1.1 --- a/Makefile	Tue Sep 20 09:43:29 2005 +0000
     1.2 +++ b/Makefile	Tue Sep 20 09:43:46 2005 +0000
     1.3 @@ -166,27 +166,25 @@ uninstall: D=$(DESTDIR)
     1.4  uninstall:
     1.5  	[ -d $(D)/etc/xen ] && mv -f $(D)/etc/xen $(D)/etc/xen.old-`date +%s`
     1.6  	rm -rf $(D)/etc/init.d/xend*
     1.7 -	rm -rf $(D)/usr/$(LIBDIR)/libxc* $(D)/usr/$(LIBDIR)/libxutil*
     1.8 -	rm -rf $(D)/usr/$(LIBDIR)/python/xen $(D)/usr/include/xen
     1.9 -	rm -rf $(D)/usr/$(LIBDIR)/share/xen $(D)/usr/$(LIBDIR)/libxenstore*
    1.10 +	rm -rf $(D)/etc/hotplug/xen-backend.agent
    1.11  	rm -rf $(D)/var/run/xen* $(D)/var/lib/xen*
    1.12 -	rm -rf $(D)/usr/include/xcs_proto.h $(D)/usr/include/xc.h
    1.13 -	rm -rf $(D)/usr/include/xs_lib.h $(D)/usr/include/xs.h
    1.14 -	rm -rf $(D)/usr/sbin/xcs $(D)/usr/sbin/xcsdump $(D)/usr/sbin/xen*
    1.15 -	rm -rf $(D)/usr/sbin/netfix
    1.16 -	rm -rf $(D)/usr/sbin/xfrd $(D)/usr/sbin/xm
    1.17 -	rm -rf $(D)/usr/share/doc/xen  $(D)/usr/man/man*/xentrace*
    1.18 -	rm -rf $(D)/usr/bin/xen* $(D)/usr/bin/miniterm
    1.19  	rm -rf $(D)/boot/*xen*
    1.20  	rm -rf $(D)/lib/modules/*xen*
    1.21 +	rm -rf $(D)/usr/bin/xen* $(D)/usr/bin/lomount
    1.22  	rm -rf $(D)/usr/bin/cpuperf-perfcntr $(D)/usr/bin/cpuperf-xen
    1.23  	rm -rf $(D)/usr/bin/xc_shadow
    1.24 -	rm -rf $(D)/usr/share/xen $(D)/usr/libexec/xen
    1.25 +	rm -rf $(D)/usr/include/xenctrl.h
    1.26 +	rm -rf $(D)/usr/include/xs_lib.h $(D)/usr/include/xs.h
    1.27 +	rm -rf $(D)/usr/include/xen
    1.28 +	rm -rf $(D)/usr/$(LIBDIR)/libxenctrl* $(D)/usr/$(LIBDIR)/libxenguest*
    1.29 +	rm -rf $(D)/usr/$(LIBDIR)/libxenstore*
    1.30 +	rm -rf $(D)/usr/$(LIBDIR)/python/xen $(D)/usr/$(LIBDIR)/xen 
    1.31 +	rm -rf $(D)/usr/libexec/xen
    1.32 +	rm -rf $(D)/usr/sbin/xen* $(D)/usr/sbin/netfix $(D)/usr/sbin/xm
    1.33 +	rm -rf $(D)/usr/share/doc/xen
    1.34 +	rm -rf $(D)/usr/share/xen
    1.35  	rm -rf $(D)/usr/share/man/man1/xen*
    1.36  	rm -rf $(D)/usr/share/man/man8/xen*
    1.37 -	rm -rf $(D)/usr/lib/xen
    1.38 -	rm -rf $(D)/etc/hotplug.d/xen-backend
    1.39 -	rm -rf $(D)/etc/hotplug/xen-backend.agent
    1.40  
    1.41  # Legacy targets for compatibility
    1.42  linux24:

     2.1 --- a/docs/Makefile	Tue Sep 20 09:43:29 2005 +0000
     2.2 +++ b/docs/Makefile	Tue Sep 20 09:43:46 2005 +0000
     2.3 @@ -12,7 +12,7 @@ DOXYGEN		:= doxygen
     2.4  
     2.5  pkgdocdir	:= /usr/share/doc/xen
     2.6  
     2.7 -DOC_TEX		:= $(wildcard src/*.tex)
     2.8 +DOC_TEX		:= src/user.tex src/interface.tex
     2.9  DOC_PS		:= $(patsubst src/%.tex,ps/%.ps,$(DOC_TEX))
    2.10  DOC_PDF		:= $(patsubst src/%.tex,pdf/%.pdf,$(DOC_TEX))
    2.11  DOC_HTML	:= $(patsubst src/%.tex,html/%/index.html,$(DOC_TEX))

     3.1 --- a/docs/src/interface.tex	Tue Sep 20 09:43:29 2005 +0000
     3.2 +++ b/docs/src/interface.tex	Tue Sep 20 09:43:46 2005 +0000
     3.3 @@ -87,1084 +87,23 @@ itself, allows the Xen framework to sepa
     3.4  mechanism and policy within the system.
     3.5  
     3.6  
     3.7 -
     3.8 -\chapter{Virtual Architecture}
     3.9 -
    3.10 -On a Xen-based system, the hypervisor itself runs in {\it ring 0}.  It
    3.11 -has full access to the physical memory available in the system and is
    3.12 -responsible for allocating portions of it to the domains.  Guest
    3.13 -operating systems run in and use {\it rings 1}, {\it 2} and {\it 3} as
    3.14 -they see fit. Segmentation is used to prevent the guest OS from
    3.15 -accessing the portion of the address space that is reserved for
    3.16 -Xen. We expect most guest operating systems will use ring 1 for their
    3.17 -own operation and place applications in ring 3.
    3.18 -
    3.19 -In this chapter we consider the basic virtual architecture provided 
    3.20 -by Xen: the basic CPU state, exception and interrupt handling, and
    3.21 -time. Other aspects such as memory and device access are discussed 
    3.22 -in later chapters. 
    3.23 -
    3.24 -\section{CPU state}
    3.25 -
    3.26 -All privileged state must be handled by Xen.  The guest OS has no
    3.27 -direct access to CR3 and is not permitted to update privileged bits in
    3.28 -EFLAGS. Guest OSes use \emph{hypercalls} to invoke operations in Xen; 
    3.29 -these are analogous to system calls but occur from ring 1 to ring 0. 
    3.30 -
    3.31 -A list of all hypercalls is given in Appendix~\ref{a:hypercalls}. 
    3.32 -
    3.33 -
    3.34 -
    3.35 -\section{Exceptions}
    3.36 -
    3.37 -A virtual IDT is provided --- a domain can submit a table of trap
    3.38 -handlers to Xen via the {\tt set\_trap\_table()} hypercall.  Most trap
    3.39 -handlers are identical to native x86 handlers, although the page-fault
    3.40 -handler is somewhat different.
    3.41 -
    3.42 -
    3.43 -\section{Interrupts and events}
    3.44 -
    3.45 -Interrupts are virtualized by mapping them to \emph{events}, which are
    3.46 -delivered asynchronously to the target domain using a callback
    3.47 -supplied via the {\tt set\_callbacks()} hypercall.  A guest OS can map
    3.48 -these events onto its standard interrupt dispatch mechanisms.  Xen is
    3.49 -responsible for determining the target domain that will handle each
    3.50 -physical interrupt source. For more details on the binding of event
    3.51 -sources to events, see Chapter~\ref{c:devices}. 
    3.52 -
    3.53 -
    3.54 -
    3.55 -\section{Time}
    3.56 -
    3.57 -Guest operating systems need to be aware of the passage of both real
    3.58 -(or wallclock) time and their own `virtual time' (the time for
    3.59 -which they have been executing). Furthermore, Xen has a notion of 
    3.60 -time which is used for scheduling. The following notions of 
    3.61 -time are provided: 
    3.62 -
    3.63 -\begin{description}
    3.64 -\item[Cycle counter time.]
    3.65 -
    3.66 -This provides a fine-grained time reference.  The cycle counter time is
    3.67 -used to accurately extrapolate the other time references.  On SMP machines
    3.68 -it is currently assumed that the cycle counter time is synchronized between
    3.69 -CPUs.  The current x86-based implementation achieves this within inter-CPU
    3.70 -communication latencies.
    3.71 -
    3.72 -\item[System time.]
    3.73 -
    3.74 -This is a 64-bit counter which holds the number of nanoseconds that
    3.75 -have elapsed since system boot.
    3.76 -
    3.77 -
    3.78 -\item[Wall clock time.]
    3.79 -
    3.80 -This is the time of day in a Unix-style {\tt struct timeval} (seconds
    3.81 -and microseconds since 1 January 1970, adjusted by leap seconds).  An
    3.82 -NTP client hosted by {\it domain 0} can keep this value accurate.  
    3.83 -
    3.84 -
    3.85 -\item[Domain virtual time.]
    3.86 -
    3.87 -This progresses at the same pace as system time, but only while a
    3.88 -domain is executing --- it stops while a domain is de-scheduled.
    3.89 -Therefore the share of the CPU that a domain receives is indicated by
    3.90 -the rate at which its virtual time increases.
    3.91 -
    3.92 -\end{description}
    3.93 -
    3.94 -
    3.95 -Xen exports timestamps for system time and wall-clock time to guest
    3.96 -operating systems through a shared page of memory.  Xen also provides
    3.97 -the cycle counter time at the instant the timestamps were calculated,
    3.98 -and the CPU frequency in Hertz.  This allows the guest to extrapolate
    3.99 -system and wall-clock times accurately based on the current cycle
   3.100 -counter time.
   3.101 -
   3.102 -Since all time stamps need to be updated and read \emph{atomically}
   3.103 -two version numbers are also stored in the shared info page. The 
   3.104 -first is incremented prior to an update, while the second is only
   3.105 -incremented afterwards. Thus a guest can be sure that it read a consistent 
   3.106 -state by checking the two version numbers are equal. 
   3.107 -
   3.108 -Xen includes a periodic ticker which sends a timer event to the
   3.109 -currently executing domain every 10ms.  The Xen scheduler also sends a
   3.110 -timer event whenever a domain is scheduled; this allows the guest OS
   3.111 -to adjust for the time that has passed while it has been inactive.  In
   3.112 -addition, Xen allows each domain to request that they receive a timer
   3.113 -event sent at a specified system time by using the {\tt
   3.114 -set\_timer\_op()} hypercall.  Guest OSes may use this timer to
   3.115 -implement timeout values when they block.
   3.116 -
   3.117 -
   3.118 -
   3.119 -%% % akw: demoting this to a section -- not sure if there is any point
   3.120 -%% % though, maybe just remove it.
   3.121 -
   3.122 -\section{Xen CPU Scheduling}
   3.123 -
   3.124 -Xen offers a uniform API for CPU schedulers.  It is possible to choose
   3.125 -from a number of schedulers at boot and it should be easy to add more.
   3.126 -The BVT, Atropos and Round Robin schedulers are part of the normal
   3.127 -Xen distribution.  BVT provides proportional fair shares of the CPU to
   3.128 -the running domains.  Atropos can be used to reserve absolute shares
   3.129 -of the CPU for each domain.  Round-robin is provided as an example of
   3.130 -Xen's internal scheduler API.
   3.131 -
   3.132 -\paragraph*{Note: SMP host support}
   3.133 -Xen has always supported SMP host systems.  Domains are statically assigned to
   3.134 -CPUs, either at creation time or when manually pinning to a particular CPU.
   3.135 -The current schedulers then run locally on each CPU to decide which of the
   3.136 -assigned domains should be run there. The user-level control software 
   3.137 -can be used to perform coarse-grain load-balancing between CPUs. 
   3.138 +%% chapter Virtual Architecture moved to architecture.tex
   3.139 +\include{src/interface/architecture}
   3.140  
   3.141 -
   3.142 -%% More information on the characteristics and use of these schedulers is
   3.143 -%% available in {\tt Sched-HOWTO.txt}.
   3.144 -
   3.145 -
   3.146 -\section{Privileged operations}
   3.147 -
   3.148 -Xen exports an extended interface to privileged domains (viz.\ {\it
   3.149 -  Domain 0}). This allows such domains to build and boot other domains 
   3.150 -on the server, and provides control interfaces for managing 
   3.151 -scheduling, memory, networking, and block devices. 
   3.152 -
   3.153 -
   3.154 -\chapter{Memory}
   3.155 -\label{c:memory} 
   3.156 -
   3.157 -Xen is responsible for managing the allocation of physical memory to
   3.158 -domains, and for ensuring safe use of the paging and segmentation
   3.159 -hardware.
   3.160 -
   3.161 -
   3.162 -\section{Memory Allocation}
   3.163 -
   3.164 -
   3.165 -Xen resides within a small fixed portion of physical memory; it also
   3.166 -reserves the top 64MB of every virtual address space. The remaining
   3.167 -physical memory is available for allocation to domains at a page
   3.168 -granularity.  Xen tracks the ownership and use of each page, which
   3.169 -allows it to enforce secure partitioning between domains.
   3.170 -
   3.171 -Each domain has a maximum and current physical memory allocation. 
   3.172 -A guest OS may run a `balloon driver' to dynamically adjust its 
   3.173 -current memory allocation up to its limit. 
   3.174 -
   3.175 -
   3.176 -%% XXX SMH: I use machine and physical in the next section (which 
   3.177 -%% is kinda required for consistency with code); wonder if this 
   3.178 -%% section should use same terms? 
   3.179 -%%
   3.180 -%% Probably. 
   3.181 -%%
   3.182 -%% Merging this and below section at some point prob makes sense. 
   3.183 -
   3.184 -\section{Pseudo-Physical Memory}
   3.185 -
   3.186 -Since physical memory is allocated and freed on a page granularity,
   3.187 -there is no guarantee that a domain will receive a contiguous stretch
   3.188 -of physical memory. However most operating systems do not have good
   3.189 -support for operating in a fragmented physical address space. To aid
   3.190 -porting such operating systems to run on top of Xen, we make a
   3.191 -distinction between \emph{machine memory} and \emph{pseudo-physical
   3.192 -memory}.
   3.193 -
   3.194 -Put simply, machine memory refers to the entire amount of memory
   3.195 -installed in the machine, including that reserved by Xen, in use by
   3.196 -various domains, or currently unallocated. We consider machine memory
   3.197 -to comprise a set of 4K \emph{machine page frames} numbered
   3.198 -consecutively starting from 0. Machine frame numbers mean the same
   3.199 -within Xen or any domain.
   3.200 -
   3.201 -Pseudo-physical memory, on the other hand, is a per-domain
   3.202 -abstraction. It allows a guest operating system to consider its memory
   3.203 -allocation to consist of a contiguous range of physical page frames
   3.204 -starting at physical frame 0, despite the fact that the underlying
   3.205 -machine page frames may be sparsely allocated and in any order.
   3.206 -
   3.207 -To achieve this, Xen maintains a globally readable {\it
   3.208 -machine-to-physical} table which records the mapping from machine page
   3.209 -frames to pseudo-physical ones. In addition, each domain is supplied
   3.210 -with a {\it physical-to-machine} table which performs the inverse
   3.211 -mapping. Clearly the machine-to-physical table has size proportional
   3.212 -to the amount of RAM installed in the machine, while each
   3.213 -physical-to-machine table has size proportional to the memory
   3.214 -allocation of the given domain.
   3.215 -
   3.216 -Architecture dependent code in guest operating systems can then use
   3.217 -the two tables to provide the abstraction of pseudo-physical
   3.218 -memory. In general, only certain specialized parts of the operating
   3.219 -system (such as page table management) needs to understand the
   3.220 -difference between machine and pseudo-physical addresses.
   3.221 -
   3.222 -\section{Page Table Updates}
   3.223 -
   3.224 -In the default mode of operation, Xen enforces read-only access to
   3.225 -page tables and requires guest operating systems to explicitly request
   3.226 -any modifications.  Xen validates all such requests and only applies
   3.227 -updates that it deems safe.  This is necessary to prevent domains from
   3.228 -adding arbitrary mappings to their page tables.
   3.229 -
   3.230 -To aid validation, Xen associates a type and reference count with each
   3.231 -memory page. A page has one of the following
   3.232 -mutually-exclusive types at any point in time: page directory ({\sf
   3.233 -PD}), page table ({\sf PT}), local descriptor table ({\sf LDT}),
   3.234 -global descriptor table ({\sf GDT}), or writable ({\sf RW}). Note that
   3.235 -a guest OS may always create readable mappings of its own memory 
   3.236 -regardless of its current type. 
   3.237 -%%% XXX: possibly explain more about ref count 'lifecyle' here?
   3.238 -This mechanism is used to
   3.239 -maintain the invariants required for safety; for example, a domain
   3.240 -cannot have a writable mapping to any part of a page table as this
   3.241 -would require the page concerned to simultaneously be of types {\sf
   3.242 -  PT} and {\sf RW}.
   3.243 -
   3.244 -
   3.245 -%\section{Writable Page Tables}
   3.246 -
   3.247 -Xen also provides an alternative mode of operation in which guests be
   3.248 -have the illusion that their page tables are directly writable.  Of
   3.249 -course this is not really the case, since Xen must still validate
   3.250 -modifications to ensure secure partitioning. To this end, Xen traps
   3.251 -any write attempt to a memory page of type {\sf PT} (i.e., that is
   3.252 -currently part of a page table).  If such an access occurs, Xen
   3.253 -temporarily allows write access to that page while at the same time
   3.254 -{\em disconnecting} it from the page table that is currently in
   3.255 -use. This allows the guest to safely make updates to the page because
   3.256 -the newly-updated entries cannot be used by the MMU until Xen
   3.257 -revalidates and reconnects the page.
   3.258 -Reconnection occurs automatically in a number of situations: for
   3.259 -example, when the guest modifies a different page-table page, when the
   3.260 -domain is preempted, or whenever the guest uses Xen's explicit
   3.261 -page-table update interfaces.
   3.262 -
   3.263 -Finally, Xen also supports a form of \emph{shadow page tables} in
   3.264 -which the guest OS uses a independent copy of page tables which are
   3.265 -unknown to the hardware (i.e.\ which are never pointed to by {\tt
   3.266 -cr3}). Instead Xen propagates changes made to the guest's tables to the
   3.267 -real ones, and vice versa. This is useful for logging page writes
   3.268 -(e.g.\ for live migration or checkpoint). A full version of the shadow
   3.269 -page tables also allows guest OS porting with less effort.
   3.270 -
   3.271 -\section{Segment Descriptor Tables}
   3.272 +%% chapter Memory moved to memory.tex
   3.273 +\include{src/interface/memory}
   3.274  
   3.275 -On boot a guest is supplied with a default GDT, which does not reside
   3.276 -within its own memory allocation.  If the guest wishes to use other
   3.277 -than the default `flat' ring-1 and ring-3 segments that this GDT
   3.278 -provides, it must register a custom GDT and/or LDT with Xen,
   3.279 -allocated from its own memory. Note that a number of GDT 
   3.280 -entries are reserved by Xen -- any custom GDT must also include
   3.281 -sufficient space for these entries. 
   3.282 -
   3.283 -For example, the following hypercall is used to specify a new GDT: 
   3.284 -
   3.285 -\begin{quote}
   3.286 -int {\bf set\_gdt}(unsigned long *{\em frame\_list}, int {\em entries})
   3.287 -
   3.288 -{\em frame\_list}: An array of up to 16 machine page frames within
   3.289 -which the GDT resides.  Any frame registered as a GDT frame may only
   3.290 -be mapped read-only within the guest's address space (e.g., no
   3.291 -writable mappings, no use as a page-table page, and so on).
   3.292 -
   3.293 -{\em entries}: The number of descriptor-entry slots in the GDT.  Note
   3.294 -that the table must be large enough to contain Xen's reserved entries;
   3.295 -thus we must have `{\em entries $>$ LAST\_RESERVED\_GDT\_ENTRY}\ '.
   3.296 -Note also that, after registering the GDT, slots {\em FIRST\_} through
   3.297 -{\em LAST\_RESERVED\_GDT\_ENTRY} are no longer usable by the guest and
   3.298 -may be overwritten by Xen.
   3.299 -\end{quote}
   3.300 -
   3.301 -The LDT is updated via the generic MMU update mechanism (i.e., via 
   3.302 -the {\tt mmu\_update()} hypercall. 
   3.303 -
   3.304 -\section{Start of Day} 
   3.305 -
   3.306 -The start-of-day environment for guest operating systems is rather
   3.307 -different to that provided by the underlying hardware. In particular,
   3.308 -the processor is already executing in protected mode with paging
   3.309 -enabled.
   3.310 -
   3.311 -{\it Domain 0} is created and booted by Xen itself. For all subsequent
   3.312 -domains, the analogue of the boot-loader is the {\it domain builder},
   3.313 -user-space software running in {\it domain 0}. The domain builder 
   3.314 -is responsible for building the initial page tables for a domain  
   3.315 -and loading its kernel image at the appropriate virtual address. 
   3.316 -
   3.317 -
   3.318 -
   3.319 -\chapter{Devices}
   3.320 -\label{c:devices}
   3.321 -
   3.322 -Devices such as network and disk are exported to guests using a
   3.323 -split device driver.  The device driver domain, which accesses the
   3.324 -physical device directly also runs a {\em backend} driver, serving
   3.325 -requests to that device from guests.  Each guest will use a simple
   3.326 -{\em frontend} driver, to access the backend.  Communication between these
   3.327 -domains is composed of two parts:  First, data is placed onto a shared
   3.328 -memory page between the domains.  Second, an event channel between the
   3.329 -two domains is used to pass notification that data is outstanding.
   3.330 -This separation of notification from data transfer allows message
   3.331 -batching, and results in very efficient device access.  
   3.332 -
   3.333 -Event channels are used extensively in device virtualization; each
   3.334 -domain has a number of end-points or \emph{ports} each of which
   3.335 -may be bound to one of the following \emph{event sources}:
   3.336 -\begin{itemize} 
   3.337 -  \item a physical interrupt from a real device, 
   3.338 -  \item a virtual interrupt (callback) from Xen, or 
   3.339 -  \item a signal from another domain 
   3.340 -\end{itemize}
   3.341 -
   3.342 -Events are lightweight and do not carry much information beyond 
   3.343 -the source of the notification. Hence when performing bulk data
   3.344 -transfer, events are typically used as synchronization primitives
   3.345 -over a shared memory transport. Event channels are managed via 
   3.346 -the {\tt event\_channel\_op()} hypercall; for more details see
   3.347 -Section~\ref{s:idc}. 
   3.348 -
   3.349 -This chapter focuses on some individual device interfaces
   3.350 -available to Xen guests. 
   3.351 -
   3.352 -\section{Network I/O}
   3.353 -
   3.354 -Virtual network device services are provided by shared memory
   3.355 -communication with a backend domain.  From the point of view of
   3.356 -other domains, the backend may be viewed as a virtual ethernet switch
   3.357 -element with each domain having one or more virtual network interfaces
   3.358 -connected to it.
   3.359 -
   3.360 -\subsection{Backend Packet Handling}
   3.361 -
   3.362 -The backend driver is responsible for a variety of actions relating to
   3.363 -the transmission and reception of packets from the physical device.
   3.364 -With regard to transmission, the backend performs these key actions:
   3.365 -
   3.366 -\begin{itemize}
   3.367 -\item {\bf Validation:} To ensure that domains do not attempt to
   3.368 -  generate invalid (e.g. spoofed) traffic, the backend driver may
   3.369 -  validate headers ensuring that source MAC and IP addresses match the
   3.370 -  interface that they have been sent from.
   3.371 -
   3.372 -  Validation functions can be configured using standard firewall rules
   3.373 -  ({\small{\tt iptables}} in the case of Linux).
   3.374 -  
   3.375 -\item {\bf Scheduling:} Since a number of domains can share a single
   3.376 -  physical network interface, the backend must mediate access when
   3.377 -  several domains each have packets queued for transmission.  This
   3.378 -  general scheduling function subsumes basic shaping or rate-limiting
   3.379 -  schemes.
   3.380 -  
   3.381 -\item {\bf Logging and Accounting:} The backend domain can be
   3.382 -  configured with classifier rules that control how packets are
   3.383 -  accounted or logged.  For example, log messages might be generated
   3.384 -  whenever a domain attempts to send a TCP packet containing a SYN.
   3.385 -\end{itemize}
   3.386 -
   3.387 -On receipt of incoming packets, the backend acts as a simple
   3.388 -demultiplexer:  Packets are passed to the appropriate virtual
   3.389 -interface after any necessary logging and accounting have been carried
   3.390 -out.
   3.391 -
   3.392 -\subsection{Data Transfer}
   3.393 -
   3.394 -Each virtual interface uses two ``descriptor rings'', one for transmit,
   3.395 -the other for receive.  Each descriptor identifies a block of contiguous
   3.396 -physical memory allocated to the domain.  
   3.397 -
   3.398 -The transmit ring carries packets to transmit from the guest to the
   3.399 -backend domain.  The return path of the transmit ring carries messages
   3.400 -indicating that the contents have been physically transmitted and the
   3.401 -backend no longer requires the associated pages of memory.
   3.402 +%% chapter Devices moved to devices.tex
   3.403 +\include{src/interface/devices}
   3.404  
   3.405 -To receive packets, the guest places descriptors of unused pages on
   3.406 -the receive ring.  The backend will return received packets by
   3.407 -exchanging these pages in the domain's memory with new pages
   3.408 -containing the received data, and passing back descriptors regarding
   3.409 -the new packets on the ring.  This zero-copy approach allows the
   3.410 -backend to maintain a pool of free pages to receive packets into, and
   3.411 -then deliver them to appropriate domains after examining their
   3.412 -headers.
   3.413 -
   3.414 -%
   3.415 -%Real physical addresses are used throughout, with the domain performing 
   3.416 -%translation from pseudo-physical addresses if that is necessary.
   3.417 -
   3.418 -If a domain does not keep its receive ring stocked with empty buffers then 
   3.419 -packets destined to it may be dropped.  This provides some defence against 
   3.420 -receive livelock problems because an overload domain will cease to receive
   3.421 -further data.  Similarly, on the transmit path, it provides the application
   3.422 -with feedback on the rate at which packets are able to leave the system.
   3.423 -
   3.424 -
   3.425 -Flow control on rings is achieved by including a pair of producer
   3.426 -indexes on the shared ring page.  Each side will maintain a private
   3.427 -consumer index indicating the next outstanding message.  In this
   3.428 -manner, the domains cooperate to divide the ring into two message
   3.429 -lists, one in each direction.  Notification is decoupled from the
   3.430 -immediate placement of new messages on the ring; the event channel
   3.431 -will be used to generate notification when {\em either} a certain
   3.432 -number of outstanding messages are queued, {\em or} a specified number
   3.433 -of nanoseconds have elapsed since the oldest message was placed on the
   3.434 -ring.
   3.435 -
   3.436 -% Not sure if my version is any better -- here is what was here before:
   3.437 -%% Synchronization between the backend domain and the guest is achieved using 
   3.438 -%% counters held in shared memory that is accessible to both.  Each ring has
   3.439 -%% associated producer and consumer indices indicating the area in the ring
   3.440 -%% that holds descriptors that contain data.  After receiving {\it n} packets
   3.441 -%% or {\t nanoseconds} after receiving the first packet, the hypervisor sends
   3.442 -%% an event to the domain. 
   3.443 -
   3.444 -\section{Block I/O}
   3.445 -
   3.446 -All guest OS disk access goes through the virtual block device VBD
   3.447 -interface.  This interface allows domains access to portions of block
   3.448 -storage devices visible to the the block backend device.  The VBD
   3.449 -interface is a split driver, similar to the network interface
   3.450 -described above.  A single shared memory ring is used between the
   3.451 -frontend and backend drivers, across which read and write messages are
   3.452 -sent.
   3.453 -
   3.454 -Any block device accessible to the backend domain, including
   3.455 -network-based block (iSCSI, *NBD, etc), loopback and LVM/MD devices,
   3.456 -can be exported as a VBD.  Each VBD is mapped to a device node in the
   3.457 -guest, specified in the guest's startup configuration.
   3.458 -
   3.459 -Old (Xen 1.2) virtual disks are not supported under Xen 2.0, since
   3.460 -similar functionality can be achieved using the more complete LVM
   3.461 -system, which is already in widespread use.
   3.462 -
   3.463 -\subsection{Data Transfer}
   3.464 -
   3.465 -The single ring between the guest and the block backend supports three
   3.466 -messages:
   3.467 -
   3.468 -\begin{description}
   3.469 -\item [{\small {\tt PROBE}}:] Return a list of the VBDs available to this guest
   3.470 -  from the backend.  The request includes a descriptor of a free page
   3.471 -  into which the reply will be written by the backend.
   3.472 -
   3.473 -\item [{\small {\tt READ}}:] Read data from the specified block device.  The
   3.474 -  front end identifies the device and location to read from and
   3.475 -  attaches pages for the data to be copied to (typically via DMA from
   3.476 -  the device).  The backend acknowledges completed read requests as
   3.477 -  they finish.
   3.478 -
   3.479 -\item [{\small {\tt WRITE}}:] Write data to the specified block device.  This
   3.480 -  functions essentially as {\small {\tt READ}}, except that the data moves to
   3.481 -  the device instead of from it.
   3.482 -\end{description}
   3.483 -
   3.484 -% um... some old text
   3.485 -%% In overview, the same style of descriptor-ring that is used for
   3.486 -%% network packets is used here.  Each domain has one ring that carries
   3.487 -%% operation requests to the hypervisor and carries the results back
   3.488 -%% again.
   3.489 -
   3.490 -%% Rather than copying data, the backend simply maps the domain's buffers
   3.491 -%% in order to enable direct DMA to them.  The act of mapping the buffers
   3.492 -%% also increases the reference counts of the underlying pages, so that
   3.493 -%% the unprivileged domain cannot try to return them to the hypervisor,
   3.494 -%% install them as page tables, or any other unsafe behaviour.
   3.495 -%% %block API here 
   3.496 -
   3.497 -
   3.498 -\chapter{Further Information} 
   3.499 -
   3.500 -
   3.501 -If you have questions that are not answered by this manual, the
   3.502 -sources of information listed below may be of interest to you.  Note
   3.503 -that bug reports, suggestions and contributions related to the
   3.504 -software (or the documentation) should be sent to the Xen developers'
   3.505 -mailing list (address below).
   3.506 -
   3.507 -\section{Other documentation}
   3.508 -
   3.509 -If you are mainly interested in using (rather than developing for)
   3.510 -Xen, the {\em Xen Users' Manual} is distributed in the {\tt docs/}
   3.511 -directory of the Xen source distribution.  
   3.512 -
   3.513 -% Various HOWTOs are also available in {\tt docs/HOWTOS}.
   3.514 -
   3.515 -\section{Online references}
   3.516 -
   3.517 -The official Xen web site is found at:
   3.518 -\begin{quote}
   3.519 -{\tt http://www.cl.cam.ac.uk/Research/SRG/netos/xen/}
   3.520 -\end{quote}
   3.521 -
   3.522 -This contains links to the latest versions of all on-line 
   3.523 -documentation. 
   3.524 -
   3.525 -\section{Mailing lists}
   3.526 -
   3.527 -There are currently four official Xen mailing lists:
   3.528 -
   3.529 -\begin{description}
   3.530 -\item[xen-devel@lists.xensource.com] Used for development
   3.531 -discussions and bug reports.  Subscribe at: \\
   3.532 -{\small {\tt http://lists.xensource.com/xen-devel}}
   3.533 -\item[xen-users@lists.xensource.com] Used for installation and usage
   3.534 -discussions and requests for help.  Subscribe at: \\
   3.535 -{\small {\tt http://lists.xensource.com/xen-users}}
   3.536 -\item[xen-announce@lists.xensource.com] Used for announcements only.
