Commit 8a12f8836145 ("net: hso: fix null-ptr-deref during tty device
unregistration") fixed the racy minor allocation reported by syzbot, but
introduced an unconditional NULL-pointer dereference on every disconnect
instead.
Specifically, the serial device table must no longer be accessed after
the minor has been released by hso_serial_tty_unregister().
Fixes: 8a12f8836145 ("net: hso: fix null-ptr-deref during tty device unregistration")
Cc: [email protected]
Cc: Anirudh Rayabharam <[email protected]>
Reported-by: Leonardo Antoniazzi <[email protected]>
Signed-off-by: Johan Hovold <[email protected]>
---
drivers/net/usb/hso.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/usb/hso.c b/drivers/net/usb/hso.c
index 9bc58e64b5b7..3ef4b2841402 100644
--- a/drivers/net/usb/hso.c
+++ b/drivers/net/usb/hso.c
@@ -3104,7 +3104,7 @@ static void hso_free_interface(struct usb_interface *interface)
cancel_work_sync(&serial_table[i]->async_put_intf);
cancel_work_sync(&serial_table[i]->async_get_intf);
hso_serial_tty_unregister(serial);
- kref_put(&serial_table[i]->ref, hso_serial_ref_free);
+ kref_put(&serial->parent->ref, hso_serial_ref_free);
}
}
--
2.26.3
On Mon, Apr 26, 2021 at 11:29:11AM +0200, Leonardo Antoniazzi wrote:
> On Mon, 26 Apr 2021 10:11:49 +0200
> Johan Hovold <[email protected]> wrote:
>
> > Commit 8a12f8836145 ("net: hso: fix null-ptr-deref during tty device
> > unregistration") fixed the racy minor allocation reported by syzbot, but
> > introduced an unconditional NULL-pointer dereference on every disconnect
> > instead.
> >
> > Specifically, the serial device table must no longer be accessed after
> > the minor has been released by hso_serial_tty_unregister().
> >
> > Fixes: 8a12f8836145 ("net: hso: fix null-ptr-deref during tty device unregistration")
> > Cc: [email protected]
> > Cc: Anirudh Rayabharam <[email protected]>
> > Reported-by: Leonardo Antoniazzi <[email protected]>
> > Signed-off-by: Johan Hovold <[email protected]>
> > ---
> the patch fix the problem
Thanks for confirming.
Next time, please keep the CC list intact when replying so that everyone
sees that you've tested it.
Johan
On Mon, Apr 26, 2021 at 11:35:46AM +0200, Johan Hovold wrote:
> On Mon, Apr 26, 2021 at 11:29:11AM +0200, Leonardo Antoniazzi wrote:
> > On Mon, 26 Apr 2021 10:11:49 +0200
> > Johan Hovold <[email protected]> wrote:
> >
> > > Commit 8a12f8836145 ("net: hso: fix null-ptr-deref during tty device
> > > unregistration") fixed the racy minor allocation reported by syzbot, but
> > > introduced an unconditional NULL-pointer dereference on every disconnect
> > > instead.
> > >
> > > Specifically, the serial device table must no longer be accessed after
> > > the minor has been released by hso_serial_tty_unregister().
> > >
> > > Fixes: 8a12f8836145 ("net: hso: fix null-ptr-deref during tty device unregistration")
> > > Cc: [email protected]
> > > Cc: Anirudh Rayabharam <[email protected]>
> > > Reported-by: Leonardo Antoniazzi <[email protected]>
> > > Signed-off-by: Johan Hovold <[email protected]>
> > > ---
>
> > the patch fix the problem
>
> Thanks for confirming.
>
> Next time, please keep the CC list intact when replying so that everyone
> sees that you've tested it.
Wonderful. Johan, thanks for the quick fix.
netdev maintainers, mind if I take this fix through my tree to Linus
this week, or can you all get it to him before -rc1 through the
networking tree?
thanks,
greg k-h
On Mon, Apr 26, 2021 at 10:11:49AM +0200, Johan Hovold wrote:
> Commit 8a12f8836145 ("net: hso: fix null-ptr-deref during tty device
> unregistration") fixed the racy minor allocation reported by syzbot, but
> introduced an unconditional NULL-pointer dereference on every disconnect
> instead.
>
> Specifically, the serial device table must no longer be accessed after
> the minor has been released by hso_serial_tty_unregister().
>
> Fixes: 8a12f8836145 ("net: hso: fix null-ptr-deref during tty device unregistration")
> Cc: [email protected]
> Cc: Anirudh Rayabharam <[email protected]>
> Reported-by: Leonardo Antoniazzi <[email protected]>
> Signed-off-by: Johan Hovold <[email protected]>
> ---
> drivers/net/usb/hso.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/net/usb/hso.c b/drivers/net/usb/hso.c
> index 9bc58e64b5b7..3ef4b2841402 100644
> --- a/drivers/net/usb/hso.c
> +++ b/drivers/net/usb/hso.c
> @@ -3104,7 +3104,7 @@ static void hso_free_interface(struct usb_interface *interface)
> cancel_work_sync(&serial_table[i]->async_put_intf);
> cancel_work_sync(&serial_table[i]->async_get_intf);
> hso_serial_tty_unregister(serial);
> - kref_put(&serial_table[i]->ref, hso_serial_ref_free);
> + kref_put(&serial->parent->ref, hso_serial_ref_free);
> }
> }
Ah, my bad. Thanks Johan for the fix!
Reviewed-by: Anirudh Rayabharam <[email protected]>
- Anirudh.
Hello:
This patch was applied to netdev/net-next.git (refs/heads/master):
On Mon, 26 Apr 2021 10:11:49 +0200 you wrote:
> Commit 8a12f8836145 ("net: hso: fix null-ptr-deref during tty device
> unregistration") fixed the racy minor allocation reported by syzbot, but
> introduced an unconditional NULL-pointer dereference on every disconnect
> instead.
>
> Specifically, the serial device table must no longer be accessed after
> the minor has been released by hso_serial_tty_unregister().
>
> [...]
Here is the summary with links:
- net: hso: fix NULL-deref on disconnect regression
https://git.kernel.org/netdev/net-next/c/2ad5692db728
You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
From: Greg Kroah-Hartman <[email protected]>
Date: Mon, 26 Apr 2021 11:50:44 +0200
> netdev maintainers, mind if I take this fix through my tree to Linus
> this week, or can you all get it to him before -rc1 through the
> networking tree?
I tossed this into net-next so Linus will see it later this week.
Thanks.
On Mon, Apr 26, 2021 at 01:09:19PM -0700, David Miller wrote:
> From: Greg Kroah-Hartman <[email protected]>
> Date: Mon, 26 Apr 2021 11:50:44 +0200
>
> > netdev maintainers, mind if I take this fix through my tree to Linus
> > this week, or can you all get it to him before -rc1 through the
> > networking tree?
>
> I tossed this into net-next so Linus will see it later this week.
Wonderful, thanks!
greg k-h