WHAT IS SLAB ALLOCATION ?<\/p>\n
Slab allocation is a memory management mechanism intended for the efficient memory allocation of kernel objects. It eliminates fragmentation caused by allocations and deallocations. The technique is used to retain allocated memory that contains a data object of a certain type for reuse upon subsequent allocations of objects of the same type. It is analogous to an object pool, but only applies to memory, not other resources.\n<\/pre>\n\nInstead of parsing the highly verbose \/proc\/slabinfo file manually, the \/usr\/bin\/slabtop program displays kernel slab cache information in real time. This program allows for custom configurations, including column sorting and screen refreshing.\n<\/pre>\nRELATED SHELL SESSION EXPOSURE
\n[bash]
\n$sudo slabtop -o
\n Active \/ Total Objects (% used) : 535097 \/ 607019 (88.2%)
\n Active \/ Total Slabs (% used) : 21678 \/ 21678 (100.0%)
\n Active \/ Total Caches (% used) : 101 \/ 124 (81.5%)
\n Active \/ Total Size (% used) : 116716.47K \/ 136870.05K (85.3%)
\n Minimum \/ Average \/ Maximum Object : 0.01K \/ 0.22K \/ 8.00K<\/p>\nOBJS ACTIVE USE OBJ SIZE SLABS OBJ\/SLAB CACHE SIZE NAME
\n118209 104139 88% 0.10K 3031 39 12124K buffer_head
\n103840 103648 99% 0.25K 6490 16 25960K skbuff_head_cache
\n 43225 41001 94% 0.20K 2275 19 9100K vm_area_struct
\n 37590 30689 81% 0.19K 1790 21 7160K dentry
\n 32512 28745 88% 0.03K 254 128 1016K kmalloc-32
\n 32384 27081 83% 0.06K 506 64 2024K anon_vma_chain
\n 29010 28898 99% 0.13K 967 30 3868K kernfs_node_cache
\n 20272 12148 59% 0.57K 724 28 11584K radix_tree_node
\n 18752 16591 88% 0.06K 293 64 1172K kmalloc-64
\n 17342 14540 83% 0.09K 377 46 1508K anon_vma
\n 16578 14490 87% 0.59K 614 27 9824K inode_cache
\n 14576 11558 79% 0.25K 911 16 3644K filp
\n 12800 11906 93% 0.02K 50 256 200K kmalloc-16
\n 12390 9609 77% 1.05K 413 30 13216K ext4_inode_cache
\n 10200 6294 61% 0.05K 120 85 480K ftrace_event_field
\n 8704 7034 80% 0.01K 17 512 68K kmalloc-8
\n 4704 3944 83% 0.09K 112 42 448K kmalloc-96
\n 4256 4200 98% 0.07K 76 56 304K Acpi-Operand
\n 4074 3120 76% 0.19K 194 21 776K cred_jar
\n 3668 3665 99% 0.14K 131 28 524K ext4_groupinfo_4k
\n 3570 3570 100% 0.04K 35 102 140K ext4_extent_status
\n 3450 2422 70% 0.69K 150 23 2400K shmem_inode_cache
\n 3312 2611 78% 0.66K 138 24 2208K proc_inode_cache
\n 3200 2599 81% 0.12K 100 32 400K pid
\n 2754 2751 99% 0.04K 27 102 108K Acpi-Namespace
\n 2688 2432 90% 0.03K 21 128 84K fscrypt_info
\n 2550 1633 64% 0.04K 25 102 100K pde_opener
\n 2448 2137 87% 0.50K 153 16 1224K kmalloc-512
\n 2394 2134 89% 0.19K 114 21 456K kmalloc-192
\n 2384 2171 91% 1.00K 149 16 2384K kmalloc-1024
\n 2176 1898 87% 0.12K 68 32 272K kmalloc-128
\n 2139 1734 81% 0.69K 93 23 1488K sock_inode_cache
\n 2112 2112 100% 0.06K 33 64 132K kmem_cache_node
\n 2037 2037 100% 0.38K 97 21 776K kmem_cache
\n 1932 1006 52% 0.56K 69 28 1104K i915_vma
\n 1840 1819 98% 0.09K 40 46 160K trace_event_file
\n 1568 1386 88% 0.07K 28 56 112K eventpoll_pwq
\n 1488 1382 92% 2.00K 93 16 2976K kmalloc-2048
\n 1472 1200 81% 1.00K 92 16 1472K UNIX
\n 1323 688 52% 0.75K 63 21 1008K xfrm_state
\n 1275 896 70% 0.08K 25 51 100K inotify_inode_mark
\n 945 903 95% 0.19K 45 21 180K proc_dir_entry
\n 896 836 93% 0.25K 56 16 224K kmalloc-256
\n 884 782 88% 0.12K 26 34 104K jbd2_journal_head
\n 882 777 88% 0.38K 42 21 336K mnt_cache
\n 837 663 79% 3.38K 93 9 2976K task_struct
\n 810 590 72% 1.06K 27 30 864K mm_struct
\n 780 618 79% 1.06K 26 30 832K signal_cache
\n 720 593 82% 4.00K 90 8 2880K kmalloc-4096
\n 680 563 82% 0.02K 4 170 16K numa_policy
\n 640 510 79% 0.25K 40 16 160K pool_workqueue
\n 511 292 57% 0.05K 7 73 28K mbcache
\n 506 388 76% 0.69K 22 23 352K files_cache
\n 450 367 81% 2.06K 30 15 960K sighand_cache
\n 288 195 67% 0.12K 9 32 36K secpath_cache
\n 256 2 0% 0.02K 1 256 4K jbd2_revoke_table_s
\n 234 234 100% 0.10K 6 39 24K blkdev_ioc
\n 225 177 78% 1.25K 9 25 288K UDPv6
\n 224 224 100% 0.12K 7 32 28K scsi_sense_cache
\n 219 152 69% 0.05K 3 73 12K Acpi-Parse
\n 208 144 69% 0.25K 13 16 52K dquot
\n 172 171 99% 8.00K 43 4 1376K kmalloc-8192
\n 170 170 100% 0.05K 2 85 8K fscrypt_ctx
\n 168 100 59% 1.12K 6 28 192K RAWv6
\n 165 150 90% 2.12K 11 15 352K TCP
\n 128 128 100% 0.06K 2 64 8K dmaengine-unmap-2
\n 128 128 100% 0.03K 1 128 4K sd_ext_cdb
\n 128 128 100% 0.06K 2 64 8K ext4_io_end
\n 112 82 73% 0.50K 7 16 56K skbuff_fclone_cache
\n 102 102 100% 0.08K 2 51 8K Acpi-State
\n 96 87 90% 0.32K 4 24 32K taskstats
\n 85 68 80% 0.23K 5 17 20K tw_sock_TCP
\n 85 42 49% 0.94K 5 17 80K RAW
\n 80 80 100% 0.20K 4 20 16K file_lock_cache
\n 76 44 57% 0.81K 4 19 64K bdev_cache
\n 75 73 97% 0.62K 3 25 48K task_group
\n 72 60 83% 8.00K 18 4 576K biovec-max
\n 64 42 65% 2.00K 4 16 128K biovec-128
\n 64 64 100% 0.12K 2 32 8K ext4_allocation_context
\n 63 46 73% 0.19K 3 21 12K dmaengine-unmap-16
\n 50 50 100% 0.16K 2 25 8K sigqueue
\n 50 50 100% 0.62K 2 25 32K i915_request
\n 42 42 100% 2.25K 3 14 96K TCPv6
\n 40 20 50% 0.39K 2 20 16K fuse_request
\n 38 19 50% 0.20K 2 19 8K ip4-frags
\n 36 0 0% 0.43K 2 18 16K uts_namespace
\n 34 17 50% 0.23K 2 17 8K posix_timers_cache
\n 34 1 2% 0.94K 2 17 32K mqueue_inode_cache
\n 32 16 50% 4.00K 4 8 128K names_cache
\n 32 32 100% 1.00K 2 16 32K biovec-64
\n 30 30 100% 1.06K 1 30 32K dmaengine-unmap-128
\n 26 26 100% 0.61K 1 26 16K hugetlbfs_inode_cache
\n 25 25 100% 0.62K 1 25 16K rpc_inode_cache
\n 21 21 100% 0.75K 1 21 16K dax_cache
\n 19 19 100% 0.81K 1 19 16K fuse_inode
\n 17 0 0% 0.45K 1 17 8K user_namespace
\n 16 0 0% 0.50K 1 16 8K dma-kmalloc-512
\n 15 6 40% 6.19K 3 5 96K net_namespace
\n 15 15 100% 2.06K 1 15 32K dmaengine-unmap-256
\n 13 13 100% 2.40K 1 13 32K request_queue
\n 8 2 25% 4.00K 1 8 32K sgpool-128
\n 0 0 0% 0.09K 0 42 0K dma-kmalloc-96
\n 0 0 0% 0.19K 0 21 0K dma-kmalloc-192
\n 0 0 0% 0.01K 0 512 0K dma-kmalloc-8
\n 0 0 0% 0.02K 0 256 0K dma-kmalloc-16
\n 0 0 0% 0.03K 0 128 0K dma-kmalloc-32
\n 0 0 0% 0.06K 0 64 0K dma-kmalloc-64
\n 0 0 0% 0.12K 0 32 0K dma-kmalloc-128
\n 0 0 0% 0.25K 0 16 0K dma-kmalloc-256
\n 0 0 0% 1.00K 0 16 0K dma-kmalloc-1024
\n 0 0 0% 2.00K 0 16 0K dma-kmalloc-2048
