Navigation on GSB - Droid Eris General

Downloaded the new Google Maps update praying it would fix the FC's no dice. Are all Cyanogen ROM users having this problem? Having one more problem, my phone will notify me of E-mail; but when I go to my email it isn't showing my latest email. I have to refresh it to have it show up. Not Gmail but my Comcast account? Any help would be most appreciated.
Thanks

---

Just FYI to anyone with this same problem, I just rooted my eris for the first time 2 days ago. Maps nav was freezing on my non-rooted 2.1 in the same exact way. Something seems to be wrong with maps in general...but I'm not sure if it's just with the eris. But I do know that maps worked within the 3-4 previous updates. Can anyone provide a way to "downgrade" maps to an update that is a little older? I think this might provide everyone with a temp solution until they provide us with an update that works with the eris...just an idea

---

i think i read somewhere that some people tried to downgrade maps but it still FC'd on them...

---

Macrodroid just posted that he had success with downgrading his maps, check his EverVolv or Macrom thread.
Sent from my MacRom MR7 using Tapatalk

---

The only issue I've had is when i received a call during navigation. While it was rerouting, I received a FC. Starting it again and didn't have any issues.

---

Downloaded the apk that Mac posted last night. I drive 50 miles to work every morning. It force closed and rebooted my phone after 20 minutes. This occurs while driving down the interstate (not while making lots of turns). Going on a trip soon, might just have to download Zach's xtr rom. I've used it in the past month and navigation worked fine.

---

If anybody finds a way to deterministically crash the Eris using Maps/Nav, let us know; it would be very helpful for testing purposes. Note that you don't need to be driving around for a crash to occur (although it is conceivable that will help cause the bug to occur) - the Nav app will cause a crash sometimes with your Eris just sitting motionless on a desk.
I've observed a couple of full Android restarts so far (phone connected to the PC with Nav running and "logcat" running), and so far the things I've noticed are:
- the problem is hard to reproduce. Sometimes a crash will happen in 10 minutes, sometimes in 1 hour... sometimes not after several hours. (This is the the problem with "I tried Nav for 30 minutes and everything seemed fine" reports - it isn't easy to prove that a specific fix has actually worked, 'cuz it might be hours before the bug rears it's ugly head).
- All the cases where I've seen an (Android) crash so far (v5713/gapps 20110613), I saw a SIGSEGV (null pointer fault) of the "system_server" process - the same "null string passed to strlen() via libdvm.so" error that has been previously noted elsewhere (e.g. in Tazz' thread and some time ago in other CM7 ROMs (Hero CDMA) ). This obviously will cause a full Android layer restart, but strictly speaking is not exactly a Maps/Nav issue... but probably something about the way that the new Maps/Nav behaves that makes the occurrence of this bug favorable.
I'm still looking at this, but not making much progress. The only thing that I (seem) to have noticed is that turning on checkJNI in the zygote causes Nav to hang (ANR) so that Android offers a FC to the user. I suppose this could be a completely separate bug - the main thread seems to hang waiting for something in the GL.
e.g.
Code:
adb shell stop
adb shell setprop dalvik.vm.checkjni true
adb shell start
... then do your testing.

---

My mothers Eris was experiencing a similar issue as was my TB. a full wipe did the trick for me using the latest available CM versions and gapps for both phones. No issues thus far... fingers crossed.
Rick

---

bftb0,
Back when I was trying to figure out the Maps/Nav issues, I had requested people get the kernel log of a crash. I posted here what was interesting to me from someone that had posted one and then later decided to look at the source code and follow the routines mentioned. I remember looking it over but don't think I ever posted what I'd found or was thinking and can't find any notes about it right now. Might look into that again to see if the source holds any clues.

