26455c4ae0
Introduce new SelectionDAG node opcodes VEXTRACT_SUBVECTOR and
...
VCONCAT_VECTORS. Use these for CopyToReg and CopyFromReg legalizing in
the case that the full register is to be split into subvectors instead
of scalars. This replaces uses of VBIT_CONVERT to present values as
vector-of-vector types in order to make whole subvectors accessible via
BUILD_VECTOR and EXTRACT_VECTOR_ELT.
This is in preparation for adding extended ValueType values, where
having vector-of-vector types is undesirable.
llvm-svn: 37569
2007-06-13 15:12:02 +00:00
cbd51c8b60
When creating CopyFromReg nodes, always use legal types. And use the
...
correct types for the result vector, even though it is currently bitcasted
to a different type immediately.
llvm-svn: 37568
2007-06-13 14:55:16 +00:00
97f7236e70
The fix that was applied for PR1224 stops the compiler
...
crashing but breaks exception handling. The problem
described in PR1224 is that invoke is a terminator that
can produce a value. The value may be needed in other
blocks. The code that writes to registers values needed
in other blocks runs before terminators are lowered (in
this case invoke) so asserted because the value was not
yet available. The fix that was applied was to do invoke
lowering earlier, before writing values to registers.
The problem this causes is that the code to copy values
to registers can be output after the invoke call. If
an exception is raised and control is passed to the
landing pad then this copy-code will never execute. If
the value is needed in some code path reached via the
landing pad then that code will get something bogus.
So revert the original fix and simply skip invoke values
in the general copying to registers code. Instead copy
the invoke value to a register in the invoke lowering code.
llvm-svn: 37567
2007-06-13 05:51:31 +00:00
9a4d987a5f
Do not change the size of function arguments. PR 1489.
...
llvm-svn: 37496
2007-06-07 21:07:15 +00:00
61166501a1
Additional fix for PR1422: make sure the landing pad label is placed in the
...
correct machine basic block - do not rely on the eh.exception intrinsic
being in the landing pad: the loop optimizers can move it out.
llvm-svn: 37463
2007-06-06 10:05:18 +00:00
b4c2690446
Pass the DAG to SDNode::dump to let it do more detailed dumps in some cases.
...
llvm-svn: 37413
2007-06-04 16:17:33 +00:00
92a7f3a65e
Resolve implicit alignment before computing the FoldingSet information so
...
that the CSE map always contains explicit alignment information. This allows
more loads to be CSE'd when there is a mix of explicit-alignment loads and
implicit-alignment loads.
Also, in SelectionDAG::FindModifiedNodeSlot, add the operands to the
FoldingSetNodeID before the load/store information instead of after, so
that it matches what is done elsewhere.
llvm-svn: 37411
2007-06-04 15:49:41 +00:00
c063f5f362
Integrate exception filter support and exception catch support. This
...
simplifies the code in DwarfWriter, allows for multiple filters and
makes it trivial to specify filters accompanied by cleanups or catch-all
specifications (see next patch). What a deal! Patch blessed by Anton.
llvm-svn: 37398
2007-06-02 16:53:42 +00:00
706421e712
Since TypeInfos are passed as i8 pointers, a NULL TypeInfo should be passed
...
as a null i8 pointer not as a 0 i32.
llvm-svn: 37383
2007-06-01 08:18:30 +00:00
397c4d9ef6
Fix CodeGen/PowerPC/2007-05-30-dagcombine-miscomp.ll, and PR1473.
...
llvm-svn: 37362
2007-05-30 16:30:06 +00:00
4698083b96
tighten up recursion depth again
...
llvm-svn: 37330
2007-05-25 02:19:06 +00:00
30978078bf
Minor comment cleanups.
...
llvm-svn: 37321
2007-05-24 14:36:04 +00:00
703e0f8608
Add explicit qualification for namespace MVT members.
...
llvm-svn: 37320
2007-05-24 14:33:05 +00:00
a4d187b8ce
Fix a typo that caused combiner to create mal-formed pre-indexed store where value store is the same as the base pointer.
