| blob | b16cbe4152ea3560e32a98624c53cc77b22fc659 |
1 The following is a list of files and features that are going to be
2 removed in the kernel source tree. Every entry should contain what
3 exactly is going away, why it is happening, and who is going to be doing
4 the work. When the feature is removed from the kernel, it should also
5 be removed from this file.
7 ---------------------------
9 What: PRISM54
10 When: 2.6.34
12 Why: prism54 FullMAC PCI / Cardbus devices used to be supported only by the
13 prism54 wireless driver. After Intersil stopped selling these
14 devices in preference for the newer more flexible SoftMAC devices
15 a SoftMAC device driver was required and prism54 did not support
16 them. The p54pci driver now exists and has been present in the kernel for
17 a while. This driver supports both SoftMAC devices and FullMAC devices.
18 The main difference between these devices was the amount of memory which
19 could be used for the firmware. The SoftMAC devices support a smaller
20 amount of memory. Because of this the SoftMAC firmware fits into FullMAC
21 devices's memory. p54pci supports not only PCI / Cardbus but also USB
22 and SPI. Since p54pci supports all devices prism54 supports
23 you will have a conflict. I'm not quite sure how distributions are
24 handling this conflict right now. prism54 was kept around due to
25 claims users may experience issues when using the SoftMAC driver.
26 Time has passed users have not reported issues. If you use prism54
27 and for whatever reason you cannot use p54pci please let us know!
28 E-mail us at: linux-wireless@vger.kernel.org
30 For more information see the p54 wiki page:
32 http://wireless.kernel.org/en/users/Drivers/p54
34 Who: Luis R. Rodriguez <lrodriguez@atheros.com>
36 ---------------------------
38 What: IRQF_SAMPLE_RANDOM
39 Check: IRQF_SAMPLE_RANDOM
40 When: July 2009
42 Why: Many of IRQF_SAMPLE_RANDOM users are technically bogus as entropy
43 sources in the kernel's current entropy model. To resolve this, every
44 input point to the kernel's entropy pool needs to better document the
45 type of entropy source it actually is. This will be replaced with
46 additional add_*_randomness functions in drivers/char/random.c
48 Who: Robin Getz <rgetz@blackfin.uclinux.org> & Matt Mackall <mpm@selenic.com>
50 ---------------------------
52 What: Deprecated snapshot ioctls
53 When: 2.6.36
55 Why: The ioctls in kernel/power/user.c were marked as deprecated long time
56 ago. Now they notify users about that so that they need to replace
57 their userspace. After some more time, remove them completely.
59 Who: Jiri Slaby <jirislaby@gmail.com>
61 ---------------------------
63 What: The ieee80211_regdom module parameter
64 When: March 2010 / desktop catchup
66 Why: This was inherited by the CONFIG_WIRELESS_OLD_REGULATORY code,
67 and currently serves as an option for users to define an
68 ISO / IEC 3166 alpha2 code for the country they are currently
69 present in. Although there are userspace API replacements for this
70 through nl80211 distributions haven't yet caught up with implementing
71 decent alternatives through standard GUIs. Although available as an
72 option through iw or wpa_supplicant its just a matter of time before
73 distributions pick up good GUI options for this. The ideal solution
74 would actually consist of intelligent designs which would do this for
75 the user automatically even when travelling through different countries.
76 Until then we leave this module parameter as a compromise.
78 When userspace improves with reasonable widely-available alternatives for
79 this we will no longer need this module parameter. This entry hopes that
80 by the super-futuristically looking date of "March 2010" we will have
81 such replacements widely available.
83 Who: Luis R. Rodriguez <lrodriguez@atheros.com>
85 ---------------------------
87 What: dev->power.power_state
88 When: July 2007
89 Why: Broken design for runtime control over driver power states, confusing
90 driver-internal runtime power management with: mechanisms to support
91 system-wide sleep state transitions; event codes that distinguish
92 different phases of swsusp "sleep" transitions; and userspace policy
93 inputs. This framework was never widely used, and most attempts to
94 use it were broken. Drivers should instead be exposing domain-specific
95 interfaces either to kernel or to userspace.
96 Who: Pavel Machek <pavel@ucw.cz>
98 ---------------------------
100 What: Video4Linux API 1 ioctls and from Video devices.
101 When: July 2009
102 Files: include/linux/videodev.h
103 Check: include/linux/videodev.h
104 Why: V4L1 AP1 was replaced by V4L2 API during migration from 2.4 to 2.6
105 series. The old API have lots of drawbacks and don't provide enough
106 means to work with all video and audio standards. The newer API is
107 already available on the main drivers and should be used instead.
108 Newer drivers should use v4l_compat_translate_ioctl function to handle
109 old calls, replacing to newer ones.
110 Decoder iocts are using internally to allow video drivers to
111 communicate with video decoders. This should also be improved to allow
112 V4L2 calls being translated into compatible internal ioctls.
113 Compatibility ioctls will be provided, for a while, via
114 v4l1-compat module.
115 Who: Mauro Carvalho Chehab <mchehab@infradead.org>
117 ---------------------------
119 What: sys_sysctl
120 When: September 2010
121 Option: CONFIG_SYSCTL_SYSCALL
122 Why: The same information is available in a more convenient from
123 /proc/sys, and none of the sysctl variables appear to be
124 important performance wise.
126 Binary sysctls are a long standing source of subtle kernel
127 bugs and security issues.
129 When I looked several months ago all I could find after
130 searching several distributions were 5 user space programs and
131 glibc (which falls back to /proc/sys) using this syscall.
133 The man page for sysctl(2) documents it as unusable for user
134 space programs.
136 sysctl(2) is not generally ABI compatible to a 32bit user
137 space application on a 64bit and a 32bit kernel.
139 For the last several months the policy has been no new binary
140 sysctls and no one has put forward an argument to use them.
142 Binary sysctls issues seem to keep happening appearing so
143 properly deprecating them (with a warning to user space) and a
144 2 year grace warning period will mean eventually we can kill
145 them and end the pain.
147 In the mean time individual binary sysctls can be dealt with
148 in a piecewise fashion.
150 Who: Eric Biederman <ebiederm@xmission.com>
152 ---------------------------
154 What: /proc/<pid>/oom_adj
155 When: August 2012
156 Why: /proc/<pid>/oom_adj allows userspace to influence the oom killer's
157 badness heuristic used to determine which task to kill when the kernel
158 is out of memory.
160 The badness heuristic has since been rewritten since the introduction of
161 this tunable such that its meaning is deprecated. The value was
162 implemented as a bitshift on a score generated by the badness()
163 function that did not have any precise units of measure. With the
164 rewrite, the score is given as a proportion of available memory to the
165 task allocating pages, so using a bitshift which grows the score
166 exponentially is, thus, impossible to tune with fine granularity.
168 A much more powerful interface, /proc/<pid>/oom_score_adj, was
169 introduced with the oom killer rewrite that allows users to increase or
170 decrease the badness() score linearly. This interface will replace
171 /proc/<pid>/oom_adj.
173 A warning will be emitted to the kernel log if an application uses this
174 deprecated interface. After it is printed once, future warnings will be
175 suppressed until the kernel is rebooted.
177 ---------------------------
179 What: remove EXPORT_SYMBOL(kernel_thread)
180 When: August 2006
181 Files: arch/*/kernel/*_ksyms.c
182 Check: kernel_thread
183 Why: kernel_thread is a low-level implementation detail. Drivers should
184 use the <linux/kthread.h> API instead which shields them from
185 implementation details and provides a higherlevel interface that
186 prevents bugs and code duplication
187 Who: Christoph Hellwig <hch@lst.de>
189 ---------------------------
191 What: Unused EXPORT_SYMBOL/EXPORT_SYMBOL_GPL exports
192 (temporary transition config option provided until then)
193 The transition config option will also be removed at the same time.
194 When: before 2.6.19
195 Why: Unused symbols are both increasing the size of the kernel binary
196 and are often a sign of "wrong API"
197 Who: Arjan van de Ven <arjan@linux.intel.com>
199 ---------------------------
201 What: PHYSDEVPATH, PHYSDEVBUS, PHYSDEVDRIVER in the uevent environment
202 When: October 2008
203 Why: The stacking of class devices makes these values misleading and
204 inconsistent.
205 Class devices should not carry any of these properties, and bus
206 devices have SUBSYTEM and DRIVER as a replacement.
207 Who: Kay Sievers <kay.sievers@suse.de>
209 ---------------------------
211 What: ACPI procfs interface
212 When: July 2008
213 Why: ACPI sysfs conversion should be finished by January 2008.
214 ACPI procfs interface will be removed in July 2008 so that
215 there is enough time for the user space to catch up.
