On 10/05/2020 05:41, Tony Lindgren wrote:
> * Tony Lindgren <tony@???> [200509 20:15]:
>> * Pavel Machek <pavel@???> [200509 19:38]:
>>> 3) I can prepare patched kernel for droid 4, based on 5.7-rc. It will use latest Tony's
>>> patches, where available, and my/Tony's old hacks where not. Goal: enough support for
>>> calls/SMS/data.
>>
>> I put together initial droid4-pending-v5.7 branch yesterday but need to
>> sort out few more audio bugs with the current serdev-ngsm. Might be able
>> to push it out on Sunday hopefully.
>
> So I got things behaving again and made a brief test call and heard
> some feedback noise between the two phonese so it seems to work still :)
>
> Below are the initial v5.7 kernel based branches at [0][1][2][3][4].

Great!

> We now use the generic serdev-ngsm driver with the standard /dev/gsmtty*
> devices spun up by n_gsm instead of the custom /dev/motmdm* devices like
> we did earlier. This means all the commands must prefix the Motorola
> custom packet ID to the commands like:
>
> printf "U1234AT+CFUN?\r" > /dev/gsmtty1
>
> This is based on the comments on LKML. The advantage is that the kernel
> code is simpler and more generic, and we should be a few steps closer
> to getting merged to the mainline kernel.
>
> Merlijn, I'm not sure what to do with the ofono package dependencies
> while we're changing ofono over to use the standard interface?
>
> Should we just rename the old ofono hacks to something like ofono-motmdm
> until it's no longer needed with the v5.6 kernel? There's probably no
> reason to try to support both the old and new interfaces in ofono as the
> old interface won't get merged upstream for kernel or for ofono.

I think there is no reason to support the old code if we have something
that we want to use going forward, so I would build the latest ofono
with patches, and the kernel linked below, and push that to -devel. Once
we're happy with that, we can push that to non-devel as well.

> Also there are changes are needed for udev to set the permissions,
> I'm using these now, not sure if there's a better way for these
> individual gsmtty* entries:
>
> KERNEL=="gsmtty1", GROUP="dialout", MODE="0660"
> KERNEL=="gsmtty3", GROUP="dialout", MODE="0660"
> KERNEL=="gsmtty5", GROUP="dialout", MODE="0660"
> KERNEL=="gsmtty6", GROUP="dialout", MODE="0660"
> KERNEL=="gsmtty7", GROUP="dialout", MODE="0660"
> KERNEL=="gsmtty8", GROUP="dialout", MODE="0660"
> KERNEL=="gsmtty9", GROUP="dialout", MODE="0660"
> KERNEL=="gsmtty10", GROUP="dialout", MODE="0660"
> KERNEL=="gsmtty11", GROUP="dialout", MODE="0660"
> KERNEL=="gsmtty12", GROUP="dialout", MODE="0660"
>
> # legacy motmdm interface
> #SUBSYSTEMS=="motmdm", GROUP="dialout", MODE="0660"
>
> SUBSYSTEMS=="gnss", GROUP="root", MODE="0644"
>
> # Pavel's ofono branch still needs this I belive
> 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"

I guess we should just start adding these to our ofono package at this
point.

Cheers,
Merlijn