   3.537 -Subscribe at: \\
   3.538 -{\small {\tt http://lists.xensource.com/xen-announce}}
   3.539 -\item[xen-changelog@lists.xensource.com]  Changelog feed
   3.540 -from the unstable and 2.0 trees - developer oriented.  Subscribe at: \\
   3.541 -{\small {\tt http://lists.xensource.com/xen-changelog}}
   3.542 -\end{description}
   3.543 -
   3.544 -Of these, xen-devel is the most active.
   3.545 -
   3.546 -
   3.547 +%% chapter Further Information moved to further_info.tex
   3.548 +\include{src/interface/further_info}
   3.549  
   3.550  
   3.551  \appendix
   3.552  
   3.553 -%\newcommand{\hypercall}[1]{\vspace{5mm}{\large\sf #1}}
   3.554 -
   3.555 -
   3.556 -
   3.557 -
   3.558 -
   3.559 -\newcommand{\hypercall}[1]{\vspace{2mm}{\sf #1}}
   3.560 -
   3.561 -
   3.562 -
   3.563 -
   3.564 -
   3.565 -
   3.566 -\chapter{Xen Hypercalls}
   3.567 -\label{a:hypercalls}
   3.568 -
   3.569 -Hypercalls represent the procedural interface to Xen; this appendix 
   3.570 -categorizes and describes the current set of hypercalls. 
   3.571 -
   3.572 -\section{Invoking Hypercalls} 
   3.573 -
   3.574 -Hypercalls are invoked in a manner analogous to system calls in a
   3.575 -conventional operating system; a software interrupt is issued which
   3.576 -vectors to an entry point within Xen. On x86\_32 machines the
   3.577 -instruction required is {\tt int \$82}; the (real) IDT is setup so
   3.578 -that this may only be issued from within ring 1. The particular 
   3.579 -hypercall to be invoked is contained in {\tt EAX} --- a list 
   3.580 -mapping these values to symbolic hypercall names can be found 
   3.581 -in {\tt xen/include/public/xen.h}. 
   3.582 -
   3.583 -On some occasions a set of hypercalls will be required to carry
   3.584 -out a higher-level function; a good example is when a guest 
   3.585 -operating wishes to context switch to a new process which 
   3.586 -requires updating various privileged CPU state. As an optimization
   3.587 -for these cases, there is a generic mechanism to issue a set of 
   3.588 -hypercalls as a batch: 
   3.589 -
   3.590 -\begin{quote}
   3.591 -\hypercall{multicall(void *call\_list, int nr\_calls)}
   3.592 -
   3.593 -Execute a series of hypervisor calls; {\tt nr\_calls} is the length of
   3.594 -the array of {\tt multicall\_entry\_t} structures pointed to be {\tt
   3.595 -call\_list}. Each entry contains the hypercall operation code followed
   3.596 -by up to 7 word-sized arguments.
   3.597 -\end{quote}
   3.598 -
   3.599 -Note that multicalls are provided purely as an optimization; there is
   3.600 -no requirement to use them when first porting a guest operating
   3.601 -system.
   3.602 -
   3.603 -
   3.604 -\section{Virtual CPU Setup} 
   3.605 -
   3.606 -At start of day, a guest operating system needs to setup the virtual
   3.607 -CPU it is executing on. This includes installing vectors for the
   3.608 -virtual IDT so that the guest OS can handle interrupts, page faults,
   3.609 -etc. However the very first thing a guest OS must setup is a pair 
   3.610 -of hypervisor callbacks: these are the entry points which Xen will
   3.611 -use when it wishes to notify the guest OS of an occurrence. 
   3.612 -
   3.613 -\begin{quote}
   3.614 -\hypercall{set\_callbacks(unsigned long event\_selector, unsigned long
   3.615 -  event\_address, unsigned long failsafe\_selector, unsigned long
   3.616 -  failsafe\_address) }
   3.617 -
   3.618 -Register the normal (``event'') and failsafe callbacks for 
   3.619 -event processing. In each case the code segment selector and 
   3.620 -address within that segment are provided. The selectors must
   3.621 -have RPL 1; in XenLinux we simply use the kernel's CS for both 
   3.622 -{\tt event\_selector} and {\tt failsafe\_selector}.
   3.623 -
   3.624 -The value {\tt event\_address} specifies the address of the guest OSes
   3.625 -event handling and dispatch routine; the {\tt failsafe\_address}
   3.626 -specifies a separate entry point which is used only if a fault occurs
   3.627 -when Xen attempts to use the normal callback. 
   3.628 -\end{quote} 
   3.629 -
   3.630 -
   3.631 -After installing the hypervisor callbacks, the guest OS can 
   3.632 -install a `virtual IDT' by using the following hypercall: 
   3.633 -
   3.634 -\begin{quote} 
   3.635 -\hypercall{set\_trap\_table(trap\_info\_t *table)} 
   3.636 -
   3.637 -Install one or more entries into the per-domain 
   3.638 -trap handler table (essentially a software version of the IDT). 
   3.639 -Each entry in the array pointed to by {\tt table} includes the 
   3.640 -exception vector number with the corresponding segment selector 
   3.641 -and entry point. Most guest OSes can use the same handlers on 
   3.642 -Xen as when running on the real hardware; an exception is the 
   3.643 -page fault handler (exception vector 14) where a modified 
   3.644 -stack-frame layout is used. 
   3.645 -
   3.646 -
   3.647 -\end{quote} 
   3.648 -
   3.649 -
   3.650 -
   3.651 -\section{Scheduling and Timer}
   3.652 -
   3.653 -Domains are preemptively scheduled by Xen according to the 
   3.654 -parameters installed by domain 0 (see Section~\ref{s:dom0ops}). 
   3.655 -In addition, however, a domain may choose to explicitly 
   3.656 -control certain behavior with the following hypercall: 
   3.657 -
   3.658 -\begin{quote} 
   3.659 -\hypercall{sched\_op(unsigned long op)} 
   3.660 -
   3.661 -Request scheduling operation from hypervisor. The options are: {\it
   3.662 -yield}, {\it block}, and {\it shutdown}.  {\it yield} keeps the
   3.663 -calling domain runnable but may cause a reschedule if other domains
   3.664 -are runnable.  {\it block} removes the calling domain from the run
   3.665 -queue and cause is to sleeps until an event is delivered to it.  {\it
   3.666 -shutdown} is used to end the domain's execution; the caller can
   3.667 -additionally specify whether the domain should reboot, halt or
   3.668 -suspend.
   3.669 -\end{quote} 
   3.670 -
   3.671 -To aid the implementation of a process scheduler within a guest OS,
   3.672 -Xen provides a virtual programmable timer:
   3.673 -
   3.674 -\begin{quote}
   3.675 -\hypercall{set\_timer\_op(uint64\_t timeout)} 
   3.676 -
   3.677 -Request a timer event to be sent at the specified system time (time 
   3.678 -in nanoseconds since system boot). The hypercall actually passes the 
   3.679 -64-bit timeout value as a pair of 32-bit values. 
   3.680 -
   3.681 -\end{quote} 
   3.682 -
   3.683 -Note that calling {\tt set\_timer\_op()} prior to {\tt sched\_op} 
   3.684 -allows block-with-timeout semantics. 
   3.685 -
   3.686 -
   3.687 -\section{Page Table Management} 
   3.688 -
   3.689 -Since guest operating systems have read-only access to their page 
   3.690 -tables, Xen must be involved when making any changes. The following
   3.691 -multi-purpose hypercall can be used to modify page-table entries, 
   3.692 -update the machine-to-physical mapping table, flush the TLB, install 
   3.693 -a new page-table base pointer, and more.
   3.694 -
   3.695 -\begin{quote} 
   3.696 -\hypercall{mmu\_update(mmu\_update\_t *req, int count, int *success\_count)} 
   3.697 -
   3.698 -Update the page table for the domain; a set of {\tt count} updates are
   3.699 -submitted for processing in a batch, with {\tt success\_count} being 
   3.700 -updated to report the number of successful updates.  
   3.701 -
   3.702 -Each element of {\tt req[]} contains a pointer (address) and value; 
   3.703 -the least significant 2-bits of the pointer are used to distinguish 
   3.704 -the type of update requested as follows:
   3.705 -\begin{description} 
   3.706 -
   3.707 -\item[\it MMU\_NORMAL\_PT\_UPDATE:] update a page directory entry or
   3.708 -page table entry to the associated value; Xen will check that the
   3.709 -update is safe, as described in Chapter~\ref{c:memory}.
   3.710 -
   3.711 -\item[\it MMU\_MACHPHYS\_UPDATE:] update an entry in the
   3.712 -  machine-to-physical table. The calling domain must own the machine
   3.713 -  page in question (or be privileged).
   3.714 -
   3.715 -\item[\it MMU\_EXTENDED\_COMMAND:] perform additional MMU operations.
   3.716 -The set of additional MMU operations is considerable, and includes
   3.717 -updating {\tt cr3} (or just re-installing it for a TLB flush),
   3.718 -flushing the cache, installing a new LDT, or pinning \& unpinning
   3.719 -page-table pages (to ensure their reference count doesn't drop to zero
   3.720 -which would require a revalidation of all entries).
   3.721 -
   3.722 -Further extended commands are used to deal with granting and 
   3.723 -acquiring page ownership; see Section~\ref{s:idc}. 
   3.724 -
   3.725 -
   3.726 -\end{description}
   3.727 -
   3.728 -More details on the precise format of all commands can be 
   3.729 -found in {\tt xen/include/public/xen.h}. 
   3.730 -
   3.731 -
   3.732 -\end{quote}
   3.733 -
   3.734 -Explicitly updating batches of page table entries is extremely
   3.735 -efficient, but can require a number of alterations to the guest
   3.736 -OS. Using the writable page table mode (Chapter~\ref{c:memory}) is
   3.737 -recommended for new OS ports.
   3.738 -
   3.739 -Regardless of which page table update mode is being used, however,
   3.740 -there are some occasions (notably handling a demand page fault) where
   3.741 -a guest OS will wish to modify exactly one PTE rather than a
   3.742 -batch. This is catered for by the following:
   3.743 -
   3.744 -\begin{quote} 
   3.745 -\hypercall{update\_va\_mapping(unsigned long page\_nr, unsigned long
   3.746 -val, \\ unsigned long flags)}
   3.747 -
   3.748 -Update the currently installed PTE for the page {\tt page\_nr} to 
   3.749 -{\tt val}. As with {\tt mmu\_update()}, Xen checks the modification 
   3.750 -is safe before applying it. The {\tt flags} determine which kind
   3.751 -of TLB flush, if any, should follow the update. 
   3.752 -
   3.753 -\end{quote} 
   3.754 -
   3.755 -Finally, sufficiently privileged domains may occasionally wish to manipulate 
   3.756 -the pages of others: 
   3.757 -\begin{quote}
   3.758 -
   3.759 -\hypercall{update\_va\_mapping\_otherdomain(unsigned long page\_nr,
   3.760 -unsigned long val, unsigned long flags, uint16\_t domid)}
   3.761 -
   3.762 -Identical to {\tt update\_va\_mapping()} save that the pages being
   3.763 -mapped must belong to the domain {\tt domid}. 
   3.764 -
   3.765 -\end{quote}
   3.766 -
   3.767 -This privileged operation is currently used by backend virtual device
   3.768 -drivers to safely map pages containing I/O data. 
   3.769 -
   3.770 -
   3.771 -
   3.772 -\section{Segmentation Support}
   3.773 -
   3.774 -Xen allows guest OSes to install a custom GDT if they require it; 
   3.775 -this is context switched transparently whenever a domain is 
   3.776 -[de]scheduled.  The following hypercall is effectively a 
   3.777 -`safe' version of {\tt lgdt}: 
   3.778 -
   3.779 -\begin{quote}
   3.780 -\hypercall{set\_gdt(unsigned long *frame\_list, int entries)} 
   3.781 -
   3.782 -Install a global descriptor table for a domain; {\tt frame\_list} is
   3.783 -an array of up to 16 machine page frames within which the GDT resides,
   3.784 -with {\tt entries} being the actual number of descriptor-entry
   3.785 -slots. All page frames must be mapped read-only within the guest's
   3.786 -address space, and the table must be large enough to contain Xen's
   3.787 -reserved entries (see {\tt xen/include/public/arch-x86\_32.h}).
   3.788 -
   3.789 -\end{quote}
   3.790 -
   3.791 -Many guest OSes will also wish to install LDTs; this is achieved by
   3.792 -using {\tt mmu\_update()} with an extended command, passing the
   3.793 -linear address of the LDT base along with the number of entries. No
   3.794 -special safety checks are required; Xen needs to perform this task
   3.795 -simply since {\tt lldt} requires CPL 0.
   3.796 -
   3.797 -
   3.798 -Xen also allows guest operating systems to update just an 
   3.799 -individual segment descriptor in the GDT or LDT:  
   3.800 -
   3.801 -\begin{quote}
   3.802 -\hypercall{update\_descriptor(unsigned long ma, unsigned long word1,
   3.803 -unsigned long word2)}
   3.804 -
   3.805 -Update the GDT/LDT entry at machine address {\tt ma}; the new
   3.806 -8-byte descriptor is stored in {\tt word1} and {\tt word2}.
   3.807 -Xen performs a number of checks to ensure the descriptor is 
   3.808 -valid. 
   3.809 -
   3.810 -\end{quote}
   3.811 -
   3.812 -Guest OSes can use the above in place of context switching entire 
   3.813 -LDTs (or the GDT) when the number of changing descriptors is small. 
   3.814 -
   3.815 -\section{Context Switching} 
   3.816 -
   3.817 -When a guest OS wishes to context switch between two processes, 
   3.818 -it can use the page table and segmentation hypercalls described
   3.819 -above to perform the the bulk of the privileged work. In addition, 
   3.820 -however, it will need to invoke Xen to switch the kernel (ring 1) 
   3.821 -stack pointer: 
   3.822 -
   3.823 -\begin{quote} 
   3.824 -\hypercall{stack\_switch(unsigned long ss, unsigned long esp)} 
   3.825 -
   3.826 -Request kernel stack switch from hypervisor; {\tt ss} is the new 
   3.827 -stack segment, which {\tt esp} is the new stack pointer. 
   3.828 -
   3.829 -\end{quote} 
   3.830 -
   3.831 -A final useful hypercall for context switching allows ``lazy'' 
   3.832 -save and restore of floating point state: 
   3.833 -
   3.834 -\begin{quote}
   3.835 -\hypercall{fpu\_taskswitch(void)} 
   3.836 -
   3.837 -This call instructs Xen to set the {\tt TS} bit in the {\tt cr0}
   3.838 -control register; this means that the next attempt to use floating
   3.839 -point will cause a trap which the guest OS can trap. Typically it will
   3.840 -then save/restore the FP state, and clear the {\tt TS} bit. 
   3.841 -\end{quote} 
   3.842 -
   3.843 -This is provided as an optimization only; guest OSes can also choose
   3.844 -to save and restore FP state on all context switches for simplicity. 
   3.845 -
   3.846 -
   3.847 -\section{Physical Memory Management}
   3.848 -
   3.849 -As mentioned previously, each domain has a maximum and current 
   3.850 -memory allocation. The maximum allocation, set at domain creation 
   3.851 -time, cannot be modified. However a domain can choose to reduce 
   3.852 -and subsequently grow its current allocation by using the
   3.853 -following call: 
   3.854 -
   3.855 -\begin{quote} 
   3.856 -\hypercall{dom\_mem\_op(unsigned int op, unsigned long *extent\_list,
   3.857 -  unsigned long nr\_extents, unsigned int extent\_order)}
   3.858 -
   3.859 -Increase or decrease current memory allocation (as determined by 
   3.860 -the value of {\tt op}). Each invocation provides a list of 
   3.861 -extents each of which is $2^s$ pages in size, 
   3.862 -where $s$ is the value of {\tt extent\_order}. 
   3.863 -
   3.864 -\end{quote} 
   3.865 -
   3.866 -In addition to simply reducing or increasing the current memory
   3.867 -allocation via a `balloon driver', this call is also useful for 
   3.868 -obtaining contiguous regions of machine memory when required (e.g. 
   3.869 -for certain PCI devices, or if using superpages).  
   3.870 -
   3.871 -
   3.872 -\section{Inter-Domain Communication}
   3.873 -\label{s:idc} 
   3.874 -
   3.875 -Xen provides a simple asynchronous notification mechanism via
   3.876 -\emph{event channels}. Each domain has a set of end-points (or
   3.877 -\emph{ports}) which may be bound to an event source (e.g. a physical
   3.878 -IRQ, a virtual IRQ, or an port in another domain). When a pair of
   3.879 -end-points in two different domains are bound together, then a `send'
   3.880 -operation on one will cause an event to be received by the destination
   3.881 -domain.
   3.882 -
   3.883 -The control and use of event channels involves the following hypercall: 
   3.884 -
   3.885 -\begin{quote}
   3.886 -\hypercall{event\_channel\_op(evtchn\_op\_t *op)} 
   3.887 -
   3.888 -Inter-domain event-channel management; {\tt op} is a discriminated 
   3.889 -union which allows the following 7 operations: 
   3.890 -
   3.891 -\begin{description} 
   3.892 -
   3.893 -\item[\it alloc\_unbound:] allocate a free (unbound) local
   3.894 -  port and prepare for connection from a specified domain. 
   3.895 -\item[\it bind\_virq:] bind a local port to a virtual 
   3.896 -IRQ; any particular VIRQ can be bound to at most one port per domain. 
   3.897 -\item[\it bind\_pirq:] bind a local port to a physical IRQ;
   3.898 -once more, a given pIRQ can be bound to at most one port per
   3.899 -domain. Furthermore the calling domain must be sufficiently
   3.900 -privileged.
   3.901 -\item[\it bind\_interdomain:] construct an interdomain event 
   3.902 -channel; in general, the target domain must have previously allocated 
   3.903 -an unbound port for this channel, although this can be bypassed by 
   3.904 -privileged domains during domain setup. 
   3.905 -\item[\it close:] close an interdomain event channel. 
   3.906 -\item[\it send:] send an event to the remote end of a 
   3.907 -interdomain event channel. 
   3.908 -\item[\it status:] determine the current status of a local port. 
   3.909 -\end{description} 
   3.910 -
   3.911 -For more details see
   3.912 -{\tt xen/include/public/event\_channel.h}. 
   3.913 -
   3.914 -\end{quote} 
   3.915 -
   3.916 -Event channels are the fundamental communication primitive between 
   3.917 -Xen domains and seamlessly support SMP. However they provide little
   3.918 -bandwidth for communication {\sl per se}, and hence are typically 
   3.919 -married with a piece of shared memory to produce effective and 
   3.920 -high-performance inter-domain communication. 
   3.921 -
   3.922 -Safe sharing of memory pages between guest OSes is carried out by
   3.923 -granting access on a per page basis to individual domains. This is
   3.924 -achieved by using the {\tt grant\_table\_op()} hypercall.
   3.925 -
   3.926 -\begin{quote}
   3.927 -\hypercall{grant\_table\_op(unsigned int cmd, void *uop, unsigned int count)}
   3.928 -
   3.929 -Grant or remove access to a particular page to a particular domain. 
   3.930 -
   3.931 -\end{quote} 
   3.932 -
   3.933 -This is not currently widely in use by guest operating systems, but 
   3.934 -we intend to integrate support more fully in the near future. 
   3.935 -
   3.936 -\section{PCI Configuration} 
   3.937 -
   3.938 -Domains with physical device access (i.e.\ driver domains) receive
   3.939 -limited access to certain PCI devices (bus address space and
   3.940 -interrupts). However many guest operating systems attempt to 
   3.941 -determine the PCI configuration by directly access the PCI BIOS, 
   3.942 -which cannot be allowed for safety. 
   3.943 -
   3.944 -Instead, Xen provides the following hypercall: 
   3.945 -
   3.946 -\begin{quote}
   3.947 -\hypercall{physdev\_op(void *physdev\_op)}
   3.948 -
   3.949 -Perform a PCI configuration option; depending on the value 
   3.950 -of {\tt physdev\_op} this can be a PCI config read, a PCI config 
   3.951 -write, or a small number of other queries. 
   3.952 -
   3.953 -\end{quote} 
   3.954 -
   3.955 -
   3.956 -For examples of using {\tt physdev\_op()}, see the 
   3.957 -Xen-specific PCI code in the linux sparse tree. 
   3.958 -
   3.959 -\section{Administrative Operations}
   3.960 -\label{s:dom0ops}
   3.961 -
   3.962 -A large number of control operations are available to a sufficiently
   3.963 -privileged domain (typically domain 0). These allow the creation and
   3.964 -management of new domains, for example. A complete list is given 
   3.965 -below: for more details on any or all of these, please see 
   3.966 -{\tt xen/include/public/dom0\_ops.h} 
   3.967 -
   3.968 -
   3.969 -\begin{quote}
   3.970 -\hypercall{dom0\_op(dom0\_op\_t *op)} 
   3.971 -
   3.972 -Administrative domain operations for domain management. The options are:
   3.973 -
   3.974 -\begin{description} 
   3.975 -\item [\it DOM0\_CREATEDOMAIN:] create a new domain
   3.976 -
   3.977 -\item [\it DOM0\_PAUSEDOMAIN:] remove a domain from the scheduler run 
   3.978 -queue. 
   3.979 -
   3.980 -\item [\it DOM0\_UNPAUSEDOMAIN:] mark a paused domain as schedulable
   3.981 -  once again. 
   3.982 -
   3.983 -\item [\it DOM0\_DESTROYDOMAIN:] deallocate all resources associated
   3.984 -with a domain
   3.985 -
   3.986 -\item [\it DOM0\_GETMEMLIST:] get list of pages used by the domain
   3.987 -
   3.988 -\item [\it DOM0\_SCHEDCTL:]
   3.989 -
   3.990 -\item [\it DOM0\_ADJUSTDOM:] adjust scheduling priorities for domain
   3.991 -
   3.992 -\item [\it DOM0\_BUILDDOMAIN:] do final guest OS setup for domain
   3.993 -
   3.994 -\item [\it DOM0\_GETDOMAINFO:] get statistics about the domain
   3.995 -
   3.996 -\item [\it DOM0\_GETPAGEFRAMEINFO:] 
   3.997 -
   3.998 -\item [\it DOM0\_GETPAGEFRAMEINFO2:]
   3.999 -
  3.1000 -\item [\it DOM0\_IOPL:] set I/O privilege level
  3.1001 -
  3.1002 -\item [\it DOM0\_MSR:] read or write model specific registers
  3.1003 -
  3.1004 -\item [\it DOM0\_DEBUG:] interactively invoke the debugger
  3.1005 -
  3.1006 -\item [\it DOM0\_SETTIME:] set system time
  3.1007 -
  3.1008 -\item [\it DOM0\_READCONSOLE:] read console content from hypervisor buffer ring
  3.1009 -
  3.1010 -\item [\it DOM0\_PINCPUDOMAIN:] pin domain to a particular CPU
  3.1011 -
  3.1012 -\item [\it DOM0\_GETTBUFS:] get information about the size and location of
  3.1013 -                      the trace buffers (only on trace-buffer enabled builds)
  3.1014 -
  3.1015 -\item [\it DOM0\_PHYSINFO:] get information about the host machine
  3.1016 -
  3.1017 -\item [\it DOM0\_PCIDEV\_ACCESS:] modify PCI device access permissions
  3.1018 -
  3.1019 -\item [\it DOM0\_SCHED\_ID:] get the ID of the current Xen scheduler
  3.1020 -
  3.1021 -\item [\it DOM0\_SHADOW\_CONTROL:] switch between shadow page-table modes
  3.1022 -
  3.1023 -\item [\it DOM0\_SETDOMAININITIALMEM:] set initial memory allocation of a domain
  3.1024 -
  3.1025 -\item [\it DOM0\_SETDOMAINMAXMEM:] set maximum memory allocation of a domain
  3.1026 -
  3.1027 -\item [\it DOM0\_SETDOMAINVMASSIST:] set domain VM assist options
  3.1028 -\end{description} 
  3.1029 -\end{quote} 
  3.1030 -
  3.1031 -Most of the above are best understood by looking at the code 
  3.1032 -implementing them (in {\tt xen/common/dom0\_ops.c}) and in 
  3.1033 -the user-space tools that use them (mostly in {\tt tools/libxc}). 
  3.1034 -
  3.1035 -\section{Debugging Hypercalls} 
  3.1036 -
  3.1037 -A few additional hypercalls are mainly useful for debugging: 
  3.1038 -
  3.1039 -\begin{quote} 
  3.1040 -\hypercall{console\_io(int cmd, int count, char *str)}
  3.1041 -
  3.1042 -Use Xen to interact with the console; operations are:
  3.1043 -
  3.1044 -{\it CONSOLEIO\_write}: Output count characters from buffer str.
  3.1045 -
  3.1046 -{\it CONSOLEIO\_read}: Input at most count characters into buffer str.
  3.1047 -\end{quote} 
  3.1048 -
  3.1049 -A pair of hypercalls allows access to the underlying debug registers: 
  3.1050 -\begin{quote}
  3.1051 -\hypercall{set\_debugreg(int reg, unsigned long value)}
  3.1052 -
  3.1053 -Set debug register {\tt reg} to {\tt value} 
  3.1054 -
  3.1055 -\hypercall{get\_debugreg(int reg)}
  3.1056 -
  3.1057 -Return the contents of the debug register {\tt reg}
  3.1058 -\end{quote}
  3.1059 -
  3.1060 -And finally: 
  3.1061 -\begin{quote}
  3.1062 -\hypercall{xen\_version(int cmd)}
  3.1063 -
  3.1064 -Request Xen version number.
  3.1065 -\end{quote} 
  3.1066 -
  3.1067 -This is useful to ensure that user-space tools are in sync 
  3.1068 -with the underlying hypervisor. 
  3.1069 -
  3.1070 -\section{Deprecated Hypercalls}
  3.1071 -
  3.1072 -Xen is under constant development and refinement; as such there 
  3.1073 -are plans to improve the way in which various pieces of functionality 
  3.1074 -are exposed to guest OSes. 
  3.1075 -
  3.1076 -\begin{quote} 
  3.1077 -\hypercall{vm\_assist(unsigned int cmd, unsigned int type)}
  3.1078 -
  3.1079 -Toggle various memory management modes (in particular wrritable page
  3.1080 -tables and superpage support). 
  3.1081 -
  3.1082 -\end{quote} 
  3.1083 -
  3.1084 -This is likely to be replaced with mode values in the shared 
  3.1085 -information page since this is more resilient for resumption 
  3.1086 -after migration or checkpoint. 
  3.1087 -
  3.1088 -
  3.1089 -
  3.1090 -
  3.1091 -
  3.1092 -
  3.1093 +%% chapter hypercalls moved to hypercalls.tex
  3.1094 +\include{src/interface/hypercalls}
  3.1095  
  3.1096  
  3.1097  %% 
  3.1098 @@ -1173,279 +112,9 @@ after migration or checkpoint.
  3.1099  %% new scheduler... not clear how many of them there are...
  3.1100  %%
  3.1101  
  3.1102 -\begin{comment}
  3.1103 -
  3.1104 -\chapter{Scheduling API}  
  3.1105 -
  3.1106 -The scheduling API is used by both the schedulers described above and should
  3.1107 -also be used by any new schedulers.  It provides a generic interface and also
  3.1108 -implements much of the ``boilerplate'' code.
  3.1109 -
  3.1110 -Schedulers conforming to this API are described by the following
  3.1111 -structure:
  3.1112 -
  3.1113 -\begin{verbatim}
  3.1114 -struct scheduler
  3.1115 -{
  3.1116 -    char *name;             /* full name for this scheduler      */
  3.1117 -    char *opt_name;         /* option name for this scheduler    */
  3.1118 -    unsigned int sched_id;  /* ID for this scheduler             */
  3.1119 -
  3.1120 -    int          (*init_scheduler) ();
  3.1121 -    int          (*alloc_task)     (struct task_struct *);
  3.1122 -    void         (*add_task)       (struct task_struct *);
  3.1123 -    void         (*free_task)      (struct task_struct *);
  3.1124 -    void         (*rem_task)       (struct task_struct *);
  3.1125 -    void         (*wake_up)        (struct task_struct *);
  3.1126 -    void         (*do_block)       (struct task_struct *);
  3.1127 -    task_slice_t (*do_schedule)    (s_time_t);
  3.1128 -    int          (*control)        (struct sched_ctl_cmd *);
  3.1129 -    int          (*adjdom)         (struct task_struct *,
  3.1130 -                                    struct sched_adjdom_cmd *);
  3.1131 -    s32          (*reschedule)     (struct task_struct *);
  3.1132 -    void         (*dump_settings)  (void);
  3.1133 -    void         (*dump_cpu_state) (int);
  3.1134 -    void         (*dump_runq_el)   (struct task_struct *);
  3.1135 -};
  3.1136 -\end{verbatim}
  3.1137 -
  3.1138 -The only method that {\em must} be implemented is
  3.1139 -{\tt do\_schedule()}.  However, if there is not some implementation for the
  3.1140 -{\tt wake\_up()} method then waking tasks will not get put on the runqueue!
  3.1141 -
  3.1142 -The fields of the above structure are described in more detail below.
  3.1143 -
  3.1144 -\subsubsection{name}
  3.1145 -
  3.1146 -The name field should point to a descriptive ASCII string.
  3.1147 -
  3.1148 -\subsubsection{opt\_name}
  3.1149 -
  3.1150 -This field is the value of the {\tt sched=} boot-time option that will select
  3.1151 -this scheduler.
  3.1152 -
  3.1153 -\subsubsection{sched\_id}
  3.1154 -
  3.1155 -This is an integer that uniquely identifies this scheduler.  There should be a
  3.1156 -macro corrsponding to this scheduler ID in {\tt <xen/sched-if.h>}.
  3.1157 -
  3.1158 -\subsubsection{init\_scheduler}
  3.1159 -
  3.1160 -\paragraph*{Purpose}
  3.1161 -
  3.1162 -This is a function for performing any scheduler-specific initialisation.  For
  3.1163 -instance, it might allocate memory for per-CPU scheduler data and initialise it
  3.1164 -appropriately.
  3.1165 -
  3.1166 -\paragraph*{Call environment}
  3.1167 -
  3.1168 -This function is called after the initialisation performed by the generic
  3.1169 -layer.  The function is called exactly once, for the scheduler that has been
  3.1170 -selected.
  3.1171 -
  3.1172 -\paragraph*{Return values}
  3.1173 -
  3.1174 -This should return negative on failure --- this will cause an
  3.1175 -immediate panic and the system will fail to boot.
  3.1176 -
  3.1177 -\subsubsection{alloc\_task}
  3.1178 -
  3.1179 -\paragraph*{Purpose}
  3.1180 -Called when a {\tt task\_struct} is allocated by the generic scheduler
  3.1181 -layer.  A particular scheduler implementation may use this method to
  3.1182 -allocate per-task data for this task.  It may use the {\tt
  3.1183 -sched\_priv} pointer in the {\tt task\_struct} to point to this data.
  3.1184 -
  3.1185 -\paragraph*{Call environment}
  3.1186 -The generic layer guarantees that the {\tt sched\_priv} field will
  3.1187 -remain intact from the time this method is called until the task is
  3.1188 -deallocated (so long as the scheduler implementation does not change
  3.1189 -it explicitly!).
  3.1190 -
  3.1191 -\paragraph*{Return values}
  3.1192 -Negative on failure.
  3.1193 -
  3.1194 -\subsubsection{add\_task}
  3.1195 -
  3.1196 -\paragraph*{Purpose}
  3.1197 -
  3.1198 -Called when a task is initially added by the generic layer.
