On 27-3-2017 11:24, Arend Van Spriel wrote:
> + Johannes
>
> On 6-3-2017 11:48, Arend Van Spriel wrote:
>> + linux-wireless
>>
>> On 6-3-2017 8:04, Daniel J Blueman wrote:
>>> When resuming from suspend with a BCM43602 on Ubuntu 16.04 with
>>> 4.9.13, we see use after free [1].
>>>
>>> We see the struct cfg80211_ops is accessed in the resume path, after
>>> it was previously freed:
>>>
>>> (gdb) list *(brcmf_cfg80211_attach+0x10b)
>>> 0x1d77b is in brcmf_cfg80211_attach
>>> (drivers/net/wireless/broadcom/brcm80211/brcmfmac/cfg80211.c:6861).
>>> 6856 brcmf_err("ndev is invalid\n");
>>> 6857 return NULL;
>>> 6858 }
>>> 6859
>>> 6860 ops = kmemdup(&brcmf_cfg80211_ops, sizeof(*ops), GFP_KERNEL);
>>> 6861 if (!ops)
>>> 6862 return NULL;
>>> 6863
>>> 6864 ifp = netdev_priv(ndev);
>>> 6865 #ifdef CONFIG_PM
>>>
>>> (gdb) list *(wiphy_resume+0x591)
>>> 0xb751 is in wiphy_resume (net/wireless/sysfs.c:133).
>>> 128 int ret = 0;
>>> 129
>>> 130 /* Age scan results with time spent in suspend */
>>> 131 cfg80211_bss_age(rdev, get_seconds() - rdev->suspend_at);
>>> 132
>>> 133 if (rdev->ops->resume) {
>>> 134 rtnl_lock();
>>> 135 if (rdev->wiphy.registered)
>>> 136 ret = rdev_resume(rdev);
>>> 137 rtnl_unlock();
>>>
>>> I'm unsure if this relates to the ordering of callbacks processed by
>>> dpm_run_callback.
>>
>> The problem is that our driver can not access the device as it has been
>> powered off during suspend. So in the resume we cleanup everything
>> calling wiphy_unregister() and wiphy_free(). This means the rdev in
>> wiphy_resume() above is already freed. Not sure how to handle this
>> properly. Probably we should do a proper rebind.
>>
>> Regards,
>> Arend
>>
>>> Thanks,
>>> Daniel
>>>
>>> -- [1]
>>>
>>> BUG: KASAN: use-after-free in wiphy_resume+0x591/0x5a0 [cfg80211] at
>>> addr ffff8803fefebb30
>>> Read of size 8 by task kworker/u16:15/3066
>>> CPU: 0 PID: 3066 Comm: kworker/u16:15 Not tainted 4.9.13-debug+ #7
>>> Hardware name: Dell Inc. XPS 15 9550/0N7TVV, BIOS 1.2.19 12/22/2016
>>> Workqueue: events_unbound async_run_entry_fn
>>> ffff8803bffdf9d8 ffffffff880db6e1 ffff88042740ef00 ffff8803fefebb28
>>> ffff8803bffdfa00 ffffffff87a4d941 ffff8803bffdfa98 ffff8803fefebb20
>>> ffff88042740ef00 ffff8803bffdfa88 ffffffff87a4dbda ffff8803fb132360
>>> Call Trace:
>>> [<ffffffff880db6e1>] dump_stack+0x85/0xc4
>>> [<ffffffff87a4d941>] kasan_object_err+0x21/0x70
>>> [<ffffffff87a4dbda>] kasan_report_error+0x1fa/0x500
>>> [<ffffffff8768d47e>] ? trace_hardirqs_on_caller+0x3fe/0x580
>>> [<ffffffffc151783a>] ? cfg80211_bss_age+0x9a/0xc0 [cfg80211]
>>> [<ffffffff8768d60d>] ? trace_hardirqs_on+0xd/0x10
>>> [<ffffffffc14f31c0>] ? wiphy_suspend+0xc70/0xc70 [cfg80211]
>>> [<ffffffff87a4e091>] __asan_report_load8_noabort+0x61/0x70
>>> [<ffffffffc14f3100>] ? wiphy_suspend+0xbb0/0xc70 [cfg80211]
>>> [<ffffffffc14f3751>] ? wiphy_resume+0x591/0x5a0 [cfg80211]
