Commit Graph

863 Commits

Author SHA1 Message Date
Dale Johannesen 9d559cfff5 Tone down an overzealous optimization.
llvm-svn: 42582
2007-10-03 17:45:27 +00:00
Duncan Sands d31649bc59 Improve comment.
llvm-svn: 42132
2007-09-19 10:25:38 +00:00
Duncan Sands 56df7dec2b A global variable with external weak linkage can be null, while
an alias could alias such a global variable.

llvm-svn: 42130
2007-09-19 10:10:31 +00:00
Dan Gohman 2ac2652779 Instcombine x-((x/y)*y) into a remainder operator.
llvm-svn: 42035
2007-09-17 17:31:57 +00:00
Duncan Sands 6d5da71288 Factor the trampoline transformation into a subroutine.
llvm-svn: 42021
2007-09-17 10:26:40 +00:00
Dale Johannesen 98d3a08d8f Remove the assumption that FP's are either float or
double from some of the many places in the optimizers
it appears, and do something reasonable with x86
long double.
Make APInt::dump() public, remove newline, use it to
dump ConstantSDNode's.
Allow APFloats in FoldingSet.
Expand X86 backend handling of long doubles (conversions
to/from int, mostly).

llvm-svn: 41967
2007-09-14 22:26:36 +00:00
Chris Lattner d9111b88d1 silence a bogus gcc warning.
llvm-svn: 41949
2007-09-14 03:07:24 +00:00
Duncan Sands 9204663bcb Turn calls to trampolines into calls to the underlying
nested function.

llvm-svn: 41844
2007-09-11 14:35:41 +00:00
Chris Lattner e804567cd8 remove some dead code, this is handled by constant folding.
llvm-svn: 41819
2007-09-10 23:46:29 +00:00
Chris Lattner 85a51e0060 Don't zap back to back volatile load/stores
llvm-svn: 41759
2007-09-07 05:33:03 +00:00
Dale Johannesen bed9dc423c Next round of APFloat changes.
Use APFloat in UpgradeParser and AsmParser.
Change all references to ConstantFP to use the
APFloat interface rather than double.  Remove
the ConstantFP double interfaces.
Use APFloat functions for constant folding arithmetic
and comparisons.
(There are still way too many places APFloat is
just a wrapper around host float/double, but we're
getting there.)

llvm-svn: 41747
2007-09-06 18:13:44 +00:00
Nick Lewycky 0c5c47944a Use isTrueWhenEqual. Thanks Chris!
llvm-svn: 41741
2007-09-06 02:40:25 +00:00
Nick Lewycky b0b066eaaa When the two operands of an icmp are equal, there are five possible predicates
that would make the icmp true. Fixes PR1637.

llvm-svn: 41740
2007-09-06 01:10:22 +00:00
Chuck Rose III 2320323647 Forgot to obey 80 column rule. Fixing that.
llvm-svn: 41725
2007-09-05 20:36:41 +00:00
Chuck Rose III e58572233d Added default parameters to GetElementPtrInstr constructor call. Visual Studio 2k5 was getting confused and was unable to compile it. Suspected compiler error.
llvm-svn: 41721
2007-09-05 16:54:38 +00:00
David Greene c656cbb8c2 Update GEP constructors to use an iterator interface to fix
GLIBCXX_DEBUG issues.

llvm-svn: 41697
2007-09-04 15:46:09 +00:00
Chris Lattner 0e258b8518 Cut off crazy computation. This helps PR1622 slightly.
llvm-svn: 41522
2007-08-28 04:23:55 +00:00
David Greene 703623d571 Update InvokeInst to work like CallInst
llvm-svn: 41506
2007-08-27 19:04:21 +00:00
Chris Lattner 99c8ee2977 Transform a load from an undef/zero global into an undef/global even if we
have complex pointer manipulation going on.  This allows us to compile
stuff like this:

__m128i foo(__m128i x){
                static const unsigned int c_0[4] = { 0, 0, 0, 0 };
                __m128i v_Zero = _mm_loadu_si128((__m128i*)c_0);
                x  = _mm_unpacklo_epi8(x,  v_Zero);
                return x;
}

into:

_foo:
        xorps   %xmm1, %xmm1
        punpcklbw       %xmm1, %xmm0
        ret

llvm-svn: 41022
2007-08-11 18:48:48 +00:00
Chris Lattner a8e4b4bc7b when we see a unaligned load from an insufficiently aligned global or
alloca, increase the alignment of the load, turning it into an aligned load.

