| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
|
|
|
|
|
|
|
|
| |
Suggested by Ted, since string literal comparison is at least slightly more
sensible than comparison of runtime literals. (Ambiguous language on
developer.apple.com implies that strings are guaranteed to be uniqued within
a translation unit and possibly across a linked binary.)
llvm-svn: 160378
|
|
|
|
|
|
|
|
| |
Recovering as if the user had actually called -isEqual: is a bit too far from
the semantics of the program as written, /even though/ it's probably what they
intended.
llvm-svn: 160377
|
|
|
|
|
|
|
|
| |
as an array of its base class TemplateArgument. Switch the const
TemplateArgument* parameters of InstantiatingTemplate's constructors to
ArrayRef<TemplateArgument> to prevent this from happening again in the future.
llvm-svn: 160245
|
|
|
|
|
|
|
|
|
|
|
| |
Chris pointed out that while the comparison is certainly problematic
and does not have well-defined behavior, it isn't any worse than some
of the other abuses that we merely warn about and doesn't need to make
the compilation fail.
Revert the release notes change (r159766) now that this is just a new warning.
llvm-svn: 159939
|
|
|
|
|
|
|
|
| |
-ftemplate-depth limit. There are various ways to get an infinite (or merely
huge) stack of substitutions with no intervening instantiations. This is also
consistent with gcc's behavior.
llvm-svn: 159907
|
|
|
|
|
|
|
|
| |
consistency a bit.
(cf -Wunused-private-field and several other existing -field diagnostics.)
llvm-svn: 159633
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In C, enum constants have the type of the enum's underlying integer type,
rather than the type of the enum. (This is not true in C++.) Thus, when a
block's return type is inferred from an enum constant, it is incompatible
with expressions that return the enum type.
In r158899, I told block returns to pretend that enum constants have enum
type, like in C++. Doug Gregor pointed out that this can break existing code.
Now, we don't check the types of return statements until the end of the block.
This lets us go back and add implicit casts in blocks with mixed enum
constants and enum-typed expressions.
<rdar://problem/11662489> (again)
llvm-svn: 159591
|
|
|
|
|
|
| |
No functionality change.
llvm-svn: 159436
|
|
|
|
| |
llvm-svn: 159394
|
|
|
|
| |
llvm-svn: 159292
|
|
|
|
|
|
| |
before, so we don't incorrectly think arguments of function type are non-POD.
llvm-svn: 159290
|
|
|
|
|
|
| |
being updated. Will fix that in a second.
llvm-svn: 159280
|
|
|
|
| |
llvm-svn: 159279
|
|
|
|
|
|
|
|
| |
to see if we had an underlying final class or method, but we would then
use the cast type to do the call, resulting in a direct call to the wrong
method.
llvm-svn: 159212
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
resulted in it being reverted. A test for that bug was added in r158950.
Original comment:
If an object (such as a std::string) with an appropriate c_str() member function
is passed to a variadic function in a position where a format string indicates
that c_str()'s return type is desired, provide a note suggesting that the user
may have intended to call the c_str() member.
Factor the non-POD-vararg checking out of DefaultVariadicArgumentPromotion and
move it to SemaChecking in order to facilitate this. Factor the call checking
out of function call checking and block call checking, and extend it to cover
constructor calls too.
Patch by Sam Panzer!
llvm-svn: 159159
|
|
|
|
|
|
|
|
|
|
| |
Heavily based on a patch from
Aaron Wishnick <aaron.s.wishnick@gmail.com>.
I'll clean up the duplicated function in CodeGen as
a follow-up, later today or tomorrow.
llvm-svn: 159060
|
|
|
|
|
|
|
|
| |
when the calling site is a member function template.
Effectively reverts r111675.
llvm-svn: 159004
|
|
|
|
|
|
|
|
| |
Revert "If an object (such as a std::string) with an appropriate c_str() member function"
This reverts commit 7d96f6106bfbd85b1af06f34fdbf2834aad0e47e.
llvm-svn: 158949
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Also, don't warn if the used function is __attribute__((const)), in which case
it's not supposed to use global variables anyway.
The inline-in-inline thing is a heuristic, and one that's possibly incorrect
fairly often because the function being inlined could definitely use global
variables. However, even some C standard library functions are written using
other (trivial) static-inline functions in the headers, and we definitely don't
want to be warning on that (or on anything that /uses/ these trivial inline
functions). So we're using "inlined" as a marker for "fairly trivial".
(Note that __attribute__((pure)) does /not/ guarantee safety like ((const),
because ((const)) does not guarantee that global variables are not being used,
and the warning is about globals not being shared across TUs.)
llvm-svn: 158898
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
is passed to a variadic function in a position where a format string indicates
that c_str()'s return type is desired, provide a note suggesting that the user
may have intended to call the c_str() member.
