ZipitForge - Trac

I would say to start from there. BTW, the MAC address of the Zipit (inside battery box) is NOT the MAC that will show up for the WIFI router. It IS a different MAC.

[11:11:12]

this

[11:11:14]

is good to know

[11:11:15]

<T0mW>

it looks like the wifi MAC is hardcoded into the gspxxx module

[11:11:28]

but doesn'T explain, why i cannot "iwconfig essid fnord"..

[11:11:42]

<T0mW>

yeah

[11:11:51]

which is my problem atm.. :/

[11:11:53]

<T0mW>

did you try an open access point?

[11:12:07]

not yet

[11:12:16]

<g1powermac_PB>

unrest: I never got wep working on mine

[11:12:26]

<T0mW>

start small, get that working.

[11:12:26]

tried to set up one of my notebooks wireless cards to ap mode, but it crashed my system ^^

[11:12:35]

<g1powermac_PB>

open access worked fine

[11:12:40]

<Limp_Trizkit>

same

[11:12:40]

<T0mW>

not sure if wpa-supplicant is in the current image or not.

[11:12:42]

madwifi bug afaik..

[11:12:52]

T0mW, it isn't but i already installed it

[11:13:14]

g1powermac, Limp_Trizkit: do you get the error bryguy[lap] and I get when trying to connect to a WEPed accespoint?

[11:13:24]

<T0mW>

MACHINE_ESSENTIAL_EXTRA_RDEPENDS

[11:13:33]

<g1powermac_PB>

unrest: whats the error again?

[11:13:35]

yeah, i know :)

[11:13:45]

<bryguy[lap]>

I'll be back later

[11:13:49]

<Limp_Trizkit>

unrest: i don't have a WEP-enabled AP

[11:13:51]

<g1powermac_PB>

unrest: oh yea, I got that

[11:13:57]

g1powermac, "wireless request "set essid" failed ; connection refused" or similar

[11:13:58]

<Limp_Trizkit>

i can do testing tonight

[11:14:04]

Limp_Trizkit, that'd be nice

[11:14:04]

<Limp_Trizkit>

i have to install wpa_supplicant first

[11:14:13]

<g1powermac_PB>

unrest: yup, yup, definitely got that with wep

[11:14:32]

<g1powermac_PB>

there's a bug somewhere causing issues

[11:14:34]

if you want to, mail me the results of the testing to: unrest@zee-aylienz.de please :)

[11:14:58]

<g1powermac_PB>

I also tried to use wpa_supplicant to do the wep stuff, that didn't work either

[11:14:59]

<Limp_Trizkit>

if i remember i will

[11:15:38]

g1powermac, which (imho) sucks ^^ i'll try to fix the issues with my driver..

[11:15:42]

Limp_Trizkit, thanks =)

[11:15:54]

...and then set up a nonencrypted "ap" for my z2

[11:16:12]

<T0mW>

I would like to get something put together to bring that USB port out to a TYPE-A connector. That way I can hook a USB ethernet to the zipit to do troubleshooting...

[11:16:22]

<g1powermac_PB>

btw, there has been work done on an OSS driver for the wifi module

[11:16:28]

drmikecrowe_, is zipitwireless intending to fix the driver issues?

[11:16:47]

<g1powermac_PB>

T0mW: only USB device is brought out

[11:16:53]

tom, i was told that it's usb __out__

[11:16:57]

<T0mW>

oh, no host eh?

[11:17:01]

<g1powermac_PB>

USB host or otg is not brought out anywhere on the board

[11:17:07]

<T0mW>

bah

[11:17:18]

but there's the serial pins on the board afaik

[11:17:23]

<g1powermac_PB>

yes

[11:17:28]

<T0mW>

then it will have to be an i2c uart then.

[11:17:30]

so we could get a serial interface...

[11:17:40]

<g1powermac_PB>

and you could do usb gadget ethernet

[11:17:54]

<T0mW>

g1powermac: yes

[11:18:05]

<T0mW>

gadget would work

[11:18:09]

<g1powermac_PB>

which was right where I left off

[11:18:27]

<g1powermac_PB>

cause you need to replace the kernel to get usb gadget working

[11:18:41]

<Limp_Trizkit>

afk

[11:18:42]

<g1powermac_PB>

and to do that well, you really need jtag/serial port

[11:18:54]

<T0mW>

ok, has anyone tried the JTAG yet?

