Commit Graph

901 Commits

Author SHA1 Message Date
Bob Wilson
56519aac04 PR9139: Specify ARM/Darwin triple for vector-DAGCombine.ll test.
The i64_buildvector test in this file relies on the alignment of i64 and
f64 types being the same, which is true for Darwin but not AAPCS.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@125525 91177308-0d34-0410-b5e6-96231b3b80d8
2011-02-14 22:12:50 +00:00
Nate Begeman
7973f350b7 Implement sdiv & udiv for <4 x i16> and <8 x i8> NEON vector types.
This avoids moving each element to the integer register file and calling __divsi3 etc. on it.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@125402 91177308-0d34-0410-b5e6-96231b3b80d8
2011-02-11 20:53:29 +00:00
Evan Cheng
c143dd4f63 Fix buggy fcopysign lowering.
This
define float @foo(float %x, float %y) nounwind readnone {
entry:
  %0 = tail call float @copysignf(float %x, float %y) nounwind readnone
  ret float %0
}

Was compiled to:
    vmov     s0, r1
    bic      r0, r0, #-2147483648
    vmov     s1, r0
    vcmpe.f32    s0, #0
    vmrs         apsr_nzcv, fpscr
    it           lt
    vneglt.f32   s1, s1
    vmov         r0, s1
    bx           lr

This fails to copy the sign of -0.0f because it's lost during the float to int
conversion. Also, it's sub-optimal when the inputs are in GPR registers.

Now it uses integer and + or operations when it's profitable. And it's correct!
    lsrs    r1, r1, #31
    bfi     r0, r1, #31, #1
    bx      lr
rdar://8984306


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@125357 91177308-0d34-0410-b5e6-96231b3b80d8
2011-02-11 02:28:55 +00:00
Andrew Trick
651c4ba80c PostRA antidependence breaker unit test for PR8986.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@125091 91177308-0d34-0410-b5e6-96231b3b80d8
2011-02-08 17:42:05 +00:00
Andrew Trick
e4f34cd4db PostRA antidependence breaker unit test for rdar://8959122.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@125090 91177308-0d34-0410-b5e6-96231b3b80d8
2011-02-08 17:41:12 +00:00
Evan Cheng
aa26102db4 Fix an obvious typo which caused an isel assertion. rdar://8964854.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@125023 91177308-0d34-0410-b5e6-96231b3b80d8
2011-02-07 18:50:47 +00:00
Bob Wilson
1c3ef90cab Add codegen support for using post-increment NEON load/store instructions.
The vld1-lane, vld1-dup and vst1-lane instructions do not yet support using
post-increment versions, but all the rest of the NEON load/store instructions
should be handled now.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@125014 91177308-0d34-0410-b5e6-96231b3b80d8
2011-02-07 17:43:21 +00:00
Jason W Kim
f009a961ca Rework some .ARM.attribute work for improved gcc compatibility.
Unified EmitTextAttribute for both Asm and Obj emission (.cpu only)
Added necessary cortex-A8 related attrs for codegen compat tests.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@124995 91177308-0d34-0410-b5e6-96231b3b80d8
2011-02-07 00:49:53 +00:00
Evan Cheng
31959b19a7 Given a pair of floating point load and store, if there are no other uses of
the load, then it may be legal to transform the load and store to integer
load and store of the same width.

This is done if the target specified the transformation as profitable. e.g.
On arm, this can transform:
vldr.32 s0, []
vstr.32 s0, []

to

ldr r12, []
str r12, []

rdar://8944252


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@124708 91177308-0d34-0410-b5e6-96231b3b80d8
2011-02-02 01:06:55 +00:00
Eric Christopher
f382f3b4dd Add a testcase for my last checkin.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@124358 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-27 06:01:17 +00:00
Evan Cheng
7cfa656ad8 Don't merge restore with tail call instruction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@124167 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-25 01:28:33 +00:00
Evan Cheng
53519f015e Last round of fixes for movw + movt global address codegen.
1. Fixed ARM pc adjustment.
2. Fixed dynamic-no-pic codegen
3. CSE of pc-relative load of global addresses.

It's now enabled by default for Darwin.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123991 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-21 18:55:51 +00:00
Andrew Trick
d1dace8aea Enable support for precise scheduling of the instruction selection
DAG. Disable using "-disable-sched-cycles".

