c482a9e31a
Implement Transforms/InstCombine/bswap.ll, turning common shift/and/or bswap
...
idioms into bswap intrinsics.
llvm-svn: 28803
2006-06-15 19:07:26 +00:00
0c4f5a655a
Fix Transforms/LoopUnswitch/2006-06-13-SingleEntryPHI.ll, a loop unswitch
...
bug exposed by the recent lcssa work.
llvm-svn: 28779
2006-06-14 04:46:17 +00:00
fd0a3d6e5c
Reapply my 6/9 changes. The bug Evan saw no longer occurs.
...
llvm-svn: 28759
2006-06-12 21:49:21 +00:00
1b6e310e6f
Back out Owen's 6/9 changes. They broke MultiSource/Benchmarks/Prolangs-C/bison (and perhaps others).
...
llvm-svn: 28747
2006-06-11 09:32:57 +00:00
b1dc1d44f8
Add LCSSA as a requirement for LoopUnswitch, and assert that LoopUnswitch preserves
...
LCSSA.
llvm-svn: 28739
2006-06-09 18:40:32 +00:00
398f70292c
RewriteExpr, either the new PHI node of induction variable or the
...
post-increment value, should be first cast to the appropriated type (to the
type of the common expr). Otherwise, the rewrite of a use based on (common +
iv) may end up with an incorrect type.
llvm-svn: 28735
2006-06-09 00:12:42 +00:00
d4b795902c
Fix a spello in a comment.
...
llvm-svn: 28714
2006-06-07 21:24:10 +00:00
95cebb082f
Fix a bug in a recent patch. This fixes UnitTests/Vector/Altivec/casts.c on
...
PPC/altivec
llvm-svn: 28698
2006-06-06 22:26:02 +00:00
540886f0ae
Remove unneeded hook. Patch by Anton K. Thanks!
...
llvm-svn: 28664
2006-06-02 19:11:46 +00:00
f905a7b994
Silence a -pedantic warning.
...
llvm-svn: 28632
2006-06-01 17:16:21 +00:00
1df0e98ac2
Swap the order of operands created here. For +&|^, the order doesn't matter,
...
but for sub, it really does! Fix fixes a miscompilation of fibheap_cut in
llvmgcc4.
llvm-svn: 28600
2006-05-31 21:14:00 +00:00
dab43b2b0e
Implement Transforms/InstCombine/store.ll:test2.
...
llvm-svn: 28503
2006-05-26 19:19:20 +00:00
0e47716e69
Transform things like (splat(splat)) -> splat
...
llvm-svn: 28490
2006-05-26 00:29:06 +00:00
12249be286
Introduce a helper function that simplifies interpretation of shuffle masks.
...
No functionality change.
llvm-svn: 28489
2006-05-25 23:48:38 +00:00
99155be33f
Turn (cast (shuffle (cast)) -> shuffle (cast) if it reduces the # casts in
...
the program. This exposes more opportunities for the instcombiner, and implements
vec_shuffle.ll:test6
llvm-svn: 28487
2006-05-25 23:24:33 +00:00
83f6578b0c
extract element from a shuffle vector can be trivially turned into an
...
extractelement from the SV's source. This implement vec_shuffle.ll:test[45]
llvm-svn: 28485
2006-05-25 22:53:38 +00:00
d0622b6894
Silence a bogus gcc warning
...
llvm-svn: 28422
2006-05-20 23:14:03 +00:00
e4cb4768fa
Declare that lowerinvoke doesn't interact with other lowering passes.
...
Patch written by Domagoj Babic!
llvm-svn: 28367
2006-05-17 21:05:27 +00:00
18d0438148
Backing out last check-in for now. It's causing an infinite loop gccas lencode.
...
llvm-svn: 28284
2006-05-14 06:46:03 +00:00
3987a8532d
Add/Sub/Mul are safe to promote here as well. Incrementing a single-bit
...
bitfield now gives this code:
_plus:
lwz r2, 0(r3)
rlwimi r2, r2, 0, 1, 31
xoris r2, r2, 32768
stw r2, 0(r3)
blr
instead of this:
_plus:
lwz r2, 0(r3)
srwi r4, r2, 31
slwi r4, r4, 31
addis r4, r4, -32768
rlwimi r2, r4, 0, 0, 0
stw r2, 0(r3)
blr
this can obviously still be improved.
llvm-svn: 28275
2006-05-13 02:16:08 +00:00
1ebbe6a22e
Implement simple promotion for cast elimination in instcombine. This is
...
currently very limited, but can be extended in the future. For example,
we now compile:
uint %test30(uint %c1) {
%c2 = cast uint %c1 to ubyte
%c3 = xor ubyte %c2, 1
%c4 = cast ubyte %c3 to uint
ret uint %c4
}
to:
_xor:
movzbl 4(%esp), %eax
xorl $1, %eax
ret
instead of:
_xor:
movb $1, %al
xorb 4(%esp), %al
movzbl %al, %eax
ret
More impressively, we now compile:
struct B { unsigned bit : 1; };
void xor(struct B *b) { b->bit = b->bit ^ 1; }
To (X86/PPC):
_xor:
movl 4(%esp), %eax
xorl $-2147483648, (%eax)
ret
_xor:
lwz r2, 0(r3)
xoris r2, r2, 32768
stw r2, 0(r3)
blr
instead of (X86/PPC):
_xor:
movl 4(%esp), %eax
movl (%eax), %ecx
movl %ecx, %edx
shrl $31, %edx
# TRUNCATE movb %dl, %dl
xorb $1, %dl
movzbl %dl, %edx
andl $2147483647, %ecx
shll $31, %edx
orl %ecx, %edx
movl %edx, (%eax)
ret
_xor:
lwz r2, 0(r3)
srwi r4, r2, 31
xori r4, r4, 1
rlwimi r2, r4, 31, 0, 0
stw r2, 0(r3)
blr
This implements InstCombine/cast.ll:test30.
llvm-svn: 28273
2006-05-13 02:06:03 +00:00
1443bc52be
Refactor some code, making it simpler.
...
When doing the initial pass of constant folding, if we get a constantexpr,
simplify the constant expr like we would do if the constant is folded in the
normal loop.
This fixes the missed-optimization regression in
Transforms/InstCombine/getelementptr.ll last night.
llvm-svn: 28224
2006-05-11 17:11:52 +00:00
a36ee4ea34
Two changes:
...
1. Implement InstCombine/deadcode.ll by not adding instructions in unreachable
blocks (due to constants in conditional branches/switches) to the worklist.
This causes them to be deleted before instcombine starts up, leading to
better optimization.
2. In the prepass over instructions, do trivial constprop/dce as we go. This
has the effect of improving the effectiveness of #1 . In addition, it
*significantly* speeds up instcombine on test cases with large amounts of
constant folding code (for example, that produced by code specialization
or partial evaluation). In one example, it speeds up instcombine from
0.0589s to 0.0224s with a release build (a 2.6x speedup).
llvm-svn: 28215
2006-05-10 19:00:36 +00:00
4fe87d67c4
Patch to make some xforms preserve each other. Patch contributed by
...
Domagoj Babic!
llvm-svn: 28181
2006-05-09 04:13:41 +00:00
1d441adfbf
Move some code around.
...
Make the "fold (and (cast A), (cast B)) -> (cast (and A, B))" transformation
only apply when both casts really will cause code to be generated. If one or
both doesn't, then this xform doesn't remove a cast.
This fixes Transforms/InstCombine/2006-05-06-Infloop.ll
llvm-svn: 28141
2006-05-06 09:00:16 +00:00
e745c7de0e
Fix an infinite loop compiling oggenc last night.
...
llvm-svn: 28128
2006-05-05 20:51:30 +00:00
3af1053488
Implement InstCombine/cast.ll:test29
...
llvm-svn: 28126
2006-05-05 06:39:07 +00:00
fb29692055
Fix Transforms/InstCombine/2006-05-04-DemandedBitCrash.ll
...
llvm-svn: 28101
2006-05-04 17:33:35 +00:00
2d3a02725d
Add pass ID's for various passes, so they can be AddRequiredID. Patch by
...
Domagoj Babic!
llvm-svn: 28048
2006-05-02 04:24:36 +00:00
655d08fda8
Fix InstCombine/2006-04-28-ShiftShiftLongLong.ll
...
llvm-svn: 28019
2006-04-28 22:21:41 +00:00
e63d808b6e
Fix Transforms/Reassociate/2006-04-27-ReassociateVector.ll
...
llvm-svn: 28007
2006-04-28 04:14:49 +00:00
b6cb64b7e6
Add support for inserting undef into a vector. This implements
...
Transforms/InstCombine/vec_insert_to_shuffle.ll
llvm-svn: 27997
2006-04-27 21:14:21 +00:00
dae49df407
Fix Transforms/ScalarRepl/2006-04-20-PromoteCrash.ll
...
llvm-svn: 27912
2006-04-20 20:48:50 +00:00
f89e630b2f
Make code match cvs commit message :)
...
llvm-svn: 27881
2006-04-20 15:41:37 +00:00
61eae29ad6
If we can convert the return pointer type into an integer that IntPtrType
...
can be converted to losslessly, we can continue the conversion to a direct call.
llvm-svn: 27880
2006-04-20 14:56:47 +00:00
36dd7c98d1
Turn x86 unaligned load/store intrinsics into aligned load/store instructions
...
if the pointer is known aligned.
llvm-svn: 27781
2006-04-17 22:26:56 +00:00
9095186deb
Fix a bug in the 'shuffle(undef,x,mask) -> shuffle(x, undef,mask')' xform
...
Make the insert/extract elt -> shuffle code more aggressive.
This fixes CodeGen/PowerPC/vec_shuffle.ll
llvm-svn: 27728
2006-04-16 00:51:47 +00:00
34cebe785d
Canonicalize shuffle(undef,x,mask) -> shuffle(x, undef,mask').
...
llvm-svn: 27727
2006-04-16 00:03:56 +00:00
39fac448d6
significant cleanups to code that uses insert/extractelt heavily. This builds
...
maximal shuffles out of them where possible.
llvm-svn: 27717
2006-04-15 01:39:45 +00:00
3323ce165d
Teach scalarrepl to promote unions of vectors and floats, producing
...
insert/extractelement operations. This implements
Transforms/ScalarRepl/vector_promote.ll
llvm-svn: 27710
2006-04-14 21:42:41 +00:00
13a1a7a4a6
Get rid of a signed/unsigned compare warning.
...
llvm-svn: 27625
2006-04-12 19:28:15 +00:00
b19a5c661b
Turn casts into getelementptr's when possible. This enables SROA to be more
...
aggressive in some cases where LLVMGCC 4 is inserting casts for no reason.
This implements InstCombine/cast.ll:test27/28.
llvm-svn: 27620
2006-04-12 18:09:35 +00:00
2d37f920ad
Implement vec_shuffle.ll:test3
...
llvm-svn: 27573
2006-04-10 23:06:36 +00:00
fbb77a408b
Implement InstCombine/vec_shuffle.ll:test[12]
...
llvm-svn: 27571
2006-04-10 22:45:52 +00:00
17bd60588c
Add supprot for shufflevector
...
llvm-svn: 27513
2006-04-08 01:19:12 +00:00
e79d249c29
Lower vperm(x,y, mask) -> shuffle(x,y,mask) if mask is constant. This allows
...
us to compile oh-so-realistic stuff like this:
vec_vperm(A, B, (vector unsigned char){14});
to:
vspltb v0, v0, 14
instead of:
vspltisb v0, 14
vperm v0, v2, v1, v0
llvm-svn: 27452
2006-04-06 19:19:17 +00:00
caba72b6ff
vector casts of casts are eliminable. Transform this:
...
%tmp = cast <4 x uint> %tmp to <4 x int> ; <<4 x int>> [#uses=1]
%tmp = cast <4 x int> %tmp to <4 x float> ; <<4 x float>> [#uses=1]
into:
%tmp = cast <4 x uint> %tmp to <4 x float> ; <<4 x float>> [#uses=1]
llvm-svn: 27355
2006-04-02 05:43:13 +00:00
ebca476b27
Allow transforming this:
...
%tmp = cast <4 x uint>* %testData to <4 x int>* ; <<4 x int>*> [#uses=1]
%tmp = load <4 x int>* %tmp ; <<4 x int>> [#uses=1]
to this:
%tmp = load <4 x uint>* %testData ; <<4 x uint>> [#uses=1]
%tmp = cast <4 x uint> %tmp to <4 x int> ; <<4 x int>> [#uses=1]
llvm-svn: 27353
2006-04-02 05:37:12 +00:00
f42d0aeda1
Turn altivec lvx/stvx intrinsics into loads and stores. This allows the
...
elimination of one load from this:
int AreSecondAndThirdElementsBothNegative( vector float *in ) {
#define QNaN 0x7FC00000
const vector unsigned int testData = (vector unsigned int)( QNaN, 0, 0, QNaN );
vector float test = vec_ld( 0, (float*) &testData );
return ! vec_any_ge( test, *in );
}
Now generating:
_AreSecondAndThirdElementsBothNegative:
mfspr r2, 256
oris r4, r2, 49152
mtspr 256, r4
li r4, lo16(LCPI1_0)
lis r5, ha16(LCPI1_0)
addi r6, r1, -16
lvx v0, r5, r4
stvx v0, 0, r6
lvx v1, 0, r3
vcmpgefp. v0, v0, v1
mfcr r3, 2
rlwinm r3, r3, 27, 31, 31
xori r3, r3, 1
cntlzw r3, r3
srwi r3, r3, 5
mtspr 256, r2
blr
llvm-svn: 27352
2006-04-02 05:30:25 +00:00
6cf4914fd4
Fix InstCombine/2006-04-01-InfLoop.ll
...
llvm-svn: 27330
2006-04-01 22:05:01 +00:00
dcd0792622
Fold A^(B&A) -> (B&A)^A
...
Fold (B&A)^A == ~B & A
This implements InstCombine/xor.ll:test2[56]
llvm-svn: 27328
2006-04-01 08:03:55 +00:00
8d1d8d364c
If we can look through vector operations to find the scalar version of an
...
extract_element'd value, do so.
llvm-svn: 27323
2006-03-31 23:01:56 +00:00
92346c315e
extractelement(undef,x) -> undef
...
llvm-svn: 27300
2006-03-31 18:25:14 +00:00
612fa8e6f3
Fix Transforms/InstCombine/2006-03-30-ExtractElement.ll
...
llvm-svn: 27261
2006-03-30 22:02:40 +00:00
d70d9f5b24
Don't crash on packed logical ops
...
llvm-svn: 27125
2006-03-25 21:58:26 +00:00
f365f5f0c1
Fix spello
...
llvm-svn: 27052
2006-03-24 07:14:34 +00:00
5821a6a17a
add the actual cost to the debug info
...
llvm-svn: 27051
2006-03-24 07:14:00 +00:00
83f99115db
Can't combine anymore - we don't have a chain through llvm.dbg intrinsics.
...
llvm-svn: 26992
2006-03-23 18:10:42 +00:00
7d80b4f366
silence a bogus gcc warning
...
llvm-svn: 26953
2006-03-22 17:27:24 +00:00
d783c76c18
Teach cee to propagate through switch statements. This implements
...
Transforms/CorrelatedExprs/switch.ll
Patch contributed by Eric Kidd!
llvm-svn: 26872
2006-03-19 19:37:24 +00:00
c28282bd87
- Fixed a bogus if condition.
...
- Added more debugging info.
- Allow reuse of IV of negative stride. e.g. -4 stride == 2 * iv of -2 stride.
llvm-svn: 26841
2006-03-18 08:03:12 +00:00
f09f0ebd48
Sort StrideOrder so we can process the smallest strides first. This allows
...
for more IV reuses.
llvm-svn: 26837
2006-03-18 00:44:49 +00:00
4520698820
Allow users of iv / stride to be rewritten with expression that is a multiply
...
of a smaller stride even if they have a common loop invariant expression part.
llvm-svn: 26828
2006-03-17 19:52:23 +00:00
3df447d354
For each loop, keep track of all the IV expressions inserted indexed by
...
stride. For a set of uses of the IV of a stride which is a multiple
of another stride, do not insert a new IV expression. Rather, reuse the
previous IV and rewrite the uses as uses of IV expression multiplied by
the factor.
e.g.
x = 0 ...; x ++
y = 0 ...; y += 4
then use of y can be rewritten as use of 4*x for x86.
llvm-svn: 26803
2006-03-16 21:53:05 +00:00
c5f866bb4a
Implement a FIXME, recusively reassociating
...
A*A*B + A*A*C --> A*(A*B+A*C) --> A*(A*(B+C))
This implements Reassociate/mul-factor3.ll
llvm-svn: 26757
2006-03-14 16:04:29 +00:00
2fc319d444
extract some code into a method, no functionality change
...
llvm-svn: 26755
2006-03-14 07:11:11 +00:00
d6bde46d85
Promote shifts by a constant to multiplies so that we can reassociate
...
(x<<1)+(y<<1) -> (X+Y)<<1. This implements
Transforms/Reassociate/shift-factor.ll
llvm-svn: 26753
2006-03-14 06:55:18 +00:00
c567c4efbb
Added target lowering hooks which LSR consults to make more intelligent
...
transformation decisions.
llvm-svn: 26738
2006-03-13 23:14:23 +00:00
fc34f8bb48
Fix a miscompilation of 188.ammp with the new CFE. 188.ammp is accessing
...
arrays out of range in a horrible way, but we shouldn't break it anyway.
Details in the comments.
llvm-svn: 26606
2006-03-08 01:05:29 +00:00
53ef5a032c
Teach the alignment handling code to look through constant expr casts and GEPs
...
llvm-svn: 26580
2006-03-07 01:28:57 +00:00
82f2ef20b6
Teach instcombine to increase the alignment of memset/memcpy/memmove when
...
the pointer is known to come from either a global variable, alloca or
malloc. This allows us to compile this:
P = malloc(28);
memset(P, 0, 28);
into explicit stores on PPC instead of a memset call.
llvm-svn: 26577
2006-03-06 20:18:44 +00:00
6bc98653c2
Make vector narrowing more effective, implementing
...
Transforms/InstCombine/vec_narrow.ll. This add support for narrowing
extract_element(insertelement) also.
llvm-svn: 26538
2006-03-05 00:22:33 +00:00
4c065091d8
Add factoring of multiplications, e.g. turning A*A+A*B into A*(A+B).
...
Testcase here: Transforms/Reassociate/mulfactor.ll
llvm-svn: 26524
2006-03-04 09:31:13 +00:00
32c01df299
Canonicalize (X+C1)*C2 -> X*C2+C1*C2
...
This implements Transforms/InstCombine/add.ll:test31
llvm-svn: 26519
2006-03-04 06:04:02 +00:00
681ef2f083
Change this to work with renamed intrinsics.
...
llvm-svn: 26484
2006-03-03 01:34:17 +00:00
85dda9a2bd
Generalize the REM folding code to handle another case Nick Lewycky
...
pointed out: realize the AND can provide factors and look through Casts.
llvm-svn: 26469
2006-03-02 06:50:58 +00:00
c5b6c9a12a
Fix a regression in a patch from a couple of days ago. This fixes
...
Transforms/InstCombine/2006-02-28-Crash.ll
llvm-svn: 26427
2006-02-28 19:47:20 +00:00
b70f141893
Implement rem.ll:test[7-9] and PR712
...
llvm-svn: 26415
2006-02-28 05:49:21 +00:00
2a7c7b8bab
Simplify some code now that the RHS of a rem can't be 0
...
llvm-svn: 26413
2006-02-28 05:40:55 +00:00
0de4a8d7b7
Rearrange some code, fold "rem X, 0", implementing rem.ll:test6
...
llvm-svn: 26411
2006-02-28 05:30:45 +00:00
c7bfed0f7b
Merge two almost-identical pieces of code.
...
Make this code more powerful by using ComputeMaskedBits instead of looking
for an AND operand. This lets us fold this:
int %test23(int %a) {
%tmp.1 = and int %a, 1
%tmp.2 = seteq int %tmp.1, 0
%tmp.3 = cast bool %tmp.2 to int ;; xor tmp1, 1
ret int %tmp.3
}
into: xor (and a, 1), 1
llvm-svn: 26396
2006-02-27 02:38:23 +00:00
f5c8a0b83f
Fold (A^B) == A -> B == 0
...
and (A-B) == A -> B == 0
llvm-svn: 26394
2006-02-27 01:44:11 +00:00
f78df7c14d
Fold (X|C1)^C2 -> X^(C1|C2) when possible. This implements
...
InstCombine/or.ll:test23.
llvm-svn: 26385
2006-02-26 19:57:54 +00:00
b580d26e7d
Fix a problem that Nate noticed that boils down to an over conservative check
...
in the code that does "select C, (X+Y), (X-Y) --> (X+(select C, Y, (-Y)))".
We now compile this loop:
LBB1_1: ; no_exit
add r6, r2, r3
subf r3, r2, r3
cmpwi cr0, r2, 0
addi r7, r5, 4
lwz r2, 0(r5)
addi r4, r4, 1
blt cr0, LBB1_4 ; no_exit
LBB1_3: ; no_exit
mr r3, r6
LBB1_4: ; no_exit
cmpwi cr0, r4, 16
mr r5, r7
bne cr0, LBB1_1 ; no_exit
into this instead:
LBB1_1: ; no_exit
srawi r6, r2, 31
add r2, r2, r6
xor r6, r2, r6
addi r7, r5, 4
lwz r2, 0(r5)
addi r4, r4, 1
add r3, r3, r6
cmpwi cr0, r4, 16
mr r5, r7
bne cr0, LBB1_1 ; no_exit
llvm-svn: 26356
2006-02-24 18:05:58 +00:00
e5521db5bc
Fix Regression/Transforms/LoopUnswitch/2006-02-22-UnswitchCrash.ll, which
...
caused SPASS to fail building last night.
We can't trivially unswitch a loop if the exit block has phi nodes in it,
because we don't know which predecessor to use.
llvm-svn: 26320
2006-02-22 23:55:00 +00:00
8a5a324dac
Add some comments, simplify some code, and fix a bug that caused rewriting
...
to rewrite with the wrong value.
llvm-svn: 26311
2006-02-22 06:37:14 +00:00
c2e3a7a4ce
improved support for branch folding, still not enabled.
...
llvm-svn: 26289
2006-02-18 07:57:38 +00:00
0add83e969
Fix bugs identified by VC++.
...
llvm-svn: 26287
2006-02-18 03:20:33 +00:00
19fa8ac938
Implement deletion of dead blocks, currently disabled.
...
llvm-svn: 26285
2006-02-18 02:42:34 +00:00
cb853de534
a previous patch completely disabled trivial unswitching, this fixees it.
...
Thanks to nate for pointing this out :)
llvm-svn: 26280
2006-02-18 01:32:04 +00:00
29f771ba21
initial trivial support for folding branches that have now-constant destinations.
...
llvm-svn: 26279
2006-02-18 01:27:45 +00:00
8e44ff50b0
When unswitching a loop, make sure to update loop info with exit blocks in
...
the right loop.
llvm-svn: 26277
2006-02-18 00:55:32 +00:00
baddba41c7
Fix loops where the header has an exit, fixing a loop-unswitch crash on crafty
...
llvm-svn: 26258
2006-02-17 06:39:56 +00:00
6fd136239b
start of some new simplification code, not thoroughly tested, use at your own
...
risk :)
llvm-svn: 26248
2006-02-17 00:31:07 +00:00
8a77efe4f7
Rework the SelectionDAG-based implementations of SimplifyDemandedBits
...
and ComputeMaskedBits to match the new improved versions in instcombine.
Tested against all of multisource/benchmarks on ppc.
llvm-svn: 26238
2006-02-16 21:11:51 +00:00
fa335f6083
Change SplitBlock to increment a BasicBlock::iterator, not an Instruction*. Apparently they do different things :)
...
This fixes a testcase that nate reduced from spass.
Also included are a couple minor code changes that don't affect the generated
code at all.
llvm-svn: 26235
2006-02-16 19:36:22 +00:00
55f63f1b53
Fix VC++ warning.
...
llvm-svn: 26228
2006-02-16 04:07:37 +00:00
ff42e81028
fix a bug where we unswitched the wrong way
...
llvm-svn: 26225
2006-02-16 01:24:41 +00:00
fdff0bb43e
Implement trivial unswitching for switch stmts. This allows us to trivial
...
unswitch this loop on 2 before sweating to unswitch on 1/3.
void test4(int N, int i, int C, int*P, int*Q) {
int j;
for (j = 0; j < N; ++j) {
switch (C) { // general unswitching.
default: P[i+j] = 0; break;
case 1: Q[i+j] = 0; break;
case 3: P[i+j] = Q[i+j]; break;
case 2: break; // TRIVIAL UNSWITCH on C==2
}
}
}
llvm-svn: 26223
2006-02-15 22:52:05 +00:00
e5cb76d744
make "trivial" unswitching significantly more general. It can now handle
...
this for example:
for (j = 0; j < N; ++j) { // trivial unswitch
if (C)
P[i+j] = 0;
}
turning it into the obvious code without bothering to duplicate an empty loop.
llvm-svn: 26220
2006-02-15 22:03:36 +00:00
65152d80ec
Checking the wrong value. This caused us to emit silly code like
...
Y = seteq bool X, true
instead of just using X :)
llvm-svn: 26215
2006-02-15 19:05:52 +00:00
01db04efb0
more refactoring, no functionality change.
...
llvm-svn: 26194
2006-02-15 01:44:42 +00:00
b0cbe7106e
pull some code out into a function
...
llvm-svn: 26191
2006-02-15 00:07:43 +00:00
0b8ec1a132
Use statistics to keep track of what flavors of loops we are unswitching
...
llvm-svn: 26157
2006-02-14 01:01:41 +00:00
8b10ab3002
Implement Instcombine/and.ll:test34
...
llvm-svn: 26155
2006-02-13 23:07:23 +00:00
7d8522884b
If any of the sign extended bits are demanded, the input sign bit is demanded
...
for a sign extension.
