On Apr 01 2017 or thereabouts, Dmitry Torokhov wrote:
> On Fri, Mar 31, 2017 at 11:37:05AM +0200, Benjamin Tissoires wrote:
> > Hi Dmitry,
> > 
> > On Mar 10 2017 or thereabouts, Dmitry Torokhov wrote:
> > > From: Benjamin Tissoires <[email protected]>
> > > 
> > > This provides glue between PS/2 devices that enumerate the RMI4 devices
> > > and Elan touchpads to the RMI4 (or Elan) SMBus driver.
> > > 
> > > The SMBus devices keep their PS/2 connection alive. If the initialization
> > > process goes too far (psmouse_activate called), the device disconnects
> > > from the I2C bus and stays on the PS/2 bus, that is why we explicitly
> > > disable PS/2 device reporting (by calling psmouse_deactivate) before
> > > trying to register SMBus companion device.
> > > 
> > > The HID over I2C devices are enumerated through the ACPI DSDT, and
> > > their PS/2 device also exports the InterTouch bit in the extended
> > > capability 0x0C. However, the firmware keeps its I2C connection open
> > > even after going further in the PS/2 initialization. We don't need
> > > to take extra precautions with those device, especially because they
> > > block their PS/2 communication when HID over I2C is used.
> > > 
> > > Signed-off-by: Benjamin Tissoires <[email protected]>
> > > Signed-off-by: Dmitry Torokhov <[email protected]>
> > > ---
> > >  drivers/input/mouse/Kconfig         |   4 +
> > >  drivers/input/mouse/Makefile        |   2 +
> > >  drivers/input/mouse/psmouse-base.c  |  16 +-
> > >  drivers/input/mouse/psmouse-smbus.c | 291 
> > > ++++++++++++++++++++++++++++++++++++
> > >  drivers/input/mouse/psmouse.h       |  37 +++++
> > >  5 files changed, 348 insertions(+), 2 deletions(-)
> > >  create mode 100644 drivers/input/mouse/psmouse-smbus.c
> > > 
> > > diff --git a/drivers/input/mouse/Kconfig b/drivers/input/mouse/Kconfig
> > > index 096abb4ad5cd..87bde8a210c7 100644
> > > --- a/drivers/input/mouse/Kconfig
> > > +++ b/drivers/input/mouse/Kconfig
> > > @@ -171,6 +171,10 @@ config MOUSE_PS2_VMMOUSE
> > >  
> > >     If unsure, say N.
> > >  
> > > +config MOUSE_PS2_SMBUS
> > > + bool
> > > + depends on MOUSE_PS2
> > > +
> > >  config MOUSE_SERIAL
> > >   tristate "Serial mouse"
> > >   select SERIO
> > > diff --git a/drivers/input/mouse/Makefile b/drivers/input/mouse/Makefile
> > > index 6168b134937b..56bf0ad877c6 100644
> > > --- a/drivers/input/mouse/Makefile
> > > +++ b/drivers/input/mouse/Makefile
> > > @@ -39,6 +39,8 @@ psmouse-$(CONFIG_MOUSE_PS2_TOUCHKIT)    += 
> > > touchkit_ps2.o
> > >  psmouse-$(CONFIG_MOUSE_PS2_CYPRESS)      += cypress_ps2.o
> > >  psmouse-$(CONFIG_MOUSE_PS2_VMMOUSE)      += vmmouse.o
> > >  
> > > +psmouse-$(CONFIG_MOUSE_PS2_SMBUS)        += psmouse-smbus.o
> > > +
> > >  elan_i2c-objs := elan_i2c_core.o
> > >  elan_i2c-$(CONFIG_MOUSE_ELAN_I2C_I2C)    += elan_i2c_i2c.o
> > >  elan_i2c-$(CONFIG_MOUSE_ELAN_I2C_SMBUS)  += elan_i2c_smbus.o
> > > diff --git a/drivers/input/mouse/psmouse-base.c 
> > > b/drivers/input/mouse/psmouse-base.c
> > > index 40f09ce84f14..bdb48b2acc57 100644
> > > --- a/drivers/input/mouse/psmouse-base.c
> > > +++ b/drivers/input/mouse/psmouse-base.c
> > > @@ -1996,16 +1996,27 @@ static int __init psmouse_init(void)
> > >   synaptics_module_init();
> > >   hgpk_module_init();
> > >  
> > > + err = psmouse_smbus_module_init();
> > > + if (err)
> > > +         return err;
> > > +
> > >   kpsmoused_wq = alloc_ordered_workqueue("kpsmoused", 0);
> > >   if (!kpsmoused_wq) {
> > >           pr_err("failed to create kpsmoused workqueue\n");
> > > -         return -ENOMEM;
> > > +         err = -ENOMEM;
> > > +         goto err_smbus_exit;
> > >   }
> > >  
> > >   err = serio_register_driver(&psmouse_drv);
> > >   if (err)
> > > -         destroy_workqueue(kpsmoused_wq);
> > > +         goto err_destroy_wq;
> > >  
> > > + return 0;
> > > +
> > > +err_destroy_wq:
> > > + destroy_workqueue(kpsmoused_wq);
> > > +err_smbus_exit:
> > > + psmouse_smbus_module_exit();
> > >   return err;
> > >  }
> > >  
> > > @@ -2013,6 +2024,7 @@ static void __exit psmouse_exit(void)
> > >  {
> > >   serio_unregister_driver(&psmouse_drv);
> > >   destroy_workqueue(kpsmoused_wq);
> > > + psmouse_smbus_module_exit();
> > >  }
> > >  
> > >  module_init(psmouse_init);
> > > diff --git a/drivers/input/mouse/psmouse-smbus.c 
> > > b/drivers/input/mouse/psmouse-smbus.c
> > > new file mode 100644
> > > index 000000000000..cc716977fbf9
> > > --- /dev/null
> > > +++ b/drivers/input/mouse/psmouse-smbus.c
