linux-kernel - Re: [PATCH] Input: psmouse - add resync_on

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

Message-ID: <ZcZ2oG1Rls-oR593@google.com>
Date: Fri, 9 Feb 2024 11:01:52 -0800
From: Dmitry Torokhov <dmitry.torokhov@...il.com>
To: Jonathan Denose <jdenose@...omium.org>
Cc: LKML <linux-kernel@...r.kernel.org>, jefferymiller@...gle.com,
	Jonathan Denose <jdenose@...gle.com>,
	Raul Rangel <rrangel@...omium.org>, linux-input@...r.kernel.org
Subject: Re: [PATCH] Input: psmouse - add resync_on_resume dmi check

On Wed, Feb 07, 2024 at 10:39:03AM -0600, Jonathan Denose wrote:
> Hi Dmitry,
> 
> Thanks for your reply.
> 
> On Tue, Feb 6, 2024 at 4:04 PM Dmitry Torokhov
> <dmitry.torokhov@...il.com> wrote:
> >
> > Hi Jonathan,
> >
> > On Thu, Nov 02, 2023 at 07:52:47AM -0500, Jonathan Denose wrote:
> > > Some elantech touchpads consistently fail after resuming from
> > > suspend at sanity_check in elantech_packet_check_v4. This means
> > > the touchpad is completely unusable after suspend resume.
> > >
> > > With different permutations of i8042 nomux, nopnp, reset, and noloop
> > > kernel options enabled, and with crc_enabled the touchpad fails in
> > > the same way.
> > >
> > > Resyncing the touchpad after receiving the
> > > PACKET_UNKNOWN/PSMOUSE_BAD_DATA return code allows the touchpad to
> > > function correctly on resume. The touchpad fails to reconnect with
> > > the serio reconnect no matter how many times it retries, so this
> > > change skips over that retry sequence and goes directly to resync.
> >
> > Why can't we do this in elantech_reconnect()? I am sure we can make it
> > simpler and more robust than what the generic handler is trying to do
> > with polling and everything.
> >
> > Thanks.
> >
> > --
> > Dmitry
> 
> I am fine with anything that would be simpler and more robust, though
> I'm not sure how to implement what you are describing.
> 
> Are you suggesting that in this PSMOUSE_BAD_DATA case, instead of
> using psmouse_set_state and psmouse_queue_work to call
> psmouse->reconnect (which calls elantech_reconnect)?

No. From the description it sounds like the device sends wrong/extra
data right after resume. I think you can handle it in
elantech_reconnect() method by draining the buffer or issuing poll
request or something similar.

Can you post the i8042 data stream that happens on suspend/resume?
Toggling i8042.debug option will cause the driver to dump the data to
dmesg.

Thanks.

-- 
Dmitry