Date: Sat, 20 Jun 2015 09:21:35 -0400 From: John Baldwin <jhb@FreeBSD.org> To: Gleb Smirnoff <glebius@FreeBSD.org>, Mateusz Guzik <mjg@FreeBSD.org> Cc: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org Subject: Re: svn commit: r284211 - in head/sys: kern sys Message-ID: <558568DF.3060902@FreeBSD.org> In-Reply-To: <20150616134352.GI73119@FreeBSD.org> References: <201506100934.t5A9YpoN093925@svn.freebsd.org> <20150610154445.GJ73119@FreeBSD.org> <20150616134352.GI73119@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 6/16/15 9:43 AM, Gleb Smirnoff wrote: > Mateusz, > > have you noticed this email? > > On Wed, Jun 10, 2015 at 06:44:45PM +0300, Gleb Smirnoff wrote: > T> On Wed, Jun 10, 2015 at 09:34:51AM +0000, Mateusz Guzik wrote: > T> M> Author: mjg > T> M> Date: Wed Jun 10 09:34:50 2015 > T> M> New Revision: 284211 > T> M> URL: https://svnweb.freebsd.org/changeset/base/284211 > T> M> > T> M> Log: > T> M> fd: use atomics to manage fd_refcnt and fd_holcnt > T> M> > T> M> This gets rid of fdesc_mtx. > T> .. > T> M> Modified: head/sys/sys/filedesc.h > T> M> ============================================================================== > T> M> --- head/sys/sys/filedesc.h Wed Jun 10 05:39:48 2015 (r284210) > T> M> +++ head/sys/sys/filedesc.h Wed Jun 10 09:34:50 2015 (r284211) > T> M> @@ -83,8 +83,8 @@ struct filedesc { > T> M> int fd_lastfile; /* high-water mark of fd_ofiles */ > T> M> int fd_freefile; /* approx. next free file */ > T> M> u_short fd_cmask; /* mask for file creation */ > T> M> - u_short fd_refcnt; /* thread reference count */ > T> M> - u_short fd_holdcnt; /* hold count on structure + mutex */ > T> M> + int fd_refcnt; /* thread reference count */ > T> M> + int fd_holdcnt; /* hold count on structure + mutex */ > T> M> struct sx fd_sx; /* protects members of this struct */ > T> M> struct kqlist fd_kqlist; /* list of kqueues on this filedesc */ > T> M> int fd_holdleaderscount; /* block fdfree() for shared close() */ > T> > T> Shouldn't they be volatile u_int? Agreed. -- John Baldwin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?558568DF.3060902>