Commit Graph

446 Commits

Author SHA1 Message Date
Chris Lattner fdd8790718 strength reduce a ton of type equality tests to check the typeid (Through
the new predicates I added) instead of going through a context and doing a
pointer comparison.  Besides being cheaper, this allows a smart compiler
to turn the if sequence into a switch.

llvm-svn: 83297
2009-10-05 05:54:46 +00:00
Nick Lewycky ae4617c974 Requires element types in a constant initializer to match the element types of
of the constant. This reverts r6544 and r7428.

llvm-svn: 83270
2009-10-03 19:30:43 +00:00
Jeffrey Yasskin 8ce67f83bc Assert that ConstantArrays are created with correctly-typed elements.
llvm-svn: 83168
2009-09-30 21:08:08 +00:00
Duncan Sands 129de48982 For the NSWSub support in the builder to actually be useable,
there need to be corresponding changes to the constant folders,
done in this patch.

llvm-svn: 82862
2009-09-26 15:35:35 +00:00
Dan Gohman 394468dc8e Rename ConstantFP's getInf to getInfinity.
llvm-svn: 82823
2009-09-25 23:40:21 +00:00
Dan Gohman feb5021134 Add a ConstantFP::getInf utility function for creating infinity ConstantFPs.
llvm-svn: 82818
2009-09-25 23:00:48 +00:00
Dan Gohman e4532f30b1 When a constant's type is refined, update the constant in place
instead of cloning and RAUWing it.

 - Make AbstractTypeUser a friend of Value so that it can offer
   its subclasses a way to update a Value's type in place. This
   is better than a universally visible setType method on Value,
   and it's sufficient for the immediate need.

 - Eliminate the constant "convert" functions. This eliminates a
   lot of logic duplication, and fixes a complicated bug where a
   constant can't actually be cloned during the type refinement
   process because some of the types that its folder needs are
   half-destroyed, being in the middle of refinement themselves.

 - Move the getValType functions from being static overloaded
   functions in Constants.cpp to be members of class template
   specializations in ConstantsContext.h. This means that the
   code ends up getting instantiated twice, however it also
   makes it possible to eliminate all "convert" functions, so
   it's not a big net code size increase. And if desired, the
   duplicate instantiations could be eliminated with some
   reorganization.

llvm-svn: 81861
2009-09-15 15:58:07 +00:00
Dan Gohman 21c6216c87 Teach lib/VMCore/ConstantFold.cpp how to set the inbounds keyword and
how to fold notionally-out-of-bounds array getelementptr indices instead
of just doing these in lib/Analysis/ConstantFolding.cpp, because it can
be done in a fairly general way without TargetData, and because not all
constants are visited by lib/Analysis/ConstantFolding.cpp. This enables
more constant folding.

Also, set the "inbounds" flag when the getelementptr indices are
one-past-the-end.

llvm-svn: 81483
2009-09-11 00:04:14 +00:00
Dan Gohman 7190d48075 Factor out the code for checking that all indices in a getelementptr are
within the notional bounds of the static type of the getelementptr (which
is not the same as "inbounds") from GlobalOpt into a utility routine,
and use it in ConstantFold.cpp to check whether there are any mis-behaved
indices.

llvm-svn: 81478
2009-09-10 23:37:55 +00:00
Dan Gohman 1b84908f92 Reappy r80998, now that the GlobalOpt bug that it exposed on MiniSAT is fixed.
llvm-svn: 81172
2009-09-07 23:54:19 +00:00
Daniel Dunbar 10ea8bb8e0 Revert "Include optional subclass flags, such as inbounds, nsw, etc., ...", this
breaks MiniSAT on x86_64.

llvm-svn: 81098
2009-09-06 00:11:24 +00:00
Dan Gohman 0c2477c26b Include optional subclass flags, such as inbounds, nsw, etc., in the
Constant uniquing tables. This allows distinct ConstantExpr objects
with the same operation and different flags.

Even though a ConstantExpr "a + b" is either always overflowing or
never overflowing (due to being a ConstantExpr), it's still necessary
to be able to represent it both with and without overflow flags at
the same time within the IR, because the safety of the flag may
depend on the context of the use. If the constant really does overflow,
it wouldn't ever be safe to use with the flag set, however the use
may be in code that is never actually executed.

This also makes it possible to merge all the flags tests into a single test.

llvm-svn: 80998
2009-09-04 12:08:11 +00:00
Dan Gohman e4ca02da1b Revert 80959. It isn't sufficient to solve the full problem. And it
introduced regressions in the Ocaml bindings tests.

llvm-svn: 80969
2009-09-03 23:34:49 +00:00
Dan Gohman 2a53b30f6d Remove the API for creating ConstantExprs with the nsw, nuw, inbounds,
and exact flags. Because ConstantExprs are uniqued, creating an
expression with this flag causes all expressions with the same operands
to have the same flag, which may not be safe. Add, sub, mul, and sdiv
ConstantExprs are usually folded anyway, so the main interesting flag
here is inbounds, and the constant folder already knows how to set the
inbounds flag automatically in most cases, so there isn't an urgent need
for the API support.

This can be reconsidered in the future, but for now just removing these
API bits eliminates a source of potential trouble with little downside.

llvm-svn: 80959
2009-09-03 22:17:40 +00:00
Devang Patel f7188325ef Now Bitcode reader bug is fixed. Reapply 80839.
Use CallbackVH, instead of WeakVH, to hold MDNode elements.
Use FoldingSetNode to unique MDNodes in a context.
Use CallbackVH hooks to update context's MDNodeSet appropriately.

llvm-svn: 80868
2009-09-03 01:39:20 +00:00
Devang Patel c180029be4 Revert 80839 for now. It causes test failures.
llvm-svn: 80841
2009-09-02 21:49:26 +00:00
Devang Patel 36309cd391 Use CallbackVH, instead of WeakVH, to hold MDNode elements.
Use FoldingSetNode to unique MDNodes in a context.
Use CallbackVH hooks to update context's MDNodeSet appropriately.

llvm-svn: 80839
2009-09-02 21:22:09 +00:00
Chris Lattner 78683a7013 switch a couple things off std::ostream
llvm-svn: 79816
2009-08-23 04:02:03 +00:00
Dan Gohman 16f5415f5b Rename hasNoUnsignedOverflow and hasNoSignedOverflow to hasNoUnsignedWrap
and hasNoSignedWrap, for consistency with the nuw and nsw properties.

llvm-svn: 79539
2009-08-20 17:11:38 +00:00
Erick Tryzelaar fc2280d874 Add helper functions to ConstantInt and ConstantFP to accept strings.
llvm-svn: 79212
2009-08-16 23:36:33 +00:00
Dan Gohman ff3af725ea Add a getOffsetOf, for building a target-independent expression for
offsetof, similar to getSizeOf for sizeof.

llvm-svn: 79208
2009-08-16 21:26:11 +00:00
Owen Anderson 55f1c09e31 Push LLVMContexts through the IntegerType APIs.
llvm-svn: 78948
2009-08-13 21:58:54 +00:00
Dan Gohman 77ad32ab0e Add convenience functions for creating nsw add operators.
llvm-svn: 78707
2009-08-11 20:20:39 +00:00
Dan Gohman 221a2c711d Don't set the isexact flag if an sdiv operator has been folded into
something else.

llvm-svn: 78702
2009-08-11 19:56:00 +00:00
Daniel Dunbar 919fe2bd2f Simplify ConstantExpr::getInBoundsGetElementPtr and fix a possible crash, if
constant folding eliminated the GEP instruction.
 - clang was hitting this on its test suite (for x86_64, at least).

llvm-svn: 78698
2009-08-11 18:28:09 +00:00
Dan Gohman 50c09d0ca3 Add convenience functions for creating inbounds GEPs.
llvm-svn: 78695
2009-08-11 17:57:01 +00:00
Dan Gohman 58f4d89a10 Add convenience functions for creating exact sdiv operators, and
use them in CreatePtrDiff.

