]> www.infradead.org Git - users/jedix/linux-maple.git/commitdiff
maple_tree: calculate new_end when needed
authorWei Yang <richard.weiyang@gmail.com>
Thu, 17 Oct 2024 01:58:08 +0000 (01:58 +0000)
committerAndrew Morton <akpm@linux-foundation.org>
Fri, 1 Nov 2024 04:29:13 +0000 (21:29 -0700)
Patch series "Following cleanup after introduce mas_wr_store_type()", v2.

Patch 1 postpone new_end calculation when needed.
Patch 2 removes a unnecessary sanity check in mas_wr_slot_store().

This patch (of 2):

For wr_exact_fit/wr_new_root, we don't need to calculate new_end.

Let's postpone it until necessary.

Link: https://lkml.kernel.org/r/20241017015809.23392-1-richard.weiyang@gmail.com
Link: https://lkml.kernel.org/r/20241017015809.23392-2-richard.weiyang@gmail.com
Signed-off-by: Wei Yang <richard.weiyang@gmail.com>
Reviewed-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Reviewed-by: Liam R. Howlett <Liam.Howlett@Oracle.com>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
lib/maple_tree.c

index 667120a445709afa64905a69023095ea0bd0ed7d..bc30e99d6cf0ced6f2028655107363be02e7ffc2 100644 (file)
@@ -4211,13 +4211,13 @@ static inline enum store_type mas_wr_store_type(struct ma_wr_state *wr_mas)
        if (!wr_mas->entry)
                mas_wr_extend_null(wr_mas);
 
-       new_end = mas_wr_new_end(wr_mas);
        if ((wr_mas->r_min == mas->index) && (wr_mas->r_max == mas->last))
                return wr_exact_fit;
 
        if (unlikely(!mas->index && mas->last == ULONG_MAX))
                return wr_new_root;
 
+       new_end = mas_wr_new_end(wr_mas);
        /* Potential spanning rebalance collapsing a node */
        if (new_end < mt_min_slots[wr_mas->type]) {
                if (!mte_is_root(mas->node) && !(mas->mas_flags & MA_STATE_BULK))