Age | Commit message (Collapse) | Author |
|
apply last stable patchset
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
With commit d37f6b5 "tslib: remove recipe" tslib was moved from oe-core
to meta-oe. Add dependency on that layer to fix:
ERROR: Required build target 'core-image-base' has no buildable providers.
Missing or unbuildable dependency chain was:
['core-image-base', 'packagegroup-base-extended', 'zaurusd', 'tslib']
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Since 4.4.67 with MTD pending patches (sharpslpaart).
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
remove PR
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
remove PR
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
remove PR
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Adds partition resizing and erasing.
Sanity check in case of 128M repartitioned clamshells.
Tested on SL-5600, SL-C860, SL-C3200
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
u-boot is not supported and was removed with commit id be6a11d.
If one wants, it is possible to flash from console with nandlogical.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Only for 128M clamshell models.
This is necessary if you want to change rootfs size over 64 MB.
It is flashed in the first 256k of the nand and overwrites the first 64k of the
Parameter Area (AdjValue, BootFlag, Version, Clock).
Version is included so verchg is not necessary.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
In corner cases, if you have repartitioned one unit with 128M flash like
SL-C860 without flashing the modified mainte_fix.bin provided i.e. by Cacko
the loader screws and the maintenance kernel does default to the standard
partitioning for 64M models.
dev: size erasesize name
mtd0: 006d0000 00020000 "Filesystem"
mtd1: 00700000 00004000 "NAND flash partition 0"
mtd2: 01e00000 00004000 "NAND flash partition 1"
mtd3: 01b00000 00004000 "NAND flash partition 2"
In any case, RO_MTD_LINE is the one ending with 1 while RW_MTD_LINE ends with 2
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
This is a copy and paste error, obviously refers to FS_RW (home).
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
The task is unnecessary after commit:
3211d43 kernel.bbclass: move in addtask kernel_link_images statement
from linux-yocto.inc
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
This approach is a little more robust than pinning specific versions from
other layers.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
This machine doesn't specify it but it expects a jffs2 image.
Add tar.gz for easy testing on removable media devices.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Somehow the configuration doesn't explicit the variable.
Restore it as it was in oe-classic and add tar.gz for easy test on removable
media devices.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Rename the variable as done in oe-core (6d969ba).
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Rename the variable as done in oe-core (6d969ba).
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
After the changes in oe-core (6d969bac) the recipes inherit staging.bbclass
have to stage in DEPLOYDIR and not in DEPLOY_DIR_IMAGE otherwise sstate can
not work.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
As seen rebuilding zaurus-installer from sstate there are issues:
WARNING: zaurus-installer-1.0-r7 do_install: Manifest ...
sstatee-control/manifest-allarch-linux-kexecboot.populate_sysroot not found?
With Recipe Specific Sysroot we can safely re-enable the disabled tasks:
just a little more time spent for populating sysroot.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Disable the creation of packages in a more robust (and elegant) way
inheriting nopackages.bbclass.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Disable the creation of packages in a more robust (and elegant) way
inheriting nopackages.bbclass.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Older 4.0 cannot be built with current gcc 6.3.0 as of Jann 2017.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
The old 4.0 kernel doesn't build with gcc 6.3.0 as of Jan 2017.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
After commit 849b67b in oe-core the variable KERNEL_OUTPUT has been replaced by
KERNEL_OUTPUT_DIR. It is worth noting the new variable contains just the path.
Modify gcw0.conf accordingly and use defaults from kernel.bbclass.
KERNEL_OUTPUT_DIR ?= "arch/${ARCH}/boot"
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
After commit 849b67b in oe-core it is possible to build
multiple types of kernel at one time.
This requires KERNEL_IMAGE_BASE_NAME and KERNEL_IMAGE_SYMLINK_NAME
must not start with KERNEL_IMAGETYPE prefix as before.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
After commit 849b67b in oe-core it is possible to build
multiple types of kernel at one time.
This requires KERNEL_IMAGE_BASE_NAME and KERNEL_IMAGE_SYMLINK_NAME
must not start with KERNEL_IMAGETYPE prefix as before.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Backport a pxa_dma pending patch from 4.5.4 stable.
Ref. http://www.spinics.net/lists/stable/msg131621.html
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Kernel's size is bloating: save some headroom for future needs.
Runtime tested on collie, poodle and corgi
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Prepare patch for upstream submission
Minor changes to defconfigs
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
On collie F2FS has been removed to fit in the max kernel size.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Using a single compressor to save as much space as possible.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Use a single compressor to squeeze a couple of KB for linux-kexecboot.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
The user doesn't need to open the keyboard to press return.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
After the rework of the locomokbd driver the key is mapped to KEY_F15.
Remap it to KEY_F5 as done by the 2.6 locomokbd hack.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Squeeze the last bits of space...
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Tested first on c7x0
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
pocketworkstation.org mirror is dead, point to sources.openembedded.org.
Note: checksum of the .gz changed but the file inside has the same one.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
The recipe is now using linux-kexecboot from meta-handheld.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|