Factor the non-POD-vararg checking out of DefaultVariadicArgumentPromotion and
move it to SemaChecking in order to facilitate this. Factor the call checking
out of function call checking and block call checking, and extend it to cover
constructor calls too.
Patch by Sam Panzer!
llvm-svn: 158887
|
|
|
|
|
|
|
|
|
|
| |
in microsoft mode. Fixes PR12701.
The code for this was already in 2 of the 3 branches of a
conditional and missing in the 3rd branch, so lift it above
the conditional.
llvm-svn: 158842
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
It's very easy for anonymous external linkage to propagate in C++ through
return types and parameter types. Likewise, it's possible that a template
containing an inline function is only used with parameters that have internal
linkage. Actually diagnosing where the internal linkage comes from is fairly
difficult (both to locate and then to print nicely). Finally, since we only
have one translation unit available, we can't even prove that any of this
violates the ODR.
This warning needs better-defined behavior in C++ before it can really go in.
Rewording of the C warning (which /is/ specified by C99) coming shortly.
llvm-svn: 158836
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
target Objective-C runtime down to the frontend: break this
down into a single target runtime kind and version, and compute
all the relevant information from that. This makes it
relatively painless to add support for new runtimes to the
compiler. Make the new -cc1 flag, -fobjc-runtime=blah-x.y.z,
available at the driver level as a better and more general
alternative to -fgnu-runtime and -fnext-runtime. This new
concept of an Objective-C runtime also encompasses what we
were previously separating out as the "Objective-C ABI", so
fragile vs. non-fragile runtimes are now really modelled as
different kinds of runtime, paving the way for better overall
differentiation.
As a sort of special case, continue to accept the -cc1 flag
-fobjc-runtime-has-weak, as a sop to PLCompatibilityWeak.
I won't go so far as to say "no functionality change", even
ignoring the new driver flag, but subtle changes in driver
semantics are almost certainly not intended.
llvm-svn: 158793
|
|
|
|
|
|
|
|
|
|
| |
Per post-commit review, it's not appropriate to use ExtWarn in C++, because
we can't prove that the inline function will actually be defined in more than
one place (and thus we can't prove that this violates the ODR).
This removes the warning entirely from uses in the main source file in C++.
llvm-svn: 158689
|
|
|
|
|
|
|
|
| |
This includes treating anonymous namespaces like internal linkage, and allowing
const variables to be used even if internal. The whole thing's been broken out
into a separate function to avoid nested ifs.
llvm-svn: 158683
|
|
|
|
|
|
|
|
|
|
|
| |
This handles the very common case of people writing inline functions in their
main source files and not tagging them as inline. These cases should still
behave as the user intended. (The diagnostic is still emitted as an extension.)
I'm reworking this code anyway to account for C++'s equivalent restriction in
[basic.def.odr]p6, but this should get some bots back to green.
llvm-svn: 158666
|
|
|
|
|
|
|
|
| |
semantics when promotions are involved.
(As far as I can tell, this only affects some edge cases.)
llvm-svn: 158591
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is explicitly forbidden in C99 6.7.4p3. This is /not/ forbidden in C++,
probably because by default file-scope const/constexpr variables have internal
linkage, while functions have external linkage. There's also the issue of
anonymous namespaces to consider. Nevertheless, there should probably be a
similar warning, since the semantics of inlining a function that references
a variable with internal linkage do not seem well-defined.
<rdar://problem/11577619>
llvm-svn: 158531
|
|
|
|
|
|
|
| |
for unexpanded parameter packs. Fixes the crash-on-invalid in
PR13117.
llvm-svn: 158525
|
|
|
|
|
|
|
| |
initialization, and use that information to produce the right kind of
initialization during template instantiation.
llvm-svn: 158288
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Objective-C literals conceptually always create new objects, but may be
optimized by the compiler or runtime (constant folding, singletons, etc).
Comparing addresses of these objects is relying on this optimization
behavior, which is really an implementation detail.
In the case of == and !=, offer a fixit to a call to -isEqual:, if the
method is available. This fixit is directly on the error so that it is
automatically applied.
Most of the time, this is really a newbie mistake, hence the fixit.
llvm-svn: 158230
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The integral APSInt value is now stored in a decomposed form and the backing
store for large values is allocated via the ASTContext. This way its not
leaked as TemplateArguments are never destructed when they are allocated in
the ASTContext. Since the integral data is immutable it is now shared between
instances, making copying TemplateArguments a trivial operation.
