On Thu, Oct 14, 2021 at 9:50 AM Jeff Layton <jlayton@xxxxxxxxxx> wrote: > > As Greg pointed out, if we get a mangled mdsmap or fsmap, then something > has gone very wrong, and we should avoid doing any activity on the > filesystem. > > When this occurs, shut down the mount the same way we would with a > forced umount by calling ceph_umount_begin when decoding fails on either > map. This causes most operations done against the filesystem to return > an error. Any dirty data or caps in the cache will be dropped as well. Hurray, simple fixes! You have my acked-by, but I'm really not familiar with the kernel shutdown mechanics so I shouldn't provide anything stronger. :) > > The effect is not reversible, so the only remedy is to umount. > > URL: https://tracker.ceph.com/issues/52303 > Cc: Greg Farnum <gfarnum@xxxxxxxxxx> > Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxx> > --- > fs/ceph/mds_client.c | 6 ++++-- > fs/ceph/super.c | 2 +- > fs/ceph/super.h | 1 + > 3 files changed, 6 insertions(+), 3 deletions(-) > > diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c > index 598425ccd020..5490f3422ae2 100644 > --- a/fs/ceph/mds_client.c > +++ b/fs/ceph/mds_client.c > @@ -5011,7 +5011,8 @@ void ceph_mdsc_handle_fsmap(struct ceph_mds_client *mdsc, struct ceph_msg *msg) > return; > > bad: > - pr_err("error decoding fsmap\n"); > + pr_err("error decoding fsmap. Shutting down mount.\n"); > + ceph_umount_begin(mdsc->fsc->sb); > err_out: > mutex_lock(&mdsc->mutex); > mdsc->mdsmap_err = err; > @@ -5078,7 +5079,8 @@ void ceph_mdsc_handle_mdsmap(struct ceph_mds_client *mdsc, struct ceph_msg *msg) > bad_unlock: > mutex_unlock(&mdsc->mutex); > bad: > - pr_err("error decoding mdsmap %d\n", err); > + pr_err("error decoding mdsmap %d. Shutting down mount.\n", err); > + ceph_umount_begin(mdsc->fsc->sb); > return; > } > > diff --git a/fs/ceph/super.c b/fs/ceph/super.c > index 580dad2c832e..fea6e69b94a0 100644 > --- a/fs/ceph/super.c > +++ b/fs/ceph/super.c > @@ -969,7 +969,7 @@ static void __ceph_umount_begin(struct ceph_fs_client *fsc) > * ceph_umount_begin - initiate forced umount. Tear down the > * mount, skipping steps that may hang while waiting for server(s). > */ > -static void ceph_umount_begin(struct super_block *sb) > +void ceph_umount_begin(struct super_block *sb) > { > struct ceph_fs_client *fsc = ceph_sb_to_client(sb); > > diff --git a/fs/ceph/super.h b/fs/ceph/super.h > index 49ca2106f853..7c3990cd3c3b 100644 > --- a/fs/ceph/super.h > +++ b/fs/ceph/super.h > @@ -947,6 +947,7 @@ extern void ceph_put_snapid_map(struct ceph_mds_client* mdsc, > struct ceph_snapid_map *sm); > extern void ceph_trim_snapid_map(struct ceph_mds_client *mdsc); > extern void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc); > +void ceph_umount_begin(struct super_block *sb); > > > /* > -- > 2.31.1 >
