2012/4/20 Ilya Averyanov <averyanovin+ub...@gmail.com>
> > > 2012/4/20 Marek Vasut <ma...@denx.de> > >> Dear Ilya Averyanov, >> >> > 2012/4/20 Marek Vasut <ma...@denx.de> >> > >> > > Dear Ilya Averyanov, >> > > >> > > [...] >> > > >> > > > > > > > > > .word 0x32 >> > > > > > > > > > + .word 0x30 >> > > > > > > > > > + .word 0x30 >> > > > > > > > > >> > > > > > > > > What this undocumented stuff? >> > > > > > > > >> > > > > > > > SoC specific, I pasted it from smdk2410, it's for initialize >> > > > > > > > the >> > > > > >> > > > > DRAM. >> > > > > >> > > > > > > Can't you init DRAM in some more fitting place? Also document >> > > > > > > this? >> > > > > > >> > > > > > I don't know, this part is a bit difficult for me, but I'm >> going to >> > > >> > > try >> > > >> > > > > > to document it with the help of the SoC documentation. >> > > > > > >> > > > > > For the place, should I move it to mini2440.c in >> arch_cpu_init() or >> > > > > > board_early_init_f() ? I don't really know the difference. >> > > > > >> > > > > dram_init() maybe ? >> > > > >> > > > You propose to initialize the SDRAM controller in dram_init ()? >> > > >> > > Well yes, ain't it more fitting ? >> > > >> > > Best regards, >> > > Marek Vasut >> > >> > No. We have to two reasons why this is not the case. >> > 1) Open u-boot/arch/arm/cpu/arm920t/start.S:328 >> > cpu_init_crit >> > ...... >> > line 346 >> > /* >> > * before relocating, we have to setup RAM timing >> > * because memory timing is board-dependend, you will >> > * find a lowlevel_init.S in your board directory. >> > */ >> >> Well don't setup run in lowlevel_init.S then ? >> >> > 2) NAND controller must be initialized before C code run. >> >> Why? >> >> Best regards, >> Marek Vasut >> > mmm... > Because we need a stack. > The stack is in RAM. > >2) NAND controller must be initialized before C code run. Sorry made a mistake. SDRAM! Not NAND!
_______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot