IRC log of #cubox of Fri 05 Jan 2018. All times are in CET < Back to index

12:44 suihkulokki> Think I found the error causing the D-I hang.
12:53 Ke> but you are not going to tell us, what it is?
13:00 suihkulokki> Ke: it's much more fun to keep you in suspension ;)
13:01 suihkulokki> Just thought I'd wait until Artox re-appears
13:07 topi`> jnettlet[m]: so you're back in business after the holidays?
13:07 topi`> I'm a bit late here, but going to have a look at u-boot configuration
13:07 jnettlet[m]> Yep
13:07 jnettlet[m]> Okay
13:07 topi`> do you have any idea how far along our build is?
13:09 jnettlet[m]> I do not. I believe production is down next week.
13:23 topi`> btw, I think Hummingboards are susceptible to Spectre
13:23 topi`> the Cortex-A9 has out-of-order execution
13:25 Ke> topi`: yes https://developer.arm.com/support/security-update
13:26 Ke> variant3 is not applicable to A9 though
13:43 Ke> A53 not affected also, which is nice for some people
14:08 Ke> btw. just realized that it's a shame not to have LITTLE cores on Armada 8040 to run sensitive stuff
16:33 suihkulokki> Ke: s/sensitive/untrusted/
16:34 Ke> are you sure?
16:34 Ke> either is probably ok though
16:35 suihkulokki> Say, the browser JS runs on A72, it can do speculative attacks
16:36 suihkulokki> And running the kernel on A53 won't stop the browser JS from speculating
16:37 suihkulokki> but running the browser in A53 the JS can't speculate since everything is in-order
16:39 suihkulokki> Now since Artox isn't appearing: The D-I issue is that ethernet and mmc share a clock, but only ethernet requests it.
16:39 suihkulokki> the MMC driver was fixed by 4.14 but the relavant DTB update slipped
16:39 Ke> in dtc?
16:39 Ke> yes
16:40 Ke> or dts
16:41 suihkulokki> https://patchwork.kernel.org/patch/10146687/
16:51 Ke> thanks
19:47 Ke> is there some mcbin pin that would allow me to send reset signal to external watchdog
19:48 Ke> it seems they all just want to have something like VCC for at least 100ns or something and then reset to GND or the reverse
19:49 Ke> like GPIO, which sadly was not available
19:50 vpeter> Last resort is some USB to serial adapter and using pin from there.
19:54 Ke> mcbin has serial headers
19:54 Ke> I guess they are 3.3V, but that should be sufficient
19:57 Ke> perhaps just sending 0 for a long time might do it
19:57 vpeter> Then try RTS pin from there with one small Linux app to control it.
20:08 Ke> I guess I could just go for the arduino
20:10 vpeter> Did you look for any special watchdog IC's ? My friend is using them in combination with MCU's watchdogs (to be even more robust).