216 Who: Zhang Rui <rui.zhang@intel.com>
218 ---------------------------
220 What: /proc/acpi/button
221 When: August 2007
222 Why: /proc/acpi/button has been replaced by events to the input layer
223 since 2.6.20.
224 Who: Len Brown <len.brown@intel.com>
226 ---------------------------
228 What: /proc/acpi/event
229 When: February 2008
230 Why: /proc/acpi/event has been replaced by events via the input layer
231 and netlink since 2.6.23.
232 Who: Len Brown <len.brown@intel.com>
234 ---------------------------
236 What: i386/x86_64 bzImage symlinks
237 When: April 2010
239 Why: The i386/x86_64 merge provides a symlink to the old bzImage
240 location so not yet updated user space tools, e.g. package
241 scripts, do not break.
242 Who: Thomas Gleixner <tglx@linutronix.de>
244 ---------------------------
246 What: GPIO autorequest on gpio_direction_{input,output}() in gpiolib
247 When: February 2010
248 Why: All callers should use explicit gpio_request()/gpio_free().
249 The autorequest mechanism in gpiolib was provided mostly as a
250 migration aid for legacy GPIO interfaces (for SOC based GPIOs).
251 Those users have now largely migrated. Platforms implementing
252 the GPIO interfaces without using gpiolib will see no changes.
253 Who: David Brownell <dbrownell@users.sourceforge.net>
254 ---------------------------
256 What: b43 support for firmware revision < 410
257 When: The schedule was July 2008, but it was decided that we are going to keep the
258 code as long as there are no major maintanance headaches.
259 So it _could_ be removed _any_ time now, if it conflicts with something new.
260 Why: The support code for the old firmware hurts code readability/maintainability
261 and slightly hurts runtime performance. Bugfixes for the old firmware
262 are not provided by Broadcom anymore.
263 Who: Michael Buesch <mb@bu3sch.de>
265 ---------------------------
267 What: /sys/o2cb symlink
268 When: January 2010
269 Why: /sys/fs/o2cb is the proper location for this information - /sys/o2cb
270 exists as a symlink for backwards compatibility for old versions of
271 ocfs2-tools. 2 years should be sufficient time to phase in new versions
272 which know to look in /sys/fs/o2cb.
273 Who: ocfs2-devel@oss.oracle.com
275 ---------------------------
277 What: Ability for non root users to shm_get hugetlb pages based on mlock
278 resource limits
279 When: 2.6.31
280 Why: Non root users need to be part of /proc/sys/vm/hugetlb_shm_group or
281 have CAP_IPC_LOCK to be able to allocate shm segments backed by
282 huge pages. The mlock based rlimit check to allow shm hugetlb is
283 inconsistent with mmap based allocations. Hence it is being
284 deprecated.
285 Who: Ravikiran Thirumalai <kiran@scalex86.org>
287 ---------------------------
289 What: CONFIG_THERMAL_HWMON
290 When: January 2009
291 Why: This option was introduced just to allow older lm-sensors userspace
292 to keep working over the upgrade to 2.6.26. At the scheduled time of
293 removal fixed lm-sensors (2.x or 3.x) should be readily available.
294 Who: Rene Herman <rene.herman@gmail.com>
296 ---------------------------
298 What: Code that is now under CONFIG_WIRELESS_EXT_SYSFS
299 (in net/core/net-sysfs.c)
300 When: After the only user (hal) has seen a release with the patches
301 for enough time, probably some time in 2010.
302 Why: Over 1K .text/.data size reduction, data is available in other
303 ways (ioctls)
304 Who: Johannes Berg <johannes@sipsolutions.net>
306 ---------------------------
308 What: sysfs ui for changing p4-clockmod parameters
309 When: September 2009
310 Why: See commits 129f8ae9b1b5be94517da76009ea956e89104ce8 and
311 e088e4c9cdb618675874becb91b2fd581ee707e6.
312 Removal is subject to fixing any remaining bugs in ACPI which may
313 cause the thermal throttling not to happen at the right time.
