Chris Lattner
32eae5daa5
Fix PR1907, a nasty miscompilation because instcombine didn't
...
realize that ne & sgt was a signed comparison (it was only
looking at whether the left compare was signed).
llvm-svn: 45937
2008-01-13 20:59:02 +00:00
Duncan Sands
7414cc131b
When turning a call to a bitcast function into a direct call,
...
if this becomes a varargs call then deal correctly with any
parameter attributes on the newly vararg call arguments.
llvm-svn: 45931
2008-01-13 08:02:44 +00:00
Chris Lattner
67f581b344
Implement PR1795, an instcombine hack for forming GEPs with integer pointer arithmetic.
...
llvm-svn: 45745
2008-01-08 07:23:51 +00:00
Duncan Sands
7955cf0cd7
Small cleanup for handling of type/parameter attribute
...
incompatibility.
llvm-svn: 45704
2008-01-07 17:16:06 +00:00
Gordon Henriksen
f0803127c6
Deleting an empty file. Thanks, /usr/bin/patch!
...
llvm-svn: 45675
2008-01-07 02:29:04 +00:00
Gordon Henriksen
db4f51e1b9
With this patch, the LowerGC transformation becomes the
...
ShadowStackCollector, which additionally has reduced overhead with
no sacrifice in portability.
Considering a function @fun with 8 loop-local roots,
ShadowStackCollector introduces the following overhead
(x86):
; shadowstack prologue
movl L_llvm_gc_root_chain$non_lazy_ptr, %eax
movl (%eax), %ecx
movl $___gc_fun, 20(%esp)
movl $0, 24(%esp)
movl $0, 28(%esp)
movl $0, 32(%esp)
movl $0, 36(%esp)
movl $0, 40(%esp)
movl $0, 44(%esp)
movl $0, 48(%esp)
movl $0, 52(%esp)
movl %ecx, 16(%esp)
leal 16(%esp), %ecx
movl %ecx, (%eax)
; shadowstack loop overhead
(none)
; shadowstack epilogue
movl 48(%esp), %edx
movl %edx, (%ecx)
; shadowstack metadata
.align 3
___gc_fun: # __gc_fun
.long 8
.space 4
In comparison to LowerGC:
; lowergc prologue
movl L_llvm_gc_root_chain$non_lazy_ptr, %eax
movl (%eax), %ecx
movl %ecx, 48(%esp)
movl $8, 52(%esp)
movl $0, 60(%esp)
movl $0, 56(%esp)
movl $0, 68(%esp)
movl $0, 64(%esp)
movl $0, 76(%esp)
movl $0, 72(%esp)
movl $0, 84(%esp)
movl $0, 80(%esp)
movl $0, 92(%esp)
movl $0, 88(%esp)
movl $0, 100(%esp)
movl $0, 96(%esp)
movl $0, 108(%esp)
movl $0, 104(%esp)
movl $0, 116(%esp)
movl $0, 112(%esp)
; lowergc loop overhead
leal 44(%esp), %eax
movl %eax, 56(%esp)
leal 40(%esp), %eax
movl %eax, 64(%esp)
leal 36(%esp), %eax
movl %eax, 72(%esp)
leal 32(%esp), %eax
movl %eax, 80(%esp)
leal 28(%esp), %eax
movl %eax, 88(%esp)
leal 24(%esp), %eax
movl %eax, 96(%esp)
leal 20(%esp), %eax
movl %eax, 104(%esp)
leal 16(%esp), %eax
movl %eax, 112(%esp)
; lowergc epilogue
movl 48(%esp), %edx
movl %edx, (%ecx)
; lowergc metadata
(none)
llvm-svn: 45670
2008-01-07 01:30:53 +00:00
Duncan Sands
fd975e4b3d
The transform that tries to turn calls to bitcast functions into
...
direct calls bails out unless caller and callee have essentially
equivalent parameter attributes. This is illogical - the callee's
attributes should be of no relevance here. Rework the logic, which
incidentally fixes a crash when removed arguments have attributes.
llvm-svn: 45658
2008-01-06 18:27:01 +00:00
Duncan Sands
b8489f09a2
When transforming a call to a bitcast function into
...
a direct call with cast parameters and cast return
value (if any), instcombine was prepared to cast any
non-void return value into any other, whether castable
or not. Add a new predicate for testing whether casting
is valid, and check it both for the return value and
(as a cleanup) for the parameters.
llvm-svn: 45657
2008-01-06 10:12:28 +00:00
Chris Lattner
7e1c3aa702
remove a couple more unsafe xforms in the face of overflow.
...
llvm-svn: 45613
2008-01-05 01:22:42 +00:00
Chris Lattner
983697dfac
remove the (x-y) < 0 comparison xform, it miscompiles
...
things that are not equality comparisons, for example:
(2147479553+4096)-2147479553 < 0 != (2147479553+4096) < 2147479553
llvm-svn: 45612
2008-01-05 01:18:20 +00:00
Wojciech Matyjewicz
9ec15f974f
fix typo
...
llvm-svn: 45594
2008-01-04 20:02:18 +00:00
Chris Lattner
ad9a6ccb83
Remove attribution from file headers, per discussion on llvmdev.
...
llvm-svn: 45418
2007-12-29 20:36:04 +00:00
Chris Lattner
8193d4af33
remove attribution from lib Makefiles.
...
llvm-svn: 45415
2007-12-29 20:09:26 +00:00
Christopher Lamb
dfad5f19b4
Disable null pointer folding transforms for non-generic address spaces. This should probably be a target-specific predicate based on address space. That way for targets where this isn't applicable the predicate can be optimized away.
...
llvm-svn: 45403
2007-12-29 07:56:53 +00:00
Owen Anderson
ebd3e9c500
Repair a transform that Chris noticed a bug in. Thanks to Nicholas for pointing out my stupid mistakes when writing this patch. :-)
...
llvm-svn: 45384
2007-12-28 07:42:12 +00:00
Chris Lattner
2456399ce5
disable this instcombine xform, it miscompiles:
...
define i32 @main() {
entry:
%z = alloca i32 ; <i32*> [#uses=2]
store i32 0, i32* %z
%tmp = load i32* %z ; <i32> [#uses=1]
%sub = sub i32 %tmp, 1 ; <i32> [#uses=1]
%cmp = icmp ult i32 %sub, 0 ; <i1> [#uses=1]
%retval = select i1 %cmp, i32 1, i32 0 ; <i32> [#uses=1]
ret i32 %retval
}
into ret 1, instead of ret 0.
Christopher, please investigate.
llvm-svn: 45383
2007-12-28 06:24:31 +00:00
Chris Lattner
90df7f7424
Don't break critical edges for single-bb loops, this helps with PR1877, though
...
it is only a partial fix. This change is noise for most programs, but
speeds up Shootout-C++/matrix by 20%, Ptrdist/ks by 24%, smg2000 by 8%,
hexxagon by 9%, bzip2 by 9% (not sure I trust this), ackerman by 13%, etc.
OTOH, it slows down Shootout/fib2 by 40% (I'll update PR1877 with this info).
llvm-svn: 45354
2007-12-25 19:06:45 +00:00
Chris Lattner
7e1e1f2933
add a -backedge-hack llc-beta option to codegenprepare.
...
When specified, don't split backedges of single-bb loops.
