Commit Graph

2008 Commits

Author SHA1 Message Date
Chris Lattner
bfd68a7858 Turn X < 0 -> TEST X,X js
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30294 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-13 17:04:54 +00:00
Chris Lattner
7a6366de86 The sense of this branch was inverted :(
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30293 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-13 16:56:12 +00:00
Chris Lattner
0f27fc34f7 Fix a regression in the 32-bit port from the 64-bit port landing.
We now compile CodeGen/X86/lea-2.ll into:

_test:
        movl 4(%esp), %eax
        movl 8(%esp), %ecx
        leal -5(%ecx,%eax,4), %eax
        ret

instead of:

_test:
        movl 4(%esp), %eax
        leal (,%eax,4), %eax
        addl 8(%esp), %eax
        addl $4294967291, %eax
        ret


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30288 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-13 04:45:25 +00:00
Chris Lattner
d1468d3332 new note
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30286 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-13 04:19:50 +00:00
Chris Lattner
95af34e33f new note
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30285 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-13 03:54:54 +00:00
Chris Lattner
f95705163f Compile X > -1 -> text X,X; js dest
This implements CodeGen/X86/jump_sign.ll.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30283 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-13 03:22:10 +00:00
Evan Cheng
c356a572e3 Reflects MachineConstantPoolEntry changes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30279 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-12 21:04:05 +00:00
Chris Lattner
adaa38f13d add a note
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30271 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-12 06:36:01 +00:00
Chris Lattner
0bfd7fde43 Testcase noticed from PR906
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30269 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-11 23:00:56 +00:00
Chris Lattner
2a33a3f08d add compilable testcase
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30268 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-11 22:57:51 +00:00
Evan Cheng
f47d167c3b Updates.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30245 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-11 05:35:17 +00:00
Evan Cheng
0f4aa6ee20 Update README file.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30244 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-11 05:25:15 +00:00
Evan Cheng
734503be59 X86ISD::CMP now produces a chain as well as a flag. Make that the chain
operand of a conditional branch to allow load folding into CMP / TEST
instructions.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30241 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-11 02:19:56 +00:00
Evan Cheng
6f34b43292 Fixed a FuseTwoAddrInst() bug: consider GlobalAddress and JumpTableIndex
in addition to immediate operands.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30205 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-08 21:08:13 +00:00
Chris Lattner
40f4ba5e71 Use __USER_LABEL_PREFIX__ to get the prefix added by the current host.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30190 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-08 17:03:56 +00:00
Evan Cheng
36978b9c61 Remove TEST64mr. It's same as TEST64rm since and is commutative.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30178 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-08 06:56:55 +00:00
Evan Cheng
25ab690a43 Committing X86-64 support.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30177 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-08 06:48:29 +00:00
Evan Cheng
8cf723d8eb - Identify a vector_shuffle that can be turned into an undef, e.g.
shuffle V1, <undef>, <undef, undef, 4, 5>
- Fix some suspicious logic into LowerVectorShuffle that cause less than
  optimal code by failing to identify MOVL (move to lowest element of a
  vector).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30171 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-08 01:50:06 +00:00
Jim Laskey
fde1b3bb2f 1. Remove condition on delete.
2. Protect and outline createTargetAsmInfo.

3. Misc. kruft.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30169 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-07 23:39:26 +00:00
Chris Lattner
2dd538c2d2 add a new value for the command line optn
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30165 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-07 22:32:28 +00:00
Chris Lattner
cdb341dcfa Fix a cross-build issue. The asmsyntax shouldn't be affected by the build
host, it should be affected by the target.  Allow the command line option to
override in either case.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30164 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-07 22:29:41 +00:00
Jim Laskey
a0f3d17daa Make target asm info a property of the target machine.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30162 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-07 22:06:40 +00:00
Jim Laskey
8e8de8f776 Break out target asm info into separate files.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30161 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-07 22:05:02 +00:00
Chris Lattner
ffc0b2663e Eliminate X86ISD::TEST, using X86ISD::CMP instead. Match X86ISD::CMP patterns
using test, which provides nice simplifications like:

-       movl %edi, %ecx
-       andl $2, %ecx
-       cmpl $0, %ecx
+       testl $2, %edi
        je LBB1_11      #cond_next90

