linux-kernel - Re: [PATCH] HID: wacom: ask for a in-prox report when it was missed

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 linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20150304160037.GA25087@mail.corp.redhat.com>
Date:	Wed, 4 Mar 2015 11:00:37 -0500
From:	Benjamin Tissoires <benjamin.tissoires@...hat.com>
To:	Jason Gerecke <killertofu@...il.com>
Cc:	Jiri Kosina <jkosina@...e.cz>, Ping Cheng <pinglinux@...il.com>,
	linux-input@...r.kernel.org, linux-kernel@...r.kernel.org,
	Peter Hutterer <peter.hutterer@...-t.net>
Subject: Re: [PATCH] HID: wacom: ask for a in-prox report when it was missed

On Mar 03 2015 or thereabouts, Jason Gerecke wrote:
> On 3/3/2015 9:20 AM, Benjamin Tissoires wrote:
> >If noone listens to the input device when a tool comes in proximity,
> >the tablet does not send the in-prox event when a client becomes available.
> >That means that no events will be sent until the tool is taken out of
> >proximity.
> >
> >In this situation, ask for the report WACOM_REPORT_INTUOSREAD which will
> >read the corresponding feature and generate an in-prox event.
> >
> >We don't schedule this read while we are in an IO interrupt because we
> >know that usbhid will do it asynchronously. If this is triggered by uhid
> >then this is obviously a client side bug :)
> >
> >Signed-off-by: Benjamin Tissoires <benjamin.tissoires@...hat.com>
> Ping and I were both a little confused by this patch. Our first impression
> was that this wouldn't accomplish anything since the driver would see events
> from the tablet regardless of if a client were connected or not. Testing
> proved us wrong though, with the events clearly going to that great
> /dev/null in the sky. Is this behavior different from the USB and HID
> subsystems? IIRC, the prox code didn't use to behave like this...

I made further tests this morning, and I found out that one of my
patches adds a more consistent (though a defect) behavior:
b905811a49bcd ("HID: usbhid: prevent unwanted events to be sent when
re-opening")

Without this patch applied, and with an Intuos Pro, I can reproduce the
bug, but sometimes it doesn't trigger. If I let the tool down on the
sensor long enough when no client is listening, then the in-prox event
is not sent to the kernel. But if the time between the landing of the
tool and the open of the device node is small enough, the in-prox event
is seen.

So I installed a v3.16 kernel and can reproduce the same behavior
observed previously.

My analysis is:
- the HID subsystem uses PM in the same way wacom_sys.c used to do in
  v3.16
- v3.16 showed the bugs too, but we never noticed it
- with b905811a49bcd, the bug is more obvious and reliable given that we
  drain the first few milliseconds when we open the device (so we miss
  the in-prox event if it was sent)
- I can not reproduce the second bug mentioned in
  http://lists.freedesktop.org/archives/wayland-devel/2015-March/020361.html
  so it must be device dependent

I think we still need to fetch the current tool in case we receive
events while no in-prox event has been seen. This gives a more reliable
behavior and may also help in some other corner cases.

I think we also still need to keep b905811a49bcd because it should
prevent the second problem Peter observed. The out-of-prox event should
disappear with b905811a49bcd, but we need this patch to actually retrieve
the current tool given that we might have dropped the in-prox event.
 
> Under the assumption that USB events are indeed being dropped until a client
> is connected, then this patch looks fairly reasonable. It doesn't, however,
> seem to work reliably across tablets. An Intuos Pro reacted as I'd expect
> (with the patch in place, already-in-prox tools actually send events once
> e.g. evemu-record is run) but didn't seem to do anything for an Intuos5 or
> Intuos3 (nothing comes through evemu-record until I removed the tool from
> prox and then put it back in).

I'll test on such devices tomorrow. The spec says that the report ID 5
was there since Intuos 2 at least, so I guess there must be a way to
retrieve the info correctly on old devices too.

Cheers,
Benjamin

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/