Commit Graph

3041 Commits

Author SHA1 Message Date
Chris Lattner 897a140e2a cleanups.
llvm-svn: 118228
2010-11-04 01:55:23 +00:00
Chris Lattner 7108dad130 replace SrcOpNum with SrcOpName, eliminating a numering dependency
on the incoming operand list.  This also makes the code simpler.

llvm-svn: 118225
2010-11-04 01:42:59 +00:00
Chris Lattner 4779e3e94a strength reduce some code, resolving a fixme.
llvm-svn: 118219
2010-11-04 00:57:06 +00:00
Chris Lattner 743081d097 take a big step to making aliases more general and less of a hack:
now matchables contain an explicit list of how to populate each
operand in the result instruction instead of having them somehow
magically be correlated to the input inst.

llvm-svn: 118217
2010-11-04 00:43:46 +00:00
Jim Grosbach e4e6bf49f3 Support generating an MC'ized CodeEmitter directly. Maintain a reference to the
Fixups list for the instruction so the operand encoders can add to it as
needed.

llvm-svn: 118206
2010-11-03 23:38:14 +00:00
Chris Lattner 896cf04885 rename Operand -> AsmOperand for clarity.
llvm-svn: 118190
2010-11-03 19:47:34 +00:00
Duncan Sands f5dda01f33 Inside the calling convention logic LocVT is always a simple
value type, so there is no point in passing it around using
an EVT.  Use the simpler MVT everywhere.  Rather than trying
to propagate this information maximally in all the code that
using the calling convention stuff, I chose to do a mainly
low impact change instead.

llvm-svn: 118167
2010-11-03 11:35:31 +00:00
Jim Grosbach c6af2b4066 Break ARM addrmode4 (load/store multiple base address) into its constituent
parts. Represent the operation mode as an optional operand instead.
rdar://8614429

llvm-svn: 118137
2010-11-03 01:01:43 +00:00
Chris Lattner cc5dce89d4 Completely reject instructions that have an operand in their
ins/outs list that isn't specified by their asmstring.  Previously
the asmmatcher would just force a 0 register into it, which clearly
isn't right.  Mark a bunch of ARM instructions that use this as 
isCodeGenOnly.  Some of them are clearly pseudo instructions (like
t2TBB) others use a weird hasExtraSrcRegAllocReq thing that will
either need to be removed or the asmmatcher will need to be taught
about it (someday).

llvm-svn: 118119
2010-11-02 23:40:41 +00:00
Chris Lattner 4d23eb2f0e make MatchableInfo::Validate reject instructions (like LDR_PRE in ARM)
that have complicated tying going on.

llvm-svn: 118112
2010-11-02 23:18:43 +00:00
Chris Lattner e032dbfd9e rewrite EmitConvertToMCInst to iterate over the MCInst operands,
filling them in one at a time.  Previously this iterated over the
asmoperands, which left the problem of "holes".  The new approach
simplifies things.

llvm-svn: 118104
2010-11-02 22:55:03 +00:00
Chris Lattner 5cf8a4a909 merge two large parallel loops in EmitConvertToMCInst, no change
in the generated .inc files.

llvm-svn: 118083
2010-11-02 21:49:44 +00:00
Chris Lattner 77d3ead281 a bunch of random cleanup, move a helper to CGT where it belongs.
llvm-svn: 118031
2010-11-02 18:10:06 +00:00
Jim Grosbach 0b7fda23cc Revert r114340 (improvements in Darwin function prologue/epilogue), as it broke
assumptions about stack layout. Specifically, LR must be saved next to FP.

llvm-svn: 118026
2010-11-02 17:35:25 +00:00
Chris Lattner a7a903e706 add and update comments.
llvm-svn: 118025
2010-11-02 17:34:28 +00:00
Chris Lattner 28ea9b18cf refactor/cleanup MatchableInfo by eliminating the Tokens array,
merging it into a Token field in Operand, and moving the first
token to an explicit mnemonic field.  These were parallel
arrays before (except for the mnemonic) which kept confusing me.

llvm-svn: 118024
2010-11-02 17:30:52 +00:00
Chris Lattner d64b7c0685 rename operands -> asmoperands to be more descriptive.
llvm-svn: 117993
2010-11-02 01:03:43 +00:00
Jim Grosbach 9a33835caa Tidy up.
llvm-svn: 117987
2010-11-02 00:16:39 +00:00
Chris Lattner e3c48deff5 fix computation of ambiguous instructions to not ignore the mnemonic.
FWIW, X86 has 254 ambiguous instructions.

