Okay i take it back... flashed the latest CWM and back to the stupid bootloops. I'm sticking with 3.0.2.8 folks!
[CM7] Help track down Sleep of Death (NEW Second form, Apps and Widgets!)
- Thread starter khaytsus
- Start date
Okay i take it back... flashed the latest CWM and back to the stupid bootloops. I'm sticking with 3.0.2.8 folks!
Clockwork cannot cause boot loops in CM7, it is an entirely different kernel and ramdisk.
I think we've disagreed on this before.
There's a lot of documentation of the problem in the CM7 user thread when 3.2.0.0 was intriduced. Also the OP for the CWM thread mentions this issue too.Perhaps we are thinking of something different - what i am describing is that the NC boots normally (i.e. what you're getting at i think, that normal boot does not involve CWM) unless you have asked it to boot into recovery, at which point it either fails and boots normally, or it bootloops.
It looks like once 3.2.x.x is flashed, even though you can flash 3.0.2.8 there is some kind of fight... there's a few threads popping up where smarter people than me are adb-ing into the boot directory and seeing files that shouldn't be there reappear.
Anyways, for those of us who are not android devs, the solution i have found was to boot from my CWM bootable SD (because any attempt to boot to recovery failed). Reflash the nightly to overwrite the /boot. Then in rom manager, reflash 3.2.0.1 - even though this is so instantaneous it looks like nothing is being flashed, it magically fixes the problem
Just reflashing 3.2.0.1from rom manager without reflashing the nightly's /boot does not fix the problemGo figure
I flashed n97 + dal's new kernel 6/9 last night and was really hoping for a fix to my SOD and NDS issues. all is running well, new kernel is so smooth and zippy. but I'm getting lots of SODs. I think I'm getting deep sleep but cant be for sure because of SOD. after rebooting all logs in cpu spy has been re-updated. I think I'll flash n98 without OC kernel and see what happens.
cm7 n97 + OC 6/9 conservative gov min300 600 800 1000 max1200
cm7 n97 + OC 6/9 conservative gov min300 600 800 1000 max1200
Last edited:
so, I just want to make sure that re-flashing to original B&N rom 1.1 or 1.2 and then flash back to a cm7 nightly has worked for you in fixing your SOD and NDS issues. I have read this before but is a little reluctant to try because of all the work that is needed. do you think a factory reset can also work? or is it the same thing. maybe I'll just need to bite the bullet and just do it.
btw: how is your nook running now? is all still well.
so, I just want to make sure that re-flashing to original B&N rom 1.1 or 1.2 and then flash back to a cm7 nightly has worked for you in fixing your SOD and NDS issues. I have read this before but is a little reluctant to try because of all the work that is needed. do you think a factory reset can also work? or is it the same thing. maybe I'll just need to bite the bullet and just do it.
btw: how is your nook running now? is all still well.
Well everything was perfect until i installed dal's new kernel today. Then i had a SOD almost immediately afterwards. Did a little experimenting based on the circulating theory that SOD is actually awake but no screen. Plugged it into my PC - the PC ding-dongs and pulls up the nook's drive (suggesting that the nook is awake). I then held power on the nook for a hard shut down, the PC dong-dings as the nook disconnects. Then a regular startup rebooted. no problems since.
Honestly, it was misery for me before (i.e. SOD every time i turned my back on the damn thing), and well worth the effort. Once you have everything downloaded the whole process takes about 10-15 minutes of flashing (not least cos you can do it in one session) and then 5-10 minutes of restoring everything with TB. After that the only minor pain is resetting any widgets you had (TB does a better job of restoring your ADW desktop than ADW does... with TB restoring ADW with data gets everything but the widgets back).
The biggest heartache was the complications with CWM... I'm now not having any problems with it on 3.0.2.1 (i always use the manual restore though).
I had tried just a factory reset before, but that was not enough to fix the SOD. There's something constipating either /boot or /system that only seems to get cleaned out by taking it completely back to stock. If i was feeling particularly masochistic i guess i could restore old /systems or old /boots from backups i have of 96 which suffered SOD, but right now i'm happy and if it ain't broke...
I've just upgraded to 98 and flashed 06/09 over the top of that so i will keep you posted.
