ia64/linux-2.6.18-xen.hg

view Documentation/feature-removal-schedule.txt @ 897:329ea0ccb344

balloon: try harder to balloon up under memory pressure.

Currently if the balloon driver is unable to increase the guest's
reservation it assumes the failure was due to reaching its full
allocation, gives up on the ballooning operation and records the limit
it reached as the "hard limit". The driver will not try again until
the target is set again (even to the same value).

However it is possible that ballooning has in fact failed due to
memory pressure in the host and therefore it is desirable to keep
attempting to reach the target in case memory becomes available. The
most likely scenario is that some guests are ballooning down while
others are ballooning up and therefore there is temporary memory
pressure while things stabilise. You would not expect a well behaved
toolstack to ask a domain to balloon to more than its allocation nor
would you expect it to deliberately over-commit memory by setting
balloon targets which exceed the total host memory.

This patch drops the concept of a hard limit and causes the balloon
driver to retry increasing the reservation on a timer in the same
manner as when decreasing the reservation.

Also if we partially succeed in increasing the reservation
(i.e. receive less pages than we asked for) then we may as well keep
those pages rather than returning them to Xen.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
author Keir Fraser <keir.fraser@citrix.com>
date Fri Jun 05 14:01:20 2009 +0100 (2009-06-05)
parents 831230e53067
children
line source
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: RAW driver (CONFIG_RAW_DRIVER)
10 When: December 2005
11 Why: declared obsolete since kernel 2.6.3
12 O_DIRECT can be used instead
13 Who: Adrian Bunk <bunk@stusta.de>
15 ---------------------------
17 What: drivers that were depending on OBSOLETE_OSS_DRIVER
18 (config options already removed)
19 When: before 2.6.19
20 Why: OSS drivers with ALSA replacements
21 Who: Adrian Bunk <bunk@stusta.de>
23 ---------------------------
25 What: raw1394: requests of type RAW1394_REQ_ISO_SEND, RAW1394_REQ_ISO_LISTEN
26 When: November 2006
27 Why: Deprecated in favour of the new ioctl-based rawiso interface, which is
28 more efficient. You should really be using libraw1394 for raw1394
29 access anyway.
30 Who: Jody McIntyre <scjody@modernduck.com>
32 ---------------------------
34 What: sbp2: module parameter "force_inquiry_hack"
35 When: July 2006
36 Why: Superceded by parameter "workarounds". Both parameters are meant to be
37 used ad-hoc and for single devices only, i.e. not in modprobe.conf,
38 therefore the impact of this feature replacement should be low.
39 Who: Stefan Richter <stefanr@s5r6.in-berlin.de>
41 ---------------------------
43 What: Video4Linux API 1 ioctls and video_decoder.h from Video devices.
44 When: July 2006
45 Why: V4L1 AP1 was replaced by V4L2 API. during migration from 2.4 to 2.6
46 series. The old API have lots of drawbacks and don't provide enough
47 means to work with all video and audio standards. The newer API is
48 already available on the main drivers and should be used instead.
49 Newer drivers should use v4l_compat_translate_ioctl function to handle
50 old calls, replacing to newer ones.
51 Decoder iocts are using internally to allow video drivers to
52 communicate with video decoders. This should also be improved to allow
53 V4L2 calls being translated into compatible internal ioctls.
54 Who: Mauro Carvalho Chehab <mchehab@brturbo.com.br>
56 ---------------------------
58 What: PCMCIA control ioctl (needed for pcmcia-cs [cardmgr, cardctl])
59 When: November 2005
60 Files: drivers/pcmcia/: pcmcia_ioctl.c
61 Why: With the 16-bit PCMCIA subsystem now behaving (almost) like a
62 normal hotpluggable bus, and with it using the default kernel
63 infrastructure (hotplug, driver core, sysfs) keeping the PCMCIA
64 control ioctl needed by cardmgr and cardctl from pcmcia-cs is
65 unnecessary, and makes further cleanups and integration of the
66 PCMCIA subsystem into the Linux kernel device driver model more
67 difficult. The features provided by cardmgr and cardctl are either
68 handled by the kernel itself now or are available in the new
69 pcmciautils package available at
70 http://kernel.org/pub/linux/utils/kernel/pcmcia/
