Age | Commit message (Collapse) | Author | Files | Lines |
|
Adopt a more conventional directory hierarchy. meta-phosphor is still
a _long_ way from suitable for hosting on yoctoproject.org but things
like this don't help.
(From meta-phosphor rev: 471cfcefa74b8c7ceb704cb670e6d915cf27c63b)
Change-Id: I3f106b2f6cdc6cec734be28a6090800546f362eb
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
Spelling errors found using github.com/lucasdemarchi/codespell
A tool to fix common misspellings.
This tool is licensed under GNU General Public License, version 2.
(From meta-phosphor rev: 38f27112a9bc38337d8f4c63c333d2a1f28945fa)
Change-Id: I834b99d63a260f21c60b58c91ec328cf04b683fe
Signed-off-by: Gunnar Mills <gmills@us.ibm.com>
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
Update external subtrees to latest Yocto sumo.
Change-Id: I8364f32bef079841c6e57f1c587f4b1bedf62fef
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
Phosphor specific overrides to systemd should be stored
within its own recipe and pulled in via the phosphor layer.
This helps distinguish systemd specific overrides with
phosphor overrides.
Tested: Verified service-restart-policy.conf was still
pulled in as expected in witherspoon QEMU
Change-Id: I32110fc2edd3abd2bb0008c63dd4ad7828b762b8
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
Upstream busybox has a change to ignore the -c option, which systemd
uses when unmounting during shutdown/reboot. Currently, if -c is used,
umount fails. This results in extensive umount failures during shutdown.
See http://lists.busybox.net/pipermail/busybox/2018-January/086045.html
Change-Id: I4630935dc5ec61263a7af9e24b70ca8f1f8b5359
Signed-off-by: Eddie James <eajames@us.ibm.com>
|
|
Tested: Performed system boots and reboots.
Change-Id: I1052cb4cd86a5815df1f1021a37f539184078d82
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
Move OpenBMC to Yocto 2.5(sumo)
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
Change-Id: I5c5ad6904a16e14c1c397f0baf10c9d465594a78
|
|
1e4452449 added a systemd tmpfile config file for the software image
upload path but the config file was added to the systemd recipe.
tmpfile config files should be added to the recipes/config files that
use them. For example on an RPM based distro:
$ rpm -qf /usr/lib/tmpfiles.d/samba.conf
samba-common-4.7.7-0.fc27.noarch
Tested: Validated software.conf present in Witherspoon rootfs
Change-Id: I47e8c8c44eb19440cc98d0380e9634f688eccb21
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
Somewhere between systemd 234 and 237 the install process stopped making
/usr/lib/systemd/network for us. When we move to Yocto 2.5 this will
result in an install failure.
bitbake.conf provides a variable for /lib/systemd. Use it, as it's the
correct location for these configuration/unit files anyhow.
Tested: Booted QEMU witherspoon image to shell
Change-Id: Ia863473539378ef07f652ae3820b9c22ce5f6900
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
This reverts commit bbf2cb8e1e4d490acf80ce3a5b162d7db5f3e4e2.
This change doesn't seem to be required anymore.
Tested: Booted Palmetto system in QEMU
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
Change-Id: I34b3dd17d1fe7427ec12f84c6ad9bb5367f73cf6
|
|
These were removed from SRC_URI with the bump to Yocto 2.4, which bumps
systemd to v234, which contains said patches.
Change-Id: If0584df10a028b444e49ca0fb7d7a9ca2e66dc3e
Fixes: d7bf8c17eca8 ("Yocto 2.4")
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
|
|
The BMC CPU performance is already challenged. When a service is
failing and a core dump is being generated and collected into a dump,
it's even more challenged. Recent failures have shown situations where
the service does not fail again until 15-20 seconds after the initial
failure which means the default of 10s for this results in the service
being restarted indefinitely. Change this to 30s to only allow a
service to be restarted StartLimitBurst times within a 30s interval
before being put in a fail state.
Testing: Verified that killing an application 3 times within 30 seconds
results in the service being in a fail state (and not restarted).
