15:52 <sjg1> Tartarus: I fixed it yesterday. It seemed to need a reboot as systemd did not like running out of disk space. We should fix the 'conf' tool to give up when it sees EOF on stdin (or perhaps after prompting 100 times)

21:39 <vagrantc> i noticed imx8 boards require the bl31.bin to be present in the top-level directory of u-boot to build ... how feasible would it be to support a BL31 variable like done with rockchip and sunxi boards?

21:40 <vagrantc> when building multiple boards from the same sources, this ... makes it difficult to ensure the correct bl31.bin is present at build time

22:29 <marex> vagrantc: I'll be talking at fosdem about removing that bl31 dependency altogether ;-)

22:30 <vagrantc> marex: infinitely better! :)

22:34 <marex> vagrantc: yeah ;-)

22:34 <marex> vagrantc: although, I practiced the talk today and I need to slim it down, it is already lengthy

22:34 <marex> vagrantc: you coming for fosdem btw ?

22:37 <vagrantc> marex: unfortunately no :/

22:38 <marex> oh well, check the recording or patches posted to the ML a few weeks ago ;-)

23:41 <sjg1> marex: That is great to see. If it is just some C code with no great magic, why didn't ARM do this ages ago?

23:47 <marex> sjg1: I suspect because they wanted BSD-licensed bootloader where others would contribute and big companies could take without giving back

23:54 <sjg1> marex: Sad. Does it completely replace BL31 or just parts of it? I suppose it depends on the SoC

23:55 <marex> on mx8m , the bl31 blob really only provides the PSCI and NXP custom extensions on which the vendor kernel depends

23:55 <marex> if you dont use vendorkernel, the u-boot support is enough

23:55 <marex> the PSCI should really only provide CPU core on/off , power off, system reset ... functionality

23:55 <marex> and that isn't hard to implement in u-boot

23:58 <sjg1> marex: I see. You have the code in arch/arm/mach-imx/imx8m/psci.c - it is not actually generic to all ARM64?

23:59 <marex> nope

23:59 <marex> that part is SoC specific

23:59 <marex> there is another one in r8a779a0 , different one

23:59 <marex> and another one for stm32mp15xx

23:59 <marex> and so on

23:59 <sjg1> marex: OK, that's a shame. I had hoped it would be standard