From 749c249dc3b656fea69149f8f7ad3e1a7f935986 Mon Sep 17 00:00:00 2001 From: Doug Moore Date: Mon, 3 Jun 2024 13:31:19 -0500 Subject: [PATCH] subr_pctrie: use ilog2(x) instead of fls(x)-1 In three instances where fls(x)-1 is used, the compiler does not know that x is nonzero and so adds needless zero checks. Using ilog(x) instead saves, in each instance, about 4 instructions, including a conditional, and 16 or so bytes, on an amd64 build. Reviewed by: alc Differential Revision: https://reviews.freebsd.org/D45330 --- sys/kern/subr_pctrie.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/sys/kern/subr_pctrie.c b/sys/kern/subr_pctrie.c index 85df0a9bf9e4..76f4ee17a8ca 100644 --- a/sys/kern/subr_pctrie.c +++ b/sys/kern/subr_pctrie.c @@ -350,7 +350,7 @@ pctrie_insert_node(void *parentp, struct pctrie_node *parent, uint64_t *val) "uint64 too wide"); _Static_assert(sizeof(uint64_t) * NBBY <= (1 << (sizeof(parent->pn_clev) * NBBY)), "pn_clev too narrow"); - parent->pn_clev = rounddown(flsll(index ^ newind) - 1, PCTRIE_WIDTH); + parent->pn_clev = rounddown(ilog2(index ^ newind), PCTRIE_WIDTH); parent->pn_owner = PCTRIE_COUNT; parent->pn_owner = index & -(parent->pn_owner << parent->pn_clev); @@ -546,14 +546,14 @@ pctrie_lookup_le(struct pctrie *ptree, uint64_t index) KASSERT((pred->pn_popmap & ((1 << slot) - 1)) != 0, ("%s: no popmap siblings before slot %d in node %p", __func__, slot, pred)); - slot = fls(pred->pn_popmap & ((1 << slot) - 1)) - 1; + slot = ilog2(pred->pn_popmap & ((1 << slot) - 1)); pred = pctrie_node_load(&pred->pn_child[slot], NULL, PCTRIE_LOCKED); } while (!pctrie_isleaf(pred)) { KASSERT(pred->pn_popmap != 0, ("%s: no popmap children in node %p", __func__, pred)); - slot = fls(pred->pn_popmap) - 1; + slot = ilog2(pred->pn_popmap); pred = pctrie_node_load(&pred->pn_child[slot], NULL, PCTRIE_LOCKED); }