ia64/linux-2.6.18-xen.hg

view lib/Kconfig.debug @ 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
2 config PRINTK_TIME
3 bool "Show timing information on printks"
4 help
5 Selecting this option causes timing information to be
6 included in printk output. This allows you to measure
7 the interval between kernel operations, including bootup
8 operations. This is useful for identifying long delays
9 in kernel startup.
12 config MAGIC_SYSRQ
13 bool "Magic SysRq key"
14 depends on !UML
15 help
16 If you say Y here, you will have some control over the system even
17 if the system crashes for example during kernel debugging (e.g., you
18 will be able to flush the buffer cache to disk, reboot the system
19 immediately or dump some status information). This is accomplished
20 by pressing various keys while holding SysRq (Alt+PrintScreen). It
21 also works on a serial console (on PC hardware at least), if you
22 send a BREAK and then within 5 seconds a command keypress. The
23 keys are documented in <file:Documentation/sysrq.txt>. Don't say Y
24 unless you really know what this hack does.
26 config UNUSED_SYMBOLS
27 bool "Enable unused/obsolete exported symbols"
28 default y if X86
29 help
30 Unused but exported symbols make the kernel needlessly bigger. For
31 that reason most of these unused exports will soon be removed. This
32 option is provided temporarily to provide a transition period in case
33 some external kernel module needs one of these symbols anyway. If you
34 encounter such a case in your module, consider if you are actually
35 using the right API. (rationale: since nobody in the kernel is using
36 this in a module, there is a pretty good chance it's actually the
37 wrong interface to use). If you really need the symbol, please send a
38 mail to the linux kernel mailing list mentioning the symbol and why
39 you really need it, and what the merge plan to the mainline kernel for
40 your module is.
42 config DEBUG_KERNEL
43 bool "Kernel debugging"
44 help
45 Say Y here if you are developing drivers or trying to debug and
46 identify kernel problems.
48 config LOG_BUF_SHIFT
49 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)" if DEBUG_KERNEL
50 range 12 21
51 default 17 if S390 || LOCKDEP
52 default 16 if X86_NUMAQ || IA64
53 default 15 if SMP
54 default 14
55 help
56 Select kernel log buffer size as a power of 2.
57 Defaults and Examples:
58 17 => 128 KB for S/390
59 16 => 64 KB for x86 NUMAQ or IA-64
60 15 => 32 KB for SMP
61 14 => 16 KB for uniprocessor
62 13 => 8 KB
63 12 => 4 KB
65 config DETECT_SOFTLOCKUP
66 bool "Detect Soft Lockups"
67 depends on DEBUG_KERNEL
68 default y
69 help
70 Say Y here to enable the kernel to detect "soft lockups",
71 which are bugs that cause the kernel to loop in kernel
72 mode for more than 10 seconds, without giving other tasks a
73 chance to run.
75 When a soft-lockup is detected, the kernel will print the
76 current stack trace (which you should report), but the
77 system will stay locked up. This feature has negligible
78 overhead.
80 (Note that "hard lockups" are separate type of bugs that
81 can be detected via the NMI-watchdog, on platforms that
82 support it.)
84 config SCHEDSTATS
85 bool "Collect scheduler statistics"
86 depends on DEBUG_KERNEL && PROC_FS
87 help
88 If you say Y here, additional code will be inserted into the
89 scheduler and related routines to collect statistics about
90 scheduler behavior and provide them in /proc/schedstat. These
91 stats may be useful for both tuning and debugging the scheduler
92 If you aren't debugging the scheduler or trying to tune a specific
93 application, you can say N to avoid the very slight overhead
94 this adds.
96 config DEBUG_SLAB
97 bool "Debug slab memory allocations"
98 depends on DEBUG_KERNEL && SLAB
99 help
100 Say Y here to have the kernel do limited verification on memory
101 allocation as well as poisoning memory on free to catch use of freed
102 memory. This can make kmalloc/kfree-intensive workloads much slower.
104 config DEBUG_SLAB_LEAK
105 bool "Memory leak debugging"
106 depends on DEBUG_SLAB
108 config DEBUG_PREEMPT
109 bool "Debug preemptible kernel"
110 depends on DEBUG_KERNEL && PREEMPT && TRACE_IRQFLAGS_SUPPORT
111 default y
112 help
113 If you say Y here then the kernel will use a debug variant of the
114 commonly used smp_processor_id() function and will print warnings
115 if kernel code uses it in a preemption-unsafe way. Also, the kernel
116 will detect preemption count underflows.