> > > @@ -0,0 +1,291 @@
> > > +/*
> > > + * Copyright (c) 2017 Red Hat, Inc
> > > + *
> > > + * This program is free software; you can redistribute it and/or modify 
> > > it
> > > + * under the terms of the GNU General Public License version 2 as 
> > > published by
> > > + * the Free Software Foundation.
> > > + */
> > > +
> > > +#define pr_fmt(fmt)              KBUILD_MODNAME ": " fmt
> > > +
> > > +#include <linux/kernel.h>
> > > +#include <linux/module.h>
> > > +#include <linux/libps2.h>
> > > +#include <linux/i2c.h>
> > > +#include <linux/serio.h>
> > > +#include <linux/slab.h>
> > > +#include <linux/workqueue.h>
> > > +#include "psmouse.h"
> > > +
> > > +struct psmouse_smbus_dev {
> > > + struct i2c_board_info board;
> > > + struct psmouse *psmouse;
> > > + struct i2c_client *client;
> > > + struct list_head node;
> > > + bool dead;
> > > +};
> > > +
> > > +static LIST_HEAD(psmouse_smbus_list);
> > > +static DEFINE_MUTEX(psmouse_smbus_mutex);
> > > +
> > > +static void psmouse_smbus_check_adapter(struct i2c_adapter *adapter)
> > > +{
> > > + struct psmouse_smbus_dev *smbdev;
> > > +
> > > + if (!i2c_check_functionality(adapter, I2C_FUNC_SMBUS_HOST_NOTIFY))
> > > +         return;
> > > +
> > > + mutex_lock(&psmouse_smbus_mutex);
> > > +
> > > + list_for_each_entry(smbdev, &psmouse_smbus_list, node) {
> > > +         if (smbdev->dead)
> > > +                 continue;
> > > +
> > > +         if (smbdev->client)
> > > +                 continue;
> > > +
> > > +         if (i2c_probe_func_quick_read(adapter, smbdev->board.addr) < 0)
> > > +                 continue;
> > > +
> > > +         /* Device seems to be there, let's try switching over */
> > > +         psmouse_dbg(smbdev->psmouse,
> > > +                     "SMBus companion appeared, triggering rescan\n");
> > > +         serio_rescan(smbdev->psmouse->ps2dev.serio);
> > > + }
> > > +
> > > + mutex_unlock(&psmouse_smbus_mutex);
> > > +}
> > > +
> > > +static void psmouse_smbus_detach_i2c_client(struct i2c_client *client)
> > > +{
> > > + struct psmouse_smbus_dev *smbdev;
> > > +
> > > + mutex_lock(&psmouse_smbus_mutex);
> > > +
> > > + list_for_each_entry(smbdev, &psmouse_smbus_list, node) {
> > > +         if (smbdev->client == client) {
> > > +                 psmouse_dbg(smbdev->psmouse,
> > > +                             "Marking SMBus companion %s as gone\n",
> > > +                             dev_name(&smbdev->client->dev));
> > > +                 smbdev->client = NULL;
> > > +                 smbdev->dead = true;
> > > +                 serio_rescan(smbdev->psmouse->ps2dev.serio);
> > > +         }
> > > + }
> > > +
> > > + kfree(client->dev.platform_data);
> > > + client->dev.platform_data = NULL;
> > 
> > I realized (thanks to an oops) that this is a little bit too much :)
> > 
> > We are freeing whatever platform_data for *all* I2C devices :/
> 
> Doh!
> 
> > 
> > I would say the following patch should be working, but I haven't spent too
> > much time on it yet:
> > 
> > ---
> > 
> > diff --git a/drivers/input/mouse/psmouse-smbus.c 
> > b/drivers/input/mouse/psmouse-smbus.c
> > index 061c1cc..b77263e 100644
> > --- a/drivers/input/mouse/psmouse-smbus.c
> > +++ b/drivers/input/mouse/psmouse-smbus.c
> > @@ -61,24 +61,30 @@ static void psmouse_smbus_check_adapter(struct 
> > i2c_adapter *adapter)
> >  
> >  static void psmouse_smbus_detach_i2c_client(struct i2c_client *client)
> >  {
> > -   struct psmouse_smbus_dev *smbdev;
> > +   struct psmouse_smbus_dev *smbdev, *tmp;
> >  
> >     mutex_lock(&psmouse_smbus_mutex);
> >  
> > -   list_for_each_entry(smbdev, &psmouse_smbus_list, node) {
> > +   list_for_each_entry_safe(smbdev, tmp, &psmouse_smbus_list, node) {
> >             if (smbdev->client == client) {
> >                     psmouse_dbg(smbdev->psmouse,
> >                                 "Marking SMBus companion %s as gone\n",
> >                                 dev_name(&smbdev->client->dev));
> >                     smbdev->client = NULL;
> >                     smbdev->dead = true;
> > +
> > +                   kfree(client->dev.platform_data);
> > +                   client->dev.platform_data = NULL;
> 
> The client may continue using platform data until it dies, so we can't
> do it here.
> 
> > +
> > +                   mutex_lock(&psmouse_smbus_mutex);
> > +                   list_del(&smbdev->node);
> > +                   mutex_unlock(&psmouse_smbus_mutex);
> 
> There is double-lock right here.

