linux-kernel - Re: [PATCH 5/6] Input: psmouse - factor out common protocol probing code

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: <20151202191839.GD15531@dtor-ws>
Date:	Wed, 2 Dec 2015 11:18:39 -0800
From:	Dmitry Torokhov <dmitry.torokhov@...il.com>
To:	Hans de Goede <hdegoede@...hat.com>
Cc:	linux-input@...r.kernel.org,
	Benjamin Tissoires <benjamin.tissoires@...hat.com>,
	Thomas Hellstrom <thellstrom@...are.com>, pali.rohar@...il.com,
	jckeerthan@...il.com, till2.schaefer@...-dortmund.de,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH 5/6] Input: psmouse - factor out common protocol probing
 code

Hi Hans,

On Wed, Dec 02, 2015 at 04:20:48PM +0100, Hans de Goede wrote:
> Hi,
> 
> Thanks for splitting out the series, patches 1 - 4 look good to me and are:
> 
> Reviewed-by: Hans de Goede <hdegoede@...hat.com>
> 
> I've some comments inline for this one.

Thanks for spending time on reviewing this.

> >@@ -1025,68 +1053,48 @@ static int psmouse_extensions(struct psmouse *psmouse,
> >  	 * Trackpads.
> >  	 */
> >  	if (max_proto > PSMOUSE_IMEX &&
> >-			cypress_detect(psmouse, set_properties) == 0) {
> >-		if (IS_ENABLED(CONFIG_MOUSE_PS2_CYPRESS)) {
> >-			if (cypress_init(psmouse) == 0)
> >-				return PSMOUSE_CYPRESS;
> >-
> >-			/*
> >-			 * Finger Sensing Pad probe upsets some modules of
> >-			 * Cypress Trackpad, must avoid Finger Sensing Pad
> >-			 * probe if Cypress Trackpad device detected.
> >-			 */
> >-			return PSMOUSE_PS2;
> >-		}
> >-
> >-		max_proto = PSMOUSE_IMEX;
> >+	    psmouse_try_protocol(psmouse, PSMOUSE_CYPRESS, &max_proto,
> >+				 set_properties, true)) {
> >+		return PSMOUSE_CYPRESS;
> >  	}
> 
> The protocols array has the CYPRESS entry wrapped in "#ifdef CONFIG_MOUSE_PS2_CYPRESS"
> so this bit of the patches changes behavior of the probe order if
> CONFIG_MOUSE_PS2_CYPRESS is not enabled. Before this patch the max_proto would
> get limited to PSMOUSE_IMEX, but now the:
> 
> 	proto = __psmouse_protocol_by_type(type);
> 	if (!proto)
> 		return false;
> 
> Part of psmouse_try_protocol will trigger and then max_proto will stay unmodified.
> 
> I think this can best be fixed by always including CYPRESS in the proto array,
> and have init be a stub which always fails when CYPRESS is not actually enabled.

I would not want to do that as that would make cypress be in the list of
supported protocols whereas it is actually compiled out.

BUT, there is actually no problem, the original code was misleading.
cypress_detect() is already a stub returning -ENOSYS when
CONFIG_MOUSE_PS2_CYPRESS is not enabled, so we would never went into
that "if" body in that case.

I'll make a patch cleaning it up, but won't repost the whole series so
as to not clutter the list.

> >
> >  	if (max_proto > PSMOUSE_IMEX) {
> >-		if (psmouse_do_detect(genius_detect,
> >-				      psmouse, set_properties) == 0)
> >+		if (psmouse_try_protocol(psmouse, PSMOUSE_GENPS,
> >+					 &max_proto, set_properties, true))
> >  			return PSMOUSE_GENPS;
> >
> >-		if (psmouse_do_detect(ps2pp_init,
> >-				      psmouse, set_properties) == 0)
> >+		if (psmouse_try_protocol(psmouse, PSMOUSE_PS2PP,
> >+					 &max_proto, set_properties, true))
> >  			return PSMOUSE_PS2PP;
> 
> The PS2PP entry in the protocols table has an init function, by passing
> in true here you are causing this to get called, where as it was not
> being called before.

No, it has detect function only (but called ps2pp_init), I'll post the
patch renaming it to ps2pp_detect.

> 
> p.s.
> 
> After this change, a whole lot of the code has the form of:
> 
> 	if (max_proto >= PSMOUSE_FOO &&
> 	    psmouse_try_protocol(psmouse, ...)) {
> 		return PSMOUSE_BAR;
> 	}
> 
> Maybe you can do a follow-up which makes PSMOUSE_FOO a parameter to
> psmouse_try_protocol and move the test into psmouse_try_protocol?

I considered it, but some protocols we detect unconditionally and some
conditionally, passing a parameter woudl somewhat hide it. Plus
try_protocol() already has a lot of parameters.

> 
> Also you may want to consider to drop the {} around the single return
> statement most of this code-blocks now have. Maybe best to do
> this in a followup patch to keep the diff readable.

I like to keep the braces if "if" condition is multi-line to better see
where condition ends and body starts.

Thanks.

-- 
Dmitry
--
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/