JFFS on a powerpc platform.

JFFS on a powerpc platform.

[Subodh Nijsure]

Hello,

Question I have is -- has anyone tested JFFS on powerpc platform.

I have retrieved the latest mtd code from CVS repository and 
the data read back from flash is not "correct".

The platform I am running on MPC860 based board having
intel flash  which mtd is able to recognize correctly.

Here are messages during the boot to indicate that cfi probe has
detected the flash. 

JFFS version 1.0, (C) 1999, 2000  Axis Communications AB
physmap.c physmap flash device: size 400000 at address 28000000
cfi_probe_new_chip map->buswidth 4
Custom flash device: Found 2 x16 CFI devices at location 0 in 16 bit mode
JEDEC ID: B0 D0

The /proc/mtd shows following output which matches the partition table I
have setup.

sh-2.04# cat /proc/mtd
dev:    size   erasesize  name
mtd0: 00400000 00020000 "Custom flash device"
mtd1: 00080000 00020000 "Bootrom Area"
mtd2: 00040000 00020000 "Custom Flash Bank #2 2x128KB"
mtd3: 00040000 00020000 "Custom Flash Bank #2 2x128KB"
mtd4: 00300000 00020000 "Custom Flash left overs from 4MB Flash" 

Now I execute command  and I see the following output
 
	mount -t jffs /dev/mtdblock4 /tmp/jffs

	cp /etc/inetd.conf /tmp/jffs/inetd.conf

	sh-2.04# cp /etc/inetd.conf /tmp/jffs
	sh-2.04# ls -l /etc/inetd.conf /tmp/jffs
	-rw-r-xr--   1 root     root         2880 Jan 10  2001 /etc/inetd.conf

	/tmp/jffs:
	total 3
	-rw-r-xr--   1 root     root         2880 Dec 31 16:05 inetd.conf  

However if I try to see contents of this /tmp/jffs/inetd.conf I see 
following output instead of my inetd.conf, 


sh-2.04# more /tmp/jffs/inetd.conf
#
#tenioc.dT      fn sihelifsed birct ses ehivre sectahtliw eb lava bali#
elht            guorht hNI e DTE/PCTs PIrepures .revoT  -er fnocrugi    #
eeht    nur gninENI p DTecor ,sstideiht if s ,elnehtnes ht d    #
eENI    p DTecora ssGIS  PUHngis

Also now if I reboot the system and remount the device /dev/mtdblock4 I 
don't see the file inetd.conf, what am I doing wrong?

/Subodh Nijsure


To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org

AW: JFFS on a powerpc platform.

[Florian Schirmer / TayTron]

Hi!

>Question I have is -- has anyone tested JFFS on powerpc platform.
>
>I have retrieved the latest mtd code from CVS repository and
>the data read back from flash is not "correct".
>
>The platform I am running on MPC860 based board having
>intel flash  which mtd is able to recognize correctly.

Yes the code runs fine - now. Please check wether you are using at least
version 1.10 of physmap.c. We fixed exactly this problem mhhh let me say
around 4 days ago. Did you checked out the source after that? Please
checkout again.

Hope that helps - if not please provide a DEBUG_CFI enabled output.

Happy hacking
   Florian Schirmer



To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org

Re: AW: JFFS on a powerpc platform.

[Subodh Nijsure]

> Hi!
> 
> >Question I have is -- has anyone tested JFFS on powerpc platform.
> >
> >I have retrieved the latest mtd code from CVS repository and
> >the data read back from flash is not "correct".
> >
> >The platform I am running on MPC860 based board having
> >intel flash  which mtd is able to recognize correctly.
> 
> Yes the code runs fine - now. Please check wether you are using at least
> version 1.10 of physmap.c. We fixed exactly this problem mhhh let me say
> around 4 days ago. Did you checked out the source after that? Please
> checkout again.
> 
> Hope that helps - if not please provide a DEBUG_CFI enabled output.

I got the latest code and one of the problem is fixed now. 

i.e. if I copy a file to jffs partition I can read back the content.

However if I mount partiton, copy a file to it, then unmount it and again
mount it, I don't see the file I copied.

e.g. I do

mount -t jffs /dev/mtdblock4 /tmp/jffs 
cp /etc/inetd.conf /tmp/jffs/inetd.conf

Now I do diff between /etc/inetd.conf and /tmp/jffs/inetd.conf both are
same files

Now I do umount and re-mount 
i.e.
umount jffs /dev/mtdblock4 
mount -t jffs /dev/mtdblock4 /tmp/jffs 

Now when I try to do ls -l /tmp/jffs I don't see the file I had copied earlier,
anybody seen this?

Thanks.

/Subodh Nijsure



To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org

Re: JFFS on a powerpc platform.

[Florian Schirmer / TayTron]

Hi!

>I got the latest code and one of the problem is fixed now.
>i.e. if I copy a file to jffs partition I can read back the content.
>However if I mount partiton, copy a file to it, then unmount it and again
>mount it, I don't see the file I copied.

Yes this was actually a problem with the endianess problem you described
before. (jffs did not find its header infos anymore, since they where
byteswapped). Please try to eraseall the partition before you mount it. You
can find this tool under /utils of the mtd cvs tree.

eraseall /dev/mtd/0
mount -t jffs /dev/mtdblock/0 /mnt
cp foo /mnt/bar
umount /mnt
mount /dev/mtdblock/0 /mnt
ls /mnt

Gives me correct infos. So it works here!

Bye
   Florian Schirmer



To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org

Re: AW: JFFS on a powerpc platform.

[Subodh Nijsure]

I still have that problem now its slightly different problem. 

It looks like problem with 'interleaving' 

This is how my mtd memory space looks like -- 

cat /proc/mtd
mtd0: 00400000 00020000 "Physically mapped flash"
mtd1: 00080000 00020000 "Bootrom Area"
mtd2: 00040000 00020000 "Custom Flash Bank #2 2x128KB"
mtd3: 00040000 00020000 "Custom Flash Bank #2 2x128KB"
mtd4: 00300000 00020000 "Custom Flash left overs from 4MB Flash" 

Now the first device /dev/mtd1 contains the bootrom so I copy /dev/mtd1
to file and it looks like this --

cat < /dev/mtd1 > /tmp/bootrom 
'V|~?&tcN |~N MTDSkyStream Bootrom (ver 0.1) (Mar  8 2001 - 10:26:50) 

doc_loadbios /dev/mtd4 bootrom 

Now if I try to dump the /dev/mtd4 I get following output 
cat < /dev/mtd4 | more
'&?~|ct N&| NSDTMtSykmaerooB mortev( .0 r( )1 raM2 9  10061 -:52:)44  


Now if I do eraseall /dev/mtd0 I see data in flash as

FFFF0080 FFFF0080.. 

instead of 
FFFFFFFF FFFFFFFF...

So it looks like for some reason one of the chip is not getting erased?

In my kernel config as interleave '2' as I have two chips giving me 32 bit
access.

CONFIG_MTD_CFI=y
CONFIG_MTD_CFI_GEOMETRY=y
# CONFIG_MTD_CFI_B1 is not set
# CONFIG_MTD_CFI_B2 is not set
CONFIG_MTD_CFI_B4=y
# CONFIG_MTD_CFI_I1 is not set
CONFIG_MTD_CFI_I2=y
# CONFIG_MTD_CFI_I4 is not set
CONFIG_MTD_CFI_INTELEXT=y 

Interleaving factor is '2' as my CPU A1 is connected to A0 of the chip.

Any more hints on why I would be getting this "flipping" of bits?


Thanks!!

/Subodh



To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org