buffer write error (status 0x90)

buffer write error (status 0x90)

[Jon Ringle]

Hello,

We are replacing the discontinued Intel StrataFlash P30 flash
JS28F256P30T95 with the Numonyx P30-65nm flash JS28F256P30TF on our
boards and have a few boards with sample chips. During some testing
one of the boards with the Numonyx flash chip is now reporting the
following buffer write error (status 0x90) on every boot:

[    4.720000] IXP4XX-Flash.0: buffer write error (status 0x90)
[    4.720000] UBI error: nor_erase_prepare: cannot invalidate PEB 62,
write returned -22 read returned 2
[    5.160000] UBI error: erase_worker: failed to erase PEB 62, error -5
[    5.180000] UBI error: erase_worker: bad physical eraseblock 62 detected
[    5.180000] UBI warning: ubi_ro_mode: switch to read-only mode
[    5.200000] UBI error: do_work: work failed with error code -5
[    5.200000] UBI error: ubi_thread: ubi_bgt0d: work failed with error code -5

The chipstatus that is report (0x90) comes from:
		unsigned long chipstatus = MERGESTATUS(status);

But I can't make sense of what 0x90 means.

Thanks,
Jon

Re: buffer write error (status 0x90)

[Adam Yergovich]

Jon Ringle wrote:
> We are replacing the discontinued Intel StrataFlash P30 flash
> JS28F256P30T95 with the Numonyx P30-65nm flash JS28F256P30TF on our
> boards and have a few boards with sample chips. During some testing
> one of the boards with the Numonyx flash chip is now reporting the
> following buffer write error (status 0x90) on every boot:

I have tested swapping the lower density part and haven't had any 
problems with MTD or otherwise.  I suspect you've just got a bad part or 
something else happened.

-Adam

-- 
Adam Yergovich
Engineer
JK Microsystems

1403 Fifth St. Suite D
Davis, CA 95616

Tel:(530) 297-6073

Re: buffer write error (status 0x90)

[Jon Ringle]

On Mon, Aug 9, 2010 at 5:21 PM, Adam Yergovich <ayergo@jkmicro.com> wrote:
> Jon Ringle wrote:
>>
>> We are replacing the discontinued Intel StrataFlash P30 flash
>> JS28F256P30T95 with the Numonyx P30-65nm flash JS28F256P30TF on our
>> boards and have a few boards with sample chips. During some testing
>> one of the boards with the Numonyx flash chip is now reporting the
>> following buffer write error (status 0x90) on every boot:
>
> I have tested swapping the lower density part and haven't had any problems
> with MTD or otherwise.  I suspect you've just got a bad part or something
> else happened.

Maybe, but we have seen other incompatibilities that occurred on all
of our samples. We use Redboot as the bootloader and we found that a
Redboot image in ROM mode will hang when trying to unlock and erase a
sector. I resolved that problem by building a ROMRAM Redboot image.

Re: buffer write error (status 0x90)

[Shane Volpe]

Jon,
I'm also replacing the obsolete P30 part and ran into an interesting
issue.  It turns out the new replacement part has a larger buffer size
so if you mix and match the obsolete part with the newer part you will
have issues because if the new part is in the lower nibble 0:15 the
CFI will read out its buffer size and since that buffer size is larger
than the obsolete P30 sitting in the higher nibble 16:31 then the its
buffer will get over run and not work.  So what I'm saying is that
even though the replacement part is almost the same as the old part do
not mix and match as the buffer sizes will cause issues.  Although you
can patch the kernel and hard code the buffer size to that of the
legacy P30 or even better patch it to read the buffer sizes out of
both ICs and use the smaller size.

Regards,
Shane

On Mon, Aug 9, 2010 at 11:46 PM, Jon Ringle <jon@ringle.org> wrote:
> On Mon, Aug 9, 2010 at 5:21 PM, Adam Yergovich <ayergo@jkmicro.com> wrote:
>> Jon Ringle wrote:
>>>
>>> We are replacing the discontinued Intel StrataFlash P30 flash
>>> JS28F256P30T95 with the Numonyx P30-65nm flash JS28F256P30TF on our
>>> boards and have a few boards with sample chips. During some testing
>>> one of the boards with the Numonyx flash chip is now reporting the
>>> following buffer write error (status 0x90) on every boot:
>>
>> I have tested swapping the lower density part and haven't had any problems
>> with MTD or otherwise.  I suspect you've just got a bad part or something
>> else happened.
>
> Maybe, but we have seen other incompatibilities that occurred on all
> of our samples. We use Redboot as the bootloader and we found that a
> Redboot image in ROM mode will hang when trying to unlock and erase a
> sector. I resolved that problem by building a ROMRAM Redboot image.
>
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/
>

