07:35 | mWaltari | ja 19:45 pienemmässä salissa. muta liput pitäs varmaan ostella per heti... |
07:35 | mWaltari | Oops... |
11:25 | topi` | does anyone have instructions on how to try out Ubuntu Core (with snappy packaging) on a Hummingboard Gate? |
11:34 | vpeter | I was only playing with Snappy long time back. |
12:15 | topi` | we're trying to see if Snappy is the answer to our requirements |
12:16 | topi` | problem with .deb packages is that if something goes wrong, it's hard to restore the original system state (before the bunch of .debs were installed) |
12:18 | vpeter | Didn't jnettlet mentioned some other OS also? From Intel I think. |
12:18 | topi` | lots of OS'es nowadays :) |
12:19 | topi` | but we are currently based on Debian, and going to Ubuntu wouldn't be such a big departure (I guess) |
13:34 | Artox | [09:43:41] wbx: if you encounter a system that doesn't have a working default boot script, please move it to distro boot in upstream u-boot |
13:34 | Artox | would you be interested to review one such patch? |
13:35 | Artox | I have had one on my disk for months, but my confidence in it isn't high enough |
13:35 | agraf | Artox: sure |
13:37 | Artox | http://paste.debian.net/hidden/da624609/ |
13:37 | Artox | there |
13:37 | Artox | if I remember right, initrd_high and fdt_high need to be reviewed |
13:37 | Artox | maybe we need neither in mainline? |
13:37 | Artox | we need both on marvells ugly fork |
13:39 | Artox | topi`: I guess ubuntu core isn't hard work |
13:39 | Artox | mostly jon isn't happy to have me put my time into buntu |
13:40 | Artox | somethin along the lines of canonical profits, so they should at least involve themselves |
13:44 | vpeter | Artox: how far is your LEDE project for clearfog? Usable? |
13:44 | Artox | Useable |
13:44 | Artox | the solidrun repo is usable without any wifi drivers |
13:44 | Artox | my latest fork is usable at least with ath10k |
13:44 | Artox | there have been reports of toruble with ath9k |
13:45 | Artox | and generally speaking, wifi drivers finally build |
13:45 | vpeter | Good, will build it then. Any special instructions needed? Probably not. |
13:45 | Artox | found my git repo? |
13:46 | Artox | https://github.com/Artox/lede-project/tree/mine_v2 |
13:46 | Artox | this is teh current development location |
13:46 | vpeter | jnettlet point it to me. |
13:46 | Artox | great |
13:46 | Artox | want a .config to start with? |
13:46 | vpeter | I'm using openwrt (from marvel) and noticed that they didn't include bunch of packages. Like nfs server. |
13:47 | vpeter | Yes please :) |
13:47 | Artox | there is a good reason why they don't include many packages |
13:47 | Artox | random build failures creeping up |
13:47 | Artox | and it eats much time |
13:47 | vpeter | I add another feed from openwrt git and my required packages build and working :) |
13:47 | vpeter | Needed only nfs/tftp for network booting. |
13:48 | vpeter | Do you know if VyOS is usable with clearfog? And everything supported. |
13:48 | Artox | I don't know |
13:49 | vpeter | I already asked on a forum. Let's wait for answer. |
13:49 | Artox | http://pastebin.com/UAbcsd3V |
13:50 | Artox | warning: before using sysupgrade, find out how it is implemented |
13:51 | Artox | it mightb e smarter to just flash new .squashfs file to p2 |
13:51 | wbx | agraf: is Artox patch fine? |
13:54 | agraf | oh, no highlight |
13:54 | malte_ | UnicronNL : hows vyos working on clearfog? :) |
13:54 | Artox | actually I think fdt_high and initrd_high should be just set to 0xffffffff |
13:54 | agraf | sorry - didn't see messages flowing in |
13:54 | UnicronNL | malte_: on pro it works great! |
13:54 | UnicronNL | base also, only official dts |
13:55 | malte_ | great |
13:55 | vpeter | UnicronNL: everything works? |
14:10 | UnicronNL | vpeter: yes, vyos is beta version, but all major stuff works |
14:10 | UnicronNL | for pro switch config also works |
14:11 | UnicronNL | but all interfaces are separate, to make it function as switch you have to put it in bridge |
14:13 | vpeter | Looking the features and don't see it can run for example nfs and tftp server. Or I don't see it correctly. And does it have bunch of other packages like openwrt or this is different beast focused only on routing? |
