Multiple banks of flash?

Multiple banks of flash?

[Andrew Carlisle]

I have two banks of flash that are physically located at different
locations.  I have successfully mapped one bank.  I can also map the other
bank, but the problem is that I can't map both the banks simultaneously.

I went into physmap.c and modified init_physmap() so that it basically does
the same thing twice, two probes, two partition searches, but it doesn't
seem to work mapping two banks which are located at different address mapped
in at the same time.

I am using the Monta Vista 2.6 version of the kernel.

 
Thanks,

Andy

Re: Multiple banks of flash?

[Josh Boyer]

On Mon, 2004-11-29 at 09:15, Andrew Carlisle wrote:
> I have two banks of flash that are physically located at different
> locations.  I have successfully mapped one bank.  I can also map the other
> bank, but the problem is that I can't map both the banks simultaneously.
> 
> I went into physmap.c and modified init_physmap() so that it basically does
> the same thing twice, two probes, two partition searches, but it doesn't
> seem to work mapping two banks which are located at different address mapped
> in at the same time.
> 
> I am using the Monta Vista 2.6 version of the kernel.

Joern Engel wrote a mphysmap driver some time ago.  It works fairly
well.  You might be able to find it if you google for it.  Not sure why
it wasn't included in MTD CVS.

Otherwise, you could always write your own mapping driver.  There are
multiple examples that deal with two or more flash banks in CVS.

josh

Re: Multiple banks of flash?

[Jörn Engel]

On Mon, 29 November 2004 09:38:32 -0600, Josh Boyer wrote:
> On Mon, 2004-11-29 at 09:15, Andrew Carlisle wrote:
> > I have two banks of flash that are physically located at different
> > locations.  I have successfully mapped one bank.  I can also map the other
> > bank, but the problem is that I can't map both the banks simultaneously.
> > 
> > I went into physmap.c and modified init_physmap() so that it basically does
> > the same thing twice, two probes, two partition searches, but it doesn't
> > seem to work mapping two banks which are located at different address mapped
> > in at the same time.
> > 
> > I am using the Monta Vista 2.6 version of the kernel.
> 
> Joern Engel wrote a mphysmap driver some time ago.  It works fairly
> well.  You might be able to find it if you google for it.  Not sure why
> it wasn't included in MTD CVS.

Iirc it was included, but in some branch that noone ever noticed or
cared about.  Should have gotten some testing, never did, nothing
happened.  Nothing unusual.

Jörn

-- 
Don't patch bad code, rewrite it.
-- Kernigham and Pike, according to Rusty

Re: Multiple banks of flash?

[Rainer Weikusat]

Josh Boyer <jdub@us.ibm.com> writes:
> On Mon, 2004-11-29 at 09:15, Andrew Carlisle wrote:
>> I have two banks of flash that are physically located at different
>> locations.  I have successfully mapped one bank.  I can also map the other
>> bank, but the problem is that I can't map both the banks simultaneously.
>> 
>> I went into physmap.c and modified init_physmap() so that it basically does
>> the same thing twice, two probes, two partition searches, but it doesn't
>> seem to work mapping two banks which are located at different address mapped
>> in at the same time.
>> 
>> I am using the Monta Vista 2.6 version of the kernel.
>
> Joern Engel wrote a mphysmap driver some time ago.  It works fairly
> well.  You might be able to find it if you google for it.  Not sure why
> it wasn't included in MTD CVS.

Various people have written something like this at different times
:->.

Re: Multiple banks of flash?

[Josh Boyer]

On Mon, 2004-11-29 at 10:29, Rainer Weikusat wrote:
> Josh Boyer <jdub@us.ibm.com> writes:
> > On Mon, 2004-11-29 at 09:15, Andrew Carlisle wrote:
> >> I have two banks of flash that are physically located at different
> >> locations.  I have successfully mapped one bank.  I can also map the other
> >> bank, but the problem is that I can't map both the banks simultaneously.
> >> 
> >> I went into physmap.c and modified init_physmap() so that it basically does
> >> the same thing twice, two probes, two partition searches, but it doesn't
> >> seem to work mapping two banks which are located at different address mapped
> >> in at the same time.
> >> 
> >> I am using the Monta Vista 2.6 version of the kernel.
> >
> > Joern Engel wrote a mphysmap driver some time ago.  It works fairly
> > well.  You might be able to find it if you google for it.  Not sure why
> > it wasn't included in MTD CVS.
> 
> Various people have written something like this at different times
> :->.

Indeed.  Joern's just happened to be the one I am most familiar with. 
But since lots of different implementations have been written, it makes
one wonder why none are in CVS...

josh

Re: Multiple banks of flash?

[Rainer Weikusat]

Josh Boyer <jdub@us.ibm.com> writes:
>> > Joern Engel wrote a mphysmap driver some time ago.  It works fairly
>> > well.  You might be able to find it if you google for it.  Not sure why
>> > it wasn't included in MTD CVS.
>> 
>> Various people have written something like this at different times
>> :->.
>
> Indeed.  Joern's just happened to be the one I am most familiar with. 
> But since lots of different implementations have been written, it makes
> one wonder why none are in CVS...

Open Source Mantra: 'Not my problem'.

Re: Multiple banks of flash?

[Jörn Engel]

On Tue, 30 November 2004 03:11:15 +0100, Rainer Weikusat wrote:
> Josh Boyer <jdub@us.ibm.com> writes:
> >> > Joern Engel wrote a mphysmap driver some time ago.  It works fairly
> >> > well.  You might be able to find it if you google for it.  Not sure why
> >> > it wasn't included in MTD CVS.
> >> 
> >> Various people have written something like this at different times
> >> :->.
> >
> > Indeed.  Joern's just happened to be the one I am most familiar with. 
> > But since lots of different implementations have been written, it makes
> > one wonder why none are in CVS...
> 
> Open Source Mantra: 'Not my problem'.

And embedded seems to be extremely devoted to that mantra.  It was
quite a learning experience (buzzword of the year) to figure that one
out.  Writing code is just 1% of the work.  Debugging it is 9%.
Getting it into some standard (Linux kernel, libc, whatever) is 90%.

And with the 'Not my problem.' mentality, the whole 90% are left to
the original author, who finally gives up and moves on to something
less frustrating.  Or at least gets a job and has someone pay for it.

Jörn

-- 
He that composes himself is wiser than he that composes a book.
-- B. Franklin