IRC log of #cubox of Mon 19 Feb 2018. All times are in CET < Back to index

08:27 topi`> Ke: if you manage to create a successful embedded controller to implement physical poweroff, please share the schematics with us :)
08:27 topi`> i've wanted this, for many years
08:27 Ke> you can do this at ATX level, though you would need to figure out, hot to power the controller itself
08:27 topi`> my first attempt was with ATmega but I couldn't get a working TCP/IP stack running ont op of it
08:28 topi`> problem is, I need to initiate the reset via tcp/ip
08:28 Ke> some SBC, like rpi zero or something?
08:29 topi`> yeah
08:29 Ke> way hardest part is powering the board itself
08:29 topi`> I'm familiar with the Allwinner boot procedure and linux hacks, so I tend to prefer OrangePI zero ;)
08:29 Ke> sure
08:30 topi`> also, OrangePIs replace the finicky microUSB power connection with a real barrel jack
08:40 Ke> next on my list is rk3399-gru-kevin
08:40 Ke> I'll need to get a working kernel
08:41 topi`> if the cortex-a72 cores in rk3399 are as fast as in the 8040, that would make a very nice chip
08:41 topi`> it does have 1 MB of L2 cache, so theoretically it could go somewhere
08:42 Ke> they should "Paul" metioned that Gnome was actually useable there with sw acceleration
08:42 Ke> IIRC
08:42 topi`> wow, then there's no bigger ram bandwidth bottleneck
08:42 topi`> like what we see on the i.MX6, making sw acceleration quite unusable
08:43 Ke> obviously some people are more patient than others
08:43 topi`> nobody expects it to be lightning fast :)
08:44 Ke> I use xmonad anyway, so even old rk3288 is fast enough for me
08:44 Ke> but rk3399 might have working virtualization and 64-bit system
08:44 topi`> xmonad rocks :D we seem to have same taste when it comes to window managing
08:44 Ke> also there is a cover for the SD card slot so your rootfs does not get unmounted every now and then
08:52 topi`> I don't want a rootfs on SD card
08:52 topi`> emmc is the minimum
08:52 topi`> sub-optimal, but somewhat reliable
08:53 topi`> Ke: what kind of machine is your gru-kevin? It sounds like a chromebook...
08:53 topi`> I got myself a Lenovo N23 yoga, but it's based on the Mediatek 8173
08:54 topi`> unfortunately
09:41 Ke> topi`: Samsung chromebook plus
09:41 Ke> I prefer SD to eMMC, as SD is user replaceable
10:00 topi`> emmc has hardware protection for boot area :)
10:00 topi`> as long as UBOOT loads, everything is up for grabs
10:01 topi`> if you need an alternate kernel loading facility, use altbootcmd
10:01 Ke> perhaps, if you don't have the local access
10:01 topi`> or you can boot from a M.2 disk, they can also be swapped/replaced
10:02 topi`> do you have a Clearfog? they can nicely boot from M.2
10:03 Ke> no I don't
10:04 jnettlet[m]> The HB2 boards can also boot from m.2
11:35 vpeter> Is there any nice documentation about using gpio on clearfog pro?
12:02 jnettlet[m]> vpeter: we haven't implemented wiringX support for the Clearfogs yet. We are trying to get a universal standard mapping for the mikrobus, however that has stalled for the moment.
12:03 jnettlet[m]> do you just need to know the number mappings?
12:10 vpeter> I don't ned any lib. Just need to use one gpio on/off. Nothing special. So mapping is needed.
12:11 vpeter> Like at the end https://wiki.solid-run.com/doku.php?id=products:a38x:clearfog
12:25 jnettlet[m]> vpeter: they use the same mpp number that is in the device-tree file. the int gpio is 22 and the reset gpio is 29
12:29 vpeter> Ok, I think I got it. Will try to turn one led when back home.
17:01 vpeter> topi`: Did you ask about clearfog pro networking? Look this https://forum.armbian.com/topic/6502-clearfog-pro-41414-network-manager-fails/?tab=comments#comment-49564
18:56 vpeter> jnettlet[m]: gpio on clearfog is 3.3V or 5V ?
18:57 jnettlet[m]> 3.3
19:01 vpeter> Then I was right. But asked just in case :-)
19:15 vpeter> echo 1 > value # LED is on
19:15 vpeter> echo 0 > value # LED is off
19:15 vpeter> Phase one completed. Moving to phase 2: making a switch with transistor to turn HDD on/off. Ordering transistor :)