This allows us to compile:

#include <xmmintrin.h>
__m128i foo(__m128i x){
 static const unsigned int c_0[4] = { 0, 0, 0, 0 };
	  __m128i v_Zero = _mm_loadu_si128((__m128i*)c_0);
  x  = _mm_unpacklo_epi8(x,  v_Zero);
  return x;
}

into:

_foo:
	punpcklbw	_c_0.5944, %xmm0
	ret
	.data
	.lcomm	_c_0.5944,16,4		# c_0.5944

instead of:

_foo:
	movdqu	_c_0.5944, %xmm1
	punpcklbw	%xmm1, %xmm0
	ret
	.data
	.lcomm	_c_0.5944,16,2		# c_0.5944

llvm-svn: 40971
2007-08-09 19:05:49 +00:00
Nick Lewycky 8052019a20 It's safe to fold not of fcmp.
llvm-svn: 40870
2007-08-06 20:04:16 +00:00
Chris Lattner f0da7975ea at the end of instcombine, explicitly clear WorklistMap.
This shrinks it down to something small.  On the testcase
from PR1432, this speeds up instcombine from 0.7959s to 0.5000s,
(59%)

llvm-svn: 40840
2007-08-05 08:47:58 +00:00
Chandler Carruth 7132e00de7 This is the patch to provide clean intrinsic function overloading support in LLVM. It cleans up the intrinsic definitions and generally smooths the process for more complicated intrinsic writing. It will be used by the upcoming atomic intrinsics as well as vector and float intrinsics in the future.
This also changes the syntax for llvm.bswap, llvm.part.set, llvm.part.select, and llvm.ct* intrinsics. They are automatically upgraded by both the LLVM ASM reader and the bitcode reader. The test cases have been updated, with special tests added to ensure the automatic upgrading is supported.

llvm-svn: 40807
2007-08-04 01:51:18 +00:00
Chris Lattner dc2cf228ce Replacing a cast with another one does not reduce the number of
casts in the input.

llvm-svn: 40741
2007-08-02 17:23:38 +00:00
Chris Lattner 222b214be7 Disable an xform that causes an infinite loop. This fixes PR1594
llvm-svn: 40739
2007-08-02 16:56:32 +00:00
Chris Lattner 2740694450 wrap some long lines. Major offenders that are left include
gvn, gvnpre, dse, and predsimplify.  To see these, use:

  make check-line-length

llvm-svn: 40738
2007-08-02 16:53:43 +00:00
Chris Lattner b0418fc607 Enhance instcombine to be more aggressive about folding casts of
operations of casts.  This implements InstCombine/zext-fold.ll

llvm-svn: 40726
2007-08-02 06:11:14 +00:00
David Greene 17a5dfe6f7 New CallInst interface to address GLIBCXX_DEBUG errors caused by
indexing an empty std::vector.

Updates to all clients.

llvm-svn: 40660
2007-08-01 03:43:44 +00:00
Lauro Ramos Venancio 549e775e67 Fix a bug in GetKnownAlignment of packed structs.
llvm-svn: 40649
2007-07-31 20:13:21 +00:00
Reid Spencer dff9d69cfb Fix a typo/thinko.
llvm-svn: 40599
2007-07-30 19:53:57 +00:00
Chris Lattner 4512cd2cab completely remove a transformation that is unsafe in the face of
undefs.

llvm-svn: 40439
2007-07-23 17:10:17 +00:00
Devang Patel 5e39293e62 Apply temporary work around to fix llvm mis-compilation
reported in PR 1556.

llvm-svn: 40133
2007-07-21 00:34:29 +00:00
Chris Lattner d82e4a19cc this xform is already done by the constant folder.
llvm-svn: 40124
2007-07-20 22:06:41 +00:00
Dan Gohman e31a61eeca Optimize alignment of loads and stores.
llvm-svn: 40102
2007-07-20 16:34:21 +00:00
Dan Gohman 06c60b6032 Fix comments about vectors to use the current wording.
llvm-svn: 39921
2007-07-16 14:29:03 +00:00
Chris Lattner 640fd5124d Repair a regression in Transforms/InstCombine/mul.ll that Reid noticed.
llvm-svn: 39896
2007-07-16 04:15:34 +00:00
Chris Lattner d4fef8dbca Implement shift-simplify.ll:test[45].
First teach instcombine that sign bit checks only demand the 
sign bit, this allows simplify demanded bits to hack on 
expressions better.

