linux/Documentation/feature-removal-schedule.txt
<<
>>
Prefs
   1The following is a list of files and features that are going to be
   2removed in the kernel source tree.  Every entry should contain what
   3exactly is going away, why it is happening, and who is going to be doing
   4the work.  When the feature is removed from the kernel, it should also
   5be removed from this file.
   6
   7---------------------------
   8
   9What:   old static regulatory information and ieee80211_regdom module parameter
  10When:   2.6.29
  11Why:    The old regulatory infrastructure has been replaced with a new one
  12        which does not require statically defined regulatory domains. We do
  13        not want to keep static regulatory domains in the kernel due to the
  14        the dynamic nature of regulatory law and localization. We kept around
  15        the old static definitions for the regulatory domains of:
  16                * US
  17                * JP
  18                * EU
  19        and used by default the US when CONFIG_WIRELESS_OLD_REGULATORY was
  20        set. We also kept around the ieee80211_regdom module parameter in case
  21        some applications were relying on it. Changing regulatory domains
  22        can now be done instead by using nl80211, as is done with iw.
  23Who:    Luis R. Rodriguez <lrodriguez@atheros.com>
  24
  25---------------------------
  26
  27What:   dev->power.power_state
  28When:   July 2007
  29Why:    Broken design for runtime control over driver power states, confusing
  30        driver-internal runtime power management with:  mechanisms to support
  31        system-wide sleep state transitions; event codes that distinguish
  32        different phases of swsusp "sleep" transitions; and userspace policy
  33        inputs.  This framework was never widely used, and most attempts to
  34        use it were broken.  Drivers should instead be exposing domain-specific
  35        interfaces either to kernel or to userspace.
  36Who:    Pavel Machek <pavel@suse.cz>
  37
  38---------------------------
  39
  40What:   Video4Linux API 1 ioctls and video_decoder.h from Video devices.
  41When:   December 2008
  42Files:  include/linux/video_decoder.h include/linux/videodev.h
  43Check:  include/linux/video_decoder.h include/linux/videodev.h
  44Why:    V4L1 AP1 was replaced by V4L2 API during migration from 2.4 to 2.6
  45        series. The old API have lots of drawbacks and don't provide enough
  46        means to work with all video and audio standards. The newer API is
  47        already available on the main drivers and should be used instead.
  48        Newer drivers should use v4l_compat_translate_ioctl function to handle
  49        old calls, replacing to newer ones.
  50        Decoder iocts are using internally to allow video drivers to
  51        communicate with video decoders. This should also be improved to allow
  52        V4L2 calls being translated into compatible internal ioctls.
  53        Compatibility ioctls will be provided, for a while, via 
  54        v4l1-compat module. 
  55Who:    Mauro Carvalho Chehab <mchehab@infradead.org>
  56
  57---------------------------
  58
  59What:   PCMCIA control ioctl (needed for pcmcia-cs [cardmgr, cardctl])
  60When:   November 2005
  61Files:  drivers/pcmcia/: pcmcia_ioctl.c
  62Why:    With the 16-bit PCMCIA subsystem now behaving (almost) like a
  63        normal hotpluggable bus, and with it using the default kernel
  64        infrastructure (hotplug, driver core, sysfs) keeping the PCMCIA
  65        control ioctl needed by cardmgr and cardctl from pcmcia-cs is
  66        unnecessary, and makes further cleanups and integration of the
  67        PCMCIA subsystem into the Linux kernel device driver model more
  68        difficult. The features provided by cardmgr and cardctl are either
  69        handled by the kernel itself now or are available in the new
  70        pcmciautils package available at
  71        http://kernel.org/pub/linux/utils/kernel/pcmcia/
  72Who:    Dominik Brodowski <linux@brodo.de>
  73
  74---------------------------
  75
  76What:   sys_sysctl
  77When:   September 2010
  78Option: CONFIG_SYSCTL_SYSCALL
  79Why:    The same information is available in a more convenient from
  80        /proc/sys, and none of the sysctl variables appear to be
  81        important performance wise.
  82
  83        Binary sysctls are a long standing source of subtle kernel
  84        bugs and security issues.
  85
  86        When I looked several months ago all I could find after
  87        searching several distributions were 5 user space programs and
  88        glibc (which falls back to /proc/sys) using this syscall.
  89
  90        The man page for sysctl(2) documents it as unusable for user
  91        space programs.
  92
  93        sysctl(2) is not generally ABI compatible to a 32bit user
  94        space application on a 64bit and a 32bit kernel.
  95
  96        For the last several months the policy has been no new binary
  97        sysctls and no one has put forward an argument to use them.
  98
  99        Binary sysctls issues seem to keep happening appearing so
 100        properly deprecating them (with a warning to user space) and a
 101        2 year grace warning period will mean eventually we can kill
 102        them and end the pain.
 103
 104        In the mean time individual binary sysctls can be dealt with
 105        in a piecewise fashion.