Hope that helps
Solved my SOD issues...by moving from class 6 SD to class 2 SD
OK, so, as the title says, I've been having SOD issues while running CM7 on a class 6 SD card. Specifically I've been seeing it since we hit the mid eighties in the nightlies. Now, there's been plenty of other problems that I've attributed to the "high speed" SD card as well, things like random FCs on apps (the xda app was probably the worst, never got it to even open), slow boot times, sometimes get stuck in boot loops (restarting the skateboard animation at 5-10 seconds), etc. But worst of all was how flaky the wifi was. I would turn it on and it would be a complete crap shoot to see if it would start up or error out. It would fail reliably if the system was busy, e.g. too close to boot. As my boot process was driven by disk access speed that hints that it would fail if the disk was too busy/queued out too far. It would usually work when I waited two or three minutes before turning it on.
Sometimes wifi would crap out in the middle of a session, this was pretty normal when downloading a new nightly with firefox or dolphin, though strangely not when downloading with opera. When this happened gapps would immediately crash/FC almost without fail. The browser would also crash and then FC when I restart it, over and over, until I reboot...(I never checked if the kernel remounted the drive as read only, but it may have if there were big enough problems, I was always too focused on getting the latest nightly) Wifi would stay errored until I rebooted too.
Now, like I said, I fixed all these issues by purchasing a class 2 SD card. I made no other changes. I literally dd'ed my current setup from the class 6 to the class 2 card. It's been working like a charm for 2 days, no issues.
This suggests two things:
1) at least some of the SODs are caused by slow flash chips. This is probably true for the emmc folks too, as flash chips are like snowflakes, each one is unique. Is there a benchmark tool out there that will allow us to evaluate the emmc on sequential and random read/write speed? That test would prove or falsify this hypothesis.
2) The devs who can't reproduce the SOD could do so by purchasing a class 6 or class 10 card. Shoot, I'd be happy to mail you mine, useless thing that it is.
Edit: see http://xdaforums.com/showthread.php?p=12964262 for a discussion on SD card speeds, benchmarks, and results/symptoms.
Edit 2: a corollary of item 2 above is that a possible solution to SOD is running off a decent SD card instead of emmc. Number one rule of debugging: if something works in one place and not another, always suspect the differences.
OK, so, as the title says, I've been having SOD issues while running CM7 on a class 6 SD card. Specifically I've been seeing it since we hit the mid eighties in the nightlies. Now, there's been plenty of other problems that I've attributed to the "high speed" SD card as well, things like random FCs on apps (the xda app was probably the worst, never got it to even open), slow boot times, sometimes get stuck in boot loops (restarting the skateboard animation at 5-10 seconds), etc. But worst of all was how flaky the wifi was. I would turn it on and it would be a complete crap shoot to see if it would start up or error out. It would fail reliably if the system was busy, e.g. too close to boot. As my boot process was driven by disk access speed that hints that it would fail if the disk was too busy/queued out too far. It would usually work when I waited two or three minutes before turning it on.
Sometimes wifi would crap out in the middle of a session, this was pretty normal when downloading a new nightly with firefox or dolphin, though strangely not when downloading with opera. When this happened gapps would immediately crash/FC almost without fail. The browser would also crash and then FC when I restart it, over and over, until I reboot...(I never checked if the kernel remounted the drive as read only, but it may have if there were big enough problems, I was always too focused on getting the latest nightly) Wifi would stay errored until I rebooted too.
Now, like I said, I fixed all these issues by purchasing a class 2 SD card. I made no other changes. I literally dd'ed my current setup from the class 6 to the class 2 card. It's been working like a charm for 2 days, no issues.
This suggests two things:
1) at least some of the SODs are caused by slow flash chips. This is probably true for the emmc folks too, as flash chips are like snowflakes, each one is unique. Is there a benchmark tool out there that will allow us to evaluate the emmc on sequential and random read/write speed? That test would prove or falsify this hypothesis.
2) The devs who can't reproduce the SOD could do so by purchasing a class 6 or class 10 card. Shoot, I'd be happy to mail you mine, useless thing that it is.
Edit: see http://xdaforums.com/showthread.php?p=12964262 for a discussion on SD card speeds, benchmarks, and results/symptoms.
Edit 2: a corollary of item 2 above is that a possible solution to SOD is running off a decent SD card instead of emmc. Number one rule of debugging: if something works in one place and not another, always suspect the differences.
Last edited:
My class 6 sdcard crapped out after 5 weeks, then I saw this and ordered a class 4 from Amazon: http://xdaforums.com/showthread.php?t=1005633
There's links to some on Amazon here: http://androidforums.com/motorola-droid-x/348043-replacement-micro-sd-card-recommendations.html
There's links to some on Amazon here: http://androidforums.com/motorola-droid-x/348043-replacement-micro-sd-card-recommendations.html
VERY Cool
Wouldn't the key issue be if the diffierent NC builds have different eMMC chips - they're probably not using a random assortment for the very reasons that you point out...
