Commit Graph

288 Commits

Author SHA1 Message Date
Chris Lattner
bd564bfc63 Start eliminating temporary vectors used to create DAG nodes. Instead, pass
in the start of an array and a count of operands where applicable.  In many
cases, the number of operands is known, so this static array can be allocated
on the stack, avoiding the heap.  In many other cases, a SmallVector can be
used, which has the same benefit in the common cases.

I updated a lot of code calling getNode that takes a vector, but ran out of
time.  The rest of the code should be updated, and these methods should be
removed.

We should also do the same thing to eliminate the methods that take a
vector of MVT::ValueTypes.

It would be extra nice to convert the dagiselemitter to avoid creating vectors
for operands when calling getTargetNode.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29566 91177308-0d34-0410-b5e6-96231b3b80d8
2006-08-08 02:23:42 +00:00
Evan Cheng
9bdca0302a Making TableGen'd instruction selection code non-recursive. This fixes PR805.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29548 91177308-0d34-0410-b5e6-96231b3b80d8
2006-08-07 22:17:58 +00:00
Chris Lattner
58d63567ea Remove some extraneous newlines.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29492 91177308-0d34-0410-b5e6-96231b3b80d8
2006-08-03 00:42:26 +00:00
Chris Lattner
32501a0e33 Really add support for compilers without noinline
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29491 91177308-0d34-0410-b5e6-96231b3b80d8
2006-08-03 00:26:13 +00:00
Evan Cheng
d3f8429f01 Pass N by reference to select functions to prevent gcc from allocating more stack.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29423 91177308-0d34-0410-b5e6-96231b3b80d8
2006-08-01 01:07:48 +00:00
Evan Cheng
6966c48767 Remove an unneeded match condition: the type check for root node has been
moved to outside the actual select routine.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29415 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-31 19:01:58 +00:00
Evan Cheng
b6ee7f73e5 Split each select function for a particular opcode into multiple ones. One
per possible ValueType of the node. e.g. Select_add is split into Select_add_i8,
Select_add_i16, etc.

For opcodes which do not produce a non-chain result, it is split on the
ValueType of its first non-chain operand. e.g. Select_store.

On X86 / Mac OS X, Select_store used to be the largest function. It had a stack
frame size of 8.5k. Now the largest one is Store_i32 with a frame size of 3.1k.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29404 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-28 22:51:01 +00:00
Evan Cheng
dc7a66a081 Clean up.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29378 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-28 01:19:22 +00:00
Evan Cheng
f2dfafcbc1 Rename IsFoldableBy to CanBeFoldedleBy
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29376 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-28 01:03:48 +00:00
Evan Cheng
2641cad180 Remove InFlightSet hack. No longer needed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29373 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-28 00:47:19 +00:00
Evan Cheng
3aac4d58e4 Remove dead code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29359 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-27 19:59:34 +00:00
Evan Cheng
b2c121ac14 Let each target specific isel provide routine to check if a chain producing node is foldable by another.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29335 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-27 06:36:49 +00:00
Evan Cheng
233baf1c4b Fix for bug 840. Only use noinline attribute if gcc version >= 3.4
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29311 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-26 23:06:27 +00:00
Evan Cheng
e6389935a9 Removed a hack intended to allow (store (op (load))) folding. Will handle this with preprocessing.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29258 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-21 22:19:51 +00:00
Evan Cheng
eb1f40d3af Also checks for noResults field.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29235 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-20 23:36:20 +00:00
Evan Cheng
e6f32034db Add code size to target instruction use it as the 3rd isel sorting tie-breaker.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29193 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-19 00:24:41 +00:00
Evan Cheng
fb878a37ba Use __attribute__((noinline)) only if compiled by gcc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29161 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-16 06:14:37 +00:00
Evan Cheng
f872940dfc Parameterize target node ValueType to allow more sharing of emit functions.
Also reduce the number of arguments passed to emit functions and removed a
hack.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29160 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-16 06:12:52 +00:00
Evan Cheng
fceb57a917 Reduce instruction selection code size and stack frame size by factoring
code that emit target specific nodes into emit functions that are uniquified
and shared among selection routines.
e.g. This reduces X86ISelDAGToDAG.o (release) from ~2M to ~1.5M. Stack frame
size of Select_store from ~13k down to ~8k.
This is the first step. Further work to enable more sharing will follow.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29158 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-15 08:45:20 +00:00
Jim Laskey
16d42c6ac6 It was pointed out that DEBUG() is only available with -debug.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29106 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-11 18:25:13 +00:00
Jim Laskey
e37fe9b3a1 Ensure that dump calls that are associated with asserts are removed from
non-debug build.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29105 91177308-0d34-0410-b5e6-96231b3b80d8
2006-07-11 17:58:07 +00:00
Evan Cheng
322812e603 Ugly hack! Add helper functions InsertInFlightSetEntry and
RemoveInFlightSetEntry. They are used in place of direct set operators to
reduce instruction selection function stack size.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28987 91177308-0d34-0410-b5e6-96231b3b80d8
2006-06-29 23:57:05 +00:00
Chris Lattner
02cdb377ec Fix an error message regression. Print:
LI8:    (LI8:i64 (imm:i64):$imm)
instead of:
  LI8:    (LI8:MVT::i64 (imm:MVT::i64):$imm)