15:05 | agraf | wbx: Artox: so the patch looks ok as a first shot to me |
15:05 | agraf | you don't need the non-_r variables |
15:05 | agraf | scriptaddr was something different too iirc |
15:06 | agraf | and you onbiously need the other drivers enabled as boot targets |
15:06 | agraf | also, you want to set CONFIG_DISTRO_DEFAULTS |
15:09 | Artox | agraf: I needed tjnem all or pieces would crash |
15:10 | Artox | but then this piece is close to a year old |
15:11 | Artox | I will look into it |
15:11 | Artox | thanks |
15:11 | Artox | about other boot targets |
15:11 | Artox | afaik network is still broken |
15:12 | Artox | so I only need sata boot target |
15:12 | Artox | emmc is sdcard |
15:12 | Artox | it is either or here |
15:15 | Artox | I specifically remember needing scriptaddr btw, and from reading where it was used it seemed save to me to reuse loadaddr |
15:22 | agraf | yeah, i remmeber that there was a second variable |
15:22 | agraf | i just thought it also had some _r or so suffix |
15:22 | agraf | maybe i'm misremembering |
15:26 | Artox | I will let you know when I reworked it a bit |
15:26 | Artox | d you have any hints on the *_high variables? |
15:26 | Artox | is u-boot smart enough not to need them? |
15:26 | Artox | or do I have to find sane values |
15:27 | Artox | this is slightly too low-level for my mind at this point |
15:30 | Artox | is there an order to BOOT_TARGET_DEVICES? |
15:30 | Artox | and is there a preferred order to it? |
15:35 | agraf | Artox: iirc the _high variables are needed for kernels that can only use so much offset from their load address |
15:35 | agraf | Artox: but for efi boot, you don't need them :) |
15:35 | agraf | Artox: the preferred order is "whatever makes sense" |
15:36 | agraf | Artox: zynqmp for example determines the device u-boot was loaded from and puts that as first in the list |
15:36 | agraf | Artox: in house we usually set DHCP as first and have a grub binary that exits as default boot entry |
15:39 | Artox | aha |
15:39 | Artox | well I would too choose sdcard first |
15:39 | Artox | then sata |
15:39 | Artox | then usb |
15:39 | Artox | then network |
15:39 | Artox | what is the last parameter for network? |
15:39 | Artox | on https://patchwork.ozlabs.org/patch/695928/ it is set to "na" |
15:40 | Artox | and I will try dropping the _high parts then |
15:47 | vpeter | Artox: seems every build of openwrt or lede has some build issues :( http://pastebin.com/TmjMFhvD |
15:49 | Artox | did I forget to push sth? |
15:49 | Artox | hang on |
15:50 | Artox | ah yes I did |
15:50 | Artox | sorry for that |
15:50 | Artox | vpeter: pushed |
15:50 | Artox | this is because lede patches both the kernel, AND backports |
15:50 | Artox | and we do neither |
15:50 | Artox | well, we do patch the kernel, but no lede patches |
15:50 | Artox | just board support |
15:51 | vpeter | let's see. |
15:51 | Artox | it is quite a pain that they aren't using mainline functionality |
15:51 | vpeter | Such a nice hw but sw is totally different case. |
15:51 | vpeter | At least for end users. |
15:51 | Artox | yep |
15:52 | Artox | there seems to be quite a lack of cooperation in so many places |
15:52 | vpeter | Agree. |
15:52 | Artox | e.g. cryptodev with distros with board vendors |
15:52 | Artox | with upstream ssl |
15:54 | vpeter | Board is available for how long? More than a 6 months? I woud expect all this distros openwrt, lede, ... to be build by one small script automatically. |
15:55 | vpeter | Even marvel's openwrt didn't build for me. Build failed at some point and I need to copy one file manually somewhere in build folder. |
16:13 | vpeter | Artox: now it builds ok. Thanks. |
16:19 | Artox | Great |
16:42 | Artox | agraf: not as easy as I had hoped |
16:42 | Artox | I get the feared include/config_distro_bootcmd.h:165:2: error: expected '}' before 'BOOT_TARGET_DEVICES_references_SATA_without_CONFIG_CMD_SATA' |
16:42 | Artox | but comments in bootcmd.h say "A note on error handling: It is possible for BOOT_TARGET_DEVICES to reference a device that is not enabled in the U-Boot configuration" |
16:42 | Artox | I may have understood this wrong |
16:43 | Artox | I think it means to say this configuration can be done by hand, but is wrong |
16:44 | Artox | it looks like sata support is not there yet for the clearfog |
