This summon-arm didn't work right out of the box due to missing tarballs (lame mirror?) and complaints about missing patchfiles. Did I miss the patchfiles in the setup instructions? Here's what I did which worked but in retrospect may not have been the easiest thing...
* Diff the summon-arm from this attachment and make a few judgements about what should work
* Realize that I still need patches to google them so create a folder "patches" and copy in the following:
(I would put this in the script but my lame osx wget doesn't like github's https cert)
(I recall there was another but can't seem to find it in my history)
Results: Though I couldn't build the unified branch, I was able to build 550d. After boldly dropping in the autoexec.bin, I booted the camera..... SUCCESS!
Anyway my final summon-arm DIFFers from esden's one on github:
--- summon-arm-toolchain
+++ summon-arm-toolchain
@@ -19,28 +19,30 @@
# You probably want to customize those
# You can also pass them as parameters to the script
##############################################################################
-TARGET=arm-none-eabi # Or: TARGET=arm-elf
-PREFIX=${HOME}/sat # Install location of your final toolchain
+TARGET=arm-elf # Or: TARGET=arm-elf
+PREFIX=${HOME}/arm-toolchain # Install location of your final toolchain
DARWIN_OPT_PATH=/opt/local # Path in which MacPorts or Fink is installed
# Set to 'sudo' if you need superuser privileges while installing
SUDO=
# Set to 1 to be quieter while running
QUIET=0
# Set to 1 to use linaro gcc instead of the FSF gcc
-USE_LINARO=1
+USE_LINARO=0
# Set to 1 to enable building of OpenOCD
-OOCD_EN=1
+OOCD_EN=0
# Set to 'master' or a git revision number to use instead of stable version
OOCD_GIT=
# Set to 1 to build libstm32 provided by ST
LIBSTM32_EN=0
# Set to 1 to build libopencm3 an open source library for Cortex M3 and simalar
# chips
-LIBOPENCM3_EN=1
+LIBOPENCM3_EN=0
# Make the gcc default to Cortex-M3
DEFAULT_TO_CORTEX_M3=0
# Override automatic detection of cpus to compile on
CPUS=
+# Mirror to use for downloading GNU software
##############################################################################
# Parsing command line parameters
@@ -84,6 +86,9 @@
CPUS=*)
CPUS=$(echo $1 | sed 's,^CPUS=,,')
;;
+ GNUMIRROR=*)
+ GNUMIRROR=$(echo $1 | sed 's,^GNUMIRROR=,,')
+ ;;
*)
echo "Unknown parameter: $1"
exit 1
@@ -114,7 +119,7 @@
# For FSF GCC:
GCCVERSION=4.5.2
GCC=gcc-${GCCVERSION}
+ GCCURL=${GNUMIRROR}/gcc/${GCC}/${GCC}.tar.gz
else
# For the Linaro GCC:
GCCRELEASE=4.5-2011.02-0
@@ -158,7 +163,7 @@
BINUTILFLAGS=
if [ ${DEFAULT_TO_CORTEX_M3} == 0 ] ; then
- GCCFLAGS=
+ GCCFLAGS="--with-float=soft"
else
# To default to the Cortex-M3:
GCCFLAGS="--with-arch=armv7-m --with-mode=thumb --with-float=soft"
@@ -316,10 +321,10 @@
cd ${SOURCES}
+fetch ${BINUTILS} ${GNUMIRROR}/binutils/${BINUTILS}.tar.bz2
fetch ${GCC} ${GCCURL}
+fetch ${GDB} ${GNUMIRROR}/gdb/${GDB}.tar.bz2
if [ ${OOCD_EN} != 0 ]; then
if [ "x${OOCD_GIT}" == "x" ]; then
> Following Chris' suggestion, I'm increasing ML version number in order
> to prevent further thread splitting problems :)
>
> I've upgraded to GCC 4.6.0, with the attached summon-arm script
>
> * With the new GCC, autoexec.bin shrunk by 8k (which I think it's pretty good).
>
> * There was a strange structure alignment problem with picture style
> data, and maybe with bitrate structure too: 0x12345678 was displayed
> as 0x56781234. The issue was only present with -Os; it was working
> fine with -O1 and -O2 (but with larger autoexec.bin). Adding the
> "aligned" attribute to the buggy structure fixed the problem. There
> might be more issues like this, but I didn't notice anything so far.
>
> * String trimming problems seem to have disappeared as far as I could
> notice (yay!).
>
> Code is pretty much the same as with previous builds, except the
>
> Before doing further changes, I'd like to know if there are more (or
> less) side effects from upgrading the compiler, so please report any
> issues you find. I'm looking especially for issues which only happen
> with only one of last two builds (but not with the other). Thanks.
>
> P.S. This message should have arrived yesterday, but I've looked it up
> in the mailing list and it's not there, so I'm reposting it. That's
> why this build is still marked as Jun07.
>
> summon-arm
> 19KViewDownload
>
> magiclantern-2011Jun07.550d.fw109.alex.zip
> 1832KViewDownload