[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: touchpad

On 1 June 2016 at 01:54, Geert Stappers <stappers@stappers.nl> wrote:
> On Tue, May 31, 2016 at 10:21:16PM -0400, MY wrote:
>> Dear Team,
>> I apologise for writing to the list,  as this is very minor,  but since
>> I had thought linux 4.5 would support touchpad ELAN1000:00 04F3:0401,
>>  I was very surprised to find a motionless pointer upon trying the live
>> non free firmware iso.   This being "unofficial" iso,  I then used the
>> official cd, as graphical install, with the same result.  Did not
>> complete install as I am drowning in work this week, but have seen this
>> with most current install media from Arch to Ubuntu.

I bet this can be fixed with a one-line patch to hid-core.c.  I'm not
sure if :0401 is in the version linux linux-4.5.x you have, but it's
definitely in the linux-4.4.11 I'm using.  Alternatively ELAN1000:00
04F3: is somehow not matching "{ HID_I2C_DEVICE(USB_VENDOR_ID_ELAN,"
so it might be a two line kernel patch.

Does the unofficial iso ship with both xserver-xorg-input-synaptics
1.8.3-2, and xserver-xorg-input-multitouch 1.0~rc2+git20110312-2?

Concerning *-input-multitouch, I just noticed that
xf86-input-multitouch v1.0-rc3 is the most recent upstream version,
from 2012.  I'm not sure if that driver is specific to Macbooks, but
I've filed a new version available bug.

>> Also,  and I have no idea why this would be true,  but, since
>> purchasing a bleeding edge laptop,  I have tried many distros with
>> kernels greater than 4.2 which would appear to support hardware  and
>> one behaviour is always the same,  and I am sure you are all aware of
>> it,  but I wish merely to re-iterate it:

Out of curiousity, does the touchpad work with the latest OpenSUSE
Tumbleweed (bleeding edge rolling release) LiveCD?  It's a great way
to check to see if a bleeding-edge laptop will work with a bleeding
edge software stack.  In particular, I'd use the GNOME one, because
I've found KDE5 to be buggy with Intel graphics.  If everything works,
then Debian can be made to work, with a little help ;-)  My experience
with Arch and Ubuntu is that they have, at most 3/5 to 4/5 of
OpenSUSE's out-of-the-box laptop support.


>> As to the touchpad,  the hid controller seems to prevent the
>> recognition of the actual physical touchpad as a real device,  and this
>> has been true in all live trials as well as my current ubuntu install.
>>  (I have made a kludge to type quickly....)

So the touchpad is detected by the kernel after all?  Was the
"motionless pointer" issue only a problem on the installation media?
I'm not sure if the following is still applicable for fully enabling
Elan touchpads, because I think the driver has been integrated into
the kernel, but these links are possibly worthwhile:


>> If there is anything I can send you to help,  please let me know,
> For starters name and shame the vendor that told
> you "Yes, it has full Linux support".
> That is based upon asking before buying.

Geert, if you're joking, the joke was above my head...If it works with
an out-of-tree driver, I think it's a grey area for what they can
claim :-p  On the other hand, if it's a Linux-specialised vendor like
Emperor Linux, or Zareason, I totally agree...but only if the
out-of-tree driver they integrated is flaky, or if they didn't bother
to integrate one into the installation they shipped the laptop with.
And then, only if they're not willing to fix the flakyness, because
that would be a breach of the TOS--the expectation being that paying a
premium for out-of-box functioning provides reliable out-of-box

> Consider returning the laptop.
> Make the problem a problem of the vendor.

If it works reliably with an out-of-tree driver, or reliably with a
driver in staging, maybe not?  Maybe I'm cynical, but given the
purportedly extremely low profit margins for all commodity hardware
(Apple and possibly Panasonic excluded), I would expect the vendor to
conclude "those demanding Linux users aren't worth it, frack 'em, they
can figure this out on their own."  IBM purportedly sold off both
their Thinkpad and Thinkserver lines due to the razor-thin profit

Now for braindead ACPI and/or UEFI firmware, yes, make a problem of
the vendor!  Also for firmware that enables DMA for the WLAN chipset
in such a way that it corrupts arbitrary segments of memory, vilify
them, and spread the word.  Or a cooling solution that is so utterly
dependent on OS drivers that taking a long phone call after entering
the BIOS will permanently damage the hardware?  Yeah in these cases, a
storm of bad PR is not only just, but maybe even a duty. ;-)  A storm
of bad PR is never just between the buyer and the vendor...

But for that grey-area of companies who are half-way-not-quite-there,
but don't have the resources to certify their hardware works out of
the box without quirks, and who can't afford a dedicated Linux team?
1) the quirks should be documented  2) the quirks should be reported
when found, both to the vendor and the community  3) the quirks should
be worked-around by anyone capable of working around them.

Incidentally, that's pretty much the way it works right now...could be
better of course, always a question of manpower and free time vs often
poor documentation.


Reply to: