| 05:43 | jnettlet[m]> | Ke: The issues that used to exist for virtualization were fixed in a patch to ATF, this should fix things for both u-boot and uefi based systems. |
| 07:10 | Ke> | jnettlet[m]: when were they fixed? |
| 07:11 | jnettlet[m]> | Ke: I think a couple of months ago now, if you are using Marvell's fork of ATF |
| 07:11 | Ke> | yes |
| 07:11 | Ke> | 17.08 now |
| 07:12 | Ke> | which could be older that a couple of months ago |
| 07:12 | Ke> | does upstream even work? |
| 07:13 | jnettlet[m]> | I have not been tracking Marvell's commits upstream. I am not even sure how they are handling that. |
| 07:36 | topi`> | back to work, folks |
| 07:36 | topi`> | hope the Santa dude brought you some shiny new SBCs :) |
| 07:36 | topi`> | what kind of boot sector does UBOOT use on the HB/HB2? there's this in the uboot src, envtools: |
| 07:37 | topi`> | #define HAVE_REDUND /* For systems with 2 env sectors */ |
| 07:37 | topi`> | #define DEVICE1_NAME "/dev/mtd1" |
| 07:37 | topi`> | does it just simply mean that on some systems, you have /dev/mtd1 and /dev/mtd2 for reasons of redundancy? |
| 07:47 | topi`> | I think I want to setup a dedicated board for just compiling uboot and envtools :) |
| 07:48 | topi`> | my cross compiler setup works for compiling kernels and the uboot itself, but fails miserably when targeting the debian userspace |
| 09:12 | Ke> | Artox: do you have some spec of what were you running |
| 13:21 | suihkulokki> | Artox: did some tests - same kernel works every from my installed emmc partition and hangs every time when booting the installer |
| 13:25 | Ke> | could it be explained by failing to mount the rootfs |
| 13:25 | Ke> | in a way that hangs the hw |
| 13:25 | Ke> | eg. firmware initializing the USB incorrectly |
| 13:33 | Artox> | suihkulokki: :( |
| 13:33 | Artox> | but thats good news, could be worse |
| 13:45 | Ke> | my mcbin is now in initial production testing |
| 13:45 | Ke> | reliable boot would be extremely important |
| 13:49 | suihkulokki> | Artox: did your EFI build have updated or old dtb ? |
| 14:14 | suihkulokki> | Artox, ke: http://paste.debian.net/1003256/ |
| 14:15 | suihkulokki> | the key difference appears phy/mdio drivers get probed before udev starts scanning for devices |
| 14:26 | Ke> | but that actually shows nothing about device internal state |
| 14:34 | Artox> | suihkulokki: Marvells DTB, I didn't bother replacing it |
| 14:57 | suihkulokki> | Artox: it's probably worth testing if the problem goes away by just updating DTB |
| 15:10 | suihkulokki | 15:10 * suihkulokki notes EFI has menu item for choosing between DTB/ACPI |
| 15:10 | suihkulokki> | this would be a great place to choose between DTB versions |
| 21:36 | Ke> | suihkulokki: thanks, qemu works now even with kvm |
| 21:36 | Ke> | at least the installer boots |
| 21:39 | Ke> | jnettlet[m]: it seems also that 17.08 release has working kvm at least to some extent |