Commit Graph

530 Commits

Author SHA1 Message Date
ddf89566a9 This commit changes:
1. Legalize now always promotes truncstore of i1 to i8. 
2. Remove patterns and gunk related to truncstore i1 from targets.
3. Rename the StoreXAction stuff to TruncStoreAction in TLI.
4. Make the TLI TruncStoreAction table a 2d table to handle from/to conversions.
5. Mark a wide variety of invalid truncstores as such in various targets, e.g.
   X86 currently doesn't support truncstore of any of its integer types.
6. Add legalize support for truncstores with invalid value input types.
7. Add a dag combine transform to turn store(truncate) into truncstore when
   safe.

The later allows us to compile CodeGen/X86/storetrunc-fp.ll to:

_foo:
	fldt	20(%esp)
	fldt	4(%esp)
	faddp	%st(1)
	movl	36(%esp), %eax
	fstps	(%eax)
	ret

instead of:

_foo:
	subl	$4, %esp
	fldt	24(%esp)
	fldt	8(%esp)
	faddp	%st(1)
	fstps	(%esp)
	movl	40(%esp), %eax
	movss	(%esp), %xmm0
	movss	%xmm0, (%eax)
	addl	$4, %esp
	ret



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@46140 91177308-0d34-0410-b5e6-96231b3b80d8
2008-01-17 19:59:44 +00:00
84bc5427d6 Rename SSARegMap -> MachineRegisterInfo in keeping with the idea
that "machine" classes are used to represent the current state of
the code being compiled.  Given this expanded name, we can start 
moving other stuff into it.  For now, move the UsedPhysRegs and
LiveIn/LoveOuts vectors from MachineFunction into it.

Update all the clients to match.

This also reduces some needless #includes, such as MachineModuleInfo
from MachineFunction.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45467 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-31 04:13:23 +00:00
4ee451de36 Remove attribution from file headers, per discussion on llvmdev.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@45418 91177308-0d34-0410-b5e6-96231b3b80d8
2007-12-29 20:36:04 +00:00
f1b1c5ed1a implement a trivial readme entry.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44380 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-27 22:36:16 +00:00
27a6c7380f Several changes:
1) Change the interface to TargetLowering::ExpandOperationResult to 
   take and return entire NODES that need a result expanded, not just
   the value.  This allows us to handle things like READCYCLECOUNTER,
   which returns two values.
2) Implement (extremely limited) support in LegalizeDAG::ExpandOp for MERGE_VALUES.
3) Reimplement custom lowering in LegalizeDAGTypes in terms of the new
   ExpandOperationResult.  This makes the result simpler and fully 
   general.
4) Implement (fully general) expand support for MERGE_VALUES in LegalizeDAGTypes.
5) Implement ExpandOperationResult support for ARM f64->i64 bitconvert and ARM
   i64 shifts, allowing them to work with LegalizeDAGTypes.
6) Implement ExpandOperationResult support for X86 READCYCLECOUNTER and FP_TO_SINT,
   allowing them to work with LegalizeDAGTypes.

LegalizeDAGTypes now passes several more X86 codegen tests when enabled and when
type legalization in LegalizeDAG is ifdef'd out.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44300 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-24 07:07:01 +00:00
0f8d9c04d9 Unify CALLSEQ_{START,END}. They take 4 parameters: the chain, two stack
adjustment fields, and an optional flag. If there is a "dynamic_stackalloc" in
the code, make sure that it's bracketed by CALLSEQ_START and CALLSEQ_END. If
not, then there is the potential for the stack to be changed while the stack's
being used by another instruction (like a call).

This can only result in tears...


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@44037 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-13 00:44:25 +00:00
e0cb36b9fb [ARM] Implement __builtin_thread_pointer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@43892 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-08 17:20:05 +00:00
f1ba1cad38 Move the LowerMEMCPY and LowerMEMCPYCall to a common place.
Thanks for the suggestions Bill :-)



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@43742 91177308-0d34-0410-b5e6-96231b3b80d8
2007-11-05 23:12:20 +00:00
e0703c84dd Make ARM and X86 LowerMEMCPY identical by moving the isThumb check into getMaxInlineSizeThreshold
and by restructuring the X86 version.