 106
 107Who:    Eric Biederman <ebiederm@xmission.com>
 108
 109---------------------------
 110
 111What:   remove EXPORT_SYMBOL(kernel_thread)
 112When:   August 2006
 113Files:  arch/*/kernel/*_ksyms.c
 114Check:  kernel_thread
 115Why:    kernel_thread is a low-level implementation detail.  Drivers should
 116        use the <linux/kthread.h> API instead which shields them from
 117        implementation details and provides a higherlevel interface that
 118        prevents bugs and code duplication
 119Who:    Christoph Hellwig <hch@lst.de>
 120
 121---------------------------
 122
 123What:   Unused EXPORT_SYMBOL/EXPORT_SYMBOL_GPL exports
 124        (temporary transition config option provided until then)
 125        The transition config option will also be removed at the same time.
 126When:   before 2.6.19
 127Why:    Unused symbols are both increasing the size of the kernel binary
 128        and are often a sign of "wrong API"
 129Who:    Arjan van de Ven <arjan@linux.intel.com>
 130
 131---------------------------
 132
 133What:   PHYSDEVPATH, PHYSDEVBUS, PHYSDEVDRIVER in the uevent environment
 134When:   October 2008
 135Why:    The stacking of class devices makes these values misleading and
 136        inconsistent.
 137        Class devices should not carry any of these properties, and bus
 138        devices have SUBSYTEM and DRIVER as a replacement.
 139Who:    Kay Sievers <kay.sievers@suse.de>
 140
 141---------------------------
 142
 143What:   ACPI procfs interface
 144When:   July 2008
 145Why:    ACPI sysfs conversion should be finished by January 2008.
 146        ACPI procfs interface will be removed in July 2008 so that
 147        there is enough time for the user space to catch up.
 148Who:    Zhang Rui <rui.zhang@intel.com>
 149
 150---------------------------
 151
 152What:   /proc/acpi/button
 153When:   August 2007
 154Why:    /proc/acpi/button has been replaced by events to the input layer
 155        since 2.6.20.
 156Who:    Len Brown <len.brown@intel.com>
 157
 158---------------------------
 159
 160What:   /proc/acpi/event
 161When:   February 2008
 162Why:    /proc/acpi/event has been replaced by events via the input layer
 163        and netlink since 2.6.23.
 164Who:    Len Brown <len.brown@intel.com>
 165
 166---------------------------
 167
 168What: libata spindown skipping and warning
 169When: Dec 2008
 170Why:  Some halt(8) implementations synchronize caches for and spin
 171      down libata disks because libata didn't use to spin down disk on
 172      system halt (only synchronized caches).
 173      Spin down on system halt is now implemented.  sysfs node
 174      /sys/class/scsi_disk/h:c:i:l/manage_start_stop is present if
 175      spin down support is available.
 176      Because issuing spin down command to an already spun down disk
 177      makes some disks spin up just to spin down again, libata tracks
 178      device spindown status to skip the extra spindown command and
 179      warn about it.
 180      This is to give userspace tools the time to get updated and will
 181      be removed after userspace is reasonably updated.
 182Who:  Tejun Heo <htejun@gmail.com>
 183
 184---------------------------
 185
 186What:   i386/x86_64 bzImage symlinks
 187When:   April 2010
 188
 189Why:    The i386/x86_64 merge provides a symlink to the old bzImage
 190        location so not yet updated user space tools, e.g. package
 191        scripts, do not break.
 192Who:    Thomas Gleixner <tglx@linutronix.de>
 193
 194---------------------------
 195
 196What (Why):
 197        - include/linux/netfilter_ipv4/ipt_TOS.h ipt_tos.h header files
 198          (superseded by xt_TOS/xt_tos target & match)
 199
 200        - "forwarding" header files like ipt_mac.h in
 201          include/linux/netfilter_ipv4/ and include/linux/netfilter_ipv6/
 202
 203        - xt_CONNMARK match revision 0
 204          (superseded by xt_CONNMARK match revision 1)
 205
 206        - xt_MARK target revisions 0 and 1
 207          (superseded by xt_MARK match revision 2)
 208
 209        - xt_connmark match revision 0
 210          (superseded by xt_connmark match revision 1)
 211
 212        - xt_conntrack match revision 0
 213          (superseded by xt_conntrack match revision 1)
 214
 215        - xt_iprange match revision 0,
 216          include/linux/netfilter_ipv4/ipt_iprange.h
 217          (superseded by xt_iprange match revision 1)
 218
 219        - xt_mark match revision 0
 220          (superseded by xt_mark match revision 1)
 221
 222        - xt_recent: the old ipt_recent proc dir
 223          (superseded by /proc/net/xt_recent)
 224
 225When:   January 2009 or Linux 2.7.0, whichever comes first
 226Why:    Superseded by newer revisions or modules
 227Who:    Jan Engelhardt <jengelh@computergmbh.de>
 228
 229---------------------------
 230
 231What:   b43 support for firmware revision < 410
 232When:   July 2008
 233Why:    The support code for the old firmware hurts code readability/maintainability
 234        and slightly hurts runtime performance. Bugfixes for the old firmware
 235        are not provided by Broadcom anymore.
 236Who:    Michael Buesch <mb@bu3sch.de>
 237
 238---------------------------
 239
 240What:   usedac i386 kernel parameter
 241When:   2.6.27
 242Why:    replaced by allowdac and no dac combination
 243Who:    Glauber Costa <gcosta@redhat.com>
 244
 245---------------------------
 246
 247What:   remove HID compat support
 248When:   2.6.29
 249Why:    needed only as a temporary solution until distros fix themselves up
 250Who:    Jiri Slaby <jirislaby@gmail.com>
 251
 252---------------------------
 253
 254What: print_fn_descriptor_symbol()
 255When: October 2009
 256Why:  The %pF vsprintf format provides the same functionality in a
 257      simpler way.  print_fn_descriptor_symbol() is deprecated but
 258      still present to give out-of-tree modules time to change.
 259Who:  Bjorn Helgaas <bjorn.helgaas@hp.com>
 260
 261---------------------------
 262
 263What:   /sys/o2cb symlink
 264When:   January 2010
 265Why:    /sys/fs/o2cb is the proper location for this information - /sys/o2cb
 266        exists as a symlink for backwards compatibility for old versions of
 267        ocfs2-tools. 2 years should be sufficient time to phase in new versions
 268        which know to look in /sys/fs/o2cb.
 269Who:    ocfs2-devel@oss.oracle.com
 270
 271---------------------------
 272
 273What:   SCTP_GET_PEER_ADDRS_NUM_OLD, SCTP_GET_PEER_ADDRS_OLD,
 274        SCTP_GET_LOCAL_ADDRS_NUM_OLD, SCTP_GET_LOCAL_ADDRS_OLD
 275When:   June 2009
 276Why:    A newer version of the options have been introduced in 2005 that
 277        removes the limitions of the old API.  The sctp library has been
 278        converted to use these new options at the same time.  Any user
 279        space app that directly uses the old options should convert to using
 280        the new options.
 281Who:    Vlad Yasevich <vladislav.yasevich@hp.com>
 282
 283---------------------------
 284
 285What:   CONFIG_THERMAL_HWMON
 286When:   January 2009
 287Why:    This option was introduced just to allow older lm-sensors userspace
 288        to keep working over the upgrade to 2.6.26. At the scheduled time of
 289        removal fixed lm-sensors (2.x or 3.x) should be readily available.
 290Who:    Rene Herman <rene.herman@gmail.com>
 291
 292---------------------------
 293
 294What:   Code that is now under CONFIG_WIRELESS_EXT_SYSFS
 295        (in net/core/net-sysfs.c)
 296When:   After the only user (hal) has seen a release with the patches
 297        for enough time, probably some time in 2010.
 298Why:    Over 1K .text/.data size reduction, data is available in other
 299        ways (ioctls)
 300Who:    Johannes Berg <johannes@sipsolutions.net>
 301
 302---------------------------
 303
 304What: CONFIG_NF_CT_ACCT
 305When: 2.6.29
 306Why:  Accounting can now be enabled/disabled without kernel recompilation.
 307      Currently used only to set a default value for a feature that is also
 308      controlled by a kernel/module/sysfs/sysctl parameter.
 309Who:  Krzysztof Piotr Oledzki <ole@ans.pl>
 310
 311---------------------------
 312
 313What:   i2c_attach_client(), i2c_detach_client(), i2c_driver->detach_client()
 314When:   2.6.29 (ideally) or 2.6.30 (more likely)
 315Why:    Deprecated by the new (standard) device driver binding model. Use
 316        i2c_driver->probe() and ->remove() instead.
 317Who:    Jean Delvare <khali@linux-fr.org>
 318
 319---------------------------
 320
 321What:   fscher and fscpos drivers
 322When:   June 2009
 323Why:    Deprecated by the new fschmd driver.
 324Who:    Hans de Goede <hdegoede@redhat.com>
 325        Jean Delvare <khali@linux-fr.org>
 326
 327---------------------------
 328
 329What:   SELinux "compat_net" functionality
 330When:   2.6.30 at the earliest
 331Why:    In 2.6.18 the Secmark concept was introduced to replace the "compat_net"
 332        network access control functionality of SELinux.  Secmark offers both
 333        better performance and greater flexibility than the "compat_net"
 334        mechanism.  Now that the major Linux distributions have moved to
 335        Secmark, it is time to deprecate the older mechanism and start the
 336        process of removing the old code.
 337Who:    Paul Moore <paul.moore@hp.com>
 338---------------------------
 339
 340What:   sysfs ui for changing p4-clockmod parameters
 341When:   September 2009
 342Why:    See commits 129f8ae9b1b5be94517da76009ea956e89104ce8 and
 343        e088e4c9cdb618675874becb91b2fd581ee707e6.
 344        Removal is subject to fixing any remaining bugs in ACPI which may
 345        cause the thermal throttling not to happen at the right time.
 346Who:    Dave Jones <davej@redhat.com>, Matthew Garrett <mjg@redhat.com>
 347