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:   x86 floppy disable_hlt
  10When:   2012
  11Why:    ancient workaround of dubious utility clutters the
  12        code used by everybody else.
  13Who:    Len Brown <len.brown@intel.com>
  14
  15---------------------------
  16
  17What:   CONFIG_APM_CPU_IDLE, and its ability to call APM BIOS in idle
  18When:   2012
  19Why:    This optional sub-feature of APM is of dubious reliability,
  20        and ancient APM laptops are likely better served by calling HLT.
  21        Deleting CONFIG_APM_CPU_IDLE allows x86 to stop exporting
  22        the pm_idle function pointer to modules.
  23Who:    Len Brown <len.brown@intel.com>
  24
  25----------------------------
  26
  27What:   x86_32 "no-hlt" cmdline param
  28When:   2012
  29Why:    remove a branch from idle path, simplify code used by everybody.
  30        This option disabled the use of HLT in idle and machine_halt()
  31        for hardware that was flakey 15-years ago.  Today we have
  32        "idle=poll" that removed HLT from idle, and so if such a machine
  33        is still running the upstream kernel, "idle=poll" is likely sufficient.
  34Who:    Len Brown <len.brown@intel.com>
  35
  36----------------------------
  37
  38What:   x86 "idle=mwait" cmdline param
  39When:   2012
  40Why:    simplify x86 idle code
  41Who:    Len Brown <len.brown@intel.com>
  42
  43----------------------------
  44
  45What:   PRISM54
  46When:   2.6.34
  47
  48Why:    prism54 FullMAC PCI / Cardbus devices used to be supported only by the
  49        prism54 wireless driver. After Intersil stopped selling these
  50        devices in preference for the newer more flexible SoftMAC devices
  51        a SoftMAC device driver was required and prism54 did not support
  52        them. The p54pci driver now exists and has been present in the kernel for
  53        a while. This driver supports both SoftMAC devices and FullMAC devices.
  54        The main difference between these devices was the amount of memory which
  55        could be used for the firmware. The SoftMAC devices support a smaller
  56        amount of memory. Because of this the SoftMAC firmware fits into FullMAC
  57        devices's memory. p54pci supports not only PCI / Cardbus but also USB
  58        and SPI. Since p54pci supports all devices prism54 supports
  59        you will have a conflict. I'm not quite sure how distributions are
  60        handling this conflict right now. prism54 was kept around due to
  61        claims users may experience issues when using the SoftMAC driver.
  62        Time has passed users have not reported issues. If you use prism54
  63        and for whatever reason you cannot use p54pci please let us know!
  64        E-mail us at: linux-wireless@vger.kernel.org
  65
  66        For more information see the p54 wiki page:
  67
  68        http://wireless.kernel.org/en/users/Drivers/p54
  69
  70Who:    Luis R. Rodriguez <lrodriguez@atheros.com>
  71
  72---------------------------
  73
  74What:   IRQF_SAMPLE_RANDOM
  75Check:  IRQF_SAMPLE_RANDOM
  76When:   July 2009
  77
  78Why:    Many of IRQF_SAMPLE_RANDOM users are technically bogus as entropy
  79        sources in the kernel's current entropy model. To resolve this, every
  80        input point to the kernel's entropy pool needs to better document the
  81        type of entropy source it actually is. This will be replaced with
  82        additional add_*_randomness functions in drivers/char/random.c
  83
  84Who:    Robin Getz <rgetz@blackfin.uclinux.org> & Matt Mackall <mpm@selenic.com>
  85
  86---------------------------
  87
  88What:   Deprecated snapshot ioctls
  89When:   2.6.36
  90
  91Why:    The ioctls in kernel/power/user.c were marked as deprecated long time
  92        ago. Now they notify users about that so that they need to replace
  93        their userspace. After some more time, remove them completely.
  94
  95Who:    Jiri Slaby <jirislaby@gmail.com>
  96
  97---------------------------
  98
  99What:   The ieee80211_regdom module parameter
 100When:   March 2010 / desktop catchup
 101
 102Why:    This was inherited by the CONFIG_WIRELESS_OLD_REGULATORY code,
 103        and currently serves as an option for users to define an
 104        ISO / IEC 3166 alpha2 code for the country they are currently
 105        present in. Although there are userspace API replacements for this
 106        through nl80211 distributions haven't yet caught up with implementing
 107        decent alternatives through standard GUIs. Although available as an
 108        option through iw or wpa_supplicant its just a matter of time before
 109        distributions pick up good GUI options for this. The ideal solution
 110        would actually consist of intelligent designs which would do this for
 111        the user automatically even when travelling through different countries.