This fixes InstCombine/2006-02-13-DemandedMiscompile.ll and Ptrdist/bc.
llvm-svn: 26152
2006-02-13 22:41:07 +00:00
68e7475777
Be careful not to request or look at bits shifted in from outside the size
...
of the input. This fixes the mediabench/gsm/toast failure last night.
llvm-svn: 26138
2006-02-13 06:09:08 +00:00
f5b4ef7f58
remove some more dead special case code
...
llvm-svn: 26135
2006-02-12 08:07:37 +00:00
5b2edb1fca
Eliminate special case hacks that are superceded by general purpose hacks
...
llvm-svn: 26134
2006-02-12 08:02:11 +00:00
ee0f280743
Three changes:
...
1. Teach GetConstantInType to handle boolean constants.
2. Teach instcombine to fold (compare X, CST) when X has known 0/1 bits.
Testcase here: set.ll:test22
3. Improve the "(X >> c1) & C2 == 0" folding code to allow a noop cast
between the shift and and. More aggressive bitfolding for other reasons
was turning signed shr's into unsigned shr's, leaving the noop cast in
the way.
llvm-svn: 26131
2006-02-12 02:07:56 +00:00
0157e7f55b
Port the recent innovations in ComputeMaskedBits to SimplifyDemandedBits.
...
This allows us to simplify on conditions where bits are not known, but they
are not demanded either! This also fixes a couple of bugs in
ComputeMaskedBits that were exposed during this work.
In the future, swaths of instcombine should be removed, as this code
subsumes a bunch of ad-hockery.
llvm-svn: 26122
2006-02-11 09:31:47 +00:00
fbadd7e1ee
implement unswitching of loops with switch stmts and selects in them
...
llvm-svn: 26114
2006-02-11 00:43:37 +00:00
f1b151684d
Update PHI nodes in successors of exit blocks.
...
llvm-svn: 26113
2006-02-10 23:26:14 +00:00
fe4151efe7
Reform the unswitching code in terms of edge splitting, not block splitting.
...
llvm-svn: 26112
2006-02-10 23:16:39 +00:00
ec6b40a093
Fix a case where UnswitchTrivialCondition broke critical edges with
...
phi's in the successors
llvm-svn: 26108
2006-02-10 19:08:15 +00:00
6e263155a6
add some notes, move some code around. Implement unswitching of loops
...
with branches on partially invariant computations.
llvm-svn: 26104
2006-02-10 02:30:37 +00:00
4935417a84
Move code around to be more logical, no functionality change.
...
llvm-svn: 26103
2006-02-10 02:01:22 +00:00
3fc3148b85
When unswitching a trivial loop, do admit we are doing it! :)
...
llvm-svn: 26102
2006-02-10 01:36:35 +00:00
ed7a67b0de
Implement unconditional unswitching of 'trivial' loops, those loops that contain
...
branches in their entry block that control whether or not the loop is a noop or not.
llvm-svn: 26101
2006-02-10 01:24:09 +00:00
4f0e66df6a
Simplify control flow a bit, note that unswitch preserves canonical loop form
...
llvm-svn: 26098
2006-02-09 22:15:42 +00:00
8976219850
Make the threshold a parameter
...
llvm-svn: 26093
2006-02-09 20:15:48 +00:00
2826e0511b
Simplify the loop-unswitch pass, by not even trying to unswitch loops with
...
uses of loop values outside the loop. We need loop-closed SSA form to do
this right, or to use SSA rewriting if we really care.
llvm-svn: 26089
2006-02-09 19:14:52 +00:00
24cd2fa269
Fix 80-column violations
...
llvm-svn: 26088
2006-02-09 07:41:14 +00:00
4534dd59a3
Enhance MVIZ in three ways:
...
1. Teach it new tricks: in particular how to propagate through signed shr and sexts.
2. Teach it to return a bitset of known-1 and known-0 bits, instead of just zero.
3. Teach instcombine (AND X, C) to fold when we know all C bits of X.
This implements Regression/Transforms/InstCombine/bittest.ll, and allows
future things to be simplified.
llvm-svn: 26087
2006-02-09 07:38:58 +00:00
ab2dc4d70d
Simplify some code, reducing calls to MaskedValueIsZero. Implement a minor
...
optimization where we reduce the number of bits in AND masks when possible.
llvm-svn: 26056
2006-02-08 07:34:50 +00:00
5997cf9381
Use EraseInstFromFunction in a few cases to put the uses of the removed
...
instruction onto the worklist (in case they are now dead).
Add a really trivial local DSE implementation to help out bitfield code.
We now fold this:
struct S {
unsigned char a : 1, b : 1, c : 1, d : 2, e : 3;
S();
};
S::S() : a(0), b(0), c(1), d(0), e(6) {}
to this:
void %_ZN1SC1Ev(%struct.S* %this) {
entry:
%tmp.1 = getelementptr %struct.S* %this, int 0, uint 0
store ubyte 38, ubyte* %tmp.1
ret void
}
much earlier (in gccas instead of only in gccld after DSE runs).
llvm-svn: 26050
2006-02-08 03:25:32 +00:00
06a0ed1ee0
Implement some more interesting select sccp cases. This implements:
...
test/Regression/Transforms/SCCP/select.ll
llvm-svn: 26049
2006-02-08 02:38:11 +00:00
ddba3289b5
Fix a problem in my patch yesterday, causing a miscompilation of 176.gcc
...
llvm-svn: 26045
2006-02-08 01:20:23 +00:00
44314827d6
Fix Transforms/InstCombine/2006-02-07-SextZextCrash.ll
...
llvm-svn: 26040
2006-02-07 19:07:40 +00:00
92a6865321
Generalize MaskedValueIsZero into a ComputeMaskedNonZeroBits function, which
...
is just as efficient as MVIZ and is also more general.
Fix a few minor bugs introduced in recent patches
llvm-svn: 26036
2006-02-07 08:05:22 +00:00
c3ebf40031
Make MaskedValueIsZero take a uint64_t instead of a ConstantIntegral as a
...
mask. This allows the code to be simpler and more efficient.
Also, generalize some of the cases in MVIZ a bit, making it slightly more aggressive.
llvm-svn: 26035
2006-02-07 07:27:52 +00:00
77defbae0a
Use Type::getIntegralTypeMask() to simplify some code
...
llvm-svn: 26034
2006-02-07 07:00:41 +00:00
2590e511d8
Implement the beginnings of a facility for simplifying expressions based on
...
'demanded bits', inspired by Nate's work in the dag combiner. This isn't
complete, but needs to unrelated instcombiner changes to continue.
llvm-svn: 26033
2006-02-07 06:56:34 +00:00
2e90b732fa
Turn A % (C << N), where C is 2^k, into A & ((C << N)-1) [urem only].
...
Turn A / (C1 << N), where C1 is "1<<C2" into A >> (N+C2) [udiv only].
Tested with: rem.ll:test5, div.ll:test10
llvm-svn: 26003
2006-02-05 07:54:04 +00:00
d30c4991a1
Use SCEVExpander::InsertCastOfTo instead of our own code. This reduces
...
#LLVM LOC, and auto-cse's cast instructions.
llvm-svn: 25974
2006-02-04 09:52:43 +00:00
2959f0003e
Fix two significant bugs in LSR:
...
1. When rewriting code in outer loops, sometimes we would insert code into
inner loops that is invariant in that loop.
2. Notice that 4*(2+x) is 8+4*x and use that to simplify expressions.
This is a performance neutral change.
llvm-svn: 25964
2006-02-04 07:36:50 +00:00
15a8c15a1f
Improve compatibility with VC2005, patch by Morten Ofstad!
...
llvm-svn: 25661
2006-01-26 20:41:32 +00:00
c0f633a598
Fix Regression/Transforms/ScalarRepl/2006-01-24-IllegalUnionPromoteCrash.ll
...
llvm-svn: 25587
2006-01-24 19:36:27 +00:00
c597b8a55e
Make iostream #inclusion explicit
...
llvm-svn: 25514
2006-01-22 23:32:06 +00:00
e154abf9b3
Implement casts.ll:test26: a cast from float -> double -> integer, doesn't
...
need the float->double part.
llvm-svn: 25452
2006-01-19 07:40:22 +00:00
6dce25019d
Lowerpacked and SCCP support for the insertelement operation.
...
llvm-svn: 25406
2006-01-17 20:06:55 +00:00
307b7ea15f
fix a crash due to missing parens
...
llvm-svn: 25363
2006-01-16 19:47:21 +00:00
0de2c7d3d8
This pass has never worked correctly. Remove.
...
llvm-svn: 25349
2006-01-16 01:06:00 +00:00
ef530c24c1
FunctionPass's cannot do IPO things.
...
llvm-svn: 25315
2006-01-14 19:30:35 +00:00
a83529678e
Added instcombine support for extractelement.
...
llvm-svn: 25299
2006-01-13 22:48:06 +00:00
503221f5c5
Do a simple instcombine xforms to delete llvm.stackrestore cases.
...
llvm-svn: 25294
2006-01-13 21:28:09 +00:00
c66b223b28
Simplify this a tiny bit by using the new IntrinsicInst functionality.
...
llvm-svn: 25292
2006-01-13 20:11:04 +00:00
cb36710ff9
Switch these to using ETForest instead of DominatorSet to compute itself.
...
Patch written by Daniel Berlin!
llvm-svn: 25202
2006-01-11 05:10:20 +00:00
48e4a2ebd8
Switch this to using ETForest instead of DominatorSet to compute itself.
...
Patch written by Daniel Berlin!
llvm-svn: 25201
2006-01-11 05:09:40 +00:00
bd518d153b
Added lower packed support for the extractelement operation.
...
llvm-svn: 25180
2006-01-10 19:05:05 +00:00
9cbfbc21bb
fix some 176.gcc miscompilation from my previous patch.
...
llvm-svn: 25137
2006-01-07 01:32:28 +00:00
330628a6d8
silence some bogus gcc warnings on fenris
...
llvm-svn: 25130
2006-01-06 17:59:59 +00:00
eb372a0276
Enhance the shift-shift folding code to allow a no-op cast to occur in between
...
the shifts.
This allows us to fold this (which is the 'integer add a constant' sequence
from cozmic's scheme compmiler):
int %x(uint %anf-temporary776) {
%anf-temporary777 = shr uint %anf-temporary776, ubyte 1
%anf-temporary800 = cast uint %anf-temporary777 to int
%anf-temporary804 = shl int %anf-temporary800, ubyte 1
%anf-temporary805 = add int %anf-temporary804, -2
%anf-temporary806 = or int %anf-temporary805, 1
ret int %anf-temporary806
}
into this:
int %x(uint %anf-temporary776) {
%anf-temporary776 = cast uint %anf-temporary776 to int
%anf-temporary776.mask1 = add int %anf-temporary776, -2
%anf-temporary805 = or int %anf-temporary776.mask1, 1
ret int %anf-temporary805
}
note that instcombine already knew how to eliminate the AND that the two
shifts fold into. This is tested by InstCombine/shift.ll:test26
-Chris
llvm-svn: 25128
2006-01-06 07:52:12 +00:00
b330939d90
Simplify the code a bit more
...
llvm-svn: 25126
2006-01-06 07:22:22 +00:00
145539343f
Extract a bunch of code out of visitShiftInst into FoldShiftByConstant. No
...
functionality changes.
llvm-svn: 25125
2006-01-06 07:12:35 +00:00
7a3ad6cae2
getting there...
...
llvm-svn: 25021
2005-12-26 13:48:44 +00:00
8c9e14620f
Fix Transforms/ScalarRepl/2005-12-14-UnionPromoteCrash.ll, a crash on undefined
...
behavior in 126.gcc on big-endian systems.
llvm-svn: 24708
2005-12-14 17:23:59 +00:00
3b0a62d8a5
Implement a little hack for parity with GCC on crafty. This speeds up
...
186.crafty by about 16% (from 15.109s to 13.045s) on my system.
This turns allocas with unions/casts into scalars. For example crafty has
something like this:
union doub {
unsigned short i[4];
long long d;
};
int f(long long a) {
return ((union doub){.d=a}).i[1];
}
Instead of generating loads and stores to an alloca, we now promote the
whole thing to a scalar long value.
This implements: Transforms/ScalarRepl/AggregatePromote.ll
llvm-svn: 24667
2005-12-12 07:19:13 +00:00
077200737c
getRawValue zero extens for unsigned values, use getsextvalue so that we
...
know that small negative values fit into the immediate field of addressing
modes.
llvm-svn: 24608
2005-12-05 18:23:57 +00:00
dc4ffef633
Fix a bug where we didn't realize that vaarg reads memory. This fixes
...
Transforms/DeadStoreElimination/2005-11-30-vaarg.ll
llvm-svn: 24545
2005-11-30 19:38:22 +00:00
5fc3794e71
since reg2mem requires it, might as well mention that it preserves it
...
llvm-svn: 24491
2005-11-25 16:04:54 +00:00
061029dee2
Reg2Mem is something a pass may depend on, so allow that
...
llvm-svn: 24488
2005-11-22 22:14:23 +00:00
71b09bbb07
turns out, demotion and invokes and critical edges don't mix
...
llvm-svn: 24487
2005-11-22 21:45:19 +00:00
9c37f23645
Fix a crash building 176.gcc due to my recent patch, which only fixed
...
half the problem.
llvm-svn: 24414
2005-11-18 18:30:47 +00:00
bca0be812d
This was checking the wrong GEP expression. Fixing this fixes a gccas crash
...
compiling mysql reported by Ted Kremenek.
llvm-svn: 24402
2005-11-17 19:35:42 +00:00
d9c13b1336
the pain isn't gone unless the phinodes are spilled too
...
llvm-svn: 24288
2005-11-10 19:39:09 +00:00
8e66c0c8a9
this works with backedges to the existing entry block alot better
...
llvm-svn: 24270
2005-11-10 17:35:34 +00:00
4130a4f061
The pass everyone has been waiting for!
...
Reg2Mem
for fun you can opt -reg2mem -mem2reg
llvm-svn: 24267
2005-11-10 01:58:38 +00:00
848622f87f
Add support alignment of allocation instructions.
...
Add support for specifying alignment and size of setjmp jmpbufs.
No targets currently do anything with this information, nor is it presrved
in the bytecode representation. That's coming up next.
llvm-svn: 24196
2005-11-05 09:21:28 +00:00
16b29e9562
Implement Transforms/TailCallElim/return-undef.ll, a trivial case
...
that has been sitting in my inbox since May 18. :)
llvm-svn: 24194
2005-11-05 08:21:11 +00:00
dd0c174082
Turn sdiv into udiv if both operands have a clear sign bit. This occurs
...
a few times in crafty:
OLD: %tmp.36 = div int %tmp.35, 8 ; <int> [#uses=1]
NEW: %tmp.36 = div uint %tmp.35, 8 ; <uint> [#uses=0]
OLD: %tmp.19 = div int %tmp.18, 8 ; <int> [#uses=1]
NEW: %tmp.19 = div uint %tmp.18, 8 ; <uint> [#uses=0]
OLD: %tmp.117 = div int %tmp.116, 8 ; <int> [#uses=1]
NEW: %tmp.117 = div uint %tmp.116, 8 ; <uint> [#uses=0]
OLD: %tmp.92 = div int %tmp.91, 8 ; <int> [#uses=1]
NEW: %tmp.92 = div uint %tmp.91, 8 ; <uint> [#uses=0]
Which all turn into shrs.
llvm-svn: 24190
2005-11-05 07:40:31 +00:00
e9ff0eaf5b
Turn srem -> urem when neither input has their sign bit set. This triggers
...
8 times in vortex, allowing the srems to be turned into shrs:
OLD: %tmp.104 = rem int %tmp.5.i37, 16 ; <int> [#uses=1]
NEW: %tmp.104 = rem uint %tmp.5.i37, 16 ; <uint> [#uses=0]
OLD: %tmp.98 = rem int %tmp.5.i24, 16 ; <int> [#uses=1]
NEW: %tmp.98 = rem uint %tmp.5.i24, 16 ; <uint> [#uses=0]
OLD: %tmp.91 = rem int %tmp.5.i19, 8 ; <int> [#uses=1]
NEW: %tmp.91 = rem uint %tmp.5.i19, 8 ; <uint> [#uses=0]
OLD: %tmp.88 = rem int %tmp.5.i14, 8 ; <int> [#uses=1]
NEW: %tmp.88 = rem uint %tmp.5.i14, 8 ; <uint> [#uses=0]
OLD: %tmp.85 = rem int %tmp.5.i9, 1024 ; <int> [#uses=2]
NEW: %tmp.85 = rem uint %tmp.5.i9, 1024 ; <uint> [#uses=0]
OLD: %tmp.82 = rem int %tmp.5.i, 512 ; <int> [#uses=2]
NEW: %tmp.82 = rem uint %tmp.5.i1, 512 ; <uint> [#uses=0]
OLD: %tmp.48.i = rem int %tmp.5.i.i161, 4 ; <int> [#uses=1]
NEW: %tmp.48.i = rem uint %tmp.5.i.i161, 4 ; <uint> [#uses=0]
OLD: %tmp.20.i2 = rem int %tmp.5.i.i, 4 ; <int> [#uses=1]
NEW: %tmp.20.i2 = rem uint %tmp.5.i.i, 4 ; <uint> [#uses=0]
it also occurs 9 times in gcc, but with odd constant divisors (1009 and 61)
so the payoff isn't as great.
llvm-svn: 24189
2005-11-05 07:28:37 +00:00
662295587d
make this 64 bit clean, fixed test30 of /Regression/Transforms/InstCombine/add.ll
...
llvm-svn: 24158
2005-11-02 18:35:40 +00:00
09efd4e5b6
Limit the search depth of MaskedValueIsZero to 6 instructions, to avoid
...
bad cases. This fixes Markus's second testcase in PR639, and should
seal it for good.
llvm-svn: 24123
2005-10-31 18:35:52 +00:00
27d351f159
This pass is now obsolete since all targets have moved to the SelectionDAG
...
infrastructure and the simple isels have been removed.
llvm-svn: 24090
2005-10-29 05:33:46 +00:00
8f663e8bbc
Pull some code out into a function, give it the ability to see through +.
...
This allows us to turn code like malloc(4*x+4) -> malloc int, (x+1)
llvm-svn: 24081
2005-10-29 04:36:15 +00:00
8270c33606
Remove a special case, allowing the general case to handle it. No functionality
...
change.
llvm-svn: 24076
2005-10-29 03:19:53 +00:00
b9d3ca5c3c
Fix a bit of backwards logic that broke exptree and smg2000
...
llvm-svn: 24056
2005-10-28 16:27:35 +00:00
c4f67e67d2
Do not sink any instruction with side effects, including vaarg. This fixes
...
PR640
llvm-svn: 24046
2005-10-27 17:13:11 +00:00
c6372cca78
Fix typo
...
llvm-svn: 24033
2005-10-27 06:26:26 +00:00
0fe7551bc0
Teach instcombine to promote stuff like (cast (malloc sbyte, 8*X) to int*)
...
into: malloc int, (2*X)
llvm-svn: 24032
2005-10-27 06:24:46 +00:00
b3ecf96900
Promote cases like cast (malloc sbyte, 100) to int* into
...
(malloc [25 x int]) directly without having to convert to
(malloc [100 x sbyte]) first.
llvm-svn: 24031
2005-10-27 06:12:00 +00:00
bb17180a23
Minor change to this file to support obscure cases with constant array amounts
...
llvm-svn: 24030
2005-10-27 05:53:56 +00:00
38a1b00a0f
fold nested and's early to avoid inefficiencies in MaskedValueIsZero. This
...
fixes a very slow compile in PR639.
llvm-svn: 24011
2005-10-26 17:18:16 +00:00
2b8cbf319c
Update Visual Studio projects to reflect moved file.
...
llvm-svn: 23998
2005-10-26 05:36:51 +00:00
46705b2f2d
Handle allocations that, even after removing dead uses, still have more than
...
one use (but one is a cast). This handles the very common case of:
X = alloc [n x byte]
Y = cast X to somethingbetter
seteq X, null
In order to avoid infinite looping when there are multiple casts, we only
allow this if the xform is strictly increasing the alignment of the
allocation.
llvm-svn: 23961
2005-10-24 06:35:18 +00:00
355ecc09f8
Fix a bug where we would 'promote' an allocation from one type to another
...
where the second has less alignment required. If we had explicit alignment
support in the IR, we could handle this case, but we can't until we do.
llvm-svn: 23960
2005-10-24 06:26:18 +00:00
ac87beb03a
Before promoting a malloc type, remove dead uses. This makes instcombine
...
more effective at promoting these allocations, catching them earlier in the
compile process.
llvm-svn: 23959
2005-10-24 06:22:12 +00:00
216be91817
Pull some code out into a function, no functionality change
...
llvm-svn: 23958
2005-10-24 06:03:58 +00:00
bde3845548
DONT_BUILD_RELINKED is gone and implied by BUILD_ARCHIVE now
...
llvm-svn: 23940
2005-10-24 02:26:13 +00:00
8c087e962c
Only build .a file versions of these libraries, instead of .a and .o versions.
...
This should speed up build times.
llvm-svn: 23933
2005-10-24 01:59:48 +00:00
bd77fac034
Make sure that anything using the ADCE pass pulls in the UnifyFunctionExitNodes
...
code
llvm-svn: 23931
2005-10-24 01:40:23 +00:00
11e26b52b2
When a function takes a variable number of pointer arguments, with a zero
...
pointer marking the end of the list, the zero *must* be cast to the pointer
type. An un-cast zero is a 32-bit int, and at least on x86_64, gcc will
not extend the zero to 64 bits, thus allowing the upper 32 bits to be
random junk.
The new END_WITH_NULL macro may be used to annotate a such a function
so that GCC (version 4 or newer) will detect the use of un-casted zero
at compile time.
llvm-svn: 23888
2005-10-23 04:37:20 +00:00
5df0e36e98
My previous patch was too conservative. Reject FP and void types, but do
...
allow pointer types.
llvm-svn: 23859
2005-10-21 05:45:41 +00:00
0c0b38bb4c
Do NOT touch FP ops with LSR. This fixes a testcase Nate sent me from an
...
inner loop like this:
LBB_RateConvertMono8AltiVec_2: ; no_exit
lis r2, ha16(.CPI_RateConvertMono8AltiVec_0)
lfs f3, lo16(.CPI_RateConvertMono8AltiVec_0)(r2)
fmr f3, f3
fadd f0, f2, f0
fadd f3, f0, f3
fcmpu cr0, f3, f1
bge cr0, LBB_RateConvertMono8AltiVec_2 ; no_exit
to an inner loop like this:
LBB_RateConvertMono8AltiVec_1: ; no_exit
fsub f2, f2, f1
fcmpu cr0, f2, f1
fmr f0, f2
bge cr0, LBB_RateConvertMono8AltiVec_1 ; no_exit
Doh! good catch!
llvm-svn: 23838
2005-10-20 04:47:10 +00:00
da1b152c43
Make this work for FP constantexprs
...
llvm-svn: 23773
2005-10-17 20:18:38 +00:00
7fde91e365
Oops, X+0.0 isn't foldable, but X+-0.0 is.
...
llvm-svn: 23772
2005-10-17 17:56:38 +00:00
32979336a7
relax this a bit, as we only support the default rounding mode
...
llvm-svn: 23771
2005-10-17 17:49:32 +00:00
192cd18f53
Fix (hopefully the last) issue where LSR is nondeterminstic. When pulling
...
out CSE's of base expressions it could build a result whose order was
nondet.
llvm-svn: 23698
2005-10-11 18:41:04 +00:00
5c9d63da31
Fix another problem where LSR was being nondeterminstic. Also remove elements
...
from the end of a vector instead of the beginning
llvm-svn: 23697
2005-10-11 18:30:57 +00:00
b7a3894e7c
Fix another lsr-is-nondeterministic case
...
llvm-svn: 23695
2005-10-11 18:17:57 +00:00
03b9eb506c
Make MaskedValueIsZero a bit more aggressive
...
llvm-svn: 23677
2005-10-09 22:08:50 +00:00
62010c450f
Fix funky xcode indentation
...
llvm-svn: 23674
2005-10-09 06:36:35 +00:00
eb4be8b942
Hrm, you didn't see this.
...
llvm-svn: 23673
2005-10-09 06:24:02 +00:00
4ea0a3eaac
Fix a source of non-determinism in the backend: the order of processing
...
IV strides dependend on the pointer order of the strides in memory.
Non-determinism is bad.
llvm-svn: 23672
2005-10-09 06:20:55 +00:00
572910c9a2
Remove useless variable.