...
llvm-svn: 37318
2007-05-24 02:35:39 +00:00
3b327826db
Mark all calls as "could throw", when exceptions are enabled. Emit necessary LP info too. This fixes PR1439
...
llvm-svn: 37311
2007-05-23 11:08:31 +00:00
6509c0673f
prevent exponential recursion in isNegatibleForFree
...
llvm-svn: 37310
2007-05-23 07:35:22 +00:00
1fa8276e70
same patch as the previous one, but the symmetric case
...
llvm-svn: 37249
2007-05-19 00:46:51 +00:00
b08cbbd737
Disable the (A == (B-A)) -> 2*A == B xform when the sub has multiple uses (in
...
this case, the xform introduces an extra operation). This compiles
PowerPC/compare-duplicate.ll into:
_test:
subf r2, r3, r4
cmplw cr0, r2, r3
bne cr0, LBB1_2 ;F
instead of:
_test:
slwi r2, r3, 1
subf r3, r3, r4
cmplw cr0, r4, r2
bne cr0, LBB1_2 ;F
This is target independent of course.
llvm-svn: 37246
2007-05-19 00:43:44 +00:00
b539df3389
Qualify calls to getTypeForValueType with MVT:: too.
...
llvm-svn: 37233
2007-05-18 18:41:29 +00:00
1796f1f8e9
Qualify several calls to functions in the MVT namespace, for consistency.
...
llvm-svn: 37230
2007-05-18 17:52:13 +00:00
0184f88deb
disable MaskedValueIsZero, ComputeMaskedBits, and SimplifyDemandedBits for
...
i128 integers. The 64-bit masks are not wide enough to represent the results.
These should be converted to APInt someday.
llvm-svn: 37169
2007-05-17 18:19:23 +00:00
2135bc08d6
add expand support for ADDC/SUBC/ADDE/SUBE so we can codegen 128-bit add/sub on 32-bit (or less) targets
...
llvm-svn: 37168
2007-05-17 18:15:41 +00:00
429178d727
Add target hook to specify block size limit for if-conversion.
...
llvm-svn: 37134
2007-05-16 23:45:53 +00:00
7a6c175e7a
Don't fold bitconvert(load) for preinc/postdec loads. Likewise stores.
...
llvm-svn: 37130
2007-05-16 22:45:30 +00:00
48fb92f75d
Use a ptr set instead of a linear search to unique TokenFactor operands.
...
This fixes PR1423
llvm-svn: 37102
2007-05-16 06:37:59 +00:00
288f133c71
Bug fix: should check ABI alignment, not pref. alignment.
...
llvm-svn: 37094
2007-05-16 02:04:50 +00:00
3f142cbca2
Fix an infinite recursion in GetNegatedExpression.
...
llvm-svn: 37086
2007-05-15 17:05:43 +00:00
c7596efdad
Fix some subtle issues handling immediate values. This fixes
...
test/CodeGen/ARM/2007-05-14-InlineAsmCstCrash.ll
llvm-svn: 37069
2007-05-15 01:33:58 +00:00
e49c974a7c
implement a simple fneg optimization/propagation thing. This compiles:
...
CodeGen/PowerPC/fneg.ll into:
_t4:
fmul f0, f3, f4
fmadd f1, f1, f2, f0
blr
instead of:
_t4:
fneg f0, f3
fmul f0, f0, f4
fmsub f1, f1, f2, f0
blr
llvm-svn: 37054
2007-05-14 22:04:50 +00:00
f325c2a65e
Can't fold the bit_convert is the store is a truncating store.
...
llvm-svn: 36962
2007-05-09 21:49:47 +00:00
192d09c2d9
Do not assert, when case range split metric is zero and JTs are not allowed: just emit binary tree in this case. This
...
fixes PR1403.
llvm-svn: 36959
2007-05-09 20:07:08 +00:00
562e45692e
Forgot a check.
...
llvm-svn: 36910
2007-05-07 21:36:06 +00:00
a4cf58a103
Enable a couple of xforms:
...