Resolves openbmc/openbmc#3131
Change-Id: I58c874c1c5b618c147946195af65f42371d683c8
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
The phosphor-logging interface has been updated to call a local interface to
sync the journal instead of making a D-Bus call:
https://gerrit.openbmc-project.xyz/#/c/9910/
Delete the journald patches since they're not needed anymore.
Tested: Built witherspoon image.
Signed-off-by: Adriana Kobylak <anoo@us.ibm.com>
Change-Id: Ie6b4387bf53ea4c29a7bbef6483ef723c333dfdf
|
|
upstream: https://github.com/systemd/systemd/pull/7352
Change-Id: I62df65eeec786890c0a7b923fd0455f7869f38b9
Signed-off-by: Edward A. James <eajames@us.ibm.com>
|
|
Upstream busybox has a change to ignore the -c option, which systemd
uses when unmounting during shutdown/reboot. Currently, if -c is used,
umount fails. This results in extensive umount failures during shutdown.
See http://lists.busybox.net/pipermail/busybox/2018-January/086045.html
Resolves openbmc/openbmc#2511
Change-Id: Ic32628bf22a134012c61cb817479a86d5187f9e9
Signed-off-by: Eddie James <eajames@us.ibm.com>
|
|
Requirements have arisen for OpenBMC firmware to run
certain services only during fresh power on operations.
To achieve that, break the obmc-host-start action target into
two targets. The existing obmc-host-start target will now
call the new obmc-host-startmin target. The startmin target
will be the minimum services required to start the host.
The obmc-host-start target is where services that should
only be called during a fresh power on can be placed.
The initial use case for this is the obmc-host-reboot target.
It will call this new obmc-host-startmin target to do the bare
minimum on a reboot operation.
This then allows the use case for OpenBMC to only reset the host
reboot count on a fresh power on operation. A new service will
be added to the obmc-host-start target to do this. If a reboot
operation is issued, or a host watchdog is triggered to cause a
reboot, this service would not be started, ensuring the reboot
count is properly decremented.
Tested: Verified on/off/reboot and ran CT regression suite.
https://gerrit.openbmc-project.xyz/#/c/9821/ is required
for reboots to work properly.
Change-Id: I86df03e6d671178c1525f852e196c4102bb73cb6
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
There were a number of places that inherited this class but did not
use it. A number of classes that did use it did not inherit it.
This was possible because the class was pulled in by the distro
configuration but that has been removed here as well.
Tested: Built and verified a witherspoon image
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
Change-Id: I73511cdcc609875b4fe016967c1b84c5200fb703
|
|
USE_LDCONFIG has been deprecated in favor of the ldconfig distro
variable, which we already don't use. Drop bbappend.
Change-Id: I343427c1c8fa012e4847c0cd340859506d3a2e54
Tested: Built a witherspoon image
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
Move OpenBMC to Yocto 2.4(rocko)
Tested: Built and verified Witherspoon and Palmetto images
Change-Id: I12057b18610d6fb0e6903c60213690301e9b0c67
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
Move OpenBMC to Yocto 2.3(pyro).
Tested: Built and verified Witherspoon and Palmetto images
Change-Id: I50744030e771f4850afc2a93a10d3507e76d36bc
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
Resolves: openbmc/openbmc#2461
|
|
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>
|
|
All the fan services associated with controlling, monitoring and
presence detection of fans need to remain active until a poweroff
completes. This helps ensure thermal safety while a system is in
transition to a powered off state.
With the introduction of this new target (which runs last in
the power off path), need to move the host reset service to
running after it to ensure the power on in the reboot path
does not start until the last power off target runs.
Resolves openbmc/openbmc#2762
Change-Id: I4ca671d2c7de66e8bfa4bba607b06c88c0a6e7f9
Signed-off-by: Matthew Barth <msbarth@us.ibm.com>
|
|
We carry a systemd patch that adds a new systemd configure option
--enable-journald-dbus. Add it to PACKAGECONFIG and select it.
Previously the patch selected the option by default.
Add new files that are installed by the patch to FILES_${PN} to
avoid unpackaged files errors from bitbake due to upstream systemd
recipe changes in Rocko.
Tested: Built and verified witherspoon and palmetto images
Change-Id: Iaf72c2b039706287a00126035eafe2f8631ca473
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
On an ssh connection, we are getting Warning messages similar to below:
dropbear[3956]: Failed loading /etc/dropbear/dropbear_rsa_host_key
dropbear[3956]: Failed loading /etc/dropbear/dropbear_dss_host_key
dropbear[3956]: Failed loading /etc/dropbear/dropbear_ecdsa_host_key
This is because dropbearkey.service creates the RSA host key in
/var/lib/dropbear and does not create any DSS and ECDSA host keys.
And the dropbear service is started with the -r option which
points to /var/lib/dropbear/dropbear_rsa_host_key as the key to
load. However, dropbear will attempt to load any key specified
by the -r option as well as all 3 keys from the default path
/etc/dropbear. This is a change to dropbear to not load the 3
keys from /etc/dropbear by default if a key and path is specified
by the dropbear -r option. This will get rid of the above Warning
messages which can fill up the log buffer in a Continuous Test
environment where many ssh connections are made.
This change has been upstreamed.
Resolves openbmc/openbmc#1340
Resolves openbmc/openbmc#1998
Change-Id: I58f8290f68a6eba7e3a77986a8ecb0ebdf321352
Signed-off-by: CamVan Nguyen <ctnguyen@us.ibm.com>
|
|
On each ssh connection, we are getting a Warning message similar to below:
dropbear[3956]: Failed loading /var/lib/dropbear/dropbear_rsa_host_key
This is because the dropbear service is started with the -r option which
points to /var/lib/dropbear/dropbear_rsa_host_key as the RSA host key to
load. However, the dropbearkey.service creates the key in /etc/dropbear
instead. There is an environment file, /etc/default/dropbear, which
contains the path to the RSA host key. This path is set by
rootfs-postcommands.bbclass. At build time, the .bbclass file checks if
the /etc/dropbear directory exists and if the RSA host key exists in
this directory. If the key does exist it sets the path to /etc/dropbear
else it sets it to /var/lib/dropbear. The dropbear service reads this
environment file to determine from which path to load the RSA host key.
This fix is to change dropbearkey.service to have similar logic to read
the file to determine which path to create the RSA host key. This will
get rid of the above Warning message, which can fill up the log buffer
in a Continuous Test environment where many ssh connection are made.
Change-Id: Iae37a3dfa9aa8c56687560f0f6560114c1e9b85a
Signed-off-by: CamVan Nguyen <ctnguyen@us.ibm.com>
|
|
This saves the download and building of the following packages:
libxslt
xmlto
docbook-xml-dtd4
docbook-xsl-stylesheets
Change-Id: Ib2488375f642ce72e18c57ee6b77cdb71251d7c6
Signed-off-by: Joel Stanley <joel@jms.id.au>
|
|
Required by perf(1), which is now in the debug tools tarball.
expand(1) expands /bin/busybox.nosuid by 468 bytes and adds one symlink
to the filesystem:
$ size a/bin/busybox.nosuid
text data bss dec hex filename
582385 1845 8816 593046 90c96 a/bin/busybox.nosuid
$ size b/bin/busybox.nosuid
text data bss dec hex filename
582853 1845 8816 593514 90e6a b/bin/busybox.nosuid
Resolves: openbmc/openbmc#2879
Change-Id: I999a12768ab7e0d740bed710b01dab49aa2a8019
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
|
|
The current uuid generation method just export the machine-id,
it does not match the systemd description
(https://www.freedesktop.org/software/systemd/man/machine-id.html):
This ID uniquely identifies the host. It should be considered
"confidential", and must not be exposed in untrusted environments,
in particular on the network. If a stable unique identifier that is
tied to the machine is needed for some application, the machine ID
or any part of it must not be used directly.
Backport the new sd_id128_get_machine_app_specific from systemd 233,
to replace the old uuid generation method.
Change-Id: I44e7b32ca951a9fa591761c3a3e4f27da04d2b6f
Signed-off-by: Yong Li <yong.b.li@linux.intel.com>
|
|
Further testing has shown that with core files enabled,
it's possible to exceed the 10s default interval time
when doing 3 restarts. Since all testing has shown
either 1 restart fixing the issue, or no restarts fixing
the issue, reduce the restart limit from 3 to 2 to
ensure even if the service is core dumping, the restart
limit will be hit within the 10s.
