[<prev] [next>] [day] [month] [year] [list]
Message-ID: <1408049748.4951.134.camel@eris.loria.fr>
Date: Thu, 14 Aug 2014 22:55:48 +0200
From: Jens Gustedt <jens.gustedt@...ia.fr>
To: musl@...ts.openwall.com
Subject: c threads, v.5.9
Hello,
this is now a version that I am quite confident of. It is a special
edition that doesn't use spinlocks. The only "principal" thing it
still has that people might not like is dynamic allocation of
auxiliary structures for mtx_t and cnd_t. For the mtx_t, this could
perhaps be avoided, but for the cnd_t it is really convenient, because
it clearly separates different batches of cnd-waiters that should
never be mixed.
In summary:
- (almost) all is private futex
- the internal locks use a minimal test-and-set logic that is modeled
after C's atomic_flag
- mutexes and conditions are simpler (I hope) than their pthread
counterparts because they don't need process sharing or robustness
Best
Jens
--
:: INRIA Nancy Grand Est ::: AlGorille ::: ICube/ICPS :::
:: ::::::::::::::: office Strasbourg : +33 368854536 ::
:: :::::::::::::::::::::: gsm France : +33 651400183 ::
:: ::::::::::::::: gsm international : +49 15737185122 ::
:: http://icube-icps.unistra.fr/index.php/Jens_Gustedt ::
View attachment "thrd11-v5.9.patch" of type "text/x-patch" (53138 bytes)
Download attachment "signature.asc" of type "application/pgp-signature" (199 bytes)
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