  3.1199 -
  3.1200 -\paragraph*{Call environment}
  3.1201 -
  3.1202 -The fields in the {\tt task\_struct} are now filled out and available for use.
  3.1203 -Schedulers should implement appropriate initialisation of any per-task private
  3.1204 -information in this method.
  3.1205 -
  3.1206 -\subsubsection{free\_task}
  3.1207 -
  3.1208 -\paragraph*{Purpose}
  3.1209 -
  3.1210 -Schedulers should free the space used by any associated private data
  3.1211 -structures.
  3.1212 -
  3.1213 -\paragraph*{Call environment}
  3.1214 -
  3.1215 -This is called when a {\tt task\_struct} is about to be deallocated.
  3.1216 -The generic layer will have done generic task removal operations and
  3.1217 -(if implemented) called the scheduler's {\tt rem\_task} method before
  3.1218 -this method is called.
  3.1219 -
  3.1220 -\subsubsection{rem\_task}
  3.1221 -
  3.1222 -\paragraph*{Purpose}
  3.1223 -
  3.1224 -This is called when a task is being removed from scheduling (but is
  3.1225 -not yet being freed).
  3.1226 -
  3.1227 -\subsubsection{wake\_up}
  3.1228 -
  3.1229 -\paragraph*{Purpose}
  3.1230 -
  3.1231 -Called when a task is woken up, this method should put the task on the runqueue
  3.1232 -(or do the scheduler-specific equivalent action).
  3.1233 -
  3.1234 -\paragraph*{Call environment}
  3.1235 -
  3.1236 -The task is already set to state RUNNING.
  3.1237 -
  3.1238 -\subsubsection{do\_block}
  3.1239 -
  3.1240 -\paragraph*{Purpose}
  3.1241 -
  3.1242 -This function is called when a task is blocked.  This function should
  3.1243 -not remove the task from the runqueue.
  3.1244 -
  3.1245 -\paragraph*{Call environment}
  3.1246 -
  3.1247 -The EVENTS\_MASTER\_ENABLE\_BIT is already set and the task state changed to
  3.1248 -TASK\_INTERRUPTIBLE on entry to this method.  A call to the {\tt
  3.1249 -  do\_schedule} method will be made after this method returns, in
  3.1250 -order to select the next task to run.
  3.1251 -
  3.1252 -\subsubsection{do\_schedule}
  3.1253 -
  3.1254 -This method must be implemented.
  3.1255 -
  3.1256 -\paragraph*{Purpose}
  3.1257 -
  3.1258 -The method is called each time a new task must be chosen for scheduling on the
  3.1259 -current CPU.  The current time as passed as the single argument (the current
  3.1260 -task can be found using the {\tt current} macro).
  3.1261 -
  3.1262 -This method should select the next task to run on this CPU and set it's minimum
  3.1263 -time to run as well as returning the data described below.
  3.1264 -
  3.1265 -This method should also take the appropriate action if the previous
  3.1266 -task has blocked, e.g. removing it from the runqueue.
  3.1267 -
  3.1268 -\paragraph*{Call environment}
  3.1269 -
  3.1270 -The other fields in the {\tt task\_struct} are updated by the generic layer,
  3.1271 -which also performs all Xen-specific tasks and performs the actual task switch
  3.1272 -(unless the previous task has been chosen again).
  3.1273 -
  3.1274 -This method is called with the {\tt schedule\_lock} held for the current CPU
  3.1275 -and local interrupts disabled.
  3.1276 -
  3.1277 -\paragraph*{Return values}
  3.1278 -
  3.1279 -Must return a {\tt struct task\_slice} describing what task to run and how long
  3.1280 -for (at maximum).
  3.1281 -
  3.1282 -\subsubsection{control}
  3.1283 -
  3.1284 -\paragraph*{Purpose}
  3.1285 -
  3.1286 -This method is called for global scheduler control operations.  It takes a
  3.1287 -pointer to a {\tt struct sched\_ctl\_cmd}, which it should either
  3.1288 -source data from or populate with data, depending on the value of the
  3.1289 -{\tt direction} field.
  3.1290 -
  3.1291 -\paragraph*{Call environment}
  3.1292 -
  3.1293 -The generic layer guarantees that when this method is called, the
  3.1294 -caller selected the correct scheduler ID, hence the scheduler's
  3.1295 -implementation does not need to sanity-check these parts of the call.
  3.1296 -
  3.1297 -\paragraph*{Return values}
  3.1298 -
  3.1299 -This function should return the value to be passed back to user space, hence it
  3.1300 -should either be 0 or an appropriate errno value.
  3.1301 -
  3.1302 -\subsubsection{sched\_adjdom}
  3.1303 -
  3.1304 -\paragraph*{Purpose}
  3.1305 -
  3.1306 -This method is called to adjust the scheduling parameters of a particular
  3.1307 -domain, or to query their current values.  The function should check
  3.1308 -the {\tt direction} field of the {\tt sched\_adjdom\_cmd} it receives in
  3.1309 -order to determine which of these operations is being performed.
  3.1310 -
  3.1311 -\paragraph*{Call environment}
  3.1312 -
  3.1313 -The generic layer guarantees that the caller has specified the correct
  3.1314 -control interface version and scheduler ID and that the supplied {\tt
  3.1315 -task\_struct} will not be deallocated during the call (hence it is not
  3.1316 -necessary to {\tt get\_task\_struct}).
  3.1317 -
  3.1318 -\paragraph*{Return values}
  3.1319 -
  3.1320 -This function should return the value to be passed back to user space, hence it
  3.1321 -should either be 0 or an appropriate errno value.
  3.1322 -
  3.1323 -\subsubsection{reschedule}
  3.1324 -
  3.1325 -\paragraph*{Purpose}
  3.1326 -
  3.1327 -This method is called to determine if a reschedule is required as a result of a
  3.1328 -particular task.
  3.1329 -
  3.1330 -\paragraph*{Call environment}
  3.1331 -The generic layer will cause a reschedule if the current domain is the idle
  3.1332 -task or it has exceeded its minimum time slice before a reschedule.  The
  3.1333 -generic layer guarantees that the task passed is not currently running but is
  3.1334 -on the runqueue.
  3.1335 -
  3.1336 -\paragraph*{Return values}
  3.1337 -
  3.1338 -Should return a mask of CPUs to cause a reschedule on.
  3.1339 -
  3.1340 -\subsubsection{dump\_settings}
  3.1341 -
  3.1342 -\paragraph*{Purpose}
  3.1343 -
  3.1344 -If implemented, this should dump any private global settings for this
  3.1345 -scheduler to the console.
  3.1346 -
  3.1347 -\paragraph*{Call environment}
  3.1348 -
  3.1349 -This function is called with interrupts enabled.
  3.1350 -
  3.1351 -\subsubsection{dump\_cpu\_state}
  3.1352 -
  3.1353 -\paragraph*{Purpose}
  3.1354 -
  3.1355 -This method should dump any private settings for the specified CPU.
  3.1356 -
  3.1357 -\paragraph*{Call environment}
  3.1358 -
  3.1359 -This function is called with interrupts disabled and the {\tt schedule\_lock}
  3.1360 -for the specified CPU held.
  3.1361 -
  3.1362 -\subsubsection{dump\_runq\_el}
  3.1363 -
  3.1364 -\paragraph*{Purpose}
  3.1365 -
  3.1366 -This method should dump any private settings for the specified task.
  3.1367 -
  3.1368 -\paragraph*{Call environment}
  3.1369 -
  3.1370 -This function is called with interrupts disabled and the {\tt schedule\_lock}
  3.1371 -for the task's CPU held.
  3.1372 -
  3.1373 -\end{comment} 
  3.1374 -
  3.1375 +%% \include{src/interface/scheduling}
  3.1376 +%% scheduling information moved to scheduling.tex
  3.1377 +%% still commented out
  3.1378  
  3.1379  
  3.1380  
  3.1381 @@ -1457,74 +126,9 @@ for the task's CPU held.
  3.1382  %% (and/or kip's stuff?) and write about that instead? 
  3.1383  %%
  3.1384  
  3.1385 -\begin{comment} 
  3.1386 -
  3.1387 -\chapter{Debugging}
  3.1388 -
  3.1389 -Xen provides tools for debugging both Xen and guest OSes.  Currently, the
  3.1390 -Pervasive Debugger provides a GDB stub, which provides facilities for symbolic
  3.1391 -debugging of Xen itself and of OS kernels running on top of Xen.  The Trace
  3.1392 -Buffer provides a lightweight means to log data about Xen's internal state and
  3.1393 -behaviour at runtime, for later analysis.
  3.1394 -
  3.1395 -\section{Pervasive Debugger}
  3.1396 -
  3.1397 -Information on using the pervasive debugger is available in pdb.txt.
  3.1398 -
  3.1399 -
  3.1400 -\section{Trace Buffer}
  3.1401 -
  3.1402 -The trace buffer provides a means to observe Xen's operation from domain 0.
  3.1403 -Trace events, inserted at key points in Xen's code, record data that can be
  3.1404 -read by the {\tt xentrace} tool.  Recording these events has a low overhead
  3.1405 -and hence the trace buffer may be useful for debugging timing-sensitive
  3.1406 -behaviours.
  3.1407 -
  3.1408 -\subsection{Internal API}
  3.1409 -
  3.1410 -To use the trace buffer functionality from within Xen, you must {\tt \#include
  3.1411 -<xen/trace.h>}, which contains definitions related to the trace buffer.  Trace
  3.1412 -events are inserted into the buffer using the {\tt TRACE\_xD} ({\tt x} = 0, 1,
  3.1413 -2, 3, 4 or 5) macros.  These all take an event number, plus {\tt x} additional
  3.1414 -(32-bit) data as their arguments.  For trace buffer-enabled builds of Xen these
  3.1415 -will insert the event ID and data into the trace buffer, along with the current
  3.1416 -value of the CPU cycle-counter.  For builds without the trace buffer enabled,
  3.1417 -the macros expand to no-ops and thus can be left in place without incurring
  3.1418 -overheads.
  3.1419 -
  3.1420 -\subsection{Trace-enabled builds}
  3.1421 -
  3.1422 -By default, the trace buffer is enabled only in debug builds (i.e. {\tt NDEBUG}
  3.1423 -is not defined).  It can be enabled separately by defining {\tt TRACE\_BUFFER},
  3.1424 -either in {\tt <xen/config.h>} or on the gcc command line.
  3.1425 -
  3.1426 -The size (in pages) of the per-CPU trace buffers can be specified using the
  3.1427 -{\tt tbuf\_size=n } boot parameter to Xen.  If the size is set to 0, the trace
  3.1428 -buffers will be disabled.
  3.1429 -
  3.1430 -\subsection{Dumping trace data}
  3.1431 -
  3.1432 -When running a trace buffer build of Xen, trace data are written continuously
  3.1433 -into the buffer data areas, with newer data overwriting older data.  This data
  3.1434 -can be captured using the {\tt xentrace} program in domain 0.
  3.1435 -
  3.1436 -The {\tt xentrace} tool uses {\tt /dev/mem} in domain 0 to map the trace
  3.1437 -buffers into its address space.  It then periodically polls all the buffers for
  3.1438 -new data, dumping out any new records from each buffer in turn.  As a result,
  3.1439 -for machines with multiple (logical) CPUs, the trace buffer output will not be
  3.1440 -in overall chronological order.
  3.1441 -
  3.1442 -The output from {\tt xentrace} can be post-processed using {\tt
  3.1443 -xentrace\_cpusplit} (used to split trace data out into per-cpu log files) and
  3.1444 -{\tt xentrace\_format} (used to pretty-print trace data).  For the predefined
  3.1445 -trace points, there is an example format file in {\tt tools/xentrace/formats }.
  3.1446 -
  3.1447 -For more information, see the manual pages for {\tt xentrace}, {\tt
  3.1448 -xentrace\_format} and {\tt xentrace\_cpusplit}.
  3.1449 -
  3.1450 -\end{comment} 
  3.1451 -
  3.1452 -
  3.1453 +%% \include{src/interface/debugging}
  3.1454 +%% debugging information moved to debugging.tex
  3.1455 +%% still commented out
  3.1456  
  3.1457  
  3.1458  \end{document}

     4.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     4.2 +++ b/docs/src/interface/architecture.tex	Tue Sep 20 09:43:46 2005 +0000
     4.3 @@ -0,0 +1,140 @@
     4.4 +\chapter{Virtual Architecture}
     4.5 +
     4.6 +On a Xen-based system, the hypervisor itself runs in {\it ring 0}.  It
     4.7 +has full access to the physical memory available in the system and is
     4.8 +responsible for allocating portions of it to the domains.  Guest
     4.9 +operating systems run in and use {\it rings 1}, {\it 2} and {\it 3} as
    4.10 +they see fit. Segmentation is used to prevent the guest OS from
    4.11 +accessing the portion of the address space that is reserved for Xen.
    4.12 +We expect most guest operating systems will use ring 1 for their own
    4.13 +operation and place applications in ring 3.
    4.14 +
    4.15 +In this chapter we consider the basic virtual architecture provided by
    4.16 +Xen: the basic CPU state, exception and interrupt handling, and time.
    4.17 +Other aspects such as memory and device access are discussed in later
    4.18 +chapters.
    4.19 +
    4.20 +
    4.21 +\section{CPU state}
    4.22 +
    4.23 +All privileged state must be handled by Xen.  The guest OS has no
    4.24 +direct access to CR3 and is not permitted to update privileged bits in
    4.25 +EFLAGS. Guest OSes use \emph{hypercalls} to invoke operations in Xen;
    4.26 +these are analogous to system calls but occur from ring 1 to ring 0.
    4.27 +
    4.28 +A list of all hypercalls is given in Appendix~\ref{a:hypercalls}.
    4.29 +
    4.30 +
    4.31 +\section{Exceptions}
    4.32 +
    4.33 +A virtual IDT is provided --- a domain can submit a table of trap
    4.34 +handlers to Xen via the {\tt set\_trap\_table()} hypercall.  Most trap
    4.35 +handlers are identical to native x86 handlers, although the page-fault
    4.36 +handler is somewhat different.
    4.37 +
    4.38 +
    4.39 +\section{Interrupts and events}
    4.40 +
    4.41 +Interrupts are virtualized by mapping them to \emph{events}, which are
    4.42 +delivered asynchronously to the target domain using a callback
    4.43 +supplied via the {\tt set\_callbacks()} hypercall.  A guest OS can map
    4.44 +these events onto its standard interrupt dispatch mechanisms.  Xen is
    4.45 +responsible for determining the target domain that will handle each
    4.46 +physical interrupt source. For more details on the binding of event
    4.47 +sources to events, see Chapter~\ref{c:devices}.
    4.48 +
    4.49 +
    4.50 +\section{Time}
    4.51 +
    4.52 +Guest operating systems need to be aware of the passage of both real
    4.53 +(or wallclock) time and their own `virtual time' (the time for which
    4.54 +they have been executing). Furthermore, Xen has a notion of time which
    4.55 +is used for scheduling. The following notions of time are provided:
    4.56 +
    4.57 +\begin{description}
    4.58 +\item[Cycle counter time.]
    4.59 +
    4.60 +  This provides a fine-grained time reference.  The cycle counter time
    4.61 +  is used to accurately extrapolate the other time references.  On SMP
    4.62 +  machines it is currently assumed that the cycle counter time is
    4.63 +  synchronized between CPUs.  The current x86-based implementation
    4.64 +  achieves this within inter-CPU communication latencies.
    4.65 +
    4.66 +\item[System time.]
    4.67 +
    4.68 +  This is a 64-bit counter which holds the number of nanoseconds that
    4.69 +  have elapsed since system boot.
    4.70 +
    4.71 +\item[Wall clock time.]
    4.72 +
    4.73 +  This is the time of day in a Unix-style {\tt struct timeval}
    4.74 +  (seconds and microseconds since 1 January 1970, adjusted by leap
    4.75 +  seconds).  An NTP client hosted by {\it domain 0} can keep this
    4.76 +  value accurate.
    4.77 +
    4.78 +\item[Domain virtual time.]
    4.79 +
    4.80 +  This progresses at the same pace as system time, but only while a
    4.81 +  domain is executing --- it stops while a domain is de-scheduled.
    4.82 +  Therefore the share of the CPU that a domain receives is indicated
    4.83 +  by the rate at which its virtual time increases.
    4.84 +
    4.85 +\end{description}
    4.86 +
    4.87 +
    4.88 +Xen exports timestamps for system time and wall-clock time to guest
    4.89 +operating systems through a shared page of memory.  Xen also provides
    4.90 +the cycle counter time at the instant the timestamps were calculated,
    4.91 +and the CPU frequency in Hertz.  This allows the guest to extrapolate
    4.92 +system and wall-clock times accurately based on the current cycle
    4.93 +counter time.
    4.94 +
    4.95 +Since all time stamps need to be updated and read \emph{atomically}
    4.96 +two version numbers are also stored in the shared info page. The first
    4.97 +is incremented prior to an update, while the second is only
    4.98 +incremented afterwards. Thus a guest can be sure that it read a
    4.99 +consistent state by checking the two version numbers are equal.
   4.100 +
   4.101 +Xen includes a periodic ticker which sends a timer event to the
   4.102 +currently executing domain every 10ms.  The Xen scheduler also sends a
   4.103 +timer event whenever a domain is scheduled; this allows the guest OS
   4.104 +to adjust for the time that has passed while it has been inactive.  In
   4.105 +addition, Xen allows each domain to request that they receive a timer
   4.106 +event sent at a specified system time by using the {\tt
   4.107 +  set\_timer\_op()} hypercall.  Guest OSes may use this timer to
   4.108 +implement timeout values when they block.
   4.109 +
   4.110 +
   4.111 +
   4.112 +%% % akw: demoting this to a section -- not sure if there is any point
   4.113 +%% % though, maybe just remove it.
   4.114 +
   4.115 +\section{Xen CPU Scheduling}
   4.116 +
   4.117 +Xen offers a uniform API for CPU schedulers.  It is possible to choose
   4.118 +from a number of schedulers at boot and it should be easy to add more.
   4.119 +The BVT, Atropos and Round Robin schedulers are part of the normal Xen
   4.120 +distribution.  BVT provides proportional fair shares of the CPU to the
   4.121 +running domains.  Atropos can be used to reserve absolute shares of
   4.122 +the CPU for each domain.  Round-robin is provided as an example of
   4.123 +Xen's internal scheduler API.
   4.124 +
   4.125 +\paragraph*{Note: SMP host support}
   4.126 +Xen has always supported SMP host systems.  Domains are statically
   4.127 +assigned to CPUs, either at creation time or when manually pinning to
   4.128 +a particular CPU.  The current schedulers then run locally on each CPU
   4.129 +to decide which of the assigned domains should be run there. The
   4.130 +user-level control software can be used to perform coarse-grain
   4.131 +load-balancing between CPUs.
   4.132 +
   4.133 +
   4.134 +%% More information on the characteristics and use of these schedulers
   4.135 +%% is available in {\tt Sched-HOWTO.txt}.
   4.136 +
   4.137 +
   4.138 +\section{Privileged operations}
   4.139 +
   4.140 +Xen exports an extended interface to privileged domains (viz.\ {\it
   4.141 +  Domain 0}). This allows such domains to build and boot other domains
   4.142 +on the server, and provides control interfaces for managing
   4.143 +scheduling, memory, networking, and block devices.

     5.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     5.2 +++ b/docs/src/interface/debugging.tex	Tue Sep 20 09:43:46 2005 +0000
     5.3 @@ -0,0 +1,62 @@
     5.4 +\chapter{Debugging}
     5.5 +
     5.6 +Xen provides tools for debugging both Xen and guest OSes.  Currently, the
     5.7 +Pervasive Debugger provides a GDB stub, which provides facilities for symbolic
     5.8 +debugging of Xen itself and of OS kernels running on top of Xen.  The Trace
     5.9 +Buffer provides a lightweight means to log data about Xen's internal state and
    5.10 +behaviour at runtime, for later analysis.
    5.11 +
    5.12 +\section{Pervasive Debugger}
    5.13 +
    5.14 +Information on using the pervasive debugger is available in pdb.txt.
    5.15 +
    5.16 +
    5.17 +\section{Trace Buffer}
    5.18 +
    5.19 +The trace buffer provides a means to observe Xen's operation from domain 0.
    5.20 +Trace events, inserted at key points in Xen's code, record data that can be
    5.21 +read by the {\tt xentrace} tool.  Recording these events has a low overhead
    5.22 +and hence the trace buffer may be useful for debugging timing-sensitive
    5.23 +behaviours.
    5.24 +
    5.25 +\subsection{Internal API}
    5.26 +
    5.27 +To use the trace buffer functionality from within Xen, you must {\tt \#include
    5.28 +<xen/trace.h>}, which contains definitions related to the trace buffer.  Trace
    5.29 +events are inserted into the buffer using the {\tt TRACE\_xD} ({\tt x} = 0, 1,
    5.30 +2, 3, 4 or 5) macros.  These all take an event number, plus {\tt x} additional
    5.31 +(32-bit) data as their arguments.  For trace buffer-enabled builds of Xen these
    5.32 +will insert the event ID and data into the trace buffer, along with the current
    5.33 +value of the CPU cycle-counter.  For builds without the trace buffer enabled,
    5.34 +the macros expand to no-ops and thus can be left in place without incurring
    5.35 +overheads.
    5.36 +
    5.37 +\subsection{Trace-enabled builds}
    5.38 +
    5.39 +By default, the trace buffer is enabled only in debug builds (i.e. {\tt NDEBUG}
    5.40 +is not defined).  It can be enabled separately by defining {\tt TRACE\_BUFFER},
    5.41 +either in {\tt <xen/config.h>} or on the gcc command line.
    5.42 +
    5.43 +The size (in pages) of the per-CPU trace buffers can be specified using the
    5.44 +{\tt tbuf\_size=n } boot parameter to Xen.  If the size is set to 0, the trace
    5.45 +buffers will be disabled.
    5.46 +
    5.47 +\subsection{Dumping trace data}
    5.48 +
    5.49 +When running a trace buffer build of Xen, trace data are written continuously
    5.50 +into the buffer data areas, with newer data overwriting older data.  This data
    5.51 +can be captured using the {\tt xentrace} program in domain 0.
    5.52 +
    5.53 +The {\tt xentrace} tool uses {\tt /dev/mem} in domain 0 to map the trace
    5.54 +buffers into its address space.  It then periodically polls all the buffers for
    5.55 +new data, dumping out any new records from each buffer in turn.  As a result,
    5.56 +for machines with multiple (logical) CPUs, the trace buffer output will not be
    5.57 +in overall chronological order.
    5.58 +
    5.59 +The output from {\tt xentrace} can be post-processed using {\tt
    5.60 +xentrace\_cpusplit} (used to split trace data out into per-cpu log files) and
    5.61 +{\tt xentrace\_format} (used to pretty-print trace data).  For the predefined
    5.62 +trace points, there is an example format file in {\tt tools/xentrace/formats }.
    5.63 +
    5.64 +For more information, see the manual pages for {\tt xentrace}, {\tt
    5.65 +xentrace\_format} and {\tt xentrace\_cpusplit}.

     6.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     6.2 +++ b/docs/src/interface/devices.tex	Tue Sep 20 09:43:46 2005 +0000
     6.3 @@ -0,0 +1,178 @@
     6.4 +\chapter{Devices}
     6.5 +\label{c:devices}
     6.6 +
     6.7 +Devices such as network and disk are exported to guests using a split
     6.8 +device driver.  The device driver domain, which accesses the physical
     6.9 +device directly also runs a \emph{backend} driver, serving requests to
    6.10 +that device from guests.  Each guest will use a simple \emph{frontend}
    6.11 +driver, to access the backend.  Communication between these domains is
    6.12 +composed of two parts: First, data is placed onto a shared memory page
    6.13 +between the domains.  Second, an event channel between the two domains
    6.14 +is used to pass notification that data is outstanding.  This
    6.15 +separation of notification from data transfer allows message batching,
    6.16 +and results in very efficient device access.
    6.17 +
    6.18 +Event channels are used extensively in device virtualization; each
    6.19 +domain has a number of end-points or \emph{ports} each of which may be
    6.20 +bound to one of the following \emph{event sources}:
    6.21 +\begin{itemize}
    6.22 +  \item a physical interrupt from a real device, 
    6.23 +  \item a virtual interrupt (callback) from Xen, or 
    6.24 +  \item a signal from another domain 
    6.25 +\end{itemize}
    6.26 +
    6.27 +Events are lightweight and do not carry much information beyond the
    6.28 +source of the notification. Hence when performing bulk data transfer,
    6.29 +events are typically used as synchronization primitives over a shared
    6.30 +memory transport. Event channels are managed via the {\tt
    6.31 +  event\_channel\_op()} hypercall; for more details see
    6.32 +Section~\ref{s:idc}.
    6.33 +
    6.34 +This chapter focuses on some individual device interfaces available to
    6.35 +Xen guests.
    6.36 +
    6.37 +
    6.38 +\section{Network I/O}
    6.39 +
    6.40 +Virtual network device services are provided by shared memory
    6.41 +communication with a backend domain.  From the point of view of other
    6.42 +domains, the backend may be viewed as a virtual ethernet switch
    6.43 +element with each domain having one or more virtual network interfaces
    6.44 +connected to it.
    6.45 +
    6.46 +\subsection{Backend Packet Handling}
    6.47 +
    6.48 +The backend driver is responsible for a variety of actions relating to
    6.49 +the transmission and reception of packets from the physical device.
    6.50 +With regard to transmission, the backend performs these key actions:
    6.51 +
    6.52 +\begin{itemize}
    6.53 +\item {\bf Validation:} To ensure that domains do not attempt to
    6.54 +  generate invalid (e.g. spoofed) traffic, the backend driver may
    6.55 +  validate headers ensuring that source MAC and IP addresses match the
    6.56 +  interface that they have been sent from.
    6.57 +
    6.58 +  Validation functions can be configured using standard firewall rules
    6.59 +  ({\small{\tt iptables}} in the case of Linux).
    6.60 +  
    6.61 +\item {\bf Scheduling:} Since a number of domains can share a single
    6.62 +  physical network interface, the backend must mediate access when
    6.63 +  several domains each have packets queued for transmission.  This
    6.64 +  general scheduling function subsumes basic shaping or rate-limiting
    6.65 +  schemes.
    6.66 +  
    6.67 +\item {\bf Logging and Accounting:} The backend domain can be
    6.68 +  configured with classifier rules that control how packets are
    6.69 +  accounted or logged.  For example, log messages might be generated
    6.70 +  whenever a domain attempts to send a TCP packet containing a SYN.
    6.71 +\end{itemize}
    6.72 +
    6.73 +On receipt of incoming packets, the backend acts as a simple
    6.74 +demultiplexer: Packets are passed to the appropriate virtual interface
    6.75 +after any necessary logging and accounting have been carried out.
    6.76 +
    6.77 +\subsection{Data Transfer}
    6.78 +
    6.79 +Each virtual interface uses two ``descriptor rings'', one for
    6.80 +transmit, the other for receive.  Each descriptor identifies a block
    6.81 +of contiguous physical memory allocated to the domain.
    6.82 +
    6.83 +The transmit ring carries packets to transmit from the guest to the
    6.84 +backend domain.  The return path of the transmit ring carries messages
    6.85 +indicating that the contents have been physically transmitted and the
    6.86 +backend no longer requires the associated pages of memory.
    6.87 +
    6.88 +To receive packets, the guest places descriptors of unused pages on
    6.89 +the receive ring.  The backend will return received packets by
    6.90 +exchanging these pages in the domain's memory with new pages
    6.91 +containing the received data, and passing back descriptors regarding
    6.92 +the new packets on the ring.  This zero-copy approach allows the
    6.93 +backend to maintain a pool of free pages to receive packets into, and
    6.94 +then deliver them to appropriate domains after examining their
    6.95 +headers.
    6.96 +
    6.97 +% Real physical addresses are used throughout, with the domain
    6.98 +% performing translation from pseudo-physical addresses if that is
    6.99 +% necessary.
   6.100 +
   6.101 +If a domain does not keep its receive ring stocked with empty buffers
   6.102 +then packets destined to it may be dropped.  This provides some
   6.103 +defence against receive livelock problems because an overload domain
   6.104 +will cease to receive further data.  Similarly, on the transmit path,
   6.105 +it provides the application with feedback on the rate at which packets
   6.106 +are able to leave the system.
   6.107 +
   6.108 +Flow control on rings is achieved by including a pair of producer
   6.109 +indexes on the shared ring page.  Each side will maintain a private
   6.110 +consumer index indicating the next outstanding message.  In this
   6.111 +manner, the domains cooperate to divide the ring into two message
   6.112 +lists, one in each direction.  Notification is decoupled from the
   6.113 +immediate placement of new messages on the ring; the event channel
   6.114 +will be used to generate notification when {\em either} a certain
   6.115 +number of outstanding messages are queued, {\em or} a specified number
   6.116 +of nanoseconds have elapsed since the oldest message was placed on the
   6.117 +ring.
   6.118 +
   6.119 +%% Not sure if my version is any better -- here is what was here
   6.120 +%% before: Synchronization between the backend domain and the guest is
   6.121 +%% achieved using counters held in shared memory that is accessible to
   6.122 +%% both.  Each ring has associated producer and consumer indices
   6.123 +%% indicating the area in the ring that holds descriptors that contain
   6.124 +%% data.  After receiving {\it n} packets or {\t nanoseconds} after
   6.125 +%% receiving the first packet, the hypervisor sends an event to the
   6.126 +%% domain.
   6.127 +
   6.128 +
   6.129 +\section{Block I/O}
   6.130 +
   6.131 +All guest OS disk access goes through the virtual block device VBD
   6.132 +interface.  This interface allows domains access to portions of block
   6.133 +storage devices visible to the the block backend device.  The VBD
   6.134 +interface is a split driver, similar to the network interface
   6.135 +described above.  A single shared memory ring is used between the
   6.136 +frontend and backend drivers, across which read and write messages are
   6.137 +sent.
   6.138 +
   6.139 +Any block device accessible to the backend domain, including
   6.140 +network-based block (iSCSI, *NBD, etc), loopback and LVM/MD devices,
   6.141 +can be exported as a VBD.  Each VBD is mapped to a device node in the
   6.142 +guest, specified in the guest's startup configuration.
   6.143 +
   6.144 +Old (Xen 1.2) virtual disks are not supported under Xen 2.0, since
   6.145 +similar functionality can be achieved using the more complete LVM
   6.146 +system, which is already in widespread use.
   6.147 +
   6.148 +\subsection{Data Transfer}
   6.149 +
   6.150 +The single ring between the guest and the block backend supports three
   6.151 +messages:
   6.152 +
   6.153 +\begin{description}
   6.154 +\item [{\small {\tt PROBE}}:] Return a list of the VBDs available to
   6.155 +  this guest from the backend.  The request includes a descriptor of a
   6.156 +  free page into which the reply will be written by the backend.
   6.157 +
   6.158 +\item [{\small {\tt READ}}:] Read data from the specified block
   6.159 +  device.  The front end identifies the device and location to read
   6.160 +  from and attaches pages for the data to be copied to (typically via
   6.161 +  DMA from the device).  The backend acknowledges completed read
   6.162 +  requests as they finish.
   6.163 +
   6.164 +\item [{\small {\tt WRITE}}:] Write data to the specified block
   6.165 +  device.  This functions essentially as {\small {\tt READ}}, except
   6.166 +  that the data moves to the device instead of from it.
   6.167 +\end{description}
   6.168 +
   6.169 +%% um... some old text: In overview, the same style of descriptor-ring
   6.170 +%% that is used for network packets is used here.  Each domain has one
   6.171 +%% ring that carries operation requests to the hypervisor and carries
   6.172 +%% the results back again.
