================
@@ -2489,6 +2489,81 @@ are listed below.
$ clang -fuse-ld=lld -Oz -Wl,--icf=safe -fcodegen-data-use code.cc
+Strict Aliasing
+---------------
+
+The C and C++ standards require accesses to objects in memory to use l-values
of
+an appropriate type for the object. This is called *strict aliasing* or
+*type-based alias analysis*. Strict aliasing enhances a variety of powerful
+memory optimizations, including reordering, combining, and eliminating memory
+accesses. These optimizations can lead to unexpected behavior in code that
+violates the strict aliasing rules. For example:
+
+.. code-block:: c++
+
+ void advance(size_t *index, double *data) {
+ double value = data[*index];
+ /* Clang may assume that this store does not change the contents of
`data`. */
+ *index += 1;
+ /* Clang may assume that this store does not change the contents of
`index`. */
+ data[*index] = value;
+ /* Either of these facts may create significant optimization
opportunities
+ if Clang is able to inline this function. */
+ }
+
+Strict aliasing can be explicitly enabled with ``-fstrict-aliasing`` and
+disabled with ``-fno-strict-aliasing``. ``clang-cl`` defaults to
+``-fno-strict-aliasing``; see :ref:`Strict aliasing in clang-cl.
+<clang_cl_strict_aliasing>`. Otherwise, Clang defaults to
``-fstrict-aliasing``.
+
+C and C++ specify slightly different rules for strict aliasing. To improve
+language interoperability, Clang allows two types to alias if either language
+would permit it. This includes applying the C++ similar types rule to C,
+allowing ``int **`` to alias ``int const * const *``. Clang also relaxes the
+standard aliasing rules in the following ways:
+
+* All integer types of the same size are permitted to alias each other,
+ including signed and unsigned types.
+* ``void*`` is permitted to alias any pointer type, ``void**`` is permitted to
+ alias any pointer to pointer type, and so on.
----------------
rjmccall wrote:
Please review the code to see if there are any other relaxations we want to
document here. We shouldn't document anything that we might want to strengthen
in the future (so, places where we're just being a bit lazy about TBAA
fidelity), but anything that we've made a policy decision not to distinguish in
TBAA should probably go in this list.
https://github.com/llvm/llvm-project/pull/122116
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
