Should sycl::free synchronize-with sycl::malloc?

[Pennycook]

The definition of std::free in C++ redirects to the definition of free in C, which says (in Section 7.22.3 of the C11 draft, "Memory management functions"):

A call to free or realloc that deallocates a region p of memory synchronizes with any allocation call that allocates all or part of the region p. This synchronization occurs after any access of p by the deallocating function, and before any such access by the allocating function.

Should this guarantee hold for sycl::free as well?

I cannot immediately see how a developer could detect or rely upon this behavior, but it does seem reasonable to say that sycl::free on a particular memory region always "happens before" any sycl::malloc call that returns the same memory, and that simply accessing the pointer returned by sycl::malloc cannot lead to a race condition.

[TApplencourt]

Free and Maloc are blocking. And all our API are thread-safe.

SYCL guarantees that all the member functions and special member functions of the SYCL classes described are thread safe.

So I don't understand how any of this can be a problem... But the C people are far smarter than me, so I guess they put it for a reason; so why not copy them : )

---

Side note for our editor @gmlueck before i forgot. We use both "thread-safe" and "thread safe", I guess we may want to converge to one.