Kaiser Kernel Thread - ONLY FOR KERNEL AND BUGS

scooter1556 · Feb 9, 2011

clemsyn said:
I meant the yaffs driver, the Nand driver I was working on somehow got better in detecting bad blocks (or it could be an err on my part) and is obviously not the way to go.

No, it didn't sound too hopeful, i take it the NAND chip has had it? I made a change to the initrd the other day which clears another directory on the data partition which is used for dalvik cache. Since then, whenever data has become corrupt clearing dalvik cache in the installer menu has recovered the system and kept data mostly, so i would assume this directory is probably written to frequently so gets corrupt more often than not, or it could be an issue with dalvik that is being overlooked.

l1q1d · Feb 9, 2011

clemsyn said:
I meant the yaffs driver, the Nand driver I was working on somehow got better in detecting bad blocks (or it could be an err on my part) and is obviously not the way to go.

I test that configuration and it seems quite stable (10 power cycle and many apps installed). I think that the bug is caused by the ecc:
- Nand driver doesn't handle the ecc
- yaffs create a new ecc from the data of nand driver
result:
nand corruption

with that configuration:
- Nand driver doesn't handle the ecc
- Yaffs doesn't create ecc list
- the erase chunck avoid problem (hopefully)

the bad news is: it's quite slow!

I will release tomorrow a new kernel with ipv6 and this changes (need some more testing) only for brave testers.

cycloid · Feb 9, 2011

I've been thinking for a few weeks now that cell-data-connection may be related to stability. As a bit of background info i live on a hill surrounded by trees, the connection is generally only 2 bars and if it's windy/rainy the signal can drop, but whenever I have signal it's Edge. Where I work (some 20 miles away) the voice signal is *always* a full 4 bars, but G3/GPRS is non existent.... and this is where it gets really interesting. Some days I have a full data signal, other days there's no data, but on some days there's data at one end of my desk but none at the other! I noticed this before I switched to android and on my previous WiMo phone, it must be atmospheric or something.

Anyway, I have had quite a few lockups or restarts, and I've been thinking that the sheer stress of inconsistent data connections - and my attempts to use them while at work, where we have a mean firewall so I have to use my phone for email/facebook - is messing with Android. I.e. one second the phone checks the data connection and thinks all is good, the next second the data connection is gone but the phone is now attempting to sync...Other times the phone can be stubbornly convinced that there's no data connection but if I pull the battery and reboot it finds the one I know for a fact is there (because I'm somewhere else and can see the mast) as soon as it wakes up.

clemsyn · Feb 9, 2011

l1q1d said:
I test that configuration and it seems quite stable (10 power cycle and many apps installed). I think that the bug is caused by the ecc:
- Nand driver doesn't handle the ecc
- yaffs create a new ecc from the data of nand driver
result:
nand corruption

with that configuration:
- Nand driver doesn't handle the ecc
- Yaffs doesn't create ecc list
- the erase chunck avoid problem (hopefully)

the bad news is: it's quite slow!

I will release tomorrow a new kernel with ipv6 and this changes (need some more testing) only for brave testers.

I do agree, the nand driver gives 10 bytes of ecc but I think yaffs gives 12 bytes of ecc. When I changed the nand driver to 12 bytes, all hell broke lose and found all kinds bad blocks. If you look closely at the nand driver, I think it has the software ecc disabled. Another way to go around this is to enable software ecc in the nand driver and let yaffs do its own ecc.

clemsyn · Feb 9, 2011

cycloid said:
Just a quick thumbs up, was on the #63 kernel and today upgraded to the #253 .. or whatever those numbers are

This latest build does seem smoother, faster and prettier. But that could be a placebo from reading all the other comments. Am system on NAND, data and swap on SD card partitions.

Main thing is that since Xmas when I put the #63 kernel on I've had zero, zip, nada data corruption, despite almost daily fallovers, lockups, white screens and swapping batteries. (which I'm hoping will all be a lot less now as I think I've finally figured out that I'm panel type 3!)

Keep it up people, thanks for your efforts.

If you are putting data on SD, I suggest you try my new kernel with Ext4 support in SD partition. I will update it soon to use barrier=1 for better data security while keeping the speed reasonable.

l1q1d · Feb 10, 2011

i added also the support for ext4 but i need the changes to the initrd, could you please send me them?

The new changelog will be:
- ext2/ext4 support
- adb working on windows
- ipv6 support
- bluetooth modules
- nand stable

i wait for your changes on ext4.

ferr2 · Feb 10, 2011

hangups

Hi there,

another hint. i noticed if i have data connection enable constantly, i have to reset the phone 2-4 times/day. if the data is turned off, it can alive days.

l1q1d · Feb 10, 2011

ferr2 said:
Hi there,

another hint. i noticed if i have data connection enable constantly, i have to reset the phone 2-4 times/day. if the data is turned off, it can alive days.

Ril problem

clemsyn · Feb 10, 2011

L1q1d, pls try these yaffs2 drivers.

Here is the link for my initrd

http://rapidshare.com/files/447222591/initrd.lzma
http://rapidshare.com/files/447223644/initrd.gz

Also with this initrd, I am unable to format the third partition with ext4 via installsys. The command I used were /bin/mke2fs -F -t ext4 *** it gives "could not invalid blocks count: error

l1q1d · Feb 10, 2011

sure, send me also the changes of initrd.

clemsyn · Feb 10, 2011

l1q1d said:
sure, send me also the changes of initrd.

OK, posted it with yaffs2 drivers.

scooter1556 · Feb 10, 2011

clemsyn said:
OK, posted it with yaffs2 drivers.

