]> granicus.if.org Git - clang/commit
Append new attributes to the end of an AttributeList.
authorMichael Kruse <llvm@meinersbur.de>
Fri, 3 Aug 2018 01:21:16 +0000 (01:21 +0000)
committerMichael Kruse <llvm@meinersbur.de>
Fri, 3 Aug 2018 01:21:16 +0000 (01:21 +0000)
commit915cb4dbc21080a91a9a6277c2bc81c2c6f79d09
tree92b74928a866cfb4ba25ff1b989d70e163c9303e
parent22d36eedd43d53bfc3a55acca8b2a54578a993ed
Append new attributes to the end of an AttributeList.

Recommit of r335084 after revert in r335516.

... instead of prepending it at the beginning (the original behavior
since implemented in r122535 2010-12-23). This builds up an
AttributeList in the the order in which the attributes appear in the
source.

The reverse order caused nodes for attributes in the AST (e.g. LoopHint)
to be in the reverse order, and therefore printed in the wrong order in
-ast-dump. Some TODO comments mention this. The order was explicitly
reversed for enable_if attribute overload resolution and name mangling,
which is not necessary anymore with this patch.

The change unfortunately has some secondary effect, especially on
diagnostic output. In the simplest cases, the CHECK lines or expected
diagnostic were changed to the the new output. If the kind of
error/warning changed, the attributes' order was changed instead.

This unfortunately causes some 'previous occurrence here' hints to be
textually after the main marker. This typically happens when attributes
are merged, but are incompatible to each other. Interchanging the role
of the the main and note SourceLocation will also cause the case where
two different declaration's attributes (in contrast to multiple
attributes of the same declaration) are merged to be reverse. There is
no easy fix because sometimes previous attributes are merged into a new
declaration's attribute list, sometimes new attributes are added to a
previous declaration's attribute list. Since 'previous occurrence here'
pointing to locations after the main marker is not rare, I left the
markers as-is; it is only relevant when the attributes are declared in
the same declaration anyway.

Differential Revision: https://reviews.llvm.org/D48100

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@338800 91177308-0d34-0410-b5e6-96231b3b80d8
35 files changed:
include/clang/Sema/ParsedAttr.h
lib/AST/ItaniumMangle.cpp
lib/Parse/ParseDeclCXX.cpp
lib/Parse/ParseObjc.cpp
lib/Sema/SemaAttr.cpp
lib/Sema/SemaOverload.cpp
lib/Sema/SemaType.cpp
lib/Serialization/ASTReaderDecl.cpp
test/Index/annotate-comments-availability-attrs.cpp
test/Index/complete-with-annotations.cpp
test/Misc/ast-print-pragmas.cpp
test/PCH/pragma-loop.cpp
test/Parser/pragma-loop-safety.cpp
test/Parser/pragma-loop.cpp
test/Parser/pragma-unroll-and-jam.cpp
test/Parser/pragma-unroll.cpp
test/Sema/attr-availability-tvos.c
test/Sema/attr-availability.c
test/Sema/attr-coldhot.c
test/Sema/attr-disable-tail-calls.c
test/Sema/attr-long-call.c
test/Sema/attr-micromips.c
test/Sema/attr-notail.c
test/Sema/attr-ownership.c
test/Sema/attr-ownership.cpp
test/Sema/attr-print.c
test/Sema/attr-visibility.c
test/Sema/internal_linkage.c
test/Sema/mips-interrupt-attr.c
test/Sema/nullability.c
test/SemaCXX/attr-print.cpp
test/SemaCXX/ms-uuid.cpp
test/SemaObjC/nullability.m
test/SemaOpenCL/address-spaces.cl
test/SemaTemplate/attributes.cpp