Hm, you're right; does this patch fix it? (Sorry for the top-posting,
folks)
The validate_byte check logic was backwards; it should return true for
an *invalid* packet. Thanks to Jeremy Katz for spotting this one.
Signed-off-by: Andres Salomon <[email protected]>
---
drivers/input/mouse/hgpk.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/input/mouse/hgpk.c b/drivers/input/mouse/hgpk.c
index e82d342..88f04bf 100644
--- a/drivers/input/mouse/hgpk.c
+++ b/drivers/input/mouse/hgpk.c
@@ -125,7 +125,7 @@ static void hgpk_spewing_hack(struct psmouse *psmouse,
*/
static int hgpk_validate_byte(unsigned char *packet)
{
- return (packet[0] & 0x0C) == 0x08;
+ return (packet[0] & 0x0C) != 0x08;
}
static void hgpk_process_packet(struct psmouse *psmouse)
--
1.5.6.5
On Mon, 10 Nov 2008 14:23:27 -0500
Jeremy Katz <[email protected]> wrote:
> I'm seeing it and someone else is across different XOs, so I don't
> think it's hardware. And I'm seeing it even if I don't touch the
> touchpad at all
>
> Looking through the diffs, the only thing that isn't certainly
> cosmetic is the attached and applying it fixes things. I haven't
> split to see which of the two chunks is the actual fix yet, but will
> probably try to do so later if I can avoid other fires popping up.
>
> Jeremy
>
> On Fri, 2008-11-07 at 03:36 -0800, Andres Salomon wrote:
> > Hi,
> >
> > Either you're seeing transient errors that are occurring due to
> > hardware problems (and thus have nothing to do with the updated
> > driver/kernel), or the miscalibration detection is being triggered
> > accidentally. Once the touchpad starts recalibrating, it will
> > continue until you've removed your fingers from the touchpad for a
> > few seconds.. so, if it starts doing that, just remove your hands
> > and within a few secs it should be fixed.
> >
> >
> > On Fri, 07 Nov 2008 16:26:05 -0500
> > Jeremy Katz <[email protected]> wrote:
> >
> > > Hey -- I backported what you sent upstream for the OLPC touchpad
> > > support to the Fedora 2.6.27.x kernel and while it identifies
> > > fine, I see boatloads of
> > > psmouse serio1: >100px jump detected (0,227)
> > > psmouse serio1: hgpk_process_byte: (1) 09 f9 08
> > > psmouse.c: HGPK at isa0060/serio1/input lost sync at byte 1
> > > psmouse.c: HGPK at isa0060/serio1/input0 - driver resynched.
> > > in dmesg and the mouse cursor just jumps around pretty much
> > > uncontrollably
> > >
> > > The only thing I'm seeing in the diff between olpc-2.6 and what's
> > > now upstream that might be relevant is the different conditions
> > > for hgpk_validate_byte(). But before I go grafting things
> > > together, I figured it was worth asking if you knew of anything
> > > specific to look at.
> > >
> > > Thanks,
> > >
> > > Jeremy
> > >
On Sun, Nov 09, 2008 at 07:40:57AM -0800, Andres Salomon wrote:
> Hm, you're right; does this patch fix it? (Sorry for the top-posting,
> folks)
>
>
>
>
> The validate_byte check logic was backwards; it should return true for
> an *invalid* packet. Thanks to Jeremy Katz for spotting this one.
>
> Signed-off-by: Andres Salomon <[email protected]>
Applied, thank you Andres.
--
Dmitry