a60b86a3fa
Make EmptySubobjectMap::CanPlaceBaseAtOffset take a BaseSubobjectInfo as well.
...
llvm-svn: 105113
2010-05-29 20:49:49 +00:00
d7f3fcf248
Change RecordLayoutBuilder::LayoutBase to take a BaseSubobjectInfo. No functionality change.
...
llvm-svn: 105112
2010-05-29 20:47:33 +00:00
6b0d914ad0
Rework the way virtual primary bases are added when laying out classes. Instead of doing it as a separate step, we now use the BaseSubobjectInfo and use it when laying out the bases. This fixes a bug where we would either not add a primary virtual base at all, or add it at the wrong offset.
...
llvm-svn: 105110
2010-05-29 19:44:50 +00:00
d6ff5d7787
Change LayoutVirtualBase to also take a BaseSubobjectInfo.
...
llvm-svn: 105104
2010-05-29 17:48:36 +00:00
bb0e6787b0
Change RecordLayoutBuilder::LayoutNonVirtualBase to take a BaseSubobjectInfo. No functionality change.
...
llvm-svn: 105103
2010-05-29 17:42:25 +00:00
e3c24c7033
Move computing the base subobject info for a class into the RecordLayoutBuilder because we're going to need it to be able to correctly add offsets for primary virtual bases.
...
llvm-svn: 105102
2010-05-29 17:35:14 +00:00
a7f3cdb0f2
Rename BaseInfo to BaseSubobjectInfo.
...
llvm-svn: 105007
2010-05-28 21:24:37 +00:00
056818f8a2
Move BaseInfo outside of the EmptySubobject class.
...
llvm-svn: 105002
2010-05-28 21:13:31 +00:00
db319761e7
More work on the empty subobject map. This code is not yet used.
...
llvm-svn: 104861
2010-05-27 18:20:57 +00:00
6da1098b4b
Sema: Add initial support for '#pragma options align=mac68k'.
...
- Docs are fairly sketchy, if someone wants to pore through gcc to look for
holes I'd appreciate any failing test cases!
llvm-svn: 104809
2010-05-27 05:45:51 +00:00
439edd1aea
More work on the new empty subobject map.
...
llvm-svn: 104808
2010-05-27 05:41:06 +00:00
592a85c137
Strip trailing whitespace.
...
llvm-svn: 104801
2010-05-27 02:25:46 +00:00
401304462a
AST: Rename PragmaPackAttr to MaxFieldAlignmentAttr, which is more accurate.
...
llvm-svn: 104795
2010-05-27 01:12:46 +00:00
c121b4e59f
More work on the empty subobjects map.
...
llvm-svn: 104787
2010-05-27 00:07:01 +00:00
c5ca1f7799
Move ComputeEmptySubobjectSizes to EmptySubobjectMap.
...
llvm-svn: 104702
2010-05-26 15:54:25 +00:00
f58de11b61
Stub out the EmptySubobjectsMap class.
...
llvm-svn: 104701
2010-05-26 15:32:58 +00:00
ce7f77729a
Inline the RecordLayoutBuilder constructor.
...
llvm-svn: 104700
2010-05-26 15:20:08 +00:00
c28a6c9251
Add a Layout overload that takes a CXXRecordDecl.
...
llvm-svn: 104695
2010-05-26 15:10:00 +00:00
c7656cdf83
Move RecordLayoutBuilder into an anonymous namespace.
...
llvm-svn: 104693
2010-05-26 09:58:31 +00:00
c222620adf
Rename ASTRecordLayoutBuilder to RecordLayoutBuilder.
...
llvm-svn: 104688
2010-05-26 05:58:59 +00:00
35a36eb78d
Move the ASTRecordLayoutBuilder class declaration into the .cpp file.
...
llvm-svn: 104686
2010-05-26 05:41:04 +00:00
b48414f4cb
Fix thinko and remove another unused function.
...
llvm-svn: 104683
2010-05-26 05:31:23 +00:00
1d1e4cf04d
Fix build.
