i wonder if #linux #kernel could have a make target for #LSP, i.e. just for generating compile_commands.json for everything, like you had local #bootlin #elixir. in the current form i rather use git grep because it is at least more universal, whereas the json file is tied to the current .config.

While people tend to have complex #vim and #neovim configurations I just have this tiny one, less than 100 lines :-) It has served me for years with minor adjustments…

vim.cmd("filetype plugin indent on")
vim.cmd [[packadd packer.nvim]]

local function plugins(use)
  use 'wbthomason/packer.nvim'
  use 'ap/vim-buftabline'
  use {
    'nvim-lualine/lualine.nvim',
    requires = { 'kyazdani42/nvim-web-devicons', opt = true }
  }
  use 'Mofiqul/dracula.nvim'
  use 'mmarchini/bpftrace.vim'
  use 'vim-scripts/git_patch_tags.vim'
  use {
    'nvim-treesitter/nvim-treesitter',
    run = ':TSUpdate'
  }
  use {
    'VonHeikemen/lsp-zero.nvim',
    branch = 'v2.x',
    requires = {
      {'neovim/nvim-lspconfig'},
      {
        'williamboman/mason.nvim',
        run = function()
          pcall(vim.cmd, 'MasonUpdate')
        end,
      },
      {'williamboman/mason-lspconfig.nvim'},
      {'hrsh7th/nvim-cmp'},
      {'hrsh7th/cmp-nvim-lsp'},
      {'L3MON4D3/LuaSnip'},
    }
  }
end

require('packer').startup(plugins)

vim.g.mapleader = ","
vim.opt.autoindent = true
vim.opt.backspace = "indent,eol,start"
vim.opt.backup = false
vim.opt.clipboard = "unnamedplus"
vim.opt.compatible = false
vim.opt.completeopt = "menuone,noinsert,noselect"
vim.opt.hidden = true
vim.opt.history = 2000
vim.opt.hlsearch = true
vim.opt.incsearch = true
vim.opt.listchars="tab:>-,trail:·"
vim.opt.mouse = ""
vim.opt.path:append("**")
vim.opt.ruler = true
vim.opt.shortmess:append("c")
vim.opt.showmatch = true
vim.opt.swapfile = false
vim.opt.tags = "./tags;/"
vim.opt.termguicolors = true
vim.opt.wildmenu = true
vim.opt.wrap = false

if vim.g.neovide then
  vim.o.guifont = "FiraCode Nerd Font Mono:h12"
end

vim.api.nvim_create_autocmd(
  { "FocusGained", "BufEnter" },
  { pattern = "*", command = [[:checktime]],
    group = vim.api.nvim_create_augroup("BufCheck", { clear = true }) }
)

local function nmap(shortcut, command)
  local opts = { noremap = true, silent = true }
  vim.api.nvim_set_keymap('n', shortcut, command, opts)
end

nmap("<silent>", "<C-l> :nohl<CR><C-l>")
nmap("<leader>cd", ":lcd %:p:h<CR>:pwd<CR>")

vim.cmd[[colorscheme dracula]]
require('lualine').setup {
  options = {
    theme = 'dracula-nvim'
  }
}

in-kernel #keystone #driver in-progress (actually it is functioning but I need to separate from research project codebase) https://github.com/keystone-enclave/keystone/pull/328 #linux #kernel #riscv

My first impressions of RISC-V CoVE: I don't see why it couldn't made to leverage #keystone for attestation.

It already has a community and legit use cases, which cannot be satisfied by CoVE. Why create this divergence and not to engineer things work together, when you have a chance? #riscv #linux #kernel #cove

#notetomyself: download and extract #riscv #toolchain:

echo elf glibc musl | xargs -n1 | xargs -I{} echo https://github.com/riscv-collab/riscv-gnu-toolchain/releases/download/2022.08.25/riscv64-{}-ubuntu-20.04-nightly-2022.08.25-nightly.tar.gz | xargs curl -sSL | tar zx -i

thinking of #chatgpt, i'm more interested what people can do with it, which previously was not possible, rather than what it can do previously performed by human labour.

doing factors faster previous science is not overwhelmingly interesting in the end.

this is sad and true unfortunately. https://lunduke.substack.com/p/microsofts-growing-control-of-linux #microsoft #linux #opensource #proprietary

