1===================================================================== 2Everything you never wanted to know about kobjects, ksets, and ktypes 3===================================================================== 4 5:Author: Greg Kroah-Hartman <gregkh@linuxfoundation.org> 6:Last updated: December 19, 2007 7 8Based on an original article by Jon Corbet for lwn.net written October 1, 92003 and located at http://lwn.net/Articles/51437/ 10 11Part of the difficulty in understanding the driver model - and the kobject 12abstraction upon which it is built - is that there is no obvious starting 13place. Dealing with kobjects requires understanding a few different types, 14all of which make reference to each other. In an attempt to make things 15easier, we'll take a multi-pass approach, starting with vague terms and 16adding detail as we go. To that end, here are some quick definitions of 17some terms we will be working with. 18 19 - A kobject is an object of type struct kobject. Kobjects have a name 20 and a reference count. A kobject also has a parent pointer (allowing 21 objects to be arranged into hierarchies), a specific type, and, 22 usually, a representation in the sysfs virtual filesystem. 23 24 Kobjects are generally not interesting on their own; instead, they are 25 usually embedded within some other structure which contains the stuff 26 the code is really interested in. 27 28 No structure should EVER have more than one kobject embedded within it. 29 If it does, the reference counting for the object is sure to be messed 30 up and incorrect, and your code will be buggy. So do not do this. 31 32 - A ktype is the type of object that embeds a kobject. Every structure 33 that embeds a kobject needs a corresponding ktype. The ktype controls 34 what happens to the kobject when it is created and destroyed. 35 36 - A kset is a group of kobjects. These kobjects can be of the same ktype 37 or belong to different ktypes. The kset is the basic container type for 38 collections of kobjects. Ksets contain their own kobjects, but you can 39 safely ignore that implementation detail as the kset core code handles 40 this kobject automatically. 41 42 When you see a sysfs directory full of other directories, generally each 43 of those directories corresponds to a kobject in the same kset. 44 45We'll look at how to create and manipulate all of these types. A bottom-up 46approach will be taken, so we'll go back to kobjects. 47 48 49Embedding kobjects 50================== 51 52It is rare for kernel code to create a standalone kobject, with one major 53exception explained below. Instead, kobjects are used to control access to 54a larger, domain-specific object. To this end, kobjects will be found 55embedded in other structures. If you are used to thinking of things in 56object-oriented terms, kobjects can be seen as a top-level, abstract class 57from which other classes are derived. A kobject implements a set of 58capabilities which are not particularly useful by themselves, but which are 59nice to have in other objects. The C language does not allow for the 60direct expression of inheritance, so other techniques - such as structure 61embedding - must be used. 62 63(As an aside, for those familiar with the kernel linked list implementation, 64this is analogous as to how "list_head" structs are rarely useful on 65their own, but are invariably found embedded in the larger objects of 66interest.) 67 68So, for example, the UIO code in drivers/uio/uio.c has a structure that 69defines the memory region associated with a uio device:: 70 71 struct uio_map { 72 struct kobject kobj; 73 struct uio_mem *mem; 74 }; 75 76If you have a struct uio_map structure, finding its embedded kobject is 77just a matter of using the kobj member. Code that works with kobjects will 78often have the opposite problem, however: given a struct kobject pointer, 79what is the pointer to the containing structure? You must avoid tricks 80(such as assuming that the kobject is at the beginning of the structure) 81and, instead, use the container_of() macro, found in <linux/kernel.h>:: 82 83 container_of(pointer, type, member) 84 85where: 86 87 * "pointer" is the pointer to the embedded kobject, 88 * "type" is the type of the containing structure, and 89 * "member" is the name of the structure field to which "pointer" points. 