...
llvm-svn: 104682
2010-05-26 05:25:15 +00:00
d20e7cd3c6
No need to use the PrimaryBaseInfo struct in the builder.
...
llvm-svn: 104681
2010-05-26 05:20:58 +00:00
87209bb54b
Remove FIXME.
...
llvm-svn: 104674
2010-05-26 05:11:02 +00:00
d295486012
Fold the other Layout overload into its sole call site.
...
llvm-svn: 104673
2010-05-26 05:10:47 +00:00
6ed3a9a494
Fold the ASTRecordLayoutBuilder::ComputeLayout overload that takes an ObjCInterfaceDecl pointer into its only callsite.
...
llvm-svn: 104672
2010-05-26 05:04:25 +00:00
df291d8d8b
Move the relevant ASTContext member functions that deal with ASTRecordLayoutBuilder into RecordLayoutBuilder.cpp. This matches the way we interact with other builder classes (CGRecordLayoutBuilder, VTableBuilder and VTTBuilder) and it also allows for making ASTRecordLayoutBuilder a private class without a header file.
...
llvm-svn: 104671
2010-05-26 04:56:53 +00:00
5773205a8d
When recording empty subobjects we should always look at the primary virtual base.
...
llvm-svn: 104464
2010-05-23 18:14:24 +00:00
5f1a1704a2
Add an UpdateVBases parameter to UpdateEmptyClassOffsets. Not used just yet.
...
llvm-svn: 103403
2010-05-10 15:28:59 +00:00
d626cb74d1
More work on handling empty classes.
...
llvm-svn: 103402
2010-05-10 15:26:14 +00:00
82828a31f6
Cleanup.
...
llvm-svn: 103370
2010-05-09 05:03:38 +00:00
15837f8f63
Actually compute the empty subobject sizes. No functionality change yet.
...
llvm-svn: 103363
2010-05-08 23:06:26 +00:00
2357bfd9d0
Add a SizeOfLargestEmptySubobject member to ASTRecordLayout. For C++ classes this will hold the largest empty subobject or 0 if the class doesn't have any empty subobjects.
...
llvm-svn: 103359
2010-05-08 22:35:05 +00:00
ccabe48235
AST: Dump ASTRecordLayout objects when they are created with -fdump-record-layouts.
...
llvm-svn: 101815
2010-04-19 20:44:53 +00:00
145f3f1e83
Fix -Wcast-qual warnings.
...
llvm-svn: 101786
2010-04-19 16:39:44 +00:00
aad5fa85d0
If a wide bit-field is inside a union its offset should always be 0.
...
llvm-svn: 101668
2010-04-17 20:21:41 +00:00
b11416d061
Add raw_ostream operators to NamedDecl for convenience. Switch over all users of getNameAsString on a stream.
...
The next step is to print the name directly into the stream, avoiding a temporary std::string copy.
llvm-svn: 101632
2010-04-17 09:33:03 +00:00
d9abbe56a5
Remove printfs.
...
llvm-svn: 101470
2010-04-16 15:59:41 +00:00
5723516ce9
More work on wide bit-fields, WIP.
...
llvm-svn: 101467
2010-04-16 15:57:11 +00:00
5efc56e36b
Rename the ASTContext member 'Context'.
...
llvm-svn: 101462
2010-04-16 15:07:51 +00:00
ea7b18298e
Split adding the primary virtual base offsets out into a separate pass. This fixes a bug where we would lay out virtual bases in the wrong order.
...
llvm-svn: 101373
2010-04-15 16:12:58 +00:00
1da6511b99
Tweak spelling (Bitfield -> BitField)
...
llvm-svn: 101369
2010-04-15 15:06:18 +00:00
3d9289c736
Add TargetInfo::useBitfieldTypeAlignment().
...
- Used to determine whether the alignment of the type in a bit-field is
respected when laying out structures. The default is true, targets can
override this as needed.
- This is designed to correspond to the PCC_BITFIELD_TYPE_MATTERS macro in
gcc. The AST/Sema implementation only affects one line, unless I have
forgotten something. I'd appreciate further review.
