| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Begin to remove content from this non-sensical file. Any BSP
configuration should be in a BSP layer, not the phosphor layer.
KERNEL_EXTRA_ARGS: should be set by BSP, and this isn't required at
all at the moment by the currently supported BSPs.
UBOOT_SUFFIX: should be set by a BSP or distro configuration.
MACHINEOVERRIDES: this was nonsensical in the bitbake sense - there
is no such thing as an openbmc machine. If there is a need for
an openbmc override in the future it can be added as a distro
override (there is no such need at the moment).
MACHINE_FEATURES_BACKFILL_CONSIDERED: BSPs should be considering
backfilled machine features.
Tested: Built witherspoon and palmetto images
Change-Id: Idf252c92cb718d076ea556f2b389f15c13716fd2
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
|
| |
Install u-boot-fw-utils via dependency, rather than explicitly
attempting to include it in all images.
Tested: Built witherspoon, qemux86 images
Change-Id: I371cb4b848d67d75816a6d0d3b35bc7615f2ec39
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
For most incremental builds, the critical path is generating a new
installed obmc-phosphor-image fs, then compressing it to put into a
.lzma.u-boot container. However, the latter step doesn't seem necessary
except for the out-of-kernel initrd (obmc-phosphor-initramfs) because
obmc-phosphor-image is packaged as a squashfs-xz and not seen by U-Boot.
qemu.inc has its own IMAGE_FSTYPE appends, so qemuarm builds should not
be affected.
This reduces incremental build time by about a third, when doing
'bitbake obmc-phosphor-image' with no changes.
before:
150.74s user 66.01s system 148% cpu 2:25.96 total
after:
94.78s user 64.87s system 159% cpu 1:39.97 total
This reflects the removal of the single-threaded, CPU-bound 'lzma'
invocation from the critical path.
Netbootable images can still by generated by adding:
IMAGE_FSTYPES += "cpio.${INITRAMFS_CTYPE}.u-boot"
to local.conf.
Signed-off-by: Xo Wang <xow@google.com>
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
Change-Id: I867500589ffe0a36cbe1193843907d7882239227
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
As with MRW the use of the UBI image scheme is distro policy.
Convert the existing machine feature to a distro feature.
Enable the new distro feature on the systems that use
it(Witherspoon).
Add a distro override and fix-up MF checks to use this override
instead for improved readability.
Tested: Built a Witherspoon image and validated image
Change-Id: I8ab03115bbfc2ecc77cff5c9eb8628903ae88051
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
| |
Change-Id: I06f511350e025144a8612c8077427a1d35cdfe10
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
|
|
|
|
|
| |
Change-Id: Ie064ea9595742b78ccb7fe8a59c923b147ebfff6
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
More refactoring in preparation for ubi based mtd images.
Promote tar file generation to a real image_fstype. This allows tar
generation to be turned on or off as needed.
Move image-overlay function to image_types_phosphor. Perhaps phosphor
is not the best name but for now we only have a handful of image types and
the generation logic is to interdependant to warrant seperate classes
for each.
Attempt to improve readability throughout.
Drop ext4 override to discourage mtd block emulation.
Change-Id: Idbdc03444549b955ef2180622d133904cb191b28
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
|
| |
Solve the problem resulting in the addition of this variable
differently, such that it can be removed and image type
selection can be selected in the standard way using IMAGE_FSTYPES.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
Change-Id: Ice94e8f278f00b8b4d344244bfb87dddeb173a23
|
|
|
|
|
|
|
|
|
|
|
|
| |
Removed a handful of unused variables.
Use upstream style conventions, which seem to be:
- pep8 for python
- tabs for shellcode
- 8 space indent everywhere else
Change-Id: I4de549ecf02472096a54234379cd7cce8c219c48
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
| |
Instead of packaging the initramfs separately in the flash image, use
the FIT image that includes kernel+fdt+initramfs.
Change-Id: I0a3315cfc635acbfaf7b603f5dca7da1ddeae1c1
Signed-off-by: Rick Altherr <raltherr@google.com>
|
|
|
|
|
|
|
|
|
| |
All the infrastructure is provided by upstream Yocto. The resulting FIT
image includes (kernel+fdt+initramfs). This can replace image-overlay's
manual inclusion of an initramfs in the flash image.