16:44 | Artox | :( |
16:45 | Artox | no big deal though for now |
16:51 | Artox | so usb and network aren't working either |
16:51 | Artox | so it makes no sense to even have them in the boot order, right? |
17:33 | agraf | Artox: did you patch Kconfig? |
17:43 | Artox | agraf: I patched teh defconfig |
17:43 | Artox | I was going to test it first with grub befpre posting the new patch |
17:43 | Artox | only to find that I hadn othing to test with |
17:44 | Artox | http://paste.debian.net/897715/ |
17:44 | Artox | agraf: this is what it looksl ike now |
17:45 | Artox | but it really needs to be tested |
17:45 | Artox | I have not booted anything from that yet |
17:46 | agraf | Artox: please patch the defconfig the other way around :) |
17:46 | Artox | currently busy converting my debian image script to create an efi partition |
17:46 | agraf | Artox: so don't call the defconfig - patch the kconfig |
17:46 | agraf | s/call/patch/ |
17:48 | Artox | is it correct that I should install grub as /EFI/BOOT/bootarm.efi |
17:48 | Artox | ? |
17:51 | Artox | and then |
17:51 | Artox | grub2-mkstandalone -O arm-efi? |
17:52 | Artox | or arm-uboot, agraf? |
17:55 | agraf | Artox: grub2-mkimage |
17:55 | agraf | Artox: and yes, efi/boot/bootarm.efi |
17:55 | agraf | Artox: https://patchwork.ozlabs.org/patch/696559/ |
17:56 | agraf | Artox: ( cd ~/grub2-arm; wget -O - http://download.opensuse.org/ports/armv7hl/distribution/leap/42.2/repo/oss/suse/armv7hl/grub2-arm-efi-2.02~beta2-87.1.armv7hl.rpm | rpm2cpio | cpio -di ) |
17:56 | agraf | :) |
18:04 | Artox | so suse ships an actual grub image |
18:04 | Artox | unlike debian where Iwill have to call that mkstandalone applicaion |
18:04 | Artox | will try to boot grub2 asap |
18:15 | agraf | Artox: yeah, no guarantees that it contains all modules you need though :) |
18:15 | agraf | Artox: it's missing lsefisystab for example |
18:19 | Artox | we shall see |
18:19 | Artox | I will try the suse bianry now, while in the background my new dbeian image is cooking |
18:20 | Artox | well, grub loaded |
18:23 | Artox | apparently it lacks ext* |
18:25 | Artox | weird, the debian grub is huge, and can't even see the partitions |
18:29 | Artox | ah no, suse version works perfect |
18:30 | Artox | even with ext4 |
18:33 | agraf | *shrug* :) |
18:33 | agraf | grub modules can be tricky |
18:33 | Artox | time to give it some rootfs |
18:45 | Artox | hm |
18:45 | Artox | I am a bit lost how to boot |
18:45 | Artox | agraf: |
18:46 | Artox | do I just run the 2 commands devicetree and linux ? |
18:46 | Artox | devicetree causes the board to reset for some reasn |
18:48 | Artox | ah no |
18:48 | Artox | it just resets when I copypaste the cmd to my terminal |
18:48 | Artox | have to type by hand :< |
18:49 | Artox | and it boots |
18:49 | Artox | very neat |
19:12 | Artox | so now grub has to somehow pick up the kernel in grub.cfg |
19:14 | Artox | even the fsck code in initrd works |
19:14 | Artox | sweet |
19:37 | Artox | enough for today |
19:37 | Artox | I can't get grub to auto-load the grub.cfg |
19:37 | Artox | and I can't get grub-mkconfig to detect any system |
19:37 | Artox | other thn that manual booting works |
19:45 | Artox | agraf: the other mvebu based boards do not configure distro support |
19:45 | Artox | should I enable it for ARCH_MVEBU anyway? |
20:00 | Artox | used TARGET_CLEARFOG now |
20:04 | Artox | agraf: http://paste.debian.net/897772/ v3 |
21:32 | agraf | Artox: sorry, it was my turn tonight to bring our son to bed :) |
21:32 | agraf | Artox: auto-loading grub.cfg is dark magic :) |
21:33 | agraf | Artox: basically what happens is that bootefi passes the device and path that the grub.efi binary was loaded from to grub on bootefi |
21:33 | agraf | Artox: grub takes that, extracts the device identifier from it, adds the partition snippet and its prefix to it |
21:33 | agraf | Artox: and then uses that as $prefix variable |
21:33 | agraf | Artox: so if it doesn't find grub.cfg, run the "set" command |
21:34 | agraf | Artox: and check the $prefix variable |
21:35 | agraf | Artox: the patch looks reasonable, it would be great to add the other boot methods though :) |
21:35 | agraf | Artox: I think we can easily take it to the mailing list from here though |