linux-kernel - Re: [PATCH] lis3lv02d: add axes knowledge of HP Pavilion dv5 models

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 for Android: free password hash cracker in your pocket

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

Message-ID: <49917650.1060208@tudelft.nl>
Date:	Tue, 10 Feb 2009 13:42:56 +0100
From:	Éric Piel <E.A.B.Piel@...elft.nl>
To:	Giuseppe Bilotta <giuseppe.bilotta@...il.com>
Cc:	Éric Piel <eric.piel@...mplin-utc.net>,
	Palatis Tseng <palatis@...il.com>,
	Pavel Machek <pavel@...e.cz>,
	LKML <linux-kernel@...r.kernel.org>,
	Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: [PATCH] lis3lv02d: add axes knowledge of HP Pavilion dv5 models

Giuseppe Bilotta schreef:
:
> 
> With my patch applied:
> 
> Left raised: 0 -> 32k
> Right raised: 0 -> -32k
> 
> Front raised: 1 -> -32k
> back raised: 1 -> 32k
> 
> Upside down: 2 -> 32k
> 
> So it seems that my model inverts the Z axis too, I'll have to fix
> that. However, this last test is tricky because unless the laptop is
> completely upside down the Z axis can go from -32k to +32k with just a
> couple of degrees (around the 90°)
> 
> I'll update my patch to invert the X and Z axis for my model.
Let's have a look at this once the "big noise" problem is fixed, because
for now it's rather impossible to deduce anything meaningful.

Andrew, could you drop the patch in the mean time?

> 
>> Palatis, could you let us know if you also have a lot of "noise" in the
>> values. That is, if when using jstest /dev/input/js0 you see big changes
>> in the values. Is it like on Giuseppe's laptop of around 8000?
>>
>> Moreover I'm wondering if this could be an error on LSB/MSB ordering.
>> Maybe looking at the raw values (cat
>> /sys/devices/platform/lis3lv02d/position) would give more insight...
> 
> amd64 architecture here. Laptop flat:
> 
> oblomov@...omov:~$ cat /sys/devices/platform/lis3lv02d/position
> (-256,256,14592)
> oblomov@...omov:~$ cat /sys/devices/platform/lis3lv02d/position
> (0,256,14848)
> oblomov@...omov:~$ cat /sys/devices/platform/lis3lv02d/position
> (-256,256,14848)
> oblomov@...omov:~$ cat /sys/devices/platform/lis3lv02d/position
> (-256,0,14336)
> oblomov@...omov:~$ cat /sys/devices/platform/lis3lv02d/position
> (-256,512,14336)
> oblomov@...omov:~$ cat /sys/devices/platform/lis3lv02d/position
> (256,256,14592)
> 
> Left edge raised:
> oblomov@...omov:~$ cat /sys/devices/platform/lis3lv02d/position
> (5120,512,13312)
> oblomov@...omov:~$ cat /sys/devices/platform/lis3lv02d/position
> (5632,512,13312)
> 
> You might be onto something.
Yes, beautiful! All are multiples of 2⁸, so much a sign of MSB/LSB
inversion! And the good news is that the device just happens to have a
register to set the endianess: CTRL2/BLE . For now, in the driver, we
expect the device to be little endian (which is the default according to
the manual). So at initialization we could force the endianess and see
if that fix the problem.

Can you try something like this in lis3lv02d_poweron():
	adev.read(handle, CTRL_REG2, &val);
	val |= CTRL2_BDU | CTRL2_IEN;
+	val &= ~CTRL2_BLE;
	adev.write(handle, CTRL_REG2, val);

Eric



View attachment "E_A_B_Piel.vcf" of type "text/x-vcard" (380 bytes)