314 Who: Dave Jones <davej@redhat.com>, Matthew Garrett <mjg@redhat.com>
316 -----------------------------
318 What: __do_IRQ all in one fits nothing interrupt handler
319 When: 2.6.32
320 Why: __do_IRQ was kept for easy migration to the type flow handlers.
321 More than two years of migration time is enough.
322 Who: Thomas Gleixner <tglx@linutronix.de>
324 -----------------------------
326 What: fakephp and associated sysfs files in /sys/bus/pci/slots/
327 When: 2011
328 Why: In 2.6.27, the semantics of /sys/bus/pci/slots was redefined to
329 represent a machine's physical PCI slots. The change in semantics
330 had userspace implications, as the hotplug core no longer allowed
331 drivers to create multiple sysfs files per physical slot (required
332 for multi-function devices, e.g.). fakephp was seen as a developer's
333 tool only, and its interface changed. Too late, we learned that
334 there were some users of the fakephp interface.
336 In 2.6.30, the original fakephp interface was restored. At the same
337 time, the PCI core gained the ability that fakephp provided, namely
338 function-level hot-remove and hot-add.
340 Since the PCI core now provides the same functionality, exposed in:
342 /sys/bus/pci/rescan
343 /sys/bus/pci/devices/.../remove
344 /sys/bus/pci/devices/.../rescan
346 there is no functional reason to maintain fakephp as well.
348 We will keep the existing module so that 'modprobe fakephp' will
349 present the old /sys/bus/pci/slots/... interface for compatibility,
350 but users are urged to migrate their applications to the API above.
352 After a reasonable transition period, we will remove the legacy
353 fakephp interface.
354 Who: Alex Chiang <achiang@hp.com>
356 ---------------------------
358 What: CONFIG_RFKILL_INPUT
359 When: 2.6.33
360 Why: Should be implemented in userspace, policy daemon.
361 Who: Johannes Berg <johannes@sipsolutions.net>
363 ----------------------------
365 What: sound-slot/service-* module aliases and related clutters in
366 sound/sound_core.c
367 When: August 2010
368 Why: OSS sound_core grabs all legacy minors (0-255) of SOUND_MAJOR
369 (14) and requests modules using custom sound-slot/service-*
370 module aliases. The only benefit of doing this is allowing
371 use of custom module aliases which might as well be considered
372 a bug at this point. This preemptive claiming prevents
373 alternative OSS implementations.
375 Till the feature is removed, the kernel will be requesting
376 both sound-slot/service-* and the standard char-major-* module
377 aliases and allow turning off the pre-claiming selectively via
378 CONFIG_SOUND_OSS_CORE_PRECLAIM and soundcore.preclaim_oss
379 kernel parameter.
381 After the transition phase is complete, both the custom module
382 aliases and switches to disable it will go away. This removal
383 will also allow making ALSA OSS emulation independent of
384 sound_core. The dependency will be broken then too.
385 Who: Tejun Heo <tj@kernel.org>
387 ----------------------------
389 What: Support for VMware's guest paravirtuliazation technique [VMI] will be
390 dropped.
391 When: 2.6.37 or earlier.
392 Why: With the recent innovations in CPU hardware acceleration technologies
393 from Intel and AMD, VMware ran a few experiments to compare these
394 techniques to guest paravirtualization technique on VMware's platform.
395 These hardware assisted virtualization techniques have outperformed the
396 performance benefits provided by VMI in most of the workloads. VMware
397 expects that these hardware features will be ubiquitous in a couple of
398 years, as a result, VMware has started a phased retirement of this
399 feature from the hypervisor. We will be removing this feature from the
400 Kernel too. Right now we are targeting 2.6.37 but can retire earlier if
401 technical reasons (read opportunity to remove major chunk of pvops)
402 arise.
404 Please note that VMI has always been an optimization and non-VMI kernels
405 still work fine on VMware's platform.
406 Latest versions of VMware's product which support VMI are,
407 Workstation 7.0 and VSphere 4.0 on ESX side, future maintainence
408 releases for these products will continue supporting VMI.
410 For more details about VMI retirement take a look at this,
411 http://blogs.vmware.com/guestosguide/2009/09/vmi-retirement.html
413 Who: Alok N Kataria <akataria@vmware.com>
415 ----------------------------
417 What: Support for lcd_switch and display_get in asus-laptop driver
418 When: March 2010
419 Why: These two features use non-standard interfaces. There are the
420 only features that really need multiple path to guess what's
421 the right method name on a specific laptop.
423 Removing them will allow to remove a lot of code an significantly
424 clean the drivers.
426 This will affect the backlight code which won't be able to know
427 if the backlight is on or off. The platform display file will also be
428 write only (like the one in eeepc-laptop).
430 This should'nt affect a lot of user because they usually know
431 when their display is on or off.
