IRC log of #cubox of Mon 08 Jan 2018. All times are in CET < Back to index

09:32 heap_> Artox: ?
09:32 Humpelst1lzchen> Any chance for Spectre fixes for i.MX6 kernels?
09:39 Ke> Humpelst1lzchen: can't you use upstream kernel?
09:49 Humpelst1lzchen> Ke: freescale upstream? Do they have the dtbs?
09:49 jnettlet[m]> Humpelst1lzchen: ARM is releasing more information tomorrow. We have working code that demonstrates the vulnerability however the only existing patches that have been released are for Aarch64 and x86
09:50 Humpelst1lzchen> jnettlet[m]: great
09:50 jnettlet[m]> Google has some fixes for all architectures in their latest security patch for Android but those changes have not been released to AOSP yet.
09:51 jnettlet[m]> so fixes are needed, and I am closing tracking the work being done. It will be a busy week but my plan is to get fixes pushed for all our platforms as soon as there is a good general consensus that they are the proper fixes.
09:52 Humpelst1lzchen> good luck then
09:52 jnettlet[m]> I will also benchmark general workloads and do a writeup on what I find.
10:18 Ke> jnettlet[m]: I guess mcbin is not "your" platform here?
10:19 jnettlet[m]> Ke: we have reached out Marvell and are waiting to hear about their strategy. If I don't hear anything by end of day tomorrow, then I will also release patched repositories for the ATF and kernel.
10:20 Ke> thanks
10:20 Ke> kernel is of no interest though
10:21 Ke> I would obviously hope that ATF would be of no interest either, but upstreaming there is quite a bit further aawy
10:21 Ke> away
10:21 Ke> kernel with version number 4.4.52 or whatever looks absolutely scary anyway, no matter which fixes are backported
10:24 Ke> I started planning arduino watchdog btw. https://users.aalto.fi/~jkarlson/files/arduino.svg
10:25 Ke> though I need 5V inputs for the arduino, so I'll perhaps use 5V from molex for the buttons
10:26 Ke> also I am unsure whether 5V signal from arduino can control the presumably 12V reset on mcbin
10:26 Ke> with FET
10:29 Ke> arduino seems to have almost everything I need anyway
10:30 Ke> btw. buying molex connectors is about the hardest things in electronics
10:51 Ke> also FET requires pullup
11:36 Artox> heap_, sorry, sadly not. Maybe suihkulokki can update us if he looked into getting dove support enabled
11:38 Artox> it really is just a matter of adding to a) debian-installer, and b) kernel .config; both of which have patches available
11:39 Artox> I just haven't figured out how to speed up getting those into debian, yet.
13:04 suihkulokki> Artox, heap_, had a look earlier but forgot about it
13:05 suihkulokki> Artox: the only potential problem is, that this add some size into vmlinuz, and kernel size is limited on some platforms
13:07 Ke> isn't the point to put the extra drivers to modules and load the necessary ones from initrd?
13:07 Ke> just like on x86?
13:07 suihkulokki> Artox: btw the D-I boot hang is solved with https://patchwork.kernel.org/patch/10146687/
13:09 suihkulokki> Ke: yes, but some things can't be modules. The cpu on Dove cubox is different enough that some low-level bits need to added to the kernel.
13:09 Ke> ok
13:12 suihkulokki 13:12 * suihkulokki wished a public TRM for Armada 8K existed
13:16 jnettlet[m]> suihkulokki: hopefully it will get released this millenium :) Getting anything through Marvell's legal department is a very slow tedious process. Took nearly a year for the a38x reference
14:37 Artox> suihkulokki, Hmm :( is debian in a situation where this issue has already surfaced on a platform?
14:37 Artox> and then wouldn't that limited platform deserve its own kernel flavour?
14:38 Artox> anyway, should I do anything to push debian on dove forward?
14:39 Artox> imo it works far too well with mainline to ignore
14:42 Artox> suihkulokki, so we need a patched DTB?
14:42 Artox> looks like it is worth switching to the mainline dtb
14:47 suihkulokki> Artox: I'll propose Dove/Debian once the main kernel developers have time from meltdown/spectre crisis
14:47 Artox> Hah, sure, thanks
14:54 Ke> jnettlet[m]: so they are going to release?
16:09 jnettlet[m]> Ke: no guarantees
16:09 jnettlet[m]> hopefully