00:38 <apritzel> hexdump0815: thanks for the "u-boot v2021.10 results in a less stable system" report, but it would be good to either bisect or at least narrow down the instability

00:41 <apritzel> and btw. we are pretty sure that HDMI on the Eachlink needs a command line override to detect a monitor, at least I came to that conclusion some months ago

00:42 <hexdump0815> apritzel: i plan to look closer into this instability - bisecting might be a bit painful but i was thinking about putting the ram timing code from the working old version into the new one to see if this is the problematic area or now

00:43 <apritzel> hexdump0815: which old version? from earlier mainline U-Boot?

00:43 <hexdump0815> v2020.07 with a few patches on top

00:47 <hexdump0815> apritzel: i just checked - the working one should be v2020.07 plus this patch on top: https://github.com/hexdump0815/u-boot-misc/blob/5f03288dbfe47a8f291f751cc07d6581817d0e87/misc.h6/misc-tests/test-15.patch

00:47 <hexdump0815> at the top of the patch is a link to the irc discussion from back then

00:49 <hexdump0815> this patch got my other no-name box working and is the u-boot i was using since then - so maybe something of this is not yet in mainline?

00:57 <apritzel> looks like this patch got merged into v2021.04: ec9cdaaa13d5

09:55 <hexdump0815> apritzel: looks like you are offline, but just in case you are reading it via irc log - looks like the u-boot problem comes from atf - new u-boot works with v2.3 and is broken with v2.6

10:48 <jernej> hexdump0815: same procedure, bisecting would be best

16:39 <hexdump0815> jernej: yes i'll do that in case nobody has an idea off mind which of the atf changes between v2.3 and v2.6 might cause those instabilities

17:24 <jernej> hexdump0815: how quickly can you reproduce the issues? even if testing takes 15 minutes, it's still the quickest way to get answers

17:38 <hexdump0815> jernej: its no problem to do it - just want to have a little break - too much trying out things yesterday :)

18:26 <apritzel> hexdump0815: from a quick browse over the changes in TF-A there is nothing that really sticks out when it comes to stability

18:31 <apritzel> there is a lot of refactoring, significant changes in how we do core power management, using RSB for talking to the PMIC, enabling the SCP, and enabling a Cortex-A53 erratum workaround (1530924)

18:37 <hexdump0815> apritzel: ok - will do the bisecting during the next days then

18:38 <apritzel> hexdump0815: how did the instability manifest? And was that on the Eachlink H6 Mini?

18:40 <hexdump0815> apritzel: i think i saw it both on the qplus and on the eachlink - the test was to run "7z b" which puts quite a bit of stress to the system and then it usually hangs quite fast (within a minute or maybe max a few minuntes)

18:43 <apritzel> hexdump0815: thanks, I might give it a try on my Eachlink then

18:46 <hexdump0815> apritzel: would be interesting to see if you can reproduce it as well

18:47 <apritzel> that would be the first step, yes

21:23 <gamiee> Hi, is Bootrom in V3s or other similar SoC capable of loading u-boot from SDC1?

21:25 <gamiee> (looks like not, so second question is..) Is u-boot capable of treating SDC1 as SD Card? Basically it should be same as SDC0, but I might miss something

23:12 <apritzel> gamiee: it should work for U-Boot proper, by just adding the right DT nodes

23:12 <apritzel> loading U-Boot proper from there is another story though, but since you need something else for the SPL anyway, that's probably not a concern?