433 Who: Corentin Chary <corentin.chary@gmail.com>
435 ----------------------------
437 What: sysfs-class-rfkill state file
438 When: Feb 2014
439 Files: net/rfkill/core.c
440 Why: Documented as obsolete since Feb 2010. This file is limited to 3
441 states while the rfkill drivers can have 4 states.
442 Who: anybody or Florian Mickler <florian@mickler.org>
444 ----------------------------
446 What: sysfs-class-rfkill claim file
447 When: Feb 2012
448 Files: net/rfkill/core.c
449 Why: It is not possible to claim an rfkill driver since 2007. This is
450 Documented as obsolete since Feb 2010.
451 Who: anybody or Florian Mickler <florian@mickler.org>
453 ----------------------------
455 What: capifs
456 When: February 2011
457 Files: drivers/isdn/capi/capifs.*
458 Why: udev fully replaces this special file system that only contains CAPI
459 NCCI TTY device nodes. User space (pppdcapiplugin) works without
460 noticing the difference.
461 Who: Jan Kiszka <jan.kiszka@web.de>
463 ----------------------------
465 What: KVM paravirt mmu host support
466 When: January 2011
467 Why: The paravirt mmu host support is slower than non-paravirt mmu, both
468 on newer and older hardware. It is already not exposed to the guest,
469 and kept only for live migration purposes.
470 Who: Avi Kivity <avi@redhat.com>
472 ----------------------------
474 What: iwlwifi 50XX module parameters
475 When: 2.6.40
476 Why: The "..50" modules parameters were used to configure 5000 series and
477 up devices; different set of module parameters also available for 4965
478 with same functionalities. Consolidate both set into single place
479 in drivers/net/wireless/iwlwifi/iwl-agn.c
481 Who: Wey-Yi Guy <wey-yi.w.guy@intel.com>
483 ----------------------------
485 What: iwl4965 alias support
486 When: 2.6.40
487 Why: Internal alias support has been present in module-init-tools for some
488 time, the MODULE_ALIAS("iwl4965") boilerplate aliases can be removed
489 with no impact.
491 Who: Wey-Yi Guy <wey-yi.w.guy@intel.com>
493 ---------------------------
495 What: xt_NOTRACK
496 Files: net/netfilter/xt_NOTRACK.c
497 When: April 2011
498 Why: Superseded by xt_CT
499 Who: Netfilter developer team <netfilter-devel@vger.kernel.org>
501 ---------------------------
503 What: video4linux /dev/vtx teletext API support
504 When: 2.6.35
505 Files: drivers/media/video/saa5246a.c drivers/media/video/saa5249.c
506 include/linux/videotext.h
507 Why: The vtx device nodes have been superseded by vbi device nodes
508 for many years. No applications exist that use the vtx support.
509 Of the two i2c drivers that actually support this API the saa5249
510 has been impossible to use for a year now and no known hardware
511 that supports this device exists. The saa5246a is theoretically
512 supported by the old mxb boards, but it never actually worked.
514 In summary: there is no hardware that can use this API and there
515 are no applications actually implementing this API.
517 The vtx support still reserves minors 192-223 and we would really
518 like to reuse those for upcoming new functionality. In the unlikely
519 event that new hardware appears that wants to use the functionality
520 provided by the vtx API, then that functionality should be build
521 around the sliced VBI API instead.
522 Who: Hans Verkuil <hverkuil@xs4all.nl>
524 ----------------------------
526 What: IRQF_DISABLED
527 When: 2.6.36
528 Why: The flag is a NOOP as we run interrupt handlers with interrupts disabled
529 Who: Thomas Gleixner <tglx@linutronix.de>
531 ----------------------------
533 What: old ieee1394 subsystem (CONFIG_IEEE1394)
534 When: 2.6.37
535 Files: drivers/ieee1394/ except init_ohci1394_dma.c
536 Why: superseded by drivers/firewire/ (CONFIG_FIREWIRE) which offers more
537 features, better performance, and better security, all with smaller
538 and more modern code base
539 Who: Stefan Richter <stefanr@s5r6.in-berlin.de>
541 ----------------------------
543 What: The acpi_sleep=s4_nonvs command line option
544 When: 2.6.37
545 Files: arch/x86/kernel/acpi/sleep.c
546 Why: superseded by acpi_sleep=nonvs
547 Who: Rafael J. Wysocki <rjw@sisk.pl>
549 ----------------------------