Intentionally or not, this post demonstrates one of the things that makes safer abstractions in C less desirable: the shared pointer implementation uses a POSIX mutex, which means it’s (1) not cross platform, and (2) pays the mutex overhead even in provably single-threaded contexts. In other words, it’s not a zero-cost abstraction.

C++’s shared pointer has the same problem; Rust avoids it by having two types (Rc and Arc) that the developer can select from (and which the compiler will prevent you from using unsafely).

▲ kouteiheika 4 hours ago | parent | next [-]

> the shared pointer implementation uses a POSIX mutex [...] C++’s shared pointer has the same problem

It doesn't. C++'s shared pointers use atomics, just like Rust's Arc does. There's no good reason (unless you have some very exotic requirements, into which I won't get into here) to implement shared pointers with mutexes. The implementation in the blog post here is just suboptimal.

(But it's true that C++ doesn't have Rust's equivalent of Rc, which means that if you just need a reference counted pointer then using std::shared_ptr is not a zero cost abstraction.)

▲ woodruffw 4 hours ago | parent | next [-]

To be clear, the “same problem” is that it’s not a zero-cost abstraction, not that it uses the same specific suboptimal approach as this blog post.

▲

kouteiheika 4 hours ago | parent [-]

I think that's an orthogonal issue. It's not that C++'s shared pointer is not a zero cost abstraction (it's as much a zero cost abstraction as in Rust), but that it only provides one type of a shared pointer.

But I suppose we're wasting time on useless nitpicking. So, fair enough.

▲

woodruffw 4 hours ago | parent [-]

I think they’re one and the same: C++ doesn’t have program-level thread safety by construction, so primitives like shared pointers need to be defensive by default instead of letting the user pick the right properties for their use case.

Edit: in other words C++ could provide an equivalent of Rc, but we’d see no end of people complaining when they shoot themselves in the foot with it.

(This is what “zero cost abstraction” means: it doesn’t mean no cost, just that the abstraction’s cost is no greater than the semantically equivalent version written by the user. So both Arc and shared_ptr are zero-cost in a MT setting, but only Rust has a zero-cost abstraction in a single-threaded setting.)

▲

kouteiheika 4 hours ago | parent [-]

I can't say I agree with this? If C++ had an Rc equivalent (or if you'd write one yourself) it would be just as zero cost as it is in Rust, both in a single-threaded setting and in a multithreaded-setting. "Zero cost abstraction" doesn't mean that it cannot be misused or that it doesn't have any cognitive overhead to use correctly, just that it matches whatever you'd write without the abstraction in place. Plenty of "zero cost" features in C++ still need to you pay attention to not accidentally blow you leg off.

Simply put, just as a `unique_ptr` (`Box`) is an entirely different abstraction than `shared_ptr` (`Arc`), an `Rc` is also an entirely different abstraction than `Arc`, and C++ simply happens to completely lack `Rc` (at least in the standard; Boost of course has one). But if it had one you could use it with exactly the same cost as in Rust, you'd just have to manually make sure to not use it across threads (which indeed is easier said than done, which is why it's not in the standard), exactly the same as if you'd manually maintain the reference count without the nice(er) abstraction. Hence "zero cost abstraction".

	▲	woodruffw 3 hours ago \| parent [-]
		Sorry, I realized I’m mixing two things in a confusing way: you’re right that C++ could easily have a standard zero-cost Rc equivalent; I’m saying that it can’t have a safe one. I think this is relevant given the weight OP gives to both performance and safety.

▲ cogman10 4 hours ago | parent | prev [-]

> very exotic requirements

I'd be interested to know what you are thinking.

The primary exotic thing I can imagine is an architecture lacking the ability to do atomic operations. But even in that case, C11 has atomic operations [1] built in. So worst case, the C library for the target architecture would likely boil down to mutex operations.

[1] https://en.cppreference.com/w/c/atomic.html

▲ kouteiheika 3 hours ago | parent | next [-]

Well, basically, yeah, if your platform lacks support for atomics, or if you'd need some extra functionality around the shared pointer like e.g. logging the shared pointer refcounts while enforcing consistent ordering of logs (which can be useful if you're unfortunate enough to have to debug a race condition where you need to pay attention to refcounts, assuming the extra mutex won't make your heisenbug disappear), or synchronizing something else along with the refcount (basically a "fat", custom shared pointer that does more than just shared-pointering).

▲

colonwqbang 2 hours ago | parent [-]

Does there exist any platform which has multithreading but not atomics? Such a platform would be quite impractical as you can't really implement locks or any other threading primitive without atomics.

▲

addaon 16 minutes ago | parent | next [-]

> Does there exist any platform which has multithreading but not atomics?

Yes. Also, almost every platform I know that supports multi threading and atomics doesn’t support atomics between /all/ possible masters. Consider a microcontroller with, say, two Arm cores (multithreaded, atomic-supporting) and a DMA engine.

▲

cogman10 2 hours ago | parent | prev [-]

Certainly such systems can pretty readily exist. You merely need atomic reads/writes in order to implement locks.

You can't create userspace locks which is a bummer, but the OS has the capability of enforcing locks. That's basically how early locking worked.

The main thing needed to make a correct lock is interrupt protection. Something every OS has.

To go fast, you need atomic operations. It especially becomes important if you are dealing with multiple cores. However, for a single core system atomics aren't needed for the OS to create locks.

	▲	colonwqbang 39 minutes ago \| parent [-]
		I wrote "multithreaded" but I really meant "multicore". If two cores are contending for a lock I don't see how irq protection help. As long as there is only one core, I agree.

▲ goalieca 3 hours ago | parent | prev [-]

Which platforms might that be? Even MIPS has atomics (at least pointer sized last i checked).

▲ cogman10 2 hours ago | parent [-]

AFIAK, and I'm not MIPS expert, but I believe it doesn't have the ability to add a value directly to a memory address. You have to do something like

    // Not real MIPS, just what I've gleaned from a brief look at some docs
    LOAD addr, register
    ADD 1, register
    STORE register, addr

The LOAD and STORE are atomic, but the `ADD` happens out of band.

That's a problem if any sort of interrupt happens (if you are multi-threading then a possibility). If it happens at the load, then a separate thread can update "addr" which mean the later STORE will stomp on what's there.

x86 and ARM can do

    ADD 1, addr

as well as other instructions like "compare and swap"

    LOAD addr, register
    MOV register, register2
    ADD 1, register2
    COMPARE_AND_SWAP addr, register, register2
    if (cas_failed) { try again }

	▲	unnah an hour ago \| parent [-]
		On MIPS you can simulate atomics with a load-linked/store-conditional (LL/SC) loop. If another processor has changed the same address between the LL and SC instructions, the SC fails to store the result and you have to retry. The underlying idea is that the processors would have to communicate memory accesses to each other via the cache coherence protocol anyway, so they can easily detect conflicting writes between the LL and SC instructions. It gets more complicated with out-of-order execution... `loop: LL r2, (r1) ADD r3, r2, 1 SC r3, (r1) BEQ r3, 0, loop NOP`

▲ spacedcowboy 4 hours ago | parent | prev | next [-]

The number of times I might want to write something in C and have it less likely to crash absolutely dwarfs the number of times I care about that code being cross-platform.

Sure, cross-platform is desirable, if there's no cost involved, and mandatory if you actually need it, but it's a "nice to have" most of the time, not a "needs this".

As for mutex overheads, yep, that's annoying, but really, how annoying ? Modern CPUs are fast. Very very fast. Personally I'm far more likely to use an os_unfair_lock_t than a pthread_mutex_t (see the previous point) which minimizes the locking to a memory barrier, but even if locking were slow, I think I'd prefer safe.

Rust is, I'm sure, great. It's not something I'm personally interested in getting involved with, but it's not necessary for C (or even this extra header) to do everything that Rust can do, for it to be an improvement on what is available.

There's simply too much out there written in C to say "just use Rust, or Swift, or ..." - too many libraries, too many resources, too many tutorials, etc. You pays your money and takes your choice.

	▲	woodruffw 4 hours ago \| parent \| next [-]
		That’s all reasonable, but here’s one of the primary motivations from the post: > We love its raw speed, its direct connection to the metal If this is a strong motivating factor (versus, say, refactoring risk), then C’s lack of safe zero-cost abstractions is a valid concern.
	▲	lelanthran 2 hours ago \| parent \| prev [-]
		> As for mutex overheads, yep, that's annoying, but really, how annoying ? For this use-case, you might not notice. ISTR, when examing the pthreads source code for some platform, that mutexes only do a context switch as a fallback, if the lock cannot be acquired. So, for most use-cases of this header, you should not see any performance impact. You'll see some bloat, to be sure.

▲ accelbred an hour ago | parent | prev | next [-]

Unfortunately, for C++, thats not true. At least with glibc and libstdc++, if you do not link with pthreads, then shared pointers are not thread-safe. At runtime it will do a symbol lookup for a pthreads symbol, and based off the result, the shared pointer code will either take the atomic or non-atomic path.

I'd much rather it didnt try to be zero-cost and it always used atomics...

	▲	woodruffw 44 minutes ago \| parent [-]
		This is, impressively, significantly worse than I realized!

▲ layer8 20 minutes ago | parent | prev | next [-]

The shared-pointer implementation isn’t actually shown (i.e. shared_ptr_copy), and the SharedPtr type doesn’t use a pthread_mutex_t.

▲ lelanthran 2 hours ago | parent | prev | next [-]

> Intentionally or not, this post demonstrates one of the things that makes safer abstractions in C less desirable: the shared pointer implementation uses a POSIX mutex, which means it’s (1) not cross platform, and (2) pays the mutex overhead even in provably single-threaded contexts. In other words, it’s not a zero-cost abstraction.

It's an implementation detail. They could have used atomic load/store (since c11) to implement the increment/decrement.

TBH I'm not sure what a mutex buys you in this situation (reference counting)

▲ kev009 39 minutes ago | parent | prev | next [-]

C11 has a mutex API (threads.h), so why would it rely on POSIX? Are you sure it's not an runtime detail on one platform? https://devblogs.microsoft.com/cppblog/c11-threads-in-visual...

▲ aidenn0 an hour ago | parent | prev | next [-]

> the shared pointer implementation uses a POSIX mutex

Do you have a source for this? I couldn't find the implementation in TFA nor a link to safe_c.h

▲ saurik 4 hours ago | parent | prev [-]

I'd think a POSIX mutex--a standard API that I not only could implement anywhere, but which has already been implemented all over the place--is way more "cross platform" than use of atomics.

	▲	woodruffw 3 hours ago \| parent [-]
		To lift things up a level: I think a language’s abstractions have failed if we even need to have a conversation around what “cross platform” really means :-)