llvm-svn: 78682
2009-08-11 17:05:24 +00:00
Owen Anderson 13234f83ac Change the MDNode uniquing to a ValueMap, at Devang's request.
llvm-svn: 78577
2009-08-10 18:16:08 +00:00
Owen Anderson 03cb69fbd1 Privatize the StructType table, which unfortunately involves routing contexts through a number of APIs.
llvm-svn: 78258
2009-08-05 23:16:16 +00:00
Owen Anderson 1584a29536 Privatize the last bit of Constant-creation state.
llvm-svn: 78097
2009-08-04 20:25:11 +00:00
Dan Gohman f011f5a8a2 Add a new Constant::getIntegerValue helper function, and convert a
few places in InstCombine to use it, to fix problems handling pointer
types. This fixes the recent llvm-gcc bootstrap error.

llvm-svn: 78005
2009-08-03 22:07:33 +00:00
Owen Anderson c8c302644c Privatize all but one of the remaining constant tables.
llvm-svn: 77748
2009-07-31 22:45:43 +00:00
Owen Anderson 5a1acd9912 Move a few more APIs back to 2.5 forms. The only remaining ones left to change back are
metadata related, which I'm waiting on to avoid conflicting with Devang.

llvm-svn: 77721
2009-07-31 20:28:14 +00:00
Owen Anderson 23a204d91b Move getTrue() and getFalse() to 2.5-like APIs.
llvm-svn: 77685
2009-07-31 17:39:07 +00:00
Owen Anderson b292b8ce70 Move more code back to 2.5 APIs.
llvm-svn: 77635
2009-07-30 23:03:37 +00:00
Owen Anderson 4056ca9568 Move types back to the 2.5 API.
llvm-svn: 77516
2009-07-29 22:17:13 +00:00
Owen Anderson 487375e9a2 Move ConstantExpr to 2.5 API.
llvm-svn: 77494
2009-07-29 18:55:55 +00:00
Devang Patel a4f43fb5dd Rename MDNode.h header. It defines MDnode and other metadata classes.
New name is Metadata.h.

llvm-svn: 77370
2009-07-28 21:49:47 +00:00
Owen Anderson 4aa3295a65 Return ConstantVector to 2.5 API.
llvm-svn: 77366
2009-07-28 21:19:26 +00:00
Owen Anderson c2c7932c64 Change ConstantArray to 2.5 API.
llvm-svn: 77347
2009-07-28 18:32:17 +00:00
Owen Anderson 45308b578b Move ConstantStruct back to 2.5 API.
llvm-svn: 77266
2009-07-27 22:29:26 +00:00
Daniel Dunbar ae15479eb3 Unbreak build.
llvm-svn: 77256
2009-07-27 21:47:07 +00:00
Owen Anderson 69c464dec4 Move ConstantFP construction back to the 2.5-ish API.
llvm-svn: 77247
2009-07-27 20:59:43 +00:00
Owen Anderson edb4a70325 Revert the ConstantInt constructors back to their 2.5 forms where possible, thanks to contexts-on-types. More to come.
llvm-svn: 77011
2009-07-24 23:12:02 +00:00
Chris Lattner 5cd4dd391f make Constant::getRelocationInfo return an enum, as suggested by Duncan.
llvm-svn: 76938
2009-07-24 03:27:21 +00:00
Owen Anderson 0348a13cc6 Privatize the ConstantVector tables.
llvm-svn: 76922
2009-07-24 00:36:24 +00:00
Owen Anderson 909f6001e9 Privatize the ConstantStruct table.
llvm-svn: 76912
2009-07-23 23:25:33 +00:00
Devang Patel e059ba6ed2 Derive MDNode from MetadataBase instead of Constant. Emit MDNodes into METADATA_BLOCK in bitcode file.
llvm-svn: 76834
2009-07-23 01:07:34 +00:00
Devang Patel 7428d8acec Introduce MetadataBase, a base class for MDString and MDNode.
Derive MDString directly from MetadataBase. 
Introduce new bitcode block to hold metadata.

