1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-31 07:52:55 +01:00
Commit Graph

2888 Commits

Author SHA1 Message Date
Nuno Lopes
11d6ecb6db objectsize: add a few more tests and fix a bug
llvm-svn: 156625
2012-05-11 18:25:29 +00:00
Eli Friedman
1746bfc50e Fix a minor logic mistake transforming compares in instcombine. PR12514.
llvm-svn: 156600
2012-05-11 01:32:59 +00:00
Nuno Lopes
415911a5c7 objectsize: add support for GEPs with non-constant indexes
add an additional parameter to InstCombiner::EmitGEPOffset() to force it to *not* emit operations with NUW flag

llvm-svn: 156585
2012-05-10 23:17:35 +00:00
Dan Gohman
8b1a3cec89 Teach DeadStoreElimination to eliminate exit-block stores with phi addresses.
llvm-svn: 156558
2012-05-10 18:57:38 +00:00
Nuno Lopes
ea7b37e3ae teach DSE and isInstructionTriviallyDead() about calloc
llvm-svn: 156553
2012-05-10 17:14:00 +00:00
Joel Jones
cc8aa34ea8 formatting change: strip debug info from test
llvm-svn: 156551
2012-05-10 16:55:31 +00:00
Joel Jones
305ecb3495 Fix a problem with incomplete equality testing of PHINodes in
Instruction::IsIdenticalToWhenDefined.

This manifested itself when inlining two calls to the same function.  The 
inlined function had a switch statement that returned one of a set of 
global variables.  Without this modification, the two phi instructions that 
chose values from the branches of the switch instruction inlined from the 
callee were considered equivalent and jump-threading replaced a load for the 
first switch value with a phi selecting from the second switch, thereby 
producing incorrect code.

This patch has been tested with "make check-all", "lnt runteste nt", and 
llvm self-hosted, and on the original program that had this problem, 
wireshark.

<rdar://problem/11025519>

llvm-svn: 156548
2012-05-10 15:59:41 +00:00
Dan Gohman
9e72870dd1 Fix the objc_storeStrong recognizer to stop before walking off the
end of a basic block if there's no store.

llvm-svn: 156520
2012-05-09 23:08:33 +00:00
Nuno Lopes
3d7a8137ee objectsize:
refactor code a bit to enable future changes to support run-time information
add support to compute allocation sizes at run-time if penalty > 1 (e.g., malloc(x), calloc(x, y), and VLAs)

llvm-svn: 156515
2012-05-09 21:30:57 +00:00
Nuno Lopes
e8880a9916 change the objectsize intrinsic signature: add a 3rd parameter to denote the maximum runtime performance penalty that the user is willing to accept.
This commit only adds the parameter. Code taking advantage of it will follow.

llvm-svn: 156473
2012-05-09 15:52:43 +00:00
Filipe Cabecinhas
fe00fb1f06 Fixed a typo
llvm-svn: 156471
2012-05-09 14:43:50 +00:00
Dan Gohman
b47d02f929 Fix objc_storeStrong pattern matching to catch a potential use of the
old value after the store but before it is released.
This fixes rdar:/11116986.

llvm-svn: 156442
2012-05-08 23:34:08 +00:00
Duncan Sands
c9f011a85b Calling ReassociateExpression recursively is extremely dangerous since it will
replace the operands of expressions with only one use with undef and generate
a new expression for the original without using RAUW to update the original.
Thus any copies of the original expression held in a vector may end up
referring to some bogus value - and using a ValueHandle won't help since there
is no RAUW.  There is already a mechanism for getting the effect of recursion
non-recursively: adding the value to be recursed on to RedoInsts.  But it wasn't
being used systematically.  Have various places where recursion had snuck in at
some point use the RedoInsts mechanism instead.  Fixes PR12169.

llvm-svn: 156379
2012-05-08 12:16:05 +00:00
Owen Anderson
1e7a4f0f91 Teach reassociate to commute FMul's and FAdd's in order to canonicalize the order of their operands across instructions. This allows for greater CSE opportunities.
llvm-svn: 156323
2012-05-07 20:47:23 +00:00
Stepan Dyatkovskiy
469935e0ae Small fix in InstCombineCasts.cpp. Restored "alloca + bitcast" reducing for case when alloca's size is calculated within the "add/sub/... nsw".
Also added fix to 2011-06-13-nsw-alloca.ll test.

