Commit Graph

884 Commits

Author SHA1 Message Date
Chris Lattner
e10c0b9d66 Transforming -A + -B --> -(A + B) isn't safe for FP, thanks
to Dale for noticing this!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47276 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-18 17:50:16 +00:00
Chris Lattner
bf1d8a7099 optimize away stackrestore calls that have no intervening alloca or call.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47258 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-18 06:12:38 +00:00
Chris Lattner
dd12f96c5e Fold (-x + -y) -> -(x+y) which promotes better association, fixing
the second half of PR2047


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47244 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-17 21:03:36 +00:00
Dan Gohman
7649127ec1 Rename APInt's isPositive to isNonNegative, to reflect what it
actually does.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47090 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-13 22:09:18 +00:00
Chris Lattner
7c95debb49 Fix a bug compiling PR1978 (perhaps not the only one though) which
was incorrectly simplifying "x == (gep x, 1, i)" into false, even 
though i could be negative.  As it turns out, all the code to 
handle this already existed, we just need to disable the incorrect
optimization case and let the general case handle it.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46739 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-05 04:45:32 +00:00
Nick Lewycky
5be2920149 There are some cases where icmp(add) can be folded into a new icmp. Handle them.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46687 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-03 16:33:09 +00:00
Nick Lewycky
0c2c3f6e81 Hack on vectors too.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46684 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-03 08:19:11 +00:00
Nick Lewycky
b6eabff169 Fold away one multiply in instcombine. This would normally be caught in
reassociate anyways, but they could be generated during instcombine's run.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46683 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-03 07:42:09 +00:00
Chris Lattner
2454a2e0c3 eliminate additions of 0.0 when they are obviously dead. This has to be careful to
avoid turning -0.0 + 0.0 -> -0.0 which is incorrect.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46499 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-29 06:52:45 +00:00
Nick Lewycky
4189a538e4 Handle some more combinations of extend and icmp. Fixes PR1940.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46431 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-28 03:48:02 +00:00
Chris Lattner
25feae555d Fix PR1932 by disabling an xform invalid for fdiv.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46429 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-28 00:58:18 +00:00
Chris Lattner
b753065f6d Fold fptrunc(add (fpextend x), (fpextend y)) -> add(x,y), as GCC does.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46406 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-27 05:29:54 +00:00
Nick Lewycky
6410510224 Enable the fix I just checked in, silly me.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46247 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-22 05:42:02 +00:00
Nick Lewycky
e6b0c0036a Multiply can be evaluated in a different type, so long as the target type has
a smaller bitwidth.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46244 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-22 05:08:48 +00:00
Duncan Sands
b0c9b93bb4 I noticed that the trampoline straightening transformation could
drop attributes on varargs call arguments.  Also, it could generate
invalid IR if the transformed call already had the 'nest' attribute
somewhere (this can never happen for code coming from llvm-gcc,
but it's a theoretical possibility).  Fix both problems.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45973 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-14 19:52:09 +00:00
Chris Lattner
37ac608623 Turn a memcpy from a double* into a load/store of double instead of
a load/store of i64.  The later prevents promotion/scalarrepl of the
source and dest in many cases.

