Bad sectors

[Takács Áron]

Hello,

I have the following problem: I am using a PXA27x based module with two 
16bit flash chips (Intel P33) parallel on it. We have been using these 
modules without any problems with linux-2.6 kernels and jffs2 filesystem 
on top of mtdblock.
Recently I moved to the 3.6-rc4 kernel. Since then the following error 
happened to two different modules: they were working for a while (some 
days, week), and then suddenly some blocks of the flash became 
uneraseable. I get a lot of such error messages while trying to dd the 
jffs2 image to /dev/mtdblockX:

[  478.796166] Buffer I/O error on device mtdblock2, logical block 2085
[  478.802472] lost page write due to I/O error on mtdblock2
[  479.068366] mtdblock: erase of region [0x300000, 0x40000] on 
"Filesystem" failed
[  479.075760] end_request: I/O error, dev mtdblock2, sector 16688

I have tried to erase the block with openocd, u-boot and WinCE too, but 
all of them failed.

The error occurred after a few writes to the flash (far less than 100k), 
so I don't think it was a wearing issue.

Is it theoretically possible the damage sectors by any software error 
(bad kernel config)? Or do I have to think of hardware defects (maybe ESD)?

Thank you very much for your help! It is very important to me to exclude 
software problems.

Best regards:

Áron