Is it expected that the Ctrl+C stop event will not be responded to during the reasoning process?

[xxnuo]

python3 -m cyllama.sd txt2img       --diffusion-model ./z_image_turbo-Q8_0.gguf       --vae ./vae.sft       --llm ./Qwen3-4B-Instruct-2507-Q4_K_M.gguf       --cfg-scale 1.0 -v       --diffusion-fa       -H 1024 -W 512       -p "a lovely plump cat"
Loading model: ./z_image_turbo-Q8_0.gguf
[DEBUG] stable-diffusion.cpp:189  - Using CPU backend
[INFO] stable-diffusion.cpp:234  - loading diffusion model from './z_image_turbo-Q8_0.gguf'
[INFO] model.cpp:378  - load ./z_image_turbo-Q8_0.gguf using gguf format
[DEBUG] model.cpp:420  - init from './z_image_turbo-Q8_0.gguf'
[INFO] stable-diffusion.cpp:281  - loading llm from './Qwen3-4B-Instruct-2507-Q4_K_M.gguf'
[INFO] model.cpp:378  - load ./Qwen3-4B-Instruct-2507-Q4_K_M.gguf using gguf format
[DEBUG] model.cpp:420  - init from './Qwen3-4B-Instruct-2507-Q4_K_M.gguf'
[INFO] stable-diffusion.cpp:295  - loading vae from './vae.sft'
[INFO] model.cpp:381  - load ./vae.sft using safetensors format
[DEBUG] model.cpp:511  - init from './vae.sft', prefix = 'vae.'
[INFO] stable-diffusion.cpp:318  - Version: Z-Image 
[INFO] stable-diffusion.cpp:346  - Weight type stat:                      f32: 640  |    q8_0: 202  |    q4_K: 216  |    q6_K: 37   
[INFO] stable-diffusion.cpp:347  - Conditioner weight type stat:          f32: 145  |    q4_K: 216  |    q6_K: 37   
[INFO] stable-diffusion.cpp:348  - Diffusion model weight type stat:      f32: 251  |    q8_0: 202  
[INFO] stable-diffusion.cpp:349  - VAE weight type stat:                  f32: 244  
[DEBUG] stable-diffusion.cpp:351  - ggml tensor size = 400 bytes
[DEBUG] llm.hpp:285  - merges size 151387
[DEBUG] llm.hpp:317  - vocab size: 151665
[INFO] stable-diffusion.cpp:535  - Using flash attention in the diffusion model
[DEBUG] ggml_extend.hpp:1873 - qwen3 params backend buffer size =  7672.62 MB(RAM) (398 tensors)
[DEBUG] ggml_extend.hpp:1873 - z_image params backend buffer size =  11745.36 MB(RAM) (453 tensors)
[DEBUG] ggml_extend.hpp:1873 - vae params backend buffer size =  94.57 MB(RAM) (138 tensors)
[DEBUG] stable-diffusion.cpp:683  - loading weights
[DEBUG] model.cpp:1359 - using 6 threads for model loading
[DEBUG] model.cpp:1381 - loading tensors from ./z_image_turbo-Q8_0.gguf
  |====================>                             | 453/1095 - 62.53it/s
[DEBUG] model.cpp:1381 - loading tensors from ./Qwen3-4B-Instruct-2507-Q4_K_M.gguf
  |======================================>           | 851/1095 - 65.08it/s
[DEBUG] model.cpp:1381 - loading tensors from ./vae.sft
  |==================================================| 1095/1095 - 82.47it/s
[INFO] model.cpp:1584 - loading tensors completed, taking 13.28s (process: 0.00s, read: 1.29s, memcpy: 0.00s, convert: 11.58s, copy_to_backend: 0.00s)
[INFO] stable-diffusion.cpp:767  - total params memory size = 19512.56MB (VRAM 0.00MB, RAM 19512.56MB): text_encoders 7672.62MB(RAM), diffusion_model 11745.36MB(RAM), vae 94.57MB(RAM), controlnet 0.00MB(VRAM), pmid 0.00MB(RAM)
[INFO] stable-diffusion.cpp:883  - running in FLOW mode
[DEBUG] stable-diffusion.cpp:908  - finished loaded file
Model loaded in 13.74s
Generating 1 image(s)...
  Prompt: a lovely plump cat
  Size: 512x1024
  Steps: 20, CFG: 1.0
[DEBUG] stable-diffusion.cpp:3138 - generate_image 512x1024
[INFO] stable-diffusion.cpp:3169 - sampling using Euler method
[INFO] denoiser.hpp:364  - get_sigmas with discrete scheduler
[INFO] stable-diffusion.cpp:3282 - TXT2IMG
[INFO] stable-diffusion.cpp:1167 - apply at runtime
[DEBUG] conditioner.hpp:1701 - parse '<|im_start|>user
a lovely plump cat<|im_end|>
<|im_start|>assistant
' to [['<|im_start|>user
', 1], ['a lovely plump cat', 1], ['<|im_end|>
<|im_start|>assistant
', 1], ]
[DEBUG] llm.hpp:259  - split prompt "<|im_start|>user
" to tokens ["<|im_start|>", "user", "Ċ", ]
[DEBUG] llm.hpp:259  - split prompt "a lovely plump cat" to tokens ["a", "Ġlovely", "Ġpl", "ump", "Ġcat", ]
[DEBUG] llm.hpp:259  - split prompt "<|im_end|>
<|im_start|>assistant
" to tokens ["<|im_end|>", "Ċ", "<|im_start|>", "assistant", "Ċ", ]
[DEBUG] ggml_extend.hpp:1688 - qwen3 compute buffer size: 1.42 MB(RAM)
[DEBUG] conditioner.hpp:1896 - computing condition graph completed, taking 512 ms
[INFO] stable-diffusion.cpp:2917 - get_learned_condition completed, taking 516 ms
[INFO] stable-diffusion.cpp:3028 - generating image: 1/1 - seed 1548866554
[DEBUG] ggml_extend.hpp:1688 - z_image compute buffer size: 276.85 MB(RAM)
  |==>                                               | 1/20 - 190.45s/it^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C^C

[shakfu]

Thanks for reporting this @xxnuo.

Short answer: This is a limitation of the stable-diffusion.cpp API, not something cyllama can fix directly.

Explanation: When Python calls into native C++ code (like generate_image()), Python's signal handlers are deferred until control returns to Python. The stable-diffusion.cpp library doesn't expose an abort/cancellation mechanism in its public API.

The underlying ggml library has an abort callback mechanism (ggml_abort_callback), but stable-diffusion.cpp doesn't expose it. The progress callback (sd_progress_cb_t) is defined as:

typedef void (*sd_progress_cb_t)(int step, int steps, float time, void* data);

Since it returns void, there's no way to signal "please stop" back to the library.

Potential fix: This would need to be addressed upstream in stable-diffusion.cpp. Options include:

1. Change the progress callback signature to return bool (false = abort)
2. Expose ggml's abort callback via sd_set_abort_callback()
3. Add an explicit sd_cancel() function

I'll open an issue upstream requesting this feature.

Workaround for now: Unfortunately there isn't a clean workaround. You can:

• Use shorter --steps values during testing
• Kill the process with kill -9from another terminal
• Set a timeout externally

[shakfu]

The issue is posted in the stable-diffusion.cpp issue tracker.