Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 12 Sep 2019 13:58:51 +0200
From:      Hans Petter Selasky <hps@selasky.org>
To:        Jeff Roberson <jeff@FreeBSD.org>, src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org
Subject:   Re: svn commit: r351188 - in head/sys: kern sys
Message-ID:  <7fc855eb-ef02-72a7-071f-4b0e13c8db30@selasky.org>
In-Reply-To: <201908181143.x7IBhxZO011254@repo.freebsd.org>
References:  <201908181143.x7IBhxZO011254@repo.freebsd.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On 2019-08-18 13:43, Jeff Roberson wrote:
> Author: jeff
> Date: Sun Aug 18 11:43:58 2019
> New Revision: 351188
> URL: https://svnweb.freebsd.org/changeset/base/351188
> 
> Log:
>    Add a blocking wait bit to refcount.  This allows refs to be used as a simple
>    barrier.
>    
>    Reviewed by:	markj, kib
>    Discussed with:	jhb
>    Sponsored by:	Netflix
>    Differential Revision:	https://reviews.freebsd.org/D21254
> 
> Modified:
>    head/sys/kern/kern_synch.c
>    head/sys/sys/refcount.h
> 

Hi,

I would like to decouple the wait/wakeup logic from the release 
function, if possible. The main objection here is that we are woken up 
before any destructor is executed. Also this function is not Giant safe 
and does neither assert that it can sleep.

--HPS



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?7fc855eb-ef02-72a7-071f-4b0e13c8db30>