linux/Documentation/stable_api_nonsense.txt
<<
>>
Prefs
   1.. _stable_api_nonsense:
   2
   3The Linux Kernel Driver Interface
   4==================================
   5
   6(all of your questions answered and then some)
   7
   8Greg Kroah-Hartman <greg@kroah.com>
   9
  10This is being written to try to explain why Linux **does not have a binary
  11kernel interface, nor does it have a stable kernel interface**.
  12
  13.. note::
  14
  15  Please realize that this article describes the **in kernel** interfaces, not
  16  the kernel to userspace interfaces.
  17
  18  The kernel to userspace interface is the one that application programs use,
  19  the syscall interface.  That interface is **very** stable over time, and
  20  will not break.  I have old programs that were built on a pre 0.9something
  21  kernel that still work just fine on the latest 2.6 kernel release.
  22  That interface is the one that users and application programmers can count
  23  on being stable.
  24
  25
  26Executive Summary
  27-----------------
  28You think you want a stable kernel interface, but you really do not, and
  29you don't even know it.  What you want is a stable running driver, and
  30you get that only if your driver is in the main kernel tree.  You also
  31get lots of other good benefits if your driver is in the main kernel
  32tree, all of which has made Linux into such a strong, stable, and mature
  33operating system which is the reason you are using it in the first
  34place.
  35
  36
  37Intro
  38-----
  39
  40It's only the odd person who wants to write a kernel driver that needs
  41to worry about the in-kernel interfaces changing.  For the majority of
  42the world, they neither see this interface, nor do they care about it at
  43all.
  44
  45First off, I'm not going to address **any** legal issues about closed
  46source, hidden source, binary blobs, source wrappers, or any other term
  47that describes kernel drivers that do not have their source code
  48released under the GPL.  Please consult a lawyer if you have any legal
  49questions, I'm a programmer and hence, I'm just going to be describing
  50the technical issues here (not to make light of the legal issues, they
  51are real, and you do need to be aware of them at all times.)
  52
  53So, there are two main topics here, binary kernel interfaces and stable
  54kernel source interfaces.  They both depend on each other, but we will
  55discuss the binary stuff first to get it out of the way.
  56
  57
  58Binary Kernel Interface
  59-----------------------
  60Assuming that we had a stable kernel source interface for the kernel, a
  61binary interface would naturally happen too, right?  Wrong.  Please
  62consider the following facts about the Linux kernel:
  63
  64  - Depending on the version of the C compiler you use, different kernel
  65    data structures will contain different alignment of structures, and
  66    possibly include different functions in different ways (putting
  67    functions inline or not.)  The individual function organization
  68    isn't that important, but the different data structure padding is
  69    very important.
  70
  71  - Depending on what kernel build options you select, a wide range of
  72    different things can be assumed by the kernel:
  73
  74      - different structures can contain different fields
  75      - Some functions may not be implemented at all, (i.e. some locks
  76        compile away to nothing for non-SMP builds.)
  77      - Memory within the kernel can be aligned in different ways,
  78        depending on the build options.
  79
  80  - Linux runs on a wide range of different processor architectures.
  81    There is no way that binary drivers from one architecture will run
  82    on another architecture properly.
  83
  84Now a number of these issues can be addressed by simply compiling your
  85module for the exact specific kernel configuration, using the same exact
  86C compiler that the kernel was built with.  This is sufficient if you
  87want to provide a module for a specific release version of a specific
  88Linux distribution.  But multiply that single build by the number of
  89different Linux distributions and the number of different supported
  90releases of the Linux distribution and you quickly have a nightmare of
  91different build options on different releases.  Also realize that each
  92Linux distribution release contains a number of different kernels, all
  93tuned to different hardware types (different processor types and
  94different options), so for even a single release you will need to create
  95multiple versions of your module.
  96
  97Trust me, you will go insane over time if you try to support this kind
  98of release, I learned this the hard way a long time ago...
  99
 100
 101Stable Kernel Source Interfaces
 102-------------------------------
 103
 104This is a much more "volatile" topic if you talk to people who try to
 105keep a Linux kernel driver that is not in the main kernel tree up to
 106date over time.
