latest MTD snapshot

latest MTD snapshot

[Luca Contini]

Hi,
I've downloaded the latest snapshot from the CVS:
mtd-snapshot-20030828.tar.bz2
Since I need only the mtdblock device working I've just copied the snapshot
/drivers/mtd content to my /linux/drivers/mtd directory, and the blktrans.h
to /include/linux/mtd/.
When I try to compile (using arm-linux-gcc) mtdblock.c I get this warning:
"mtdblock.c:251: warning: implicit declaration of function `unlikely'"
and when I try to insert the mtdblock.o module I get:

[arm@Botolo2 mtd]$insmod mtdblock.o
mtdblock.o: unresolved symbol unlikely
mtdblock.o: unresolved symbol register_mtd_blktrans
mtdblock.o: unresolved symbol add_mtd_blktrans_dev
mtdblock.o: unresolved symbol deregister_mtd_blktrans
mtdblock.o: unresolved symbol del_mtd_blktrans_dev
[arm@Botolo2 mtd]$

What Am I missing?

Regards

Luca

Re: latest MTD snapshot

[J.D. Bakker]

At 15:28 +0200 29-08-2003, Luca Contini wrote:
>Hi,
>I've downloaded the latest snapshot from the CVS:
>mtd-snapshot-20030828.tar.bz2
>Since I need only the mtdblock device working I've just copied the snapshot
>/drivers/mtd content to my /linux/drivers/mtd directory, and the blktrans.h
>to /include/linux/mtd/.
>When I try to compile (using arm-linux-gcc) mtdblock.c I get this warning:
>"mtdblock.c:251: warning: implicit declaration of function `unlikely'"
>and when I try to insert the mtdblock.o module I get:
>
>[arm@Botolo2 mtd]$insmod mtdblock.o
>mtdblock.o: unresolved symbol unlikely
>mtdblock.o: unresolved symbol register_mtd_blktrans
>mtdblock.o: unresolved symbol add_mtd_blktrans_dev
>mtdblock.o: unresolved symbol deregister_mtd_blktrans
>mtdblock.o: unresolved symbol del_mtd_blktrans_dev
>[arm@Botolo2 mtd]$
>
>What Am I missing?

1) Which kernel version ?
2) Which compiler version (arm-linux-gcc -v) ?
3) Any particular reason why you don't use the patchin.sh script ? I 
know, you 'only' need mtdblock, but partial installs tend to create 
no end of trouble.

HTH,

JDB.
-- 
LART. 250 MIPS under one Watt. Free hardware design files.
http://www.lart.tudelft.nl/

Re: latest MTD snapshot

[Luca Contini]

> At 15:28 +0200 29-08-2003, Luca Contini wrote:
> >Hi,
> >I've downloaded the latest snapshot from the CVS:
> >mtd-snapshot-20030828.tar.bz2
> >Since I need only the mtdblock device working I've just copied the
snapshot
> >/drivers/mtd content to my /linux/drivers/mtd directory, and the
blktrans.h
> >to /include/linux/mtd/.
> >When I try to compile (using arm-linux-gcc) mtdblock.c I get this
warning:
> >"mtdblock.c:251: warning: implicit declaration of function `unlikely'"
> >and when I try to insert the mtdblock.o module I get:
> >
> >[arm@Botolo2 mtd]$insmod mtdblock.o
> >mtdblock.o: unresolved symbol unlikely
> >mtdblock.o: unresolved symbol register_mtd_blktrans
> >mtdblock.o: unresolved symbol add_mtd_blktrans_dev
> >mtdblock.o: unresolved symbol deregister_mtd_blktrans
> >mtdblock.o: unresolved symbol del_mtd_blktrans_dev
> >[arm@Botolo2 mtd]$
> >
> >What Am I missing?
>
> 1) Which kernel version ?

2.4.18-rmk6-swl6

> 2) Which compiler version (arm-linux-gcc -v) ?

2.95.2

> 3) Any particular reason why you don't use the patchin.sh script ? I
> know, you 'only' need mtdblock, but partial installs tend to create
> no end of trouble.

I just thought it was easiest to do it manually just for mtdblock device.
I've copied the snapshot to ../linux/drivers/mtd/util/snapshot and tried the
patchin.sh:
[root@Lab patches]# patchin.sh
Directory  does not exist or is not a kernel source directory

this is the result

May be the location I copied the snapshot to is wrong.

