summaryrefslogtreecommitdiff
path: root/lib/kobject.c
AgeCommit message (Collapse)Author
2013-05-07kref: minor cleanupAnatol Pomozov
- make warning smp-safe - result of atomic _unless_zero functions should be checked by caller to avoid use-after-free error - trivial whitespace fix. Link: https://lkml.org/lkml/2013/4/12/391 Tested: compile x86, boot machine and run xfstests Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com> [ Removed line-break, changed to use WARN_ON_ONCE() - Linus ] Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-13kobject: fix kset_find_obj() race with concurrent last kobject_put()Linus Torvalds
Anatol Pomozov identified a race condition that hits module unloading and re-loading. To quote Anatol: "This is a race codition that exists between kset_find_obj() and kobject_put(). kset_find_obj() might return kobject that has refcount equal to 0 if this kobject is freeing by kobject_put() in other thread. Here is timeline for the crash in case if kset_find_obj() searches for an object tht nobody holds and other thread is doing kobject_put() on the same kobject: THREAD A (calls kset_find_obj()) THREAD B (calls kobject_put()) splin_lock() atomic_dec_return(kobj->kref), counter gets zero here ... starts kobject cleanup .... spin_lock() // WAIT thread A in kobj_kset_leave() iterate over kset->list atomic_inc(kobj->kref) (counter becomes 1) spin_unlock() spin_lock() // taken // it does not know that thread A increased counter so it remove obj from list spin_unlock() vfree(module) // frees module object with containing kobj // kobj points to freed memory area!! kobject_put(kobj) // OOPS!!!! The race above happens because module.c tries to use kset_find_obj() when somebody unloads module. The module.c code was introduced in commit 6494a93d55fa" Anatol supplied a patch specific for module.c that worked around the problem by simply not using kset_find_obj() at all, but rather than make a local band-aid, this just fixes kset_find_obj() to be thread-safe using the proper model of refusing the get a new reference if the refcount has already dropped to zero. See examples of this proper refcount handling not only in the kref documentation, but in various other equivalent uses of this pattern by grepping for atomic_inc_not_zero(). [ Side note: the module race does indicate that module loading and unloading is not properly serialized wrt sysfs information using the module mutex. That may require further thought, but this is the correct fix at the kobject layer regardless. ] Reported-analyzed-and-tested-by: Anatol Pomozov <anatol.pomozov@gmail.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: stable@vger.kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-05-07kobject: fix the uncorrect commentZhi Yong Wu
Signed-off-by: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-05-02Merge 3.4-rc5 into driver-core-nextGreg Kroah-Hartman
This was done to resolve a merge issue with the init/main.c file. Reported-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-04-23lib/kobject.c : Remove redundant check in create_diryan
create_dir is a static function used only in kobject_add_internal. There's no need to do check here, for kobject_add_internal will reject kobject with invalid name. Signed-off-by: Yan Hong <clouds.yan@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-04-10kobject: provide more diagnostic info for kobject_add_internal() failuresDan Williams
1/ convert open-coded KERN_ERR+dump_stack() to WARN(), so that automated tools pick up this warning. 2/ include the 'child' and 'parent' kobject names. This information was useful for tracking down the case where scsi invoked device_del() on a parent object and subsequently invoked device_add() on a child. Now the warning looks like: kobject_add_internal failed for target8:0:16 (error: -2 parent: end_device-8:0:24) Pid: 2942, comm: scsi_scan_8 Not tainted 3.3.0-rc7-isci+ #2 Call Trace: [<ffffffff8125e551>] kobject_add_internal+0x1c1/0x1f3 [<ffffffff81075149>] ? trace_hardirqs_on+0xd/0xf [<ffffffff8125e659>] kobject_add_varg+0x41/0x50 [<ffffffff8125e723>] kobject_add+0x64/0x66 [<ffffffff8131124b>] device_add+0x12d/0x63a [<ffffffff8125e0ef>] ? kobject_put+0x4c/0x50 [<ffffffff8132f370>] scsi_sysfs_add_sdev+0x4e/0x28a [<ffffffff8132dce3>] do_scan_async+0x9c/0x145 Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: James Bottomley <JBottomley@parallels.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-03-07lib: reduce the use of module.h wherever possiblePaul Gortmaker
For files only using THIS_MODULE and/or EXPORT_SYMBOL, map them onto including export.h -- or if the file isn't even using those, then just delete the include. Fix up any implicit include dependencies that were being masked by module.h along the way. Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-12-21kobject: remove kset_find_obj_hinted()Kay Sievers
Now that there are no in-kernel users of this function, remove it as it is no longer needed. Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2011-06-12Delay struct net freeing while there's a sysfs instance refering to itAl Viro