>>> [<ffffffffc14f3751>] wiphy_resume+0x591/0x5a0 [cfg80211]
>>> [<ffffffffc14f31c0>] ? wiphy_suspend+0xc70/0xc70 [cfg80211]
>>> [<ffffffff885b220e>] dpm_run_callback+0x6e/0x4f0
>>> [<ffffffff885b3352>] device_resume+0x1c2/0x670
>>> [<ffffffff885b381d>] async_resume+0x1d/0x50
>>> [<ffffffff875ee9ee>] async_run_entry_fn+0xfe/0x610
>>> [<ffffffff875d0806>] process_one_work+0x716/0x1a50
>>> [<ffffffff875d0769>] ? process_one_work+0x679/0x1a50
>>> [<ffffffff88fd7bad>] ? _raw_spin_unlock_irq+0x3d/0x60
>>> [<ffffffff875d00f0>] ? pwq_dec_nr_in_flight+0x2b0/0x2b0
>>> [<ffffffff875d1c20>] worker_thread+0xe0/0x1460
>>> [<ffffffff875d1b40>] ? process_one_work+0x1a50/0x1a50
>>> [<ffffffff875e5662>] kthread+0x222/0x2e0
>>> [<ffffffff875e5440>] ? kthread_park+0x80/0x80
>>> [<ffffffff875e5440>] ? kthread_park+0x80/0x80
>>> [<ffffffff875e5440>] ? kthread_park+0x80/0x80
>>> [<ffffffff88fd86ea>] ret_from_fork+0x2a/0x40
>
> So the wiphy instance is unregistered *and* freed in our
> brcmf_pcie_resume() callback. However, the wiphy_resume() for this
> instance still comes right after that. I tried several fixes using API
> calls into driver core, but to no avail. So I came up with the fix
> below.
>
> Regards,
> Arend
> ---
resending with proper indentation (I hope).
---
diff --git a/net/wireless/core.c b/net/wireless/core.c
index d888613..3af9c2e 100644
--- a/net/wireless/core.c
+++ b/net/wireless/core.c
@@ -1096,6 +1096,17 @@ void cfg80211_stop_iface(struct wiphy *wiphy,
struct wireless_dev *wdev,
}
EXPORT_SYMBOL(cfg80211_stop_iface);
+bool cfg80211_dev_in_list_rtnl(struct cfg80211_registered_device *rdev)
+{
+ struct cfg80211_registered_device *item;
+
+ list_for_each_entry(item, &cfg80211_rdev_list, list) {
+ if (item == rdev)
+ return true;
+ }
+ return false;
+}
+
static int cfg80211_netdev_notifier_call(struct notifier_block *nb,
unsigned long state, void *ptr)
{
diff --git a/net/wireless/core.h b/net/wireless/core.h
index e9afbc7..0407841 100644
--- a/net/wireless/core.h
+++ b/net/wireless/core.h
@@ -286,6 +286,7 @@ struct cfg80211_cqm_config {
int cfg80211_dev_rename(struct cfg80211_registered_device *rdev,
char *newname);
+bool cfg80211_dev_in_list_rtnl(struct cfg80211_registered_device *rdev);
void ieee80211_set_bitrate_flags(struct wiphy *wiphy);
diff --git a/net/wireless/sysfs.c b/net/wireless/sysfs.c
index 16b6b59..e92459f 100644
--- a/net/wireless/sysfs.c
+++ b/net/wireless/sysfs.c
@@ -129,16 +129,19 @@ static int wiphy_resume(struct device *dev)
struct cfg80211_registered_device *rdev = dev_to_rdev(dev);
int ret = 0;
+ rtnl_lock();
+ if (!cfg80211_dev_in_list_rtnl(rdev))
+ goto done;
+
/* Age scan results with time spent in suspend */
cfg80211_bss_age(rdev, get_seconds() - rdev->suspend_at);
if (rdev->ops->resume) {
- rtnl_lock();
if (rdev->wiphy.registered)
ret = rdev_resume(rdev);
- rtnl_unlock();
}
-
+done:
+ rtnl_unlock();
return ret;
}
