Date: Tue, 21 May 2013 04:36:38 +0000 From: Orit Moskovich <oritm@mellanox.com> To: John Baldwin <jhb@freebsd.org>, "freebsd-arch@freebsd.org" <freebsd-arch@freebsd.org> Subject: RE: FreeBSD spinlock - compatibility layer Message-ID: <981733489AB3BD4DB24B48340F53E0A55B0D091F@MTLDAG01.mtl.com> In-Reply-To: <201305200950.26834.jhb@freebsd.org> References: <981733489AB3BD4DB24B48340F53E0A55B0CFD79@MTLDAG01.mtl.com> <201305200950.26834.jhb@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
That's not the case when using taskqueues for deferring execution of an in= terrupt handler. Tasks can be delayed using the global taskqueue taskqueue_swi, which execut= es its tasks in the context of an interrupt. In this case sleep is forbidden, and using spin mutex is not (although migh= t be not recommended). -----Original Message----- From: John Baldwin [mailto:jhb@freebsd.org]=20 Sent: Monday, May 20, 2013 11:42 PM To: freebsd-arch@freebsd.org Cc: Orit Moskovich Subject: Re: FreeBSD spinlock - compatibility layer On Tuesday, May 14, 2013 6:04:21 am Orit Moskovich wrote: > Hi, >=20 > I read about the FreeBSD mutex implementation for spinlock in the compatibility layer. > I might be wrong, but I noticed a code section that might be problematic: >=20 > Taken from http://svn.freebsd.org/base/release/9.1.0/sys/ofed/include/linux/spinlock.h= : >=20 > static inline void > spin_lock_init(spinlock_t *lock) > { >=20 > memset(&lock->m, 0, sizeof(lock->m)); > mtx_init(&lock->m, "lnxspin", NULL, MTX_DEF | MTX_NOWITNESS);=20 > } >=20 > But MTX_DEF initializes mutex as a sleep mutex: >=20 > By default, MTX_DEF mutexes will context switch when they are already >=20 > held. >=20 >=20 > There is a flag MTX_SPIN Which I think is the right one in this case . >=20 >=20 >=20 > I'd appreciate your take on this issue. Since FreeBSD uses a different approach to interrupt handlers (they run in = threads, not in the bottom half), a regular mutex may in fact give the clos= est match to the same semantics. Regular mutexes are also cheaper and in g= eneral preferable to spin mutexes whenever possible. -- John Baldwin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?981733489AB3BD4DB24B48340F53E0A55B0D091F>