Commit Graph

7182 Commits

Author SHA1 Message Date
Dale Johannesen 82810c8a13 Make tail merging the default, except on powerPC. There was no prior art
for a target-dependent default with a command-line override; this way
should be generally usable.

llvm-svn: 37285
2007-05-22 17:14:46 +00:00
Bill Wendling 3fb7fdfded We only need to specify the most-implied feature for an architecture.
llvm-svn: 37275
2007-05-22 05:15:37 +00:00
Evan Cheng 8c8afb27d7 Fix some -march=thumb regressions. tBR_JTr is not predicable.
llvm-svn: 37272
2007-05-21 23:17:32 +00:00
Dale Johannesen d1de276c16 Use AXI3 not AXI2 for appropriate PIC PC-relative loads and stores. Cosmetic.
llvm-svn: 37271
2007-05-21 22:42:04 +00:00
Dale Johannesen 7d55f3733e Add some patterns for PIC PC-relative loads and stores.
llvm-svn: 37269
2007-05-21 22:14:33 +00:00
Evan Cheng 147b334b6a BlockHasNoFallThrough() now returns true if block ends with a return instruction; AnalyzeBranch() should ignore predicated instructionsd.
llvm-svn: 37268
2007-05-21 18:56:31 +00:00
Evan Cheng fc94eb66d2 BlockHasNoFallThrough() now returns true if block ends with a return instruction.
llvm-svn: 37266
2007-05-21 18:44:17 +00:00
Dan Gohman c12dd5207d Apply this patch:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070514/049845.html

llvm-svn: 37240
2007-05-18 23:21:46 +00:00
Chris Lattner 7ea2df6e2a add a note
llvm-svn: 37239
2007-05-18 20:18:14 +00:00
Dan Gohman eefa83e67b Use MVT::FIRST_VECTOR_VALUETYPE and MVT::LAST_VECTOR_VALUETYPE.
llvm-svn: 37234
2007-05-18 18:44:07 +00:00
Evan Cheng 4ae1840d21 Mark calls non-predicable for now. Need to ensure it's the last instruction in the if-converted block or make sure it preserve condition code.
llvm-svn: 37199
2007-05-18 01:53:54 +00:00
Evan Cheng ea623560f8 Silence some compilation warnings.
llvm-svn: 37197
2007-05-18 01:19:57 +00:00
Evan Cheng 6addd65914 Set ARM if-conversion block size threshold to 10 instructions for now.
llvm-svn: 37194
2007-05-18 00:19:34 +00:00
Evan Cheng e20dd92792 RemoveBranch() and InsertBranch() now returns number of instructions deleted / inserted.
llvm-svn: 37193
2007-05-18 00:18:17 +00:00
Evan Cheng 99be49dd9b RemoveBranch() and InsertBranch() now returns number of instructions deleted / inserted.
llvm-svn: 37192
2007-05-18 00:05:48 +00:00
Dale Johannesen 58698d2534 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.

llvm-svn: 37179
2007-05-17 21:31:21 +00:00
Evan Cheng afa1cb6da3 Fix a bogus check that prevented folding VECTOR_SHUFFLE to UNDEF; add an optimization to fold VECTOR_SHUFFLE to a zero vector.
llvm-svn: 37173
2007-05-17 18:45:50 +00:00
Evan Cheng 632c3f01ed Added missing patterns for UNPCKH* and PUNPCKH*.
llvm-svn: 37172
2007-05-17 18:44:37 +00:00
Chris Lattner dade607f19 This is the correct fix for PR1427. This fixes mmx-shuffle.ll and doesn't
cause other regressions.