[11:19:05]

<g1powermac_PB>

yes, GPSFan has got it working

[11:19:10]

<T0mW>

good

[11:19:31]

g1powermac, did he document his works someplace online?

[11:19:32]

<T0mW>

of course, Ken would do that while he was waiting for the grass to grow.

[11:19:50]

<T0mW>

GPSFan: nudge

[11:19:55]

<g1powermac_PB>

yea, he's going to if he hasn't already

[11:21:28]

<T0mW>

I did brick one unit by attempting to replace the kernel via the MTDBLOCK stuff. Did not realize it was an uncompressed kernel that was required.

[11:23:11]

g1powermac, who's developing such a driver?

[11:23:18]

<T0mW>

we should be able to use the MTD drivers to replace the kernel + bootloader. However, JTAG would be needed to recover a bricked unit.

[11:23:57]

<T0mW>

IIRC, gadget has been backported to 2.6.22?

[11:25:43]

<g1powermac_PB>

yea, thats what I was working on

[11:26:13]

<g1powermac_PB>

http://zipit2system.sourceforge.net/usbgadget.html

[11:34:14]

<T0mW>

well, time to go rip out more carpet.

[11:35:36]

<g1powermac_PB>

ahh, now thats fun stuff :-)

[11:36:22]

<T0mW`gone>

g1powermac_PB: yes, until I rip out the ceiling in the one bedroom to find out what is leaking in the upstairs bathroom.

[11:36:32]

<g1powermac_PB>

ahh

[11:36:45]

<g1powermac_PB>

one tip, make sure you wear a respirator

[11:36:50]

<T0mW`gone>

I have only one thing more to say: PEX

[11:37:01]

<g1powermac_PB>

there most likely is a fairly bad mold infestation

[11:37:18]

<g1powermac_PB>

and you really don't want to breathe that stuff in

[11:37:56]

<g1powermac_PB>

make sure its a respirator rated for mold spores too

[11:38:23]

<T0mW`gone>

g1powermac_PB: I have some surgical masks I purchased to use when I grind / cut fiberglass..

[11:38:42]

<T0mW`gone>

good point though, thanks

[11:39:00]

<g1powermac_PB>

yea, not sure if that'll be good enough

[11:39:14]

<g1powermac_PB>

we've been through a major mold problem in the past

[11:39:35]

<g1powermac_PB>

and have felt the effects of the stuff, its not fun

[11:40:33]

<g1powermac_PB>

we had bought a new house, but the crawl space foundation was leaking

[11:40:53]

<T0mW>

well, I will have to get something like scaffolding to get up there to work. Probably pick up a Little Giant ladder, or just sawhorses and some planking. That will happen a little later this week. Ripping out the floor covering is the job right now.

[11:41:06]

<g1powermac_PB>

we complained to the builder and he decided to spray cellulose insulation on it thinking it was going to fix the problem

[11:41:16]

<T0mW>

yeah, right

[11:41:48]

<g1powermac_PB>

well, that created the biggest stachybactrous (spelling is not right) infestation which completely destroyed the house

[11:41:48]

<T0mW>

g1powermac_PB: what was your final solution?

[11:42:49]

<g1powermac_PB>

we left the house and sued the builder

[11:42:52]

<g1powermac_PB>

the suit fell through as we just didn't have the money to continue and we eventually went bankrupt

[11:42:54]

<T0mW>

nod

[11:43:16]

<g1powermac_PB>

not a fun time of my life

[11:43:26]

<T0mW>

no, there goes your dream.

[11:44:06]

<g1powermac_PB>

we also lost alot of our possessions as the stuff needed to be decontaminated

[11:44:16]

<g1powermac_PB>

and we just couldn't get it done

[11:45:35]

<T0mW>

I had a friend who had rented a trailer and she had a massive problem with mold. It really took a physical toll on her: resperatory problems, hair falling out.

[11:45:43]

<g1powermac_PB>

yup

[11:46:01]

<g1powermac_PB>

it'll also play with your neurological system

[11:46:11]

<T0mW>

the walls and ceilings were infested with mold

[11:47:02]

<g1powermac_PB>

yea, not good

[11:47:28]

<T0mW>

I will need to do some landscape work and thin out some trees on the property to open the space up so that the house gets sunlight and can keep dry.

[11:47:41]

<g1powermac_PB>

yea, thats a good thing

[11:47:57]

<T0mW>

well, off to get some work done. later.

[17:58:46]

To get the z2shell pieces working, what filesystem needs to be on the card?