...
llvm-svn: 23656
2005-10-07 05:28:29 +00:00
f07a587c79
Make IVUseShouldUsePostIncValue more aggressive when the use is a PHI. In
...
particular, it should realize that phi's use their values in the pred block
not the phi block itself. This change turns our em3d loop from this:
_test:
cmpwi cr0, r4, 0
bgt cr0, LBB_test_2 ; entry.no_exit_crit_edge
LBB_test_1: ; entry.loopexit_crit_edge
li r2, 0
b LBB_test_6 ; loopexit
LBB_test_2: ; entry.no_exit_crit_edge
li r6, 0
LBB_test_3: ; no_exit
or r2, r6, r6
lwz r6, 0(r3)
cmpw cr0, r6, r5
beq cr0, LBB_test_6 ; loopexit
LBB_test_4: ; endif
addi r3, r3, 4
addi r6, r2, 1
cmpw cr0, r6, r4
blt cr0, LBB_test_3 ; no_exit
LBB_test_5: ; endif.loopexit.loopexit_crit_edge
addi r3, r2, 1
blr
LBB_test_6: ; loopexit
or r3, r2, r2
blr
into:
_test:
cmpwi cr0, r4, 0
bgt cr0, LBB_test_2 ; entry.no_exit_crit_edge
LBB_test_1: ; entry.loopexit_crit_edge
li r2, 0
b LBB_test_5 ; loopexit
LBB_test_2: ; entry.no_exit_crit_edge
li r6, 0
LBB_test_3: ; no_exit
lwz r2, 0(r3)
cmpw cr0, r2, r5
or r2, r6, r6
beq cr0, LBB_test_5 ; loopexit
LBB_test_4: ; endif
addi r3, r3, 4
addi r6, r6, 1
cmpw cr0, r6, r4
or r2, r6, r6
blt cr0, LBB_test_3 ; no_exit
LBB_test_5: ; loopexit
or r3, r2, r2
blr
Unfortunately, this is actually worse code, because the register coallescer
is getting confused somehow. If it were doing its job right, it could turn the
code into this:
_test:
cmpwi cr0, r4, 0
bgt cr0, LBB_test_2 ; entry.no_exit_crit_edge
LBB_test_1: ; entry.loopexit_crit_edge
li r6, 0
b LBB_test_5 ; loopexit
LBB_test_2: ; entry.no_exit_crit_edge
li r6, 0
LBB_test_3: ; no_exit
lwz r2, 0(r3)
cmpw cr0, r2, r5
beq cr0, LBB_test_5 ; loopexit
LBB_test_4: ; endif
addi r3, r3, 4
addi r6, r6, 1
cmpw cr0, r6, r4
blt cr0, LBB_test_3 ; no_exit
LBB_test_5: ; loopexit
or r3, r6, r6
blr
... which I'll work on next. :)
llvm-svn: 23604
2005-10-03 02:50:05 +00:00
e4ed42a426
Refactor some code into a function
...
llvm-svn: 23603
2005-10-03 01:04:44 +00:00
360928dbed
This break is bogus and I have no idea why it was there. Basically it prevents
...
memoizing code when IV's are used by phinodes outside of loops. In a simple
example, we were getting this code before (note that r6 and r7 are isomorphic
IV's):
li r6, 0
or r7, r6, r6
LBB_test_3: ; no_exit
lwz r2, 0(r3)
cmpw cr0, r2, r5
or r2, r7, r7
beq cr0, LBB_test_5 ; loopexit
LBB_test_4: ; endif
addi r2, r7, 1
addi r7, r7, 1
addi r3, r3, 4
addi r6, r6, 1
cmpw cr0, r6, r4
blt cr0, LBB_test_3 ; no_exit
Now we get:
li r6, 0
LBB_test_3: ; no_exit
or r2, r6, r6
lwz r6, 0(r3)
cmpw cr0, r6, r5
beq cr0, LBB_test_6 ; loopexit
LBB_test_4: ; endif
addi r3, r3, 4
addi r6, r2, 1
cmpw cr0, r6, r4
blt cr0, LBB_test_3 ; no_exit
this was noticed in em3d.
llvm-svn: 23602
2005-10-03 00:37:33 +00:00
8fcce170cf
when checking if we should move a split edge block outside of a loop,
...
check the presplit pred, not the post-split pred. This was causing us
to make the wrong decision in some cases, leaving the critical edge block
in the loop.
llvm-svn: 23601
2005-10-03 00:31:52 +00:00
f8a5e5ae6e
Fix VC++ warnings.
...
llvm-svn: 23579
2005-10-01 03:57:14 +00:00
a554c9470b
Insert stores after phi nodes in the normal dest. This fixes
...
LowerInvoke/2005-08-03-InvokeWithPHI.ll
llvm-svn: 23525
2005-09-29 17:44:20 +00:00
3b63bb375c
add a note about a way to improve this code further, that I won't be getting
...
to right now.
llvm-svn: 23485
2005-09-27 22:44:59 +00:00
e285f5ed8f
Avoid spilling stack slots... to stack slots.
...
llvm-svn: 23478
2005-09-27 21:33:12 +00:00
87eb249300
Completely rewrite 'correct' eh support. This changes how setjmp insertion
...
is performed so it is only at most once per function that contains an invoke
instead of once per invoke in the function. This patch has the following perks:
1. It fixes PR631, which complains about slowness.
2. If fixes PR240, which complains about non-volatile vars being live across
setjmp/longjmps.
3. It improves (but does not fix) the jmpbuf alignment issue on itanium by not
forcing the jmpbufs to always be 8-bytes off the alignment of the structure.
4. It speeds up 253.perlbmk from 338s to 13.70s (a 25x improvement!), making us
now about 4% faster than GCC.
Further improvements are also possible.
llvm-svn: 23477
2005-09-27 21:18:17 +00:00
92233d2175
Make the pass name simpler
...
llvm-svn: 23476
2005-09-27 21:10:32 +00:00
02ae21e1e0
Eliminate GetGEPGlobalInitializer in favor of the more powerful
...
ConstantFoldLoadThroughGEPConstantExpr function in the utils lib.
llvm-svn: 23446
2005-09-26 05:28:52 +00:00
0b011ec8e2
Factor the GetGEPGlobalInitializer out of this pass and into Transforms/Utils
...
as ConstantFoldLoadThroughGEPConstantExpr.
llvm-svn: 23445
2005-09-26 05:28:06 +00:00
0b3557f54a
Move MaskedValueIsZero up.
...
Match a bunch of idioms for sign extensions, implementing InstCombine/signext.ll
llvm-svn: 23428
2005-09-24 23:43:33 +00:00
b4b2530a1a
Refactor this code a bit and make it more general. This now compiles:
...
struct S { unsigned int i : 6, j : 11, k : 15; } b;
void plus2 (unsigned int x) { b.j += x; }
To:
_plus2:
lis r2, ha16(L_b$non_lazy_ptr)
lwz r2, lo16(L_b$non_lazy_ptr)(r2)
lwz r4, 0(r2)
slwi r3, r3, 6
add r3, r4, r3
rlwimi r3, r4, 0, 26, 14
stw r3, 0(r2)
blr
instead of:
_plus2:
lis r2, ha16(L_b$non_lazy_ptr)
lwz r2, lo16(L_b$non_lazy_ptr)(r2)
lwz r4, 0(r2)
rlwinm r5, r4, 26, 21, 31
add r3, r5, r3
rlwimi r4, r3, 6, 15, 25
stw r4, 0(r2)
blr
by eliminating an 'and'.
I'm pretty sure this is as small as we can go :)
llvm-svn: 23386
2005-09-18 07:22:02 +00:00
797dee7705
Compile
...
struct S { unsigned int i : 6, j : 11, k : 15; } b;
void plus2 (unsigned int x) {
b.j += x;
}
to:
plus2:
mov %EAX, DWORD PTR [b]
mov %ECX, %EAX
and %ECX, 131008
mov %EDX, DWORD PTR [%ESP + 4]
shl %EDX, 6
add %EDX, %ECX
and %EDX, 131008
and %EAX, -131009
or %EDX, %EAX
mov DWORD PTR [b], %EDX
ret
instead of:
plus2:
mov %EAX, DWORD PTR [b]
mov %ECX, %EAX
shr %ECX, 6
and %ECX, 2047
add %ECX, DWORD PTR [%ESP + 4]
shl %ECX, 6
and %ECX, 131008
and %EAX, -131009
or %ECX, %EAX
mov DWORD PTR [b], %ECX
ret
llvm-svn: 23385
2005-09-18 06:30:59 +00:00
01f56c68e9
Generalize this transform, using MaskedValueIsZero, allowing us to compile:
...
struct S { unsigned int i : 6, j : 11, k : 15; } b;
void plus3 (unsigned int x) { b.k += x; }
To:
plus3:
mov %EAX, DWORD PTR [%ESP + 4]
shl %EAX, 17
add DWORD PTR [b], %EAX
ret
instead of:
plus3:
mov %EAX, DWORD PTR [%ESP + 4]
shl %EAX, 17
mov %ECX, DWORD PTR [b]
add %EAX, %ECX
and %EAX, -131072
and %ECX, 131071
or %ECX, %EAX
mov DWORD PTR [b], %ECX
ret
llvm-svn: 23384
2005-09-18 06:02:59 +00:00
4ebc8ab4e0
fix typeo
...
llvm-svn: 23383
2005-09-18 05:25:20 +00:00
e5b23a6d67
Remove unintentionally committed code
...
llvm-svn: 23382
2005-09-18 05:12:51 +00:00
27cb9dbd35
implement shift.ll:test25. This compiles:
...
struct S { unsigned int i : 6, j : 11, k : 15; } b;
void plus3 (unsigned int x) {
b.k += x;
}
to:
_plus3:
lis r2, ha16(L_b$non_lazy_ptr)
lwz r2, lo16(L_b$non_lazy_ptr)(r2)
lwz r3, 0(r2)
rlwinm r4, r3, 0, 0, 14
add r4, r4, r3
rlwimi r4, r3, 0, 15, 31
stw r4, 0(r2)
blr
instead of:
_plus3:
lis r2, ha16(L_b$non_lazy_ptr)
lwz r2, lo16(L_b$non_lazy_ptr)(r2)
lwz r4, 0(r2)
srwi r5, r4, 17
add r3, r5, r3
slwi r3, r3, 17
rlwimi r3, r4, 0, 15, 31
stw r3, 0(r2)
blr
llvm-svn: 23381
2005-09-18 05:12:10 +00:00
af517574ce
Implement add.ll:test29. Codegening:
...
struct S { unsigned int i : 6, j : 11, k : 15; } b;
void plus1 (unsigned int x) {
b.i += x;
}
as:
_plus1:
lis r2, ha16(L_b$non_lazy_ptr)
lwz r2, lo16(L_b$non_lazy_ptr)(r2)
lwz r4, 0(r2)
add r3, r4, r3
rlwimi r3, r4, 0, 0, 25
stw r3, 0(r2)
blr
instead of:
_plus1:
lis r2, ha16(L_b$non_lazy_ptr)
lwz r2, lo16(L_b$non_lazy_ptr)(r2)
lwz r4, 0(r2)
rlwinm r5, r4, 0, 26, 31
add r3, r5, r3
rlwimi r3, r4, 0, 0, 25
stw r3, 0(r2)
blr
llvm-svn: 23379
2005-09-18 04:24:45 +00:00
027eaf01cf
remove debug output
...
llvm-svn: 23377
2005-09-18 03:50:25 +00:00
1521298993
Implement or.ll:test21. This teaches instcombine to be able to turn this:
...
struct {
unsigned int bit0:1;
unsigned int ubyte:31;
} sdata;
void foo() {
sdata.ubyte++;
}
into this:
foo:
add DWORD PTR [sdata], 2
ret
instead of this:
foo:
mov %EAX, DWORD PTR [sdata]
mov %ECX, %EAX
add %ECX, 2
and %ECX, -2
and %EAX, 1
or %EAX, %ECX
mov DWORD PTR [sdata], %EAX
ret
llvm-svn: 23376
2005-09-18 03:42:07 +00:00
a393e4d4b3
Fix the regression last night compiling povray
...
llvm-svn: 23348
2005-09-14 17:32:56 +00:00
2a8932960d
Add a simple xform to simplify array accesses with casts in the way.
...
This is useful for 178.galgel where resolution of dope vectors (by the
optimizer) causes the scales to become apparent.
llvm-svn: 23328
2005-09-13 18:36:04 +00:00
fd018c8dfe
Fix an issue where LSR would miss rewriting a use of an IV expression by a PHI node that is not the original PHI.
...
This fixes up a dot-product loop in galgel, speeding it up from 18.47s to
16.13s.
llvm-svn: 23327
2005-09-13 02:09:55 +00:00
567b81f0d2
Add a helper function, allowing us to simplify some code a bit, changing
...
indentation, no functionality change
llvm-svn: 23325
2005-09-13 00:40:14 +00:00
219175c84d
Implement a simple xform to turn code like this:
...
if () { store A -> P; } else { store B -> P; }
into a PHI node with one store, in the most trival case. This implements
load.ll:test10.
llvm-svn: 23324
2005-09-12 23:23:25 +00:00
e0bfdf1485
Another load-peephole optimization: do gcse when two loads are next to
...
each other. This implements InstCombine/load.ll:test9
llvm-svn: 23322
2005-09-12 22:21:03 +00:00
b990f7d8ed
Implement a trivial form of store->load forwarding where the store and the
...
load are exactly consequtive. This is picked up by other passes, but this
triggers thousands of times in fortran programs that use static locals
(and is thus a compile-time speedup).
llvm-svn: 23320
2005-09-12 22:00:15 +00:00
8048b85e8f
Fix a regression from last night, which caused this pass to create invalid
...
code for IV uses outside of loops that are not dominated by the latch block.
We should only convert these uses to use the post-inc value if they ARE
dominated by the latch block.
Also use a new LoopInfo method to simplify some code.
This fixes Transforms/LoopStrengthReduce/2005-09-12-UsesOutOutsideOfLoop.ll
llvm-svn: 23318
2005-09-12 17:11:27 +00:00
a67648396a
_test:
...
li r2, 0
LBB_test_1: ; no_exit.2
li r5, 0
stw r5, 0(r3)
addi r2, r2, 1
addi r3, r3, 4
cmpwi cr0, r2, 701
blt cr0, LBB_test_1 ; no_exit.2
LBB_test_2: ; loopexit.2.loopexit
addi r2, r2, 1
stw r2, 0(r4)
blr
[zion ~/llvm]$ cat > ~/xx
Uses of IV's outside of the loop should use hte post-incremented version
of the IV, not the preincremented version. This helps many loops (e.g. in sixtrack)
which used to generate code like this (this is the code from the
dont-hoist-simple-loop-constants.ll testcase):
_test:
li r2, 0 **** IV starts at 0
LBB_test_1: ; no_exit.2
or r5, r2, r2 **** Copy for loop exit
li r2, 0
stw r2, 0(r3)
addi r3, r3, 4
addi r2, r5, 1
addi r6, r5, 2 **** IV+2
cmpwi cr0, r6, 701
blt cr0, LBB_test_1 ; no_exit.2
LBB_test_2: ; loopexit.2.loopexit
addi r2, r5, 2 **** IV+2
stw r2, 0(r4)
blr
And now generated code like this:
_test:
li r2, 1 *** IV starts at 1
LBB_test_1: ; no_exit.2
li r5, 0
stw r5, 0(r3)
addi r2, r2, 1
addi r3, r3, 4
cmpwi cr0, r2, 701 *** IV.postinc + 0
blt cr0, LBB_test_1
LBB_test_2: ; loopexit.2.loopexit
stw r2, 0(r4) *** IV.postinc + 0
blr
llvm-svn: 23313
2005-09-12 06:04:47 +00:00
530fe6ab30
implement Transforms/LoopStrengthReduce/dont-hoist-simple-loop-constants.ll.
...
We used to emit this code for it:
_test:
li r2, 1 ;; Value tying up a register for the whole loop
li r5, 0
LBB_test_1: ; no_exit.2
or r6, r5, r5
li r5, 0
stw r5, 0(r3)
addi r5, r6, 1
addi r3, r3, 4
add r7, r2, r5 ;; should be addi r7, r5, 1
cmpwi cr0, r7, 701
blt cr0, LBB_test_1 ; no_exit.2
LBB_test_2: ; loopexit.2.loopexit
addi r2, r6, 2
stw r2, 0(r4)
blr
now we emit this:
_test:
li r2, 0
LBB_test_1: ; no_exit.2
or r5, r2, r2
li r2, 0
stw r2, 0(r3)
addi r3, r3, 4
addi r2, r5, 1
addi r6, r5, 2 ;; whoa, fold those adds!
cmpwi cr0, r6, 701
blt cr0, LBB_test_1 ; no_exit.2
LBB_test_2: ; loopexit.2.loopexit
addi r2, r5, 2
stw r2, 0(r4)
blr
more improvement coming.
llvm-svn: 23306
2005-09-10 01:18:45 +00:00
b5e381a8cf
Fix a problem that Dan Berlin noticed, where reassociation would not succeed
...
in building maximal expressions before simplifying them. In particular, i
cases like this:
X-(A+B+X)
the code would consider A+B+X to be a maximal expression (not understanding
that the single use '-' would be turned into a + later), simplify it (a noop)
then later get simplified again.
Each of these simplify steps is where the cost of reassociation comes from,
so this patch should speed up the already fast pass a bit.
Thanks to Dan for noticing this!
llvm-svn: 23214
2005-09-02 07:07:58 +00:00
9fe263aa75
Avoid creating garbage instructions, just move the old add instruction
...
to where we need it when converting -(A+B+C) -> -A + -B + -C.
llvm-svn: 23213
2005-09-02 06:38:04 +00:00
d1325da091
add some assertions and fix problems where reassociate could access the
...
Ops vector out of range
llvm-svn: 23211
2005-09-02 05:23:22 +00:00
8ca5b2a6d2
Fix Regression/Transforms/Reassociate/2005-08-24-Crash.ll
...
llvm-svn: 23019
2005-08-24 17:55:32 +00:00
ea7dfd53d6
Fix Transforms/LoopStrengthReduce/2005-08-17-OutOfLoopVariant.ll, a crash
...
on 177.mesa
llvm-svn: 22843
2005-08-17 21:22:41 +00:00
2bf7cb5213
Use a new helper to split critical edges, making the code simpler.
...
Do not claim to not change the CFG. We do change the cfg to split critical
edges. This isn't causing us a problem now, but could likely do so in the
future.
llvm-svn: 22824
2005-08-17 06:35:16 +00:00
5cf983ee0f
Fix a bad case in gzip where we put lots of things in registers across the
...
loop, because a IV-dependent value was used outside of the loop and didn't
have immediate-folding capability
llvm-svn: 22798
2005-08-16 00:38:11 +00:00
47d3ec3525
Ooops, don't forget to clear this. The real inner loop is now:
...
.LBB_foo_3: ; no_exit.1
lfd f2, 0(r9)
lfd f3, 8(r9)
fmul f4, f1, f2
fmadd f4, f0, f3, f4
stfd f4, 8(r9)
fmul f3, f1, f3
fmsub f2, f0, f2, f3
stfd f2, 0(r9)
addi r9, r9, 16
addi r8, r8, 1
cmpw cr0, r8, r4
ble .LBB_foo_3 ; no_exit.1
llvm-svn: 22782
2005-08-13 07:42:01 +00:00
5949d49032
Recursively scan scev expressions for common subexpressions. This allows us
...
to handle nested loops much better, for example, by being able to tell that
these two expressions:
{( 8 + ( 16 * ( 1 + %Tmp11 + %Tmp12)) + %c_),+,( 16 * %Tmp 12)}<loopentry.1>
{(( 16 * ( 1 + %Tmp11 + %Tmp12)) + %c_),+,( 16 * %Tmp12)}<loopentry.1>
Have the following common part that can be shared:
{(( 16 * ( 1 + %Tmp11 + %Tmp12)) + %c_),+,( 16 * %Tmp12)}<loopentry.1>
This allows us to codegen an important inner loop in 168.wupwise as:
.LBB_foo_4: ; no_exit.1
lfd f2, 16(r9)
fmul f3, f0, f2
fmul f2, f1, f2
fadd f4, f3, f2
stfd f4, 8(r9)
fsub f2, f3, f2
stfd f2, 16(r9)
addi r8, r8, 1
addi r9, r9, 16
cmpw cr0, r8, r4
ble .LBB_foo_4 ; no_exit.1
instead of:
.LBB_foo_3: ; no_exit.1
lfdx f2, r6, r9
add r10, r6, r9
lfd f3, 8(r10)
fmul f4, f1, f2
fmadd f4, f0, f3, f4
stfd f4, 8(r10)
fmul f3, f1, f3
fmsub f2, f0, f2, f3
stfdx f2, r6, r9
addi r9, r9, 16
addi r8, r8, 1
cmpw cr0, r8, r4
ble .LBB_foo_3 ; no_exit.1
llvm-svn: 22781
2005-08-13 07:27:18 +00:00
79396539d3
remove dead code. The exit block list is computed on demand, thus does not
...
need to be updated. This code is a relic from when it did.
llvm-svn: 22775
2005-08-13 01:30:36 +00:00
8447b49526
When splitting critical edges, make sure not to leave the new block in the
...
middle of the loop. This turns a critical loop in gzip into this:
.LBB_test_1: ; loopentry
or r27, r28, r28
add r28, r3, r27
lhz r28, 3(r28)
add r26, r4, r27
lhz r26, 3(r26)
cmpw cr0, r28, r26
bne .LBB_test_8 ; loopentry.loopexit_crit_edge
.LBB_test_2: ; shortcirc_next.0
add r28, r3, r27
lhz r28, 5(r28)
add r26, r4, r27
lhz r26, 5(r26)
cmpw cr0, r28, r26
bne .LBB_test_7 ; shortcirc_next.0.loopexit_crit_edge
.LBB_test_3: ; shortcirc_next.1
add r28, r3, r27
lhz r28, 7(r28)
add r26, r4, r27
lhz r26, 7(r26)
cmpw cr0, r28, r26
bne .LBB_test_6 ; shortcirc_next.1.loopexit_crit_edge
.LBB_test_4: ; shortcirc_next.2
add r28, r3, r27
lhz r26, 9(r28)
add r28, r4, r27
lhz r25, 9(r28)
addi r28, r27, 8
cmpw cr7, r26, r25
mfcr r26, 1
rlwinm r26, r26, 31, 31, 31
add r25, r8, r27
cmpw cr7, r25, r7
mfcr r25, 1
rlwinm r25, r25, 29, 31, 31
and. r26, r26, r25
bne .LBB_test_1 ; loopentry
instead of this:
.LBB_test_1: ; loopentry
or r27, r28, r28
add r28, r3, r27
lhz r28, 3(r28)
add r26, r4, r27
lhz r26, 3(r26)
cmpw cr0, r28, r26
beq .LBB_test_3 ; shortcirc_next.0
.LBB_test_2: ; loopentry.loopexit_crit_edge
add r2, r30, r27
add r8, r29, r27
b .LBB_test_9 ; loopexit
.LBB_test_3: ; shortcirc_next.0
add r28, r3, r27
lhz r28, 5(r28)
add r26, r4, r27
lhz r26, 5(r26)
cmpw cr0, r28, r26
beq .LBB_test_5 ; shortcirc_next.1
.LBB_test_4: ; shortcirc_next.0.loopexit_crit_edge
add r2, r11, r27
add r8, r12, r27
b .LBB_test_9 ; loopexit
.LBB_test_5: ; shortcirc_next.1
add r28, r3, r27
lhz r28, 7(r28)
add r26, r4, r27
lhz r26, 7(r26)
cmpw cr0, r28, r26
beq .LBB_test_7 ; shortcirc_next.2
.LBB_test_6: ; shortcirc_next.1.loopexit_crit_edge
add r2, r9, r27
add r8, r10, r27
b .LBB_test_9 ; loopexit
.LBB_test_7: ; shortcirc_next.2
add r28, r3, r27
lhz r26, 9(r28)
add r28, r4, r27
lhz r25, 9(r28)
addi r28, r27, 8
cmpw cr7, r26, r25
mfcr r26, 1
rlwinm r26, r26, 31, 31, 31
add r25, r8, r27
cmpw cr7, r25, r7
mfcr r25, 1
rlwinm r25, r25, 29, 31, 31
and. r26, r26, r25
bne .LBB_test_1 ; loopentry
Next up, improve the code for the loop.
llvm-svn: 22769
2005-08-12 22:22:17 +00:00
4fec86d348
Fix a FIXME: if we are inserting code for a PHI argument, split the critical
...
edge so that the code is not always executed for both operands. This
prevents LSR from inserting code into loops whose exit blocks contain
PHI uses of IV expressions (which are outside of loops). On gzip, for
example, we turn this ugly code:
.LBB_test_1: ; loopentry
add r27, r3, r28
lhz r27, 3(r27)
add r26, r4, r28
lhz r26, 3(r26)
add r25, r30, r28 ;; Only live if exiting the loop
add r24, r29, r28 ;; Only live if exiting the loop
cmpw cr0, r27, r26
bne .LBB_test_5 ; loopexit
into this:
.LBB_test_1: ; loopentry
or r27, r28, r28
add r28, r3, r27
lhz r28, 3(r28)
add r26, r4, r27
lhz r26, 3(r26)
cmpw cr0, r28, r26
beq .LBB_test_3 ; shortcirc_next.0
.LBB_test_2: ; loopentry.loopexit_crit_edge
add r2, r30, r27
add r8, r29, r27
b .LBB_test_9 ; loopexit
.LBB_test_2: ; shortcirc_next.0
...
blt .LBB_test_1
into this:
.LBB_test_1: ; loopentry
or r27, r28, r28
add r28, r3, r27
lhz r28, 3(r28)
add r26, r4, r27
lhz r26, 3(r26)
cmpw cr0, r28, r26
beq .LBB_test_3 ; shortcirc_next.0
.LBB_test_2: ; loopentry.loopexit_crit_edge
add r2, r30, r27
add r8, r29, r27
b .LBB_t_3: ; shortcirc_next.0
.LBB_test_3: ; shortcirc_next.0
...
blt .LBB_test_1
Next step: get the block out of the loop so that the loop is all
fall-throughs again.
llvm-svn: 22766
2005-08-12 22:06:11 +00:00
62df798919
remove some trickiness that broke yacr2 and some other programs last night
...
llvm-svn: 22751
2005-08-10 17:15:20 +00:00
f83ce5faee
Make loop-simplify produce better loops by turning PHI nodes like X = phi [X, Y]
...
into just Y. This often occurs when it seperates loops that have collapsed loop
headers. This implements LoopSimplify/phi-node-simplify.ll
llvm-svn: 22746
2005-08-10 02:07:32 +00:00
677d85784a
Allow indvar simplify to canonicalize ANY affine IV, not just affine IVs with
...
constant stride. This implements Transforms/IndVarsSimplify/variable-stride-ivs.ll
llvm-svn: 22744
2005-08-10 01:12:06 +00:00
edff91a49a
Teach LSR to strength reduce IVs that have a loop-invariant but non-constant stride.