llvm-svn: 37160
2007-05-17 17:13:13 +00:00
Anton Korobeynikov 1ad4618715 Revert patch for PR1427. It breaks almost all vector tests.
llvm-svn: 37159
2007-05-17 07:50:14 +00:00
Chris Lattner 3e549e9d5f add support for 128-bit add/sub on ppc64
llvm-svn: 37158
2007-05-17 06:52:46 +00:00
Chris Lattner 13f4bf5c5e add support for 128-bit integer add/sub
llvm-svn: 37154
2007-05-17 06:35:11 +00:00
Chris Lattner 6a5a46322f Fix PR1427 and test/CodeGen/X86/mmx-shuffle.ll
llvm-svn: 37141
2007-05-17 03:29:42 +00:00
Evan Cheng 2db22024cf Remove. Not needed.
llvm-svn: 37139
2007-05-17 00:11:35 +00:00
Evan Cheng 733b4bd8ae Default implementation of TargetInstrInfo::getBlockSize().
llvm-svn: 37138
2007-05-16 23:53:44 +00:00
Evan Cheng 1634e7186b ARM::tB is also predicable.
llvm-svn: 37125
2007-05-16 21:53:43 +00:00
Evan Cheng dcff2eb0e8 PredicateInstruction returns true if the operation was successful.
llvm-svn: 37124
2007-05-16 21:53:07 +00:00
Evan Cheng 973c3739b0 Add default implementation of PredicateInstruction().
llvm-svn: 37123
2007-05-16 21:20:37 +00:00
Evan Cheng 4423687831 Move if-conversion after all passes that may use register scavenger.
llvm-svn: 37120
2007-05-16 20:52:46 +00:00
Evan Cheng e2762c3d68 Removed isPredicable().
llvm-svn: 37119
2007-05-16 20:50:23 +00:00
Evan Cheng dcd6cdf896 Make ARM::B isPredicable; Make Bcc and MOVCC condition option a normal operand so they are not predicable.
llvm-svn: 37118
2007-05-16 20:50:01 +00:00
Evan Cheng cc33218607 Added isPredicable bit to class Instruction.
llvm-svn: 37117
2007-05-16 20:47:01 +00:00
Evan Cheng 01a4227ed1 Conditional branch is not a barrier.
llvm-svn: 37103
2007-05-16 07:45:54 +00:00
Chris Lattner 888653cdba implement the missing maskmovq mmx intrinsic that akor hit.
llvm-svn: 37100
2007-05-16 06:08:17 +00:00
Evan Cheng c95f95b6da Fix comment.
llvm-svn: 37098
2007-05-16 05:14:06 +00:00
Evan Cheng ad3aac71ce Hooks for predication support.
llvm-svn: 37093
2007-05-16 02:01:49 +00:00
Chris Lattner 0b7472da6f fix some subtle inline asm selection issues
llvm-svn: 37067
2007-05-15 01:31:05 +00:00
Evan Cheng 0f7cbe8370 Add PredicateOperand to all ARM instructions that have the condition field.
llvm-svn: 37066
2007-05-15 01:29:07 +00:00
Chris Lattner c8798d085c fix subtle bugs in inline asm operand selection
llvm-svn: 37065
2007-05-15 01:28:08 +00:00
Lauro Ramos Venancio 1b8d46ab21 Fix previous patch. GOTOFF can be used only when the symbol has internal
linkage or hidden visibility.

llvm-svn: 37055
2007-05-14 23:20:21 +00:00
Lauro Ramos Venancio d705f5d51d Optimize PIC implementation. GOTOFF can be used when the symbol is defined
and used in the same module.

llvm-svn: 37044
2007-05-14 18:46:23 +00:00
Lauro Ramos Venancio 3551928a2b Enable aliases on arm-linux.
llvm-svn: 37042
2007-05-14 18:32:56 +00:00
Chris Lattner 1904e45d0c Fix PR1413
llvm-svn: 37023
2007-05-13 22:19:27 +00:00
Anton Korobeynikov 13da17843c More DWARF-related things cleanup:
1. Fix PR1380
2. Apply Duncan's patch from PR1410
3. Insert workaround for "one personality function per module" as noted in PR1414
4. Emit correct debug frames for x86/linux. This partly fixes DebugInfo/2006-11-06-StackTrace.cpp: stack trace is
shown correctly, but arguments for function on top of stack are displayed incorrectly.

