:: Re: [maemo-leste] Droid 4 in 5.3-rc…
Top Page
Delete this message
Reply to this message
Author: Tony Lindgren
Date:  
To: Merlijn Wajer
CC: Pavel Machek, maemo-leste, Ivaylo Dimitrov, Sebastian Reichel
Subject: Re: [maemo-leste] Droid 4 in 5.3-rc1, development boards
Hi,

* Merlijn Wajer <merlijn@???> [190717 11:33]:
> On 15/07/2019 08:54, Tony Lindgren wrote:
> > * Merlijn Wajer <merlijn@???> [190714 00:48]:
> >> On 13/07/2019 21:47, Pavel Machek wrote:
> >>> 5.3-rc1 should be released next week, and it should be interesting
> >>> release for Droid 4. AFAICT it should be possible to get video to work
> >>> there with minimal changes.
> >>
> >> Great. Our current Droid4 kernels are based on branches prepared by
> >> Tony. If we can follow mainline without patches, that would ultimately
> >> be better.
> >
> > Yeah I'll do v5.2 based branch with the dss backported and with the
> > pending modem stuff. I'll be also working more on getting the n_gsm
> > modem support upstreamed, I think that's now the only set of pending
> > patches :)
>
> Cool, please let me know when you've done that. I'd be glad to get it in
> our repo.


OK just pushed out droid4-pending-v5.2 with all the drm stuff for v5.3
merged and the same old pending mdm patches merged in. Also included
is the usb option config and the serdev PM issue I noticed with no
serdev drivers loaded.

Here's also one new udev rule to idle the mdm usb, here's what I
currently have based on what Pavel listed earlier for ofono:

SUBSYSTEMS=="motmdm", GROUP="dialout", MODE="0664"
SUBSYSTEMS=="gnss", GROUP="root", MODE="0644"

# ofono needs these
KERNEL=="ttyUSB4", ENV{OFONO_DRIVER}="g1"
KERNEL=="motmdm1", ENV{OFONO_DRIVER}="motmdm"

# mdm6600 needs autoidle enabled, otherwise ohci will block deeper pm states for soc
ACTION=="add", SUBSYSTEM=="usb", ATTRS{idVendor}=="22b8", ATTRS{idProduct}=="2a70", TEST=="power/control" ATTR{power/control}="auto"

Oh and the new backlight module comes up blanked right now so it
needs to be manually enabled via sysfs.

> > FYI the powervr kernel driver glue layer just needs to call runtime PM
> > hopefully on droid4 to get the powervr initialized. Well I've only briefly
> > tested via /sys by changing power from auto to on and then dumping out
> > some registers:
> >
> > # echo on > /sys/devices/platform/44000000.ocp/5601fc00.target-module/power/control
> > # rwmem 0x5601fc00
> > 0x5601fc00 = 0000000000    # sgx revision register
> > # echo auto > /sys/devices/platform/44000000.ocp/5601fc00.target-module/power/control

> >
> > The same applies for the other omap variant socs, I'll try to post
> > similar dts patches for few other variants soonish.
>
> That would be great. I'm keen to trying to get PowerVR to work on the
> Droid4 or Pandaboard ES (figured it was similar to the droid4 but easier
> to play with, so got that board a while ago too), but I'm also trying to
> some userspace written...
>
> Should the revision register be zeros, though?


Hmm yeah looks like there might be two revision registers, one
for sgx and one for interconnect target module I guess:

# rwmem 0x56000024
0x56000024 = 0x00010200
# rwmem 0x5601fc00
0x5601fc00 = 0000000000

Regards,

Tony