90 91The return value from container_of() is a pointer to the corresponding 92container type. So, for example, a pointer "kp" to a struct kobject 93embedded *within* a struct uio_map could be converted to a pointer to the 94*containing* uio_map structure with:: 95 96 struct uio_map *u_map = container_of(kp, struct uio_map, kobj); 97 98For convenience, programmers often define a simple macro for "back-casting" 99kobject pointers to the containing type. Exactly this happens in the 100earlier drivers/uio/uio.c, as you can see here:: 101 102 struct uio_map { 103 struct kobject kobj; 104 struct uio_mem *mem; 105 }; 106 107 #define to_map(map) container_of(map, struct uio_map, kobj) 108 109where the macro argument "map" is a pointer to the struct kobject in 110question. That macro is subsequently invoked with:: 111 112 struct uio_map *map = to_map(kobj); 113 114 115Initialization of kobjects 116========================== 117 118Code which creates a kobject must, of course, initialize that object. Some 119of the internal fields are setup with a (mandatory) call to kobject_init():: 120 121 void kobject_init(struct kobject *kobj, struct kobj_type *ktype); 122 123The ktype is required for a kobject to be created properly, as every kobject 124must have an associated kobj_type. After calling kobject_init(), to 125register the kobject with sysfs, the function kobject_add() must be called:: 126 127 int kobject_add(struct kobject *kobj, struct kobject *parent, 128 const char *fmt, ...); 129 130This sets up the parent of the kobject and the name for the kobject 131properly. If the kobject is to be associated with a specific kset, 132kobj->kset must be assigned before calling kobject_add(). If a kset is 133associated with a kobject, then the parent for the kobject can be set to 134NULL in the call to kobject_add() and then the kobject's parent will be the 135kset itself. 136 137As the name of the kobject is set when it is added to the kernel, the name 138of the kobject should never be manipulated directly. If you must change 139the name of the kobject, call kobject_rename():: 140 141 int kobject_rename(struct kobject *kobj, const char *new_name); 142 143kobject_rename does not perform any locking or have a solid notion of 144what names are valid so the caller must provide their own sanity checking 145and serialization. 146 147There is a function called kobject_set_name() but that is legacy cruft and 148is being removed. If your code needs to call this function, it is 149incorrect and needs to be fixed. 150 151To properly access the name of the kobject, use the function 152kobject_name():: 153 154 const char *kobject_name(const struct kobject * kobj); 155 156There is a helper function to both initialize and add the kobject to the 157kernel at the same time, called surprisingly enough kobject_init_and_add():: 158 159 int kobject_init_and_add(struct kobject *kobj, struct kobj_type *ktype, 160 struct kobject *parent, const char *fmt, ...); 161 162The arguments are the same as the individual kobject_init() and 163kobject_add() functions described above. 164 165 166Uevents 167======= 168 169After a kobject has been registered with the kobject core, you need to 170announce to the world that it has been created. This can be done with a 171call to kobject_uevent():: 172 173 int kobject_uevent(struct kobject *kobj, enum kobject_action action); 174 175Use the KOBJ_ADD action for when the kobject is first added to the kernel. 176This should be done only after any attributes or children of the kobject 177have been initialized properly, as userspace will instantly start to look 178for them when this call happens. 179 180When the kobject is removed from the kernel (details on how to do that are 181below), the uevent for KOBJ_REMOVE will be automatically created by the 182kobject core, so the caller does not have to worry about doing that by 183hand. 184 185 186Reference counts 187================ 188 189One of the key functions of a kobject is to serve as a reference counter 190for the object in which it is embedded. As long as references to the object 191exist, the object (and the code which supports it) must continue to exist. 192The low-level functions for manipulating a kobject's reference counts are:: 193 194 struct kobject *kobject_get(struct kobject *kobj); 195 void kobject_put(struct kobject *kobj); 196 197A successful call to kobject_get() will increment the kobject's reference 198counter and return the pointer to the kobject. 199 200When a reference is released, the call to kobject_put() will decrement the 201reference count and, possibly, free the object. Note that kobject_init() 202sets the reference count to one, so the code which sets up the kobject will 203need to do a kobject_put() eventually to release that reference. 204 205Because kobjects are dynamic, they must not be declared statically or on 206the stack, but instead, always allocated dynamically. Future versions of 207the kernel will contain a run-time check for kobjects that are created 208statically and will warn the developer of this improper usage. 