Second, teach instcombine that ashr is useless if only the
sign bit is demanded.

llvm-svn: 39880
2007-07-15 20:54:51 +00:00
Chris Lattner 06205d5567 Implement shift-simplify.ll:test3, turning:
(X << 31) <s 0  --> (X&1) != 0

This happens dozens of times in the CFE.

llvm-svn: 39879
2007-07-15 20:42:37 +00:00
Chris Lattner fb032b176b Significantly improve the documentation of the instcombine divide/compare
transformation.  Also, keep track of which end of the integer interval overflows
occur on.  This fixes Transforms/InstCombine/2007-06-21-DivCompareMiscomp.ll
and rdar://5278853, a miscompilation of perl.

llvm-svn: 37692
2007-06-21 18:11:19 +00:00
Chris Lattner 3bbec59e8b refactor a bunch of code out of visitICmpInstWithInstAndIntCst into its own
routine.

llvm-svn: 37679
2007-06-20 23:46:26 +00:00
Chris Lattner 09a33a4f64 silence a bogus warning Duraid ran into.
llvm-svn: 37649
2007-06-19 05:43:49 +00:00
Chris Lattner 373389260f Generalize many transforms to work on ~ of vectors in addition to ~ of
integer ops.  This implements Transforms/InstCombine/and-or-not.ll
test3/test4, and finishes off PR1510

llvm-svn: 37589
2007-06-15 06:23:19 +00:00
Chris Lattner 481e28b1f5 Implement two xforms:
1. ~(~X | Y) === (X & ~Y)
2. (A|B) & ~(A&B) -> A^B

This allows us to transform  ~(~(a|b) | (a&b)) -> a^b.

This implements PR1510 for scalar values.

llvm-svn: 37584
2007-06-15 05:58:24 +00:00
Chris Lattner f14e5175ed delete some obviously dead vector operations, which deletes a few thousand
operations from Duraids example.

llvm-svn: 37582
2007-06-15 05:26:55 +00:00
Lauro Ramos Venancio 368e8872db Fix PR1499.
llvm-svn: 37472
2007-06-06 17:08:48 +00:00
Chris Lattner f79577d314 fix a miscompilation when passing a float through varargs
llvm-svn: 37297
2007-05-23 01:17:04 +00:00
Chris Lattner a655a157a0 Fix Transforms/InstCombine/2007-05-18-CastFoldBug.ll, a bug that devastates
objc code due to the way the FE lowers objc message sends.

llvm-svn: 37256
2007-05-19 06:51:32 +00:00
Chris Lattner 234f96daa8 Fix Transforms/InstCombine/2007-05-14-Crash.ll
llvm-svn: 37057
2007-05-15 00:16:00 +00:00
Dan Gohman b5650ebd6a Fix typos.
llvm-svn: 36994
2007-05-11 21:10:54 +00:00
Chris Lattner 600db3eb96 fix regressions from my previous checking, including
Transforms/InstCombine/2006-12-08-ICmp-Combining.ll

llvm-svn: 36989
2007-05-11 16:58:45 +00:00
Chris Lattner fe2b44de9f fix Transforms/InstCombine/2007-05-10-icmp-or.ll
llvm-svn: 36984
2007-05-11 05:55:56 +00:00
Nick Lewycky e7da2d6ac3 Fix typo in comment.
llvm-svn: 36873
2007-05-06 13:37:16 +00:00
Chris Lattner 9b35b3e863 Fix a bug in my previous patch
llvm-svn: 36857
2007-05-06 07:24:03 +00:00
Chris Lattner 5aa73fe34c Implement Transforms/InstCombine/cast_ptr.ll
llvm-svn: 36809
2007-05-05 22:41:33 +00:00
Chris Lattner 361e981415 wrap long lines
llvm-svn: 36807
2007-05-05 22:32:24 +00:00
Chris Lattner 5c827bda0d Fix InstCombine/2007-05-04-Crash.ll and PR1384
llvm-svn: 36775
2007-05-05 01:59:31 +00:00
Devang Patel 8c78a0bff0 Drop 'const'
llvm-svn: 36662
2007-05-03 01:11:54 +00:00
Devang Patel e95c6ad802 Use 'static const char' instead of 'static const int'.
Due to darwin gcc bug, one version of darwin linker coalesces
static const int, which defauts PassID based pass identification.

llvm-svn: 36652
2007-05-02 21:39:20 +00:00
Devang Patel 09f162ca6a Do not use typeinfo to identify pass in pass manager.
llvm-svn: 36632
2007-05-01 21:15:47 +00:00
Chris Lattner 089e35cc57 fix a bug triggered by 403.gcc
llvm-svn: 36527
2007-04-28 05:27:36 +00:00
Chris Lattner 6e880871e9 Fix several latent bugs in EmitGEPOffset that didn't manifest with its
previous clients.  This fixes MallocBench/gs

llvm-svn: 36525
2007-04-28 04:52:43 +00:00
Chris Lattner c753800800 uhn zap cvs
llvm-svn: 36523
2007-04-28 03:50:56 +00:00
Chris Lattner acbf6a401d Implement PR1345 and Transforms/InstCombine/bitcast-gep.ll
llvm-svn: 36521
2007-04-28 00:57:34 +00:00
Chris Lattner 1db224db92 refactor some code relating to pointer cast xforms, pulling it out of the codepath
for unrelated casts.

llvm-svn: 36511
2007-04-27 17:44:50 +00:00
Zhou Sheng aafe4e216e Make use of ConstantInt::isZero instead of ConstantInt::isNullValue.
llvm-svn: 36261
2007-04-19 05:39:12 +00:00
Chris Lattner 4a6e0cbd41 Extend store merging to support the 'if/then' version in addition to if/then/else.
This sinks the two stores in this example into a single store in cond_next.  In this
case, it allows elimination of the load as well:

        store double 0.000000e+00, double* @s.3060
        %tmp3 = fcmp ogt double %tmp1, 5.000000e-01             ; <i1> [#uses=1]
        br i1 %tmp3, label %cond_true, label %cond_next
cond_true:              ; preds = %entry
        store double 1.000000e+00, double* @s.3060
        br label %cond_next
cond_next:              ; preds = %entry, %cond_true
        %tmp6 = load double* @s.3060            ; <double> [#uses=1]

This implements Transforms/InstCombine/store-merge.ll:test2

llvm-svn: 36040
2007-04-15 01:02:18 +00:00
Chris Lattner 14a251b937 refactor some code, no functionality change.
llvm-svn: 36037
2007-04-15 00:07:55 +00:00
Chris Lattner 28d921d04f fix long lines
llvm-svn: 36031
2007-04-14 23:32:02 +00:00
Chris Lattner 7bfdd0abe1 Implement Transforms/InstCombine/vec_extract_elt.ll, transforming:
define i32 @test(float %f) {
        %tmp7 = insertelement <4 x float> undef, float %f, i32 0
        %tmp17 = bitcast <4 x float> %tmp7 to <4 x i32>
        %tmp19 = extractelement <4 x i32> %tmp17, i32 0
        ret i32 %tmp19
}

into:

define i32 @test(float %f) {
        %tmp19 = bitcast float %f to i32                ; <i32> [#uses=1]
        ret i32 %tmp19
}

On PPC, this is the difference between:

_test:
        mfspr r2, 256
        oris r3, r2, 8192
        mtspr 256, r3
        stfs f1, -16(r1)
        addi r3, r1, -16
        addi r4, r1, -32
        lvx v2, 0, r3
        stvx v2, 0, r4
        lwz r3, -32(r1)
        mtspr 256, r2
        blr

and:

_test:
        stfs f1, -4(r1)
        nop
        nop
        nop
        lwz r3, -4(r1)
        blr

llvm-svn: 36025
2007-04-14 23:02:14 +00:00
Chris Lattner b37fb6a0da Implement InstCombine/vec_demanded_elts.ll:test2. This allows us to turn
unsigned test(float f) {
 return _mm_cvtsi128_si32( (__m128i) _mm_set_ss( f*f ));
}

into:

_test:
        movss 4(%esp), %xmm0
        mulss %xmm0, %xmm0
        movd %xmm0, %eax
        ret

instead of:

_test:
        movss 4(%esp), %xmm0
        mulss %xmm0, %xmm0
        xorps %xmm1, %xmm1
        movss %xmm0, %xmm1
        movd %xmm1, %eax
        ret

GCC gets:

_test:
        subl    $28, %esp
        movss   32(%esp), %xmm0
        mulss   %xmm0, %xmm0
        xorps   %xmm1, %xmm1
        movss   %xmm0, %xmm1
        movaps  %xmm1, %xmm0
        movd    %xmm0, 12(%esp)
        movl    12(%esp), %eax
        addl    $28, %esp
        ret

llvm-svn: 36020
2007-04-14 22:29:23 +00:00
Chris Lattner efb33d28c6 Implement PR1201 and test/Transforms/InstCombine/malloc-free-delete.ll
llvm-svn: 35981
2007-04-14 00:20:02 +00:00
Chris Lattner 74ff60ff84 Turn stuff like:
icmp slt i32 %X, 0              ; <i1>:0 [#uses=1]
        sext i1 %0 to i32               ; <i32>:1 [#uses=1]

into:

        %X.lobit = ashr i32 %X, 31              ; <i32> [#uses=1]

This implements InstCombine/icmp.ll:test[34]

llvm-svn: 35891
2007-04-11 06:57:46 +00:00
Chris Lattner d0f7942e23 Simplify some comparisons to arithmetic, this implements:
Transforms/InstCombine/icmp.ll

llvm-svn: 35890
2007-04-11 06:53:04 +00:00
Chris Lattner 20f2372a7c canonicalize (x <u 2147483648) -> (x >s -1) and (x >u 2147483647) -> (x <s 0)
llvm-svn: 35886
2007-04-11 06:12:58 +00:00
Chris Lattner 7ddbff090a fix a miscompilation of:
define i32 @test(i32 %X) {
entry:
        %Y = and i32 %X, 4              ; <i32> [#uses=1]
        icmp eq i32 %Y, 0               ; <i1>:0 [#uses=1]
        sext i1 %0 to i32               ; <i32>:1 [#uses=1]
        ret i32 %1
}

by moving code out of commonIntCastTransforms into visitZExt.  Simplify the
APInt gymnastics in it etc.

llvm-svn: 35885
2007-04-11 05:45:39 +00:00
Chris Lattner 467b69cabb Strengthen the boundary conditions of this fold, implementing
InstCombine/set.ll:test25

llvm-svn: 35852
2007-04-09 23:52:13 +00:00
Chris Lattner a87c9f6114 Fix PR1304 and Transforms/InstCombine/2007-04-08-SingleEltVectorCrash.ll
llvm-svn: 35792
2007-04-09 01:37:55 +00:00
Chris Lattner 4ca9cbb170 Eliminate useless insertelement instructions. This implements
Transforms/InstCombine/vec_insertelt.ll and fixes PR1286.

We now compile the code from that bug into:

_foo:
        movl 4(%esp), %eax
        movdqa (%eax), %xmm0
        movl 8(%esp), %ecx
        psllw (%ecx), %xmm0
        movdqa %xmm0, (%eax)
        ret

instead of:

_foo:
        subl $4, %esp
        movl %ebp, (%esp)
        movl %esp, %ebp
        movl 12(%ebp), %eax
        movdqa (%eax), %xmm0
        #IMPLICIT_DEF %eax
        pinsrw $2, %eax, %xmm0
        xorl %ecx, %ecx
        pinsrw $3, %ecx, %xmm0
        pinsrw $4, %eax, %xmm0
        pinsrw $5, %ecx, %xmm0
        pinsrw $6, %eax, %xmm0
        pinsrw $7, %ecx, %xmm0
        movl 8(%ebp), %eax
        movdqa (%eax), %xmm1
        psllw %xmm0, %xmm1
        movdqa %xmm1, (%eax)
        movl %ebp, %esp
        popl %ebp
        ret

