| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
|
|
|
|
|
|
|
| |
This is much faster than the current introspection based
mechanism.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
Change-Id: I1a737cab00311e9e0c0b63f4d8a487085cdf44e0
|
|
|
|
|
|
|
|
| |
Remove the busname field from the 'GetObject' method responses.
Applications should get the busname information from the mapper.
Change-Id: Id548f3d5d6afda8334ad1163d66f69832591511a
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
|
| |
This is a reaction to a /org/openbmc/managers/System API update.
The update removes the bus name field from the 'GetObjectByxxx'
method responses.
Change-Id: I061f8071097b4e870031198a509abe48a4128d82
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
| |
Remove a reference to 'BMC_INIT' state which isn't used.
Remove heartbeat code that isn't used.
Change-Id: I2de0c881372085519a15235703881054c5a9adfa
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
| |
Change-Id: I976228399027420f255f09622b02352e1799fe91
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
| |
The download path for temporary files isn't a property
of the board. Hardcode /tmp into the application.
Change-Id: Ib375ddee9a4c50aa8817980562fbd5b1191b5542
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
|
|
| |
Remove ObjectManager signals emitted before a bus name is owned.
This isn't necessary now that the system_manager can detect objects via
NameOwnerChanged.
Change-Id: Ia2b626a926392b50d20c26474e6c9a9fdb3ca2f0
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
|
| |
The default behavior in pyphosphor changed such that ObjectManager
and Properties signals are not emitted by default. Enable signals
just prior to bus name grab.
Change-Id: I27a2de4b60482dd843702396f33c7cdccaf17277
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
|
| |
Avoid race conditions where applications expecting an object to be
present upon recipt of NameOwnerChanged find that the object hasn't
been created yet.
Change-Id: I4236a3898055a84b00e1eaf37f32e0192b683a0f
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
|
| |
Add a bus name handler so the system manager can detect
objects implemented by services that create objects before
owning a bus name.
Change-Id: Ic311133a8e532e7f2441102689632f5821dbf589
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
| |
Add a re-usable try_next_state function.
A handful of idiomatic python improvements.
Change-Id: I435aaaeb5760a677f8fd292f396cdbdae1195674
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
| |
Change-Id: I0c795151f723fdfa41aa24d9213ffb8490b03349
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
| |
To be consistant with the other gdbus app implementations.
Change-Id: Ib427a61c4ef9ae0cf593aec1deafd9d3e7aac586
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
|
|
|
|
|
|
|
| |
Firestone, Garrison, and Witherspoon were missing the checkstop
gpio handling. Add the checkstop handler process and the gpio
pin number per the system schematics.
Change-Id: Id53e2f8888a7c706bf30a95fe56223628a2ba70a
Signed-off-by: Adriana Kobylak <anoo@us.ibm.com>
|
|\
| |
| | |
Flash bmc updates
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This commit adds methods to Validate (not exposed), Apply,
GetUpdateProgress, and Abort along with an auto_apply attribute.
A PrepareForUpdate method is also added that will write bootloader
environment to instruct the initramfs init script to copy necessary
file system state into RAM to allow the flash to be updated.
Together these methods take the prior Update method and chain
on the Validate phase which checks the image can be applied
concurrently (and could be extended to check signatures). If
validation fails auto_apply is cleared. If auto_apply remains
true then Apply is called which will perform the check again
(implicitly in the update script) and update the flash contents,
saving its output to a temporary file.
The GetUpdateProgress will return the status variable followed
by the output from the update script, with status that was
overwritten with a carriage return stripped. As each new phase
(erase, write, and verify) of each image is encountered the
message will be extended.
An Abort method will kill an in-progress update, erase the logfile,
and remove any pending images. A new image can be unpacked at this
time.
Note: the current method to clear all persistent files removes the
whitelist and all future calls will clear all files from the read
write layer. This maybe addressed in a future update.
Signed-off-by: Milton Miller <miltonm@us.ibm.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The whitelist is moveing from a file to a set of files in a directory.
Use try when deleting whitelist entries.
It would be better to call update without saving the files in the
first place but this will suffice for now.
Signed-off-by: Milton Miller <miltonm@us.ibm.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Copy the version saved by our initfs and pointed to by our fw_setenv
tools via /etc/fw_setenv. Do this copy by default.
The /etc/fw_env.cnf config file points to /run/fw_env in our systems.
The bmc_update object was copying /dev/mtd2. That could be very
system specific, and it misses any changes made from elsewhere
in the system.
This variable causes the u-boot environment to be written back
after writing the whole image. This will save the ethernet mac
which for systems which don't store it elsewhere.
Signed-off-by: Milton Miller <miltonm@us.ibm.com>
|
|\ \
| | |
| | | |
Disable watchdog timer if debug_mode is set
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
resolves openbmc/openbmc#246
debug_mode is used when the host is not going to be started.
In this case, we do not want the watchdog timer to be started
because there is no one to keep it from resetting.
This code change prevents the watchdog from being started
during poweron.
Potential side affects could be if we do start the host
in this code path. The watchdog is un-tested when it's
not started but software comes in and tries to reset it.
The code in control_host_obj.c currently enforces the
no-host-power-on when debug_mode is set.
|
| | |
| | |
| | |
| | | |
This commit is formatting only, no functional changes.
|
|\ \ \
| | | |
| | | | |
Remove Chassis and Product area from IO Board
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Remove Product and Chassis area for IO board vpd
since only board area is present in the latest
version of the IO eeprom.
Signed-off-by: Adriana Kobylak <anoo@us.ibm.com>
|
|\ \ \ \
| |_|_|/
|/| | | |
hwmon: attribute is not polled after failure
|
|/ / /
| | |
| | |
| | |
| | |
| | |
| | | |
This line of code was removed with revert bd01c24d (original efc68970)
but should not have been. The line was originally added with 6956bbd4.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|\ \ \
| |_|/
|/| | |
Add the missing binary into Palmetto config file
|
|/ /
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Palmetto.py did not have discover_system_state.py and this patch is
to update that. discover_system_state.py is responsible for looking
at power restoration policy set by the user and then taking appropriate
actions when BMC comes back from any reboot / power cycle and reaches
BMC_READY state. Actions could be :
1) Leave the system in power off state.
2) Power on the box.
3) Restore the last state which can be either #1 or #2 above.
|
|\ \
| |/
|/| |
GCC 5 build fixes
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Fixes the following error:
host_checkstop_obj.c: In function ‘is_host_booted’:
host_checkstop_obj.c:46:10: error: implicit declaration of function
‘strcmp’ [-Werror=implicit-function-declaration]
if ((strcmp(system_state, "HOST_BOOTED") == 0) ||
Signed-off-by: Joel Stanley <joel@jms.id.au>
|
|/
|
|
|
|
|
|
|
|
| |
Fixes the following error:
In file included from button_reset_obj.c:4:0:
../libopenbmc_intf/openbmc.h:39:15: error: return type defaults to ‘int’ [-Werror=implicit-int]
static inline devmem(uint32_t val, uint32_t reg)
Signed-off-by: Joel Stanley <joel@jms.id.au>
|
|\
| |
| | |
More skeleton refactoring
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This patch structures the make rules such that building from the top
will build the in-tree libopenbmc_intf and link to that. For example:
$make
$make bmcctl
would both result in ../libopenbmc_intf being built and the applications
will link to that.
Alternatively when building applications directly like:
$make -C bmcctl
the default compiler/linker search paths will be used.
The latter facilitates package managers and the former retains a level
of flexibility for developers working from the repository.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
| |
| |
| |
| |
| |
| |
| | |
Address an issue where the targets specified in the BINS variable
are rebuilt every time.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Don't use rules.mk for libopenbmc_intf. There isn't enough
overlap here to warrant the extra complexity.
Fix an issue where the library is rebuilt for every make invocation.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Create a libopenbmc_intf symlink in the build target to
enable things like:
-L ../libopenbmc_intf -lopenbmc_intf
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
| |
| |
| |
| |
| |
| |
| | |
Update codegen to reflect new path xml/output locations.
Run updated codegen.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
| |
| |
| |
| |
| |
| | |
To the libopenbmc_intf directory.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Remove the interfaces prefix from applications including openbmc_intf.h.
Use brackets rather than quotes in prep for a libopenbmc_intf SDK
package.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
| |
| |
| |
| |
| |
| | |
This allows 'make install' to be invoked more than once.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|/
|
|
|
|
|
| |
Install the libopenbmc_intf includes in prep for a devel
package.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|\
| |
| | |
dbus.Double and revert sensor cache
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
At some point a floating point value was added to a sensor
object.
python-json doesn't handle the dbus wrapper types like
dbus.Double. Cast dbus.Double to a float, similar to integers.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This reverts commit efc689707d21f36c0ed8f0b1af474d2087ef8f95.
This enhancement breaks the ability to set the sensor value
via user-interface.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|/
|
|
|
|
| |
This reverts commit 8224ff87c914861d28ed0dc6a251ba213a9ee607.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|\
| |
| | |
rules.mk: Fix parallel build for 'make install'
|
| |
| |
| |
| |
| |
| | |
install-bins was missing a dependency on the binaries themselves.
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
|
|\ \
| |/
|/| |
remove org.openbmc.Object.Enumerate interface
|
| |
| |
| |
| |
| |
| | |
This is the result of running gdbus-codegen with the latest XML.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
| |
| |
| |
| |
| |
| |
| | |
All applications are using freedesktop ObjectManager now so this
can be removed.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
| |
| |
| |
| |
| |
| |
| | |
The org.openbmc.Object.Enumerate interface is deprecated in favor
of the freedesktop ObjectManager.
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|