| 10:10 | topi`> | Ke: why do you need to reset the board? supposedly some linux boots just get stuck, or what? |
| 10:11 | topi`> | my mcbin might also be in exactly that state. Won't know until I can install a raspberry next to it to monitor the serial console. |
| 10:39 | Ke> | topi`: yes, linux never boots on the first cold boot for me |
| 10:39 | Ke> | topi`: having a watchdog from u-boot would workaround my problem completely |
| 10:40 | Ke> | but without watchdog router that never boots after power cycle is no-go |
| 10:40 | Ke> | unless I get an UPS |
| 10:41 | Ke> | and UPS almost never works, because power cuts are so rare that when they happen, the UPS is already broken |
| 14:28 | topi`> | what's the usual size for the SPL in i.mx6? I'm trying to take a shasum to compare where the bits went wrong (we have one HB that doesn't boot, doesn't even show the 2 lines of text on UART from spl) |
| 14:29 | topi`> | Ke: sounds like some very obscure DRAM problem you're seeing |
| 14:29 | topi`> | I think I've also seen that, maybe once or twice? |
| 14:29 | topi`> | then I compiled my own mv-ddr and installed that and since then it has worked |
| 14:31 | Ke> | it's possible, could be anything else as well |
| 14:31 | topi`> | our process of flashing the HBs seems to be completely broken. We consistently end up with boards that do not boot at all. |
| 14:31 | Ke> | I have tried about 10 different bootloader configs and they all fail the same way |
| 14:31 | topi`> | maybe should try to place the emmc chip into reset to see if it accepts usb-otg |
| 14:32 | topi`> | Ke: that sounds frustrating, or rather, as they say, business as usual for the poor early adopter |
| 14:32 | Ke> | sure |
| 14:32 | Ke> | if I could be sure the hw is not broken, it would be completely fine |
| 14:32 | topi`> | well if it "mostly works", then they just tell you the HW is okay :) |
| 14:33 | topi`> | and anything that is only occasional behaviour, is of course blamed on the software |
| 14:33 | topi`> | I worked for Nokia in the dark past, and that's how it always was :D |
| 14:33 | jnettlet[m]> | hey guys. Sorry I am traveling today so a bit on and off |
| 14:33 | topi`> | lay the blame on the SW guys |
| 14:34 | jnettlet[m]> | topi`: how are you loading SPL? |
| 14:34 | topi`> | eFuses at 0x1020 IIRC |
| 14:34 | jnettlet[m]> | luckily SR is not responsible for mv_ddr. |
| 14:35 | jnettlet[m]> | but we have recently uncovered an issue with the mainline mv_ddr vs Marvell's u-boot for the a38x |
| 14:35 | jnettlet[m]> | so I am trying to track that down. |
| 14:35 | jnettlet[m]> | topi`: and you are setting the active boot hardware partition via the mmc interface? |
| 14:36 | topi`> | how does that happen? all I know is the same SPL code works if we write it to a SD card instead |
| 14:37 | topi`> | oh, you mean the mmcblk2bootx devices |
| 14:37 | topi`> | we don't use those |
| 14:37 | jnettlet[m]> | so you are just writing to the raw mmcblk0 device? |
| 14:37 | topi`> | we've managed to clone HB installations by just reading /dev/mmcblk2 with dd and writing it onto a new device, and it boots |
| 14:37 | topi`> | yes |
| 14:38 | topi`> | I think it does load the SPL from emmc, but then that SPL tries to load UBOOT from SDcard and we have to place a properly formatted SD card into the slot |
| 14:38 | topi`> | s/formatted/prepared/ |
| 14:39 | jnettlet[m]> | oh wait. which u-boot are you using? You are probably missing the mmc hotplug detection patch. |
| 14:39 | topi`> | something I downloaded from the soild-build.xyz repos |
| 14:40 | topi`> | on the jessie repo there were the spl + uboot files |
| 14:40 | topi`> | jnettlet[m]: can you prepare us a shell script you would use for prepping a blank HB for emmc boot? |
| 14:41 | topi`> | if we get that to work, it would mean one less woe for us |
| 14:42 | jnettlet[m]> | topi`: I will need to check with Artox on that. I am not 100% sure which build that is. |
| 14:44 | jnettlet[m]> | topi`: yes but it won't happen until Wed. All day tomorrow I am at a customers. If I get settled in Paris early enough tonight and the wifi and/or LTE doesn't suck I will put something together. I think I have 1 unfused SOM travelling with me :) |
| 14:47 | jnettlet[m]> | hope that all made it through. The airport wifi just gave up on me. |
| 15:09 | topi`> | ok |
| 15:09 | topi`> | but, anyhow, my point was that we have successfully booted with that binary, and even deployed the pilot project with that, but now writing that same image to a new board gave us a dead silence on the serial line... |
| 20:10 | Artox> | jnettlet[m], topi` as to imx6 emmc boot: I have two problems at this point: something breaks when booting stretch, and jessie is missing libfdisk (which I use for the new filesystem expansion script) |
| 20:11 | Artox> | and, of course, passing root=UUID to kernel on jessie. I guess I can make it happen, I was hoping to skip it altogether for jessie and throw away my image building code. |
| 20:14 | Artox> | u-boot age is April 6. |
| 20:15 | Artox> | it is same as HEAD here: https://github.com/SolidRun/u-boot-imx6 |
| 20:16 | Artox> | not imx6-next! |