...did you image from one uSD to the other? If so then this is something very different from the eMMC SOD I've seen at least, where you can transfer the SOD problem (or remove it) with a CWM backup (which is an image, essentially).
Does SOD occurrence correlate with different NC builds? I was under the impression that it didn't. And you don't just see variations in performance from a random assortment of chips, you see significant variations in performance from chips from the same WAFER, much less the same manufacturing lot. It's inherent in the manufacturing process. Fabs just work around it in flash chips by making them slightly oversized, testing them heavily, and simply marking each area with flaws as a bad block. That's why every SD card has a slightly different size. And there are flaws that affect speed directly, you just have to bottleneck the wrong circuit and it will affect a few blocks, a lot of blocks, or the whole chip. Would B&N's spec to their emmc chip provider have ruled all these out? Maybe. But these specs are usually specified with a mean and standard deviation, and there's always SOMEBODY that ends up with a 3 sigma part...
All I'm saying is that there's a lot of variation in hardware, and though we can work around a lot of it in software, it's not guaranteed that any specific workaround will always work.
Anyway.
Yes, I first tried to image directly from one card to the other, but the new card was slightly smaller than the old card (both marked 8G), so that failed. So I set up the new card with verygreen's installer, installed the latest nightly to get the partitions setup, then dd'ed partitions 1, 2, and 3 from the old SD directly over 1, 2, and 3 on the new card. I think those are /boot, /system, and /data. I then mounted both partition 4s (the new one was empty) and copied the files over directly. That partition is /sdcard, and simply gets all the space that's left after the other three partitions are setup, so that's the one that was a different size. So, yes, very close to a direct image, but not exactly. The devil may be in the details, of course.
Do you have a CWM backup that consistently gives SOD? You could MAIL that to dalingrin or fattire or somebody as something reproducible. Or, to save on postage, find some site that allows huge file transfers...
J Disk Benchmark 3.0 seems to be closer to what I'm looking for. Will do both emmc and sd card, can vary file sizes, but the smallest file size is 10 MB. I doubt that counts as small block size...shoot this would be a simple shell script. dd from /dev/random to a file, nested in a do loop...I guess it would be easier in python...
WiFi off
(Quoting myself below)
I have been having several SOD per day. I found I had to plug into charger to
revive NC - not much fun to use the NC this way.
I turned off WiFi using the system settings menu. I have not had a SOD since
(no SOD for ~48 hours). I turn on Wifi to use it, then turn it off when I am done.
I did see other posts with wifi off and no SOD.
[CPU spy reports 23 hours of deepsleep of about 25 hours since unplugged.
SystemPanel reported cpu usage with screen off has dropped, so I'm sure
applications updating/syncing are behaving very differently too.]
Hope this helps track down the problems.
(Quoting myself below)
I have been having several SOD per day. I found I had to plug into charger to
revive NC - not much fun to use the NC this way.
I turned off WiFi using the system settings menu. I have not had a SOD since
(no SOD for ~48 hours). I turn on Wifi to use it, then turn it off when I am done.
I did see other posts with wifi off and no SOD.
[CPU spy reports 23 hours of deepsleep of about 25 hours since unplugged.
SystemPanel reported cpu usage with screen off has dropped, so I'm sure
applications updating/syncing are behaving very differently too.]
Hope this helps track down the problems.
Anyone had SOD on Daligrin's 0609 OC kernel? I haven't had SOD for the 2 or so days since I started using it, and I regularly get 1 SOD a day.
Not sure it matters, but I have it set to 200/300/600/800/1100
Not sure it matters, but I have it set to 200/300/600/800/1100
I was having numerous SOD issues per day until yesterday when I changed the CPU governor from Conservative to OnDemand. Since then I have not had one SOD.
Deep sleep is still working. Left the nook last night at 95% battery remaining, was still at 95% this morning. WiFi is set to turn off with the screen, and I have a good number of widgets running.
Deep sleep is still working. Left the nook last night at 95% battery remaining, was still at 95% this morning. WiFi is set to turn off with the screen, and I have a good number of widgets running.
