Re: [ibm-acpi-devel] HKEY version 0x200

[Dennis W. <den...@se...>]

Hi,

On 13.04.2016 16:48, Henrique de Moraes Holschuh wrote:
> On Wed, Apr 13, 2016, at 09:28, Dennis Wassenberg wrote:
>> I would like to contribute to this topic as well.
> 
> Thanks!
> 
>> I believe (don't know, just guessing) that we should not make the
>> feature "has_adaptive_kbd" depend on hkeyv 0x200. Instead we should use
>> the "MHKA" method to evaluate if there is a flags mask set for the
>> adaptive keyboard. I assume that if we pass "1" to the MHKA method we
>> will get the hotkey_all_mask. If we pass "2" to the MHKA mask we will
>> get the hotkey_all_adaptive_mask. Thats my assumption which is based on
>> the observation of current Lenovo Skylake devices and the Thinkpad X1
>> Carbon (2014).
> 
> It might need to depend on both.  Are there any adaptive keyboards with
> HKEY 0x100?
> 
> If so, do they have the full ACPI interface for the adaptative keyboard,
> or do they like something, e.g. MHKA(0x2) ?

I don't know if there are any adaptive keyboards with HKEY 0x100. The
only adaptive keyboard which I know is the X1 Carbon 2014. As far as I
know this was the first HKEY 0x200 device. Thats why I assumed that the
HKEY 0x200 is only a necessary condition. If HKEY 0x200 is used it is
possible to query method MHKA with parameter 0x2. If there is something
set != 0 this would be the sufficient condition for adaptive keyboards.

But there might be other adaptive keybaords with HKEY 0x100, which I
don't know?

> 
> Because HKEY returning 0x200 should actually be the marker that the MHK*
> methods have changed to support the adaptive keyboard better...
> 
>> I checked it with T460, T460p, T460s, T560, X260 and X1 Tablet (2016).
>> All of these devices return 0 in that case. Only the X1 Carbon (2014)
>> will return 0x1FFFFFFF. The Thinkpads L460 and L560 are not covered by
>> this because they have hkeyv 0x100.
> 
> Seems like it would also work if we detect by HKEY 0x200 + MHKA(2) !=
> 0...
> 
Unfortunately, I don't know the correct interpretation of MHKA(2). HKEY
0x200 + MHKA(2) != 0 is what I used in the patch I send yesterday.

>> I prepared a patch which does what I described above (attachment). This
>> works fine in my case with kernel >= 4.3 until 4.6-rc3 for T460, T460p,
>> T460s, T560, X260. All hkeyv 0x100 Thinkpads should not be affected by
>> this change. Whats your opinion about this patch?
> 
> I will likely be able to only look at it in detail during the weekend.
> 
>> Testing this with Lenovo Thinkpad X1 Tablet (2016) will make the hotkeys
>> generate an event but currently there seems to be a mapping issue
>> because the most events are "KEY_UNKNOWN".
> 
> Ok.  This needs a fix, but I will only know which one after I have time
> to look over the current adaptive code and your patch in detail.
> 
> We likely need to extend the keymap.
>
Ok, I will let you know the results of my investigation regarding the
key mapping during the next days.

>> Additionally I observed that the new Skylake Thinkpads have no HKEY
>> methods GDFR, STRW, GTRW, SBMI, SVOI, GVOI, GCGI, GHOV, NALL, SAPI,
>> GAPI, SMNC, GMNC, SCMI, GCMI, SCMD, GCMD, GMMS, UAND, DGBL, SDHK, GDHK,
>> SDFL, GDFL, SDAM, GDAM in contrast to X1 Carbon (2014) with HKEY 0x200.
> 
> Can you send to me in private mail the full acpidump (i.e. every table)
> and dmidecode output for all these thinkpads you tested with?  Please
> cross-out any UUIDs and serial numbers...
>