Skip to contentSkip to conversation

USB: gadget: Fix CFI failure during usb config switch.

PATCH In review Jiantao Zhang 2 participants 2 messages Nov 7, 2022 USB

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 USB: gadget: Fix CFI failure during usb config switch.

This reverts commit 0a55187a1ec8c

In the process of switching USB config from rndis to other config, if function gadget->ops->pullup() return an error,it will inevitably cause a CFI failure(Linux version 5.10.43).

analysis as follows:
======================================================
(1) write /config/usb_gadget/g1/UDC "none" (init.usb.configfs.rc:2)

gether_disconnect+0x2c/0x1f8     (dev->port_usb = NULL)
rndis_disable+0x4c/0x74
composite_disconnect+0x74/0xb0
configfs_composite_disconnect+0x60/0x7c
usb_gadget_disconnect+0x70/0x124
usb_gadget_unregister_driver+0xc8/0x1d8
gadget_dev_desc_UDC_store+0xec/0x1e4

in function usb_gadget_disconnect(),gadget->udc->driver->disconnect()
will not be called when gadget->ops->pullup() return an error, therefore, pointer dev->port will not be set to NULL.

(2) rm /config/usb_gadget/g1/configs/b.1/f1 (init.usb.configfs.rc:8)

    (f1 -> ../../../../usb_gadget/g1/functions/rndis.gs4)
rndis_deregister+0x28/0x54
rndis_free+0x44/0x7c
usb_put_function+0x14/0x1c
config_usb_cfg_unlink+0xc4/0xe0
configfs_unlink+0x124/0x1c8
vfs_unlink+0x114/0x1dc

(3) rmdir /config/usb_gadget/g1/functions/rndis.gs4

    (init.usb.configfs.rc:11)

CFI failure (target: [<ffffff814bc20c00>] 0000000068f50078):

CPU: 2 PID: 1 Comm: init VIP: 00 Tainted: G   W  O   5.10.43 #1
Call trace:
 __cfi_slowpath+0x300/0x3bc
 rndis_signal_disconnect+0x1e0/0x204
 rndis_close+0x24/0x2c
 eth_stop+0xd0/0x234           (if dev->port_usb != NULL, call rndis_close)
 __dev_close_many+0x204/0x2d4
 dev_close_many+0x48/0x2c8
 rollback_registered_many+0x184/0xdac
 unregister_netdevice_queue+0xf8/0x24c
 rndis_free_inst+0x78/0xc8
 rndis_attr_release+0x3c/0x84
 config_item_release+0x6c/0x180
 configfs_rmdir+0x7e0/0xca0

Since the rndis memory has been freed in step2, function rndis_close cannot be called here. In function eth_stop(), if pointer dev->port_usb is NULL, function rndis_close() will not be called. So, if gadget->ops->pullup() return an error in step1, a CFI failure will be
caused here.
======================================================
Through above analysis, i think gadget->udc->driver->disconnect() need to be called even if gadget->udc->driver->disconnect() return an error.

---
 drivers/usb/gadget/udc/core.c | 20 +++++++++++---------
 1 file changed, 11 insertions(+), 9 deletions(-)
drivers/usb/gadget/udc/core.c +0 -3
@@ -707,9 +707,6 @@ EXPORT_SYMBOL_GPL(usb_gadget_connect);
  * as a disconnect (when a VBUS session is active).  Not all systems
  * support software pullup controls.
  *
- * Following a successful disconnect, invoke the ->disconnect() callback
- * for the current gadget driver so that UDC drivers don't need to.
- *
  * Returns zero on success, else negative errno.
  */
 int usb_gadget_disconnect(struct usb_gadget *gadget) @@ -734,13 +731,8 @@ int usb_gadget_disconnect(struct usb_gadget *gadget)