---

MongooseHelix said:
bftb0,
Back when I was trying to figure out the Maps/Nav issues, I had requested people get the kernel log of a crash. I posted here what was interesting to me from someone that had posted one and then later decided to look at the source code and follow the routines mentioned. I remember looking it over but don't think I ever posted what I'd found or was thinking and can't find any notes about it right now. Might look into that again to see if the source holds any clues.
Click to expand...
Click to collapse
I looked at the link you provided - if you thought that somehow the kernel module for WiFi was a source of trouble, you could certainly just move /system/lib/modules/wlan.ko out of the modules folder and reboot (for testing purposes - obviously in the long run you want to have WiFi capability).
My own testing is moving me in a different direction, though. So far, all the Android crashes that I have seen so far have been segfaults of the "system_server" process. I just watched one a couple of minutes ago, using both logcat and strace simultaneously.
The stack trace at the segfault of system_server in the logcat (if it is actually to be believed) looks something like
Code:
strlen() [libc.so] <- segfault here
??? [libdvm.so]
??? [libdvm.so]
??? [libdvm.so]
dvmLockObject() [libdvm.so]
??? [libdvm.so]
dvmMterpStd() [libdvm.so]
dvmInterpret() [libdvm.so]
dvmCallMethodV() [libdvm.so]
libdvm is the Dalvik VM parser - which you would think/hope would be extremely robust code (and thus makes me wonder if the stack backtrace above the strlen() call is just nonsense - the debugger is not guaranteed to get it correct).
here's the last couple of lines from an strace of system_server as it was crashing:
Code:
11:51:40 sched_setscheduler(157, SCHED_OTHER, { 0 }) = 0
11:51:40 ioctl(10, 0xc0186201, 0xbedce340) = 0
11:51:48 sched_setscheduler(157, SCHED_OTHER, { 0 }) = 0
11:51:48 getpriority(PRIO_PROCESS, 157) = 20
11:51:48 recv(606568, 0x81, 1, MSG_CTRUNC|MSG_EOR|MSG_WAITALL|MSG_ERRQUEUE|MSG_DONTWAIT|MSG_FIN|0xfffb0000) = 1
11:51:48 sched_setscheduler(157, SCHED_OTHER, { 0 }) = 0
11:51:48 ioctl(10, 0xc0186201 <unfinished ...>
11:52:04 +++ killed by SIGSEGV +++
Ugh... not very informative, frankly.
This might be a fairly difficult thing to pin down - after a little bit of testing, I've noticed that it is easier to cause this particular type of crash with a phone experiencing a decent (but reasonable) load than a lightly loaded phone - as if it is more likely to occur during a low resource condition, or is due to a timing/timeout issue. I have some difficulty getting it to occur on a "stripped" GSB 3.7 (Odex) ROM, where I removed almost everything non-essential, and didn't load any apps; but on my GSB 3.7 (Odex) "daily driver" ROM, it happens more frequently, and especially if I try and make the phone jump through other hoops while Maps/Nav is running.
... still looking...
bftb0
[ Edit ] 10 minutes after I made this post I flashed my "stripped down" GSB v3.7 Odex ROM, and set it up to watch it with Nav running. It crashed about 24 minutes later. I guess I should recant what I said above about load dependencies.

---

Just for kicks I reinstalled 2.1 (ota version) it didn't force close but the map quit updating my location. I had the latest maps installed and once again it crashed/stopped responding after ~20 minutes.

---

Just for kicks I restored a fairly old version of maps - v5127 a.k.a. "5.1.0" to my GSB v3.7 Odex ROM.
It crashed my phone (an Android restart, rather than a kernel reboot) within 10 minutes of starting up Nav.
Because all of these "crashes" seem to be in the system_server process, rather than the Maps/Nav app, I'm not convinced that the problems are due to the Maps app - we certainly were not experiencing restarts on older versions of GSB way back when maps 5.1.0 came on the scene. (Although we were experiencing the "GPS signal lost" after 20 minutes bug)
I don't know what to think at this time; maybe there is more than one bug involved. Perhaps I should try the non-odexed GSB version to see if it behaves any differently.
bftb0

---

I went driving around yesterday and was using navigation and had it crap out a few times. One time it said that acore had crashed. Two other times it just locked up for about 30 seconds and then I saw the cyanogen animation. Sort of like a partial reboot because it was really fast.
Sent from my GSB v3.6 ODEXed using XDA App

---

bftb0 said:
Because all of these "crashes" seem to be in the system_server process, rather than the Maps/Nav app, I'm not convinced that the problems are due to the Maps app ...
bftb0
Click to expand...
Click to collapse
I see what you're getting at but if the crashes are not Maps related why would non-CM7 phones also freeze up? Since this is happening on xtr and stock ROMs also is it possible that something in Maps is causing the system_server process to crash?

