In the past, I had thought it didn't make sense to have multiple
Action classes with identical triggers within ActionManager::actions_,
and opted to instead combine these into a single action. In theory,
it should reduce memory overhead as only one copy of the triggers
needs to be stored.
In practice, this ends up not being a good idea.
Most importantly, given a file with the below three sections in this
same order:
on boot
setprop a b
on boot && property:true=true
setprop c d
on boot
setprop e f
Assuming that property 'true' == 'true', when the `boot` event
happens, the order of the setprop commands will actually be:
setprop a b
setprop e f
setprop c d
instead of the more intuitive order of:
setprop a b
setprop c d
setprop e f
This is a mistake and this CL fixes it. It also documents this order.
Secondly, with a given 'Action' now spanning multiple files, in order
to keep track of which file a command is run from, the 'Command'
itself needs to store this. Ironically to the original intention,
this increases total ram usage. This change now only stores the file
name in each 'Action' instead of each 'Command'. All in all this is a
negligible trade off of ram usage.
Thirdly, this requires a bunch of extra code and assumptions that
don't help anything else. In particular it forces to keep property triggers
sorted for easy comparison, which I'm using an std::map for currently,
but that is not the best data structure to contain them.
Lastly, I added the filename and line number to the 'processing
action' LOG(INFO) message.
Test: Boot bullhead, observe above changes
Test: Boot sailfish, observe no change in boot time
Change-Id: I3fbcac4ee677351314e33012c758145be82346e9
If the input file isn't a multiple of block_size, count the trailing
bit as a block in the output, otherwise bootloader will try to write
past the last declared block in the sparse file.
Test: Run sparse_test
Bug: 37339998
Change-Id: Iec143107d15795c6a6f4d3ea7b7515f6f16adcae
ro.logd.kernel, ro.config.low_ram, ro.logd.timestamp and ro.debuggable
need to be retrieved prior to logd start in order for the service to
behave in a configured manner. Other essential services are also
dependent on these system properties as well, so it just makes sense
to pick them all up first in 'on fs'.
Test: smoke test
Bug: 37425809
Change-Id: I33ad185f397ee527ed3c84cc2bcb40ff8ca785b5
- Do not use -f if it was cleanly shutdown.
- For unclean shutdown or other operation failures like
mount, tune2fs failure, run full check.
- Still old image will run full check once in 5 reboots
while new image will not run full check unless something
fails.
- Add retry for final mount. If mount fails once, run full fsck
once and try again.
bug: 32246772
bug: 35366616
Test: many reboots
Change-Id: I86949732ffe1955636ac179d553c91e52910f73e
When BOOL_DEFAULT_FLAG_ENG and/or BOOL_DEFAULT_FLAG_SVELTE is set for
flags parameter in __android_logger_property_get_bool, they can not be
overridden by a supplied property value. Reset these two flags if
there is something in the specified property.
Test: gTest liblog-unit-tests, logcat-unit-tests and logd-unit-tests
Bug: 37425809
Change-Id: I5fc2d727e8c30bc1e781f8b365f44ec62dd36b4d
Helpful instrumentation to determine who is waiting for logger data.
Test: manual
Bug: 37274132
Bug: 37378309
Change-Id: I14fb1d9d15ae413930121048b770852359f06682
Switch vnsprintf to StringPrintf to avoid off-by-one, and switch
sparse_read.c to cpp.
Test: fastboot flash normal image with smaller sparse limit
Change-Id: Ia399b167625deb271bfd0ee3273071306d71c4d4
Currently if a process sets the sys.powerctl property, init adds this
property change into the event queue, just like any other property.
The actual logic to shutdown the device is not executed until init
gets to the action associated with the property change.
This is bad for multiple reasons, but explicitly causes deadlock in
the follow scenario:
A service is started with `exec` or `exec_start`
The same service sets sys.powerctl indicating to the system to
shutdown
The same service then waits infinitely
In this case, init doesn't process any further commands until the exec
service completes, including the command to reboot the device.