* new refcount in struct net, controlling actual freeing of the memory * new method in kobj_ns_type_operations (->drop_ns()) * ->current_ns() semantics change - it's supposed to be followed by corresponding ->drop_ns(). For struct net in case of CONFIG_NET_NS it bumps the new refcount; net_drop_ns() decrements it and calls net_free() if the last reference has been dropped. Method renamed to ->grab_current_ns(). * old net_free() callers call net_drop_ns() instead. * sysfs_exit_ns() is gone, along with a large part of callchain leading to it; now that the references stored in ->ns[...] stay valid we do not need to hunt them down and replace them with NULL. That fixes problems in sysfs_lookup() and sysfs_readdir(), along with getting rid of sb->s_instances abuse. Note that struct net *shutdown* logics has not changed - net_cleanup() is called exactly when it used to be called. The only thing postponed by having a sysfs instance refering to that struct net is actual freeing of memory occupied by struct net. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2010-10-22kobject: Introduce kset_find_obj_hinted.Robin Holt
One call chain getting to kset_find_obj is: link_mem_sections() find_mem_section() kset_find_obj() This is done during boot. The memory sections were added in a linearly increasing order and link_mem_sections tends to utilize them in that same linear order. Introduce a kset_find_obj_hinted which is passed the result of the previous kset_find_obj which it uses for a quick "is the next object our desired object" check before falling back to the old behavior. Signed-off-by: Robin Holt <holt@sgi.com> To: Robert P. J. Day <rpjday@crashcourse.ca> Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-05-21sysfs: Comment sysfs directory tagging logicSerge E. Hallyn
Add some in-line comments to explain the new infrastructure, which was introduced to support sysfs directory tagging with namespaces. I think an overall description someplace might be good too, but it didn't really seem to fit into Documentation/filesystems/sysfs.txt, which appears more geared toward users, rather than maintainers, of sysfs. (Tejun, please let me know if I can make anything clearer or failed altogether to comment something that should be commented.) Signed-off-by: Serge E. Hallyn <serue@us.ibm.com> Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-05-21sysfs: Implement sysfs tagged directory support.Eric W. Biederman
The problem. When implementing a network namespace I need to be able to have multiple network devices with the same name. Currently this is a problem for /sys/class/net/*, /sys/devices/virtual/net/*, and potentially a few other directories of the form /sys/ ... /net/*. What this patch does is to add an additional tag field to the sysfs dirent structure. For directories that should show different contents depending on the context such as /sys/class/net/, and /sys/devices/virtual/net/ this tag field is used to specify the context in which those directories should be visible. Effectively this is the same as creating multiple distinct directories with the same name but internally to sysfs the result is nicer. I am calling the concept of a single directory that looks like multiple directories all at the same path in the filesystem tagged directories. For the networking namespace the set of directories whose contents I need to filter with tags can depend on the presence or absence of hotplug hardware or which modules are currently loaded. Which means I need a simple race free way to setup those directories as tagged. To achieve a reace free design all tagged directories are created and managed by sysfs itself. Users of this interface: - define a type in the sysfs_tag_type enumeration. - call sysfs_register_ns_types with the type and it's operations - sysfs_exit_ns when an individual tag is no longer valid - Implement mount_ns() which returns the ns of the calling process so we can attach it to a sysfs superblock. - Implement ktype.namespace() which returns the ns of a syfs kobject. Everything else is left up to sysfs and the driver layer. For the network namespace mount_ns and namespace() are essentially one line functions, and look to remain that. Tags are currently represented a const void * pointers as that is both generic, prevides enough information for equality comparisons, and is trivial to create for current users, as it is just the existing namespace pointer. The work needed in sysfs is more extensive. At each directory or symlink creating I need to check if the directory it is being created in is a tagged directory and if so generate the appropriate tag to place on the sysfs_dirent. Likewise at each symlink or directory removal I need to check if the sysfs directory it is being removed from is a tagged directory and if so figure out which tag goes along with the name I am deleting. Currently only directories which hold kobjects, and symlinks are supported. There is not enough information in the current file attribute interfaces to give us anything to discriminate on which makes it useless, and there are no potential users which makes it an uninteresting problem to solve. Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Benjamin Thery <benjamin.thery@bull.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-05-21kobj: Add basic infrastructure for dealing with namespaces.Eric W. Biederman
