Originally Posted by clemsyn

its on /system/bin/userinit.sh

You sure ? i only find a remount there ... to fix su

Has anyone recently synchronized our yaffs2 tree with the main one on www.yaffs.net? I noticed there are quite a few differences. I'm testing, too soon to tell if there are any notable differences.

I think scooter synchronised it a little over a month ago? I know its been about that long.

Originally Posted by daedric

You sure ? i only find a remount there ... to fix su

you can add this to userinit.sh

busybox mount -o remount,rw,nouid,nodiratime,nodev /data

I think it depends on the rom but my ROM have those options. I'll upload my userinit.sh

Here is a copy of my userinit.sh

BTW, I added noatime on 2.6.25 kernel (which is suppose to be stable) and I got a Reboot and data corruption. I think if we mount 2.6.32 like 2.6.25's default mount, it could be better but it has to be done in the kernel itself coz 2.6.32 defaults to relatime.

Originally Posted by aceoyame

I think scooter synchronised it a little over a month ago? I know its been about that long.

I synchronized my yaffs sources yesterday. It survived several battery-pulls yesterday with no corruption, but one single clean shutdown modified email settings for one account and made Maps FC. A dalvik clear cured maps but not the email settings.

Yaffs has data recovery capabilities, as proven by the battery pulls that gave no corruption at all.
So i think yaffs is good and nand is good. The data corruption problem lies outside the kernel and is a subject for a different thread.

Sent from my Android on HTC Kaiser/Kaiser/Polaris using XDA App

Originally Posted by n2rjt

I synchronized my yaffs sources yesterday. It survived several battery-pulls yesterday with no corruption, but one single clean shutdown modified email settings for one account and made Maps FC. A dalvik clear cured maps but not the email settings.

Yaffs has data recovery capabilities, as proven by the battery pulls that gave no corruption at all.
So i think yaffs is good and nand is good. The data corruption problem lies outside the kernel and is a subject for a different thread.

Sent from my Android on HTC Kaiser/Kaiser/Polaris using XDA App

I read this quote about atime since you had problems with email on reboot

1. atime – This option causes Linux to record the last (or latest) time when a particular file was accessed. This information is particularly helpful for sysadmins or some programs (specially mail programs) to know when a particular file was last used/accessed.

So I think here is were atime option would help (email programs)

Originally Posted by clemsyn

I read this quote about atime since you had problems with email on reboot

1. atime – This option causes Linux to record the last (or latest) time when a particular file was accessed. This information is particularly helpful for sysadmins or some programs (specially mail programs) to know when a particular file was last used/accessed.

So I think here is were atime option would help (email programs)

I think you're confused... the e-mail problem n2rjt was having was complete random. He had problems in e-mail and maps.. could be browser and settings. It's random.

Now:

1. atime – This option causes Linux to record the last (or latest) time when a particular file was accessed. This information is particularly helpful for sysadmins or some programs (specially mail programs) to know when a particular file was last used/accessed.

2. noatime – This option stops recording the last file access time when the file is just read. The noatime option eliminates all the writes to the disk each time a file was just read which previously used to happen with atime. However a write is made to a disk in case if a file is being changed/written.

3. relatime – A filesystem mount with this option causes the access time to be updated if they are (before the update has occurred) earlier than the modification time. This significantly cuts down the writes caused by atime updates. However not many people use this option because they are simply not aware of it.

source: http://linux.koolsolutions.com/2009/...mount-options/

Indeed relatime is the most usefull option for us.

Yet... reading this for the 2nd time:

Yaffs has data recovery capabilities, as proven by the battery pulls that gave no corruption at all.
So i think yaffs is good and nand is good. The data corruption problem lies outside the kernel and is a subject for a different thread.

Could it be... a proc atempting to unmount /data or remount,ro that causes a defect ?

One other thing... you're sync'ing to yaffs.net, would it be too bad/dificult to use the yaffs2 provided with usual HTC sources ?
I'm looking at Supersonic... it uses our 2.6.32 kernel....

Originally Posted by daedric

One other thing... you're sync'ing to yaffs.net, would it be too bad/dificult to use the yaffs2 provided with usual HTC sources ?

The yaffs in the kernel at the minute is from the official android source, so what all other HTC devices are using

I tried the yaffs.net latest just as a comparison. It didn't change anything one way or the other.

Sent from my Android on HTC Kaiser/Kaiser/Polaris using XDA App

Trying to setup 2.6.32 kernel over 2.6.25 on my tytnII with Warm Donut on board.
Using Atools. Doing everything step-by-step, but when i reach this steps:

33 then flash from sd (gray)
34 reset device when it finished

it's booting from sd, (gray), show "Loading.." and never complete :(
have wait more than hour.

What's possible reason?
Using zImage-11-27-10 from l1qu1d sourceforge.net site.

Thanks in advance for help.