This change causes init to immediately handle sys.powerctl and reboot
the device regardless of the state of the event queue, wait for exec,
or wait for property conditions.
Bug: 37209359
Bug: 37415192
Test: Init reboots normally
Test: Update verifier can reboot the system
Change-Id: Iff2295aed970840f47e56c4bacc93001b791fa35
Add checking for impossible(tm) scenarios within LogBuffer::flushTo:
1) When iterating through the log entries, check if the iterator
returns two identical element references and break out of the loop.
2) Cap the maximum number of log entries we will skip while holding
the iterator lock at 4194304, break out of the loop.
We print a message to the kernel logs if we hit these cases.
ToDo: Remove this paranoia at some future date.
Test: gTest liblog-unit-tests logcat-unit-tests and logd-unit-tests
Bug: 37378309
Change-Id: I789594649db14093238828b9f6d1daeca8b780c2
Deal with a regression introduced in commit
5a34d6ea43 (logd: drop mSequence from
LogBufferElement) where log_time was compared against nsec() time
miscalculating the watermark boundary. When dealing with logcat
-t/-T, or any tail reading, add a margin to prune to back off by a
period of 3 seconds (pruneMargin).
Test: gTest liblog-unit-tests logcat-unit-tests and logd-unit-tests
Bug: 37378309
Change-Id: I72ea858e4e7b5fa91741ea84c40d2e7c3c4aa031
- moved __android_log_is_debuggable to a new public header
(log_properties.h)
- vendor version of sched_policy uses ALOG* instead SLOG*
Test: (sanity) liblog-unit-tests
Test: (sanity) libcutils_test (noting b/b/32972117, two tests continue
to fail)
Test: system/core as a whole makes with BOARD_VNDK_VERSION := current
now with no problems.
Test: boots/works on internal marlin
Bug: 33241851
(cherry picked from commit 1f83aa424f)
Merged-In: I5bc1f348dc0f0c8814bec5b5c3d2c52c825ab640
Change-Id: I5bc1f348dc0f0c8814bec5b5c3d2c52c825ab640
We can't reuse the GID range for internal cache files, otherwise
we don't have a way to tease apart the difference when deciding if
it's safe to move apps.
Test: builds, boots
Bug: 37193650
Change-Id: I22c4e575cd557636e74c5c73035adb1d4dcbb7f7
Create classes for dev and sys permissions and store these permissions
in std::vector instead of the C list.
Test: boot bullhead
Test: init unit tests
Change-Id: I874039a3db29b4c70149506da8e407123ab7eca2
Adding methods to queue and download flashable images by fd instead of
by pointer, so that we can deal with sending large (up to 4GB) files
on windows and linux. This gets past limitations on linux to read
more than 2GB from a file at a time, as well as memory limitations
on win32, in order to download up to 4GB in a single transfer.
Test: fastboot -w
Test: "flash-all" from nexus factory images site (incl. fastboot -w update)
Test: fastboot flash with large and small image, large and small max-download-size
Test: Sanity check flashing on win32, darwin, linux.
Test: Sanity check 3GB image download (with 3GB max-download-size)
on win32, darwin, linux.
Bug: 36810152
Change-Id: I528d739d344eb080d59d721dadf3b3b34d4b375e
Set ro.boot.avb_version to "AVB_VERSION_MAJOR.AVB_VERSION_MINOR".
During Treble OTA match, the major version must be the same as that in
the avb metadata on disk, while the minor version can be equal or
greater to that in the avb metadata on disk.
See how avb versioning work on the following link:
https://android-review.googlesource.com/#/c/342757/
Also renames AvbHashtreeDisabled() -> hashtree_disabled().
Bug: 35322304
Test: Early mount with AVB, checks [ro.boot.avb_version]: [1.0] exists.
Test: Not enable AVB, checks [ro.boot.avb_version] doesn't exists.