 112        Until then we leave this module parameter as a compromise.
 113
 114        When userspace improves with reasonable widely-available alternatives for
 115        this we will no longer need this module parameter. This entry hopes that
 116        by the super-futuristically looking date of "March 2010" we will have
 117        such replacements widely available.
 118
 119Who:    Luis R. Rodriguez <lrodriguez@atheros.com>
 120
 121---------------------------
 122
 123What:   dev->power.power_state
 124When:   July 2007
 125Why:    Broken design for runtime control over driver power states, confusing
 126        driver-internal runtime power management with:  mechanisms to support
 127        system-wide sleep state transitions; event codes that distinguish
 128        different phases of swsusp "sleep" transitions; and userspace policy
 129        inputs.  This framework was never widely used, and most attempts to
 130        use it were broken.  Drivers should instead be exposing domain-specific
 131        interfaces either to kernel or to userspace.
 132Who:    Pavel Machek <pavel@ucw.cz>
 133
 134---------------------------
 135
 136What:   sys_sysctl
 137When:   September 2010
 138Option: CONFIG_SYSCTL_SYSCALL
 139Why:    The same information is available in a more convenient from
 140        /proc/sys, and none of the sysctl variables appear to be
 141        important performance wise.
 142
 143        Binary sysctls are a long standing source of subtle kernel
 144        bugs and security issues.
 145
 146        When I looked several months ago all I could find after
 147        searching several distributions were 5 user space programs and
 148        glibc (which falls back to /proc/sys) using this syscall.
 149
 150        The man page for sysctl(2) documents it as unusable for user
 151        space programs.
 152
 153        sysctl(2) is not generally ABI compatible to a 32bit user
 154        space application on a 64bit and a 32bit kernel.
 155
 156        For the last several months the policy has been no new binary
 157        sysctls and no one has put forward an argument to use them.
 158
 159        Binary sysctls issues seem to keep happening appearing so
 160        properly deprecating them (with a warning to user space) and a
 161        2 year grace warning period will mean eventually we can kill
 162        them and end the pain.
 163
 164        In the mean time individual binary sysctls can be dealt with
 165        in a piecewise fashion.
 166
 167Who:    Eric Biederman <ebiederm@xmission.com>
 168
 169---------------------------
 170
 171What:   /proc/<pid>/oom_adj
 172When:   August 2012
 173Why:    /proc/<pid>/oom_adj allows userspace to influence the oom killer's
 174        badness heuristic used to determine which task to kill when the kernel
 175        is out of memory.
 176
 177        The badness heuristic has since been rewritten since the introduction of
 178        this tunable such that its meaning is deprecated.  The value was
 179        implemented as a bitshift on a score generated by the badness()
 180        function that did not have any precise units of measure.  With the
 181        rewrite, the score is given as a proportion of available memory to the
 182        task allocating pages, so using a bitshift which grows the score
 183        exponentially is, thus, impossible to tune with fine granularity.
 184
 185        A much more powerful interface, /proc/<pid>/oom_score_adj, was
 186        introduced with the oom killer rewrite that allows users to increase or
 187        decrease the badness score linearly.  This interface will replace
 188        /proc/<pid>/oom_adj.
 189
 190        A warning will be emitted to the kernel log if an application uses this
 191        deprecated interface.  After it is printed once, future warnings will be
 192        suppressed until the kernel is rebooted.