llvm-svn: 117979
2010-11-01 23:57:23 +00:00
Chris Lattner c28e1db284 give MatchableInfo::Operand a constructor
llvm-svn: 117968
2010-11-01 23:08:02 +00:00
Chris Lattner dd3b09c234 Implement enough of the missing instalias support to get
aliases installed and working.  They now work when the
matched pattern and the result instruction have exactly
the same operand list.

This is now enough for us to define proper aliases for
movzx and movsx, implementing rdar://8017633 and PR7459.

Note that we do not accept instructions like:
  movzx 0(%rsp), %rsi

GAS accepts this instruction, but it doesn't make any
sense because we don't know the size of the memory
operand.  It could be 8/16/32 bits.

llvm-svn: 117901
2010-11-01 05:34:34 +00:00
Chris Lattner ad77681253 rename InstructionInfo -> MatchableInfo since it now
represents InstAliases as well.  Rename 
isAssemblerInstruction -> Validate since that is what
it does (modulo the ARM $lane hack).

llvm-svn: 117899
2010-11-01 05:06:45 +00:00
Chris Lattner ba465f9a8a refactor initialization of InstructionInfo to be sharable between
instructions and InstAliases.  Start creating InstructionInfo's
for Aliases.

llvm-svn: 117898
2010-11-01 04:53:48 +00:00
Chris Lattner 178f4bb62d make the asm matcher emitter reject instructions that have comments
in their asmstring.  Fix the two x86 "NOREX" instructions that have them.
If these comments are important, the instlowering stuff can print them.

llvm-svn: 117897
2010-11-01 04:44:29 +00:00
Chris Lattner 39bc53b33a refactor InstructionInfo to not have a pointer to CodeGenInstruction
member, and make isAssemblerInstruction() a method (pushing some code
around inside it).

llvm-svn: 117895
2010-11-01 04:34:44 +00:00
Chris Lattner 488c201bb7 define a new CodeGenInstAlias. It has an asmstring and operand list for now,
todo: the result field.

llvm-svn: 117894
2010-11-01 04:05:41 +00:00
Chris Lattner d8adec70f3 factor the operand list (and related fields/operations) out of
CodeGenInstruction into its own helper class.  No functionality change.

llvm-svn: 117893
2010-11-01 04:03:32 +00:00
Chris Lattner 86e1c9484f avoid needless throw/catch/rethrow, stringref'ize some simple stuff.
llvm-svn: 117892
2010-11-01 03:19:09 +00:00
Chris Lattner a397716081 eliminate the old InstFormatName which is always "AsmString",
simplify CodeGenInstruction. No functionality change.

llvm-svn: 117891
2010-11-01 02:15:23 +00:00
Chris Lattner 517dc95d47 all predicates on an MnemonicAlias must be AssemblerPredicates.
llvm-svn: 117890
2010-11-01 02:09:21 +00:00
Chris Lattner f7a01e9f46 change the singleton register handling code to be based on Record*'s
instead of strings, simplifying it.

llvm-svn: 117889
2010-11-01 01:47:07 +00:00
Chris Lattner b80ab36179 Give AsmMatcherInfo a CodeGenTarget, which simplifies a bunch of
argument passing.  Consolidate all SingletonRegister detection
and handling into a new 
InstructionInfo::getSingletonRegisterForToken method instead of
having it scattered about.  No change in generated .inc files.

llvm-svn: 117888
2010-11-01 01:37:30 +00:00
Chris Lattner 25d9c7fa2e move FlattenVariants out of AsmMatcherEmitter into a shared
CodeGenInstruction::FlattenAsmStringVariants method.  Use it
to simplify the code in AsmWriterInst, which now no longer 
needs to worry about variants.

llvm-svn: 117886
2010-11-01 01:07:14 +00:00
Chris Lattner 40dd3f0939 add a FIXME, $lane in ARM is an issue that needs to be resolved before
this can start rejecting instructions.

