Rev 1702 | Rev 1760 | Go to most recent revision | Show entire file | Regard whitespace | Details | Blame | Last modification | View Log | RSS feed
| Rev 1702 | Rev 1757 | ||
|---|---|---|---|
| Line 81... | Line 81... | ||
| 81 | * releases slabs from cpu-shared magazine-list, until at least 1 slab |
81 | * releases slabs from cpu-shared magazine-list, until at least 1 slab |
| 82 | * is deallocated in each cache (this algorithm should probably change). |
82 | * is deallocated in each cache (this algorithm should probably change). |
| 83 | * The brutal reclaim removes all cached objects, even from CPU-bound |
83 | * The brutal reclaim removes all cached objects, even from CPU-bound |
| 84 | * magazines. |
84 | * magazines. |
| 85 | * |
85 | * |
| 86 | * TODO:@n |
86 | * @todo |
| 87 | * For better CPU-scaling the magazine allocation strategy should |
87 | * For better CPU-scaling the magazine allocation strategy should |
| 88 | * be extended. Currently, if the cache does not have magazine, it asks |
88 | * be extended. Currently, if the cache does not have magazine, it asks |
| 89 | * for non-cpu cached magazine cache to provide one. It might be feasible |
89 | * for non-cpu cached magazine cache to provide one. It might be feasible |
| 90 | * to add cpu-cached magazine cache (which would allocate it's magazines |
90 | * to add cpu-cached magazine cache (which would allocate it's magazines |
| 91 | * from non-cpu-cached mag. cache). This would provide a nice per-cpu |
91 | * from non-cpu-cached mag. cache). This would provide a nice per-cpu |
| 92 | * buffer. The other possibility is to use the per-cache |
92 | * buffer. The other possibility is to use the per-cache |
| 93 | * 'empty-magazine-list', which decreases competing for 1 per-system |
93 | * 'empty-magazine-list', which decreases competing for 1 per-system |
| 94 | * magazine cache. |
94 | * magazine cache. |
| 95 | * |
95 | * |
| - | 96 | * @todo |
|
| 96 | * @li it might be good to add granularity of locks even to slab level, |
97 | * it might be good to add granularity of locks even to slab level, |
| 97 | * we could then try_spinlock over all partial slabs and thus improve |
98 | * we could then try_spinlock over all partial slabs and thus improve |
| 98 | * scalability even on slab level |
99 | * scalability even on slab level |
| 99 | */ |
100 | */ |
| 100 | 101 | ||
| 101 | #include <synch/spinlock.h> |
102 | #include <synch/spinlock.h> |
| Line 907... | Line 908... | ||
| 907 | _slab_free(slab->cache, obj, slab); |
908 | _slab_free(slab->cache, obj, slab); |
| 908 | } |
909 | } |
| 909 | 910 | ||
| 910 | /** @} |
911 | /** @} |
| 911 | */ |
912 | */ |
| 912 | - | ||