71 Who: Dominik Brodowski <linux@brodo.de>
73 ---------------------------
75 What: ip_queue and ip6_queue (old ipv4-only and ipv6-only netfilter queue)
76 When: December 2005
77 Why: This interface has been obsoleted by the new layer3-independent
78 "nfnetlink_queue". The Kernel interface is compatible, so the old
79 ip[6]tables "QUEUE" targets still work and will transparently handle
80 all packets into nfnetlink queue number 0. Userspace users will have
81 to link against API-compatible library on top of libnfnetlink_queue
82 instead of the current 'libipq'.
83 Who: Harald Welte <laforge@netfilter.org>
85 ---------------------------
87 What: remove EXPORT_SYMBOL(kernel_thread)
88 When: August 2006
89 Files: arch/*/kernel/*_ksyms.c
90 Why: kernel_thread is a low-level implementation detail. Drivers should
91 use the <linux/kthread.h> API instead which shields them from
92 implementation details and provides a higherlevel interface that
93 prevents bugs and code duplication
94 Who: Christoph Hellwig <hch@lst.de>
96 ---------------------------
98 What: CONFIG_FORCED_INLINING
99 When: June 2006
100 Why: Config option is there to see if gcc is good enough. (in january
101 2006). If it is, the behavior should just be the default. If it's not,
102 the option should just go away entirely.
103 Who: Arjan van de Ven
105 ---------------------------
107 What: START_ARRAY ioctl for md
108 When: July 2006
109 Files: drivers/md/md.c
110 Why: Not reliable by design - can fail when most needed.
111 Alternatives exist
112 Who: NeilBrown <neilb@suse.de>
114 ---------------------------
116 What: eepro100 network driver
117 When: January 2007
118 Why: replaced by the e100 driver
119 Who: Adrian Bunk <bunk@stusta.de>
121 ---------------------------
123 What: drivers depending on OSS_OBSOLETE_DRIVER
124 When: options in 2.6.20, code in 2.6.22
125 Why: OSS drivers with ALSA replacements
126 Who: Adrian Bunk <bunk@stusta.de>
128 ---------------------------
130 What: pci_module_init(driver)
131 When: January 2007
132 Why: Is replaced by pci_register_driver(pci_driver).
133 Who: Richard Knutsson <ricknu-0@student.ltu.se> and Greg Kroah-Hartman <gregkh@suse.de>
135 ---------------------------
137 What: Usage of invalid timevals in setitimer
138 When: March 2007
139 Why: POSIX requires to validate timevals in the setitimer call. This
140 was never done by Linux. The invalid (e.g. negative timevals) were
141 silently converted to more or less random timeouts and intervals.
142 Until the removal a per boot limited number of warnings is printed
143 and the timevals are sanitized.
145 Who: Thomas Gleixner <tglx@linutronix.de>
147 ---------------------------
149 What: I2C interface of the it87 driver
150 When: January 2007
151 Why: The ISA interface is faster and should be always available. The I2C
152 probing is also known to cause trouble in at least one case (see
153 bug #5889.)
154 Who: Jean Delvare <khali@linux-fr.org>
156 ---------------------------
158 What: Unused EXPORT_SYMBOL/EXPORT_SYMBOL_GPL exports
159 (temporary transition config option provided until then)
160 The transition config option will also be removed at the same time.
161 When: before 2.6.19
162 Why: Unused symbols are both increasing the size of the kernel binary
163 and are often a sign of "wrong API"
164 Who: Arjan van de Ven <arjan@linux.intel.com>
166 ---------------------------
168 What: mount/umount uevents
169 When: February 2007
170 Why: These events are not correct, and do not properly let userspace know
171 when a file system has been mounted or unmounted. Userspace should
172 poll the /proc/mounts file instead to detect this properly.
173 Who: Greg Kroah-Hartman <gregkh@suse.de>
175 ---------------------------
177 What: USB driver API moves to EXPORT_SYMBOL_GPL
178 When: Febuary 2008
179 Files: include/linux/usb.h, drivers/usb/core/driver.c
180 Why: The USB subsystem has changed a lot over time, and it has been
181 possible to create userspace USB drivers using usbfs/libusb/gadgetfs
182 that operate as fast as the USB bus allows. Because of this, the USB
183 subsystem will not be allowing closed source kernel drivers to
184 register with it, after this grace period is over. If anyone needs
185 any help in converting their closed source drivers over to use the