...
For code like this:
void foo(float *a, float *b, int n, int stride_a, int stride_b) {
int i;
for (i=0; i<n; i++)
a[i*stride_a] = b[i*stride_b];
}
we now emit:
.LBB_foo2_2: ; no_exit
lfs f0, 0(r4)
stfs f0, 0(r3)
addi r7, r7, 1
add r4, r2, r4
add r3, r6, r3
cmpw cr0, r7, r5
blt .LBB_foo2_2 ; no_exit
instead of:
.LBB_foo_2: ; no_exit
mullw r8, r2, r7 ;; multiply!
slwi r8, r8, 2
lfsx f0, r4, r8
mullw r8, r2, r6 ;; multiply!
slwi r8, r8, 2
stfsx f0, r3, r8
addi r2, r2, 1
cmpw cr0, r2, r5
blt .LBB_foo_2 ; no_exit
loops with variable strides occur pretty often. For example, in SPECFP2K
there are 317 variable strides in 177.mesa, 3 in 179.art, 14 in 188.ammp,
56 in 168.wupwise, 36 in 172.mgrid.
Now we can allow indvars to turn functions written like this:
void foo2(float *a, float *b, int n, int stride_a, int stride_b) {
int i, ai = 0, bi = 0;
for (i=0; i<n; i++)
{
a[ai] = b[bi];
ai += stride_a;
bi += stride_b;
}
}
into code like the above for better analysis. With this patch, they generate
identical code.
llvm-svn: 22740
2005-08-10 00:45:21 +00:00
dde7dc525e
Fix Regression/Transforms/LoopStrengthReduce/phi_node_update_multiple_preds.ll
...
by being more careful about updating PHI nodes
llvm-svn: 22739
2005-08-10 00:35:32 +00:00
c6c4d99a21
Fix some 80 column violations.
...
Once we compute the evolution for a GEP, tell SE about it. This allows users
of the GEP to know it, if the users are not direct. This allows us to compile
this testcase:
void fbSolidFillmmx(int w, unsigned char *d) {
while (w >= 64) {
*(unsigned long long *) (d + 0) = 0;
*(unsigned long long *) (d + 8) = 0;
*(unsigned long long *) (d + 16) = 0;
*(unsigned long long *) (d + 24) = 0;
*(unsigned long long *) (d + 32) = 0;
*(unsigned long long *) (d + 40) = 0;
*(unsigned long long *) (d + 48) = 0;
*(unsigned long long *) (d + 56) = 0;
w -= 64;
d += 64;
}
}
into:
.LBB_fbSolidFillmmx_2: ; no_exit
li r2, 0
stw r2, 0(r4)
stw r2, 4(r4)
stw r2, 8(r4)
stw r2, 12(r4)
stw r2, 16(r4)
stw r2, 20(r4)
stw r2, 24(r4)
stw r2, 28(r4)
stw r2, 32(r4)
stw r2, 36(r4)
stw r2, 40(r4)
stw r2, 44(r4)
stw r2, 48(r4)
stw r2, 52(r4)
stw r2, 56(r4)
stw r2, 60(r4)
addi r4, r4, 64
addi r3, r3, -64
cmpwi cr0, r3, 63
bgt .LBB_fbSolidFillmmx_2 ; no_exit
instead of:
.LBB_fbSolidFillmmx_2: ; no_exit
li r11, 0
stw r11, 0(r4)
stw r11, 4(r4)
stwx r11, r10, r4
add r12, r10, r4
stw r11, 4(r12)
stwx r11, r9, r4
add r12, r9, r4
stw r11, 4(r12)
stwx r11, r8, r4
add r12, r8, r4
stw r11, 4(r12)
stwx r11, r7, r4
add r12, r7, r4
stw r11, 4(r12)
stwx r11, r6, r4
add r12, r6, r4
stw r11, 4(r12)
stwx r11, r5, r4
add r12, r5, r4
stw r11, 4(r12)
stwx r11, r2, r4
add r12, r2, r4
stw r11, 4(r12)
addi r4, r4, 64
addi r3, r3, -64
cmpwi cr0, r3, 63
bgt .LBB_fbSolidFillmmx_2 ; no_exit
llvm-svn: 22737
2005-08-09 23:39:36 +00:00
02742710f3
SCEVAddExpr::get() of an empty list is invalid.
...
llvm-svn: 22724
2005-08-09 01:13:47 +00:00
a091ff1764
Implement: LoopStrengthReduce/share_ivs.ll
...
Two changes:
* Only insert one PHI node for each stride. Other values are live in
values. This cannot introduce higher register pressure than the
previous approach, and can take advantage of reg+reg addressing modes.
* Factor common base values out of uses before moving values from the
base to the immediate fields. This improves codegen by starting the
stride-specific PHI node out at a common place for each IV use.
As an example, we used to generate this for a loop in swim:
.LBB_main_no_exit_2E_6_2E_i_no_exit_2E_7_2E_i_2: ; no_exit.7.i
lfd f0, 0(r8)
stfd f0, 0(r3)
lfd f0, 0(r6)
stfd f0, 0(r7)
lfd f0, 0(r2)
stfd f0, 0(r5)
addi r9, r9, 1
addi r2, r2, 8
addi r5, r5, 8
addi r6, r6, 8
addi r7, r7, 8
addi r8, r8, 8
addi r3, r3, 8
cmpw cr0, r9, r4
bgt .LBB_main_no_exit_2E_6_2E_i_no_exit_2E_7_2E_i_1
now we emit:
.LBB_main_no_exit_2E_6_2E_i_no_exit_2E_7_2E_i_2: ; no_exit.7.i
lfdx f0, r8, r2
stfdx f0, r9, r2
lfdx f0, r5, r2
stfdx f0, r7, r2
lfdx f0, r3, r2
stfdx f0, r6, r2
addi r10, r10, 1
addi r2, r2, 8
cmpw cr0, r10, r4
bgt .LBB_main_no_exit_2E_6_2E_i_no_exit_2E_7_2E_i_1
As another more dramatic example, we used to emit this:
.LBB_main_L_90_no_exit_2E_0_2E_i16_no_exit_2E_1_2E_i19_2: ; no_exit.1.i19
lfd f0, 8(r21)
lfd f4, 8(r3)
lfd f5, 8(r27)
lfd f6, 8(r22)
lfd f7, 8(r5)
lfd f8, 8(r6)
lfd f9, 8(r30)
lfd f10, 8(r11)
lfd f11, 8(r12)
fsub f10, f10, f11
fadd f5, f4, f5
fmul f5, f5, f1
fadd f6, f6, f7
fadd f6, f6, f8
fadd f6, f6, f9
fmadd f0, f5, f6, f0
fnmsub f0, f10, f2, f0
stfd f0, 8(r4)
lfd f0, 8(r25)
lfd f5, 8(r26)
lfd f6, 8(r23)
lfd f9, 8(r28)
lfd f10, 8(r10)
lfd f12, 8(r9)
lfd f13, 8(r29)
fsub f11, f13, f11
fadd f4, f4, f5
fmul f4, f4, f1
fadd f5, f6, f9
fadd f5, f5, f10
fadd f5, f5, f12
fnmsub f0, f4, f5, f0
fnmsub f0, f11, f3, f0
stfd f0, 8(r24)
lfd f0, 8(r8)
fsub f4, f7, f8
fsub f5, f12, f10
fnmsub f0, f5, f2, f0
fnmsub f0, f4, f3, f0
stfd f0, 8(r2)
addi r20, r20, 1
addi r2, r2, 8
addi r8, r8, 8
addi r10, r10, 8
addi r12, r12, 8
addi r6, r6, 8
addi r29, r29, 8
addi r28, r28, 8
addi r26, r26, 8
addi r25, r25, 8
addi r24, r24, 8
addi r5, r5, 8
addi r23, r23, 8
addi r22, r22, 8
addi r3, r3, 8
addi r9, r9, 8
addi r11, r11, 8
addi r30, r30, 8
addi r27, r27, 8
addi r21, r21, 8
addi r4, r4, 8
cmpw cr0, r20, r7
bgt .LBB_main_L_90_no_exit_2E_0_2E_i16_no_exit_2E_1_2E_i19_1
we now emit:
.LBB_main_L_90_no_exit_2E_0_2E_i16_no_exit_2E_1_2E_i19_2: ; no_exit.1.i19
lfdx f0, r21, r20
lfdx f4, r3, r20
lfdx f5, r27, r20
lfdx f6, r22, r20
lfdx f7, r5, r20
lfdx f8, r6, r20
lfdx f9, r30, r20
lfdx f10, r11, r20
lfdx f11, r12, r20
fsub f10, f10, f11
fadd f5, f4, f5
fmul f5, f5, f1
fadd f6, f6, f7
fadd f6, f6, f8
fadd f6, f6, f9
fmadd f0, f5, f6, f0
fnmsub f0, f10, f2, f0
stfdx f0, r4, r20
lfdx f0, r25, r20
lfdx f5, r26, r20
lfdx f6, r23, r20
lfdx f9, r28, r20
lfdx f10, r10, r20
lfdx f12, r9, r20
lfdx f13, r29, r20
fsub f11, f13, f11
fadd f4, f4, f5
fmul f4, f4, f1
fadd f5, f6, f9
fadd f5, f5, f10
fadd f5, f5, f12
fnmsub f0, f4, f5, f0
fnmsub f0, f11, f3, f0
stfdx f0, r24, r20
lfdx f0, r8, r20
fsub f4, f7, f8
fsub f5, f12, f10
fnmsub f0, f5, f2, f0
fnmsub f0, f4, f3, f0
stfdx f0, r2, r20
addi r19, r19, 1
addi r20, r20, 8
cmpw cr0, r19, r7
bgt .LBB_main_L_90_no_exit_2E_0_2E_i16_no_exit_2E_1_2E_i19_1
llvm-svn: 22722
2005-08-09 00:18:09 +00:00
37c24cc98c
Suck the base value out of the UsersToProcess vector into the BasedUser
...
class to simplify the code. Fuse two loops.
llvm-svn: 22721
2005-08-08 22:56:21 +00:00
37ed895bf1
Split MoveLoopVariantsToImediateField out from MoveImmediateValues. The
...
first is a correctness thing, and the later is an optzn thing. This also
is needed to support a future change.
llvm-svn: 22720
2005-08-08 22:32:34 +00:00
9f269e40c9
Use the new 'moveBefore' method to simplify some code. Really, which is
...
easier to understand? :)
llvm-svn: 22706
2005-08-08 19:11:57 +00:00
14203e85b2
Not all constants are legal immediates in load/store instructions.
...
llvm-svn: 22704
2005-08-08 06:25:50 +00:00
c70bbc0c41
Implement LoopStrengthReduce/share_code_in_preheader.ll by having one
...
rewriter for all code inserted into the preheader, which is never flushed.
llvm-svn: 22702
2005-08-08 05:47:49 +00:00
9bfa6f8784
Implement a simple optimization for the termination condition of the loop.
...
The termination condition actually wants to use the post-incremented value
of the loop, not a new indvar with an unusual base.
On PPC, for example, this allows us to compile
LoopStrengthReduce/exit_compare_live_range.ll to:
_foo:
li r2, 0
.LBB_foo_1: ; no_exit
li r5, 0
stw r5, 0(r3)
addi r2, r2, 1
cmpw cr0, r2, r4
bne .LBB_foo_1 ; no_exit
blr
instead of:
_foo:
li r2, 1 ;; IV starts at 1, not 0
.LBB_foo_1: ; no_exit
li r5, 0
stw r5, 0(r3)
addi r5, r2, 1
cmpw cr0, r2, r4
or r2, r5, r5 ;; Reg-reg copy, extra live range
bne .LBB_foo_1 ; no_exit
blr
This implements LoopStrengthReduce/exit_compare_live_range.ll
llvm-svn: 22699
2005-08-08 05:28:22 +00:00
2c14cf7b74
Add some simple folds that occur in bitfield cases. Fix a minor bug in
...
isHighOnes, where it would consider 0 to have high ones.
llvm-svn: 22693
2005-08-07 07:03:10 +00:00
134ebd0801
Fix typoCVS: ----------------------------------------------------------------------
...
llvm-svn: 22692
2005-08-07 07:00:52 +00:00
f4dd8c445c
* Use the new PHINode::hasConstantValue method to simplify some code
...
* Teach this code to move allocas out of the loop when tail call eliminating
a call marked 'tail'. This implements TailCallElim/move_alloca_for_tail_call.ll
* Do not perform this transformation if a call is marked 'tail' and if there
are allocas that we cannot move out of the loop in #2 . Doing so would increase
the stack usage of the function. This implements fixes
PR615 and TailCallElim/dont-tce-tail-marked-call.ll.
llvm-svn: 22690
2005-08-07 04:27:41 +00:00
11e7a5eda7
Make sure to clean CastedPointers after casts are potentially deleted.
...
This fixes LSR crashes on 301.apsi, 191.fma3d, and 189.lucas
llvm-svn: 22673
2005-08-05 01:30:11 +00:00
9f9c260b8c
now that hasConstantValue defaults to only returning values that dominate
...
the PHI node, this ugly code can vanish.
llvm-svn: 22672
2005-08-05 01:04:30 +00:00
257efb2ad3
This code can handle non-dominating instructions
...
llvm-svn: 22667
2005-08-05 00:57:45 +00:00
b392321cae
Fix a fixme in CondPropagate.cpp by moving a PhiNode optimization into
...
BasicBlock's removePredecessor routine. This requires shuffling around
the definition and implementation of hasContantValue from Utils.h,cpp into
Instructions.h,cpp
llvm-svn: 22664
2005-08-04 23:24:19 +00:00
45f8b6e7aa
Modify how immediates are removed from base expressions to deal with the fact
...
that the symbolic evaluator is not always able to use subtraction to remove
expressions. This makes the code faster, and fixes the last crash on 178.galgel.
Finally, add a statistic to see how many phi nodes are inserted.
On 178.galgel, we get the follow stats:
2562 loop-reduce - Number of PHIs inserted
3927 loop-reduce - Number of GEPs strength reduced
llvm-svn: 22662
2005-08-04 22:34:05 +00:00
a6d7c355bc
* Refactor some code into a new BasedUser::RewriteInstructionToUseNewBase
...
method.
* Fix a crash on 178.galgel, where we would insert expressions before PHI
nodes instead of into the PHI node predecessor blocks.
llvm-svn: 22657
2005-08-04 20:03:32 +00:00
0f7c0fa2a7
Fix a case that caused this to crash on 178.galgel
...
llvm-svn: 22653
2005-08-04 19:26:19 +00:00
acc42c4df1
Teach LSR about loop-variant expressions, such as loops like this:
...
for (i = 0; i < N; ++i)
A[i][foo()] = 0;
here we still want to strength reduce the A[i] part, even though foo() is
l-v.
This also simplifies some of the 'CanReduce' logic.
This implements Transforms/LoopStrengthReduce/ops_after_indvar.ll
llvm-svn: 22652
2005-08-04 19:08:16 +00:00
456044b724
Remove some more dead code.
...
llvm-svn: 22650
2005-08-04 18:13:56 +00:00
eaf24725b2
Refactor this code substantially with the following improvements:
...
1. We only analyze instructions once, guaranteed
2. AnalyzeGetElementPtrUsers has been ripped apart and replaced with
something much simpler.
The next step is to handle expressions that are not all indvar+loop-invariant
values (e.g. handling indvar+loopvariant).
llvm-svn: 22649
2005-08-04 17:40:30 +00:00
6f286b760f
refactor some code
...
llvm-svn: 22643
2005-08-04 01:19:13 +00:00
6510749050
invert to if's to make the logic simpler
...
llvm-svn: 22641
2005-08-04 00:40:47 +00:00
a0102fbc4f
When processing outer loops and we find uses of an IV in inner loops, make
...
sure to handle the use, just don't recurse into it.
This permits us to generate this code for a simple nested loop case:
.LBB_foo_0: ; entry
stwu r1, -48(r1)
stw r29, 44(r1)
stw r30, 40(r1)
mflr r11
stw r11, 56(r1)
lis r2, ha16(L_A$non_lazy_ptr)
lwz r30, lo16(L_A$non_lazy_ptr)(r2)
li r29, 1
.LBB_foo_1: ; no_exit.0
bl L_bar$stub
li r2, 1
or r3, r30, r30
.LBB_foo_2: ; no_exit.1
lfd f0, 8(r3)
stfd f0, 0(r3)
addi r4, r2, 1
addi r3, r3, 8
cmpwi cr0, r2, 100
or r2, r4, r4
bne .LBB_foo_2 ; no_exit.1
.LBB_foo_3: ; loopexit.1
addi r30, r30, 800
addi r2, r29, 1
cmpwi cr0, r29, 100
or r29, r2, r2
bne .LBB_foo_1 ; no_exit.0
.LBB_foo_4: ; return
lwz r11, 56(r1)
mtlr r11
lwz r30, 40(r1)
lwz r29, 44(r1)
lwz r1, 0(r1)
blr
instead of this:
_foo:
.LBB_foo_0: ; entry
stwu r1, -48(r1)
stw r28, 44(r1) ;; uses an extra register.
stw r29, 40(r1)
stw r30, 36(r1)
mflr r11
stw r11, 56(r1)
li r30, 1
li r29, 0
or r28, r29, r29
.LBB_foo_1: ; no_exit.0
bl L_bar$stub
mulli r2, r28, 800 ;; unstrength-reduced multiply
lis r3, ha16(L_A$non_lazy_ptr) ;; loop invariant address computation
lwz r3, lo16(L_A$non_lazy_ptr)(r3)
add r2, r2, r3
mulli r4, r29, 800 ;; unstrength-reduced multiply
addi r3, r3, 8
add r3, r4, r3
li r4, 1
.LBB_foo_2: ; no_exit.1
lfd f0, 0(r3)
stfd f0, 0(r2)
addi r5, r4, 1
addi r2, r2, 8 ;; multiple stride 8 IV's
addi r3, r3, 8
cmpwi cr0, r4, 100
or r4, r5, r5
bne .LBB_foo_2 ; no_exit.1
.LBB_foo_3: ; loopexit.1
addi r28, r28, 1 ;;; Many IV's with stride 1
addi r29, r29, 1
addi r2, r30, 1
cmpwi cr0, r30, 100
or r30, r2, r2
bne .LBB_foo_1 ; no_exit.0
.LBB_foo_4: ; return
lwz r11, 56(r1)
mtlr r11
lwz r30, 36(r1)
lwz r29, 40(r1)
lwz r28, 44(r1)
lwz r1, 0(r1)
blr
llvm-svn: 22640
2005-08-04 00:14:11 +00:00
fc62470466
Teach loop-reduce to see into nested loops, to pull out immediate values
...
pushed down by SCEV.
In a nested loop case, this allows us to emit this:
lis r3, ha16(L_A$non_lazy_ptr)
lwz r3, lo16(L_A$non_lazy_ptr)(r3)
add r2, r2, r3
li r3, 1
.LBB_foo_2: ; no_exit.1
lfd f0, 8(r2) ;; Uses offset of 8 instead of 0
stfd f0, 0(r2)
addi r4, r3, 1
addi r2, r2, 8
cmpwi cr0, r3, 100
or r3, r4, r4
bne .LBB_foo_2 ; no_exit.1
instead of this:
lis r3, ha16(L_A$non_lazy_ptr)
lwz r3, lo16(L_A$non_lazy_ptr)(r3)
add r2, r2, r3
addi r3, r3, 8
li r4, 1
.LBB_foo_2: ; no_exit.1
lfd f0, 0(r3)
stfd f0, 0(r2)
addi r5, r4, 1
addi r2, r2, 8
addi r3, r3, 8
cmpwi cr0, r4, 100
or r4, r5, r5
bne .LBB_foo_2 ; no_exit.1
llvm-svn: 22639
2005-08-03 23:44:42 +00:00
bb78c97e24
improve debug output
...
llvm-svn: 22638
2005-08-03 23:30:08 +00:00
db23c74e5e
Move from Stage 0 to Stage 1.
...
Only emit one PHI node for IV uses with identical bases and strides (after
moving foldable immediates to the load/store instruction).
This implements LoopStrengthReduce/dont_insert_redundant_ops.ll, allowing
us to generate this PPC code for test1:
or r30, r3, r3
.LBB_test1_1: ; Loop
li r2, 0
stw r2, 0(r30)
stw r2, 4(r30)
bl L_pred$stub
addi r30, r30, 8
cmplwi cr0, r3, 0
bne .LBB_test1_1 ; Loop
instead of this code:
or r30, r3, r3
or r29, r3, r3
.LBB_test1_1: ; Loop
li r2, 0
stw r2, 0(r29)
stw r2, 4(r30)
bl L_pred$stub
addi r30, r30, 8 ;; Two iv's with step of 8
addi r29, r29, 8
cmplwi cr0, r3, 0
bne .LBB_test1_1 ; Loop
llvm-svn: 22635
2005-08-03 22:51:21 +00:00
430d0022df
Rename IVUse to IVUsersOfOneStride, use a struct instead of a pair to
...
unify some parallel vectors and get field names more descriptive than
"first" and "second". This isn't lisp afterall :)
llvm-svn: 22633
2005-08-03 22:21:05 +00:00
84e9baa925
Fix a nasty dangling pointer issue. The ScalarEvolution pass would keep a
...
map from instruction* to SCEVHandles. When we delete instructions, we have
to tell it about it. We would run into nasty cases where new instructions
were reallocated at old instruction addresses and get the old map values.
Bad bad bad :(
llvm-svn: 22632
2005-08-03 21:36:09 +00:00
3de05cc930
The correct fix for PR612, which also fixes
...
Transforms/LowerInvoke/2005-08-03-InvokeWithPHIUse.ll
llvm-svn: 22628
2005-08-03 18:51:44 +00:00
f8a81a9886
When inserting code, make sure not to insert it before PHI nodes. This
...
fixes PR612 and Transforms/LowerInvoke/2005-08-03-InvokeWithPHI.ll
llvm-svn: 22626
2005-08-03 18:34:29 +00:00
22d00a8e90
Update to use the new MathExtras.h support for log2 computation.
...
Patch contributed by Jim Laskey!
llvm-svn: 22592
2005-08-02 19:16:58 +00:00
351b891cbc
Like the comment says, do not insert cast instructions before phi nodes
...
llvm-svn: 22586
2005-08-02 03:31:14 +00:00
75a44e154e
add a comment, make a check more lenient
...
llvm-svn: 22581
2005-08-02 02:52:02 +00:00
dcce49e006
Simplify for loop, clear a per-loop map after processing each loop
...
llvm-svn: 22580
2005-08-02 02:44:31 +00:00
9ef1294210
Add a comment
...
Make LSR ignore GEP's that have loop variant base values, as we currently
cannot codegen them
llvm-svn: 22576
2005-08-02 01:32:29 +00:00
564900e5e5
Fix an iterator invalidation problem
...
llvm-svn: 22575
2005-08-02 00:41:11 +00:00
546fd5944e
Keep tabs and trailing spaces out.
...
llvm-svn: 22565
2005-07-30 18:33:25 +00:00
c500991055
Fix VC++ build problems.
...
llvm-svn: 22564
2005-07-30 18:22:27 +00:00
17a0e2afea
Ack, typo
...
llvm-svn: 22560
2005-07-30 00:21:31 +00:00
e68bcd1946
Commit a new LoopStrengthReduce pass that can use scalar evolutions and
...
target data to decide which loop induction variables to strength reduce
and how to do so. This work is mostly by Chris Lattner, with tweaks by
me to get it working on some of MultiSource.
llvm-svn: 22558
2005-07-30 00:15:07 +00:00
2bca4d9b7b
Break SCEVExpander out of IndVarSimplify into its own .h/.cpp file so that
...
other passes may use it.
llvm-svn: 22557
2005-07-30 00:12:19 +00:00
5f4ef3c5a8
Eliminate all remaining tabs and trailing spaces.
...
llvm-svn: 22523
2005-07-27 06:12:32 +00:00
18aa4d8196
Do not let MaskedValueIsZero consider undef to be zero, for reasons
...
explained in the comment.
This fixes UnitTests/2003-09-18-BitFieldTest on darwin
llvm-svn: 22483
2005-07-20 18:49:28 +00:00
247aef884c
When transforming &A[i] < &A[j] -> i < j, make sure to perform the comparison
...
as a signed compare. This patch may fix PR597, but is correct in any case.
llvm-svn: 22465
2005-07-18 23:07:33 +00:00
4ed40f7c6f
Fix a problem that instcombine would hit when dealing with unreachable code.
...
Because the instcombine has to scan the entire function when it starts up
to begin with, we might as well do it in DFO so we can nuke unreachable code.
This fixes: Transforms/InstCombine/2005-07-07-DeadPHILoop.ll
llvm-svn: 22348
2005-07-07 20:40:38 +00:00
cf52eb2b99
prevent va_arg from being hoisted from a loop
...
llvm-svn: 22265
2005-06-20 13:36:33 +00:00
9144ec4764
core changes for varargs
...
llvm-svn: 22254
2005-06-18 18:34:52 +00:00
4fdd96c4e0
Clean up some uninitialized variables and missing return statements that
...
