Jim Grosbach 0e536ee4ca Legalize: Improve legalization of long vector extends.
When an extend more than doubles the size of the elements (e.g., a zext
from v16i8 to v16i32), the normal legalization method of splitting the
vectors will run into problems as by the time the destination vector is
legal, the source vector is illegal. The end result is the operation
often becoming scalarized, with the typical horrible performance. For
example, on x86_64, the simple input of:
define void @bar(<16 x i8> %a, <16 x i32>* %p) nounwind {
  %tmp = zext <16 x i8> %a to <16 x i32>
  store <16 x i32> %tmp, <16 x i32>*%p
  ret void
}

Generates:
  .section  __TEXT,__text,regular,pure_instructions
  .section  __TEXT,__const
  .align  5
LCPI0_0:
  .long 255                     ## 0xff
  .long 255                     ## 0xff
  .long 255                     ## 0xff
  .long 255                     ## 0xff
  .long 255                     ## 0xff
  .long 255                     ## 0xff
  .long 255                     ## 0xff
  .long 255                     ## 0xff
  .section  __TEXT,__text,regular,pure_instructions
  .globl  _bar
  .align  4, 0x90
_bar:
  vpunpckhbw  %xmm0, %xmm0, %xmm1
  vpunpckhwd  %xmm0, %xmm1, %xmm2
  vpmovzxwd %xmm1, %xmm1
  vinsertf128 $1, %xmm2, %ymm1, %ymm1
  vmovaps LCPI0_0(%rip), %ymm2
  vandps  %ymm2, %ymm1, %ymm1
  vpmovzxbw %xmm0, %xmm3
  vpunpckhwd  %xmm0, %xmm3, %xmm3
  vpmovzxbd %xmm0, %xmm0
  vinsertf128 $1, %xmm3, %ymm0, %ymm0
  vandps  %ymm2, %ymm0, %ymm0
  vmovaps %ymm0, (%rdi)
  vmovaps %ymm1, 32(%rdi)
  vzeroupper
  ret

So instead we can check if there are legal types that enable us to split
more cleverly when the input vector is already legal such that we don't
turn it into an illegal type. If the extend is such that it's more than
doubling the size of the input we check if
  - the number of vector elements is even,
  - the source type is legal,
  - the type of a split source is illegal,
  - the type of an extended (by doubling element size) source is legal, and
  - the type of that extended source when split is legal.
If the conditions are met, instead of just splitting both the
destination and the source types, we create an extend that only goes up
one "step" (doubling the element width), and the continue legalizing the
rest of the operation normally. The result is that this operates as a
new, more effecient, termination condition for the loop of "split the
operation until the destination type is legal."

With this change, the above example now compiles to:
_bar:
  vpxor %xmm1, %xmm1, %xmm1
  vpunpcklbw  %xmm1, %xmm0, %xmm2
  vpunpckhwd  %xmm1, %xmm2, %xmm3
  vpunpcklwd  %xmm1, %xmm2, %xmm2
  vinsertf128 $1, %xmm3, %ymm2, %ymm2
  vpunpckhbw  %xmm1, %xmm0, %xmm0
  vpunpckhwd  %xmm1, %xmm0, %xmm3
  vpunpcklwd  %xmm1, %xmm0, %xmm0
  vinsertf128 $1, %xmm3, %ymm0, %ymm0
  vmovaps %ymm0, 32(%rdi)
  vmovaps %ymm2, (%rdi)
  vzeroupper
  ret

This generalizes a custom lowering that was added a while back to the
ARM backend. That lowering is no longer necessary, and is removed. The
testcases for it, however, provide excellent ARM tests for this change
and so remain.

rdar://14735100

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193727 91177308-0d34-0410-b5e6-96231b3b80d8
2013-10-31 00:20:48 +00:00
..
2013-03-25 13:47:46 +00:00
2013-10-11 15:40:14 +00:00
2013-07-25 17:52:30 +00:00
2013-10-08 16:47:11 +00:00
2013-10-04 16:52:58 +00:00
2013-07-16 18:26:15 +00:00
2013-08-23 17:48:53 +00:00

//===---------------------------------------------------------------------===//

Common register allocation / spilling problem:

        mul lr, r4, lr
        str lr, [sp, #+52]
        ldr lr, [r1, #+32]
        sxth r3, r3
        ldr r4, [sp, #+52]
        mla r4, r3, lr, r4

can be:

        mul lr, r4, lr
        mov r4, lr
        str lr, [sp, #+52]
        ldr lr, [r1, #+32]
        sxth r3, r3
        mla r4, r3, lr, r4

and then "merge" mul and mov:

        mul r4, r4, lr
        str r4, [sp, #+52]
        ldr lr, [r1, #+32]
        sxth r3, r3
        mla r4, r3, lr, r4

It also increase the likelihood the store may become dead.