Ouch

> 
> > +
> > +                   kfree(smbdev);
> >                     serio_rescan(smbdev->psmouse->ps2dev.serio);
> 
> And use-after-free...

Doh!

Clearly not tested enough :)

Thanks for pushing out a better patch.

Cheers,
Benjamin

> 
> >             }
> >     }
> >  
> > -   kfree(client->dev.platform_data);
> > -   client->dev.platform_data = NULL;
> > -
> >     mutex_unlock(&psmouse_smbus_mutex);
> >  }
> >  
> > @@ -161,17 +167,19 @@ static void psmouse_smbus_disconnect(struct psmouse 
> > *psmouse)
> >  {
> >     struct psmouse_smbus_dev *smbdev = psmouse->private;
> >  
> > -   mutex_lock(&psmouse_smbus_mutex);
> > -   list_del(&smbdev->node);
> > -   mutex_unlock(&psmouse_smbus_mutex);
> > -
> >     if (smbdev->client) {
> >             psmouse_dbg(smbdev->psmouse,
> >                         "posting removal request for SMBus companion %s\n",
> >                         dev_name(&smbdev->client->dev));
> >             psmouse_smbus_schedule_remove(smbdev->client);
> > +           psmouse->private = NULL;
> > +           return;
> >     }
> >  
> > +   mutex_lock(&psmouse_smbus_mutex);
> > +   list_del(&smbdev->node);
> > +   mutex_unlock(&psmouse_smbus_mutex);
> > +
> >     kfree(smbdev);
> >     psmouse->private = NULL;
> >  }
> > 
> > ---
> > 
> > I am just dumping my diff that seems to be OK here. Feel free to
> > integrate in a proper patch or wait next week for a more formal
> > submission from me.
> 
> I think we simply need a 2nd list with "dead" smbclient, let me see if I
> can write something.
> 
> Thanks.
> 
> -- 
> Dmitry

Reply via email to