llvm-svn: 37015
2007-05-12 22:36:25 +00:00
Chris Lattner 623c738fe9 add some notes
llvm-svn: 36965
2007-05-10 00:08:04 +00:00
Evan Cheng 9c031c0ddf Switch BCC, MOVCCr, etc. to PredicateOperand.
llvm-svn: 36948
2007-05-08 21:08:43 +00:00
Evan Cheng d194a8603d PredicateOperand can be used as a normal operand for isel.
llvm-svn: 36947
2007-05-08 21:06:08 +00:00
Evan Cheng f50ba31ec6 R0 is a sub-register of X0, etc.
llvm-svn: 36939
2007-05-08 17:03:51 +00:00
Lauro Ramos Venancio 744350b131 Fix PR1390 in a better way.
llvm-svn: 36916
2007-05-07 23:15:16 +00:00
Evan Cheng d37c23745f This is no longer needed after enabling the DAG combiner xform.
llvm-svn: 36909
2007-05-07 21:29:41 +00:00
Chris Lattner 5992f4709c add this back
llvm-svn: 36892
2007-05-06 20:31:17 +00:00
Anton Korobeynikov c892cf466d Update MSIL BE. This patch fixes most weird glitches outlined in
README.txt. Patch by Roman Samoilov!

llvm-svn: 36890
2007-05-06 20:13:33 +00:00
Bill Wendling f985c492e1 3DNowA implies 3DNow. 64-bit implies SSE1, SSE2, and I assume MMX.
llvm-svn: 36860
2007-05-06 07:56:19 +00:00
Nate Begeman 4060c7ac63 Reference correct header
llvm-svn: 36834
2007-05-06 04:00:55 +00:00
Lauro Ramos Venancio 34b2735f20 Fix PR1390.
Don't spill extra register to align the stack.

llvm-svn: 36814
2007-05-05 23:44:41 +00:00
Chris Lattner 9958b82733 add a note
llvm-svn: 36811
2007-05-05 22:44:08 +00:00
Chris Lattner 2cca31f77f the mason example is implemented. Move some examples out of llvm/test,
upgrade the syntax of some other examples.

llvm-svn: 36806
2007-05-05 22:29:06 +00:00
Chris Lattner cc43c3fe6b implement anyextend from i1 -> i64
llvm-svn: 36802
2007-05-05 22:17:00 +00:00
Chris Lattner be8f99ecbb move CodeGen/X86/overlap-add.ll here.
llvm-svn: 36799
2007-05-05 22:10:24 +00:00
Anton Korobeynikov 4db0090339 Emit sections/directives in the proper order. This fixes PR1376. Also,
some small cleanup was made.

llvm-svn: 36780
2007-05-05 09:04:50 +00:00
Lauro Ramos Venancio 8f8199086f Add a processor.
llvm-svn: 36765
2007-05-04 22:16:30 +00:00
Bill Wendling e6182267d7 Add an "implies" field to features. This indicates that, if the current
feature is set, then the features in the implied list should be set also.
The opposite is also enforced: if a feature in the implied list isn't set,
then the feature that owns that implies list shouldn't be set either.

llvm-svn: 36756
2007-05-04 20:38:40 +00:00
Evan Cheng 33c9886001 On Mac OS X, GV requires an extra load only when relocation-model is non-static.
llvm-svn: 36718
2007-05-04 00:26:58 +00:00
Evan Cheng 23040754b0 Should never see an indexed load / store with zero offset.
llvm-svn: 36714
2007-05-03 23:30:36 +00:00
Dale Johannesen 89200ce0f0 Evan's patch to avoid FPreg->intreg copy for cvt; store to mem
llvm-svn: 36693
2007-05-03 20:54:42 +00:00
Lauro Ramos Venancio 83930198dd Debug support for arm-linux.
Patch by Raul Herbster.