- (store (bitconvert v)) -> (store v) if resultant store does not require
higher alignment
- (bitconvert (load v)) -> (load (bitconvert*)v) if resultant load does not
require higher alignment
llvm-svn: 36908
2007-05-07 21:27:48 +00:00
671e8c4444
Parameter attributes on invoke calls were being lost due to the wrong
...
attribute index being used. Fix proposed by Anton Korobeynikov, who
asked me to implement and commit it for him. This is PR1398.
llvm-svn: 36906
2007-05-07 20:49:28 +00:00
a8fd7fdc25
Detabify
...
llvm-svn: 36891
2007-05-06 20:14:21 +00:00
07e6f3257c
Propagate alignment/volatility in two places.
...
Implement support for expanding a bitcast from an illegal vector type to
a legal one (e.g. 4xi32 -> 4xf32 in SSE1). This fixes PR1371 and
CodeGen/X86/2007-05-05-VecCastExpand.ll
llvm-svn: 36787
2007-05-05 19:39:05 +00:00
4cb9eb81ef
A bitcast of a global variable may have been constant folded to a GEP -
...
handle this case too.
llvm-svn: 36745
2007-05-04 17:12:26 +00:00
044a0a8cfb
Don't create indexed load / store with zero offset!
...
llvm-svn: 36716
2007-05-03 23:52:19 +00:00
44a2ed66b1
Allow i/s to match (gv+c). This fixes CodeGen/PowerPC/2007-05-03-InlineAsm-S-Constraint.ll
...
and PR1382
llvm-svn: 36672
2007-05-03 16:54:34 +00:00
8c78a0bff0
Drop 'const'
...
llvm-svn: 36662
2007-05-03 01:11:54 +00:00
11940fbba3
Properly set arguments bitwidth of EHSELECT node
...
llvm-svn: 36654
2007-05-02 22:15:48 +00:00
e95c6ad802
Use 'static const char' instead of 'static const int'.
...
Due to darwin gcc bug, one version of darwin linker coalesces
static const int, which defauts PassID based pass identification.
llvm-svn: 36652
2007-05-02 21:39:20 +00:00
09f162ca6a
Do not use typeinfo to identify pass in pass manager.
...
llvm-svn: 36632
2007-05-01 21:15:47 +00:00
b68343cdd8
Forgot about chain result; also UNDEF cannot have multiple values.
...
llvm-svn: 36622
2007-05-01 08:53:39 +00:00
a684cd23a5
* Only turn a load to UNDEF if all of its outputs have no uses (indexed loads
...
produce two results.)
* Do not touch volatile loads.
llvm-svn: 36604
2007-05-01 00:38:21 +00:00
8cfd33b647
Continue refactoring inline asm code. If there is an earlyclobber output
...
register, preallocate all input registers and the early clobbered output.
This fixes PR1357 and CodeGen/PowerPC/2007-04-30-InlineAsmEarlyClobber.ll
llvm-svn: 36599
2007-04-30 21:11:17 +00:00
4333f8b1cf
refactor GetRegistersForValue to take OpInfo as an argument instead of various
...
pieces of it. No functionality change.
llvm-svn: 36592
2007-04-30 17:29:31 +00:00
ef07332504
refactor some code, no functionality change
...
llvm-svn: 36590
2007-04-30 17:16:27 +00:00
412d61af43
generalize aggregate handling
...
llvm-svn: 36568
2007-04-29 18:58:03 +00:00
401d8db381
memory operands that have a direct operand should have their stores created
...
before the copies into physregs are done. This avoids having flag operands
skip the store, causing cycles in the dag at sched time. This fixes infinite
loops on these tests:
test/CodeGen/Generic/2007-04-08-MultipleFrameIndices.ll for PR1308
test/CodeGen/PowerPC/2007-01-29-lbrx-asm.ll
test/CodeGen/PowerPC/2007-01-31-InlineAsmAddrMode.ll
test/CodeGen/X86/2006-07-12-InlineAsmQConstraint.ll for PR828
llvm-svn: 36547
2007-04-28 21:12:06 +00:00
de339fa55d
eliminate more redundant constraint type analysis
...
llvm-svn: 36546
2007-04-28 21:03:16 +00:00
b2e55562ed
merge constraint type analysis stuff together.