llvm-svn: 156231
2012-05-05 07:09:40 +00:00
Nuno Lopes
2762496a1a remove calls to calloc if the allocated memory is not used (it was already being done for malloc)
fix a few typos found by Chad in my previous commit

llvm-svn: 156110
2012-05-03 22:08:19 +00:00
Nuno Lopes
26239aeb99 add support for calloc to objectsize lowering
llvm-svn: 156102
2012-05-03 21:19:58 +00:00
Bill Wendling
4cb38868b9 The value held in the vector may be RAUW'ed by some of the canonicalization
methods. Use a weak value handle to keep up with this.
PR12245

llvm-svn: 155984
2012-05-02 09:59:45 +00:00
Nick Lewycky
fd4342c2f1 An instruction in a loop is not guaranteed to be executed just because the loop
has no exit blocks. Fixes PR12706!

llvm-svn: 155884
2012-05-01 04:03:01 +00:00
Lang Hames
26d71c9d0a Add support for llvm.arm.neon.vmull* intrinsics to InstCombine. Fixes
<rdar://problem/11291436>.

This is a second attempt at a fix for this, the first was r155468. Thanks
to Chandler, Bob and others for the feedback that helped me improve this.

llvm-svn: 155866
2012-05-01 00:20:38 +00:00
Duncan Sands
8fa4166239 Just mark the sign bit as known zero, rather than any other irrelevant bits
known zero in the LHS.  Fixes PR12541.

llvm-svn: 155818
2012-04-30 11:56:58 +00:00
Bill Wendling
5a1a6421ca Second attempt at PR12573:
Allow the "SplitCriticalEdge" function to split the edge to a landing pad. If
the pass is *sure* that it thinks it knows what it's doing, then it may go ahead
and specify that the landing pad can have its critical edge split. The loop
unswitch pass is one of these passes. It will split the critical edges of all
edges coming from a loop to a landing pad not within the loop. Doing so will
retain important loop analysis information, such as loop simplify.

llvm-svn: 155817
2012-04-30 10:44:54 +00:00
Rafael Espindola
314a1a477a Make sure HoistInsertPosition finds a position that is dominated by all
inputs.

llvm-svn: 155809
2012-04-30 03:53:06 +00:00
Hal Finkel
a565d03d78 Don't vectorize target-specific types (ppc_fp128, x86_fp80, etc.).
Target specific types should not be vectorized. As a practical matter,
these types are already register matched (at least in the x86 case),
and codegen does not always work correctly (at least in the ppc case,
and this is not worth fixing because ppc_fp128 is currently broken and
will probably go away soon).

llvm-svn: 155729
2012-04-27 19:34:00 +00:00
Dan Gohman
25a863dcf7 Reapply r155682, making constant folding more consistent, with a fix to work
properly with how the code handles all-undef PHI nodes.

llvm-svn: 155721
2012-04-27 17:50:22 +00:00
NAKAMURA Takumi
a28147f072 Revert r155682, "Use ConstantExpr::getExtractElement when constant-folding vectors"
It broke stage2 build. stage1/clang sometimes crashed.

llvm-svn: 155699
2012-04-27 07:59:20 +00:00
Dan Gohman
a72b2f97a6 Use ConstantExpr::getExtractElement when constant-folding vectors
instead of getAggregateElement. This has the advantage of being
more consistent and allowing higher-level constant folding to
procede even if an inner extract element cannot be folded.

Make ConstantFoldInstruction call ConstantFoldConstantExpression
on the instruction's operands, making it more consistent with 
ConstantFoldConstantExpression itself. This makes sure that
ConstantExprs get TargetData-aware folding before being handed
off as operands for further folding.

This causes more expressions to be folded, but due to a known
shortcoming in constant folding, this currently has the side effect
of stripping a few more nuw and inbounds flags in the non-targetdata
side of constant-fold-gep.ll. This is mostly harmless.

This fixes rdar://11324230.

llvm-svn: 155682
2012-04-27 00:54:36 +00:00
Chad Rosier
f3d4646377 Add instcombine patterns for the following transformations:
(x & y) | (x ^ y) -> x | y 
 (x & y) + (x ^ y) -> x | y 

Patch by Manman Ren.
rdar://10770603

llvm-svn: 155674
2012-04-26 23:29:14 +00:00
Chandler Carruth
587c136c31 Teach the reassociate pass to fold chains of multiplies with repeated
elements to minimize the number of multiplies required to compute the
final result. This uses a heuristic to attempt to form near-optimal
binary exponentiation-style multiply chains. While there are some cases
it misses, it seems to at least a decent job on a very diverse range of
inputs.