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28868 91177308-0d34-0410-b5e6-96231b3b80d8
2006-06-20 00:56:37 +00:00
Chris Lattner
186fb7d131 Don't require src/dst patterns to be able to fully resolve their types,
because information about one can help refine the other.  This allows us to
write:

def : Pat<(i32 (extload xaddr:$src, i8)),
          (LBZX xaddr:$src)>;

as:

def : Pat<(extload xaddr:$src, i8),
          (LBZX xaddr:$src)>;

because tblgen knows LBZX returns i32.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28865 91177308-0d34-0410-b5e6-96231b3b80d8
2006-06-20 00:31:27 +00:00
Chris Lattner
a354849c27 Make sure to use the result of the pattern to infer the result type of the
instruction, and the result type of the instruction to refine the pattern.
This allows us to write things like this:

def : Pat<(v2i64 (bitconvert (v16i8 VR128:$src))), (v2i64 VR128:$src)>;

as:
def : Pat<(v2i64 (bitconvert (v16i8 VR128:$src))), (VR128:$src)>

and fixes a ppc64 issue.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28863 91177308-0d34-0410-b5e6-96231b3b80d8
2006-06-20 00:18:02 +00:00
Chris Lattner
8f60d54fb1 Improve a comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28833 91177308-0d34-0410-b5e6-96231b3b80d8
2006-06-16 18:25:06 +00:00
Evan Cheng
d7c2c86239 Allow more use of iPTR in patterns.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28790 91177308-0d34-0410-b5e6-96231b3b80d8
2006-06-15 00:16:37 +00:00
Evan Cheng
e945f4d3d4 Added support for variable_ops.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28788 91177308-0d34-0410-b5e6-96231b3b80d8
2006-06-14 22:22:20 +00:00
Evan Cheng
bc6b86a89e Fix support for optional input flag.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28784 91177308-0d34-0410-b5e6-96231b3b80d8
2006-06-14 19:27:50 +00:00
Evan Cheng
50c997e26a getOperandNum(): error if specified operand number is out of range.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28775 91177308-0d34-0410-b5e6-96231b3b80d8
2006-06-13 21:47:27 +00:00
Chris Lattner
11bcd28dff Wrap to 80 cols
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28743 91177308-0d34-0410-b5e6-96231b3b80d8
2006-06-09 23:59:44 +00:00
Evan Cheng
553ef1be94 Can't trust NodeDepth when checking for possibility of load folding creating
a cycle. This increase the search space and will increase compile time (in
practice it appears to be small, e.g. 176.gcc goes from 62 sec to 65 sec)
that will be addressed later.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28476 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-25 20:16:55 +00:00
Evan Cheng
b0793f9741 Fixed a really ugly bug. The TableGen'd isel is not freeing the "inflight set"
correctly. That is causing non-deterministic behavior (and possibly preventing
some load folding from happening).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28458 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-25 00:21:44 +00:00
Evan Cheng
d15531b430 Now that iPTR is a fully resolved type. We end up losing the type check for
patterns that look like this:

def : Pat<(i32 (X86Wrapper tconstpool  :$dst)), (MOV32ri tconstpool  :$dst)>;

InsertOneTypeCheck should copy the type from the resolved pattern to the
unresolved one as long as there types are different.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28389 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-19 07:24:32 +00:00
Evan Cheng
2618d07765 Remove PointerType from target definition. Use abstract type MVT::iPTR to
represent pointer type.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28363 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-17 20:37:59 +00:00
Evan Cheng
44a65fa552 Allow patterns to refer to physical registers that belong to multiple
register classes.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28323 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-16 07:05:30 +00:00
Evan Cheng
d9d4518a86 Unused instruction
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28240 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-12 07:42:01 +00:00
Evan Cheng
3eff89bf89 Watch out for the following case:
1. Use expects a chain output.
2. Node is expanded into multiple target ops.
3. One of the inner node produces a chain, the outer most one doesn't.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28209 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-10 02:47:57 +00:00
Evan Cheng
045953c37a Fix a load folding bug. It is exposed by a multi- resulting instructions
def : Pat<> pattern.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28208 91177308-0d34-0410-b5e6-96231b3b80d8
2006-05-10 00:05:46 +00:00
Evan Cheng
55d0fa1bfa Remove the temporary option: -no-isel-fold-inflight
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28012 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-28 18:54:11 +00:00
Evan Cheng
61a0209a8a When isel'ing a node, mark its operands "InFlight" before selecting them. These
nodes should not be folded into other nodes.
This fixes the miscompilation of PR 749.
Temporarily under flag control.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@28002 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-28 02:08:10 +00:00
Nate Begeman
37efe67645 JumpTable support! What this represents is working asm and jit support for
x86 and ppc for 100% dense switch statements when relocations are non-PIC.
This support will be extended and enhanced in the coming days to support
PIC, and less dense forms of jump tables.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27947 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-22 18:53:45 +00:00
Evan Cheng
c81d2a041d Rename AddedCost to AddedComplexity.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27841 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-19 20:36:09 +00:00
Evan Cheng
5941320c0d Allow "let AddedCost = n in" to increase pattern complexity.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27834 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-19 18:07:24 +00:00
Chris Lattner
91ded08224 Infer element types for shuffle masks
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27456 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-06 20:36:51 +00:00
Chris Lattner
52793e2c41 rename a method, to avoid confusion with llvm intrinsics.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27455 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-06 20:19:52 +00:00
Chris Lattner
ffa4fdcd17 Allow bits init values to be used in patterns, turn them into ints.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27286 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-31 05:25:56 +00:00
Chris Lattner
8c06318821 Implement Regression/TableGen/DagDefSubst.ll
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27263 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-30 22:50:40 +00:00
Chris Lattner
b026e70462 Tblgen doesn't like multiple SDNode<> definitions that map to the sameenum value. Split them into separate enums.
Also, don't emit dynamic checks when we can compute them statically


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27202 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-28 00:41:33 +00:00
Chris Lattner
2c4e65d460 Print error messages like this:
tblgen: In STVEBX: Intrinsic 'llvm.ppc.altivec.stvebx' expects 3 operands, not 2 operands!

instead of like this:
tblgen: In STVEBX: Intrinsic 'intrinsic_void expects 3 operands, not 2 operands!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27185 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-27 22:21:18 +00:00
Chris Lattner
5c4c774c75 Add a missing check which cause an invalid .td file to crash tblgen
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27126 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-25 22:12:44 +00:00
Chris Lattner
9bf2d3e6d6 When failing selection for an intrinsic, print this:
Cannot yet select: intrinsic %llvm.ppc.altivec.lvx

instead of this:

Cannot yet select: 0x9b047e0: v4i32,ch = INTRINSIC 0x9b04540:1, 0x9b04710, 0x9b04790, 0x9b04540


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27110 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-25 06:47:53 +00:00
Chris Lattner
5a1df389a6 Change approach so that we get codegen for free for intrinsics. With this,
intrinsics that don't take pointer arguments now work.  For example, we can
compile this:

int test3( __m128d *A) {
  return _mm_movemask_pd(*A);
}
int test4( __m128 *A) {
  return _mm_movemask_ps(*A);
}

to this:

_test3:
        movl 4(%esp), %eax
        movapd (%eax), %xmm0
        movmskpd %xmm0, %eax
        ret
_test4:
        movl 4(%esp), %eax
        movaps (%eax), %xmm0
        movmskps %xmm0, %eax
        ret


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27090 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-24 23:10:39 +00:00
Chris Lattner
947604b0e2 fix 80 column violations
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27084 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-24 21:52:20 +00:00
Chris Lattner
550525e3cf Parse intrinsics correctly and perform type propagation. This doesn't currently
emit the code to select intrinsics, but that is next :)


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27082 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-24 21:48:51 +00:00
Evan Cheng
83e1a6ab95 Allow result node to be a simple leaf node. This enable bitconvert patterns
like this:
def : Pat<(v4i32 (bitconvert (v4f32 VR128:$src))), (v4i32 VR128:$src)>;


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26968 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-23 02:35:32 +00:00
Evan Cheng
3a7a14b0ae Don't forget to promote xform function to an explicit node for def : Pat<>
patterns.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26929 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-21 20:44:17 +00:00
Chris Lattner
4e2f54d99e minor code simplification
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26918 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-21 06:42:58 +00:00
Evan Cheng
863bf5afb0 The node wrapped in PatLeaf<> should be treated as a leaf even if it isn't
one, i.e. don't select it.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26909 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-20 22:53:06 +00:00
Evan Cheng
f26ba69725 It should be ok for a xform output type to be different from input type.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26899 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-20 08:09:17 +00:00
Evan Cheng
420132e1eb Copy matching pattern's output type info to instruction result pattern.
The instruction patterns do not contain enough information to resolve the
exact type of the destination if it of a generic vector type.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26892 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-20 06:04:09 +00:00
Chris Lattner
697f88456c Add a new SDTCisIntVectorOfSameSize type constraint
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26890 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-20 05:39:48 +00:00
Evan Cheng
ed66e855ec Temporary hack to enable more (store (op (load ...))) folding. This makes
it possible when a TokenFactor is between the load and store. But is still
missing some cases due to ordering issue.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26638 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-09 08:19:11 +00:00
Evan Cheng
6f8aaf2cb0 Don't generate silly matching code like this:
if (N1.getOpcode() == ISD::ADD &&
     ...)
   if (... &&
       (N1.getNumOperands() == 1 || !isNonImmUse(N1.Val, N10.Val))) &&
       ...)

TableGen knows N1 must have more than one operand.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26592 91177308-0d34-0410-b5e6-96231b3b80d8
2006-03-07 08:31:27 +00:00
Chris Lattner
fd105d464a Select inline asm memory operands.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26349 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-24 02:13:31 +00:00
Evan Cheng
fbad70808c Bump up pattern cost if the resulting instruction is marked
usesCustomDAGSchedInserter.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26282 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-18 02:33:09 +00:00
Evan Cheng
67212a05e6 Call InsertISelMapEntry rather than map insertion operator to prevent overly
aggrssive inlining. This reduces Select_store frame size from 24k to 10k.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26095 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-09 22:12:27 +00:00
Evan Cheng
d7805a741d Match getTargetNode() changes (now returns SDNode* instead of SDOperand).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26084 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-09 07:16:09 +00:00
Evan Cheng
34167215a8 Change Select() from
SDOperand Select(SDOperand N);
to
void Select(SDOperand &Result, SDOperand N);


