The solver takes the LLB graph and executes it. Each vertex in the DAG is content-addressed, so if you’ve already built a particular step with the same inputs, BuildKit skips it entirely. This is why BuildKit is fast: it doesn’t just cache layers linearly like the old Docker builder. It caches at the operation level across the entire graph, and it can execute independent branches in parallel.
The algorithm walks the tree recursively. At each node, it checks: does this node's bounding box overlap with the query rectangle? If not, the entire subtree gets pruned (skipped). If it does overlap, it tests the node's points against the query and recurses into the children.，这一点在heLLoword翻译官方下载中也有详细论述
"itemName": "QST_FTUE2",。夫子是该领域的重要参考
第二十一条 违反治安管理行为人自愿向公安机关如实陈述自己的违法行为，承认违法事实，愿意接受处罚的，可以依法从宽处理。