   6.173 +
   6.174 +%% Rather than copying data, the backend simply maps the domain's
   6.175 +%% buffers in order to enable direct DMA to them.  The act of mapping
   6.176 +%% the buffers also increases the reference counts of the underlying
   6.177 +%% pages, so that the unprivileged domain cannot try to return them to
   6.178 +%% the hypervisor, install them as page tables, or any other unsafe
   6.179 +%% behaviour.
   6.180 +%%
   6.181 +%% % block API here

     7.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     7.2 +++ b/docs/src/interface/further_info.tex	Tue Sep 20 09:43:46 2005 +0000
     7.3 @@ -0,0 +1,49 @@
     7.4 +\chapter{Further Information}
     7.5 +
     7.6 +If you have questions that are not answered by this manual, the
     7.7 +sources of information listed below may be of interest to you.  Note
     7.8 +that bug reports, suggestions and contributions related to the
     7.9 +software (or the documentation) should be sent to the Xen developers'
    7.10 +mailing list (address below).
    7.11 +
    7.12 +
    7.13 +\section{Other documentation}
    7.14 +
    7.15 +If you are mainly interested in using (rather than developing for)
    7.16 +Xen, the \emph{Xen Users' Manual} is distributed in the {\tt docs/}
    7.17 +directory of the Xen source distribution.
    7.18 +
    7.19 +% Various HOWTOs are also available in {\tt docs/HOWTOS}.
    7.20 +
    7.21 +
    7.22 +\section{Online references}
    7.23 +
    7.24 +The official Xen web site is found at:
    7.25 +\begin{quote}
    7.26 +{\tt http://www.cl.cam.ac.uk/Research/SRG/netos/xen/}
    7.27 +\end{quote}
    7.28 +
    7.29 +This contains links to the latest versions of all on-line
    7.30 +documentation.
    7.31 +
    7.32 +
    7.33 +\section{Mailing lists}
    7.34 +
    7.35 +There are currently four official Xen mailing lists:
    7.36 +
    7.37 +\begin{description}
    7.38 +\item[xen-devel@lists.xensource.com] Used for development
    7.39 +  discussions and bug reports.  Subscribe at: \\
    7.40 +  {\small {\tt http://lists.xensource.com/xen-devel}}
    7.41 +\item[xen-users@lists.xensource.com] Used for installation and usage
    7.42 +  discussions and requests for help.  Subscribe at: \\
    7.43 +  {\small {\tt http://lists.xensource.com/xen-users}}
    7.44 +\item[xen-announce@lists.xensource.com] Used for announcements only.
    7.45 +  Subscribe at: \\
    7.46 +  {\small {\tt http://lists.xensource.com/xen-announce}}
    7.47 +\item[xen-changelog@lists.xensource.com] Changelog feed
    7.48 +  from the unstable and 2.0 trees - developer oriented.  Subscribe at: \\
    7.49 +  {\small {\tt http://lists.xensource.com/xen-changelog}}
    7.50 +\end{description}
    7.51 +
    7.52 +Of these, xen-devel is the most active.

     8.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     8.2 +++ b/docs/src/interface/hypercalls.tex	Tue Sep 20 09:43:46 2005 +0000
     8.3 @@ -0,0 +1,524 @@
     8.4 +
     8.5 +\newcommand{\hypercall}[1]{\vspace{2mm}{\sf #1}}
     8.6 +
     8.7 +\chapter{Xen Hypercalls}
     8.8 +\label{a:hypercalls}
     8.9 +
    8.10 +Hypercalls represent the procedural interface to Xen; this appendix 
    8.11 +categorizes and describes the current set of hypercalls. 
    8.12 +
    8.13 +\section{Invoking Hypercalls} 
    8.14 +
    8.15 +Hypercalls are invoked in a manner analogous to system calls in a
    8.16 +conventional operating system; a software interrupt is issued which
    8.17 +vectors to an entry point within Xen. On x86\_32 machines the
    8.18 +instruction required is {\tt int \$82}; the (real) IDT is setup so
    8.19 +that this may only be issued from within ring 1. The particular 
    8.20 +hypercall to be invoked is contained in {\tt EAX} --- a list 
    8.21 +mapping these values to symbolic hypercall names can be found 
    8.22 +in {\tt xen/include/public/xen.h}. 
    8.23 +
    8.24 +On some occasions a set of hypercalls will be required to carry
    8.25 +out a higher-level function; a good example is when a guest 
    8.26 +operating wishes to context switch to a new process which 
    8.27 +requires updating various privileged CPU state. As an optimization
    8.28 +for these cases, there is a generic mechanism to issue a set of 
    8.29 +hypercalls as a batch: 
    8.30 +
    8.31 +\begin{quote}
    8.32 +\hypercall{multicall(void *call\_list, int nr\_calls)}
    8.33 +
    8.34 +Execute a series of hypervisor calls; {\tt nr\_calls} is the length of
    8.35 +the array of {\tt multicall\_entry\_t} structures pointed to be {\tt
    8.36 +call\_list}. Each entry contains the hypercall operation code followed
    8.37 +by up to 7 word-sized arguments.
    8.38 +\end{quote}
    8.39 +
    8.40 +Note that multicalls are provided purely as an optimization; there is
    8.41 +no requirement to use them when first porting a guest operating
    8.42 +system.
    8.43 +
    8.44 +
    8.45 +\section{Virtual CPU Setup} 
    8.46 +
    8.47 +At start of day, a guest operating system needs to setup the virtual
    8.48 +CPU it is executing on. This includes installing vectors for the
    8.49 +virtual IDT so that the guest OS can handle interrupts, page faults,
    8.50 +etc. However the very first thing a guest OS must setup is a pair 
    8.51 +of hypervisor callbacks: these are the entry points which Xen will
    8.52 +use when it wishes to notify the guest OS of an occurrence. 
    8.53 +
    8.54 +\begin{quote}
    8.55 +\hypercall{set\_callbacks(unsigned long event\_selector, unsigned long
    8.56 +  event\_address, unsigned long failsafe\_selector, unsigned long
    8.57 +  failsafe\_address) }
    8.58 +
    8.59 +Register the normal (``event'') and failsafe callbacks for 
    8.60 +event processing. In each case the code segment selector and 
    8.61 +address within that segment are provided. The selectors must
    8.62 +have RPL 1; in XenLinux we simply use the kernel's CS for both 
    8.63 +{\tt event\_selector} and {\tt failsafe\_selector}.
    8.64 +
    8.65 +The value {\tt event\_address} specifies the address of the guest OSes
    8.66 +event handling and dispatch routine; the {\tt failsafe\_address}
    8.67 +specifies a separate entry point which is used only if a fault occurs
    8.68 +when Xen attempts to use the normal callback. 
    8.69 +\end{quote} 
    8.70 +
    8.71 +
    8.72 +After installing the hypervisor callbacks, the guest OS can 
    8.73 +install a `virtual IDT' by using the following hypercall: 
    8.74 +
    8.75 +\begin{quote} 
    8.76 +\hypercall{set\_trap\_table(trap\_info\_t *table)} 
    8.77 +
    8.78 +Install one or more entries into the per-domain 
    8.79 +trap handler table (essentially a software version of the IDT). 
    8.80 +Each entry in the array pointed to by {\tt table} includes the 
    8.81 +exception vector number with the corresponding segment selector 
    8.82 +and entry point. Most guest OSes can use the same handlers on 
    8.83 +Xen as when running on the real hardware; an exception is the 
    8.84 +page fault handler (exception vector 14) where a modified 
    8.85 +stack-frame layout is used. 
    8.86 +
    8.87 +
    8.88 +\end{quote} 
    8.89 +
    8.90 +
    8.91 +
    8.92 +\section{Scheduling and Timer}
    8.93 +
    8.94 +Domains are preemptively scheduled by Xen according to the 
    8.95 +parameters installed by domain 0 (see Section~\ref{s:dom0ops}). 
    8.96 +In addition, however, a domain may choose to explicitly 
    8.97 +control certain behavior with the following hypercall: 
    8.98 +
    8.99 +\begin{quote} 
   8.100 +\hypercall{sched\_op(unsigned long op)} 
   8.101 +
   8.102 +Request scheduling operation from hypervisor. The options are: {\it
   8.103 +yield}, {\it block}, and {\it shutdown}.  {\it yield} keeps the
   8.104 +calling domain runnable but may cause a reschedule if other domains
   8.105 +are runnable.  {\it block} removes the calling domain from the run
   8.106 +queue and cause is to sleeps until an event is delivered to it.  {\it
   8.107 +shutdown} is used to end the domain's execution; the caller can
   8.108 +additionally specify whether the domain should reboot, halt or
   8.109 +suspend.
   8.110 +\end{quote} 
   8.111 +
   8.112 +To aid the implementation of a process scheduler within a guest OS,
   8.113 +Xen provides a virtual programmable timer:
   8.114 +
   8.115 +\begin{quote}
   8.116 +\hypercall{set\_timer\_op(uint64\_t timeout)} 
   8.117 +
   8.118 +Request a timer event to be sent at the specified system time (time 
   8.119 +in nanoseconds since system boot). The hypercall actually passes the 
   8.120 +64-bit timeout value as a pair of 32-bit values. 
   8.121 +
   8.122 +\end{quote} 
   8.123 +
   8.124 +Note that calling {\tt set\_timer\_op()} prior to {\tt sched\_op} 
   8.125 +allows block-with-timeout semantics. 
   8.126 +
   8.127 +
   8.128 +\section{Page Table Management} 
   8.129 +
   8.130 +Since guest operating systems have read-only access to their page 
   8.131 +tables, Xen must be involved when making any changes. The following
   8.132 +multi-purpose hypercall can be used to modify page-table entries, 
   8.133 +update the machine-to-physical mapping table, flush the TLB, install 
   8.134 +a new page-table base pointer, and more.
   8.135 +
   8.136 +\begin{quote} 
   8.137 +\hypercall{mmu\_update(mmu\_update\_t *req, int count, int *success\_count)} 
   8.138 +
   8.139 +Update the page table for the domain; a set of {\tt count} updates are
   8.140 +submitted for processing in a batch, with {\tt success\_count} being 
   8.141 +updated to report the number of successful updates.  
   8.142 +
   8.143 +Each element of {\tt req[]} contains a pointer (address) and value; 
   8.144 +the least significant 2-bits of the pointer are used to distinguish 
   8.145 +the type of update requested as follows:
   8.146 +\begin{description} 
   8.147 +
   8.148 +\item[\it MMU\_NORMAL\_PT\_UPDATE:] update a page directory entry or
   8.149 +page table entry to the associated value; Xen will check that the
   8.150 +update is safe, as described in Chapter~\ref{c:memory}.
   8.151 +
   8.152 +\item[\it MMU\_MACHPHYS\_UPDATE:] update an entry in the
   8.153 +  machine-to-physical table. The calling domain must own the machine
   8.154 +  page in question (or be privileged).
   8.155 +
   8.156 +\item[\it MMU\_EXTENDED\_COMMAND:] perform additional MMU operations.
   8.157 +The set of additional MMU operations is considerable, and includes
   8.158 +updating {\tt cr3} (or just re-installing it for a TLB flush),
   8.159 +flushing the cache, installing a new LDT, or pinning \& unpinning
   8.160 +page-table pages (to ensure their reference count doesn't drop to zero
   8.161 +which would require a revalidation of all entries).
   8.162 +
   8.163 +Further extended commands are used to deal with granting and 
   8.164 +acquiring page ownership; see Section~\ref{s:idc}. 
   8.165 +
   8.166 +
   8.167 +\end{description}
   8.168 +
   8.169 +More details on the precise format of all commands can be 
   8.170 +found in {\tt xen/include/public/xen.h}. 
   8.171 +
   8.172 +
   8.173 +\end{quote}
   8.174 +
   8.175 +Explicitly updating batches of page table entries is extremely
   8.176 +efficient, but can require a number of alterations to the guest
   8.177 +OS. Using the writable page table mode (Chapter~\ref{c:memory}) is
   8.178 +recommended for new OS ports.
   8.179 +
   8.180 +Regardless of which page table update mode is being used, however,
   8.181 +there are some occasions (notably handling a demand page fault) where
   8.182 +a guest OS will wish to modify exactly one PTE rather than a
   8.183 +batch. This is catered for by the following:
   8.184 +
   8.185 +\begin{quote} 
   8.186 +\hypercall{update\_va\_mapping(unsigned long page\_nr, unsigned long
   8.187 +val, \\ unsigned long flags)}
   8.188 +
   8.189 +Update the currently installed PTE for the page {\tt page\_nr} to 
   8.190 +{\tt val}. As with {\tt mmu\_update()}, Xen checks the modification 
   8.191 +is safe before applying it. The {\tt flags} determine which kind
   8.192 +of TLB flush, if any, should follow the update. 
   8.193 +
   8.194 +\end{quote} 
   8.195 +
   8.196 +Finally, sufficiently privileged domains may occasionally wish to manipulate 
   8.197 +the pages of others: 
   8.198 +\begin{quote}
   8.199 +
   8.200 +\hypercall{update\_va\_mapping\_otherdomain(unsigned long page\_nr,
   8.201 +unsigned long val, unsigned long flags, uint16\_t domid)}
   8.202 +
   8.203 +Identical to {\tt update\_va\_mapping()} save that the pages being
   8.204 +mapped must belong to the domain {\tt domid}. 
   8.205 +
   8.206 +\end{quote}
   8.207 +
   8.208 +This privileged operation is currently used by backend virtual device
   8.209 +drivers to safely map pages containing I/O data. 
   8.210 +
   8.211 +
   8.212 +
   8.213 +\section{Segmentation Support}
   8.214 +
   8.215 +Xen allows guest OSes to install a custom GDT if they require it; 
   8.216 +this is context switched transparently whenever a domain is 
   8.217 +[de]scheduled.  The following hypercall is effectively a 
   8.218 +`safe' version of {\tt lgdt}: 
   8.219 +
   8.220 +\begin{quote}
   8.221 +\hypercall{set\_gdt(unsigned long *frame\_list, int entries)} 
   8.222 +
   8.223 +Install a global descriptor table for a domain; {\tt frame\_list} is
   8.224 +an array of up to 16 machine page frames within which the GDT resides,
   8.225 +with {\tt entries} being the actual number of descriptor-entry
   8.226 +slots. All page frames must be mapped read-only within the guest's
   8.227 +address space, and the table must be large enough to contain Xen's
   8.228 +reserved entries (see {\tt xen/include/public/arch-x86\_32.h}).
   8.229 +
   8.230 +\end{quote}
   8.231 +
   8.232 +Many guest OSes will also wish to install LDTs; this is achieved by
   8.233 +using {\tt mmu\_update()} with an extended command, passing the
   8.234 +linear address of the LDT base along with the number of entries. No
   8.235 +special safety checks are required; Xen needs to perform this task
   8.236 +simply since {\tt lldt} requires CPL 0.
   8.237 +
   8.238 +
   8.239 +Xen also allows guest operating systems to update just an 
   8.240 +individual segment descriptor in the GDT or LDT:  
   8.241 +
   8.242 +\begin{quote}
   8.243 +\hypercall{update\_descriptor(unsigned long ma, unsigned long word1,
   8.244 +unsigned long word2)}
   8.245 +
   8.246 +Update the GDT/LDT entry at machine address {\tt ma}; the new
   8.247 +8-byte descriptor is stored in {\tt word1} and {\tt word2}.
   8.248 +Xen performs a number of checks to ensure the descriptor is 
   8.249 +valid. 
   8.250 +
   8.251 +\end{quote}
   8.252 +
   8.253 +Guest OSes can use the above in place of context switching entire 
   8.254 +LDTs (or the GDT) when the number of changing descriptors is small. 
   8.255 +
   8.256 +\section{Context Switching} 
   8.257 +
   8.258 +When a guest OS wishes to context switch between two processes, 
   8.259 +it can use the page table and segmentation hypercalls described
   8.260 +above to perform the the bulk of the privileged work. In addition, 
   8.261 +however, it will need to invoke Xen to switch the kernel (ring 1) 
   8.262 +stack pointer: 
   8.263 +
   8.264 +\begin{quote} 
   8.265 +\hypercall{stack\_switch(unsigned long ss, unsigned long esp)} 
   8.266 +
   8.267 +Request kernel stack switch from hypervisor; {\tt ss} is the new 
   8.268 +stack segment, which {\tt esp} is the new stack pointer. 
   8.269 +
   8.270 +\end{quote} 
   8.271 +
   8.272 +A final useful hypercall for context switching allows ``lazy'' 
   8.273 +save and restore of floating point state: 
   8.274 +
   8.275 +\begin{quote}
   8.276 +\hypercall{fpu\_taskswitch(void)} 
   8.277 +
   8.278 +This call instructs Xen to set the {\tt TS} bit in the {\tt cr0}
   8.279 +control register; this means that the next attempt to use floating
   8.280 +point will cause a trap which the guest OS can trap. Typically it will
   8.281 +then save/restore the FP state, and clear the {\tt TS} bit. 
   8.282 +\end{quote} 
   8.283 +
   8.284 +This is provided as an optimization only; guest OSes can also choose
   8.285 +to save and restore FP state on all context switches for simplicity. 
   8.286 +
   8.287 +
   8.288 +\section{Physical Memory Management}
   8.289 +
   8.290 +As mentioned previously, each domain has a maximum and current 
   8.291 +memory allocation. The maximum allocation, set at domain creation 
   8.292 +time, cannot be modified. However a domain can choose to reduce 
   8.293 +and subsequently grow its current allocation by using the
   8.294 +following call: 
   8.295 +
   8.296 +\begin{quote} 
   8.297 +\hypercall{dom\_mem\_op(unsigned int op, unsigned long *extent\_list,
   8.298 +  unsigned long nr\_extents, unsigned int extent\_order)}
   8.299 +
   8.300 +Increase or decrease current memory allocation (as determined by 
   8.301 +the value of {\tt op}). Each invocation provides a list of 
   8.302 +extents each of which is $2^s$ pages in size, 
   8.303 +where $s$ is the value of {\tt extent\_order}. 
   8.304 +
   8.305 +\end{quote} 
   8.306 +
   8.307 +In addition to simply reducing or increasing the current memory
   8.308 +allocation via a `balloon driver', this call is also useful for 
   8.309 +obtaining contiguous regions of machine memory when required (e.g. 
   8.310 +for certain PCI devices, or if using superpages).  
   8.311 +
   8.312 +
   8.313 +\section{Inter-Domain Communication}
   8.314 +\label{s:idc} 
   8.315 +
   8.316 +Xen provides a simple asynchronous notification mechanism via
   8.317 +\emph{event channels}. Each domain has a set of end-points (or
   8.318 +\emph{ports}) which may be bound to an event source (e.g. a physical
   8.319 +IRQ, a virtual IRQ, or an port in another domain). When a pair of
   8.320 +end-points in two different domains are bound together, then a `send'
   8.321 +operation on one will cause an event to be received by the destination
   8.322 +domain.
   8.323 +
   8.324 +The control and use of event channels involves the following hypercall: 
   8.325 +
   8.326 +\begin{quote}
   8.327 +\hypercall{event\_channel\_op(evtchn\_op\_t *op)} 
   8.328 +
   8.329 +Inter-domain event-channel management; {\tt op} is a discriminated 
   8.330 +union which allows the following 7 operations: 
   8.331 +
   8.332 +\begin{description} 
   8.333 +
   8.334 +\item[\it alloc\_unbound:] allocate a free (unbound) local
   8.335 +  port and prepare for connection from a specified domain. 
   8.336 +\item[\it bind\_virq:] bind a local port to a virtual 
   8.337 +IRQ; any particular VIRQ can be bound to at most one port per domain. 
   8.338 +\item[\it bind\_pirq:] bind a local port to a physical IRQ;
   8.339 +once more, a given pIRQ can be bound to at most one port per
   8.340 +domain. Furthermore the calling domain must be sufficiently
   8.341 +privileged.
   8.342 +\item[\it bind\_interdomain:] construct an interdomain event 
   8.343 +channel; in general, the target domain must have previously allocated 
   8.344 +an unbound port for this channel, although this can be bypassed by 
   8.345 +privileged domains during domain setup. 
   8.346 +\item[\it close:] close an interdomain event channel. 
   8.347 +\item[\it send:] send an event to the remote end of a 
   8.348 +interdomain event channel. 
   8.349 +\item[\it status:] determine the current status of a local port. 
   8.350 +\end{description} 
   8.351 +
   8.352 +For more details see
   8.353 +{\tt xen/include/public/event\_channel.h}. 
   8.354 +
   8.355 +\end{quote} 
   8.356 +
   8.357 +Event channels are the fundamental communication primitive between 
   8.358 +Xen domains and seamlessly support SMP. However they provide little
   8.359 +bandwidth for communication {\sl per se}, and hence are typically 
   8.360 +married with a piece of shared memory to produce effective and 
   8.361 +high-performance inter-domain communication. 
   8.362 +
   8.363 +Safe sharing of memory pages between guest OSes is carried out by
   8.364 +granting access on a per page basis to individual domains. This is
   8.365 +achieved by using the {\tt grant\_table\_op()} hypercall.
   8.366 +
   8.367 +\begin{quote}
   8.368 +\hypercall{grant\_table\_op(unsigned int cmd, void *uop, unsigned int count)}
   8.369 +
   8.370 +Grant or remove access to a particular page to a particular domain. 
   8.371 +
   8.372 +\end{quote} 
   8.373 +
   8.374 +This is not currently widely in use by guest operating systems, but 
   8.375 +we intend to integrate support more fully in the near future. 
   8.376 +
   8.377 +\section{PCI Configuration} 
   8.378 +
   8.379 +Domains with physical device access (i.e.\ driver domains) receive
   8.380 +limited access to certain PCI devices (bus address space and
   8.381 +interrupts). However many guest operating systems attempt to 
   8.382 +determine the PCI configuration by directly access the PCI BIOS, 
   8.383 +which cannot be allowed for safety. 
   8.384 +
   8.385 +Instead, Xen provides the following hypercall: 
   8.386 +
   8.387 +\begin{quote}
   8.388 +\hypercall{physdev\_op(void *physdev\_op)}
   8.389 +
   8.390 +Perform a PCI configuration option; depending on the value 
   8.391 +of {\tt physdev\_op} this can be a PCI config read, a PCI config 
   8.392 +write, or a small number of other queries. 
   8.393 +
   8.394 +\end{quote} 
   8.395 +
   8.396 +
   8.397 +For examples of using {\tt physdev\_op()}, see the 
   8.398 +Xen-specific PCI code in the linux sparse tree. 
   8.399 +
   8.400 +\section{Administrative Operations}
   8.401 +\label{s:dom0ops}
   8.402 +
   8.403 +A large number of control operations are available to a sufficiently
   8.404 +privileged domain (typically domain 0). These allow the creation and
   8.405 +management of new domains, for example. A complete list is given 
   8.406 +below: for more details on any or all of these, please see 
   8.407 +{\tt xen/include/public/dom0\_ops.h} 
   8.408 +
   8.409 +
   8.410 +\begin{quote}
   8.411 +\hypercall{dom0\_op(dom0\_op\_t *op)} 
   8.412 +
   8.413 +Administrative domain operations for domain management. The options are:
   8.414 +
   8.415 +\begin{description} 
   8.416 +\item [\it DOM0\_CREATEDOMAIN:] create a new domain
   8.417 +
   8.418 +\item [\it DOM0\_PAUSEDOMAIN:] remove a domain from the scheduler run 
   8.419 +queue. 
   8.420 +
   8.421 +\item [\it DOM0\_UNPAUSEDOMAIN:] mark a paused domain as schedulable
   8.422 +  once again. 
   8.423 +
   8.424 +\item [\it DOM0\_DESTROYDOMAIN:] deallocate all resources associated
   8.425 +with a domain
   8.426 +
   8.427 +\item [\it DOM0\_GETMEMLIST:] get list of pages used by the domain
   8.428 +
   8.429 +\item [\it DOM0\_SCHEDCTL:]
   8.430 +
   8.431 +\item [\it DOM0\_ADJUSTDOM:] adjust scheduling priorities for domain
   8.432 +
   8.433 +\item [\it DOM0\_BUILDDOMAIN:] do final guest OS setup for domain
   8.434 +
   8.435 +\item [\it DOM0\_GETDOMAINFO:] get statistics about the domain
   8.436 +
   8.437 +\item [\it DOM0\_GETPAGEFRAMEINFO:] 
   8.438 +
   8.439 +\item [\it DOM0\_GETPAGEFRAMEINFO2:]
   8.440 +
   8.441 +\item [\it DOM0\_IOPL:] set I/O privilege level
   8.442 +
   8.443 +\item [\it DOM0\_MSR:] read or write model specific registers
   8.444 +
   8.445 +\item [\it DOM0\_DEBUG:] interactively invoke the debugger
   8.446 +
   8.447 +\item [\it DOM0\_SETTIME:] set system time
   8.448 +
   8.449 +\item [\it DOM0\_READCONSOLE:] read console content from hypervisor buffer ring
   8.450 +
   8.451 +\item [\it DOM0\_PINCPUDOMAIN:] pin domain to a particular CPU
   8.452 +
   8.453 +\item [\it DOM0\_GETTBUFS:] get information about the size and location of
   8.454 +                      the trace buffers (only on trace-buffer enabled builds)
   8.455 +
   8.456 +\item [\it DOM0\_PHYSINFO:] get information about the host machine
   8.457 +
   8.458 +\item [\it DOM0\_PCIDEV\_ACCESS:] modify PCI device access permissions
   8.459 +
   8.460 +\item [\it DOM0\_SCHED\_ID:] get the ID of the current Xen scheduler
   8.461 +
   8.462 +\item [\it DOM0\_SHADOW\_CONTROL:] switch between shadow page-table modes
   8.463 +
   8.464 +\item [\it DOM0\_SETDOMAININITIALMEM:] set initial memory allocation of a domain
   8.465 +
   8.466 +\item [\it DOM0\_SETDOMAINMAXMEM:] set maximum memory allocation of a domain
   8.467 +
   8.468 +\item [\it DOM0\_SETDOMAINVMASSIST:] set domain VM assist options
   8.469 +\end{description} 
   8.470 +\end{quote} 
   8.471 +
   8.472 +Most of the above are best understood by looking at the code 
   8.473 +implementing them (in {\tt xen/common/dom0\_ops.c}) and in 
   8.474 +the user-space tools that use them (mostly in {\tt tools/libxc}). 
   8.475 +
   8.476 +\section{Debugging Hypercalls} 
   8.477 +
   8.478 +A few additional hypercalls are mainly useful for debugging: 
   8.479 +
   8.480 +\begin{quote} 
   8.481 +\hypercall{console\_io(int cmd, int count, char *str)}
   8.482 +
   8.483 +Use Xen to interact with the console; operations are:
   8.484 +
   8.485 +{\it CONSOLEIO\_write}: Output count characters from buffer str.
   8.486 +
   8.487 +{\it CONSOLEIO\_read}: Input at most count characters into buffer str.
   8.488 +\end{quote} 
   8.489 +
   8.490 +A pair of hypercalls allows access to the underlying debug registers: 
   8.491 +\begin{quote}
   8.492 +\hypercall{set\_debugreg(int reg, unsigned long value)}
   8.493 +
   8.494 +Set debug register {\tt reg} to {\tt value} 
   8.495 +
   8.496 +\hypercall{get\_debugreg(int reg)}
   8.497 +
   8.498 +Return the contents of the debug register {\tt reg}
   8.499 +\end{quote}
   8.500 +
   8.501 +And finally: 
   8.502 +\begin{quote}
   8.503 +\hypercall{xen\_version(int cmd)}
   8.504 +
   8.505 +Request Xen version number.
   8.506 +\end{quote} 
   8.507 +
   8.508 +This is useful to ensure that user-space tools are in sync 
   8.509 +with the underlying hypervisor. 
   8.510 +
   8.511 +\section{Deprecated Hypercalls}
   8.512 +
   8.513 +Xen is under constant development and refinement; as such there 
   8.514 +are plans to improve the way in which various pieces of functionality 
   8.515 +are exposed to guest OSes. 
   8.516 +
   8.517 +\begin{quote} 
   8.518 +\hypercall{vm\_assist(unsigned int cmd, unsigned int type)}
   8.519 +
   8.520 +Toggle various memory management modes (in particular wrritable page
   8.521 +tables and superpage support). 
   8.522 +
   8.523 +\end{quote} 
   8.524 +
   8.525 +This is likely to be replaced with mode values in the shared 
   8.526 +information page since this is more resilient for resumption 
   8.527 +after migration or checkpoint. 

     9.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     9.2 +++ b/docs/src/interface/memory.tex	Tue Sep 20 09:43:46 2005 +0000
     9.3 @@ -0,0 +1,162 @@
     9.4 +\chapter{Memory}
     9.5 +\label{c:memory} 
     9.6 +
     9.7 +Xen is responsible for managing the allocation of physical memory to
     9.8 +domains, and for ensuring safe use of the paging and segmentation
     9.9 +hardware.
    9.10 +
    9.11 +
    9.12 +\section{Memory Allocation}
    9.13 +
    9.14 +Xen resides within a small fixed portion of physical memory; it also
    9.15 +reserves the top 64MB of every virtual address space. The remaining
    9.16 +physical memory is available for allocation to domains at a page
    9.17 +granularity.  Xen tracks the ownership and use of each page, which
    9.18 +allows it to enforce secure partitioning between domains.
    9.19 +
    9.20 +Each domain has a maximum and current physical memory allocation.  A
    9.21 +guest OS may run a `balloon driver' to dynamically adjust its current
    9.22 +memory allocation up to its limit.
    9.23 +
    9.24 +
    9.25 +%% XXX SMH: I use machine and physical in the next section (which is
    9.26 +%% kinda required for consistency with code); wonder if this section
    9.27 +%% should use same terms?
    9.28 +%%
    9.29 +%% Probably. 
    9.30 +%%
    9.31 +%% Merging this and below section at some point prob makes sense.
    9.32 +
    9.33 +\section{Pseudo-Physical Memory}
    9.34 +
    9.35 +Since physical memory is allocated and freed on a page granularity,
    9.36 +there is no guarantee that a domain will receive a contiguous stretch
    9.37 +of physical memory. However most operating systems do not have good
    9.38 +support for operating in a fragmented physical address space. To aid
    9.39 +porting such operating systems to run on top of Xen, we make a
    9.40 +distinction between \emph{machine memory} and \emph{pseudo-physical
    9.41 +  memory}.
    9.42 +
    9.43 +Put simply, machine memory refers to the entire amount of memory
    9.44 +installed in the machine, including that reserved by Xen, in use by
    9.45 +various domains, or currently unallocated. We consider machine memory
    9.46 +to comprise a set of 4K \emph{machine page frames} numbered
    9.47 +consecutively starting from 0. Machine frame numbers mean the same
    9.48 +within Xen or any domain.
    9.49 +
    9.50 +Pseudo-physical memory, on the other hand, is a per-domain
    9.51 +abstraction. It allows a guest operating system to consider its memory
    9.52 +allocation to consist of a contiguous range of physical page frames
    9.53 +starting at physical frame 0, despite the fact that the underlying
    9.54 +machine page frames may be sparsely allocated and in any order.
    9.55 +
    9.56 +To achieve this, Xen maintains a globally readable {\it
    9.57 +  machine-to-physical} table which records the mapping from machine
    9.58 +page frames to pseudo-physical ones. In addition, each domain is
    9.59 +supplied with a {\it physical-to-machine} table which performs the
    9.60 +inverse mapping. Clearly the machine-to-physical table has size
    9.61 +proportional to the amount of RAM installed in the machine, while each
    9.62 +physical-to-machine table has size proportional to the memory
    9.63 +allocation of the given domain.
