From mboxrd@z Thu Jan 1 00:00:00 1970 From: NeilBrown Subject: Re: [PATCH] FIX: Remove error message during reshape restart Date: Wed, 21 Sep 2011 17:58:56 +1000 Message-ID: <20110921175856.195e94cb@notabene.brown> References: <20110916135637.10460.87995.stgit@gklab-128-013.igk.intel.com> <20110921130935.563effd1@notabene.brown> <79556383A0E1384DB3A3903742AAC04A032491@IRSMSX101.ger.corp.intel.com> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=PGP-SHA1; boundary="Sig_/WYdR3CHu55o_52bzFMUCO0O"; protocol="application/pgp-signature" Return-path: In-Reply-To: <79556383A0E1384DB3A3903742AAC04A032491@IRSMSX101.ger.corp.intel.com> Sender: linux-raid-owner@vger.kernel.org To: "Kwolek, Adam" Cc: "linux-raid@vger.kernel.org" , "Ciechanowski, Ed" , "Labun, Marcin" List-Id: linux-raid.ids --Sig_/WYdR3CHu55o_52bzFMUCO0O Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable On Wed, 21 Sep 2011 07:20:33 +0000 "Kwolek, Adam" wrote: >=20 >=20 > > -----Original Message----- > > From: NeilBrown [mailto:neilb@suse.de] > > Sent: Wednesday, September 21, 2011 5:10 AM > > To: Kwolek, Adam > > Cc: linux-raid@vger.kernel.org; Ciechanowski, Ed; Labun, Marcin > > Subject: Re: [PATCH] FIX: Remove error message during reshape restart > >=20 > > On Fri, 16 Sep 2011 15:56:38 +0200 Adam Kwolek > > wrote: > >=20 > > > When reshape is restarted freeze() is called on blocked already array. > > > Meaningless error message is displayed and reshape is continued > > > without interruption. > >=20 > > Maybe I'm just having a bad day, but again I cannot understand the situ= ation > > that you are trying to fix. >=20 > In assemble_container_content () one of the first command for reshape cas= e is:=20 > if (content->reshape_active) > block_subarray(content); >=20 > This means that call block_monitor () will produce error message while ar= ray is frozen. >=20 > I think, that in fact for reshape continue case block_monitor () should s= kip frozen > arrays and freeze all other /not frozen yet/. I've got the following chan= ge in my sent Q (this will replace suppressing error message patch): >=20 > instead: > if (sra->text_version[0] =3D=3D '-') > break; >=20 > I think should be: >=20 > if (sra->text_version[0] =3D=3D '-') { > if (restart) > continue; > else > break; > } Why not just get rid of it completely - it doesn't seem to be necessary. NeilBrown >=20 > This will freeze whole container during reshape continuation (just like f= or reshape start). >=20 > A few more fixes for freezing are in my Q also. I'll send them soon. >=20 >=20 > >=20 > > When we restart a reshape the array should not be frozen should it? Ma= ybe > > it is after your new series of patches, in which case if you think this= is still > > relevant, please add this to that series. > >=20 > > If a reshaping array is started while in the 'initrd', then I think the= only way > > that it should be frozen is that sync_max should be set to where the re= shape > > is currently up to. >=20 > This patch is independent to 'initrd' problem, but I think about it. >=20 > BR > Adam >=20 >=20 >=20 >=20 > >=20 > > NeilBrown > >=20 > >=20 > > > > > > Disable error message for reshape restart displayed by freeze() > > > function. > > > > > > Signed-off-by: Adam Kwolek > > > --- > > > > > > Grow.c | 17 +++++++++-------- > > > msg.c | 7 ++++--- > > > msg.h | 2 +- > > > 3 files changed, 14 insertions(+), 12 deletions(-) > > > > > > diff --git a/Grow.c b/Grow.c > > > index 04e6679..15d4a65 100644 > > > --- a/Grow.c > > > +++ b/Grow.c > > > @@ -478,7 +478,7 @@ static int check_idle(struct supertype *st) > > > return is_idle; > > > } > > > > > > -static int freeze_container(struct supertype *st) > > > +static int freeze_container(struct supertype *st, int restart) > > > { > > > int container_dev =3D (st->container_dev !