GCC 4.0.0 compiler (sometimes incorrectly) warns about under release build.
llvm-svn: 22249
2005-06-18 17:37:34 +00:00
2ceb6ee576
This is not true: (X != 13 | X < 15) -> X < 15
...
It is actually always true. This fixes PR586 and
Transforms/InstCombine/2005-06-16-SetCCOrSetCCMiscompile.ll
llvm-svn: 22236
2005-06-17 03:59:17 +00:00
73bcba5f61
Don't crash when dealing with INTMIN. This fixes PR585 and
...
Transforms/InstCombine/2005-06-16-RangeCrash.ll
llvm-svn: 22234
2005-06-17 02:05:55 +00:00
c53cb9d3ff
avoid constructing out of range shift amounts.
...
llvm-svn: 22230
2005-06-17 01:29:28 +00:00
89dc4f16f5
Fix PR583 and testcase Transforms/InstCombine/2005-06-15-DivSelectCrash.ll
...
llvm-svn: 22227
2005-06-16 04:55:52 +00:00
252a845e30
Fix PR571, removing code that does just the WRONG thing :)
...
llvm-svn: 22225
2005-06-16 03:00:08 +00:00
104002bee3
Fix a bug in my previous patch. Do not get the shift amount type (which
...
is always ubyte, get the type being shifted). This unbreaks espresso
llvm-svn: 22224
2005-06-16 01:52:07 +00:00
df81539278
Fix PR582. The rewriter can move casts around, which invalidated the
...
BB iterator. This fixes Transforms/IndVarsSimplify/2005-06-15-InstMoveCrash.ll
llvm-svn: 22221
2005-06-15 21:29:31 +00:00
19b57f55aa
Fix PR577 and testcase InstCombine/2005-06-15-ShiftSetCCCrash.ll.
...
Do not perform undefined out of range shifts.
llvm-svn: 22217
2005-06-15 20:53:31 +00:00
a299d6f701
Put the hack back in that removes features, causes regressions to fail, but
...
allows test programs to succeed. Actual fix for this is forthcoming.
llvm-svn: 22213
2005-06-15 18:25:30 +00:00
6d231e55fa
Unbreak several InstCombine regression checks introduced by a hack to
...
fix the bzip2 test. A better hack is needed.
llvm-svn: 22209
2005-06-13 06:41:26 +00:00
1609a541cd
Fix a 64-bit problem, passing (int)0 through ... instead of (void*)0
...
llvm-svn: 22206
2005-06-09 03:32:54 +00:00
ffe65458e7
hack to fix bzip2 (bug 571)
...
llvm-svn: 22192
2005-06-04 12:43:56 +00:00
05c703ea85
preserve calling conventions when hacking on code
...
llvm-svn: 22024
2005-05-14 12:25:32 +00:00
61d9d81770
calling a function with the wrong CC is undefined, turn it into an unreachable
...
instruction. This is useful for catching optimizers that don't preserve
calling conventions
llvm-svn: 21928
2005-05-13 07:09:09 +00:00
ca968393ab
When lowering invokes to calls, amke sure to preserve the calling conv. This
...
fixes Ptrdist/anagram with x86 llcbeta
llvm-svn: 21925
2005-05-13 06:27:02 +00:00
ae186e012c
Prefer int 0 instead of long 0 for GEP arguments.
...
llvm-svn: 21924
2005-05-13 06:10:12 +00:00
31c667e234
Fix Reassociate/shifttest.ll
...
llvm-svn: 21839
2005-05-10 03:39:25 +00:00
bfc796f622
If a function contains no allocas, all of the calls in it are trivially
...
suitable for tail calls.
llvm-svn: 21836
2005-05-09 23:51:13 +00:00
b62f5082c5
implement and.ll:test33
...
llvm-svn: 21809
2005-05-09 04:58:36 +00:00
df3332660f
Implement Reassociate/mul-neg-add.ll
...
llvm-svn: 21788
2005-05-08 21:41:35 +00:00
c4f8e2b0ed
Bail out earlier
...
llvm-svn: 21786
2005-05-08 21:33:47 +00:00
877b114037
Teach reassociate that 0-X === X*-1
...
llvm-svn: 21785
2005-05-08 21:28:52 +00:00
9f284e0a3c
Fix PR557 and basictest[34].ll.
...
This makes reassociate realize that loads should be treated as unmovable, and
gives distinct ranks to distinct values defined in the same basic block, allowing
reassociate to do its thing.
llvm-svn: 21783
2005-05-08 20:57:04 +00:00
9187f3905e
Add debugging information
...
llvm-svn: 21781
2005-05-08 20:09:57 +00:00
08582be283
eliminate gotos
...
llvm-svn: 21780
2005-05-08 19:48:43 +00:00
5847e5e10c
Improve reassociation handling of inverses, implementing inverses.ll.
...
llvm-svn: 21778
2005-05-08 18:59:37 +00:00
4922118dc4
clean up and modernize this pass.
...
llvm-svn: 21776
2005-05-08 18:45:26 +00:00
b18dbbfff5
Strength reduce SAR into SHR if there is no way sign bits could be shifted
...
in. This tends to get cases like this:
X = cast ubyte to int
Y = shr int X, ...
Tested by: shift.ll:test24
llvm-svn: 21775
2005-05-08 17:34:56 +00:00
e1850b86b6
Refactor some code
...
llvm-svn: 21772
2005-05-08 00:19:31 +00:00
6e2086d7e4
Handle some simple cases where we can see that values get annihilated.
...
llvm-svn: 21771
2005-05-08 00:08:33 +00:00
4294cec0f1
Fix a miscompilation of crafty by clobbering the "A" variable.
...
llvm-svn: 21770
2005-05-07 23:49:08 +00:00
1e5065052a
Rewrite the guts of the reassociate pass to be more efficient and logical. Instead
...
of trying to do local reassociation tweaks at each level, only process an expression
tree once (at its root). This does not improve the reassociation pass in any real way.
llvm-svn: 21768
2005-05-07 21:59:39 +00:00
cea579932d
Convert shifts to muls to assist reassociation. This implements
...
Reassociate/shifttest.ll
llvm-svn: 21761
2005-05-07 04:24:13 +00:00
f43e974abd
Simplify the code and rearrange it. No major functionality changes here.
...
llvm-svn: 21759
2005-05-07 04:08:02 +00:00
6aacb0f9da
Preserve tail marker
...
llvm-svn: 21737
2005-05-06 06:48:21 +00:00
ef298a3b8a
Teach instcombine propagate zeroness through shl instructions, implementing
...
and.ll:test31
llvm-svn: 21717
2005-05-06 04:53:20 +00:00
873804168e
Implement shift.ll:test23. If we are shifting right then immediately truncating
...
the result, turn signed shift rights into unsigned shift rights if possible.
This leads to later simplification and happens *often* in 176.gcc. For example,
this testcase:
struct xxx { unsigned int code : 8; };
enum codes { A, B, C, D, E, F };
int foo(struct xxx *P) {
if ((enum codes)P->code == A)
bar();
}
used to be compiled to:
int %foo(%struct.xxx* %P) {
%tmp.1 = getelementptr %struct.xxx* %P, int 0, uint 0 ; <uint*> [#uses=1]
%tmp.2 = load uint* %tmp.1 ; <uint> [#uses=1]
%tmp.3 = cast uint %tmp.2 to int ; <int> [#uses=1]
%tmp.4 = shl int %tmp.3, ubyte 24 ; <int> [#uses=1]
%tmp.5 = shr int %tmp.4, ubyte 24 ; <int> [#uses=1]
%tmp.6 = cast int %tmp.5 to sbyte ; <sbyte> [#uses=1]
%tmp.8 = seteq sbyte %tmp.6, 0 ; <bool> [#uses=1]
br bool %tmp.8, label %then, label %UnifiedReturnBlock
Now it is compiled to:
%tmp.1 = getelementptr %struct.xxx* %P, int 0, uint 0 ; <uint*> [#uses=1]
%tmp.2 = load uint* %tmp.1 ; <uint> [#uses=1]
%tmp.2 = cast uint %tmp.2 to sbyte ; <sbyte> [#uses=1]
%tmp.8 = seteq sbyte %tmp.2, 0 ; <bool> [#uses=1]
br bool %tmp.8, label %then, label %UnifiedReturnBlock
which is the difference between this:
foo:
subl $4, %esp
movl 8(%esp), %eax
movl (%eax), %eax
shll $24, %eax
sarl $24, %eax
testb %al, %al
jne .LBBfoo_2
and this:
foo:
subl $4, %esp
movl 8(%esp), %eax
movl (%eax), %eax
testb %al, %al
jne .LBBfoo_2
This occurs 3243 times total in the External tests, 215x in povray,
6x in each f2c'd program, 1451x in 176.gcc, 7x in crafty, 20x in perl,
25x in gap, 3x in m88ksim, 25x in ijpeg.
Maybe this will cause a little jump on gcc tommorow :)
llvm-svn: 21715
2005-05-06 04:18:52 +00:00
7208616ec0
Implement xor.ll:test22
...
llvm-svn: 21713
2005-05-06 02:07:39 +00:00
4c2d3781aa
implement and.ll:test30 and set.ll:test21
...
llvm-svn: 21712
2005-05-06 01:53:19 +00:00
dd1e562ec3
implement or.ll:test20
...
llvm-svn: 21709
2005-05-06 00:58:50 +00:00
809dfac421
Instcombine: cast (X != 0) to int, cast (X == 1) to int -> X iff X has only the low bit set.
...
This implements set.ll:test20.
This triggers 2x on povray, 9x on mesa, 11x on gcc, 2x on crafty, 1x on eon,
6x on perlbmk and 11x on m88ksim.
It allows us to compile these two functions into the same code:
struct s { unsigned int bit : 1; };
unsigned foo(struct s *p) {
if (p->bit)
return 1;
else
return 0;
}
unsigned bar(struct s *p) { return p->bit; }
llvm-svn: 21690
2005-05-04 19:10:26 +00:00
f42ed7bdaf
Fixed a comment.
...
llvm-svn: 21653
2005-05-02 14:47:42 +00:00
a816eee427
Implement getelementptr.ll:test11
...
llvm-svn: 21647
2005-05-01 04:42:15 +00:00
a9d84e3388
Check for volatile loads only once.
...
Implement load.ll:test7
llvm-svn: 21645
2005-05-01 04:24:53 +00:00
bd43b9db9d
Fix the compile failures from last night.
...
llvm-svn: 21565
2005-04-26 14:40:41 +00:00
a21bf8d1be
implement getelementptr.ll:test10
...
llvm-svn: 21541
2005-04-25 20:17:30 +00:00
2f1457fd83
Eliminate cases where we could << by 64, which is undefined in C.
...
llvm-svn: 21500
2005-04-24 17:46:05 +00:00
d6f636a340
Implement xor.ll:test21: select (not C), A, B -> select C, B, A
...
llvm-svn: 21495
2005-04-24 07:30:14 +00:00
d1f46d3bf9
Use getPrimitiveSizeInBits() instead of getPrimitiveSize()*8
...
Completely rework the 'setcc (cast x to larger), y' code. This code has
the advantage of implementing setcc.ll:test19 (being more general than
the previous code) and being correct in all cases.
This allows us to unxfail 2004-11-27-SetCCForCastLargerAndConstant.ll,
and close PR454.
llvm-svn: 21491
2005-04-24 06:59:08 +00:00
82639853c0
Eliminate tabs and trailing spaces
...
llvm-svn: 21480
2005-04-23 21:38:35 +00:00
77c32c34d7
Generalize the setcc -> PHI and Select folding optimizations to work with
...
any constant RHS, not just a constant integer RHS. This implements
select.ll:test17
llvm-svn: 21470
2005-04-23 15:31:55 +00:00
b1c9317bb4
Remove trailing whitespace
...
llvm-svn: 21427
2005-04-21 23:48:37 +00:00
374e659466
Instcombine this:
...
%shortcirc_val = select bool %tmp.1, bool true, bool %tmp.4 ; <bool> [#uses=1]
%tmp.6 = cast bool %shortcirc_val to int ; <int> [#uses=1]
into this:
%shortcirc_val = or bool %tmp.1, %tmp.4 ; <bool> [#uses=1]
%tmp.6 = cast bool %shortcirc_val to int ; <int> [#uses=1]
not this:
%tmp.4.cast = cast bool %tmp.4 to int ; <int> [#uses=1]
%tmp.6 = select bool %tmp.1, int 1, int %tmp.4.cast ; <int> [#uses=1]
llvm-svn: 21389
2005-04-21 05:43:13 +00:00
8cb10a1775
Wrap some long lines.
...
Make IPSCCP strip off dead constant exprs that are using functions, making
them appear as though their address is taken. This allows us to propagate
some more pool descriptors, lowering the overhead of pool alloc.
llvm-svn: 21363
2005-04-19 19:16:19 +00:00
5c219469a0
Eliminate a broken transformation, fixing PR548
...
llvm-svn: 21354
2005-04-19 06:04:18 +00:00
ee84413730
silence a bogus warning
...
llvm-svn: 21320
2005-04-18 05:26:21 +00:00
16a50fd0a0
a new simple pass, which will be extended to be more useful in the future.
...
This pass forward branches through conditions when it can show that the
conditions is either always true or false for a predecessor. This currently
only handles the most simple cases of this, but is successful at threading
across 2489 branches and 65 switch instructions in 176.gcc, which isn't bad.
llvm-svn: 21306
2005-04-15 19:28:32 +00:00
4236261930
Fix bug: InstCombine/2005-05-07-UDivSelectCrash.ll
...
llvm-svn: 21152
2005-04-08 04:03:26 +00:00
4706046e68
Implement the following xforms:
...
(X-Y)-X --> -Y
A + (B - A) --> B
(B - A) + A --> B
llvm-svn: 21138
2005-04-07 17:14:51 +00:00
c7f3c1a00e
Implement InstCombine/add.ll:test28, transforming C1-(X+C2) --> (C1-C2)-X.
...
This occurs several dozen times in specint2k, particularly in crafty and gcc
apparently.
llvm-svn: 21136
2005-04-07 16:28:01 +00:00
a9be4490d8
Transform X-(X+Y) == -Y and X-(Y+X) == -Y
...
llvm-svn: 21134
2005-04-07 16:15:25 +00:00
ecfa9b5810
disable this transformation in the one obscure case that really pessimizes
...
pointer analysis.
llvm-svn: 20916
2005-03-29 06:37:47 +00:00
9ead0d7b4c
Rename createPromoteMemoryToRegister() to
...
createPromoteMemoryToRegisterPass() to be consistent with other pass
creation functions.
llvm-svn: 20885
2005-03-28 02:01:12 +00:00
514e843e89
Enhance loopsimplify to preserve alias analysis instead of clobbering it.
...
This prevents crashes on some programs when using -ds-aa -licm.
llvm-svn: 20831
2005-03-25 06:37:22 +00:00
faf7791fea
Fix a bug where LICM was not updating AA information properly when sinking
...
a pointer value out of a loop causing it to be duplicated.
llvm-svn: 20828
2005-03-25 00:22:36 +00:00
1c790bf656
enable -debug-only=licm
...
llvm-svn: 20788
2005-03-23 21:00:12 +00:00
531f9e92d4
This mega patch converts us from using Function::a{iterator|begin|end} to
...
using Function::arg_{iterator|begin|end}. Likewise Module::g* -> Module::global_*.
This patch is contributed by Gabor Greif, thanks!
llvm-svn: 20597
2005-03-15 04:54:21 +00:00
8c79559443
fix a bug where we thought arguments were constants :(
...
llvm-svn: 20506
2005-03-06 22:52:29 +00:00
2ce303b406
Fix Regression/Transforms/LoopStrengthReduce/dont_insert_redundant_ops.ll,
...
hopefully not breaking too many other things.
llvm-svn: 20505
2005-03-06 22:36:12 +00:00
45403e5052
implement Transforms/LoopStrengthReduce/invariant_value_first_arg.ll
...
llvm-svn: 20501
2005-03-06 22:06:22 +00:00
d3874fad44
minor simplifications of the code.
...
llvm-svn: 20497
2005-03-06 21:58:22 +00:00
dd3ec92085
trivial simplification
...
llvm-svn: 20494
2005-03-06 21:35:38 +00:00
238f6df546
Fix a bug where we could corrupt a parent loop's header info if we unrolled
...
a nested loop. This fixes Transforms/LoopUnroll/2005-03-06-BadLoopInfoUpdate.ll
and PR532
llvm-svn: 20493
2005-03-06 20:57:32 +00:00
4abcea3a69
Reformat comments to fix 80 columns.
...
llvm-svn: 20467
2005-03-05 22:45:40 +00:00
be37fa07fd
Reuse induction variables created for strength-reduced GEPs by other similar GEPs.
...
llvm-svn: 20466
2005-03-05 22:40:34 +00:00
cfe2822cdf
Do not compute 1ULL << 64, which is undefined. This fixes Ptrdist/ks on the
...
sparc, and testcase Regression/Transforms/InstCombine/2005-03-04-ShiftOverflow.ll
llvm-svn: 20445
2005-03-04 23:21:33 +00:00
a2c59b7423
Add support for not strength reducing GEPs where the element size is a small
...
power of two. This emphatically includes the zeroeth power of two.
llvm-svn: 20429
2005-03-04 04:04:26 +00:00
ef1e989e4f
Add an optional argument to lower to a specific constant value instead of
...
to a "sizeof" expression.
llvm-svn: 20414
2005-03-03 01:03:43 +00:00
8ea6f9e821
Fixed the following LSR bugs:
...
* Loop invariant code does not dominate the loop header, but rather
the end of the loop preheader.
* The base for a reduced GEP isn't a constant unless all of its
operands (preceding the induction variable) are constant.
* Allow induction variable elimination for the simple case after all.
Also made changes recommended by Chris for properly deleting
instructions.
llvm-svn: 20383
2005-03-01 03:46:11 +00:00
dcaa48b5c4
Fix crash in LSR due to attempt to remove original induction variable. However,
...
for reasons explained in the comments, I also deactivated this code as it needs
more thought.
llvm-svn: 20367
2005-02-28 00:08:56 +00:00
fd63d3af0d
PHI nodes were incorrectly placed when more than one GEP is reduced in a loop.
...
llvm-svn: 20360
2005-02-27 21:08:04 +00:00
39751c3b7c
First pass at improved Loop Strength Reduction. Still not yet ready for prime time.
...
llvm-svn: 20358
2005-02-27 19:37:07 +00:00
52e931b37d
Remove use of bind_obj
...
llvm-svn: 20276
2005-02-22 23:22:58 +00:00
7b5d9e2217
Do not mark obviously unreachable blocks live when processing PHI nodes,
...
and handle incomplete control dependences correctly. This fixes:
Regression/Transforms/ADCE/dead-phi-edge.ll
-> a missed optimization
Regression/Transforms/ADCE/dead-phi-edge.ll
-> a compiler crash distilled from QT4
llvm-svn: 20227
2005-02-17 19:28:49 +00:00
31f3382b3b
Fix the second bug attached to PR504.
...
llvm-svn: 20181
2005-02-14 20:11:45 +00:00
e616fea3bc
Fix for testcase Transforms/IndVarsSimplify/2005-02-11-InvokeCrash.ll
...
and PR504.
llvm-svn: 20129
2005-02-12 03:26:49 +00:00
82b42c5d85
API change.
...
llvm-svn: 19959
2005-02-01 01:23:49 +00:00
72684fecf8
Implement InstCombine/cast.ll:test25, a case that occurs many times
...
in spec
llvm-svn: 19953
2005-01-31 05:51:45 +00:00
31f486c775
Implement the trivial cases in InstCombine/store.ll
...
llvm-svn: 19950
2005-01-31 05:36:43 +00:00
fe1b0b8b24
Implement Transforms/InstCombine/cast-load-gep.ll, which allows us to devirtualize
...
11 indirect calls in perlbmk.
llvm-svn: 19947
2005-01-31 04:50:46 +00:00
d8e20188c6
Adjust to changes in instruction interfaces.
...
llvm-svn: 19900
2005-01-29 00:39:08 +00:00
cd517ff0c7
* add some DEBUG statements
...
* Properly compile this:
struct a {};
int test() {
struct a b[2];
if (&b[0] != &b[1])
abort ();
return 0;
}
to 'return 0', not abort().
llvm-svn: 19875
2005-01-28 19:32:01 +00:00
9e2c7facb2
Get rid of a several dozen more and instructions in specint.
...
llvm-svn: 19786
2005-01-23 20:26:55 +00:00
fc4429e7c1
Handle comparisons of gep instructions that have different typed indices
...
as long as they are the same size.
llvm-svn: 19734
2005-01-21 23:06:49 +00:00
411336fe04
Add two optimizations. The first folds (X+Y)-X -> Y
...
The second folds operations into selects, e.g. (select C, (X+Y), (Y+Z))
-> (Y+(select C, X, Z)
This occurs a few times across spec, e.g.
select add/sub
mesa: 83 0
povray: 5 2
gcc 4 2
parser 0 22
perlbmk 13 30
twolf 0 3
llvm-svn: 19706
2005-01-19 21:50:18 +00:00
715364364b
Delete PHI nodes that are not dead but are locked in a cycle of single
...
useness.
llvm-svn: 19629
2005-01-17 05:10:15 +00:00
03f06f11aa
Move code out of indentation one level to make it easier to read.
...
Disable the xform for < > cases. It turns out that the following is being
miscompiled:
bool %test(sbyte %S) {
%T = cast sbyte %S to uint
%V = setgt uint %T, 255
ret bool %V
}
llvm-svn: 19628
2005-01-17 03:20:02 +00:00
51726c47fe
Fix some bugs in an xform added yesterday. This fixes Prolangs-C/allroots.
...
llvm-svn: 19553
2005-01-14 17:35:12 +00:00
7aa41cfa88
Fix a compile crash on spiff
...
llvm-svn: 19552
2005-01-14 17:17:59 +00:00
4fa89827e2
if two gep comparisons only differ by one index, compare that index directly.
...
This allows us to better optimize begin() -> end() comparisons in common cases.
llvm-svn: 19542
2005-01-14 00:20:05 +00:00
d35d210ea0
Do not overrun iterators. This fixes a 176.gcc crash
...
llvm-svn: 19541
2005-01-13 23:26:48 +00:00
a04c904c4c
Turn select C, (X+Y), (X-Y) --> (X+(select C, Y, (-Y))). This occurs in
...
the 'sim' program and probably elsewhere. In sim, it comes up for cases
like this:
#define round(x) ((x)>0.0 ? (x)+0.5 : (x)-0.5)
double G;
void T(double X) { G = round(X); }
(it uses the round macro a lot). This changes the LLVM code from:
%tmp.1 = setgt double %X, 0.000000e+00 ; <bool> [#uses=1]
%tmp.4 = add double %X, 5.000000e-01 ; <double> [#uses=1]
%tmp.6 = sub double %X, 5.000000e-01 ; <double> [#uses=1]
%mem_tmp.0 = select bool %tmp.1, double %tmp.4, double %tmp.6
store double %mem_tmp.0, double* %G
to:
%tmp.1 = setgt double %X, 0.000000e+00 ; <bool> [#uses=1]
%mem_tmp.0.p = select bool %tmp.1, double 5.000000e-01, double -5.000000e-01
%mem_tmp.0 = add double %mem_tmp.0.p, %X
store double %mem_tmp.0, double* %G
ret void
llvm-svn: 19537
2005-01-13 22:52:24 +00:00
81e8417614
Implement an optimization for == and != comparisons like this:
...
_Bool test2(int X, int Y) {
return &arr[X][Y] == arr;
}
instead of generating this:
bool %test2(int %X, int %Y) {
%tmp.3.idx = mul int %X, 160 ; <int> [#uses=1]
%tmp.3.idx1 = shl int %Y, ubyte 2 ; <int> [#uses=1]
%tmp.3.offs2 = sub int 0, %tmp.3.idx ; <int> [#uses=1]
%tmp.7 = seteq int %tmp.3.idx1, %tmp.3.offs2 ; <bool> [#uses=1]
ret bool %tmp.7
}
generate this:
bool %test2(int %X, int %Y) {
seteq int %X, 0 ; <bool>:0 [#uses=1]
seteq int %Y, 0 ; <bool>:1 [#uses=1]
%tmp.7 = and bool %0, %1 ; <bool> [#uses=1]
ret bool %tmp.7
}
This idiom occurs in C++ programs when iterating from begin() to end(),
in a vector or array. For example, we now compile this:
void test(int X, int Y) {
for (int *i = arr; i != arr+100; ++i)
foo(*i);
}
to this:
no_exit: ; preds = %entry, %no_exit
...
%exitcond = seteq uint %indvar.next, 100 ; <bool> [#uses=1]
br bool %exitcond, label %return, label %no_exit
instead of this:
no_exit: ; preds = %entry, %no_exit
...
%inc5 = getelementptr [100 x [40 x int]]* %arr, int 0, int 0, int %inc.rec ; <int*> [#uses=1]
%tmp.8 = seteq int* %inc5, getelementptr ([100 x [40 x int]]* %arr, int 0, int 100, int 0) ; <bool> [#uses=1]
%indvar.next = add uint %indvar, 1 ; <uint> [#uses=1]
br bool %tmp.8, label %return, label %no_exit
llvm-svn: 19536
2005-01-13 22:25:21 +00:00
4cb9fa373b
Fix some bugs in code I didn't mean to check in.
...
llvm-svn: 19534
2005-01-13 20:40:58 +00:00
0798af33a5
Fix a crash compiling 129.compress
...
llvm-svn: 19533
2005-01-13 20:14:25 +00:00
fdfe3e49fe
Fix uint64_t -> unsigned VS warnings.
...
llvm-svn: 19381
2005-01-08 19:42:22 +00:00
47f395cd85
Silence VS warnings.