This helps address PR1877
llvm-svn: 45344
2007-12-24 19:32:55 +00:00
Chris Lattner
d64df490ca
implement InstCombine/shift-trunc-shift.ll. This allows
...
us to compile:
#include <math.h>
int t1(double d) { return signbit(d); }
into:
_t1:
movd %xmm0, %rax
shrq $63, %rax
ret
instead of:
_t1:
movd %xmm0, %rax
shrq $32, %rax
shrl $31, %eax
ret
on x86-64.
llvm-svn: 45311
2007-12-22 09:07:47 +00:00
Christopher Lamb
7ca648a7b1
Implement review feedback, including additional transforms
...
(icmp slt (sub A B) 1) -> (icmp sle A B)
icmp sgt (sub A B) -1) -> (icmp sge A B)
and add testcase.
llvm-svn: 45256
2007-12-20 07:21:11 +00:00
Evan Cheng
eb07401701
Clean up previous patch: PHI uses should not prevent iv reuse if all other uses are addresses. This trades a constant multiply for one fewer iv.
...
llvm-svn: 45251
2007-12-20 02:20:53 +00:00
Chris Lattner
1a386cbdae
simplify this code with the new m_Zero() pattern. Make sure the select only
...
has a single use, and generalize it to not require N to be a constant.
llvm-svn: 45250
2007-12-20 01:56:58 +00:00
Evan Cheng
cf7b6b419a
Allow iv reuse if the user is a PHI node which is in turn used as addresses.
...
llvm-svn: 45230
2007-12-19 23:33:23 +00:00
Duncan Sands
56f3add5b7
When inlining through an 'nounwind' call, mark inlined
...
calls 'nounwind'. It is important for correct C++
exception handling that nounwind markings do not get
lost, so this transformation is actually needed for
correctness.
llvm-svn: 45218
2007-12-19 21:13:37 +00:00
Christopher Lamb
be0cbc7e92
Fold subtracts into integer compares vs. zero. This improves generate code for this case on X86
...
from
_foo:
movl $99, %ecx
movl 4(%esp), %eax
subl %eax, %ecx
xorl %edx, %edx
testl %ecx, %ecx
cmovs %edx, %eax
ret
to
_foo:
xorl %ecx, %ecx
movl 4(%esp), %eax
cmpl $99, %eax
cmovg %ecx, %eax
ret
llvm-svn: 45173
2007-12-18 21:32:20 +00:00
Christopher Lamb
051a1320e8
Fix comments
...
llvm-svn: 45170
2007-12-18 20:33:11 +00:00
Christopher Lamb
d56318b885
Remove an orthogonal transformation of the selection condition from my most recent submission.
...
llvm-svn: 45169
2007-12-18 20:30:28 +00:00
Duncan Sands
242f80be86
Rename isNoReturn to doesNotReturn, and isNoUnwind to
...
doesNotThrow.
llvm-svn: 45160
2007-12-18 09:59:50 +00:00
Christopher Lamb
437b4d229e
Fix typos.
...
llvm-svn: 45159
2007-12-18 09:45:40 +00:00
Christopher Lamb
aeb76743dc
Fold certain additions through selects (and their compares) so as to eliminate subtractions. This code is often produced by the SMAX expansion in SCEV.
...
This implements test/Transforms/InstCombine/2007-12-18-AddSelCmpSub.ll
llvm-svn: 45158
2007-12-18 09:34:41 +00:00
David Greene
3f3a521a33
Get rid of annoying spaces.
...
llvm-svn: 45100
2007-12-17 17:40:29 +00:00
David Greene
8fda00ca05
Fix GLIBCXX_DEBUG errors. Erase invalidates std::vector iterators
...
passed the erased element.
llvm-svn: 45099
2007-12-17 17:39:51 +00:00
Christopher Lamb
a608afb52e
Change the PointerType api for creating pointer types. The old functionality of PointerType::get() has become PointerType::getUnqual(), which returns a pointer in the generic address space. The new prototype of PointerType::get() requires both a type and an address space.
...
llvm-svn: 45082
2007-12-17 01:12:55 +00:00
Duncan Sands
bf62f62058
Make instcombine promote inline asm calls to 'nounwind'
...
calls. Remove special casing of inline asm from the
inliner. There is a potential problem: the verifier
rejects invokes of inline asm (not sure why). If an
asm call is not marked "nounwind" in some .ll, and
instcombine is not run, but the inliner is run, then
an illegal module will be created. This is bad but
I'm not sure what the best approach is. I'm tempted
to remove the check in the verifier...
llvm-svn: 45073
2007-12-16 15:51:49 +00:00
Evan Cheng
4acb4eb95e
Fix typo.
...
llvm-svn: 44997
2007-12-13 07:50:36 +00:00
Evan Cheng
a152909956
Be extra careful with extension use optimation. Now turned on by default.
...
llvm-svn: 44981
2007-12-13 03:32:53 +00:00
Wojciech Matyjewicz
8bb1d9e67c
1. "Upgrage" comments.
...
2. Using zero-extended value of Scale and unsigned division is safe provided
that Scale doesn't have the sign bit set.
Previously these 2 instructions:
%p = bitcast [100 x {i8,i8,i8}]* %x to i8*
%q = getelementptr i8* %p, i32 -4
were combined into:
%q = getelementptr [100 x { i8, i8, i8 }]* %x, i32 0,
i32 1431655764, i32 0
what was incorrect.
llvm-svn: 44936
2007-12-12 15:21:32 +00:00
Evan Cheng
bb418c4551
Don't muck with phi nodes; bug fixes.
...
llvm-svn: 44905
2007-12-12 02:53:41 +00:00
Evan Cheng
d6ecb2e58d
Bug fix. Only safe to perform extension uses optimization if the source of extension is also defined in the same BB as the extension.
...
llvm-svn: 44896
2007-12-12 00:51:06 +00:00
Duncan Sands
b7ac459292
Make PruneEH update the nounwind/noreturn attributes
...
on functions as it calculates them.
llvm-svn: 44802
2007-12-10 19:09:40 +00:00
Owen Anderson
e3de18ac1d
Fix several cache coherence bugs in MemDep/GVN that were found. Also add some (disabled) debugging code
...
to make such problems easier to diagnose in the future, written by Duncan Sands.
llvm-svn: 44695
2007-12-08 01:37:09 +00:00
Chris Lattner
861df2f4e9
simplify some code.
...
llvm-svn: 44655
2007-12-06 06:25:04 +00:00
Chris Lattner
1f2a96f9c1
move some ashr-specific code out of commonShiftTransforms into visitAShr.
...
llvm-svn: 44650
2007-12-06 01:59:46 +00:00
Evan Cheng
9e69c0ada8
If both result of the {s|z}xt and its source are live out, rewrite all uses of the source with result of extension.
...
llvm-svn: 44643
2007-12-05 23:58:20 +00:00
Duncan Sands
1e2e4972ff
Rather than having special rules like "intrinsics cannot
...
throw exceptions", just mark intrinsics with the nounwind
attribute. Likewise, mark intrinsics as readnone/readonly
and get rid of special aliasing logic (which didn't use
anything more than this anyway).
llvm-svn: 44544
2007-12-03 20:06:50 +00:00
Chris Lattner
0c41f32943
update file comment.
...
llvm-svn: 44543
2007-12-03 19:43:18 +00:00
Devang Patel
c47f68e747
If ExitValue operand is also defined in Loop header then
...
insert new ExitValue after this operand definition.
