xiaomi-sm8450/android_kernel_xiaomi_sm8450
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

IB/rdmavt: Use per-CPU reference count for MRs

Browse Source 
Having per-CPU reference count for each MR prevents
cache-line bouncing across the system. Thus, it
prevents bottlenecks. Use per-CPU reference counts
per MR.

The per-CPU reference count for FMRs is used in
atomic mode to allow accurate testing of the busy
state. Other MR types run in per-CPU mode MR until
they're freed.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Sebastian Sanchez <sebastian.sanchez@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
This commit is contained in:
Sebastian Sanchez
2017-02-08 05:26:31 -08:00
committed by Doug Ledford
parent f3e862cb68
commit 338adfdddf
2 changed files with 43 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
include/rdma/rdmavt_mr.h
View File

@@ -52,6 +52,7 @@
* For Memory Regions. This stuff should probably be moved into rdmavt/mr.h once
* drivers no longer need access to the MR directly.
*/
#include <linux/percpu-refcount.h>
/*
* A segment is a linear region of low physical memory.
@@ -79,11 +80,11 @@ struct rvt_mregion {
int access_flags;
u32 max_segs; /* number of rvt_segs in all the arrays */
u32 mapsz; /* size of the map array */
atomic_t lkey_invalid; /* true if current lkey is invalid */
u8 page_shift; /* 0 - non unform/non powerof2 sizes */
u8 lkey_published; /* in global table */
atomic_t lkey_invalid; /* true if current lkey is invalid */
struct percpu_ref refcount;
struct completion comp; /* complete when refcount goes to zero */
atomic_t refcount;
struct rvt_segarray *map[0]; /* the segments */
};
@@ -123,13 +124,12 @@ struct rvt_sge_state {
static inline void rvt_put_mr(struct rvt_mregion *mr)
{
if (unlikely(atomic_dec_and_test(&mr->refcount)))
complete(&mr->comp);
percpu_ref_put(&mr->refcount);
}
static inline void rvt_get_mr(struct rvt_mregion *mr)
{
atomic_inc(&mr->refcount);
percpu_ref_get(&mr->refcount);
}
static inline void rvt_put_ss(struct rvt_sge_state *ss)