...
llvm-svn: 19380
2005-01-08 19:37:20 +00:00
ce274ce93d
Silence warnings
...
llvm-svn: 19379
2005-01-08 19:34:41 +00:00
677babc4d4
Add more missing createXxxPass functions.
...
llvm-svn: 19370
2005-01-08 17:21:40 +00:00
eca0d0f2da
Put createLoopUnswitchPass() into proper namespace
...
llvm-svn: 19306
2005-01-06 05:47:18 +00:00
86102b8ad5
This is a bulk commit that implements the following primary improvements:
...
* We can now fold cast instructions into select instructions that
have at least one constant operand.
* We now optimize expressions more aggressively based on bits that are
known to be zero. These optimizations occur a lot in code that uses
bitfields even in simple ways.
* We now turn more cast-cast sequences into AND instructions. Before we
would only do this if it if all types were unsigned. Now only the
middle type needs to be unsigned (guaranteeing a zero extend).
* We transform sign extensions into zero extensions in several cases.
This corresponds to these test/Regression/Transforms/InstCombine testcases:
2004-11-22-Missed-and-fold.ll
and.ll: test28-29
cast.ll: test21-24
and-or-and.ll
cast-cast-to-and.ll
zeroext-and-reduce.ll
llvm-svn: 19220
2005-01-01 16:22:27 +00:00
13516fe2e7
Fix PR491 and testcase Transforms/DeadStoreElimination/2004-12-28-PartialStore.ll
...
llvm-svn: 19180
2004-12-29 04:36:02 +00:00
b17f3e13ec
Adjust to new interfaces
...
llvm-svn: 18958
2004-12-15 07:22:25 +00:00
9ad0d55025
Constant exprs are not efficiently negatable in practice. This disables
...
turning X - (constantexpr) into X + (-constantexpr) among other things.
llvm-svn: 18935
2004-12-14 20:08:06 +00:00
8f430a3b59
Get rid of getSizeOf, using ConstantExpr::getSizeOf instead.
...
do not insert a prototype for malloc of: void* malloc(uint): on 64-bit u
targets this is not correct. Instead of prototype it as void *malloc(...),
and pass the correct intptr_t through the "...".
Finally, fix Regression/CodeGen/SparcV9/2004-12-13-MallocCrash.ll, by not
forming constantexpr casts from pointer to uint.
llvm-svn: 18908
2004-12-13 20:00:02 +00:00
a199e3c1e2
Change indentation of a whole bunch of code, no real changes here.
...
llvm-svn: 18843
2004-12-12 23:49:37 +00:00
14d07db44d
More substantial simplifications and speedups. This makes ADCE about 20% faster
...
in some cases.
llvm-svn: 18842
2004-12-12 23:40:17 +00:00
9115eb3024
More minor microoptimizations
...
llvm-svn: 18841
2004-12-12 22:44:30 +00:00
d4298781c1
Remove some more set operations
...
llvm-svn: 18840
2004-12-12 22:22:18 +00:00
a538439bf0
Reduce number of set operations.
...
llvm-svn: 18839
2004-12-12 22:16:13 +00:00
bf5b7cf638
Optimize div/rem + select combinations more.
...
In particular, implement div.ll:test10 and rem.ll:test4.
llvm-svn: 18838
2004-12-12 21:48:58 +00:00
88deefa303
Simplify code and do not invalidate iterators.
...
This fixes a crash compiling TimberWolfMC that was exposed due to recent
optimizer changes.
llvm-svn: 18831
2004-12-12 18:23:20 +00:00
cbc0161d1f
If one side of and/or is known to be 0/-1, it doesn't matter
...
if the other side is overdefined.
This allows us to fold conditions like: if (X < Y || Y > Z) in some cases.
llvm-svn: 18807
2004-12-11 23:15:19 +00:00
2f687fd9d6
Two bug fixes:
...
1. Actually increment the Statistic for the GV elim optzn
2. When resolving undef branches, only resolve branches in executable blocks,
avoiding marking a bunch of completely dead blocks live. This has a big
impact on the quality of the generated code.
With this patch, we positively rip up vortex, compiling Ut_MoveBytes to a
single memcpy call. In vortex we get this:
12 ipsccp - Number of globals found to be constant
986 ipsccp - Number of arguments constant propagated
1378 ipsccp - Number of basic blocks unreachable
8919 ipsccp - Number of instructions removed
llvm-svn: 18796
2004-12-11 06:05:53 +00:00
8525ebe465
Do not delete the entry block to a function.
...
llvm-svn: 18795
2004-12-11 05:32:19 +00:00
91dbae6fee
Implement Transforms/SCCP/ipsccp-gvar.ll, by tracking values stored to
...
non-address-taken global variables.
llvm-svn: 18790
2004-12-11 05:15:59 +00:00
99e1295645
Fix a bug where we could delete dead invoke instructions with uses.
...
In functions where we fully constant prop the return value, replace all
ret instructions with 'ret undef'.
llvm-svn: 18786
2004-12-11 02:53:57 +00:00
bae4b64553
Implement SCCP/ipsccp-conditional.ll, by totally deleting dead blocks.
...
llvm-svn: 18781
2004-12-10 22:29:08 +00:00
7285f43836
Fix SCCP/2004-12-10-UndefBranchBug.ll
...
llvm-svn: 18776
2004-12-10 20:41:50 +00:00
b439464c61
This is the initial implementation of IPSCCP, as requested by Brian.
...
This implements SCCP/ipsccp-basic.ll, rips apart Olden/mst (as described in
PR415), and does other nice things.
There is still more to come with this, but it's a start.
llvm-svn: 18752
2004-12-10 08:02:06 +00:00
36d39cecb4
note to self: Do not check in debugging code!
...
llvm-svn: 18693
2004-12-09 07:15:52 +00:00
f17a2fb849
Implement trivial sinking for load instructions. This causes us to sink 567 loads in spec
...
llvm-svn: 18692
2004-12-09 07:14:34 +00:00
39c98bb31c
Do extremely simple sinking of instructions when they are only used in a
...
successor block. This turns cases like this:
x = a op b
if (c) {
use x
}
into:
if (c) {
x = a op b
use x
}
This triggers 3965 times in spec, and is tested by
Regression/Transforms/InstCombine/sink_instruction.ll
This appears to expose a bug in the X86 backend for 177.mesa, which I'm
looking in to.
llvm-svn: 18677
2004-12-08 23:43:58 +00:00
a1291a0679
Fix this regression and remove the XFAIL from this test.
...
llvm-svn: 18674
2004-12-08 23:10:30 +00:00
8f30caf549
Fix Transforms/InstCombine/2004-12-08-RemInfiniteLoop.ll
...
llvm-svn: 18670
2004-12-08 22:20:34 +00:00
9273d480ad
For PR387:\
...
Add doInitialization method to avoid overloaded virtuals
llvm-svn: 18602
2004-12-07 08:11:36 +00:00
a4c9808603
This pass is moving to lib IPO
...
llvm-svn: 18439
2004-12-02 21:24:40 +00:00
951673a94c
This pass is completely broken.
...
llvm-svn: 18387
2004-11-30 17:09:06 +00:00
6e455608e2
Allow hoisting loads of globals and alloca's in conditionals.
...
llvm-svn: 18363
2004-11-29 21:26:12 +00:00
279fa256a2
Fix for PR454:
...
* Make sure we handle signed to unsigned conversion correctly
* Move this visitSetCondInst case to its own method.
llvm-svn: 18312
2004-11-28 21:31:15 +00:00
6ea2888832
Make DSE potentially more aggressive by being more specific about alloca sizes.
...
llvm-svn: 18309
2004-11-28 20:44:37 +00:00
14f3cdc227
Implement Regression/Transforms/InstCombine/getelementptr_cast.ll, which
...
occurs many times in crafty
llvm-svn: 18273
2004-11-27 17:55:46 +00:00
b137409926
Provide size information when checking to see if we can LICM a load, this
...
allows us to hoist more loads in some cases.
llvm-svn: 18265
2004-11-26 21:20:09 +00:00
540e5f92b4
Do not count debugger intrinsics in size estimation.
...
llvm-svn: 18110
2004-11-22 17:23:57 +00:00
6d048a0d32
Do not consider debug intrinsics in the size computations for loop unrolling.
...
Patch contributed by Michael McCracken!
llvm-svn: 18108
2004-11-22 17:18:36 +00:00
446948e094
Fix the exposed prototype for the lower packed pass, thanks to
...
Morten Ofstad.
llvm-svn: 17996
2004-11-19 16:49:34 +00:00
953075442d
Delete stoppoints that occur for the same source line.
...
llvm-svn: 17970
2004-11-18 21:41:39 +00:00
c08ac110df
Check in hook that I forgot
...
llvm-svn: 17956
2004-11-18 17:24:20 +00:00
27af257ea0
Do not delete dead invoke instructions!
...
llvm-svn: 17897
2004-11-16 16:32:28 +00:00
9339638e9c
Remove unused variable for compilation by VC++.
...
Patch contributed by Morten Ofstad.
llvm-svn: 17830
2004-11-15 17:29:41 +00:00
1890f94413
Minor cleanups. There is no reason for SCCP to derive from instvisitor anymore.
...
llvm-svn: 17825
2004-11-15 07:15:04 +00:00
9a038a3a5e
Count more accurately
...
llvm-svn: 17824
2004-11-15 07:02:42 +00:00
97013636cd
Quiet warnings on the persephone tester
...
llvm-svn: 17821
2004-11-15 05:54:07 +00:00
d18c16b842
Two minor improvements:
...
1. Speedup getValueState by having it not consider Arguments. It's better
to just add them before we start SCCP'ing.
2. SCCP can delete the contents of dead blocks. No really, it's ok! This
reduces the size of the IR for subsequent passes, even though
simplifycfg would do the same job. In practice, simplifycfg does not
run until much later than sccp in gccas
llvm-svn: 17820
2004-11-15 05:45:33 +00:00
4f0316229c
rename InstValue to LatticeValue, as it holds for more than instructions.
...
llvm-svn: 17818
2004-11-15 05:03:30 +00:00
074be1f6e4
Substantially refactor the SCCP class into an SCCP pass and an SCCPSolver
...
class. The only changes are minor:
* Do not try to SCCP instructions that return void in the rewrite loop.
This is silly and fool hardy, wasting a map lookup and adding an entry
to the map which is never used.
* If we decide something has an undefined value, rewrite it to undef,
potentially leading to further simplications.
llvm-svn: 17816
2004-11-15 04:44:20 +00:00
46dd5a6304
This optimization makes MANY phi nodes that all have the same incoming value.
...
If this happens, detect it early instead of relying on instcombine to notice
it later. This can be a big speedup, because PHI nodes can have many
incoming values.
llvm-svn: 17741
2004-11-14 19:29:34 +00:00
7515cabe2a
Implement instcombine/phi.ll:test6 - pulling operations through PHI nodes.
...
This exposes subsequent optimization possiblities and reduces code size.
This triggers 1423 times in spec.
llvm-svn: 17740
2004-11-14 19:13:23 +00:00
15ff1e1885
Transform this:
...
%X = alloca ...
%Y = alloca ...
X == Y
into false. This allows us to simplify some stuff in eon (and probably
many other C++ programs) where operator= was checking for self assignment.
Folding this allows us to SROA several additional structs.
llvm-svn: 17735
2004-11-14 07:33:16 +00:00
fe3f4e6ebd
Teach SROA how to promote an array index that is variable, if the dimension
...
of the array is just two. This occurs 8 times in gcc, 6 times in crafty, and
12 times in 099.go.
This implements ScalarRepl/sroa_two.ll
llvm-svn: 17727
2004-11-14 05:00:19 +00:00
8881912d71
Rearrange some code, no functionality changes.
...
llvm-svn: 17724
2004-11-14 04:24:28 +00:00
8c3e7b92af
Simplify handling of shifts to be the same as we do for adds. Add support
...
for (X * C1) + (X * C2) (where * can be mul or shl), allowing us to fold:
Y+Y+Y+Y+Y+Y+Y+Y
into
%tmp.8 = shl long %Y, ubyte 3 ; <long> [#uses=1]
instead of
%tmp.4 = shl long %Y, ubyte 2 ; <long> [#uses=1]
%tmp.12 = shl long %Y, ubyte 2 ; <long> [#uses=1]
%tmp.8 = add long %tmp.4, %tmp.12 ; <long> [#uses=1]
This implements add.ll:test25
Also add support for (X*C1)-(X*C2) -> X*(C1-C2), implementing sub.ll:test18
llvm-svn: 17704
2004-11-13 19:50:12 +00:00
4efe20a103
Fold:
...
(X + (X << C2)) --> X * ((1 << C2) + 1)
((X << C2) + X) --> X * ((1 << C2) + 1)
This means that we now canonicalize "Y+Y+Y" into:
%tmp.2 = mul long %Y, 3 ; <long> [#uses=1]
instead of:
%tmp.10 = shl long %Y, ubyte 1 ; <long> [#uses=1]
%tmp.6 = add long %Y, %tmp.10 ; <long> [#uses=1]
llvm-svn: 17701
2004-11-13 19:31:40 +00:00
2858e17538
Lazily create the abort message, so only translation units that use unwind
...
will actually get it.
llvm-svn: 17700
2004-11-13 19:07:32 +00:00
5c1d84c769
Simplify handling of constant initializers
...
llvm-svn: 17696
2004-11-12 22:42:57 +00:00
595016d090
This is V9 specific, move it there.
...
llvm-svn: 17545
2004-11-07 00:39:26 +00:00
33eb909939
Fix some warnings on VC++
...
llvm-svn: 17481
2004-11-05 04:45:43 +00:00
96f6616479
* Rearrange code slightly
...
* Disable broken transforms for simplifying (setcc (cast X to larger), CI)
where CC is not != or ==
llvm-svn: 17422
2004-11-02 03:50:32 +00:00
8af7424920
Speed up the tail duplication pass on the testcase below from 68.2s to 1.23s:
...
#define CL0(a) case a: f(); goto c;
#define CL1(a) CL0(a##0) CL0(a##1) CL0(a##2) CL0(a##3) CL0(a##4) CL0(a##5) \
CL0(a##6) CL0(a##7) CL0(a##8) CL0(a##9)
#define CL2(a) CL1(a##0) CL1(a##1) CL1(a##2) CL1(a##3) CL1(a##4) CL1(a##5) \
CL1(a##6) CL1(a##7) CL1(a##8) CL1(a##9)
#define CL3(a) CL2(a##0) CL2(a##1) CL2(a##2) CL2(a##3) CL2(a##4) CL2(a##5) \
CL2(a##6) CL2(a##7) CL2(a##8) CL2(a##9)
#define CL4(a) CL3(a##0) CL3(a##1) CL3(a##2) CL3(a##3) CL3(a##4) CL3(a##5) \
CL3(a##6) CL3(a##7) CL3(a##8) CL3(a##9)
void f();
void a() {
int b;
c: switch (b) {
CL4(1)
}
}
This comes from GCC PR 15524
llvm-svn: 17390
2004-11-01 07:05:07 +00:00
57cbe39d1e
Change Library Names Not To Conflict With Others When Installed
...
llvm-svn: 17286
2004-10-27 23:18:45 +00:00
7dfc2d29ac
Convert 'struct' to 'class' in various places to adhere to the coding standards
...
and work better with VC++. Patch contributed by Morten Ofstad!
llvm-svn: 17281
2004-10-27 16:14:51 +00:00
70c2039b39
Hrm, this code was severely botched. As it turns out, this patch:
...
http://mail.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20041018/019708.html
exposed ANOTHER latent bug in this xform, which caused Prolangs-C/bison to fill
the zion nightly tester disk up and make the tester barf.
This is obviously not a good thing, so lets fix this bug shall we? :)
llvm-svn: 17276
2004-10-27 05:57:15 +00:00
845afe9b20
Initialize with the correct constant type
...
llvm-svn: 17270
2004-10-27 03:55:24 +00:00
d57638c4a7
Fix compatibility with MSVC, patch by Morten Ofstad
...
llvm-svn: 17218
2004-10-25 18:45:16 +00:00
5c3c21e10a
Fix a bug Nate noticed, where we miscompiled a simple testcase
...
llvm-svn: 17157
2004-10-22 04:53:16 +00:00
c1c320c335
We won't use automake
...
llvm-svn: 17155
2004-10-22 03:35:04 +00:00
257b284038
Hrm, some people complain when the compiler cheerfully tells them what it's
...
doing... I guess they're right.
llvm-svn: 17142
2004-10-19 06:33:16 +00:00
6a11a75f31
Initial automake generated Makefile template
...
llvm-svn: 17136
2004-10-18 23:55:41 +00:00
b18121e6a9
Initial implementation of the strength reduction for GEP instructions in
...
loops. This optimization is not turned on by default yet, but may be run
with the opt tool's -loop-reduce flag. There are many FIXMEs listed in the
code that will make it far more applicable to a wide range of code, but you
have to start somewhere :)
This limited version currently triggers on the following tests in the
MultiSource directory:
pcompress2: 7 times
cfrac: 5 times
anagram: 2 times
ks: 6 times
yacr2: 2 times
llvm-svn: 17134
2004-10-18 21:08:22 +00:00
ce0783318b
Correction to allow compilation with Visual C++.
...
Patch contributed by Morten Ofstad. Thanks Morten!
llvm-svn: 17123
2004-10-18 14:38:48 +00:00
a67dd32004
Turn store -> null/undef into the LLVM unreachable instruction! This simple
...
change hacks off 10K of bytecode from perlbmk (.5%) even though the front-end
is not generating them yet and we are not optimizing the resultant code.
This isn't too bad.
llvm-svn: 17111
2004-10-18 03:00:50 +00:00
8ba9ec9bbb
Turn things with obviously undefined semantics into 'store -> null'
...
llvm-svn: 17110
2004-10-18 02:59:09 +00:00
3b92f17165
My friend the invoke instruction does not dominate all basic blocks if it
...
occurs in the entry node of a function
llvm-svn: 17109
2004-10-18 01:48:31 +00:00
6a792feb02
Getting ADCE to interact well with unreachable instructions seems like a nontrivial
...
exercise that I'm not interested in tackling right now. Just punt and treat them
like unwind's.
This 'fixes' test/Regression/Transforms/ADCE/unreachable-function.ll
llvm-svn: 17106
2004-10-17 23:45:06 +00:00
107c15c33d
Remove printout, realize that instructions in the entry block dominate all
...
other blocks.
llvm-svn: 17099
2004-10-17 21:31:34 +00:00
e29d634a94
hasConstantValue will soon return instructions that don't dominate the PHI node,
...
so prepare for this.
llvm-svn: 17095
2004-10-17 21:22:38 +00:00
67f0545daf
Fix a type violation
...
llvm-svn: 17069
2004-10-16 23:28:04 +00:00
684c5c6587
Kill the bogon that slipped into my buffer before I committed.
...
llvm-svn: 17067
2004-10-16 19:46:33 +00:00
6580e09fef
Implement InstCombine/getelementptr.ll:test9, which is the source of many
...
ugly and giant constnat exprs in some programs.
llvm-svn: 17066
2004-10-16 19:44:59 +00:00
81a7a23494
Optimize instructions involving undef values. For example X+undef == undef.
...
llvm-svn: 17047
2004-10-16 18:11:37 +00:00
646354bae1
Handle undef values as undefined on the constant lattice
...
ignore unreachable instructions
llvm-svn: 17044
2004-10-16 18:09:41 +00:00
6ac3ef950d
Add note
...
llvm-svn: 17043
2004-10-16 18:09:25 +00:00
ace94df71f
Update to reflect changes in Makefile rules.
...
llvm-svn: 16950
2004-10-13 11:46:52 +00:00
00648e1f86
Transform memmove -> memcpy when the source is obviously constant memory.
...
llvm-svn: 16932
2004-10-12 04:52:52 +00:00
7cabf6f87a
Fix a REALLY obscure bug in my previous checkin, which was splicing the END
...
marker from one ilist into the middle of another basic block!
llvm-svn: 16925
2004-10-12 01:02:29 +00:00
9776f7259b
Handle a common case more carefully. In particular, instead of transforming
...
pointer recurrences into expressions from this:
%P_addr.0.i.0 = phi sbyte* [ getelementptr ([8 x sbyte]* %.str_1, int 0, int 0), %entry ], [ %inc.0.i, %no_exit.i ]
%inc.0.i = getelementptr sbyte* %P_addr.0.i.0, int 1 ; <sbyte*> [#uses=2]
into this:
%inc.0.i = getelementptr sbyte* getelementptr ([8 x sbyte]* %.str_1, int 0, int 0), int %inc.0.i.rec
Actually create something nice, like this:
%inc.0.i = getelementptr [8 x sbyte]* %.str_1, int 0, int %inc.0.i.rec
llvm-svn: 16924
2004-10-11 23:06:50 +00:00
a92af96c56
Reenable the transform, turning X/-10 < 1 into X > -10
...
llvm-svn: 16918
2004-10-11 19:40:04 +00:00
97327f05fc
Initial version of automake Makefile.am file.
...
llvm-svn: 16893
2004-10-10 22:20:40 +00:00
5c91c8f18b
Use DEBUG instead of DebugFlag directly, as DebugFlag does not respect
...
-debug-only!
llvm-svn: 16868
2004-10-09 19:30:36 +00:00
4ad08352b4
Implement sub.ll:test17, -X/C -> X/-C
...
llvm-svn: 16863
2004-10-09 02:50:40 +00:00
0b41e861b6
Temporarily disable a buggy transformation until it can be fixed. This fixes
...
254.gap.
llvm-svn: 16853
2004-10-08 19:15:44 +00:00
bff91d9a2e
Instcombine (X & FF00) + xx00 -> (X+xx00) & FF00, implementing and.ll:test27
...
This comes up when doing adds to bitfield elements.
llvm-svn: 16836
2004-10-08 05:07:56 +00:00
44bd392cbf
Little patch to turn (shl (add X, 123), 4) -> (add (shl X, 4), 123 << 4)
...
This triggers in cases of bitfield additions, opening opportunities for
future improvements.
llvm-svn: 16834
2004-10-08 03:46:20 +00:00
0aee4b7947
Instcombine: -(X sdiv C) -> (X sdiv -C), tested by sub.ll:test16
...
llvm-svn: 16769
2004-10-06 15:08:25 +00:00
2ce32df8b0
Reduce code growth implied by the tail duplication pass by not duplicating
...
an instruction if it can be hoisted to a common dominator of the block.
This implements: test/Regression/Transforms/TailDup/MergeTest.ll
llvm-svn: 16758
2004-10-06 03:27:37 +00:00
abae776b18
Hrm, debugging printouts do not need to be in here
...
llvm-svn: 16598
2004-09-29 21:21:14 +00:00
6862fbd2cf
* Pull range optimization code out into new InsertRangeTest function.
...
* SubOne/AddOne functions always return ConstantInt, declare them as such
* Pull code for handling setcc X, cst, where cst is at the end of the range,
or cc is LE or GE up earlier in visitSetCondInst. This reduces #iterations
in some cases.
* Fold: (div X, C1) op C2 -> range check, implementing div.ll:test6 - test9.
llvm-svn: 16588
2004-09-29 17:40:11 +00:00
6a4adcda4c
Fold binary expressions and casts into PHI nodes that have all constant inputs.
...
This takes something like this:
%A = phi int [ 3, %cond_false.0 ], [ 2, %endif.0.i ], [ 2, %endif.1.i ]
%B = div int %tmp.243, 4
and turns it into:
%A = phi int [ 3/4, %cond_false.0 ], [ 2/4, %endif.0.i ], [ 2/4, %endif.1.i ]
which is later simplified (in this case) into %A = 0.
This triggers thousands of times in spec, for example, 269 times in 176.gcc.
This is tested by InstCombine/add.ll:test23 and set.ll:test18.
llvm-svn: 16582
2004-09-29 05:07:12 +00:00
c949128b2f
Hrm, really, all tests passed without this, but it is scary to think how...
...
llvm-svn: 16568
2004-09-29 03:16:24 +00:00
be7a69ebd8
Remove debugging printout
...
Instcombine (setcc (truncate X), C1).
This occurs THOUSANDS of times in many benchmarks. Particularlly common
seem to be things like (seteq (cast bool X to int), int 0)
This turns it into (seteq bool %X, false), which then becomes (not %X).
llvm-svn: 16567
2004-09-29 03:09:18 +00:00
dcf756ec22
Fold (X setcc C1) | (X setcc C2)
...
This implements or.ll:test1[89]
llvm-svn: 16561
2004-09-28 22:33:08 +00:00
623826c888
Fold (and (setcc X, C1), (setcc X, C2))
...
This is important for several reasons:
1. Benchmarks have lots of code that looks like this (perlbmk in particular):
%tmp.2.i = setne int %tmp.0.i, 128 ; <bool> [#uses=1]
%tmp.6343 = seteq int %tmp.0.i, 1 ; <bool> [#uses=1]
%tmp.63 = and bool %tmp.2.i, %tmp.6343 ; <bool> [#uses=1]
we now fold away the setne, a clear improvement.
2. In the more important cases, such as (X >= 10) & (X < 20), we now produce
smaller code: (X-10) < 10.
3. Perhaps the nicest effect of this patch is that it really helps out the
code generators. In particular, for a 'range test' like the above,
instead of generating this on X86 (the difference on PPC is even more
pronounced):
cmp %EAX, 50
setge %CL
cmp %EAX, 100
setl %AL
and %CL, %AL
cmp %CL, 0
we now generate this:
add %EAX, -50
cmp %EAX, 50
Furthermore, this causes setcc's to be folded into branches more often.
