| 11 Mar 2026 |
 dramforever | notthe most representative but on reddit on coremark they got -30% perf improvement adding vectors on k3. yes that's negative thirty percent, a downgrade | 00:31:57 |
 JamieMagee | That sounds somewhat promising. | 00:34:45 |
| JamieMagee | I tried running hydra using QEMU for RISC-V support, and ran headfirst into: https://github.com/NixOS/nixpkgs/issues/460825#issuecomment-3994959331
TL;DR RISC-V bootstrap binutils is predates support for zca, while GCC 15 now implies support for zca :/ | 00:39:27 |
| dramforever | that's an unrelated failure | 00:45:37 |
| dramforever | if convenient can you open a new issue for that? thanks | 00:46:01 |
| JamieMagee | dramforever: I opened https://github.com/NixOS/nixpkgs/issues/498689 | 01:01:39 |
| JamieMagee | I was going to wait for the bootstrap overhaul, but who knows how long that will take to land. So this should resolve the issue for now: https://github.com/NixOS/nixpkgs/pull/498702 | 02:13:47 |
| 12 Mar 2026 |
 Alister Galpin | Does anyone have an example of the minimum needed to get nix onto a device with uboot was just thinking about it and a lot of the previous pain I had was compiling, so thinking I should just get it booting and able to build/switch | 07:01:31 |
| dramforever | nixos? or just nix | 07:02:14 |
| Alister Galpin | Nixos | 07:02:27 |
| dramforever | also we don't really have prebuilt stuff for riscv so maybe it's not going to be much less pain | 07:02:36 |
| Alister Galpin | True, but might make the iterations faster, last attempt took 2 days of compiling 😅 | 07:03:42 |
| Alister Galpin | And then the image didn't boot ran out of time and now that's 6 months ago at least | 07:04:47 |
| 13 Mar 2026 |
|  eljamm joined the room. | 10:30:39 |
| eljamm | Hello, can someone please help test this program? the executable seems to be working, but I have no clue how to test the module. | 10:34:42 |
 Alex | In reply to @eljamm:matrix.org
Hello, can someone please help test this program? the executable seems to be working, but I have no clue how to test the module. It's not 100% reliable, but you can try in qEMU.
Otherwise, for the most reliable results, you will need to wait a few days for someone to do a full native build of your branch. | 18:05:40 |
| 16 Mar 2026 |
| eljamm | Thanks, I'm looking into doing this right now with this NixOS test. Things are a bit cursed, because I had to enable binfmt for riscv64-linux in my system to get the test to even start, but inside that I'm also supposed to test Linux emulation for RISC-V 😅
Now, the test is just hanging here:
machine: Guest root shell did not produce any data yet...
machine: To debug, enter the VM and run 'systemctl status backdoor.service'.
machine: Guest root shell did not produce any data yet...
machine: To debug, enter the VM and run 'systemctl status backdoor.service'.
| 12:44:05 |
| eljamm | * Thanks, I'm looking into doing this right now with this NixOS test. Things are a bit cursed, because I had to enable binfmt for riscv64-linux in my system to get the test to even start, but inside that I'm also supposed to test Linux emulation for RISC-V 😅
This said, the test is just hanging here:
machine: Guest root shell did not produce any data yet...
machine: To debug, enter the VM and run 'systemctl status backdoor.service'.
machine: Guest root shell did not produce any data yet...
machine: To debug, enter the VM and run 'systemctl status backdoor.service'.
| 12:44:47 |
| dramforever | oh, dear... | 15:53:26 |
| Alex | That's not a good sign.
The backdoor service that the test driver uses to control the VM is not printing the sentinel value that the test driver recognises as "VM is ready". There are a number of ways for this to go wrong, none of which are easy to identify without logs.
It might not be writing system logs to a visible location.
Can you try adding this to the VM's system configuration?
boot.kernelParams = [ "console=ttyS0" "console=hvc0" ];
One of these console options might allow you to see kernel and early userspace logs. | 15:58:21 |
| dramforever | i don't know what other logs are showing up, but if that's all that's shown maybe the vm didn't work at all | 15:59:49 |
| Alex | True, and if necessary one could also try the early console option. | 16:00:56 |
| Alex | Is anyone able to build the latest Nixpkgs natively?
My build is failing during xgcc.
checking for suffix of object files... configure: error: in `/build/build/riscv64-unknown-linux-gnu/libgcc':
configure: error: cannot compute suffix of object files: cannot compile
| 16:01:19 |
| dramforever | maybe powerline10k | 16:03:42 |
| dramforever | * maybe https://github.com/NixOS/nixpkgs/pull/498702 | 16:03:52 |
| Alex | I could try the parent of that merge commit.
The only other option is bisecting, which would probably take days/weeks to find the problematic commit. | 16:04:54 |
| dramforever | huh, i'd have thought that would have fixed it | 16:06:23 |
| Alex | Ah nevermind, my nixpkgs is older, so I should probably try that commit.
dramforever thanks. | 16:06:24 |
| dramforever | do you have more logs | 16:06:25 |
| eljamm | Download felix86-with-params.log | 16:42:54 |