 193
 194---------------------------
 195
 196What:   remove EXPORT_SYMBOL(kernel_thread)
 197When:   August 2006
 198Files:  arch/*/kernel/*_ksyms.c
 199Check:  kernel_thread
 200Why:    kernel_thread is a low-level implementation detail.  Drivers should
 201        use the <linux/kthread.h> API instead which shields them from
 202        implementation details and provides a higherlevel interface that
 203        prevents bugs and code duplication
 204Who:    Christoph Hellwig <hch@lst.de>
 205
 206---------------------------
 207
 208What:   Unused EXPORT_SYMBOL/EXPORT_SYMBOL_GPL exports
 209        (temporary transition config option provided until then)
 210        The transition config option will also be removed at the same time.
 211When:   before 2.6.19
 212Why:    Unused symbols are both increasing the size of the kernel binary
 213        and are often a sign of "wrong API"
 214Who:    Arjan van de Ven <arjan@linux.intel.com>
 215
 216---------------------------
 217
 218What:   PHYSDEVPATH, PHYSDEVBUS, PHYSDEVDRIVER in the uevent environment
 219When:   October 2008
 220Why:    The stacking of class devices makes these values misleading and
 221        inconsistent.
 222        Class devices should not carry any of these properties, and bus
 223        devices have SUBSYTEM and DRIVER as a replacement.
 224Who:    Kay Sievers <kay.sievers@suse.de>
 225
 226---------------------------
 227
 228What:   ACPI procfs interface
 229When:   July 2008
 230Why:    ACPI sysfs conversion should be finished by January 2008.
 231        ACPI procfs interface will be removed in July 2008 so that
 232        there is enough time for the user space to catch up.
 233Who:    Zhang Rui <rui.zhang@intel.com>
 234
 235---------------------------
 236
 237What:   CONFIG_ACPI_PROCFS_POWER
 238When:   2.6.39
 239Why:    sysfs I/F for ACPI power devices, including AC and Battery,
 240        has been working in upstream kernel since 2.6.24, Sep 2007.
 241        In 2.6.37, we make the sysfs I/F always built in and this option
 242        disabled by default.
 243        Remove this option and the ACPI power procfs interface in 2.6.39.
 244Who:    Zhang Rui <rui.zhang@intel.com>
 245
 246---------------------------
 247
 248What:   /proc/acpi/event
 249When:   February 2008
 250Why:    /proc/acpi/event has been replaced by events via the input layer
 251        and netlink since 2.6.23.
 252Who:    Len Brown <len.brown@intel.com>
 253
 254---------------------------
 255
 256What:   i386/x86_64 bzImage symlinks
 257When:   April 2010
 258
 259Why:    The i386/x86_64 merge provides a symlink to the old bzImage
 260        location so not yet updated user space tools, e.g. package
 261        scripts, do not break.
 262Who:    Thomas Gleixner <tglx@linutronix.de>
 263
 264---------------------------
 265
 266What:   GPIO autorequest on gpio_direction_{input,output}() in gpiolib
 267When:   February 2010
 268Why:    All callers should use explicit gpio_request()/gpio_free().
 269        The autorequest mechanism in gpiolib was provided mostly as a
 270        migration aid for legacy GPIO interfaces (for SOC based GPIOs).
 271        Those users have now largely migrated.  Platforms implementing
 272        the GPIO interfaces without using gpiolib will see no changes.
 273Who:    David Brownell <dbrownell@users.sourceforge.net>
 274---------------------------
 275
 276What:   b43 support for firmware revision < 410
 277When:   The schedule was July 2008, but it was decided that we are going to keep the
 278        code as long as there are no major maintanance headaches.
 279        So it _could_ be removed _any_ time now, if it conflicts with something new.
 280Why:    The support code for the old firmware hurts code readability/maintainability
 281        and slightly hurts runtime performance. Bugfixes for the old firmware
 282        are not provided by Broadcom anymore.
 283Who:    Michael Buesch <m@bues.ch>
 284
 285---------------------------
 286
 287What:   Ability for non root users to shm_get hugetlb pages based on mlock
 288        resource limits
 289When:   2.6.31
 290Why:    Non root users need to be part of /proc/sys/vm/hugetlb_shm_group or
 291        have CAP_IPC_LOCK to be able to allocate shm segments backed by
 292        huge pages.  The mlock based rlimit check to allow shm hugetlb is
 293        inconsistent with mmap based allocations.  Hence it is being
 294        deprecated.
 295Who:    Ravikiran Thirumalai <kiran@scalex86.org>
 296
 297---------------------------
 298
 299What:   Code that is now under CONFIG_WIRELESS_EXT_SYSFS
 300        (in net/core/net-sysfs.c)
 301When:   After the only user (hal) has seen a release with the patches
 302        for enough time, probably some time in 2010.