How old is the yaffs driver, is it from an old .2x kernel?

l1q1d · Feb 10, 2011

clemsyn said:
OK, posted it with yaffs2 drivers.

You didn't solved the mke2fs problem

busybox haven't the support for ext4 so we need to create mke2fs from the original sources.

tiagoclc · Feb 10, 2011

"Googling" for yaffs2 nand ecc, I found this:

Title:NAND + YAFFS2 - bad blocks
Link:http://lists.infradead.org/pipermail/linux-mtd/2009-May/025462.html
(UC Browser)

I think that might be useful...

Sent from my CyanogenMod Kaiser/Kaiser using XDA App

l1q1d · Feb 10, 2011

probably the nand issue is solved.
Test the latest version from here:
http://xdaforums.com/showthread.php?t=949387

tiagoclc · Feb 10, 2011

Thanks!
But I have bad news...
Lost my contacts after reboot using soft reset button.

Sent from my CyanogenMod Kaiser/Kaiser using XDA App

msichal · Mar 15, 2011

The main issues I have:
-sometimes can't wake from sleep
-is the backup/recovery even working? It never worked for me
-battery usage, of course.

Is the CPU frequency changed when going sleep? I think it should, maybe it should be compiled in kernel or something? Maybe a smart roguetools-like app to dynamically change the freq?

masteryoda · Apr 4, 2011

is this working with ipv6?

Heiligkuh · May 5, 2011

I think you guys may be on to something suspecting the data connection having something to do with the data corruption. I have been running Not So Super Froyo on a Kaiser for the past 4 months or so and noticed something. As with everyone else I have suffered with the corruption of the data partition on NAND every couple of days.

I switched to Clemsyn's 2.6.32.9 kernel from post 61 a couple months ago and it was better but still corrupted the data partition a little more often than I like although I learned to live with it by backing up important info regularly.

Anyway, recently I was able to get my phone to go a whole 12 days without corruption by turning off background data and disabling backup my data in privacy settings. I always disable the cellular data connection because the sim card I use does not have a data plan (company phone).

I'm thinking maybe Android is trying to do backups or send data in the background when there is no data connection and it is corrupting the data partition somehow.

It may be related to the problems we have with this kernel not wanting to wake up from standby at times and requiring a reset.

I did notice that with background data and automatic backups disabled the phone would recover from a lockup with little if any corruption. Over the 12 days it ran without corruption I would say I rebooted the phone on average twice a day due to lockups.

Before this a reboot would almost certainly mean loss of data if not total corruption of the data partition.

Anyone else notice that the Kaiser tends to lock up more on cloudy/rainy days or when you are in an area where cell coverage is poor, especially if you use GPS?

How about issues with bluetooth especially when disconnecting and re-connecting?

Does turning bluetooth on/off mess with your WiFi connection?

At any rate keep up the good work, you guys have done a great job breathing new life into the old Kaiser.

Krazy-Killa · May 5, 2011

Heiligkuh said:
I think you guys may be on to something suspecting the data connection having something to do with the data corruption. I have been running Not So Super Froyo on a Kaiser for the past 4 months or so and noticed something. As with everyone else I have suffered with the corruption of the data partition on NAND every couple of days.

I switched to Clemsyn's 2.6.32.9 kernel from post 61 a couple months ago and it was better but still corrupted the data partition a little more often than I like although I learned to live with it by backing up important info regularly.

Anyway, recently I was able to get my phone to go a whole 12 days without corruption by turning off background data and disabling backup my data in privacy settings. I always disable the cellular data connection because the sim card I use does not have a data plan (company phone).

I'm thinking maybe Android is trying to do backups or send data in the background when there is no data connection and it is corrupting the data partition somehow.

It may be related to the problems we have with this kernel not wanting to wake up from standby at times and requiring a reset.

I did notice that with background data and automatic backups disabled the phone would recover from a lockup with little if any corruption. Over the 12 days it ran without corruption I would say I rebooted the phone on average twice a day due to lockups.

Before this a reboot would almost certainly mean loss of data if not total corruption of the data partition.

Anyone else notice that the Kaiser tends to lock up more on cloudy/rainy days or when you are in an area where cell coverage is poor, especially if you use GPS?

How about issues with bluetooth especially when disconnecting and re-connecting?

Does turning bluetooth on/off mess with your WiFi connection?

At any rate keep up the good work, you guys have done a great job breathing new life into the old Kaiser.

Data corruption is because of a buggy NAND driver, not because of something in Android. We know of the issue, we just don't have spare phones to break to attempt fixing the NAND driver. Also we've been working on temporary work arounds to help with data corruption (ECC checking, filesystem ECC checking, etc), but so far they've yielded minimal help because the actual NAND driver is not reading/writing data properly, even though it thinks it is.

As far as everything else. I believe is just coincidence, as the kernels are still buggy, and there will be crashes, so the fact that it "crashes more on cloudy/rainy days" is just superstition.

Bluetooth, at least with my kernel is because the phone sleeps it shuts EVERYTHING down including Bluetooth when it's suppose to stay on. Like I said, most of the issues we are aware of, and we are fixing them, and the only one I believe won't be fixed anytime soon is the NAND driver as it's buggy, and anytime one of us attempts to fix, we corrupt our NAND chips to the point our phone's are not even usable.

Kaiser Kernel Thread - ONLY FOR KERNEL AND BUGS

Senior Member

Retired Recognized Developer

Senior Member

Senior Member

Senior Member

Retired Recognized Developer

New member

Retired Recognized Developer

Senior Member

Attachments

Retired Recognized Developer

Senior Member

Senior Member

Retired Recognized Developer

Senior Member

Retired Recognized Developer

Senior Member

Member

Member

Member

Senior Member

Similar threads

Top Liked Posts