to allow packages which assumes that LINUX26_DIR
and the /lib/modules/LINUX_VERSION use
the consistent version numbering to build.
Create a file ".linux-version" containing
the real linux version in $(PROJECT_B UILD_DIR)
to allow packages to read the correct linux version
if an uImage is to be built for arm, avr32, blackfin and sh.
PATH has therefore been set to have $(STAGING_DIR)/usr/bin
as the first item allowing linux to find mkimage.
This patch will, if an uImage is built, create a "u-boot-toóls"
directory in the PROJECT_BUILD_DIR directory.
$(STAGING_DIR)/usr/bin/mkimage is copied to this directory
and the PATH variable will now include "u-boot-tools"
If an uImage is built, then the linux build will now depend on "u-boot".
As discussed on the list.
- No sense in depending on BR2_TARGET_UBOOT when the configurations are
within a if BR2_TARGET_UBOOT conditional.
- n is default, so remove 'default n'
- clarify nested conditionals
to configure linux using the
make <board>_defconfig
You have to supply the $(BOARD_NAME) variable.
Add this info in the help text for
target/linux/Makefile.in.advanced
----------------
When you do not set BR2_PACKAGE_LINUX_USE_XCONFIG
to configure linux using "$make <XXXX>config
The choices "$make menuconfig" or "$make xconfig"
should not be visible.
Add dependency
Provide link using "u-boot.bin"
Add BR2_TARGET_UBOOT_DEFAULT_ENV config
If this is set, Buildroot will generate
a default environment to allow easy downloading
of the kernel and root file system
Both as compile time data and an autoscript.
Make u-boot depend on $(U_BOOT_TARGETS)
Add "u-boot-autoscript" to U_BOOT_TARGETS if
BR2_TARGET_UBOOT_DEFAULT_ENV is set.
Take kernel name from Linux build.
Use correct ARCH when mkimaging the autoscript.
Do not use TARGET_ATMEL_COPYTO for generic u-boot.
Hide dataflash memory configurations when
different boot memory is used.
More u-boot build unbreaking. r24646 added U_BOOT_TARGET which uses the
nonexisting BOARDNAME variable, which means that the filename expands
to -u-boot-*, confusing ln.
There's limited point in having that symlink in the first please, so simply
revert that hunk.
git-svn (and git) doesn't handle empty directories, so add .empty files
to those dirs like elsewhere in buildroot.
Those empty directories are normally not a big deal, but the recent changes
to u-boot broke the build.
------------------------------------------
Replace $(BOARDNAME)_PATH with BOARD_PATH
Set U-Boot targets for 1.2.0-atmel as well as 2009.01-rc1
Remove hardwiring of BusyBox configuration files.
Remove older BusyBox configuration files
Remove "at91sam9260dfc-linux-2.6.28-rc6.config" from at91sam9g20dfc directory
Update "at91rm9200df_defconfig"
Add "at91rm9200df-linux-2.6.28.config"
Update "at91sam9260dfc_defconfig"
mkimage the autoscript to autoscript.$(PROJECT)-
Move build of u-boot-1.2.0-atmel to target/u-boot
Only visible for AT91 targets.
Fetch prepatched 1.2.0-atmel tarball from $(BR2_ATMEL_MIRROR)
Rest from normal location.
Introduce U_BOOT_CONFIGURE_OPTS
Set to NOSOFT_FLOAT for u-boot-1.2.0-atmel
Disable target/device/Atmel/u-boot
Keep files in target/device/Atmel/u-boot during a test period.
Files:
M target/device/Atmel/Makefile.in
M target/device/Atmel/Config.in
M target/u-boot/Makefile.in
M target/u-boot/Config.in
A target/u-boot/1.2.0-atmel
msh is currently the only good choice for uClinux but it doesn't
understands shell functions. In this case there is nothing to gain
by using shell functions anyway.
Signed-off-by: Nicolas Pitre <nico@cam.org>
Currently, BR2_TARGET_ROOTFS_INITRAMFS doesn't actually do anything,
as it doesn't alter $(TARGETS)
This change hooks up the TARGETS variable to INITRAMFS_TARGET (which
is only set if BR2_TARGET_ROOTFS_INITRAMFS=y), so that the initramfs
list is built.
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
cc1: warnings being treated as errors
buildroot/buildroot/build_powerpc/makedevs/makedevs.c: In function main:
buildroot/buildroot/build_powerpc/makedevs/makedevs.c:531: error:
ignoring return value of system, declared with attribute warn_unused_result
This change adds a simple warning if system() fails.
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
While mkcramfs was being patched to support a device table, the target device table was not being passed to the mkcramfs options. This patch fixes that
Signed-off-by: Jared Holzman <j.holzman@genesysdesign.com.au>
This patch will make the installation of modules rule depend on .depend_done instead of .configured to make sure make prepare is run before modules are installed.
Make kernelversion does not work before make prepare has been run.
Signed-off-by: Hans-Christian Egtvedt <hans-christian.egtvedt@atmel.com>
This patch prevents the user from select "linux (Same version as linux
headers)" as a choice for building the kernel when an external binary
toolchain is used, since "same version as linux headers" doesn't make
sense when an external toolchain is used.
It fixes the issue encountered by Hartley <hartleys@visionengravers.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
This patch will check if CONFIG_MODULES is set in the kernel .config, if not
yes then it will skip installing the modules.
Signed-off-by: Hans-Christian Egtvedt <hans-christian.egtvedt@atmel.com>
This patch will ask the kernel for the specific version before doing
modules_install, that way Buildroot will know which directories to remove and
what parameters to use for depmod. Needed when compiling a kernel grabbed from
GIT or vendor specific releases.
Signed-off-by: Hans-Christian Egtvedt <hans-christian.egtvedt@atmel.com>
From: Grant Likely <grant.likely@secretlab.ca>
The Xilinx UART Lite and MPC5200 Programmable Serial Controller (PSC)
devices have dedicated major/minor numbers. Add them to the device
table data file so they get are present in the root filesystem
This patch fixes bug report #5204. The Busybox config was not copied from the
correct place.
Signed-off-by: Hans-Christian Egtvedt <hans-christian.egtvedt@atmel.com>
Patch by Markus Heidelberg <markus.heidelberg@web.de>
* remove needless help in invisible config options
* correct wrong indenting
* remove unused BR2 variables for default pagesize and devfile
Patch by Markus Heidelberg <markus.heidelberg@web.de>
Some $(call insert_define, ..., ...) calls were lacking the comma after
the variable. That didn't lead to a syntax error but it didn't have any
effect on the output u-boot/include/config.h
U_BOOT_ARCH seems to race with KERNEL_ARCH in some situations,
so it end up empty. It's only used once, so fix it by using a
use-time evaluated variable instead (=).