    9.64 +
    9.65 +Architecture dependent code in guest operating systems can then use
    9.66 +the two tables to provide the abstraction of pseudo-physical memory.
    9.67 +In general, only certain specialized parts of the operating system
    9.68 +(such as page table management) needs to understand the difference
    9.69 +between machine and pseudo-physical addresses.
    9.70 +
    9.71 +
    9.72 +\section{Page Table Updates}
    9.73 +
    9.74 +In the default mode of operation, Xen enforces read-only access to
    9.75 +page tables and requires guest operating systems to explicitly request
    9.76 +any modifications.  Xen validates all such requests and only applies
    9.77 +updates that it deems safe.  This is necessary to prevent domains from
    9.78 +adding arbitrary mappings to their page tables.
    9.79 +
    9.80 +To aid validation, Xen associates a type and reference count with each
    9.81 +memory page. A page has one of the following mutually-exclusive types
    9.82 +at any point in time: page directory ({\sf PD}), page table ({\sf
    9.83 +  PT}), local descriptor table ({\sf LDT}), global descriptor table
    9.84 +({\sf GDT}), or writable ({\sf RW}). Note that a guest OS may always
    9.85 +create readable mappings of its own memory regardless of its current
    9.86 +type.
    9.87 +
    9.88 +%%% XXX: possibly explain more about ref count 'lifecyle' here?
    9.89 +This mechanism is used to maintain the invariants required for safety;
    9.90 +for example, a domain cannot have a writable mapping to any part of a
    9.91 +page table as this would require the page concerned to simultaneously
    9.92 +be of types {\sf PT} and {\sf RW}.
    9.93 +
    9.94 +
    9.95 +% \section{Writable Page Tables}
    9.96 +
    9.97 +Xen also provides an alternative mode of operation in which guests be
    9.98 +have the illusion that their page tables are directly writable.  Of
    9.99 +course this is not really the case, since Xen must still validate
   9.100 +modifications to ensure secure partitioning. To this end, Xen traps
   9.101 +any write attempt to a memory page of type {\sf PT} (i.e., that is
   9.102 +currently part of a page table).  If such an access occurs, Xen
   9.103 +temporarily allows write access to that page while at the same time
   9.104 +\emph{disconnecting} it from the page table that is currently in use.
   9.105 +This allows the guest to safely make updates to the page because the
   9.106 +newly-updated entries cannot be used by the MMU until Xen revalidates
   9.107 +and reconnects the page.  Reconnection occurs automatically in a
   9.108 +number of situations: for example, when the guest modifies a different
   9.109 +page-table page, when the domain is preempted, or whenever the guest
   9.110 +uses Xen's explicit page-table update interfaces.
   9.111 +
   9.112 +Finally, Xen also supports a form of \emph{shadow page tables} in
   9.113 +which the guest OS uses a independent copy of page tables which are
   9.114 +unknown to the hardware (i.e.\ which are never pointed to by {\tt
   9.115 +  cr3}). Instead Xen propagates changes made to the guest's tables to
   9.116 +the real ones, and vice versa. This is useful for logging page writes
   9.117 +(e.g.\ for live migration or checkpoint). A full version of the shadow
   9.118 +page tables also allows guest OS porting with less effort.
   9.119 +
   9.120 +
   9.121 +\section{Segment Descriptor Tables}
   9.122 +
   9.123 +On boot a guest is supplied with a default GDT, which does not reside
   9.124 +within its own memory allocation.  If the guest wishes to use other
   9.125 +than the default `flat' ring-1 and ring-3 segments that this GDT
   9.126 +provides, it must register a custom GDT and/or LDT with Xen, allocated
   9.127 +from its own memory. Note that a number of GDT entries are reserved by
   9.128 +Xen -- any custom GDT must also include sufficient space for these
   9.129 +entries.
   9.130 +
   9.131 +For example, the following hypercall is used to specify a new GDT:
   9.132 +
   9.133 +\begin{quote}
   9.134 +  int {\bf set\_gdt}(unsigned long *{\em frame\_list}, int {\em
   9.135 +    entries})
   9.136 +
   9.137 +  \emph{frame\_list}: An array of up to 16 machine page frames within
   9.138 +  which the GDT resides.  Any frame registered as a GDT frame may only
   9.139 +  be mapped read-only within the guest's address space (e.g., no
   9.140 +  writable mappings, no use as a page-table page, and so on).
   9.141 +
   9.142 +  \emph{entries}: The number of descriptor-entry slots in the GDT.
   9.143 +  Note that the table must be large enough to contain Xen's reserved
   9.144 +  entries; thus we must have `{\em entries $>$
   9.145 +    LAST\_RESERVED\_GDT\_ENTRY}\ '.  Note also that, after registering
   9.146 +  the GDT, slots \emph{FIRST\_} through
   9.147 +  \emph{LAST\_RESERVED\_GDT\_ENTRY} are no longer usable by the guest
   9.148 +  and may be overwritten by Xen.
   9.149 +\end{quote}
   9.150 +
   9.151 +The LDT is updated via the generic MMU update mechanism (i.e., via the
   9.152 +{\tt mmu\_update()} hypercall.
   9.153 +
   9.154 +\section{Start of Day}
   9.155 +
   9.156 +The start-of-day environment for guest operating systems is rather
   9.157 +different to that provided by the underlying hardware. In particular,
   9.158 +the processor is already executing in protected mode with paging
   9.159 +enabled.
   9.160 +
   9.161 +{\it Domain 0} is created and booted by Xen itself. For all subsequent
   9.162 +domains, the analogue of the boot-loader is the {\it domain builder},
   9.163 +user-space software running in {\it domain 0}. The domain builder is
   9.164 +responsible for building the initial page tables for a domain and
   9.165 +loading its kernel image at the appropriate virtual address.

    10.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
    10.2 +++ b/docs/src/interface/scheduling.tex	Tue Sep 20 09:43:46 2005 +0000
    10.3 @@ -0,0 +1,268 @@
    10.4 +\chapter{Scheduling API}  
    10.5 +
    10.6 +The scheduling API is used by both the schedulers described above and should
    10.7 +also be used by any new schedulers.  It provides a generic interface and also
    10.8 +implements much of the ``boilerplate'' code.
    10.9 +
   10.10 +Schedulers conforming to this API are described by the following
   10.11 +structure:
   10.12 +
   10.13 +\begin{verbatim}
   10.14 +struct scheduler
   10.15 +{
   10.16 +    char *name;             /* full name for this scheduler      */
   10.17 +    char *opt_name;         /* option name for this scheduler    */
   10.18 +    unsigned int sched_id;  /* ID for this scheduler             */
   10.19 +
   10.20 +    int          (*init_scheduler) ();
   10.21 +    int          (*alloc_task)     (struct task_struct *);
   10.22 +    void         (*add_task)       (struct task_struct *);
   10.23 +    void         (*free_task)      (struct task_struct *);
   10.24 +    void         (*rem_task)       (struct task_struct *);
   10.25 +    void         (*wake_up)        (struct task_struct *);
   10.26 +    void         (*do_block)       (struct task_struct *);
   10.27 +    task_slice_t (*do_schedule)    (s_time_t);
   10.28 +    int          (*control)        (struct sched_ctl_cmd *);
   10.29 +    int          (*adjdom)         (struct task_struct *,
   10.30 +                                    struct sched_adjdom_cmd *);
   10.31 +    s32          (*reschedule)     (struct task_struct *);
   10.32 +    void         (*dump_settings)  (void);
   10.33 +    void         (*dump_cpu_state) (int);
   10.34 +    void         (*dump_runq_el)   (struct task_struct *);
   10.35 +};
   10.36 +\end{verbatim}
   10.37 +
   10.38 +The only method that {\em must} be implemented is
   10.39 +{\tt do\_schedule()}.  However, if there is not some implementation for the
   10.40 +{\tt wake\_up()} method then waking tasks will not get put on the runqueue!
   10.41 +
   10.42 +The fields of the above structure are described in more detail below.
   10.43 +
   10.44 +\subsubsection{name}
   10.45 +
   10.46 +The name field should point to a descriptive ASCII string.
   10.47 +
   10.48 +\subsubsection{opt\_name}
   10.49 +
   10.50 +This field is the value of the {\tt sched=} boot-time option that will select
   10.51 +this scheduler.
   10.52 +
   10.53 +\subsubsection{sched\_id}
   10.54 +
   10.55 +This is an integer that uniquely identifies this scheduler.  There should be a
   10.56 +macro corrsponding to this scheduler ID in {\tt <xen/sched-if.h>}.
   10.57 +
   10.58 +\subsubsection{init\_scheduler}
   10.59 +
   10.60 +\paragraph*{Purpose}
   10.61 +
   10.62 +This is a function for performing any scheduler-specific initialisation.  For
   10.63 +instance, it might allocate memory for per-CPU scheduler data and initialise it
   10.64 +appropriately.
   10.65 +
   10.66 +\paragraph*{Call environment}
   10.67 +
   10.68 +This function is called after the initialisation performed by the generic
   10.69 +layer.  The function is called exactly once, for the scheduler that has been
   10.70 +selected.
   10.71 +
   10.72 +\paragraph*{Return values}
   10.73 +
   10.74 +This should return negative on failure --- this will cause an
   10.75 +immediate panic and the system will fail to boot.
   10.76 +
   10.77 +\subsubsection{alloc\_task}
   10.78 +
   10.79 +\paragraph*{Purpose}
   10.80 +Called when a {\tt task\_struct} is allocated by the generic scheduler
   10.81 +layer.  A particular scheduler implementation may use this method to
   10.82 +allocate per-task data for this task.  It may use the {\tt
   10.83 +sched\_priv} pointer in the {\tt task\_struct} to point to this data.
   10.84 +
   10.85 +\paragraph*{Call environment}
   10.86 +The generic layer guarantees that the {\tt sched\_priv} field will
   10.87 +remain intact from the time this method is called until the task is
   10.88 +deallocated (so long as the scheduler implementation does not change
   10.89 +it explicitly!).
   10.90 +
   10.91 +\paragraph*{Return values}
   10.92 +Negative on failure.
   10.93 +
   10.94 +\subsubsection{add\_task}
   10.95 +
   10.96 +\paragraph*{Purpose}
   10.97 +
   10.98 +Called when a task is initially added by the generic layer.
   10.99 +
  10.100 +\paragraph*{Call environment}
  10.101 +
  10.102 +The fields in the {\tt task\_struct} are now filled out and available for use.
  10.103 +Schedulers should implement appropriate initialisation of any per-task private
  10.104 +information in this method.
  10.105 +
  10.106 +\subsubsection{free\_task}
  10.107 +
  10.108 +\paragraph*{Purpose}
  10.109 +
  10.110 +Schedulers should free the space used by any associated private data
  10.111 +structures.
  10.112 +
  10.113 +\paragraph*{Call environment}
  10.114 +
  10.115 +This is called when a {\tt task\_struct} is about to be deallocated.
  10.116 +The generic layer will have done generic task removal operations and
  10.117 +(if implemented) called the scheduler's {\tt rem\_task} method before
  10.118 +this method is called.
  10.119 +
  10.120 +\subsubsection{rem\_task}
  10.121 +
  10.122 +\paragraph*{Purpose}
  10.123 +
  10.124 +This is called when a task is being removed from scheduling (but is
  10.125 +not yet being freed).
  10.126 +
  10.127 +\subsubsection{wake\_up}
  10.128 +
  10.129 +\paragraph*{Purpose}
  10.130 +
  10.131 +Called when a task is woken up, this method should put the task on the runqueue
  10.132 +(or do the scheduler-specific equivalent action).
  10.133 +
  10.134 +\paragraph*{Call environment}
  10.135 +
  10.136 +The task is already set to state RUNNING.
  10.137 +
  10.138 +\subsubsection{do\_block}
  10.139 +
  10.140 +\paragraph*{Purpose}
  10.141 +
  10.142 +This function is called when a task is blocked.  This function should
  10.143 +not remove the task from the runqueue.
  10.144 +
  10.145 +\paragraph*{Call environment}
  10.146 +
  10.147 +The EVENTS\_MASTER\_ENABLE\_BIT is already set and the task state changed to
  10.148 +TASK\_INTERRUPTIBLE on entry to this method.  A call to the {\tt
  10.149 +  do\_schedule} method will be made after this method returns, in
  10.150 +order to select the next task to run.
  10.151 +
  10.152 +\subsubsection{do\_schedule}
  10.153 +
  10.154 +This method must be implemented.
  10.155 +
  10.156 +\paragraph*{Purpose}
  10.157 +
  10.158 +The method is called each time a new task must be chosen for scheduling on the
  10.159 +current CPU.  The current time as passed as the single argument (the current
  10.160 +task can be found using the {\tt current} macro).
  10.161 +
  10.162 +This method should select the next task to run on this CPU and set it's minimum
  10.163 +time to run as well as returning the data described below.
  10.164 +
  10.165 +This method should also take the appropriate action if the previous
  10.166 +task has blocked, e.g. removing it from the runqueue.
  10.167 +
  10.168 +\paragraph*{Call environment}
  10.169 +
  10.170 +The other fields in the {\tt task\_struct} are updated by the generic layer,
  10.171 +which also performs all Xen-specific tasks and performs the actual task switch
  10.172 +(unless the previous task has been chosen again).
  10.173 +
  10.174 +This method is called with the {\tt schedule\_lock} held for the current CPU
  10.175 +and local interrupts disabled.
  10.176 +
  10.177 +\paragraph*{Return values}
  10.178 +
  10.179 +Must return a {\tt struct task\_slice} describing what task to run and how long
  10.180 +for (at maximum).
  10.181 +
  10.182 +\subsubsection{control}
  10.183 +
  10.184 +\paragraph*{Purpose}
  10.185 +
  10.186 +This method is called for global scheduler control operations.  It takes a
  10.187 +pointer to a {\tt struct sched\_ctl\_cmd}, which it should either
  10.188 +source data from or populate with data, depending on the value of the
  10.189 +{\tt direction} field.
  10.190 +
  10.191 +\paragraph*{Call environment}
  10.192 +
  10.193 +The generic layer guarantees that when this method is called, the
  10.194 +caller selected the correct scheduler ID, hence the scheduler's
  10.195 +implementation does not need to sanity-check these parts of the call.
  10.196 +
  10.197 +\paragraph*{Return values}
  10.198 +
  10.199 +This function should return the value to be passed back to user space, hence it
  10.200 +should either be 0 or an appropriate errno value.
  10.201 +
  10.202 +\subsubsection{sched\_adjdom}
  10.203 +
  10.204 +\paragraph*{Purpose}
  10.205 +
  10.206 +This method is called to adjust the scheduling parameters of a particular
  10.207 +domain, or to query their current values.  The function should check
  10.208 +the {\tt direction} field of the {\tt sched\_adjdom\_cmd} it receives in
  10.209 +order to determine which of these operations is being performed.
  10.210 +
  10.211 +\paragraph*{Call environment}
  10.212 +
  10.213 +The generic layer guarantees that the caller has specified the correct
  10.214 +control interface version and scheduler ID and that the supplied {\tt
  10.215 +task\_struct} will not be deallocated during the call (hence it is not
  10.216 +necessary to {\tt get\_task\_struct}).
  10.217 +
  10.218 +\paragraph*{Return values}
  10.219 +
  10.220 +This function should return the value to be passed back to user space, hence it
  10.221 +should either be 0 or an appropriate errno value.
  10.222 +
  10.223 +\subsubsection{reschedule}
  10.224 +
  10.225 +\paragraph*{Purpose}
  10.226 +
  10.227 +This method is called to determine if a reschedule is required as a result of a
  10.228 +particular task.
  10.229 +
  10.230 +\paragraph*{Call environment}
  10.231 +The generic layer will cause a reschedule if the current domain is the idle
  10.232 +task or it has exceeded its minimum time slice before a reschedule.  The
  10.233 +generic layer guarantees that the task passed is not currently running but is
  10.234 +on the runqueue.
  10.235 +
  10.236 +\paragraph*{Return values}
  10.237 +
  10.238 +Should return a mask of CPUs to cause a reschedule on.
  10.239 +
  10.240 +\subsubsection{dump\_settings}
  10.241 +
  10.242 +\paragraph*{Purpose}
  10.243 +
  10.244 +If implemented, this should dump any private global settings for this
  10.245 +scheduler to the console.
  10.246 +
  10.247 +\paragraph*{Call environment}
  10.248 +
  10.249 +This function is called with interrupts enabled.
  10.250 +
  10.251 +\subsubsection{dump\_cpu\_state}
  10.252 +
  10.253 +\paragraph*{Purpose}
  10.254 +
  10.255 +This method should dump any private settings for the specified CPU.
  10.256 +
  10.257 +\paragraph*{Call environment}
  10.258 +
  10.259 +This function is called with interrupts disabled and the {\tt schedule\_lock}
  10.260 +for the specified CPU held.
  10.261 +
  10.262 +\subsubsection{dump\_runq\_el}
  10.263 +
  10.264 +\paragraph*{Purpose}
  10.265 +
  10.266 +This method should dump any private settings for the specified task.
  10.267 +
  10.268 +\paragraph*{Call environment}
  10.269 +
  10.270 +This function is called with interrupts disabled and the {\tt schedule\_lock}
  10.271 +for the task's CPU held.

    11.1 --- a/docs/src/user.tex	Tue Sep 20 09:43:29 2005 +0000
    11.2 +++ b/docs/src/user.tex	Tue Sep 20 09:43:46 2005 +0000
    11.3 @@ -59,1803 +59,36 @@ Contributions of material, suggestions a
    11.4  \renewcommand{\floatpagefraction}{.8}
    11.5  \setstretch{1.1}
    11.6  
    11.7 +
    11.8  \part{Introduction and Tutorial}
    11.9 -\chapter{Introduction}
   11.10 -
   11.11 -Xen is a {\em paravirtualising} virtual machine monitor (VMM), or
   11.12 -`hypervisor', for the x86 processor architecture.  Xen can securely
   11.13 -execute multiple virtual machines on a single physical system with
   11.14 -close-to-native performance.  The virtual machine technology
   11.15 -facilitates enterprise-grade functionality, including:
   11.16 -
   11.17 -\begin{itemize}
   11.18 -\item Virtual machines with performance close to native
   11.19 -  hardware.
   11.20 -\item Live migration of running virtual machines between physical hosts.
   11.21 -\item Excellent hardware support (supports most Linux device drivers).
   11.22 -\item Sandboxed, restartable device drivers.
   11.23 -\end{itemize}
   11.24 -
   11.25 -Paravirtualisation permits very high performance virtualisation,
   11.26 -even on architectures like x86 that are traditionally
   11.27 -very hard to virtualise.
   11.28 -The drawback of this approach is that it requires operating systems to
   11.29 -be {\em ported} to run on Xen.  Porting an OS to run on Xen is similar
   11.30 -to supporting a new hardware platform, however the process
   11.31 -is simplified because the paravirtual machine architecture is very
   11.32 -similar to the underlying native hardware. Even though operating system
   11.33 -kernels must explicitly support Xen, a key feature is that user space
   11.34 -applications and libraries {\em do not} require modification.
   11.35 -
   11.36 -Xen support is available for increasingly many operating systems:
   11.37 -right now, Linux 2.4, Linux 2.6 and NetBSD are available for Xen 2.0.
   11.38 -A FreeBSD port is undergoing testing and will be incorporated into the
   11.39 -release soon. Other OS ports, including Plan 9, are in progress.  We
   11.40 -hope that that arch-xen patches will be incorporated into the
   11.41 -mainstream releases of these operating systems in due course (as has
   11.42 -already happened for NetBSD).
   11.43 -
   11.44 -Possible usage scenarios for Xen include:
   11.45 -\begin{description}
   11.46 -\item [Kernel development.] Test and debug kernel modifications in a
   11.47 -      sandboxed virtual machine --- no need for a separate test
   11.48 -      machine.
   11.49 -\item [Multiple OS configurations.] Run multiple operating systems
   11.50 -      simultaneously, for instance for compatibility or QA purposes.
   11.51 -\item [Server consolidation.] Move multiple servers onto a single
   11.52 -      physical host with performance and fault isolation provided at
   11.53 -      virtual machine boundaries. 
   11.54 -\item [Cluster computing.] Management at VM granularity provides more
   11.55 -      flexibility than separately managing each physical host, but
   11.56 -      better control and isolation than single-system image solutions, 
   11.57 -      particularly by using live migration for load balancing. 
   11.58 -\item [Hardware support for custom OSes.] Allow development of new OSes
   11.59 -      while benefiting from the wide-ranging hardware support of
   11.60 -      existing OSes such as Linux.
   11.61 -\end{description}
   11.62 -
   11.63 -\section{Structure of a Xen-Based System}
   11.64 -
   11.65 -A Xen system has multiple layers, the lowest and most privileged of
   11.66 -which is Xen itself. 
   11.67 -Xen in turn may host multiple {\em guest} operating systems, each of
   11.68 -which is executed within a secure virtual machine (in Xen terminology,
   11.69 -a {\em domain}). Domains are scheduled by Xen to make effective use of
   11.70 -the available physical CPUs.  Each guest OS manages its own
   11.71 -applications, which includes responsibility for scheduling each
   11.72 -application within the time allotted to the VM by Xen.
   11.73 -
   11.74 -The first domain, {\em domain 0}, is created automatically when the
   11.75 -system boots and has special management privileges. Domain 0 builds
   11.76 -other domains and manages their virtual devices. It also performs
   11.77 -administrative tasks such as suspending, resuming and migrating other
   11.78 -virtual machines.
   11.79 -
   11.80 -Within domain 0, a process called \emph{xend} runs to manage the system.
   11.81 -\Xend is responsible for managing virtual machines and providing access
   11.82 -to their consoles.  Commands are issued to \xend over an HTTP
   11.83 -interface, either from a command-line tool or from a web browser.
   11.84 -
   11.85 -\section{Hardware Support}
   11.86 -
   11.87 -Xen currently runs only on the x86 architecture, requiring a `P6' or
   11.88 -newer processor (e.g. Pentium Pro, Celeron, Pentium II, Pentium III,
   11.89 -Pentium IV, Xeon, AMD Athlon, AMD Duron).  Multiprocessor machines are
   11.90 -supported, and we also have basic support for HyperThreading (SMT),
   11.91 -although this remains a topic for ongoing research. A port
   11.92 -specifically for x86/64 is in progress, although Xen already runs on
   11.93 -such systems in 32-bit legacy mode. In addition a port to the IA64
   11.94 -architecture is approaching completion. We hope to add other
   11.95 -architectures such as PPC and ARM in due course.
   11.96 -
   11.97 -
   11.98 -Xen can currently use up to 4GB of memory.  It is possible for x86
   11.99 -machines to address up to 64GB of physical memory but there are no
  11.100 -current plans to support these systems: The x86/64 port is the
  11.101 -planned route to supporting larger memory sizes.
  11.102 -
  11.103 -Xen offloads most of the hardware support issues to the guest OS
  11.104 -running in Domain~0.  Xen itself contains only the code required to
  11.105 -detect and start secondary processors, set up interrupt routing, and
  11.106 -perform PCI bus enumeration.  Device drivers run within a privileged
  11.107 -guest OS rather than within Xen itself. This approach provides
  11.108 -compatibility with the majority of device hardware supported by Linux.
  11.109 -The default XenLinux build contains support for relatively modern
  11.110 -server-class network and disk hardware, but you can add support for
  11.111 -other hardware by configuring your XenLinux kernel in the normal way.
  11.112 -
  11.113 -\section{History}
  11.114 -
  11.115 -Xen was originally developed by the Systems Research Group at the
  11.116 -University of Cambridge Computer Laboratory as part of the XenoServers
  11.117 -project, funded by the UK-EPSRC.
  11.118 -XenoServers aim to provide a `public infrastructure for
  11.119 -global distributed computing', and Xen plays a key part in that,
  11.120 -allowing us to efficiently partition a single machine to enable
  11.121 -multiple independent clients to run their operating systems and
  11.122 -applications in an environment providing protection, resource
  11.123 -isolation and accounting.  The project web page contains further
  11.124 -information along with pointers to papers and technical reports:
  11.125 -\path{http://www.cl.cam.ac.uk/xeno} 
  11.126 -
  11.127 -Xen has since grown into a fully-fledged project in its own right,
  11.128 -enabling us to investigate interesting research issues regarding the
  11.129 -best techniques for virtualising resources such as the CPU, memory,
  11.130 -disk and network.  The project has been bolstered by support from
  11.131 -Intel Research Cambridge, and HP Labs, who are now working closely
  11.132 -with us.
  11.133 -
  11.134 -Xen was first described in a paper presented at SOSP in
  11.135 -2003\footnote{\tt
  11.136 -http://www.cl.cam.ac.uk/netos/papers/2003-xensosp.pdf}, and the first
  11.137 -public release (1.0) was made that October.  Since then, Xen has
  11.138 -significantly matured and is now used in production scenarios on
  11.139 -many sites.
  11.140 -
  11.141 -Xen 2.0 features greatly enhanced hardware support, configuration
  11.142 -flexibility, usability and a larger complement of supported operating
  11.143 -systems. This latest release takes Xen a step closer to becoming the 
  11.144 -definitive open source solution for virtualisation.
  11.145 -
  11.146 -\chapter{Installation}
  11.147 -
  11.148 -The Xen distribution includes three main components: Xen itself, ports
  11.149 -of Linux 2.4 and 2.6 and NetBSD to run on Xen, and the user-space
  11.150 -tools required to manage a Xen-based system.  This chapter describes
  11.151 -how to install the Xen 2.0 distribution from source.  Alternatively,
  11.152 -there may be pre-built packages available as part of your operating
  11.153 -system distribution.
  11.154 -
  11.155 -\section{Prerequisites}
  11.156 -\label{sec:prerequisites}
  11.157 -
  11.158 -The following is a full list of prerequisites.  Items marked `$\dag$'
  11.159 -are required by the \xend control tools, and hence required if you
  11.160 -want to run more than one virtual machine; items marked `$*$' are only
  11.161 -required if you wish to build from source.
  11.162 -\begin{itemize}
  11.163 -\item A working Linux distribution using the GRUB bootloader and
  11.164 -running on a P6-class (or newer) CPU.
  11.165 -\item [$\dag$] The \path{iproute2} package. 
  11.166 -\item [$\dag$] The Linux bridge-utils\footnote{Available from 
  11.167 -{\tt http://bridge.sourceforge.net}} (e.g., \path{/sbin/brctl})
  11.168 -\item [$\dag$] An installation of Twisted v1.3 or
  11.169 -above\footnote{Available from {\tt
  11.170 -http://www.twistedmatrix.com}}. There may be a binary package
  11.171 -available for your distribution; alternatively it can be installed by
  11.172 -running `{\sl make install-twisted}' in the root of the Xen source
  11.173 -tree.
  11.174 -\item [$*$] Build tools (gcc v3.2.x or v3.3.x, binutils, GNU make).
  11.175 -\item [$*$] Development installation of libcurl (e.g., libcurl-devel) 
  11.176 -\item [$*$] Development installation of zlib (e.g., zlib-dev).
  11.177 -\item [$*$] Development installation of Python v2.2 or later (e.g., python-dev).
  11.178 -\item [$*$] \LaTeX and transfig are required to build the documentation.
  11.179 -\end{itemize}
  11.180 -
  11.181 -Once you have satisfied the relevant prerequisites, you can 
  11.182 -now install either a binary or source distribution of Xen. 
  11.183 -
  11.184 -\section{Installing from Binary Tarball} 
  11.185 -
  11.186 -Pre-built tarballs are available for download from the Xen 
  11.187 -download page
  11.188 -\begin{quote} 
  11.189 -{\tt http://xen.sf.net}
  11.190 -\end{quote} 
  11.191 -
  11.192 -Once you've downloaded the tarball, simply unpack and install: 
  11.193 -\begin{verbatim}
  11.194 -# tar zxvf xen-2.0-install.tgz
  11.195 -# cd xen-2.0-install
  11.196 -# sh ./install.sh 
  11.197 -\end{verbatim} 
  11.198 -
  11.199 -Once you've installed the binaries you need to configure
  11.200 -your system as described in Section~\ref{s:configure}. 
  11.201 -
  11.202 -\section{Installing from Source} 
  11.203 -
  11.204 -This section describes how to obtain, build, and install 
  11.205 -Xen from source. 
  11.206 -
  11.207 -\subsection{Obtaining the Source} 
  11.208 -
  11.209 -The Xen source tree is available as either a compressed source tar
  11.210 -ball or as a clone of our master BitKeeper repository.
  11.211 -
  11.212 -\begin{description} 
  11.213 -\item[Obtaining the Source Tarball]\mbox{} \\  
  11.214 -Stable versions (and daily snapshots) of the Xen source tree are
  11.215 -available as compressed tarballs from the Xen download page
  11.216 -\begin{quote} 
  11.217 -{\tt http://xen.sf.net}
  11.218 -\end{quote} 
  11.219 -
  11.220 -\item[Using BitKeeper]\mbox{} \\  
  11.221 -If you wish to install Xen from a clone of our latest BitKeeper
  11.222 -repository then you will need to install the BitKeeper tools.
  11.223 -Download instructions for BitKeeper can be obtained by filling out the
  11.224 -form at:
  11.225 -
  11.226 -\begin{quote} 
  11.227 -{\tt http://www.bitmover.com/cgi-bin/download.cgi}
  11.228 -\end{quote}
  11.229 -The public master BK repository for the 2.0 release lives at: 
  11.230 -\begin{quote}
  11.231 -{\tt bk://xen.bkbits.net/xen-2.0.bk}  
  11.232 -\end{quote} 
  11.233 -You can use BitKeeper to
  11.234 -download it and keep it updated with the latest features and fixes.
  11.235 -
  11.236 -Change to the directory in which you want to put the source code, then
  11.237 -run:
  11.238 -\begin{verbatim}
  11.239 -# bk clone bk://xen.bkbits.net/xen-2.0.bk
  11.240 -\end{verbatim}
  11.241 -
  11.242 -Under your current directory, a new directory named \path{xen-2.0.bk}
  11.243 -has been created, which contains all the source code for Xen, the OS
  11.244 -ports, and the control tools. You can update your repository with the
  11.245 -latest changes at any time by running:
  11.246 -\begin{verbatim}
  11.247 -# cd xen-2.0.bk # to change into the local repository
  11.248 -# bk pull       # to update the repository
  11.249 -\end{verbatim}
  11.250 -\end{description} 
  11.251 -
  11.252 -%\section{The distribution}
  11.253 -%
  11.254 -%The Xen source code repository is structured as follows:
  11.255 -%
  11.256 -%\begin{description}
  11.257 -%\item[\path{tools/}] Xen node controller daemon (Xend), command line tools, 
  11.258 -%  control libraries
  11.259 -%\item[\path{xen/}] The Xen VMM.
  11.260 -%\item[\path{linux-*-xen-sparse/}] Xen support for Linux.
  11.261 -%\item[\path{linux-*-patches/}] Experimental patches for Linux.
  11.262 -%\item[\path{netbsd-*-xen-sparse/}] Xen support for NetBSD.
  11.263 -%\item[\path{docs/}] Various documentation files for users and developers.
  11.264 -%\item[\path{extras/}] Bonus extras.
  11.265 -%\end{description}
  11.266 -
  11.267 -\subsection{Building from Source} 
  11.268 -
  11.269 -The top-level Xen Makefile includes a target `world' that will do the
  11.270 -following:
  11.271 -
  11.272 -\begin{itemize}
  11.273 -\item Build Xen
  11.274 -\item Build the control tools, including \xend
  11.275 -\item Download (if necessary) and unpack the Linux 2.6 source code,
  11.276 -      and patch it for use with Xen
  11.277 -\item Build a Linux kernel to use in domain 0 and a smaller
  11.278 -      unprivileged kernel, which can optionally be used for
  11.279 -      unprivileged virtual machines.