 303Why:    Over 1K .text/.data size reduction, data is available in other
 304        ways (ioctls)
 305Who:    Johannes Berg <johannes@sipsolutions.net>
 306
 307---------------------------
 308
 309What:   sysfs ui for changing p4-clockmod parameters
 310When:   September 2009
 311Why:    See commits 129f8ae9b1b5be94517da76009ea956e89104ce8 and
 312        e088e4c9cdb618675874becb91b2fd581ee707e6.
 313        Removal is subject to fixing any remaining bugs in ACPI which may
 314        cause the thermal throttling not to happen at the right time.
 315Who:    Dave Jones <davej@redhat.com>, Matthew Garrett <mjg@redhat.com>
 316
 317-----------------------------
 318
 319What:   fakephp and associated sysfs files in /sys/bus/pci/slots/
 320When:   2011
 321Why:    In 2.6.27, the semantics of /sys/bus/pci/slots was redefined to
 322        represent a machine's physical PCI slots. The change in semantics
 323        had userspace implications, as the hotplug core no longer allowed
 324        drivers to create multiple sysfs files per physical slot (required
 325        for multi-function devices, e.g.). fakephp was seen as a developer's
 326        tool only, and its interface changed. Too late, we learned that
 327        there were some users of the fakephp interface.
 328
 329        In 2.6.30, the original fakephp interface was restored. At the same
 330        time, the PCI core gained the ability that fakephp provided, namely
 331        function-level hot-remove and hot-add.
 332
 333        Since the PCI core now provides the same functionality, exposed in:
 334
 335                /sys/bus/pci/rescan
 336                /sys/bus/pci/devices/.../remove
 337                /sys/bus/pci/devices/.../rescan
 338
 339        there is no functional reason to maintain fakephp as well.
 340
 341        We will keep the existing module so that 'modprobe fakephp' will
 342        present the old /sys/bus/pci/slots/... interface for compatibility,
 343        but users are urged to migrate their applications to the API above.
 344
 345        After a reasonable transition period, we will remove the legacy
 346        fakephp interface.
 347Who:    Alex Chiang <achiang@hp.com>
 348
 349---------------------------
 350
 351What:   CONFIG_RFKILL_INPUT
 352When:   2.6.33
 353Why:    Should be implemented in userspace, policy daemon.
 354Who:    Johannes Berg <johannes@sipsolutions.net>
 355
 356----------------------------
 357
 358What:   sound-slot/service-* module aliases and related clutters in
 359        sound/sound_core.c
 360When:   August 2010
 361Why:    OSS sound_core grabs all legacy minors (0-255) of SOUND_MAJOR
 362        (14) and requests modules using custom sound-slot/service-*
 363        module aliases.  The only benefit of doing this is allowing
 364        use of custom module aliases which might as well be considered
 365        a bug at this point.  This preemptive claiming prevents
 366        alternative OSS implementations.
 367
 368        Till the feature is removed, the kernel will be requesting
 369        both sound-slot/service-* and the standard char-major-* module
 370        aliases and allow turning off the pre-claiming selectively via
 371        CONFIG_SOUND_OSS_CORE_PRECLAIM and soundcore.preclaim_oss
 372        kernel parameter.
 373
 374        After the transition phase is complete, both the custom module
 375        aliases and switches to disable it will go away.  This removal
 376        will also allow making ALSA OSS emulation independent of
 377        sound_core.  The dependency will be broken then too.
 378Who:    Tejun Heo <tj@kernel.org>
 379
 380----------------------------
 381
 382What:   sysfs-class-rfkill state file
 383When:   Feb 2014
 384Files:  net/rfkill/core.c
 385Why:    Documented as obsolete since Feb 2010. This file is limited to 3
 386        states while the rfkill drivers can have 4 states.