llvm-svn: 76759
2009-07-22 17:43:22 +00:00
Chris Lattner 4565ef5b65 reimplement Constant::ContainsRelocations as
Constant::getRelocationInfo(), which has a much simpler
to use API.  It still should not be part of libvmcore, but
is better than it was.  Also teach it to be smart about 
hidden visibility.

llvm-svn: 76700
2009-07-22 00:05:44 +00:00
Owen Anderson 3d34492c1f Privatize the ConstantArray table.
llvm-svn: 76639
2009-07-21 20:55:28 +00:00
Owen Anderson 39ede7b8d1 Privatize the first of the value maps.
llvm-svn: 76634
2009-07-21 20:13:12 +00:00
Owen Anderson 2ad52176f9 Move a bit more state over to the LLVMContext.
llvm-svn: 76533
2009-07-21 02:47:59 +00:00
Dan Gohman 33a3fd0b9c Revert the addition of hasNoPointerOverflow to GEPOperator.
Getelementptrs that are defined to wrap are virtually useless to
optimization, and getelementptrs that are undefined on any kind
of overflow are too restrictive -- it's difficult to ensure that
all intermediate addresses are within bounds. I'm going to take
a different approach.

Remove a few optimizations that depended on this flag.

llvm-svn: 76437
2009-07-20 17:43:30 +00:00
Dan Gohman 7d82e1338e Make GetElementPtr ConstantExprs default to having no pointer overflow.
llvm-svn: 76280
2009-07-18 01:49:22 +00:00
Daniel Dunbar e974dc3eb1 Fix compile warning.
llvm-svn: 76210
2009-07-17 18:33:52 +00:00
Owen Anderson 4118ddeeed Privatize the MDNode uniquing table.
llvm-svn: 76126
2009-07-16 23:44:30 +00:00
Owen Anderson 69ab416d66 Privatize the MDString uniquing table.
llvm-svn: 76113
2009-07-16 22:11:26 +00:00
Owen Anderson c277dc408b Privatize the ConstantFP table. I'm on a roll!
llvm-svn: 76097
2009-07-16 19:05:41 +00:00
Owen Anderson 20b34ac794 Move the ConstantInt uniquing table into LLVMContextImpl. This exposed a number of issues in
our current context-passing stuff, which is also fixed here

llvm-svn: 76089
2009-07-16 18:04:31 +00:00
Owen Anderson f945a9ed07 Move a few more convenience factory functions from Constant to LLVMContext.
llvm-svn: 75840
2009-07-15 21:51:10 +00:00
Owen Anderson 8ee7e405dc Move the ConstantStruct factory methods over to LLVMContext.
llvm-svn: 75830
2009-07-15 21:00:46 +00:00
Rafael Espindola f5d53d46b9 Revert 75798 to fix llvm build.
llvm-svn: 75805
2009-07-15 17:40:42 +00:00
Duncan Sands c5928d2fad The static function TypeToFloatSemantics is now
unused - remove it.

llvm-svn: 75798
2009-07-15 17:19:24 +00:00
Owen Anderson b6b2530000 Move EVER MORE stuff over to LLVMContext.
llvm-svn: 75703
2009-07-14 23:09:55 +00:00
Torok Edwin fbcc663cbf llvm_unreachable->llvm_unreachable(0), LLVM_UNREACHABLE->llvm_unreachable.
This adds location info for all llvm_unreachable calls (which is a macro now) in
!NDEBUG builds.
In NDEBUG builds location info and the message is off (it only prints
"UREACHABLE executed").

