titaniumtown/nixos
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Compare commits

base: titaniumtown:fdeda20373f6f497beccb7f5b6d27dc7c34ed017
Branches Tags
titaniumtown:main
..
compare: titaniumtown:90f2c27c2c6099a37293d3224600cd13f03d17a4
Branches Tags
titaniumtown:main

3 Commits
fdeda20373 ... 90f2c27c2c

Author SHA1 Message Date
Simon Gardling
90f2c27c2c DISABLE KMSCON
Some checks failed
Build and Deploy / mreow (push) Successful in 7m39s
Details
Build and Deploy / yarn (push) Successful in 1m5s
Details
Build and Deploy / muffin (push) Failing after 36s
Details 
THIS is what caused issues with greetd, nothing kernel related
 2026-04-25 19:20:24 -04:00
Simon Gardling
450b77140b pi: apply omp patches via prePatch (bun2nix.hook overrides patchPhase) 
`bun2nix.hook` (used by upstream omp's package.nix) sets

  patchPhase = bunPatchPhase

at the end of its setup-hook unless `dontUseBunPatch` is already set.
`bunPatchPhase` only runs `patchShebangs` plus a HOME mktemp; it never
iterates over `$patches`. The standard nixpkgs `patches` attribute
therefore went into the derivation env but was silently ignored at
build time, leaving the deployed omp binary unpatched.

Switch to applying the two patches via `prePatch` (which `bunPatchPhase`
does call). Verified with strings(1) over the rebuilt binary that both
patch hunks land:

  /wrong_api_format|...|invalid tool parameters/  (patch 0001)
  stubsReasoningContent ... thinkingFormat == "openrouter"  (patch 0002)
 2026-04-25 19:20:08 -04:00
Simon Gardling
318373c09c pi: patch omp to require reasoning_content for OpenRouter reasoning models 
DeepSeek V4 Pro (and similar reasoning models reached via OpenRouter) reject
multi-turn requests in thinking mode with:

  400 The `reasoning_content` in the thinking mode must be passed back
  to the API.

omp's existing kimi placeholder injection (`requiresReasoningContentForToolCalls`)
covered this requirement only for `thinkingFormat == "openai"`. OpenRouter
sets `thinkingFormat == "openrouter"`, so the gate never fired even though
the underlying providers behind OpenRouter (DeepSeek, Kimi, etc.) all enforce
the same invariant.

This patch:

1. Extends `requiresReasoningContentForToolCalls` detection: any
   reasoning-capable model fronted by OpenRouter now sets the flag.
2. Extends the placeholder gate in `convertMessages` to accept
   `thinkingFormat == "openrouter"` alongside `"openai"`.

Cross-provider continuations are the dominant trigger: a conversation
warmed up by Anthropic Claude (whose reasoning is redacted/encrypted on
the wire) followed by a switch to DeepSeek V4 Pro via OpenRouter. omp
cannot synthesize plaintext `reasoning_content` from Anthropic's
encrypted blocks, so the placeholder satisfies DeepSeek's validator
without fabricating a reasoning trace. Real captured reasoning, when
present, short-circuits the placeholder via `hasReasoningField` and
survives intact.

Side benefit: also closes a latent gap where Kimi-via-OpenRouter
(`thinkingFormat == "openrouter"`) had the compat flag set but the
placeholder gate silently rejected it.

Applies cleanly on top of patch 0001.
 2026-04-25 19:20:05 -04:00
2 changed files with 18 additions and 2 deletions
Expand all files Collapse all files

2
modules/common.nix
View File

@@ -58,8 +58,6 @@
];
};
services.kmscon.enable = true;
environment.systemPackages = with pkgs; [
doas-sudo-shim
];

18
modules/desktop-common.nix
View File

@@ -166,12 +166,30 @@
DRM_NOUVEAU = lib.mkForce no;
# other gpus not present
DRM_RADEON = lib.mkForce no;
DRM_GMA500 = lib.mkForce no;
DRM_AST = lib.mkForce no;
DRM_MGAG200 = lib.mkForce no;
DRM_HISI_HIBMC = lib.mkForce no;
DRM_APPLETBDRM = lib.mkForce no;
# legacy AMD IP blocks. hosts are Navi 32 RDNA3 dGPU (7800 XT, yarn,
# 2023, gfx1101, DCN 3.2) and Krackan Point RDNA 3.5 iGPU (mreow,
# 2024, gfx1150, DCN 3.5). everything below pre-dates those by a
# decade. upstream only exposes per-generation toggles for SI and
# CIK — no switch for VI/Polaris/Vega/Navi1x, those stay in amdgpu.
DRM_AMDGPU_SI = lib.mkForce no; # Southern Islands / GCN 1 (2012): HD 7950/7970, R9 280/280X, R7 260X
DRM_AMDGPU_CIK = lib.mkForce no; # Sea Islands / GCN 2 (2013): R9 290/290X/390, Kaveri APUs (A10-7850K), Steam Machine Bonaire
DRM_AMD_SECURE_DISPLAY = lib.mkForce no; # HDCP region-CRC debugfs helper, needs custom DMCU firmware
# early-boot framebuffer chain: drop every alternative to amdgpu so
# the console never transitions simpledrm -> dummy -> amdgpu (visible
# as a flash + scrolled dmesg). amdgpu owns the display from initrd
# onward; pre-amdgpu kernel output stays in the printk ring buffer.
DRM_SIMPLEDRM = lib.mkForce no;
FB_EFI = lib.mkForce no;
FB_VESA = lib.mkForce no;
# intel cpu / platform
INTEL_IOMMU = lib.mkForce no;
INTEL_IDLE = lib.mkForce no;