For ARM, this enables a framework for modeling the cpu pipeline and
counting stalls. It also activates several heuristics to drive
scheduling based on the model. Scheduling is inherently imprecise at
this stage, and until spilling is improved it may defeat attempts to
schedule. However, this framework provides greater control over
tuning codegen.

Although the flag is not target-specific, it should have very little
affect on the default scheduler used by x86. The only two changes that
affect x86 are:
- scheduling a high-latency operation bumps the current cycle so independent
  operations can have their latency covered. i.e. two independent 4
  cycle operations can produce results in 4 cycles, not 8 cycles.
- Two operations with equal register pressure impact and no
  latency-based stalls on their uses will be prioritized by depth before height
  (height is irrelevant if no stalls occur in the schedule below this point).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123971 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-21 06:19:05 +00:00
Andrew Trick
c8bfd1d78f Convert -enable-sched-cycles and -enable-sched-hazard to -disable
flags. They are still not enable in this revision.

Added TargetInstrInfo::isZeroCost() to fix a fundamental problem with
the scheduler's model of operand latency in the selection DAG.

Generalized unit tests to work with sched-cycles.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123969 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-21 05:51:33 +00:00
Evan Cheng
d7e3cc840b Don't be overly aggressive with CSE of "ldr constantpool". If it's a pc-relative
value, the "add pc" must be CSE'ed at the same time. We could follow the same
approach as T2 by adding pseudo instructions that combine the ldr + "add pc".
But the better approach is to use movw + movt (which I will enable soon), so
I'll leave this as a TODO.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123949 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-20 23:55:07 +00:00
Evan Cheng
ab56e31424 Add test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123906 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-20 08:38:21 +00:00
Evan Cheng
9fe2009956 Sorry, several patches in one.
TargetInstrInfo:
Change produceSameValue() to take MachineRegisterInfo as an optional argument.
When in SSA form, targets can use it to make more aggressive equality analysis.

Machine LICM:
1. Eliminate isLoadFromConstantMemory, use MI.isInvariantLoad instead.
2. Fix a bug which prevent CSE of instructions which are not re-materializable.
3. Use improved form of produceSameValue.

ARM:
1. Teach ARM produceSameValue to look pass some PIC labels.
2. Look for operands from different loads of different constant pool entries
   which have same values.
3. Re-implement PIC GA materialization using movw + movt. Combine the pair with
   a "add pc" or "ldr [pc]" to form pseudo instructions. This makes it possible
   to re-materialize the instruction, allow machine LICM to hoist the set of
   instructions out of the loop and make it possible to CSE them. It's a bit
   hacky, but it significantly improve code quality.
4. Some minor bug fixes as well.

With the fixes, using movw + movt to materialize GAs significantly outperform the
load from constantpool method. 186.crafty and 255.vortex improved > 20%, 254.gap
and 176.gcc ~10%.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123905 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-20 08:34:58 +00:00
Eric Christopher
38a18261b9 If we can, lower the multiply part of a umulo/smulo call to a libcall
with an invalid type then split the result and perform the overflow check
normally.

Fixes the 32-bit parts of rdar://8622122 and rdar://8774702.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123864 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-20 00:29:24 +00:00
Devang Patel
e449d1f189 Fix debug info for merged global.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123862 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-20 00:02:16 +00:00
Evan Cheng
5de5d4b6d0 Materialize GA addresses with movw + movt pairs for Darwin in PIC mode. e.g.
movw    r0, :lower16:(L_foo$non_lazy_ptr-(LPC0_0+4))
        movt    r0, :upper16:(L_foo$non_lazy_ptr-(LPC0_0+4))
LPC0_0:
        add     r0, pc, r0

It's not yet enabled by default as some tests are failing. I suspect bugs in
down stream tools.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123619 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-17 08:03:18 +00:00
Eric Christopher
94bb7b561c Even if we don't have 7 bytes of stack space we may need to save and
restore the stack pointer from the frame pointer on thumbv6.