llvm-svn: 36690
2007-05-03 20:28:35 +00:00
Dan Gohman 2675a651b0 Indent the .text, .data, and .bss directives in assembly output, so that
they are consistent with the other directives.

llvm-svn: 36687
2007-05-03 18:46:30 +00:00
Chris Lattner 83df45a959 Fix two classes of bugs:
1. x86 backend rejected (&gv+c) for the 'i' constraint when in static mode.
  2. the matcher didn't correctly reject and accept some global addresses.
     the right predicate is GVRequiresExtraLoad, not "relomodel = pic".

llvm-svn: 36670
2007-05-03 16:52:29 +00:00
Chris Lattner 9a8c7cf00b add support for printing offset from global
llvm-svn: 36669
2007-05-03 16:42:23 +00:00
Chris Lattner 90bb4fc96b revert accidental commit
llvm-svn: 36668
2007-05-03 16:40:25 +00:00
Chris Lattner c1a2a3b344 add support for printing offset of global
llvm-svn: 36667
2007-05-03 16:39:48 +00:00
Dan Gohman e27e6e6fa8 Sets the section names for fixed-size constants and use the mergeable
flag for ELF on x86 so that duplicate constants can be eliminated by
the linker. This matches what GCC does with its -fmerge-constants
option, which is enabled at most -O levels.

llvm-svn: 36666
2007-05-03 16:38:57 +00:00
Chris Lattner 4995f92cbc revert reid's patch to fix these failures:
test/CodeGen/CBackend/2007-01-08-ParamAttr-ICmp.ll for PR1099 [DEJAGNU]
Applications/SPASS/SPASS [CBE]
Regression/C/2004-03-15-IndirectGoto [CBE]

llvm-svn: 36664
2007-05-03 02:57:13 +00:00
Evan Cheng bef131de68 Typo. It's checking if V is multiple of 4, not multiple of 3. :-)
llvm-svn: 36663
2007-05-03 02:00:18 +00:00
Devang Patel 8c78a0bff0 Drop 'const'
llvm-svn: 36662
2007-05-03 01:11:54 +00:00
Chris Lattner 1c1082133c match a reassociated form of fnmul. This implements CodeGen/ARM/fnmul.ll
llvm-svn: 36660
2007-05-03 00:32:00 +00:00
Chris Lattner f65f18b2f9 expose HonorSignDependentRoundingFPMathOption to .td files
llvm-svn: 36658
2007-05-03 00:27:11 +00:00
Chris Lattner 6ac3a37fb6 Add a new option.
llvm-svn: 36657
2007-05-03 00:16:07 +00:00
Bill Wendling b5ce7c5466 Non-algorithmic change. Moved definitions around into separate sections
for SSE1, SSE2, SSE3, and SSSE3.

llvm-svn: 36656
2007-05-02 23:11:52 +00:00
Bill Wendling ba3b7ee030 Update.
llvm-svn: 36653
2007-05-02 21:42:20 +00:00
Devang Patel e95c6ad802 Use 'static const char' instead of 'static const int'.
Due to darwin gcc bug, one version of darwin linker coalesces
static const int, which defauts PassID based pass identification.

llvm-svn: 36652
2007-05-02 21:39:20 +00:00
Lauro Ramos Venancio 41223586a2 Fix build error.
llvm-svn: 36648
2007-05-02 20:37:47 +00:00
Anton Korobeynikov f1dcf69fc3 Emit correct register move information in eh frames for X86. This allows Shootout-C++/except to pass on x86/linux
with non-llvm-compiled (e.g. "native") unwind runtime.

llvm-svn: 36647
2007-05-02 19:53:33 +00:00
Anton Korobeynikov 073ad20459 Emit correct DWARF reg # for RA (return address) register
llvm-svn: 36646
2007-05-02 08:46:03 +00:00
Reid Spencer f3aa932824 Make sign extension work correctly for unusual bit widths.
llvm-svn: 36635
2007-05-02 02:17:41 +00:00
Dale Johannesen 9ce17f1e33 Add some support for (Darwin) code-generating directives in getInlineAsmLength.
Support is incomplete, but more accurate than gcc's.

