Richard Smith [Sat, 8 Mar 2014 00:03:56 +0000 (00:03 +0000)]
Module [extern_c] attribute: inherit to submodules, don't write 'extern "C"'
blocks when building in C mode, and serialize and deserialize the attribute.
Aaron Ballman [Fri, 7 Mar 2014 22:17:20 +0000 (22:17 +0000)]
In my tests, I'm finding that declaring iterators in terms of ranges can sometimes have dangerous side-effects where the range temporary is destroyed, taking the underlying iterators out with it.
This changes the iterators so that they are no longer implemented in terms of ranges (so it's a very partial revert of the existing rangification efforts).
Aaron Ballman [Fri, 7 Mar 2014 19:56:05 +0000 (19:56 +0000)]
[C++11] Replacing DeclBase iterators decls_begin() and decls_end() with iterator_range decls(). The same is true for the noload versions of these APIs. Updating all of the usages of the iterators with range-based for loops.
Aaron Ballman [Fri, 7 Mar 2014 18:11:58 +0000 (18:11 +0000)]
[C++11] Replacing IndirectFieldDecl iterators chain_begin() and chain_end() with iterator_range chains(). Updating all of the usages of the iterators with range-based for loops.
Aaron Ballman [Fri, 7 Mar 2014 17:50:17 +0000 (17:50 +0000)]
[C++11] Replacing ObjCMethodDecl iterators param_begin() and param_end() with iterator_range params(). Updating all of the usages of the iterators with range-based for loops.
Aaron Ballman [Fri, 7 Mar 2014 16:40:17 +0000 (16:40 +0000)]
[C++11] Adding an iterator_range accessor for parameter declarations to the CaptureDecl class; however, the only usage of the iterators cannot be converted to using the range yet. This change is for consistency with other Decls exposing parameters.
Aaron Ballman [Fri, 7 Mar 2014 16:09:59 +0000 (16:09 +0000)]
[C++11] Replacing BlockDecl iterators param_begin() and param_end() with iterator_range params(). Updating all of the usages of the iterators with range-based for loops.
Aaron Ballman [Fri, 7 Mar 2014 15:12:56 +0000 (15:12 +0000)]
[C++11] Replacing FunctionDecl iterators param_begin() and param_end() with iterator_range params(). Updating all of the usages of the iterators with range-based for loops.
Aaron Ballman [Fri, 7 Mar 2014 14:09:15 +0000 (14:09 +0000)]
[C++11] Replacing iterators ddiag_begin() and ddiag_end() with iterator_range ddiags(). Updating all of the usages of the iterators with range-based for loops.
Aaron Ballman [Fri, 7 Mar 2014 12:50:00 +0000 (12:50 +0000)]
[C++11] Replacing iterators attr_begin() and attr_end() with iterator_range attrs(). Updating all of the usages of the iterators with range-based for loops.
Ted Kremenek [Fri, 7 Mar 2014 07:14:36 +0000 (07:14 +0000)]
Fix recent regressions in -Wreturn-type caused by heuristics to -Wunreachable-code.
I had forgotten that the same reachability code is used by both -Wreturn-type
and -Wunreachable-code, so the heuristics applied to the latter were indirectly
impacting the former.
To address this, the reachability code is more refactored so that whiled
the logic at its core is shared, the intention of the clients are better
captured and segregated in helper APIs.
Fixes PR19074, and also some false positives reported offline to me
by Nick Lewycky.
Ben Langmuir [Fri, 7 Mar 2014 06:40:32 +0000 (06:40 +0000)]
Add dependencies from imported modules with -MD
Add module dependencies to the dependency files created by -MD/-MMD/etc.
by attaching an ASTReaderListener that will call into the dependency
file generator when a module input file is seen in the serialized AST.
Yunzhong Gao [Fri, 7 Mar 2014 00:23:36 +0000 (00:23 +0000)]
The Visual Studio IDE changed behavior in VS2012. It used to be the case that
the clang diagnostic has to report a column number one less than the correct
value in order for the IDE to move the cursor to the expected location. This
behavior is changed in VS2012 and VS2013 so that the IDE is now expecting the
column number to match the actual source location.
Before: source(line, column-1): type: message
After: source(line, column): type: message
This patch changes -fdiagnostics-format=msvc to match the new VS2012 and VS2013
when fmsc-version is 1700 or greater.
Aaron Ballman [Fri, 7 Mar 2014 00:10:58 +0000 (00:10 +0000)]
Partial revert of r203179. The build bots are telling me that some out-of-tree builds are using redecls_begin()/redecls_end(), so adding some machinery back in to support them.
Aaron Ballman [Thu, 6 Mar 2014 23:45:36 +0000 (23:45 +0000)]
[C++11] Replacing iterators redecls_begin() and redecls_end() with iterator_range redecls(). Updating all of the usages of the iterators with range-based for loops, which allows the begin/end forms to be removed entirely.
C. Compare vector sizes using their raw element size instead of
getTypeSize (which rounds up sizes) in order to issue diagnostics
when casting to mismatched vector sizes; instead of crashing in IRGen.