Move complete knowledge of namespaces into the kobject layer so we can use that information when reporting kobjects to userspace. Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-03-07Driver core: Constify struct sysfs_ops in struct kobj_typeEmese Revfy
Constify struct sysfs_ops. This is part of the ops structure constification effort started by Arjan van de Ven et al. Benefits of this constification: * prevents modification of data that is shared (referenced) by many other structure instances at runtime * detects/prevents accidental (but not intentional) modification attempts on archs that enforce read-only kernel data at runtime * potentially better optimized code as the compiler can assume that the const data cannot be changed * the compiler/linker move const data into .rodata and therefore exclude them from false sharing Signed-off-by: Emese Revfy <re.emese@gmail.com> Acked-by: David Teigland <teigland@redhat.com> Acked-by: Matt Domsch <Matt_Domsch@dell.com> Acked-by: Maciej Sosnowski <maciej.sosnowski@intel.com> Acked-by: Hans J. Koch <hjk@linutronix.de> Acked-by: Pekka Enberg <penberg@cs.helsinki.fi> Acked-by: Jens Axboe <jens.axboe@oracle.com> Acked-by: Stephen Hemminger <shemminger@vyatta.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-03-07kobject: Constify struct kset_uevent_opsEmese Revfy
Constify struct kset_uevent_ops. This is part of the ops structure constification effort started by Arjan van de Ven et al. Benefits of this constification: * prevents modification of data that is shared (referenced) by many other structure instances at runtime * detects/prevents accidental (but not intentional) modification attempts on archs that enforce read-only kernel data at runtime * potentially better optimized code as the compiler can assume that the const data cannot be changed * the compiler/linker move const data into .rodata and therefore exclude them from false sharing Signed-off-by: Emese Revfy <re.emese@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2009-06-15kobject: make kset_create check kobject_set_name return valueDave Young
kset_create should check the kobject_set_name return value. Add the return value checking code. Signed-off-by: Dave Young <hidave.darkstar@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2009-04-20driver: dont update dev_name via device_add pathKay Sievers
notice one system /proc/iomem some entries missed the name for pci_devices it turns that dev->dev.kobj name is changed after device_add. for pci code: via acpi_pci_root_driver.ops.add (aka acpi_pci_root_add) ==> pci_acpi_scan_root is used to scan pci bus/device, and at the same time we read the resource for pci_dev in the pci_read_bases, we have res->name = pci_name(pci_dev); pci_name is calling dev_name. later via acpi_pci_root_driver.ops.start (aka acpi_pci_root_start) ==> pci_bus_add_device to add all pci_dev in kobj tree. pci_bus_add_device will call device_add. actually in device_add /* first, register with generic layer. */ error = kobject_add(&dev->kobj, dev->kobj.parent, "%s", dev_name(dev)); if (error) goto Error; will get one new name for that kobj, old name is freed. [Impact: fix corrupted names in /proc/iomem ] Signed-off-by: Yinghai Lu <yinghai@kernel.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-03-24driver core: get rid of struct device's bus_id string arrayKay Sievers
Now that all users of bus_id is gone, we can remove it from struct device. Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-10-16kobject: Cleanup kobject_rename and !CONFIG_SYSFSEric W. Biederman
It finally dawned on me what the clean fix to sysfs_rename_dir calling kobject_set_name is. Move the work into kobject_rename where it belongs. The callers serialize us anyway so this is safe. Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Acked-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-10-16kobject: Fix kobject_rename and !CONFIG_SYSFSEric W. Biederman
When looking at kobject_rename I found two bugs with that exist when sysfs support is disabled in the kernel. kobject_rename does not change the name on the kobject when sysfs support is not compiled in. kobject_rename without locking attempts to check the validity of a rename operation, which the kobject layer simply does not have the infrastructure to do. This patch documents the previously unstated requirement of kobject_rename that is the responsibility of the caller to provide mutual exclusion and to be certain that the new_name for the kobject is valid. This patch modifies sysfs_rename_dir in !CONFIG_SYSFS case to call kobject_set_name to actually change the kobject_name. This patch removes the bogus and misleading check in kobject_rename that attempts to see if a rename is valid. The check is bogus because we do not have the proper locking. The check is misleading because it looks like we can and do perform checking at the kobject level that we don't. Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-08-21kobject: Replace ALL occurrences of '/' with '!' instead of only the first one.Ingo Oeser
