Commit Graph

5648 Commits

Author SHA1 Message Date
Dan Gohman
a23643d5b7 Rename ConstantFP's getInf to getInfinity.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82823 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-25 23:40:21 +00:00
Dan Gohman
79cb840136 Transform pow(x, 0.5) to (x == -inf ? inf : fabs(sqrt(x))), which is
typically faster then doing a general pow.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82819 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-25 23:10:17 +00:00
Torok Edwin
c338499525 Constant propagating byval pointer is safe if function is readonly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82700 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 18:33:42 +00:00
Torok Edwin
30a94e3b42 Don't constant propagate byval pointers, since they are not really pointers, but
rather structs passed by value.
This fixes PR5038.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82689 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-24 09:47:18 +00:00
Dale Johannesen
d5405a67d5 A minor improvment in accuracy to inline cost
computation, and some cosmetics.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82660 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-23 22:05:24 +00:00
Chris Lattner
2b723a5e3d tidy up
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82488 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-21 22:26:02 +00:00
Chris Lattner
19ad784dac big endian systems shift by bits too, hopefully this will fix the ppc
bootstrap problems.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82464 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-21 17:55:47 +00:00
Dan Gohman
471fbf9949 Nick pointed out that DominanceFrontier and DominanceTree are preserved
by setPreservesCFG().


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82463 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-21 17:54:42 +00:00
Dan Gohman
5ee98e8713 Remove the special-case for constants in PHI nodes; it's not really
helpful, and it didn't correctly handle the case of constants input
to PHIs for backedges.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82462 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-21 17:53:35 +00:00
Chris Lattner
8b2bc3d574 fix PR5016, a crash I introduced in GVN handing first class
arrays and structs, which cannot be bitcast to integers.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82460 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-21 17:24:04 +00:00
Chris Lattner
4fbd14e80e enable non-local analysis and PRE of large store -> little load.
This doesn't kick in too much because of phi translation issues,
but this can be resolved in the future.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82447 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-21 06:48:08 +00:00
Chris Lattner
879135145f convert an std::pair to an explicit struct.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82446 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-21 06:30:24 +00:00
Chris Lattner
ca74940fee move some functions, add a comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82444 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-21 06:24:16 +00:00
Chris Lattner
1ce0829e83 split HandleLoadFromClobberingStore in two pieces: one that does the
analysis, one that does the xform.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82443 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-21 06:22:46 +00:00
Chris Lattner
eed919b1ba Improve GVN to be able to forward substitute a small load
from a piece of a large store when both are in the same block.

This allows clang to compile the testcase in PR4216 to this code:

_test_bitfield:
	movl	4(%esp), %eax
	movl	%eax, %ecx
	andl	$-65536, %ecx
	orl	$32962, %eax
	andl	$40186, %eax
	orl	%ecx, %eax
	ret

This is not ideal, but is a whole lot better than the code produced
by llvm-gcc:

_test_bitfield:
	movw	$-32574, %ax
	orw	4(%esp), %ax
	andw	$-25350, %ax
	movw	%ax, 4(%esp)
	movw	7(%esp), %cx
	shlw	$8, %cx
	movzbl	6(%esp), %edx
	orw	%cx, %dx
	movzwl	%dx, %ecx
	shll	$16, %ecx
	movzwl	%ax, %eax
	orl	%ecx, %eax
	ret

and dramatically better than that produced by gcc 4.2:

_test_bitfield:
	pushl	%ebx
	call	L3
"L00000000001$pb":
L3:
	popl	%ebx
	movl	8(%esp), %eax
	leal	0(,%eax,4), %edx
	sarb	$7, %dl
	movl	%eax, %ecx
	andl	$7168, %ecx
	andl	$-7201, %ebx
	movzbl	%dl, %edx
	andl	$1, %edx
	sall	$5, %edx
	orl	%ecx, %ebx
	orl	%edx, %ebx
	andl	$24, %eax
	andl	$-58336, %ebx
	orl	%eax, %ebx
	orl	$32962, %ebx
	movl	%ebx, %eax
	popl	%ebx
	ret