Change-Id: I5aaf476ca53c4fe817779518ba14b68ebcfdc6d6
This is definitely a FAQ, and it doesn't make sense for me to maintain
this on Google+ when I have the power to just check it in (and now have
a sensible place to do so).
Bug: N/A
Test: N/A
Change-Id: I649803ac846917a45f4b7b89cffe5b450179c479
Starting zygote early requires cpuset to be initialized to all cores for
foreground cpuset. Change to expolit all cores by default at boot and
let device manufacturers override to proper values in device specific
init script.
Bug: 36576280
Test: marlin boot fast and checked cpuset during early boot
Change-Id: I2c1ce0630e58a7b04d1a453c6740d3f0bce9de9f
* changes:
liblog: allow event tags to include some punctuations
liblog: logprint supports number of seconds time event field
logcat: test: standardize rest() to let logs land when injecting
This is required for large (>INT_MAX) sparse limit reported by
the target.
Also, patch up return chains of "int" that need to deal with sizes
bigger than 2GB as well as return negative error codes.
Test: -S works with large max-download-size
Test: Flash 3GB system.img with max-download-size 2.5GB
Bug: 36810152
Change-Id: I562a50eabd706bd5b97c71a1aef07c1ffd1a2e5c
fastboot uses AdbWinUsbApi on Windows, let's keep it required.
Test: rm -rf out/host; mma fastboot and confirm AdbWinUsbApi is in out/
Bug: 36810152
Change-Id: Ica8b27cb1d0bca260f716dc61fdcea2ccc282623
event_log_tag parser complains about a period (.) in the name,
we would consider such an enhancement to the tag names possible.
I expect we would want to be able to support alphanumerics,
underscore (_), period (.), minus (-), at (@) and comma (,) for
starters as they are present in the other text log buffer tags.
We introduce a local endOfTag function that is used during parsing and
during android_lookupEventTagNum for submitting new tags. This
function caused us to enforce const char more closely. By filtering
in both places we resolve an issue that could have plagued us if
garbage requests were made.
Test: gTest liblog-unit-tests, logd-unit-tests & logcat-unit-tests
Bug: 31456426
Change-Id: I596b8706e843719ddac07ec40e1cd2875c214bed
Add s to report time in seconds. The time could be a period, duration
or monotonic, expanded to seconds, minutes, hours and days. gTest has
to acquire a dynamic tag allocation as there are no users of this
feature yet.
Looking to the future, audio media logging has binary content similar
to the binary events structures Android logging uses and they have
a definition of a duration field in their internal binary logging, so
may be of use when we unify the logs.
Test: gTest logcat-unit-tests --gtest_filter=*.descriptive
Bug: 31456426
Change-Id: I262c03775983b3bc7b1b00227ce2bb2b0f357bec
Add an internal rest() function to set the standard for how long we
should wait for a log message to land end to end into the logs. Add
a retry on -EBUSY for all descriptive tests along with a rest() after
log injection so the tests can try to survive heavy Denial Of Service
(DOS) loads. Some of the resting was done at a coarse 1 second, when
200ms will do just fine.
Fix a few pesky issues surrounding test reliability and correctness:
- stdint.h include missing.
- missing a logcat_executable, means logcatd.descriptive testing was
testing logcat instead.
- count of interfering tests is 3 now: logcat, liblogcat and logcatd, if
they all run in the same PID, we have to not fail.
- Added a missed opportunity to add uniqueness to logcat.descriptive sync
test.
Test: gTest logcat-unit-tests
Bug: 31456426
Change-Id: Iaf2bee97878957f654613bb2e78e32379bb8d1b7
Also simplify this code a bit.
There's only one consumer that removes the /devices/platform prefix,
so have them handle it instead of storing two copies of the string.
Remove an unneeded search for '/' in get_character_device_symlinks()
as a / will always be the next character after a parent path, by
nature of FindPlatformDevice().
Test: boot bullhead
Test: init unit tests
Change-Id: I9d0482d137b1342ae7509ae993ff99198be814f0
Treehugger Robot