---

PieceKeepr said:
I see what you're getting at but if the crashes are not Maps related why would non-CM7 phones also freeze up? Since this is happening on xtr and stock ROMs also is it possible that something in Maps is causing the system_server process to crash?
Click to expand...
Click to collapse
My best guess is that there is more than one bug involved. Personally, I've observed three different behaviors, from most frequent to least frequent:
1- Android layer restart (always a system_server segmentation fault)
2- Full reboot (kernel restart)
3- Maps ANR (App Not Responding)
#1 is absolutely, positively a bug in the system_server code; plainly stated, native code should never segfault. What is odd about this is that having Maps running seems to cause this fault to occur - it doesn't seem to happen when Maps/Nav is not running. What is extremely challenging about diagnosing this is that (if the debuggerd stack traces are to be believed) it appears that it is the Dalvik VM code itself (libdvm.so) that makes the fatal call into libc.so - a simple strlen() call that gets a null pointer. That seems to imply that figuring this out requires understanding:
- how an unrelated app (Maps/Nav) is affecting java code of the system_server
- how the Dalvik compiler/optimizer/interpreter could possibly have a bug
Note that (again, if the debuggerd stack traces of the system_server crash appearing in the logcat are to be believed) if there really is a Dalvik VM bug involved, I suppose that also means that occurence of the bug could be dependent on:
- JIT on vs. off
- Optimizer settings (and odex'ed vs. non-odex'ed ROMs)
Perhaps this explains some of the variability in the various reports; most of my experiences with the bug(s) have been with the GSB 3.7 Odex ROM (JIT on, no compcache or dithering, various VM heap sizes).
I full-wipe installed GSB 3.7 non-odexed yesterday, and am going to do some more testing.
As for crash cases #2 and #3 - I haven't experienced enough of them to say anything meaningful.
bftb0
PS Over the weekend, I drove about 3 hours on both Sat and Sun, and I didn't want my phone blowing up on me, so I rolled back my phone to celb 4.1 (Froyo) and used Nav continously. The version of Maps in that celb backup was 5127 (aka "5.1.0"). I experienced the "GPS Signal Lost after 20 minutes" bug - which I expected - but using GPS status would allow Nav to re-acquire signal. But, no other trouble. The reason that I mention this is that this same *very old* version of Maps (5127) can readily cause the system_server process to crash in GSB v3.7 Odex. This kind of data is what leads me to believe that (at least for the "Android soft-restart" bug expression) while Maps/Nav can *cause the problem to occur*, this particular bug actually is not in the Maps app itself.
Anyway, that's a few of my observations; no solutions unfortunately.
More testing to come.

---

I have experienced all of those, but not in that order. Rearranged by frequency for me they would be ...
3- Maps ANR (App Not Responding)
1- Android layer restart
2- Full reboot (kernel restart)
I am on GSB 3.6 ODEXed (3.7 was sluggish for me and I was getting odd and random app force closes so I rolled back).
Settings:
CPU - Ondemand, 710/245
Compcache - disabled
JIT - off
Surface dithering - off
Purging of assets - off
Lock home in memory - off
VM heap size - 32mb (default)
V6 Supercharger - multitasking or sometimes Balanced 1 with launcher HTK
By far the most common problem for me was the Maps ANR. I did get the Full reboot, and several Android layer restarts before I started playing with settings. Unfortunately I don't remember what I started with and can't say what all I did. I was running Maps 3.6.2 (I'm pretty sure) at the time and I kept getting the Maps ANR. I found a copy of Maps 5.5.0 I had forgot about in MyBackup Pro so I wiped Maps data, then uninstalled before restoring just the Maps 5.5.0 apk.
Since then I have only had one problem, which happened two days ago. I had been using Navigation for about an hour and a half, maybe approaching two hours, by the time I reached my third destination. It showed me I had reached my destination but it froze. None of the app buttons were responsive so I tried the phone soft buttons and found they were unresponsive as well. Then the phone did an Android layer restart. At the time the phone was VERY HOT to the point of being uncomfortable to hold in your hand. I have never seen my phone that hot before. It was plugged in at the time because the battery had gotten low so I was charging it.
My wife's Eris is running xtrSense 5.0.1 and I took it out and tried it a while back. It never did a full reboot or an Android layer restart but Maps ANR occurred within five minutes. I tried several times and got the same result each time.
Hopefully some others who have experienced problems will see this and chime in. Maybe if we can narrow down specific occurrences with stock and other ROMs it will help decipher this. Thanks bftb0 for breaking this down into categories and for your work on the problem. If there is any data I can capture and provide let me know.

