caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
104,478 Commits 1 Branch 0 Tags
1c229c0472edfd59ce61ba063a93517df609ceb4
Commit Graph

2290 Commits

Author SHA1 Message Date
Cameron Zwarich
ca4c633489 Fix another case of <rdar://problem/9184212> that only occurs with code 
generated by llvm-gcc, since llvm-gcc uses 2 i64s for passing a 4 x float
vector on ARM rather than an i64 array like Clang.

llvm-svn: 129878
 2011-04-20 21:48:38 +00:00
Frits van Bommel
d097212a08 Add test cases for Jay's r129641 and fix a 32-bit-centric testcase in a file with a 64-bit datalayout. 
llvm-svn: 129643
 2011-04-16 14:31:50 +00:00
Chris Lattner
0ab5e2cded Fix a ton of comment typos found by codespell. Patch by 
Luis Felipe Strano Moraes!

llvm-svn: 129558
 2011-04-15 05:18:47 +00:00
Eli Friedman
2395626605 Add an instcombine for constructs like a | -(b != c); a select is more 
canonical, and generally leads to better code.  Found while looking at
an article about saturating arithmetic.

llvm-svn: 129545
 2011-04-14 22:41:27 +00:00
Owen Anderson
92651ec374 Fix an infinite alternation in JumpThreading where two transforms would repeatedly undo each other. The solution is to perform more aggressive constant folding to make one of the edges just folded away rather than trying to thread it. 
Fixes <rdar://problem/9284786>.

Discovered with CSmith.

llvm-svn: 129538
 2011-04-14 21:35:50 +00:00
Mon P Wang
2e5528f0b2 Vectors with different number of elements of the same element type can have 
the same allocation size but different primitive sizes(e.g., <3xi32> and
<4xi32>).  When ScalarRepl promotes them, it can't use a bit cast but
should use a shuffle vector instead.

llvm-svn: 129472
 2011-04-13 21:40:02 +00:00
Dan Gohman
1c6c34834b Fix reassociate to use a worklist instead of recursing when new 
reassociation opportunities are exposed. This fixes a bug where
the nested reassociation expects to be the IR to be consistent,
but it isn't, because the outer reassociation has disconnected
some of the operands.  rdar://9167457

llvm-svn: 129324
 2011-04-12 00:11:56 +00:00
Chris Lattner
e81d045d94 remove the StructRetPromotion pass. It is unused, not maintained and 
has some bugs.  If this is interesting functionality, it should be 
reimplemented in the argpromotion pass.

llvm-svn: 129314
 2011-04-11 23:09:44 +00:00
Eli Friedman
9cca0715aa Add back a couple checks removed by r129128; the fact that an intitializer 
is an array of structures doesn't imply it's a ConstantArray of
ConstantStruct.

llvm-svn: 129207
 2011-04-09 09:11:09 +00:00
Chris Lattner
88974f4625 fix PR9523, a crash in looprotate on a non-canonical loop made out of indirectbr. 
llvm-svn: 129203
 2011-04-09 07:25:58 +00:00
Eli Friedman
17822fcde9 PR9604; try to deal with RAUW updates correctly in the AST. I'm not convinced 
it's completely safe to cache the AST across LICM runs even with this fix,
but this fix can't hurt.

llvm-svn: 129198
 2011-04-09 06:55:46 +00:00
Eli Friedman
4db39cefdb Test for r129190. 
llvm-svn: 129197
 2011-04-09 06:39:43 +00:00
Devang Patel
bc3d8b212f Do not let debug info interfer with branch folding. 
llvm-svn: 129114
 2011-04-07 23:11:25 +00:00
Devang Patel
197c35298a While hoisting common code from if/else, hoist debug info intrinsics if they match. 
llvm-svn: 129078
 2011-04-07 17:27:36 +00:00
Eli Friedman
c5f22a7815 PR9634: Don't unconditionally tell the AliasSetTracker that the PreheaderLoad 
is equivalent to any other relevant value; it isn't true in general.
If it is equivalent, the LoopPromoter will tell the AST the equivalence.
Also, delete the PreheaderLoad if it is unused.

Chris, since you were the last one to make major changes here, can you check
that this is sane?

llvm-svn: 129049
 2011-04-07 01:35:06 +00:00
Nadav Rotem
cc771acd77 This testcase passed even without the fix. Added the target info to make the 
test fail (without the fix). Thanks Dan.

llvm-svn: 128999
 2011-04-06 11:18:29 +00:00
Nadav Rotem
a069c6ce05 InstCombine optimizes gep(bitcast(x)) even when the bitcasts casts away address 
space info. We crash with an assert in this case. This change checks that the
address space of the bitcasted pointer is the same as the gep ptr.

llvm-svn: 128884
 2011-04-05 14:29:52 +00:00
Eli Friedman
17bf4922c9 PR9446: RecursivelyDeleteTriviallyDeadInstructions can delete the instruction 
after the given instruction; make sure to handle that case correctly.
(It's difficult to trigger; the included testcase involves a dead 
block, but I don't think that's a requirement.) 