Change-Id: I6f3580c5e564539960e287a9fae3161cb425a534
Signed-off-by: Rick Altherr <raltherr@google.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Before c7248eb11c and 794bd3d4ac, flash image generation was split
between obmc-phosphor-image_types_uboot.bbclass and
image-overlay.bbclass. This was because generate_flash() relied on
calling oe_mkimage() which was also defined in
obmc-phosphor-image_types_uboot.bbclass. With that dependency removed,
all of the flash image generation can now be consolidated into
image-overlay.bbclass. This location makes more sense as the .uboot
image type is for generating U-Boot legacy ramdisks while the overlay
image type is specifically for OpenBMC and its flash layout.
obmc-phopshor-image_types_uboot.bbclass is now empty and has been
removed.
Change-Id: Idcf38d3ef6f419780a7c54c77f1195bf3b5f5667
Signed-off-by: Rick Altherr <raltherr@google.com>
|
|
|
|
|
|
|
|
|
|
| |
U-Boot has support for FIT images enabled. Yocto has been building both
cuImage and fitImage for a while now. Dropping cuImage gets us closer
to mainline Yocto. Using a FIT image clears the path for bundling the
initramfs into the kernel image.
Change-Id: I46f1e3d4196bab12b733c59efca40e9e50ca264d
Signed-off-by: Rick Altherr <raltherr@google.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The netboot image consists of the initrd and rootfs packaged together
into a single U-Boot legacy ramdisk image. The initrd is already built
as a U-Boot legacy image via INITRAMFS_IMAGE_TYPES. Similarly, the
rootfs is already built as a U-Boot legacy image via IMAGE_TYPES. In
getting rid of the redundancy, using built-in mechanisms is preferred
over custom code thus netboot image is being removed.
Note that removing the netboot image generation means the raw initramfs
CPIO archive is not explicitly needed so it has been removed from
INITRAMFS_IMAGE_TYPES. It will still be generated as the U-Boot ramdisk
generation requires it as an implicit dependency.
Change-Id: I0370ccc9575000de3238af07f7f14faa283edc15
Signed-off-by: Rick Altherr <raltherr@google.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
do_generate_flash() was manually packaging the initramfs cpio archive as
a u-boot ramdisk legacy image. Add the u-boot type to
INITRAMFS_FSTYPES to use the existing infrastructure instead.
NOTE: if you have an existing build directory, this change may cause the
build to fail while creating symlinks. This is caused by the .u-boot
file changing from an actual file to a symlink.
Change-Id: I9fa69620f97452671f89675768b1d1f33dbe97f2
Signed-off-by: Rick Altherr <raltherr@google.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
As a known issue, the introspection data cannot be built on ppc64le.
https://lists.yoctoproject.org/pipermail/yocto/2016-April/029582.html.
To work around, I disable the use of qemu when we build openbmc
by modifying the configuration files:
Add
MACHINE_FEATURES_BACKFILL_CONSIDERED = "qemu-usermode"
to the end line.
Resolve: openbmc/openbmc#640
Change-Id: Idb6584ca43a8d4f005aa365f62aa1d46d2a00fd5
Signed-off-by: Nan Li <william.bjlinan@hotmail.com>
|
|
|
|
|
| |
Change-Id: I2333097ce2a17a2e2d932384373ccbae6724a861
Signed-off-by: Rick Altherr <raltherr@google.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
sample.inc has been reduced to simply appending some
server related machine features.
For the sake of correctness, don't assume anything about
the machine features desired to be simulated by the phosphor
layer client. Accomplish this by removing sample.inc from
local.conf(.sample).
To simulate additional machine features on fake platforms
like evaluation boards or the default qemu models, append
additional features to OBMC_MACHINE_FEATURES in local.conf.
Change-Id: I8bf3e0b428a3d0c20136340720aa500e1312607e
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
| |
Move the fake, example implementations from sample.inc
to phosphor-defaults. Other layers override as needed.
Change-Id: Ic024b13c39bf1750f5c76bc0fac1d18092e8305d
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
| |
Add a distro/machine feature for Host IPMI.
Change-Id: I873e5a37236ab408890e967109f16bc1926503be
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
| |
The desired providers were previously expressed in a
virtual package. Move them to a packagegroup instead.
Change-Id: I7343b64871b655dc7f567ecb8412b6b47b791d33
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
| |
These variables don't do anything functionally at the moment
and the idea is better implemented with a packagegroup anyway.
Change-Id: I026989df7a1169d44ff1a532731c97930c95d293
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
|
|
|
| |
The value added by these classes is minimal and lowers the flexibility
available to clients using them when it comes to packaging.