[18:02:56]

Interesting. Second time around it seems to work

[18:18:25]

Hm. Doesn't work on first boot, but works on reboot. Yuck.

[18:23:37]

<g1powermac_PB>

crazymonkeypants: was your zipit actually off the first time?

[18:24:02]

<g1powermac_PB>

it may look like its off though it isn't

[18:27:40]

I took the battery out

[18:27:46]

I'm fairly sure it was off

[18:27:53]

If I power down then back up quickly, it works

[18:28:08]

I suspect my $15 miniSD card is not getting along perfectly

[18:28:18]

I saw in the logs earlier that someone was having issues with WPA

[18:28:32]

It works fine for me talking to an AEBS

[18:29:45]

<g1powermac_PB>

hmm, could be with the sd card

[18:31:27]

It gives up very quickly; I could easily see the card taking a while to get its act together.

[18:31:56]

I got the serial console working, thanks muchly for the pictures

[18:32:02]

Trying to sort out JTAG now

[18:32:56]

Do you know if anyone has actually sourced the back-panel connector?

[18:34:47]

Oh, hah. It's a stock Hirose part

[18:34:53]

And Digikey have them.

[18:35:51]

<g1powermac_PB>

yea

[18:36:06]

<g1powermac_PB>

its all in the wiki in the topic

[18:37:50]

By "the wiki" you mean the Aeronix one?

[18:38:37]

<g1powermac_PB>

http://linux.zipitwireless.com

[18:39:25]

Right

[18:43:00]

Speaking of which, was there a reason you started your own site rather than expanding the existing wiki?

[18:43:13]

e.g. are there things that Aeronix don't want published there?

[18:43:32]

I was going to take a shot at decoding the GPIO pad config registers and making a table of what's where

[18:44:43]

<g1powermac_PB>

yup, that was me :-)

[18:45:36]

<g1powermac_PB>

http://zipit2system.sourceforge.net

[18:46:21]

I realised that from your tag at the bottom of the page 8)

[18:47:31]

<g1powermac_PB>

cool, if you wanna try that, that'd be great

[18:48:05]

<g1powermac_PB>

the zipit guys have made the pinout for the back connector available on their website but you have to agree to their terms to view it

[18:48:37]

<g1powermac_PB>

if someone wants to find it without contaminating themselves by agreeing to it, that'd be awesome

[18:49:39]

Too late for me, sadly

[18:49:57]

crazymonkeypants: the GPIO's are initially setup in blob, but after the zipit app runs, it re-defines some of them. you will need to use devmem2 or some such while the zipit app is running to figure them out.

[18:50:24]

That was what I had in mind

[18:50:53]

The pxa270 pad config bits are fairly straightforward

[18:51:15]

I note their gpio driver is a module, so I expect there's no source/documentation

[18:52:01]

I built a sd rootfs that booted up with dropbear setup so I could log in. it then ran the zipit app. there were a few tricks to get it to run properly once booted from the sd card.

[18:52:30]

I was just going to hack the startup script on the jffs

[18:54:14]

I don't see anything that looks like signatures or integrity checking, so it looks like it's open for business

[18:54:16]

you could do that too, but you need a way to interact with devmem2 since gpio's are used for things like charging the batt and all the led's

[18:54:55]

the z2 app won't run if the system is booted from sd card.

[18:55:30]

Mounts are wrong?

[18:56:02]

don't know for sure. but I had to trick it into thinking it was booted from the flash.

[18:56:27]

Interesting. Perhaps it's the fbcons stuff?

[18:56:42]

it's been a while since I did it, I'll have to re-educate myself on what I did... ;>)

[18:57:04]

8)

[18:58:18]

Found something more interesting in the meantime? 8)

[18:59:34]

no, just day to day stuff, I did that back in early march.

Main: i=104

Main: i=104

Main: i=104

Seems to run OK

Apart from the endless "Main: i=255" spew

[19:07:30]

And the cursor flashing in the lower right

[19:07:43]

^C kills it too

[19:08:57]

<g1powermac_PB>

are you running that from ssh?

[19:47:10]

serial console

[19:50:40]

weird

[19:50:40]

<g1powermac_PB>

ahh

[19:50:51]

There are actually two hooks for z2script.sh

[19:51:10]

one in rc.sysinit, then another in start.sh

[20:01:11]

crazymonkeypants: the serial console isn't set up in the z2's inittab (or what passes for it) so you need to start a getty on it manually

[20:05:58]

Er

[20:06:02]

Mine works fine, no changes

[20:12:28]

you can log in?

