feat: sync llama.cpp to b10951 - #394
Merged
Merged
Conversation
…atch for b10951
llama.cpp b10951 moved the specialized chat template parsers out of
common/chat.cpp into common/parsers/ (one file each) and added
common/json-schema.{h,cpp}, so the chat.cpp patch that added the
FunctionGemma parser no longer applied and the tree did not compile.
- sync-vendor.sh: export common/parsers and common/json-schema.*; prune
parsers/sources.cmake like the other upstream build files
- rnllama-sources.cmake: glob common/parsers/*.cpp (the podspec already
takes common/**)
- FunctionGemma parser becomes common/parsers/function-gemma.cpp with
its declaration in parsers/parsers.h, following the upstream split, and
uses common_chat_tool_parameters() since the common_schema refactor
dropped common_grammar_builder::resolve_refs
- chat.cpp.patch keeps only the detection hook, content-part extra
fields and template caps/variant helpers
- sync-vendor.sh: record the resolved commit before applying patches and
drop .orig/.rej leftovers even when a patch fails, so a partial sync
committed for fixing has a correct pin and no junk files
- drop the .orig/.rej files the failed CI sync committed and fix the
stale LLAMA_CPP_COMMIT pin
jhen0409
force-pushed
the
auto/sync-llama.cpp
branch
from
September 14, 2026 06:32
831ae52 to
0b1cbb7
Compare
…peg-parser.h includes it)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
🤖 Automated llama.cpp sync
This PR was automatically created/updated by the daily sync workflow.
Changes:
b10829tob10951Verification:
auto/sync-llama.cppdirectly: while this PR stays open, later sync runs build on top of the branch instead of resetting it.📋 llama.cpp changes (b10829 → b10951)
093a2f8common : move llama_n_rs_seq to before llama_decode (#28749)ad6c668ggml-cuda: fallback to F32 on device without BF16 hardware acceleration (#28846)7a16a6cgrammar : coalesce find + insert into a single insert and adjust move/copy mechanics (#26885)5f436ddtests : exclude HY_V4 from WebGPU test-llama-archs tests (#28855)e49d2c2models : guard the expert FFN size fallback in nemotron-h against a zero divisor (#28779)6978052ggml-cpu(s390x): guard VXE-only repack helpers (#28775)243a308tests : fix typo in test-quant-type-selection for nemotron 3 nano (#28835)b6b003dsycl : Fix get mem error (#28227)c95f8e4ci : run editorconfig and code-style checks on ubuntu-slim (#28854)bc52a12pi : prefer PI_MODEL_NAME env var for model disclosure (#28853)4a89937tests : reduce FA test sizes (#28842)37b3a9eci : remove leftover command (#28839)002a12aci : cap test-backend-ops parallel jobs at 2 and add a 3600s timeout (#28833)f1e44dcvulkan: workaround NV queuesubmit driver bug (#28830)56b9eb2opencl: apply the noshuffle row-alignment rule to q4_K, q5_K and q8_0, not just q6_K (#28575)790cf51chat : improve parsing of complex types in qwen3-coder (#28742)8e33095common: add LOG_JSON macro to log structured data (#28586)acecd56common : implement common_schema internal representation for JSON schemas (#28736)ae9afffjinja : support dot property integer literals (#28817)737e098cmake: leave the timestamp out of precompiled headers on clang (#28816)3057bb6ui : add cache (#28802)56381e4server : allow model downloads at model limit fix issue #26809 (#28530)c8edcebggml-cuda: hip add specific config table for AMD GCN (#27841)e192abbserver : add missing headers (#28795)718f7b4vendor : update cpp-httplib to 0.56.0 (#28787)2a3005csyscl : Handle (fail gracefully) unsupported tq1_0 quants (#28681)f3a33dfrpc : fix linking when compiling with BUILD_SHARED_LIBS=OFF (#28492)c069aa7server: frame the router child state command as a whole line (#28747)8a56aedopencl: fix several bugs where the backend aborts (#27630)07fc977opencl: add bin kernelkernel_gemm_noshuffle_q4_k_f32_32b_trans_ila_a8_bin(#28677)3f5e94dwebgpu: align tensor bindings to the type block size (#28382)eafe15ahexagon: support for multi-device model split (aka row-split) (#28589)d3146f2ggml-webgpu: Update to a recent version of Dawn (#28683)82d6bb2server: refactor subproc handling (#28555)8ea2902cmake : skip PCH for llama-server PCH when using MSVC (#28763)b78a39aci : run test-backend-ops as a dedicated ci/run.sh test (#28740)982937atests: extend test-quantize-fns to test nrc=2 (i8mm) kernels (#16234)8172e65tests: tolerate a shared pool abort in test_completion_unified (#28759)43f3ddaggml: skip 0-sized ids tensor when offloading selected experts (#28739)5bda51bmetal : skip the empty half of the mul_mm_id token tile (#28301)3bcfeb7cmake : add PCH and unity build to improve build times (#28091)1dfe94ecommon : fix typo in speculative.cpp comment [no ci] (#28750)a2878d3metal : single-source fusion table + fusion debug rework (#28164)aac8102metal : fix idle threads in the remaining iq mul_mv kernels for ne00 < 1024 (#28692)5cdd3d1model : fix MTP context kv cache allocation for deepseek2, glm4moe, … (#28630)b0dcb81server: fix speculation after an image (#28715)16378d9CUDA/HIP: Flash Attention tuning (gfx1201) (#28102)451b89bci : key cache to sanitizer matrix (#28708)481c65fvulkan: fix data race and OOB access in argsort(large) (#28705)df03399opencl: add A8 Q4_0 mm binary kernel support (#28268)Please review and merge if all checks pass.