This fixes PR1828.
llvm-svn: 44539
2007-12-03 19:17:21 +00:00
Duncan Sands
14f11d6836
Integrate the readonly/readnone logic more deeply
...
into alias analysis. This meant updating the API
which now has versions of the getModRefBehavior,
doesNotAccessMemory and onlyReadsMemory methods
which take a callsite parameter. These should be
used unless the callsite is not known, since in
general they can do a better job than the versions
that take a function. Also, users should no longer
call the version of getModRefBehavior that takes
both a function and a callsite. To reduce the
chance of misuse it is now protected.
llvm-svn: 44487
2007-12-01 07:51:45 +00:00
Owen Anderson
3b893e1f84
Fix a miscompilation in spiff on PPC.
...
llvm-svn: 44437
2007-11-29 18:02:22 +00:00
Duncan Sands
1b0feb42e2
Add some convenience methods for querying attributes, and
...
use them.
llvm-svn: 44403
2007-11-28 17:07:01 +00:00
Duncan Sands
3602011bec
Fix PR1146: parameter attributes are longer part of
...
the function type, instead they belong to functions
and function calls. This is an updated and slightly
corrected version of Reid Spencer's original patch.
The only known problem is that auto-upgrading of
bitcode files doesn't seem to work properly (see
test/Bitcode/AutoUpgradeIntrinsics.ll). Hopefully
a bitcode guru (who might that be? :) ) will fix it.
llvm-svn: 44359
2007-11-27 13:23:08 +00:00
Owen Anderson
43d4a82d4b
Make LoopInfoBase more generic, in preparation for having MachineLoopInfo. This involves a small interface change.
...
llvm-svn: 44348
2007-11-27 03:43:35 +00:00
Owen Anderson
2aa0218a6c
Fix another bug that was causing siod to fail.
...
llvm-svn: 44325
2007-11-26 07:17:19 +00:00
Owen Anderson
05275b7b14
Allow GVN to eliminate read-only function calls when it can detect that they are redundant.
...
llvm-svn: 44323
2007-11-26 02:26:36 +00:00
Anton Korobeynikov
14246a2eba
Remove another leak. Due to some reason AliasSetTracker didn't had any dtor...
...
llvm-svn: 44320
2007-11-25 23:52:02 +00:00
Chris Lattner
d1e03b5387
Implement PR1822
...
llvm-svn: 44318
2007-11-25 21:27:53 +00:00
Duncan Sands
114968a3e8
Fix PR1816. If a bitcast of a function only exists because of a
...
trivial difference in function attributes, allow calls to it to
be converted to direct calls. Based on a patch by Török Edwin.
While there, move the various lists of mutually incompatible
parameters etc out of the verifier and into ParameterAttributes.h.
llvm-svn: 44315
2007-11-25 14:10:56 +00:00
Chris Lattner
642ae99085
add a comment.
...
llvm-svn: 44293
2007-11-23 22:35:18 +00:00
Duncan Sands
0973ae4c8c
Remove some logic I thoughtlessly copied over
...
from the old ADCE implementation (there it was
correct because the transform was being done
for read-only functions).
llvm-svn: 44287
2007-11-23 09:10:17 +00:00
Chris Lattner
bff48b8f0d
Fix PR1817.
...
llvm-svn: 44284
2007-11-22 23:47:13 +00:00
Duncan Sands
e79932aed4
Turn invokes of nounwind functions into ordinary calls.
...
llvm-svn: 44280
2007-11-22 22:24:59 +00:00
Duncan Sands
6ec98952e0
Readonly/readnone functions are allowed to throw
...
exceptions, so don't turn invokes of them into
calls.
llvm-svn: 44278
2007-11-22 21:40:06 +00:00
Nick Lewycky
a734aa84f3
typo
...
llvm-svn: 44262
2007-11-21 05:21:54 +00:00
Dan Gohman
760d574313
Add explicit keywords.
...
llvm-svn: 44234
2007-11-19 15:30:20 +00:00
Dale Johannesen
c5032e5366
Remove indeterminism from a loop. We think this will
...
fix an occasional nonrepeatable bootstrap failure we've
been seeing on Darwin.
llvm-svn: 44202
2007-11-17 02:48:01 +00:00
Chris Lattner
5574ba5ce6
Fix PR1800 by correcting mistaken logic.
...
llvm-svn: 44188
2007-11-16 06:04:17 +00:00
Chris Lattner
3b66875602
Implement PR1796 and Transforms/SimplifyCFG/noreturn-call.ll
...
by inserting unreachable after no-return calls.
llvm-svn: 44099
2007-11-14 06:19:25 +00:00
Chris Lattner
2d15a52df0
Implement PR1786 by iterating between dead cycle elimination
...
and simplifycfg in the rare cases when it is needed.
llvm-svn: 44044
2007-11-13 07:32:38 +00:00
Andrew Lenharth
83adca5075
Better check
...
llvm-svn: 43897
2007-11-08 18:45:15 +00:00
Andrew Lenharth
310a65171d
Fix PR1780
...
llvm-svn: 43893
2007-11-08 17:39:28 +00:00
Chris Lattner
8e982efdc5
fix const correctness, BB is const, so its predecessors are too
...
llvm-svn: 43780
2007-11-06 22:07:40 +00:00
Chris Lattner
2a909d32b6
don't put erase or query for non-allocainst pointers in an set of allocainsts*'s
...
llvm-svn: 43779
2007-11-06 22:07:22 +00:00
Chris Lattner
907b9b92fe
Implement PR1777 by detecting dependent phis that
...
all compute the same value.
llvm-svn: 43777
2007-11-06 21:52:06 +00:00
Duncan Sands
59b08debe3
At the point of calculating the shift amount, the
...
type of SV has changed from what it originally was.
However we need the store width of the original.
llvm-svn: 43775
2007-11-06 20:39:11 +00:00
Chris Lattner
e194944b29
wrap long lines
...
llvm-svn: 43745
2007-11-06 01:15:27 +00:00
Dan Gohman
adc21ed938
Fix an abort in instcombine when folding creates a vector rem instruction.
...
llvm-svn: 43743
2007-11-05 23:16:33 +00:00
Devang Patel
963e54c2cf
If a value is incoming from outside the loop then the value does not need remapping and the value is never tracked through LastValueMap.
...
llvm-svn: 43728
2007-11-05 19:32:30 +00:00
Duncan Sands
7c0f410b42
If a long double is in a packed struct, it may be
...
that there is no padding.
llvm-svn: 43691
2007-11-05 00:35:07 +00:00
Gordon Henriksen
4d157a1bc6
Finishing initial docs for all transformations in Passes.html.
...
Also cleaned up some comments in source files.
llvm-svn: 43674
2007-11-04 16:15:04 +00:00
Duncan Sands
662fb070a7
Change uses of getTypeSize to getABITypeSize, getTypeStoreSize
...
or getTypeSizeInBits as appropriate in ScalarReplAggregates.
The right change to make was not always obvious, so it would
be good to have an sroa guru review this. While there I noticed
some bugs, and fixed them: (1) arrays of x86 long double have
holes due to alignment padding, but this wasn't being spotted
by HasStructPadding (renamed to HasPadding). The same goes
for arrays of oddly sized ints. Vectors also suffer from this,
in fact the problem for vectors is much worse because basic
vector assumptions seem to be broken by vectors of type with
alignment padding. I didn't try to fix any of these vector
problems. (2) The code for extracting smaller integers from
larger ones (in the "int union" case) was wrong on big-endian
machines for integers with size not a multiple of 8, like i1.