118 config DEBUG_RT_MUTEXES
119 bool "RT Mutex debugging, deadlock detection"
120 depends on DEBUG_KERNEL && RT_MUTEXES
121 help
122 This allows rt mutex semantics violations and rt mutex related
123 deadlocks (lockups) to be detected and reported automatically.
125 config DEBUG_PI_LIST
126 bool
127 default y
128 depends on DEBUG_RT_MUTEXES
130 config RT_MUTEX_TESTER
131 bool "Built-in scriptable tester for rt-mutexes"
132 depends on DEBUG_KERNEL && RT_MUTEXES
133 help
134 This option enables a rt-mutex tester.
136 config DEBUG_SPINLOCK
137 bool "Spinlock and rw-lock debugging: basic checks"
138 depends on DEBUG_KERNEL
139 help
140 Say Y here and build SMP to catch missing spinlock initialization
141 and certain other kinds of spinlock errors commonly made. This is
142 best used in conjunction with the NMI watchdog so that spinlock
143 deadlocks are also debuggable.
145 config DEBUG_MUTEXES
146 bool "Mutex debugging: basic checks"
147 depends on DEBUG_KERNEL
148 help
149 This feature allows mutex semantics violations to be detected and
150 reported.
152 config DEBUG_RWSEMS
153 bool "RW-sem debugging: basic checks"
154 depends on DEBUG_KERNEL
155 help
156 This feature allows read-write semaphore semantics violations to
157 be detected and reported.
159 config DEBUG_LOCK_ALLOC
160 bool "Lock debugging: detect incorrect freeing of live locks"
161 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
162 select DEBUG_SPINLOCK
163 select DEBUG_MUTEXES
164 select DEBUG_RWSEMS
165 select LOCKDEP
166 help
167 This feature will check whether any held lock (spinlock, rwlock,
168 mutex or rwsem) is incorrectly freed by the kernel, via any of the
169 memory-freeing routines (kfree(), kmem_cache_free(), free_pages(),
170 vfree(), etc.), whether a live lock is incorrectly reinitialized via
171 spin_lock_init()/mutex_init()/etc., or whether there is any lock
172 held during task exit.
174 config PROVE_LOCKING
175 bool "Lock debugging: prove locking correctness"
176 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
177 select LOCKDEP
178 select DEBUG_SPINLOCK
179 select DEBUG_MUTEXES
180 select DEBUG_RWSEMS
181 select DEBUG_LOCK_ALLOC
182 default n
183 help
184 This feature enables the kernel to prove that all locking
185 that occurs in the kernel runtime is mathematically
186 correct: that under no circumstance could an arbitrary (and
187 not yet triggered) combination of observed locking
188 sequences (on an arbitrary number of CPUs, running an
189 arbitrary number of tasks and interrupt contexts) cause a
190 deadlock.
192 In short, this feature enables the kernel to report locking
193 related deadlocks before they actually occur.
195 The proof does not depend on how hard and complex a
196 deadlock scenario would be to trigger: how many
197 participant CPUs, tasks and irq-contexts would be needed
198 for it to trigger. The proof also does not depend on
199 timing: if a race and a resulting deadlock is possible
200 theoretically (no matter how unlikely the race scenario
201 is), it will be proven so and will immediately be
202 reported by the kernel (once the event is observed that
203 makes the deadlock theoretically possible).
205 If a deadlock is impossible (i.e. the locking rules, as
206 observed by the kernel, are mathematically correct), the
207 kernel reports nothing.
209 NOTE: this feature can also be enabled for rwlocks, mutexes
210 and rwsems - in which case all dependencies between these
211 different locking variants are observed and mapped too, and
212 the proof of observed correctness is also maintained for an
213 arbitrary combination of these separate locking variants.
215 For more details, see Documentation/lockdep-design.txt.
217 config LOCKDEP
218 bool
219 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
220 select STACKTRACE
221 select FRAME_POINTER
222 select KALLSYMS
223 select KALLSYMS_ALL
225 config DEBUG_LOCKDEP
226 bool "Lock dependency engine debugging"
227 depends on DEBUG_KERNEL && LOCKDEP
228 help
229 If you say Y here, the lock dependency engine will do
230 additional runtime checks to debug itself, at the price
231 of more runtime overhead.
233 config TRACE_IRQFLAGS
234 depends on DEBUG_KERNEL
235 bool
236 default y
237 depends on TRACE_IRQFLAGS_SUPPORT
238 depends on PROVE_LOCKING
240 config DEBUG_SPINLOCK_SLEEP
241 bool "Spinlock debugging: sleep-inside-spinlock checking"
242 depends on DEBUG_KERNEL
243 help
244 If you say Y here, various routines which may sleep will become very
245 noisy if they are called with a spinlock held.