 387Who:    anybody or Florian Mickler <florian@mickler.org>
 388
 389----------------------------
 390
 391What:   sysfs-class-rfkill claim file
 392When:   Feb 2012
 393Files:  net/rfkill/core.c
 394Why:    It is not possible to claim an rfkill driver since 2007. This is
 395        Documented as obsolete since Feb 2010.
 396Who:    anybody or Florian Mickler <florian@mickler.org>
 397
 398----------------------------
 399
 400What:   KVM paravirt mmu host support
 401When:   January 2011
 402Why:    The paravirt mmu host support is slower than non-paravirt mmu, both
 403        on newer and older hardware.  It is already not exposed to the guest,
 404        and kept only for live migration purposes.
 405Who:    Avi Kivity <avi@redhat.com>
 406
 407----------------------------
 408
 409What:   iwlwifi 50XX module parameters
 410When:   3.0
 411Why:    The "..50" modules parameters were used to configure 5000 series and
 412        up devices; different set of module parameters also available for 4965
 413        with same functionalities. Consolidate both set into single place
 414        in drivers/net/wireless/iwlwifi/iwl-agn.c
 415
 416Who:    Wey-Yi Guy <wey-yi.w.guy@intel.com>
 417
 418----------------------------
 419
 420What:   iwl4965 alias support
 421When:   3.0
 422Why:    Internal alias support has been present in module-init-tools for some
 423        time, the MODULE_ALIAS("iwl4965") boilerplate aliases can be removed
 424        with no impact.
 425
 426Who:    Wey-Yi Guy <wey-yi.w.guy@intel.com>
 427
 428---------------------------
 429
 430What:   xt_NOTRACK
 431Files:  net/netfilter/xt_NOTRACK.c
 432When:   April 2011
 433Why:    Superseded by xt_CT
 434Who:    Netfilter developer team <netfilter-devel@vger.kernel.org>
 435
 436----------------------------
 437
 438What:   IRQF_DISABLED
 439When:   2.6.36
 440Why:    The flag is a NOOP as we run interrupt handlers with interrupts disabled
 441Who:    Thomas Gleixner <tglx@linutronix.de>
 442
 443----------------------------
 444
 445What:   PCI DMA unmap state API
 446When:   August 2012
 447Why:    PCI DMA unmap state API (include/linux/pci-dma.h) was replaced
 448        with DMA unmap state API (DMA unmap state API can be used for
 449        any bus).
 450Who:    FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
 451
 452----------------------------
 453
 454What:   iwlwifi disable_hw_scan module parameters
 455When:   3.0
 456Why:    Hareware scan is the prefer method for iwlwifi devices for
 457        scanning operation. Remove software scan support for all the
 458        iwlwifi devices.
 459
 460Who:    Wey-Yi Guy <wey-yi.w.guy@intel.com>
 461
 462----------------------------
 463
 464What:   Legacy, non-standard chassis intrusion detection interface.
 465When:   June 2011
 466Why:    The adm9240, w83792d and w83793 hardware monitoring drivers have
 467        legacy interfaces for chassis intrusion detection. A standard
 468        interface has been added to each driver, so the legacy interface
 469        can be removed.
 470Who:    Jean Delvare <khali@linux-fr.org>
 471
 472----------------------------
 473
 474What:   xt_connlimit rev 0
 475When:   2012
 476Who:    Jan Engelhardt <jengelh@medozas.de>
 477Files:  net/netfilter/xt_connlimit.c
 478
 479----------------------------
 480
 481What:   ipt_addrtype match include file
 482When:   2012
 483Why:    superseded by xt_addrtype
 484Who:    Florian Westphal <fw@strlen.de>
 485Files:  include/linux/netfilter_ipv4/ipt_addrtype.h
 486
 487----------------------------
 488
 489What:   i2c_driver.attach_adapter
 490        i2c_driver.detach_adapter
 491When:   September 2011
 492Why:    These legacy callbacks should no longer be used as i2c-core offers
 493        a variety of preferable alternative ways to instantiate I2C devices.
 494Who:    Jean Delvare <khali@linux-fr.org>
 495
 496----------------------------
 497
 498What:   Support for UVCIOC_CTRL_ADD in the uvcvideo driver
 499When:   3.2
 500Why:    The information passed to the driver by this ioctl is now queried
 501        dynamically from the device.
 502Who:    Laurent Pinchart <laurent.pinchart@ideasonboard.com>
 503
 504----------------------------
 505
 506What:   Support for UVCIOC_CTRL_MAP_OLD in the uvcvideo driver
 507When:   3.2
 508Why:    Used only by applications compiled against older driver versions.