209 210If all that you want to use a kobject for is to provide a reference counter 211for your structure, please use the struct kref instead; a kobject would be 212overkill. For more information on how to use struct kref, please see the 213file Documentation/kref.txt in the Linux kernel source tree. 214 215 216Creating "simple" kobjects 217========================== 218 219Sometimes all that a developer wants is a way to create a simple directory 220in the sysfs hierarchy, and not have to mess with the whole complication of 221ksets, show and store functions, and other details. This is the one 222exception where a single kobject should be created. To create such an 223entry, use the function:: 224 225 struct kobject *kobject_create_and_add(char *name, struct kobject *parent); 226 227This function will create a kobject and place it in sysfs in the location 228underneath the specified parent kobject. To create simple attributes 229associated with this kobject, use:: 230 231 int sysfs_create_file(struct kobject *kobj, struct attribute *attr); 232 233or:: 234 235 int sysfs_create_group(struct kobject *kobj, struct attribute_group *grp); 236 237Both types of attributes used here, with a kobject that has been created 238with the kobject_create_and_add(), can be of type kobj_attribute, so no 239special custom attribute is needed to be created. 240 241See the example module, samples/kobject/kobject-example.c for an 242implementation of a simple kobject and attributes. 243 244 245 246ktypes and release methods 247========================== 248 249One important thing still missing from the discussion is what happens to a 250kobject when its reference count reaches zero. The code which created the 251kobject generally does not know when that will happen; if it did, there 252would be little point in using a kobject in the first place. Even 253predictable object lifecycles become more complicated when sysfs is brought 254in as other portions of the kernel can get a reference on any kobject that 255is registered in the system. 256 257The end result is that a structure protected by a kobject cannot be freed 258before its reference count goes to zero. The reference count is not under 259the direct control of the code which created the kobject. So that code must 260be notified asynchronously whenever the last reference to one of its 261kobjects goes away. 262 263Once you registered your kobject via kobject_add(), you must never use 264kfree() to free it directly. The only safe way is to use kobject_put(). It 265is good practice to always use kobject_put() after kobject_init() to avoid 266errors creeping in. 267 268This notification is done through a kobject's release() method. Usually 269such a method has a form like:: 270 271 void my_object_release(struct kobject *kobj) 272 { 273 struct my_object *mine = container_of(kobj, struct my_object, kobj); 274 275 /* Perform any additional cleanup on this object, then... */ 276 kfree(mine); 277 } 278 279One important point cannot be overstated: every kobject must have a 280release() method, and the kobject must persist (in a consistent state) 281until that method is called. If these constraints are not met, the code is 282flawed. Note that the kernel will warn you if you forget to provide a 283release() method. Do not try to get rid of this warning by providing an 284"empty" release function; you will be mocked mercilessly by the kobject 285maintainer if you attempt this. 286 287Note, the name of the kobject is available in the release function, but it 288must NOT be changed within this callback. Otherwise there will be a memory 289leak in the kobject core, which makes people unhappy. 290 291Interestingly, the release() method is not stored in the kobject itself; 292instead, it is associated with the ktype. So let us introduce struct 293kobj_type:: 294 295 struct kobj_type { 296 void (*release)(struct kobject *kobj); 297 const struct sysfs_ops *sysfs_ops; 298 struct attribute **default_attrs; 299 const struct kobj_ns_type_operations *(*child_ns_type)(struct kobject *kobj); 300 const void *(*namespace)(struct kobject *kobj); 301 }; 302 303This structure is used to describe a particular type of kobject (or, more 304correctly, of containing object). Every kobject needs to have an associated 305kobj_type structure; a pointer to that structure must be specified when you 306call kobject_init() or kobject_init_and_add(). 307 308The release field in struct kobj_type is, of course, a pointer to the 309release() method for this type of kobject. The other two fields (sysfs_ops 310and default_attrs) control how objects of this type are represented in 311sysfs; they are beyond the scope of this document. 312 313The default_attrs pointer is a list of default attributes that will be 314automatically created for any kobject that is registered with this ktype. 315 316 317ksets 318===== 319 320A kset is merely a collection of kobjects that want to be associated with 321each other. There is no restriction that they be of the same ktype, but be 322very careful if they are not. 323 324A kset serves these functions: 325 326 - It serves as a bag containing a group of objects. A kset can be used by 327 the kernel to track "all block devices" or "all PCI device drivers." 