Initial benchmarks show no interesting regressions, and an 8%
improvement on SPASS. Let me know if any other interesting results (in
either direction) crop up!

Credit to Richard Smith for the core algorithm, and helping code the
patch itself.

llvm-svn: 155616
2012-04-26 05:30:30 +00:00
Chandler Carruth
8821a79947 Actually delete now-empty file.
llvm-svn: 155532
2012-04-25 02:30:00 +00:00
Lang Hames
7f69fbca29 Reverting r155468. Chris and Chandler have convinced me that it's dangerous and
in poor taste.

Talking through some alternate solutions with Chandler.

llvm-svn: 155530
2012-04-25 02:16:54 +00:00
Nadav Rotem
3c817bb807 ConstantFoldSelectInstruction swapped the operands of the select.
Fix 12592. Patch by Matt Pharr.

llvm-svn: 155480
2012-04-24 20:18:49 +00:00
Lang Hames
08eb5f2340 Add support for llvm.arm.neon.vmull* intrinsics to InstCombine. This fixes
<rdar://problem/11291436>.

llvm-svn: 155468
2012-04-24 18:58:36 +00:00
Chandler Carruth
eb9c5df516 Fix a crash on valid (if UB) bitcode that is produced for some global
constants in C++11 mode. I have no idea why it required such particular
circumstances to get here, the code seems clearly to rely upon unchecked
assumptions.

Specifically, when we decide to form an index into a struct type, we may
have gone through (at least one) zero-length array indexing round, which
would have left the offset un-adjusted, and thus not necessarily valid
for use when indexing the struct type.

This is just an canonicalization step, so the correct thing is to refuse
to canonicalize nonsensical GEPs of this form. Implemented, and test
case added.

Fixes PR12642. Pair debugged and coded with Richard Smith. =] I credit
him with most of the debugging, and preventing me from writing the wrong
code.

llvm-svn: 155466
2012-04-24 18:42:47 +00:00
Jakob Stoklund Olesen
6c1440cf27 Reapply r155136 after fixing PR12599.
Original commit message:

Defer some shl transforms to DAGCombine.

The shl instruction is used to represent multiplication by a constant
power of two as well as bitwise left shifts. Some InstCombine
transformations would turn an shl instruction into a bit mask operation,
making it difficult for later analysis passes to recognize the
constsnt multiplication.

Disable those shl transformations, deferring them to DAGCombine time.
An 'shl X, C' instruction is now treated mostly the same was as 'mul X, C'.

These transformations are deferred:

  (X >>? C) << C   --> X & (-1 << C)  (When X >> C has multiple uses)
  (X >>? C1) << C2 --> X << (C2-C1) & (-1 << C2)   (When C2 > C1)
  (X >>? C1) << C2 --> X >>? (C1-C2) & (-1 << C2)  (When C1 > C2)

The corresponding exact transformations are preserved, just like
div-exact + mul:

  (X >>?,exact C) << C   --> X
  (X >>?,exact C1) << C2 --> X << (C2-C1)
  (X >>?,exact C1) << C2 --> X >>?,exact (C1-C2)

The disabled transformations could also prevent the instruction selector
from recognizing rotate patterns in hash functions and cryptographic
primitives. I have a test case for that, but it is too fragile.

llvm-svn: 155362
2012-04-23 17:39:52 +00:00
Chandler Carruth
a4f8aa5231 Tidy up this test more:
1) Make the checked assertions a bit more precise. We really want the
   canonical forms coming out of reassociate to be exactly what is
   expected.
2) Remove other passes, and switch the test to actually directly check
   that reassociate makes the important transforms and
   canonicalizations.
3) Fold in a related test case now that we're using FileCheck. Make the
   same tidying changes to it.

llvm-svn: 155311
2012-04-22 10:11:26 +00:00
Chandler Carruth
038c36b06c FileCheck-ize a test, and tidy it up a touch.
llvm-svn: 155310
2012-04-22 10:11:23 +00:00
Jakob Stoklund Olesen
3d22f26e88 Revert r155136 "Defer some shl transforms to DAGCombine."
While the patch was perfect and defect free, it exposed a really nasty
bug in X86 SelectionDAG that caused an llc crash when compiling lencod.

