OK, I'll explain again then. We start at the first block on the freelist. If it cannot take the new row we either unlink it from the freelist or walk across it, depending on whether it is more or less full than PCTUSED. The two parameters limit how many times we will do these two things respectively. They are mutually exclusive.

In fact I have already seen that.

Please check the answers about Data Blocks and Freelists.

_walk_insert_threshold defines the number of datablocks to walk in the freelist chain during an insert to find an eligible block in the freelist chain before bumping the high-water mark. Then what does _release_insert_threshold do?