Tone it down for everyone, as it's been running smoothly for a while
now, but add a dedicated extra warning on buggy boards that it might
randomly implode.
Basically:
* Use `extend` for class definitions
* Use `new` for object instantiations
That includes some minor code cleanups along the way:
* Updated `Widget`'s docs to make the semantics clearer.
* Removed `should_restrict_JIT` (it's been dead code since https://github.com/koreader/android-luajit-launcher/pull/283)
* Minor refactoring of LuaSettings/LuaData/LuaDefaults/DocSettings to behave (mostly, they are instantiated via `open` instead of `new`) like everything else and handle inheritance properly (i.e., DocSettings is now a proper LuaSettings subclass).
* Default to `WidgetContainer` instead of `InputContainer` for stuff that doesn't actually setup key/gesture events.
* Ditto for explicit `*Listener` only classes, make sure they're based on `EventListener` instead of something uselessly fancier.
* Unless absolutely necessary, do not store references in class objects, ever; only values. Instead, always store references in instances, to avoid both sneaky inheritance issues, and sneaky GC pinning of stale references.
* ReaderUI: Fix one such issue with its `active_widgets` array, with critical implications, as it essentially pinned *all* of ReaderUI's modules, including their reference to the `Document` instance (i.e., that was a big-ass leak).
* Terminal: Make sure the shell is killed on plugin teardown.
* InputText: Fix Home/End/Del physical keys to behave sensibly.
* InputContainer/WidgetContainer: If necessary, compute self.dimen at paintTo time (previously, only InputContainers did, which might have had something to do with random widgets unconcerned about input using it as a baseclass instead of WidgetContainer...).
* OverlapGroup: Compute self.dimen at *init* time, because for some reason it needs to do that, but do it directly in OverlapGroup instead of going through a weird WidgetContainer method that it was the sole user of.
* ReaderCropping: Under no circumstances should a Document instance member (here, self.bbox) risk being `nil`ed!
* Kobo: Minor code cleanups.
* UIManager: Support more specialized update modes for corner-cases:
* A2, which we'll use for the VirtualKeyboards keys (they'd... inadvertently switched to UI with the highlight refactor).
* NO_MERGE variants of ui & partial (for sunxi). Use `[ui]` in ReaderHighlight's popup, because of a Sage kernel bug that could otherwise make it translucent, sometimes completely so (*sigh*).
* UIManager: Assorted code cleanups & simplifications.
* Logger & dbg: Unify logging style, and code cleanups.
* SDL: Unbreak suspend/resume outside of the emulator (fix#9567).
* NetworkMgr: Cache the network status, and allow it to be queried. (Used by AutoSuspend to avoid repeatedly poking the system when computing the standby schedule delay).
* OneTimeMigration: Don't forget about `NETWORK_PROXY` & `STARDICT_DATA_DIR` when migrating `defaults.persistent.lua` (fix#9573)
* WakeupMgr: Workaround an apparent limitation of the RTC found on i.MX5 Kobo devices, where setting a wakealarm further than UINT16_MAX seconds in the future would apparently overflow and wraparound... (fix#8039, many thanks to @yfede for the extensive deep-dive and for actually accurately pinpointing the issue!).
* Kobo: Handle standby transitions at full CPU clock speeds, in order to limit the latency hit.
* UIManager: Properly quit on reboot & exit. This ensures our exit code is preserved, as we exit on our own terms (instead of being killed by the init system). This is important on platforms where exit codes are semantically meaningful (e.g., Kobo).
* UIManager: Speaking of reboot & exit, make sure the Screensaver shows in all circumstances (e.g., autoshutdown, re: #9542)), and that there aren't any extraneous refreshes triggered. (Additionally, fix a minor regression since #9448 about tracking this very transient state on Kobo & Cervantes).
* Kindle: ID the upcoming Scribe.
* Bump base (https://github.com/koreader/koreader-base/pull/1524)
* AutoSuspend: Use the canSuspend devcap check instead of reinventing the wheel.
* Device & UIManager: Cleanup canSuspend devcap check related stuff to avoid boilerplate code.