Resolves openbmc/openbmc#2559
Change-Id: Ie1dac96eb1b4ced0bf166c12419db26924104ab4
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
/tmp is deleted every 10 days
/usr/lib/tmpfiles.d/tmp.conf does this cleanup with this entry,
"q /tmp 1777 root root 10d".
Exclude /tmp/images from the cleanup, by adding "x /tmp/images"
to /usr/lib/tmpfiles.d/software.conf, if a feature of the machine is
obmc-ubi-fs or openpower-ubi-fs.
Resolves openbmc/openbmc#2544
Change-Id: I3ead88724adcd63ebfd809d310109a19232b1def
Signed-off-by: Gunnar Mills <gmills@us.ibm.com>
|
|
With this patch, the build system supports the tools-profile feature,
We can add it into the EXTRA_IMAGE_FEATURES in conf/local.conf,
to add more debug/profile tools into the image
Change-Id: Ic3c3811d9bca3f99a9e94d7c9da5a71ce1321d02
Signed-off-by: Yong Li <yong.b.li@linux.intel.com>
|
|
Change-Id: Icd951d2ed4075b7b04393157b3c420ef644b3bc5
Signed-off-by: Gunnar Mills <gmills@us.ibm.com>
|
|
Systemd requires /etc mounted and ready at init time. Previously, we
were mounting /etc as an overlay from the persistant filesystem after
systemd started. This caused a variety of issues with systemd and
associated applications (journal, etc).
This change also adds factory reset functionality.
Resolves openbmc/openbmc#2542
Change-Id: Ib8a38e99df39f833026fe4475670a7ec2b9c8d4d
Signed-off-by: Edward A. James <eajames@us.ibm.com>
|
|
The service was getting called twice, resulting in a removal of the
overlayfs workdir while the overlay was mounted. Solve this by checking
if the /etc overlay is mounted yet before removing the workdir. Need to
add mountpoint to busybox for this.
Resolves openbmc/openbmc#2429
Change-Id: I407a07dcc3f3856af12a7f3a52966fe7c3d756c3
Signed-off-by: Edward A. James <eajames@us.ibm.com>
|
|
This reverts commit 2a284a012d902bfa399bc14309337c3c1fe1a8ff.
Change-Id: I88f1cc9275d9dd1c39fda7e4fe218941fd5ddd42
Signed-off-by: Jayanth Othayoth <ojayanth@in.ibm.com>
|
|
Install elog-gen.py and mako script by default in the
obmc-phosphor-image sdk.
Change-Id: Ibfb94fc4a645bcca66f7b93656edd88a1af5c4f6
Signed-off-by: Marri Devender Rao <devenrao@in.ibm.com>
|
|
This commit adds a service file requirement to fstab for mounting
the ubi0:rwfs volume to /var. The required file is
obmc-flash-bmc-reset.service, which, when called, will perform a
BMC factory reset if the corresponding environment variable is set.
This functionality was moved out of obmc-flash-bmc-ubirw.service because
the factory reset is not the only procedure that calls that service.
Moving this function to a new service avoids unexpected behaviors.
Resolves openbmc/openbmc#2326
Change-Id: If2a9c650f315cff23040194ec87f16c05622c85e
Signed-off-by: Michael Tritz <mtritz@us.ibm.com>
|
|
Reduce number of service restarts and increase wait
time between restarts.
- Change the StartLimitBurst to 3
Five just seems excessive for our services in openbmc. In all fail
scenarios seen so far (other then with phosphor-hwmon), either
restarting once does the job or restarting all 5 times does not help
and we just end up hitting the 5 limit anyway.
- Change the RestartSec from 100ms to 1s.
When a service hits a failure, our new debug collection service kicks
in. When a core file is involved, it's been found that generating 5 core
files within ~500ms puts a huge strain on the BMC. Also, if the bmc is
going to get a fix on a restart of a service, the more time the better
(think retries on device driver scenarios).
Change-Id: I0e5bb3f32022cdb72d00f1a861a69653ef937cf1
Signed-off-by: Andrew Geissler <andrewg@us.ibm.com>
|
|
This reverts commit 3b900a42acf55606bf28d7e83cabe1a545e9178d.