Probably this is impossible to hit via llvm-gcc, but I fixed
it anyway while there and added a testcase. I also got rid of
some trailing whitespace and changed a function name which
had an obvious typo in it.
llvm-svn: 43672
2007-11-04 14:43:57 +00:00
Chris Lattner
493f83eeb1
Disable tail duplication of call instructions. The cost
...
metric is way off for these in general, and this works around
buggy code like that in PR1764. we'll see if there is a big
performance impact of this. If so, I'll revert it tomorrow.
llvm-svn: 43668
2007-11-04 06:37:55 +00:00
Duncan Sands
eb464e976f
Executive summary: getTypeSize -> getTypeStoreSize / getABITypeSize.
...
The meaning of getTypeSize was not clear - clarifying it is important
now that we have x86 long double and arbitrary precision integers.
The issue with long double is that it requires 80 bits, and this is
not a multiple of its alignment. This gives a primitive type for
which getTypeSize differed from getABITypeSize. For arbitrary precision
integers it is even worse: there is the minimum number of bits needed to
hold the type (eg: 36 for an i36), the maximum number of bits that will
be overwriten when storing the type (40 bits for i36) and the ABI size
(i.e. the storage size rounded up to a multiple of the alignment; 64 bits
for i36).
This patch removes getTypeSize (not really - it is still there but
deprecated to allow for a gradual transition). Instead there is:
(1) getTypeSizeInBits - a number of bits that suffices to hold all
values of the type. For a primitive type, this is the minimum number
of bits. For an i36 this is 36 bits. For x86 long double it is 80.
This corresponds to gcc's TYPE_PRECISION.
(2) getTypeStoreSizeInBits - the maximum number of bits that is
written when storing the type (or read when reading it). For an
i36 this is 40 bits, for an x86 long double it is 80 bits. This
is the size alias analysis is interested in (getTypeStoreSize
returns the number of bytes). There doesn't seem to be anything
corresponding to this in gcc.
(3) getABITypeSizeInBits - this is getTypeStoreSizeInBits rounded
up to a multiple of the alignment. For an i36 this is 64, for an
x86 long double this is 96 or 128 depending on the OS. This is the
spacing between consecutive elements when you form an array out of
this type (getABITypeSize returns the number of bytes). This is
TYPE_SIZE in gcc.
Since successive elements in a SequentialType (arrays, pointers
and vectors) need to be aligned, the spacing between them will be
given by getABITypeSize. This means that the size of an array
is the length times the getABITypeSize. It also means that GEP
computations need to use getABITypeSize when computing offsets.
Furthermore, if an alloca allocates several elements at once then
these too need to be aligned, so the size of the alloca has to be
the number of elements multiplied by getABITypeSize. Logically
speaking this doesn't have to be the case when allocating just
one element, but it is simpler to also use getABITypeSize in this
case. So alloca's and mallocs should use getABITypeSize. Finally,
since gcc's only notion of size is that given by getABITypeSize, if
you want to output assembler etc the same as gcc then getABITypeSize
is the size you want.
Since a store will overwrite no more than getTypeStoreSize bytes,
and a read will read no more than that many bytes, this is the
notion of size appropriate for alias analysis calculations.
In this patch I have corrected all type size uses except some of
those in ScalarReplAggregates, lib/Codegen, lib/Target (the hard
cases). I will get around to auditing these too at some point,
but I could do with some help.
Finally, I made one change which I think wise but others might
consider pointless and suboptimal: in an unpacked struct the
amount of space allocated for a field is now given by the ABI
size rather than getTypeStoreSize. I did this because every
other place that reserves memory for a type (eg: alloca) now
uses getABITypeSize, and I didn't want to make an exception
for unpacked structs, i.e. I did it to make things more uniform.
This only effects structs containing long doubles and arbitrary
precision integers. If someone wants to pack these types more
tightly they can always use a packed struct.
llvm-svn: 43620
2007-11-01 20:53:16 +00:00
Owen Anderson
594b0fe9e2
Fix test/Transforms/DeadStoreElimination/PartialStore.ll, which had been
...
silently failing because of an incorrect run line for some time.
llvm-svn: 43605
2007-11-01 05:29:16 +00:00
Chris Lattner
d29624e11a
Fix InstCombine/2007-10-31-RangeCrash.ll
...
llvm-svn: 43596
2007-11-01 02:18:41 +00:00
Dan Gohman
51cadc3b59
Fix a typo in a comment.
...
llvm-svn: 43553
2007-10-31 14:35:39 +00:00
Evan Cheng
06ec64fcda
At end of LSR, replace uses of now constant (as result of SplitCriticalEdge) PHI node with the constant value.
...
llvm-svn: 43533
2007-10-30 23:45:15 +00:00
Evan Cheng
5e058e94b5
It's not safe to tell SplitCriticalEdge to merge identical edges. It may delete the phi instruction that's being processed.
...
llvm-svn: 43524
2007-10-30 22:27:26 +00:00
Evan Cheng
633cd3e84d
- Bug fixes.
...
- Allow icmp rewrite using an iv / stride of a smaller integer type.
llvm-svn: 43480
2007-10-29 22:07:18 +00:00
Dan Gohman
1c499173c8
Don't bitcast from pointer-to-vector to pointer-to-array when
...
lowering load and store instructions.
llvm-svn: 43468
2007-10-29 20:34:35 +00:00
Dan Gohman
8f74d7f5c0
Use an array instead of a fixed-length std::vector.
...
llvm-svn: 43467
2007-10-29 20:24:00 +00:00
Dan Gohman
0e9e5b6534
Do a real assert if there is an unhandled vector instruction instead
...
of just printing to cerr.
llvm-svn: 43466
2007-10-29 20:14:29 +00:00
Dan Gohman
a309c59972
Update a comment to reflect the current code.
...
llvm-svn: 43463
2007-10-29 19:32:39 +00:00
Dan Gohman
459bb8cbd2
Remove an unused function argument.
...
llvm-svn: 43462
2007-10-29 19:31:25 +00:00
Dan Gohman
30b8bd2ad4
Fix a typo in a comment.
...
llvm-svn: 43461
2007-10-29 19:26:14 +00:00
Dan Gohman
957fd1704a
Avoid calling ValidStride when not all uses are addresses.
...
llvm-svn: 43460
2007-10-29 19:23:53 +00:00
Evan Cheng
c8adcda731
A number of LSR fixes:
...
- ChangeCompareStride only reuse stride that is larger than current stride. It
will let the general reuse mechanism to try to reuse a smaller stride.
- Watch out for multiplication overflow in ChangeCompareStride.
- Replace std::set with SmallPtrSet.
llvm-svn: 43408
2007-10-26 23:08:19 +00:00
Evan Cheng
53b2e7f3ca
Fix a crash. Make sure TLI is not null.
...
llvm-svn: 43384
2007-10-26 17:24:46 +00:00
Evan Cheng
53696b7e9f
Loosen up iv reuse to allow reuse of the same stride but a larger type when truncating from the larger type to smaller type is free.
...
e.g.