...
llvm-svn: 36545
2007-04-28 21:01:43 +00:00
d7e3b6c442
Significant refactoring of the inline asm stuff, to support future changes.
...
No functionality change.
llvm-svn: 36544
2007-04-28 20:49:53 +00:00
1deacd61f4
memory inputs to an inline asm are required to have an address available.
...
If the operand is not already an indirect operand, spill it to a constant
pool entry or a stack slot.
This fixes PR1356 and CodeGen/X86/2007-04-27-InlineAsm-IntMemInput.ll
llvm-svn: 36536
2007-04-28 06:42:38 +00:00
d102ed0ac6
Fix CodeGen/Generic/2007-04-27-LargeMemObject.ll and
...
CodeGen/Generic/2007-04-27-InlineAsm-X-Dest.ll
llvm-svn: 36534
2007-04-28 06:08:13 +00:00
4df3e8093b
Fix this to match change to InlineAsm class.
...
llvm-svn: 36524
2007-04-28 04:05:59 +00:00
1cbe208cda
Fix incorrect legalization of EHSELECTOR. This fixes
...
CodeGen/Generic/2007-04-14-EHSelectorCrash.ll and PR1326
llvm-svn: 36510
2007-04-27 17:12:52 +00:00
bf535fc8bd
Expand UINT_TO_FP in turns of SINT_TO_FP when UINTTOFP_* libcalls are not available.
...
llvm-svn: 36501
2007-04-27 07:33:31 +00:00
784fe9dbbb
improve EH global handling, patch by Duncan Sands.
...
llvm-svn: 36499
2007-04-27 01:20:11 +00:00
8131ab7c0f
enable Anton's shift/and switch lowering stuff! It now passes ppc bootstrap
...
successfully! woohoo...
llvm-svn: 36496
2007-04-26 21:09:43 +00:00
d7ae7f1659
Fixx off-by-one bug, which prevents llvm-gcc bootstrap on ppc32
...
llvm-svn: 36490
2007-04-26 20:44:04 +00:00
e131e3ac02
Fix a typo in a comment.
...
llvm-svn: 36485
2007-04-26 19:40:56 +00:00
15f269afa3
This was lefted out. Fixed sumarray-dbl.
...
llvm-svn: 36445
2007-04-25 18:33:21 +00:00
cb0ed0cfbd
allow support for 64-bit stack objects
...
llvm-svn: 36420
2007-04-25 04:08:28 +00:00
01a26c74ae
Be more careful about folding op(x, undef) when we have vector operands.
...
This fixes CodeGen/X86/2007-04-24-VectorCrash.ll
llvm-svn: 36413
2007-04-25 00:00:45 +00:00
47917b697f
Assertion when using a 1-element vector for an add operation. Get the
...
real vector type in this case.
llvm-svn: 36402
2007-04-24 21:13:23 +00:00
4cfa616cee
Use '-1U' where '-1UL' is obvious overkill, eliminating gcc warnings about
...
tests always being true in the process.
llvm-svn: 36387
2007-04-24 01:24:20 +00:00
8af6d5896f
PR400 phase 2. Propagate attributed load/store information through DAGs.
...
llvm-svn: 36356
2007-04-22 23:15:30 +00:00
4e91908f17
X86 TLS: Implement review feedback.
...
llvm-svn: 36318
2007-04-21 20:56:26 +00:00
0c1349e6bc
Revert Christopher Lamb's load/store alignment changes.
...
llvm-svn: 36309
2007-04-21 18:36:27 +00:00
bff50208c8
add support for alignment attributes on load/store instructions
...
llvm-svn: 36301
2007-04-21 08:16:25 +00:00
94314be0e0
Allow the lowering of ISD::GLOBAL_OFFSET_TABLE.
...
llvm-svn: 36290
2007-04-20 23:02:39 +00:00
2518889872
Implement "general dynamic", "initial exec" and "local exec" TLS models for
...
X86 32 bits.
llvm-svn: 36283
2007-04-20 21:38:10 +00:00
f03c90bee6
allow SRL to simplify its operands, as it doesn't demand all bits as input.
