Commit Graph

1707 Commits

Author SHA1 Message Date
Rafael Espindola 273fd772ab Fix our handling of #pragma GCC visibility.
Now the lexer just produces a token and the parser is the one responsible for
activating it.
This fixes problem like the one pr11797 where the lexer and the parser were not
in sync. This also let us be more strict on where in the file we accept
these pragmas.

llvm-svn: 149014
2012-01-26 02:02:57 +00:00
Eli Friedman 9562f39e2f Don't stack-allocate an IntegerLiteral which can be referred to after the current method returns. PR11744, part 2.
llvm-svn: 148995
2012-01-25 23:20:27 +00:00
Eli Friedman 5995489d27 Fix a crash for an edge case of the GNU ?: extension.
llvm-svn: 148923
2012-01-25 05:04:17 +00:00
Eric Christopher f3ed1bc8ac Testcase for PR11345.
llvm-svn: 148585
2012-01-20 22:10:18 +00:00
Sebastian Pop 422377cfd3 rename -ccc-host-triple into -target
llvm-svn: 148582
2012-01-20 22:01:23 +00:00
Richard Smith f8379a0fc3 constexpr: converted constant expression handling for enumerator values, case
values and non-type template arguments of integral and enumeration types.

This change causes some legal C++98 code to no longer compile in C++11 mode, by
enforcing the C++11 rule that narrowing integral conversions are not permitted
in the final implicit conversion sequence for the above cases.

llvm-svn: 148439
2012-01-18 23:55:52 +00:00
Fariborz Jahanian 07b5f4ca03 output body of folded case again.
llvm-svn: 148361
2012-01-17 23:55:19 +00:00
Fariborz Jahanian 9ce39e299d Folding away unreachable case statement.
patch (slightly revised) by Aaron Ballman.

llvm-svn: 148359
2012-01-17 23:39:50 +00:00
Fariborz Jahanian 1352eea8c9 Another test for 148243
non-constant-folded-switch containing a constant-folded switch.

llvm-svn: 148247
2012-01-16 18:22:26 +00:00
Fariborz Jahanian 7c4dcf4b93 A little more elaborate test for r148243
llvm-svn: 148245
2012-01-16 18:07:45 +00:00
Fariborz Jahanian 60bb82a584 Patch for r148243 which was left behind.
llvm-svn: 148244
2012-01-16 17:52:22 +00:00
Richard Smith dafff94759 constexpr irgen: Add irgen support for APValue::Struct, APValue::Union,
APValue::Array and APValue::MemberPointer. All APValue values can now be emitted
as constants.

Add new CGCXXABI entry point for emitting an APValue MemberPointer. The other
entrypoints dealing with constant member pointers are no longer necessary and
will be removed in a later change.

Switch codegen from using EvaluateAsRValue/EvaluateAsLValue to
VarDecl::evaluateValue. This performs caching and deals with the nasty cases in
C++11 where a non-const object's initializer can refer indirectly to
previously-initialized fields within the same object.

Building the intermediate APValue object incurs a measurable performance hit on
pathological testcases with huge initializer lists, so we continue to build IR
directly from the Expr nodes for array and record types outside of C++11.

llvm-svn: 148178
2012-01-14 04:30:29 +00:00
Rafael Espindola 2f869a3f4f Remember if a type has its visibility set explicitly or implicitly.
With that, centralize the way we merge visibility, always preferring explicit over
implicit and then picking the most restrictive one.
Fixes pr10113 and pr11690.

llvm-svn: 148163
2012-01-14 00:30:36 +00:00
Eli Friedman d749c6bf2e Revert r148138; it's causing test failures.
llvm-svn: 148141
2012-01-13 21:33:06 +00:00
Sebastian Pop 9a8d528ddf rename -ccc-host-triple into -target
llvm-svn: 148138
2012-01-13 20:37:02 +00:00
Eli Friedman 09d272d025 A base subobject type doesn't make sense for unions; don't try to compute it. Based on patch by Yin Ma. Fixes PR11751.
llvm-svn: 148093
2012-01-13 03:58:31 +00:00
Richard Smith 02ab9c2940 Allow constant-folding of references which were formed in a manner not permitted
in a constant expression, for compatibility with g++.

llvm-svn: 148020
2012-01-12 06:08:57 +00:00
John McCall 8ba2f20bc0 Fix the mangling of class template arguments in a particular
dependent case.  Thanks to Jason Merrill for pointing this out.

llvm-svn: 147653
2012-01-06 05:06:35 +00:00
Eli Friedman b1bc368ca7 Address Richard's review comments on r147561 (Evaluate support for address-of-label differences).
llvm-svn: 147631
2012-01-05 23:59:40 +00:00
Eli Friedman 50c9d08fa5 Fix test on Release builds.
llvm-svn: 147565
2012-01-04 23:41:09 +00:00
Eli Friedman fd5e54da2d Add an APValue representation for the difference between two address-of-label expressions. Add support to Evaluate and CGExprConstant for generating/handling them. Remove the special-case for such differences in Expr::isConstantInitializer.
With that done, remove a bunch of buggy code from CGExprConstant for handling scalar expressions which is no longer necessary.

Fixes PR11705.

llvm-svn: 147561
2012-01-04 23:13:47 +00:00
Rafael Espindola 2aa7acfadb Process attributes in explicit function template instantiations. Fixes part of
PR11690.

llvm-svn: 147523
2012-01-04 05:40:59 +00:00
Rafael Espindola 0b0620776a Process attributes in explicit template instatiation definitions. Fixes PR11690.
llvm-svn: 147441
2012-01-03 06:04:21 +00:00
Rafael Espindola eeb9d9fef6 Small cosmetic cleanups in code I will change anyway.
llvm-svn: 147424
2012-01-02 06:26:22 +00:00
Rafael Espindola 0f960a0a5a Consider visibility attributes in namespaces as being explicit. I.e., they
take precedence over command line options. Fixes PR10113.

llvm-svn: 147405
2012-01-01 18:06:40 +00:00
Richard Smith fddd384b73 Unrevert r147271, reverted in r147361.
Also temporarily remove the assumption from IR gen that we can emit IR for every
constant we can fold, since it isn't currently true in C++11, to fix PR11676.

Original comment from r147271:

constexpr: perform zero-initialization prior to / instead of performing a
constructor call when appropriate. Thanks to Eli for spotting this.

llvm-svn: 147384
2011-12-30 21:15:51 +00:00
Rafael Espindola f49e9ce849 Add test for PRr11676.
llvm-svn: 147363
2011-12-30 03:13:44 +00:00
Rafael Espindola fafe4b7572 Revert r147271. This fixes PR11676.
llvm-svn: 147362
2011-12-30 03:11:50 +00:00
Rafael Espindola 0b79046c5c Fix the visibility of methods of explicit template instantiation definition
when using -fvisibility-inlines-hidden. This matches gcc's behavior and
documentation.

Fixes PR11642.

llvm-svn: 147295
2011-12-27 21:15:28 +00:00
Richard Smith 6e52514c5d constexpr: support for evaluation and codegen of typeid constants.
llvm-svn: 147290
2011-12-27 12:18:28 +00:00
Richard Smith 771c4a1775 constexpr: perform zero-initialization prior to / instead of performing a
constructor call when appropriate. Thanks to Eli for spotting this.

llvm-svn: 147271
2011-12-25 20:00:17 +00:00
Richard Smith 966c1fb0c8 Always implicitly declare move assignment operations for dynamic classes, in
case they override virtual functions from a base class. Also fix -print-stats
counting of move assignment/construction.

llvm-svn: 147258
2011-12-24 21:56:24 +00:00
Richard Smith 0f8ede1654 Revert r146766, and add a testcase for which it introduced a wrong-code bug.
llvm-svn: 146961
2011-12-20 04:00:21 +00:00
Richard Smith 764d2fe666 Unlike in C++03, a constant-expression is not an unevaluated operand in C++11.
Split out a new ExpressionEvaluationContext flag for this case, and don't treat
it as unevaluated in C++11. This fixes some crash-on-invalids where we would
allow references to class members in potentially-evaluated constant expressions
in static member functions, and also fixes half of PR10177.

The fix to PR10177 exposed a case where template instantiation failed to provide
a source location for a diagnostic, so TreeTransform has been tweaked to supply
source locations when transforming a type. The source location is still not very
good, but MarkDeclarationsReferencedInType would need to operate on a TypeLoc to
improve it further.

Also fix MarkDeclarationReferenced in C++98 mode to trigger instantiation for
static data members of class templates which are used in constant expressions.
This fixes a link-time problem, but we still incorrectly treat the member as
non-constant. The rest of the fix for that issue is blocked on PCH support for
early-instantiated static data members, which will be added in a subsequent
patch.

llvm-svn: 146955
2011-12-20 02:08:33 +00:00
Eli Friedman e321795c60 Remove a non-gcc-compatible extension that would apply attributes on declarations without a declarator to structs. Add a warning for ignored attributes. Patch by Michael Han.
llvm-svn: 146796
2011-12-17 00:36:09 +00:00
Eric Christopher 0f79160bb7 Add a triple to this test to hopefully get it passing on i386-freebsd.
llvm-svn: 146795
2011-12-17 00:35:01 +00:00
Eric Christopher cd88813dbf When in a limited debug information situation if we can find the original
declaration for the type then go ahead and use that, it's still smaller
than creating an all new derived type.

Part of rdar://10335756 and others.

llvm-svn: 146779
2011-12-16 23:40:18 +00:00
Eric Christopher ea225fb412 Start down the path of getting clang to internally agree on structs versus
classes.

Part of rdar://10520586 and a couple others.

llvm-svn: 146778
2011-12-16 23:40:14 +00:00
Nick Lewycky f43e681ed7 Testcase for LLVM commit r146475.
llvm-svn: 146476
2011-12-13 05:09:22 +00:00
Nick Lewycky 304584f5a5 s/%clang-cc1/%clang_cc1/ for conformity.
llvm-svn: 146454
2011-12-13 00:32:15 +00:00
Eli Friedman 3c840aad54 Make CGRecordLayoutBuilder correctly switch over to a packed class when a class has a base whose alignment will break the class layout. <rdar://problem/10551376>.
llvm-svn: 146443
2011-12-12 23:13:20 +00:00
Douglas Gregor c7f46f20d1 When we manage to re-use an expression during tree transformation (=
template instantiation), and that expression might produce a
temporary, invoke MaybeBindToTemporary. Otherwise, we forget to
destroy objects, release objects, etc. Fixes <rdar://problem/10531073>.

llvm-svn: 146301
2011-12-10 00:23:21 +00:00
Eli Friedman de6a86b4f6 Make array new on a pointer to data member type work correctly. PR11523.
llvm-svn: 146291
2011-12-09 23:05:37 +00:00
Eli Friedman dae858aca3 Make sure we correctly zero-initialize unions containing a pointer to data member as the first field. PR11487.
llvm-svn: 146009
2011-12-07 01:30:11 +00:00
Douglas Gregor fb9968d6f1 Fix mangling substitutions for address-space-qualified class
types. Patch from Dmitri Rubinstein!

llvm-svn: 145776
2011-12-03 18:24:43 +00:00
Richard Smith d572f82e49 Fix wrong-code bug when a const automatic variable of struct type has both a
mutable member and a constant initializer. We'd previously promoted such
variables to global constants, resulting in nasal demons if the mutable member
was modified.

This is only a temporary fix. The subtle interplay between isConstantInitializer
and CGExprConstant is very bug-prone; there are some other issues in this area
which I will be addressing in subsequent, more major reworking of this code.

llvm-svn: 145654
2011-12-02 00:30:33 +00:00
Michael J. Spencer 0567b8ec58 Add Microsoft mangling of constructors and destructors. Patch by Dmitry!
llvm-svn: 145581
2011-12-01 09:55:00 +00:00
Douglas Gregor 7372248a98 When checking a call to a builtin atomic operation, be sure to
consider the _<width> variants as well, which we'll see if we're
performing the type checking in a template instantiation where the
call expression itself was originally not type-dependent. Fixes
PR11411.

llvm-svn: 145248
2011-11-28 16:30:08 +00:00
Peter Collingbourne fbef4c8657 In Sema::MaybeBindToTemporary, create a CXXBindTemporaryExpr for an
array of objects with non-trivial destructors.  PR11365.

llvm-svn: 145203
2011-11-27 22:09:28 +00:00
Rafael Espindola 32d8a275b9 Add inreg attributes to reference arguments.
llvm-svn: 145189
2011-11-27 18:35:39 +00:00
Sebastian Redl 29526f09ce Reference initialization with initializer lists.
This supports single-element initializer lists for references according to DR1288, as well as creating temporaries and binding to them for other initializer lists.

llvm-svn: 145186
2011-11-27 16:50:07 +00:00
Richard Smith 514f7e9cb9 Don't check alignment, it's irrelevant to this test.
llvm-svn: 145093
2011-11-23 02:12:00 +00:00
Richard Smith 2d988f0f05 Use static storage duration for file-scope compound literals, even when they
appear in non-constant initializers in C++.

llvm-svn: 145087
2011-11-22 22:48:32 +00:00
Eli Friedman ee94534662 Simplify code for returning a struct for Darwin x86-32 ABI. Use a better type for a function returning a struct containing only a pointer. Handle the edge case of a struct containing only a float or double plus some dead padding instead of asserting.
llvm-svn: 144960
2011-11-18 01:25:50 +00:00
Richard Smith 027bf11eb3 Constant expression evaluation: add support for evaluation of member pointers
and base-to-derived casts, and add proper handling of temporaries.

llvm-svn: 144926
2011-11-17 22:56:20 +00:00
Eli Friedman aa9512c484 Fix test on Release builds.
llvm-svn: 144761
2011-11-16 02:19:03 +00:00
Eli Friedman d20adbdce3 Fix a bunch of really nasty bugs in how we compute alignment for reference lvalues. PR11376.
llvm-svn: 144745
2011-11-16 00:42:57 +00:00
Peter Collingbourne fcc764d140 When reference binding array rvalues, such as those created by compound
literals of array type, materialise a temporary.

llvm-svn: 144483
2011-11-13 00:51:30 +00:00
Richard Smith ce40ad677e Represent an APValue based on a Decl as that Decl, rather than a DeclRefExpr
or MemberExpr which refers to it. As a side-effect, MemberExprs which refer to
static member functions and static data members are now emitted as constant
expressions.

llvm-svn: 144468
2011-11-12 22:28:03 +00:00
Eli Friedman b29c8d55b9 Make test compatible with ARM hosts.
llvm-svn: 144428
2011-11-11 23:36:04 +00:00
John McCall 03318c1dcf Don't apply NRVO to over-aligned variables. The caller only
guarantees alignment up to the ABI alignment of the return type.

llvm-svn: 144364
2011-11-11 03:57:31 +00:00
John McCall f1a3c2aee1 Be sure to insulate block literals from any cleanups in their
enclosing full-expressions.  It is somewhat amazing that 
this hasn't come up as a problem before.

llvm-svn: 144362
2011-11-11 03:19:12 +00:00
John McCall f4beacd059 Whenever explicitly activating or deactivating a cleanup, we
need to provide a 'dominating IP' which is guaranteed to
dominate the (de)activation point but which cannot be avoided
along any execution path from the (de)activation point to
the push-point of the cleanup.  Using the entry block is
bad mojo.

llvm-svn: 144276
2011-11-10 10:43:54 +00:00
John McCall e63abb5d2b Fix a subtle bug with cleanups: when activating
a previously-inactive cleanup, not only do we need a
flag variable, but we should also force the cleanup to
query the flag variable.  However, we only need to do
this when we're activating in a context that's
conditionally executed;  otherwise, we may safely
assume that the cleanup is dominated by the activation
point.

llvm-svn: 144271
2011-11-10 09:22:44 +00:00
Benjamin Kramer 3757be395e Update test.
llvm-svn: 144185
2011-11-09 12:43:30 +00:00
Eli Friedman 1307652185 This test isn't portable to non-x86 platforms; specify a triple.
llvm-svn: 144112
2011-11-08 19:55:32 +00:00
Daniel Dunbar b3675dd660 tests: Remove some stray and unused -cc1 flags. These tests appear to be written
to go through the driver.

llvm-svn: 143791
2011-11-05 05:41:57 +00:00
Devang Patel 384dfa4f10 Enable -flimit-debug-info by default. Now, clang lazily emits debug info for structs. Original behavior can be restored using -fno-limit-debug-info.
llvm-svn: 143733
2011-11-04 20:05:58 +00:00
Douglas Gregor 10f939c3db Don't build member initializers for zero-length or incomplete arrays,
and don't try to destroy them, either. Fixes
<rdar://problem/10228639>.

llvm-svn: 143584
2011-11-02 23:04:16 +00:00
Fariborz Jahanian 10317eaa28 back out changes in r143399 and r143475.
rvale-references are captured by reference
in blocks. // rdar://9971124.

llvm-svn: 143583
2011-11-02 22:53:43 +00:00
Fariborz Jahanian ab92b753d6 Find copy constructor needed to copy an rvalue reference
c++ object into block descriptor. // rdar://9971124

llvm-svn: 143475
2011-11-01 18:57:34 +00:00
Eli Friedman fcec630a57 Fix the representation of wide strings in the AST and IR so that it uses the native representation of integers for the elements. This fixes a bunch of nastiness involving
treating wide strings as a series of bytes.

Patch by Seth Cantrell.

llvm-svn: 143417
2011-11-01 02:23:42 +00:00
Fariborz Jahanian f0cda63975 Adds IRGen support for captured rvalue references in blocks.
In this case, temporary value is copied into block descriptor
as their own copy to work on. // rdar://9971124

llvm-svn: 143399
2011-10-31 23:44:33 +00:00
Devang Patel fa59ac363e In case of template specialization, do not try to delay emitting debug info for concrete type in -flimit-debug-info mode. This fixes some of the failures from bs15503.exp tests in gdb testsuite.
llvm-svn: 143227
2011-10-28 21:12:13 +00:00
Richard Smith 11562c5e10 Reinstate r142844 (reverted in r142872) now that lvalue-to-rvalue conversions
are present in all the necessary places:

In constant expression evaluation, evaluate lvalues as lvalues and rvalues as
rvalues. Remove special case for caching reference initialization and fix a
cyclic initialization crash in the process.

llvm-svn: 143204
2011-10-28 17:51:58 +00:00
John McCall 9a43e12117 Be sure to build a dependent expression when we see
a binary operator involving a dependently-typed overload set.

llvm-svn: 143172
2011-10-28 01:04:34 +00:00
Nick Lewycky b25abd1a54 Move these test from checking assembly to checking LLVM IR. Should fix fallout
from r143097.

llvm-svn: 143098
2011-10-27 07:11:43 +00:00
Devang Patel 242ce91ecc Do not drop type qualifiers in -flimit-debug-info mode.
llvm-svn: 142873
2011-10-24 23:15:17 +00:00
Eli Friedman 9190e19b66 Don't try to emit CK_LValueBitCast casts as constants. PR9558.
llvm-svn: 142863
2011-10-24 22:25:55 +00:00
Eric Christopher 4b4beb2fe7 Fix PR11073 by adding the argument type information to the decl we construct
for the function type. Update a testcase accordingly.

Patch initially by Anders Waldenborg!

llvm-svn: 142700
2011-10-21 23:30:10 +00:00
Fariborz Jahanian 47f9a73f51 c++: support gcc's application of weak attribute on
class declaration which forces any such class and any 
class that inherits from such a class to have their 
typeinfo symbols be marked as weak.
// rdar://10246395
A    test/CodeGenCXX/weak-extern-typeinfo.cpp
M    lib/Sema/SemaDeclCXX.cpp
M    lib/Sema/SemaDeclAttr.cpp
M    lib/CodeGen/CGRTTI.cpp

llvm-svn: 142693
2011-10-21 22:27:12 +00:00
Eli Friedman 27ef75b0be Handle an edge case involving the conditional operator and throw expressions. PR10582.
llvm-svn: 142047
2011-10-15 02:10:40 +00:00
Anton Korobeynikov f82017f3fe Missed tests for half FP support
llvm-svn: 142017
2011-10-14 23:32:50 +00:00
Eli Friedman fde961dbf3 PR11124: Don't overwrite memory outside of a base class when performing zero-initialization before running its constructor.
llvm-svn: 141933
2011-10-14 02:27:24 +00:00
Richard Smith 9ca5c42582 Update all tests other than Driver/std.cpp to use -std=c++11 rather than
-std=c++0x. Patch by Ahmed Charles!

llvm-svn: 141900
2011-10-13 22:29:44 +00:00
Douglas Gregor c7a3107baf When performing a user-defined conversion via a constructor, be sure
to check whether the constructor is accessible. Fixes
<rdar://problem/10202900>.

llvm-svn: 141588
2011-10-10 22:41:00 +00:00
John McCall 0710e551ef Record layout requires not just a definition, but a complete
definition.  Assert this.  Change IR generation to not try to
aggressively emit the IR translation of a record during its
own definition.  Fixes PR10912.

llvm-svn: 141350
2011-10-07 02:39:22 +00:00
Eli Friedman ed47406871 Move test, so it actually tests what it is supposed to (given that we don't have an AST verifier).
llvm-svn: 140620
2011-09-27 18:55:06 +00:00
Peter Collingbourne 085095f1ac Move vtable dumper call to VTableBuilder ctor
llvm-svn: 140508
2011-09-26 01:57:00 +00:00
Peter Collingbourne 71c26936a0 Remove CodeGenVTables::ComputeVTableRelatedInformation dependency on CodeGen
llvm-svn: 140503
2011-09-26 01:56:36 +00:00
Sebastian Redl 12757ab4cb Treat list-initialization of scalars as a first-class citizen in C++11.
Allow empty initializer lists for scalars, which mean value-initialization.
Constant evaluation for single-element and empty initializer lists for scalars.
Codegen for empty initializer lists for scalars.
Test case comes in next commit.

llvm-svn: 140459
2011-09-24 17:48:14 +00:00
Douglas Gregor 8a556ce512 Explicitly-defaulted copy/move constructors are not "implicit", but
they still need the logic to cope with array member
initialization. Fixes PR10720.

llvm-svn: 140302
2011-09-22 15:15:51 +00:00
John McCall 30909031a7 Enforce access control for conversion operators used in contextual
conversions (rather than just call-arguments).

llvm-svn: 140244
2011-09-21 08:36:56 +00:00
Bill Wendling f0724e8e06 Throw the switch to convert clang to the new exception handling model!
This model uses the 'landingpad' instruction, which is pinned to the top of the
landing pad. (A landing pad is defined as the destination of the unwind branch
of an invoke instruction.) All of the information needed to generate the correct
exception handling metadata during code generation is encoded into the
landingpad instruction.

The new 'resume' instruction takes the place of the llvm.eh.resume intrinsic
call. It's lowered in much the same way as the intrinsic is.

llvm-svn: 140049
2011-09-19 20:31:14 +00:00
John McCall 28a5b326e1 In apple-kext mode, use external linkage for explicit template instantiations
instead of internal linkage.

llvm-svn: 140030
2011-09-19 18:05:26 +00:00
Richard Smith 20104048be PR10304: Do not call destructors for data members from union destructors. Prior to C++11, this
has no effect since any such destructors must be trivial, and in C++11 such destructors must not
be called.

llvm-svn: 139997
2011-09-18 12:11:43 +00:00
Richard Smith 4037235429 Fix PR10531. Attach an initializer to anonymous unions, since the default constructor might not be trivial (if there is an in-class initializer for some member) and might be deleted.
llvm-svn: 139991
2011-09-18 00:06:34 +00:00
Devang Patel ecaf9ac0bb Emit debug info for c++0x nullptr.
llvm-svn: 139752
2011-09-14 23:14:14 +00:00
Eli Friedman 84d2812111 Re-commit r139643.
Make clang use Acquire loads and Release stores where necessary.

llvm-svn: 139650
2011-09-13 22:21:56 +00:00
Eli Friedman acca089617 Revert r139643 while I look into it; it's breaking selfhost.
llvm-svn: 139648
2011-09-13 22:08:16 +00:00
Eli Friedman f92b2e0714 Make clang use Acquire loads and Release stores where necessary.
llvm-svn: 139643
2011-09-13 21:31:32 +00:00
Devang Patel ba15240e83 Update test to fix windows buildbot.
llvm-svn: 139513
2011-09-12 18:11:52 +00:00
Devang Patel 98ca8aeca6 Fix debug info encodings for char16_t and char32_t.
llvm-svn: 139502
2011-09-12 17:11:58 +00:00
Devang Patel 964d758d17 Emit debug info for wchar_t.
llvm-svn: 139443
2011-09-10 00:44:49 +00:00
Douglas Gregor e0e9630e07 When extracting the callee declaration from a call expression, be sure
to look through SubstNonTypeTemplateParmExprs. Then, update the IR
generation of CallExprs to actually use CallExpr::getCalleeDecl()
rather than attempting to mimick its behavior (badly).

Fixes <rdar://problem/10063539>.

llvm-svn: 139185
2011-09-06 21:41:04 +00:00
Eli Friedman cf9b1f6524 Rearrange code so that we pass the right pointer to delete[] when an exception is thrown constructing the array elements in an array new expression. Fixes PR10870.
llvm-svn: 139158
2011-09-06 18:53:03 +00:00
Richard Smith 3f9211d322 PR10506: Extend test for temporary cleanups in range-based for loop to cover the dependent case.
llvm-svn: 139109
2011-09-04 23:52:03 +00:00
Nick Lewycky 0112b11f5c Don't try to emit unsupported templated friend declarations. They're unsupported
and may very well be dependent-types, triggering an assertion in debug info
codegen.

llvm-svn: 138970
2011-09-01 21:49:51 +00:00
John McCall 3480ef24d1 The size of struct UnwindException varies by platform with no
apparent general rule.  Just special-case it as appropriate.
PR10789.

llvm-svn: 138792
2011-08-30 01:42:09 +00:00
John McCall 154a2fd3cb Be sure to emit lvalue-to-rvalue casts for loads from x-values.
Doing this happens to disrupt the pattern that ARC was looking for
for move optimizations, so we need to fix that simultaneously.

llvm-svn: 138789
2011-08-30 00:57:29 +00:00
John McCall 29ad95b232 The lvalue-to-rvalue on structs in C++ is actually part
of default argument promotion and needs to happen unconditionally.
This is particularly semantically important in C++0x.

llvm-svn: 138691
2011-08-27 01:09:30 +00:00
John McCall 46759f4f46 Since the 'is aliased' bit is critical for correctness in C++, it
really shouldn't be optional.  Fix the remaining place where a
temporary was being passed as potentially-aliased memory.

Fixes PR10756.

llvm-svn: 138627
2011-08-26 07:31:35 +00:00
John McCall a8a39bc346 An initialization does not alias.
llvm-svn: 138624
2011-08-26 05:38:08 +00:00
John McCall 35e4f0cd0f The allocated exception slot does not alias anything; should fix self-host.
llvm-svn: 138615
2011-08-26 00:46:38 +00:00
John McCall a5efa7386a Track whether an AggValueSlot is potentially aliased, and do not
emit call results into potentially aliased slots.  This allows us
to properly mark indirect return slots as noalias, at the cost
of requiring an extra memcpy when assigning an aggregate call
result into a l-value.  It also brings us into compliance with
the x86-64 ABI.

llvm-svn: 138599
2011-08-25 23:04:34 +00:00
Chandler Carruth 4e6bd3955f Remove this test. It requires C++ standard library headers, which we'd
like to avoid within the Clang test suite, it doesn't verify the output
at all so it only servers as a test that Clang doesn't crash, and
finally all it does is declare a function that returns a vector and call
it. Probably the biggest thing being tested here is Clang's parsing of
the vector template, and we have lots of good template parsing tests. We
don't need another in codegen.

llvm-svn: 138197
2011-08-20 06:30:57 +00:00
NAKAMURA Takumi d0dc49e640 test/CodeGenCXX/2003-11-25-ReturningOpaqueByValue.cpp: Mark XFAIL: win32.
FIXME: Avoid using system headers. clang cannot handle MS <vector> yet.
llvm-svn: 138190
2011-08-20 05:38:29 +00:00
Eric Christopher 50507663a6 Remove this file, it's not much of a test and string headers cause
problems on windows.

llvm-svn: 138166
2011-08-20 00:22:40 +00:00
Eric Christopher aab7dffa12 Migrate, FileCheckize and update:
2003-11-02-WeakLinkage.cpp
2003-11-18-PtrMemConstantInitializer.cpp
2003-11-25-ReturningOpaqueByValue.cpp
2003-11-27-MultipleInheritanceThunk.cpp
2003-11-29-DuplicatedCleanupTest.cpp
2003-12-08-ArrayOfPtrToMemberFunc.cpp
2004-01-11-DynamicInitializedConstant.cpp

from llvm/test/FrontendC++.

llvm-svn: 138162
2011-08-20 00:17:18 +00:00
Eric Christopher 6672b33511 Migrate, FileCheckize and update:
2004-03-08-ReinterpretCastCopy.cpp
2004-03-09-UnmangledBuiltinMethods.cpp
2004-03-15-CleanupsAndGotos.cpp
2004-06-08-LateTemplateInstantiation.cpp
2004-09-27-CompilerCrash.cpp
2004-09-27-DidntEmitTemplate.cpp
2004-11-27-ExceptionCleanupAssertion.cpp
2004-11-27-FriendDefaultArgCrash.cpp
2005-01-03-StaticInitializers.cpp

from llvm/test/FrontendC++.

llvm-svn: 138157
2011-08-20 00:09:39 +00:00
Eric Christopher 19eb735e27 Modify the check line to be happier on windows.
llvm-svn: 138152
2011-08-19 23:46:18 +00:00
Eric Christopher 862f8ecf55 Remove dead file.
llvm-svn: 138150
2011-08-19 23:45:03 +00:00
Eric Christopher 3bf8310d42 Migrate, FileCheckize and update:
2005-02-11-AnonymousUnion.cpp
2005-02-13-BadDynamicInit.cpp
2005-02-14-BitFieldOffset.cpp
2005-02-19-BitfieldStructCrash.cpp
2005-02-19-UnnamedVirtualThunkArgument.cpp
2005-02-20-BrokenReferenceTest.cpp
2006-03-01-GimplifyCrash.cpp
2006-03-06-C++RecurseCrash.cpp
2006-09-12-OpaqueStructCrash.cpp
2006-10-30-ClassBitfield.cpp
2006-11-20-GlobalSymbols.cpp
2006-11-20-GlobalSymbols.ll
2006-11-30-ConstantExprCrash.cpp

from llvm/test/FrontendC++.

llvm-svn: 138148
2011-08-19 23:41:35 +00:00
Eric Christopher 925d593b60 Try removing xfail and xtarget to see if this brings back a couple of bots.
llvm-svn: 138139
2011-08-19 23:17:46 +00:00
Eric Christopher a8cd70e21a Migrate 2007-01-02-UnboundedArray.cpp from llvm/test/FrontendC++.
llvm-svn: 138138
2011-08-19 23:15:04 +00:00
Eric Christopher 17db038966 Migrate:
2007-01-06-PtrMethodInit.cpp
2007-04-05-PackedBitFields-1.cpp
2007-04-05-PackedBitFieldsOverlap-2.cpp
2007-04-05-PackedBitFieldsOverlap.cpp
2007-04-05-PackedBitFieldsSmall.cpp
2007-04-05-StructPackedFieldUnpacked.cpp
2007-04-10-PackedUnion.cpp
2007-04-14-FNoBuiltin.cpp
2007-05-03-VectorInit.cpp
2007-07-29-RestrictPtrArg.cpp
2007-07-29-RestrictRefArg.cpp
2007-09-10-RecursiveTypeResolution.cpp

from llvm/test/FrontendC++ and FileCheckize where appropriate.

llvm-svn: 138134
2011-08-19 23:08:33 +00:00
Eric Christopher 7e88fdfecc Remove 2008-10-29-WrongOffset.cpp since it requires a header file on all
platforms.