Turns this loop:
LBB1_1: # entry.bb_crit_edge
xorl %ecx, %ecx
xorw %dx, %dx
movw %dx, %si
LBB1_2: # bb
movl L_X$non_lazy_ptr, %edi
movw %si, (%edi)
movl L_Y$non_lazy_ptr, %edi
movw %dx, (%edi)
addw $4, %dx
incw %si
incl %ecx
cmpl %eax, %ecx
jne LBB1_2 # bb
into
LBB1_1: # entry.bb_crit_edge
xorl %ecx, %ecx
xorw %dx, %dx
LBB1_2: # bb
movl L_X$non_lazy_ptr, %esi
movw %cx, (%esi)
movl L_Y$non_lazy_ptr, %esi
movw %dx, (%esi)
addw $4, %dx
incl %ecx
cmpl %eax, %ecx
jne LBB1_2 # bb
llvm-svn: 43375
2007-10-26 01:56:11 +00:00
Evan Cheng
d7eab3a984
Do not rewrite compare instruction using iv of a different stride if the new
...
stride may be rewritten using the stride of the compare instruction.
llvm-svn: 43367
2007-10-25 22:45:20 +00:00
Evan Cheng
c25c4276a6
Remove code that's commented out.
...
llvm-svn: 43356
2007-10-25 18:38:24 +00:00
Evan Cheng
66cbf54030
If a loop termination compare instruction is the only use of its stride,
...
and the compaison is against a constant value, try eliminate the stride
by moving the compare instruction to another stride and change its
constant operand accordingly. e.g.
loop:
...
v1 = v1 + 3
v2 = v2 + 1
if (v2 < 10) goto loop
=>
loop:
...
v1 = v1 + 3
if (v1 < 30) goto loop
llvm-svn: 43336
2007-10-25 09:11:16 +00:00
Chris Lattner
ae9cfd2fb0
simplify some code by using the new isNaN predicate
...
llvm-svn: 43305
2007-10-24 18:54:45 +00:00
Chris Lattner
483c471daa
Implement a couple of foldings for ordered and unordered comparisons,
...
implementing cases related to PR1738.
llvm-svn: 43289
2007-10-24 05:38:08 +00:00
Dan Gohman
df1f166e4a
Strength reduction improvements.
...
- Avoid attempting stride-reuse in the case that there are users that
aren't addresses. In that case, there will be places where the
multiplications won't be folded away, so it's better to try to
strength-reduce them.
- Several SSE intrinsics have operands that strength-reduction can
treat as addresses. The previous item makes this more visible, as
any non-address use of an IV can inhibit stride-reuse.
- Make ValidStride aware of whether there's likely to be a base
register in the address computation. This prevents it from thinking
that things like stride 9 are valid on x86 when the base register is
already occupied.
Also, XFAIL the 2007-08-10-LEA16Use32.ll test; the new logic to avoid
stride-reuse elimintes the LEA in the loop, so the test is no longer
testing what it was intended to test.
llvm-svn: 43231
2007-10-22 20:40:42 +00:00
Dan Gohman
68fc6d7395
Move the SCEV object factors from being static members of the individual
...
SCEV subclasses to being non-static member functions of the ScalarEvolution
class.
llvm-svn: 43224
2007-10-22 18:31:58 +00:00
Anton Korobeynikov
bcee4726bf
Reg2Mem cleanup and optimizations:
...
- enable phi instructions demotion to stack
- create alloca instructions in the entry block
llvm-svn: 43208
2007-10-21 23:05:16 +00:00
Devang Patel
eff4619cc8
Try again.
...
Instead of loading small global string from memory, use
integer constant.
llvm-svn: 43148
2007-10-18 19:52:32 +00:00
Owen Anderson
f0e040a0c7
Allow GVN to eliminate redundant calls to functions without side effects.
...
llvm-svn: 43147
2007-10-18 19:39:33 +00:00
Evan Cheng
1c34d807ce
Reverting r43070 for now. It's causing llc test failures.
...
llvm-svn: 43103
2007-10-17 23:51:13 +00:00
Devang Patel
cf2f9d6daa
Apply "Instead of loading small c string constant, use integer constant directly" transformation while processing load instruction.
...
llvm-svn: 43070
2007-10-17 07:24:40 +00:00
Devang Patel
c3d0477a0e
Use immediate stores.
...
llvm-svn: 43055
2007-10-16 23:44:18 +00:00
Devang Patel
7d1d5d6bf6
Achieve same result but use fewer lines of code.
...
llvm-svn: 42985
2007-10-15 15:31:35 +00:00
Devang Patel
f65c028dad
Dest type is always i8 *. This allows some simplification.
...
Do not filter memmove.
llvm-svn: 42930
2007-10-12 20:10:21 +00:00
Chris Lattner
3af877f26a
Fix a bug in my patch last night that broke InstCombine/2007-10-12-Crash.ll
...
llvm-svn: 42920
2007-10-12 18:05:47 +00:00
Gabor Greif
cbfb655705
eliminate warning
...
llvm-svn: 42892
2007-10-12 07:44:54 +00:00
Chris Lattner
3c23c37233
Fix some 80 column violations.
...
Fix DecomposeSimpleLinearExpr to handle simple constants better.
Don't nuke gep(bitcast(allocation)) if the bitcast(allocation) will
fold the allocation. This fixes PR1728 and Instcombine/malloc3.ll
llvm-svn: 42891
2007-10-12 05:30:59 +00:00
Devang Patel
15d6257fa8
Lower memcpy if it makes sense.
...
llvm-svn: 42864
2007-10-11 17:21:57 +00:00
Devang Patel
b13057acf6
Do not walk invalid iterator.
...
llvm-svn: 42812
2007-10-09 21:31:36 +00:00
Devang Patel
36b68478cb
Fix bug in updating dominance frontier after loop
...
unswitch when frontier includes basic blocks that
are not inside loop.
llvm-svn: 42654
2007-10-05 22:29:34 +00:00
Devang Patel
5efcf79cf3
Fix 80 col violation.
...
llvm-svn: 42591
2007-10-03 21:17:43 +00:00
Devang Patel
1e2cced8b9
Refactor code in a separate method.
...
llvm-svn: 42590
2007-10-03 21:16:08 +00:00
Dan Gohman
30ba45b569
Use empty() member functions when that's what's being tested for instead
...
of comparing begin() and end().
llvm-svn: 42585
2007-10-03 19:26:29 +00:00
Dale Johannesen
529cc16893
Tone down an overzealous optimization.
...
llvm-svn: 42582
2007-10-03 17:45:27 +00:00
Dale Johannesen
d94f00234f
Fix stride computations for long double arrays.
...
llvm-svn: 42508
2007-10-01 23:08:35 +00:00
Devang Patel
42f006a51a
Relax unsafe use check. If there is one unconditional use inside the loop then it is safe to promote value even if there is another conditional use inside the loop.
...
llvm-svn: 42493
2007-10-01 18:12:58 +00:00
Dale Johannesen
412575891e
Don't do SRA for unions with long double fields.
...
Fixes a SWB crash.
llvm-svn: 42422
2007-09-28 00:21:38 +00:00
Devang Patel
d98abb62ce
Handle multiple induction variables.
...
This fixes PR714.
llvm-svn: 42309
2007-09-25 18:24:48 +00:00
Devang Patel
ab58843813
Do not reserve DOM check for GetElementPtrInst.
...
llvm-svn: 42306
2007-09-25 17:55:50 +00:00
Devang Patel
f35e6c1181
doh..
...
llvm-svn: 42300
2007-09-25 17:43:08 +00:00
Devang Patel
de9d1c3654
Add transformation to update loop interation space. Now,
...
for (i=A; i<N; i++) {
if (i < X && i > Y)
do_something();
}
is transformed into
U=min(N,X); L=max(A,Y);
for (i=L;i<U;i++)
do_somethihg();
llvm-svn: 42299
2007-09-25 17:31:19 +00:00
Devang Patel
65f8d0c2d7
Do not promote null values because it may be unsafe to do so.