I'll put the patch back in after fixing the SelectionDAG problem.

llvm-svn: 155181
2012-04-20 00:38:45 +00:00
Dan Gohman
f4472e9a1f Avoid a bug in the path count computation, preventing an infinite
loop repeatedlt making the same change. This is for rdar://11256239.

llvm-svn: 155160
2012-04-19 21:50:46 +00:00
Jakob Stoklund Olesen
1507d20c57 Defer some shl transforms to DAGCombine.
The shl instruction is used to represent multiplication by a constant
power of two as well as bitwise left shifts. Some InstCombine
transformations would turn an shl instruction into a bit mask operation,
making it difficult for later analysis passes to recognize the
constsnt multiplication.

Disable those shl transformations, deferring them to DAGCombine time.
An 'shl X, C' instruction is now treated mostly the same was as 'mul X, C'.

These transformations are deferred:

  (X >>? C) << C   --> X & (-1 << C)  (When X >> C has multiple uses)
  (X >>? C1) << C2 --> X << (C2-C1) & (-1 << C2)   (When C2 > C1)
  (X >>? C1) << C2 --> X >>? (C1-C2) & (-1 << C2)  (When C1 > C2)

The corresponding exact transformations are preserved, just like
div-exact + mul:

  (X >>?,exact C) << C   --> X
  (X >>?,exact C1) << C2 --> X << (C2-C1)
  (X >>?,exact C1) << C2 --> X >>?,exact (C1-C2)

The disabled transformations could also prevent the instruction selector
from recognizing rotate patterns in hash functions and cryptographic
primitives. I have a test case for that, but it is too fragile.

llvm-svn: 155136
2012-04-19 16:46:26 +00:00
Jakob Stoklund Olesen
3377b8ac85 Extract the broken part of XFAILed test into its own file.
llvm-svn: 155081
2012-04-19 00:20:38 +00:00
Jakob Stoklund Olesen
0ad7ee539b FileCheckize
llvm-svn: 155010
2012-04-18 17:01:26 +00:00
Jakob Stoklund Olesen
330da655a7 Nobody likes shifty instructions, but that was a bit strong.
llvm-svn: 155009
2012-04-18 16:44:44 +00:00
Joe Groff
1674d0c68d FileCheckify, un-XFAIL SimplifyLibCalls/floor test
Fixes build on MSVC

llvm-svn: 154970
2012-04-18 00:36:07 +00:00
Joe Groff
aff3c9d60d Move win32 SimplifyLibcall test under Transforms
llvm-svn: 154967
2012-04-18 00:07:45 +00:00
Chandler Carruth
728acc9bd9 Flip the new block-placement pass to be on by default.
This is mostly to test the waters. I'd like to get results from FNT
build bots and other bots running on non-x86 platforms.

This feature has been pretty heavily tested over the last few months by
me, and it fixes several of the execution time regressions caused by the
inlining work by preventing inlining decisions from radically impacting
block layout.

I've seen very large improvements in yacr2 and ackermann benchmarks,
along with the expected noise across all of the benchmark suite whenever
code layout changes. I've analyzed all of the regressions and fixed
them, or found them to be impossible to fix. See my email to llvmdev for
more details.

I'd like for this to be in 3.1 as it complements the inliner changes,
but if any failures are showing up or anyone has concerns, it is just
a flag flip and so can be easily turned off.

I'm switching it on tonight to try and get at least one run through
various folks' performance suites in case SPEC or something else has
serious issues with it. I'll watch bots and revert if anything shows up.

llvm-svn: 154816
2012-04-16 13:49:17 +00:00
Hal Finkel
028d6e153e Fix an error in BBVectorize important for vectorizing pointer types.
When vectorizing pointer types it is important to realize that potential
pairs cannot be connected via the address pointer argument of a load or store.
This is because even after vectorization, the address is still a scalar because
the address of the higher half of the pair is implicit from the address of the
lower half (it need not be, and should not be, explicitly computed).

llvm-svn: 154735
2012-04-14 07:32:50 +00:00
Hal Finkel
c55edb7b35 Enhance BBVectorize to more-properly handle pointer values and vectorize GEPs.
llvm-svn: 154734
2012-04-14 07:32:43 +00:00
Hal Finkel
12b4c41203 Add support to BBVectorize for vectorizing selects.
llvm-svn: 154700
2012-04-13 20:45:45 +00:00
Dan Gohman
d5743c7fd0 Consider ObjC runtime calls objc_storeWeak and others which make a copy of
their argument as "escape" points for objc_retainBlock optimization.
This fixes rdar://11229925.

llvm-svn: 154682
2012-04-13 18:28:58 +00:00