While I'm here, get rid of the unnecessary warning about
SimplifyInstructionsInBlock, since it should work correctly as far as I know.

llvm-svn: 128782
 2011-04-02 22:45:17 +00:00
Benjamin Kramer
d121765e64 InstCombine: Turn icmp + sext into bitwise/integer ops when the input has only one unknown bit. 
int test1(unsigned x) { return (x&8) ? 0 : -1; }
int test3(unsigned x) { return (x&8) ? -1 : 0; }

before (x86_64):
_test1:
	andl	$8, %edi
	cmpl	$1, %edi
	sbbl	%eax, %eax
	ret
_test3:
	andl	$8, %edi
	cmpl	$1, %edi
	sbbl	%eax, %eax
	notl	%eax
	ret

after:
_test1:
	shrl	$3, %edi
	andl	$1, %edi
	leal	-1(%rdi), %eax
	ret
_test3:
	shll	$28, %edi
	movl	%edi, %eax
	sarl	$31, %eax
	ret

llvm-svn: 128732
 2011-04-01 20:09:10 +00:00
Nadav Rotem
d74b72b8a9 Instcombile optimization: extractelement(cast) -> cast(extractelement) 
llvm-svn: 128683
 2011-03-31 22:57:29 +00:00
Benjamin Kramer
5291054ef1 InstCombine: APFloat can't perform arithmetic on PPC double doubles, don't even try. 
Thanks Eli!

llvm-svn: 128676
 2011-03-31 21:35:49 +00:00
Benjamin Kramer
be209ab8a2 InstCombine: Fix transform to use the swapped predicate. 
Thanks Frits!

llvm-svn: 128628
 2011-03-31 10:46:03 +00:00
Benjamin Kramer
d159d94644 InstCombine: fold fcmp (fneg x), (fneg y) -> fcmp x, y 
llvm-svn: 128627
 2011-03-31 10:12:22 +00:00
Benjamin Kramer
a8c5d0872d InstCombine: fold fcmp pred (fneg x), C -> fcmp swap(pred) x, -C 
llvm-svn: 128626
 2011-03-31 10:12:15 +00:00
Benjamin Kramer
cbb18e91a8 InstCombine: Shrink "fcmp (fpext x), C" to "fcmp x, C" if C can be losslessly converted to the type of x. 
Fixes PR9592.

llvm-svn: 128625
 2011-03-31 10:12:07 +00:00
Benjamin Kramer
2ccfbc8b71 InstCombine: fold fcmp (fpext x), (fpext y) -> fcmp x, y. 
llvm-svn: 128624
 2011-03-31 10:11:58 +00:00
Bill Wendling
5034159c5f * The DSE code that tested for overlapping needed to take into account the fact 
that one of the numbers is signed while the other is unsigned. This could lead
  to a wrong result when the signed was promoted to an unsigned int.

* Add the data layout line to the testcase so that it will test the appropriate
  thing.

Patch by David Terei!

llvm-svn: 128577
 2011-03-30 21:37:19 +00:00
Benjamin Kramer
af0ed953c5 Avoid turning a floating point division with a constant power of two into a denormal multiplication. 
Some platforms may treat denormals as zero, on other platforms multiplication
with a subnormal is slower than dividing by a normal.

llvm-svn: 128555
 2011-03-30 17:02:54 +00:00
Benjamin Kramer
8564e0de96 InstCombine: If the divisor of an fdiv has an exact inverse, turn it into an fmul. 
Fixes PR9587.

llvm-svn: 128546
 2011-03-30 15:42:35 +00:00
Benjamin Kramer
272f2b0044 InstCombine: Add a few missing combines for ANDs and ORs of sign bit tests. 
On x86 we now compile "if (a < 0 && b < 0)" into
	testl	%edi, %esi
	js	IF.THEN

llvm-svn: 128496
 2011-03-29 22:06:41 +00:00
Cameron Zwarich
ff811cc475 Do some simple copy propagation through integer loads and stores when promoting 
vector types. This helps a lot with inlined functions when using the ARM soft
float ABI. Fixes <rdar://problem/9184212>.

llvm-svn: 128453
 2011-03-29 05:19:52 +00:00
Nick Lewycky
8544228d5a Teach the transformation that moves binary operators around selects to preserve 
the subclass optional data.

llvm-svn: 128388
 2011-03-27 19:51:23 +00:00
Frits van Bommel
0bb2ad2cf7 Constant folding support for calls to umul.with.overflow(), basically identical to the smul.with.overflow() code. 
llvm-svn: 128379
 2011-03-27 14:26:13 +00:00
Nick Lewycky
83167df787 Add a small missed optimization: turn X == C ? X : Y into X == C ? C : Y. This 
removes one use of X which helps it pass the many hasOneUse() checks.