This fixes the 300% performance regression of the byval stuff on 
stepanov_v1p2.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45945 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-14 00:28:35 +00:00
Chris Lattner
f497b028b8 factor memcpy/memmove simplification out to its own SimplifyMemTransfer
method, no functionality change.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45944 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-13 23:50:23 +00:00
Chris Lattner
88b39744ec simplify some code. If we can infer alignment for source and dest that are
greater than memcpy alignment, and if we lower to load/store, use the best 
alignment info we have.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45943 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-13 22:30:28 +00:00
Chris Lattner
6d0339d451 simplify some code by adding a InsertBitCastBefore method,
make memmove->memcpy conversion a bit simpler.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45942 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-13 22:23:22 +00:00
Chris Lattner
ee2b7a4530 Fix PR1907, a nasty miscompilation because instcombine didn't
realize that ne & sgt  was a signed comparison (it was only 
looking at whether the left compare was signed).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45937 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-13 20:59:02 +00:00
Duncan Sands
e1e520f601 When turning a call to a bitcast function into a direct call,
if this becomes a varargs call then deal correctly with any
parameter attributes on the newly vararg call arguments.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45931 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-13 08:02:44 +00:00
Chris Lattner
f9d9e45225 Implement PR1795, an instcombine hack for forming GEPs with integer pointer arithmetic.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45745 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-08 07:23:51 +00:00
Duncan Sands
6c3470efdc Small cleanup for handling of type/parameter attribute
incompatibility.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45704 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-07 17:16:06 +00:00
Duncan Sands
ad9a9e1559 The transform that tries to turn calls to bitcast functions into
direct calls bails out unless caller and callee have essentially
equivalent parameter attributes.  This is illogical - the callee's
attributes should be of no relevance here.  Rework the logic, which
incidentally fixes a crash when removed arguments have attributes.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45658 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-06 18:27:01 +00:00
Duncan Sands
a9d0c9dc58 When transforming a call to a bitcast function into
a direct call with cast parameters and cast return
value (if any), instcombine was prepared to cast any
non-void return value into any other, whether castable
or not.  Add a new predicate for testing whether casting
is valid, and check it both for the return value and
(as a cleanup) for the parameters.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45657 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-06 10:12:28 +00:00
Chris Lattner
5a62d0a038 remove a couple more unsafe xforms in the face of overflow.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45613 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-05 01:22:42 +00:00
Chris Lattner
b656601943 remove the (x-y) < 0 comparison xform, it miscompiles
things that are not equality comparisons, for example:
   (2147479553+4096)-2147479553 < 0    !=   (2147479553+4096) < 2147479553


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45612 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-05 01:18:20 +00:00
Chris Lattner
4ee451de36 Remove attribution from file headers, per discussion on llvmdev.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45418 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-29 20:36:04 +00:00
Christopher Lamb
b15147ea4c Disable null pointer folding transforms for non-generic address spaces. This should probably be a target-specific predicate based on address space. That way for targets where this isn't applicable the predicate can be optimized away.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45403 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-29 07:56:53 +00:00
Owen Anderson
f5783f825d Repair a transform that Chris noticed a bug in. Thanks to Nicholas for pointing out my stupid mistakes when writing this patch. :-)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45384 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-28 07:42:12 +00:00
Chris Lattner
c48b138869 disable this instcombine xform, it miscompiles:
define i32 @main() {
entry:
	%z = alloca i32		; <i32*> [#uses=2]
	store i32 0, i32* %z
	%tmp = load i32* %z		; <i32> [#uses=1]
	%sub = sub i32 %tmp, 1		; <i32> [#uses=1]
	%cmp = icmp ult i32 %sub, 0		; <i1> [#uses=1]
	%retval = select i1 %cmp, i32 1, i32 0		; <i32> [#uses=1]
	ret i32 %retval
}

into ret 1, instead of ret 0.

Christopher, please investigate.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45383 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-28 06:24:31 +00:00
Chris Lattner
8999dd3c68 implement InstCombine/shift-trunc-shift.ll. This allows
us to compile:
#include <math.h>
int t1(double d) { return signbit(d); }

into:

_t1:
	movd	%xmm0, %rax
	shrq	$63, %rax
	ret

instead of:

_t1:
	movd	%xmm0, %rax
	shrq	$32, %rax
	shrl	$31, %eax
	ret

on x86-64.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45311 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-22 09:07:47 +00:00
Christopher Lamb
103e1a3118 Implement review feedback, including additional transforms
(icmp slt (sub A B) 1) -> (icmp sle A B)
icmp sgt (sub A B) -1) -> (icmp sge A B)

and add testcase.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45256 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-20 07:21:11 +00:00
Chris Lattner
42790486a8 simplify this code with the new m_Zero() pattern. Make sure the select only
has a single use, and generalize it to not require N to be a constant.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45250 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-20 01:56:58 +00:00
Duncan Sands
f0c3354d99 When inlining through an 'nounwind' call, mark inlined
calls 'nounwind'.  It is important for correct C++
exception handling that nounwind markings do not get
lost, so this transformation is actually needed for
correctness.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45218 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-19 21:13:37 +00:00
Christopher Lamb
7a0678cae6 Fold subtracts into integer compares vs. zero. This improves generate code for this case on X86
from
_foo:
        movl    $99, %ecx
        movl    4(%esp), %eax
        subl    %eax, %ecx
        xorl    %edx, %edx
        testl   %ecx, %ecx
        cmovs   %edx, %eax
        ret

to
_foo:
        xorl    %ecx, %ecx
        movl    4(%esp), %eax
        cmpl    $99, %eax
        cmovg   %ecx, %eax
        ret