These combinations trigger dozens of times in the spec benchmarks, particularly
in 176.gcc, 186.crafty, 253.perlbmk, 254.gap, & 099.go.
llvm-svn: 16559
2004-09-28 21:48:02 +00:00
272d5ca9e0
Implement X / C1 / C2 folding
...
Implement (setcc (shl X, C1), C2) folding.
The second one occurs several dozen times in spec. The first was added
just in case. :)
These are tested by shift.ll:test2[12], and div.ll:test5
llvm-svn: 16549
2004-09-28 18:22:15 +00:00
6afc02f816
shl is always zero extending, so always use a zero extending shift right.
...
This latent bug was exposed by recent changes, and is tested as:
llvm/test/Regression/Transforms/InstCombine/2004-09-28-BadShiftAndSetCC.llx
llvm-svn: 16546
2004-09-28 17:54:07 +00:00
3ce42ec7ee
Pull assignment out of for loop conditional in order for this to
...
compile under windows. Patch contributed by Paolo Invernizzi!
llvm-svn: 16534
2004-09-28 02:40:37 +00:00
bfff18a869
Fix two bugs: one where a condition was mistakenly swapped, and another
...
where we folded (X & 254) -> X < 1 instead of X < 2. These problems were
latent problems exposed by the latest patch.
llvm-svn: 16528
2004-09-27 19:29:18 +00:00
1023b8726e
Fold: (setcc (shr X, ShAmt), CI), where 'cc' is eq or ne. This xform
...
triggers often, for example:
6x in povray, 1x in gzip, 279x in gcc, 1x in crafty, 8x in eon, 11x in perlbmk,
362x in gap, 4x in vortex, 14 in m88ksim, 211x in 126.gcc, 1x in compress,
11x in ijpeg, and 4x in 147.vortex.
llvm-svn: 16521
2004-09-27 16:18:50 +00:00
7e794273f5
Implement shift-and combinations, implementing InstCombine/and.ll:test19-21
...
These combinations trigger 4 times in povray, 7x in gcc, 4x in gap, and 2x in bzip2.
llvm-svn: 16508
2004-09-24 15:21:34 +00:00
e1b4d2a470
Move LHSI->hasOneUse() into the arms of the conditional, reindenting code.
...
No functionality changes here.
llvm-svn: 16505
2004-09-23 21:52:49 +00:00
8fc5af4da9
Implement Transforms/InstCombine/and.ll:test18, a case that occurs 20 times
...
in perlbmk
llvm-svn: 16504
2004-09-23 21:46:38 +00:00
bdcf41a8a2
Implement select.ll:test16: fold load (select C, X, null) -> load X
...
llvm-svn: 16499
2004-09-23 15:46:00 +00:00
b121ae1cec
Do not fold (X + C1 != C2) if there are other users of the add. Doing
...
this transformation used to take a loop like this:
int Array[1000];
void test(int X) {
int i;
for (i = 0; i < 1000; ++i)
Array[i] += X;
}
Compiled to LLVM is:
no_exit: ; preds = %entry, %no_exit
%indvar = phi uint [ 0, %entry ], [ %indvar.next, %no_exit ] ; <uint> [#uses=2]
%tmp.4 = getelementptr [1000 x int]* %Array, int 0, uint %indvar ; <int*> [#uses=2]
%tmp.7 = load int* %tmp.4 ; <int> [#uses=1]
%tmp.9 = add int %tmp.7, %X ; <int> [#uses=1]
store int %tmp.9, int* %tmp.4
*** %indvar.next = add uint %indvar, 1 ; <uint> [#uses=2]
*** %exitcond = seteq uint %indvar.next, 1000 ; <bool> [#uses=1]
br bool %exitcond, label %return, label %no_exit
and turn it into a loop like this:
no_exit: ; preds = %entry, %no_exit
%indvar = phi uint [ 0, %entry ], [ %indvar.next, %no_exit ] ; <uint> [#uses=3]
%tmp.4 = getelementptr [1000 x int]* %Array, int 0, uint %indvar ; <int*> [#uses=2]
%tmp.7 = load int* %tmp.4 ; <int> [#uses=1]
%tmp.9 = add int %tmp.7, %X ; <int> [#uses=1]
store int %tmp.9, int* %tmp.4
*** %indvar.next = add uint %indvar, 1 ; <uint> [#uses=1]
*** %exitcond = seteq uint %indvar, 999 ; <bool> [#uses=1]
br bool %exitcond, label %return, label %no_exit
Note that indvar.next and indvar can no longer be coallesced. In machine
code terms, this patch changes this code:
.LBBtest_1: # no_exit
mov %EDX, OFFSET Array
mov %ESI, %EAX
add %ESI, DWORD PTR [%EDX + 4*%ECX]
mov %EDX, OFFSET Array
mov DWORD PTR [%EDX + 4*%ECX], %ESI
mov %EDX, %ECX
inc %EDX
cmp %ECX, 999
mov %ECX, %EDX
jne .LBBtest_1 # no_exit
into this:
.LBBtest_1: # no_exit
mov %EDX, OFFSET Array
mov %ESI, %EAX
add %ESI, DWORD PTR [%EDX + 4*%ECX]
mov %EDX, OFFSET Array
mov DWORD PTR [%EDX + 4*%ECX], %ESI
inc %ECX
cmp %ECX, 1000
jne .LBBtest_1 # no_exit
We need better instruction selection to get this:
.LBBtest_1: # no_exit
add DWORD PTR [Array + 4*%ECX], EAX
inc %ECX
cmp %ECX, 1000
jne .LBBtest_1 # no_exit
... but at least there is less register juggling
llvm-svn: 16473
2004-09-21 21:35:23 +00:00
42618551d5
Fix potential miscompilations: InstCombine/2004-09-20-BadLoadCombine*.llx
...
llvm-svn: 16447
2004-09-20 10:15:10 +00:00
d59cebf87a
Fix loop condition so that we don't decrement off the beginning of the
...
list.
llvm-svn: 16440
2004-09-20 06:42:58 +00:00
3e86084641
Prototype these functions more accurately
...
llvm-svn: 16432
2004-09-20 04:43:15 +00:00
e6f13093e6
Make isSafeToLoadUnconditionally a bit smarter, implementing PR362 and
...
Regression/Transforms/InstCombine/CPP_min_max.llx
llvm-svn: 16409
2004-09-19 19:18:10 +00:00
f62ea8ef4b
Make instruction combining a bit more aggressive in the face of volatile
...
loads, and implement two new transforms: InstCombine/load.ll:test[56].
llvm-svn: 16404
2004-09-19 18:43:46 +00:00
6614946443
Convert code to compile with vc7.1.
...
Patch contributed by Paolo Invernizzi. Thanks Paolo!
llvm-svn: 16368
2004-09-15 17:06:42 +00:00
f11216d24f
Fix a bug in the previous checkin that broke 255.vortex
...
llvm-svn: 16355
2004-09-15 02:34:40 +00:00
a346578d92
Make sure to update alias analysis information as we transform the function.
...
This fixes PR420 and Regression/Transforms/LICM/2004-09-14-AliasAnalysisInvalidate.llx
llvm-svn: 16348
2004-09-15 01:04:07 +00:00
f41b80a05f
Remove a long-dead pass. Actually, this pass was never used at all.
...
llvm-svn: 16337
2004-09-14 16:33:01 +00:00
a5c04ee50f
Fixes to make LLVM compile with vc7.1.
...
Patch contributed by Paolo Invernizzi!
llvm-svn: 16152
2004-09-03 18:19:51 +00:00
7c16caa336
Changes For Bug 352
...
Move include/Config and include/Support into include/llvm/Config,
include/llvm/ADT and include/llvm/Support. From here on out, all LLVM
public header files must be under include/llvm/.
llvm-svn: 16137
2004-09-01 22:55:40 +00:00
f39f66e3ef
Initial checkin of a pass to lower packed operations to scalars operations.
...
This also registers the pass with opt with a -lower-packed command line
option.
Patch contributed by Brad Jones.
llvm-svn: 15987
2004-08-21 21:39:24 +00:00
4456da6a4c
Fix InstCombine/2004-08-10-BoolSetCC.ll, a bug that is miscompiling
...
176.gcc. Note that this is apparently not the only bug miscompiling gcc
though. :(
llvm-svn: 15639
2004-08-11 00:50:51 +00:00
8e7260652b
Fix InstCombine/2004-08-09-RemInfLoop.llx
...
This should go into the 1.3 branch
llvm-svn: 15593
2004-08-09 21:05:48 +00:00
832437255d
Stop using getValues().
...
llvm-svn: 15487
2004-08-04 08:44:43 +00:00
7aa2d4747a
Fix a regression in InstCombine/xor.ll
...
llvm-svn: 15410
2004-08-01 19:42:59 +00:00
9c003d8f65
Fix De Morgan's name.
...
llvm-svn: 15343
2004-07-30 12:50:08 +00:00
d4252a7c64
Start using the PatternMatcher a bit.
...
llvm-svn: 15342
2004-07-30 07:50:03 +00:00
63b38bd2ed
Fix #includes of i*.h => Instructions.h as per PR403.
...
llvm-svn: 15334
2004-07-29 17:30:56 +00:00
2b3387a6d9
Fix #includes of i*.h => Instructions.h as per PR403.
...
llvm-svn: 15328
2004-07-29 17:05:13 +00:00
7b5b86cd0f
This change fixed a bug in the function visitMul. The prior version
...
assumed that a constant on the RHS of a multiplication was either an
IntConstant or an FPConstant. It checked for an IntConstant and then,
if it did not find one, did a hard cast to an FPConstant. That code
would crash if the RHS were a ConstantExpr that was neither an
IntConstant nor an FPConstant. This version replaces the hard cast
with a dyn_cast. It performs the same way for IntConstants and
FPConstants but does nothing, instead of crashing, for constant
expressions.
The regression test for this change is 2004-07-27-ConstantExprMul.ll.
llvm-svn: 15291
2004-07-27 21:02:21 +00:00
38b79e8fbc
Make the create...() functions for some of these passes return a FunctionPass *.
...
llvm-svn: 15276
2004-07-27 17:43:21 +00:00
50eb771d37
Fix hoisting of void typed values, e.g. calls
...
llvm-svn: 15263
2004-07-27 07:38:32 +00:00
f29807169a
Implement DeadStoreElim/alloca.llx by observing that allocas are dead at the
...
end of the function (either return or unwind)
llvm-svn: 15232
2004-07-26 06:14:11 +00:00
e5ad26dbb3
Throttle back indvar substitution from creating multiplies in loops. This is bad bad bad.
...
llvm-svn: 15227
2004-07-26 02:47:12 +00:00
7b25bcdf52
* Substantially simplify how free instructions are handled (potentially fixing
...
a bug in DSE).
* Delete dead operand uses iteratively instead of recursively, using a
SetVector.
* Defer deletion of dead operand uses until the end of processing, which means
we don't have to bother with updating the AliasSetTracker. This speeds up
DSE substantially.
llvm-svn: 15204
2004-07-25 11:09:56 +00:00
4c1c1ac7e4
Free instructions kill values too. This implements DeadStoreElim/free.llx
...
llvm-svn: 15199
2004-07-25 07:58:38 +00:00
bad6478b00
obvious fix
...
llvm-svn: 15162
2004-07-24 07:51:27 +00:00
3844c300de
This is a trivial dead store elimination pass. It very very simple and
...
can be improved in many ways. But: stop laughing, even with -basicaa it
deletes 15% of the stores in 252.eon :)
llvm-svn: 15101
2004-07-22 08:00:28 +00:00
51f7c9e56d
Update GC intrinsics to take a pointer to the object as well as a pointer
...
to the field being updated. Patch contributed by Tobias Nurmiranta
llvm-svn: 15097
2004-07-22 05:51:13 +00:00
d8f5e2ccac
* Further cleanup.
...
* Test for whether bits are shifted out during the optzn.
If so, the fold is illegal, though it can be handled explicitly for setne/seteq
This fixes the miscompilation of 254.gap last night, which was a latent bug
exposed by other optimizer improvements.
llvm-svn: 15085
2004-07-21 20:14:10 +00:00
1638de4499
Make cast-cast code a bit more defensive
...
"simplify" a bit of code for comparison/and folding
llvm-svn: 15082
2004-07-21 19:50:44 +00:00
4fbad968f8
Remove special casing of pointers and treat them generically as integers of
...
the appopriate size. This gives us the ability to eliminate int -> ptr -> int
llvm-svn: 15063
2004-07-21 04:27:24 +00:00
11ffd59e37
Implement Transforms/InstCombine/IntPtrCast.ll
...
llvm-svn: 15029
2004-07-20 05:21:00 +00:00
44d0b9502a
Implement InstCombine/GEPIdxCanon.ll
...
llvm-svn: 15024
2004-07-20 01:48:15 +00:00
4e2dbc6b4a
Rewrite cast->cast elimination code completely based on the information we
...
actually care about. Someday when the cast instruction is gone, we can do
better here, but this will do for now. This implements
instcombine/cast.ll:test17/18 as well.
llvm-svn: 15018
2004-07-20 00:59:32 +00:00
f3edc49ae2
Minor cleanup, no functionality change
...
llvm-svn: 14972
2004-07-18 18:59:44 +00:00
f0a5bcaae4
Delete a redundant if branch.
...
llvm-svn: 14967
2004-07-18 08:34:52 +00:00
c44cb6bd9f
Expand the coercion of constants to include the newly constant Globals.
...
llvm-svn: 14966
2004-07-18 08:34:19 +00:00
539429d9b5
Delete a no-op loop.
...
llvm-svn: 14965
2004-07-18 08:32:43 +00:00
6c2b627e23
Expand the scope to include global values because they are now constants
...
too.
llvm-svn: 14964
2004-07-18 08:32:10 +00:00
cb3fb5d4f5
bug 122:
...
- Replace ConstantPointerRef usage with GlobalValue usage
llvm-svn: 14953
2004-07-18 00:44:37 +00:00
874368790f
bug 122:
...
- Replace ConstantPointerRef usage with GlobalValue usage
- Minimize redundant isa<GlobalValue> usage
- Correct isa<Constant> for GlobalValue subclass
llvm-svn: 14950
2004-07-18 00:38:32 +00:00
ef784f01dd
bug 122:
...
- Minimize redundant isa<GlobalValue> usage
llvm-svn: 14948
2004-07-18 00:32:14 +00:00
c5afc9512b
bug 122:
...
- Replace ConstantPointerRef usage with GlobalValue usage
- Correct isa<Constant> for GlobalValue subclass
llvm-svn: 14947
2004-07-18 00:31:05 +00:00
9e855c6832
bug 122:
...
- Minimize redundant isa<GlobalValue> usage
- Correct isa<Constant> for GlobalValue subclass
llvm-svn: 14946
2004-07-18 00:29:57 +00:00
d79334df33
This patch was contributed by Daniel Berlin!
...
Speed up SCCP substantially by processing overdefined values quickly. This
patch speeds up SCCP by about 30-40% on large testcases.
llvm-svn: 14861
2004-07-15 23:36:43 +00:00
f2c018c0c1
Fix PR404 try #2
...
This version takes about 1s longer than the previous one (down to 2.35s),
but on the positive side, it actually works :)
llvm-svn: 14856
2004-07-15 08:20:22 +00:00
daa12135da
Revert previous patch until I get a bug fixed
...
llvm-svn: 14853
2004-07-15 05:36:31 +00:00
70177e402d
Fix PR404: Loop simplify is really slow on 252.eon
...
This eliminates an N*N*logN algorithm from the loop simplify pass, replacing
it with a much simpler and faster alternative. In a debug build, this reduces
gccas time on eon from 85s to 42s.
llvm-svn: 14851
2004-07-15 04:27:04 +00:00
9a63520b1a
Fixes working towards PR341
...
llvm-svn: 14839
2004-07-15 01:50:47 +00:00
ba7aef39fd
Now that we codegen the portable "sizeof" efficiently, we can use it for
...
malloc lowering. This means that lowerallocations doesn't need targetdata
anymore. yaay.
llvm-svn: 14835
2004-07-15 01:08:08 +00:00
35e24774eb
Factor some code to handle "load (constantexpr cast foo)" just like
...
"load (cast foo)". This allows us to compile C++ code like this:
class Bclass {
public: virtual int operator()() { return 666; }
};
class Dclass: public Bclass {
public: virtual int operator()() { return 667; }
} ;
int main(int argc, char** argv) {
Dclass x;
return x();
}
Into this:
int %main(int %argc, sbyte** %argv) {
entry:
call void %__main( )
ret int 667
}
Instead of this:
int %main(int %argc, sbyte** %argv) {
entry:
%x = alloca "struct.std::bad_typeid" ; <"struct.std::bad_typeid"*> [#uses=3]
call void %__main( )
%tmp.1.i.i = getelementptr "struct.std::bad_typeid"* %x, uint 0, uint 0, uint 0 ; <int (...)***> [#uses=1]
store int (...)** getelementptr ([3 x int (...)*]* %vtable for Bclass, int 0, long 2), int (...)*** %tmp.1.i.i
%tmp.3.i = getelementptr "struct.std::bad_typeid"* %x, int 0, uint 0, uint 0 ; <int (...)***> [#uses=1]
store int (...)** getelementptr ([3 x int (...)*]* %vtable for Dclass, int 0, long 2), int (...)*** %tmp.3.i
%tmp.5 = load int ("struct.std::bad_typeid"*)** cast (int (...)** getelementptr ([3 x int (...)*]* %vtable for Dclass, int 0, long 2) to int
("struct.std::bad_typeid"*)**) ; <int ("struct.std::bad_typeid"*)*> [#uses=1]
%tmp.6 = call int %tmp.5( "struct.std::bad_typeid"* %x ) ; <int> [#uses=1]
ret int %tmp.6
ret int 0
}
In order words, we now resolve the virtual function call.
llvm-svn: 14783
2004-07-13 01:49:43 +00:00
9eb9ccd9f6
Check to make sure types are sized before calling getTypeSize on them.
...
llvm-svn: 14649
2004-07-06 19:28:42 +00:00
a501be556f
It doesn't matter what the 2nd operand is; if the GEP has 2 operands and
...
the first is a zero, we should leave it alone.
llvm-svn: 14648
2004-07-06 19:24:47 +00:00
0e0fe8a2e9
Add helper function.
...
Don't touch GEPs for which DecomposeArrayRef is not going to do anything
special (e.g., < 2 indices, or 2 indices and the last one is a constant.)
llvm-svn: 14647
2004-07-06 18:15:39 +00:00
23b47b6af9
Implement rem.ll:test3
...
llvm-svn: 14640
2004-07-06 07:38:18 +00:00
98c6bdf251
Fix a minor bug where we would go into infinite loops on some constants
...
llvm-svn: 14638
2004-07-06 07:11:42 +00:00
7fd5f0745a
Implement InstCombine/sub.ll:test15: X % -Y === X % Y
...
Also, remove X % -1 = 0, because it's not true for unsigneds, and the
signed case is superceeded by this new handling.
llvm-svn: 14637
2004-07-06 07:01:22 +00:00
eb04d9bcb4
Add #include <iostream> since Value.h does not #include it any more.
...
llvm-svn: 14622
2004-07-04 12:19:56 +00:00
4c9c20af28
Implement add.ll:test22, a common case in MSIL files
...
llvm-svn: 14587
2004-07-03 00:26:11 +00:00
49df6cefa5
Do not call getTypeSize on a type that has no size
...
llvm-svn: 14584
2004-07-02 22:55:47 +00:00
e1a136fb4b
Get rid of a dead variable, and fix a typo in a comment.
...
llvm-svn: 14560
2004-07-02 05:30:01 +00:00
163c87fc32
Make this pass use a more specific debug message than "Processing:".
...
llvm-svn: 14541
2004-07-01 19:27:10 +00:00
6e07936ed2
Implement InstCombine/add.ll:test21
...
llvm-svn: 14443
2004-06-27 22:51:36 +00:00
7f4222237d
New constant expression lowering pass to simplify your instruction selection needs.
...
Contributed by Vladimir Prus!
llvm-svn: 14399
2004-06-25 07:48:09 +00:00
7a002d6010
Two fixes. First, stop using the ugly shouldSubstituteIndVar method.
...
Second, disable substitution of quadratic addrec expressions to avoid putting
multiplies in loops!
llvm-svn: 14358
2004-06-24 06:49:18 +00:00
c9e06336ab
Make use of BinaryOperator::create* methods to shrinkify code.
...
llvm-svn: 14262
2004-06-20 05:04:01 +00:00
42ad646104
Now that dominator tree children are built in determinstic order, this horrible code
...
can go away
llvm-svn: 14254
2004-06-19 20:23:35 +00:00
4027500e1c
Fix a nasty bug, noticed by Reid
...
llvm-svn: 14249
2004-06-19 18:15:50 +00:00
ec2d34cc19
Fix one source of nondeterminism in the -licm pass: the hoist pass
...
was processing blocks in whatever order they happened to end up in the
dominator tree data structure. Force an ordering.
llvm-svn: 14248
2004-06-19 08:56:43 +00:00
b5f8eb8315
Do not loop over uses as we delete them. This causes iterators to be
...
invalidated out from under us. This bug goes back to revision 1.1: scary.
llvm-svn: 14242
2004-06-19 02:02:22 +00:00
023a483c76
Implement Transforms/InstCombine/and.ll:test17, a common case that
...
occurs due to unordered comparison macros in math.h
llvm-svn: 14221
2004-06-18 06:07:51 +00:00
97bfcea262
Rename Type::PrimitiveID to TypeId and ::getPrimitiveID() to ::getTypeID()
...
Delete two functions that are now methods on the Type class
llvm-svn: 14200
2004-06-17 18:16:02 +00:00
661963c63f
Fix typo in DEBUG printout.
...
llvm-svn: 14196
2004-06-17 07:26:52 +00:00
ee59d4bf04
Fix a bug in my checkin from last night that caused miscompilations of
...
186.crafty, fhourstones and 132.ijpeg.
Bugpoint makes really nasty miscompilations embarassingly easy to find. It
narrowed it down to the instcombiner and this testcase (from fhourstones):
bool %l7153_l4706_htstat_loopentry_2E_4_no_exit_2E_4(int* %i, [32 x int]* %works, int* %tmp.98.out) {
newFuncRoot:
%tmp.96 = load int* %i ; <int> [#uses=1]
%tmp.97 = getelementptr [32 x int]* %works, long 0, int %tmp.96 ; <int*> [#uses=1]
%tmp.98 = load int* %tmp.97 ; <int> [#uses=2]
%tmp.99 = load int* %i ; <int> [#uses=1]
%tmp.100 = and int %tmp.99, 7 ; <int> [#uses=1]
%tmp.101 = seteq int %tmp.100, 7 ; <bool> [#uses=2]
%tmp.102 = cast bool %tmp.101 to int ; <int> [#uses=0]
br bool %tmp.101, label %codeRepl4.exitStub, label %codeRepl3.exitStub
codeRepl4.exitStub: ; preds = %newFuncRoot
store int %tmp.98, int* %tmp.98.out
ret bool true
codeRepl3.exitStub: ; preds = %newFuncRoot
store int %tmp.98, int* %tmp.98.out
ret bool false
}
... which only has one combination performed on it:
$ llvm-as < t.ll | opt -instcombine -debug | llvm-dis
IC: Old = %tmp.101 = seteq int %tmp.100, 7 ; <bool> [#uses=1]
New = setne int %tmp.100, 0 ; <bool>:<badref> [#uses=0]
IC: MOD = br bool %tmp.101, label %codeRepl3.exitStub, label %codeRepl4.exitStub
IC: MOD = %tmp.97 = getelementptr [32 x int]* %works, uint 0, int %tmp.96 ; <int*> [#uses=1]
It doesn't get much better than this. :)
llvm-svn: 14109
2004-06-10 02:33:20 +00:00
c8e7e298c1
More minor cleanups
...
llvm-svn: 14108
2004-06-10 02:12:35 +00:00
df20a4d589
Eliminate many occurrances of Instruction::
...
llvm-svn: 14107
2004-06-10 02:07:29 +00:00
35167c3087
Implement InstCombine/select.ll:test15*
...
llvm-svn: 14095
2004-06-09 07:59:58 +00:00
396dbfe327
Be more careful about the order we put stuff onto the worklist. This allow us to
...
collapse this:
bool %le(int %A, int %B) {
%c1 = setgt int %A, %B
%tmp = select bool %c1, int 1, int 0
%c2 = setlt int %A, %B
%result = select bool %c2, int -1, int %tmp
%c3 = setle int %result, 0
ret bool %c3
}
into:
bool %le(int %A, int %B) {
%c3 = setle int %A, %B ; <bool> [#uses=1]
ret bool %c3
}
which is handy, because the Java FE makes these sequences all over the place.
This is tested as: test/Regression/Transforms/InstCombine/JavaCompare.ll
llvm-svn: 14086
2004-06-09 05:08:07 +00:00
2dd017402b
Implement select.ll:test14*
...
llvm-svn: 14083
2004-06-09 04:24:29 +00:00
523d3e6674
Fix one of the major things that is causing the C Backend to infinite loop
...
llvm-svn: 13872
2004-05-28 05:02:13 +00:00
ed79d8af53
Fix InstCombine/load.ll & PR347.
...
This code hadn't been updated after the "structs with more than 256 elements"
related changes to the GEP instruction. Also it was not handling the
ConstantAggregateZero class.
Now it does!
llvm-svn: 13834
2004-05-27 17:30:27 +00:00
297d7fe7e6
Remove unused header file.
...
llvm-svn: 13750
2004-05-25 08:51:36 +00:00
1cc31f264f
Make this pass simply invoke SymbolTable::strip().
...
llvm-svn: 13749
2004-05-25 08:51:25 +00:00
e1e10e1883
Implement InstCombine:shift.ll:test16, which turns (X >> C1) & C2 != C3
...
into (X & (C2 << C1)) != (C3 << C1), where the shift may be either left or
right and the compare may be any one.
