Differences

This shows you the differences between two versions of the page.

--- build [2011/08/09 16:02] – original firmware info ray
+++ build [2013/02/19 14:25] (current) – [4. Deploy] added instructions for manual firmware deployment trapicki
@@ Line 1: / Line 1: @@
 ====== Build environment ======
+<note tip>
+See point 5 on how to rebuild a near-original firmware for your r0ket. The original firmware that comes with your badge contains secret keys that you will not find in git. If you reflash your firmware, those will be lost - but that's not as bad as it sounds unless you like properitary stuff. WE RECOMMEND THAT YOU READ POINT 5 BELOW BEFORE YOU REFLASH YOUR FIRMWARE THOUGH. Meanwhile the keys are leaked, so rebuilding an original firmware is possible. Guess we will also check them into git soon... </note>
 ===== 0. Work in Progress =====
 After all it's r0ket science. If you actually start developing something for the r0ket now, we'd recommend to join our Mailing List and IRC channel - see [[contact|contact]].
@@ Line 6: / Line 10: @@
 ===== 1. Compiler =====
+<note tip>
+As of November 2012, CodeSourcery got bought by [[http://www.mentor.com/embedded-software/codesourcery|Mentor Graphics]]. The old "Sourcery G++ Lite" is now [[http://www.mentor.com/embedded-software/sourcery-tools/sourcery-codebench/editions/lite-edition/|Sourcery CodeBench Lite Edition]]. The links in this page to the old toolchain don't work; if it's legal, I could upload my copy of arm-2011.03-42-arm-none-eabi.bin somewhere for everybody to download. Otherwise, you could try to register on mentor.com  and download the freely available [[http://www.mentor.com/embedded-software/sourcery-tools/sourcery-codebench/editions/lite-edition/arm-eabi/|Sourcery CodeBench Lite Edition (EABI Release)]] and see if the old installation instructions still make sense.\\
+[[mailto:bruno@anche.no|White_Rabbit]] </note>
 <note tip>
 The CodeSourcery G++ package provides a GNU cross-compiler toolchain for ARM.
@@ Line 37: / Line 44: @@
 bash
 </code>
+<note tip>
+Running on FreeBSD8.2 with gcc4.5.2
+</note>
+<code>
+cd /usr/ports/devel/
+sudo cp -r ./arm-rtms-gcc ./arm-none-eabi-gcc
+cd arm-none-eabi-gcc
+vi Makfile
+> change line 11 "TGTABI = rtems" to "TGTABI = none-eabi"
+> uncommit line 12 "WITH_RTEMS_PATCHES= yes"
+sudo make install clean
+# at me one c-file dont found: "sys/sysctl.h"
+# fix it in this c-file :"#include <sys/sysctl.h>" to "#include '/usr/include/sys/sysctl.h'"
+</code>
+Someone asked me to write this down, so [[building_with_a_free_toolchain|here]]'s how I built the firmware with stock binutils, gcc, newlib and gdb.
 ===== 2. Source =====
@@ Line 73: / Line 99: @@
 </code>
+<note tip>
+For FreeBSD
+</note>
+<code bash>
+gmake APP=hello
+</code>
 ===== 4. Deploy =====
 <note tip>
-The LPC1343 has a USB-ISP, it can be booted as an USB storage device, and the firmware can be copied as a file. Unfortunately, this only works on Windows. On Linux, the firmware has to be directly written to the device. Additionally, a bug in the ROM causes the CPU to hang for 20 seconds when entering the bootloader. To make development more convenient, the autoflash tool monitors the USB for r0kets in ISP-Mode and automatically flashes the last built firmware. An "ISPAndReset()"-function in the library enables the user to reset the r0ket to ISP-mode and circumvent the 20 second delay.
+The LPC1343 has a USB-ISP, it can be booted as an USB storage device, and the firmware can be copied as a file. Unfortunately, this only works on Windows. On Linux, the firmware has to be directly written to the device, see [[#manual|below]]. Additionally, a bug in the ROM causes the CPU to hang for 20 seconds when entering the bootloader. To make development more convenient, the autoflash tool monitors the USB for r0kets in ISP-Mode and automatically flashes the last built firmware. An "ISPAndReset()"-function in the library enables the user to reset the r0ket to ISP-mode and circumvent the 20 second delay.