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45173 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-18 21:32:20 +00:00
Christopher Lamb
1d87cb1c12 Fix comments
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45170 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-18 20:33:11 +00:00
Christopher Lamb
fd8b1fd7f6 Remove an orthogonal transformation of the selection condition from my most recent submission.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45169 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-18 20:30:28 +00:00
Duncan Sands
2b0e8990ab Rename isNoReturn to doesNotReturn, and isNoUnwind to
doesNotThrow.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45160 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-18 09:59:50 +00:00
Christopher Lamb
f311f53a0f Fix typos.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45159 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-18 09:45:40 +00:00
Christopher Lamb
30f017ad89 Fold certain additions through selects (and their compares) so as to eliminate subtractions. This code is often produced by the SMAX expansion in SCEV.
This implements test/Transforms/InstCombine/2007-12-18-AddSelCmpSub.ll


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45158 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-18 09:34:41 +00:00
Christopher Lamb
43ad6b3e0d Change the PointerType api for creating pointer types. The old functionality of PointerType::get() has become PointerType::getUnqual(), which returns a pointer in the generic address space. The new prototype of PointerType::get() requires both a type and an address space.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45082 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-17 01:12:55 +00:00
Duncan Sands
ece2c04d53 Make instcombine promote inline asm calls to 'nounwind'
calls.  Remove special casing of inline asm from the
inliner.  There is a potential problem: the verifier
rejects invokes of inline asm (not sure why).  If an
asm call is not marked "nounwind" in some .ll, and
instcombine is not run, but the inliner is run, then
an illegal module will be created.  This is bad but
I'm not sure what the best approach is.  I'm tempted
to remove the check in the verifier...


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45073 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-16 15:51:49 +00:00
Wojciech Matyjewicz
ed2232502e 1. "Upgrage" comments.
2. Using zero-extended value of Scale and unsigned division is safe provided
   that Scale doesn't have the sign bit set.
   Previously these 2 instructions:
        %p = bitcast [100 x {i8,i8,i8}]* %x to i8*
        %q = getelementptr i8* %p, i32 -4
   were combined into:
        %q = getelementptr [100 x { i8, i8, i8 }]* %x, i32 0,
               i32 1431655764, i32 0
   what was incorrect.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44936 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-12 15:21:32 +00:00
Chris Lattner
c95ba44b30 simplify some code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44655 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-06 06:25:04 +00:00
Chris Lattner
348f665ee2 move some ashr-specific code out of commonShiftTransforms into visitAShr.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44650 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-06 01:59:46 +00:00
Duncan Sands
dc024674ff Fix PR1146: parameter attributes are longer part of
the function type, instead they belong to functions
and function calls.  This is an updated and slightly
corrected version of Reid Spencer's original patch.
The only known problem is that auto-upgrading of
bitcode files doesn't seem to work properly (see
test/Bitcode/AutoUpgradeIntrinsics.ll).  Hopefully
a bitcode guru (who might that be? :) ) will fix it.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44359 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-27 13:23:08 +00:00
Chris Lattner
cfa597569d Implement PR1822
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44318 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-25 21:27:53 +00:00
Duncan Sands
827cde1c83 Fix PR1816. If a bitcast of a function only exists because of a
trivial difference in function attributes, allow calls to it to
be converted to direct calls.  Based on a patch by Török Edwin.
While there, move the various lists of mutually incompatible
parameters etc out of the verifier and into ParameterAttributes.h.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44315 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-25 14:10:56 +00:00
Chris Lattner
148083a529 add a comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44293 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-23 22:35:18 +00:00