186 userspace filesystems, please contact the
187 linux-usb-devel@lists.sourceforge.net mailing list, and the developers
188 there will be glad to help you out.
189 Who: Greg Kroah-Hartman <gregkh@suse.de>
191 ---------------------------
193 What: find_trylock_page
194 When: January 2007
195 Why: The interface no longer has any callers left in the kernel. It
196 is an odd interface (compared with other find_*_page functions), in
197 that it does not take a refcount to the page, only the page lock.
198 It should be replaced with find_get_page or find_lock_page if possible.
199 This feature removal can be reevaluated if users of the interface
200 cannot cleanly use something else.
201 Who: Nick Piggin <npiggin@suse.de>
203 ---------------------------
205 What: Support for the MIPS EV96100 evaluation board
206 When: September 2006
207 Why: Does no longer build since at least November 15, 2003, apparently
208 no userbase left.
209 Who: Ralf Baechle <ralf@linux-mips.org>
211 ---------------------------
213 What: Support for the Momentum / PMC-Sierra Jaguar ATX evaluation board
214 When: September 2006
215 Why: Does no longer build since quite some time, and was never popular,
216 due to the platform being replaced by successor models. Apparently
217 no user base left. It also is one of the last users of
218 WANT_PAGE_VIRTUAL.
219 Who: Ralf Baechle <ralf@linux-mips.org>
221 ---------------------------
223 What: Support for the Momentum Ocelot, Ocelot 3, Ocelot C and Ocelot G
224 When: September 2006
225 Why: Some do no longer build and apparently there is no user base left
226 for these platforms.
227 Who: Ralf Baechle <ralf@linux-mips.org>
229 ---------------------------
231 What: Support for MIPS Technologies' Altas and SEAD evaluation board
232 When: September 2006
233 Why: Some do no longer build and apparently there is no user base left
234 for these platforms. Hardware out of production since several years.
235 Who: Ralf Baechle <ralf@linux-mips.org>
237 ---------------------------
239 What: Support for the IT8172-based platforms, ITE 8172G and Globespan IVR
240 When: September 2006
241 Why: Code does no longer build since at least 2.6.0, apparently there is
242 no user base left for these platforms. Hardware out of production
243 since several years and hardly a trace of the manufacturer left on
244 the net.
245 Who: Ralf Baechle <ralf@linux-mips.org>
247 ---------------------------
249 What: Interrupt only SA_* flags
250 When: Januar 2007
251 Why: The interrupt related SA_* flags are replaced by IRQF_* to move them
252 out of the signal namespace.
254 Who: Thomas Gleixner <tglx@linutronix.de>
256 ---------------------------
258 What: i2c-ite and i2c-algo-ite drivers
259 When: September 2006
260 Why: These drivers never compiled since they were added to the kernel
261 tree 5 years ago. This feature removal can be reevaluated if
262 someone shows interest in the drivers, fixes them and takes over
263 maintenance.
264 http://marc.theaimsgroup.com/?l=linux-mips&m=115040510817448
265 Who: Jean Delvare <khali@linux-fr.org>
267 ---------------------------
269 What: Bridge netfilter deferred IPv4/IPv6 output hook calling
270 When: January 2007
271 Why: The deferred output hooks are a layering violation causing unusual
272 and broken behaviour on bridge devices. Examples of things they
273 break include QoS classifation using the MARK or CLASSIFY targets,
274 the IPsec policy match and connection tracking with VLANs on a
275 bridge. Their only use is to enable bridge output port filtering
276 within iptables with the physdev match, which can also be done by
277 combining iptables and ebtables using netfilter marks. Until it
278 will get removed the hook deferral is disabled by default and is
279 only enabled when needed.
281 Who: Patrick McHardy <kaber@trash.net>
283 ---------------------------
285 What: frame diverter
286 When: November 2006
287 Why: The frame diverter is included in most distribution kernels, but is
288 broken. It does not correctly handle many things:
289 - IPV6
290 - non-linear skb's
291 - network device RCU on removal
292 - input frames not correctly checked for protocol errors
293 It also adds allocation overhead even if not enabled.
294 It is not clear if anyone is still using it.
295 Who: Stephen Hemminger <shemminger@osdl.org>