- IRgen still needs to be updated to fully support this (which is effectively
PR5591).
llvm-svn: 101356
2010-04-15 06:18:39 +00:00
fe53aaef51
Remove unnecessary cast.
...
llvm-svn: 101176
2010-04-13 20:52:05 +00:00
7a148f7141
Fix another vbase layout bug.
...
llvm-svn: 100952
2010-04-10 21:35:33 +00:00
5b441d73b7
Add a simple debug-only verification pass to the record layout builder.
...
llvm-svn: 100951
2010-04-10 21:24:48 +00:00
291279e3a6
Simplify the virtual base layout code and fix a bug where we wouldn't store the offset for a virtual base.
...
llvm-svn: 100940
2010-04-10 18:42:27 +00:00
aa423afffc
AST: Move C++ record layout dumping to ASTContext::DumpRecordLayout.
...
llvm-svn: 100746
2010-04-08 02:59:49 +00:00
fe9009635d
More record layout builder cleanup and simplification.
...
llvm-svn: 98238
2010-03-11 05:42:17 +00:00
de710c9125
Get rid of the PrimaryBase parameter from LayoutVirtualBases.
...
llvm-svn: 98233
2010-03-11 04:33:54 +00:00
f7b7a1e781
More cleanup towards fixing the real bug.
...
llvm-svn: 98232
2010-03-11 04:24:02 +00:00
6a84889d3f
Replace the class offset vectors in RecordLayoutBuilder with maps instead so we'll have faster lookup and so we can detect duplicates.
...
llvm-svn: 98231
2010-03-11 04:10:39 +00:00
f2fa75bdbb
More cleanup.
...
llvm-svn: 98230
2010-03-11 03:39:12 +00:00
c4c00ec41b
More record layout builder cleanup.
...
llvm-svn: 98229
2010-03-11 02:41:30 +00:00
aa87b4eab1
Remove the IndirectPrimary parameter in LayoutVirtualBases; it's already there as a member variable.
...
llvm-svn: 98211
2010-03-11 00:21:21 +00:00
8630b5b09b
More cleanup and simplification of the record layout builder.
...
llvm-svn: 98208
2010-03-11 00:15:35 +00:00
0d0b588fb2
Get rid of the LayoutBaseNonVirtually; it was used to lay out a base either as a non-virtual base or a virtual base.
...
llvm-svn: 98198
2010-03-10 22:26:24 +00:00
09ffa32b92
More shuffling.
...
llvm-svn: 98197
2010-03-10 22:21:28 +00:00
f08e681ba8
Cleanups, no functionality change yet.
...
llvm-svn: 98196
2010-03-10 22:16:06 +00:00
c3015a914f
Allocate ASTRecordLayout objects using the allocator associated with ASTContext.
...
This allows them to be allocated using a BumpPtrAllocated
in the common case.
llvm-svn: 97978
2010-03-08 20:56:29 +00:00
aef662206b
Patch removes IVars list from ObjCInterfaceDecl and
...
instead relies on their DeclContext for iteration, etc.
llvm-svn: 96638
2010-02-19 00:31:17 +00:00
ec345b7ecb
Simplify code. If we can reach here, the base must be virtual.
...
llvm-svn: 96211
2010-02-15 04:28:35 +00:00
0a2754ed8a
Be sure to select primary bases among the nearly empties in preorder,
...
not just among the direct bases. Before we where missing nearly
empties that were bases of virtual base classes.
llvm-svn: 94208
2010-01-22 20:27:17 +00:00
a318efd1f2
Improve key-function computation for templates. In particular:
...
- All classes can have a key function; templates don't change that.
non-template classes when computing the key function.
- We always mark all of the virtual member functions of class
template instantiations.
- The vtable for an instantiation of a class template has weak
linkage.
We could probably use available_externally linkage for vtables of
classes instantiated by explicit instantiation declarations (extern
templates), but GCC doesn't do this and I'm not 100% that the ABI
permits it.
llvm-svn: 92753
2010-01-05 19:06:31 +00:00
f2c79b6b9c
Misc key function fixes.
