Problem with Hynix HY27UF081G2A 1Gb NAND flash?

Problem with Hynix HY27UF081G2A 1Gb NAND flash?

[Ricard Wanderlof]

I'm testing the above mentioned NAND flash chip in a design which has 
previously used the corresponding 1Gb parts from Samsung and Numonyx (ST) 
without problems. However, with the Hynix chip, mtd won't even read the ID 
code, it's just as if the chip wasn't there (it fails with

nand_get_flash_type: second ID read did not match 01,ec against ec,7a .

Before deciding on a course of action, I was wondering if anyone else has 
had a similar problem. I've skimmed through the data sheet and cannot find 
anything that is obviously different (save for the manufacturer's ID code 
which is expected of course).

Kernel version is 2.6.26.

/Ricard
--
Ricard Wolf Wanderlöf                           ricardw(at)axis.com
Axis Communications AB, Lund, Sweden            www.axis.com
Phone +46 46 272 2016                           Fax +46 46 13 61 30
    "With Free Software you are employing the best programmers on the planet"

Re: Problem with Hynix HY27UF081G2A 1Gb NAND flash?

[Jason Liu]

please send reset command(0xFF) to nand flash first then read id.

2008/9/19 Ricard Wanderlof <ricard.wanderlof@axis.com>:
>
> I'm testing the above mentioned NAND flash chip in a design which has
> previously used the corresponding 1Gb parts from Samsung and Numonyx (ST)
> without problems. However, with the Hynix chip, mtd won't even read the ID
> code, it's just as if the chip wasn't there (it fails with
>
> nand_get_flash_type: second ID read did not match 01,ec against ec,7a .
>
> Before deciding on a course of action, I was wondering if anyone else has
> had a similar problem. I've skimmed through the data sheet and cannot find
> anything that is obviously different (save for the manufacturer's ID code
> which is expected of course).
>
> Kernel version is 2.6.26.
>
> /Ricard
> --
> Ricard Wolf Wanderlöf                           ricardw(at)axis.com
> Axis Communications AB, Lund, Sweden            www.axis.com
> Phone +46 46 272 2016                           Fax +46 46 13 61 30
>    "With Free Software you are employing the best programmers on the planet"
>
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/
>

Re: Problem with Hynix HY27UF081G2A 1Gb NAND flash?

[Enrico Scholz]

Ricard Wanderlof <ricard.wanderlof@axis.com> writes:

> I'm testing the above mentioned NAND flash chip in a design
> which has previously used the corresponding 1Gb parts from
> Samsung and Numonyx (ST) without problems. However, with the
> Hynix chip, mtd won't even read the ID code, it's just as if
> the chip wasn't there (it fails with

Some Hynix NAND chips seem to have problems with bus arbitration;
e.g. they continue to shift out data when chipselect is
deasserted during a operation (e.g. because ethernet controller
was accessed in this time).  We saw this with HY27UG088G5M on
PXA320 and read-page operations.


Enrico

Re: Problem with Hynix HY27UF081G2A 1Gb NAND flash?

[Ricard Wanderlof]

On Fri, 19 Sep 2008, Enrico Scholz wrote:

> Ricard Wanderlof <ricard.wanderlof@axis.com> writes:
>
>> I'm testing the above mentioned NAND flash chip in a design
>> which has previously used the corresponding 1Gb parts from
>> Samsung and Numonyx (ST) without problems. However, with the
>> Hynix chip, mtd won't even read the ID code, it's just as if
>> the chip wasn't there (it fails with
>
> Some Hynix NAND chips seem to have problems with bus arbitration;
> e.g. they continue to shift out data when chipselect is
> deasserted during a operation (e.g. because ethernet controller
> was accessed in this time).  We saw this with HY27UG088G5M on
> PXA320 and read-page operations.

Interesting. We don't have any other devices on the bus that the NAND chip 
is connected to however.

/Ricard
--
Ricard Wolf Wanderlöf                           ricardw(at)axis.com
Axis Communications AB, Lund, Sweden            www.axis.com
Phone +46 46 272 2016                           Fax +46 46 13 61 30
    "With Free Software you are employing the best programmers on the planet"

Re: Problem with Hynix HY27UF081G2A 1Gb NAND flash?

[Ricard Wanderlof]

On Mon, 22 Sep 2008, Ricard Wanderlof wrote:

>
> On Fri, 19 Sep 2008, Enrico Scholz wrote:
>
>> Some Hynix NAND chips seem to have problems with bus arbitration;
>> e.g. they continue to shift out data when chipselect is
>> deasserted during a operation (e.g. because ethernet controller
>> was accessed in this time).  We saw this with HY27UG088G5M on
>> PXA320 and read-page operations.
>
> Interesting. We don't have any other devices on the bus that the NAND 
> chip is connected to however.

Sorry, I was wrong, we do share the bus with the RAM (mixed up two designs 
here). Did you only see the problem with page read operations, or was it a 
general problem. We don't even seem to get by the ID read stage.

/Ricard
--
Ricard Wolf Wanderlöf                           ricardw(at)axis.com
Axis Communications AB, Lund, Sweden            www.axis.com
Phone +46 46 272 2016                           Fax +46 46 13 61 30
    "With Free Software you are employing the best programmers on the planet"

Re: Problem with Hynix HY27UF081G2A 1Gb NAND flash?

[Enrico Scholz]

Ricard Wanderlof <ricard.wanderlof@axis.com> writes:

>>> Some Hynix NAND chips seem to have problems with bus arbitration;
>>> e.g. they continue to shift out data when chipselect is
>>> deasserted during a operation (e.g. because ethernet controller
>>> was accessed in this time).  We saw this with HY27UG088G5M on
>>> PXA320 and read-page operations.
>> ...
> Did you only see the problem with page read operations, or was
> it a general problem. We don't even seem to get by the ID read
> stage.

We saw this only during page-read operations.  But access is
handled completely by the PXA320 which reads data from NAND in
bursts. Hence, nCS won't change during small transfer so that the
brokeness is not triggered at read-id operations.


Enrico

Re: Problem with Hynix HY27UF081G2A 1Gb NAND flash?

[Ricard Wanderlof]

On Fri, 19 Sep 2008, Karl Beldan wrote:

>> ...
>> nand_get_flash_type: second ID read did not match 01,ec against ec,7a .
>> ...

> I had the same problem with Microns's.
> Their datasheet mentionned the need of resetting the chip after
> power-up though.
> My patch : http://patchwork.ozlabs.org/patch/279/
> cf http://lists.infradead.org/pipermail/linux-mtd/2008-September/023009.html.
>
> Hope it'll help.

Thanks for the patch. Applied it, but it didn't seem to make any 
difference (except that the read bytes are 01,bf and 0b,0b respectively
instead of the values given above). Still, it's a lead to work on.

/Ricard
--
Ricard Wolf Wanderlöf                           ricardw(at)axis.com
Axis Communications AB, Lund, Sweden            www.axis.com
Phone +46 46 272 2016                           Fax +46 46 13 61 30
    "With Free Software you are employing the best programmers on the planet"