+++ David Bisset [2010-09-24 14:26 +0100]:
> Firstly I've fixed the FPGA build so NAND now works.
wahey!
> I will check in VHDL changes once I'm happy it's all working.
> Once a I get a system to test I'll take a look at PCMCIA as well.
>
> In fact both FPGA and CPLD were completely broken. It appears the VLIO
> changes had crept back into Head.
> I suspect that the only reason this hasn't come to light on CPLD is that TCL
> aren't using HEAD for their CPLD builds.
Hmm. Obviously I'm a little out of the loop but so far as I know tcl
thinks it is using rc1.0 which was pretty-much the same as HEAD last
time I looked. (real change has been kept to the menconfig branch)
Certainly when we tested we used the same code to build CPLD and FPGA
and the CPLD stuff worked but the FPGA didn't (or I am hopelessly
confused).
> With my new code and working with an old 3v32 board with 4 x 2G NAND the
> time taken to get to a boot prompt after pressing reset (and a space bar) is
> 17s. Booting from NOR takes ages and from NAND painfully long.
> It's just spent 2 hours loading root from a USB stick. (frustratingly a
> broken root as it happens.).
> I'm guessing a 10 fold speed loss. As far as I can tell this is not VHDL
> related.
>
> I'm suspecting YAFFS has slowed down. This may be my build being wrong
> somehow, although my experience of incompatible YAFFS version is that it
> just breaks.
> The code I'm using is from the menuconfig system and kernel 2.6.29-1.
>
> Is it possible that in fixing stuff for 8G NAND and large page sizes we've
> slowed it down for 2G small page size?
I've forgotten the details now, but that seems very unlikely.
Is it still fast on a 8G NAND board then?
Wookey
--
Principal hats: Linaro, Emdebian, Wookware, Balloonboard, ARM
http://wookware.org/
(text/plain)