Re: [PATCH] mac80211: fix race in ieee80211_register_hw()

[Krishna Chaitanya]

On Mon, Apr 6, 2020 at 7:31 PM Johannes Berg <johannes@xxxxxxxxxxxxxxxx> wrote:
>
> On Mon, 2020-04-06 at 19:25 +0530, Krishna Chaitanya wrote:
> > On Mon, Apr 6, 2020 at 6:57 PM Kalle Valo <kvalo@xxxxxxxxxxxxxx> wrote:
> > > Sumit Garg <sumit.garg@xxxxxxxxxx> writes:
> > >
> > > > On Mon, 6 Apr 2020 at 18:38, Johannes Berg <johannes@xxxxxxxxxxxxxxxx> wrote:
> > > > > On Mon, 2020-04-06 at 16:04 +0300, Kalle Valo wrote:
> > > > > > Johannes Berg <johannes@xxxxxxxxxxxxxxxx> writes:
> > > > > >
> > > > > > > On Mon, 2020-04-06 at 15:52 +0300, Kalle Valo wrote:
> > > > > > > > Johannes Berg <johannes@xxxxxxxxxxxxxxxx> writes:
> > > > > > > >
> > > > > > > > > On Mon, 2020-04-06 at 15:44 +0300, Kalle Valo wrote:
> > > > > > > > > > >     user-space  ieee80211_register_hw()  RX IRQ
> > > > > > > > > > >     +++++++++++++++++++++++++++++++++++++++++++++
> > > > > > > > > > >        |                    |             |
> > > > > > > > > > >        |<---wlan0---wiphy_register()      |
> > > > > > > > > > >        |----start wlan0---->|             |
> > > > > > > > > > >        |                    |<---IRQ---(RX packet)
> > > > > > > > > > >        |              Kernel crash        |
> > > > > > > > > > >        |              due to unallocated  |
> > > > > > > > > > >        |              workqueue.          |
> > > > > > > > >
> > > > > > > > > [snip]
> > > > > > > > >
> > > > > > > > > > I have understood that no frames should be received until mac80211 calls
> > > > > > > > > > struct ieee80211_ops::start:
> > > > > > > > > >
> > > > > > > > > >  * @start: Called before the first netdevice attached to the hardware
> > > > > > > > > >  *         is enabled. This should turn on the hardware and must turn on
> > > > > > > > > >  *         frame reception (for possibly enabled monitor interfaces.)
> > > > > > > > >
> > > > > > > > > True, but I think he's saying that you can actually add and configure an
> > > > > > > > > interface as soon as the wiphy is registered?
> > > > > > > >
> > > > > > > > With '<---IRQ---(RX packet)' I assumed wcn36xx is delivering a frame to
> > > > > > > > mac80211 using ieee80211_rx(), but of course I'm just guessing here.
> > > > > > >
> > > > > > > Yeah, but that could be legitimate?
> > > > > >
> > > > > > Ah, I misunderstood then. The way I have understood is that no rx frames
> > > > > > should be delivered (= calling ieee80211_rx()_ before start() is called,
> > > > > > but if that's not the case please ignore me :)
> > > > >
> > > > > No no, that _is_ the case. But I think the "start wlan0" could end up
> > > > > calling it?
>
> > I am still confused, without ieee80211_if_add how can the userspace
> > bring up the interface?
>
> It can add its own interface. Maybe that won't be 'wlan0' but something
> else?
>
> like
>
> iw phy0 interface add wlan0 type station
> ip link set wlan0 up
Ah okay, got it, thanks. Very narrow window though :-) as the
alloc_ordered_workqueue
doesn't need RTNL and there is a long way to go to do if_add() from
user and setup
the driver for interrupts. Again depends on the driver though, it
should properly handle
pending ieee80211_register_hw() with start().