llvm-svn: 117885
2010-11-01 00:51:32 +00:00
Chris Lattner 941c19b7ba reject instructions that contain a \n in their asmstring. Mark
various X86 and ARM instructions that are bitten by this as isCodeGenOnly,
as they are.

llvm-svn: 117884
2010-11-01 00:46:16 +00:00
Chris Lattner d689026899 fix a crash on:
let Constraints = "$val = $dst", Defs = [EFLAGS] in, isCodeGenOnly = 1 {

we now get:

X86InstrCompiler.td:653:52: error: Expected class, def, defm, multiclass or let definition
let Constraints = "$val = $dst", Defs = [EFLAGS] in, isCodeGenOnly = 1 {
                                                   ^

llvm-svn: 117863
2010-10-31 19:27:15 +00:00
Chris Lattner 7ff334687d fix the !eq operator in tblgen to return a bit instead of an int.
Use this to make the X86 and ARM targets set isCodeGenOnly=1 
automatically for their instructions that have Format=Pseudo,
resolving a hack in tblgen.

llvm-svn: 117862
2010-10-31 19:22:57 +00:00
Chris Lattner 9492c17baf two changes: make the asmmatcher generator ignore ARM pseudos properly,
and make it a hard error for instructions to not have an asm string.
These instructions should be marked isCodeGenOnly.

llvm-svn: 117861
2010-10-31 19:15:18 +00:00
Chris Lattner 33fc3e095b reapply r117858 with apparent editor malfunction fixed (somehow I
got a dulicated line).

llvm-svn: 117860
2010-10-31 19:10:56 +00:00
Chris Lattner e59eef3dd1 revert r117858 while I check out a failure I missed.
llvm-svn: 117859
2010-10-31 19:05:32 +00:00
Chris Lattner 9293008e90 the asm matcher can't handle operands with modifiers (like ${foo:bar}).
Instead of silently ignoring these instructions, emit a hard error and
force the target author to either refactor the target or mark the 
instruction 'isCodeGenOnly'.

Mark a few instructions in ARM and MBlaze as isCodeGenOnly the are 
doing this.

llvm-svn: 117858
2010-10-31 18:48:12 +00:00
Chris Lattner 43690071cf have GetAliasRequiredFeatures get its features from
AsmMatcherInfo so we don't have two places that know the
feature -> enum mapping.  No functionality change.

llvm-svn: 117845
2010-10-30 20:15:02 +00:00
Chris Lattner a0e871901b simplify code that creates SubtargetFeatureInfo, ensuring that features
that are only used by MnemonicAliases will be found.

llvm-svn: 117844
2010-10-30 20:07:57 +00:00
Chris Lattner 25896af4bb fix a fixme in stringmatcher, having it generate nice looking code if the
'tomatch' code contains \n's.

llvm-svn: 117843
2010-10-30 19:57:17 +00:00
Chris Lattner f9ec2fb34a fix typos and some serious bugs in feature handling (but not for
cases that are currently exercised).  Thanks to Frits van Bommel for
the great review!

llvm-svn: 117840
2010-10-30 19:47:49 +00:00
Chris Lattner aac142cc06 Resolve a terrible hack in tblgen: instead of hardcoding
"In32BitMode" and "In64BitMode" into tblgen, allow any
predicate that inherits from AssemblerPredicate.

llvm-svn: 117831
2010-10-30 19:38:20 +00:00
Chris Lattner 2cb092dc55 Implement (and document!) support for MnemonicAlias's to have Requires
directives, allowing things like this:

def : MnemonicAlias<"pop", "popl">, Requires<[In32BitMode]>;
def : MnemonicAlias<"pop", "popq">, Requires<[In64BitMode]>;

Move the rest of the X86 MnemonicAliases over to the .td file.

llvm-svn: 117830
2010-10-30 19:23:13 +00:00
Chris Lattner ec56397eb4 fix build problem
llvm-svn: 117828
2010-10-30 18:57:07 +00:00
Chris Lattner cf9b6e3107 diagnose targets that define two alises with the same 'from' mnemonic
with a useful error message instead of having tblgen explode with an 
assert.

llvm-svn: 117827
2010-10-30 18:56:12 +00:00
Chris Lattner 477fba4f54 emit the mnemonic aliases in their own helper function instead of
inline into MatchInstructionImpl.