llvm-svn: 138118
2011-08-19 22:25:04 +00:00
Eric Christopher 5478583ac1 Migrate 2007-10-01-StructResize.cpp from llvm/test/FrontendC++.
llvm-svn: 138117
2011-08-19 22:25:02 +00:00
Eric Christopher cae25b58fc Migrate 2008-01-12-VecInit.cpp from llvm/test/FrontendC++.
llvm-svn: 138114
2011-08-19 22:16:30 +00:00
Eric Christopher 88279b7eb6 Migrate 2008-05-07-CrazyOffsetOf.cpp from llvm/test/FrontendC++.
llvm-svn: 138113
2011-08-19 22:16:28 +00:00
Eric Christopher fd4d7156ea Migrate 2008-10-29-WrongOffset.cpp from llvm/test/FrontendC++.
llvm-svn: 138112
2011-08-19 22:16:25 +00:00
Eric Christopher a82c55ecf0 Migrate 2009-03-17-dbg.cpp from llvm/test/FrontendC++.
llvm-svn: 138111
2011-08-19 22:16:23 +00:00
Eric Christopher 20429e90d7 Migrate 2009-04-23-bool2.cpp from llvm/test/FrontendC++.
llvm-svn: 138110
2011-08-19 22:16:21 +00:00
Eric Christopher b991a6ae87 Remove this test. It's failing and it's not that good of a test.
llvm-svn: 138106
2011-08-19 22:00:44 +00:00
Eric Christopher b52ebc3e29 Migrate 2009-06-16-DebugInfoCrash.cpp from llvm/test/FrontendC++.
llvm-svn: 138099
2011-08-19 21:51:07 +00:00
Eric Christopher 0c5f2a3bf1 Migrate 2009-07-16-PrivateCopyConstructor.cpp from llvm/test/FrontendC++.
llvm-svn: 138098
2011-08-19 21:51:05 +00:00
Eric Christopher 69d6c3d31a Migrate 2009-08-05-ZeroInitWidth.cpp from llvm/test/FrontendC++.
llvm-svn: 138097
2011-08-19 21:51:03 +00:00
Eric Christopher 336a2bf4dd Migrate 2009-07-16-Using.cpp from llvm/test/FrontendC++.
llvm-svn: 138096
2011-08-19 21:51:01 +00:00
Eric Christopher 824e2e702f Migrate 2009-08-11-VectorRetTy.cpp from llvm/test/FrontendC++.
llvm-svn: 138095
2011-08-19 21:50:59 +00:00
Eric Christopher 1c241f741e Migrate 2009-09-09-packed-layout.cpp test from llvm/test/FrontendC++.
llvm-svn: 138090
2011-08-19 21:24:41 +00:00
Eric Christopher 65534ba4a4 Migrate 2009-10-27-crash.cpp from llvm/test/FrontendC++.
llvm-svn: 138089
2011-08-19 21:24:39 +00:00
Eric Christopher 16bfebc42c Migrate 2009-12-23-MissingSext.cpp from test/FrontendC++ and modify
for sext -> and difference.

llvm-svn: 138088
2011-08-19 21:24:37 +00:00
Nick Lewycky d78cfd57eb Do not depend on the standard library on the system. Replace the #include with
the relevant section of libcxx.

As an aside, I am not at all confident that this test is still testing that it's
supposed to.

llvm-svn: 138075
2011-08-19 20:43:24 +00:00
Chandler Carruth 48b2831d8a Always mark friend function declarations in class templates as
implicitly instantiable, even if we don't see a body on the friend
function declaration. The body may simply have not yet been attached.
This fixes PR10666.

There may be an alternate, preferred implementation strategy, see my
FIXME. Review would definitely be appreciated Doug. =D

llvm-svn: 137934
2011-08-18 09:09:59 +00:00
John McCall 7d92c70e6f Somehow I fouled up this test.
llvm-svn: 137882
2011-08-17 22:04:02 +00:00
John McCall 75f92b519c Gather cleanups correctly in block return statements.
Thanks to Ted for finding this with magic tools.

llvm-svn: 137877
2011-08-17 21:34:14 +00:00
Eric Christopher 43b1188a71 Migrate test from llvm/test/FrontendC++.
llvm-svn: 137773
2011-08-16 21:41:54 +00:00
Eric Christopher ed78181605 Migrate from llvm/test/FrontendC++ and FileCheckize.
llvm-svn: 137772
2011-08-16 21:41:52 +00:00
Eric Christopher 77214ad172 Migrate from llvm/test/FrontendC++.
llvm-svn: 137771
2011-08-16 21:41:50 +00:00
Eric Christopher 38a7c55cae Migrate test from llvm/test/FrontendC++ and FileCheckize.
llvm-svn: 137770
2011-08-16 21:41:48 +00:00
Eric Christopher 88fa0f0eb7 Migrate test from llvm/test/FrontendC++.
llvm-svn: 137769
2011-08-16 21:41:46 +00:00
Eric Christopher 3f98c99b07 Migrate from llvm/test/FrontendC++.
llvm-svn: 137768
2011-08-16 21:41:44 +00:00
Eric Christopher 82da54d872 Migrate from llvm/test/FrontendC++.
llvm-svn: 137767
2011-08-16 21:41:41 +00:00
Eric Christopher 2e4877fcb2 Migrate test from llvm/test/FrontendC++.
llvm-svn: 137766
2011-08-16 21:41:40 +00:00
Eric Christopher f1b912b88e Migrate from llvm/test/FrontendC++.
llvm-svn: 137765
2011-08-16 21:41:38 +00:00
Eric Christopher 74733d2e12 Migrate varargs.cpp from llvm/test/FrontendC++.
llvm-svn: 137764
2011-08-16 21:41:35 +00:00
Eric Christopher 9eeae42a6e Migrate weak-external.cpp from llvm/test/FrontendC++.
llvm-svn: 137763
2011-08-16 21:41:33 +00:00
Eric Christopher 148bdfa58a Migrate x86-64-abi-sret-vs-2word-struct-param.cpp from
llvm/test/FrontendC++ and FileCheckize.

llvm-svn: 137762
2011-08-16 21:41:31 +00:00
Eric Christopher 8a24ed567c Migrate thunk-linkonce-odr.cpp from llvm/test/FrontendC++.
llvm-svn: 137761
2011-08-16 21:41:26 +00:00
Eric Christopher d9afc5c749 FileCheckize this test.
llvm-svn: 137671
2011-08-15 22:48:14 +00:00
Eric Christopher bf005ecd9c 'pure' and 'const' functions should also be marked nounwind. Migrate
test over from llvm/test/FrontendC++ and update others to account for
the change.

llvm-svn: 137669
2011-08-15 22:38:22 +00:00
Eli Friedman 0be3970804 Fix r137086 to actually work properly in general. PR10650.
llvm-svn: 137574
2011-08-14 04:50:34 +00:00
Eli Friedman 729ed799c7 Update clang tests for r137527.
llvm-svn: 137535
2011-08-12 23:33:52 +00:00
John McCall 8e4c74bb7c Simplify EH control flow by observing that EH scopes form a simple
hierarchy of delegation, and that EH selector values are meaningful
function-wide (good thing, too, or inlining wouldn't work).
2,3d
1a
hierarchy of delegation and that EH selector values have the same
meaning everywhere in the function instead of being meaningful only
in the context of a specific selector.

This removes the need for routing edges through EH cleanups,
since a cleanup simply always branches to its enclosing scope.

llvm-svn: 137293
2011-08-11 02:22:43 +00:00
Douglas Gregor 493627ba8b Rewrite default initialization of anonymous structs/unions within a
constructor. Previously, we did some bogus recursion into the fields
of anonymous structs (recursively), which ended up building invalid
ASTs that would cause CodeGen to crash due to invalid GEPs.

Now, we instead build the default initializations based on the
indirect field declarations at the top level, which properly generates
the sequence of GEPs needed to initialize the proper member. Fixes
PR10512 and <rdar://problem/9924046>.

llvm-svn: 137212
2011-08-10 15:22:55 +00:00
Douglas Gregor 34f6c6db26 Hand materialization of temporary expressions when emitting a scalar
expression. Fxies PR10592.

llvm-svn: 137086
2011-08-09 00:37:14 +00:00
Fariborz Jahanian c1a151bed9 Don't emit memcpy for copying fields of arrays of volatile elements.
Use the the path that generates a loop. This fixes
bogus error that clang puts out. // rdar://9894548

llvm-svn: 137080
2011-08-09 00:26:11 +00:00
John McCall f82bdf6dd1 Be sure to destroy the normal entry block of a cleanup that we
aren't actually going to make a normal cleanup for.  Sometimes
we optimistically create branches to such blocks for fixups,
and then we resolve the fixup to somewhere within the cleanup's
scope, and then the cleanup is actually not reachable for some
reason.  The process of resolving the fixup leaves us with
switches whose default edge leads to the cleanup;  we can
replace that with unreachable, then (in many cases) turn
the switch into an unconditional branch.