llvm-svn: 75640
2009-07-14 16:55:14 +00:00
Owen Anderson 3e42e9f2fc Move more functionality over to LLVMContext.
llvm-svn: 75559
2009-07-13 23:50:59 +00:00
Owen Anderson 13c240a4c1 Move a bit more functionality to LLVMContext, which apparently wasn't being used anyways.
llvm-svn: 75546
2009-07-13 23:16:26 +00:00
Owen Anderson e4dcecd006 As Chris pointed out, this doesn't actually need an LLVMContext to operate.
llvm-svn: 75508
2009-07-13 21:27:19 +00:00
Owen Anderson 542619e6d5 Move more functionality over to LLVMContext.
llvm-svn: 75497
2009-07-13 20:58:05 +00:00
Owen Anderson 53a52215b5 Begin the painful process of tearing apart the rat'ss nest that is Constants.cpp and ConstantFold.cpp.
This involves temporarily hard wiring some parts to use the global context.  This isn't ideal, but it's
the only way I could figure out to make this process vaguely incremental.

llvm-svn: 75445
2009-07-13 04:09:18 +00:00
Torok Edwin 69208f0f9e Remove extra \n from LLVM_UNREACHABLE calls.
llvm-svn: 75416
2009-07-12 07:15:17 +00:00
Torok Edwin 56d0659726 assert(0) -> LLVM_UNREACHABLE.
Make llvm_unreachable take an optional string, thus moving the cerr<< out of
line.
LLVM_UNREACHABLE is now a simple wrapper that makes the message go away for
NDEBUG builds.

llvm-svn: 75379
2009-07-11 20:10:48 +00:00
Torok Edwin ccb29cd290 Convert more assert(0)+abort() -> LLVM_UNREACHABLE,
and abort()/exit() -> llvm_report_error().

llvm-svn: 75363
2009-07-11 13:10:19 +00:00
Nick Lewycky a21d3daadc Remove the vicmp and vfcmp instructions. Because we never had a release with
these instructions, no autoupgrade or backwards compatibility support is
provided.

llvm-svn: 74991
2009-07-08 03:04:38 +00:00
Owen Anderson 5c96ef7c4e Have scoped mutexes take referenes instead of pointers.
llvm-svn: 74931
2009-07-07 18:33:04 +00:00
Owen Anderson f92b432385 Make this const.
llvm-svn: 74317
2009-06-26 20:21:18 +00:00
Devang Patel 4c563168d5 Add constructor to create MDString using std::string
llvm-svn: 74133
2009-06-24 22:42:39 +00:00
Owen Anderson 0d2de8c35e Revert r73790, and replace it with a significantly less ugly solution. Rather than trying to make the global reader-writer lock work,
create separate recursive mutexes for each value map.  The recursive-ness fixes the double-acquiring issue, which having one per ValueMap
lets us continue to maintain some concurrency.

llvm-svn: 73801
2009-06-20 00:24:58 +00:00
Owen Anderson b07dd959a4 Fix a serious bug that would cause deadlock during abstract type refinement. The constant creation
gets involved, and we end up trying to recursively acquire a writer lock.  The fix for this is slightly horrible,
and involves passing a boolean "locked" parameter around in Constants.cpp, but it's better than having locked and
unlocked versions of most of the code.

llvm-svn: 73790
2009-06-19 23:16:19 +00:00
Owen Anderson 59ba814f39 Forgot to remove some explicit locking when it became implicit in the ValueMap.
llvm-svn: 73778
2009-06-19 18:34:09 +00:00
Owen Anderson d830eb8365 Simplify a lot of code by using a R/W mutex that becomes a no-op when multithreading is disabled.
llvm-svn: 73714
2009-06-18 19:10:19 +00:00
Owen Anderson 7d42b95b6c Move Threading.[h|cpp] from Support to System.
llvm-svn: 73707
2009-06-18 16:54:52 +00:00
Owen Anderson f89c38cace Factor out some common code.
llvm-svn: 73631
2009-06-17 20:43:39 +00:00
Owen Anderson 65c5cd728e Add an RAII ScopedWriter, which allows one to acquire a writer lock for the duration of a scope. Simplify a lot of uses of
writer locks in Constants.cpp by using it.

llvm-svn: 73630
2009-06-17 20:34:43 +00:00
Owen Anderson 61794049e4 Simplify the locking on the Constants tables, and make it more efficient, by pushing it into the ValueMap from the callers.
Document those ValueMap functions that are _not_ locked, so that callers are aware that they need to do the locking themselves.