...
llvm-svn: 36245
2007-04-18 03:06:49 +00:00
bf14f20632
When replacing a node in SimplifyDemandedBits, if the old node used any
...
single-use nodes, they will be dead soon. Make sure to remove them before
processing other nodes. This implements CodeGen/X86/shl_elim.ll
llvm-svn: 36244
2007-04-18 03:05:22 +00:00
15c1b820cc
fix a pasto
...
llvm-svn: 36242
2007-04-18 03:01:40 +00:00
4aff52bf3d
Fix a bug in my previous patch, grabbing the shift amount width from the
...
wrong operand.
llvm-svn: 36223
2007-04-17 22:53:02 +00:00
9a861a8550
Fold (x << c1)>> c2 into a single shift if the bits shifted out aren't used.
...
This compiles:
int baz(long long a) { return (short)(((int)(a >>24)) >> 9); }
into:
_baz:
srwi r2, r3, 1
extsh r3, r2
blr
on PPC, instead of:
_baz:
slwi r2, r3, 8
srwi r2, r2, 9
extsh r3, r2
blr
GCC produces:
_baz:
srwi r10,r4,24
insrwi r10,r3,24,0
srawi r9,r3,24
srawi r3,r10,9
extsh r3,r3
blr
This implements CodeGen/PowerPC/shl_elim.ll
llvm-svn: 36221
2007-04-17 21:14:16 +00:00
9ad5915559
SIGN_EXTEND_INREG does not demand its top bits. Give SimplifyDemandedBits
...
a chance to hack on it. This compiles:
int baz(long long a) { return (short)(((int)(a >>24)) >> 9); }
into:
_baz:
slwi r2, r3, 8
srwi r2, r2, 9
extsh r3, r2
blr
instead of:
_baz:
srwi r2, r4, 24
rlwimi r2, r3, 8, 0, 23
srwi r2, r2, 9
extsh r3, r2
blr
This implements CodeGen/PowerPC/sign_ext_inreg1.ll
llvm-svn: 36212
2007-04-17 19:03:21 +00:00
fb80151c42
Removed tabs everywhere except autogenerated & external files. Add make
...
target for tabs checking.
llvm-svn: 36146
2007-04-16 18:10:23 +00:00
6bd7b7b30b
disable switch lowering using shift/and. It still breaks ppc bootstrap for
...
some reason. :( Will investigate.
llvm-svn: 36011
2007-04-14 19:39:41 +00:00
8a1a84f96e
Fix PR1325: Case range optimization was performed in the case it
...
shouldn't. Also fix some "latent" bug on 64-bit platforms
llvm-svn: 35990
2007-04-14 13:25:55 +00:00
7196f09edc
disable shift/and lowering to work around PR1325 for now.
...
llvm-svn: 35985
2007-04-14 02:26:56 +00:00
e288040abf
Fix PR1323 : we haven't updated phi nodes in good manner :)
...
llvm-svn: 35963
2007-04-13 06:53:51 +00:00
5111499136
the result of an inline asm copy can be an arbitrary VT that the register
...
class supports. In the case of vectors, this means we often get the wrong
type (e.g. we get v4f32 instead of v8i16). Make sure to convert the vector
result to the right type. This fixes CodeGen/X86/2007-04-11-InlineAsmVectorResult.ll
llvm-svn: 35944
2007-04-12 06:00:20 +00:00
a77cb3ce68
fold noop vbitconvert instructions
...
llvm-svn: 35943
2007-04-12 05:58:43 +00:00
784a68a702
Fix weirdness handling single element vectors.
...
llvm-svn: 35941
2007-04-12 04:44:28 +00:00
c6251a7dfd
For PR1284:
...
Implement the "part_set" intrinsic.
llvm-svn: 35938
2007-04-12 02:48:46 +00:00
18e4ac4107
fix an infinite loop compiling ldecod, notice by JeffC.
...
llvm-svn: 35910
2007-04-11 16:51:53 +00:00
a083ffcad7
Fix this harder.