\n 0 0 0% 4.00K 0 8 0K dma-kmalloc-4096
\n 0 0 0% 8.00K 0 4 0K dma-kmalloc-8192
\n 0 0 0% 0.12K 0 34 0K iint_cache
\n 0 0 0% 0.30K 0 26 0K blkdev_requests
\n 0 0 0% 0.30K 0 26 0K request_sock_TCP
\n 0 0 0% 0.94K 0 17 0K PING
\n 0 0 0% 0.31K 0 25 0K xfrm_dst_cache
\n 0 0 0% 0.03K 0 128 0K dnotify_struct
\n 0 0 0% 0.19K 0 21 0K userfaultfd_ctx_cache
\n 0 0 0% 0.12K 0 34 0K cfq_io_cq
\n 0 0 0% 0.30K 0 26 0K request_sock_TCPv6
\n 0 0 0% 0.23K 0 17 0K tw_sock_TCPv6
\n 0 0 0% 1.12K 0 28 0K PINGv6
\n$<\/p>\n[\/bash]
\nRELATED CODE PARTS FROM LINUX KERNEL<\/p>\nA
\n[c]
\nstruct slabinfo {
\n\tunsigned long active_objs;
\n\tunsigned long num_objs;
\n\tunsigned long active_slabs;
\n\tunsigned long num_slabs;
\n\tunsigned long shared_avail;
\n\tunsigned int limit;
\n\tunsigned int batchcount;
\n\tunsigned int shared;
\n\tunsigned int objects_per_slab;
\n\tunsigned int cache_order;
\n};<\/p>\nvoid get_slabinfo(struct kmem_cache *s, struct slabinfo *sinfo);
\nvoid slabinfo_show_stats(struct seq_file *m, struct kmem_cache *s);
\nssize_t slabinfo_write(struct file *file, const char __user *buffer,
\n\t\t size_t count, loff_t *ppos);
\n[\/c]
\nB
\n[c]
\nvoid get_slabinfo(struct kmem_cache *cachep, struct slabinfo *sinfo)
\n{
\n\tunsigned long active_objs, num_objs, active_slabs;
\n\tunsigned long total_slabs = 0, free_objs = 0, shared_avail = 0;
\n\tunsigned long free_slabs = 0;
\n\tint node;
\n\tstruct kmem_cache_node *n;<\/p>\n\tfor_each_kmem_cache_node(cachep, node, n) {
\n\t\tcheck_irq_on();
\n\t\tspin_lock_irq(&n->list_lock);<\/p>\n\t\ttotal_slabs += n->total_slabs;
\n\t\tfree_slabs += n->free_slabs;
\n\t\tfree_objs += n->free_objects;<\/p>\n\t\tif (n->shared)
\n\t\t\tshared_avail += n->shared->avail;<\/p>\n\t\tspin_unlock_irq(&n->list_lock);
\n\t}
\n\tnum_objs = total_slabs * cachep->num;
\n\tactive_slabs = total_slabs – free_slabs;
\n\tactive_objs = num_objs – free_objs;<\/p>\n\tsinfo->active_objs = active_objs;
\n\tsinfo->num_objs = num_objs;
\n\tsinfo->active_slabs = active_slabs;
\n\tsinfo->num_slabs = total_slabs;
\n\tsinfo->shared_avail = shared_avail;
\n\tsinfo->limit = cachep->limit;
\n\tsinfo->batchcount = cachep->batchcount;
\n\tsinfo->shared = cachep->shared;
\n\tsinfo->objects_per_slab = cachep->num;
\n\tsinfo->cache_order = cachep->gfporder;
\n}<\/p>\n[\/c]<\/p>\n
LINKS
\nhttps:\/\/lkml.org\/lkml\/2017\/3\/27\/370<\/a>
\nhttp:\/\/www.secretmango.com\/jimb\/Whitepapers\/slabs\/slab.html<\/a>
\nhttps:\/\/access.redhat.com\/documentation\/en-US\/Red_Hat_Enterprise_Linux\/4\/html\/Reference_Guide\/s2-proc-slabinfo.html<\/a>
\nhttps:\/\/en.wikipedia.org\/wiki\/Slab_allocation<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"WHAT IS SLAB ALLOCATION ? Slab allocation is a memory management mechanism intended for the efficient memory allocation of kernel objects. It eliminates fragmentation caused by allocations and deallocations. The technique is used to retain allocated memory that contains a data object of a certain type for reuse upon subsequent allocations of objects of the … <\/p>\n