Re: i'm back, port status

From: Michael Opdenacker <michael_at_free-electrons.com>
Date: Tue, 31 Aug 2004 21:31:08 +0200

Hi Szabolcs!

>
> Hi!
>
> I'm back from afrika.

You mean an unfortunate land with very rare access to the Internet (not
counting lots of worse problems) :-(

> I'm just checking out the last couple of thousand mails on the list ;)
>
> As I see on the summary we still missing some parts. I have now an
> offical
> project to make on h2210/linux. So it's my job now to participate in the
> project since I got payed for this :)

That's very cool! Congratulations :-)

>
> The parts I really intrested in to solve is the NAND write, LAB,
> battery management
> and calibration.

NAND write: just needs testing, using the write_enable parameter of
shamcop_nand. You may tried by raw writing to a safe area (one of your
files in iPAQ file store, for example, with special markers to identify
them).

LAB: Joshua

Reflashing: Joshua, Jamey (see Joshua's thread and explanations last week)

Battery management: No one at the moment (initial driver from Matthew)

Calibration: I took care of the linearization driver, but we also need
to check that calibration works fine with the newly created root
filesystems from OpenEmbedded.

> Who is dealing with these parts now? My goal is to run linux on h2200
> from its flash
> as the core OS.
>
> We planned to deal with the HTC bootloader, are we? What about that?

Not sure, see my attachment.

Actually, I made a status last week on achieved and remaining tasks,
with time left estimates for each task. The trouble is I'm horribly busy
preparing a training in 3 weeks: I can hardly read my mail, and I had no
time to update the Wiki with this status information (see below). If
someone can do it, that'd be great.

Here it is! Good luck everyone! I'll definitely be back in 4 weeks (with
free documentation training materials).

:-)

Michael.

Achieved
========

<>Linux booting (started in RAM from PocketPC), serial console
Display: LCD, backlight
Touchscreen
USB device
PCMCIA (CompactFlash slot)
Audio: playing with Alsa
LEDS
NAND flash access

Pending work with time estimates
====================================

(assuming 1 full-time person)

          Touchscreen calibration

*1 week* - The raw touchscreen driver works fine. However, need to build
and check the latest Opie and GPE binaries available through OpenEmbedded.

          DMA - MMC / SD

*3 weeks* - Internal ASIC very similar to the one in iPAQ H5xxx, which
is already supported. Will definitely help to implement the DMA and
MMC/SD drivers. Also need to implement improvements in the H5400 DMA
driver to achieve full 4bit speed (will get help from somebody who
already implemented a proprietary driver with these features).

          SDIO

*Difficult to assess*. There is very little SDIO support in Linux,
mainly because of a lack of public specifications. Would need to work on
a common Linux SDIO interface and then implement drivers for SDIO
devices as well. However, can be replaced by well supported CF (pcmcia)
devices, while SD is used for storage.

          Booting Linux from internal flash

*Up to 6 weeks.* Can be a difficult task because it hasn't been done yet
on any PDA with Linux 2.6. Options are either:

    *

      To keep the original bootloader from HTC, and replace the Windows
      part of the flash ROM by Linux. It is safe because as long as this
      bootloader is intact, the device can be reflashed. However, it can
      be quite complex as we'd have to understand (and perhaps
      disassemble the bootloader) to find out how it boots a system. In
      addition, as the bootloader doesn't understand jffs2 filesystems,
      we would probably need to implement a 2-stage boot.

    *

      To use the new LAB (Linux As a Bootloader) bootloader which is
      under development. This would result in a much simpler
      implementation. Unfortunately, replacing the existing bootloader
      by a faulty one would “brick” the device. We'd need to find out
      how to reflash the device through JTAG then (work has already
      started on this topic). Fortunately, LAB can be first booted from
      RAM as a regular kernel with HaRET
      Using LAB could bring lots of new options, such as booting from a
      storage card, dual booting with PocketPC...

          Access to battery and power information

*1 week*. Not very difficult to implement from specifications, but need
to port the Linux 2.4 driver to new 2.6 standards (no longer using
/proc/hal/battery).

          Power management

*2 weeks*. Need to fit existing drivers in the new Linux 2.6 power
management interface. Need to properly manage and test the Real Time
Clock as well as wake-up (including alarms) and suspend.

          Bluetooth

*2 weeks* - Chipset probably already supported by existing Linux drivers
(hci_uart?). However, work needed to manage on the leds user space
interface and probably to get full speed.

          Audio recording

*1 week* - Just not tested yet. May be working with the current
implementation. Otherwise, a few days learning about Alsa and pxa sound
may be needed (unless Giorgio takes care of this)

          Opie validation

*1 week*. Probably represents more work, but lot of testing and patching
will certainly be done in parallel by the community.

          GTK validation

*1 week*. Probably represents more work, but lot of testing and patching
will certainly be done in parallel by the community.

-- 
Michael Opdenacker
http://free-electrons.com
Received on Tue Aug 31 2004 - 15:51:08 EDT

This archive was generated by hypermail 2.2.0 : Mon Jul 25 2005 - 17:19:29 EDT