(It also now defaults to no, and is explicitly set by device implementations where supported).
* AutoSuspend: Re-engage suspend/shutdown timers when fully charged.
This restores the existing behavior pre #9036
(c.f., https://github.com/koreader/koreader/pull/9258#issuecomment-1167672356)
* SDL: Unbreak the fake suspend behavior so that it actually works.
Tweak the default screensaver message to remind users that Power is bound to F2.
(Fix#9262)
* AutoSuspend: Re-engage suspend/shutdown timers on unplug.
This matters on Kobo, because the unexpected wakeup guard might have stopped the suspend timer.
This was originally designed for finer-grainbed control (specifically,
to correct by the amount of time spent in standby), but the final fix
only ever sets this to 0.
Simplify and rename (consumeInputEarlyAfterPM) to avoid useless computations.
Long story short: the LeaveStandby event is sent via `tickAfterNext`, so if we tear down the plugin right after calling it (in this case, that means that the very input event that wakes the device up from suspend is one that kills ReaderUI or FileManager), what's in UIManager's task queue isn't the actual function, but the anonymous nextTick wrapper constructed by `tickAfterNext` (c.f.,
https://github.com/koreader/koreader/issues/9112#issuecomment-1133999385).
Tweak `UIManager:tickAfterNext` to return a reference to said wrapper, so that we can store it and unschedule that one, too, in `AutoSuspend:onCloseWidget`.
Fix#9112 (many thanks to [@boredhominid](https://github.com/boredhominid) for his help in finding a repro for this ;)).
Re: #8638, as the extra debugging facilities (i.e., ebb81b9845) added during testing might help pinpoint the root issue for that one, too.
Also includes a minor simplification to `UIManager:_checkTasks`, and various other task queue related codepaths (e.g., `WakeupMgr`) ;).
* Disable all non power management related input during suspend. (This prevents wonky touch events from being tripped when closing a sleep cover on an already-in-suspend device, among other things).
* Kobo: Use our WakeupMgr instance, not the class.
* WakupMgr: split `removeTask` in two:
* `removeTask`, which *only* takes a queue index as input, and only removes a single task. Greatly simplifies the function (i.e., it's just a `table.remove`).
* `removeTasks`, which takes an epoch or a cb ref, and removes *every* task that matches.
* Both of these will also *always* re-schedule the next task (if any) on exit, since we can have multiple WakeupMgr tasks queued, but we can only have a single RTC wake alarm set ;).
* `wakeupAction` now takes a `proximity` argument, which it passes on to its `validateWakeupAlarmByProximity` call, allowing call sites to avoir having to duplicate that call themselves when they want to use a custom proximity window.
* `wakeupAction` now re-schedules the next task (if any) on exit.
* Simplify `Kobo:checkUnexpectedWakeup`, by removing the duplicate `WakerupMgr:validateWakeupAlarmByProximity` call, now that we can pass a proximity window to `WakeuoMgr:wakeupAction`.
* The various network activity timeouts are now halved when autostandby is enabled.
* Autostandby: get rid of the dummy deadline_guard task, as it's no longer necessary since #9009.
* UIManager: The previous change allows us to simplify `getNextTaskTimes` into a simpler `getNextTaskTime` variant, getting rid of a table & a loop.
* ReaderFooter & ReaderHeader: Make sure we only perform a single refresh when exiting standby.
* Kobo: Rewrite sysfs writes to use ANSI C via FFI instead of stdio via Lua, as it obscured some common error cases (e.g., EBUSY on /sys/power/state).
* Kobo: Simplify `suspend`, now that we have sane error handling in sysfs writes.
* Kobo.powerd: Change `isCharging` & `isAuxCharging` behavior to match the behavior of the NTX ioctl (i.e., Charging == Plugged-in). This has the added benefit of making the AutoSuspend checks behave sensibly in the "fully-charged but still plugged in" scenario (because being plugged in is enough to break PM on `!canPowerSaveWhileCharging` devices).