Re: buffer write error (status 0x90)

[Joakim Tjernlund]

>
> Jon,
> I'm also replacing the obsolete P30 part and ran into an interesting
> issue.  It turns out the new replacement part has a larger buffer size
> so if you mix and match the obsolete part with the newer part you will
> have issues because if the new part is in the lower nibble 0:15 the
> CFI will read out its buffer size and since that buffer size is larger
> than the obsolete P30 sitting in the higher nibble 16:31 then the its
> buffer will get over run and not work.  So what I'm saying is that
> even though the replacement part is almost the same as the old part do
> not mix and match as the buffer sizes will cause issues.  Although you
> can patch the kernel and hard code the buffer size to that of the
> legacy P30 or even better patch it to read the buffer sizes out of
> both ICs and use the smaller size.
>
> Regards,
> Shane
>
> On Mon, Aug 9, 2010 at 11:46 PM, Jon Ringle <jon@ringle.org> wrote:
> > On Mon, Aug 9, 2010 at 5:21 PM, Adam Yergovich <ayergo@jkmicro.com> wrote:
> >> Jon Ringle wrote:
> >>>
> >>> We are replacing the discontinued Intel StrataFlash P30 flash
> >>> JS28F256P30T95 with the Numonyx P30-65nm flash JS28F256P30TF on our
> >>> boards and have a few boards with sample chips. During some testing
> >>> one of the boards with the Numonyx flash chip is now reporting the
> >>> following buffer write error (status 0x90) on every boot:

Check your samples. We did this too some time ago and the
samples we got were old and had errors not present in production parts.

> >>
> >> I have tested swapping the lower density part and haven't had any problems
> >> with MTD or otherwise.  I suspect you've just got a bad part or something
> >> else happened.
> >
> > Maybe, but we have seen other incompatibilities that occurred on all
> > of our samples. We use Redboot as the bootloader and we found that a
> > Redboot image in ROM mode will hang when trying to unlock and erase a
> > sector. I resolved that problem by building a ROMRAM Redboot image.

Most likely due to a know bug in the chips. See
"Numonyx Axcell P33/P30 256-Mbit Specification Update"

It states :
When customer uses [...] block unlock, the block lock status might be
altered inadvertently. Lock status might be set to either 01h or 03h
unexpectedly (00h as expected data), which leads to program/erase failure
on certain blocks.

Easy to work around.

 Jocke

Re: buffer write error (status 0x90)

[Shane Volpe]

On Tue, Aug 24, 2010 at 7:18 PM, Joakim Tjernlund
<joakim.tjernlund@transmode.se> wrote:
> It states :
> When customer uses [...] block unlock, the block lock status might be
> altered inadvertently. Lock status might be set to either 01h or 03h
> unexpectedly (00h as expected data), which leads to program/erase failure
> on certain blocks.
>
> Easy to work around.

Joakim,
Do you have a patch or an example on how to work around this?
Regards,
Shane

Re: buffer write error (status 0x90)

[Joakim Tjernlund]

Shane Volpe <shanevolpe@gmail.com> wrote on 2010/08/25 13:58:06:
>
> On Tue, Aug 24, 2010 at 7:18 PM, Joakim Tjernlund
> <joakim.tjernlund@transmode.se> wrote:
> > It states :
> > When customer uses [...] block unlock, the block lock status might be
> > altered inadvertently. Lock status might be set to either 01h or 03h
> > unexpectedly (00h as expected data), which leads to program/erase failure
> > on certain blocks.
> >
> > Easy to work around.
>
> Joakim,
> Do you have a patch or an example on how to work around this?
> Regards,
> Shane

See u-boot:
http://lists.denx.de/pipermail/u-boot/2010-August/075727.html

Re: buffer write error (status 0x90)

[Shane Volpe]

>
> See u-boot:
> http://lists.denx.de/pipermail/u-boot/2010-August/075727.html
>
>

Joakim,
Great, thanks.  Do you know if this issue has been addressed in the
Linux kernel?
Regards,
Shane

Re: buffer write error (status 0x90)

[Joakim Tjernlund]

Shane Volpe <shanevolpe@gmail.com> wrote on 2010/08/25 16:00:07:
>
> >
> > See u-boot:
> > http://lists.denx.de/pipermail/u-boot/2010-August/075727.html
> >
> >
>
> Joakim,
> Great, thanks.  Do you know if this issue has been addressed in the
> Linux kernel?

It hasn't but it hasn't occurred there either(yet)