...
llvm-svn: 90831
2009-12-08 03:56:49 +00:00
8e0317bf05
Instantiated or specialized class templates never have a key function. This (and the previous check-in) fixes PR5557.
...
llvm-svn: 90753
2009-12-07 08:29:39 +00:00
5ebf8b44e9
Move key functions to a separate map.
...
llvm-svn: 90745
2009-12-07 04:35:11 +00:00
71a26d8f82
Move helper onto CXXMethodDecl.
...
llvm-svn: 90716
2009-12-06 20:50:05 +00:00
0846d52f70
Tweak "key function" rules so that they work for templates with virtual
...
inline functions.
llvm-svn: 90645
2009-12-05 04:55:55 +00:00
f98849eb8a
In Sema, whenever we think that a function is going to cause a vtable to be generated, we mark any virtual implicit member functions as referenced.
...
llvm-svn: 90327
2009-12-02 17:15:43 +00:00
b1d3f7c909
Have ASTRecordLayout keep track of the key function, in preparation of fixing a synthetic ctor/dtor bug.
...
llvm-svn: 90168
2009-11-30 23:41:22 +00:00
a30c0d3ace
Use a PointerIntPair for the PrimaryBaseInfo. Yay for clever LLVM data structures!
...
llvm-svn: 90020
2009-11-27 22:14:40 +00:00
03ff379911
Add a new PrimaryBaseInfo struct that combines the record decl of a primary base with whether it's virtual or not.
...
llvm-svn: 90018
2009-11-27 22:05:05 +00:00
ba958400a2
When laying out bitfields, make sure that the data size is always aligned to a byte. This fixes PR5580.
...
llvm-svn: 89611
2009-11-22 19:13:51 +00:00
072094407d
Move bit-field layout out into a separate function. No functionality change.
...
llvm-svn: 89604
2009-11-22 17:37:31 +00:00
96d5c76498
Added rudimentary C++0x attribute support.
...
The following attributes are currently supported in C++0x attribute
lists (and in GNU ones as well):
- align() - semantics believed to be conformant to n3000, except for
redeclarations and what entities it may apply to
- final - semantics believed to be conformant to CWG issue 817's proposed
wording, except for redeclarations
- noreturn - semantics believed to be conformant to n3000, except for
redeclarations
- carries_dependency - currently ignored (this is an optimization hint)
llvm-svn: 89543
2009-11-21 08:43:09 +00:00
4e16d05dc4
Fixup spacing.
...
llvm-svn: 86792
2009-11-11 02:49:00 +00:00
2b84dd39be
Refine layout for indirect virtual base classes.
...
llvm-svn: 86116
2009-11-05 04:02:15 +00:00
1054faed32
Audit the code for places where it is assumed that every base specifier refers to a RecordType. Add assertions or conditions as appropriate. This fixes another crash in the Apache stdlib vector.
...
llvm-svn: 85055
2009-10-25 17:03:50 +00:00
534b0639db
Address comment from Daniel.
...
llvm-svn: 82865
2009-09-26 16:55:29 +00:00
47680d8040
Rename NextOffset to DataSize.
...
llvm-svn: 82832
2009-09-26 01:34:51 +00:00
7f78d85be4
Start at NextOffset when laying out bases as well.
...
llvm-svn: 82831
2009-09-26 01:29:10 +00:00
a6d95b5a3e
Use NextOffset when laying out a field.
...
llvm-svn: 82828
2009-09-26 01:09:35 +00:00
d7d358a65e
Even more work on empty classes.
...
llvm-svn: 82770
2009-09-25 15:39:00 +00:00
e1883100a3
Who would have thought that empty classes were so tricky? Handle cases where an empty virtual base class needs to be moved aside because it conflicts with the first field.
...
llvm-svn: 82746
2009-09-25 01:54:38 +00:00
4bf82141e7
Handle array fields that contain empty structs.
...
llvm-svn: 82744
2009-09-25 01:23:32 +00:00
6f95c705ea
More work on empty classes.
...
llvm-svn: 82736
2009-09-25 00:02:51 +00:00
bb66bc8e84
More improvements with laying out empty bases.