llvm-svn: 73628
2009-06-17 20:10:08 +00:00
Owen Anderson 2d7231d4c3 Type safety for Constants.cpp! Some of this is temporary, as I'm planning to push some of the R/W locking into FoldingSet.
llvm-svn: 73624
2009-06-17 18:40:29 +00:00
Dan Gohman 7889f2b5e2 Use Type::isIntOrIntVector and Type::isFPOrFPVector.
llvm-svn: 73433
2009-06-15 22:25:12 +00:00
Dan Gohman 7ccc52f131 Support vector casts in more places, fixing a variety of assertion
failures.

To support this, add some utility functions to Type to help support
vector/scalar-independent code. Change ConstantInt::get and
ConstantFP::get to support vector types, and add an overload to
ConstantInt::get that uses a static IntegerType type, for
convenience.

Introduce a new getConstant method for ScalarEvolution, to simplify
common use cases.

llvm-svn: 73431
2009-06-15 22:12:54 +00:00
Dan Gohman a5b9645c4b Split the Add, Sub, and Mul instruction opcodes into separate
integer and floating-point opcodes, introducing
FAdd, FSub, and FMul.

For now, the AsmParser, BitcodeReader, and IRBuilder all preserve
backwards compatability, and the Core LLVM APIs preserve backwards
compatibility for IR producers. Most front-ends won't need to change
immediately.

This implements the first step of the plan outlined here:
http://nondot.org/sabre/LLVMNotes/IntegerOverflow.txt

llvm-svn: 72897
2009-06-04 22:49:04 +00:00
Nick Lewycky adbc284666 Give embedded metadata its own type instead of relying on EmptyStructTy.
llvm-svn: 72610
2009-05-30 05:06:04 +00:00
Duncan Sands d334aca93f Add a getAlignOf helper for getting the ABI alignment of a
type as a target independent constant expression.  I confess
that I didn't check that this method works as intended (though
I did test the equivalent hand-written IR a little).  But what
could possibly go wrong!

llvm-svn: 72213
2009-05-21 15:52:21 +00:00
Nick Lewycky b8f9b7a965 Make MDNode use CallbackVH. Also change MDNode to store Value* instead of
Constant* in preperation of a future change to support holding non-Constants
in an MDNode.

llvm-svn: 71407
2009-05-10 20:57:05 +00:00
Nick Lewycky 49f891958f Add support for embedded metadata to LLVM. This introduces two new types of
Constant, MDString and MDNode which can only be used by globals with a name
that starts with "llvm." or as arguments to a function with the same naming
restriction.

llvm-svn: 68420
2009-04-04 07:22:01 +00:00
Anton Korobeynikov 255a3cbfb4 Fix infinite looping
llvm-svn: 68034
2009-03-30 15:28:21 +00:00
Anton Korobeynikov d5e8e93a92 Properly propagate Kind.
llvm-svn: 68033
2009-03-30 15:28:00 +00:00
Anton Korobeynikov 7437b59caf Extend the relocation tracker handler, so we can filter on different 'kinds' of relocations required.
llvm-svn: 68004
2009-03-29 17:13:18 +00:00
Chris Lattner cbeda87da1 add a fastpath to ConstantExpr::getBitCast to handle the case when an obviously
unneeded bitcast is requested.  This is common for frontends who just unconditionally
cast even if the target is often the right type already.  THis prevents going into
getFoldedCast which switches on the opcode and does a bunch of other stuff before
doing the same opzn.

llvm-svn: 67435
2009-03-21 06:55:54 +00:00
Dan Gohman 79975d5ffb Apply a patch by Micah Villmow to fix AsmParser to accept vector
shift constant expressions, and add support for folding vector
shift constant expressions. This fixes PR3802.

llvm-svn: 67010
2009-03-14 17:09:17 +00:00
Chris Lattner e408601cec fix typo
llvm-svn: 66367
2009-03-08 04:06:26 +00:00