Fixes PR10467.

llvm-svn: 137011
2011-08-06 06:53:52 +00:00
Eli Friedman b23533db13 PR10566: Make sure codegen for deleting an pointer to an incomplete type actually works.
llvm-svn: 136703
2011-08-02 18:05:30 +00:00
Eli Friedman 9d155aa061 Fix typo in test.
llvm-svn: 136251
2011-07-27 20:09:36 +00:00
Eli Friedman b2f3695e9a Make CodeGen for array delete involving incomplete class work without crashing. Should fix regression on g++.dg/init/delete1.C.
llvm-svn: 136241
2011-07-27 18:54:57 +00:00
Douglas Gregor 8771796493 Disable the optimization that skips emission of complete, non-virtual
destructors of abstract classes. It's undefined behavior to actually
call the destructor (e.g., via delete), but the presence of code that
calls this destructor doesn't make the program
ill-formed. Fixes <rdar://problem/9819242>.

llvm-svn: 136180
2011-07-26 23:18:30 +00:00
Chris Lattner e135b089d8 fix PR10384: C++ allows external arrays of incomplete type as well.
Many thanks to Eli for reducing this great testcase.

llvm-svn: 135752
2011-07-22 06:27:26 +00:00
Chris Lattner 43cae02689 rename test.
llvm-svn: 135751
2011-07-22 06:26:12 +00:00
Chris Lattner 71bd0c3263 fix PR10395 - array decay can produce an interesting type when
decaying an array of incomplete type (which has type [0 x i8]*) to a
normal pointer (which has incompletetype*).

llvm-svn: 135565
2011-07-20 04:31:01 +00:00
Bruno Cardoso Lopes 98154a76fd Reapply r134946 with fixes. Tested on Benjamin testcase and other test-suite failures.
llvm-svn: 135091
2011-07-13 21:58:55 +00:00
John McCall 928a25714e Enforce access control for the destructor in a new[] expression and mark
it as used.  Otherwise, we can fail to instantiate or validate the destructor,
which can lead to crashes in IR gen like PR10351.

llvm-svn: 135073
2011-07-13 20:12:57 +00:00
John McCall 97eab0a271 Okay, that rule about zero-length arrays applies to destroying
them, too.

llvm-svn: 135038
2011-07-13 08:09:46 +00:00
John McCall 6549b31fe9 Arrays are permitted to be zero-length in some situations.
llvm-svn: 135036
2011-07-13 07:37:11 +00:00
John McCall f677a8e99e Convert the standard default-construction loops to use phis and
partial destruction.

llvm-svn: 135033
2011-07-13 06:10:41 +00:00
Douglas Gregor d29fde248e Eliminate an incomplete/incorrect attempt to provide support for C++0x
unrestricted unions, which ended up attempting to initialize objects
in a union (which CodeGen isn't prepared for). Fixes PR9683.

llvm-svn: 135027
2011-07-13 02:14:02 +00:00
John McCall ca2c56f20b Switch delete[] IR-generation over to the destroy framework,
which implicitly makes it EH-safe as well.

llvm-svn: 135025
2011-07-13 01:41:37 +00:00
Douglas Gregor 1c2e20d73d When compiling ::delete for a class with a virtual destructor, call
the complete destructor and then invoke the global delete
operator. Previously, we would invoke the deleting destructor, which
calls the wrong delete operator. Fixes PR10341.

llvm-svn: 135021
2011-07-13 00:54:47 +00:00
Douglas Gregor 2207ec273a Improve name mangling for instantiation-dependent types that are not
dependent. This covers an odd class of types such as

  int (&)[sizeof(sizeof(T() + T()))];

which involve template parameters but, because of some trick typically
involving a form of expression that is never type-dependent, resolve
down to a non-dependent type. Such types need to be mangled
essentially as they were written in the source code (involving
template parameters), rather than via their canonical type.

In general, instantiation-dependent types should be mangled as
they were written in the source. However, since we can't do that now
without non-trivial refactoring of the AST (see the new FIXME), I've
gone for this partial solution: only use the as-written-in-the-source
mangling for these strange types that are instantiation-dependent but
not dependent. This provides better compatibility with previous
incarnations of Clang and with GCC. In the future, we'd like to get
this right.

Fixes <rdar://problem/9663282>.

llvm-svn: 134984
2011-07-12 15:18:55 +00:00
Douglas Gregor e5802234f9 Implement name mangling for sizeof...(function parameter pack).
llvm-svn: 134974
2011-07-12 07:03:48 +00:00
Chris Lattner 3f32d69699 Fix a problem Eli ran into where we now reject incomplete arrays of
uncompleted struct types.  We now do what llvm-gcc does and compile
them into [i8 x 0].  If the type is later completed, we make sure that
it is appropriately cast.

We compile the terrible example to something like this now:

%struct.A = type { i32, i32, i32 }

@g = external global [0 x i8]

define void @_Z1fv() nounwind {
entry:
  call void @_Z3fooP1A(%struct.A* bitcast ([0 x i8]* @g to %struct.A*))
  ret void
}

declare void @_Z3fooP1A(%struct.A*)

define %struct.A* @_Z2f2v() nounwind {
entry:
  ret %struct.A* getelementptr inbounds ([0 x %struct.A]* bitcast ([0 x i8]* @g to [0 x %struct.A]*), i32 0, i64 1)
}

llvm-svn: 134972
2011-07-12 06:52:18 +00:00
Chris Lattner 73e3004e75 fix an unintended behavior change in the type system rewrite, which caused us to compile
stuff like this:

typedef struct {
 int x, y, z; 
} foo_t;

foo_t g;

into:
%"struct.<anonymous>" = type { i32, i32, i32 }
we now get:
%struct.foo_t = type { i32, i32, i32 }

This doesn't change the behavior of the compiler, but makes the IR much easier to read.

llvm-svn: 134969
2011-07-12 05:53:08 +00:00
Douglas Gregor 1cf9dd21dc Mangle dependent template names of unknown arity
llvm-svn: 134967
2011-07-12 05:06:05 +00:00
Douglas Gregor 6ef06cc48d Implement the Itanium C++ ABI's mangling rule for
non-instantiation-dependent sizeof and alignof expressions.

llvm-svn: 134963
2011-07-12 04:47:20 +00:00
Douglas Gregor 2e12550e37 Allow us to compute linkage et al for instantiation-dependent types.
llvm-svn: 134932
2011-07-11 22:38:07 +00:00
John McCall 3b935d33a1 Emit partial destruction of structs with initializer lists.
llvm-svn: 134913
2011-07-11 19:35:02 +00:00
John McCall 178360e1cd Fix a lot of problems with the partial destruction of arrays:
- an off-by-one error in emission of irregular array limits for
   InitListExprs
 - use an EH partial-destruction cleanup within the normal
   array-destruction cleanup
 - get the branch destinations right for the empty check
Also some refactoring which unfortunately obscures these changes.

llvm-svn: 134890
2011-07-11 08:38:19 +00:00
Chris Lattner 709af79e0a relax this test a bit to hopefully work with VC++
llvm-svn: 134870
2011-07-10 14:52:52 +00:00
Chris Lattner 18ca0475d7 implement a nice new optimization: CodeGenTypes::UpdateCompletedType
is called whenever a tag type is completed.  We previously used that
as the sign to layout the codegen representation for the tag type,
which worked but meant that we laid out *every* completed type, whether
it was used or not.