}

ret = gadget->ops->pullup(gadget, 0);
- if (!ret) {
+ if (!ret)
gadget->connected = 0;
- mutex_lock(&udc_lock);
- if (gadget->udc->driver)
- gadget->udc->driver->disconnect(gadget);
- mutex_unlock(&udc_lock);
- }

out:
trace_usb_gadget_disconnect(gadget, ret); @@ -1532,6 +1524,11 @@ static void gadget_unbind_driver(struct device *dev) kobject_uevent(&udc->dev.kobj, KOBJ_CHANGE);

usb_gadget_disconnect(gadget);
+
+ mutex_lock(&udc_lock);
+ udc->driver->disconnect(udc->gadget);
+ mutex_unlock(&udc_lock);
+
usb_gadget_disable_async_callbacks(udc);
if (gadget->irq)
synchronize_irq(gadget->irq);
@@ -1626,6 +1623,11 @@ static ssize_t soft_connect_store(struct device *dev,
usb_gadget_connect(udc->gadget);
} else if (sysfs_streq(buf, "disconnect")) {
usb_gadget_disconnect(udc->gadget);
+
+ mutex_lock(&udc_lock);
+ udc->driver->disconnect(udc->gadget);
+ mutex_unlock(&udc_lock);
+
usb_gadget_udc_stop(udc);
} else {
dev_err(dev, "unsupported command '%s'\n", buf);
--
2.17.1

#1 lore raw
Quoted 1 line · Jiantao Zhang

This reverts commit 0a55187a1ec8c

Please show the full hash for this line.

Quoted 2 lines

In the process of switching USB config from rndis to other config, if function gadget->ops->pullup() return an error,it will inevitably cause a CFI failure(Linux version 5.10.43).

Please wrap the line properly.

And CFI is not in 5.10, so why is this an issue?

Quoted 45 lines

analysis as follows:
======================================================
(1) write /config/usb_gadget/g1/UDC "none" (init.usb.configfs.rc:2)

gether_disconnect+0x2c/0x1f8     (dev->port_usb = NULL)
rndis_disable+0x4c/0x74
composite_disconnect+0x74/0xb0
configfs_composite_disconnect+0x60/0x7c
usb_gadget_disconnect+0x70/0x124
usb_gadget_unregister_driver+0xc8/0x1d8
gadget_dev_desc_UDC_store+0xec/0x1e4

in function usb_gadget_disconnect(),gadget->udc->driver->disconnect()
will not be called when gadget->ops->pullup() return an error, therefore, pointer dev->port will not be set to NULL.

(2) rm /config/usb_gadget/g1/configs/b.1/f1 (init.usb.configfs.rc:8)

    (f1 -> ../../../../usb_gadget/g1/functions/rndis.gs4)
rndis_deregister+0x28/0x54
rndis_free+0x44/0x7c
usb_put_function+0x14/0x1c
config_usb_cfg_unlink+0xc4/0xe0
configfs_unlink+0x124/0x1c8
vfs_unlink+0x114/0x1dc

(3) rmdir /config/usb_gadget/g1/functions/rndis.gs4

    (init.usb.configfs.rc:11)

CFI failure (target: [<ffffff814bc20c00>] 0000000068f50078):

CPU: 2 PID: 1 Comm: init VIP: 00 Tainted: G   W  O   5.10.43 #1
Call trace:
 __cfi_slowpath+0x300/0x3bc
 rndis_signal_disconnect+0x1e0/0x204
 rndis_close+0x24/0x2c
 eth_stop+0xd0/0x234           (if dev->port_usb != NULL, call rndis_close)
 __dev_close_many+0x204/0x2d4
 dev_close_many+0x48/0x2c8
 rollback_registered_many+0x184/0xdac
 unregister_netdevice_queue+0xf8/0x24c
 rndis_free_inst+0x78/0xc8
 rndis_attr_release+0x3c/0x84
 config_item_release+0x6c/0x180
 configfs_rmdir+0x7e0/0xca0
Quoted 5 lines

Since the rndis memory has been freed in step2, function rndis_close cannot be called here. In function eth_stop(), if pointer dev->port_usb is NULL, function rndis_close() will not be called. So, if gadget->ops->pullup() return an error in step1, a CFI failure will be
caused here.
======================================================
Through above analysis, i think gadget->udc->driver->disconnect() need to be called even if gadget->udc->driver->disconnect() return an error.

I really do not understand, sorry. What does CFI have to do with anything here? What functions are mis-matched that CFI trips over it?

And by reverting this change, don't you now have the original problem that this commit was trying to solve?

Quoted 3 lines

No Fixes: tag?

No cc: stable?

thanks,

greg k-h