llvm-svn: 36634
2007-05-02 01:02:40 +00:00
Anton Korobeynikov b538f67b1a Fix couple of bugs connected with eh info:
1. Correct output offsets on Linux
2. Fix "style" of personality function. It shouldn't be indirect.

llvm-svn: 36633
2007-05-01 22:23:12 +00:00
Devang Patel 09f162ca6a Do not use typeinfo to identify pass in pass manager.
llvm-svn: 36632
2007-05-01 21:15:47 +00:00
Evan Cheng fa34bc9623 Doh. PC displacement is between the constantpool and the add instruction.
llvm-svn: 36630
2007-05-01 20:27:19 +00:00
Anton Korobeynikov 76c8c95466 Use correct PC symbol
llvm-svn: 36628
2007-05-01 10:19:31 +00:00
Anton Korobeynikov 2ac2197a0f Adjust correct EH-related sections
llvm-svn: 36627
2007-05-01 10:16:06 +00:00
Evan Cheng 5662b21db1 eliminateFrameIndex() change.
llvm-svn: 36626
2007-05-01 09:13:03 +00:00
Nate Begeman 27a625a74b llvm bug #1350, parts 1, 2, and 3.
llvm-svn: 36618
2007-05-01 05:57:02 +00:00
Evan Cheng 77c545e6b8 Under normal circumstances, when a frame pointer is not required, we reserve
argument space for call sites in the function immediately on entry to the
current function. This eliminates the need for add/sub sp brackets around call
sites. However, this is not always a good idea. If the "call frame" is large and
the target load / store instructions have small immediate field to encode sp
offset, this can cause poor codegen. In the worst case, this can make it
impossible to scavenge a register if the reserved spill slot is pushed too far
apart from sp / fp.

llvm-svn: 36607
2007-05-01 00:52:08 +00:00
Dale Johannesen 8dd36a51fc Remove item: thumb padding in constant islands
llvm-svn: 36586
2007-04-30 00:32:06 +00:00
Dale Johannesen 71577f3848 remove unused variable
llvm-svn: 36585
2007-04-30 00:30:48 +00:00
Lauro Ramos Venancio 2c673ad7ef Enable protected visibility on ARM.
llvm-svn: 36583
2007-04-30 00:23:51 +00:00
Dale Johannesen 4a00cf3fc4 Rewrite of Thumb constant islands handling (exact allowance for padding
around islands and jump tables).

llvm-svn: 36573
2007-04-29 19:19:30 +00:00
Dale Johannesen 1ee29dd026 Make ARM-specific version of getInlineAsmLength
llvm-svn: 36572
2007-04-29 19:17:45 +00:00
Anton Korobeynikov 39f3cffbe3 Implement protected visibility. This partly implements PR1363. Linker
should be taught to deal with protected symbols.

llvm-svn: 36565
2007-04-29 18:35:00 +00:00
Dale Johannesen 7e7280b538 change per review
llvm-svn: 36519
2007-04-28 00:36:37 +00:00
Dale Johannesen 29c05756b5 Prevent Thumb code from generating ARM instructions
llvm-svn: 36518
2007-04-27 22:17:18 +00:00
Lauro Ramos Venancio 99cb63029a add parenthesis.
llvm-svn: 36514
2007-04-27 20:10:08 +00:00
Evan Cheng 559c546453 Doh. ARM::LEApcrel is a single instruction MI.
llvm-svn: 36513
2007-04-27 18:27:13 +00:00
Lauro Ramos Venancio 62f65468bc In Thumb mode, the frame register must be R7.
llvm-svn: 36512
2007-04-27 17:58:03 +00:00
Lauro Ramos Venancio c39c12a3fa ARM TLS: implement "general dynamic", "initial exec" and "local exec" models.
llvm-svn: 36506
2007-04-27 13:54:47 +00:00
Evan Cheng c9f22fd1a4 Darwin runtime library does not have these.
llvm-svn: 36505
2007-04-27 08:15:43 +00:00
Evan Cheng d9990f061f Special handling of LEApcrel and tLEApcrel.
llvm-svn: 36504
2007-04-27 08:14:15 +00:00
Evan Cheng e54018687d Back out previous check-in. Incorrect.
llvm-svn: 36503
2007-04-27 07:50:02 +00:00
Evan Cheng 68ec63b3d7 tLEApcrel is a AddrModeTs, i.e. pc relative.
llvm-svn: 36502
2007-04-27 07:37:22 +00:00
Dan Gohman a30eabdd6e Fix PR1339 and CodeGen/X86/dollar-name.ll
llvm-svn: 36495
2007-04-26 21:07:05 +00:00
Bill Wendling c8264ca457 Have MMX registers clobbered in x86-64 too.
llvm-svn: 36494
2007-04-26 21:06:48 +00:00
Evan Cheng 910c80851e Rename findRegisterUseOperand to findRegisterUseOperandIdx to avoid confusion.
llvm-svn: 36483
2007-04-26 19:00:32 +00:00
Evan Cheng ce6e6db704 Fix for PR1348. If stack inc / dec amount is > 32-bits, issue a series of add / sub instructions.
llvm-svn: 36456
2007-04-26 01:09:28 +00:00
Evan Cheng 0ba174534c Match MachineFunction::UsedPhysRegs changes.
llvm-svn: 36452
2007-04-25 22:13:27 +00:00
Bill Wendling 157d7ee7e5 Add SSSE3 as a feature of Core2. Add MMX registers to the list of registers
clobbered by a call.

llvm-svn: 36448
2007-04-25 21:31:48 +00:00
Chris Lattner d20cd6658a do the multiplication as signed, so that 2*-2 == -4 instead of 4294967292
when promoted to 64-bits

llvm-svn: 36442
2007-04-25 17:23:53 +00:00
Lauro Ramos Venancio 4ae9425bb6 remember to emit weak reference in one more case.
llvm-svn: 36438
2007-04-25 14:50:40 +00:00
Anton Korobeynikov a97b694c82 Implement aliases. This fixes PR1017 and it's dependent bugs. CFE part
will follow.

llvm-svn: 36435
2007-04-25 14:27:10 +00:00
Evan Cheng 8cd224e81c Relex assertions to account for additional implicit def / use operands.
llvm-svn: 36430
2007-04-25 07:12:14 +00:00
Chris Lattner b975bebec1 support for >4G stack frames
llvm-svn: 36425
2007-04-25 04:30:24 +00:00
Chris Lattner 1ef35a2721 support >4G stack frames
llvm-svn: 36423
2007-04-25 04:25:10 +00:00
Chris Lattner cb35c61a4b Fix PR1351 and CodeGen/PowerPC/2007-04-24-InlineAsm-I-Modifier.ll
llvm-svn: 36410
2007-04-24 22:51:03 +00:00
Bill Wendling a784d875be Update.
llvm-svn: 36407
2007-04-24 21:20:03 +00:00
Bill Wendling b3b6c35beb Add the PADDQ to the list.
llvm-svn: 36406
2007-04-24 21:19:14 +00:00
Bill Wendling 5c7f25632e Add the final MMX instructions. Correct a few wrong patterns.
llvm-svn: 36405
2007-04-24 21:18:37 +00:00
Bill Wendling e2324ca17d Remove some invalid instructions from this check.
llvm-svn: 36404
2007-04-24 21:17:46 +00:00
Bill Wendling 591eab8844 Support for the special case of a vector with the canonical form:
vector_shuffle v1, v2, <2, 6, 3, 7>

I.e.

         vector_shuffle v, undef, <2, 2, 3, 3>

MMX only has a shuffle for v4i16 vectors. It needs to use the unpackh for
this type of operation.

