Re: FreeBSD's Google Summer of Code 2016

[Pedro Giffuni <pfg@FreeBSD.org>]

On 03/05/16 19:25, Rich Felker wrote:
> On Sat, Mar 05, 2016 at 07:14:34PM -0500, Pedro Giffuni wrote:
>>
>>
>> On 03/05/16 18:32, Rich Felker wrote:
>>> On Sat, Mar 05, 2016 at 05:41:25PM -0500, Pedro Giffuni wrote:
>>>> First of all, great to hear there is interest on the musl side too.
>>>>
>>>> I think the biggest precedent of porting linux-oriented C libraries
>>>> came from Debian's kFreeBSD. We accomodated a little by for them
>>>> by defining __FreeBSD_kernel__ in sys/param.h.
>>>>
>>>> While using the optional linux-abi futex in FreeBSD could be an option,
>>>> it is not really the cleanest option. The Debian guys did a port of
>>>> NPTL using regular pthreads:
>>>>
>>
>> Of course I ahould have meant "based on regular FreeBSD kernel services".
>>
>>>> http://thread.gmane.org/gmane.linux.debian.ports.bsd/11702
>>>>
>>>> I am certain this will require more research but it would be useful
>>>> for other ports as well.
>>>
>>
>> We could ask Petr Salinger for the details, but I am pretty sure
>> FreeBSD has the required functionality natively.
>>
>>> Glibc/NPTL has a lot of what I'd call "gratuitous abstraction" (like
>>> the lll stuff) in their pthread primitives which makes this
>>> "possible". I call it gratuitous because it's really really hard to
>>> achieve correct implementations of the pthread sync primitives that
>>> don't have serious corner-case bugs, and it's unlikely that their
>>> abstractions actually suffice to make correct alternate
>>> implementations.
>>>
>>> musl does not have any such abstraction. We require a compare-and-swap
>>> operation or equivalent on which arbitrary atomic operations can be
>>> constructed, a futex or equivalent operation that's roughly
>>> while(*addr==expected) sleep(), and implement all the sync primitives
>>> just once on top of these.
>>>
>>
>> I am not a threading expert (or even a CS guy), but it sounds like
>> mutex(9) with condvar(9) would do [1]:
>
> No, they don't satisfy the needs of musl; they have their own
> additional storage requirements and are probably not AS-safe. It might
> be possible to use them to implement a userspace-emulated futex queue
> (only if they are AS-safe), but I don't see a way to extend that to
> the process-shared case.
>

OK, it looks like sema(9) may be nearer (and also simpler but slower).
For the process-shared case libthr(2) uses the stuff in sys/utmx.h,
shich should be looked at but it is not documented[1].

Luckily Ed, the developer that would be mentoring the project, knows 
this stuff better than I do.

Pedro.

[1] http://fxr.watson.org/fxr/source/sys/umtx.h?v=FREEBSD8