[PATCH] seq_file: don't set read position for invalid iterator

[PATCH] seq_file: don't set read position for invalid iterator

[Tomasz Majchrzak]

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.

Don't set read position if valid iterator has not been returned.

Signed-off-by: Tomasz Majchrzak <tomasz.majchrzak@intel.com>
---
 fs/seq_file.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/fs/seq_file.c b/fs/seq_file.c
index 19f532e..893db43 100644
--- a/fs/seq_file.c
+++ b/fs/seq_file.c
@@ -242,7 +242,8 @@ 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 (p && !IS_ERR(p))
+				m->index = pos;
 			continue;
 		}
 		if (m->count < m->size)
-- 
1.8.3.1

Re: [PATCH] seq_file: don't set read position for invalid iterator

[Dan Williams]

[ adding Al ]

On Tue, Aug 23, 2016 at 2:19 AM, Tomasz Majchrzak
<tomasz.majchrzak@intel.com> 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.
>
> Don't set read position if valid iterator has not been returned.

Can you say a bit about how you found this?  So we have an idea about
userspace tooling is broken by this bug.

>
> Signed-off-by: Tomasz Majchrzak <tomasz.majchrzak@intel.com>

I would add "Cc: <stable@vger.kernel.org>" to the next version of this patch.

>  fs/seq_file.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/fs/seq_file.c b/fs/seq_file.c
> index 19f532e..893db43 100644
> --- a/fs/seq_file.c
> +++ b/fs/seq_file.c
> @@ -242,7 +242,8 @@ 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 (p && !IS_ERR(p))

Lets change this to use the IS_ERR_OR_NULL() helper, and do the same
with the earlier if (!p || IS_ERR()) a few lines up in the loop.

Other than that, this looks good to me.

Reviewed-by: Dan Williams <dan.j.williams@intel.com>