On Wed, Oct 12, 2016 at 2:07 PM, Tomasz Majchrzak <[email protected]> wrote: > If kernfs file is empty on a first read, successive read operations > using the same file descriptor will return no data, even when data is > available. Default kernfs 'seq_next' implementation advances iterator > position even when next object is not there. Kernfs 'seq_start' for > following requests will not return iterator as position is already on > the second object. > > This bug doesn't allow to monitor badblocks sysfs files from MD raid. > They are initially empty but if data appears at some stage, userspace is > not able to read it. It doesn't affect any released applications but it > is necessary for upcoming bad block support for external metadata in MD > raid.
What is the expectation from the userspace application? Should seq_file support "tail -f" as more data is added to the file? AFAICS this patch doesn't address that generally, just the empty->nonempty transition (with the single record case). Why does userspace app not do open+read+close each time it polls the badblocks file? Thanks, Miklos > > Signed-off-by: Tomasz Majchrzak <[email protected]> > Reviewed-by: Dan Williams <[email protected]> > --- > fs/seq_file.c | 5 +++-- > 1 file changed, 3 insertions(+), 2 deletions(-) > diff --git a/fs/seq_file.c b/fs/seq_file.c > index 6dc4296..74197f4 100644 > --- a/fs/seq_file.c > +++ b/fs/seq_file.c > @@ -235,7 +235,7 @@ ssize_t seq_read(struct file *file, char __user *buf, > size_t size, loff_t *ppos) > p = m->op->start(m, &pos); > while (1) { > err = PTR_ERR(p); > - if (!p || IS_ERR(p)) > + if (IS_ERR_OR_NULL(p)) > break; > err = m->op->show(m, p); > if (err < 0) > @@ -244,7 +244,8 @@ ssize_t seq_read(struct file *file, char __user *buf, > size_t size, loff_t *ppos) > m->count = 0; > if (unlikely(!m->count)) { > p = m->op->next(m, p, &pos); > - m->index = pos; > + if (!IS_ERR_OR_NULL(p)) > + m->index = pos; > continue; > } > if (m->count < m->size) > -- > 1.8.3.1 >