 </note>
+==== automatic =====
   * Install udev rule
@@ Line 88: / Line 123: @@
 <note important>
-On ubuntu, nautilus will automount the USB-Storage device, and our automouter will not work. Disable the nautilus-automounter:
+On ubuntu, nautilus will automount the USB-Storage device, and our automounter will not work. Disable the nautilus-automounter:
 <code>
 gconf-editor
@@ Line 112: / Line 147: @@
 </note>
+==== manual ====
+Due to some incompatibility in the Linux vfat driver, just mounting and copying the firmware does not work. The firmware has to be dd'ed to the r0ket:
+  - connect r0ket in ISP mode
+  - make sure that the device is not mounted
+  - dd if=firmware.bin of=/dev/<devicename> bs=1024 skip=2
+The right device is 34 kiByte in size.
 ===== 5. Building the original Firmware =====
 To rebuild the original firmware you can use the release-all script in firmware/. It will build the firmware and all additional programs and put them into /release. You have to copy SECRETS to SECRETS.release first.
-Note: the SECRETS in the git contain different keys like you had on your badge originally. That way the original firmware can only execute l0dables signed by the vendor (us!!), which makes it at least as cool as an iBadge. So if you reflash your badge you will lose this "feature" forever. Also you will leave the mesh of sheeps and enter the mesh of hackers, as the keys are different. Don't trust the INVADERS highscore on the hackers mesh though. Finally openAMD will stop encrypting messages - so if you have tracking enabled people can sniff and fake your ID. And of course you can always try to break the read protection of a working badge and extract it's crypto keys and release them on wikileaks. We'd love to see that. But wait for day 2 with that so we can see the mesh working before for some time...
+Note: the SECRETS in the git contain different keys than those you had on your badge originally. That way the original firmware can only execute l0dables signed by the vendor (us!!), which makes it at least as cool as an iBadge. So if you reflash your badge you will lose this "feature" forever. Also you will leave the mesh of sheeps and enter the mesh of hackers, as the keys are different. Don't trust the INVADERS highscore on the hackers mesh though. Finally openAMD will stop encrypting messages - so if you have tracking enabled people can sniff and fake your ID. And of course you can always try to break the read protection of a working badge and extract its crypto keys and release them on wikileaks. We'd love to see that. But wait for day 2 with that so we can see the mesh working before for some time...
+After running the release-all script, copy the release/final.bin to firmware/firmware.bin and flash it using autoflasher:
+<code>
+firmware/$> cp SECRETS SECRETS.release
+firmware/$> ./release-all
+firmware/$> cp ../release/final.bin firmware.bin
+firmware/$> ../tools/bootloader/autoflash
+</code>
+If you get some message the firmware is too large, it might be an accident in our source, but more probably you were using a different compiler than exactly the version we used.
+Now hold joystick to the left and power on, release joystick after a second or so. LEDs will light dim, after about 20 seconds the autoflasher should tell you flashed sdX...
+Then start your badge holding joystick down (not press center but down), that starts the USB massstorage. Mount it (should be a 512K volume). Erase all files (except your own ones, like received stuff from others if you have that) and/or copy the files from ../release/files/ onto the bage. The new .c0d files are now signed with the git-"SECRETS" and therefore will run again on your firmware, as well as you'll be able to run l0dables others built using the same keys.
-After running the release-all script, copy the release/final.bin to firmware/firmware.bin and flash it like written above. Then start your badge, start the USB massstorage and mount it. Erase all files (except your own ones, like received stuff from others) and/or copy the files from /release/files/ onto the bage. The new .c0d files are now signed with the git-"SECRETS" and therefore will run again on your firmware, as well as you'll be able to run l0dables others built using the same keys.
+If your filesystem is corrupted, you can download http://y.nu/r0ket-df.dd and dd it onto the device before mounting it. Or you can flash the initial.bin created by the above process before flashing final.bin, starting initial.bin format the dataflash and starts massstorage.