Thanks for your advice, it was helpful
well, I was just about ready to re-flash to stock B&N and start over, but lots of new stuff came out this weekend. I flash 3 new nightlys and dal's new kernel and app. so far my SOD has been to a minimum with some deep sleep. playing with the wifi on and off seems to all so help a bit. but can't seem to pin anything down though. there's talk about the new OC kernel being the reason for SOD and a modify cfs version of the kernel is available to try here:
http://xdaforums.com/showthread.php?t=925451&page=330
see post 3298
I'll try this tonight and report back.
hi guys,
this might seem a tad weird but i have completely eliminated SOD from my Nook Color by remapping the home button to "back" instead of "home". Prior to this, I kept getting SOD every other time I try to wake up the nook after leaving it alone for more than an hour
i read somewhere in the xda forums that people who always use their power button to wake their nook up never experience SOD and that i always hated the fact that the biggest key on the nook was useless ("home" is a useless key for me).
this gave me the inspiration to re-mapped the home button to correspond to the "back" key instead by editing /system/usr/keylayout/qwerty.kl (it's a linux text file. i used jota editor to edit it) since i cannot train myself to use the power button instead of the "n" button to wake the nook --->
EDIT: my signature lies. I'm using CM7 96N.
this might seem a tad weird but i have completely eliminated SOD from my Nook Color by remapping the home
button to "back" instead of "home". Prior to this, I kept getting SOD every other time I try to wake up the nook after leaving it alone for more than an houri read somewhere in the xda forums that people who always use their power button to wake their nook up never experience SOD and that i always hated the fact that the biggest key on the nook was useless ("home" is a useless key for me).
this gave me the inspiration to re-mapped the home
button to correspond to the "back" key instead by editing /system/usr/keylayout/qwerty.kl (it's a linux text file. i used jota editor to edit it) since i cannot train myself to use the power button instead of the "n" button to wake the nook --->EDIT: my signature lies. I'm using CM7 96N.
A
ace7196
Guest
Yep. No computer near, so only a last kmsg.
pastebin.com/AtMSvr7v
So fyi. Had troubles with 6/9 kernel, so went to 5/23. Had sod, so flashed back to 6/9 (which I think is captured in the log). Will try things out. Had sod since nightly 98. Only new app has been system panel. If I get sod again, will uninstall system and try again.
Last edited:
Similar threads
- Locked
Top Liked Posts
-
There are no posts matching your filters.
-
3Hey guys, per Dalingrin's posts in the CM7 Dev or General thread, he's been unable to reproduce Sleep of Death. A lot of folks have seen it but it's hard to pin down a pattern in forum posts, so I created a google spreadsheet form to track builds and configuration people are using.
Dal has looked over the form and gives his thumbs up, so let's fill it out! The idea here is to fill out for EVERY SOD you have, filling out the latest information. If you remember the details, feel free to fill out recent SOD info.
If anyone sees anything I should update/modify or add, let me know. Also, if anyone wants an Excel file from either or both forms for analysis, please let me know and I'll send it. One thing to note here is it's impossible to see folks who are saying No on the Results form vs their answers, so the raw data could be useful.
Anywho, link is here, please fill it out guys! This form is to query you on what build you're using and how you have some things set up. Please fill it out every time you have SOD, and folks who do NOT have SOD are also encouraged to fill it out!
https://spreadsheets.google.com/spr...ormkey=dFZrS0tOdTVwZ0pnaWpqSTQwaDFmcmc6MQ&ifq
And here's a summary page of the responses so far:
https://spreadsheets.google.com/viewanalytics?formkey=dFZrS0tOdTVwZ0pnaWpqSTQwaDFmcmc6MQ
New Form!
And there's also a second form to track some answers to Widgets and Apps that sync/run all the time below. I did set the answers as No/Yes so you should just be able to click the ones you have installed/use and the rest you can just skip over and it'll default to No. Probably only fill out this second form once, as this is just to query the apps/widgets folks are running.
I am looking for feedback on additional apps, but since I have SOD on this set of "stuff" I figure it's good enough to start with. Honestly my confidence that this will be what figures out SOD is fairly low, but it should just take a few minutes, unless you're really aligned with me and have a bunch of the same stuff
Like the previous form, folks who have and have not experienced SOD are encouraged to submit their data.
Appreciate it guys, hopefully it continues to help out the CM7 gang understand why we're seeing SOD.
https://spreadsheets.google.com/spr...US&formkey=dHRMVklTSTgzT1htRkZRYWFfTmQ5WHc6MQ
And here's the direct link to the results so far..
https://spreadsheets.google.com/viewanalytics?formkey=dHRMVklTSTgzT1htRkZRYWFfTmQ5WHc6MQ1so, I just want to make sure that re-flashing to original B&N rom 1.1 or 1.2 and then flash back to a cm7 nightly has worked for you in fixing your SOD and NDS issues. I have read this before but is a little reluctant to try because of all the work that is needed. do you think a factory reset can also work? or is it the same thing. maybe I'll just need to bite the bullet and just do it.
btw: how is your nook running now? is all still well.
