Date: Wed, 26 Jan 2011 16:31:16 -0700 From: Warner Losh <imp@bsdimp.com> To: John Baldwin <jhb@freebsd.org> Cc: svn-src-projects@freebsd.org, src-committers@freebsd.org Subject: Re: svn commit: r217907 - projects/graid/head/sys/geom/raid Message-ID: <4D40AEC4.4070607@bsdimp.com> In-Reply-To: <201101261654.51397.jhb@freebsd.org> References: <201101262034.p0QKYGkB089319@svn.freebsd.org> <201101261654.51397.jhb@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 01/26/2011 14:54, John Baldwin wrote: > On Wednesday, January 26, 2011 3:34:16 pm Warner Losh wrote: >> Author: imp >> Date: Wed Jan 26 20:34:16 2011 >> New Revision: 217907 >> URL: http://svn.freebsd.org/changeset/base/217907 >> >> Log: >> Set arg before timeout function pointer to avoid a 1 instruction race. >> >> Modified: >> projects/graid/head/sys/geom/raid/tr_raid1.c >> >> Modified: projects/graid/head/sys/geom/raid/tr_raid1.c >> > ============================================================================== >> --- projects/graid/head/sys/geom/raid/tr_raid1.c Wed Jan 26 20:29:28 2011 > (r217906)+++ projects/graid/head/sys/geom/raid/tr_raid1.c Wed Jan 26 20:34:16 2011 > (r217907) >> @@ -259,8 +259,8 @@ g_raid_tr_update_state_raid1(struct g_ra >> G_RAID_EVENT_VOLUME); >> if (s == G_RAID_VOLUME_S_DEGRADED) { >> g_raid_tr_raid1_rebuild_start(vol->v_tr, vol); >> - vol->v_timeout = g_raid_tr_raid1_idle_rebuild; >> vol->v_to_arg = trs; >> + vol->v_timeout = g_raid_tr_raid1_idle_rebuild; >> } else { >> vol->v_timeout = 0; >> vol->v_to_arg = 0; > Err, this can't solve anything without a memory barrier given the ability of > modern CPUs on at least some architectures to reorder writes True. I thought this was the cause of a crash, but it turns out it was elsewhere. I'm reworking this, but thanks for the info. Warner
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4D40AEC4.4070607>