---

All -
I've been testing latest map version on a number of roms as of late and offer a few observations:
I have consistently experienced Android layer restart and full reboot on GSB roms, stock settings, versions 3.4 through current. I experienced a reboot on Condemned soul's version 14 rom only once. I have not ever experienced a navigation reboot on OMGB 1.1, Evervolv,3, or 666th Sense (March OTA based rom).
The only ROM I have run that has not resulted in a map ANR is OMGB 1.1, but GPS signal is lost within a minute on that rom. Navigation continues to wait for a signal but does not force close.
Condemned soul's v14, Evervolv 3, and 666th Sense all display the same behavior: navigation runs for between 2-10 minutes and map display freezes while voice navigation continues running as a separate service. Eventually a FC is offered and navigation can be easily resumed from the notification bar. It will FC within another 10-20 minutes.
I've been doing some testing and trying a few tweaks - all dead ends and none of them have worked, but I'll repeat them here in case they are of any use to this group:
First, I tried flashing old versions of gapps to roll back the com.google.android.maps.jar file to see if there was anything there affecting navigation. At first I thought it was working differently, but alas the old jar (if they are actually even different) files really behaved the same -- they locked up the navigation moving map while partially allowing the navigation voice(s) to continue.
After repeated testing, It *seems* that the moving map is hanging up for me in the SAME PLACES. Now this is all very subjective since I cannot pay all that much attention while driving, but when I leave my office for home it seems that the maps hangs at the first right turn, 1/2 mile or 2 minutes away. If I restart it, it will continue for another 10 minutes or so until after a major highway merge.
So, I began to notice the dual-voice navigation some others have mentioned and deduced that one voice is TTS and one is google's backup non-TTS navigation sounds. I suspected, perhaps, that TTS might be crashing or being force closed by the system. A little routing around and I found that if you uninstall TTS that navigation simply will not load but will FC immediately. I also found that you can disable TTS by removing/renaming the libttspico.so file. This way, you can actually force the second navigation voice - its rather pleasant but does not pronounce street names, just says "turn left" etc.
This did not fix the problem either. But it did allow me to rule out that the phone was temporarily losing GPS signal leading to the FC. Without TTS, navigation alerts you that GPS signal was lost with a chime sound. Maps was not hanging on a lost GPS signal. since voice directions continue despite visual map hanging.
So this leads me to my next observation. Without the distracting TTS voice I notice that the maps seems to hang precisely at a point where I am supposed to be alerted to an upcoming change. For instance, after making my first right leaving the office, it hangs when (1) I turn right forcing the map to pan and draw the route for the next leg AND SIMULTANEOUSLY (2) when the GPS wants to tell me the next direction "In a half mile, turn right on to the ramp"
I also thought that there may be a memory problem - like GB is trying to kill one of the Navigation related processes. As far as I can tell, navigation requires that the navigation map, a separate navigation voice service, and PICO TTS all be running at the same time on top of a launcher, maps, email, and other services. I tried tweaking V6 supercharger to gaming settings with no luck.

---

klobkelosh said:
...First, I tried flashing old versions of gapps to roll back the com.google.android.maps.jar file to see if there was anything there affecting navigation. At first I thought it was working differently, but alas the old jar (if they are actually even different) files really behaved the same -- they locked up the navigation moving map while partially allowing the navigation voice(s) to continue...
Click to expand...
Click to collapse
If the jar files are the same, they should have the same md5sum so it might be worthwhile to compare them and see if each jar has matching checksums.

---

*POSSIBLE* Navigation Fix
I may have found a fix. I was able to navigate on my daily commute without a FC or reboot (that's not happened in a while). BUT I need your help in testing it more to be sure it is not a fluke. I'll also be testing over coming days
The attached file is NOT a flashable zip. Extract the com.google.android.maps.jar file and push it to /system/framework replacing the gapps gb version. If you use root explorer/es file manager be sure to set permissions to rwrr.
This jar version is from the May 2011 MDPI gapps package. It is nearly double the filesize of the gb version so clearly not the usual framework.
Hope this works in the long run....

---

I'm driving from Paducah, KY to Oscoda, MI. Should be a good test, I hope this works