ia64/linux-2.6.18-xen.hg

view Documentation/irqflags-tracing.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 IRQ-flags state tracing
3 started by Ingo Molnar <mingo@redhat.com>
5 the "irq-flags tracing" feature "traces" hardirq and softirq state, in
6 that it gives interested subsystems an opportunity to be notified of
7 every hardirqs-off/hardirqs-on, softirqs-off/softirqs-on event that
8 happens in the kernel.
10 CONFIG_TRACE_IRQFLAGS_SUPPORT is needed for CONFIG_PROVE_SPIN_LOCKING
11 and CONFIG_PROVE_RW_LOCKING to be offered by the generic lock debugging
12 code. Otherwise only CONFIG_PROVE_MUTEX_LOCKING and
13 CONFIG_PROVE_RWSEM_LOCKING will be offered on an architecture - these
14 are locking APIs that are not used in IRQ context. (the one exception
15 for rwsems is worked around)
17 architecture support for this is certainly not in the "trivial"
18 category, because lots of lowlevel assembly code deal with irq-flags
19 state changes. But an architecture can be irq-flags-tracing enabled in a
20 rather straightforward and risk-free manner.
22 Architectures that want to support this need to do a couple of
23 code-organizational changes first:
25 - move their irq-flags manipulation code from their asm/system.h header
26 to asm/irqflags.h
28 - rename local_irq_disable()/etc to raw_local_irq_disable()/etc. so that
29 the linux/irqflags.h code can inject callbacks and can construct the
30 real local_irq_disable()/etc APIs.
32 - add and enable TRACE_IRQFLAGS_SUPPORT in their arch level Kconfig file
34 and then a couple of functional changes are needed as well to implement
35 irq-flags-tracing support:
37 - in lowlevel entry code add (build-conditional) calls to the
38 trace_hardirqs_off()/trace_hardirqs_on() functions. The lock validator
39 closely guards whether the 'real' irq-flags matches the 'virtual'
40 irq-flags state, and complains loudly (and turns itself off) if the
41 two do not match. Usually most of the time for arch support for
42 irq-flags-tracing is spent in this state: look at the lockdep
43 complaint, try to figure out the assembly code we did not cover yet,
44 fix and repeat. Once the system has booted up and works without a
45 lockdep complaint in the irq-flags-tracing functions arch support is
46 complete.
47 - if the architecture has non-maskable interrupts then those need to be
48 excluded from the irq-tracing [and lock validation] mechanism via
49 lockdep_off()/lockdep_on().
51 in general there is no risk from having an incomplete irq-flags-tracing
52 implementation in an architecture: lockdep will detect that and will
53 turn itself off. I.e. the lock validator will still be reliable. There
54 should be no crashes due to irq-tracing bugs. (except if the assembly
55 changes break other code by modifying conditions or registers that
56 shouldnt be)