Permalink
Commits on May 3, 2015
  1. print exact commands to write the image to flash

    eddyp committed May 3, 2015
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
Commits on Apr 26, 2015
  1. Revert "Use RedBoot's command buffer instead of a local one"

    eddyp committed Apr 26, 2015
    This reverts commit f1fd575.
  2. drop the useless ip_address command setting the server ip

    eddyp committed Apr 26, 2015
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  3. Todo wrt the (hacked) reboot command

    eddyp committed Apr 25, 2015
    The current implementation of the Apex 'redboot' command is not ideal
    and there are several improvements that can be done. Note what those are
    in case more interest is gathered around this topic in the future.
    
    OTOH, the current implementation is enough to make TFTP+NFS NetBSD
    booting possible by just powering the device (after making the kernel
    available through TFTP and a root file system over NFS).
    
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  4. netbsd-nfs-nslu2_config: hacked apex config to netboot NetBSD

    eddyp committed Apr 25, 2015
    Added the configuration used to build a hacked Apex bootloaded for NSLU2
    which can be used to boot NetBSD over TFTP+NFS. The 'hack' in Apex is to
    provide a new 'redboot' command which can execute one of multiple
    predefined RedBoot commands (see variable chain_cmds in
    src/apex/cmd-redboot.c to see more details).
    
    This config relies on a having on the NSLU2 a specific version of the
    RedBoot bootloader, namely the version which reports:
    
      RedBoot(tm) bootstrap and debug environment [ROMRAM]
      Red Hat certified release, version 1.92 - built 15:16:07, Feb  3 2004
    
      Platform: IXDP425 Development Platform (XScale)
      Copyright (C) 2000, 2001, 2002, Red Hat, Inc.
    
    The memory configuration of the Linksys NSLU2 on which this hacked Apex
    was tested is:
    
      RAM: 0x00000000-0x02000000, 0x000723a0-0x01ff3000 available
      FLASH: 0x50000000 - 0x50800000, 64 blocks of 0x00020000 bytes each.
    
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  5. empty_nslu2.flash: minimal/fake flash image

    eddyp committed Apr 25, 2015
    RedBoot expects a valid sercom header at 0x50160000 for the ramdisk,
    otherwise it will crash. Of course, since at 0x50060000 we would load a
    second stage hacked Apex bootloader to boot via TFTP the NetBSD kernel,
    and since for NetBSD booting there is no ramdisk used, we will simply
    create a small/dummy ramdisk which we can flash at 0x50160000 so RedBoot
    does not crash.
    
    Create the fake image with 'make empty_nslu2.flash'.
    
    The resulting image can be written to flash directly from RedBoot with
    the following command (assuming the image has been loaded already at
    0x200000):
    
    fis write -f 0x50160000 -b 0x200000 -l 32
    
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  6. Makefile: apex_nslu2.flash: Apex image with sercomm header

    eddyp committed Apr 21, 2015
    RedBoot expects a Sercomm header to be able to boot using the Sercomm
    sequence. The Sercomm header contains the size of the rest of the image.
    
    We create an Apex image appropriate for flashing into the NSLU2's flash
    at address 0x50060000 (instead of kernel).
    
    The '+16' bit is copied from the Debian package flash-kernel code in
    /usr/share/flash-kernel/functions and is prbably meant to make sure
    there are no alignment issues, incomplete writes to flash, since,
    according to http://www.nslu2-linux.org/wiki/Info/BootFlash the size is
    for the rest of the data bytes excluding the 16 bytes header. Since
    there is no harm in RedBoot copying a few more bytes during the Sercomm
    boot sequence, we add 16 to our size in the Sercom header, too.
    
    Writing into NSLU2's flash can be done at address 0x50060000 with the
    command below (assuming the apex_nslu2.flash image was loaded in RAM ad
    0x200000):
    
    fis write -f 0x50060000 -b 0x200000 -l <size>
    
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
Commits on Apr 25, 2015
  1. increase the buffer sizes for redboot commands

    eddyp committed Apr 21, 2015
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
Commits on Apr 19, 2015
  1. typo fixes

    eddyp committed Apr 19, 2015
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  2. cmd_redboot: better debug message for redboot parsing

    eddyp committed Apr 19, 2015
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  3. Prepare APEX for NetBSD booting

    eddyp committed Apr 19, 2015
    - Drop the built in RedBoot version command; as a reference,
      the following version was tested with this code:
    
        RedBoot(tm) bootstrap and debug environment [ROMRAM]
        Red Hat certified release, version 1.92 - built 15:16:07, Feb  3 2004
    
        Platform: IXDP425 Development Platform (XScale)
        Copyright (C) 2000, 2001, 2002, Red Hat, Inc.
    
