3.19 versus 4.0 dtb-rebuilder

Hi all.
Is dtb-rebuilder
git clone https://github.com/RobertCNelson/dtb-rebuilder

and
git checkout origin/3.19.x -b tmp

compatible with 4.0 kernel?
Kind Regards

The intial sync up of the 4.0.x branch is now out..

The biggest delta, ttyOx -> ttySx transition for the serial port..

Regards,

If your using "systemd", you'll never notice it, as systemd fixes up
the getty call for serial console..

If your not using systemd, then you need the udev rule installed by
"bb-customizations"...

https://github.com/rcn-ee/repos/blob/master/bb-customizations/suite/jessie/debian/60-omap-tty.rules

Regards,

Hi all.
Is dtb-rebuilder
git clone https://github.com/RobertCNelson/dtb-rebuilder
and
git checkout origin/3.19.x -b tmp
compatible with 4.0 kernel?

The intial sync up of the 4.0.x branch is now out..

The biggest delta, ttyOx -> ttySx transition for the serial port..

Why was this done, again. Didn't we go to ttyS before. This is bad
dejavu. Not the good kind.

Different driver.

There is this:

SERIAL_8250_OMAP_TTYO_FIXUP

but still have to make host changes

Maybe I'm missing something but if you want to boot into
different kernels this is a PITA.

The omap-serial driver (aka, ttyOx) still works. TBH, I'm not
sure why Robert switched to the omap_8250 driver so soon, because
the userspace breakage was to be expected.

Regards,
Peter Hurley

The omap-serial driver (aka, ttyOx) still works. TBH, I’m not
sure why Robert switched to the omap_8250 driver so soon, because
the userspace breakage was to be expected.

Because I think Robert loves breaking stuff . . .:wink:

Seriously, I must admit when stuff like this changes it annoys me. When someone else does all the work however . . . Yeah I just tend to deal with it. It’s either that, or do everything myself and spend far less time doing the things I want / need to be doing.

That said, 4.0 is still a work in progress yeah ? So whats all the worry about ?

The v3.8.x tree is still /dev/ttyOx :wink:

I was going to phase it in on mainline, but then ti made the switch on
their v3.14.x-ti tree...

So we bit the bullet.

For systemd users, systemd fixed it up on boot..

For udev users, we now having working /dev/ttySx -> /dev/ttyOx rule...

For everything else... We i didn't test.. :wink:

If it's any consolation, i did yell at the developers on linux-omap,
how "fxxxd" up this was, specially since i was here during the ttyS ->
ttyO mess.

But i think we are now ready in 26-ish+ kernel releases to move back
to ttyO... :wink:

Regards,

Hi all.
Is dtb-rebuilder
git clone https://github.com/RobertCNelson/dtb-rebuilder
and
git checkout origin/3.19.x -b tmp
compatible with 4.0 kernel?

The intial sync up of the 4.0.x branch is now out..

The biggest delta, ttyOx -> ttySx transition for the serial port..

Why was this done, again. Didn't we go to ttyS before. This is bad
dejavu. Not the good kind.

Different driver.

There is this:

SERIAL_8250_OMAP_TTYO_FIXUP

but still have to make host changes

Maybe I'm missing something but if you want to boot into
different kernels this is a PITA.

The omap-serial driver (aka, ttyOx) still works. TBH, I'm not
sure why Robert switched to the omap_8250 driver so soon, because
the userspace breakage was to be expected.

The v3.8.x tree is still /dev/ttyOx :wink:

Oh, it doesn't bother me; I reviewed the 8250_omap driver for
inclusion in mainline.

I was going to phase it in on mainline, but then ti made the switch on
their v3.14.x-ti tree...

I hate vendor trees.

So we bit the bullet.

For systemd users, systemd fixed it up on boot..

For udev users, we now having working /dev/ttySx -> /dev/ttyOx rule...

That works.

What systemd does is parse the list of consoles emitted by the tty core
to sysfs in /sys/class/tty/console/active and start getty on those.

For everything else... We i didn't test.. :wink:

If it's any consolation, i did yell at the developers on linux-omap,
how "fxxxd" up this was, specially since i was here during the ttyS ->
ttyO mess.

I'm sorry I missed your email.

But i think we are now ready in 26-ish+ kernel releases to move back
to ttyO... :wink:

Hopefully before then, we'll have a namespace solution that will
allow pain-free serial driver migrations (and allow multiple drivers
to share the ttyS namespace).

Regards,
Peter Hurley

Oh, it doesn't bother me; I reviewed the 8250_omap driver for
inclusion in mainline.

With all the generic improvements done to the base 8250 since the
omap_serial driver came out, it is a big improvement. But with v4.0.x
+ 8250_omap does has some issues with omap3/dm3730 thou... Haven't had
a chance to test v4.1-rc1 yet to see that's fixed with the xM,
hopefully give that a test tomorrow...

Regards,

There were a couple of issues during development that required
alternate DMA handling. Although it wasn't clear at the time, this may
have been
a driver bug. A set of 3 patches was recently posted to linux-serial which
I haven't had time to review or test, but it may be related.

In any event, please let us know your test results on linux-omap and
linux-serial
so that we can resolve outstanding driver issues.

Regards,
Peter Hurley