328 329 - A kset is also a subdirectory in sysfs, where the associated kobjects 330 with the kset can show up. Every kset contains a kobject which can be 331 set up to be the parent of other kobjects; the top-level directories of 332 the sysfs hierarchy are constructed in this way. 333 334 - Ksets can support the "hotplugging" of kobjects and influence how 335 uevent events are reported to user space. 336 337In object-oriented terms, "kset" is the top-level container class; ksets 338contain their own kobject, but that kobject is managed by the kset code and 339should not be manipulated by any other user. 340 341A kset keeps its children in a standard kernel linked list. Kobjects point 342back to their containing kset via their kset field. In almost all cases, 343the kobjects belonging to a kset have that kset (or, strictly, its embedded 344kobject) in their parent. 345 346As a kset contains a kobject within it, it should always be dynamically 347created and never declared statically or on the stack. To create a new 348kset use:: 349 350 struct kset *kset_create_and_add(const char *name, 351 struct kset_uevent_ops *u, 352 struct kobject *parent); 353 354When you are finished with the kset, call:: 355 356 void kset_unregister(struct kset *kset); 357 358to destroy it. This removes the kset from sysfs and decrements its reference 359count. When the reference count goes to zero, the kset will be released. 360Because other references to the kset may still exist, the release may happen 361after kset_unregister() returns. 362 363An example of using a kset can be seen in the 364samples/kobject/kset-example.c file in the kernel tree. 365 366If a kset wishes to control the uevent operations of the kobjects 367associated with it, it can use the struct kset_uevent_ops to handle it:: 368 369 struct kset_uevent_ops { 370 int (*filter)(struct kset *kset, struct kobject *kobj); 371 const char *(*name)(struct kset *kset, struct kobject *kobj); 372 int (*uevent)(struct kset *kset, struct kobject *kobj, 373 struct kobj_uevent_env *env); 374 }; 375 376 377The filter function allows a kset to prevent a uevent from being emitted to 378userspace for a specific kobject. If the function returns 0, the uevent 379will not be emitted. 380 381The name function will be called to override the default name of the kset 382that the uevent sends to userspace. By default, the name will be the same 383as the kset itself, but this function, if present, can override that name. 384 385The uevent function will be called when the uevent is about to be sent to 386userspace to allow more environment variables to be added to the uevent. 387 388One might ask how, exactly, a kobject is added to a kset, given that no 389functions which perform that function have been presented. The answer is 390that this task is handled by kobject_add(). When a kobject is passed to 391kobject_add(), its kset member should point to the kset to which the 392kobject will belong. kobject_add() will handle the rest. 393 394If the kobject belonging to a kset has no parent kobject set, it will be 395added to the kset's directory. Not all members of a kset do necessarily 396live in the kset directory. If an explicit parent kobject is assigned 397before the kobject is added, the kobject is registered with the kset, but 398added below the parent kobject. 399 400 401Kobject removal 402=============== 403 404After a kobject has been registered with the kobject core successfully, it 405must be cleaned up when the code is finished with it. To do that, call 406kobject_put(). By doing this, the kobject core will automatically clean up 407all of the memory allocated by this kobject. If a KOBJ_ADD uevent has been 408sent for the object, a corresponding KOBJ_REMOVE uevent will be sent, and 409any other sysfs housekeeping will be handled for the caller properly. 410 411If you need to do a two-stage delete of the kobject (say you are not 412allowed to sleep when you need to destroy the object), then call 413kobject_del() which will unregister the kobject from sysfs. This makes the 414kobject "invisible", but it is not cleaned up, and the reference count of 415the object is still the same. At a later time call kobject_put() to finish 416the cleanup of the memory associated with the kobject. 417 418kobject_del() can be used to drop the reference to the parent object, if 419circular references are constructed. It is valid in some cases, that a 420parent objects references a child. Circular references _must_ be broken 421with an explicit call to kobject_del(), so that a release functions will be 422called, and the objects in the former circle release each other. 423 424 425Example code to copy from 426========================= 427 428For a more complete example of using ksets and kobjects properly, see the 429example programs samples/kobject/{kobject-example.c,kset-example.c}, 430which will be built as loadable modules if you select CONFIG_SAMPLE_KOBJECT. 431