Change-Id: Ia19c92e6724068e84989f62d1673c9effb287f83
Signed-off-by: Jayanth Othayoth <ojayanth@in.ibm.com>
|
|
Need /home to be writable and persistent.
Resolves openbmc/openbmc#2250
Change-Id: I1b3477945b3364a9a5053fd085ff8fd591ad6949
Signed-off-by: Eddie James <eajames@us.ibm.com>
|
|
systemd-tmpfiles-setup.service tries to create a temporary
dir /srv but due to the read-only root file system it was
failing.
Resolves openbmc/openbmc#2231
Change-Id: I429b16a7736959583f24bf7fb4c6766cda24130a
Signed-off-by: Saqib Khan <khansa@us.ibm.com>
|
|
/media will be a tmpfs.
This is done so that /media is writeable in a
read-only filesystem to be able to mount the
ubi volumes for the systems that implement the
software interfaces.
Resolves openbmc/openbmc#2072
Change-Id: Id6163fe43843f9d17ef49760dee1caee7cc17ceb
Signed-off-by: Leonel Gonzalez <lgonzalez@us.ibm.com>
|
|
Append base-files to add new fstab and depend on mount-overlay recipe.
Add systemd services to help mounting overlay fs.
Resolves openbmc/openbmc#2134
Change-Id: Ib56fcdde9f7364fca56a08e1ce72e7cafaac35b0
Signed-off-by: Edward A. James <eajames@us.ibm.com>
|
|
Resolves openbmc/openbmc#1938
This reverts commit 45af9a22aabf38474a00698f511e334c14d84c76.
Change-Id: I482d8b14ce1cfcb9022973a9412adff46f458f5f
Signed-off-by: Charles Paul Hofer <Charles.Hofer@ibm.com>
|
|
Package includes systemd tools for storing and retrieving
coredumps.
Change-Id: I7d71b9a0f055213c6293c3cbd1af26a662e87e2a
Signed-off-by: Jayanth Othayoth <ojayanth@in.ibm.com>
|
|
Currently building causes a warning about a postinstall hook that is
unable to run, due to needing qemu, which causes the postinstall hook to
be delayed until first boot. When moving to a read-only-rootfs, we
cannot have delayed postinstall actions (otherwise bitbake fails).
The glib package does not actually have any GIO modules that need the
postinstall action executed for, so remove it entirely via bbappend.
This is not appropriate to do upstream for the glib package because,
depending on glib options, this may become required. We only use glib
for the python dbus libraries, so we are unlikely to ever be affected
by this. (Even if we are it just causes a small performance hit on
process start-up due to a gio cache not being populated.)
Change-Id: I217d4f21be3e2af20895b51267830f4fda376d2c
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
|
|
The Synchronize method would be called when an error log is
committed to get all the entries flushed into the journal.
These patches have been submitted upstream and can be removed
once the changes are integrated into systemd.
Change-Id: I89059d438870803b41bafdb4de3fb11eba3c5fd2
Signed-off-by: Adriana Kobylak <anoo@us.ibm.com>
|
|
The state management code on the BMC is designed to
utilize systemd targets as much as possible. The host
reboot was always the odd duck that had some special
code written to support it. This special code would
basically issue a power off, look for it to complete
and then issue a power on. This is very error prone
code and has resulted in a few issues.
Moving to a single systemd target (the new reboot target)
defined here will allow the host reboots to occur
solely through a single systemd target now.
openbmc/openbmc#2032 is one of those bugs being
addressed with this change.
Change-Id: If33454aa4f15580e45f888a9b1602dec24a46fca
Signed-off-by: Andrew Geissler <andrewg@us.ibm.com>
|
|
Put os-release in the sysroot so it can be accessed by other
recipes.
Change-Id: Id97d44bcec0272d6b7dc29831a056e419579adc9
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
Update the os-release recipe such that it is not rebuilt unless
git reports something different.
To do this both the recipe and sstate caches must be disabled
but the impact with such a small recipe is low enough to be
acceptable.
Resolves: openbmc/openbmc#124
Change-Id: I53343410d973ab4a60d9437704c34dbec7120ede
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|