Well everything was perfect until i installed dal's new kernel today. Then i had a SOD almost immediately afterwards. Did a little experimenting based on the circulating theory that SOD is actually awake but no screen. Plugged it into my PC - the PC ding-dongs and pulls up the nook's drive (suggesting that the nook is awake). I then held power on the nook for a hard shut down, the PC dong-dings as the nook disconnects. Then a regular startup rebooted. no problems since.
Honestly, it was misery for me before (i.e. SOD every time i turned my back on the damn thing), and well worth the effort. Once you have everything downloaded the whole process takes about 10-15 minutes of flashing (not least cos you can do it in one session) and then 5-10 minutes of restoring everything with TB. After that the only minor pain is resetting any widgets you had (TB does a better job of restoring your ADW desktop than ADW does... with TB restoring ADW with data gets everything but the widgets back).
The biggest heartache was the complications with CWM... I'm now not having any problems with it on 3.0.2.1 (i always use the manual restore though).
I had tried just a factory reset before, but that was not enough to fix the SOD. There's something constipating either /boot or /system that only seems to get cleaned out by taking it completely back to stock. If i was feeling particularly masochistic i guess i could restore old /systems or old /boots from backups i have of 96 which suffered SOD, but right now i'm happy and if it ain't broke...
I've just upgraded to 98 and flashed 06/09 over the top of that so i will keep you posted.
Hope that helps1
Does SOD occurrence correlate with different NC builds? I was under the impression that it didn't. And you don't just see variations in performance from a random assortment of chips, you see significant variations in performance from chips from the same WAFER, much less the same manufacturing lot. It's inherent in the manufacturing process. Fabs just work around it in flash chips by making them slightly oversized, testing them heavily, and simply marking each area with flaws as a bad block. That's why every SD card has a slightly different size. And there are flaws that affect speed directly, you just have to bottleneck the wrong circuit and it will affect a few blocks, a lot of blocks, or the whole chip. Would B&N's spec to their emmc chip provider have ruled all these out? Maybe. But these specs are usually specified with a mean and standard deviation, and there's always SOMEBODY that ends up with a 3 sigma part...
All I'm saying is that there's a lot of variation in hardware, and though we can work around a lot of it in software, it's not guaranteed that any specific workaround will always work.
Anyway.
Yes, I first tried to image directly from one card to the other, but the new card was slightly smaller than the old card (both marked 8G), so that failed. So I set up the new card with verygreen's installer, installed the latest nightly to get the partitions setup, then dd'ed partitions 1, 2, and 3 from the old SD directly over 1, 2, and 3 on the new card. I think those are /boot, /system, and /data. I then mounted both partition 4s (the new one was empty) and copied the files over directly. That partition is /sdcard, and simply gets all the space that's left after the other three partitions are setup, so that's the one that was a different size. So, yes, very close to a direct image, but not exactly. The devil may be in the details, of course.
Do you have a CWM backup that consistently gives SOD? You could MAIL that to dalingrin or fattire or somebody as something reproducible. Or, to save on postage, find some site that allows huge file transfers...1I was having numerous SOD issues per day until yesterday when I changed the CPU governor from Conservative to OnDemand. Since then I have not had one SOD.
Deep sleep is still working. Left the nook last night at 95% battery remaining, was still at 95% this morning. WiFi is set to turn off with the screen, and I have a good number of widgets running.1hi guys,
this might seem a tad weird but i have completely eliminated SOD from my Nook Color by remapping the home button to "back" instead of "home". Prior to this, I kept getting SOD every other time I try to wake up the nook after leaving it alone for more than an hour
i read somewhere in the xda forums that people who always use their power button to wake their nook up never experience SOD and that i always hated the fact that the biggest key on the nook was useless ("home" is a useless key for me).
this gave me the inspiration to re-mapped the home button to correspond to the "back" key instead by editing /system/usr/keylayout/qwerty.kl (it's a linux text file. i used jota editor to edit it) since i cannot train myself to use the power button instead of the "n" button to wake the nook --->
EDIT: my signature lies. I'm using CM7 96N.
New posts
-
-
-
[UNOFFICIAL][ROM][14] LineageOS 21- Latest: Skyblueborb