[20:12:31]

Nifty. They appear to be using sqlite to manage the mp3 list

[20:12:37]

I just get a shell, no questions asked

[20:12:45]

hmm.

[20:13:15]

I'll have to try mine out, IIRC all I got was the random messages like you showed above.

[20:13:24]

tsk tsk

[20:13:32]

cygwin... that's not very hardcore

[20:13:46]

/cygdrive/c/Projects/Zipit2/AerMedia/Core/Build/Linux/Linux-2.6.21_ArmV5L/Systems/../../../../Src/Systems/AerHeapManager.cpp

[20:14:05]

<g1powermac_PB>

heh

[20:14:29]

it's a windoze world don'ch know...

[20:15:39]

Holy toolchains batman...

[20:16:31]

GCC: (GNU) 3.4.1 20040714 (TimeSys 3.4.1-7)

[20:16:31]

GCC: (GNU) 3.4.1 20040714 (TimeSys 3.4.1-7)

[20:16:31]

GCC: (GNU) 3.3.1 (MontaVista 3.3.1-7.0.38.0600024 2006-01-08)

[20:16:31]

ARM C++ Compiler, ADS1.0.1 [Build 187]

[20:16:31]

ARM/Thumb Macro Assembler vsn ADS1.0.1 [Build 187]

[20:16:31]

ARM C Compiler, ADS1.0.1 [Build 187]

[20:16:47]

GCC: (GNU) 4.1.0 20060304 (TimeSys 4.1.0-3)

[20:16:57]

<g1powermac_PB>

interesting

[20:19:26]

nope, once my z2 is booted into the zipit app, the serial port just says Main: i=10 when I hit return.

[20:19:55]

Before you start the app, what about then?

[20:20:58]

If I boot with the miniSD installed, I have a terminal on the serial console

[20:21:11]

I can get into blob no problem, I'll try banging on the serial console throughout the boot....

[20:21:12]

I can launch the Zipit2 app at the point and "stuff" happens

[20:21:27]

Trying to get binutils built

[20:22:41]

no, sending returns, or ctrl-C has no effect after blob starts to load the kernel.

[20:23:22]

what's on the sd card?

[20:24:41]

The most recent z2shell archive off the wiki

[20:25:27]

It ends with own-tty /dev/tty0 /bin/sh -sh &

[20:25:41]

ah, when you start the zipit app by hand does it work?

[20:25:55]

It drops into the charging screen, yes

[20:26:14]

It's fairly mad, though, I think

[20:26:26]

can you get the zipit app to give you the normal gui?

[20:26:34]

It came up at least partially

[20:26:42]

The fbcons cursor was still blinking

[20:26:56]

I was considering dropping all that stuff out of the z2start.sh script to see if it behaved better

[20:27:07]

but I got distracted building binutils to look at the blob binary

[20:27:28]

It's been long enough I don't remember how to tell it what file format to use

[20:28:07]

Ah

[20:28:12]

You can upset it more to get more help

[20:28:14]

How very GNU

[20:28:17]

when I did the work in march, it was long before we had the serial port & jtag pads figured out.

[20:31:22]

Marvellous

[20:31:22]

once we found the serial stuff, I played around with blob for a while. I was able to rebuild it. the first stage was byte-for-byte the same but it divirged in main.c. Possibly compiler/binutils related.

[20:31:28]

binutils just segfaults trying to disassemble it

[20:32:27]

there is lots of data mixed with the code. 0x0000 in the blob binary is the int vector table.

[20:32:34]

It

[20:32:36]

It's code

[20:32:44]

ARM vectors are just branch instructions

[20:33:04]

00000000 32 00 00 ea 78 00 00 ea 78 00 00 ea 77 00 00 ea

[20:34:14]

yeah those first few are jumps to the int routines 0x0000 is a jump to the reset code at 0x000000d0

[20:34:47]

Right... so objdump should be able to disassemble it

[20:34:52]

Instead, the smoke comes out

[20:35:54]

which corresponds to reset in xlli/mainstone/start.s

[20:36:09]

Yup

[20:36:20]

I'm just trying to correlate what I see in the blob sources with the binary

[20:36:34]

Since those Z2 files in there look pertinent, but I can't see them being built from the makefiles

[20:36:39]

I han no luck with objdump on the straight binary.