Regards

>
> HTH,
>
> JDB.
> -- 
> LART. 250 MIPS under one Watt. Free hardware design files.
> http://www.lart.tudelft.nl/
>
>

Re: latest MTD snapshot

[J.D. Bakker]

At 15:56 +0200 29-08-2003, Luca Contini wrote:
>  > At 15:28 +0200 29-08-2003, Luca Contini wrote:
>  > >What Am I missing?
>>
>>  1) Which kernel version ?
>
>2.4.18-rmk6-swl6

That's a pretty ancient kernel (almost 18 months old). You may want 
to retry with 2.4.21-rmk1 (works fine on the LART, which is also 
arm-based).

>  > 3) Any particular reason why you don't use the patchin.sh script ? I
>>  know, you 'only' need mtdblock, but partial installs tend to create
>>  no end of trouble.
>
>I just thought it was easiest to do it manually just for mtdblock device.
>I've copied the snapshot to ../linux/drivers/mtd/util/snapshot and tried the
>patchin.sh:
>[root@Lab patches]# patchin.sh
>Directory  does not exist or is not a kernel source directory

You must give it the path to your kernel tree as an argument (as 
documented on http://www.linux-mtd.infradead.org/).

HTH,

JDB.
-- 
LART. 250 MIPS under one Watt. Free hardware design files.
http://www.lart.tudelft.nl/

Re: latest MTD snapshot

[David Woodhouse]

On Fri, 2003-08-29 at 16:07 +0200, J.D. Bakker wrote:
> That's a pretty ancient kernel (almost 18 months old). You may want 
> to retry with 2.4.21-rmk1 (works fine on the LART, which is also 
> arm-based).

Just "#define unlikely(x) (x)" will suffice, although you may find other
problems with a kernel that old.

> >[root@Lab patches]# patchin.sh
> >Directory  does not exist or is not a kernel source directory
> 
> You must give it the path to your kernel tree as an argument (as 
> documented on http://www.linux-mtd.infradead.org/).

Urgh. It'll also _tell_ you that for itself now...

hades /home/dwmw2/working/mtd/patches $ ./patchin.sh
usage:  ./patchin.sh [-c] [-j] kernelpath
   -c  -- copy files to kernel tree instead of building links
   -j  -- include jffs2 filesystem


-- 
dwmw2

Re: latest MTD snapshot

[Luca Contini]

> At 15:56 +0200 29-08-2003, Luca Contini wrote:
> >  > At 15:28 +0200 29-08-2003, Luca Contini wrote:
> >  > >What Am I missing?
> >>
> >>  1) Which kernel version ?
> >
> >2.4.18-rmk6-swl6
>
> That's a pretty ancient kernel (almost 18 months old). You may want
> to retry with 2.4.21-rmk1 (works fine on the LART, which is also
> arm-based).

I'll try. But I need a patch from samsung (my board is Samsung SMDK2410 with
ARM920T) and I don't know whether the current patch works fine with a newer
kernel.
>
> >  > 3) Any particular reason why you don't use the patchin.sh script ? I
> >>  know, you 'only' need mtdblock, but partial installs tend to create
> >>  no end of trouble.
> >
> >I just thought it was easiest to do it manually just for mtdblock device.
> >I've copied the snapshot to ../linux/drivers/mtd/util/snapshot and tried
the
> >patchin.sh:
> >[root@Lab patches]# patchin.sh
> >Directory  does not exist or is not a kernel source directory
>
> You must give it the path to your kernel tree as an argument (as
> documented on http://www.linux-mtd.infradead.org/).

You are right. It was my mistake. I've correctly (I believe) patched the
kernel.
But next, I tried to recompile the kernel and got this message:

drivers/mtd/mtdlink.o: In function `put_mtd_device':
drivers/mtd/mtdlink.o(.text+0x4b8): undefined reference to `BUG_ON'
drivers/mtd/mtdlink.o(.text+0x4b8): relocation truncated to fit: R_ARM_PC24
BUG_ON
drivers/mtd/mtdlink.o: In function `nand_scan':
drivers/mtd/mtdlink.o(.text+0x406c): undefined reference to `nand_flash_ids'
drivers/mtd/mtdlink.o(.text+0x4070): undefined reference to `nand_manuf_ids'
make: *** [vmlinux] Error 1
[root@Lab linux]#

Moreover after the patchin.sh was applied, the menuconfig has changed.
I cannot see anymore the s3c2410 nand driver. What happened to it?