  11.280 -\end{itemize}
  11.281 -
  11.282 -
  11.283 -After the build has completed you should have a top-level 
  11.284 -directory called \path{dist/} in which all resulting targets 
  11.285 -will be placed; of particular interest are the two kernels 
  11.286 -XenLinux kernel images, one with a `-xen0' extension
  11.287 -which contains hardware device drivers and drivers for Xen's virtual
  11.288 -devices, and one with a `-xenU' extension that just contains the
  11.289 -virtual ones. These are found in \path{dist/install/boot/} along
  11.290 -with the image for Xen itself and the configuration files used
  11.291 -during the build. 
  11.292  
  11.293 -The NetBSD port can be built using: 
  11.294 -\begin{quote}
  11.295 -\begin{verbatim}
  11.296 -# make netbsd20
  11.297 -\end{verbatim} 
  11.298 -\end{quote} 
  11.299 -NetBSD port is built using a snapshot of the netbsd-2-0 cvs branch.
  11.300 -The snapshot is downloaded as part of the build process, if it is not
  11.301 -yet present in the \path{NETBSD\_SRC\_PATH} search path.  The build
  11.302 -process also downloads a toolchain which includes all the tools
  11.303 -necessary to build the NetBSD kernel under Linux.
  11.304 -
  11.305 -To customize further the set of kernels built you need to edit
  11.306 -the top-level Makefile. Look for the line: 
  11.307 -
  11.308 -\begin{quote}
  11.309 -\begin{verbatim}
  11.310 -KERNELS ?= mk.linux-2.6-xen0 mk.linux-2.6-xenU
  11.311 -\end{verbatim} 
  11.312 -\end{quote} 
  11.313 -
  11.314 -You can edit this line to include any set of operating system kernels
  11.315 -which have configurations in the top-level \path{buildconfigs/}
  11.316 -directory, for example \path{mk.linux-2.4-xenU} to build a Linux 2.4
  11.317 -kernel containing only virtual device drivers.
  11.318 -
  11.319 -%% Inspect the Makefile if you want to see what goes on during a build.
  11.320 -%% Building Xen and the tools is straightforward, but XenLinux is more
  11.321 -%% complicated.  The makefile needs a `pristine' Linux kernel tree to which
  11.322 -%% it will then add the Xen architecture files.  You can tell the
  11.323 -%% makefile the location of the appropriate Linux compressed tar file by
  11.324 -%% setting the LINUX\_SRC environment variable, e.g. \\
  11.325 -%% \verb!# LINUX_SRC=/tmp/linux-2.6.11.tar.bz2 make world! \\ or by
  11.326 -%% placing the tar file somewhere in the search path of {\tt
  11.327 -%% LINUX\_SRC\_PATH} which defaults to `{\tt .:..}'.  If the makefile
  11.328 -%% can't find a suitable kernel tar file it attempts to download it from
  11.329 -%% kernel.org (this won't work if you're behind a firewall).
  11.330 -
  11.331 -%% After untaring the pristine kernel tree, the makefile uses the {\tt
  11.332 -%% mkbuildtree} script to add the Xen patches to the kernel. 
  11.333 -
  11.334 -
  11.335 -%% The procedure is similar to build the Linux 2.4 port: \\
  11.336 -%% \verb!# LINUX_SRC=/path/to/linux2.4/source make linux24!
  11.337 -
  11.338 -
  11.339 -%% \framebox{\parbox{5in}{
  11.340 -%% {\bf Distro specific:} \\
  11.341 -%% {\it Gentoo} --- if not using udev (most installations, currently), you'll need
  11.342 -%% to enable devfs and devfs mount at boot time in the xen0 config.
  11.343 -%% }}
  11.344 -
  11.345 -\subsection{Custom XenLinux Builds}
  11.346 -
  11.347 -% If you have an SMP machine you may wish to give the {\tt '-j4'}
  11.348 -% argument to make to get a parallel build.
  11.349 -
  11.350 -If you wish to build a customized XenLinux kernel (e.g. to support
  11.351 -additional devices or enable distribution-required features), you can
  11.352 -use the standard Linux configuration mechanisms, specifying that the
  11.353 -architecture being built for is \path{xen}, e.g:
  11.354 -\begin{quote}
  11.355 -\begin{verbatim} 
  11.356 -# cd linux-2.6.11-xen0 
  11.357 -# make ARCH=xen xconfig 
  11.358 -# cd ..
  11.359 -# make
  11.360 -\end{verbatim} 
  11.361 -\end{quote} 
  11.362 -
  11.363 -You can also copy an existing Linux configuration (\path{.config}) 
  11.364 -into \path{linux-2.6.11-xen0} and execute:  
  11.365 -\begin{quote}
  11.366 -\begin{verbatim} 
  11.367 -# make ARCH=xen oldconfig 
  11.368 -\end{verbatim} 
  11.369 -\end{quote} 
  11.370 -
  11.371 -You may be prompted with some Xen-specific options; we 
  11.372 -advise accepting the defaults for these options.
  11.373 -
  11.374 -Note that the only difference between the two types of Linux kernel
  11.375 -that are built is the configuration file used for each.  The "U"
  11.376 -suffixed (unprivileged) versions don't contain any of the physical
  11.377 -hardware device drivers, leading to a 30\% reduction in size; hence
  11.378 -you may prefer these for your non-privileged domains.  The `0'
  11.379 -suffixed privileged versions can be used to boot the system, as well
  11.380 -as in driver domains and unprivileged domains.
  11.381 -
  11.382 -
  11.383 -\subsection{Installing the Binaries}
  11.384 -
  11.385 -
  11.386 -The files produced by the build process are stored under the
  11.387 -\path{dist/install/} directory. To install them in their default
  11.388 -locations, do:
  11.389 -\begin{quote}
  11.390 -\begin{verbatim}
  11.391 -# make install
  11.392 -\end{verbatim} 
  11.393 -\end{quote}
  11.394 -
  11.395 -
  11.396 -Alternatively, users with special installation requirements may wish
  11.397 -to install them manually by copying the files to their appropriate
  11.398 -destinations.
  11.399 -
  11.400 -%% Files in \path{install/boot/} include:
  11.401 -%% \begin{itemize}
  11.402 -%% \item \path{install/boot/xen-2.0.gz} Link to the Xen 'kernel'
  11.403 -%% \item \path{install/boot/vmlinuz-2.6-xen0}  Link to domain 0 XenLinux kernel
  11.404 -%% \item \path{install/boot/vmlinuz-2.6-xenU}  Link to unprivileged XenLinux kernel
  11.405 -%% \end{itemize}
  11.406 -
  11.407 -The \path{dist/install/boot} directory will also contain the config files
  11.408 -used for building the XenLinux kernels, and also versions of Xen and
  11.409 -XenLinux kernels that contain debug symbols (\path{xen-syms-2.0.6} and
  11.410 -\path{vmlinux-syms-2.6.11.11-xen0}) which are essential for interpreting crash
  11.411 -dumps.  Retain these files as the developers may wish to see them if
  11.412 -you post on the mailing list.
  11.413 -
  11.414 -
  11.415 -
  11.416 -
  11.417 -
  11.418 -\section{Configuration}
  11.419 -\label{s:configure}
  11.420 -Once you have built and installed the Xen distribution, it is 
  11.421 -simple to prepare the machine for booting and running Xen. 
  11.422 -
  11.423 -\subsection{GRUB Configuration}
  11.424 -
  11.425 -An entry should be added to \path{grub.conf} (often found under
  11.426 -\path{/boot/} or \path{/boot/grub/}) to allow Xen / XenLinux to boot.
  11.427 -This file is sometimes called \path{menu.lst}, depending on your
  11.428 -distribution.  The entry should look something like the following:
  11.429 -
  11.430 -{\small
  11.431 -\begin{verbatim}
  11.432 -title Xen 2.0 / XenLinux 2.6
  11.433 -  kernel /boot/xen-2.0.gz dom0_mem=131072
  11.434 -  module /boot/vmlinuz-2.6-xen0 root=/dev/sda4 ro console=tty0
  11.435 -\end{verbatim}
  11.436 -}
  11.437 +%% Chapter Introduction moved to introduction.tex
  11.438 +\include{src/user/introduction}
  11.439  
  11.440 -The kernel line tells GRUB where to find Xen itself and what boot
  11.441 -parameters should be passed to it (in this case, setting domain 0's
  11.442 -memory allocation in kilobytes and the settings for the serial port). For more
  11.443 -details on the various Xen boot parameters see Section~\ref{s:xboot}. 
  11.444 -
  11.445 -The module line of the configuration describes the location of the
  11.446 -XenLinux kernel that Xen should start and the parameters that should
  11.447 -be passed to it (these are standard Linux parameters, identifying the
  11.448 -root device and specifying it be initially mounted read only and
  11.449 -instructing that console output be sent to the screen).  Some
  11.450 -distributions such as SuSE do not require the \path{ro} parameter.
  11.451 -
  11.452 -%% \framebox{\parbox{5in}{
  11.453 -%% {\bf Distro specific:} \\
  11.454 -%% {\it SuSE} --- Omit the {\tt ro} option from the XenLinux kernel
  11.455 -%% command line, since the partition won't be remounted rw during boot.
  11.456 -%% }}
  11.457 -
  11.458 -
  11.459 -If you want to use an initrd, just add another \path{module} line to
  11.460 -the configuration, as usual:
  11.461 -{\small
  11.462 -\begin{verbatim}
  11.463 -  module /boot/my_initrd.gz
  11.464 -\end{verbatim}
  11.465 -}
  11.466 -
  11.467 -As always when installing a new kernel, it is recommended that you do
  11.468 -not delete existing menu options from \path{menu.lst} --- you may want
  11.469 -to boot your old Linux kernel in future, particularly if you
  11.470 -have problems.
  11.471 -
  11.472 -
  11.473 -\subsection{Serial Console (optional)}
  11.474 -
  11.475 -%%   kernel /boot/xen-2.0.gz dom0_mem=131072 com1=115200,8n1
  11.476 -%%   module /boot/vmlinuz-2.6-xen0 root=/dev/sda4 ro 
  11.477 -
  11.478 -
  11.479 -In order to configure Xen serial console output, it is necessary to add 
  11.480 -an boot option to your GRUB config; e.g. replace the above kernel line 
  11.481 -with: 
  11.482 -\begin{quote}
  11.483 -{\small
  11.484 -\begin{verbatim}
  11.485 -   kernel /boot/xen.gz dom0_mem=131072 com1=115200,8n1
  11.486 -\end{verbatim}}
  11.487 -\end{quote}
  11.488 -
  11.489 -This configures Xen to output on COM1 at 115,200 baud, 8 data bits, 
  11.490 -1 stop bit and no parity. Modify these parameters for your set up. 
  11.491 -
  11.492 -One can also configure XenLinux to share the serial console; to 
  11.493 -achieve this append ``\path{console=ttyS0}'' to your 
  11.494 -module line. 
  11.495 -
  11.496 -
  11.497 -If you wish to be able to log in over the XenLinux serial console it
  11.498 -is necessary to add a line into \path{/etc/inittab}, just as per 
  11.499 -regular Linux. Simply add the line:
  11.500 -\begin{quote}
  11.501 -{\small 
  11.502 -{\tt c:2345:respawn:/sbin/mingetty ttyS0}
  11.503 -}
  11.504 -\end{quote} 
  11.505 -
  11.506 -and you should be able to log in. Note that to successfully log in 
  11.507 -as root over the serial line will require adding \path{ttyS0} to
  11.508 -\path{/etc/securetty} in most modern distributions. 
  11.509 -
  11.510 -\subsection{TLS Libraries}
  11.511 -
  11.512 -Users of the XenLinux 2.6 kernel should disable Thread Local Storage
  11.513 -(e.g.\ by doing a \path{mv /lib/tls /lib/tls.disabled}) before
  11.514 -attempting to run with a XenLinux kernel\footnote{If you boot without first
  11.515 -disabling TLS, you will get a warning message during the boot
  11.516 -process. In this case, simply perform the rename after the machine is
  11.517 -up and then run \texttt{/sbin/ldconfig} to make it take effect.}.  You can
  11.518 -always reenable it by restoring the directory to its original location
  11.519 -(i.e.\ \path{mv /lib/tls.disabled /lib/tls}).
  11.520 -
  11.521 -The reason for this is that the current TLS implementation uses
  11.522 -segmentation in a way that is not permissible under Xen.  If TLS is
  11.523 -not disabled, an emulation mode is used within Xen which reduces
  11.524 -performance substantially.
  11.525 -
  11.526 -We hope that this issue can be resolved by working with Linux
  11.527 -distribution vendors to implement a minor backward-compatible change
  11.528 -to the TLS library.
  11.529 -
  11.530 -\section{Booting Xen} 
  11.531 -
  11.532 -It should now be possible to restart the system and use Xen.  Reboot
  11.533 -as usual but choose the new Xen option when the Grub screen appears.
  11.534 -
  11.535 -What follows should look much like a conventional Linux boot.  The
  11.536 -first portion of the output comes from Xen itself, supplying low level
  11.537 -information about itself and the machine it is running on.  The
  11.538 -following portion of the output comes from XenLinux.
  11.539 -
  11.540 -You may see some errors during the XenLinux boot.  These are not
  11.541 -necessarily anything to worry about --- they may result from kernel
  11.542 -configuration differences between your XenLinux kernel and the one you
  11.543 -usually use.
  11.544 -
  11.545 -When the boot completes, you should be able to log into your system as
  11.546 -usual.  If you are unable to log in to your system running Xen, you
  11.547 -should still be able to reboot with your normal Linux kernel.
  11.548 -
  11.549 -
  11.550 -\chapter{Starting Additional Domains}
  11.551 -
  11.552 -The first step in creating a new domain is to prepare a root
  11.553 -filesystem for it to boot off.  Typically, this might be stored in a
  11.554 -normal partition, an LVM or other volume manager partition, a disk
  11.555 -file or on an NFS server.  A simple way to do this is simply to boot
  11.556 -from your standard OS install CD and install the distribution into
  11.557 -another partition on your hard drive.
  11.558 -
  11.559 -To start the \xend control daemon, type
  11.560 -\begin{quote}
  11.561 -\verb!# xend start!
  11.562 -\end{quote}
  11.563 -If you
  11.564 -wish the daemon to start automatically, see the instructions in
  11.565 -Section~\ref{s:xend}. Once the daemon is running, you can use the
  11.566 -\path{xm} tool to monitor and maintain the domains running on your
  11.567 -system. This chapter provides only a brief tutorial: we provide full
  11.568 -details of the \path{xm} tool in the next chapter. 
  11.569 -
  11.570 -%\section{From the web interface}
  11.571 -%
  11.572 -%Boot the Xen machine and start Xensv (see Chapter~\ref{cha:xensv} for
  11.573 -%more details) using the command: \\
  11.574 -%\verb_# xensv start_ \\
  11.575 -%This will also start Xend (see Chapter~\ref{cha:xend} for more information).
  11.576 -%
  11.577 -%The domain management interface will then be available at {\tt
  11.578 -%http://your\_machine:8080/}.  This provides a user friendly wizard for
  11.579 -%starting domains and functions for managing running domains.
  11.580 -%
  11.581 -%\section{From the command line}
  11.582 -
  11.583 -
  11.584 -\section{Creating a Domain Configuration File} 
  11.585 +%% Chapter Installation moved to installation.tex
  11.586 +\include{src/user/installation}
  11.587  
  11.588 -Before you can start an additional domain, you must create a
  11.589 -configuration file. We provide two example files which you 
  11.590 -can use as a starting point: 
  11.591 -\begin{itemize} 
  11.592 -  \item \path{/etc/xen/xmexample1} is a simple template configuration file
  11.593 -    for describing a single VM.
  11.594 -
  11.595 -  \item \path{/etc/xen/xmexample2} file is a template description that
  11.596 -    is intended to be reused for multiple virtual machines.  Setting
  11.597 -    the value of the \path{vmid} variable on the \path{xm} command line
  11.598 -    fills in parts of this template.
  11.599 -\end{itemize} 
  11.600 -
  11.601 -Copy one of these files and edit it as appropriate.
  11.602 -Typical values you may wish to edit include: 
  11.603 -
  11.604 -\begin{quote}
  11.605 -\begin{description}
  11.606 -\item[kernel] Set this to the path of the kernel you compiled for use
  11.607 -              with Xen (e.g.\  \path{kernel = '/boot/vmlinuz-2.6-xenU'})
  11.608 -\item[memory] Set this to the size of the domain's memory in
  11.609 -megabytes (e.g.\ \path{memory = 64})
  11.610 -\item[disk] Set the first entry in this list to calculate the offset
  11.611 -of the domain's root partition, based on the domain ID.  Set the
  11.612 -second to the location of \path{/usr} if you are sharing it between
  11.613 -domains (e.g.\ \path{disk = ['phy:your\_hard\_drive\%d,sda1,w' \%
  11.614 -(base\_partition\_number + vmid), 'phy:your\_usr\_partition,sda6,r' ]}
  11.615 -\item[dhcp] Uncomment the dhcp variable, so that the domain will
  11.616 -receive its IP address from a DHCP server (e.g.\ \path{dhcp='dhcp'})
  11.617 -\end{description}
  11.618 -\end{quote}
  11.619 -
  11.620 -You may also want to edit the {\bf vif} variable in order to choose
  11.621 -the MAC address of the virtual ethernet interface yourself.  For
  11.622 -example: 
  11.623 -\begin{quote}
  11.624 -\verb_vif = ['mac=00:06:AA:F6:BB:B3']_
  11.625 -\end{quote}
  11.626 -If you do not set this variable, \xend will automatically generate a
  11.627 -random MAC address from an unused range.
  11.628 -
  11.629 -
  11.630 -\section{Booting the Domain}
  11.631 -
  11.632 -The \path{xm} tool provides a variety of commands for managing domains.
  11.633 -Use the \path{create} command to start new domains. Assuming you've 
  11.634 -created a configuration file \path{myvmconf} based around
  11.635 -\path{/etc/xen/xmexample2}, to start a domain with virtual 
  11.636 -machine ID~1 you should type: 
  11.637 -
  11.638 -\begin{quote}
  11.639 -\begin{verbatim}
  11.640 -# xm create -c myvmconf vmid=1
  11.641 -\end{verbatim}
  11.642 -\end{quote}
  11.643 -
  11.644 -
  11.645 -The \path{-c} switch causes \path{xm} to turn into the domain's
  11.646 -console after creation.  The \path{vmid=1} sets the \path{vmid}
  11.647 -variable used in the \path{myvmconf} file. 
  11.648 -
  11.649 -
  11.650 -You should see the console boot messages from the new domain 
  11.651 -appearing in the terminal in which you typed the command, 
  11.652 -culminating in a login prompt. 
  11.653 -
  11.654 -
  11.655 -\section{Example: ttylinux}
  11.656 -
  11.657 -Ttylinux is a very small Linux distribution, designed to require very
  11.658 -few resources.  We will use it as a concrete example of how to start a
  11.659 -Xen domain.  Most users will probably want to install a full-featured
  11.660 -distribution once they have mastered the basics\footnote{ttylinux is
  11.661 -maintained by Pascal Schmidt. You can download source packages from
  11.662 -the distribution's home page: {\tt http://www.minimalinux.org/ttylinux/}}.
  11.663 -
  11.664 -\begin{enumerate}
  11.665 -\item Download and extract the ttylinux disk image from the Files
  11.666 -section of the project's SourceForge site (see 
  11.667 -\path{http://sf.net/projects/xen/}).
  11.668 -\item Create a configuration file like the following:
  11.669 -\begin{verbatim}
  11.670 -kernel = "/boot/vmlinuz-2.6-xenU"
  11.671 -memory = 64
  11.672 -name = "ttylinux"
  11.673 -nics = 1
  11.674 -ip = "1.2.3.4"
  11.675 -disk = ['file:/path/to/ttylinux/rootfs,sda1,w']
  11.676 -root = "/dev/sda1 ro"
  11.677 -\end{verbatim}
  11.678 -\item Now start the domain and connect to its console:
  11.679 -\begin{verbatim}
  11.680 -xm create configfile -c
  11.681 -\end{verbatim}
  11.682 -\item Login as root, password root.
  11.683 -\end{enumerate}
  11.684 -
  11.685 -
  11.686 -\section{Starting / Stopping Domains Automatically}
  11.687 -
  11.688 -It is possible to have certain domains start automatically at boot
  11.689 -time and to have dom0 wait for all running domains to shutdown before
  11.690 -it shuts down the system.
  11.691 -
  11.692 -To specify a domain is to start at boot-time, place its
  11.693 -configuration file (or a link to it) under \path{/etc/xen/auto/}.
  11.694 -
  11.695 -A Sys-V style init script for RedHat and LSB-compliant systems is
  11.696 -provided and will be automatically copied to \path{/etc/init.d/}
  11.697 -during install.  You can then enable it in the appropriate way for
  11.698 -your distribution.
  11.699 -
  11.700 -For instance, on RedHat:
  11.701 -
  11.702 -\begin{quote}
  11.703 -\verb_# chkconfig --add xendomains_
  11.704 -\end{quote}
  11.705 -
  11.706 -By default, this will start the boot-time domains in runlevels 3, 4
  11.707 -and 5.
  11.708 -
  11.709 -You can also use the \path{service} command to run this script
  11.710 -manually, e.g:
  11.711 -
  11.712 -\begin{quote}
  11.713 -\verb_# service xendomains start_
  11.714 -
  11.715 -Starts all the domains with config files under /etc/xen/auto/.
  11.716 -\end{quote}
  11.717 -
  11.718 -
  11.719 -\begin{quote}
  11.720 -\verb_# service xendomains stop_
  11.721 -
  11.722 -Shuts down ALL running Xen domains.
  11.723 -\end{quote}
  11.724 -
  11.725 -\chapter{Domain Management Tools}
  11.726 -
  11.727 -The previous chapter described a simple example of how to configure
  11.728 -and start a domain.  This chapter summarises the tools available to
  11.729 -manage running domains.
  11.730 -
  11.731 -\section{Command-line Management}
  11.732 -
  11.733 -Command line management tasks are also performed using the \path{xm}
  11.734 -tool.  For online help for the commands available, type:
  11.735 -\begin{quote}
  11.736 -\verb_# xm help_
  11.737 -\end{quote}
  11.738 -
  11.739 -You can also type \path{xm help $<$command$>$} for more information 
  11.740 -on a given command. 
  11.741 -
  11.742 -\subsection{Basic Management Commands}
  11.743 -
  11.744 -The most important \path{xm} commands are: 
  11.745 -\begin{quote}
  11.746 -\verb_# xm list_: Lists all domains running.\\
  11.747 -\verb_# xm consoles_ : Gives information about the domain consoles.\\
  11.748 -\verb_# xm console_: Opens a console to a domain (e.g.\
  11.749 -  \verb_# xm console myVM_
  11.750 -\end{quote}
  11.751 -
  11.752 -\subsection{\tt xm list}
  11.753 -
  11.754 -The output of \path{xm list} is in rows of the following format:
  11.755 -\begin{center}
  11.756 -{\tt name domid memory cpu state cputime console}
  11.757 -\end{center}
  11.758 -
  11.759 -\begin{quote}
  11.760 -\begin{description}
  11.761 -\item[name]  The descriptive name of the virtual machine.
  11.762 -\item[domid] The number of the domain ID this virtual machine is running in.
  11.763 -\item[memory] Memory size in megabytes.
  11.764 -\item[cpu]   The CPU this domain is running on.
  11.765 -\item[state] Domain state consists of 5 fields:
  11.766 -  \begin{description}
  11.767 -  \item[r] running
  11.768 -  \item[b] blocked
  11.769 -  \item[p] paused
  11.770 -  \item[s] shutdown
  11.771 -  \item[c] crashed
  11.772 -  \end{description}
  11.773 -\item[cputime] How much CPU time (in seconds) the domain has used so far.
  11.774 -\item[console] TCP port accepting connections to the domain's console.
  11.775 -\end{description}
  11.776 -\end{quote}
  11.777 -
  11.778 -The \path{xm list} command also supports a long output format when the
  11.779 -\path{-l} switch is used.  This outputs the fulls details of the
  11.780 -running domains in \xend's SXP configuration format.
  11.781 -
  11.782 -For example, suppose the system is running the ttylinux domain as
  11.783 -described earlier.  The list command should produce output somewhat
  11.784 -like the following:
  11.785 -\begin{verbatim}
  11.786 -# xm list
  11.787 -Name              Id  Mem(MB)  CPU  State  Time(s)  Console
  11.788 -Domain-0           0      251    0  r----    172.2        
  11.789 -ttylinux           5       63    0  -b---      3.0    9605
  11.790 -\end{verbatim}
  11.791 -
  11.792 -Here we can see the details for the ttylinux domain, as well as for
  11.793 -domain 0 (which, of course, is always running).  Note that the console
  11.794 -port for the ttylinux domain is 9605.  This can be connected to by TCP
  11.795 -using a terminal program (e.g. \path{telnet} or, better, 
  11.796 -\path{xencons}).  The simplest way to connect is to use the \path{xm console}
  11.797 -command, specifying the domain name or ID.  To connect to the console
  11.798 -of the ttylinux domain, we could use any of the following: 
  11.799 -\begin{verbatim}
  11.800 -# xm console ttylinux
  11.801 -# xm console 5
  11.802 -# xencons localhost 9605
  11.803 -\end{verbatim}
  11.804 -
  11.805 -\section{Domain Save and Restore}
  11.806 -
  11.807 -The administrator of a Xen system may suspend a virtual machine's
  11.808 -current state into a disk file in domain 0, allowing it to be resumed
  11.809 -at a later time.
  11.810 -
  11.811 -The ttylinux domain described earlier can be suspended to disk using
  11.812 -the command:
  11.813 -\begin{verbatim}
  11.814 -# xm save ttylinux ttylinux.xen
  11.815 -\end{verbatim}
  11.816 -
  11.817 -This will stop the domain named `ttylinux' and save its current state
  11.818 -into a file called \path{ttylinux.xen}.
  11.819 -
  11.820 -To resume execution of this domain, use the \path{xm restore} command:
  11.821 -\begin{verbatim}
  11.822 -# xm restore ttylinux.xen
  11.823 -\end{verbatim}
  11.824 -
  11.825 -This will restore the state of the domain and restart it.  The domain
  11.826 -will carry on as before and the console may be reconnected using the
  11.827 -\path{xm console} command, as above.
  11.828 -
  11.829 -\section{Live Migration}
  11.830 -
  11.831 -Live migration is used to transfer a domain between physical hosts
  11.832 -whilst that domain continues to perform its usual activities --- from
  11.833 -the user's perspective, the migration should be imperceptible.
  11.834 -
  11.835 -To perform a live migration, both hosts must be running Xen / \xend and
  11.836 -the destination host must have sufficient resources (e.g. memory
  11.837 -capacity) to accommodate the domain after the move. Furthermore we
  11.838 -currently require both source and destination machines to be on the 
  11.839 -same L2 subnet. 
  11.840 -
  11.841 -Currently, there is no support for providing automatic remote access
  11.842 -to filesystems stored on local disk when a domain is migrated.
  11.843 -Administrators should choose an appropriate storage solution
  11.844 -(i.e. SAN, NAS, etc.) to ensure that domain filesystems are also
  11.845 -available on their destination node. GNBD is a good method for
  11.846 -exporting a volume from one machine to another. iSCSI can do a similar
  11.847 -job, but is more complex to set up.
  11.848 -
  11.849 -When a domain migrates, it's MAC and IP address move with it, thus it
  11.850 -is only possible to migrate VMs within the same layer-2 network and IP
  11.851 -subnet. If the destination node is on a different subnet, the
  11.852 -administrator would need to manually configure a suitable etherip or
  11.853 -IP tunnel in the domain 0 of the remote node. 
  11.854 -
  11.855 -A domain may be migrated using the \path{xm migrate} command.  To
  11.856 -live migrate a domain to another machine, we would use
  11.857 -the command:
  11.858 -
  11.859 -\begin{verbatim}
  11.860 -# xm migrate --live mydomain destination.ournetwork.com
  11.861 -\end{verbatim}
  11.862 -
  11.863 -Without the \path{--live} flag, \xend simply stops the domain and
  11.864 -copies the memory image over to the new node and restarts it. Since
  11.865 -domains can have large allocations this can be quite time consuming,
  11.866 -even on a Gigabit network. With the \path{--live} flag \xend attempts
  11.867 -to keep the domain running while the migration is in progress,
  11.868 -resulting in typical `downtimes' of just 60--300ms.
  11.869 -
  11.870 -For now it will be necessary to reconnect to the domain's console on
  11.871 -the new machine using the \path{xm console} command.  If a migrated
  11.872 -domain has any open network connections then they will be preserved,
  11.873 -so SSH connections do not have this limitation.
  11.874 -
  11.875 -\section{Managing Domain Memory}
  11.876 -
  11.877 -XenLinux domains have the ability to relinquish / reclaim machine
  11.878 -memory at the request of the administrator or the user of the domain.
  11.879 +%% Chapter Starting Additional Domains  moved to start_addl_dom.tex
  11.880 +\include{src/user/start_addl_dom}
  11.881  
  11.882 -\subsection{Setting memory footprints from dom0}
  11.883 -
  11.884 -The machine administrator can request that a domain alter its memory
  11.885 -footprint using the \path{xm set-mem} command.  For instance, we can
  11.886 -request that our example ttylinux domain reduce its memory footprint
  11.887 -to 32 megabytes.
  11.888 -
  11.889 -\begin{verbatim}
  11.890 -# xm set-mem ttylinux 32
  11.891 -\end{verbatim}
  11.892 -
  11.893 -We can now see the result of this in the output of \path{xm list}:
  11.894 -
  11.895 -\begin{verbatim}
  11.896 -# xm list
  11.897 -Name              Id  Mem(MB)  CPU  State  Time(s)  Console
  11.898 -Domain-0           0      251    0  r----    172.2        
  11.899 -ttylinux           5       31    0  -b---      4.3    9605
  11.900 -\end{verbatim}
  11.901 -
  11.902 -The domain has responded to the request by returning memory to Xen. We
  11.903 -can restore the domain to its original size using the command line:
  11.904 -
  11.905 -\begin{verbatim}
  11.906 -# xm set-mem ttylinux 64
  11.907 -\end{verbatim}
  11.908 -
  11.909 -\subsection{Setting memory footprints from within a domain}
  11.910 -
  11.911 -The virtual file \path{/proc/xen/balloon} allows the owner of a
  11.912 -domain to adjust their own memory footprint.  Reading the file
  11.913 -(e.g. \path{cat /proc/xen/balloon}) prints out the current
  11.914 -memory footprint of the domain.  Writing the file
  11.915 -(e.g. \path{echo new\_target > /proc/xen/balloon}) requests
  11.916 -that the kernel adjust the domain's memory footprint to a new value.
  11.917 -
  11.918 -\subsection{Setting memory limits}
  11.919 -
  11.920 -Xen associates a memory size limit with each domain.  By default, this
  11.921 -is the amount of memory the domain is originally started with,
  11.922 -preventing the domain from ever growing beyond this size.  To permit a
  11.923 -domain to grow beyond its original allocation or to prevent a domain
  11.924 -you've shrunk from reclaiming the memory it relinquished, use the 
  11.925 -\path{xm maxmem} command.
  11.926 -
  11.927 -\chapter{Domain Filesystem Storage}
  11.928 -
  11.929 -It is possible to directly export any Linux block device in dom0 to
  11.930 -another domain, or to export filesystems / devices to virtual machines
  11.931 -using standard network protocols (e.g. NBD, iSCSI, NFS, etc).  This
  11.932 -chapter covers some of the possibilities.