woo :)

llvm-svn: 35788
2007-04-09 01:11:16 +00:00
Chris Lattner c8d3788f71 reenable this xform, whoops :)
llvm-svn: 35765
2007-04-08 08:01:49 +00:00
Chris Lattner 7621a031d8 Fix regression on Instcombine/apint-or2.ll
llvm-svn: 35763
2007-04-08 07:55:22 +00:00
Chris Lattner 1150df9cc4 Generalize the code that handles (A&B)|(A&C) to work where B/C are not constants.
Add a new xform to simplify (A&B)|(~A&C).  THis implements InstCombine/or2.ll:test1

llvm-svn: 35760
2007-04-08 07:47:01 +00:00
Chris Lattner 3dbe65f80a implement Transforms/InstCombine/malloc2.ll and PR1313
llvm-svn: 35700
2007-04-06 18:57:34 +00:00
Dale Johannesen 7c2001d014 Prevent transformConstExprCastCall from generating conversions that assert
elsewhere.

llvm-svn: 35668
2007-04-04 19:16:42 +00:00
Jeff Cohen 5a1c750f31 Fix 2007-04-04-BadFoldBitcastIntoMalloc.ll
llvm-svn: 35665
2007-04-04 16:58:57 +00:00
Duncan Sands f01a47c93c Fix comment.
llvm-svn: 35655
2007-04-04 06:42:45 +00:00
Chris Lattner e5bbb3cb1a Fix a bug I introduced with my patch yesterday which broke Qt (I converted
some constant exprs to apints).

Thanks to Anton for tracking down a small testcase that triggered this!

llvm-svn: 35633
2007-04-03 23:29:39 +00:00
Chris Lattner a74deafb13 reinstate the previous two patches, with a bugfix :)
ldecod now passes.

llvm-svn: 35626
2007-04-03 17:43:25 +00:00
Evan Cheng 7511fa280d Reverting back to 1.723. The last two commits broke JM (and possibily others) on ARM.
llvm-svn: 35620
2007-04-03 08:11:50 +00:00
Chris Lattner 64c764cebc Split a whole ton of code out of visitICmpInst into visitICmpInstWithInstAndIntCst.
llvm-svn: 35614
2007-04-03 04:46:52 +00:00
Chris Lattner 8b2ec5f506 Fix PR1253 and xor2.ll:test[01]
llvm-svn: 35612
2007-04-03 01:47:41 +00:00
Zhou Sheng 9bc8ab100d 1. Make use of APInt operation instead of using ConstantExpr::getXXX.
2. Use cheaper APInt methods.

llvm-svn: 35594
2007-04-02 13:45:30 +00:00
Zhou Sheng 56cda95658 Use uint32_t for bitwidth instead of unsigned.
llvm-svn: 35593
2007-04-02 08:20:41 +00:00
Chris Lattner 9d5aacee92 Wrap long line
llvm-svn: 35588
2007-04-02 05:48:58 +00:00
Chris Lattner 50490d54f2 use more obvious function name.
llvm-svn: 35587
2007-04-02 05:42:22 +00:00
Chris Lattner b24acc7bee simplify (x+c)^signbit as (x+c+signbit), pointed out by PR1288. This implements
test/Transforms/InstCombine/xor.ll:test28

llvm-svn: 35584
2007-04-02 05:36:22 +00:00
Chris Lattner c3eeb42809 simplify this code, make it work for ap ints
llvm-svn: 35561
2007-04-01 20:57:36 +00:00
Zhou Sheng 150f3bbab2 Avoid unnecessary APInt construction.
llvm-svn: 35555
2007-04-01 17:13:37 +00:00
Reid Spencer 6bba6c8143 For PR1297:
Support overloaded intrinsics bswap, ctpop, cttz, ctlz.

llvm-svn: 35547
2007-04-01 07:35:23 +00:00
Chris Lattner 0427799531 Fix InstCombine/2007-03-31-InfiniteLoop.ll
llvm-svn: 35536
2007-04-01 05:36:37 +00:00
Zhou Sheng 82c42284f4 Delete dead code.
llvm-svn: 35525
2007-03-31 02:50:26 +00:00