llvm-svn: 117826
2010-10-30 18:48:18 +00:00
Chris Lattner ba7b4fea97 implement (and document!) the first kind of MC assembler alias, which
just remaps one mnemonic to another.  Convert a few of the X86 aliases
from .cpp to .td code.

llvm-svn: 117815
2010-10-30 17:36:36 +00:00
Dan Gohman 96f54c9d3e Apply a patch from Frits van Bommel to fix the CMake build.
llvm-svn: 117731
2010-10-29 22:36:50 +00:00
Jim Grosbach 0eccfc2693 trailing whitespace
llvm-svn: 117724
2010-10-29 22:13:48 +00:00
Dan Gohman b75ce4f816 not is testing for a normal exit with a non-zero value. It shouldn't
return success if the child process is killed with a signal.

llvm-svn: 117701
2010-10-29 20:20:29 +00:00
Chris Lattner 1be0697ab9 fix the asmmatcher generator to handle targets with no RegisterPrefix
(like ARM) correctly.  With this change, we can now match "bx lr"
because we recognize lr as a register.

llvm-svn: 117606
2010-10-28 21:28:42 +00:00
Evan Cheng 59bbc545e0 Shifter ops are not always free. Do not fold them (especially to form
complex load / store addressing mode) when they have higher cost and
when they have more than one use.

llvm-svn: 117509
2010-10-27 23:41:30 +00:00
Owen Anderson fadb951e5b Provide correct encodings for NEON vcvt, which has its own special immediate encoding
for specifying fractional bits for fixed point conversions.

llvm-svn: 117501
2010-10-27 22:49:00 +00:00
Kevin Enderby 5e7cb5fc27 Added the x86 instruction ud2b (2nd official undefined instruction).
llvm-svn: 117485
2010-10-27 20:46:49 +00:00
Dan Gohman 5faac39061 Enable clang autocompletion by default.
llvm-svn: 117415
2010-10-26 23:24:54 +00:00
Jim Grosbach 1e4d9a17c2 First part of refactoring ARM addrmode2 (load/store) instructions to be more
explicit about the operands. Split out the different variants into separate
instructions. This gives us the ability to, among other things, assign
different scheduling itineraries to the variants. rdar://8477752.

llvm-svn: 117409
2010-10-26 22:37:02 +00:00
Benjamin Kramer 34402c4fe4 Constify another 2 disassembler tables.
llvm-svn: 117208
2010-10-23 09:28:42 +00:00
Benjamin Kramer de0a4fbf3b Make the disassembler tables const so they end up in read-only memory.
llvm-svn: 117206
2010-10-23 09:10:44 +00:00
Mikhail Glushenkov f4a6809231 Remove -llvmc-temp-hack from tblgen.
llvm-svn: 117197
2010-10-23 07:32:53 +00:00
Mikhail Glushenkov e9b78186d4 Syntax tweak in llvmc: (something [a,b,c]) -> (something a, b, c).
llvm-svn: 117196
2010-10-23 07:32:46 +00:00
Mikhail Glushenkov de68389cd3 Trailing whitespace.
llvm-svn: 117195
2010-10-23 07:32:37 +00:00
Bob Wilson dbb77154dc Do not install libBugpointPasses.dylib for Apple builds. Radar 8585383.
llvm-svn: 117172
2010-10-22 23:04:17 +00:00
Bob Wilson 1eaa381789 Fix the name of libLLVMHello.dylib so that it does not get installed for
Apple builds.  Radar 8585383.

llvm-svn: 117164
2010-10-22 22:10:57 +00:00
Benjamin Kramer 9192e7ab12 Make some symbols static, move classes into anonymous namespaces.
llvm-svn: 117111
2010-10-22 17:35:07 +00:00
Anders Carlsson 83123a47ce Add a way to emit StringSwitch of clang attribute spellings.
llvm-svn: 116899
2010-10-20 01:21:53 +00:00
Michael J. Spencer 713460114d Fix VC2010 build.
llvm-svn: 116833
2010-10-19 18:04:06 +00:00
Oscar Fuentes 889c1e7d80 Build with RTTI and exceptions disabled. Only in GCC for now.
llvm-svn: 116682
2010-10-17 02:26:16 +00:00
Michael J. Spencer c46e8cb258 KillTheDoctor: Fix 2008 build. I'm actually surprised 2010 defines all of these, many are non-standard posix/unix extensions.
llvm-svn: 116597
2010-10-15 18:13:02 +00:00
Jakob Stoklund Olesen eba55822d7 Teach FileCheck to handle trailing CHECK-NOT patterns.
A CHECK-NOT pattern without a following CHECK pattern simply checks that the
pattern doesn't match before the end of the input file.

