Re: mtd errors on boot OpenRD "ultimate"

[Rick Thomas <rbthomas@pobox.com>]

Thanks for the confirmation, Alexander!

Would there be any point in trying to generate a Linux kernel patch to  
accept both kinds of ECC?

Rick

PS:  Can you explain (for us less knowledgeable folks) why they are  
harmless for mtdblock[01] but not for mtdblock[2]?


Alexander Coulter says:

> If I am right it's because u-boot does not support (or uses a  
> different
> ECC method) for the NAND to Linux.  When Linux reads blocks from the
> NAND the ECC check fails and you get an IO error.
>
> The messages are harmless though for mtdblock0 and mtdblock1, it's  
> only
> if they appear on mtdblock2 should you get worried :)


On Aug 6, 2011, at 11:42 PM, Rick Thomas wrote:

> Hi again,
>
> Some googling on the subject offers a suggestion that is attractive,  
> even if not easily verified:
>
> The suggestion is that Uboot and Linux do different and incompatible  
> checksums on the write-blocks in the NAND flash.  So you either get  
> messages in your syslog from Linux saying it doesn't like the Uboot- 
> style checksums, or you get messages from Uboot saying it doesn't  
> like the Linux-style checksums.  The Uboot messages won't show up in  
> syslog, but they will pollute your serial console during boot-up.   
> Pay your money and take your pick.  Both are unpleasant.
>
> Does anybody out there have true and certain knowledge if this  
> suggestion is even close?
>
> If so, would the new Debian Uboot package have Linux compatible  
> checksums?
>
> Rick
>
>
> On Aug 6, 2011, at 6:02 PM, Philippe Clérié wrote:
>
> > When I first encountered those messages I was told that they were  
> > completely
> > harmless and have always been there. You'd best ignore them.
> >
> > On Saturday 06 August 2011 13:31:24 Rick Thomas wrote:
> > > For a few seconds during the boot process on my OpenRD "ultimate",
> > > running Debian Squeeze, I get hundreds of messages in syslog like  
> > > this:
> > >
> > > Buffer I/O error on device mtdblock0, logical block 0
> > > Buffer I/O error on device mtdblock0, logical block 8
> > > end_request: I/O error, dev mtdblock0, sector 0
> > > end_request: I/O error, dev mtdblock0, sector 128
> > > end_request: I/O error, dev mtdblock0, sector 16
> > > end_request: I/O error, dev mtdblock0, sector 64
> > > end_request: I/O error, dev mtdblock0, sector 8
> > > loop: module loaded
> > > uncorrectable error :
> > >
> > >
> > > Does anybody know what causes them?  Is there anything I can (or
> > > should) do about them?
> > >
> > >
> > > Thanks!
> > >
> > > Rick