  11.933 -
  11.934 -
  11.935 -\section{Exporting Physical Devices as VBDs} 
  11.936 -\label{s:exporting-physical-devices-as-vbds}
  11.937 -
  11.938 -One of the simplest configurations is to directly export 
  11.939 -individual partitions from domain 0 to other domains. To 
  11.940 -achieve this use the \path{phy:} specifier in your domain 
  11.941 -configuration file. For example a line like
  11.942 -\begin{quote}
  11.943 -\verb_disk = ['phy:hda3,sda1,w']_
  11.944 -\end{quote}
  11.945 -specifies that the partition \path{/dev/hda3} in domain 0 
  11.946 -should be exported read-write to the new domain as \path{/dev/sda1}; 
  11.947 -one could equally well export it as \path{/dev/hda} or 
  11.948 -\path{/dev/sdb5} should one wish. 
  11.949 -
  11.950 -In addition to local disks and partitions, it is possible to export
  11.951 -any device that Linux considers to be ``a disk'' in the same manner.
  11.952 -For example, if you have iSCSI disks or GNBD volumes imported into
  11.953 -domain 0 you can export these to other domains using the \path{phy:}
  11.954 -disk syntax. E.g.:
  11.955 -\begin{quote}
  11.956 -\verb_disk = ['phy:vg/lvm1,sda2,w']_
  11.957 -\end{quote}
  11.958 -
  11.959 -
  11.960 -
  11.961 -\begin{center}
  11.962 -\framebox{\bf Warning: Block device sharing}
  11.963 -\end{center}
  11.964 -\begin{quote}
  11.965 -Block devices should typically only be shared between domains in a
  11.966 -read-only fashion otherwise the Linux kernel's file systems will get
  11.967 -very confused as the file system structure may change underneath them
  11.968 -(having the same ext3 partition mounted rw twice is a sure fire way to
  11.969 -cause irreparable damage)!  \Xend will attempt to prevent you from
  11.970 -doing this by checking that the device is not mounted read-write in
  11.971 -domain 0, and hasn't already been exported read-write to another
  11.972 -domain.
  11.973 -If you want read-write sharing, export the directory to other domains
  11.974 -via NFS from domain0 (or use a cluster file system such as GFS or
  11.975 -ocfs2).
  11.976 -
  11.977 -\end{quote}
  11.978 -
  11.979 -
  11.980 -\section{Using File-backed VBDs}
  11.981 -
  11.982 -It is also possible to use a file in Domain 0 as the primary storage
  11.983 -for a virtual machine.  As well as being convenient, this also has the
  11.984 -advantage that the virtual block device will be {\em sparse} --- space
  11.985 -will only really be allocated as parts of the file are used.  So if a
  11.986 -virtual machine uses only half of its disk space then the file really
  11.987 -takes up half of the size allocated.
  11.988 -
  11.989 -For example, to create a 2GB sparse file-backed virtual block device
  11.990 -(actually only consumes 1KB of disk):
  11.991 -\begin{quote}
  11.992 -\verb_# dd if=/dev/zero of=vm1disk bs=1k seek=2048k count=1_
  11.993 -\end{quote}
  11.994 -
  11.995 -Make a file system in the disk file: 
  11.996 -\begin{quote}
  11.997 -\verb_# mkfs -t ext3 vm1disk_
  11.998 -\end{quote}
  11.999 -
 11.1000 -(when the tool asks for confirmation, answer `y')
 11.1001 -
 11.1002 -Populate the file system e.g. by copying from the current root:
 11.1003 -\begin{quote}
 11.1004 -\begin{verbatim}
 11.1005 -# mount -o loop vm1disk /mnt
 11.1006 -# cp -ax /{root,dev,var,etc,usr,bin,sbin,lib} /mnt
 11.1007 -# mkdir /mnt/{proc,sys,home,tmp}
 11.1008 -\end{verbatim}
 11.1009 -\end{quote}
 11.1010 -
 11.1011 -Tailor the file system by editing \path{/etc/fstab},
 11.1012 -\path{/etc/hostname}, etc (don't forget to edit the files in the
 11.1013 -mounted file system, instead of your domain 0 filesystem, e.g. you
 11.1014 -would edit \path{/mnt/etc/fstab} instead of \path{/etc/fstab} ).  For
 11.1015 -this example put \path{/dev/sda1} to root in fstab.
 11.1016 -
 11.1017 -Now unmount (this is important!):
 11.1018 -\begin{quote}
 11.1019 -\verb_# umount /mnt_
 11.1020 -\end{quote}
 11.1021 -
 11.1022 -In the configuration file set:
 11.1023 -\begin{quote}
 11.1024 -\verb_disk = ['file:/full/path/to/vm1disk,sda1,w']_
 11.1025 -\end{quote}
 11.1026 +%% Chapter Domain Management Tools moved to domain_mgmt.tex
 11.1027 +\include{src/user/domain_mgmt}
 11.1028  
 11.1029 -As the virtual machine writes to its `disk', the sparse file will be
 11.1030 -filled in and consume more space up to the original 2GB.
 11.1031 -
 11.1032 -{\bf Note that file-backed VBDs may not be appropriate for backing
 11.1033 -I/O-intensive domains.}  File-backed VBDs are known to experience
 11.1034 -substantial slowdowns under heavy I/O workloads, due to the I/O handling
 11.1035 -by the loopback block device used to support file-backed VBDs in dom0.
 11.1036 -Better I/O performance can be achieved by using either LVM-backed VBDs
 11.1037 -(Section~\ref{s:using-lvm-backed-vbds}) or physical devices as VBDs
 11.1038 -(Section~\ref{s:exporting-physical-devices-as-vbds}).
 11.1039 -
 11.1040 -Linux supports a maximum of eight file-backed VBDs across all domains by
 11.1041 -default.  This limit can be statically increased by using the {\em
 11.1042 -max\_loop} module parameter if CONFIG\_BLK\_DEV\_LOOP is compiled as a
 11.1043 -module in the dom0 kernel, or by using the {\em max\_loop=n} boot option
 11.1044 -if CONFIG\_BLK\_DEV\_LOOP is compiled directly into the dom0 kernel.
 11.1045 -
 11.1046 -
 11.1047 -\section{Using LVM-backed VBDs}
 11.1048 -\label{s:using-lvm-backed-vbds}
 11.1049 -
 11.1050 -A particularly appealing solution is to use LVM volumes 
 11.1051 -as backing for domain file-systems since this allows dynamic
 11.1052 -growing/shrinking of volumes as well as snapshot and other 
 11.1053 -features. 
 11.1054 -
 11.1055 -To initialise a partition to support LVM volumes:
 11.1056 -\begin{quote}
 11.1057 -\begin{verbatim} 
 11.1058 -# pvcreate /dev/sda10           
 11.1059 -\end{verbatim} 
 11.1060 -\end{quote}
 11.1061 -
 11.1062 -Create a volume group named `vg' on the physical partition:
 11.1063 -\begin{quote}
 11.1064 -\begin{verbatim} 
 11.1065 -# vgcreate vg /dev/sda10
 11.1066 -\end{verbatim} 
 11.1067 -\end{quote}
 11.1068 -
 11.1069 -Create a logical volume of size 4GB named `myvmdisk1':
 11.1070 -\begin{quote}
 11.1071 -\begin{verbatim} 
 11.1072 -# lvcreate -L4096M -n myvmdisk1 vg
 11.1073 -\end{verbatim} 
 11.1074 -\end{quote}
 11.1075 -
 11.1076 -You should now see that you have a \path{/dev/vg/myvmdisk1}
 11.1077 -Make a filesystem, mount it and populate it, e.g.:
 11.1078 -\begin{quote}
 11.1079 -\begin{verbatim} 
 11.1080 -# mkfs -t ext3 /dev/vg/myvmdisk1
 11.1081 -# mount /dev/vg/myvmdisk1 /mnt
 11.1082 -# cp -ax / /mnt
 11.1083 -# umount /mnt
 11.1084 -\end{verbatim} 
 11.1085 -\end{quote}
 11.1086 -
 11.1087 -Now configure your VM with the following disk configuration:
 11.1088 -\begin{quote}
 11.1089 -\begin{verbatim} 
 11.1090 - disk = [ 'phy:vg/myvmdisk1,sda1,w' ]
 11.1091 -\end{verbatim} 
 11.1092 -\end{quote}
 11.1093 -
 11.1094 -LVM enables you to grow the size of logical volumes, but you'll need
 11.1095 -to resize the corresponding file system to make use of the new
 11.1096 -space. Some file systems (e.g. ext3) now support on-line resize.  See
 11.1097 -the LVM manuals for more details.
 11.1098 +%% Chapter Domain Filesystem Storage moved to domain_filesystem.tex
 11.1099 +\include{src/user/domain_filesystem}
 11.1100  
 11.1101 -You can also use LVM for creating copy-on-write clones of LVM
 11.1102 -volumes (known as writable persistent snapshots in LVM
 11.1103 -terminology). This facility is new in Linux 2.6.8, so isn't as
 11.1104 -stable as one might hope. In particular, using lots of CoW LVM
 11.1105 -disks consumes a lot of dom0 memory, and error conditions such as
 11.1106 -running out of disk space are not handled well. Hopefully this
 11.1107 -will improve in future.
 11.1108 -
 11.1109 -To create two copy-on-write clone of the above file system you
 11.1110 -would use the following commands:
 11.1111 -
 11.1112 -\begin{quote}
 11.1113 -\begin{verbatim} 
 11.1114 -# lvcreate -s -L1024M -n myclonedisk1 /dev/vg/myvmdisk1
 11.1115 -# lvcreate -s -L1024M -n myclonedisk2 /dev/vg/myvmdisk1
 11.1116 -\end{verbatim} 
 11.1117 -\end{quote}
 11.1118 -
 11.1119 -Each of these can grow to have 1GB of differences from the master
 11.1120 -volume. You can grow the amount of space for storing the
 11.1121 -differences using the lvextend command, e.g.:
 11.1122 -\begin{quote}
 11.1123 -\begin{verbatim} 
 11.1124 -# lvextend +100M /dev/vg/myclonedisk1
 11.1125 -\end{verbatim} 
 11.1126 -\end{quote}
 11.1127 -
 11.1128 -Don't let the `differences volume' ever fill up otherwise LVM gets
 11.1129 -rather confused. It may be possible to automate the growing
 11.1130 -process by using \path{dmsetup wait} to spot the volume getting full
 11.1131 -and then issue an \path{lvextend}.
 11.1132 -
 11.1133 -In principle, it is possible to continue writing to the volume
 11.1134 -that has been cloned (the changes will not be visible to the
 11.1135 -clones), but we wouldn't recommend this: have the cloned volume
 11.1136 -as a `pristine' file system install that isn't mounted directly
 11.1137 -by any of the virtual machines.
 11.1138 -
 11.1139 -
 11.1140 -\section{Using NFS Root}
 11.1141 -
 11.1142 -First, populate a root filesystem in a directory on the server
 11.1143 -machine. This can be on a distinct physical machine, or simply 
 11.1144 -run within a virtual machine on the same node.
 11.1145 -
 11.1146 -Now configure the NFS server to export this filesystem over the
 11.1147 -network by adding a line to \path{/etc/exports}, for instance:
 11.1148 -
 11.1149 -\begin{quote}
 11.1150 -\begin{small}
 11.1151 -\begin{verbatim}
 11.1152 -/export/vm1root      1.2.3.4/24 (rw,sync,no_root_squash)
 11.1153 -\end{verbatim}
 11.1154 -\end{small}
 11.1155 -\end{quote}
 11.1156 -
 11.1157 -Finally, configure the domain to use NFS root.  In addition to the
 11.1158 -normal variables, you should make sure to set the following values in
 11.1159 -the domain's configuration file:
 11.1160 -
 11.1161 -\begin{quote}
 11.1162 -\begin{small}
 11.1163 -\begin{verbatim}
 11.1164 -root       = '/dev/nfs'
 11.1165 -nfs_server = '2.3.4.5'       # substitute IP address of server 
 11.1166 -nfs_root   = '/path/to/root' # path to root FS on the server
 11.1167 -\end{verbatim}
 11.1168 -\end{small}
 11.1169 -\end{quote}
 11.1170 -
 11.1171 -The domain will need network access at boot time, so either statically
 11.1172 -configure an IP address (Using the config variables \path{ip}, 
 11.1173 -\path{netmask}, \path{gateway}, \path{hostname}) or enable DHCP (
 11.1174 -\path{dhcp='dhcp'}).
 11.1175 -
 11.1176 -Note that the Linux NFS root implementation is known to have stability
 11.1177 -problems under high load (this is not a Xen-specific problem), so this
 11.1178 -configuration may not be appropriate for critical servers.
 11.1179  
 11.1180  
 11.1181  \part{User Reference Documentation}
 11.1182  
 11.1183 -\chapter{Control Software} 
 11.1184 -
 11.1185 -The Xen control software includes the \xend node control daemon (which 
 11.1186 -must be running), the xm command line tools, and the prototype 
 11.1187 -xensv web interface. 
 11.1188 -
 11.1189 -\section{\Xend (node control daemon)}
 11.1190 -\label{s:xend}
 11.1191 -
 11.1192 -The Xen Daemon (\Xend) performs system management functions related to
 11.1193 -virtual machines.  It forms a central point of control for a machine
 11.1194 -and can be controlled using an HTTP-based protocol.  \Xend must be
 11.1195 -running in order to start and manage virtual machines.
 11.1196 -
 11.1197 -\Xend must be run as root because it needs access to privileged system
 11.1198 -management functions.  A small set of commands may be issued on the
 11.1199 -\xend command line:
 11.1200 -
 11.1201 -\begin{tabular}{ll}
 11.1202 -\verb!# xend start! & start \xend, if not already running \\
 11.1203 -\verb!# xend stop!  & stop \xend if already running       \\
 11.1204 -\verb!# xend restart! & restart \xend if running, otherwise start it \\
 11.1205 -% \verb!# xend trace_start! & start \xend, with very detailed debug logging \\
 11.1206 -\verb!# xend status! & indicates \xend status by its return code
 11.1207 -\end{tabular}
 11.1208 -
 11.1209 -A SysV init script called {\tt xend} is provided to start \xend at boot
 11.1210 -time.  {\tt make install} installs this script in {\path{/etc/init.d}.
 11.1211 -To enable it, you have to make symbolic links in the appropriate
 11.1212 -runlevel directories or use the {\tt chkconfig} tool, where available.
 11.1213 -
 11.1214 -Once \xend is running, more sophisticated administration can be done
 11.1215 -using the xm tool (see Section~\ref{s:xm}) and the experimental
 11.1216 -Xensv web interface (see Section~\ref{s:xensv}).
 11.1217 -
 11.1218 -As \xend runs, events will be logged to \path{/var/log/xend.log} and, 
 11.1219 -if the migration assistant daemon (\path{xfrd}) has been started, 
 11.1220 -\path{/var/log/xfrd.log}. These may be of use for troubleshooting
 11.1221 -problems.
 11.1222 -
 11.1223 -\section{Xm (command line interface)}
 11.1224 -\label{s:xm}
 11.1225 -
 11.1226 -The xm tool is the primary tool for managing Xen from the console.
 11.1227 -The general format of an xm command line is:
 11.1228 -
 11.1229 -\begin{verbatim}
 11.1230 -# xm command [switches] [arguments] [variables]
 11.1231 -\end{verbatim}
 11.1232 -
 11.1233 -The available {\em switches} and {\em arguments} are dependent on the
 11.1234 -{\em command} chosen.  The {\em variables} may be set using
 11.1235 -declarations of the form {\tt variable=value} and command line
 11.1236 -declarations override any of the values in the configuration file
 11.1237 -being used, including the standard variables described above and any
 11.1238 -custom variables (for instance, the \path{xmdefconfig} file uses a
 11.1239 -{\tt vmid} variable).
 11.1240 -
 11.1241 -The available commands are as follows:
 11.1242 -
 11.1243 -\begin{description}
 11.1244 -\item[set-mem] Request a domain to adjust its memory footprint.
 11.1245 -\item[create] Create a new domain.
 11.1246 -\item[destroy] Kill a domain immediately.
 11.1247 -\item[list] List running domains.
 11.1248 -\item[shutdown] Ask a domain to shutdown.
 11.1249 -\item[dmesg] Fetch the Xen (not Linux!) boot output.
 11.1250 -\item[consoles] Lists the available consoles.
 11.1251 -\item[console] Connect to the console for a domain.
 11.1252 -\item[help] Get help on xm commands.
 11.1253 -\item[save] Suspend a domain to disk.
 11.1254 -\item[restore] Restore a domain from disk.
 11.1255 -\item[pause] Pause a domain's execution.
 11.1256 -\item[unpause] Unpause a domain.
 11.1257 -\item[pincpu] Pin a domain to a CPU.
 11.1258 -\item[bvt] Set BVT scheduler parameters for a domain.
 11.1259 -\item[bvt\_ctxallow] Set the BVT context switching allowance for the system.
 11.1260 -\item[atropos] Set the atropos parameters for a domain.
 11.1261 -\item[rrobin] Set the round robin time slice for the system.
 11.1262 -\item[info] Get information about the Xen host.
 11.1263 -\item[call] Call a \xend HTTP API function directly.
 11.1264 -\end{description}
 11.1265 -
 11.1266 -For a detailed overview of switches, arguments and variables to each command
 11.1267 -try
 11.1268 -\begin{quote}
 11.1269 -\begin{verbatim}
 11.1270 -# xm help command
 11.1271 -\end{verbatim}
 11.1272 -\end{quote}
 11.1273 -
 11.1274 -\section{Xensv (web control interface)}
 11.1275 -\label{s:xensv}
 11.1276 -
 11.1277 -Xensv is the experimental web control interface for managing a Xen
 11.1278 -machine.  It can be used to perform some (but not yet all) of the
 11.1279 -management tasks that can be done using the xm tool.
 11.1280 -
 11.1281 -It can be started using:
 11.1282 -\begin{quote}
 11.1283 -\verb_# xensv start_
 11.1284 -\end{quote}
 11.1285 -and stopped using: 
 11.1286 -\begin{quote}
 11.1287 -\verb_# xensv stop_
 11.1288 -\end{quote}
 11.1289 -
 11.1290 -By default, Xensv will serve out the web interface on port 8080.  This
 11.1291 -can be changed by editing 
 11.1292 -\path{/usr/lib/python2.3/site-packages/xen/sv/params.py}.
 11.1293 -
 11.1294 -Once Xensv is running, the web interface can be used to create and
 11.1295 -manage running domains.
 11.1296 -
 11.1297 -
 11.1298 -
 11.1299 -
 11.1300 -\chapter{Domain Configuration}
 11.1301 -\label{cha:config}
 11.1302 -
 11.1303 -The following contains the syntax of the domain configuration 
 11.1304 -files and description of how to further specify networking, 
 11.1305 -driver domain and general scheduling behaviour. 
 11.1306 -
 11.1307 -\section{Configuration Files}
 11.1308 -\label{s:cfiles}
 11.1309 -
 11.1310 -Xen configuration files contain the following standard variables.
 11.1311 -Unless otherwise stated, configuration items should be enclosed in
 11.1312 -quotes: see \path{/etc/xen/xmexample1} and \path{/etc/xen/xmexample2} 
 11.1313 -for concrete examples of the syntax.
 11.1314 -
 11.1315 -\begin{description}
 11.1316 -\item[kernel] Path to the kernel image 
 11.1317 -\item[ramdisk] Path to a ramdisk image (optional).
 11.1318 -% \item[builder] The name of the domain build function (e.g. {\tt'linux'} or {\tt'netbsd'}.
 11.1319 -\item[memory] Memory size in megabytes.
 11.1320 -\item[cpu] CPU to run this domain on, or {\tt -1} for
 11.1321 -  auto-allocation. 
 11.1322 -\item[console] Port to export the domain console on (default 9600 + domain ID).
 11.1323 -\item[nics] Number of virtual network interfaces.
 11.1324 -\item[vif] List of MAC addresses (random addresses are assigned if not
 11.1325 -  given) and bridges to use for the domain's network interfaces, e.g.
 11.1326 -\begin{verbatim}
 11.1327 -vif = [ 'mac=aa:00:00:00:00:11, bridge=xen-br0',
 11.1328 -        'bridge=xen-br1' ]
 11.1329 -\end{verbatim}
 11.1330 -  to assign a MAC address and bridge to the first interface and assign
 11.1331 -  a different bridge to the second interface, leaving \xend to choose
 11.1332 -  the MAC address.
 11.1333 -\item[disk] List of block devices to export to the domain,  e.g. \\
 11.1334 -  \verb_disk = [ 'phy:hda1,sda1,r' ]_ \\
 11.1335 -  exports physical device \path{/dev/hda1} to the domain 
 11.1336 -  as \path{/dev/sda1} with read-only access. Exporting a disk read-write 
 11.1337 -  which is currently mounted is dangerous -- if you are \emph{certain}
 11.1338 -  you wish to do this, you can specify \path{w!} as the mode. 
 11.1339 -\item[dhcp] Set to {\tt 'dhcp'} if you want to use DHCP to configure
 11.1340 -  networking. 
 11.1341 -\item[netmask] Manually configured IP netmask.
 11.1342 -\item[gateway] Manually configured IP gateway. 
 11.1343 -\item[hostname] Set the hostname for the virtual machine.
 11.1344 -\item[root] Specify the root device parameter on the kernel command
 11.1345 -  line. 
 11.1346 -\item[nfs\_server] IP address for the NFS server (if any). 
 11.1347 -\item[nfs\_root] Path of the root filesystem on the NFS server (if any).
 11.1348 -\item[extra] Extra string to append to the kernel command line (if
 11.1349 -  any) 
 11.1350 -\item[restart] Three possible options:
 11.1351 -  \begin{description}
 11.1352 -  \item[always] Always restart the domain, no matter what
 11.1353 -                its exit code is.
 11.1354 -  \item[never]  Never restart the domain.
 11.1355 -  \item[onreboot] Restart the domain iff it requests reboot.
 11.1356 -  \end{description}
 11.1357 -\end{description}
 11.1358 -
 11.1359 -For additional flexibility, it is also possible to include Python
 11.1360 -scripting commands in configuration files.  An example of this is the
 11.1361 -\path{xmexample2} file, which uses Python code to handle the 
 11.1362 -\path{vmid} variable.
 11.1363 -
 11.1364 -
 11.1365 -%\part{Advanced Topics}
 11.1366 -
 11.1367 -\section{Network Configuration}
 11.1368 -
 11.1369 -For many users, the default installation should work `out of the box'.
 11.1370 -More complicated network setups, for instance with multiple ethernet
 11.1371 -interfaces and/or existing bridging setups will require some
 11.1372 -special configuration.
 11.1373 -
 11.1374 -The purpose of this section is to describe the mechanisms provided by
 11.1375 -\xend to allow a flexible configuration for Xen's virtual networking.
 11.1376 -
 11.1377 -\subsection{Xen virtual network topology}
 11.1378 -
 11.1379 -Each domain network interface is connected to a virtual network
 11.1380 -interface in dom0 by a point to point link (effectively a `virtual
 11.1381 -crossover cable').  These devices are named {\tt
 11.1382 -vif$<$domid$>$.$<$vifid$>$} (e.g. {\tt vif1.0} for the first interface
 11.1383 -in domain 1, {\tt vif3.1} for the second interface in domain 3).
 11.1384 -
 11.1385 -Traffic on these virtual interfaces is handled in domain 0 using
 11.1386 -standard Linux mechanisms for bridging, routing, rate limiting, etc.
 11.1387 -Xend calls on two shell scripts to perform initial configuration of
 11.1388 -the network and configuration of new virtual interfaces.  By default,
 11.1389 -these scripts configure a single bridge for all the virtual
 11.1390 -interfaces.  Arbitrary routing / bridging configurations can be
 11.1391 -configured by customising the scripts, as described in the following
 11.1392 -section.
 11.1393 -
 11.1394 -\subsection{Xen networking scripts}
 11.1395 -
 11.1396 -Xen's virtual networking is configured by two shell scripts (by
 11.1397 -default \path{network} and \path{vif-bridge}).  These are
 11.1398 -called automatically by \xend when certain events occur, with
 11.1399 -arguments to the scripts providing further contextual information.
 11.1400 -These scripts are found by default in \path{/etc/xen/scripts}.  The
 11.1401 -names and locations of the scripts can be configured in
 11.1402 -\path{/etc/xen/xend-config.sxp}.
 11.1403 -
 11.1404 -\begin{description} 
 11.1405 -
 11.1406 -\item[network:] This script is called whenever \xend is started or
 11.1407 -stopped to respectively initialise or tear down the Xen virtual
 11.1408 -network. In the default configuration initialisation creates the
 11.1409 -bridge `xen-br0' and moves eth0 onto that bridge, modifying the
 11.1410 -routing accordingly. When \xend exits, it deletes the Xen bridge and
 11.1411 -removes eth0, restoring the normal IP and routing configuration.
 11.1412 -
 11.1413 -%% In configurations where the bridge already exists, this script could
 11.1414 -%% be replaced with a link to \path{/bin/true} (for instance).
 11.1415 -
 11.1416 -\item[vif-bridge:] This script is called for every domain virtual
 11.1417 -interface and can configure firewalling rules and add the vif 
 11.1418 -to the appropriate bridge. By default, this adds and removes 
 11.1419 -VIFs on the default Xen bridge.
 11.1420 -
 11.1421 -\end{description} 
 11.1422 -
 11.1423 -For more complex network setups (e.g. where routing is required or
 11.1424 -integrate with existing bridges) these scripts may be replaced with
 11.1425 -customised variants for your site's preferred configuration.
 11.1426 -
 11.1427 -%% There are two possible types of privileges:  IO privileges and
 11.1428 -%% administration privileges.
 11.1429 -
 11.1430 -\section{Driver Domain Configuration} 
 11.1431 -
 11.1432 -I/O privileges can be assigned to allow a domain to directly access
 11.1433 -PCI devices itself.  This is used to support driver domains.
 11.1434 -
 11.1435 -Setting backend privileges is currently only supported in SXP format
 11.1436 -config files.  To allow a domain to function as a backend for others,
 11.1437 -somewhere within the {\tt vm} element of its configuration file must
 11.1438 -be a {\tt backend} element of the form {\tt (backend ({\em type}))}
 11.1439 -where {\tt \em type} may be either {\tt netif} or {\tt blkif},
 11.1440 -according to the type of virtual device this domain will service.
 11.1441 -%% After this domain has been built, \xend will connect all new and
 11.1442 -%% existing {\em virtual} devices (of the appropriate type) to that
 11.1443 -%% backend.
 11.1444 -
 11.1445 -Note that a block backend cannot currently import virtual block
 11.1446 -devices from other domains, and a network backend cannot import
 11.1447 -virtual network devices from other domains.  Thus (particularly in the
 11.1448 -case of block backends, which cannot import a virtual block device as
 11.1449 -their root filesystem), you may need to boot a backend domain from a
 11.1450 -ramdisk or a network device.
 11.1451 -
 11.1452 -Access to PCI devices may be configured on a per-device basis.  Xen
 11.1453 -will assign the minimal set of hardware privileges to a domain that
 11.1454 -are required to control its devices.  This can be configured in either
 11.1455 -format of configuration file:
 11.1456 -
 11.1457 -\begin{itemize}
 11.1458 -\item SXP Format: Include device elements of the form: \\
 11.1459 -\centerline{  {\tt (device (pci (bus {\em x}) (dev {\em y}) (func {\em z})))}} \\
 11.1460 -  inside the top-level {\tt vm} element.  Each one specifies the address
 11.1461 -  of a device this domain is allowed to access ---
 11.1462 -  the numbers {\em x},{\em y} and {\em z} may be in either decimal or
 11.1463 -  hexadecimal format.
 11.1464 -\item Flat Format: Include a list of PCI device addresses of the
 11.1465 -  format: \\ 
 11.1466 -\centerline{{\tt pci = ['x,y,z', ...]}} \\ 
 11.1467 -where each element in the
 11.1468 -  list is a string specifying the components of the PCI device
 11.1469 -  address, separated by commas.  The components ({\tt \em x}, {\tt \em
 11.1470 -  y} and {\tt \em z}) of the list may be formatted as either decimal
 11.1471 -  or hexadecimal.
 11.1472 -\end{itemize}
 11.1473 -
 11.1474 -%% \section{Administration Domains}
 11.1475 -
 11.1476 -%% Administration privileges allow a domain to use the `dom0
 11.1477 -%% operations' (so called because they are usually available only to
 11.1478 -%% domain 0).  A privileged domain can build other domains, set scheduling
 11.1479 -%% parameters, etc.
 11.1480 -
 11.1481 -% Support for other administrative domains is not yet available...  perhaps
 11.1482 -% we should plumb it in some time
 11.1483 -
 11.1484 -
 11.1485 -
 11.1486 -
 11.1487 -
 11.1488 -\section{Scheduler Configuration}
 11.1489 -\label{s:sched} 
 11.1490 -
 11.1491 -
 11.1492 -Xen offers a boot time choice between multiple schedulers.  To select
 11.1493 -a scheduler, pass the boot parameter {\em sched=sched\_name} to Xen,
 11.1494 -substituting the appropriate scheduler name.  Details of the schedulers
 11.1495 -and their parameters are included below; future versions of the tools
 11.1496 -will provide a higher-level interface to these tools.
 11.1497 +%% Chapter Control Software moved to control_software.tex
 11.1498 +\include{src/user/control_software}
 11.1499  
 11.1500 -It is expected that system administrators configure their system to
 11.1501 -use the scheduler most appropriate to their needs.  Currently, the BVT
 11.1502 -scheduler is the recommended choice. 
 11.1503 -
 11.1504 -\subsection{Borrowed Virtual Time}
 11.1505 -
 11.1506 -{\tt sched=bvt} (the default) \\ 
 11.1507 -
 11.1508 -BVT provides proportional fair shares of the CPU time.  It has been
 11.1509 -observed to penalise domains that block frequently (e.g. I/O intensive
 11.1510 -domains), but this can be compensated for by using warping. 
 11.1511 -
 11.1512 -\subsubsection{Global Parameters}
 11.1513 -
 11.1514 -\begin{description}
 11.1515 -\item[ctx\_allow]
 11.1516 -  the context switch allowance is similar to the `quantum'
 11.1517 -  in traditional schedulers.  It is the minimum time that
 11.1518 -  a scheduled domain will be allowed to run before being
 11.1519 -  pre-empted. 
 11.1520 -\end{description}
 11.1521 -
 11.1522 -\subsubsection{Per-domain parameters}
 11.1523 -
 11.1524 -\begin{description}
 11.1525 -\item[mcuadv]
 11.1526 -  the MCU (Minimum Charging Unit) advance determines the
 11.1527 -  proportional share of the CPU that a domain receives.  It
 11.1528 -  is set inversely proportionally to a domain's sharing weight.
 11.1529 -\item[warp]
 11.1530 -  the amount of `virtual time' the domain is allowed to warp
 11.1531 -  backwards
 11.1532 -\item[warpl]
 11.1533 -  the warp limit is the maximum time a domain can run warped for
 11.1534 -\item[warpu]
 11.1535 -  the unwarp requirement is the minimum time a domain must
 11.1536 -  run unwarped for before it can warp again
 11.1537 -\end{description}
 11.1538 -
 11.1539 -\subsection{Atropos}
 11.1540 -
 11.1541 -{\tt sched=atropos} \\
 11.1542 -
 11.1543 -Atropos is a soft real time scheduler.  It provides guarantees about
 11.1544 -absolute shares of the CPU, with a facility for sharing
 11.1545 -slack CPU time on a best-effort basis. It can provide timeliness
 11.1546 -guarantees for latency-sensitive domains.
