Well, yes I think the entire NAND flash would have to be rewritten. So maybe that's not a "nice" option, but supporting two kernel versions (or at least driver versions) is kind of ugly too. Maybe there is not a good solution.
Daniel J. Allred DSP Applications Applications Processors and Floating Point -----Original Message----- From: Narnakaje, Snehaprabha Sent: Tuesday, March 03, 2009 10:23 AM To: Allred, Daniel; David Brownell Cc: Nori, Sekhar; DaVinci Subject: RE: [patch davinci-git 2/3] NAND: Fix "raw" reads with ECC syndrome layouts Daniel, Wouldn't the customers having the legacy support in LSP releases have problem migrating to the upper software layers with new standard layout? For example a DM355 based device in the field has a MV based LSP 1.xx/2.xx release that uses the legacy layout in the upper software layers requires a firmware upgrade to the latest GIT based kernel. Not sure, if they have to remain on legacy layout or can migrate easily to the new standard layout. Thanks Sneha > -----Original Message----- > From: Allred, Daniel > Sent: Tuesday, March 03, 2009 11:11 AM > To: Narnakaje, Snehaprabha; David Brownell > Cc: Nori, Sekhar; DaVinci > Subject: RE: [patch davinci-git 2/3] NAND: Fix "raw" reads with ECC > syndrome layouts > > Sneha, > > Why is this an issue? The only component that absolutely must be written > in the legacy format is the UBL. Shouldn't all the upper software layers > migrate to the new layout, since they really should never touch that UBL > portion of the NAND? Or maybe they would during a BB scan? > > Regards, > Daniel > > Daniel J. Allred > DSP Applications > Applications Processors and Floating Point > -----Original Message----- > From: Narnakaje, Snehaprabha > Sent: Tuesday, March 03, 2009 9:56 AM > To: David Brownell > Cc: Allred, Daniel; Nori, Sekhar; DaVinci > Subject: RE: [patch davinci-git 2/3] NAND: Fix "raw" reads with ECC > syndrome layouts > > David, > > > -----Original Message----- > > From: David Brownell [mailto:davi...@pacbell.net] > > Sent: Tuesday, February 24, 2009 5:48 PM > > To: Narnakaje, Snehaprabha > > Cc: Allred, Daniel; Nori, Sekhar; DaVinci > > Subject: Re: [patch davinci-git 2/3] NAND: Fix "raw" reads with ECC > > syndrome layouts > > > > On Tuesday 24 February 2009, Narnakaje, Snehaprabha wrote: > > > It would be good, if your patches go into the davinci GIT. > > > We can then come up with the follow-up patches to fix the > > > ECC issues (ECC isn't working yet, with the current set of patches). > > > > I think only the $SUBJECT patch -- already submitted > > to the black-hole-which-is-the-MTD-list -- is really > > appropriate to merge. It's clearly a nand-core bug. > > > > So far as I can tell, the current DaVinci NAND driver > > (with no support for 4-bit ECC) is OK to merge as-is, > > with followup patches to handle 4-bit ECC mode. Which > > is, again, why that code was sent into that black hole. > > > > > > But ... I thought you were abandoning NAND_ECC_HW_SYNDROME? > > The layout Daniel had described was incompatible with that. > > No, we should still be using NAND_ECC_HW_SYNDROME option, since our ECC is > syndrome engine (with syndrome and parity bits) based. However, we would > override the *_syndrome() APIs in the davinci driver, as Sekhar mentioned. > > The real question is - how can we support both DM355 legacy layout > (current silicon versions) along with the standard layout. The current > DM355 silicon versions may require an upgrade path to migrate into a GIT > based Linux Kernel. > > Thanks > Sneha > > > > > > - Dave _______________________________________________ Davinci-linux-open-source mailing list Davinci-linux-open-source@linux.davincidsp.com http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