New I just have to move this to a common place :-)



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@43554 91177308-0d34-0410-b5e6-96231b3b80d8
2007-10-31 14:39:58 +00:00
fc05f402ea Make ARM an X86 memcpy expansion more similar to each other.
Now both subtarget define getMaxInlineSizeThreshold and the expansion uses it.

This should not change generated code.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@43552 91177308-0d34-0410-b5e6-96231b3b80d8
2007-10-31 11:52:06 +00:00
4102eb57bb Fix memcpy lowering when addresses are 4-byte aligned but size is not multiple of 4.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@43234 91177308-0d34-0410-b5e6-96231b3b80d8
2007-10-22 22:11:27 +00:00
7b73a5d6de split LowerMEMCPY into LowerMEMCPYCall and LowerMEMCPYInline in the ARM backend.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@43176 91177308-0d34-0410-b5e6-96231b3b80d8
2007-10-19 14:35:17 +00:00
65a3323b0a legalizing the ret operation on f64 shouldn't introduce a new
i64 bit convert needlessly.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@43116 91177308-0d34-0410-b5e6-96231b3b80d8
2007-10-18 06:17:07 +00:00
f96e4de403 Set ISD::FPOW to Expand.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@42881 91177308-0d34-0410-b5e6-96231b3b80d8
2007-10-11 23:21:31 +00:00
525178cdbf Migrate X86 and ARM from using X86ISD::{,I}DIV and ARMISD::MULHILO{U,S} to
use ISD::{S,U}DIVREM and ISD::{S,U}MUL_HIO. Move the lowering code
associated with these operators into target-independent in LegalizeDAG.cpp
and TargetLowering.cpp.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@42762 91177308-0d34-0410-b5e6-96231b3b80d8
2007-10-08 18:33:35 +00:00
f7331b3dd7 Fold the adjust_trampoline intrinsic into
init_trampoline.  There is now only one
trampoline intrinsic.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@41841 91177308-0d34-0410-b5e6-96231b3b80d8
2007-09-11 14:10:23 +00:00
eaf089430e Enhance APFloat to retain bits of NaNs (fixes oggenc).
Use APFloat interfaces for more references, mostly
of ConstantFPSDNode.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@41632 91177308-0d34-0410-b5e6-96231b3b80d8
2007-08-31 04:03:46 +00:00
36397f5034 Support for trampolines, except for X86 codegen which is
still under discussion.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40549 91177308-0d34-0410-b5e6-96231b3b80d8
2007-07-27 12:58:54 +00:00
a99be51bf5 Here is the bulk of the sanitizing.
Almost all occurrences of "bytecode" in the sources have been eliminated.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37913 91177308-0d34-0410-b5e6-96231b3b80d8
2007-07-05 17:07:56 +00:00
0e1d37904a Reflects the chanegs made to PredicateOperand.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37898 91177308-0d34-0410-b5e6-96231b3b80d8
2007-07-05 07:18:20 +00:00
e2446c6076 Silence a warning.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37737 91177308-0d34-0410-b5e6-96231b3b80d8
2007-06-26 18:31:22 +00:00
ea859be53c Move ComputeMaskedBits, MaskedValueIsZero, and ComputeNumSignBits from
TargetLowering to SelectionDAG so that they have more convenient
access to the current DAG, in preparation for the ValueType routines
being changed from standalone functions to members of SelectionDAG for
the pre-legalize vector type changes.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37704 91177308-0d34-0410-b5e6-96231b3b80d8
2007-06-22 14:59:07 +00:00
97e604e7d8 Be more conservative of duplicating blocks.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37669 91177308-0d34-0410-b5e6-96231b3b80d8
2007-06-19 23:55:02 +00:00
277f0741c5 Allow predicated immediate ARM to ARM calls.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37659 91177308-0d34-0410-b5e6-96231b3b80d8
2007-06-19 21:05:09 +00:00
144fd1ff0f Set ARM ifcvt duplication limit to 3 for now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37385 91177308-0d34-0410-b5e6-96231b3b80d8
2007-06-01 08:28:59 +00:00
e5e7ce458a Silence some compilation warnings.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37197 91177308-0d34-0410-b5e6-96231b3b80d8
2007-05-18 01:19:57 +00:00
9f8cbd147c Set ARM if-conversion block size threshold to 10 instructions for now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37194 91177308-0d34-0410-b5e6-96231b3b80d8
2007-05-18 00:19:34 +00:00
8dd86c14d4 More effective breakdown of memcpy into repeated load/store. These are now
in the order lod;lod;lod;sto;sto;sto which means the load-store optimizer
has a better chance of producing ldm/stm.  Ideally you would get cooperation
from the RA as well but this is not there yet.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37179 91177308-0d34-0410-b5e6-96231b3b80d8
2007-05-17 21:31:21 +00:00
5d3d44a848 Fix previous patch. GOTOFF can be used only when the symbol has internal
linkage or hidden visibility.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37055 91177308-0d34-0410-b5e6-96231b3b80d8
2007-05-14 23:20:21 +00:00
930d161ba2 Optimize PIC implementation. GOTOFF can be used when the symbol is defined
and used in the same module.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@37044 91177308-0d34-0410-b5e6-96231b3b80d8
2007-05-14 18:46:23 +00:00
97c9bb5cc6 On Mac OS X, GV requires an extra load only when relocation-model is non-static.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36718 91177308-0d34-0410-b5e6-96231b3b80d8
2007-05-04 00:26:58 +00:00
e8e5495474 Debug support for arm-linux.
Patch by Raul Herbster.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36690 91177308-0d34-0410-b5e6-96231b3b80d8
2007-05-03 20:28:35 +00:00
0b0a9a90a4 Typo. It's checking if V is multiple of 4, not multiple of 3. :-)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36663 91177308-0d34-0410-b5e6-96231b3b80d8
2007-05-03 02:00:18 +00:00
64f4fa5e0e ARM TLS: implement "general dynamic", "initial exec" and "local exec" models.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36506 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-27 13:54:47 +00:00
b1df8f2750 Darwin runtime library does not have these.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36505 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-27 08:15:43 +00:00
0ae4a3357a Implement PIC for arm-linux.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36324 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-22 00:04:12 +00:00
5a3d40d88f arm has r+r*s and r+i addr modes, but no r+i+r*s addr modes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35962 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-13 06:50:55 +00:00
e11529438c Fix incorrect fall-throughs in addr mode code. This fixes CodeGen/ARM/arm-negative-stride.ll
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35909 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-11 16:17:12 +00:00
eb13d1b710 restore support for negative strides
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35859 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-10 03:48:29 +00:00
37caf8c68e remove dead target hooks
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35846 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-09 23:33:39 +00:00
b445d0cbb9 remove some dead target hooks, subsumed by isLegalAddressingMode
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35840 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-09 22:27:04 +00:00
0a7baa24d1 Typo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35639 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-04 00:06:07 +00:00
b2c594f350 Arm supports negative strides as well, add them. This lets us compile:
CodeGen/ARM/arm-negative-stride.ll to:

LBB1_2: @bb
        str r1, [r3, -r0, lsl #2]
        add r0, r0, #1
        cmp r0, r2
        bne LBB1_2      @bb


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35609 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-03 00:13:57 +00:00
6e0784da77 fix the CodeGen/ARM/2007-03-13-InstrSched.ll regression: allow IV's with scales
to be folded into non-store instructions.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35601 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-02 18:51:18 +00:00
c4e3f8e736 add support for the 'w' inline asm register class.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35598 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-02 17:24:08 +00:00
9996663fc6 - Divides the comparisons in two types: comparisons that only use N and Z
flags (ARMISD::CMPNZ) and comparisons that use all flags (ARMISD::CMP).
- Defines the instructions: TST, TEQ (ARM) and TST (Thumb).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35573 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-02 01:30:03 +00:00
3074d9df96 Add i16 address mode.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35551 91177308-0d34-0410-b5e6-96231b3b80d8
2007-04-01 08:06:46 +00:00
c9addb7488 implement the new addressing mode description hook.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35521 91177308-0d34-0410-b5e6-96231b3b80d8
2007-03-30 23:15:24 +00:00
caaf69107e Remove isLegalAddressImmediate.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35406 91177308-0d34-0410-b5e6-96231b3b80d8
2007-03-28 01:53:55 +00:00
b8a93a45f8 bugfix: sometimes the spiller puts a load between the "mov lr, pc" and "bx" of a CALL_NOLINK.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35381 91177308-0d34-0410-b5e6-96231b3b80d8
2007-03-27 16:19:21 +00:00