vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-05-23 21:13:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
koboldcpp/src
History
Daniel Elliott eeeaf6180b
llama-graph: fix null-buffer crash in llm_graph_input_attn_kv_iswa for SWA-only models (#23131) 
When a model has zero non-SWA attention layers (e.g. a SWA-only slice of Gemma 4),
the base KV cache has no layer tensors. The input tensors (self_k_idxs, self_v_idxs,
self_kq_mask) are created as graph input nodes but never consumed by any compute node,
so the backend scheduler never allocates a buffer for them. Calling
mctx->get_base()->set_input_k_idxs() on an unallocated tensor then hits
GGML_ASSERT(buffer) at ggml-backend.cpp:194.

The same scenario applies symmetrically: if a model had zero SWA layers, the SWA
tensors would be unallocated.

Fix: guard both the base and SWA set_input calls with null/buffer checks, matching
the pattern already used by llm_graph_input_mem_hybrid_iswa::set_input (line ~674)
which has the comment: 'base tensors may not be allocated if there are no non-SWA
attention layers'.

Also fix can_reuse() in the same class to skip the ne[0] and kq_mask checks for
unallocated tensors, preventing a null-dereference on the reuse path.
2026-05-21 09:20:51 +03:00
..
models metal : optimize pad + cpy (#23354) 2026-05-20 09:42:00 +03:00
CMakeLists.txt cmake: use glob to collect src/models sources (#22005) 2026-04-16 23:25:16 +02:00
llama-adapter.cpp fix: correct misspellings in code comments (#21217) 2026-03-31 13:50:51 +02:00
llama-adapter.h llama : re-enable manual LoRA adapter free (#19983) 2026-03-18 12:03:26 +02:00
llama-arch.cpp llama + spec: MTP Support (#22673) 2026-05-16 20:06:23 +08:00
llama-arch.h llama + spec: MTP Support (#22673) 2026-05-16 20:06:23 +08:00
llama-batch.cpp kv-cache : fix M-RoPE checkpoints (#20132) 2026-03-06 08:46:51 +02:00
llama-batch.h fix: correct misspellings in code comments (#21217) 2026-03-31 13:50:51 +02:00
llama-chat.cpp mtmd, model : merge HunyuanOCR into HunyuanVL and fix OCR vision precision (#23329) 2026-05-21 00:35:37 +02:00
llama-chat.h mtmd, model : merge HunyuanOCR into HunyuanVL and fix OCR vision precision (#23329) 2026-05-21 00:35:37 +02:00
llama-context.cpp Move to backend sampling for MTP draft path (#23287) 2026-05-20 22:34:45 +05:30
llama-context.h llama: avoid copying logits during prompt decode in MTP (#23198) 2026-05-17 23:30:25 +08:00
llama-cparams.cpp cparams : rename LLAMA_MAX_PARALLEL_SEQUENCES to LLAMA_MAX_SEQ (#14188) 2025-06-15 10:08:58 +03:00
llama-cparams.h llama: avoid copying logits during prompt decode in MTP (#23198) 2026-05-17 23:30:25 +08:00
llama-ext.h llama: avoid copying logits during prompt decode in MTP (#23198) 2026-05-17 23:30:25 +08:00
llama-grammar.cpp common/grammar: fix grammar parsing issues to prevent stack overflow and hangs (#18604) 2026-03-21 18:43:35 +01:00
llama-grammar.h common/grammar : replace problematic backtracking regex [\s\S]* (#18342) 2026-01-03 16:02:43 -06:00
llama-graph.cpp llama-graph: fix null-buffer crash in llm_graph_input_attn_kv_iswa for SWA-only models (#23131) 2026-05-21 09:20:51 +03:00
llama-graph.h llama : MTP clean-up (#23269) 2026-05-19 15:32:58 +03:00
llama-hparams.cpp llama + spec: MTP Support (#22673) 2026-05-16 20:06:23 +08:00
llama-hparams.h llama + spec: MTP Support (#22673) 2026-05-16 20:06:23 +08:00