Now we just lay out the type if we've already seen it somehow else.
This means that we lay out types we've used but haven't seen a body
for, but we don't lay out tons of stuff that noone cares about.

llvm-svn: 134866
2011-07-10 06:03:22 +00:00
Chris Lattner 117b724b59 keep track of whether being in a RS_StructPointer state
caused us to skip layout out a function accurately.  If
so, flush the type cache for both the function and struct
case to ensure that any pointers to the functions get
recomputed.  This is overconservative, but with this patch
clang can build itself again.

llvm-svn: 134863
2011-07-10 05:39:13 +00:00
Chris Lattner d1f1158203 Fix the clang bootstrap and Jay's testcase from llvm-dev by being completely
conservative when converting a functiontype to IR when in a "pointer within
a struct" context.  This has the unfortunate sideeffect of compiling all 
function pointers inside of structs into "{}*" which, though correct, is
ugly.  This has the positive side effect of being correct, and it is pretty
straight-forward to improve on this.

llvm-svn: 134861
2011-07-10 03:47:27 +00:00
Chris Lattner f5926b460b fix this on 32-bit hosts.
llvm-svn: 134840
2011-07-09 18:33:01 +00:00
Chris Lattner a5f58b05e8 clang side to match the LLVM IR type system rewrite patch.
llvm-svn: 134831
2011-07-09 17:41:47 +00:00
John McCall 82fe67bb82 A number of array-related IR-gen cleanups.
- Emit default-initialization of arrays that were partially initialized
    with initializer lists with a loop, rather than emitting the default
    initializer N times;
  - support destroying VLAs of non-trivial type, although this is not
    yet exposed to users; and
  - support the partial destruction of arrays initialized with
    initializer lists when an initializer throws an exception.

llvm-svn: 134784
2011-07-09 01:37:26 +00:00
Rafael Espindola ba195cff81 Use attributes from the definition (if available) when
instantiating functions.

Fixes PR10272.

llvm-svn: 134491
2011-07-06 15:46:09 +00:00
John McCall 542e7c6f64 When tree-transforming an expression sequence, always flag expanded
variadic argument pack expansions as having changed, rather than doing
it for each changed expansion, which leaves out zero-argument packs
with catastrophic consequences.

Fixes PR10260.

llvm-svn: 134483
2011-07-06 07:30:07 +00:00
Douglas Gregor fa918f69f6 Emit guard variables for any weak global that has a run-time
initializer. Previously, we only used guard variables for weak static
data members. Fixes <rdar://problem/9692249>.

llvm-svn: 134266
2011-07-01 21:54:36 +00:00
Douglas Gregor c063ac20d8 Fix testcase
llvm-svn: 134262
2011-07-01 21:19:15 +00:00
Douglas Gregor 82e1af20cb Don't zero-initialize default-initialized local variables that have
trivial default constructors. This generated-code regression was
caused by r131796, which had simplified the handling of default
initialization in Sema. Fixes <rdar://problem/9694300>.

llvm-svn: 134260
2011-07-01 21:08:19 +00:00
John McCall 2de1c33f77 Just mangle substituted template parameter types as unresolved types.
This is kindof questionable but seems to do more-or-less the right thing.
This is not a particularly friendly part of the ABI.

llvm-svn: 134227
2011-07-01 02:19:08 +00:00
Douglas Gregor 678d76c026 Introduce the notion of instantiation dependence into Clang's AST. A
type/expression/template argument/etc. is instantiation-dependent if
it somehow involves a template parameter, even if it doesn't meet the
requirements for the more common kinds of dependence (dependent type,
type-dependent expression, value-dependent expression).

When we see an instantiation-dependent type, we know we always need to
perform substitution into that instantiation-dependent type. This
keeps us from short-circuiting evaluation in places where we
shouldn't, and lets us properly implement C++0x [temp.type]p2.

In theory, this would also allow us to properly mangle
instantiation-dependent-but-not-dependent decltype types per the
Itanium C++ ABI, but we aren't quite there because we still mangle
based on the canonical type in cases like, e.g.,

  template<unsigned> struct A { };
  template<typename T>
    void f(A<sizeof(sizeof(decltype(T() + T())))>) { }
  template void f<int>(A<sizeof(sizeof(int))>);

and therefore get the wrong answer.

llvm-svn: 134225
2011-07-01 01:22:09 +00:00
John McCall 3a4a4c5228 Change the mangling of enclosing template template parameters
that serve as the base template name of an unresolved-name to
be mangled as a substitution.

llvm-svn: 134213
2011-07-01 00:04:39 +00:00
Eli Friedman 4774b7e456 We don't pass classes with a copy-constructor or destructor byval, so the address takes up an integer register (if one is available). Make sure the x86-64 ABI implementation takes that into account properly.
The fixed implementation is compatible with the implementation both gcc and llvm-gcc use.

rdar://9686430 . (This is the issue that was reported in the thread "[LLVMdev] Segfault calling LLVM libs from a clang-compiled executable".)

llvm-svn: 134059
2011-06-29 07:04:55 +00:00
John McCall 15547bbdd1 Be more thorough about mangling unresolved types.
llvm-svn: 134011
2011-06-28 16:49:23 +00:00
John McCall b8c604abfd Don't factor in visibility for templates and template arguments
for explicit specializations with their own explicit visibility.

llvm-svn: 133958
2011-06-27 23:06:04 +00:00
Douglas Gregor 3bdcff5866 When instantiating a C++ "new" expression, don't fake source locations
for the '(' and ')' around the initializer unless we actually have an
initializer. Fixes PR10197, an issue where we were value-initializing
rather than default-initializing.

llvm-svn: 133913
2011-06-27 16:55:54 +00:00
John McCall 77527a8e65 Mark the multiply which occurs as part of performing pointer
arithmetic on a VLA as 'nsw', per discussion with djg, and
implement pointer arithmetic (other than array accesses) and
pointer subtraction for VLA types.

llvm-svn: 133855
2011-06-25 01:32:37 +00:00
John McCall 8766a83a2f Two more test cases which have been long uncommitted.
llvm-svn: 133842
2011-06-24 23:30:52 +00:00
John McCall 23c29fea92 Change the IR-generation of VLAs so that we capture bounds,
not sizes;  so that we use well-typed allocas;  and so that we
properly recurse through the full set of variably-modified types.

llvm-svn: 133827
2011-06-24 21:55:10 +00:00
John McCall 575fda19b3 Fix the mangling of dependent-scope decl ref expressions so that they
use the unresolved-name production correctly.

llvm-svn: 133554
2011-06-21 22:12:46 +00:00
Douglas Gregor fe31481f68 Introduce a new AST node describing reference binding to temporaries.
MaterializeTemporaryExpr captures a reference binding to a temporary
value, making explicit that the temporary value (a prvalue) needs to
be materialized into memory so that its address can be used. The
intended AST invariant here is that a reference will always bind to a
glvalue, and MaterializeTemporaryExpr will be used to convert prvalues
into glvalues for that binding to happen. For example, given

  const int& r = 1.0;

The initializer of "r" will be a MaterializeTemporaryExpr whose
subexpression is an implicit conversion from the double literal "1.0"
to an integer value. 

IR generation benefits most from this new node, since it was
previously guessing (badly) when to materialize temporaries for the
purposes of reference binding. There are likely more refactoring and
cleanups we could perform there, but the introduction of
MaterializeTemporaryExpr fixes PR9565, a case where IR generation
would effectively bind a const reference directly to a bitfield in a
struct. Addresses <rdar://problem/9552231>.

llvm-svn: 133521
2011-06-21 17:03:29 +00:00
Chris Lattner e64d7ba153 Update to match mainline ConstantStruct::get API change. Also, use
ConvertType on InitListExprs as they are being converted.  This is
needed for a forthcoming patch, and improves the IR generated anyway
(see additional type names in testcases). 

This patch also converts a bunch of std::vector's in CGObjCMac to use
C arrays.  There are a ton more that should be converted as well.

llvm-svn: 133413
2011-06-20 04:01:35 +00:00
John McCall 5aa5259f5e Perform an acquire memory barrier on the fast path of a thread-safe
static initializer check, as required by the Itanium ABI.

llvm-svn: 133250
2011-06-17 07:33:57 +00:00
Douglas Gregor 9b71f0cfac Implement proper support for generating code for compound literals in
C++, which means:
  - binding the temporary as needed in Sema, so that we generate the
  appropriate call to the destructor, and
  - emitting the compound literal into the appropriate location for
  the aggregate, rather than trying to emit it as a temporary and
  memcpy() it.

