<unlord>
at least to get started. I would be very interested in hearing if you get a stock upstream kernel working
<palmer>
we well eventually...
<palmer>
will
<unlord>
That would be great, because then I can junk this vendor sysroot and use normal linux
<courmisch>
unlord: stock upstream, greedy much
<courmisch>
I'd be happy with recent upstream + patches
<palmer>
ya, just having something sort of sane is all I need -- 5.10 is just so old
<unlord>
courmisch: sure, that would work too
<courmisch>
even VF2 doesn't have full upstream support yet, does it?
<courmisch>
I seem to be still stuck with 5.15 on the VF2
<unlord>
courmisch: I'm sure the runtime detection of V fails on this kernel
<palmer>
I thought VF2 worked upstraem? Is smaeul around?
<courmisch>
maybe recently
<courmisch>
but for me NVME never worked with upstream
<courmisch>
and the Debian guys appear to think that HiFive is the only thing that can boot upstream kernels
<palmer>
OK, that wouldn't be super surprising -- those StarFive chips were all kind of a mess, I wouldn't be surprised if there's a bunch of lurking issues
<courmisch>
I can only guess that there's a dirty little kludge in the vendor kernel that's not in upstream
<palmer>
ya, it's probably something small
<palmer>
do you know of a kernel that actually works?
<courmisch>
on what device?
<palmer>
VF2
<courmisch>
for VF2, I'm using the vendor kernel plus some custom config
<courmisch>
release from June or July, IIRC
<courmisch>
that's derived from 5.15
<courmisch>
that does mostly work
<palmer>
OK, 5.15 is still pretty old. I thought someone had something newer running...
<courmisch>
some people run 6.something with patches, but that broke NVME for me
<courmisch>
oh well, now to install Debian on the K230
<mps>
courmisch: for VF2 I use mainline kernel with patches from aurel32 git repo, not much patches
<mps>
and hope next week when 6.7-rc1 is released number of needed patches to boot it would be very small
<courmisch>
mps: NVME keeps staling for me with those (last time I tried, admittedly a while ago)
<courmisch>
mps: maybe I missed an kludge option or a kernel cmdline trick, I dunno
<mps>
courmisch: it is not perfect but it works quite fine for building kernel with root FS on NVME
<courmisch>
mps: gets stuck way too often for me, device effectively unusable
<courmisch>
people have blamed it on my power supply, but the same hardware works with the vendor kernel :shrug:
<mps>
courmisch: could be power supply issue, some alpine users had some similar problems with NVME and when they found better power supply it started to work fine
koolazer has quit [Server closed connection]
<mps>
idk, but I'm sure it stalls and crashes when I use some not enough good power supply
koolazer has joined #riscv
<courmisch>
I know, but clearly the vendor kernel is somehow able to get around it much better. Maybe it's just more aggressively working around lost IRQs or whatever, I dunno
<mps>
could be, I have no idea
<courmisch>
I don't mind buying another power adapter, but people seem pretty dodgy (somewhat understandably) about recommending anything
<mps>
before buying I would test with good one
<mps>
why waste money if it will not help
<mps>
I first tested with chromebook and apple silicon power supplies
<palmer>
for USB-C power adapters, that's the way I'd go for reliability -- the Chromebook guys in particular used to test these heavily, IIUC anything shipped along with a chromebook should be sane
<palmer>
I also have some of those USB-C pass-through multimeters, they can be useful to see if anything odd is going on