A recent patch from Kay Sievers <kay.sievers@vrfy.org> replaced the first occurrence of '/' with '!' as needed for block devices. Now do some cheap defensive coding and replace all of them to avoid future issues in this area. Signed-off-by: Ingo Oeser <ioe-lkml@rameria.de> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-07-25Example use of WARN()Arjan van de Ven
Now that WARN() exists, we can fold some of the printk's into it. Signed-off-by: Arjan van de Ven <arjan@linux.intel.com> Cc: Greg KH <greg@kroah.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-21kobject: replace '/' with '!' in nameKay Sievers
Some (block) devices have a '/' in the name, and need special handling. Let's have that rule to the core, so we can remove it from the block class. Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-06-10PCI: ACPI PCI slot detection driverAlex Chiang
Detect all physical PCI slots as described by ACPI, and create entries in /sys/bus/pci/slots/. Not all physical slots are hotpluggable, and the acpiphp module does not detect them. Now we know the physical PCI geography of our system, without caring about hotplug. [kaneshige.kenji@jp.fujitsu.com: export-kobject_rename-for-pci_hotplug_core] Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Acked-by: Greg KH <greg@kroah.com> [akpm@linux-foundation.org: build fix] [akpm@linux-foundation.org: fix build with CONFIG_DMI=n] Signed-off-by: Alex Chiang <achiang@hp.com> Cc: Greg KH <greg@kroah.com> Cc: Kristen Carlson Accardi <kristen.c.accardi@intel.com> Cc: Len Brown <lenb@kernel.org> Acked-by: Len Brown <len.brown@intel.com> Acked-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
2008-04-30kobject: do not copy vargs, just pass them aroundKay Sievers
This prevents a few unneeded copies. Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-04-30lib: replace remaining __FUNCTION__ occurrencesHarvey Harrison
__FUNCTION__ is gcc specific, use __func__ Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-19Kobject: Replace list_for_each() with list_for_each_entry().Robert P. J. Day
Use the more concise list_for_each_entry(), which allows for the deletion of the to_kobj() routine at the same time. Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-04-19kobject: catch kobjects that are not initializedGreg Kroah-Hartman
Add warnings to kobject_put() to catch kobjects that are cleaned up but were never initialized to begin with. Cc: Kay Sievers <kay.sievers@vrfy.org> Cc: Hannes Reinecke <hare@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-03-04kobject: properly initialize ksetsGreg Kroah-Hartman
kset_initialize was calling kobject_init_internal() which didn't initialize the kobject as well as kobject_init() was. So have kobject_init() call kobject_init_internal() and move the logic to initalize the kobject there. Cc: Kay Sievers <kay.sievers@vrfy.org> Cc: Hannes Reinecke <hare@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-02-02kobject: kerneldoc comment fixDave Young
Fix kerneldoc comment of kobject_create. Signed-off-by: Dave Young <hidave.darkstar@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24Kobject: fix coding style issues in kobject c filesGreg Kroah-Hartman
Clean up the kobject.c and kobject_uevent.c files to follow the proper coding style rules. Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24Driver core: change sysdev classes to use dynamic kobject namesKay Sievers
All kobjects require a dynamically allocated name now. We no longer need to keep track if the name is statically assigned, we can just unconditionally free() all kobject names on cleanup. Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24Kobject: remove kobject_unregister() as no one uses it anymoreGreg Kroah-Hartman
There are no in-kernel users of kobject_unregister() so it should be removed. Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24Kobject: convert remaining kobject_unregister() to kobject_put()Greg Kroah-Hartman
There is no need for kobject_unregister() anymore, thanks to Kay's kobject cleanup changes, so replace all instances of it with kobject_put(). Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24Kobject: auto-cleanup on final unrefKay Sievers
We save the current state in the object itself, so we can do proper cleanup when the last reference is dropped. If the initial reference is dropped, the object will be removed from sysfs if needed, if an "add" event was sent, "remove" will be send, and the allocated resources are released. This allows us to clean up some driver core usage as well as allowing us to do other such changes to the rest of the kernel. Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24Kset: remove kset_add functionGreg Kroah-Hartman
No one is calling this anymore, so just remove it and hard-code the one internal-use of it. Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24Kobject: remove kobject_register()Greg Kroah-Hartman