You can even have only CHECK-NOT patterns to check that strings appear nowhere
in the input file.

llvm-svn: 116592
2010-10-15 17:47:12 +00:00
Jim Grosbach 68a335e185 ARM mode encoding information for UBFX and SBFX instructions.
llvm-svn: 116588
2010-10-15 17:15:16 +00:00
Jim Grosbach dba47755a0 Teach PerfectShuffle to not generate files with embedded tab characters.
llvm-svn: 116458
2010-10-14 00:12:49 +00:00
Jim Grosbach 7e72ec6626 Refactor the ARM 'setend' instruction pattern. Use a single instruction pattern
and handle the operand explicitly. Flesh out encoding information. Add an
explicit disassembler testcase for the instruction.

llvm-svn: 116432
2010-10-13 21:00:04 +00:00
Jim Grosbach 1e7db68774 Add ARM mode encoding for [SU]XT[BH] and [SU]XTA[BH] instructions.
llvm-svn: 116421
2010-10-13 19:56:10 +00:00
Jim Grosbach 51a12eb11d Allow targets to optionally specify custom binary encoder functions for
operand values. This is useful for operands which require additional trickery
to encode into the instruction. For example, the ARM shifted immediate and
shifted register operands.

llvm-svn: 116353
2010-10-12 22:21:57 +00:00
Michael J. Spencer 65c1c43c13 KillTheDoctor: Fix VS2008 build.
llvm-svn: 116330
2010-10-12 19:27:44 +00:00
Cameron Esfahani a48349f596 Fix spelling error.
llvm-svn: 116282
2010-10-12 00:21:05 +00:00
Jim Grosbach e61de930bc The assert() should reference to machine instr operand number, too.
llvm-svn: 116243
2010-10-11 21:41:31 +00:00
Jim Grosbach 11ced671be Make sure to use the machine instruction operand number. It doesn't always
map one-to-one with the CodeGenInstruction operand number.

llvm-svn: 116238
2010-10-11 21:31:22 +00:00
Michael J. Spencer 279362dd5a Add KillTheDoctor.
llvm-svn: 116216
2010-10-11 19:55:38 +00:00
Jim Grosbach 806b139bbc trailing whitespace cleanup
llvm-svn: 116215
2010-10-11 19:38:01 +00:00
Jim Grosbach 191ad7c473 When figuring out which operands match which encoding fields in an instruction,
try to match them by name first. If there is no by-name match, fall back to
assuming they are in order (this was the previous behavior).

llvm-svn: 116211
2010-10-11 18:25:51 +00:00
Chris Lattner 1ef5e84c31 Per discussion with Sanjiv, remove the PIC16 target from mainline. When/if
it comes back, it will be largely a rewrite, so keeping the old codebase
in tree isn't helping anyone.

