OK so just to give a short description I'm setting up a thread to store my ramblings and keep track of all MK908II issues and solutions. Hopefully this will give proper info not scattered between loads of threads or forums.
The start of my problems
I can run android first boot after a flash everytime but after doing not much with it apps start to not load or crash for no reason then after a few reboots it starts to hang just before desktop is reached.
I can confirm this as the keyboard is recognised (caps lock light works) and mouse pointer can be seen (when mouse is moved).
I beleive there must be data corruption on the nand but android is unaware of the bad blocks because the nand itself hasnt marked them as bad (thus avoiding using them)
Now im of the understanding that if i can format the nand it will detect bad blocks (NAND storage always has them they are just made that way) and avoid using them using its own error checking built into the NAND
I can achieve maskrom mode several ways but still a format (eraseidb) fails.
It would appear that any of the tools available to flash these sticks are unable to issue the correct commands to perform a format of the nand as erase IDB achieves nothing but reports all went well unless you dig through the logs.
Now the nand chip used on this stick appears to have a few differnces compared to the chips used in other sticks. Other sticks may use foresee memory and if they do any info would be greatly recieved.
Ive been sent the specification sheet from foresee but dont want to post on here as it says confidential on it (dont know if its relevant and dont want to upset them seeing as they look willing to help).
Ive got a few theories that may overcome this problem
The rk batch tool can write to the nand but not format it, is there a way of writing zeros to the NAND?
Could I in theory build a 6GB img file (rough size) of nothing and flash that first before a real system flash?
Could it be more than just 2 pins needs shorting to enter a TRUE maskrom mode eg:- 4 or 6 or more then a true format can be achieved?
Is a change or alteration in code needed for the flashing programs to issue the correct commands to the nand or would it require a driver tweak?
If anyone has any ideas please post i can give alot of info on the nand chip itself
I have a suspicion that the errors have come from when the flashing software has repartitioned the nand when changing to a 4GB param file but again thats just a guess.
If there's a way around this without soldering a fresh nand chip on i would do that as that makes these sticks as standard very unreliable
I've installed picuntu and don't appear to have any issues with it but havnt been running it long this has made 1 big 6GB partition. In the hope of running some disk type tools. A re-flash after this could achieve something but maybe not answer all my questions
As an additional note those that are suffering stutter on HD content on lan or wifi should take note that the spec sheet states these nand chips only write at 6MB/s and read at 12MB/s and that will be in ideal conditions. 54mb wifi theoretical max is 6.75MB and a 100mb connection is 12.5MB so im guessing theres not much room for all the io if any!!!
I may have solved the bad nand issue further testing needed but im pretty confident *8O)
Im 90% certain ive solved any non boot issues with this stick by flashing in a non standard way so far so good
I will continue to write what i find over time and edit this post accordingly
The start of my problems
I can run android first boot after a flash everytime but after doing not much with it apps start to not load or crash for no reason then after a few reboots it starts to hang just before desktop is reached.
I can confirm this as the keyboard is recognised (caps lock light works) and mouse pointer can be seen (when mouse is moved).
I beleive there must be data corruption on the nand but android is unaware of the bad blocks because the nand itself hasnt marked them as bad (thus avoiding using them)
Now im of the understanding that if i can format the nand it will detect bad blocks (NAND storage always has them they are just made that way) and avoid using them using its own error checking built into the NAND
I can achieve maskrom mode several ways but still a format (eraseidb) fails.
It would appear that any of the tools available to flash these sticks are unable to issue the correct commands to perform a format of the nand as erase IDB achieves nothing but reports all went well unless you dig through the logs.
Now the nand chip used on this stick appears to have a few differnces compared to the chips used in other sticks. Other sticks may use foresee memory and if they do any info would be greatly recieved.
Ive been sent the specification sheet from foresee but dont want to post on here as it says confidential on it (dont know if its relevant and dont want to upset them seeing as they look willing to help).
Ive got a few theories that may overcome this problem
The rk batch tool can write to the nand but not format it, is there a way of writing zeros to the NAND?
Could I in theory build a 6GB img file (rough size) of nothing and flash that first before a real system flash?
Could it be more than just 2 pins needs shorting to enter a TRUE maskrom mode eg:- 4 or 6 or more then a true format can be achieved?
Is a change or alteration in code needed for the flashing programs to issue the correct commands to the nand or would it require a driver tweak?
If anyone has any ideas please post i can give alot of info on the nand chip itself
I have a suspicion that the errors have come from when the flashing software has repartitioned the nand when changing to a 4GB param file but again thats just a guess.
If there's a way around this without soldering a fresh nand chip on i would do that as that makes these sticks as standard very unreliable
I've installed picuntu and don't appear to have any issues with it but havnt been running it long this has made 1 big 6GB partition. In the hope of running some disk type tools. A re-flash after this could achieve something but maybe not answer all my questions
As an additional note those that are suffering stutter on HD content on lan or wifi should take note that the spec sheet states these nand chips only write at 6MB/s and read at 12MB/s and that will be in ideal conditions. 54mb wifi theoretical max is 6.75MB and a 100mb connection is 12.5MB so im guessing theres not much room for all the io if any!!!
I may have solved the bad nand issue further testing needed but im pretty confident *8O)
Im 90% certain ive solved any non boot issues with this stick by flashing in a non standard way so far so good
I will continue to write what i find over time and edit this post accordingly
Comment