Fixes rdar://8819685


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123196 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-11 00:16:04 +00:00
Evan Cheng
55d4200336 Recognize inline asm 'rev /bin/bash, ' as a bswap intrinsic call.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123048 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-08 01:24:27 +00:00
Bob Wilson
11a1dfffc8 Lower some BUILD_VECTORS using VEXT+shuffle.
Patch by Tim Northover.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123035 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-07 21:37:30 +00:00
Bob Wilson
8a644f4f6f Add testcases for PR8411 (vget_low and vget_high implemented as shuffles).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122997 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-07 06:44:14 +00:00
Bob Wilson
5e8b833707 Add ARM patterns to match EXTRACT_SUBVECTOR nodes.
Also fix an off-by-one in SelectionDAGBuilder that was preventing shuffle
vectors from being translated to EXTRACT_SUBVECTOR.
Patch by Tim Northover.

The test changes are needed to keep those spill-q tests from testing aligned
spills and restores.  If the only aligned stack objects are spill slots, we
no longer realign the stack frame.  Prior to this patch, an EXTRACT_SUBVECTOR
was legalized by loading from the stack, which created an aligned frame index.
Now, however, there is nothing except the spill slot in the stack frame, so
I added an aligned alloca.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122995 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-07 04:59:04 +00:00
Bob Wilson
6819dbb6f8 PR8921: LDM/POP do not support interworking prior to v5t.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122970 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-06 19:24:41 +00:00
Anton Korobeynikov
1db58f936b Update the test
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122666 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-01 20:57:26 +00:00
Bob Wilson
3c904694fc Radar 8803471: Fix expansion of ARM BCCi64 pseudo instructions.
If the basic block containing the BCCi64 (or BCCZi64) instruction ends with
an unconditional branch, that branch needs to be deleted before appending
the expansion of the BCCi64 to the end of the block.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122521 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-23 22:45:49 +00:00
Bob Wilson
316009054e Add ARM-specific DAG combining to cast i64 vector element load/stores to f64.
Type legalization splits up i64 values into pairs of i32 values, which leads
to poor quality code when inserting or extracting i64 vector elements.
If the vector element is loaded or stored, it can be treated as an f64 value
and loaded or stored directly from a VPR register.  Use the pre-legalization
DAG combiner to cast those vector elements to f64 types so that the type
legalizer won't mess them up.  Radar 8755338.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122319 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-21 06:43:19 +00:00
Chris Lattner
78d0094e4c move this test into the ARM test so that it is only run when the arm backend
is enabled.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122163 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-19 02:58:14 +00:00
Bob Wilson
3deb45149a Fix result type of Neon floating-point comparisons against zero.
The result vector elements are always integers.  Radar 8782191.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122112 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-18 00:04:33 +00:00
Bill Wendling
976ef86689 During local stack slot allocation, the materializeFrameBaseRegister function
may be called. If the entry block is empty, the insertion point iterator will be
the "end()" value. Calling ->getParent() on it (among others) causes problems.

Modify materializeFrameBaseRegister to take the machine basic block and insert
the frame base register at the beginning of that block. (It's very similar to
what the code does all ready. The only difference is that it will always insert
at the beginning of the entry block instead of after a previous materialization
of the frame base register. I doubt that that matters here.)

<rdar://problem/8782198>


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122104 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-17 23:09:14 +00:00
Bob Wilson
d727343a40 Fix a DAGCombiner crash when folding binary vector operations with constant
BUILD_VECTOR operands where the element type is not legal.  I had previously
changed this code to insert TRUNCATE operations, but that was just wrong.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122102 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-17 23:06:49 +00:00
Bob Wilson
4c6b3d35bb Combine several vector-related DAGCombiner tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122101 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-17 23:06:46 +00:00
Bob Wilson
495de3b783 Fix crash compiling a QQQQ REG_SEQUENCE for a Neon vld3_lane operation.
Radar 8776599

