Geekmaster's adventures in K1 spelunking

[geekmaster]

Info harvested from my K1 with RUNME.sh (launched by Update_RUNME.bin, which redirects all output to /mnt/us/documents/RUNME.txt so you can view it on the kindle):

PHP Code:

#> cat /proc/cpuinfo
Processor    : XScale-PXA255 rev 6 (v5l)
BogoMIPS    : 99.32
Features    : swp half thumb fastmult edsp 
CPU implementer    : 0x69
CPU architecture: 5TE
CPU variant    : 0x0
CPU part    : 0x2d0
CPU revision    : 6
Cache type    : undefined 5
Cache clean    : undefined 5
Cache lockdown    : undefined 5
Cache format    : Harvard
I size        : 32768
I assoc        : 32
I line length    : 32
I sets        : 32
D size        : 32768
D assoc        : 32
D line length    : 32
D sets        : 32

Hardware    : Fiona Platform
Revision    : 0009
Serial        : "B101xxxxxxxxxxxx"

#> uname -m
armv5tel

#> uname -r
2.6.10-lab126

#> cat /proc/cmdline
console=ttyS2,115200n8 root=/dev/rd/0 initrd=/linuxrc rw reboot=hard 


Here is my Update_RUNME.bin (for K1), which also displays a "done" message and waits for you to press the 'R' key before restarting (so you can read any content written to the eink display by your RUNME.sh script):

PHP Code:

#!/bin/sh

export PATH=/bin:/sbin:/usr/bin:/usr/sbin:$PATH

LOG=/mnt/us/documents/RUNME.txt
RUNME=/mnt/us/RUNME.sh
DN=/dev/null

echo "" >$LOG 2>&1

eips 23 12 "Hello USBterm." >$DN 2>&1
sleep 1
  
rm -f /mnt/us/update*.bin >$DN 2>&1
rm -f /mnt/us/Update*.bin >$DN 2>&1
rm -f /mnt/mmc/update*.bin >$DN 2>&1
rm -f /mnt/mmc/Update*.bin >$DN 2>&1

if [ -f $RUNME ]; then
    sh $RUNME >>$LOG 2>&1
  else
    echo "=== $RUNME not found..." >>$LOG 2>&1
fi

eips 1 1 "Done. Goodbye. Press 'R' key to restart."  
waitforkey 19 # wait for K1 'R' keycode
sleep 1 # needed to prevent weirdness
eips 3 3 "Restarting..."    

return 0 


[geekmaster]

I built a lenny.ext3 from scratch (using his debootstrap intructions). I loop mounted it and did a qemu-static ch to it from my x86 linux mint. I installed build-essentials for architecture armeabi. I build a hello.c with "gcc -march=arm5te -o hello hello.c". It does not work, just like all the others.

Dynamic shared lib executables say "not found" when executed. Static builds seg fault. Just like always. I wonder if "not found" means it tried to run but could not find some library?

When I force the cpu-type for emulation, it still works on my x86 box (so it should on K1 unless a library issue):

PHP Code:

mint17-32 build # file hello
hello: ELF 32-bit LSB  executable, ARM, EABI4 version 1 (SYSV), statically linked, for GNU/Linux 2.6.14, not stripped
mint17-32 build # qemu-arm-static -cpu pxa255 hello
hello world! 


[knc1]

https://www.mobileread.com/forums/sho...9&postcount=38

[geekmaster]

Originally Posted by knc1

My NSLU2 is an Intel Xscale IXP420 device -
I have a complete, self-hosting, install of Debian on it -
I know, I set it up to build the entire install from source (it took 3 or 4 days IIRC).
So it is possible.

Didn't you once tell me you had an NSLU-2 also?
Use it to native compile your code for Xscale under NSLU2/Debian.

Xscale is really, really old - I can only guess beyond what I have already posted.

Note: The XScale architecture is based on the ARMv5TE ISA without the floating point instructions.
Ref: https://en.wikipedia.org/wiki/XScale

I do not have an NSLU2, though I do have root shell on most of my devices (including a Seagate NAS). I did some kernel module work on OpenWRT (mipsel) years ago. I will look at thst link again...

[geekmaster]

I just received a K2 from an online auction. SN B002*, and it boots fine.

I just received another package from a MobileRead member who read my post about wanting to trade for kindles to test on. It has a couple paperwhite models -- trying to see which serials/models, but they need to charge longer to stop rebooting, I think. Or I need to debrick one or more of them. The settings menu always reboots before displaying anything on one (and now it is stuck in a boot loop at the kindle tree screen, showing up on USB as "Lab126, Inc. USB Charger"). They need to charge overnight.

I need to package up my K5 I have been working on lately (well tested) and send it to him too, ASAP. Now I will be able to support 3 more kindle models!

One of the PW units has no response other than 4 blinks on the power LED after holding it for 8 seconds. Other than that, it does not turn off the LED even after holding the power button for a minute or more. The eink screen is stuck at about 90-percent boot progress, and never changes.

