Commit Graph

24626 Commits

Author SHA1 Message Date
Jeff Cohen
e564de2d37 Unlike Unix, Windows won't let a file be implicitly replaced via renaming without explicit permission.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28157 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-07 02:51:51 +00:00
Nate Begeman
77f361f5b3 New rlwimi implementation, which is superior to the old one. There are
still a couple missed optimizations, but we now generate all the possible
rlwimis for multiple inserts into the same bitfield.  More regression tests
to come.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28156 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-07 00:23:38 +00:00
Chris Lattner
822db93e57 Use ComputeMaskedBits to determine # sign bits as a fallback. This allows us
to handle all kinds of stuff, including silly things like:
sextinreg(setcc,i16) -> setcc.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28155 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 23:48:13 +00:00
Chris Lattner
e60351bb72 Add some more sign propagation cases
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28154 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 23:40:29 +00:00
Jeff Cohen
943b9b6651 Apply bug fix supplied by Greg Pettyjohn for a bug he found: '<invalid>' is not a legal path on Windows.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28153 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 23:25:53 +00:00
Chris Lattner
310b578023 Simplify some code, add a couple minor missed folds
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28152 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 23:06:26 +00:00
Chris Lattner
b9ebacdb4d constant fold sign_extend_inreg
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28151 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 23:05:41 +00:00
Chris Lattner
541a24f7af remove cases handled elsewhere
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28150 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 22:43:44 +00:00
Chris Lattner
d6f7fe76a6 Add some more simple sign bit propagation cases.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28149 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 22:39:59 +00:00
Jeff Cohen
d43b18d9e4 Fix some loose ends in MASM support.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28148 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 21:27:14 +00:00
Chris Lattner
a0aec0a61d new testcase we handle right now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28147 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 18:15:50 +00:00
Chris Lattner
ee4ea92358 Use the new TargetLowering::ComputeNumSignBits method to eliminate
sign_extend_inreg operations.  Though ComputeNumSignBits is still rudimentary,
this is enough to compile this:

short test(short X, short x) {
  int Y = X+x;
  return (Y >> 1);
}
short test2(short X, short x) {
  int Y = (short)(X+x);
  return Y >> 1;
}

into:

_test:
        add r2, r3, r4
        srawi r3, r2, 1
        blr
_test2:
        add r2, r3, r4
        extsh r2, r2
        srawi r3, r2, 1
        blr

instead of:

_test:
        add r2, r3, r4
        srawi r2, r2, 1
        extsh r3, r2
        blr
_test2:
        add r2, r3, r4
        extsh r2, r2
        srawi r2, r2, 1
        extsh r3, r2
        blr


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28146 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 09:30:03 +00:00
Chris Lattner
5c3e21d687 Add some really really simple code for computing sign-bit propagation.
This will certainly be enhanced in the future.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28145 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 09:27:13 +00:00
Chris Lattner
a0d513ba75 Add some new methods for computing sign bit information.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28144 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 09:26:22 +00:00
Chris Lattner
f0df8824eb When inserting casts, be careful of where we put them. We cannot insert
a cast immediately before a PHI node.

This fixes Regression/CodeGen/Generic/2006-05-06-GEP-Cast-Sink-Crash.ll


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28143 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 09:10:37 +00:00
Chris Lattner
707f41302b new testcase
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28142 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 09:09:47 +00:00
Chris Lattner
33a6113995 Move some code around.
Make the "fold (and (cast A), (cast B)) -> (cast (and A, B))" transformation
only apply when both casts really will cause code to be generated.  If one or
both doesn't, then this xform doesn't remove a cast.

This fixes Transforms/InstCombine/2006-05-06-Infloop.ll


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28141 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 09:00:16 +00:00
Chris Lattner
985906323f new testcase from ghostscript that inf looped instcombine
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28140 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 08:58:06 +00:00
Chris Lattner
80a7ecc923 Teach the X86 backend about non-i32 inline asm register classes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28139 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 00:29:37 +00:00
Chris Lattner
c93dfda905 Fold (trunc (srl x, c)) -> (srl (trunc x), c)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28138 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-06 00:11:52 +00:00
Chris Lattner
b72773bb88 Fold trunc(any_ext). This gives stuff like:
27,28c27
<       movzwl %di, %edi
<       movl %edi, %ebx
---
>       movw %di, %bx


