1Copyright 2009 Jonathan Corbet <corbet@lwn.net>
2
3Debugfs exists as a simple way for kernel developers to make information
4available to user space.  Unlike /proc, which is only meant for information
5about a process, or sysfs, which has strict one-value-per-file rules,
6debugfs has no rules at all.  Developers can put any information they want
7there.  The debugfs filesystem is also intended to not serve as a stable
8ABI to user space; in theory, there are no stability constraints placed on
9files exported there.  The real world is not always so simple, though [1];
10even debugfs interfaces are best designed with the idea that they will need
11to be maintained forever.
12
13Debugfs is typically mounted with a command like:
14
15    mount -t debugfs none /sys/kernel/debug
16
17(Or an equivalent /etc/fstab line).
18
19Note that the debugfs API is exported GPL-only to modules.
20
21Code using debugfs should include <linux/debugfs.h>.  Then, the first order
22of business will be to create at least one directory to hold a set of
23debugfs files:
24
25    struct dentry *debugfs_create_dir(const char *name, struct dentry *parent);
26
27This call, if successful, will make a directory called name underneath the
28indicated parent directory.  If parent is NULL, the directory will be
29created in the debugfs root.  On success, the return value is a struct
30dentry pointer which can be used to create files in the directory (and to
31clean it up at the end).  A NULL return value indicates that something went
32wrong.  If ERR_PTR(-ENODEV) is returned, that is an indication that the
33kernel has been built without debugfs support and none of the functions
34described below will work.
35
36The most general way to create a file within a debugfs directory is with:
37
38    struct dentry *debugfs_create_file(const char *name, mode_t mode,
39				       struct dentry *parent, void *data,
40				       const struct file_operations *fops);
41
42Here, name is the name of the file to create, mode describes the access
43permissions the file should have, parent indicates the directory which
44should hold the file, data will be stored in the i_private field of the
45resulting inode structure, and fops is a set of file operations which
46implement the file's behavior.  At a minimum, the read() and/or write()
47operations should be provided; others can be included as needed.  Again,
48the return value will be a dentry pointer to the created file, NULL for
49error, or ERR_PTR(-ENODEV) if debugfs support is missing.
50
51In a number of cases, the creation of a set of file operations is not
52actually necessary; the debugfs code provides a number of helper functions
53for simple situations.  Files containing a single integer value can be
54created with any of:
55
56    struct dentry *debugfs_create_u8(const char *name, mode_t mode,
57				     struct dentry *parent, u8 *value);
58    struct dentry *debugfs_create_u16(const char *name, mode_t mode,
59				      struct dentry *parent, u16 *value);
60    struct dentry *debugfs_create_u32(const char *name, mode_t mode,
61				      struct dentry *parent, u32 *value);
62    struct dentry *debugfs_create_u64(const char *name, mode_t mode,
63				      struct dentry *parent, u64 *value);
64
65These files support both reading and writing the given value; if a specific
66file should not be written to, simply set the mode bits accordingly.  The
67values in these files are in decimal; if hexadecimal is more appropriate,
68the following functions can be used instead:
69
70    struct dentry *debugfs_create_x8(const char *name, mode_t mode,
71				     struct dentry *parent, u8 *value);
72    struct dentry *debugfs_create_x16(const char *name, mode_t mode,
73				      struct dentry *parent, u16 *value);
74    struct dentry *debugfs_create_x32(const char *name, mode_t mode,
75				      struct dentry *parent, u32 *value);
76
77Note that there is no debugfs_create_x64().
78
79These functions are useful as long as the developer knows the size of the
80value to be exported.  Some types can have different widths on different
81architectures, though, complicating the situation somewhat.  There is a
82function meant to help out in one special case:
83
84    struct dentry *debugfs_create_size_t(const char *name, mode_t mode,
85				         struct dentry *parent,
86					 size_t *value);
87
88As might be expected, this function will create a debugfs file to represent
89a variable of type size_t.
90
91Boolean values can be placed in debugfs with:
92
93    struct dentry *debugfs_create_bool(const char *name, mode_t mode,
94				       struct dentry *parent, u32 *value);
95
96A read on the resulting file will yield either Y (for non-zero values) or
97N, followed by a newline.  If written to, it will accept either upper- or
98lower-case values, or 1 or 0.  Any other input will be silently ignored.
99
100Finally, a block of arbitrary binary data can be exported with:
101
102    struct debugfs_blob_wrapper {
103	void *data;
104	unsigned long size;
105    };
106
107    struct dentry *debugfs_create_blob(const char *name, mode_t mode,
108				       struct dentry *parent,
109				       struct debugfs_blob_wrapper *blob);
110
111A read of this file will return the data pointed to by the
112debugfs_blob_wrapper structure.  Some drivers use "blobs" as a simple way
113to return several lines of (static) formatted text output.  This function
114can be used to export binary information, but there does not appear to be
115any code which does so in the mainline.  Note that all files created with
116debugfs_create_blob() are read-only.
117
118There are a couple of other directory-oriented helper functions:
119
120    struct dentry *debugfs_rename(struct dentry *old_dir,
121    				  struct dentry *old_dentry,
122		                  struct dentry *new_dir,
123				  const char *new_name);
124
125    struct dentry *debugfs_create_symlink(const char *name,
126                                          struct dentry *parent,
127				      	  const char *target);
128
129A call to debugfs_rename() will give a new name to an existing debugfs
130file, possibly in a different directory.  The new_name must not exist prior
131to the call; the return value is old_dentry with updated information.
132Symbolic links can be created with debugfs_create_symlink().
133
134There is one important thing that all debugfs users must take into account:
135there is no automatic cleanup of any directories created in debugfs.  If a
136module is unloaded without explicitly removing debugfs entries, the result
137will be a lot of stale pointers and no end of highly antisocial behavior.
138So all debugfs users - at least those which can be built as modules - must
139be prepared to remove all files and directories they create there.  A file
140can be removed with:
141
142    void debugfs_remove(struct dentry *dentry);
143
144The dentry value can be NULL, in which case nothing will be removed.
145
146Once upon a time, debugfs users were required to remember the dentry
147pointer for every debugfs file they created so that all files could be
148cleaned up.  We live in more civilized times now, though, and debugfs users
149can call:
150
151    void debugfs_remove_recursive(struct dentry *dentry);
152
153If this function is passed a pointer for the dentry corresponding to the
154top-level directory, the entire hierarchy below that directory will be
155removed.
156
157Notes:
158	[1] http://lwn.net/Articles/309298/
159