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26067 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-09 00:37:58 +00:00
Evan Cheng
21ad392ee6 Hoist all SDOperand declarations within a Select_{opcode}() to the top level
to reduce stack memory usage. This is intended to work around the gcc bug.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26026 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-07 00:37:41 +00:00
Evan Cheng
7cd19d038f At the end of isel, select a replacement node for each handle that does not
have one. This can happen if a load's real uses are dead (i.e. they do not
have uses themselves).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26014 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-06 08:12:55 +00:00
Evan Cheng
024524fe9b Name change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26013 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-06 06:03:35 +00:00
Evan Cheng
0a83ed506c Handle HANDLENODE: just return itself.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26011 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-05 08:46:14 +00:00
Evan Cheng
e41bf82107 Allow more loads to be folded which were previously prevented from happening
due to ordering issue. i.e. they were selected for chain use first.
Now at load select time, check if it is being selected for a chain use and if
it has only a single real use. If so, return a HANDLENODE (with the load as
its operand) in its place and record it.
When it is folded or the load is selected for a real use, the isel records it
as the replacement for the HANDLENODE. The replacement is done when all nodes
are selected.
This scheme exposed a couple of problems where cycles can happen. (See comments
in EmitMatchCode() for descriptions of the problems and their workaround /
solutions.) These problems have been resolved with a small compile time
penality.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25995 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-05 06:43:12 +00:00
Evan Cheng
2216d8ad2f Re-committing the last bit of change. It shouldn't break PPC this time.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25982 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-05 05:22:18 +00:00
Chris Lattner
9ed44818c5 Temporarily revert the last change, which breaks PPC and other targets that
DO select things.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25970 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-04 09:23:06 +00:00
Evan Cheng
9c4815a036 Complex pattern's custom matcher should not call Select() on any operands.
Select them afterwards if it returns true.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25968 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-04 08:50:49 +00:00
Chris Lattner
3e1798086b node predicates add to the complexity of a pattern. This ensures that the
X86 backend attempts to match small-immediate versions of instructions before
the full size immediate versions.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25937 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-03 18:06:02 +00:00
Evan Cheng
e4a8a6e645 (store (op (load ...))) folding problem. In the generated matching code,
Chain is initially set to the chain operand of store node, when it reaches
load, if it matches the load then Chain is set to the chain operand of the
load.

However, if the matching code that follows this fails, isel moves on to the
next pattern but it does not restore Chain to the chain operand of the store.
So when it tries to match the next store / op / load pattern it would fail on
the Chain == load.getOperand(0) test.

The solution is for each chain operand to get a unique name. e.g. Chain10.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25931 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-03 06:22:41 +00:00
Evan Cheng
657416cfba If a pattern's root node is a constant, its size should be 3 rather than 2.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25870 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-01 06:06:31 +00:00
Chris Lattner
7859313bae simplify some code
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25791 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-29 20:01:35 +00:00
Andrew Lenharth
8e51773bb7 it is nice not to chop off bits for those blessed with lots of bits
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25766 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-29 05:22:37 +00:00
Andrew Lenharth
2cba57cb09 make the casts actually cast to the variable type
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25765 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-29 05:17:22 +00:00
Andrew Lenharth
5c6bdf71d4 start of the 64bit safety cleanup
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25764 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-29 05:07:04 +00:00
Chris Lattner
24e00a4ac1 Emit series of conditionals with &&, emitting stuff like this:
if (N1.getOpcode() == ISD::LOAD &&
        N1.hasOneUse() &&
        !CodeGenMap.count(N1.getValue(0)) &&
        !CodeGenMap.count(N1.getValue(1))) {

instead of this:

    if (N1.getOpcode() == ISD::LOAD) {
      if (N1.hasOneUse()) {
        if (!CodeGenMap.count(N1.getValue(0))) {
          if (!CodeGenMap.count(N1.getValue(1))) {


