2020-12-15 15:18:23

by Gao Yan

Subject: [PATCH] net: remove disc_data_lock in ppp line discipline

tty layer provide tty->ldisc_sem lock to protect tty->disc_data;
For examlpe, when cpu A is running ppp_synctty_ioctl that
hold the tty->ldisc_sem, so if cpu B calls ppp_synctty_close,
it will wait until cpu A release tty->ldisc_sem. So I think it is
unnecessary to have the disc_data_lock;

cpu A cpu B
tty_ioctl tty_reopen
->hold tty->ldisc_sem ->hold tty->ldisc_sem(write), failed
->ld->ops->ioctl ->wait...
->release tty->ldisc_sem ->wait...OK,hold tty->ldisc_sem
->tty_ldisc_reinit
->tty_ldisc_close
->ld->ops->close

Signed-off-by: Gao Yan <[email protected]>
---
drivers/net/ppp/ppp_async.c | 5 -----
drivers/net/ppp/ppp_synctty.c | 5 -----
2 files changed, 10 deletions(-)

diff --git a/drivers/net/ppp/ppp_async.c b/drivers/net/ppp/ppp_async.c
index 29a0917a8..f8cb591d6 100644
--- a/drivers/net/ppp/ppp_async.c
+++ b/drivers/net/ppp/ppp_async.c
@@ -127,17 +127,14 @@ static const struct ppp_channel_ops async_ops = {
* FIXME: this is no longer true. The _close path for the ldisc is
* now guaranteed to be sane.
*/
-static DEFINE_RWLOCK(disc_data_lock);

static struct asyncppp *ap_get(struct tty_struct *tty)
{
struct asyncppp *ap;

- read_lock(&disc_data_lock);
ap = tty->disc_data;
if (ap != NULL)
refcount_inc(&ap->refcnt);
- read_unlock(&disc_data_lock);
return ap;
}

@@ -216,10 +213,8 @@ ppp_asynctty_close(struct tty_struct *tty)
{
struct asyncppp *ap;

- write_lock_irq(&disc_data_lock);
ap = tty->disc_data;
tty->disc_data = NULL;
- write_unlock_irq(&disc_data_lock);
if (!ap)
return;

diff --git a/drivers/net/ppp/ppp_synctty.c b/drivers/net/ppp/ppp_synctty.c
index 0f338752c..8cdf7268c 100644
--- a/drivers/net/ppp/ppp_synctty.c
+++ b/drivers/net/ppp/ppp_synctty.c
@@ -129,17 +129,14 @@ ppp_print_buffer (const char *name, const __u8 *buf, int count)
*
* FIXME: Fixed in tty_io nowadays.
*/
-static DEFINE_RWLOCK(disc_data_lock);

static struct syncppp *sp_get(struct tty_struct *tty)
{
struct syncppp *ap;

- read_lock(&disc_data_lock);
ap = tty->disc_data;
if (ap != NULL)
refcount_inc(&ap->refcnt);
- read_unlock(&disc_data_lock);
return ap;
}

@@ -215,10 +212,8 @@ ppp_sync_close(struct tty_struct *tty)
{
struct syncppp *ap;

- write_lock_irq(&disc_data_lock);
ap = tty->disc_data;
tty->disc_data = NULL;
- write_unlock_irq(&disc_data_lock);
if (!ap)
return;

--
2.17.1

2020-12-17 00:51:06

by Jakub Kicinski

Subject: Re: [PATCH] net: remove disc_data_lock in ppp line discipline

On Tue, 15 Dec 2020 23:00:54 +0800 Gao Yan wrote:
> tty layer provide tty->ldisc_sem lock to protect tty->disc_data;
> For examlpe, when cpu A is running ppp_synctty_ioctl that
> hold the tty->ldisc_sem, so if cpu B calls ppp_synctty_close,
> it will wait until cpu A release tty->ldisc_sem. So I think it is
> unnecessary to have the disc_data_lock;
>
> cpu A cpu B
> tty_ioctl tty_reopen
> ->hold tty->ldisc_sem ->hold tty->ldisc_sem(write), failed
> ->ld->ops->ioctl ->wait...
> ->release tty->ldisc_sem ->wait...OK,hold tty->ldisc_sem
> ->tty_ldisc_reinit
> ->tty_ldisc_close
> ->ld->ops->close
>
> Signed-off-by: Gao Yan <[email protected]>

# Form letter - net-next is closed

We have already sent the networking pull request for 5.11 and therefore
net-next is closed for new drivers, features, code refactoring and
optimizations. We are currently accepting bug fixes only.

Please repost when net-next reopens after 5.11-rc1 is cut.

Look out for the announcement on the mailing list or check:
http://vger.kernel.org/~davem/net-next.html

RFC patches sent for review only are obviously welcome at any time.