...
llvm-svn: 82682
2009-09-24 05:21:31 +00:00
f24b18fb06
More work on empty classes.
...
llvm-svn: 82679
2009-09-24 03:22:10 +00:00
6522b05db7
Scaffolding for supporting empty bases/fields.
...
llvm-svn: 82678
2009-09-24 03:13:30 +00:00
55640546e4
If we already set a primary base, don't set it to the first nearly empty base class.
...
llvm-svn: 82563
2009-09-22 19:16:59 +00:00
96cff1fc10
Explicitly initialize the PrimaryBase and PrimaryBaseWasVirtual members.
...
llvm-svn: 82560
2009-09-22 18:21:58 +00:00
865715442e
Yes.
...
llvm-svn: 82559
2009-09-22 17:38:02 +00:00
8143069417
Store the set of indirect primary bases directly in the record layout builder.
...
llvm-svn: 82513
2009-09-22 03:02:06 +00:00
d6020c321a
Record layout builder cleanup.
...
llvm-svn: 82502
2009-09-22 00:04:45 +00:00
93b6d5e85e
Temporary band-aid for handling empty classes somewhat better.
...
llvm-svn: 82124
2009-09-17 04:42:56 +00:00
11289f4280
Remove tabs, and whitespace cleanups.
...
llvm-svn: 81346
2009-09-09 15:08:12 +00:00
996576f3aa
Refine vbase offset calculations. WIP.
...
llvm-svn: 79198
2009-08-16 19:04:13 +00:00
22ea1f8a30
Cleanups and fixups for calculating the virtual base offsets. WIP.
...
llvm-svn: 79156
2009-08-16 01:46:26 +00:00
bcf756cbac
Deconflate virtual base offsets from non-virtual base offsets.
...
Deconflate a virtual base primary from a non-virtual base.
llvm-svn: 78971
2009-08-14 01:44:03 +00:00
590a7c780c
We can't avoid doing the work to find all the indirect primary virtual
...
base classes as we'll need that to layout the virtual bases...
llvm-svn: 78954
2009-08-13 23:26:06 +00:00
c2f591b1e6
Refine vtable layout for virtual bases and keep better track of
...
primaries. WIP.
llvm-svn: 78950
2009-08-13 22:53:07 +00:00
1300754fdd
Prep for vbase layout refinements. WIP.
...
llvm-svn: 78882
2009-08-13 02:02:14 +00:00
6bb66bbd06
Refine virtual base layout. WIP.
...
llvm-svn: 78873
2009-08-13 00:34:14 +00:00
cb2752bb5d
Remove another done audit FIXME.
...
llvm-svn: 78847
2009-08-12 22:10:00 +00:00
aa08da7906
Refactor a bit and remove some FIXME audit markers, now that the code
...
has been audited for correctness.
llvm-svn: 78846
2009-08-12 22:06:55 +00:00
6f3793b406
Refine primary vbase selection ordering. WIP.
...
llvm-svn: 78844
2009-08-12 21:50:08 +00:00
78696a70bf
Implement more of the inductive case for vtable layout involving
...
virtual base primaries and improve the layout of classes with virtual
bases. WIP.
Hey, I've decided I want a change to FileCheck, I need to ensure that
the group is together, nothing in between. Can we change it to check
the match line is from the line immediately following the last matched
line, if the source for the matched line is immediately after the
source for the previously matched line?
// CHECK: 1
// CHECK: 2
// CHECK: 3
// CHECK: 4
// CHECK: 5
// CHECK: 6
would require 1 2 and 3 to be continuous in the output, and 4 5 and 6
to be continuous.
llvm-svn: 78638
2009-08-11 04:03:59 +00:00
28a5fa29f3
Take #pragma pack into account when laying out structs. Fixes rdar://problem/7095436.
...
llvm-svn: 78490
2009-08-08 19:38:24 +00:00
68e0b68289
Introduce a new PragmaPack attribute, and use it for #pragma pack. The PackedAttr now only represents __attribute__((packed)).
...
