doc2k: "formatting block"?

doc2k: "formatting block"?

[Ian Eure <Ian Eure]

while testing a 32mb doc2000, i see these messages:

-- snip --
DiskOnChip 2000 found at address 0xD0000
Flash chip found: Manufacturer ID: 98, Chip ID: 73 (Toshiba TH58V128DC)
2 flash chips found. Total DiskOnChip size: 32 Mb
Block 807: referencing block 110 already in another chain
Formatting chain at block 807
Formatting block 807
Unreferenced block 366, formatting it
-- snip --

...and later on:

-- snip --
EXT2-fs error (device nftl(93,1)): ext2_free_blocks: Freeing blocks in system zones - Block = 121, count = 1
EXT2-fs error (device nftl(93,1)): ext2_free_blocks: Freeing blocks in system zones - Block = 185, count = 1
EXT2-fs error (device nftl(93,1)): ext2_free_blocks: Freeing blocks in system zones - Block = 217, count = 1
EXT2-fs error (device nftl(93,1)): ext2_free_blocks: Freeing blocks in system zones - Block = 233, count = 1
EXT2-fs error (device nftl(93,1)): ext2_free_blocks: Freeing blocks in system zones - Block = 248, count = 1
EXT2-fs error (device nftl(93,1)): ext2_free_blocks: Freeing blocks in system zones - Block = 243, count = 1
end_request: I/O error, dev 5d:01 (unknown), sector 8274
EXT2-fs error (device nftl(93,1)): ext2_free_branches: Read failure, inode=12, block=4137
-- snip --

is this a software problem, or a bad doc?

i suppose it could be my eval board, they don't seem to be made very
well.

-- 
 ___________________________________
| Ian Eure - <ieure@insynq.com>     | "You're living in a facist world...
|         -  Developer -            | Freedom is a luxury." -Front Line
|        -  InsynQ, Inc. -          | Assembly, "Digital Tension Dementia"
| Your Internet Utility Company.tm  |________________________________________


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

Re: doc2k: "formatting block"?

[David Woodhouse]

On Tue, 6 Feb 2001, Ian Eure <Ian Eure wrote:

> while testing a 32mb doc2000, i see these messages:
> 
> -- snip --
> DiskOnChip 2000 found at address 0xD0000
> Flash chip found: Manufacturer ID: 98, Chip ID: 73 (Toshiba TH58V128DC)
> 2 flash chips found. Total DiskOnChip size: 32 Mb
> Block 807: referencing block 110 already in another chain
> Formatting chain at block 807
> Formatting block 807
> Unreferenced block 366, formatting it
> -- snip --

Eep. Can you reproduce it? Did this happen after the M-Systems drivers (or 
firmware) had been writing to the device, or is it purely our software? 
It's unlikely to be hardware.

I'm trying to grok that code - even if the block _is_ linked in another 
chain, we should be able to work out which chain it _ought_ to be in by 
its own VirtUnitNum entry. We still have to format the chain in which the 
block _isn't_ supposed to be, and there's definitely some data corruption 
happened, but we should be able to do a little better than we do ATM, I 
think.

I don't see how this could be caused by a fold in progress - we don't fold 
one chain into _another_, only into itself. But the comment implies 
otherwise...

	printk("Block %d: referencing block %d already in another chain\n", 
		block, rep_block);
	/* XXX: should handle correctly fold in progress chains */
	do_format_chain = 1;
	s->ReplUnitTable[block] = BLOCK_NIL;


-- 
dwmw2





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

Re: doc2k: "formatting block"?

[Ian Eure <Ian Eure]

David Woodhouse writes:
 > On Tue, 6 Feb 2001, Ian Eure <Ian Eure wrote:
 > 
 > > while testing a 32mb doc2000, i see these messages:
 > > 
 > > -- snip --
 > > DiskOnChip 2000 found at address 0xD0000
 > > Flash chip found: Manufacturer ID: 98, Chip ID: 73 (Toshiba TH58V128DC)
 > > 2 flash chips found. Total DiskOnChip size: 32 Mb
 > > Block 807: referencing block 110 already in another chain
 > > Formatting chain at block 807
 > > Formatting block 807
 > > Unreferenced block 366, formatting it
 > > -- snip --
 > 
 > Eep. Can you reproduce it? Did this happen after the M-Systems drivers (or 
 > firmware) had been writing to the device, or is it purely our software? 
 > It's unlikely to be hardware.
 > 
it's reproducible, but only with this one 32mb doc2000. i only have
this chip, and a 48mb doc2000 for testing right now.

the doc had been written to by the m-sys driver before.

 > I'm trying to grok that code - even if the block _is_ linked in another 
 > chain, we should be able to work out which chain it _ought_ to be in by 
 > its own VirtUnitNum entry. We still have to format the chain in which the 
 > block _isn't_ supposed to be, and there's definitely some data corruption 
 > happened, but we should be able to do a little better than we do ATM, I 
 > think.
 > 
 > I don't see how this could be caused by a fold in progress - we don't fold 
 > one chain into _another_, only into itself. But the comment implies 
 > otherwise...
 > 
 > 	printk("Block %d: referencing block %d already in another chain\n", 
 > 		block, rep_block);
 > 	/* XXX: should handle correctly fold in progress chains */
 > 	do_format_chain = 1;
 > 	s->ReplUnitTable[block] = BLOCK_NIL;
 > 
i'm an intermediate c programmer, and have no experience with kernel
programming. but, i'd like to try and help you, so i'm just gonna jump
in. some patches may (or may not) fly out in a few days or weeks. :)

-- 
 ___________________________________
| Ian Eure - <ieure@insynq.com>     | "You're living in a facist world...
|         -  Developer -            | Freedom is a luxury." -Front Line
|        -  InsynQ, Inc. -          | Assembly, "Digital Tension Dementia"
| Your Internet Utility Company.tm  |________________________________________


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

Re: doc2k: "formatting block"?

[David Woodhouse]

Ian Eure" <ieure@insynq.com said:
> it's reproducible, but only with this one 32mb doc2000. i only have
> this chip, and a 48mb doc2000 for testing right now.

Same block every time, or different blocks?

> the doc had been written to by the m-sys driver before. 

Immediately before the boot on which the trouble was seen, or at some
distant point in the past?

--
dwmw2




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

Re: doc2k: "formatting block"?

[Ian Eure <Ian Eure]

David Woodhouse writes:
 > 
 > Ian Eure" <ieure@insynq.com said:
 > > it's reproducible, but only with this one 32mb doc2000. i only have
 > > this chip, and a 48mb doc2000 for testing right now.
 > 
 > Same block every time, or different blocks?
 > 
different.

 > > the doc had been written to by the m-sys driver before. 
 > 
 > Immediately before the boot on which the trouble was seen, or at some
 > distant point in the past?
 > 
in the distant past.

-- 
 ___________________________________
| Ian Eure - <ieure@insynq.com>     | "You're living in a facist world...
|         -  Developer -            | Freedom is a luxury." -Front Line
|        -  InsynQ, Inc. -          | Assembly, "Digital Tension Dementia"
| Your Internet Utility Company.tm  |________________________________________


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