...
llvm-svn: 42270
2007-09-24 20:02:42 +00:00
Dan Gohman
ed361aa114
explicit keywords.
...
llvm-svn: 42262
2007-09-24 15:48:49 +00:00
Devang Patel
b741c56cfc
Fix PR1692
...
llvm-svn: 42209
2007-09-21 21:18:19 +00:00
Owen Anderson
9070cf6dea
Add partial caching of non-local memory dependence queries. This provides a modest
...
speedup for GVN.
llvm-svn: 42185
2007-09-21 03:53:52 +00:00
Devang Patel
632653ce8f
Update aux. info associated with an instruction before erasing instruction.
...
llvm-svn: 42180
2007-09-20 23:45:50 +00:00
Devang Patel
57b0b31257
Don't increment invalid iterator.
...
llvm-svn: 42178
2007-09-20 23:01:50 +00:00
Nick Lewycky
3baa4cde1c
Fix optimization. %x = sub %x, %y does not imply that %y is zero.
...
llvm-svn: 42157
2007-09-20 00:48:36 +00:00
Devang Patel
648d7a1074
Avoid unsafe promotion.
...
llvm-svn: 42149
2007-09-19 20:18:51 +00:00
Duncan Sands
f7abe75944
Improve comment.
...
llvm-svn: 42132
2007-09-19 10:25:38 +00:00
Duncan Sands
d88f60ed32
A global variable with external weak linkage can be null, while
...
an alias could alias such a global variable.
llvm-svn: 42130
2007-09-19 10:10:31 +00:00
Devang Patel
339bee9291
Relax loop ExitCondition predicate restriction.
...
llvm-svn: 42122
2007-09-19 00:28:47 +00:00
Devang Patel
2883f5cc94
Filter loops where split condition's false branch is not empty. For example
...
for (int i = 0; i < N; ++i) {
if (i == somevalue)
dosomething();
else
dosomethingelse();
}
llvm-svn: 42121
2007-09-19 00:15:16 +00:00
Devang Patel
03df2adc31
Bail out early, before modifying anything.
...
llvm-svn: 42120
2007-09-19 00:11:01 +00:00
Devang Patel
9806df698b
Work is incomplete. Loop is not modified at all right now.
...
llvm-svn: 42119
2007-09-19 00:08:13 +00:00
Devang Patel
4014965efb
Fix PR1657
...
llvm-svn: 42075
2007-09-18 01:54:42 +00:00
Devang Patel
578a1ae12a
Do not eliminate loop when it is invalid to do so. For example,
...
for(int i = 0; i < N; i++) {
if ( i == XYZ) {
A;
else
B;
}
C;
D;
}
llvm-svn: 42058
2007-09-17 21:01:05 +00:00
Devang Patel
14b9a01c02
Skeleton for transformations to truncate loop's iteration space.
...
llvm-svn: 42054
2007-09-17 20:39:48 +00:00
Chris Lattner
cc315726f7
Merge DenseMapKeyInfo & DenseMapValueInfo into DenseMapInfo
...
Add a new DenseMapInfo::isEqual method to allow clients to redefine
the equality predicate used when probing the hash table.
llvm-svn: 42042
2007-09-17 18:34:04 +00:00
Dan Gohman
2de5779a99
Instcombine x-((x/y)*y) into a remainder operator.
...
llvm-svn: 42035
2007-09-17 17:31:57 +00:00
Duncan Sands
901cb2662d
Factor the trampoline transformation into a subroutine.
...
llvm-svn: 42021
2007-09-17 10:26:40 +00:00
Owen Anderson
fd6ec5cf82
Be more careful when constant-folding PHI nodes.
...
llvm-svn: 41998
2007-09-16 08:04:16 +00:00
Owen Anderson
fe5adad9c4
Remove RLE. It is subsumed by GVN.
...
llvm-svn: 41968
2007-09-14 22:33:52 +00:00
Dale Johannesen
575bd6070a
Remove the assumption that FP's are either float or
...
double from some of the many places in the optimizers
it appears, and do something reasonable with x86
long double.
Make APInt::dump() public, remove newline, use it to
dump ConstantSDNode's.
Allow APFloats in FoldingSet.
Expand X86 backend handling of long doubles (conversions
to/from int, mostly).
llvm-svn: 41967
2007-09-14 22:26:36 +00:00
Chris Lattner
9c3cd36dd0
silence a bogus gcc warning.
...
llvm-svn: 41949
2007-09-14 03:07:24 +00:00
Bill Wendling
843ac5aebe
Temporary reverting r41817
...
(http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070910/053370.html ). It's
causing SPASS to fail.
llvm-svn: 41938
2007-09-14 01:13:55 +00:00
Chris Lattner
d58e0026ec
Change llvm.gcroot to not init the root to null at runtime, this prevents
...
using it for live-in values etc.
llvm-svn: 41879
2007-09-12 17:53:10 +00:00
Duncan Sands
c63fd15cd9
Turn calls to trampolines into calls to the underlying
...
nested function.
llvm-svn: 41844
2007-09-11 14:35:41 +00:00
Devang Patel
8f8fe2f0da
Avoid negative logic.
...
llvm-svn: 41829
2007-09-11 01:10:45 +00:00
Devang Patel
9bc265b94a
Refactor code into a separate method.
...
llvm-svn: 41826
2007-09-11 00:42:56 +00:00
Devang Patel
49ffe783eb
Clear split info object.
...
llvm-svn: 41823
2007-09-11 00:23:56 +00:00
Devang Patel
8394964f1a
Split condition does not have to be ICmpInst in all cases.
...
llvm-svn: 41822
2007-09-11 00:12:56 +00:00
Devang Patel
82374f7476
Check all terminators inside loop.
...
llvm-svn: 41821
2007-09-10 23:57:58 +00:00
Chris Lattner
3ace09794b
remove some dead code, this is handled by constant folding.
...
llvm-svn: 41819
2007-09-10 23:46:29 +00:00
Devang Patel
ce76c29be0
Swap exit condition operands if it works.
...
llvm-svn: 41817
2007-09-10 23:34:06 +00:00
Chris Lattner
ab8aa7d61a
Prevent tailcallelim from breaking "recursive" calls to builtins.
...
llvm-svn: 41804
2007-09-10 20:58:55 +00:00
Devang Patel
76790c1b28
Filter exit conditions which are not yet handled.
...
llvm-svn: 41800
2007-09-10 18:33:42 +00:00
Devang Patel
7a759cdfd7
Require SCEV before LCSSA.
...
llvm-svn: 41798
2007-09-10 18:08:23 +00:00
Chris Lattner
8e6c39d961
Don't zap back to back volatile load/stores
...
llvm-svn: 41759
2007-09-07 05:33:03 +00:00
Dale Johannesen
86f367a6b7
Next round of APFloat changes.
...
Use APFloat in UpgradeParser and AsmParser.
Change all references to ConstantFP to use the
APFloat interface rather than double. Remove
the ConstantFP double interfaces.
Use APFloat functions for constant folding arithmetic
and comparisons.
(There are still way too many places APFloat is
just a wrapper around host float/double, but we're
getting there.)
llvm-svn: 41747
2007-09-06 18:13:44 +00:00
Nick Lewycky
79e179ff1f
Use isTrueWhenEqual. Thanks Chris!