//===---------------------------------------------------------------------===//

bb27 ...
        ...
        %reg1037 = ADDri %reg1039, 1
        %reg1038 = ADDrs %reg1032, %reg1039, %NOREG, 10
    Successors according to CFG: 0x8b03bf0 (#5)

bb76 (0x8b03bf0, LLVM BB @0x8b032d0, ID#5):
    Predecessors according to CFG: 0x8b0c5f0 (#3) 0x8b0a7c0 (#4)
        %reg1039 = PHI %reg1070, mbb<bb76.outer,0x8b0c5f0>, %reg1037, mbb<bb27,0x8b0a7c0>

Note ADDri is not a two-address instruction. However, its result %reg1037 is an
operand of the PHI node in bb76 and its operand %reg1039 is the result of the
PHI node. We should treat it as a two-address code and make sure the ADDri is
scheduled after any node that reads %reg1039.

//===---------------------------------------------------------------------===//

Use local info (i.e. register scavenger) to assign it a free register to allow
reuse:
        ldr r3, [sp, #+4]
        add r3, r3, #3
        ldr r2, [sp, #+8]
        add r2, r2, #2
        ldr r1, [sp, #+4]  <==
        add r1, r1, #1
        ldr r0, [sp, #+4]
        add r0, r0, #2

//===---------------------------------------------------------------------===//

LLVM aggressively lift CSE out of loop. Sometimes this can be negative side-
effects:

R1 = X + 4
R2 = X + 7
R3 = X + 15

loop:
load [i + R1]
...
load [i + R2]
...
load [i + R3]

Suppose there is high register pressure, R1, R2, R3, can be spilled. We need
to implement proper re-materialization to handle this:

R1 = X + 4
R2 = X + 7
R3 = X + 15

loop:
R1 = X + 4  @ re-materialized
load [i + R1]
...
R2 = X + 7 @ re-materialized
load [i + R2]
...
R3 = X + 15 @ re-materialized
load [i + R3]

Furthermore, with re-association, we can enable sharing:

R1 = X + 4
R2 = X + 7
R3 = X + 15

loop:
T = i + X
load [T + 4]
...
load [T + 7]
...
load [T + 15]
//===---------------------------------------------------------------------===//

It's not always a good idea to choose rematerialization over spilling. If all
the load / store instructions would be folded then spilling is cheaper because
it won't require new live intervals / registers. See 2003-05-31-LongShifts for
an example.

//===---------------------------------------------------------------------===//

With a copying garbage collector, derived pointers must not be retained across
collector safe points; the collector could move the objects and invalidate the
derived pointer. This is bad enough in the first place, but safe points can
crop up unpredictably. Consider:

        %array = load { i32, [0 x %obj] }** %array_addr
        %nth_el = getelementptr { i32, [0 x %obj] }* %array, i32 0, i32 %n
        %old = load %obj** %nth_el
        %z = div i64 %x, %y
        store %obj* %new, %obj** %nth_el

If the i64 division is lowered to a libcall, then a safe point will (must)
appear for the call site. If a collection occurs, %array and %nth_el no longer
point into the correct object.

The fix for this is to copy address calculations so that dependent pointers
are never live across safe point boundaries. But the loads cannot be copied
like this if there was an intervening store, so may be hard to get right.

Only a concurrent mutator can trigger a collection at the libcall safe point.
So single-threaded programs do not have this requirement, even with a copying
collector. Still, LLVM optimizations would probably undo a front-end's careful
work.

//===---------------------------------------------------------------------===//

The ocaml frametable structure supports liveness information. It would be good
to support it.

//===---------------------------------------------------------------------===//

The FIXME in ComputeCommonTailLength in BranchFolding.cpp needs to be
revisited. The check is there to work around a misuse of directives in inline
assembly.

//===---------------------------------------------------------------------===//

It would be good to detect collector/target compatibility instead of silently
doing the wrong thing.

//===---------------------------------------------------------------------===//

It would be really nice to be able to write patterns in .td files for copies,
which would eliminate a bunch of explicit predicates on them (e.g. no side 
effects).  Once this is in place, it would be even better to have tblgen 
synthesize the various copy insertion/inspection methods in TargetInstrInfo.

//===---------------------------------------------------------------------===//

Stack coloring improvements:

1. Do proper LiveStackAnalysis on all stack objects including those which are
   not spill slots.
2. Reorder objects to fill in gaps between objects.
   e.g. 4, 1, <gap>, 4, 1, 1, 1, <gap>, 4 => 4, 1, 1, 1, 1, 4, 4

//===---------------------------------------------------------------------===//

The scheduler should be able to sort nearby instructions by their address. For
example, in an expanded memset sequence it's not uncommon to see code like this:

  movl $0, 4(%rdi)
  movl $0, 8(%rdi)
  movl $0, 12(%rdi)
  movl $0, 0(%rdi)

Each of the stores is independent, and the scheduler is currently making an
arbitrary decision about the order.

//===---------------------------------------------------------------------===//

Another opportunitiy in this code is that the $0 could be moved to a register:

  movl $0, 4(%rdi)
  movl $0, 8(%rdi)
  movl $0, 12(%rdi)
  movl $0, 0(%rdi)

This would save substantial code size, especially for longer sequences like
this. It would be easy to have a rule telling isel to avoid matching MOV32mi
if the immediate has more than some fixed number of uses. It's more involved
to teach the register allocator how to do late folding to recover from
excessive register pressure.