llvm-svn: 36403
2007-04-24 21:16:55 +00:00
Dale Johannesen fdfb757682 Make Thumb funcs containing islands 4-byte aligned. Fix bugs maintaining
BBOffsets and BBSizes when adjusting conditional branches.

llvm-svn: 36372
2007-04-23 20:09:04 +00:00
Dale Johannesen ed0ecdb7ff add Align field, and use when generating function alignment
llvm-svn: 36371
2007-04-23 20:07:25 +00:00
Dale Johannesen 8eb9bff493 add isThumb (unused as yet)
llvm-svn: 36370
2007-04-23 20:04:35 +00:00
Dale Johannesen 0a1069d526 Fix generic getInlineAsmLength
llvm-svn: 36369
2007-04-23 20:00:17 +00:00
Lauro Ramos Venancio 6db679a49a X86 TLS: optimize the implementation of "local exec" model.
llvm-svn: 36359
2007-04-23 01:28:10 +00:00
Lauro Ramos Venancio efb8077ddd X86 TLS: fix and optimize the implementation of "initial exec" model.
llvm-svn: 36355
2007-04-22 22:50:52 +00:00
Christopher Lamb 297fcb0f73 Support alignment queries for degenerate (length 1) vectors.
llvm-svn: 36352
2007-04-22 21:54:13 +00:00
Lauro Ramos Venancio ee2d164f0f Implement PIC for arm-linux.
llvm-svn: 36324
2007-04-22 00:04:12 +00:00
Lauro Ramos Venancio 4e91908f17 X86 TLS: Implement review feedback.
llvm-svn: 36318
2007-04-21 20:56:26 +00:00
Jeff Cohen 5959f42498 Comment out usage of write() for now.
llvm-svn: 36287
2007-04-20 22:40:10 +00:00
Lauro Ramos Venancio 2518889872 Implement "general dynamic", "initial exec" and "local exec" TLS models for
X86 32 bits.

llvm-svn: 36283
2007-04-20 21:38:10 +00:00
Evan Cheng e656d093a5 Specify S registers as D registers' sub-registers.
llvm-svn: 36280
2007-04-20 21:20:10 +00:00
Evan Cheng 06a164c6bc Specify sub-register relations. e.g. RAX: [EAX], EAX: [AX], AX: [AL,AH].
llvm-svn: 36279
2007-04-20 21:15:21 +00:00
Evan Cheng ac9fdde7f4 Add sub-registers sets.
llvm-svn: 36278
2007-04-20 21:13:46 +00:00
Chris Lattner 3d3f22766a add a crazy idea
llvm-svn: 36273
2007-04-20 20:18:43 +00:00
Jeff Cohen 6c673ac01c Make Microsoft assembler and linker happy.
llvm-svn: 36265
2007-04-20 00:33:54 +00:00
Chris Lattner 25e62eb63b Fix a message, patch by Christopher Lamb.
llvm-svn: 36264
2007-04-19 18:42:38 +00:00
Lauro Ramos Venancio 42cd7253b1 Fix a bug in getFrameRegister.
Reported by Raul Herbster.

llvm-svn: 36262
2007-04-19 14:09:38 +00:00
Dan Gohman 29845cd40d Fix the spelling of the prefetchnta instruction.
llvm-svn: 36256
2007-04-18 14:09:14 +00:00
Evan Cheng 5118226c6c Oops. Didn't mean to check in a quick hack.
llvm-svn: 36227
2007-04-17 23:33:39 +00:00
Chris Lattner 598bc0d9a3 dag combiner just got better at pruning bits. This fixes CodeGen/ARM/rev.ll
llvm-svn: 36222
2007-04-17 22:39:58 +00:00
Evan Cheng 2f45bf31c5 Change getAllocatableSet() so it returns allocatable registers for a specific register class.
llvm-svn: 36215
2007-04-17 20:23:34 +00:00