llvm-svn: 116190
2010-10-11 05:44:40 +00:00
Michael J. Spencer 9cc3fcc8ec syntax-highlighting: Fix module asm keyword.
llvm-svn: 116152
2010-10-09 15:44:36 +00:00
Michael J. Spencer 3431778a2b Add Kate syntax highlighting files.
llvm-svn: 116146
2010-10-09 07:11:04 +00:00
Jim Grosbach b75d0ca38e A few 80 column cleanups
llvm-svn: 116069
2010-10-08 18:13:57 +00:00
Jim Grosbach 2f0be8f404 trailing whitespace
llvm-svn: 116068
2010-10-08 18:09:59 +00:00
Daniel Dunbar ba66a81017 Fix -Asserts warning.
llvm-svn: 116030
2010-10-08 02:07:22 +00:00
Jim Grosbach a7b6d58f45 Make <target>CodeEmitter::getBinaryCodeForInstr() a const method.
llvm-svn: 116018
2010-10-08 00:21:28 +00:00
Jim Grosbach 33c1eb16e8 Move checking for t2MOVCCi16 to the right place.
llvm-svn: 115994
2010-10-07 22:14:01 +00:00
Nick Lewycky 1e00173d20 Fix typo in comment.
llvm-svn: 115986
2010-10-07 21:55:16 +00:00
Dan Gohman 0df7ea4c24 Move tool_output_file into its own file.
llvm-svn: 115973
2010-10-07 20:32:40 +00:00
Michael J. Spencer d4b62765a7 gtest: Fix warnings on MinGW.
llvm-svn: 115945
2010-10-07 18:29:44 +00:00
Michael J. Spencer 33a390e2ca CMake: Fix warning in gtest.
llvm-svn: 115935
2010-10-07 18:12:54 +00:00
Jim Grosbach daab660fb1 trailing whitespace
llvm-svn: 115923
2010-10-07 16:56:28 +00:00
Jim Grosbach 5b255c2dd6 Allow use of the 16-bit literal move instruction in CMOVs for Thumb2 mode.
llvm-svn: 115890
2010-10-07 00:53:56 +00:00
Jim Grosbach 742adc328a Allow use of the 16-bit literal move instruction in CMOVs for ARM mode.
llvm-svn: 115884
2010-10-07 00:42:42 +00:00
Jim Grosbach b270f28c1a Now that VDUPfqf and VDUPfdfare properly pseudos, nuke the special handling.
llvm-svn: 115841
2010-10-06 21:17:07 +00:00
Chris Lattner 28f034c21a Generalize tblgen's dag parsing logic to handle arbitrary expressions
as the operator of the dag.  Specifically, this allows parsing things
like (F.x 4) in addition to just (a 4).

Unfortunately, this runs afoul of an idiom being used by llvmc.  It
is using dags like (foo [1,2,3]) to represent a list of stuff being
passed into foo.  With this change, this is parsed as a [1,2,3] 
subscript on foo instead of being the first argument to the dag.
Cope with this in the short term by requiring a "-llvmc-temp-hack"
argument to tblgen to get the old parsing behavior.

llvm-svn: 115742
2010-10-06 04:55:48 +00:00
Chris Lattner e76cfcf8a8 cleanups
llvm-svn: 115739
2010-10-06 04:31:40 +00:00
Chris Lattner 9402633637 remove the !nameconcat tblgen feature. It "shorthand" and only used in 4 places
where !cast is just as short.

llvm-svn: 115722
2010-10-06 00:19:21 +00:00
Chris Lattner 61ea00b494 allow !strconcat to take more than two operands to eliminate
!strconcat(!strconcat(!strconcat(!strconcat

Simplify some x86 td files to use it.

llvm-svn: 115719
2010-10-05 23:58:18 +00:00
Chris Lattner b8ff8f0cb6 when david added support for #NAME# he didn't update the comments and
tried (but failed) to artificially constrain it to working with #NAME#.
Just allow any # in identifiers, and update the comments.

llvm-svn: 115704
2010-10-05 22:59:29 +00:00
Chris Lattner 7538ed80a9 enhance tblgen to support anonymous defm's, use this to
simplify the X86 CMOVmr's.

llvm-svn: 115702
2010-10-05 22:51:56 +00:00
Jim Grosbach c1526595b3 trailing whitespace
llvm-svn: 115664
2010-10-05 20:35:57 +00:00
Sebastian Redl c4abc7036d Update attribute reading for the changed source location code.
llvm-svn: 115624
2010-10-05 15:59:36 +00:00
Douglas Gregor 9ddb678d45 Properly deserialize Clang types that are used as attribute arguments
llvm-svn: 115616
2010-10-05 14:51:48 +00:00
Sean Callanan 8d302b2e71 Fixed the disassembler to handle two new X86
instruction forms.  Now the ENTER instruction
disassembles correctly.

llvm-svn: 115573
2010-10-04 22:45:51 +00:00
Chris Lattner 50774e8fa7 remove reference to dead script.
llvm-svn: 115413
2010-10-02 18:49:27 +00:00
Chris Lattner 250ecf0e00 nuke an old script
llvm-svn: 115412
2010-10-02 18:44:23 +00:00
Francois Pichet 8cbc86e912 Fix MSVC release mode compilation error.
llvm-svn: 115407
2010-10-02 03:26:54 +00:00
Francois Pichet 77339c7c98 Fix typo
llvm-svn: 115348
2010-10-01 21:20:39 +00:00
Dale Johannesen dd224d2333 Massive rewrite of MMX:
The x86_mmx type is used for MMX intrinsics, parameters and
return values where these use MMX registers, and is also
supported in load, store, and bitcast.