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28137 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 22:56:26 +00:00
Chris Lattner
06afe07037 Shrink shifts when possible.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28136 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 22:53:17 +00:00
Chris Lattner
fe8babf689 Implement ComputeMaskedBits/SimplifyDemandedBits for ISD::TRUNCATE
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28135 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 22:32:12 +00:00
Chris Lattner
35c3913328 Print a grouping around inline asm blocks so that we can tell when we are
using them.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28134 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 21:50:04 +00:00
Chris Lattner
7660ebc90a Print *some* grouping around inline asm blocks so we know where they are.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28133 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 21:48:50 +00:00
Chris Lattner
1c05997bd8 Indent multiline asm strings more nicely
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28132 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 21:47:05 +00:00
Chris Lattner
bd04aa5796 Teach the code generator to use cvtss2sd as extload f32 -> f64
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28131 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 21:35:18 +00:00
Chris Lattner
e564dbb51c Fold (fpext (load x)) -> (extload x)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28130 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 21:34:35 +00:00
Chris Lattner
903236468c More aggressively sink GEP offsets into loops. For example, before we
generated:

        movl 8(%esp), %eax
        movl %eax, %edx
        addl $4316, %edx
        cmpb $1, %cl
        ja LBB1_2       #cond_false
LBB1_1: #cond_true
        movl L_QuantizationTables720$non_lazy_ptr, %ecx
        movl %ecx, (%edx)
        movl L_QNOtoQuantTableShift720$non_lazy_ptr, %edx
        movl %edx, 4460(%eax)
        ret
...

Now we generate:

        movl 8(%esp), %eax
        cmpb $1, %cl
        ja LBB1_2       #cond_false
LBB1_1: #cond_true
        movl L_QuantizationTables720$non_lazy_ptr, %ecx
        movl %ecx, 4316(%eax)
        movl L_QNOtoQuantTableShift720$non_lazy_ptr, %ecx
        movl %ecx, 4460(%eax)
        ret

... which uses one fewer register.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28129 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 21:17:49 +00:00
Chris Lattner
581a7ad7c9 Fix an infinite loop compiling oggenc last night.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28128 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 20:51:30 +00:00
Evan Cheng
4713724eda Need extload patterns after Chris' DAG combiner changes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28127 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 08:23:07 +00:00
Chris Lattner
6fc205fc44 Implement InstCombine/cast.ll:test29
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28126 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 06:39:07 +00:00
Chris Lattner
a0771b8020 New testcase
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28125 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 06:38:40 +00:00
Chris Lattner
0d8dae749c Fold some common code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28124 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 06:32:04 +00:00
Chris Lattner
540121f1ec Implement:
// fold (and (sext x), (sext y)) -> (sext (and x, y))
  // fold (or  (sext x), (sext y)) -> (sext (or  x, y))
  // fold (xor (sext x), (sext y)) -> (sext (xor x, y))
  // fold (and (aext x), (aext y)) -> (aext (and x, y))
  // fold (or  (aext x), (aext y)) -> (aext (or  x, y))
  // fold (xor (aext x), (aext y)) -> (aext (xor x, y))


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28123 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 06:31:05 +00:00
Chris Lattner
a3dc3f692c Pull and through and/or/xor. This compiles some bitfield code to:
mov EAX, DWORD PTR [ESP + 4]
        mov ECX, DWORD PTR [EAX]
        mov EDX, ECX
        add EDX, EDX
        or EDX, ECX
        and EDX, -2147483648
        and ECX, 2147483647
        or EDX, ECX
        mov DWORD PTR [EAX], EDX
        ret

instead of:

        sub ESP, 4
        mov DWORD PTR [ESP], ESI
        mov EAX, DWORD PTR [ESP + 8]
        mov ECX, DWORD PTR [EAX]
        mov EDX, ECX
        add EDX, EDX
        mov ESI, ECX
        and ESI, -2147483648
        and EDX, -2147483648
        or EDX, ESI
        and ECX, 2147483647
        or EDX, ECX
        mov DWORD PTR [EAX], EDX
        mov ESI, DWORD PTR [ESP]
        add ESP, 4
        ret