    - Load the netbsd-nfs.bin image at 0x200000 instead of
      0x220000.  NetBSD must be loaded at 0x200000, is not as
      relocatable as Apex.
    
    - Using this code and the Apex settings below, the autobooting
      should, at least in theory, work:
    
        CONFIG_ENV_STARTUP_PREFIX_P=y
        CONFIG_ENV_STARTUP_PREFIX="redboot 0 ; redboot 1 ; redboot 2"
    
    NOTE: The built-in commands assume the tftp server is at
          192.168.0.2.
    
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  4. Use RedBoot's command buffer instead of a local one

    eddyp committed Apr 19, 2015
    We need to shrink the Apex footprint for flash execution and
    RedBoot does the same kind of processing we did here, anyway.
    
    The RedBoot buffer is already available, so it would be a
    shame to waste it :) .
    
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  5. less verbose debugging

    eddyp committed Apr 18, 2015
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  6. pass the command itself, RedBoot expects it

    eddyp committed Apr 19, 2015
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  7. rewrote the inital command splitting code

    eddyp committed Apr 19, 2015
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  8. add our own cmd_proc and call functions directly

    eddyp committed Apr 5, 2015
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  9. correct netbsd image filename

    eddyp committed Apr 5, 2015
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  10. some other attempts at calling RB

    eddyp committed Apr 5, 2015
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  11. WIP: hack a redboot call command (DOESN'T WORK)

    eddyp committed Mar 31, 2015
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  12. typo fixes in tftp driver

    eddyp committed Mar 31, 2015
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
Commits on Mar 29, 2015
  1. correct the pause explanation

    eddyp committed Mar 29, 2015
  2. Fix outdated usages of alias* instead of variable

    eddyp committed Mar 29, 2015
    There were several places where alias was still used instead of variable.
    
    This time I grepped through the entire sources and fixed all obvious outdated usages.
    
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
  3. Makefile: allow file to executed directly

    eddyp committed Mar 29, 2015
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
Commits on Mar 27, 2015
  1. Fix variations on NSLU2

    eddyp committed Mar 27, 2015
    Since the renaming of aliases to variables, the variations feature was
    non-woring on NSLU2 because variation-nslu2.c was still using alias_set
    instead of variable_set.
    
    This fixes the man of the NSLU2 configs using this feature, including all
    debian_nslu2*_config-s and completes the changes introduced in commit
    0e966c3.
    
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
Commits on Mar 25, 2015
  1. Merge remote-tracking branch 'cooldavids/master' into nslu2

    eddyp committed Mar 25, 2015
    Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
Commits on Aug 24, 2014
  1. nslu2_config: Tweak

    vstehle committed Aug 24, 2014
    Signed-off-by: Vincent Stehlé <vincent.stehle@laposte.net>
  2. bswapsi2: fix compilation with buildroot toolchain

    vstehle committed Aug 24, 2014
    Signed-off-by: Vincent Stehlé <vincent.stehle@laposte.net>
  3. drv-fis: fix lookup

    vstehle committed Aug 24, 2014
    Signed-off-by: Vincent Stehlé <vincent.stehle@laposte.net>
  4. Fix mixed implicit and normal rules

    cooldavid authored and vstehle committed Jun 6, 2012
    Latest GNU Make won't work with mixed implicit and normal rules
Commits on Jun 6, 2012
  1. Fix LH79520-10 CF Card Driver

    cooldavid committed Jun 6, 2012
    I spend a lot of time trying to figure out why my
    CF Card won't work both in APEX and Linux.
    
    After playing around with 0x5020XXXX addresses,
    I found that the bit offset seems to be offed by 1.
    
    After this patch my Linux and Apex works fine with CF Card.
    PS: My Linux patch seems form the same author of APEX. ;)
  2. Fix mixed implicit and normal rules

    cooldavid committed Jun 6, 2012
    Latest GNU Make won't work with mixed implicit and normal rules
Commits on Jun 5, 2012
  1. Adding missing cmd-list.c file.

    beewoolie committed Jun 5, 2012
    o Omitted because of gitignore.
    o Revised gitignore to prevent another similar mishap.
Commits on Dec 9, 2011
Commits on Dec 6, 2011
Commits on Sep 16, 2011
  1. Notes.

    beewoolie committed Sep 16, 2011