[PATCH 4/6] xfs: add xlog sysfs kobject and attribute handlers
Brian Foster
bfoster at redhat.com
Thu Jun 26 07:28:55 CDT 2014
On Thu, Jun 26, 2014 at 02:47:19PM +1000, Dave Chinner wrote:
> On Fri, Jun 06, 2014 at 09:13:32AM -0400, Brian Foster wrote:
> > Embed a kobject into the xfs log data structure (xlog). This creates a
> > 'log' subdirectory for every XFS mount instance in sysfs. The lifecycle
> > of the log kobject is tied to the lifecycle of the log.
> >
> > Also define a set of generic attribute handlers associated with the log
> > kobject in preparation for the addition of attributes.
>
> The code works fine, but....
> >
> > Signed-off-by: Brian Foster <bfoster at redhat.com>
> > ---
> > fs/xfs/xfs_log.c | 9 +++++++++
> > fs/xfs/xfs_log_priv.h | 3 +++
> > fs/xfs/xfs_sysfs.c | 53 +++++++++++++++++++++++++++++++++++++++++++++++++++
> > fs/xfs/xfs_sysfs.h | 1 +
> > 4 files changed, 66 insertions(+)
> >
> > diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
> > index 292308d..8eb10d5 100644
> > --- a/fs/xfs/xfs_log.c
> > +++ b/fs/xfs/xfs_log.c
> > @@ -34,6 +34,7 @@
> > #include "xfs_trace.h"
> > #include "xfs_fsops.h"
> > #include "xfs_cksum.h"
> > +#include "xfs_sysfs.h"
> >
> > kmem_zone_t *xfs_log_ticket_zone;
> >
> > @@ -707,6 +708,11 @@ xfs_log_mount(
> > }
> > }
> >
> > + error = xfs_sysfs_init(&mp->m_log->l_kobject, &xfs_log_ktype,
> > + &mp->m_log->l_kobject_complete, &mp->m_kobject, "log");
> > + if (error)
> > + goto out_destroy_ail;
>
> ... that's, ummm, rather verbose. At minimum, a local log variable,
> but I suspect that if the pattern of "all sysfs dirs have a kobject
> and a completion" and they are always used together that maybe
> we should have a:
>
> struct xfs_kobj {
> struct kobject kobj;
> struct completion complete;
> };
>
> And we pass them around instead. that would make this:
>
> error = xfs_sysfs_init(&log->l_kobj, &xfs_log_ktype,
> &mp->m_kobj, "log");
>
> which is much easier to read....
>
Indeed, that's a nice cleanup.
> > +
> > /* Normal transactions can now occur */
> > mp->m_log->l_flags &= ~XLOG_ACTIVE_RECOVERY;
> >
> > @@ -947,6 +953,9 @@ xfs_log_unmount(
> > xfs_log_quiesce(mp);
> >
> > xfs_trans_ail_destroy(mp);
> > +
> > + xfs_sysfs_del(&mp->m_log->l_kobject, &mp->m_log->l_kobject_complete);
>
> And that would become:
>
> xfs_sysfs_del(&log->l_kobj);
>
> > +
> > xlog_dealloc_log(mp->m_log);
> > }
> >
> > diff --git a/fs/xfs/xfs_log_priv.h b/fs/xfs/xfs_log_priv.h
> > index 9bc403a..ce1eee2 100644
> > --- a/fs/xfs/xfs_log_priv.h
> > +++ b/fs/xfs/xfs_log_priv.h
> > @@ -405,6 +405,9 @@ struct xlog {
> > struct xlog_grant_head l_reserve_head;
> > struct xlog_grant_head l_write_head;
> >
> > + struct kobject l_kobject;
> > + struct completion l_kobject_complete;
>
> struct xfs_kobj l_kobj;
>
> > +
> > /* The following field are used for debugging; need to hold icloglock */
> > #ifdef DEBUG
> > char *l_iclog_bak[XLOG_MAX_ICLOGS];
> > diff --git a/fs/xfs/xfs_sysfs.c b/fs/xfs/xfs_sysfs.c
> > index 41365fe..f837527 100644
> > --- a/fs/xfs/xfs_sysfs.c
> > +++ b/fs/xfs/xfs_sysfs.c
> > @@ -54,3 +54,56 @@ xfs_mp_release(struct kobject *kobj)
> > struct kobj_type xfs_mp_ktype = {
> > .release = xfs_mp_release,
> > };
> > +
> > +/* xlog */
> > +
> > +static struct attribute *xfs_log_attrs[] = {
> > + NULL,
> > +};
> > +
> > +STATIC ssize_t
> > +xfs_log_show(
> > + struct kobject *kobj,
> > + struct attribute *attr,
> > + char *buf)
> > +{
> > + struct xlog *log = container_of(kobj, struct xlog, l_kobject);
>
> > + struct xfs_sysfs_attr *xfs_attr = container_of(attr,
> > + struct xfs_sysfs_attr, attr);
> > +
> > + return xfs_attr->show ? xfs_attr->show(buf, log) : 0;
>
> I think this could be made much less verbose:
>
> static inline struct xlog *
> to_xlog(struct kobject *kobj)
> {
> return container_of(...)
> }
>
> static inline struct xfs_sysfs_attr *
> to_attr(struct attribute *attr)
> {
> return container_of(...)
> }
>
> so this becomes:
>
> {
> struct xlog *log = to_xlog(kobj);
> struct xfs_sysfs_attr *attr = to_attr(kattr);
>
> return attr->show ? attr->show(buf, log) : 0;
> }
>
OK.
>
> > +}
> > +
> > +STATIC ssize_t
> > +xfs_log_store(
> > + struct kobject *kobj,
> > + struct attribute *attr,
> > + const char *buf,
> > + size_t count)
> > +{
> > + struct xlog *log = container_of(kobj, struct xlog, l_kobject);
> > + struct xfs_sysfs_attr *xfs_attr = container_of(attr,
> > + struct xfs_sysfs_attr, attr);
> > +
> > + return xfs_attr->store ? xfs_attr->store(buf, count, log) : 0;
> > +}
> > +
> > +static struct sysfs_ops xfs_log_ops = {
> > + .show = xfs_log_show,
> > + .store = xfs_log_store,
> > +};
> > +
> > +STATIC void
> > +xfs_log_release(struct kobject *kobj)
> > +{
> > + struct xlog *log = container_of(kobj, struct xlog, l_kobject);
> > +
> > + complete(&log->l_kobject_complete);
> > +}
>
> If the release funtion is common with other types, then the xfs_kobj
> structure is perfect for this use - it will prevent a heap of
> duplicated release functions...
>
It's going to look virtually the same for every kobject. Unfortunately,
it needs to go from kobj->xfs_object->xfs_kobj, so each type requires a
unique definition. We might be able to just turn it into a macro or
something that takes the appropriate info and reduces the clutter. I'll
play around with it. Thanks.
Brian
> Cheers,
>
> Dave.
> --
> Dave Chinner
> david at fromorbit.com
More information about the xfs
mailing list