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82439 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-21 05:57:11 +00:00
Chris Lattner
b2412a8bec formatting cleanups, no functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82426 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-21 02:42:51 +00:00
Chris Lattner
771a5422e1 Move CoerceAvailableValueToLoadType earlier in GVN.cpp. Hook it up
so that nonlocal and partially redundant loads can use it as well.
The testcase shows examples of craziness this can handle.  This triggers
*many* times in 176.gcc.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82403 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-20 20:09:34 +00:00
Chris Lattner
6af4b7ce4d change the interface to CoerceAvailableValueToLoadType to be
more generic.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82402 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-20 19:31:14 +00:00
Chris Lattner
bb6495cc67 enhance GVN to forward substitute a stored value to a load
(and load -> load) when the base pointers must alias but when
they are different types.  This occurs very very frequently in
176.gcc and other code that uses bitfields a lot.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82399 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-20 19:03:47 +00:00
Daniel Dunbar
a279bc3da5 Tabs -> spaces, and remove trailing whitespace.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82355 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-20 02:20:51 +00:00
Nick Lewycky
c332fba828 Remove the default value for ConstantStruct::get's isPacked parameter and
update the code which was broken by this.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82327 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-19 20:30:26 +00:00
Victor Hernandez
83d63919bd Enhance transform passes so that they apply the same tranforms to malloc calls as to MallocInst.
Reviewed by Dan Gohman.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82300 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-18 22:35:49 +00:00
Victor Hernandez
88d9839d07 Update malloc call creation code (AllocType is now the element type of the malloc, not the resulting type).
In getMallocArraySize(), fix bug in the case that array size is the product of 2 constants.

Extend isArrayMalloc() and getMallocArraySize() to handle case where malloc is used as char array.

Ensure that ArraySize in LowerAllocations::runOnBasicBlock() is correct type.

Extend Instruction::isSafeToSpeculativelyExecute() to handle malloc calls.

Add verification for malloc calls.

Reviewed by Dan Gohman.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82257 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-18 19:20:02 +00:00
Daniel Dunbar
e443525f35 Update CMake.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82097 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-17 00:06:48 +00:00
Dan Gohman
6f65d79750 Add a new pass for doing late hoisting of floating-point and vector
constants out of loops. These aren't covered by the regular LICM
pass, because in LLVM IR constants don't require separate
instructions. They're not always covered by the MachineLICM pass
either, because it doesn't know how to unfold folded constant-pool
loads. This is somewhat experimental at this point, and off by
default.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82076 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-16 20:25:11 +00:00
Dan Gohman
9ad29207db Change FoldPHIArgBinOpIntoPHI to decline folding if it would introduce two
phis, similar to the FoldPHIArgGEPIntoPHI change.

Also, delete some comments that don't reflect the code.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82053 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-16 16:50:24 +00:00
Andreas Neustifter
9233272679 Reapplied r81355 with the problems fixed.
(See http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/086737.html and
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090907/086746.html)