 107
 108Linux kernel development is continuous and at a rapid pace, never
 109stopping to slow down.  As such, the kernel developers find bugs in
 110current interfaces, or figure out a better way to do things.  If they do
 111that, they then fix the current interfaces to work better.  When they do
 112so, function names may change, structures may grow or shrink, and
 113function parameters may be reworked.  If this happens, all of the
 114instances of where this interface is used within the kernel are fixed up
 115at the same time, ensuring that everything continues to work properly.
 116
 117As a specific examples of this, the in-kernel USB interfaces have
 118undergone at least three different reworks over the lifetime of this
 119subsystem.  These reworks were done to address a number of different
 120issues:
 121
 122  - A change from a synchronous model of data streams to an asynchronous
 123    one.  This reduced the complexity of a number of drivers and
 124    increased the throughput of all USB drivers such that we are now
 125    running almost all USB devices at their maximum speed possible.
 126  - A change was made in the way data packets were allocated from the
 127    USB core by USB drivers so that all drivers now needed to provide
 128    more information to the USB core to fix a number of documented
 129    deadlocks.
 130
 131This is in stark contrast to a number of closed source operating systems
 132which have had to maintain their older USB interfaces over time.  This
 133provides the ability for new developers to accidentally use the old
 134interfaces and do things in improper ways, causing the stability of the
 135operating system to suffer.
 136
 137In both of these instances, all developers agreed that these were
 138important changes that needed to be made, and they were made, with
 139relatively little pain.  If Linux had to ensure that it will preserve a
 140stable source interface, a new interface would have been created, and
 141the older, broken one would have had to be maintained over time, leading
 142to extra work for the USB developers.  Since all Linux USB developers do
 143their work on their own time, asking programmers to do extra work for no
 144gain, for free, is not a possibility.
 145
 146Security issues are also very important for Linux.  When a
 147security issue is found, it is fixed in a very short amount of time.  A
 148number of times this has caused internal kernel interfaces to be
 149reworked to prevent the security problem from occurring.  When this
 150happens, all drivers that use the interfaces were also fixed at the
 151same time, ensuring that the security problem was fixed and could not
 152come back at some future time accidentally.  If the internal interfaces
 153were not allowed to change, fixing this kind of security problem and
 154insuring that it could not happen again would not be possible.
 155
 156Kernel interfaces are cleaned up over time.  If there is no one using a
 157current interface, it is deleted.  This ensures that the kernel remains
 158as small as possible, and that all potential interfaces are tested as
 159well as they can be (unused interfaces are pretty much impossible to
 160test for validity.)
 161
 162
 163What to do
 164----------
 165
 166So, if you have a Linux kernel driver that is not in the main kernel
 167tree, what are you, a developer, supposed to do?  Releasing a binary
 168driver for every different kernel version for every distribution is a
 169nightmare, and trying to keep up with an ever changing kernel interface
 170is also a rough job.
 171
 172Simple, get your kernel driver into the main kernel tree (remember we
 173are talking about GPL released drivers here, if your code doesn't fall
 174under this category, good luck, you are on your own here, you leech
 175<insert link to leech comment from Andrew and Linus here>.)  If your
 176driver is in the tree, and a kernel interface changes, it will be fixed
 177up by the person who did the kernel change in the first place.  This
 178ensures that your driver is always buildable, and works over time, with
 179very little effort on your part.
 180
 181The very good side effects of having your driver in the main kernel tree
 182are:
 183
 184  - The quality of the driver will rise as the maintenance costs (to the
 185    original developer) will decrease.
 186  - Other developers will add features to your driver.
 187  - Other people will find and fix bugs in your driver.
 188  - Other people will find tuning opportunities in your driver.
 189  - Other people will update the driver for you when external interface
 190    changes require it.
 191  - The driver automatically gets shipped in all Linux distributions
 192    without having to ask the distros to add it.
 193
 194As Linux supports a larger number of different devices "out of the box"
 195than any other operating system, and it supports these devices on more
 196different processor architectures than any other operating system, this
 197proven type of development model must be doing something right :)
 198
 199
 200
 201------
 202
 203Thanks to Randy Dunlap, Andrew Morton, David Brownell, Hanna Linder,
 204Robert Love, and Nishanth Aravamudan for their review and comments on
 205early drafts of this paper.
 206