Re: help on udoo_neo power up on mainline uboot

[Tommaso Merciai <tomm.merciai@gmail.com>]

On Thu, Dec 30, 2021 at 02:20:07PM +0000, Peter Robinson wrote:
> On Thu, Dec 30, 2021 at 2:12 PM Fabio Estevam <festevam@gmail.com> wrote:
> >
> > Hi Peter,
> >
> > On Thu, Dec 30, 2021 at 11:08 AM Peter Robinson <pbrobinson@gmail.com> wrote:
> >
> > > That works for my Full board. Interestingly in my playing I had got to
> > > similar and had wondered about whether the checkboard function was
> > > needed.
> > >
> > > One minor query on the above is it worth assigning the board_string
> > > return to a local variable rather than called the function twice?
> > > Either way feel free to add my RB/TB or let me know if you want me to
> > > send a patch.
> > > Reviewed-by: Peter Robinson <pbrobinson@gmail.com>
> > > Tested-by: Peter Robinson <pbrobinson@gmail.com>
> >
> > Thanks for testing. I can send a formal patch after your series gets applied.
> 
> Actually, doing a full boot I get
> ERROR: invalid device tree
> Found EFI removable media binary efi/boot/bootarm.efi
> 1929216 bytes read in 92 ms (20 MiB/s)
> libfdt fdt_check_header(): FDT_ERR_BADMAGIC
> 
> And if I interrupt boot and do printenv I get fdtfile=undefined so
> while it fixes the detection and printing of the actual model it seems
> it's too load for the device tree file logic, which now why I remember
> I was looking at the show_board_info function above, the
> board_late_init seems to late for the DT logic.
> 
> P

Hi Peter, Fabio,
What do you think about this solution:

First create new static entry in gd:

+++ b/include/asm-generic/global_data.h
@@ -459,6 +459,7 @@ struct global_data {
         */
        char *smbios_version;
 #endif
+       int board_cfg;
 };
 #ifndef DO_DEPS_ONLY
 static_assert(sizeof(struct global_data) == GD_SIZE);

I know maybe is not a clean solution but we can work on it.
After, I store inside this variable the configuration
calling get_board_value() function from spl:

 static void spl_dram_init(void)
 {
-       int board = get_board_value();
+       gd->board_cfg = get_board_value();

        struct mx6_ddr_sysinfo sysinfo = {
                .dsize = 1, /* width of data bus: 1 = 32 bits */
@@ -471,7 +400,7 @@ static void spl_dram_init(void)
        };

        mx6sx_dram_iocfg(32, &mx6_ddr_ioregs, &mx6_grp_ioregs);
-       if (board == UDOO_NEO_TYPE_BASIC || board == UDOO_NEO_TYPE_BASIC_KS)
+       if (gd->board_cfg == UDOO_NEO_TYPE_BASIC || gd->board_cfg == UDOO_NEO_TYPE_BASIC_KS)
                mx6_dram_cfg(&sysinfo, &neo_basic_mmcd_calib,
                             &neo_basic_mem_ddr);


Then I use gd->board_cfg in board_string function:

 static char *board_string(void)
 {
-       switch (get_board_value()) {
+       switch (gd->board_cfg) {
        case UDOO_NEO_TYPE_BASIC:
                return "BASIC";
        case UDOO_NEO_TYPE_BASIC_KS:

On full boot I get the right environment variable for fdtfile, and the
right strings on Model and Board print:

fdtfile=imx6sx-udoo-neo-basic.dtb

Environment size: 4338/8188 bytes
=>
U-Boot SPL 2022.01-rc4-00030-gb3f84a939f-dirty (Jan 02 2022 - 00:01:42 +0100)
DEBUG FUNC = get_board_value, LINE = 274, r184 = 0
DEBUG FUNC = get_board_value, LINE = 275, r184 = 0
Trying to boot from MMC1


U-Boot 2022.01-rc4-00030-gb3f84a939f-dirty (Jan 02 2022 - 00:01:42 +0100)

CPU:   Freescale i.MX6SX rev1.2 996 MHz (running at 792 MHz)
CPU:   Extended Commercial temperature grade (-20C to 105C) at 44C
Reset cause: POR
Model: UDOO Neo Basic
Board: UDOO Neo BASIC
DRAM:  512 MiB
MMC:   FSL_SDHC: 1, FSL_SDHC: 2
Loading Environment from MMC... OK
In:    serial
Out:   serial
Err:   serial
Net:   eth0: ethernet@2188000 [PRIME]
Hit any key to stop autoboot:  0


What do you think about this solution?
Let me know.

Thanks,
Tommaso