The function is no longer used by anyone in the kernel, and it prevents the proper sending of the kobject uevent after the needed files are set up by the caller. kobject_init_and_add() can be used in its place. Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24Kobject: rename kobject_init_ng() to kobject_init()Greg Kroah-Hartman
Now that the old kobject_init() function is gone, rename kobject_init_ng() to kobject_init() to clean up the namespace. Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24Kobject: remove kobject_init() as no one uses it anymoreGreg Kroah-Hartman
The old kobject_init() function is on longer in use, so let us remove it from the public scope (kset mess in the kobject.c file still uses it, but that can be cleaned up later very simply.) Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24Kobject: rename kobject_add_ng() to kobject_add()Greg Kroah-Hartman
Now that the old kobject_add() function is gone, rename kobject_add_ng() to kobject_add() to clean up the namespace. Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24Kobject: remove kobject_add() as no one uses it anymoreGreg Kroah-Hartman
The old kobject_add() function is on longer in use, so let us remove it from the public scope (kset mess in the kobject.c file still uses it, but that can be cleaned up later very simply.) Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24Kobject: drop child->parent ref at unregistrationAlan Stern
This patch (as1015) reverts changes that were made to the driver core about four years ago. The intent back then was to avoid certain kinds of invalid memory accesses by leaving kernel objects allocated as long as any of their children were still allocated. The original and correct approach was to wait only as long as any children were still _registered_; that's what this patch reinstates. This fixes a problem in the SCSI core made visible by the class_device to regular device conversion: A reference loop (scsi_device holds reference to request_queue, which is the child of a gendisk, which is the child of the scsi_device) prevents the data structures from being released, even though they are deregistered okay. It's possible that this change will cause a few bugs to surface, things that have been hidden for several years. They can be fixed easily enough by having the child device take an explicit reference to the parent whenever needed. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24kobject: clean up debugging messagesGreg Kroah-Hartman
The kobject debugging messages are a mess. This provides a unified message that makes them actually useful. The format for new kobject debug messages should be: kobject: 'KOBJECT_NAME' (ADDRESS): FUNCTION_NAME: message.\n Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24kobject: grab the kset reference in kobject_add, not kobject_initGreg Kroah-Hartman
kobject_init should not be grabing any references, but only initializing the object. This patch fixes this, and makes the lock hold-time shorter for when a kset is present in the kobject. The current kernel tree has been audited to verify that this change should be safe. Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24kobject: remove subsystem_(un)register functionsGreg Kroah-Hartman
These functions are no longer used and are the last remants of the old subsystem crap. So delete them for good. Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24Driver Core: kill subsys_attribute and default sysfs opsKay Sievers
Remove the no longer needed subsys_attributes, they are all converted to the more sensical kobj_attributes. There is no longer a magic fallback in sysfs attribute operations, all kobjects which create simple attributes need explicitely a ktype assigned, which tells the core what was intended here. Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24Driver Core: switch all dynamic ksets to kobj_sysfs_opsKay Sievers
Switch all dynamically created ksets, that export simple attributes, to kobj_attribute from subsys_attribute. Struct subsys_attribute will be removed. Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Cc: Mike Halcrow <mhalcrow@us.ibm.com> Cc: Phillip Hellewell <phillip@hellewell.homeip.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24Driver Core: add kobj_attribute handlingKay Sievers
Add kobj_sysfs_ops to replace subsys_sysfs_ops. There is no need for special kset operations, we want to be able to use simple attribute operations at any kobject, not only ksets. The whole concept of any default sysfs attribute operations will go away with the upcoming removal of subsys_sysfs_ops. Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24kobject: get rid of kobject_kset_add_dirGreg Kroah-Hartman
kobject_kset_add_dir is only called in one place so remove it and use kobject_create() instead. Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-24kobject: get rid of kobject_add_dirGreg Kroah-Hartman
kobject_create_and_add is the same as kobject_add_dir, so drop kobject_add_dir. Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>