Fixes PR10138 / <rdar://problem/9615901>.

llvm-svn: 133235
2011-06-17 04:59:12 +00:00
John McCall 1b1a1dbbe7 When synthesizing implicit copy/move constructors and copy/move assignment
operators, don't make an initializer or sub-operation for zero-width
bitfields.

llvm-svn: 133221
2011-06-17 00:18:42 +00:00
Eli Friedman e2a79e9ce8 Fix a couple more tests with ARM ABI.
llvm-svn: 133073
2011-06-15 18:37:44 +00:00
Eli Friedman 50ed150632 Whack a bunch of tests in CodeGenCXX to work on ARM (using ARM ABI). Batch 2 of 3.
llvm-svn: 133011
2011-06-14 21:20:53 +00:00
Eli Friedman a526f275e3 Whack a bunch of tests in CodeGenCXX to work on ARM (using ARM ABI). Batch 1 of 3 or so.
llvm-svn: 132945
2011-06-13 22:51:21 +00:00
Richard Smith 938f40b5aa Implement support for C++11 in-class initialization of non-static data members.
llvm-svn: 132878
2011-06-11 17:19:42 +00:00
Eli Friedman 300f55dcad PR10120: Make CodeGenModule::getVTableLinkage use NamedDecl::getLinkage to determine whether the vtable should be externally visible, instead of a rough approximation of it which messes up with templates.
While I'm here, zap the other user of isInAnonymousNamespace outside of Decl.cpp.

llvm-svn: 132861
2011-06-10 21:53:06 +00:00
Douglas Gregor c507db4f70 Add name mangling for expr .* expr. Fixes PR9983 / <rdar://problem/9486332>.
llvm-svn: 132659
2011-06-05 05:27:58 +00:00
Galina Kistanova a6faf8e28b These tests require particular registered targets. Declared as such.
llvm-svn: 132623
2011-06-04 04:38:16 +00:00
Galina Kistanova f56b4f6fdd These tests require particular registered targets. Declared as such.
llvm-svn: 132600
2011-06-03 22:24:54 +00:00
Devang Patel 2780e4545a List c++ class type as public type in dwarf debug info output.
llvm-svn: 132357
2011-05-31 20:46:46 +00:00
Chandler Carruth e54ff6cc3e Expand the coverage of the warning for constants on the RHS of logical operands:
return f() || -1;

where the user meant to write '|'.

This bootstraps without any additional warnings.

Patch by Richard Trieu.

llvm-svn: 132327
2011-05-31 05:41:42 +00:00
John McCall 9b382dde92 Convert Clang over to resuming from landing pads with llvm.eh.resume.
It's quite likely that this will explode, but I need to know how. :)

llvm-svn: 132269
2011-05-28 21:13:02 +00:00
Eli Friedman 380b8dad6b Back out r132209; it's breaking nightly tests.
llvm-svn: 132219
2011-05-27 21:32:17 +00:00
John McCall 63fb333fa4 Implement a new, much improved version of the cleanup hack. We just need
to be careful to emit landing pads that are always prepared to handle a
cleanup path.  This is correct mostly because of the fix to the LLVM
inliner, r132200.

llvm-svn: 132209
2011-05-27 20:01:14 +00:00
Chandler Carruth 7c430c0ec2 Enhance Clang to start instantiating static data member definitions
within class templates when they are necessary to complete the type of
the member. The canonical example is code like:

  template <typename T> struct S {
    static const int arr[];
    static const int x;
    static int f();
  };

  template <typename T> const int S<T>::arr[] = { 1, 2, 3 };
  template <typename T> const int S<T>::x = sizeof(arr) / sizeof(arr[0]);
  template <typename T> int S<T>::f() { return x; }

  int x = S<int>::f();

We need to instantiate S<T>::arr's definition to pick up its initializer
and complete the array type. This involves new code to specially handle
completing the type of an expression where the type alone is
insufficient. It also requires *updating* the expression with the newly
completed type. Fortunately, all the other infrastructure is already in
Clang to do the instantiation, do the completion, and prune out the
unused bits of code that result from this instantiation.

This addresses the initial bug in PR10001, and will be a step to
fleshing out other cases where we need to work harder to complete an
expression's type. Who knew we still had missing C++03 "features"?

llvm-svn: 132172
2011-05-27 01:33:31 +00:00
Alexis Hunt 6a31f2c7c9 Delete the extraneous return statement that was causing my earlier
issues and also add a test.

We should now handle defaulted members of templates properly. No
comment as to whether or not this also holds for templated functions,
but defaulting those is kind of insane.

llvm-svn: 131938
2011-05-23 23:56:01 +00:00
Alexis Hunt 1fb4e76218 Correctly propagate defaultedness across template instantiation. This
fixes PR9965, but we're not out of the water yet, as we do not
successfully handle out-of-line definitions, due to my utter
misunderstanding of how we manage templates.

llvm-svn: 131920
2011-05-23 21:07:59 +00:00
Chris Lattner e76b95ae15 make the x86-32 backend specify a byval alignment, even when the
code generator will do it.  With this patch, clang compiles the example
in PR9794 to not have an alloca temporary.

llvm-svn: 131881
2011-05-22 23:35:00 +00:00
Douglas Gregor 347e0f26be Fix our handling of the warning when one tries to pass a
non-POD/non-trivial object throuugh a C-style varargs. The warning
itself was default-mapped to error, but can be downgraded, but we were
treating it in Sema like a hard error, silently dropping the call.

Instead, treat this problem like a warning, and do what the warning
says we do: abort at runtime. To do so, we fake up a __builtin_trap()
expression that gets evaluated as part of the argument.

llvm-svn: 131805
2011-05-21 19:26:31 +00:00
Douglas Gregor 6ab2fa8f78 Introduce Type::isSignedIntegerOrEnumerationType() and
Type::isUnsignedIntegerOrEnumerationType(), which are like
Type::isSignedIntegerType() and Type::isUnsignedIntegerType() but also
consider the underlying type of a C++0x scoped enumeration type.

Audited all callers to the existing functions, switching those that
need to also handle scoped enumeration types (e.g., those that deal
with constant values) over to the new functions. Fixes PR9923 /
<rdar://problem/9447851>.

llvm-svn: 131735
2011-05-20 16:38:50 +00:00
Sebastian Redl 623ea82a6b Reapply r121528, fixing PR9941 by delaying the exception specification check for destructors until the class is complete and destructors have been adjusted.
llvm-svn: 131632
2011-05-19 05:13:44 +00:00
Alexis Hunt d0cdd1fab1 Revert r121528 as it breaks a simple testcase, which leads to, among
other things, libcxx not building.

llvm-svn: 131573
2011-05-18 20:57:11 +00:00
Sebastian Redl b900f04ccc Implement implicit exception specifications of destructors.
llvm-svn: 131528
2011-05-18 05:20:56 +00:00
John McCall 3597b6340c Revert r131434, my commit disabling the complete-object constructor
optimization for abstract classes;  there was a misunderstanding, and
it turns out that there are no kexts which rely on this.

llvm-svn: 131489
2011-05-17 21:05:49 +00:00
Ted Kremenek c249c411c2 Fix crash on C++ code when compiling with -finstrument-functions.
llvm-svn: 131435
2011-05-16 23:49:20 +00:00
John McCall 2ccfe89b3e Emit complete-object constructors for abstract classes in kext mode for
reasons that honestly really, really need to be looked into.

llvm-svn: 131434
2011-05-16 23:34:46 +00:00
Anders Carlsson d6f1518cc9 Fix another regression from the "skip vtable pointer initialization"
optimization. Make sure to require a vtable when trying to get the address
of a VTT, otherwise we would never end up emitting the VTT.

llvm-svn: 131400
2011-05-16 04:08:36 +00:00
John McCall 7ec4b4342c Don't actually emit calls to the reserved global placement new and delete
operators;  their semantics are guaranteed by the language.

If someone wants to argue that freestanding compiles shouldn't recognize
this, I might be convinceable.

llvm-svn: 131395
2011-05-16 01:05:12 +00:00
Anders Carlsson 49c0bd2a25 Re-enable the fix for PR9181 now that all the edge cases are handled.
llvm-svn: 131385
2011-05-15 17:36:21 +00:00
John McCall 036f2f6b35 The array-size operand to a new-expression is not necessarily a size_t.
It can be larger, it can be smaller, it can be signed, whatever.  Handle
all the crazy cases with grace and spirit.

llvm-svn: 131378
2011-05-15 07:14:44 +00:00
Anders Carlsson b55c8c127e Disable the optimization until the bug noticed by Sean Hunt has been fixed.
llvm-svn: 131372
2011-05-15 01:50:14 +00:00