Hi there, my first post here, bugreport. :)
Hi there, my first post here, bugreport. 🙂
Something wrong is in v12.5.2 compared to v12.5.1.
Have HTC Desire Z (Android 2.3.3, original stock rom) with not so much free RAM and new version often lost from “back” button sequence when I returning from apps. Only “home” button runs LL again. LL is probably force closed by system, but I seen several self close and open desktop even LL only was on display. Nothing like this has ever happened before. Some memory leaks or what? Backup/restore template did not help.
Going back to v12.5.1 and everything is ok.
Please analyse, ask for more, or try to fix it, thx! 😉
]]>« Guys, does nine patches really works on LL? (Previous Post)
(Next Post) Something WIP, something a bit new. »
D5 Creation
< ![CDATA[
So where does it go when you hit the back button?
]]>
< ![CDATA[
It remains in the loop Phone -> People – > Settings, never go to default launcher.
]]>
< ![CDATA[
I observed similar problem on my second NEO N003 phone (2 GB RAM), but only once or twice immediately after update and not systematically like on HTC.
]]>
< ![CDATA[
Technically when you hit the back key, the key is sent to some app which may choose to terminate its activity, and in that case the system selects another activity to resume. Why the system doesn’t select the launcher I don’t know. Probably it has been removed from the back stack, which is managed by Android. Of course there are some flags to configure the behavior of an app regarding the back stack, but none of them is in use in Lightning, and the app doesn’t remove itself from this stack neither. I checked the source code differences between 12.5.1 and 12.5.2 and really there is nothing related to app life cycle or system configuration, so I believe that this is either a coincidence or a very nasty border effect.
]]>
< ![CDATA[
Thx Pierre, I understand all about back stack and I’m also surprised how is this bug caused. I can take some logs and film the video, if it helps. I think that some memory leak or critical api call is related. I have more Android devices and I can test more. Hope we find it, or I can try newest beta, if error is still present. Where I can find lastest apk please?
]]>
< ![CDATA[
Pierre Hébert I never made any tests, but I think the problem is when the launcher auto-restarts.
]]>
< ![CDATA[
I use LL for years and LL sometimes killed and well restarted when no free RAM, but now it seems that can’t recover rightly. Important is that now is LL killed much more often than ever before. Hence I thinking about leaks or api calls.
]]>
< ![CDATA[
Could you add a dynamic text configured as “Heap free” and see whether this number constantly decreases? In theory it shouldn’t, otherwise there is a leak. Be aware that it slows down the app a lot, don’t forget to remove it once you’re done with it.
]]>
< ![CDATA[
On v12.5.1 – heap free ~25.9 MB and was stable.
On v12.5.2 – heap free ~26.9 MB, after some using ~25.4 MB, and was stable.
After apk reinstall error still persist, tested ~9 depth level of back then returned to Phone-People-Settings loop like before, not to LL.
Path: LL Home -> LL Desktop #2 -> LL All Apps -> Link2SD -> Google Play Store -> Total Commander -> *.jpg picture -> QuickPic -> *.3gp video -> MX Player
Second try returned to LL Desktop #2 then back button do nothing.
Third try returned to LL Home, probably apps are well cached.
Heap free during back tests ~24.7 MB.
]]>
< ![CDATA[
It doesn’t seem to be a leak then, but anyway this doesn’t prevent the app to be killed. When loading big apps (play store, quick pick and mxplayer are big apps) other apps will be suspended anyway, whatever their memory use.
I just thought at something: when the system suspends an app, it should remember it was there and restore it when needed. But if the app restarts itself, then it may explain why the system don’t keep it at the same place in the stack (it just sees that an app stops and another one starts). Do you use the restart action in the app (possibly from scripting), or could it be due to timezone change (which triggers a restart in 12.5.2) ?
]]>
< ![CDATA[
The only change is 12.5.1 to 12.5.2 and error is present, I did not use any scipting (I got it disabled in settings). It does not make sense to be timezone change triggered, but who knows. I know that my HTC clock is quite crazy/unaccurate and sometime I noticed that shows 3-5 minutes late, then recync and maybe this can trigger something?
Anyway this error is really annoying, I click for received SMS only, then press back and I’m in loop, not in LL Home. 🙁
Can logs help? I think that LL Service must crash, not to be killed by OS, this may explain why LL is not in back stack.
]]>
< ![CDATA[
No, in case of a crash you would have seen something and the system would have restarted it anyway, more or less automatically (I think).
Yes logs could help, termination are logged with the reason. An hypothetical time zone changed is probably logged too.
]]>
< ![CDATA[
Tested on v12.6b2. I have logs when LL Home do self restart (LL Home -> black screen -> LL Home) without my assistance, phone only rest on desk with display on. Where you want to upload it?
]]>
< ![CDATA[
Please send it to me by email: pierrox@pierrox.net
]]>
< ![CDATA[
v12.6b3 is still wrong.
]]>
< ![CDATA[
Pierre Hébert I don’t know if this will help narrow it down any, but I have been having the same issue on 4.1 and 4.4.4
My Moto x just updated to 5.1 last night and so far it seems to be resolved.
Goes back to launcher every time.
]]>
< ![CDATA[
Chris, thank you for sharing, I’m happy I’m not alone. 🙂 Have you root and some advanced utils installed, like BusyBox Pro, GravityBox, 3C Toolbox Pro?
]]>
< ![CDATA[
Not this phone. My first gen Moto X was. I wish I could root this one, but there is not yet a way to unlock the bootloader.
]]>
< ![CDATA[
I understand, have same problem with HTC. I’m only thinking about some app collision, but my HTC not rooted, so the problem must be in LL…
]]>
< ![CDATA[
Tested all LL data erase and also tried reinstall, but LL still crashing.
Something found in logs, Pierre?
]]>
< ![CDATA[
No crash in logs, the app is stopped and restarted, most probably because of the timezone change thing. The timezone event is not explicit in the log but there are indications of time event at the same time and by checking the pid in the log the app is clearly stopped and restarted by itself, so I am 99% sure that this is the reason. The line “Scheduling restart of crashed service” doesn’t mean the app crashed but that the service has been destroyed, in this case because of the exit system call (apps don’t do that usually).
In b3 the app should only stop without restarting, I don’t know whether it makes a difference. I’ll try to filter timezone changes by skipping events when the value isn’t really modified.
]]>
< ![CDATA[
Thank you, Pierre, interesting article. Hope it helps in next beta. 🙂
In addition, I noticed that the Home press (to start LL) usually does not work on the first pressing, but mostly on the second or third.
]]>
< ![CDATA[
I tried to filter time zone changes in the new beta, hopefully it will help to solve the issue, but to be honest I am not extremely confident in the fix. On one hand the app restarts itself and there is almost no doubt that this is due to the time zone auto restart thing, but on the other hand I find very strange that the time zone is often modified without (visible) reason.
Well, let’s see how it works with v12.6b4.
]]>
< ![CDATA[
Sorry, today v12.6b4 still wrong and crashing. 🙁
Pls. send me betas between v12.5.1 and v12.5.2 by email to test, thanks!
Ops, there are not available beta as I see in change log, sorry.
So what next step to isolate bug you suggest? Can you particulary rebuild v12.5.2 with a few new changes removed (half of all)?
]]>
< ![CDATA[
When I disable “Automatic – use network-provided values” in “Date and time” settings, then my primary problem with LL (randomly force closed) is solved, so I can confirm that time change or time zone change cause the problem.
]]>
< ![CDATA[
Jaroslav Grochal great idea!
]]>