Fetching a tag from #ubuntu #linux #kernel tree:

function git-fetch-linux-ubuntu-tag {
  local tag=refs/tags/cod/mainline/$1
  git fetch --no-tags \
    git://git.launchpad.net/~ubuntu-kernel-test/ubuntu/+source/linux/+git/mainline-crack \
    $tag:$tag
}

Usage example:

$ git checkout -b tmp
Switched to a new branch 'tmp'

$ git-fetch-linux-ubuntu-tag v6.3
From git://git.launchpad.net/~ubuntu-kernel-test/ubuntu/+source/linux/+git/mainline-crack
 * [new tag]                   cod/mainline/v6.3 -> cod/mainline/v6.3

$ git merge v6.3

Building:

LANG=C fakeroot debian/rules clean
LANG=C fakeroot debian/rules binary-headers binary-generic binary-perarch

thank you #mutt 1999-2023, i've now fully moved into #aerch as my main #email client. https://aerc-mail.org/

I like virt manager otherwise, except for its crappy looking icon :-) #kvm #qemu #libvirt

#remindertomyself: fetch a #git tag for the given #ubuntu #mainline #linux #kernel #snapshot:

$ git fetch --no-tags git://git.launchpad.net/~ubuntu-kernel-test/ubuntu/+source/linux/+git/mainline-crack refs/tags/cod/mainline/v6.3-rc7:refs/tags/cod/mainline/v6.3-rc7
From git://git.launchpad.net/~ubuntu-kernel-test/ubuntu/+source/linux/+git/mainline-crack
 * [new tag]                   cod/mainline/v6.3-rc7 -> cod/mainline/v6.3-rc7

suppilovahvero in linux-tpmdd on  next took 19s
$ git tag
cod/mainline/v6.3-rc7

Compilation:

LANG=C fakeroot debian/rules clean
LANG=C fakeroot debian/rules binary-headers binary-generic binary-perarch

#libreoffice #draw is the only reason why i install it in the first place anywhere. it provides equivalent user experience and robustness to all #subscription based #diagram tools, without the subscription fees.

the only thing i would improve in it would be t separate it from the rest of the suite because i don't have any use for the other libreoffice applications.

trying out #aerc #email client

I have now kind of current state of desktop in both #x86 and #ARM ends :-)

1. i9-13900k / 64 GB / Intel ARC A-770
2. Mac Mini M2 Pro 12 12/19 core version / 32 GB

The 2nd is both the computer I do music with (i.e. my studio machine) and also #ARM test farm for #TPM. I compile ARM kernels with it and run them in VMWare Fusion, which provides nicely packaged #vTPM’s.

i find paying for #sublimetext and using it with #neovintageous much better option than #vscode, if I need a GUI editor e.g. in meetings. It is fast and uncluttered. If you write code normally in the #terminal with #neovim, vscode can feel a bit sticky, and also stressful because it has too much shit going on in the screen estate. Milliseconds do affect to the overall feeling just like they would when e.g. when playing with a MIDI keyboard.

And generally if it is not fully open source, I feel less worried to use an application where I pay a sum of money for the license. Because the profit is often acquired by some means anyway, it is the most transparent for me…

#remindertomyself

$ ps -T -o comm,policy,rtprio -p $(pgrep -w -d ',' irq) | egrep '(snd|hci)'
irq/148-xhci_hc FF      85
irq/147-ahci[00 FF      50
irq/203-snd_hda FF      90
irq/204-snd_hda FF      90

#renewed my accidentally #expired
@lwnnet #subscription. unintentionally i had used single time payment last year (before that i was using corporate subscription while i was still working at intel). now it is set to automatic payment like it should be :-)

i have now kind of "#intel of the present day" #desktop configurtion in my main #development machine: i9-13900k with 64GB of DDR and Intel Arc A770 GPU.

ive been waiting the day to get rid of shitty #nvidia #hardware, causing all the time compatibility issues...

on #macos (which I have in my #university laptop, thus learning the ways) the easiest way to do #linux #kernel development and testing is podman (brew install podman). each container will have a #QEMU instance backing it up.

that sorted out, using macos does not really make any difference, as full-screen #tmux is pretty much the same always :-) any desktop that can launch a web browser is good enough desktop for me really...

this is how i deal with #password's. i protect the critical passwords with #PGP public key. Any password where #leakage is not an existential question is offloaded to #lastpass. The keyring itself is protected by #yubikey. i feel overally pretty safe :-)