jedec_probe_chip error?

jedec_probe_chip error?

[Jeremy Jackson]

in Linux-MTD:

I'm looking at mtd/drivers/chips/jedec_probe.c:jedec_probe_chip
line 794 from current CVS.

for CFI_DEVICETYPE_X8, the unlock1 and unlock2 values
seem wrong.   They're currently 0x555 and 0x2aa, resp.
Does it work like this?  All of the small (512KiB -
128KiB) devices I've read the datasheets for, require
0x5555 and 0x2aaa here.  Are there any devices which
use the current code's values?

Also, I have discovered (see prev. post) at least one device that doesn't
respond to the aa 55 90 sequence, requiring an alternate
"old" sequence instead.  Any guidance on the best way to
implement this generically would be appreciated.

Jeremy

Re: jedec_probe_chip error?

[Eric W. Biederman]

"Jeremy Jackson" <jerj@coplanar.net> writes:

> in Linux-MTD:
> 
> I'm looking at mtd/drivers/chips/jedec_probe.c:jedec_probe_chip
> line 794 from current CVS.
> 
> for CFI_DEVICETYPE_X8, the unlock1 and unlock2 values
> seem wrong.   They're currently 0x555 and 0x2aa, resp.
> Does it work like this?  

Unless it's changed since I last looked the code auto detects
whether it should use 0x555/0x2aa or 0x5555/0x2aaa.  If one variant
doesn't work it then tries the other.

> All of the small (512KiB -
> 128KiB) devices I've read the datasheets for, require
> 0x5555 and 0x2aaa here.  Are there any devices which
> use the current code's values?

Yes. I believe it is SST that is the odd man out in this
regard. 
 
> Also, I have discovered (see prev. post) at least one device that doesn't
> respond to the aa 55 90 sequence, requiring an alternate
> "old" sequence instead.  Any guidance on the best way to
> implement this generically would be appreciated.

My hunch is to try the very old sequence last, if the newer
sequences do not work.  In the worst case you will have to write
your own probe routine.  But I suspect Jedec probe can stretch a
little more. 

Eric
 

Re: jedec_probe_chip error?

[Stefan Reinauer]

* Jeremy Jackson <jerj@coplanar.net> [020404 04:21]:
> Also, I have discovered (see prev. post) at least one device that doesn't
> respond to the aa 55 90 sequence, requiring an alternate
> "old" sequence instead.  Any guidance on the best way to
> implement this generically would be appreciated.
 
There's 3 different methods used by flash devices to identify
themselfes.
aa 55 90 is the most used one. Others (i.e. some Winbonds) need
aa 55 80 aa 55 60, whereas the third method (some other winbonds)
is aa 55 80 aa 55 20.
As far as I could estimate, the order in which you probe is crucial.
Some chips answer to aa 55 80 aa 55 60 or  aa 55 80 aa 55 20 with
their protection map instead of their id. 
Eon Flashchips don't put their device id to 0x01, but to 0x101. 
0x00 and 0x01 contain the vendor id (0x7f)

/dev/bios does that part of the probing pretty ok. I suggest you have a
look at programming.c:flash_probe_address()

get /dev/bios from the OpenBIOS cvs tree:
http://www.freiburg.linux.de/OpenBIOS/dev/cvs.html

Best regards,
  Stefan Reinauer
  
-- 
Ok hex 4666 dup negate do i 4000 dup 2* negate do " *" 0 dup 2dup 1e 0 do
 2swap * e >>a 2* 5 pick + -rot - j + dup dup * e >>a rot dup dup * e >>a 
  rot swap 2dup + 10000 > if 3drop 3drop "  " 0 dup 2dup leave then loop 
              2drop 2drop type 268 +loop cr drop 5de +loop

Re: jedec_probe_chip error?

[Richard A. Smith]

On Thu, 4 Apr 2002 12:36:04 +0200, Stefan Reinauer wrote:

>
>get /dev/bios from the OpenBIOS cvs tree:
>http://www.freiburg.linux.de/OpenBIOS/dev/cvs.html
>

The Readme lists that this should work under a 2.2.x kernel, however
I get serveral errors trying to compile with 2.2.19 

Things like:

structure has no member nameed 'resource'
'PCI_ROM_RESOURCE' undeclared

Does it sill work with 2.2.x?

I'm using the latest tarball from the above link.  I was unable to
get a CVS copy as it kept complaining that 'anonymous' was an unknown
user.

--
Richard A. Smith                         Bitworks, Inc.               
rsmith@bitworks.com               501.846.5777 x204                        
Sr. Design Engineer        http://www.bitworks.com   

Re: jedec_probe_chip error?

[Stefan Reinauer]

* Richard A. Smith <rsmith@bitworks.com> [020405 19:09]:
> >get /dev/bios from the OpenBIOS cvs tree:
> >http://www.freiburg.linux.de/OpenBIOS/dev/cvs.html
> 
> The Readme lists that this should work under a 2.2.x kernel, however
> I get serveral errors trying to compile with 2.2.19 
[..] 
> Does it sill work with 2.2.x?

Oops. I dropped 2.2 support with one of the last releases. It is tested
under 2.4 - backporting it shouldnt be too hard - drop me a note if 2.4 
is not an option for you.
 
> I'm using the latest tarball from the above link.  I was unable to
> get a CVS copy as it kept complaining that 'anonymous' was an unknown
> user.

Hmm... I'll check into that asap.

Best regards,
  Stefan Reinauer
  
-- 
Ok hex 4666 dup negate do i 4000 dup 2* negate do " *" 0 dup 2dup 1e 0 do
 2swap * e >>a 2* 5 pick + -rot - j + dup dup * e >>a rot dup dup * e >>a 
  rot swap 2dup + 10000 > if 3drop 3drop "  " 0 dup 2dup leave then loop 
              2drop 2drop type 268 +loop cr drop 5de +loop