Only the above operations generate MMX instructions, and optimizations
do not operate on or produce MMX intrinsics. 

MMX-sized vectors <2 x i32> etc. are lowered to XMM or split into
smaller pieces.  Optimizations may occur on these forms and the
result casted back to x86_mmx, provided the result feeds into a
previous existing x86_mmx operation.

The point of all this is prevent optimizations from introducing
MMX operations, which is unsafe due to the EMMS problem.

llvm-svn: 115243
2010-09-30 23:57:10 +00:00
Jim Grosbach 4a57b76eea Let a target specify whether it wants an assembly printer to be the MC version
or not. TableGen needs to generate the printInstruction() function as taking
an MCInstr* or a MachineInstr*, depending. Default to the old non-MC
version so that everything not yet using MC continues to just work without
fidding.

llvm-svn: 115126
2010-09-30 01:29:54 +00:00
Evan Cheng 4a010fd1ea Model Cortex-a9 load to SUB, RSB, ADD, ADC, SBC, RSC, CMN, MVN, or CMP
pipeline forwarding path.

llvm-svn: 115098
2010-09-29 22:42:35 +00:00
Jim Grosbach a5497345ad trailing whitespace
llvm-svn: 115096
2010-09-29 22:32:50 +00:00
Duncan Sands e9da6db67f Convert a bunch of uses of 'bytecode' into 'bitcode'. This
is not everything, but the remaining cases are less trivial.

llvm-svn: 115080
2010-09-29 20:09:55 +00:00
Daniel Dunbar 8211d31055 GetSourceVersion: Strip off svninfo extra markers, which aren't really part of the version.
llvm-svn: 115074
2010-09-29 19:30:17 +00:00
Daniel Dunbar afd363e65b utils: Add GetRepositoryPath script, patch by Jonathan Mulder!
- Also, fix indention in GetSourceVersion while in the area.

llvm-svn: 115048
2010-09-29 17:57:01 +00:00
Daniel Dunbar a9dd1998cc lit: Fix a subtle resource usage bug when executing tests using the internal
shell runner.

We would inadvertently end up holding on to handles to the temporary files
longer than we should have been. On Win32, where open handles lock some file
operations, this caused problems in tests which would try to move temporary
files around (as Clang does by default now).

Many thanks to Francois Pichet for the excellent detective work on this.

llvm-svn: 115040
2010-09-29 15:59:37 +00:00
Chris Lattner f60062fd55 add basic avx support to the disassembler, also teach it about ssmem/sdmem
operands.

With this done, we can remove the _Int suffixes from the round instructions
without the disassembler blowing up.  This allows the assembler to support
them, implementing rdar://8456376 - llvm-mc rejects 'roundss'

llvm-svn: 115019
2010-09-29 02:57:56 +00:00
Evan Cheng 0097dd0d5a Add support to model pipeline bypass / forwarding.
llvm-svn: 115005
2010-09-28 23:50:49 +00:00
Benjamin Kramer c758311025 Push twines deeper into SourceMgr's error handling methods.
llvm-svn: 114847
2010-09-27 17:42:11 +00:00
Oscar Fuentes 46d8a93005 Reverting "CMake: Don't include tools, unittets, or examples as
available targets unless LLVM_INCLUDE_X is ON. LLVM_BUILD_X implies
LLVM_INCLUDE_X"

It breaks the configuration phase when cmake is invoked without
parameters, it is too complex for the purpose and introduces an
incovenience for the user (as both LLVM_BUILD_X and LLVM_INCLUDE_X
must set to OFF for not including X on the build)

llvm-svn: 114795
2010-09-25 20:25:25 +00:00
Che-Liang Chiou 299479020a Add ret instruction to PTX backend
llvm-svn: 114788
2010-09-25 07:46:17 +00:00
Michael J. Spencer ded5f66813 Get rid of pop_macro warnings on MSVC.
llvm-svn: 114750
2010-09-24 19:48:47 +00:00
Michael J. Spencer e35a611aa5 CMake: Don't include tools, unittets, or examples as available targets
unless LLVM_INCLUDE_X is ON. LLVM_BUILD_X implies LLVM_INCLUDE_X

