (04:17 +0000)]
earlycse can do trivial with-a-block dead store
elimination as well. This deletes 60 stores in 176.gcc
that largely come from bitfield code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122736
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Mon, 3 Jan 2011 04:07:46 +0000 (04:07 +0000)]
Use a RecyclingAllocator to allocate values for MachineCSE's ScopedHashTable for
a 28% speedup of MachineCSE time on 403.gcc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122735
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Mon, 3 Jan 2011 04:01:44 +0000 (04:01 +0000)]
Permit CallGraphSCCPasses readonly access to the direct callers of the functions
in their SCC as they already have with the direct callees.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122734
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 03:53:50 +0000 (03:53 +0000)]
switch the load table to use a recycling bump pointer allocator,
speeding earlycse up by 6%.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122733
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 03:46:34 +0000 (03:46 +0000)]
now that loads are in their own table, we can implement
store->load forwarding. This allows EarlyCSE to zap 600 more
loads from 176.gcc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122732
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 03:41:27 +0000 (03:41 +0000)]
split loads and calls into separate tables. Loads are now just indexed
by their pointer instead of using MemoryValue to wrap it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122731
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 03:33:47 +0000 (03:33 +0000)]
add a testcase for readonly call CSE
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122730
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 03:28:23 +0000 (03:28 +0000)]
various cleanups, no functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122729
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Mon, 3 Jan 2011 03:19:35 +0000 (03:19 +0000)]
Add spliceFunction to the CallGraph interface. This allows users to efficiently
update a callGraph when performing the common operation of splicing the body to
a new function and updating all callers (such as via RAUW).
No users yet, though this is intended for DeadArgumentElimination as part of
PR8887.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122728
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 03:18:43 +0000 (03:18 +0000)]
Teach EarlyCSE to do trivial CSE of loads and read-only calls.
On 176.gcc, this catches 13090 loads and calls, and increases the
number of simple instructions CSE'd from 29658 to 36208.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122727
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 03:16:20 +0000 (03:16 +0000)]
add a handy typedef.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122726
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 02:20:48 +0000 (02:20 +0000)]
rename InstValue to SimpleValue, add some comments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122725
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael J. Spencer [Mon, 3 Jan 2011 02:13:05 +0000 (02:13 +0000)]
CMake: Add missing source file.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122724
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 01:42:46 +0000 (01:42 +0000)]
Allocate nodes for the scoped hash table from a recyling bump pointer
allocator. This speeds up early cse by about 20%
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122723
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 01:38:29 +0000 (01:38 +0000)]
really get this working with a custom allocator.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122722
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 01:29:37 +0000 (01:29 +0000)]
Enhance ScopedHashTable to allow it to take an allocator argument.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122721
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Jan 2011 01:10:08 +0000 (01:10 +0000)]
reduce redundancy in the hashing code and other misc cleanups.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122720
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Mon, 3 Jan 2011 00:25:16 +0000 (00:25 +0000)]
Add a new loop-instsimplify pass, with the intention of replacing the instance
of instcombine that is currently in the middle of the loop pass pipeline. This
commit only checks in the pass; it will hopefully be enabled by default later.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122719
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 2 Jan 2011 23:29:58 +0000 (23:29 +0000)]
fix some pastos
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122718
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 2 Jan 2011 23:19:45 +0000 (23:19 +0000)]
add DEBUG and -stats output to earlycse.
Teach it to CSE the rest of the non-side-effecting instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122716
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 2 Jan 2011 23:04:14 +0000 (23:04 +0000)]
Enhance earlycse to do CSE of casts, instsimplify and die.
Add a testcase.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122715
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 2 Jan 2011 22:09:33 +0000 (22:09 +0000)]
split dom frontier handling stuff out to its own DominanceFrontier header,
so that Dominators.h is *just* domtree. Also prune #includes a bit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122714
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 2 Jan 2011 21:47:05 +0000 (21:47 +0000)]
sketch out a new early cse pass. No functionality yet.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122713
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 2 Jan 2011 21:14:18 +0000 (21:14 +0000)]
fix a miscompilation of tramp3d-v4: when forming a memcpy, we have to make
sure that the loop we're promoting into a memcpy doesn't mutate the input
of the memcpy. Before we were just checking that the dest of the memcpy
wasn't mod/ref'd by the loop.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122712
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 2 Jan 2011 20:24:21 +0000 (20:24 +0000)]
If a loop iterates exactly once (has backedge count = 0) then don't
mess with it. We'd rather peel/unroll it than convert all of its
stores into memsets.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122711
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Sun, 2 Jan 2011 19:57:05 +0000 (19:57 +0000)]
Try to reuse the value when lowering memset.
This allows us to compile:
void test(char *s, int a) {
__builtin_memset(s, a, 15);
}
into 1 mul + 3 stores instead of 3 muls + 3 stores.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122710
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Sun, 2 Jan 2011 19:44:58 +0000 (19:44 +0000)]
Lower the i8 extension in memset to a multiply instead of a potentially long series of shifts and ors.
We could implement a DAGCombine to turn x * 0x0101 back into logic operations
on targets that doesn't support the multiply or it is slow (p4) if someone cares
enough.
Example code:
void test(char *s, int a) {
__builtin_memset(s, a, 4);
}
before:
_test: ## @test
movzbl 8(%esp), %eax
movl %eax, %ecx
shll $8, %ecx
orl %eax, %ecx
movl %ecx, %eax
shll $16, %eax
orl %ecx, %eax
movl 4(%esp), %ecx
movl %eax, 4(%ecx)
movl %eax, (%ecx)
ret
after:
_test: ## @test
movzbl 8(%esp), %eax
imull $
16843009, %eax, %eax ## imm = 0x1010101
movl 4(%esp), %ecx
movl %eax, 4(%ecx)
movl %eax, (%ecx)
ret
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122707
91177308-0d34-0410-b5e6-
96231b3b80d8
Oscar Fuentes [Sun, 2 Jan 2011 19:32:31 +0000 (19:32 +0000)]
A workaround for a bug in cmake 2.8.3 diagnosed on PR 8885.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122706
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Sun, 2 Jan 2011 19:16:44 +0000 (19:16 +0000)]
Also remove functions that use complex constant expressions in terms of
another function.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122705
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 2 Jan 2011 19:01:03 +0000 (19:01 +0000)]
enhance loop idiom recognition to scan *all* unconditionally executed
blocks in a loop, instead of just the header block. This makes it more
aggressive, able to handle Duncan's Ada examples.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122704
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 2 Jan 2011 18:53:08 +0000 (18:53 +0000)]
make inSubLoop much more efficient.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122703
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 2 Jan 2011 18:45:39 +0000 (18:45 +0000)]
rip out isExitBlockDominatedByBlockInLoop, calling DomTree::dominates instead.
isExitBlockDominatedByBlockInLoop is a relic of the days when domtree was
*just* a tree and didn't have DFS numbers. Checking DFS numbers is faster
and easier than "limiting the search of the tree".
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122702
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 2 Jan 2011 18:32:09 +0000 (18:32 +0000)]
add a list of opportunities for future improvement.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122701
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 2 Jan 2011 18:31:38 +0000 (18:31 +0000)]
update a bunch of entries.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122700
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan Sands [Sun, 2 Jan 2011 13:38:21 +0000 (13:38 +0000)]
Fix PR8702 by not having LoopSimplify claim to preserve LCSSA form. As described
in the PR, the pass could break LCSSA form when inserting preheaders. It probably
would be easy enough to fix this, but since currently we always go into LCSSA form
after running this pass, doing so is not urgent.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122695
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Sun, 2 Jan 2011 12:37:22 +0000 (12:37 +0000)]
Remove an unused member function.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122693
91177308-0d34-0410-b5e6-
96231b3b80d8
Oscar Fuentes [Sun, 2 Jan 2011 12:30:18 +0000 (12:30 +0000)]
Propagate to parent scope changes made to CMAKE_CXX_FLAGS.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122692
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Sun, 2 Jan 2011 12:17:10 +0000 (12:17 +0000)]
Fix a typo in a variable name.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122691
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Sun, 2 Jan 2011 10:50:14 +0000 (10:50 +0000)]
Move a load into the only branch where it is used and eliminate a temporary.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122690
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Sun, 2 Jan 2011 10:40:14 +0000 (10:40 +0000)]
Add the explanatory comment from r122680's commit message to the code itself.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122689
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Sun, 2 Jan 2011 10:10:02 +0000 (10:10 +0000)]
Tidy up indentation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122688
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Sun, 2 Jan 2011 10:06:44 +0000 (10:06 +0000)]
Fix a typo, which should also fix the failure on llvm-x86_64-linux-checks.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122687
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 2 Jan 2011 07:58:36 +0000 (07:58 +0000)]
Allow loop-idiom to run on multiple BB loops, but still only scan the loop
header for now for memset/memcpy opportunities. It turns out that loop-rotate
is successfully rotating loops, but *DOESN'T MERGE THE BLOCKS*, turning "for
loops" into 2 basic block loops that loop-idiom was ignoring.
With this fix, we form many *many* more memcpy and memsets than before, including
on the "history" loops in the viterbi benchmark, which look like this:
for (j=0; j<MAX_history; ++j) {
history_new[i][j+1] = history[2*i][j];
}
Transforming these loops into memcpy's speeds up the viterbi benchmark from
11.98s to 3.55s on my machine. Woo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122685
91177308-0d34-0410-b5e6-
96231b3b80d8
Cameron Zwarich [Sun, 2 Jan 2011 07:53:49 +0000 (07:53 +0000)]
Remove the #ifdef'd code for balancing the eval-link data structure. It doesn't
compile, and everyone's tests have shown it to be slower in practice, even for
quite large graphs.
I also hope to do an optimization that is only correct with the simpler data
structure, which would break this even further.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122684
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 2 Jan 2011 07:37:13 +0000 (07:37 +0000)]
remove debugging code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122683
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Sun, 2 Jan 2011 07:36:44 +0000 (07:36 +0000)]
add some -stats output.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122682
91177308-0d34-0410-b5e6-
96231b3b80d8