=3D NoMdDev > > > ? st->container_dev : st->devnum); @@ -486,11 > > +486,12 @@ > > > static int freeze_container(struct supertype *st) > > > > > > if (!check_idle(st)) > > > return -1; > > > - > > > + > > > fmt_devname(container, container_dev); > > > > > > - if (block_monitor(container, 1)) { > > > - fprintf(stderr, Name ": failed to freeze container\n"); > > > + if (block_monitor(container, 1, restart)) { > > > + if (!restart) > > > + fprintf(stderr, Name ": failed to freeze container\n"); > > > return -2; > > > } > > > > > > @@ -508,7 +509,7 @@ static void unfreeze_container(struct supertype > > *st) > > > unblock_monitor(container, 1); > > > } > > > > > > -static int freeze(struct supertype *st) > > > +static int freeze(struct supertype *st, int restart) > > > { > > > /* Try to freeze resync/rebuild on this array/container. > > > * Return -1 if the array is busy, > > > @@ -517,7 +518,7 @@ static int freeze(struct supertype *st) > > > * return 1 if it worked. > > > */ > > > if (st->ss->external) > > > - return freeze_container(st); > > > + return freeze_container(st, restart); > > > else { > > > struct mdinfo *sra =3D sysfs_read(-1, st->devnum, > > GET_VERSION); > > > int err; > > > @@ -1444,7 +1445,7 @@ int Grow_reshape(char *devname, int fd, int > > quiet, char *backup_file, > > > devname); > > > return 1; > > > } > > > - frozen =3D freeze(st); > > > + frozen =3D freeze(st, 0); > > > if (frozen < -1) { > > > /* freeze() already spewed the reason */ > > > return 1; > > > @@ -3846,7 +3847,7 @@ int Grow_continue(int mdfd, struct supertype > > *st, struct mdinfo *info, > > > if (st->ss->external) { > > > fmt_devname(buf, st->container_dev); > > > container =3D buf; > > > - freeze(st); > > > + freeze(st, 1); > > > > > > if (!mdmon_running(st->container_dev)) > > > start_mdmon(st->container_dev); > > > diff --git a/msg.c b/msg.c > > > index a10c930..0cae6ba 100644 > > > --- a/msg.c > > > +++ b/msg.c > > > @@ -300,7 +300,7 @@ int block_subarray(struct mdinfo *sra) > > > * As older versions of mdmon (which might run from initrd) don't > > understand > > > * this, we first check that the running mdmon is new enough. > > > */ > > > -int block_monitor(char *container, const int freeze) > > > +int block_monitor(char *container, const int freeze, const int > > > +restart) > > > { > > > int devnum =3D devname2devnum(container); > > > struct mdstat_ent *ent, *e, *e2; > > > @@ -388,8 +388,9 @@ int block_monitor(char *container, const int free= ze) > > > } > > > > > > if (e) { > > > - fprintf(stderr, Name ": failed to freeze subarray%s\n", > > > - to_subarray(e, container)); > > > + if (!restart) > > > + fprintf(stderr, Name ": failed to freeze > > subarray%s\n", > > > + to_subarray(e, container)); > > > > > > /* thaw the partially frozen container */ > > > for (e2 =3D ent; e2 && e2 !=3D e; e2 =3D e2->next) { diff --git > > a/msg.h > > > b/msg.h index c6d037d..1c8565b 100644 > > > --- a/msg.h > > > +++ b/msg.h > > > @@ -30,7 +30,7 @@ extern int ping_monitor(char *devname); extern int > > > ping_monitor_by_id(int devnum); extern int block_subarray(struct > > > mdinfo *sra); extern int unblock_subarray(struct mdinfo *sra, const > > > int unfreeze); -extern int block_monitor(char *container, const int > > > freeze); > > > +extern int block_monitor(char *container, const int freeze, const int > > > +restart); > > > extern void unblock_monitor(char *container, const int unfreeze); > > > extern int fping_monitor(int sock); extern int ping_manager(char > > > *devname); --Sig_/WYdR3CHu55o_52bzFMUCO0O Content-Type: application/pgp-signature; name=signature.asc Content-Disposition: attachment; filename=signature.asc -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.18 (GNU/Linux) iD8DBQFOeZlAG5fc6gV+Wb0RAqaLAJ9VYr1K3M9imL54l1/DP0hJVpDdDwCfSjUl C5mT1uZdgtDf0OG0Ezr3Wls= =aSy8 -----END PGP SIGNATURE----- --Sig_/WYdR3CHu55o_52bzFMUCO0O--