Rename recipe virtuals - obmc-phosphor seems redundant.
Align package virtual names with best practices (virtual-package).
Change-Id: I34c3cb03f6b1d2aaa734ed02078b388667f70917
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
| |
The policy feature doesn't have any meaning or definition
at the moment so remove in an attempt to reduce clutter.
Change-Id: I7edaeb3bc9ffb409e5f7dfeae424ffb94f6085fd
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously, obmc-bsp-common.inc added "overlay" to the IMAGE_FSTYPES
but this only has meaning if the 'image-overlay' class is used. The
obmc-phosphor-image class explicitly includes 'image-overlay' when
appropriate but no other images (ie. ones that come with yocto) do.
Changed obmc-bsp-common to instead enable an obmc-phosphor-image feature
by setting OBMC_PHOSPHOR_IMAGE_OVERLAY=1. Then in obmc-phosphor-image we
add "overlay" to the IMAGE_FSTYPES only when the feature is enabled.
This will fix both non-phosphor images on hardware platforms and all images
on simulation platforms.
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
|
|
|
|
|
|
|
|
| |
Yocto 2.1 can't seem to find the original reference package
obmc-phosphor-userd. obmc-phosphor-user is a full implementation
that works on qemu, so we will use it.
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
|
|
|
|
|
|
|
|
| |
Yocto 2.1 can't seem to find the original reference package
obmc-phosphor-eventd. obmc-phosphor-event is a full
implementation that works on qemu, so we will use it instead.
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Prior to this patch packagegroup-core-boot was not being pulled in.
core-boot is what brings in the init_manager (systemd) but
somehow systemd was added as a dependency by OE for the skeleton package
so the missing core-boot packagegroup went unnoticed.
core-boot was not being pulled in because of an assignment to
IMAGE_INSTALL in the global configuration. Yocto core provides a
mechanism to do the same thing safely - CORE_IMAGE_EXTRA_INSTALL;
use this and also adopt a similar approach for adding extra packages
to image recipes using obmc-phosphor-image.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
| |
Build a rootfs cpio uimage with everything along with the other
images.
|
|
|
|
|
|
|
|
|
|
|
| |
This attempts to address a race condition in the build where
we attempt to build a complete flash image before all the sub-images
are available (specifically the initramfs image).
Any classes in IMAGE_CLASSES are inherited by all image recipes, including
the initramfs image. OE-Core's live/hddimg recipes address this
by pulling in the class configuration in recipe scope, rather than
global scope. Copy that behavior here.
|
|
|
|
|
|
| |
Create virtual class to include user management in the image
Add the systemd patch for network support to the systemd recipe
Update the IPMI recipe to pickup the support for network configuration
|
|
|
|
|
|
|
|
|
|
|
| |
This adds the u-boot-fw-utils package to the obmc phosphor bsp
include file to include the fw_setenv and fw_printenv utilities
in the image.
This will allow us to read and update the u-boot boot environment
including setting the mac address of the ethernet.
Signed-off-by: Milton Miller <miltonm@us.ibm.com>
|
| |
|
|\
| |
| | |
stop the build break for btbridge on qemu
|
| |
| |
| |
| | |
btbridge is not even part of the qemu image
|
|/
|
|
| |
x
|
|
|
|
|
| |
Moved ipmi up to the latest level to allow the ipmid-api.h to be exported in the package
fixed bb files to ensure RECCOMNDS do not get overwritten
|
|
|
|
|
|
| |
They can still be built.
Not included in any images.
Part 2 of 2 (meta-palmetto, meta-phosphor).
|
|
|
|
| |
Set preferred providers for palmetto and qemuarm targets.
|
|
|
|
| |
This saves almost 30 percent.
|
|
|
|
|
| |
This makes all machines use the OpenBMC flash image uboot
image fstype specialization.
|
| |
|
|
|
|
|
| |
Added obmc-phosphor-example-sdbus.
Added sdbus-service class for common logic.
|
|
|
|
|
|
| |
Renamed to obmc-phosphor-example-pydbus.
Prep for forthcoming sdbus example.
Added signal example.
|
| |
|
|
git-subtree-dir: meta-phosphor
git-subtree-mainline: 21f9b84b4b729fbd7acbd465e7a3f726e4d20f91
git-subtree-split: 93fb535853a24d4c3c42ec425ac92c9d3de13294
|