Apple clang version 14.0.0 (clang-1400.0.17.3.1) fails the build
because stds.h defines `fallthrough` as a macro which is then
expanded when base.h evaluates
# if __has_attribute(fallthrough) && __clang_major__ >= 5
The macOS SDK defines `DISPATCH_FALLTHROUGH` as the macro instead
of `fallthrough`.
This change replaces the use of `fallthrough` in the tree with
`HEIM_FALLTHROUGH` and updates the declaration in configure logic
to define `HEIM_FALLTHROUGH` based upon existing definitions
(if any) of `fallthrough` or `DISPATCH_FALLTHROUGH`.
Do not leak the object when the intent is to free it.
Introduced by 40d1271094
("asn1: Expand decoration w/ C types")
Change-Id: If8cd502f61d6f9b72118630839525933911c6697
190263bb7a
("assert non-NULL ptrs before calling mem funcs") introduced
two wrong size argument warnings. These locations are not
errors since the allocation is simply to ensure that the
data pointer is non-NULL; length is zero.
Change-Id: I7b3b58247799a48da3653008c7b6d7fbbbf83e25
The definitions of memcpy(), memmove(), and memset() state that
the behaviour is undefined if any of the pointer arguments are
NULL, and some compilers are known to make use of this to
optimise away existing NULL checks in the source.
Change-Id: I489bc256e3eac7ff41d91becb0b43aba73dbb3f9
Link: https://www.imperialviolet.org/2016/06/26/nonnull.html
Assign zero to the output size parameter at the start so that
callers that use the value when an error occurs do not see
garbage that might be misinterpreted.
Change-Id: Iccfcf4f6944b1bf72789c83919901d9b9d6f9153
There remains one tough shift/reduce conflict, the warning for which is
quieted with an `%expect 1` directive.
The remaining conflict has to do with whether a constraint attaches to
the inner type that some other outer type is a SET OF or SEQUENCE OF or
tagged-type of, or whether it attaches to the outer type. The two are
really the same thing. The latter is the reduce side, so it's not used,
but if it were we could grab the constraint in the action and attach it
to the inner type anyways.
The pseudo keyword 'fallthrough' is defined such that case statement
blocks must end with any of these keywords:
* break;
* fallthrough;
* continue;
* goto <label>;
* return [expression];
*
* gcc: https://gcc.gnu.org/onlinedocs/gcc/Statement-Attributes.html#Statement-Attributes
The macro is defined either as
__attribute__((__fallthrough__))
or as
do {} while (0) /* fallthrough */
not including the semicolon.
This change implements the Linux kernel style and updates several locations
where "/*fallthrough*/ and /* FALLTHROUGH */ were not previously replaced.
Externally imported code such as libedit, libtommath and sqlite are
restored to their unaltered state.
Change-Id: I69db8167b0d5884f55d96d72de3059a0235a1ba3
First, we enlarge Member's val field to int64_t.
Then we warn about skipping 2int, int2, and parse units glue for such
things with too-large members.
And we error out when generating the template for such things with
>UINT32_MAX members.
What about too-negative members? That could be a thing for INTEGER /
ENUMERATED. We'll look at that later.
Prior to this commit only those C structs for SET and SEQUENCE types
could be decorated. Now those for CHOICE types also can be decorated.
We could further extend this to SET OF and SEQUENCE OF types if it
proves useful.
Make krb5_pac a heim_object_t and use heim_retain() (i.e. reference count +1)
as the copy constructor for the type decoration.
Note: this assumes that PACs included in naming attributes are immutable.
This is not a very good fix, though the warnings remain. Such errors
can in principle be a problem because in C there is no standard enum
sizing.
In this case we have two enums with the same elements and so the same
size, so it's clearly not a problem.
This commits allows `heim_object_t` as a type and causes the generated
code to use the `heim_retain()` and `heim_release()` functions for
copying and releasing values of such types.
Also, now one can have more than one decoration per-type.
If the decorated external type is a void *, don't include the decoration
header in the ASN.1 header, only in the template implementation. This allows
the copy constructor and destructor to be implementation private.
This adds support for asn1_compile --decorate=... variation that causes
decoration of an ASN.1 SET/SEQUENCE type with a field of a non-ASN.1
type.
This means we can now have an ASN.1 type to represent a request that can
then have a "hidden" field -- hidden in that it is neither encoded nor
decoded. This field will be copied and freed when the decoration is of
an ASN.1 type or of a external, C type that comes with copy constructor
and destructor functions. Decoration with a `void *` field which is
neither copied nor freed is also supported.
We may end up using this to, for example, replace the `hdb_entry_ex`
type by decorating `HDB_entry` with a C type that points to the `HDB` in
which the entry was found or to which it should be written.
Fix regression introduced in 9427796f, where checked in files matching asn1_*.c
would be clobbered by "make clean".
Change-Id: Ifc4bbe3c46dd357fdd642040ad964c7cfe1d395c
