Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 13 Feb 2015 22:27:45 +0000
From:      "ian (Ian Lepore)" <phabric-noreply@FreeBSD.org>
To:        freebsd-arm@freebsd.org
Subject:   [Differential] [Changed Subscribers] D1833: Add memory barriers to buf_ring
Message-ID:  <5222a64f48ca455a47b47cd0b337ecb0@localhost.localdomain>
In-Reply-To: <differential-rev-PHID-DREV-ntnotl4wj5wcj2eoukly-req@FreeBSD.org>
References:  <differential-rev-PHID-DREV-ntnotl4wj5wcj2eoukly-req@FreeBSD.org>

next in thread | previous in thread | raw e-mail | index | archive | help
ian added a subscriber: ian.
ian added a comment.

an atomic_store_rel would put the dmb() before writing the producer ring head, the opposite of what this change proposes (which reinforces my vague feeling that this change isn't quite right somehow).  On the consumer side, it's not clear what piece of data would be the subject of an atomic_load_acq(), because I can't really tell from the current placement of the rmb() what reordering it's trying to prevent.

REVISION DETAIL
  https://reviews.freebsd.org/D1833

To: zbb, kmacy, imp, rpaulo
Cc: ian, adrian, freebsd-arm



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5222a64f48ca455a47b47cd0b337ecb0>