This triggers 1546 times in 176.gcc alone, as it is a common pattern that
occurs for bitfield accesses.
llvm-svn: 13740
2004-05-25 06:32:08 +00:00
03841659a4
Implement instcombine/cast.ll:test16:
...
Canonicalize cast X to bool into a setne instruction
llvm-svn: 13736
2004-05-25 04:29:21 +00:00
99173879ad
Spelling people's names right is kinda important
...
llvm-svn: 13702
2004-05-23 21:27:29 +00:00
289ba2ac4d
Adjust to the changes in the AliasSetTracker interface
...
llvm-svn: 13690
2004-05-23 21:20:19 +00:00
e67dbc2ae2
Add support for replacement of formal arguments with simpler expressions.
...
llvm-svn: 13689
2004-05-23 21:19:55 +00:00
099c8cfe90
Implement the -lowergc pass which is used by code generators (like the CBE)
...
that do not have builtin support for garbage collection.
llvm-svn: 13688
2004-05-23 21:19:22 +00:00
0026512bac
This was not meant to be committed
...
llvm-svn: 13565
2004-05-13 20:56:34 +00:00
c12c945cc4
Fix a nasty bug that caused us to unroll EXTREMELY large loops due to overflow
...
in the size calculation.
This is not something you want to see:
Loop Unroll: F[main] Loop %no_exit Loop Size = 2 Trip Count = 2147483648 - UNROLLING!
The problem was that 2*2147483648 == 0.
Now we get:
Loop Unroll: F[main] Loop %no_exit Loop Size = 2 Trip Count = 2147483648 - TOO LARGE: 4294967296>100
Thanks to some anonymous person playing with the demo page that repeatedly
caused zion to go into swapping land. That's one way to ensure you'll get
a quick bugfix. :)
Testcase here: Transforms/LoopUnroll/2004-05-13-DontUnrollTooMuch.ll
llvm-svn: 13564
2004-05-13 20:43:31 +00:00
8ec5f88c79
Fix stupid bug in my checkin yesterday
...
llvm-svn: 13429
2004-05-08 22:41:42 +00:00
5f667a6f58
Implement folding of GEP's like:
...
%tmp.0 = getelementptr [50 x sbyte]* %ar, uint 0, int 5 ; <sbyte*> [#uses=2]
%tmp.7 = getelementptr sbyte* %tmp.0, int 8 ; <sbyte*> [#uses=1]
together. This patch actually allows us to simplify and generalize the code.
llvm-svn: 13415
2004-05-07 22:09:22 +00:00
d9e5813821
Fix PR336: The instcombine pass asserts when visiting load instruction
...
llvm-svn: 13400
2004-05-07 15:35:56 +00:00
9490849028
Do not mark instructions in unreachable sections of the function as live.
...
This fixes PR332 and ADCE/2004-05-04-UnreachableBlock.llx
llvm-svn: 13349
2004-05-04 17:00:46 +00:00
dd1a86d858
Minor efficiency tweak, suggested by Patrick Meredith
...
llvm-svn: 13341
2004-05-04 15:19:33 +00:00
63d75af920
Make sure to reprocess instructions used by deleted instructions to avoid
...
missing opportunities for combination.
llvm-svn: 13309
2004-05-01 23:27:23 +00:00
b643a9e675
Make sure the instruction combiner doesn't lose track of instructions
...
when replacing them, missing the opportunity to do simplifications
llvm-svn: 13308
2004-05-01 23:19:52 +00:00
652064e3b8
Fix a major pessimization in the instcombiner. If an allocation instruction
...
is only used by a cast, and the casted type is the same size as the original
allocation, it would eliminate the cast by folding it into the allocation.
Unfortunately, it was placing the new allocation instruction right before
the cast, which could pull (for example) alloca instructions into the body
of a function. This turns statically allocatable allocas into expensive
dynamically allocated allocas, which is bad bad bad.
This fixes the problem by placing the new allocation instruction at the same
place the old one was, duh. :)
llvm-svn: 13289
2004-04-30 04:37:52 +00:00
2d3a7a6ff0
Changes to fix up the inst_iterator to pass to boost iterator checks. This
...
patch was graciously contributed by Vladimir Prus.
llvm-svn: 13185
2004-04-27 15:13:33 +00:00
e20c334e65
Instcombine X/-1 --> 0-X
...
llvm-svn: 13172
2004-04-26 14:01:59 +00:00
83cd87efcd
Move the scev expansion code into this pass, where it belongs. There is
...
still room for cleanup, but at least the code modification is out of the
analysis now.
llvm-svn: 13135
2004-04-23 21:29:48 +00:00
c27302c79f
Disable a previous patch that was causing indvars to loop infinitely :(
...
llvm-svn: 13108
2004-04-22 15:12:36 +00:00
c1a682dda0
Fix an extremely serious thinko I made in revision 1.60 of this file.
...
llvm-svn: 13106
2004-04-22 14:59:40 +00:00
af532f27e7
Implement a todo, rewriting all possible scev expressions inside of the
...
loop. This eliminates the extra add from the previous case, but it's
not clear that this will be a performance win overall. Tommorows test
results will tell. :)
llvm-svn: 13103
2004-04-21 23:36:08 +00:00
fb9a299f68
This code really wants to iterate over the OPERANDS of an instruction, not
...
over its USES. If it's dead it doesn't have any uses! :)
Thanks to the fabulous and mysterious Bill Wendling for pointing this out. :)
llvm-svn: 13102
2004-04-21 22:29:37 +00:00
dc7cc35088
Implement a fixme. The helps loops that have induction variables of different
...
types in them. Instead of creating an induction variable for all types, it
creates a single induction variable and casts to the other sizes. This generates
this code:
no_exit: ; preds = %entry, %no_exit
%indvar = phi uint [ %indvar.next, %no_exit ], [ 0, %entry ] ; <uint> [#uses=4]
*** %j.0.0 = cast uint %indvar to short ; <short> [#uses=1]
%indvar = cast uint %indvar to int ; <int> [#uses=1]
%tmp.7 = getelementptr short* %P, uint %indvar ; <short*> [#uses=1]
store short %j.0.0, short* %tmp.7
%inc.0 = add int %indvar, 1 ; <int> [#uses=2]
%tmp.2 = setlt int %inc.0, %N ; <bool> [#uses=1]
%indvar.next = add uint %indvar, 1 ; <uint> [#uses=1]
br bool %tmp.2, label %no_exit, label %loopexit
instead of:
no_exit: ; preds = %entry, %no_exit
%indvar = phi ushort [ %indvar.next, %no_exit ], [ 0, %entry ] ; <ushort> [#uses=2]
*** %indvar = phi uint [ %indvar.next, %no_exit ], [ 0, %entry ] ; <uint> [#uses=3]
%indvar = cast uint %indvar to int ; <int> [#uses=1]
%indvar = cast ushort %indvar to short ; <short> [#uses=1]
%tmp.7 = getelementptr short* %P, uint %indvar ; <short*> [#uses=1]
store short %indvar, short* %tmp.7
%inc.0 = add int %indvar, 1 ; <int> [#uses=2]
%tmp.2 = setlt int %inc.0, %N ; <bool> [#uses=1]
%indvar.next = add uint %indvar, 1
*** %indvar.next = add ushort %indvar, 1
br bool %tmp.2, label %no_exit, label %loopexit
This is an improvement in register pressure, but probably doesn't happen that
often.
The more important fix will be to get rid of the redundant add.
llvm-svn: 13101
2004-04-21 22:22:01 +00:00
c1aa21f5a7
Fix PR325
...
llvm-svn: 13081
2004-04-20 20:26:03 +00:00
f48f777d4c
Initial checkin of a simple loop unswitching pass. It still needs work,
...
but it's a start, and seems to do it's basic job.
llvm-svn: 13068
2004-04-19 18:07:02 +00:00
bc02177fdc
Add #include
...
llvm-svn: 13057
2004-04-19 03:01:23 +00:00
fc44a25bcb
Move isLoopInvariant to the Loop class
...
llvm-svn: 13051
2004-04-18 22:46:08 +00:00
827826320d
Correct rewriting of exit blocks after my last patch
...
llvm-svn: 13048
2004-04-18 22:27:10 +00:00
35eaa55cfc
Loop exit sets are no longer explicitly held, they are dynamically computed on demand.
...
llvm-svn: 13046
2004-04-18 22:15:13 +00:00
d72c3eb54e
Change the ExitBlocks list from being explicitly contained in the Loop
...
structure to being dynamically computed on demand. This makes updating
loop information MUCH easier.
llvm-svn: 13045
2004-04-18 22:14:10 +00:00
d15250240c
Reduce the unrolling limit
...
llvm-svn: 13040
2004-04-18 18:06:14 +00:00
30ae18155d
If the preheader of the loop was the entry block of the function, make sure
...
that the exit block of the loop becomes the new entry block of the function.
This was causing a verifier assertion on 252.eon.
llvm-svn: 13039
2004-04-18 17:38:42 +00:00
230bcb6b35
Be much more careful about how we update instructions outside of the loop
...
using instructions inside of the loop. This should fix the MishaTest failure
from last night.
llvm-svn: 13038
2004-04-18 17:32:39 +00:00
4d52e1e401
After unrolling our single basic block loop, fold it into the preheader and exit
...
block. The primary motivation for doing this is that we can now unroll nested loops.
This makes a pretty big difference in some cases. For example, in 183.equake,
we are now beating the native compiler with the CBE, and we are a lot closer
with LLC.
I'm now going to play around a bit with the unroll factor and see what effect
it really has.
llvm-svn: 13034
2004-04-18 06:27:43 +00:00
f2cc841619
Fix a bug: this does not preserve the CFG!
...
While we're at it, add support for updating loop information correctly.
llvm-svn: 13033
2004-04-18 05:38:37 +00:00
946b255977
Initial checkin of a simple loop unroller. This pass is extremely basic and
...
limited. Even in it's extremely simple state (it can only *fully* unroll single
basic block loops that execute a constant number of times), it already helps improve
performance a LOT on some benchmarks, particularly with the native code generators.
llvm-svn: 13028
2004-04-18 05:20:17 +00:00
c14da9600b
Make the tail duplication threshold accessible from the command line instead of hardcoded
...
llvm-svn: 13025
2004-04-18 00:52:43 +00:00
a814080025
If the loop executes a constant number of times, try a bit harder to replace
...
exit values.
llvm-svn: 13018
2004-04-17 18:44:09 +00:00
1e9ac1a45e
Fix a HUGE pessimization on X86. The indvars pass was taking this
...
(familiar) function:
int _strlen(const char *str) {
int len = 0;
while (*str++) len++;
return len;
}
And transforming it to use a ulong induction variable, because the type of
the pointer index was left as a constant long. This is obviously very bad.
The fix is to shrink long constants in getelementptr instructions to intptr_t,
making the indvars pass insert a uint induction variable, which is much more
efficient.
Here's the before code for this function:
int %_strlen(sbyte* %str) {
entry:
%tmp.13 = load sbyte* %str ; <sbyte> [#uses=1]
%tmp.24 = seteq sbyte %tmp.13, 0 ; <bool> [#uses=1]
br bool %tmp.24, label %loopexit, label %no_exit
no_exit: ; preds = %entry, %no_exit
*** %indvar = phi uint [ %indvar.next, %no_exit ], [ 0, %entry ] ; <uint> [#uses=2]
*** %indvar = phi ulong [ %indvar.next, %no_exit ], [ 0, %entry ] ; <ulong> [#uses=2]
%indvar1 = cast ulong %indvar to uint ; <uint> [#uses=1]
%inc.02.sum = add uint %indvar1, 1 ; <uint> [#uses=1]
%inc.0.0 = getelementptr sbyte* %str, uint %inc.02.sum ; <sbyte*> [#uses=1]
%tmp.1 = load sbyte* %inc.0.0 ; <sbyte> [#uses=1]
%tmp.2 = seteq sbyte %tmp.1, 0 ; <bool> [#uses=1]
%indvar.next = add ulong %indvar, 1 ; <ulong> [#uses=1]
%indvar.next = add uint %indvar, 1 ; <uint> [#uses=1]
br bool %tmp.2, label %loopexit.loopexit, label %no_exit
loopexit.loopexit: ; preds = %no_exit
%indvar = cast uint %indvar to int ; <int> [#uses=1]
%inc.1 = add int %indvar, 1 ; <int> [#uses=1]
ret int %inc.1
loopexit: ; preds = %entry
ret int 0
}
Here's the after code:
int %_strlen(sbyte* %str) {
entry:
%inc.02 = getelementptr sbyte* %str, uint 1 ; <sbyte*> [#uses=1]
%tmp.13 = load sbyte* %str ; <sbyte> [#uses=1]
%tmp.24 = seteq sbyte %tmp.13, 0 ; <bool> [#uses=1]
br bool %tmp.24, label %loopexit, label %no_exit
no_exit: ; preds = %entry, %no_exit
*** %indvar = phi uint [ %indvar.next, %no_exit ], [ 0, %entry ] ; <uint> [#uses=3]
%indvar = cast uint %indvar to int ; <int> [#uses=1]
%inc.0.0 = getelementptr sbyte* %inc.02, uint %indvar ; <sbyte*> [#uses=1]
%inc.1 = add int %indvar, 1 ; <int> [#uses=1]
%tmp.1 = load sbyte* %inc.0.0 ; <sbyte> [#uses=1]
%tmp.2 = seteq sbyte %tmp.1, 0 ; <bool> [#uses=1]
%indvar.next = add uint %indvar, 1 ; <uint> [#uses=1]
br bool %tmp.2, label %loopexit, label %no_exit
loopexit: ; preds = %entry, %no_exit
%len.0.1 = phi int [ 0, %entry ], [ %inc.1, %no_exit ] ; <int> [#uses=1]
ret int %len.0.1
}
llvm-svn: 13016
2004-04-17 18:16:10 +00:00
885a6eb74d
Even if there are not any induction variables in the loop, if we can compute
...
the trip count for the loop, insert one so that we can canonicalize the exit
condition.
llvm-svn: 13015
2004-04-17 18:08:33 +00:00
284d3b0311
Fix some really nasty dominance bugs that were exposed by my patch to
...
make the verifier more strict. This fixes building zlib
llvm-svn: 13002
2004-04-16 18:08:07 +00:00
9e9b2b7474
Fix some of the strange CBE-only failures that happened last night.
...
llvm-svn: 12980
2004-04-16 06:03:17 +00:00
d7a559e353
Fix a bug in the previous checkin: if the exit block is not the same as
...
the back-edge block, we must check the preincremented value.
llvm-svn: 12968
2004-04-15 20:26:22 +00:00
0cec5cb92c
Change the canonical induction variable that we insert.
...
Instead of producing code like this:
Loop:
X = phi 0, X2
...
X2 = X + 1
if (X != N-1) goto Loop
We now generate code that looks like this:
Loop:
X = phi 0, X2
...
X2 = X + 1
if (X2 != N) goto Loop
This has two big advantages:
1. The trip count of the loop is now explicit in the code, allowing
the direct implementation of Loop::getTripCount()
2. This reduces register pressure in the loop, and allows X and X2 to be
put into the same register.
As a consequence of the second point, the code we generate for loops went
from:
.LBB2: # no_exit.1
...
mov %EDI, %ESI
inc %EDI
cmp %ESI, 2
mov %ESI, %EDI
jne .LBB2 # PC rel: no_exit.1
To:
.LBB2: # no_exit.1
...
inc %ESI
cmp %ESI, 3
jne .LBB2 # PC rel: no_exit.1
... which has two fewer moves, and uses one less register.
llvm-svn: 12961
2004-04-15 15:21:43 +00:00
6679e46b59
ADd a trivial instcombine: load null -> null
...
llvm-svn: 12940
2004-04-14 03:28:36 +00:00
ff9362a8da
Add SCCP support for constant folding calls, implementing:
...
test/Regression/Transforms/SCCP/calltest.ll
llvm-svn: 12921
2004-04-13 19:43:54 +00:00
d0dc6d5295
Constant propagation should remove the dead instructions
...
llvm-svn: 12917
2004-04-13 19:28:20 +00:00
89e959bb1f
Fix LoopSimplify/2004-04-13-LoopSimplifyUpdateDomFrontier.ll
...
LoopSimplify was not updating dominator frontiers correctly in some cases.
llvm-svn: 12890
2004-04-13 16:23:25 +00:00
a6e22814ab
Refactor code a bit to make it simpler and eliminate the goto
...
llvm-svn: 12888
2004-04-13 15:21:18 +00:00
8417052938
This patch addresses PR35: Loop simplify should reconstruct nested loops.
...
This is fairly straight-forward, but was a real nightmare to get just
perfect. aarg. :)
llvm-svn: 12884
2004-04-13 05:05:33 +00:00
494a685449
Add support for removing invoke instructions
...
llvm-svn: 12858
2004-04-12 05:15:13 +00:00
24cf0200c7
Fix a bug in my select transformation
...
llvm-svn: 12826
2004-04-11 01:39:19 +00:00
f16fe7206c
Update the value numbering interface.
...
llvm-svn: 12824
2004-04-10 22:33:34 +00:00
623fba1107
Implement InstCombine/select.ll:test13*
...
llvm-svn: 12821
2004-04-10 22:21:27 +00:00
cf4a996cba
Implement InstCombine/add.ll:test20
...
Canonicalize add of sign bit constant into a xor
llvm-svn: 12819
2004-04-10 22:01:55 +00:00
69c4900512
Rewrite the GCSE pass to be *substantially* simpler, a bit more efficient,
...
and a bit more powerful
llvm-svn: 12817
2004-04-10 21:11:11 +00:00
f9d9665138
Fix spurious warning in release mode
...
llvm-svn: 12816
2004-04-10 19:15:56 +00:00
d95ef7eff0
Simplify code a bit, and fix a bug that was breaking perlbmk
...
llvm-svn: 12814
2004-04-10 18:06:21 +00:00
7ebfe61dc1
Fix a bug in my checkin last night that was breaking programs using invoke.
...
llvm-svn: 12813
2004-04-10 16:53:29 +00:00
5093213c40
Fix previous patch
...
llvm-svn: 12811
2004-04-10 07:27:48 +00:00
6149ac8991
Correctly update counters
...
llvm-svn: 12810
2004-04-10 07:02:02 +00:00
cfa1adcdb8
Simplify code a bit, and use alias analysis to allow us to delete unused
...
call and invoke instructions that are known to not write to memory.
llvm-svn: 12807
2004-04-10 06:53:09 +00:00
56e4d3d8ad
Implement select.ll:test12*
...
This transforms code like this:
%C = or %A, %B
%D = select %cond, %C, %A
into:
%C = select %cond, %B, 0
%D = or %A, %C
Since B is often a constant, the select can often be eliminated. In any case,
this reduces the usage count of A, allowing subsequent optimizations to happen.
This xform applies when the operator is any of:
add, sub, mul, or, xor, and, shl, shr
llvm-svn: 12800
2004-04-09 23:46:01 +00:00
183b336a54
Fold binary operators with a constant operand into select instructions
...
that have a constant operand. This implements
add.ll:test19, shift.ll:test15*, and others that are not tested
llvm-svn: 12794
2004-04-09 19:05:30 +00:00
cf7baf3519
Implement select.ll:test11
...
llvm-svn: 12793
2004-04-09 18:19:44 +00:00
e228ee5870
Implement InstCombine/cast-propagate.ll
...
llvm-svn: 12784
2004-04-08 20:39:49 +00:00
1c631e813d
Implement InstCombine/select.ll:test[7-10]
...
llvm-svn: 12769
2004-04-08 04:43:23 +00:00
2b2412d0c8
Implement test/Regression/Transforms/InstCombine/getelementptr_index.ll
...
llvm-svn: 12762
2004-04-07 18:38:20 +00:00
4d1fcf1dcd
Fix a bug in yesterdays checkins which broke siod. siod is a great testcase! :)
...
llvm-svn: 12659
2004-04-05 16:02:41 +00:00
8953b90aaa
Fix InstCombine/2004-04-04-InstCombineReplaceAllUsesWith.ll
...
llvm-svn: 12658
2004-04-05 02:10:19 +00:00
69193f93b6
Support getelementptr instructions which use uint's to index into structure
...
types and can have arbitrary 32- and 64-bit integer types indexing into
sequential types.
llvm-svn: 12653
2004-04-05 01:30:19 +00:00
e61b67d7d5
Rewrite the indvars pass to use the ScalarEvolution analysis.
...
This also implements some new features for the indvars pass, including
linear function test replacement, exit value substitution, and it works with
a much more general class of induction variables and loops.
llvm-svn: 12620
2004-04-02 20:24:31 +00:00
59fdf74968
Remove some assertions that are now bogus with the last patch I put in
...
llvm-svn: 12595
2004-04-01 19:21:46 +00:00
146d0df5e4
Fix PR306: Loop simplify incorrectly updates dominator information
...
Testcase: LoopSimplify/2004-04-01-IncorrectDomUpdate.ll
llvm-svn: 12592
2004-04-01 19:06:07 +00:00
61fab1409d
Add warning
...
llvm-svn: 12573
2004-03-31 22:00:30 +00:00
533bc49775
Implement select.ll:test[3-6]
...
llvm-svn: 12544
2004-03-30 19:37:13 +00:00
059f390257
Add a simple select instruction lowering pass
...
llvm-svn: 12540
2004-03-30 18:41:10 +00:00
56b5051428
X % -1 == X % 1 == 0
...
llvm-svn: 12520
2004-03-26 16:11:24 +00:00
57c67b06e9
Two changes:
...
#1 is to unconditionally strip constantpointerrefs out of
instruction operands where they are absolutely pointless and inhibit
optimization. GRRR!
#2 is to implement InstCombine/getelementptr_const.ll
llvm-svn: 12519
2004-03-25 22:59:29 +00:00
abb77c9959
Teach the optimizer to delete zero sized alloca's (but not mallocs!)
...
llvm-svn: 12507
2004-03-19 06:08:10 +00:00
684fa5ac64
Be more accurate
...
llvm-svn: 12464
2004-03-17 01:59:27 +00:00
a3783a577e
Fix bug in previous checkin
...
llvm-svn: 12458
2004-03-16 23:36:49 +00:00
95057f6ad1
Okay, so there is no reasonable way for tail duplication to update SSA form,
...
as it is making effectively arbitrary modifications to the CFG and we don't
have a domset/domfrontier implementations that can handle the dynamic updates.
Instead of having a bunch of code that doesn't actually work in practice,
just demote any potentially tricky values to the stack (causing the problem
to go away entirely). Later invocations of mem2reg will rebuild SSA for us.
This fixes all of the major performance regressions with tail duplication
from LLVM 1.1. For example, this loop:
---
int popcount(int x) {
int result = 0;
while (x != 0) {
result = result + (x & 0x1);
x = x >> 1;
}
return result;
}
---
Used to be compiled into:
int %popcount(int %X) {
entry:
br label %loopentry
loopentry: ; preds = %entry, %no_exit
%x.0 = phi int [ %X, %entry ], [ %tmp.9, %no_exit ] ; <int> [#uses=3]
%result.1.0 = phi int [ 0, %entry ], [ %tmp.6, %no_exit ] ; <int> [#uses=2]
%tmp.1 = seteq int %x.0, 0 ; <bool> [#uses=1]
br bool %tmp.1, label %loopexit, label %no_exit
no_exit: ; preds = %loopentry
%tmp.4 = and int %x.0, 1 ; <int> [#uses=1]
%tmp.6 = add int %tmp.4, %result.1.0 ; <int> [#uses=1]
%tmp.9 = shr int %x.0, ubyte 1 ; <int> [#uses=1]
br label %loopentry
loopexit: ; preds = %loopentry
ret int %result.1.0
}
And is now compiled into:
int %popcount(int %X) {
entry:
br label %no_exit
no_exit: ; preds = %entry, %no_exit
%x.0.0 = phi int [ %X, %entry ], [ %tmp.9, %no_exit ] ; <int> [#uses=2]
%result.1.0.0 = phi int [ 0, %entry ], [ %tmp.6, %no_exit ] ; <int> [#uses=1]
%tmp.4 = and int %x.0.0, 1 ; <int> [#uses=1]
%tmp.6 = add int %tmp.4, %result.1.0.0 ; <int> [#uses=2]
%tmp.9 = shr int %x.0.0, ubyte 1 ; <int> [#uses=2]
%tmp.1 = seteq int %tmp.9, 0 ; <bool> [#uses=1]
br bool %tmp.1, label %loopexit, label %no_exit
loopexit: ; preds = %no_exit
ret int %tmp.6
}
llvm-svn: 12457
2004-03-16 23:29:09 +00:00
7a7b114871
Do not try to optimize PHI nodes with incredibly high degree. This reduces SCCP
...
time from 615s to 1.49s on a large testcase that has a gigantic switch statement
that all of the blocks in the function go to (an intepreter).
llvm-svn: 12442
2004-03-16 19:49:59 +00:00
a64923ad26
Do not copy gigantic switch instructions
...
llvm-svn: 12441
2004-03-16 19:45:22 +00:00
db5b8f4d6b
Fix a regression from this patch:
...
http://mail.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20040308/013095.html
Basically, this patch only updated the immediate dominatees of the header node
to tell them that the preheader also dominated them. In practice, ALL
dominatees of the header node are also dominated by the preheader.
This fixes: LoopSimplify/2004-03-15-IncorrectDomUpdate.
and PR293
llvm-svn: 12434
2004-03-16 06:00:15 +00:00
cd83282df1
Add counters for the number of calls elimianted
...
llvm-svn: 12420
2004-03-15 05:46:59 +00:00
Chris Lattner