llama.cpp b11596：合并上下文状态测试并修复 sanitizer 误报
原标题：b11596
AI 摘要
llama.cpp 发布 b11596 版本，主要将原本分散在 test-save-load-state、test-recurrent-state-rollback 和 test-state-restore-fragmented 三个测试中的上下文状态测试合并为统一的 test-llama-context 二进制，以模型为行、用例为列输出判定矩阵，并支持多线程加载模型。同时引入 llama_model_set_attn_rot 取代环境变量 LLAMA_ATTN_ROT_DISABLE，新增 llama_memory_get_buffers，并修复了 macOS 上线程 sanitizer 对 ggml_aligned_malloc 的误报问题。该版本同步提供 macOS、Linux、Android、Windows 等多平台预编译包。
正文节选
<details open> tests : consolidate context state tests into test-llama-context (#30023) * tests : consolidate the context state tests into test-llama-context test-save-load-state, test-recurrent-state-rollback and test-state-restore-fragmented check the same object from three places: the KV cache and the recurrent state of a llama_context, saved, loaded, copied, rolled back. Each had its own main, its own loop over the models, its own copy of the token and the comparison helpers, and a run of