Regards

Luca

>
> HTH,
>
> JDB.
> -- 
> LART. 250 MIPS under one Watt. Free hardware design files.
> http://www.lart.tudelft.nl/
>
>

Re: latest MTD snapshot

[Luca Contini]

> At 15:56 +0200 29-08-2003, Luca Contini wrote:
> >  > At 15:28 +0200 29-08-2003, Luca Contini wrote:
> >  > >What Am I missing?
> >>
> >>  1) Which kernel version ?
> >
> >2.4.18-rmk6-swl6
>
> That's a pretty ancient kernel (almost 18 months old). You may want
> to retry with 2.4.21-rmk1 (works fine on the LART, which is also
> arm-based).

I've tried kernel 2.4.21, but the patch I have from samsung didn't work.
So the point is: to make the new MTD subsystem work with my old 2.4.18
kernel. Is it possible and how?

>
> >  > 3) Any particular reason why you don't use the patchin.sh script ? I
> >>  know, you 'only' need mtdblock, but partial installs tend to create
> >>  no end of trouble.
> >
> >I just thought it was easiest to do it manually just for mtdblock device.
> >I've copied the snapshot to ../linux/drivers/mtd/util/snapshot and tried
the
> >patchin.sh:
> >[root@Lab patches]# patchin.sh
> >Directory  does not exist or is not a kernel source directory
>
> You must give it the path to your kernel tree as an argument (as
> documented on http://www.linux-mtd.infradead.org/).
>
> HTH,
>
> JDB.
> -- 
> LART. 250 MIPS under one Watt. Free hardware design files.
> http://www.lart.tudelft.nl/
>
>

Re: latest MTD snapshot

[David Woodhouse]

On Fri, 2003-08-29 at 15:28 +0200, Luca Contini wrote:

> [arm@Botolo2 mtd]$insmod mtdblock.o
> mtdblock.o: unresolved symbol unlikely

Should be in linux/compiler.h, which should be included from
linux/kernel.h. Ancient kernel?

> mtdblock.o: unresolved symbol register_mtd_blktrans
> mtdblock.o: unresolved symbol add_mtd_blktrans_dev
> mtdblock.o: unresolved symbol deregister_mtd_blktrans
> mtdblock.o: unresolved symbol del_mtd_blktrans_dev

insmod mtd_blkdevs.o

-- 
dwmw2

Latest mtd snapshot

[Zoltan Sogor]

Dear All,

we have a problem with the latest mtd snapshot and nandsim. (kernel
version is 2.6.14.3 (x86))

We run the following commands, but it doesn't work.
# modprobe nandsim
# flash_eraseall -j /dev/mtd0
# mount /mnt/flash/
# cp /boot/vmlinuz-2.6.14.3 /mnt/flash/
# umount /mnt/flash/
# mount /mnt/flash/
mount: /dev/mtd0: can't read superblock

kern.log:
kernel: NAND device: Manufacturer ID: 0x98, Chip ID: 0x39 (Toshiba NAND 
8MiB 1,8V 8-bit)
kernel: flash size: 8 MiB
kernel: page size: 512 bytes
kernel: OOB area size: 16 bytes
kernel: sector size: 8 KiB
kernel: pages number: 16384
kernel: pages per sector: 16
kernel: bus width: 8
kernel: bits in sector size: 13
kernel: bits in page size: 9
kernel: bits in OOB size: 4
kernel: flash size with OOB: 8448 KiB
kernel: page address bytes: 3
kernel: sector address bytes: 2
kernel: options: 0x62
kernel: Scanning device for bad blocks
kernel: Creating 1 MTD partitions on "NAND 8MiB 1,8V 8-bit":
kernel: 0x00000000-0x00800000 : "NAND simulator partition"
kernel: JFFS2 version 2.2. (NAND) (C) 2001-2003 Red Hat, Inc.
kernel: Node at 0x00529b64 with length 0x000006ba would run over the end
of the erase block
kernel: Perhaps the file system was created with the wrong erase size?
Reject to mount.

Has somebody get the same effect on a real NAND device on recent snapshot ?

It works on mtdram.

Any idea ???

Zoltan