1 ====================================
2 SLOW WORK ITEM EXECUTION THREAD POOL
3 ====================================
5 By: David Howells <dhowells@redhat.com>
7 The slow work item execution thread pool is a pool of threads for performing
8 things that take a relatively long time, such as making mkdir calls.
9 Typically, when processing something, these items will spend a lot of time
10 blocking a thread on I/O, thus making that thread unavailable for doing other
13 The standard workqueue model is unsuitable for this class of work item as that
14 limits the owner to a single thread or a single thread per CPU. For some
15 tasks, however, more threads - or fewer - are required.
17 There is just one pool per system. It contains no threads unless something
18 wants to use it - and that something must register its interest first. When
19 the pool is active, the number of threads it contains is dynamic, varying
20 between a maximum and minimum setting, depending on the load.
27 This pool support two classes of work items:
31 (*) Very slow work items.
33 The former are expected to finish much quicker than the latter.
35 An operation of the very slow class may do a batch combination of several
36 lookups, mkdirs, and a create for instance.
38 An operation of the ordinarily slow class may, for example, write stuff or
39 expand files, provided the time taken to do so isn't too long.
41 Operations of both types may sleep during execution, thus tying up the thread
45 THREAD-TO-CLASS ALLOCATION
46 --------------------------
48 Not all the threads in the pool are available to work on very slow work items.
49 The number will be between one and one fewer than the number of active threads.
50 This is configurable (see the "Pool Configuration" section).
52 All the threads are available to work on ordinarily slow work items, but a
53 percentage of the threads will prefer to work on very slow work items.
55 The configuration ensures that at least one thread will be available to work on
56 very slow work items, and at least one thread will be available that won't work
57 on very slow work items at all.
64 Firstly, a module or subsystem wanting to make use of slow work items must
65 register its interest:
67 int ret = slow_work_register_user(struct module *module);
69 This will return 0 if successful, or a -ve error upon failure. The module
70 pointer should be the module interested in using this facility (almost
71 certainly THIS_MODULE).
74 Slow work items may then be set up by:
76 (1) Declaring a slow_work struct type variable:
78 #include <linux/slow-work.h>
80 struct slow_work myitem;
82 (2) Declaring the operations to be used for this item:
84 struct slow_work_ops myitem_ops = {
85 .get_ref = myitem_get_ref,
86 .put_ref = myitem_put_ref,
87 .execute = myitem_execute,
90 [*] For a description of the ops, see section "Item Operations".
92 (3) Initialising the item:
94 slow_work_init(&myitem, &myitem_ops);
98 vslow_work_init(&myitem, &myitem_ops);
100 depending on its class.
102 A suitably set up work item can then be enqueued for processing:
104 int ret = slow_work_enqueue(&myitem);
106 This will return a -ve error if the thread pool is unable to gain a reference
107 on the item, 0 otherwise.
110 The items are reference counted, so there ought to be no need for a flush
111 operation. But as the reference counting is optional, means to cancel
112 existing work items are also included:
114 cancel_slow_work(&myitem);
116 can be used to cancel pending work. The above cancel function waits for
117 existing work to have been executed (or prevent execution of them, depending
121 When all a module's slow work items have been processed, and the
122 module has no further interest in the facility, it should unregister its
125 slow_work_unregister_user(struct module *module);
127 The module pointer is used to wait for all outstanding work items for that
128 module before completing the unregistration. This prevents the put_ref() code
129 from being taken away before it completes. module should almost certainly be
137 Each work item requires a table of operations of type struct slow_work_ops.
138 Only ->execute() is required, getting and putting of a reference are optional.
140 (*) Get a reference on an item:
142 int (*get_ref)(struct slow_work *work);
144 This allows the thread pool to attempt to pin an item by getting a
145 reference on it. This function should return 0 if the reference was
146 granted, or a -ve error otherwise. If an error is returned,
147 slow_work_enqueue() will fail.
149 The reference is held whilst the item is queued and whilst it is being
150 executed. The item may then be requeued with the same reference held, or
151 the reference will be released.
153 (*) Release a reference on an item:
155 void (*put_ref)(struct slow_work *work);
157 This allows the thread pool to unpin an item by releasing the reference on
158 it. The thread pool will not touch the item again once this has been
163 void (*execute)(struct slow_work *work);
165 This should perform the work required of the item. It may sleep, it may
166 perform disk I/O and it may wait for locks.
173 The slow-work thread pool has a number of configurables:
175 (*) /proc/sys/kernel/slow-work/min-threads
177 The minimum number of threads that should be in the pool whilst it is in
178 use. This may be anywhere between 2 and max-threads.
180 (*) /proc/sys/kernel/slow-work/max-threads
182 The maximum number of threads that should in the pool. This may be
183 anywhere between min-threads and 255 or NR_CPUS * 2, whichever is greater.
185 (*) /proc/sys/kernel/slow-work/vslow-percentage
187 The percentage of active threads in the pool that may be used to execute
188 very slow work items. This may be between 1 and 99. The resultant number
189 is bounded to between 1 and one fewer than the number of active threads.
190 This ensures there is always at least one thread that can process very
191 slow work items, and always at least one thread that won't.