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25763 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-29 04:41:05 +00:00
Chris Lattner
8bc7472a0c Factor matching code that is common between patterns. This works around
GCC not jump-threading across this common code, and produces far nicer
output.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25762 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-29 04:25:26 +00:00
Chris Lattner
2bd4dd7668 Split out code generation from analysis from emission
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25759 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-29 02:57:39 +00:00
Chris Lattner
355408bcb7 move some code around, no change in the generated code
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25758 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-29 02:43:35 +00:00
Chris Lattner
67a202bc42 now that we have control over emission of the code, emit the code using nested
"if" statements (indenting it appropriately, of course) instead of using goto's.

This inverts the logic for all of the if statements, which makes things simpler
to understand in addition to making the generated code easier to read.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25757 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-28 20:43:52 +00:00
Chris Lattner
8a0604b523 Change PatternCodeEmitter to emit code into a buffer instead of emitting it
directly to the output file.  This makes things simple because the code doesn't
have to worry about indentation or the case when there is no goto.  It also
allows us to indent the code better without touching everything :)


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25756 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-28 20:31:24 +00:00
Chris Lattner
488580c1ed Fit to 80 columns, no functionality change
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25752 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-28 19:06:51 +00:00
Jeff Cohen
9b0ffca478 Teach tablegen to generate code that is VC++ warning-free.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25709 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-27 22:22:28 +00:00
Evan Cheng
c15d18c501 (store (op (load ...)), ...)
If store's chain operand is load, then use load's chain operand instead. If
it isn't (likely a TokenFactor), then do not allow the folding.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25708 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-27 22:13:45 +00:00
Chris Lattner
fabcb7abc5 Teach the dag selectors to select InlineAsm nodes.
Aren't we happy the pattern selectors are almost all gone?


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25666 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-26 23:08:55 +00:00
Evan Cheng
1feeeecf4c Another folding problem: if a node r/w chain or flag, don't fold it if it
has already been selected. The number of use check is not strong enough since
a node can be replaced with newly created target node. e.g. If the original
node has two uses, when it is selected for one of the uses it is replaced with
another. Each node now has a single use but isel still should not fold it.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25651 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-26 19:13:45 +00:00
Evan Cheng
8eab3608de Add a FIXME comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25635 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-26 02:13:31 +00:00
Evan Cheng
545977371a Incoming (and optional) flag bugs. They may be embedded inside a inner node of
a pattern. Also, nodes which take incoming flag should not be folded if it has
more than one use.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25627 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-26 00:22:25 +00:00
Evan Cheng
9789aaad63 Fix an optional in flag bug.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25590 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-24 20:46:50 +00:00
Evan Cheng
18667c4545 Optional InFlag was not being included in node.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25588 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-24 20:07:38 +00:00
Evan Cheng
76356d9594 Prevent folding of a node with multiple uses if the node already folds a load!
Here is an example where the load ended up being done twice:

%A = global uint 0

uint %test(uint %B, ubyte %C) {
	%tmp = load uint *%A;
	%X = shl uint %tmp, ubyte %C
	%Cv = sub ubyte 32, %C
	%Y = shr uint %B, ubyte %Cv
	%Z = or uint %Y, %X
	store uint %Z, uint* %A
	ret uint %Z
}

==>

	subl $4, %esp
	movl %ebx, (%esp)
	movl 8(%esp), %edx
	movl A, %eax
	movb 12(%esp), %bl
	movb %bl, %cl
	shldl %cl, %edx, %eax
	movb %bl, %cl
	shldl %cl, %edx, A
	movl (%esp), %ebx
	addl $4, %esp
	ret


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@25471 91177308-0d34-0410-b5e6-96231b3b80d8
2006-01-20 01:11:03 +00:00