247 config DEBUG_LOCKING_API_SELFTESTS
248 bool "Locking API boot-time self-tests"
249 depends on DEBUG_KERNEL
250 help
251 Say Y here if you want the kernel to run a short self-test during
252 bootup. The self-test checks whether common types of locking bugs
253 are detected by debugging mechanisms or not. (if you disable
254 lock debugging then those bugs wont be detected of course.)
255 The following locking APIs are covered: spinlocks, rwlocks,
256 mutexes and rwsems.
258 config STACKTRACE
259 bool
260 depends on DEBUG_KERNEL
261 depends on STACKTRACE_SUPPORT
263 config DEBUG_KOBJECT
264 bool "kobject debugging"
265 depends on DEBUG_KERNEL
266 help
267 If you say Y here, some extra kobject debugging messages will be sent
268 to the syslog.
270 config DEBUG_HIGHMEM
271 bool "Highmem debugging"
272 depends on DEBUG_KERNEL && HIGHMEM
273 help
274 This options enables addition error checking for high memory systems.
275 Disable for production systems.
277 config DEBUG_BUGVERBOSE
278 bool "Verbose BUG() reporting (adds 70K)" if DEBUG_KERNEL && EMBEDDED
279 depends on BUG
280 depends on ARM || ARM26 || M32R || M68K || SPARC32 || SPARC64 || X86_32 || FRV
281 default !EMBEDDED
282 help
283 Say Y here to make BUG() panics output the file name and line number
284 of the BUG call as well as the EIP and oops trace. This aids
285 debugging but costs about 70-100K of memory.
287 config DEBUG_INFO
288 bool "Compile the kernel with debug info"
289 depends on DEBUG_KERNEL
290 help
291 If you say Y here the resulting kernel image will include
292 debugging info resulting in a larger kernel image.
293 Say Y here only if you plan to debug the kernel.
295 If unsure, say N.
297 config DEBUG_FS
298 bool "Debug Filesystem"
299 depends on SYSFS
300 help
301 debugfs is a virtual file system that kernel developers use to put
302 debugging files into. Enable this option to be able to read and
303 write to these files.
305 If unsure, say N.
307 config DEBUG_VM
308 bool "Debug VM"
309 depends on DEBUG_KERNEL
310 help
311 Enable this to turn on extended checks in the virtual-memory system
312 that may impact performance.
314 If unsure, say N.
316 config FRAME_POINTER
317 bool "Compile the kernel with frame pointers"
318 depends on DEBUG_KERNEL && (X86 || CRIS || M68K || M68KNOMMU || FRV || UML || S390)
319 default y if DEBUG_INFO && UML
320 help
321 If you say Y here the resulting kernel image will be slightly larger
322 and slower, but it might give very useful debugging information on
323 some architectures or if you use external debuggers.
324 If you don't debug the kernel, you can say N.
326 config UNWIND_INFO
327 bool "Compile the kernel with frame unwind information"
328 depends on !IA64 && !PARISC
329 depends on !MODULES || !(MIPS || PPC || SUPERH || V850)
330 help
331 If you say Y here the resulting kernel image will be slightly larger
332 but not slower, and it will give very useful debugging information.
333 If you don't debug the kernel, you can say N, but we may not be able
334 to solve problems without frame unwind information or frame pointers.
336 config STACK_UNWIND
337 bool "Stack unwind support"
338 depends on UNWIND_INFO
339 depends on X86
340 help
341 This enables more precise stack traces, omitting all unrelated
342 occurrences of pointers into kernel code from the dump.
344 config FORCED_INLINING
345 bool "Force gcc to inline functions marked 'inline'"
346 depends on DEBUG_KERNEL
347 default y
348 help
349 This option determines if the kernel forces gcc to inline the functions
350 developers have marked 'inline'. Doing so takes away freedom from gcc to
351 do what it thinks is best, which is desirable for the gcc 3.x series of
352 compilers. The gcc 4.x series have a rewritten inlining algorithm and
353 disabling this option will generate a smaller kernel there. Hopefully
354 this algorithm is so good that allowing gcc4 to make the decision can
355 become the default in the future, until then this option is there to
356 test gcc for this.
358 config RCU_TORTURE_TEST
359 tristate "torture tests for RCU"
360 depends on DEBUG_KERNEL
361 default n
362 help
363 This option provides a kernel module that runs torture tests
364 on the RCU infrastructure. The kernel module may be built
365 after the fact on the running kernel to be tested, if desired.
367 Say Y here if you want RCU torture tests to start automatically
368 at boot time (you probably don't).
369 Say M if you want the RCU torture tests to build as a module.
370 Say N if you are unsure.