...
llvm-svn: 41741
2007-09-06 02:40:25 +00:00
Nick Lewycky
2f66503c0a
When the two operands of an icmp are equal, there are five possible predicates
...
that would make the icmp true. Fixes PR1637.
llvm-svn: 41740
2007-09-06 01:10:22 +00:00
Chuck Rose III
4f602f5eba
Forgot to obey 80 column rule. Fixing that.
...
llvm-svn: 41725
2007-09-05 20:36:41 +00:00
Chuck Rose III
a1061872a7
Added default parameters to GetElementPtrInstr constructor call. Visual Studio 2k5 was getting confused and was unable to compile it. Suspected compiler error.
...
llvm-svn: 41721
2007-09-05 16:54:38 +00:00
David Greene
8cda5af2e7
Update GEP constructors to use an iterator interface to fix
...
GLIBCXX_DEBUG issues.
llvm-svn: 41697
2007-09-04 15:46:09 +00:00
Evan Cheng
9c7cff8e62
Fix a gcroot lowering bug.
...
llvm-svn: 41668
2007-09-01 02:00:51 +00:00
Chris Lattner
73aa3d62dc
Cut off crazy computation. This helps PR1622 slightly.
...
llvm-svn: 41522
2007-08-28 04:23:55 +00:00
Devang Patel
c649f24384
Use simpler test to filter loops.
...
llvm-svn: 41516
2007-08-27 21:34:31 +00:00
David Greene
5b85021be8
Update InvokeInst to work like CallInst
...
llvm-svn: 41506
2007-08-27 19:04:21 +00:00
Owen Anderson
f660f11ec4
Don't DSe volatile stores.
...
llvm-svn: 41456
2007-08-26 21:14:47 +00:00
Devang Patel
cbfa564c1c
Move exit condition and exit branch from exiting block into loop header and dominator info. This avoid execution of dead iteration. Loop is already filter in the beginning such that this change is safe.
...
llvm-svn: 41394
2007-08-25 02:39:24 +00:00
Devang Patel
d961a55c5a
Constant split values needs upper bound and lower bound check, just like any other split value.
...
llvm-svn: 41389
2007-08-25 01:09:14 +00:00
Devang Patel
027410a7aa
While calculating upper loop bound for first loop and lower loop bound for second loop, take care of edge cases.
...
llvm-svn: 41387
2007-08-25 00:56:38 +00:00
Devang Patel
c46dc61548
Fix regression that I caused yesterday night while adding logic to select appropriate split condition branch.
...
llvm-svn: 41365
2007-08-24 19:32:26 +00:00
Devang Patel
0b351a64db
It is not safe to execute split condition's true branch first all the time. If split
...
condition predicate is GT or GE then execute false branch first.
llvm-svn: 41358
2007-08-24 06:17:19 +00:00
Devang Patel
59af13f0cf
Reject ICMP_NE as index split condition.
...
llvm-svn: 41357
2007-08-24 06:02:25 +00:00
Devang Patel
4b52632179
Tightenup loop filter.
...
llvm-svn: 41356
2007-08-24 05:36:56 +00:00
Devang Patel
8b1d0a1ad2
Remove incomplete cost analysis.
...
llvm-svn: 41354
2007-08-24 05:21:13 +00:00
Devang Patel
4ad7c57729
Remove dead code.
...
llvm-svn: 41295
2007-08-22 21:07:41 +00:00
Devang Patel
798ba45c39
Fix typo.
...
llvm-svn: 41292
2007-08-22 20:55:18 +00:00
Devang Patel
59b8967a74
Cosmetic change
...
"True Loop" and "False Loop" naming terminology to refer two loops
after loop cloning is confusing. Instead just use A_Loop and B_Loop.
llvm-svn: 41287
2007-08-22 19:33:29 +00:00
Devang Patel
797fbef078
Refactor loop condition check in a separate function.
...
llvm-svn: 41282
2007-08-22 18:27:01 +00:00
Devang Patel
7e6f9f4779
Fix thinko.
...
Starting value of second loop's induction variable can not be lower
then starting value of original loop's induction variable.
llvm-svn: 41280
2007-08-22 18:07:47 +00:00
Devang Patel
60f5256cd1
Rename bunch of variables.
...
llvm-svn: 41250
2007-08-21 21:12:02 +00:00
Devang Patel
4fe5509c7d
Preserve LCSSA.
...
llvm-svn: 41246
2007-08-21 19:47:46 +00:00
Devang Patel
f06e667e9c
Use SmallVector instead of std::vector.
...
llvm-svn: 41207
2007-08-21 00:31:24 +00:00
Devang Patel
b64f73f8f0
s/ExitBlock/ExitingBlock/g
...
llvm-svn: 41204
2007-08-20 23:51:18 +00:00
Devang Patel
b6ebf8cbf3
Replace indunction variable with split value in loop body.
...
This fixes art miscompile.
llvm-svn: 41195
2007-08-20 20:49:01 +00:00
Devang Patel
6c91023dad
Do not split loops rejected by processOneIterationLoop().
...
llvm-svn: 41194
2007-08-20 20:24:15 +00:00
Nick Lewycky
3ea7d43047
Oops, remove assert that wasn't meant to be committed.
...
llvm-svn: 41170
2007-08-18 23:21:28 +00:00
Nick Lewycky
125c9f2db3
Never insert duplicate edges.
...
llvm-svn: 41169
2007-08-18 23:18:03 +00:00
Nick Lewycky
9570c4c5ca
Typo.
...
llvm-svn: 41168
2007-08-18 15:08:56 +00:00
Devang Patel
17cd20c077
Avoid spliting loops where two split condition branches are not independent.
...
llvm-svn: 41148
2007-08-18 00:00:32 +00:00
Devang Patel
fded73828f
When one branch of condition is eliminated then head of the other
...
branch is not necessary immediate dominators of merge blcok in all cases.
llvm-svn: 41144
2007-08-17 21:59:16 +00:00
Owen Anderson
5173494693
Factor out some code into a helper function.
...
llvm-svn: 41131
2007-08-16 22:51:56 +00:00
Owen Anderson
70ac81f489
Add some more comments to GVN.
...
llvm-svn: 41129
2007-08-16 22:02:55 +00:00
Devang Patel
423d18e306
Dominance frontier is now required.
...
llvm-svn: 41096
2007-08-15 03:34:53 +00:00
Devang Patel
315ea9c519
Cleanup removeBlocks.
...
Use dominance frontier to fixup incoming edges of successor blocks not domianted by DeadBB.
Use df_iterator to walk and delete basic blocks dominated by DeadBB.
llvm-svn: 41095
2007-08-15 03:31:47 +00:00
Reid Spencer
2cfe29fc0b
Remove unneeded header file.
...
llvm-svn: 41094
2007-08-15 03:01:04 +00:00
Devang Patel
b0924a406a
Avoid triangle loops.
...
llvm-svn: 41093
2007-08-15 02:14:55 +00:00
Devang Patel
d39b9d41bb
Avoid nested loops at the moment.
...
llvm-svn: 41090
2007-08-14 23:53:57 +00:00
Devang Patel
7ff808f08e
Fix dominance frontier update while removing blocks.