The other PW unit keeps rebooting, though it may sit at the desktop for several minutes. Rebooting sets backlight brightness near maximum. It remembers through spontaneous reboots that I turned on airplane mode and I turned off cover mode. We shall see how it behaves tomorrow. One bad thing -- tapping power goes to screensaver mode, and there is no grayscale, as though the eink waveform is hosed. More reason to learn how to create custom eink waveforms. It seems to be stabilizing now (though it just rebooted after I grabbed its info, and not responding to touchscreen or power button). The SN says it is a PW3 wifi, and other info:

SN G090G1*
FW 5.6.1.1 (2634130033)
Waveform 320_R046_AE3001_ED060KD1C1_TC

The other one is still dead, other than the 4 blinks after 8 seconds of power button press.

[geekmaster]

Okay, may be are having loader issues (executable "not found") because various cpu-compatible rootfs apps, and most cross-compilers, want to link agains glibc, whereas the K1 uses uclibc? Still finding myself too early to predict remaining learning curve here, and too many competing projects...

[geekmaster]

@knc1: Thanks for turning me on to binfmt. It never consciously crossed my path before. I think it is effing wonderful! Now I plan to spend more time on the linux side of the fence...

[knc1]

Originally Posted by geekmaster

I just received a K2 from an online auction. SN B002*, and it boots fine.

I just received another package from a MobileRead member who read my post about wanting to trade for kindles to test on. It has a couple paperwhite models -- trying to see which serials/models, but they need to charge longer to stop rebooting, I think. Or I need to debrick one or more of them. The settings menu always reboots before displaying anything on one (and now it is stuck in a boot loop at the kindle tree screen, showing up on USB as "Lab126, Inc. USB Charger"). They need to charge overnight.

I need to package up my K5 I have been working on lately (well tested) and send it to him too, ASAP. Now I will be able to support 3 more kindle models!

One of the PW units has no response other than 4 blinks on the power LED after holding it for 8 seconds. Other than that, it does not turn off the LED even after holding the power button for a minute or more. The eink screen is stuck at about 90-percent boot progress, and never changes.

The other PW unit keeps rebooting, though it may sit at the desktop for several minutes. Rebooting sets backlight brightness near maximum. It remembers through spontaneous reboots that I turned on airplane mode and I turned off cover mode. We shall see how it behaves tomorrow. One bad thing -- tapping power goes to screensaver mode, and there is no grayscale, as though the eink waveform is hosed. More reason to learn how to create custom eink waveforms. It seems to be stabilizing now (though it just rebooted after I grabbed its info, and not responding to touchscreen or power button). The SN says it is a PW3 wifi, and other info:
The other one is still dead, other than the 4 blinks after 8 seconds of power button press.

Don't let it connect to the Internet or it will update to something other than 5.6.5 -
You need that to use the software jailbreak -
Although it sounds like this one needs a serial port.

You can find a thread by myself where I installed the adapter board inside of my PW-3 -
The usual USB cable plugs into the bottom - the serial/USB plugs into the top.

- - - -

5.6.1.1 is what the PW-3 originally shipped with.

[geekmaster]

Originally Posted by knc1

Don't let it connect to the Internet or it will update to something other than 5.6.5 -
You need that to use the software jailbreak -
Although it sounds like this one needs a serial port.

You can find a thread by myself where I installed the adapter board inside of my PW-3 -
The usual USB cable plugs into the bottom - the serial/USB plugs into the top.

- - - -

5.6.1.1 is what the PW-3 originally shipped with.

I got it into airplane mode during a (barely) long enough moment between reboots. Thanks.

If I get these working, I will have K1 though PW3 (minus the PW1 that walked off in California). I have more "as-is" units on the way, so I will have a stack of misc. units to debrick/repair/(or maybe just recharge?) -- or for spare parts...

[donB006]

Here is an interesting project using the PXA255:
http://www.linuxjournal.com/magazine...-x-and-gumstix

I found these too although they could be big/little endian or uclibc mismatch.
http://downloads.gumstix.com/feeds/a...c/ipk/armv5te/

[geekmaster]

I tried building my hello.c in a qemu chroot of the arm-debian-lenny.ext that I built using twobob's instructions, after copying in /lib from the rootfs unpacked from the latest K1 OTA firmware update. It built fine, but STILL throws a "not found" error when executed on my K1 (just like before). Now I will try a static build... Same "seg fault" as before. I am feeling rather vexed by this build problem.

I can do interesting things on my K1 with scripts, including eink animation (but very slow). There are some things where a compiled program is the only way (and not just for execution speed). Then of course, I want to built kernel module, because the K1 has no USB networking (and other things).

[donB006]

I have installed gcc into my DXG 2.5.8 and have compiled hello world. See attached.

I found the glibc directory in addition to the above gumstix directory for the uclibc files for the armv5te.

I grabbed several .ipk files and extracted the contents using 'dpkg-deb -x file.ipk directory' command.

binutils_2.18-r1_armv5te.ipk
binutils-doc_2.18-r1_armv5te.ipk
gcc_4.1.2-r7_armv5te.ipk
gcc-symlinks_4.1.2-r7_armv5te.ipk
libc6_2.5-r7_armv5te.ipk
libc6-dev_2.5-r7_armv5te.ipk
libgcc1_4.1.2-r10_armv5te.ipk
libgcc-s-dev_4.1.2-r10_armv5te.ipk

I next put the extracted files in /mnt/us/arm-gcc directory and linked the files into the rootfs. This was very messy and just a test to see if the binaries would work with the libraries. I plan to delete all of these changes and return my DXG to normal since I was just experimenting.