This is necessary because #pragma pack and __attribute__((packed)) have different semantics. No functionality change yet, but this lays the groundwork for fixing a record layout bug.
llvm-svn: 78483
2009-08-08 18:23:56 +00:00
c266c6d797
Add ability to generate vcall offsets for primary virtual base.
...
llvm-svn: 78396
2009-08-07 19:00:50 +00:00
6b2556f829
Layout virtual bases. Work in progress.
...
llvm-svn: 78308
2009-08-06 13:41:24 +00:00
e9c85a7f58
Fix spell-o.
...
llvm-svn: 78303
2009-08-06 12:52:13 +00:00
2509480813
Fixup object layout when we have a primary base (it goes first). Start preping for
...
virtual base layout.
llvm-svn: 78265
2009-08-06 00:38:46 +00:00
d8fe7b2792
Calculate the primary base class better and use that when laying down
...
the vtable. Still a work in progress.
llvm-svn: 78252
2009-08-05 22:37:18 +00:00
bc78a728ee
Add code to setup the vtable pointer in the constructor. Work in progress.
...
llvm-svn: 77699
2009-07-31 18:25:34 +00:00
57724c2f7e
Update based upon comments. Explain why we have an assert.
...
llvm-svn: 77612
2009-07-30 18:01:44 +00:00
44b8e90369
We'll also need a vtable pointer if we have virtual bases.
...
llvm-svn: 77610
2009-07-30 17:53:53 +00:00
3dc7eb9440
Add ability to layout the vtable pointer in trivial cases. I noticed
...
that we would silently do bad things with virtual bases in the layout
code, so, we just turn them off. When people do better things with
them, we can turn them back on.
llvm-svn: 77556
2009-07-30 00:22:38 +00:00
c23c7e6a51
Change uses of:
...
Type::getAsReferenceType() -> Type::getAs<ReferenceType>()
Type::getAsRecordType() -> Type::getAs<RecordType>()
Type::getAsPointerType() -> Type::getAs<PointerType>()
Type::getAsBlockPointerType() -> Type::getAs<BlockPointerType>()
Type::getAsLValueReferenceType() -> Type::getAs<LValueReferenceType>()
Type::getAsRValueReferenceType() -> Type::getAs<RValueReferenceType>()
Type::getAsMemberPointerType() -> Type::getAs<MemberPointerType>()
Type::getAsReferenceType() -> Type::getAs<ReferenceType>()
Type::getAsTagType() -> Type::getAs<TagType>()
And remove Type::getAsReferenceType(), etc.
This change is similar to one I made a couple weeks ago, but that was partly
reverted pending some additional design discussion. With Doug's pending smart
pointer changes for Types, it seemed natural to take this approach.
llvm-svn: 77510
2009-07-29 21:53:49 +00:00
7be1ad814b
No longer need to keep base class offsets in the offset
...
table as it has its own place now.
llvm-svn: 77491
2009-07-29 18:50:06 +00:00
fc8cfa8b9f
Add a field for C++ specific data to ASTRecordLayout. Use it to store the non-virtual size and alignment + base offsets.
...
llvm-svn: 77352
2009-07-28 19:24:15 +00:00
dedf1e4b1a
ir-gen for non-virtual base class initialization
...
in constructors.
llvm-svn: 77087
2009-07-25 21:12:28 +00:00
6d9f6f326d
Handle layout of non-virtual base classes.
...
llvm-svn: 76348
2009-07-19 00:18:47 +00:00
118ce16bf5
Refactor field layout into a separate function.
...
llvm-svn: 76343
2009-07-18 21:48:39 +00:00
27b50135ee
Rename NextOffset to DataSize, which better matches the Itanium C++ ABI
...
llvm-svn: 76339
2009-07-18 21:26:44 +00:00
4f51628b02
More layout builder work.
...
llvm-svn: 76333
2009-07-18 20:50:59 +00:00
7947433411
Add a new ASTRecordLayoutBuilder class. Not used yet.
...
llvm-svn: 76330
2009-07-18 20:20:21 +00:00
Anders Carlsson