* AutoSuspend: Disable our `AllowStandby` handler when auto standby is disabled, so as to not interfere with other modules using `UIManager:allowStandby` (fix#9038).
* PowerD: Allow platforms to implement `isCharged`, indicating that the battery is full while still plugged in to a power source (battery icon becomes a power plug icon).
* Kobo.powerd: Implement `isCharged`, and kill charging LEDs once battery is full.
* Kindle.powerd: Implement `isCharged` on post-Wario devices. (`isCharging` is still true in that state, as it ought to).
* Make sure AutoStandby cohabitates nicely with AutoSuspend (regression since #8985), specifically:
* Disable standby during suspend.
* Ensure that leaving standby restores the scheduled suspend properly, with the appropriate remaining amount of time based on the last user input.
* Handle devices with an auxiliary battery better when scheduling suspend (assume it's only charging when the aux battery is charging, not the ereader's).
* Tweak debug logging to be able to remote debug corner-cases more easily without requiring code changes.
* Fix erroneous behavior when awoken from standby by something that doesn't trigger an InputEvent Event (e.g., rtc alarm, gyro, or random bullshit).
Specifically, don't forcibly unschedule/schedule on every input event, instead, let the scheduled task figure out if the deadline came to pass or not ;).
c.f., https://github.com/koreader/koreader/pull/8970#issuecomment-1092775830
Besides getting rid of some overhead, this allows proper scheduling after a task that would have blocked for longer than the standby timeout.
Allows the device to go into standby (if available in `/sys/power/state`) to save power.
Adds an entry in the device menu to tune the timeout for standby.
(Shows total standby- and suspend-time in system statistics.)
* Ensure that going from one to the other tears down the former and
its plugins before instantiating the latter and its plugins.
UIManager: Unify Event sending & broadcasting
* Make the two behave the same way (walk the widget stack from top to
bottom), and properly handle the window stack shrinking shrinking
*and* growing.
Previously, broadcasting happened bottom-to-top and didn't really
handle the list shrinking/growing, while sending only handled the list
shrinking by a single element, and hopefully that element being the one
the event was just sent to.
These two items combined allowed us to optimize suboptimal
refresh behavior with Menu and other Menu classes when
opening/closing a document.
e.g., the "opening document" Notification is now properly regional,
and the "open last doc" option no longer flashes like a crazy person
anymore.
Plugins: Allow optimizing Menu refresh with custom menus, too.
Requires moving Menu's close_callback *after* onMenuSelect, which, eh,
probably makes sense, and is probably harmless in the grand scheme of
things.
Plugins are loaded *once*, but torn-down/instantiated multiple times,
and sometimes in the reverse order.
As such, if we use a public function member as the scheduled task, we're
always pointing to the same function, and going from FM to RD
effectively *un*schedules it.
Instead, use an instance-specific closure, so that each instance
schedules & unschedules don't affect each other.
In the same vein, settings ought to be read at instantiation, not at
loading, otherwise, changing a setting in the FM, then switching to the
reader will not pick up the changes.
Kindles are not flagged canPowerOff, although that's technically not
entirely warranted, but sorta makes sense.
The Plugin already handles that sanely, and will only expose/honor the
suspend timer.
* Input: Don't create a new TimeVal object for input frame timestamps, just promote our existing table by assigning it the `TimeVal` metatable.
* TimeVal: Export (const) `zero` & `huge` TimeVal objects, because they're common enough in our codebase. (NOTE: not actually const, that's a Lua 5.4 feature ;p).
* GestureDetector: Explain the behavior of the `last_tevs` & `first_tevs` tables, and why one needs a new object and not the other.
* Speaking of, simplify the copy method for `first_tevs`, because it doesn't need to create a new TimeVal object, we can just reference the original, it's unique and re-assigned for each frame.
Requires https://github.com/koreader/koreader-base/pull/1344 & https://github.com/koreader/koreader-base/pull/1346 (fix#7485)
Assorted input fixes:
* Actually handle errors in the "there's a callback timer" input polling loop.
* Don't break timerfd when the clock probe was inconclusive.
