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  PHC 
Open Source and information security mailing list archives
 
Hash Suite for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:   Mon, 5 Oct 2020 17:44:48 +0200
From:   Marcel Holtmann <marcel@...tmann.org>
To:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Cc:     Johan Hedberg <johan.hedberg@...il.com>,
        Sathish Narsimman <sathish.narasimman@...el.com>,
        "David S. Miller" <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>,
        linux-bluetooth <linux-bluetooth@...r.kernel.org>,
        "open list:NETWORKING [GENERAL]" <netdev@...r.kernel.org>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] Revert "Bluetooth: Update resolving list when updating
 whitelist"

Hi Greg,

>>>>>>>>> This reverts commit 0eee35bdfa3b472cc986ecc6ad76293fdcda59e2 as it
>>>>>>>>> breaks all bluetooth connections on my machine.
>>>>>>>>> 
>>>>>>>>> Cc: Marcel Holtmann <marcel@...tmann.org>
>>>>>>>>> Cc: Sathish Narsimman <sathish.narasimman@...el.com>
>>>>>>>>> Fixes: 0eee35bdfa3b ("Bluetooth: Update resolving list when updating whitelist")
>>>>>>>>> Signed-off-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
>>>>>>>>> ---
>>>>>>>>> net/bluetooth/hci_request.c | 41 ++-----------------------------------
>>>>>>>>> 1 file changed, 2 insertions(+), 39 deletions(-)
>>>>>>>>> 
>>>>>>>>> This has been bugging me for since 5.9-rc1, when all bluetooth devices
>>>>>>>>> stopped working on my desktop system.  I finally got the time to do
>>>>>>>>> bisection today, and it came down to this patch.  Reverting it on top of
>>>>>>>>> 5.9-rc7 restored bluetooth devices and now my input devices properly
>>>>>>>>> work.
>>>>>>>>> 
>>>>>>>>> As it's almost 5.9-final, any chance this can be merged now to fix the
>>>>>>>>> issue?
>>>>>>>> 
>>>>>>>> can you be specific what breaks since our guys and I also think the
>>>>>>>> ChromeOS guys have been testing these series of patches heavily.
>>>>>>> 
>>>>>>> My bluetooth trackball does not connect at all.  With this reverted, it
>>>>>>> all "just works".
>>>>>>> 
>>>>>>> Same I think for a Bluetooth headset, can check that again if you really
>>>>>>> need me to, but the trackball is reliable here.
>>>>>>> 
>>>>>>>> When you run btmon does it indicate any errors?
>>>>>>> 
>>>>>>> How do I run it and where are the errors displayed?
>>>>>> 
>>>>>> you can do btmon -w trace.log and just let it run like tcdpump.
>>>>> 
>>>>> Ok, attached.
>>>>> 
>>>>> The device is not connecting, and then I open the gnome bluetooth dialog
>>>>> and it scans for devices in the area, but does not connect to my
>>>>> existing devices at all.
>>>>> 
>>>>> Any ideas?
>>>> 
>>>> the trace file is from -rc7 or from -rc7 with this patch reverted?
>>>> 
>>>> I asked, because I see no hint that anything goes wrong. However I have a suspicion if you bisected it to this patch.
>>>> 
>>>> diff --git a/net/bluetooth/hci_request.c b/net/bluetooth/hci_request.c
>>>> index e0269192f2e5..94c0daa9f28d 100644
>>>> --- a/net/bluetooth/hci_request.c
>>>> +++ b/net/bluetooth/hci_request.c
>>>> @@ -732,7 +732,7 @@ static int add_to_white_list(struct hci_request *req,
>>>>               return -1;
>>>> 
>>>>       /* White list can not be used with RPAs */
>>>> -       if (!allow_rpa && !use_ll_privacy(hdev) &&
>>>> +       if (!allow_rpa &&
>>>>           hci_find_irk_by_addr(hdev, &params->addr, params->addr_type)) {
>>>>               return -1;
>>>>       }
>>>> @@ -812,7 +812,7 @@ static u8 update_white_list(struct hci_request *req)
>>>>               }
>>>> 
>>>>               /* White list can not be used with RPAs */
>>>> -               if (!allow_rpa && !use_ll_privacy(hdev) &&
>>>> +               if (!allow_rpa &&
>>>>                   hci_find_irk_by_addr(hdev, &b->bdaddr, b->bdaddr_type)) {
>>>>                       return 0x00;
>>>>               }
>>>> 
>>>> 
>>>> If you just do the above, does thing work for you again?
>>> 
>>> Corrupted white-space issues aside, yes, it works!
>> 
>> I just pasted it from a different terminal ;)
>> 
>>> I am running 5.9-rc8 with just this change on it and my tracball works
>>> just fine.
>>> 
>>>> My suspicion is that the use_ll_privacy check is the wrong one here. It only checks if hardware feature is available, not if it is also enabled.
>>> 
>>> How would one go about enabling such a hardware feature if they wanted
>>> to?  :)
>> 
>> I need to understand what is going wrong for you. I have a suspicion,
>> but first I need to understand what kind of device you have. I hope
>> the trace file is enough.
> 
> If you need any other information, just let me know, this is a USB
> Bluetooth controller from Intel:
> 
> 	$ lsusb | grep Blue
> 	Bus 009 Device 002: ID 8087:0029 Intel Corp. AX200 Bluetooth
> 
> And the output of usb-devices for it:
> 	T:  Bus=09 Lev=01 Prnt=01 Port=04 Cnt=01 Dev#=  2 Spd=12  MxCh= 0
> 	D:  Ver= 2.01 Cls=e0(wlcon) Sub=01 Prot=01 MxPS=64 #Cfgs=  1
> 	P:  Vendor=8087 ProdID=0029 Rev=00.01
> 	C:  #Ifs= 2 Cfg#= 1 Atr=e0 MxPwr=100mA
> 	I:  If#=0x0 Alt= 0 #EPs= 3 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
> 	I:  If#=0x1 Alt= 0 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb

I already figured out that it is one of our controllers. The trace file gives it away.

So my suspicion is that the device you want to connect to uses RPA (aka random addresses). And we added support for resolving them in the firmware. Your hardware does support that, but the host side is not fully utilizing it and thus your device is filtered out.

If I am not mistaken, then the use_ll_privacy() check in these two specific places need to be replaced with LL Privacy Enabled check. And then the allow_rpa condition will do its job as expected.

We can confirm this if you send me a trace with the patch applied.

Anyhow, if my suspicion is correct, then I just need to figure out on how we fix this for 5.9-rc8 and also net-next. We might have to have two different patches. Just carrying the revert forward is going to cause some complicated merging and a broken -next.

Regards

Marcel

Powered by blists - more mailing lists