I found a slight problem where the libbfd and libopcodes were version 2.18 in my experiment while only 2.17 in the original DXG 2.5.8 except that worked anyway.

So Geekmaster, I would think you could try the same thing with your Kindle1 except using the uclibc versions instead of the libc6 versions. Your Kindle1 might be easy to convert into a full development machine using the sdcard for space. I doubt the wait would be very long for building your programs unless they were gigantic. That should depend on how much effort you want to invest in the setup. Cross-compilers are a pain.

[geekmaster]

I have had no problems compiling and running apps on my K2,DX,DXG,K3,K4,K5,PW1, and PW3. I have not been able to compile anything that runs for the K1, even when using the libs dir copied directly from the K1.

A little problem with compiling programs on the K1 is that I cannot find or compile a compiler FOR the K1. I tried compiling tcc -- no luck. I cannot even compile a simple "hello world" app that will run.

And then there is the little problem of no g_ether.ko, and therefore no USB networking. One of the FIRST things I want to compile is a g_ether.ko so I can have root shell.

But now, no shell -- only running RUNME.sh launched from my custom Update_RUNME.bin, which I launch with "Menu/Settings/Menu/Update Your Kindle". And the ONLY binary code that runs on it is code I copied from the rootfs. Even static compiled code using the /lib copied from the K1 fails (either "not found", or segfault), as I mentioned previously.

I am flummoxed and confuzzled by this untenably disquietudinous situation.

@donB006: But thanks for the suggestion. If you can think of anything I *haven't* tried, please tell me and I would be more than happy to give it a try. And I will try what you did too, just for fun, but with no expectation for success...

[donB006]

Originally Posted by geekmaster

And then there is the little problem of no g_ether.ko, and therefore no USB networking. One of the FIRST things I want to compile is a g_ether.ko so I can have root shell.

@donB006: But thanks for the suggestion. If you can think of anything I *haven't* tried, please tell me and I would be more than happy to give it a try. And I will try what you did too, just for fun, but with no expectation for success...

Wow, that is a tough circumstance. I can't imagine not having the usbnet working. Looking at the file listing I see the Kindle1 only has the usbserial.ko kernel module. I wonder if that is only for the wan. I sure wouldn't want to work on a hardwired serial console for free. For that reason, I can't think of any way to improve the Kindle1 at all. I wouldn't even try my technique without the usbnet.

New idea just now: I did see the Amazon update file version 30? has the full rootfs.img available. You could edit that file when loop mounted and then reflash the entire OS all at once. No more hope at this time. Sorry.

[geekmaster]

Originally Posted by donB006

Looking at the file listing I see the Kindle1 only has the usbserial.ko kernel module. I wonder if that is only for the wan. I sure wouldn't want to work on a hardwired serial console for free. ... New idea just now: I did see the Amazon update file version 30? has the full rootfs.img available. You could edit that file when loop mounted and then reflash the entire OS all at once. No more hope at this time. Sorry.

I can already run scripts. Reflashing would only be useful if I could add some working binaries, but then of course, I can do that with my install script.

The only partial solution I have is to export USB mass storage and then ALSO remount the filesystem for my install script, then do a "shared file terminal" (which I have already developed and tested using a windows host and linux VM with shared files). The basic idea is for the K1 to tail follow INFILE (appended to by the PC) and to append shell output to OUTFILE (tail followed by the PC). To bypass I/O buffering problems (no apps on the K1 to defeat piped STDIO buffering), I pipe tailed INFILE to "while read line, do $line; done >> OUTFILE" (typed from memory, so beware); I also echo the input command and a prompt inside that loop. It works fine for non-interactive terminal stuff, but I am hopelessly addicted to 'nano" in my kindles, and that does not work so while still line-buffering I/O... And I am a bit afraid that sharing a raw device on both sides of the USB will cause file system corruption, with script STDOUT appending. If I could compile apps, I could write to fixed size preallocated mapped files, so file system corruption should not be a possibility. But for that, I need to be able to compile code I can run in the K1... So compiling is still my bottleneck of infinite resistance (i.e. a brick wall with impenetrable pores).

I want to AT LEAST get a working tcc compiler that will run in my K1, but first a working "hello" would make me ecstatic at this juncture.

EDIT: I just did a "history" command, and copied my accurate command line that I tested (similar to what I posted above):

tail -f in|while read line;do echo "$line";$line 2>&1;echo; echo -n "#> ";done>out

I had two CYGWIN windows open on the host side. In the small window (two lines tall) I typed commands (executed on the target when I pressed ENTER):

cat >> in

In the other larger window, which displayed output from the target:

tail -f out

It worked fine for simple non-interactive stuff, but a serial port connection would be much better (and eventually usbnet).