* Lai Jiangshan ([email protected]) wrote:
> reduce memory fragment
> add a slowpath overhead
>
> Signed-off-by: Lai Jiangshan <[email protected]>
Hrm, I see what you want to do here: given we already have
synchronize_rcu calls in the fini code, you use them to guarantee
quiescence of the previous iteration instead of using call_rcu. I would
argue that at this point in the development process, I would rather
prefer to keep allocation/delayed free as decoupled as possible from the
algorithm. So I'll keep this optimisation in mind, but for now I think I
will keep the call_rcu scheme to deal with all memory free. This makes
it more easy to modify the code when the allocation/delayed free
concerns are all encapsulated through use of call_rcu.
Moreover, this patch really is an optimisation, so it should be backed
by benchmarks that shows if there are measurable performance
improvements.
Thanks!
Mathieu
> ---
> rculfhash.c | 22 ++++++++++------------
> 1 files changed, 10 insertions(+), 12 deletions(-)
>
> diff --git a/rculfhash.c b/rculfhash.c
> index e379c71..49f3637 100644
> --- a/rculfhash.c
> +++ b/rculfhash.c
> @@ -209,7 +209,6 @@ struct ht_items_count {
> } __attribute__((aligned(CAA_CACHE_LINE_SIZE)));
>
> struct rcu_level {
> - struct rcu_head head;
> struct _cds_lfht_node nodes[0];
> };
>
> @@ -712,14 +711,6 @@ unsigned long _uatomic_max(unsigned long *ptr, unsigned
> long v)
> return v;
> }
>
> -static
> -void cds_lfht_free_level(struct rcu_head *head)
> -{
> - struct rcu_level *l =
> - caa_container_of(head, struct rcu_level, head);
> - poison_free(l);
> -}
> -
> /*
> * Remove all logically deleted nodes from a bucket up to a certain node key.
> */
> @@ -1131,8 +1122,7 @@ void init_table(struct cds_lfht *ht,
> if (CMM_LOAD_SHARED(ht->t.resize_target) < (!i ? 1 : (1UL <<
> i)))
> break;
>
> - ht->t.tbl[i] = calloc(1, sizeof(struct rcu_level)
> - + (len * sizeof(struct _cds_lfht_node)));
> + ht->t.tbl[i] = calloc(1, len * sizeof(struct _cds_lfht_node));
> assert(ht->t.tbl[i]);
>
> /*
> @@ -1218,6 +1208,7 @@ void fini_table(struct cds_lfht *ht,
> unsigned long first_order, unsigned long len_order)
> {
> long i, end_order;
> + void *free_by_rcu = NULL;
>
> dbg_printf("fini table: first_order %lu end_order %lu\n",
> first_order, first_order + len_order);
> @@ -1243,6 +1234,8 @@ void fini_table(struct cds_lfht *ht,
> * return a logically removed node as insert position.
> */
> ht->cds_lfht_synchronize_rcu();
> + if (free_by_rcu)
> + free(free_by_rcu);
>
> /*
> * Set "removed" flag in dummy nodes about to be removed.
> @@ -1252,12 +1245,17 @@ void fini_table(struct cds_lfht *ht,
> */
> remove_table(ht, i, len);
>
> - ht->cds_lfht_call_rcu(&ht->t.tbl[i]->head, cds_lfht_free_level);
> + free_by_rcu = ht->t.tbl[i];
>
> dbg_printf("fini new size: %lu\n", 1UL << i);
> if (CMM_LOAD_SHARED(ht->in_progress_destroy))
> break;
> }
> +
> + if (free_by_rcu) {
> + ht->cds_lfht_synchronize_rcu();
> + free(free_by_rcu);
> + }
> }
>
> struct cds_lfht *_cds_lfht_new(cds_lfht_hash_fct hash_fct,
> --
> 1.7.4.4
>
--
Mathieu Desnoyers
Operating System Efficiency R&D Consultant
EfficiOS Inc.
http://www.efficios.com
_______________________________________________
ltt-dev mailing list
[email protected]
http://lists.casi.polymtl.ca/cgi-bin/mailman/listinfo/ltt-dev