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28122 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 06:10:43 +00:00
Chris Lattner
5ffc066912 Implement a variety of simplifications for ANY_EXTEND.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28121 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 05:58:59 +00:00
Chris Lattner
35e5c14b80 Factor some code, add these transformations:
// fold (and (trunc x), (trunc y)) -> (trunc (and x, y))
  // fold (or  (trunc x), (trunc y)) -> (trunc (or  x, y))
  // fold (xor (trunc x), (trunc y)) -> (trunc (xor x, y))


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28120 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 05:51:50 +00:00
Evan Cheng
8f7f7125e9 Better implementation of truncate. ISel matches it to a pseudo instruction
that gets emitted as movl (for r32 to i16, i8) or a movw (for r16 to i8). And
if the destination gets allocated a subregister of the source operand, then
the instruction will not be emitted at all.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28119 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 05:40:20 +00:00
Chris Lattner
55c63257f3 New note, Nate, please check to see if I'm full of it :)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28118 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 05:36:15 +00:00
Jeff Cohen
4b75e73791 Fix VC++ compilation error.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28117 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 01:47:05 +00:00
Nate Begeman
7328797b16 Somehow, I missed this part of the checkin a couple days ago
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28116 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 01:13:11 +00:00
Chris Lattner
7e598096ea Sink noop copies into the basic block that uses them. This reduces the number
of cross-block live ranges, and allows the bb-at-a-time selector to always
coallesce these away, at isel time.

This reduces the load on the coallescer and register allocator.  For example
on a codec on X86, we went from:

   1643 asm-printer           - Number of machine instrs printed
    419 liveintervals         - Number of loads/stores folded into instructions
   1144 liveintervals         - Number of identity moves eliminated after coalescing
   1022 liveintervals         - Number of interval joins performed
    282 liveintervals         - Number of intervals after coalescing
   1304 liveintervals         - Number of original intervals
     86 regalloc              - Number of times we had to backtrack
1.90232 regalloc              - Ratio of intervals processed over total intervals
     40 spiller               - Number of values reused
    182 spiller               - Number of loads added
    121 spiller               - Number of stores added
    132 spiller               - Number of register spills
      6 twoaddressinstruction - Number of instructions commuted to coalesce
    360 twoaddressinstruction - Number of two-address instructions

to:

   1636 asm-printer           - Number of machine instrs printed
    403 liveintervals         - Number of loads/stores folded into instructions
   1155 liveintervals         - Number of identity moves eliminated after coalescing
   1033 liveintervals         - Number of interval joins performed
    279 liveintervals         - Number of intervals after coalescing
   1312 liveintervals         - Number of original intervals
     76 regalloc              - Number of times we had to backtrack
1.88998 regalloc              - Ratio of intervals processed over total intervals
      1 spiller               - Number of copies elided
     41 spiller               - Number of values reused
    191 spiller               - Number of loads added
    114 spiller               - Number of stores added
    128 spiller               - Number of register spills
      4 twoaddressinstruction - Number of instructions commuted to coalesce
    356 twoaddressinstruction - Number of two-address instructions

On this testcase, this change provides a modest reduction in spill code,
regalloc iterations, and total instructions emitted.  It increases the number
of register coallesces.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28115 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 01:04:50 +00:00
Chris Lattner
71fe0f4a43 Add a helper method.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28114 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-05 00:51:42 +00:00
Chris Lattner
67ef9e4304 wrap long line
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28113 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-04 23:35:31 +00:00
Chris Lattner
b93b0347d8 Adjust to use proper TargetData copy ctor
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28112 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-04 21:18:40 +00:00
Chris Lattner
ed5b016f0e Fix this to be a proper copy ctor
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28111 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-04 21:17:35 +00:00
Chris Lattner
6e994b7492 Final pass of minor cleanups for MachineInstr
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28110 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-04 19:36:09 +00:00
Evan Cheng
14a6db8bd9 Initial support for register pressure aware scheduling. The register reduction
scheduler can go into a "vertical mode" (i.e. traversing up the two-address
chain, etc.) when the register pressure is low.
This does seem to reduce the number of spills in the cases I've looked at. But
with x86, it's no guarantee the performance of the code improves.
It can be turned on with -sched-vertically option.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28108 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-04 19:16:39 +00:00
Chris Lattner
943b5e117f Remove redundancy and a level of indirection when creating machine operands
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28107 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-04 19:14:44 +00:00