// rdar:// 16196902. Reviewed offline by John McCall.
Use llvm.compiler.used instead of llvm.used for objc symbols.
LLVM currently has a hack (shouldEmitUsedDirectiveFor) that causes it to not
print no_dead_strip for symbols starting with 'l' or 'L'. These are exactly the
ones that the clang's objc codegen is producing. The net result, is that it is
equivalent to llvm.compiler.used.
The need for putting the private symbol in llvm.compiler.used should be clear
(the objc runtime uses them). The reason for also putting the weak symbols in
it is for LTO: ld64 will not ask us to preserve the it.
Aaron Ballman [Thu, 6 Mar 2014 19:37:24 +0000 (19:37 +0000)]
Based on usage, the NamedDecl is always set when constructing a CallingContext, but none of the other optional parameters are. Removing the optional parameters, and making the NamedDecl required.
Aaron Ballman [Thu, 6 Mar 2014 14:02:27 +0000 (14:02 +0000)]
Reformatting the style used within the massive attribute semantic handling switch statement, so now there is only one style used in this block of code, instead of three or more styles.
NAKAMURA Takumi [Thu, 6 Mar 2014 07:29:30 +0000 (07:29 +0000)]
CodeGenObjC/instr-profile.m: Add -target darwin to appease bots.
This expected;
; Function Attrs: uwtable
define internal void @"\01+[A foreach:]"(i8* %self, i8* %_cmd, %0* %array) #0 {
But the Name is mangled on other hosts; (x86_64-unknown-unknown)
; Function Attrs: nounwind uwtable
define internal void @_c_A__foreach_(i8* %self, i8* %_cmd, %1* %array) #0 {
Ted Kremenek [Thu, 6 Mar 2014 06:50:46 +0000 (06:50 +0000)]
[-Wunreachable-code] don't warn about dead 'return <string literal>' dominated by a 'noreturn' call, where literal becomes an std::string.
I have mixed feelings about this one. It's used all over the codebase,
and is analogous to the current heuristic for ordinary C string literals.
This requires some ad hoc pattern matching of the AST. While the
test case mirrors what we see std::string in libc++, it's not really
testing the libc++ headers.
Bob Wilson [Thu, 6 Mar 2014 04:55:41 +0000 (04:55 +0000)]
PGO: Use the main file name to help distinguish functions with local linkage.
In addition, for all functions, use the name from the llvm::Function to
identify the function in the profile data. Compute that "function name",
including the file name for local functions, once when assigning the PGO
counters and store it in the CodeGenPGO class.
Move the code to add InlineHint and Cold attributes out of StartFunction(),
because the "function name" string isn't available at that point.
Bob Wilson [Thu, 6 Mar 2014 04:55:37 +0000 (04:55 +0000)]
PGO: Rename variables to avoid referring to the "MangledName" of a function.
For C++ functions, we will continue to use the mangled name to identify
functions in the PGO profile data, but this name is confusing for things like
Objective-C methods. For functions with local linkage, we're also going to
include the file name to help distinguish those functions, so this changes to
use more generic variable names.
Bob Wilson [Thu, 6 Mar 2014 04:55:35 +0000 (04:55 +0000)]
Refactor PGO code in preparation for handling non-C/C++ code.
Move the PGO.assignRegionCounters() call out of StartFunction, because that
function is called from many places where it does not make sense to do PGO
instrumentation (e.g., compiler-generated helper functions). Change several
functions to take a StringRef argument for the unique name associated with
a function, so that the name can be set differently for things like Objective-C
methods and block literals.
Bob Wilson [Thu, 6 Mar 2014 04:55:28 +0000 (04:55 +0000)]
PGO: don't emit counter increment if no counters have been allocated.
I hit this while debugging another issue where my sources were in an
inconsistent state, so I don't have a testcase. Regardless, this check is
simpler and more direct than checking if the option is enabled.
Richard Smith [Thu, 6 Mar 2014 03:16:27 +0000 (03:16 +0000)]
Fix crash if a submodule overrides one of its own macros, and add support for
submodule macro overriding within the same top-level module (necessary for the
testcase to be remotely reasonable). Incidentally reduces the number of libc++
testsuite regressions with modules enabled from 7 to 6.
Use private linkage for remaining GlobalVariables with private names.
This patch changes the remaining GlobalVariables using "\01L" and
"\01l" prefixes to use private linkage. What is strange about them is
that they currently use WeakAnyLinkage. There is no comment stating
why and that is really odd since the symbols are completely hidden, so
it doesn't make sense for them to be weak.
Clang revisions like r63329, r63408, r63770, r65761 set the linkage to
weak, but don't say why. I suspect they were just copying llvm-gcc.
In llvm-gcc I found r58599 and r56322 that set DECL_WEAK, but they
were just syncing from the apple gcc. I am not exactly sure what that
means, since the last commit to
svn://gcc.gnu.org/svn/gcc/branches/apple was in 2006, 2 years earlier.
In summary, I have no idea why weak linkage was being used :-(
To quote John McCall, "Let’s try without it and see" :-)
projects / clang / log