[U-Boot] U-boot and JFFS2

[U-Boot] U-boot and JFFS2

[Pillai, Manikandan]

Hi,

On the OMAP3 EVM board, I have four partition on my Onenand flash of which /dev/mtd4 is a JFFS2 partition.

After booting up linux, I use the following command to mount the jffs2 empty partition.

Linux#> mount -t jffs2 /dev/mtdblock4 /mnt

Then I do a reboot .

When u-boot boots up it gives bad erase block for all the blocks used by /dev/mtd4.

The reason is the JFFS2 filesystem is putting some signature at location 0x0 and 0x1 of OOB locations
of every block.

Is there a way this can be avoided ?



Regards
Mani

[U-Boot] U-boot and JFFS2

[Nishanth Menon]

Pillai, Manikandan said the following on 03/16/2009 01:35 PM:
> Hi,
>
> On the OMAP3 EVM board, I have four partition on my Onenand flash of which /dev/mtd4 is a JFFS2 partition.
>
> After booting up linux, I use the following command to mount the jffs2 empty partition.
>
> Linux#> mount -t jffs2 /dev/mtdblock4 /mnt
>
> Then I do a reboot .
>
> When u-boot boots up it gives bad erase block for all the blocks used by /dev/mtd4.
>
> The reason is the JFFS2 filesystem is putting some signature at location 0x0 and 0x1 of OOB locations
> of every block.
>
> Is there a way this can be avoided ?
>
>   
A)
mount -t jffs2 /dev/mtdblock4 /mnt
umount /mnt
sync
reboot

this gives the garbage collector time to put the cleanmarkers in place.

B) is the kernel's oob layout and ECC strategy (h/w vs s/w ecc) same as
that used for u-boot?

Regards,
Nishanth Menon

[U-Boot] U-boot and JFFS2

[Pillai, Manikandan]

Hi,

The cleanmarkers are being written in location 0x0 and 0x1 of the OOB space
And they are the problem since when u-boot comes up it does a scan and decided
That since these OOB locations do not contain 0xff, the blocks are bad.

Regards
Manikandan

> -----Original Message-----
> From: Nishanth Menon [mailto:menon.nishanth at gmail.com]
> Sent: Monday, March 16, 2009 5:09 PM
> To: Pillai, Manikandan
> Cc: u-boot at lists.denx.de
> Subject: Re: [U-Boot] U-boot and JFFS2
> 
> Pillai, Manikandan said the following on 03/16/2009 01:35 PM:
> > Hi,
> >
> > On the OMAP3 EVM board, I have four partition on my Onenand flash of which
> /dev/mtd4 is a JFFS2 partition.
> >
> > After booting up linux, I use the following command to mount the jffs2 empty
> partition.
> >
> > Linux#> mount -t jffs2 /dev/mtdblock4 /mnt
> >
> > Then I do a reboot .
> >
> > When u-boot boots up it gives bad erase block for all the blocks used by
> /dev/mtd4.
> >
> > The reason is the JFFS2 filesystem is putting some signature at location 0x0
> and 0x1 of OOB locations
> > of every block.
> >
> > Is there a way this can be avoided ?
> >
> >
> A)
> mount -t jffs2 /dev/mtdblock4 /mnt
> umount /mnt
> sync
> reboot
> 
> this gives the garbage collector time to put the cleanmarkers in place.
> 
> B) is the kernel's oob layout and ECC strategy (h/w vs s/w ecc) same as
> that used for u-boot?
> 
> Regards,
> Nishanth Menon

[U-Boot] U-boot and JFFS2

[Nishanth Menon]

Pillai, Manikandan said the following on 03/16/2009 01:41 PM:
> The cleanmarkers are being written in location 0x0 and 0x1 of the OOB space
> And they are the problem since when u-boot comes up it does a scan and decided
> That since these OOB locations do not contain 0xff, the blocks are bad.
>   
offset 0 and 1 in a x16 large page micron device is reserved for
badblock marker - u-boot is checking the correct location and if those
locations dont have 0xff, it is noted as bad blocks. if cleanmarker is
being written there, you'd need to fix the kernel board file etc..
Regards,
Nishanth Menon

[U-Boot] U-boot and JFFS2

[Pillai, Manikandan]

Hi,

offset 0 and 1 in a x16 large page micron device is reserved for
badblock marker - u-boot is checking the correct location and if those
locations dont have 0xff, it is noted as bad blocks - This is no issue
with u-boot as far as I can see but on the other hand the clean markers
are also supposed to be in the same location.

I am not clear on the other part of the answer. Do you mean to say I need to fix the kernel code for cleanmarker issue ? Is it like I can give an offset for the
Cleanmarker to be written ?

regards
Mani

> -----Original Message-----
> From: Nishanth Menon [mailto:menon.nishanth at gmail.com]
> Sent: Monday, March 16, 2009 5:15 PM
> To: Pillai, Manikandan
> Cc: u-boot at lists.denx.de
> Subject: Re: [U-Boot] U-boot and JFFS2
> 
> Pillai, Manikandan said the following on 03/16/2009 01:41 PM:
> > The cleanmarkers are being written in location 0x0 and 0x1 of the OOB space
> > And they are the problem since when u-boot comes up it does a scan and
> decided
> > That since these OOB locations do not contain 0xff, the blocks are bad.
> >
> offset 0 and 1 in a x16 large page micron device is reserved for
> badblock marker - u-boot is checking the correct location and if those
> locations dont have 0xff, it is noted as bad blocks. if cleanmarker is
> being written there, you'd need to fix the kernel board file etc..
> Regards,
> Nishanth Menon

[U-Boot] U-boot and JFFS2

[Nishanth Menon]

Pillai, Manikandan said the following on 03/17/2009 06:41 AM:
> with u-boot as far as I can see but on the other hand the clean markers
> are also supposed to be in the same location.
>   
Why do you say that? ooblayout free is used if I am not wrong to store
cleanmarkers. have you compared the layouts of the kernel Vs u-boot? if
there is a delta, I suppose a good idea is to fix the kernel board file
(we've had more than enough oob discussions for omap3 gpmc i suppose on
u-boot ml ;) )..
> I am not clear on the other part of the answer. Do you mean to say I need to fix the kernel code for cleanmarker issue ? Is it like I can give an offset for the
> Cleanmarker to be written ?
>   
I believe you can, essentially coz you can choose where the ecc region
is, jffs2 should store cleanmarkers in free area.
Regards,
Nishanth Menon