Hal Finkel 24517d023f Add the ability to use GEPs for address sinking in CGP ... The current memory-instruction optimization logic in CGP, which sinks parts of the address computation that can be adsorbed by the addressing mode, does this by explicitly converting the relevant part of the address computation into IR-level integer operations (making use of ptrtoint and inttoptr). For most targets this is currently not a problem, but for targets wishing to make use of IR-level aliasing analysis during CodeGen, the use of ptrtoint/inttoptr is a problem for two reasons: 1. BasicAA becomes less powerful in the face of the ptrtoint/inttoptr 2. In cases where type-punning was used, and BasicAA was used to override TBAA, BasicAA may no longer do so. (this had forced us to disable all use of TBAA in CodeGen; something which we can now enable again) This (use of GEPs instead of ptrtoint/inttoptr) is not currently enabled by default (except for those targets that use AA during CodeGen), and so aside from some PowerPC subtargets and SystemZ, there should be no change in behavior. We may be able to switch completely away from the ptrtoint/inttoptr sinking on all targets, but further testing is required. I've doubled-up on a number of existing tests that are sensitive to the address sinking behavior (including some store-merging tests that are sensitive to the order of the resulting ADD operations at the SDAG level). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206092 91177308-0d34-0410-b5e6-96231b3b80d8		2014-04-12 00:59:48 +00:00
..
2011-03-09-CPSRSpill.ll
2011-03-17-AsmPrinterCrash.ll
2011-03-21-Unaligned-Frame-Index.ll
2011-04-21-CPSRBug.ll
2011-10-18-LdStOptBug.ll
2012-01-11-ComparisonDAGCrash.ll
2012-05-07-DAGCombineVectorExtract.ll
2012-05-07-MemcpyAlignBug.ll
2012-05-09-LOADgot-bug.ll
2012-05-22-LdStOptBug.ll
2012-06-06-FPToUI.ll
2012-07-11-InstrEmitterBug.ll
2013-01-13-ffast-fcmp.ll
2013-01-23-frem-crash.ll
2013-01-23-sext-crash.ll
2013-02-12-shufv8i8.ll
aapcs.ll
abi_align.ll	[ARM64] Rename FP to the UAL-compliant 'X29'.	2014-04-09 14:43:50 +00:00
abi-varargs.ll
abi.ll	[ARM64] Rename FP to the UAL-compliant 'X29'.	2014-04-09 14:43:50 +00:00
addp.ll
addr-mode-folding.ll	Fix some doc and comment typos	2014-04-09 14:47:27 +00:00
addr-type-promotion.ll
addrmode.ll
AdvSIMD-Scalar.ll
alloc-no-stack-realign.ll
alloca-frame-pointer-offset.ll
andCmpBrToTBZ.ll
anyregcc-crash.ll
anyregcc.ll	[Stackmaps] Update the stackmap format to use 64-bit relocations for the function address and properly align all entries.	2014-03-31 22:14:04 +00:00
arith-saturating.ll
arith.ll
atomic-128.ll
atomic.ll
basic-pic.ll	ARM64: use GOT for weak symbols & PIC.	2014-04-02 14:39:11 +00:00
big-imm-offsets.ll
big-stack.ll
bitfield-extract.ll
blockaddress.ll