...
llvm-svn: 41082
2007-08-14 18:35:57 +00:00
Owen Anderson
b970937b06
Eliminate PHI nodes with constant values during normal GVN processing, even when
...
they're not related to eliminating a load.
llvm-svn: 41081
2007-08-14 18:33:27 +00:00
Owen Anderson
d9a7b61a64
Be more aggressive in pruning unnecessary PHI nodes when doing PHI construction.
...
llvm-svn: 41080
2007-08-14 18:16:29 +00:00
Owen Anderson
c855195b2f
Make GVN iterative.
...
llvm-svn: 41078
2007-08-14 18:04:11 +00:00
Owen Anderson
8d15350c4e
Fix a case where GVN was failing to return true when it had, in fact, modified
...
the function.
llvm-svn: 41077
2007-08-14 17:59:48 +00:00
Devang Patel
5d8b2b3293
Handle last value assignments.
...
llvm-svn: 41063
2007-08-14 01:30:57 +00:00
Devang Patel
55b6a8f212
StartValue is already calculated.
...
llvm-svn: 41062
2007-08-14 00:15:45 +00:00
Devang Patel
762259ba73
Preserve simple analysis.
...
llvm-svn: 41054
2007-08-13 22:22:13 +00:00
Devang Patel
e3ea62e49a
Preserve dominator info.
...
llvm-svn: 41053
2007-08-13 22:13:24 +00:00
Devang Patel
63d1affed8
Split loops and do CFG cleanup.
...
llvm-svn: 41029
2007-08-12 07:02:51 +00:00
Reid Spencer
c71871ba30
Remove unused variables.
...
llvm-svn: 41028
2007-08-12 04:45:36 +00:00
Chris Lattner
50f25115cd
Transform a load from an undef/zero global into an undef/global even if we
...
have complex pointer manipulation going on. This allows us to compile
stuff like this:
__m128i foo(__m128i x){
static const unsigned int c_0[4] = { 0, 0, 0, 0 };
__m128i v_Zero = _mm_loadu_si128((__m128i*)c_0);
x = _mm_unpacklo_epi8(x, v_Zero);
return x;
}
into:
_foo:
xorps %xmm1, %xmm1
punpcklbw %xmm1, %xmm0
ret
llvm-svn: 41022
2007-08-11 18:48:48 +00:00
Devang Patel
c8cff19c2e
Clone loop.
...
llvm-svn: 40998
2007-08-10 18:07:13 +00:00
Devang Patel
b1160475fd
Remove unncessary duplication.
...
llvm-svn: 40979
2007-08-10 00:59:03 +00:00
Devang Patel
23c5e2dcd3
Calculate exit and start value of true loop and false loop respectively.
...
llvm-svn: 40978
2007-08-10 00:53:35 +00:00
Devang Patel
51da0a465a
ExitCondition and Induction variable are loop constraints
...
not split condition constraints.
llvm-svn: 40977
2007-08-10 00:33:50 +00:00
Chris Lattner
3548932573
when we see a unaligned load from an insufficiently aligned global or
...
alloca, increase the alignment of the load, turning it into an aligned load.
This allows us to compile:
#include <xmmintrin.h>
__m128i foo(__m128i x){
static const unsigned int c_0[4] = { 0, 0, 0, 0 };
__m128i v_Zero = _mm_loadu_si128((__m128i*)c_0);
x = _mm_unpacklo_epi8(x, v_Zero);
return x;
}
into:
_foo:
punpcklbw _c_0.5944, %xmm0
ret
.data
.lcomm _c_0.5944,16,4 # c_0.5944
instead of:
_foo:
movdqu _c_0.5944, %xmm1
punpcklbw %xmm1, %xmm0
ret
.data
.lcomm _c_0.5944,16,2 # c_0.5944
llvm-svn: 40971
2007-08-09 19:05:49 +00:00
Owen Anderson
3eba1e8f5c
Make NonLocal and None const in the right way. :-)
...
llvm-svn: 40961
2007-08-09 04:42:44 +00:00
Devang Patel
652613b1db
Traverse loop blocks' terminators to find split candidates.
...
llvm-svn: 40960
2007-08-09 01:39:01 +00:00
Devang Patel
30bf54b9d3
Add cost analysis.
...
llvm-svn: 40952
2007-08-08 22:25:28 +00:00
Devang Patel
ed9b08c44f
Preserve dom info while processing one iteration loop.
...
llvm-svn: 40947
2007-08-08 21:39:47 +00:00
Owen Anderson
f988c3ee5c
Change the None and NonLocal markers in memdep to be const.
...
llvm-svn: 40946
2007-08-08 21:39:39 +00:00
Devang Patel
a74e533c5e
Clear split info.
...
llvm-svn: 40944
2007-08-08 21:18:27 +00:00
Devang Patel
4bd07b86ae
Handle multiple split conditions.
...
llvm-svn: 40941
2007-08-08 21:02:17 +00:00
Owen Anderson
04ba98d0f5
Global values also don't undead-ify pointers in our dead alloca's set.
...
llvm-svn: 40936
2007-08-08 19:12:31 +00:00
Owen Anderson
fd772509e9
Make handleEndBlock significantly faster with one trivial improvement,
...
and one hack to avoid hitting a bad case when the alias analysis is imprecise.
llvm-svn: 40935
2007-08-08 18:38:28 +00:00
Owen Anderson
930610cceb
Small improvement: if a function doesn't access memory, we don't need to scan
...
it for potentially undeading pointers.
llvm-svn: 40933
2007-08-08 17:58:56 +00:00
Owen Anderson
ecd9348bd6
Add some comments, remove a dead argument, and simplify some control flow.
...
No functionality change.
llvm-svn: 40932
2007-08-08 17:50:09 +00:00
Owen Anderson
02298f7388
A few more small cleanups.
...
llvm-svn: 40922
2007-08-08 06:06:02 +00:00
Owen Anderson
a9facda8db
First round of cleanups from Chris' feedback.
...
llvm-svn: 40919
2007-08-08 04:52:29 +00:00
Devang Patel
62f8bf41c3
Embrace patch review feedback.
...
llvm-svn: 40915
2007-08-08 01:51:27 +00:00
Devang Patel
08504a95b0
Fix new compare instruction's signness. Caught by Chris during review.
...
llvm-svn: 40912
2007-08-07 23:17:52 +00:00
Owen Anderson
2c98f8ee3c
Don't insert nearly as many redundant phi nodes.
...
llvm-svn: 40909
2007-08-07 23:12:31 +00:00
Devang Patel
92d6834fd7
Use eraseFromParent().
...
llvm-svn: 40903
2007-08-07 17:45:35 +00:00
David Greene
310b0d7e39
Fix comment typo
...
llvm-svn: 40898
2007-08-07 16:52:03 +00:00
David Greene
7eea029403
Fix GLIBCXX_DEBUG error triggered by incrementing erased iterator.
...
llvm-svn: 40897
2007-08-07 16:44:38 +00:00
Devang Patel
228dc09ff6
Begin loop index split pass.
...
llvm-svn: 40883
2007-08-07 00:25:56 +00:00
Nick Lewycky
34cf98c558
It's safe to fold not of fcmp.
...
llvm-svn: 40870
2007-08-06 20:04:16 +00:00
David Greene
02ac5cfce9
Make this code more efficient.
...
llvm-svn: 40861
2007-08-06 15:09:17 +00:00
Reid Spencer
ac9262072f
Silence some warnings from doxygen about @param argument name not matching the
...
actual argument name of the documented function.
llvm-svn: 40851
2007-08-05 19:35:22 +00:00