 509        Superseded by UVCIOC_CTRL_MAP which supports V4L2 menu controls.
 510Who:    Laurent Pinchart <laurent.pinchart@ideasonboard.com>
 511
 512----------------------------
 513
 514What:   Support for UVCIOC_CTRL_GET and UVCIOC_CTRL_SET in the uvcvideo driver
 515When:   3.2
 516Why:    Superseded by the UVCIOC_CTRL_QUERY ioctl.
 517Who:    Laurent Pinchart <laurent.pinchart@ideasonboard.com>
 518
 519----------------------------
 520
 521What:   Support for driver specific ioctls in the pwc driver (everything
 522        defined in media/pwc-ioctl.h)
 523When:   3.3
 524Why:    This stems from the v4l1 era, with v4l2 everything can be done with
 525        standardized v4l2 API calls
 526Who:    Hans de Goede <hdegoede@redhat.com>
 527
 528----------------------------
 529
 530What:   Driver specific sysfs API in the pwc driver
 531When:   3.3
 532Why:    Setting pan/tilt should be done with v4l2 controls, like with other
 533        cams. The button is available as a standard input device
 534Who:    Hans de Goede <hdegoede@redhat.com>
 535
 536----------------------------
 537
 538What:   Driver specific use of pixfmt.priv in the pwc driver
 539When:   3.3
 540Why:    The .priv field never was intended for this, setting a framerate is
 541        support using the standardized S_PARM ioctl
 542Who:    Hans de Goede <hdegoede@redhat.com>
 543
 544----------------------------
 545
 546What:   Software emulation of arbritary resolutions in the pwc driver
 547When:   3.3
 548Why:    The pwc driver claims to support any resolution between 160x120
 549        and 640x480, but emulates this by simply drawing a black border
 550        around the image. Userspace can draw its own black border if it
 551        really wants one.
 552Who:    Hans de Goede <hdegoede@redhat.com>
 553
 554----------------------------
 555
 556What:   For VIDIOC_S_FREQUENCY the type field must match the device node's type.
 557        If not, return -EINVAL.
 558When:   3.2
 559Why:    It makes no sense to switch the tuner to radio mode by calling
 560        VIDIOC_S_FREQUENCY on a video node, or to switch the tuner to tv mode by
 561        calling VIDIOC_S_FREQUENCY on a radio node. This is the first step of a
 562        move to more consistent handling of tv and radio tuners.
 563Who:    Hans Verkuil <hans.verkuil@cisco.com>
 564
 565----------------------------
 566
 567What:   Opening a radio device node will no longer automatically switch the
 568        tuner mode from tv to radio.
 569When:   3.3
 570Why:    Just opening a V4L device should not change the state of the hardware
 571        like that. It's very unexpected and against the V4L spec. Instead, you
 572        switch to radio mode by calling VIDIOC_S_FREQUENCY. This is the second
 573        and last step of the move to consistent handling of tv and radio tuners.
 574Who:    Hans Verkuil <hans.verkuil@cisco.com>
 575
 576----------------------------
 577
 578What:   g_file_storage driver
 579When:   3.8
 580Why:    This driver has been superseded by g_mass_storage.
 581Who:    Alan Stern <stern@rowland.harvard.edu>
 582
 583----------------------------
 584
 585What:   threeg and interface sysfs files in /sys/devices/platform/acer-wmi
 586When:   2012
 587Why:    In 3.0, we can now autodetect internal 3G device and already have
 588        the threeg rfkill device. So, we plan to remove threeg sysfs support
 589        for it's no longer necessary.
 590
 591        We also plan to remove interface sysfs file that exposed which ACPI-WMI
 592        interface that was used by acer-wmi driver. It will replaced by
 593        information log when acer-wmi initial.
 594Who:    Lee, Chun-Yi <jlee@novell.com>
 595
 596----------------------------
 597What:   The XFS nodelaylog mount option
 598When:   3.3
 599Why:    The delaylog mode that has been the default since 2.6.39 has proven
 600        stable, and the old code is in the way of additional improvements in
 601        the log code.
 602Who:    Christoph Hellwig <hch@lst.de>
 603
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.