llvm-svn: 114747
2010-09-24 19:10:51 +00:00
Michael J. Spencer 55581b3336 Fix line endings from my last commit.
llvm-svn: 114728
2010-09-24 09:10:21 +00:00
Michael J. Spencer d8010d6456 unittests: Support Windows.
llvm-svn: 114727
2010-09-24 09:01:34 +00:00
Michael J. Spencer 5d28c8428f Test unittests built with CMake.
llvm-svn: 114726
2010-09-24 09:01:22 +00:00
Michael J. Spencer 10d274d874 CMake: Build unittests.
llvm-svn: 114725
2010-09-24 09:01:13 +00:00
Michael J. Spencer 1444f47deb Fix whitespace.
llvm-svn: 114724
2010-09-24 09:00:56 +00:00
Owen Anderson 2c5df619c4 Revert r114703 and r114702, removing the isConditionalMove flag from instructions. After further
reflection, this isn't going to achieve the purpose I intended it for.  Back to the drawing board!

llvm-svn: 114710
2010-09-23 23:45:25 +00:00
Owen Anderson 6e0e8d7d64 Add an TargetInstrDesc bit to indicate that a given instruction is a conditional move.
Not intended functionality change, as nothing uses this yet.

llvm-svn: 114702
2010-09-23 22:44:10 +00:00
Nate Begeman b4e0cc0212 Revert r114596, it's breaking a few tests.
llvm-svn: 114659
2010-09-23 16:49:17 +00:00
Nate Begeman e9e9c08ce2 <rdar://problem/8228022> Wvector-conversions warnings in arm_neon.h
Explicitly cast arguments to the type the builtin expects, which is <vN x i8>

llvm-svn: 114596
2010-09-22 22:28:42 +00:00
Chris Lattner a9e57e0eff Rework passing parent pointers into complexpatterns, I forgot
that complex patterns are matched after the entire pattern has
a structural match, therefore the NodeStack isn't in a useful
state when the actual call to the matcher happens.

llvm-svn: 114489
2010-09-21 22:00:25 +00:00
Chris Lattner dd83548fea just like they can opt into getting the root of the pattern being
matched, allow ComplexPatterns to opt into getting the parent node
of the operand being matched.

llvm-svn: 114472
2010-09-21 20:37:12 +00:00
Chris Lattner 0e023ea02a fix a long standing wart: all the ComplexPattern's were being
passed the root of the match, even though only a few patterns
actually needed this (one in X86, several in ARM [which should
be refactored anyway], and some in CellSPU that I don't feel 
like detangling).   Instead of requiring all ComplexPatterns to
take the dead root, have targets opt into getting the root by
putting SDNPWantRoot on the ComplexPattern.

llvm-svn: 114471
2010-09-21 20:31:19 +00:00
Mikhail Glushenkov 5be6764363 Trailing whitespace, 80-col violations.
llvm-svn: 114435
2010-09-21 14:59:50 +00:00
Mikhail Glushenkov ed79d5f24d llvmc: Allow multiple output languages.
llvm-svn: 114433
2010-09-21 14:59:42 +00:00
Misha Brukman 81b7899bc8 Using regexp-opt for keyword regex declarations makes the word lists more
readable and easier to edit.

llvm-svn: 114308
2010-09-19 03:44:22 +00:00
Eric Christopher a573d19662 Handle the odd case where we only have one instruction.
llvm-svn: 114293
2010-09-18 18:50:27 +00:00
Daniel Dunbar c4f3990f01 lit: Bump version to 0.2.0dev, for no apparent reason.
llvm-svn: 114262
2010-09-18 02:28:15 +00:00
Daniel Dunbar 196024ce1b lit: Tweak setup.py.
llvm-svn: 114261
2010-09-18 02:28:12 +00:00
Daniel Dunbar 07cc4f6b33 lit: These TODOs are done(ish).
llvm-svn: 114260
2010-09-18 02:28:09 +00:00
Duncan Sands 922563cc6d Some of the tests build an executable with llvm-gcc. For this to work, llvm-gcc
needs to find some libraries, which may require searching the directories given
by LIBRARY_PATH on curiously configured systems.  So pass on LIBRARY_PATH.

llvm-svn: 114186
2010-09-17 17:05:27 +00:00