Re: [PATCH] UBIFS: don't fail on -EBADMSG when fixing free space

From: Ben Gardiner
Date: Tue May 24 2011 - 10:33:46 EST


Hi Artem,

On Fri, May 20, 2011 at 2:21 AM, Artem Bityutskiy <dedekind1@xxxxxxxxx> wrote:
> On Thu, 2011-05-19 at 09:28 -0400, Ben Gardiner wrote:
>> [...]
>>
>> Because the "ubi_io_read: error -74 (ECC error)" is precisely what I
>> am encountering on my hardware when I do not flash with a utility that
>> drops empty pages at the end of eraseblocks. I imagined that this was
>> also the case for you. But I have also read that there are
>> peculiarities of the davinci nand driver (both in u-boot and linux).
>>
>> So, at least on my hardware, the -74 error is expected when the 0xff
>> pages are not dropped and so without the 'err != -EBADMSG' exception
>> the free space fixup will cause the volume to fail mount for me:
>
> I am confused. The fix Matthew made is about the following situation:

Sorry for the confusion. In the first 'Programming ubinized images'
thread I saw that Matthew mentioned seeing lots of -74 errors so I was
assuming that his and my problems are one and the same. Now that I
have tested his patch I am seeing that this is not the case.

> 1. You have completely erased flash (MTD partition) - no one has ever
> written there. If you now read the flash, you'll get all 0xFFs with no
> errors.
>
> 2. You use a "dumb" flasher to program an UBI image. This flasher will
> write empty NAND pages "as is". If you now read the flash after the
> "dumb" programming, you should have no errors.
>
> 3. You mount UBIFS for the very first time. It tries to fix up your
> flash. Whatever eraseblock UBIFS reads, it should not encounter any
> error.

Yes I agree with the context for the patch. I can say so far that on
da850evm 1. occurs as you described whereas for 2. and 3. ECC errors
are encountered on read.

> Isn't it weird that a freshly programmed flash cannot be read without
> -EBADMSG (ECC correction failure).

Yes it is very weird. :) I had been assuming up to this point that the
-74 errors were the result of writing the 0xff pages when they should
be dropped. I can see now that this is not the case.

On Fri, May 20, 2011 at 2:29 AM, Artem Bityutskiy <dedekind1@xxxxxxxxx> wrote:
> On Thu, 2011-05-19 at 09:28 -0400, Ben Gardiner wrote:
>> So, at least on my hardware, the -74 error is expected when the 0xff
>> pages are not dropped and so without the 'err != -EBADMSG' exception
>> the free space fixup will cause the volume to fail mount for me:
>
> Still confused. If you have empty and erased flash, then you program it
> with all 0xFFs, you should be able to read it with no errors.

Right. In retrospect this seems to be a self-evident statement :)
I had been previously attributing the -74 errors to a combination of
1) writing 0xff pages when one should not and 2) lack of subpage
writing support workaround with -O 2048. But it is clearly (to me now)
a reading operation failure.

> If this is
> not the case for you, you should fix the driver. UBIFS cannot help in
> this case - we consider it as "broken flash driver" case.

Thank you. Yes, this appears to be the case and I'm glad to hear
confirmation of this from an expert such as yourself.

> I have not read your e-mail carefully because of limited amount of time,
> but with quick reading I became confused - too much information :-)

No problem. Thank you very much for what you have read and for your
insights into the ongoing NAND flash problems on da850evm.

Best Regards,
Ben Gardiner

---
Nanometrics Inc.
http://www.nanometrics.ca
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/