On Mon, 31 Jan 2022 15:12:23 +0100 Giulio Benetti giulio.benetti@benettiengineering.com wrote:
Hi Andre,
On 31/01/22 14:57, Andre Przywara wrote:
On Mon, 31 Jan 2022 14:07:54 +0100 Giulio Benetti giulio.benetti@benettiengineering.com wrote:
Hi Giulio,
Il giorno 31 gen 2022, alle ore 13:57, Andre Przywara andre.przywara@foss.arm.com ha scritto:
On Mon, 31 Jan 2022 13:17:52 +0100 Giulio Benetti giulio.benetti@benettiengineering.com wrote:
Hi All,
I've tried several times on both F1C100s and F1C200s, maybe I'm doing something wrong but it doesn't work :-/
I burn sd-card with: # sudo dd if=u-boot-sunxi-with-spl.bin of=/dev/sdb bs=1k seek=8
I've triple-checked that the file has been written there by dumping the sd-card and I see it's written at 0x2000, so 8K offset, like the mangoPI image for example. But no uart output.
I haven't looked in detail what would be needed for the SPL part of the MMC driver to really work, I just checked that the base addresses are correct, the pinmux looked alright as well. The DT nodes are not needed for the SPL, and I don't see other changes related to MMC in the original series?
Yes that’s true.
Do you see UART output from the SPL, or nothing at all?
Unfortunately nothing either with SPL
That is interesting. If there would be a problem with the MMC code, you should at least see the SPL banner, and then see it failing loading U-Boot proper.
Yes
I take it FEL boot works?
Yes, it does work on both F1C100s and 200s boards :-/
Maybe we miss something with the clock setup that the BROM does for us during FEL operation?
Yes, it sounds like that, it's a classic
Can you try to clear the area between the end of the SPL and the beginning of the U-Boot proper image (dd if=/dev/zero of=/dev/sdX bs=1k seek=32 count=8)?
I've zeroed the first 32MB of sd-card now, so we're sure nothing can interfere.
# sudo dd if=/dev/zero of=/dev/sda bs=1M count=32 # sync # sudo dd if=u-boot-sunxi-with-spl.bin of=/dev/sda bs=1k seek=8 # sync
(/dev/sda is my sd-card for real since I have m.2 ssd)
Sorry, I meant to clear those bits after writing to the card (or clear that in the image file). binman now manages the packaging, so places 0xff into the 8K gap between our 24K SPL and the 32KB offset we place U-Boot proper at. If the BROM would read beyond the 24KB, the checksum would fail, but zeros are ignored by the algorithm, so we are a bit more robust. Also it might be interesting to set the size of 32KB (fixing up the checksum). What I was after is that no output could also mean that the BROM dismissed the SPL it found on the SD card. Do you see FEL mode enabled afterwards?
Does any other branch work with SD card booting? Can you check the SPL size in there (with hexdump -C u-boot-sunxi-with-spl.bin | head, check for the halfword at 0x10)?
This is the result of the command:
00000000 16 00 00 ea 65 47 4f 4e 2e 42 54 30 64 04 71 7c |....eGON.BT0d.q|| 00000010 00 60 00 00 53 50 4c 02 00 00 00 00 00 00 00 00 |.`..SPL.........| 00000020 2c 00 00 00 00 00 00 00 00 00 00 00 73 75 6e 69 |,...........suni| 00000030 76 2d 66 31 63 31 30 30 73 2d 6c 69 63 68 65 65 |v-f1c100s-lichee| 00000040 70 69 2d 6e 61 6e 6f 00 00 00 00 00 00 00 00 00 |pi-nano.........| 00000050 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................| 00000060 0f 00 00 ea 14 f0 9f e5 14 f0 9f e5 14 f0 9f e5 |................| 00000070 14 f0 9f e5 14 f0 9f e5 14 f0 9f e5 14 f0 9f e5 |................| 00000080 a0 00 00 00 a0 00 00 00 a0 00 00 00 a0 00 00 00 |................| 00000090 a0 00 00 00 a0 00 00 00 a0 00 00 00 ef be ad de |................|And you are not using UART0 on the PortF pins, so multiplexed with the MMC, I guess?
I use PE0 and PE1 Uart0 according to dts.
Have you got such board to test it?
That's the problem: I am driving blindfolded, because I don't have hardware :-(
Yeah, but I've read you've ordered it! :-)
Indeed, but it ships from Hongkong. Even without Chinese New Year that would probably take another week to arrive.
Cheers, Andre
Jesse has it working so maybe he can check how clocks are setup after BROM using u-boot itself. I've never understood why Allwinner/Rockchip place jtag on the sd-card pins. I really hate it, it would take so few with free jtag pins to check this :-/
Ciao :-)