There are a couple of dagiselemitter deficiencies that this exposes, they will
be handled later.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30156 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-07 20:33:45 +00:00
Chris Lattner
b14ca605d7 Some notes on better load folding we could do
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30155 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-07 20:32:01 +00:00
Evan Cheng
ec3bc39413 Consistency.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30152 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-07 19:03:48 +00:00
Jim Laskey
05a059d5d8 Make the x86 asm flavor part of the subtarget info.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30146 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-07 12:23:47 +00:00
Evan Cheng
95971c51b0 Clean up.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30140 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-07 01:17:57 +00:00
Evan Cheng
48ff2fb129 Watch out for variable_ops instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30135 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-06 20:32:45 +00:00
Evan Cheng
3530bafe00 Variable ops instructions may ignore the last few operands for code emission.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30134 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-06 20:24:14 +00:00
Jim Laskey
563321a258 Separate target specific asm properties from the asm printers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30126 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-06 18:34:40 +00:00
Chris Lattner
6cdb1ea610 Bugfix to work with the two-addr changes that have been made in the tree recently
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30121 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-05 20:27:32 +00:00
Evan Cheng
2f5993bcf1 Fix a few dejagnu failures. e.g. fast-cc-merge-stack-adj.ll
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30113 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-05 08:32:49 +00:00
Evan Cheng
23b3122c44 JIT encoding bug.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30112 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-05 05:59:25 +00:00
Chris Lattner
0e42d81a83 Update the X86 JIT to make it work with the new two-addr changes. This also
adds assertions that check to make sure every operand gets emitted.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30110 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-05 02:52:35 +00:00
Chris Lattner
09e460662a Completely eliminate def&use operands. Now a register operand is EITHER a
def operand or a use operand.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30109 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-05 02:31:13 +00:00
Chris Lattner
2926869b4a Fix a long-standing wart in the code generator: two-address instruction lowering
actually *removes* one of the operands, instead of just assigning both operands
the same register.  This make reasoning about instructions unnecessarily complex,
because you need to know if you are before or after register allocation to match
up operand #'s with the target description file.

Changing this also gets rid of a bunch of hacky code in various places.

This patch also includes changes to fold loads into cmp/test instructions in
the X86 backend, along with a significant simplification to the X86 spill
folding code.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30108 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-05 02:12:02 +00:00
Chris Lattner
2130b99eb2 Fix some X86 JIT failures. This should really come from TargetJITInfo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30102 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-04 18:48:41 +00:00
Chris Lattner
1911fd4f85 Completely rearchitect the interface between targets and the pass manager.
This pass:

1. Splits TargetMachine into TargetMachine (generic targets, can be implemented
any way, like the CBE) and LLVMTargetMachine (subclass of TM that is used by
things using libcodegen and other support).
2. Instead of having each target fully populate the passmgr for file or JIT
   output, move all this to common code, and give targets hooks they can
   implement.
3. Commonalize the target population stuff between file emission and JIT
   emission.
4. All (native code) codegen stuff now happens in a FunctionPassManager, which
   paves the way for "fast -O0" stuff in the CFE later, and now LLC could
   lazily stream .bc files from disk to use less memory.
5. There are now many fewer #includes and the targets don't depend on the
   scalar xforms or libanalysis anymore (but codegen does).
6. Changing common code generator pass ordering stuff no longer requires
   touching all targets.
7. The JIT now has the option of "-fast" codegen or normal optimized codegen,
   which is now orthogonal to the fact that JIT'ing is being done.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30081 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-04 04:14:57 +00:00
Chris Lattner
dd842e12e1 Add accessor
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30080 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-04 04:08:58 +00:00
Chris Lattner
c4fa386471 Simplify target construction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30070 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-03 18:44:02 +00:00
Evan Cheng
1453de5aa9 Oops. Bad typo. Without the check of N1.hasOneUse() bad things can happen.
Suppose the TokenFactor can reach the Op:

       [Load chain]
           ^
           |
         [Load]
         ^    ^
         |    |
        /      \-
       /         |
      /          [Op]
     /          ^ ^
     |        ..  |
     |       /    |
   [TokenFactor]  |
       ^          |
       |          |
        \        /
         \      /
         [Store]

If we move the Load below the TokenFactor, we would have created a cycle in
the DAG.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@30040 91177308-0d34-0410-b5e6-96231b3b80d8
2006-09-01 22:52:28 +00:00
Evan Cheng
c9676deb23 Minor asm fix.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29965 91177308-0d34-0410-b5e6-96231b3b80d8
2006-08-29 22:14:48 +00:00
Evan Cheng
4a888584ea Remove dead code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29962 91177308-0d34-0410-b5e6-96231b3b80d8
2006-08-29 21:42:58 +00:00
Evan Cheng
780413d462 Don't performance load/op/store transformation if op produces a floating point
or vector result. X86 does not have load/mod/store variants of those
instructions.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29957 91177308-0d34-0410-b5e6-96231b3b80d8
2006-08-29 18:37:37 +00:00
Evan Cheng
e50794adda - Enable x86 isel preprocessing by default unless -fast is specified.
- Also disable isel load folding if -fast.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29956 91177308-0d34-0410-b5e6-96231b3b80d8
2006-08-29 18:28:33 +00:00
Evan Cheng
82a35b34fb Avoid making unneeded load/mod/store transformation which can hurt performance.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29952 91177308-0d34-0410-b5e6-96231b3b80d8
2006-08-29 06:44:17 +00:00
Evan Cheng
67afeceaf0 On Mac, print jump table entries after the function to work around a linker issue.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29946 91177308-0d34-0410-b5e6-96231b3b80d8
2006-08-28 22:14:16 +00:00
Evan Cheng
70e674e95e Add an optional pass to preprocess the DAG before x86 isel to allow selecting more load/mod/store instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29943 91177308-0d34-0410-b5e6-96231b3b80d8
2006-08-28 20:10:17 +00:00
Chris Lattner
a4f0b3a084 s|llvm/Support/Visibility.h|llvm/Support/Compiler.h|
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@29911 91177308-0d34-0410-b5e6-96231b3b80d8
2006-08-27 12:54:02 +00:00