Not directly related, but noticed because of duplicate onInputEvent handlers:
* HookContainer: Fix deregistration to actually deregister properly. "Regression" extant since its inception in #2933 (!).
* Made sure the three plugins (basically the trio of AutoThingies ;p) that were using HookContainer actually unschedule their task on teardown.
* ReaderDictionary: Port delay computations to TimeVal
* ReaderHighlight: Port delay computations to TimeVal
* ReaderView: Port delay computations to TimeVal
* Android: Reset gesture detection state on APP_CMD_TERM_WINDOW.
This prevents potentially being stuck in bogus gesture states when switching apps.
* GestureDetector:
* Port delay computations to TimeVal
* Fixed delay computations to handle time warps (large and negative deltas).
* Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture.
* Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1.
* Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy.
* The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events.
The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use
a clock source that is *NOT* MONOTONIC.
AFAICT, that's pretty much... PocketBook, and that's it?
* Input:
* Use the <linux/input.h> FFI module instead of re-declaring every constant
* Fixed (verbose) debug logging of input events to actually translate said constants properly.
* Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume.
* Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient.
Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts,
as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector.
* reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary.
* TimeVal:
* Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>).
* Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC.
* Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence.
* New methods:
* Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime
* Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero
* UIManager:
* Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base.
This ensures reliable and consistent scheduling, as time is ensured never to go backwards.
* Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame.
It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value,
because very few time has passed.
The only code left that does live syscalls does it because it's actually necessary for accuracy,
and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics).
* DictQuickLookup: Port delay computations to TimeVal
* FootNoteWidget: Port delay computations to TimeVal
* HTMLBoxWidget: Port delay computations to TimeVal
* Notification: Port delay computations to TimeVal
* TextBoxWidget: Port delay computations to TimeVal
* AutoSuspend: Port to TimeVal
* AutoTurn:
* Fix it so that settings are actually honored.
* Port to TimeVal
* BackgroundRunner: Port to TimeVal
* Calibre: Port benchmarking code to TimeVal
* BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly.
* All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
* Unify logging with AutoSuspend (e.g., keep ourselves to showing the delay in seconds, not the raw timestamp, as that's way harder to interpret, and the RTC module and/or logger will do that for us when the time comes).
* Speaking of, minor revamp of RTC related logging to make it more human-readable.
* On Kobo, if we hit the unexpected wakeup limit, re-engage AutoSuspend's *suspend* check, so that the device has a chance to poweroff instead of being kept awake.
* LuaSettings/DocSettings: Updated readSetting API to allow proper initialization to default.
Use it to initialize tables, e.g., fixing corner-cases in readerFooter that could prevent settings from being saved.
(Fixes an issue reported on Gitter).
* LuaSettings/DocSettings: Add simpler API than the the flip* ones to toggle boolean settings.
* Update LuaSettings/DocSettigns usage throughout the codebase to use the dedicated boolean methods wher appropriate, and clean up some of the more mind-bending uses.
* FileChooser: Implement an extended default exclusion list (fix#2360)
* ScreenSaver: Refactor to avoid the pile of kludges this was threatening to become. Code should be easier to follow and use, and fallbacks now behave as expected (fix#4418).
This allows for better energy efficiency (no more 50Hz tick poll),
as well as lower input lag / higher precision - touch events are
native linux ones.
In addition, auto off/suspend plugin is used in this mode, as we need
to trigger (timed) sleep / poweroff on our own, since the OS ones
will no longer work whenever koreader has focus.
This is for rooted devices only, and possibly somewhat FW
specific, so enabled only on PB740-2 where it's reasonably tested.
* Enable AutoSuspend plugin on rM
Fix#6769
Re: #6028
* Use the PowerEvent handler on rM
It makes much more sense than the fire & forget & hope for the best
approach copied from the Kindle platform, because we *are* controlling
suspend ourselves (mostly), unlike on Kindle ;).
Fix#6676
* Enable HW inversion on the rM
I mean, we kinda forgot to ever test that, but I don't really see why it
wouldn't work ;).
