linux-kernel - Re: [PATCH] usb: typec: altmodes/displayport: Update active state

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <Y8kMsw/wT35KN7VK@kuha.fi.intel.com>
Date:   Thu, 19 Jan 2023 11:26:11 +0200
From:   Heikki Krogerus <heikki.krogerus@...ux.intel.com>
To:     Prashant Malani <pmalani@...omium.org>
Cc:     linux-kernel@...r.kernel.org, linux-usb@...r.kernel.org,
        bleung@...omium.org, stable@...r.kernel.org
Subject: Re: [PATCH] usb: typec: altmodes/displayport: Update active state

Hi Prashant,

On Wed, Jan 18, 2023 at 10:26:21AM -0800, Prashant Malani wrote:
> Hi Heikki,
> 
> Thanks for reviewing the patch.
> 
> On Wed, Jan 18, 2023 at 1:39 AM Heikki Krogerus
> <heikki.krogerus@...ux.intel.com> wrote:
> >
> > On Wed, Jan 18, 2023 at 03:15:15AM +0000, Prashant Malani wrote:
> > > Update the altmode "active" state when we receive Acks for Enter and
> > > Exit Mode commands. Having the right state is necessary to change Pin
> > > Assignments using the 'pin_assignment" sysfs file.
> >
> > The idea was that the port drivers take care of this, not the altmode
> > drivers.
> 
> For the port's typec_altmode struct, that makes sense.
> Should the port driver be taking care of the state for the partner's altmode
> too, i.e "/sys/class/typec/port1-partner/port1-partner.0/active" ?
> 
> It seemed like the port driver should be forwarding the VDMs without snooping
> the header, or IOW, it should let the altmode driver parse the VDMs (which it's
> doing in this case) and manage the partner altmode state.
> 
> "pin_assignment_store" seems to only work if the partner's altmode
> "active" bit is set to active [1]
> 
> FWIW, I think we can make the typec_altmode_update_active() calls from
> our (cros-ec-typec) port driver too, but displayport.c is parsing the header
> anyway, so it seemed repetitive. Just wanted to clarify the intention here.

The alt modes may have been entered even if there are no drivers for
them, if for example the PD controller handles the mode entry. In
those cases the port driver needs to update the active state of the
partner alt mode.

Since the port drivers have to handle that in some cases, for the sake
of consistency I thought that they might as well take care of it in
every case.

On the other hand, it should be safe to do it in both the port driver
and the altmode driver.

If you prefer that the altmode drivers always do this, I'm not against
it. But in that case could you patch tcpm.c while at it - in the same
series:

diff --git a/drivers/usb/typec/tcpm/tcpm.c b/drivers/usb/typec/tcpm/tcpm.c
index 904c7b4ce2f0c..0f5a9d4db105a 100644
--- a/drivers/usb/typec/tcpm/tcpm.c
+++ b/drivers/usb/typec/tcpm/tcpm.c
@@ -1693,14 +1693,11 @@ static int tcpm_pd_svdm(struct tcpm_port *port, struct typec_altmode *adev,
                        }
                        break;
                case CMD_ENTER_MODE:
-                       if (adev && pdev) {
-                               typec_altmode_update_active(pdev, true);
+                       if (adev && pdev)
                                *adev_action = ADEV_QUEUE_VDM_SEND_EXIT_MODE_ON_FAIL;
-                       }
                        return 0;
                case CMD_EXIT_MODE:
                        if (adev && pdev) {
-                               typec_altmode_update_active(pdev, false);
                                /* Back to USB Operation */
                                *adev_action = ADEV_NOTIFY_USB_AND_QUEUE_VDM;
                                return 0;

That's the only driver that will definitely always requires the
altmode drivers, so perhaps it would be good to drop the calls
from it at the same time.

thanks,

-- 
heikki