[U-Boot] [PATCH 1/6] arm: socfpga: cyclone5-socdk: Enabling mtd partitioning layout

[Marek Vasut]

On Monday, December 28, 2015 at 09:51:44 AM, Chin Liang See wrote:
> On Thu, 2015-12-24 at 02:30 +0100, Marek Vasut wrote:
> > On Thursday, December 24, 2015 at 12:36:45 AM, Chin Liang See wrote:
> > > On Thu, 2015-12-24 at 00:16 +0100, Marek Vasut wrote:
> > > > On Wednesday, December 23, 2015 at 11:57:49 PM, Chin Liang See
> > > > 
> > > > wrote:
> > > > > On Wed, 2015-12-23 at 20:02 +0100, Pavel Machek wrote:
> [..]
> 
> > > For Cyclone V socdk pin mux, they are in good shape where i2c
> > > should be
> > > the last needed. I need to do the same for Arria V socdk as I am
> > > trying
> > > to find the board.
> > 
> > It's in my office, but I am at home now :-/ I _think_ I reloaded the
> > AV
> > SoCDK pinmux, so the basic functionality should at least be there.
> 
> Yah, I just tested master against arria5_socdk and they are doing good
> 
> :) For record, this include testing sdmmc, sf, i2c, usb and ethernet.

Excellent, what about QSPI ?

> > > I am still testing against other. One thing I bump yesterday is the
> > > FAT
> > > support doesn't work in SPL. That would need some tricks that I did
> > > for
> > > SOCEDS U-Boot :)
> > 
> > I think this might be because FAT is too ... well ... fat.
> 
> Hmmm... let me create some slimming pill for that :)

Well, have fun checking this stuff. I think this has to do with BSS usage.

> > It consumes too
> > much stack I think.
> 
> Yah, it took lot of space in stack and especially malloc. As FAT access
> only happen after SDRAM access, we can relocate the stack and malloc to
> SDRAM after SDRAM workng. FYI, I did that for 2013.01.01 and it works
> well :) Will try it out

The stack and malloc area are already placed into DRAM when running
board_init_r() in SPL. I suspect your FAT issues might be elsewhere.

Best regards,
Marek Vasut