 11.1547 -
 11.1548 -Every domain has an associated period and slice.  The domain should
 11.1549 -receive `slice' nanoseconds every `period' nanoseconds.  This allows
 11.1550 -the administrator to configure both the absolute share of the CPU a
 11.1551 -domain receives and the frequency with which it is scheduled. 
 11.1552 -
 11.1553 -%%  When
 11.1554 -%% domains unblock, their period is reduced to the value of the latency
 11.1555 -%% hint (the slice is scaled accordingly so that they still get the same
 11.1556 -%% proportion of the CPU).  For each subsequent period, the slice and
 11.1557 -%% period times are doubled until they reach their original values.
 11.1558 -
 11.1559 -Note: don't overcommit the CPU when using Atropos (i.e. don't reserve
 11.1560 -more CPU than is available --- the utilisation should be kept to
 11.1561 -slightly less than 100\% in order to ensure predictable behaviour).
 11.1562 -
 11.1563 -\subsubsection{Per-domain parameters}
 11.1564 -
 11.1565 -\begin{description}
 11.1566 -\item[period] The regular time interval during which a domain is
 11.1567 -  guaranteed to receive its allocation of CPU time.
 11.1568 -\item[slice]
 11.1569 -  The length of time per period that a domain is guaranteed to run
 11.1570 -  for (in the absence of voluntary yielding of the CPU). 
 11.1571 -\item[latency]
 11.1572 -  The latency hint is used to control how soon after
 11.1573 -  waking up a domain it should be scheduled.
 11.1574 -\item[xtratime] This is a boolean flag that specifies whether a domain
 11.1575 -  should be allowed a share of the system slack time.
 11.1576 -\end{description}
 11.1577 -
 11.1578 -\subsection{Round Robin}
 11.1579 -
 11.1580 -{\tt sched=rrobin} \\
 11.1581 -
 11.1582 -The round robin scheduler is included as a simple demonstration of
 11.1583 -Xen's internal scheduler API.  It is not intended for production use. 
 11.1584 -
 11.1585 -\subsubsection{Global Parameters}
 11.1586 -
 11.1587 -\begin{description}
 11.1588 -\item[rr\_slice]
 11.1589 -  The maximum time each domain runs before the next
 11.1590 -  scheduling decision is made.
 11.1591 -\end{description}
 11.1592 -
 11.1593 -
 11.1594 -
 11.1595 -
 11.1596 -
 11.1597 -
 11.1598 -
 11.1599 -
 11.1600 -
 11.1601 -
 11.1602 -
 11.1603 -
 11.1604 -\chapter{Build, Boot and Debug options} 
 11.1605 -
 11.1606 -This chapter describes the build- and boot-time options 
 11.1607 -which may be used to tailor your Xen system. 
 11.1608 -
 11.1609 -\section{Xen Build Options}
 11.1610 -
 11.1611 -Xen provides a number of build-time options which should be 
 11.1612 -set as environment variables or passed on make's command-line.  
 11.1613 -
 11.1614 -\begin{description} 
 11.1615 -\item[verbose=y] Enable debugging messages when Xen detects an unexpected condition.
 11.1616 -Also enables console output from all domains.
 11.1617 -\item[debug=y] 
 11.1618 -Enable debug assertions.  Implies {\bf verbose=y}.
 11.1619 -(Primarily useful for tracing bugs in Xen).       
 11.1620 -\item[debugger=y] 
 11.1621 -Enable the in-Xen debugger. This can be used to debug 
 11.1622 -Xen, guest OSes, and applications.
 11.1623 -\item[perfc=y] 
 11.1624 -Enable performance counters for significant events
 11.1625 -within Xen. The counts can be reset or displayed
 11.1626 -on Xen's console via console control keys.
 11.1627 -\item[trace=y] 
 11.1628 -Enable per-cpu trace buffers which log a range of
 11.1629 -events within Xen for collection by control
 11.1630 -software. 
 11.1631 -\end{description} 
 11.1632 -
 11.1633 -\section{Xen Boot Options}
 11.1634 -\label{s:xboot}
 11.1635 -
 11.1636 -These options are used to configure Xen's behaviour at runtime.  They
 11.1637 -should be appended to Xen's command line, either manually or by
 11.1638 -editing \path{grub.conf}.
 11.1639 -
 11.1640 -\begin{description}
 11.1641 -\item [noreboot ] 
 11.1642 - Don't reboot the machine automatically on errors.  This is
 11.1643 - useful to catch debug output if you aren't catching console messages
 11.1644 - via the serial line. 
 11.1645 -
 11.1646 -\item [nosmp ] 
 11.1647 - Disable SMP support.
 11.1648 - This option is implied by `ignorebiostables'. 
 11.1649 -
 11.1650 -\item [watchdog ] 
 11.1651 - Enable NMI watchdog which can report certain failures. 
 11.1652 -
 11.1653 -\item [noirqbalance ] 
 11.1654 - Disable software IRQ balancing and affinity. This can be used on
 11.1655 - systems such as Dell 1850/2850 that have workarounds in hardware for
 11.1656 - IRQ-routing issues.
 11.1657 +%% Chapter Domain Configuration moved to domain_configuration.tex
 11.1658 +\include{src/user/domain_configuration}
 11.1659  
 11.1660 -\item [badpage=$<$page number$>$,$<$page number$>$, \ldots ] 
 11.1661 - Specify a list of pages not to be allocated for use 
 11.1662 - because they contain bad bytes. For example, if your
 11.1663 - memory tester says that byte 0x12345678 is bad, you would
 11.1664 - place `badpage=0x12345' on Xen's command line. 
 11.1665 -
 11.1666 -\item [com1=$<$baud$>$,DPS,$<$io\_base$>$,$<$irq$>$
 11.1667 - com2=$<$baud$>$,DPS,$<$io\_base$>$,$<$irq$>$ ] \mbox{}\\ 
 11.1668 - Xen supports up to two 16550-compatible serial ports.
 11.1669 - For example: `com1=9600, 8n1, 0x408, 5' maps COM1 to a
 11.1670 - 9600-baud port, 8 data bits, no parity, 1 stop bit,
 11.1671 - I/O port base 0x408, IRQ 5.
 11.1672 - If some configuration options are standard (e.g., I/O base and IRQ),
 11.1673 - then only a prefix of the full configuration string need be
 11.1674 - specified. If the baud rate is pre-configured (e.g., by the
 11.1675 - bootloader) then you can specify `auto' in place of a numeric baud
 11.1676 - rate. 
 11.1677 -
 11.1678 -\item [console=$<$specifier list$>$ ] 
 11.1679 - Specify the destination for Xen console I/O.
 11.1680 - This is a comma-separated list of, for example:
 11.1681 -\begin{description}
 11.1682 - \item[vga]  use VGA console and allow keyboard input
 11.1683 - \item[com1] use serial port com1
 11.1684 - \item[com2H] use serial port com2. Transmitted chars will
 11.1685 -   have the MSB set. Received chars must have
 11.1686 -   MSB set.
 11.1687 - \item[com2L] use serial port com2. Transmitted chars will
 11.1688 -   have the MSB cleared. Received chars must
 11.1689 -   have MSB cleared.
 11.1690 -\end{description}
 11.1691 - The latter two examples allow a single port to be
 11.1692 - shared by two subsystems (e.g. console and
 11.1693 - debugger). Sharing is controlled by MSB of each
 11.1694 - transmitted/received character.
 11.1695 - [NB. Default for this option is `com1,vga'] 
 11.1696 -
 11.1697 -\item [sync\_console ]
 11.1698 - Force synchronous console output. This is useful if you system fails
 11.1699 - unexpectedly before it has sent all available output to the
 11.1700 - console. In most cases Xen will automatically enter synchronous mode
 11.1701 - when an exceptional event occurs, but this option provides a manual
 11.1702 - fallback.
 11.1703 -
 11.1704 -\item [conswitch=$<$switch-char$><$auto-switch-char$>$ ] 
 11.1705 - Specify how to switch serial-console input between
 11.1706 - Xen and DOM0. The required sequence is CTRL-$<$switch-char$>$
 11.1707 - pressed three times. Specifying the backtick character 
 11.1708 - disables switching.
 11.1709 - The $<$auto-switch-char$>$ specifies whether Xen should
 11.1710 - auto-switch input to DOM0 when it boots --- if it is `x'
 11.1711 - then auto-switching is disabled.  Any other value, or
 11.1712 - omitting the character, enables auto-switching.
 11.1713 - [NB. default switch-char is `a'] 
 11.1714 -
 11.1715 -\item [nmi=xxx ] 
 11.1716 - Specify what to do with an NMI parity or I/O error. \\
 11.1717 - `nmi=fatal':  Xen prints a diagnostic and then hangs. \\
 11.1718 - `nmi=dom0':   Inform DOM0 of the NMI. \\
 11.1719 - `nmi=ignore': Ignore the NMI. 
 11.1720 -
 11.1721 -\item [mem=xxx ]
 11.1722 - Set the physical RAM address limit. Any RAM appearing beyond this
 11.1723 - physical address in the memory map will be ignored. This parameter
 11.1724 - may be specified with a B, K, M or G suffix, representing bytes,
 11.1725 - kilobytes, megabytes and gigabytes respectively. The
 11.1726 - default unit, if no suffix is specified, is kilobytes.
 11.1727 -
 11.1728 -\item [dom0\_mem=xxx ] 
 11.1729 - Set the amount of memory to be allocated to domain0. In Xen 3.x the parameter
 11.1730 - may be specified with a B, K, M or G suffix, representing bytes,
 11.1731 - kilobytes, megabytes and gigabytes respectively; if no suffix is specified, 
 11.1732 - the parameter defaults to kilobytes. In previous versions of Xen, suffixes
 11.1733 - were not supported and the value is always interpreted as kilobytes. 
 11.1734 -
 11.1735 -\item [tbuf\_size=xxx ] 
 11.1736 - Set the size of the per-cpu trace buffers, in pages
 11.1737 - (default 1).  Note that the trace buffers are only
 11.1738 - enabled in debug builds.  Most users can ignore
 11.1739 - this feature completely. 
 11.1740 -
 11.1741 -\item [sched=xxx ] 
 11.1742 - Select the CPU scheduler Xen should use.  The current
 11.1743 - possibilities are `bvt' (default), `atropos' and `rrobin'. 
 11.1744 - For more information see Section~\ref{s:sched}. 
 11.1745 -
 11.1746 -\item [apic\_verbosity=debug,verbose ]
 11.1747 - Print more detailed information about local APIC and IOAPIC configuration.
 11.1748 -
 11.1749 -\item [lapic ]
 11.1750 - Force use of local APIC even when left disabled by uniprocessor BIOS.
 11.1751 -
 11.1752 -\item [nolapic ]
 11.1753 - Ignore local APIC in a uniprocessor system, even if enabled by the BIOS.
 11.1754 -
 11.1755 -\item [apic=bigsmp,default,es7000,summit ]
 11.1756 - Specify NUMA platform. This can usually be probed automatically.
 11.1757 -
 11.1758 -\end{description} 
 11.1759 -
 11.1760 -In addition, the following options may be specified on the Xen command
 11.1761 -line. Since domain 0 shares responsibility for booting the platform,
 11.1762 -Xen will automatically propagate these options to its command
 11.1763 -line. These options are taken from Linux's command-line syntax with
 11.1764 -unchanged semantics.
 11.1765 -
 11.1766 -\begin{description}
 11.1767 -\item [acpi=off,force,strict,ht,noirq,\ldots ] 
 11.1768 - Modify how Xen (and domain 0) parses the BIOS ACPI tables.
 11.1769 -
 11.1770 -\item [acpi\_skip\_timer\_override ]
 11.1771 - Instruct Xen (and domain 0) to ignore timer-interrupt override
 11.1772 - instructions specified by the BIOS ACPI tables.
 11.1773 -
 11.1774 -\item [noapic ]
 11.1775 - Instruct Xen (and domain 0) to ignore any IOAPICs that are present in
 11.1776 - the system, and instead continue to use the legacy PIC.
 11.1777 -
 11.1778 -\end{description} 
 11.1779 -
 11.1780 -\section{XenLinux Boot Options}
 11.1781 -
 11.1782 -In addition to the standard Linux kernel boot options, we support: 
 11.1783 -\begin{description} 
 11.1784 -\item[xencons=xxx ] Specify the device node to which the Xen virtual
 11.1785 -console driver is attached. The following options are supported:
 11.1786 -\begin{center}
 11.1787 -\begin{tabular}{l}
 11.1788 -`xencons=off': disable virtual console \\ 
 11.1789 -`xencons=tty': attach console to /dev/tty1 (tty0 at boot-time) \\
 11.1790 -`xencons=ttyS': attach console to /dev/ttyS0
 11.1791 -\end{tabular}
 11.1792 -\end{center}
 11.1793 -The default is ttyS for dom0 and tty for all other domains.
 11.1794 -\end{description} 
 11.1795 -
 11.1796 -
 11.1797 -
 11.1798 -\section{Debugging}
 11.1799 -\label{s:keys} 
 11.1800 -
 11.1801 -Xen has a set of debugging features that can be useful to try and
 11.1802 -figure out what's going on. Hit 'h' on the serial line (if you
 11.1803 -specified a baud rate on the Xen command line) or ScrollLock-h on the
 11.1804 -keyboard to get a list of supported commands.
 11.1805 -
 11.1806 -If you have a crash you'll likely get a crash dump containing an EIP
 11.1807 -(PC) which, along with an \path{objdump -d image}, can be useful in
 11.1808 -figuring out what's happened.  Debug a Xenlinux image just as you
 11.1809 -would any other Linux kernel.
 11.1810 -
 11.1811 -%% We supply a handy debug terminal program which you can find in
 11.1812 -%% \path{/usr/local/src/xen-2.0.bk/tools/misc/miniterm/}
 11.1813 -%% This should be built and executed on another machine that is connected
 11.1814 -%% via a null modem cable. Documentation is included.
 11.1815 -%% Alternatively, if the Xen machine is connected to a serial-port server
 11.1816 -%% then we supply a dumb TCP terminal client, {\tt xencons}.
 11.1817 -
 11.1818 -
 11.1819 +%% Chapter Build, Boot and Debug Options moved to build.tex
 11.1820 +\include{src/user/build}
 11.1821  
 11.1822  
 11.1823  \chapter{Further Support}
 11.1824 @@ -1875,6 +108,7 @@ directory of the Xen source distribution
 11.1825  %Various HOWTOs are available in \path{docs/HOWTOS} but this content is
 11.1826  %being integrated into this manual.
 11.1827  
 11.1828 +
 11.1829  \section{Online References}
 11.1830  
 11.1831  The official Xen web site is found at:
 11.1832 @@ -1885,6 +119,7 @@ The official Xen web site is found at:
 11.1833  This contains links to the latest versions of all on-line 
 11.1834  documentation (including the lateset version of the FAQ). 
 11.1835  
 11.1836 +
 11.1837  \section{Mailing Lists}
 11.1838  
 11.1839  There are currently four official Xen mailing lists:
 11.1840 @@ -1905,326 +140,18 @@ from the unstable and 2.0 trees - develo
 11.1841  \end{description}
 11.1842  
 11.1843  
 11.1844 +
 11.1845  \appendix
 11.1846  
 11.1847 +%% Chapter Installing Xen / XenLinux on Debian moved to debian.tex
 11.1848 +\include{src/user/debian}
 11.1849 +
 11.1850 +%% Chapter Installing Xen on Red Hat moved to redhat.tex
 11.1851 +\include{src/user/redhat}
 11.1852 +
 11.1853  
 11.1854 -\chapter{Installing Xen / XenLinux on Debian}
 11.1855 -
 11.1856 -The Debian project provides a tool called \path{debootstrap} which
 11.1857 -allows a base Debian system to be installed into a filesystem without
 11.1858 -requiring the host system to have any Debian-specific software (such
 11.1859 -as \path{apt}. 
 11.1860 -
 11.1861 -Here's some info how to install Debian 3.1 (Sarge) for an unprivileged
 11.1862 -Xen domain:
 11.1863 -
 11.1864 -\begin{enumerate}
 11.1865 -\item Set up Xen 2.0 and test that it's working, as described earlier in
 11.1866 -      this manual.
 11.1867 -
 11.1868 -\item Create disk images for root-fs and swap (alternatively, you
 11.1869 -      might create dedicated partitions, LVM logical volumes, etc. if
 11.1870 -      that suits your setup).
 11.1871 -\begin{small}\begin{verbatim}  
 11.1872 -dd if=/dev/zero of=/path/diskimage bs=1024k count=size_in_mbytes
 11.1873 -dd if=/dev/zero of=/path/swapimage bs=1024k count=size_in_mbytes
 11.1874 -\end{verbatim}\end{small}
 11.1875 -      If you're going to use this filesystem / disk image only as a
 11.1876 -      `template' for other vm disk images, something like 300 MB should
 11.1877 -      be enough.. (of course it depends what kind of packages you are
 11.1878 -      planning to install to the template)
 11.1879 -
 11.1880 -\item Create the filesystem and initialise the swap image
 11.1881 -\begin{small}\begin{verbatim}
 11.1882 -mkfs.ext3 /path/diskimage
 11.1883 -mkswap /path/swapimage
 11.1884 -\end{verbatim}\end{small}
 11.1885 -
 11.1886 -\item Mount the disk image for installation
 11.1887 -\begin{small}\begin{verbatim}
 11.1888 -mount -o loop /path/diskimage /mnt/disk
 11.1889 -\end{verbatim}\end{small}
 11.1890 -
 11.1891 -\item Install \path{debootstrap}
 11.1892 -
 11.1893 -Make sure you have debootstrap installed on the host.  If you are
 11.1894 -running Debian sarge (3.1 / testing) or unstable you can install it by
 11.1895 -running \path{apt-get install debootstrap}.  Otherwise, it can be
 11.1896 -downloaded from the Debian project website.
 11.1897 -
 11.1898 -\item Install Debian base to the disk image:
 11.1899 -\begin{small}\begin{verbatim}
 11.1900 -debootstrap --arch i386 sarge /mnt/disk  \
 11.1901 -            http://ftp.<countrycode>.debian.org/debian
 11.1902 -\end{verbatim}\end{small}
 11.1903 -
 11.1904 -You can use any other Debian http/ftp mirror you want.
 11.1905 -
 11.1906 -\item When debootstrap completes successfully, modify settings:
 11.1907 -\begin{small}\begin{verbatim}
 11.1908 -chroot /mnt/disk /bin/bash
 11.1909 -\end{verbatim}\end{small}
 11.1910 -
 11.1911 -Edit the following files using vi or nano and make needed changes:
 11.1912 -\begin{small}\begin{verbatim}
 11.1913 -/etc/hostname
 11.1914 -/etc/hosts
 11.1915 -/etc/resolv.conf
 11.1916 -/etc/network/interfaces
 11.1917 -/etc/networks
 11.1918 -\end{verbatim}\end{small}
 11.1919 -
 11.1920 -Set up access to the services, edit:
 11.1921 -\begin{small}\begin{verbatim}
 11.1922 -/etc/hosts.deny
 11.1923 -/etc/hosts.allow
 11.1924 -/etc/inetd.conf
 11.1925 -\end{verbatim}\end{small}
 11.1926 -
 11.1927 -Add Debian mirror to:   
 11.1928 -\begin{small}\begin{verbatim}
 11.1929 -/etc/apt/sources.list
 11.1930 -\end{verbatim}\end{small}
 11.1931 -
 11.1932 -Create fstab like this:
 11.1933 -\begin{small}\begin{verbatim}
 11.1934 -/dev/sda1       /       ext3    errors=remount-ro       0       1
 11.1935 -/dev/sda2       none    swap    sw                      0       0
 11.1936 -proc            /proc   proc    defaults                0       0
 11.1937 -\end{verbatim}\end{small}
 11.1938 -
 11.1939 -Logout
 11.1940 -
 11.1941 -\item      Unmount the disk image
 11.1942 -\begin{small}\begin{verbatim}
 11.1943 -umount /mnt/disk
 11.1944 -\end{verbatim}\end{small}
 11.1945 -
 11.1946 -\item Create Xen 2.0 configuration file for the new domain. You can
 11.1947 -        use the example-configurations coming with Xen as a template.
 11.1948 -
 11.1949 -        Make sure you have the following set up:
 11.1950 -\begin{small}\begin{verbatim}
 11.1951 -disk = [ 'file:/path/diskimage,sda1,w', 'file:/path/swapimage,sda2,w' ]
 11.1952 -root = "/dev/sda1 ro"
 11.1953 -\end{verbatim}\end{small}
 11.1954 -
 11.1955 -\item Start the new domain
 11.1956 -\begin{small}\begin{verbatim}
 11.1957 -xm create -f domain_config_file
 11.1958 -\end{verbatim}\end{small}
 11.1959 -
 11.1960 -Check that the new domain is running:
 11.1961 -\begin{small}\begin{verbatim}
 11.1962 -xm list
 11.1963 -\end{verbatim}\end{small}
 11.1964 -
 11.1965 -\item   Attach to the console of the new domain.
 11.1966 -        You should see something like this when starting the new domain:
 11.1967 -
 11.1968 -\begin{small}\begin{verbatim}
 11.1969 -Started domain testdomain2, console on port 9626
 11.1970 -\end{verbatim}\end{small}
 11.1971 -        
 11.1972 -        There you can see the ID of the console: 26. You can also list
 11.1973 -        the consoles with \path{xm consoles} (ID is the last two
 11.1974 -        digits of the port number.)
 11.1975 -
 11.1976 -        Attach to the console:
 11.1977 -
 11.1978 -\begin{small}\begin{verbatim}
 11.1979 -xm console 26
 11.1980 -\end{verbatim}\end{small}
 11.1981 -
 11.1982 -        or by telnetting to the port 9626 of localhost (the xm console
 11.1983 -        program works better).
 11.1984 -
 11.1985 -\item   Log in and run base-config
 11.1986 -
 11.1987 -        As a default there's no password for the root.
 11.1988 -
 11.1989 -        Check that everything looks OK, and the system started without
 11.1990 -        errors.  Check that the swap is active, and the network settings are
 11.1991 -        correct.
 11.1992 -
 11.1993 -        Run \path{/usr/sbin/base-config} to set up the Debian settings.
 11.1994 -
 11.1995 -        Set up the password for root using passwd.
 11.1996 -
 11.1997 -\item     Done. You can exit the console by pressing \path{Ctrl + ]}
 11.1998 -
 11.1999 -\end{enumerate}
 11.2000 -
 11.2001 -If you need to create new domains, you can just copy the contents of
 11.2002 -the `template'-image to the new disk images, either by mounting the
 11.2003 -template and the new image, and using \path{cp -a} or \path{tar} or by
 11.2004 -simply copying the image file.  Once this is done, modify the
 11.2005 -image-specific settings (hostname, network settings, etc).
 11.2006 -
 11.2007 -\chapter{Installing Xen / XenLinux on Redhat or Fedora Core}
 11.2008 -
 11.2009 -When using Xen / XenLinux on a standard Linux distribution there are
 11.2010 -a couple of things to watch out for:
 11.2011 -
 11.2012 -Note that, because domains>0 don't have any privileged access at all,
 11.2013 -certain commands in the default boot sequence will fail e.g. attempts
 11.2014 -to update the hwclock, change the console font, update the keytable
 11.2015 -map, start apmd (power management), or gpm (mouse cursor).  Either
 11.2016 -ignore the errors (they should be harmless), or remove them from the
 11.2017 -startup scripts.  Deleting the following links are a good start:
 11.2018 -{\path{S24pcmcia}}, {\path{S09isdn}},
 11.2019 -{\path{S17keytable}}, {\path{S26apmd}},
 11.2020 -{\path{S85gpm}}.
 11.2021 -
 11.2022 -If you want to use a single root file system that works cleanly for
 11.2023 -both domain 0 and unprivileged domains, a useful trick is to use
 11.2024 -different 'init' run levels. For example, use
 11.2025 -run level 3 for domain 0, and run level 4 for other domains. This
 11.2026 -enables different startup scripts to be run in depending on the run
 11.2027 -level number passed on the kernel command line.
 11.2028 -
 11.2029 -If using NFS root files systems mounted either from an
 11.2030 -external server or from domain0 there are a couple of other gotchas.
 11.2031 -The default {\path{/etc/sysconfig/iptables}} rules block NFS, so part
 11.2032 -way through the boot sequence things will suddenly go dead.
 11.2033 -
 11.2034 -If you're planning on having a separate NFS {\path{/usr}} partition, the
 11.2035 -RH9 boot scripts don't make life easy - they attempt to mount NFS file
 11.2036 -systems way to late in the boot process. The easiest way I found to do
 11.2037 -this was to have a {\path{/linuxrc}} script run ahead of
 11.2038 -{\path{/sbin/init}} that mounts {\path{/usr}}:
 11.2039 -
 11.2040 -\begin{quote}
 11.2041 -\begin{small}\begin{verbatim}
 11.2042 - #!/bin/bash
 11.2043 - /sbin/ipconfig lo 127.0.0.1
 11.2044 - /sbin/portmap
 11.2045 - /bin/mount /usr
 11.2046 - exec /sbin/init "$@" <>/dev/console 2>&1
 11.2047 -\end{verbatim}\end{small}
 11.2048 -\end{quote}
 11.2049 -
 11.2050 -%$ XXX SMH: font lock fix :-)  
 11.2051 -
 11.2052 -The one slight complication with the above is that
 11.2053 -{\path{/sbin/portmap}} is dynamically linked against
 11.2054 -{\path{/usr/lib/libwrap.so.0}} Since this is in
 11.2055 -{\path{/usr}}, it won't work. This can be solved by copying the
 11.2056 -file (and link) below the /usr mount point, and just let the file be
 11.2057 -'covered' when the mount happens.
 11.2058 -
 11.2059 -In some installations, where a shared read-only {\path{/usr}} is
 11.2060 -being used, it may be desirable to move other large directories over
 11.2061 -into the read-only {\path{/usr}}. For example, you might replace
 11.2062 -{\path{/bin}}, {\path{/lib}} and {\path{/sbin}} with
 11.2063 -links into {\path{/usr/root/bin}}, {\path{/usr/root/lib}}
 11.2064 -and {\path{/usr/root/sbin}} respectively. This creates other
 11.2065 -problems for running the {\path{/linuxrc}} script, requiring
 11.2066 -bash, portmap, mount, ifconfig, and a handful of other shared
 11.2067 -libraries to be copied below the mount point --- a simple
 11.2068 -statically-linked C program would solve this problem.
 11.2069 -
 11.2070 -
 11.2071 -
 11.2072 -
 11.2073 -\chapter{Glossary of Terms}
 11.2074 -
 11.2075 -\begin{description}
 11.2076 -\item[Atropos]             One of the CPU schedulers provided by Xen.
 11.2077 -                           Atropos provides domains with absolute shares
 11.2078 -                           of the CPU, with timeliness guarantees and a
 11.2079 -                           mechanism for sharing out `slack time'.
 11.2080 -
 11.2081 -\item[BVT]                 The BVT scheduler is used to give proportional
 11.2082 -                           fair shares of the CPU to domains.
 11.2083 -
 11.2084 -\item[Exokernel]           A minimal piece of privileged code, similar to
 11.2085 -                           a {\bf microkernel} but providing a more
 11.2086 -                           `hardware-like' interface to the tasks it
 11.2087 -                           manages.  This is similar to a paravirtualising
 11.2088 -                           VMM like {\bf Xen} but was designed as a new
 11.2089 -                           operating system structure, rather than
 11.2090 -                           specifically to run multiple conventional OSs.
 11.2091 -
 11.2092 -\item[Domain]              A domain is the execution context that
 11.2093 -                           contains a running {\bf virtual machine}.
 11.2094 -                           The relationship between virtual machines
 11.2095 -                           and domains on Xen is similar to that between
 11.2096 -                           programs and processes in an operating
 11.2097 -                           system: a virtual machine is a persistent
 11.2098 -                           entity that resides on disk (somewhat like
 11.2099 -                           a program).  When it is loaded for execution,
 11.2100 -                           it runs in a domain.  Each domain has a
 11.2101 -                           {\bf domain ID}.
 11.2102 -
 11.2103 -\item[Domain 0]            The first domain to be started on a Xen
 11.2104 -                           machine.  Domain 0 is responsible for managing
 11.2105 -                           the system.
 11.2106 -
 11.2107 -\item[Domain ID]           A unique identifier for a {\bf domain},
 11.2108 -                           analogous to a process ID in an operating
 11.2109 -                           system.
 11.2110 -
 11.2111 -\item[Full virtualisation] An approach to virtualisation which
 11.2112 -                           requires no modifications to the hosted
 11.2113 -                           operating system, providing the illusion of
 11.2114 -                           a complete system of real hardware devices.
 11.2115 -
 11.2116 -\item[Hypervisor]          An alternative term for {\bf VMM}, used
 11.2117 -                           because it means `beyond supervisor',
 11.2118 -                           since it is responsible for managing multiple
 11.2119 -                           `supervisor' kernels.
 11.2120 -
 11.2121 -\item[Live migration]      A technique for moving a running virtual
 11.2122 -                           machine to another physical host, without
 11.2123 -                           stopping it or the services running on it.
 11.2124 -
 11.2125 -\item[Microkernel]         A small base of code running at the highest
 11.2126 -                           hardware privilege level.  A microkernel is
 11.2127 -                           responsible for sharing CPU and memory (and
 11.2128 -                           sometimes other devices) between less
 11.2129 -                           privileged tasks running on the system.
 11.2130 -                           This is similar to a VMM, particularly a
 11.2131 -                           {\bf paravirtualising} VMM but typically
 11.2132 -                           addressing a different problem space and
 11.2133 -                           providing different kind of interface.
 11.2134 -
 11.2135 -\item[NetBSD/Xen]          A port of NetBSD to the Xen architecture.
 11.2136 -
 11.2137 -\item[Paravirtualisation]  An approach to virtualisation which requires
 11.2138 -                           modifications to the operating system in
 11.2139 -                           order to run in a virtual machine.  Xen
 11.2140 -                           uses paravirtualisation but preserves
 11.2141 -                           binary compatibility for user space
 11.2142 -                           applications.
 11.2143 -
 11.2144 -\item[Shadow pagetables]   A technique for hiding the layout of machine
 11.2145 -                           memory from a virtual machine's operating
 11.2146 -                           system.  Used in some {\bf VMMs} to provide
 11.2147 -                           the illusion of contiguous physical memory,
 11.2148 -                           in Xen this is used during
 11.2149 -                           {\bf live migration}.
 11.2150 -
 11.2151 -\item[Virtual Machine]     The environment in which a hosted operating
 11.2152 -                           system runs, providing the abstraction of a
 11.2153 -                           dedicated machine.  A virtual machine may
 11.2154 -                           be identical to the underlying hardware (as
 11.2155 -                           in {\bf full virtualisation}, or it may
 11.2156 -                           differ, as in {\bf paravirtualisation}.
 11.2157 -
 11.2158 -\item[VMM]                 Virtual Machine Monitor - the software that
 11.2159 -                           allows multiple virtual machines to be
 11.2160 -                           multiplexed on a single physical machine.
 11.2161 -
 11.2162 -\item[Xen]                 Xen is a paravirtualising virtual machine
 11.2163 -                           monitor, developed primarily by the
 11.2164 -                           Systems Research Group at the University
 11.2165 -                           of Cambridge Computer Laboratory.
 11.2166 -
 11.2167 -\item[XenLinux]            Official name for the port of the Linux kernel
 11.2168 -                           that runs on Xen.
 11.2169 -
 11.2170 -\end{description}
 11.2171 +%% Chapter Glossary of Terms moved to glossary.tex
 11.2172 +\include{src/user/glossary}
 11.2173  
 11.2174  
 11.2175  \end{document}