Currently getting the integral data out of a TemplateArgument requires creating
a new APSInt object. This is cheap when the value is small but can be expensive
if it's not. If this turns out to be an issue a more efficient accessor could
be added.
llvm-svn: 158150
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
value_type
In addition, I've made the pointer and reference typedef 'void' rather than T*
just so they can't get misused. I would've omitted them entirely but
std::distance likes them to be there even if it doesn't use them.
This rolls back r155808 and r155869.
Review by Doug Gregor incorporating feedback from Chandler Carruth.
llvm-svn: 158104
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
temporary or an array subobject of a class temporary, and the resulting value
is used to initialize a pointer which outlives the temporary. Such a pointer
is always left dangling after the initialization completes and the array's
lifetime ends.
In order to detect this situation, this change also adds an
LValueClassification of LV_ArrayTemporary for temporaries of array type which
aren't subobjects of class temporaries. These occur in C++11 T{...} and GNU C++
(T){...} expressions, when T is an array type. Previously we treated the former
as a generic prvalue and the latter as a class temporary.
llvm-svn: 157955
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Before:
test.cc:2:18: note: place parentheses around the == expression to silence this warning
if (0 == flags & 0xdd)
^
( )
Now:
test.cc:2:18: note: place parentheses around the == expression to silence this warning
if (0 == flags & 0xdd)
^
( )
llvm-svn: 157897
|
|
|
|
| |
llvm-svn: 157667
|
|
|
|
|
|
|
|
|
|
| |
instantiatiating a template.
This comes up in the begin/end calls of a range-for (see the included test
case). Other suggestions are welcome, though this seems to do the trick without
regressing anything.
llvm-svn: 157553
|
|
|
|
|
|
| |
Conceptually, this is part of -Wunused-value, but I added a separate flag -Wunused-volatile-lvalue so it doesn't get turned off by accident with -Wno-unused-value. I also made a few minor improvements to existing unused value warnings in the process. <rdar://problem/11516811>.
llvm-svn: 157362
|
|
|
|
|
|
|
|
|
| |
lambda as referring to a local in an enclosing scope if we're in the
enclosing scope of the lambda (not it's function call operator). Also,
turn the test into an IR generation test, since that's where the
crashes occurred. Really fixes PR12746 / <rdar://problem/11465120>.
llvm-svn: 156926
|
|
|
|
|
|
|
| |
blocks and lambdas, based heavily on a patch from Meador Inge. Fixes
PR12746 / <rdar://problem/11465120>.
llvm-svn: 156925
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
macro.
This fixes the included test case & was reported by Nico Weber.
It's a little bit nasty using the difference in the conversion context, but
seems to me like a not unreasonable solution. I did have to fix up the
conversion context for conditional operators (it seems correct to me to include
the context for which we're actually doing the comparison - across all the
nested conditionals, rather than the innermost conditional which might not
actually have the problematic implicit conversion at all) and template default
arguments (this is a bit of a hack, since we don't have the source location of
the '=' anymore, so I just used the start of the parameter - open to
suggestions there)
llvm-svn: 156861
|
|
|
|
|
|
| |
the sole parameter name in the diagnostic in more cases. Patch by Terry Long!
llvm-svn: 156807
|
|
|
|
|
|
|
| |
which expects exactly one argument, include the name of the argument in
the diagnostic text. Patch by Terry Long!
llvm-svn: 156607
|
|
|
|
|
|
|
|
| |
Sema::ConvertToIntegralOrEnumerationType() from PartialDiagnostics to
abstract "diagnoser" classes. Not much of a win here, but we're
-several PartialDiagnostics.
llvm-svn: 156217
|
|
|
|
| |
llvm-svn: 156180
|
|
|
|
| |
llvm-svn: 156178
|
|
|
|
|
|
|
|
|
|
|
| |
off PartialDiagnostic. PartialDiagnostic is rather heavyweight for
something that is in the critical path and is rarely used. So, switch
over to an abstract-class-based callback mechanism that delays most of
the work until a diagnostic is actually produced. Good for ~11k code
size reduction in the compiler and 1% speedup in -fsyntax-only on the
code in <rdar://problem/11004361>.
llvm-svn: 156176
|
|
|
|
| |
llvm-svn: 156123
|
|
|
|
| |
llvm-svn: 155938
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Apparently we weren't checking default arguments when they were instantiated.
This adds the check, fixes the lack of instantiation caching (which seems like
it was mostly implemented but just missed the last step), and avoids
implementing non-dependent default args (for non-dependent parameter types) as
uninstantiated default arguments (so that we don't warn once for every
instantiation when it's not instantiation dependent).
Reviewed by Richard Smith.
llvm-svn: 155838
|