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82039 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-16 11:35:50 +00:00
Andreas Neustifter
ad80981a10 Preserve ProfileInfo during CodeGenPrepare.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82034 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-16 09:26:52 +00:00
Dan Gohman
b6c3385a2d Don't sink gep operators through phi nodes if the result would require
more than one phi, since that leads to higher register pressure on
entry to the phi. This is especially problematic when the phi is in
a loop header, as it increases register pressure throughout the loop.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81993 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-16 02:01:52 +00:00
Nick Lewycky
59136251f3 Add more newlines to make up for the ones removed from the end of instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81851 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-15 07:08:25 +00:00
Chris Lattner
a51c39cc32 add a new CallGraphNode::replaceCallEdge method and use it from
argpromote to avoid invalidating an iterator.  This fixes PR4977.
All clang tests now pass with expensive checking (on my system 
at least).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81843 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-15 05:40:35 +00:00
Chris Lattner
ff1147072a add newline to debug dump
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81840 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-15 05:14:57 +00:00
Dan Gohman
264d245851 When extending a memset range past the front, set the alignment of the
memset region to the alignment of the new start address.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81810 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-14 23:39:10 +00:00
Dan Gohman
e6992f728a Factor out the code for checking that all indices in a getelementptr are
within the notional bounds of the static type of the getelementptr (which
is not the same as "inbounds") from GlobalOpt into a utility routine,
and use it in ConstantFold.cpp to check whether there are any mis-behaved
indices.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81478 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-10 23:37:55 +00:00
Nick Lewycky
fd5249e15b Correctly handle the case where a comparison is created in one BasicBlock and
used by a terminator in another.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81437 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-10 07:02:09 +00:00
Evan Cheng
fabcb9127f Add malloc call utility functions. Patch by Victor Hernandez.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81426 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-10 04:36:43 +00:00
Dan Gohman
c292caf55c Fix SplitCriticalEdge to properly update LCSSA form when splitting a
loop exit edge -- new PHIs may be needed not only for the additional
splits that are made to preserve LoopSimplify form, but also for the
original split. Factor out the code that inserts new PHIs so that it
can be used for both. Remove LoopRotation.cpp's code for manually
updating LCSSA form, as it is now redundant. This fixes PR4934.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81363 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-09 18:18:18 +00:00
Mike Stump
c02f0d7e4b Reflow comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81361 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-09 17:57:16 +00:00
Andreas Neustifter
ff5dfdff56 Preserve ProfileInfo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81360 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-09 17:53:39 +00:00
Dan Gohman
842dee44b5 Fix an 80-column violation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81354 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-09 17:17:19 +00:00
Chris Lattner
273a488cdf revert r81335, which breaks the build.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81347 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-09 16:00:57 +00:00
Andreas Neustifter
1f3b00272a Updated ProfileInfo to have clean seperation between different sentinels.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81335 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-09 12:48:26 +00:00
Owen Anderson
a0ecbe7821 Fix PR4909, patch by Jakub Staszak.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81250 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-08 19:53:15 +00:00
Chris Lattner
275a6d6639 remove an extremely dubious instcombine transformation of
extractelement(load).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81239 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-08 18:48:01 +00:00
Dan Gohman
5c89b5240c Re-apply r80926, with fixes: keep the domtree informed of new blocks
that get created during loop unswitching, and fix SplitBlockPredecessors'
LCSSA updating code to create new PHIs instead of trying to just move
existing ones.

Also, optimize Loop::verifyLoop, since it gets called a lot. Use
searches on a sorted list of blocks instead of calling the "contains"
function, as is done in other places in the Loop class, since "contains"
does a linear search. Also, don't call verifyLoop from LoopSimplify or
LCSSA, as the PassManager is already calling verifyLoop as part of
LoopInfo's verifyAnalysis.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81221 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-08 15:45:00 +00:00
Chris Lattner
687bc9bfde remove a turd
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81186 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-08 03:47:41 +00:00
Chris Lattner
4cb81bdd56 instcombine transforms vector loads that are only used by
extractelement operations into a bitcast of the pointer,
then a gep, then a scalar load.  Disable this when the vector
only has one element, because it leads to infinite loops in
instcombine (PR4908).

This transformation seems like a really bad idea to me, as it
will likely disable CSE of vector load/stores etc and can be
better done in the code generator when profitable.  This
goes all the way back to the first days of packed types,
r25299 specifically.

I'll let those people who care about the performance of vector
code decide what to do with this.





git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81185 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-08 03:44:51 +00:00
Chris Lattner
ff1e98c72a Fix PR4882, by making MemCpyOpt not dereference removed stores to get the
context for the newly created operations.
 
Patch by Jakub Staszak!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81175 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-08 00:27:14 +00:00
Dan Gohman
f8dbee7cea Reappy r80998, now that the GlobalOpt bug that it exposed on MiniSAT is fixed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81172 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-07 23:54:19 +00:00
Dan Gohman
80bdc967e9 Don't commit stores with addresses that have indices that are not
compile-time constant integers or that are out of bounds for their
corresponding static array types. These can cause aliasing that
GlobalOpt assumes won't happen.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81165 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-07 22:44:55 +00:00