In my analysis, this turns up very often where X = A >>exact B and that can't be
simplified unless X has one use (except by increasing the lifetime of A which is
generally a performance loss).

llvm-svn: 128373
 2011-03-27 07:30:57 +00:00
Cameron Zwarich
d4174ee43e Fix a typo and add a test. 
llvm-svn: 128331
 2011-03-26 04:58:50 +00:00
Bill Wendling
db40b5c899 PR9561: A store with a negative offset (via GEP) could erroniously say that it 
completely overlaps a previous store, thus mistakenly deleting that store. Check
for this condition.

llvm-svn: 128319
 2011-03-26 01:20:37 +00:00
Cameron Zwarich
10ebc189ee Fix PR9464 by correcting some math that just happened to be right in most cases 
that were hit in practice.

llvm-svn: 128146
 2011-03-23 05:25:55 +00:00
Anders Carlsson
ee6bc70d2f Add an optimization to GlobalOpt that eliminates calls to __cxa_atexit, if the function passed is empty. 
llvm-svn: 127970
 2011-03-20 17:59:11 +00:00
Andrew Trick
1c4b42d00f Avoid creating canonical induction variables for non-native types. 
For example, on 32-bit architecture, don't promote all uses of the IV
to 64-bits just because one use is a 64-bit cast.
Alternate implementation of the patch by Arnaud de Grandmaison.

llvm-svn: 127884
 2011-03-18 16:50:32 +00:00
Eli Friedman
c17c9a78aa FileCheck-ize and update test. 
llvm-svn: 127845
 2011-03-18 01:10:31 +00:00
Devang Patel
aad34d882d Try to not lose variable's debug info during instcombine. 
This is done by lowering dbg.declare intrinsic into dbg.value intrinsic.
Radar 9143931.

llvm-svn: 127834
 2011-03-17 22:18:16 +00:00
Cameron Zwarich
0454253d7a Only convert allocas to scalars if it is profitable. The profitability metric I 
chose is having a non-memcpy/memset use and being larger than any native integer
type. Originally I chose having an access of a size smaller than the total size
of the alloca, but this caused some minor issues on the spirit benchmark where
SRoA runs again after some inlining.

This fixes <rdar://problem/8613163>.

llvm-svn: 127718
 2011-03-16 00:13:44 +00:00
Cameron Zwarich
7b0f3c6a1a Add native integer type TargetData to some existing tests. 
llvm-svn: 127717
 2011-03-16 00:13:40 +00:00
Cameron Zwarich
0b8cdfb6ec Do not add PHIs with no users when creating LCSSA form. Patch by Andrew Clinton. 
llvm-svn: 127674
 2011-03-15 07:41:25 +00:00
Eli Friedman
c4414c6e92 PR9450: Make switch optimization in SimplifyCFG not dependent on the ordering 
of pointers in an std::map.

llvm-svn: 127650
 2011-03-15 02:23:35 +00:00
Eric Christopher
2139d3148f If we don't know how long a string is we can't fold an _chk version to the 
normal version.

Fixes rdar://9123638

llvm-svn: 127636
 2011-03-15 00:25:41 +00:00
Benjamin Kramer
5acc751b6f Teach ComputeMaskedBits about sub nsw. 
llvm-svn: 127548
 2011-03-12 17:18:11 +00:00
Cameron Zwarich
338d362200 Roll r127459 back in: 
Optimize trivial branches in CodeGenPrepare, which often get created from the
lowering of objectsize intrinsics. Unfortunately, a number of tests were relying
on llc not optimizing trivial branches, so I had to add an option to allow them
to continue to test what they originally tested.

This fixes <rdar://problem/8785296> and <rdar://problem/9112893>.

llvm-svn: 127498
 2011-03-11 21:52:04 +00:00
Daniel Dunbar
94ccb27b43 Revert r127459, "Optimize trivial branches in CodeGenPrepare, which often get 
created from the", it broke some GCC test suite tests.

llvm-svn: 127477
 2011-03-11 19:30:30 +00:00
Benjamin Kramer
391a946fa9 ComputeMaskedBits: sub falls through to add, and sub doesn't have the same overflow semantics as add. 
Should fix the selfhost failures that started with r127463.

llvm-svn: 127465
 2011-03-11 14:46:49 +00:00