howto

howto

[Livio Plos]

Hi,

I'm trying to build a system with flash chips instead of an
hard disk or a disk on module and this seems to be the 
most advanced project to help me doing this.

Neverthless, it is not easy to begin understanding how to 
move between sources, configuration, modules...

Can I boot totally from flash without any other storage support?

Is there any howto or high level docs to configure and compile?


Thank you very much.

Livio





To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org

Re: howto

[David Woodhouse]

livioplos@bigfoot.com said:
>  I'm trying to build a system with flash chips instead of an hard disk
> or a disk on module and this seems to be the  most advanced project to
> help me doing this.

> Neverthless, it is not easy to begin understanding how to  move
> between sources, configuration, modules...

Documentation is moving up the TODO list very slowly, as I get the code 
closer to a state where I'm happy with it.

> Can I boot totally from flash without any other storage support?

Yes, as long as you have the flash mapped where the CPU startup vector 
points, or the board's firmware has support for starting the Linux kernel
from flash (RedBoot does).

> Is there any howto or high level docs to configure and compile? 

Drop the latest code from CVS on top of your 2.4 kernel tree. Enable all the
CFI stuff, enable CONFIG_MTD_PHYSMAP and set the address, size and bus width
for your flash mapping appropriately. Boot it and see what happens.

If you want to partition the flash, as most people do, you'll want to copy
one of the other 'map' drivers (eg. pnc2000.c) and change the numbers
accordingly.

--
dwmw2




To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org

mtdchar.c and vmalloc

[Alice Hennessy]

Hello,

During  stress testing,  I ran into a data corruption problem that  went away
after I changed vmalloc to
kmalloc  in mtd_read and mtd_write functions (mtdchar.c).   I added a loop to
the code so that malloc size
doesn't exceed the kmalloc maximum of 128k bytes.
Seems to work fine now.

Does this make sense? Anyone else see this problem?   Any opinions if this is
an ok solution until
kiovec or something else is implemented?

Alice



To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org

Re: mtdchar.c and vmalloc

[David Woodhouse]

ahennessy@mvista.com said:
>   Any opinions if this is an ok solution until kiovec or something
> else is implemented? 

S'fine.

--
dwmw2




To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org