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122018 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-17 01:21:12 +00:00
Jason W Kim
f13743bb3c 1. ARM/MC/ELF: A few more ELF relocs for .o
2. Fixed EmitLocalCommonSymbol for ELF (Yes, they exist. :)
   Test added.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121951 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-16 03:12:17 +00:00
Eric Christopher
836c6245ad Don't handle -arm-long-calls in fast isel for now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121919 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-15 23:47:29 +00:00
Bob Wilson
0406356cd4 Add Neon VCVT instructions for f32 <-> f16 conversions.
Clang is now providing intrinsics for these and so we need to support them
in the backend.  Radar 8068427.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121902 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-15 22:14:12 +00:00
Evan Cheng
0c1aec1891 bfi A, (and B, C1), C2) -> bfi A, B, C2 iff C1 & C2 == C1. rdar://8458663
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121746 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-14 03:22:07 +00:00
Jason W Kim
db934e7474 fix fixme case typo :-)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121743 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-14 01:42:38 +00:00
Jason W Kim
3fa4c1dc95 First cut of ARM/MC/ELF PIC relocations.
Test has fixme, to move to .s -> .o test when AsmParser works better.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121732 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-13 23:16:07 +00:00
Evan Cheng
a9688c4b57 (or (and (shl A, #shamt), mask), B) => ARMbfi B, A, ~mask where lsb(mask) == #shamt. rdar://8752056
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121606 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-11 04:11:38 +00:00
Bob Wilson
746fa17d59 Add float patterns for Neon vld1-lane/dup and vst1-lane operations.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121583 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-10 22:13:32 +00:00
Bob Wilson
a92bac64cb Fix some invalid alignments for Neon vld-dup and vld/st-lane instructions.
Alignments smaller than the total size of the memory being loaded or stored,
unless the alignment is 8 bytes, are not allowed.  Add tests for this, too.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121506 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-10 19:37:42 +00:00
Jim Grosbach
c6f9261711 ARM stm/ldm instructions require more than one register in the register list.
Otherwise, a plain str/ldr should be used instead. Make sure we account for
that in prologue/epilogue code generation.
rdar://8745460

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121391 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-09 18:31:13 +00:00
Jason W Kim
a0871e7927 ARM/MC/ELF TPsoft is now a proper pseudo inst.
Added test to check bl __aeabi_read_tp gets emitted properly for ELF/ASM
as well as ELF/OBJ (including fixup)

Also added support for ELF::R_ARM_TLS_IE32



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121312 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-08 23:14:44 +00:00
Evan Cheng
06d65f5156 Fix a bad prologue / epilogue codegen bug where the compiler would emit illegal
vpush instructions to save / restore VFP / NEON registers like this:
vpush {d8,d10,d11}
vpop {d8,d10,d11}

vpush and vpop do not allow gaps in the register list.
rdar://8728956


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121197 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-07 23:08:38 +00:00
Devang Patel
afeaae7a94 If dbg_declare() or dbg_value() is not lowered by isel then emit DEBUG message instead of creating DBG_VALUE for undefined value in reg0.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121059 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-06 22:39:26 +00:00
Evan Cheng
48575f6ea7 Making use of VFP / NEON floating point multiply-accumulate / subtraction is
difficult on current ARM implementations for a few reasons.
1. Even though a single vmla has latency that is one cycle shorter than a pair
   of vmul + vadd, a RAW hazard during the first (4? on Cortex-a8) can cause
   additional pipeline stall. So it's frequently better to single codegen
   vmul + vadd.
2. A vmla folowed by a vmul, vmadd, or vsub causes the second fp instruction to
   stall for 4 cycles. We need to schedule them apart.
3. A vmla followed vmla is a special case. Obvious issuing back to back RAW
   vmla + vmla is very bad. But this isn't ideal either:
     vmul
     vadd
     vmla
   Instead, we want to expand the second vmla:
     vmla
     vmul
     vadd
   Even with the 4 cycle vmul stall, the second sequence is still 2 cycles
   faster.

Up to now, isel simply avoid codegen'ing fp vmla / vmls. This works well enough
but it isn't the optimial solution. This patch attempts to make it possible to
use vmla / vmls in cases where it is profitable.

A. Add missing isel predicates which cause vmla to be codegen'ed.
B. Make sure the fmul in (fadd (fmul)) has a single use. We don't want to
   compute a fmul and a fmla.
C. Add additional isel checks for vmla, avoid cases where vmla is feeding into
   fp instructions (except for the #3 exceptional case).
D. Add ARM hazard recognizer to model the vmla / vmls hazards.
E. Add a special pre-regalloc case to expand vmla / vmls when it's likely the
   vmla / vmls will trigger one of the special hazards.

Work in progress, only A+B are enabled.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120960 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-05 22:04:16 +00:00
Jason W Kim
85fed5e0c5 ARM/MC/ELF relocation "hello world" for movw/movt.
Lifted adjustFixupValue() from Darwin for sharing w ELF.
Test added
TODO:
  refactor ELFObjectWriter::RecordRelocation more.
  Possibly share more code with Darwin?
  Lots more relocations...



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120534 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-01 02:40:06 +00:00