Newly-erased Block...

Newly-erased Block...

[Calò Massimo]

Hi All,

I have an AM29LV160DB flash cip :
	snds100map flash device: 200000 at 1000000
	Search for id:(01 2249) interleave(1) type(2)
	Found: AMD AM29LV160DB
	Physically mapped flash of SNDS100: Found 1 x16 devices at 0x0 in
16-bit mode
	number of JEDEC chips: 1
	Creating 6 MTD partitions on "Physically mapped flash of SNDS100":
	0x00000000-0x00010000 : "boot loader (64K)"
	0x00010000-0x00020000 : "boot data (64K)"
	0x00020000-0x000e0000 : "kernel (768K)"
	0x000e0000-0x00100000 : "user data 00 (128K)"
	0x00100000-0x001a0000 : "root fs (512K)"
	0x001a0000-0x00200000 : "user Jffs2 (512K)"


After the mount command :  />mount -t jffs2 /dev/mtdblock5 /mnt

I receive the following msg:

	Newly-erased block contained wod xxxx at offset xxxx
	Newly-erased block contained wod xxxx at offset xxxx

And the command />df tell me that all the blocks of the partition are used :
0 available.

I tryed to compile "flash_eraseall" under my uClinux-2.4.29, but I did't
have success.

I cleared the flash using my Flash programmer, but the situation is the
same.

Can someone help me on finding the cause of this messages?

Thanks!

	Massimo.

Re: Newly-erased Block...

[Artem B. Bityuckiy]

Calò Massimo wrote:
> I receive the following msg:
> 
> 	Newly-erased block contained wod xxxx at offset xxxx
> 	Newly-erased block contained wod xxxx at offset xxxx
> 

There is an extra check in JFFS2 which makes shure that the newly 
earased flash eraseblock contains only 0xFF bytes. In fact, this is 
overkill and I'd like to move that check into the JFFS2 debugging stuff.

JFFS2 issued the erase command then found that the block contains some 
trash. That's not JFFS2's fault. Dig your flash driver/hardware.

-- 
Best Regards,
Artem B. Bityuckiy,
St.-Petersburg, Russia.

Re: Newly-erased Block...

[Calò Massimo]

> 
> JFFS2 issued the erase command then found that the block 
> contains some 
> trash. That's not JFFS2's fault. Dig your flash driver/hardware.
> 

Artem,

	where is the flash driver? Under
uClinux-dist/linux-2.4.x/drivers/mtd/chips?

Is this the low-level driver : cfi_cmdset_0002?

Under /include/linux/mtd/cfi.h I saw that "cfi_write" calls "map->write16()"
: where is write16 defined ? 

...Or do I have to check blkmem.c : flash_amd16_erase, flash_amd16_write,
flash_amd16_writeall?

I'm sorry for these questions but I don't have the possibility to debug the
kernel.

Thanks for your help!

	Massimo.

Re: Newly-erased Block...

[David Woodhouse]

On Thu, 2005-07-28 at 18:20 +0200, Calò Massimo wrote:
> I receive the following msg:
>         Newly-erased block contained wod xxxx at offset xxxx
>         Newly-erased block contained wod xxxx at offset xxxx

You lie -- it didn't say that. It would never say 'xxxx', and the
numbers it _did_ print will often give some clue as to what went wrong.
Why do you ask for help but not actually give us all the available
information?

-- 
dwmw2

Re: Newly-erased Block...

[Calò Massimo]

> You lie -- it didn't say that. It would never say 'xxxx', and 
> the numbers it _did_ print will often give some clue as to 
> what went wrong. Why do you ask for help but not actually 
> give us all the available information?

David,

	the complete messages after "/>mount -t jffs2 /dev/mtdblock5 /mnt"
are :

/>Newly-erased block contained word 0x4c004c at offest 0x00040000
Newly-erased block contained word 0x4c004c at offest 0x00030000


And this is my flash configuration:

  snds100map flash device: 200000 at 1000000
  Search for id:(01 2249) interleave(1) type(2) 
  Found: AMD AM29LV160DB
  Physically mapped flash of SNDS100: Found 1 x16 devices at 0x0 in 16-bit
mode
  number of JEDEC chips: 1
  Creating 6 MTD partitions on "Physically mapped flash of SNDS100":
  0x00000000-0x00010000 : "boot loader (64K)"
  0x00010000-0x00020000 : "boot data (64K)"
  0x00020000-0x000e0000 : "kernel (768K)"
  0x000e0000-0x00100000 : "user data 00 (128K)"
  0x00100000-0x001a0000 : "root fs (512K)"
  0x001a0000-0x00200000 : "user Jffs2 (512K)"

Thanks,

	Massimo.

Re: Newly-erased Block...

[Artem B. Bityuckiy]

On Mon, 2005-08-01 at 10:38 +0200, Calò Massimo wrote:
> Artem,
> 
> 	where is the flash driver? Under
> uClinux-dist/linux-2.4.x/drivers/mtd/chips?
> 
> Is this the low-level driver : cfi_cmdset_0002?
> 
> Under /include/linux/mtd/cfi.h I saw that "cfi_write" calls "map->write16()"
> : where is write16 defined ? 
> 
> ...Or do I have to check blkmem.c : flash_amd16_erase, flash_amd16_write,
> flash_amd16_writeall?

All that you can find in <linux>/drivers/mtd/ may be regarded as driver-
related.

> I'm sorry for these questions but I don't have the possibility to debug the
> kernel.
Well, you're in trouble. I cannot help here, sorry :-)

-- 
Best Regards,
Artem B. Bityuckiy,
St.-Petersburg, Russia.

Newly-erased Block...

[Calò Massimo]

Hi Artem,

	do you remember about my flash problem?

>>/>mount -t jffs2 /dev/mtdblock5 /mnt
>>/>Newly-erased block contained word 0x4c004c at offest 0x00040000
>>Newly-erased block contained word 0x4c004c at offest 0x00030000

I have found the cause : the Samsung S3C4510b when access to the memory use
the "cache" by default.
If you have to disable it (like while you program the flash) you have to 'or
0x4000000 to the base address.

So I have modifyed the file "cfi_cmdset_0002.c" as follow :

	for esample: "do_erase_chip" function

		...
		chip->start |= 0x4000000; //my modify!!!!!!!!!!!!!

		chip->state = FL_ERASING;

		ENABLE_VPP(map);
	
cfi_send_gen_cmd(0xAA,cfi->addr_unlock1,chip->start,map,cfi,......


I did the same for every "cfi_send_gen_cmd" : the 3rd parameter has been
ored to 0x4000000.

What do you think about this modify? Do you know a better way?

Now I still have one warning msg : 
	"Warning: DQ5 raised while program operation was in progress,
however operation completed OK"
Can I ignore this msg?

Thanks!

	Massimo.

Re: Newly-erased Block...

[Artem B. Bityutskiy]

Calò Massimo wrote:
> What do you think about this modify? Do you know a better way?
I'm sorry Carlo, I have never used CFI and have no comments.
But there are compitent people present here, the may comment this.

-- 
Best Regards,
Artem B. Bityuckiy,
St.-Petersburg, Russia.