[20:39:12]

the 2 z2 files are not used I believe, z2.s isn't compiled and looks like it has the same stuff inside as z2.inc

[20:41:04]

Right... but they have tantalising things in them... so cruel

[20:42:01]

and what is a PCF50606?

[20:42:25]

That's the TI PMU

[20:42:57]

Or I should say, "a" PMU

[20:43:25]

no that is a tps65021

[20:43:41]

Yes, sorry, my bad

[20:43:46]

Getting many things confused

[20:43:53]

The PCF* part is a Philips device

[20:45:39]

http://www.rockbox.org/twiki/pub/Main/DataSheets/pcf50606.pdf

[20:46:12]

yeah just found that it's a 56 pin device.

[20:46:27]

not in the z2 that we have...

[20:46:54]

There's one on the Mainstone board though

[20:47:51]

that was probably their dev kit. they left a lot of z2 stuff in the mainstone.xxx files rather than getting a new machine id and doing a real z2 varient.

[20:48:46]

Mainstone was the Intel name for the Bulverde development board

[20:48:55]

Bulverde being the PXA27x family

[20:49:28]

There's an Intel wiki with a bunch of mainstone/bulverde stuff on it

[20:49:42]

blink.h isn't referenced anywhere either.

[20:49:44]

there's a guy mentioned in the blob sources that's also xref'ed on the wiki

[20:49:49]

Ok

[20:50:35]

http://embedded.seattle.intel-research.net/wiki/index.php?title=Main_Page

[20:50:36]

The darwinports arm-elf-binutils seems to be able to disassemble blob just fine

[20:50:50]

That's the one

[20:51:14]

ah, do you have a link to that?

darwinports?

www.darwinports.com

duh ;P

It looks like it just built a vanilla 2.17 binutils

[20:53:25]

There's a bunch of arm cross stuff in the 'cross' category

[20:53:31]

I used binutils 2.16.1

[20:53:43]

and it doesn't like the raw binary.

[20:53:45]

I tried 2.18, which was a no-go

[20:54:26]

so you have a disassembled file of blob now?

[20:54:56]

I have something that looks vaguely like it

[20:55:50]

I suspect that the first/rest break may not look so good

[20:56:37]

the point of most interest.... ;>(

[20:56:52]

Indeed

[20:56:58]

but better than nothing...

[20:57:06]

I need to spend a little time working out how to get the BDI to talk to this thing

[20:57:23]

I suspect I just need VCC of some sort to make it happy

[20:58:08]

ah, I got the flyswatter wokring, I got vtarget from the gold pad near the tps pin 10. and connected nSRST to the little reset button.

[20:58:36]

I can step through blob one inst at a time;

[20:58:54]

tps pin 10?

[20:59:21]

the ti power manager, I was gaoing to put a red circle around it but haven't got to it yet.

[20:59:42]

That's the one closest to the audio jack?

[20:59:59]

yeah

[21:00:46]

Ok, and the live pad on the reset button was the one closest to the front corner of the unit?

[21:00:56]

Or the one closest to the centre of it?

[21:00:59]

pin 1 is lower left pin 10 lower right the gold pat just to the lower right of the chip

[21:01:42]

http://elinux.org/upload/9/91/Z2jtag.jpg

[21:02:11]

I connected a ground there, mistakenly thinking it was just the upper left pin that was switched

[21:02:13]

oops

[21:02:14]

8)

[21:02:23]

It don't boot too good when you do that

[21:02:34]

no it don't

[21:03:36]

that vcc stays on when the cpu goes to sleep, so you can do a reset after blob init's the gpio's etc and shuts off the main vcc

[21:04:09]

Hmm

[21:04:12]

Not sure I follow you there

[21:04:32]

But you're using OpenOCD, correct?

[21:06:30]

the little reset button is connected to HOT_RESET on the Ti chip which drives nRESET to the CPU. when you push it all the power comes on and blob runs the reset vector code which init's the gpio's and some other stuff and then puts the cpu to sleep, and shuts down all the non-essintial VCC's.

[21:07:04]

thus it appears as if the power is off, but it's not really.

[21:07:40]

the power button on the keyboard causes an interrupt which brings the cpu out of sleep and runs the normal_boot code.

[21:08:22]

I'am using openoce and a TCT flyswatter JTAG (fdti2232 based)

[21:08:26]

openocd

[21:08:42]

Ok, thanks

[21:08:56]

I have a BDI and the relevant firmware