block: add support for IO CPU affinity

This patch adds support for controlling the IO completion CPU of
either all requests on a queue, or on a per-request basis. We export
a sysfs variable (rq_affinity) which, if set, migrates completions
of requests to the CPU that originally submitted it. A bio helper
(bio_set_completion_cpu()) is also added, so that queuers can ask
for completion on that specific CPU.

In testing, this has been show to cut the system time by as much
as 20-40% on synthetic workloads where CPU affinity is desired.

This requires a little help from the architecture, so it'll only
work as designed for archs that are using the new generic smp
helper infrastructure.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
This commit is contained in:
Jens Axboe
2008-09-13 20:26:01 +02:00
parent 18887ad910
commit c7c22e4d5c
9 changed files with 182 additions and 60 deletions

View File

@@ -81,6 +81,8 @@ struct bio {
unsigned int bi_max_vecs; /* max bvl_vecs we can hold */
unsigned int bi_comp_cpu; /* completion CPU */
struct bio_vec *bi_io_vec; /* the actual vec list */
bio_end_io_t *bi_end_io;
@@ -105,6 +107,7 @@ struct bio {
#define BIO_BOUNCED 5 /* bio is a bounce bio */
#define BIO_USER_MAPPED 6 /* contains user pages */
#define BIO_EOPNOTSUPP 7 /* not supported */
#define BIO_CPU_AFFINE 8 /* complete bio on same CPU as submitted */
#define bio_flagged(bio, flag) ((bio)->bi_flags & (1 << (flag)))
/*
@@ -342,6 +345,14 @@ void zero_fill_bio(struct bio *bio);
extern struct bio_vec *bvec_alloc_bs(gfp_t, int, unsigned long *, struct bio_set *);
extern unsigned int bvec_nr_vecs(unsigned short idx);
/*
* Allow queuer to specify a completion CPU for this bio
*/
static inline void bio_set_completion_cpu(struct bio *bio, unsigned int cpu)
{
bio->bi_comp_cpu = cpu;
}
/*
* bio_set is used to allow other portions of the IO system to
* allocate their own private memory pools for bio and iovec structures.