llama-impl.cpp llama : correct platform-independent loading of BOOL metadata (#21428) 2026-04-06 01:40:38 +02:00
llama-impl.h llama : enable chunked fused GDN path (#20340) 2026-03-11 22:46:40 +02:00
llama-io.cpp server : avoid checkpoint data host copies (#22558) 2026-05-02 18:03:25 +03:00
llama-io.h llama : add option to save memory in device buffers (#22679) 2026-05-05 06:35:07 +03:00
llama-kv-cache-iswa.cpp (revert) kv-cache : do not quantize SWA KV cache (#21332) 2026-04-03 09:07:01 +03:00
llama-kv-cache-iswa.h llama: print memory breakdown on exit (#15860) 2025-09-24 16:53:48 +02:00
llama-kv-cache.cpp ggml : implement fast walsh-hadamard transform for kv rotation (#21352) (#22631) 2026-05-05 10:05:05 +08:00
llama-kv-cache.h kv-cache : support attention rotation for heterogeneous iSWA (#21513) 2026-04-07 20:31:28 +03:00
llama-kv-cells.h llama: store mrope data in KV cell (#16825) 2025-10-29 18:09:18 +01:00
llama-memory-hybrid-iswa.cpp llama : MTP clean-up (#23269) 2026-05-19 15:32:58 +03:00
llama-memory-hybrid-iswa.h llama + spec: MTP Support (#22673) 2026-05-16 20:06:23 +08:00
llama-memory-hybrid.cpp llama : MTP clean-up (#23269) 2026-05-19 15:32:58 +03:00
llama-memory-hybrid.h llama + spec: MTP Support (#22673) 2026-05-16 20:06:23 +08:00
llama-memory-recurrent.cpp llama : MTP clean-up (#23269) 2026-05-19 15:32:58 +03:00
llama-memory-recurrent.h llama : MTP clean-up (#23269) 2026-05-19 15:32:58 +03:00
llama-memory.cpp memory : correctly handle failure in apply() (#14438) 2025-06-30 18:03:03 +03:00
llama-memory.h llama + spec: MTP Support (#22673) 2026-05-16 20:06:23 +08:00
llama-mmap.cpp Update llama-mmap to use ftello/fseeko (#22497) 2026-04-30 14:17:52 -07:00
llama-mmap.h llama: fix llama-model-saver (#20503) 2026-03-25 12:53:16 +02:00
llama-model-loader.cpp llama + spec: MTP Support (#22673) 2026-05-16 20:06:23 +08:00
llama-model-loader.h llama + spec: MTP Support (#22673) 2026-05-16 20:06:23 +08:00
llama-model-saver.cpp model : NvFP4 quantized LM head support (#23046) 2026-05-16 11:09:27 +02:00
llama-model-saver.h llama: fix llama-model-saver (#20503) 2026-03-25 12:53:16 +02:00
llama-model.cpp llama + spec: MTP Support (#22673) 2026-05-16 20:06:23 +08:00
llama-model.h model : NvFP4 quantized LM head support (#23046) 2026-05-16 11:09:27 +02:00
llama-quant.cpp model: move load_hparams and load_tensors to per-model definition (#22004) 2026-05-04 12:36:59 +02:00
llama-quant.h llama : refactor src/llama.cpp (#10902) 2025-01-03 10:18:53 +02:00
llama-sampler.cpp llama : rename llama-sampling to llama-sampler (#19363) 2026-02-06 07:26:54 +01:00
llama-sampler.h llama : rename llama-sampling to llama-sampler (#19363) 2026-02-06 07:26:54 +01:00
llama-vocab.cpp model : add sarvam_moe architecture support (#20275) 2026-05-09 16:31:50 +02:00
llama-vocab.h model : add sarvam_moe architecture support (#20275) 2026-05-09 16:31:50 +02:00
llama.cpp llama : add missing call to ggml_backend_load_all() (#22752) 2026-05-07 08:24:47 +03:00
unicode-data.cpp server : better security control for public deployments (#9776) 2024-10-08 13:27:04 +02:00
unicode-data.h llama : reduce compile time and binary size (#9712) 2024-10-02 15:49:55 +02:00
unicode.cpp unicode,test: add Qwen3.5 non-backtracking tokenizer handler and regr… (#22110) 2026-05-14 11:03:40 +02:00
unicode.h vocab: fix Gemma4 tokenizer (#21343) 2026-04-03 10:33:03 +02:00