build-vector.ll	Fix some doc and comment typos	2014-04-09 14:47:27 +00:00
call-tailcalls.ll
cast-opt.ll
ccmp-heuristics.ll
ccmp.ll
clrsb.ll	Add ARM64 CLS patterns	2014-04-11 22:27:58 +00:00
coalesce-ext.ll
code-model-large-abs.ll
collect-loh-garbage-crash.ll
collect-loh-str.ll
collect-loh.ll
compact-unwind-unhandled-cfi.S
complex-ret.ll
const-addr.ll	[Constant Hoisting][ARM64] Enable constant hoisting for ARM64.	2014-04-08 20:39:59 +00:00
convert-v2f64-v2i32.ll
convert-v2i32-v2f64.ll
copy-tuple.ll
crc32.ll
crypto.ll
cse.ll
csel.ll
cvt.ll
dagcombiner-convergence.ll
dagcombiner-indexed-load.ll	Add the ability to use GEPs for address sinking in CGP	2014-04-12 00:59:48 +00:00
dagcombiner-load-slicing.ll
dead-register-def-bug.ll	[ARM64] Teach the ARM64DeadRegisterDefinition pass to respect implicit-defs.	2014-04-03 20:51:08 +00:00
dup.ll
early-ifcvt.ll
elf-calls.ll
elf-constpool.ll
elf-globals.ll
ext.ll
extend-int-to-fp.ll
extend.ll
extern-weak.ll	ARM64: use GOT for weak symbols & PIC.	2014-04-02 14:39:11 +00:00
extload-knownzero.ll
extract_subvector.ll
extract.ll
fast-isel-addr-offset.ll
fast-isel-alloca.ll	[ARM64] Rename FP to the UAL-compliant 'X29'.	2014-04-09 14:43:50 +00:00
fast-isel-br.ll
fast-isel-call.ll	[ARM64] Rename FP to the UAL-compliant 'X29'.	2014-04-09 14:43:50 +00:00
fast-isel-conversion.ll
fast-isel-fcmp.ll
fast-isel-gv.ll
fast-isel-icmp.ll
fast-isel-indirectbr.ll
fast-isel-intrinsic.ll
fast-isel-materialize.ll
fast-isel-noconvert.ll
fast-isel-rem.ll
fast-isel-ret.ll
fast-isel-select.ll
fast-isel.ll
fastcc-tailcall.ll
fastisel-gep-promote-before-add.ll
fcmp-opt.ll
fcopysign.ll
fixed-point-scalar-cvt-dagcombine.ll
fmadd.ll	ARM64: fix fmsub patterns which assumed accum operand was first	2014-04-08 12:23:51 +00:00
fmax.ll
fminv.ll	ARM64: shuffle patterns around for fmin/fmax & add tests.	2014-03-31 15:46:30 +00:00
fmuladd.ll
fold-address.ll
fold-lsl.ll
fp128-folding.ll
fp128.ll
fp-imm.ll	[ARM64] Fix materialization of an fp128 zero immediate. There currently	2014-03-31 00:02:10 +00:00
fp.ll
frame-index.ll
frameaddr.ll	[ARM64] Rename LR to the UAL-compliant 'X30'.	2014-04-09 14:43:59 +00:00
global-address.ll
hello.ll	[ARM64] Rename LR to the UAL-compliant 'X30'.	2014-04-09 14:43:59 +00:00
i16-subreg-extract.ll
icmp-opt.ll
illegal-float-ops.ll	ARM64: fix lowering of fp128 fptosi/fptoui	2014-04-02 14:39:07 +00:00
indexed-memory.ll
inline-asm-error-I.ll
inline-asm-error-J.ll
inline-asm-error-K.ll
inline-asm-error-L.ll
inline-asm-error-M.ll
inline-asm-error-N.ll
inline-asm-zero-reg-error.ll
inline-asm.ll
join-reserved.ll
jumptable.ll
ld1.ll	ARM64: add patterns for more lane-wise ld1/st1 operations.	2014-04-01 10:37:09 +00:00
ldp.ll
ldur.ll
ldxr-stxr.ll
leaf-compact-unwind.ll
leaf.ll
lit.local.cfg
long-shift.ll
memcpy-inline.ll
memset-inline.ll
memset-to-bzero.ll
movi.ll
mul.ll
neg.ll	ARM64: add extra scalar neg pattern & tests.	2014-03-31 15:46:42 +00:00
neon-compare-instructions.ll
neon-v1i1-setcc.ll	DAGLegalize: add last-ditch type-legalization for VSELECT.	2014-04-04 14:49:30 +00:00
patchpoint.ll	[ARM64] Rename FP to the UAL-compliant 'X29'.	2014-04-09 14:43:50 +00:00
platform-reg.ll
popcnt.ll
prefetch.ll
promote-const.ll
redzone.ll
register-offset-addressing.ll
register-pairing.ll
regress-f128csel-flags.ll
regress-interphase-shift.ll	ARM64: add regression test for r205519.	2014-04-03 09:36:05 +00:00
return-vector.ll
returnaddr.ll	[ARM64] Rename LR to the UAL-compliant 'X30'.	2014-04-09 14:43:59 +00:00
rev.ll
rounding.ll
scaled_iv.ll
scvt.ll
shifted-sext.ll
simd-scalar-to-vector.ll	ARM64: make sure first argument to INSERT_SUBVECTOR has right type.	2014-04-02 14:38:58 +00:00
simplest-elf.ll
sincos.ll	ARM64: don't generate __sincos_stret calls unless on MachO	2014-04-03 07:06:13 +00:00
sitofp-combine-chains.ll
sli-sri-opt.ll
smaxv.ll
sminv.ll
spill-lr.ll
spill.ll
st1.ll	ARM64: add patterns for more lane-wise ld1/st1 operations.	2014-04-01 10:37:09 +00:00
stack-no-frame.ll
stackmap.ll	[Stackmaps] Update the stackmap format to use 64-bit relocations for the function address and properly align all entries.	2014-03-31 22:14:04 +00:00
stacksave.ll
stp.ll
strict-align.ll
stur.ll
subvector-extend.ll
swizzle-tbl-i16-layout.ll
tbl.ll
this-return.ll
tls-darwin.ll
tls-dynamic-together.ll
tls-dynamics.ll
tls-execs.ll
trap.ll
trn.ll
trunc-store.ll
umaxv.ll
uminv.ll
umov.ll
unaligned_ldst.ll
uzp.ll
vaargs.ll
vabs.ll	ARM64: add i64 scalar pattern for @llvm.arm64.abs	2014-03-31 15:46:17 +00:00
vadd.ll
vaddlv.ll
vaddv.ll	ARM64: use regalloc-friendly COPY_TO_REGCLASS for bitcasts	2014-04-04 09:03:09 +00:00
variadic-aapcs.ll
vbitwise.ll
vclz.ll
vcmp.ll	ARM64: add pattern for <1 x i64> custom not node.	2014-04-09 06:55:39 +00:00
vcnt.ll
vcombine.ll
vcvt_f32_su32.ll
vcvt_f.ll	ARM64: convert fp16 narrowing ISel to pseudo-instruction	2014-04-02 14:38:54 +00:00
vcvt_n.ll
vcvt_su32_f32.ll
vcvt.ll
vcvtxd_f32_f64.ll
vecCmpBr.ll
vecFold.ll
vector-ext.ll
vector-imm.ll
vector-ldst.ll
vext.ll
vfloatintrinsics.ll
vhadd.ll
vhsub.ll
virtual_base.ll
vmax.ll
vminmaxnm.ll
vmovn.ll
vmul.ll	ARM64: scalarize v1i64 mul operation	2014-04-09 07:07:02 +00:00
volatile.ll
vqadd.ll	ARM64: add more scalar patterns for usqadd & suqadd.	2014-03-31 15:46:26 +00:00
vqsub.ll
vselect.ll
vsetcc_fp.ll
vshift.ll
vshr.ll	ARM64: add extra patterns for scalar shifts	2014-03-31 15:46:46 +00:00
vshuffle.ll
vsqrt.ll	ARM64: add more scalar patterns for reciprocal ops.	2014-03-31 15:46:22 +00:00
vsra.ll	ARM64: add extra patterns for scalar shifts	2014-03-31 15:46:46 +00:00
vsub.ll
weak-reference.ll
xaluo.ll
zero-cycle-regmov.ll
zero-cycle-zeroing.ll
zext.ll
zextload-unscaled.ll
zip.ll