...
llvm-svn: 35888
2007-04-11 06:50:51 +00:00
c5f85d3738
don't create shifts by zero, fix some problems with my previous patch
...
llvm-svn: 35887
2007-04-11 06:43:25 +00:00
65786b078c
Teach the codegen to turn [aez]ext (setcc) -> selectcc of 1/0, which often
...
allows other simplifications. For example, this compiles:
int isnegative(unsigned int X) {
return !(X < 2147483648U);
}
Into this code:
x86:
movl 4(%esp), %eax
shrl $31, %eax
ret
arm:
mov r0, r0, lsr #31
bx lr
thumb:
lsr r0, r0, #31
bx lr
instead of:
x86:
cmpl $0, 4(%esp)
sets %al
movzbl %al, %eax
ret
arm:
mov r3, #0
cmp r0, #0
movlt r3, #1
mov r0, r3
bx lr
thumb:
mov r2, #1
mov r1, #0
cmp r0, #0
blt LBB1_2 @entry
LBB1_1: @entry
cpy r2, r1
LBB1_2: @entry
cpy r0, r2
bx lr
Testcase here: test/CodeGen/Generic/ispositive.ll
llvm-svn: 35883
2007-04-11 05:32:27 +00:00
41189c63cc
Codegen integer abs more efficiently using the trick from the PPC CWG. This
...
improves codegen on many architectures. Tests committed as CodeGen/*/iabs.ll
X86 Old: X86 New:
_test: _test:
movl 4(%esp), %ecx movl 4(%esp), %eax
movl %ecx, %eax movl %eax, %ecx
negl %eax sarl $31, %ecx
testl %ecx, %ecx addl %ecx, %eax
cmovns %ecx, %eax xorl %ecx, %eax
ret ret
PPC Old: PPC New:
_test: _test:
cmpwi cr0, r3, -1 srawi r2, r3, 31
neg r2, r3 add r3, r3, r2
bgt cr0, LBB1_2 ; xor r3, r3, r2
LBB1_1: ; blr
mr r3, r2
LBB1_2: ;
blr
ARM Old: ARM New:
_test: _test:
rsb r3, r0, #0 add r3, r0, r0, asr #31
cmp r0, #0 eor r0, r3, r0, asr #31
movge r3, r0 bx lr
mov r0, r3
bx lr
Thumb Old: Thumb New:
_test: _test:
neg r2, r0 asr r2, r0, #31
cmp r0, #0 add r0, r0, r2
bge LBB1_2 eor r0, r2
LBB1_1: @ bx lr
cpy r0, r2
LBB1_2: @
bx lr
Sparc Old: Sparc New:
test: test:
save -96, %o6, %o6 save -96, %o6, %o6
sethi 0, %l0 sra %i0, 31, %l0
sub %l0, %i0, %l0 add %i0, %l0, %l1
subcc %i0, -1, %l1 xor %l1, %l0, %i0
bg .BB1_2 restore %g0, %g0, %g0
nop retl
.BB1_1: nop
or %g0, %l0, %i0
.BB1_2:
restore %g0, %g0, %g0
retl
nop
It also helps alpha/ia64 :)
llvm-svn: 35881
2007-04-11 05:11:38 +00:00
a472f66dd0
For PR1146:
...
Put the parameter attributes in their own ParamAttr name space. Adjust the
rest of llvm as a result.
llvm-svn: 35877
2007-04-11 02:44:20 +00:00
f269d84ca0
apparently some people commit without building the tree, or they forget to
...
commit a LOT of files.
llvm-svn: 35858
2007-04-10 03:20:39 +00:00
e0bbbd3774
No longer needed.
...
llvm-svn: 35850
2007-04-09 23:42:32 +00:00
35f0417ec1
remove dead target hooks.
...
llvm-svn: 35847
2007-04-09 23:34:08 +00:00
39f65335d5
remove some dead target hooks, subsumed by isLegalAddressingMode
...
llvm-svn: 35840
2007-04-09 22:27:04 +00:00
da964a2852
Use integer log for metric calculation
...
llvm-svn: 35834
2007-04-09 21:57:03 +00:00
0475f3b4e9
Unbreak VC++ build.
...
llvm-svn: 35817
2007-04-09 14:32:59 +00:00
Dan Gohman