ia64/xen-unstable
changeset 18685:46d7e12c4c91
merge with xen-unstable.hg
| author | Isaku Yamahata <yamahata@valinux.co.jp> |
|---|---|
| date | Wed Oct 22 11:46:55 2008 +0900 (2008-10-22) |
| parents | 6583186e5989 d2f7243fc571 |
| children | 6db3c096c244 |
| files | xen/include/asm-ia64/xenspinlock.h |
line diff
1.1 --- a/docs/xen-api/coversheet.tex Wed Oct 22 11:38:22 2008 +0900 1.2 +++ b/docs/xen-api/coversheet.tex Wed Oct 22 11:46:55 2008 +0900 1.3 @@ -51,6 +51,7 @@ Hollis Blanchard, IBM & Alastair Tse, Xe 1.4 Mike Day, IBM & Daniel Veillard, Red Hat \\ 1.5 Jim Fehlig, Novell & Tom Wilkie, University of Cambridge \\ 1.6 Jon Harrop, XenSource & Yosuke Iwamatsu, NEC \\ 1.7 +Masaki Kanno, FUJITSU \\ 1.8 \end{tabular} 1.9 \end{large} 1.10
2.1 --- a/docs/xen-api/revision-history.tex Wed Oct 22 11:38:22 2008 +0900 2.2 +++ b/docs/xen-api/revision-history.tex Wed Oct 22 11:46:55 2008 +0900 2.3 @@ -56,5 +56,14 @@ 2.4 \end{flushleft} 2.5 \end{minipage}\\ 2.6 \hline 2.7 + 1.0.7 & 20th Oct. 08 & M. Kanno & 2.8 + \begin{minipage}[t]{7cm} 2.9 + \begin{flushleft} 2.10 + Added definitions of new classes DSCSI and PSCSI. Updated the table 2.11 + and the diagram representing relationships between classes. 2.12 + Added host.PSCSIs and VM.DSCSIs fields. 2.13 + \end{flushleft} 2.14 + \end{minipage}\\ 2.15 + \hline 2.16 \end{tabular} 2.17 \end{center}
3.1 --- a/docs/xen-api/xenapi-coversheet.tex Wed Oct 22 11:38:22 2008 +0900 3.2 +++ b/docs/xen-api/xenapi-coversheet.tex Wed Oct 22 11:46:55 2008 +0900 3.3 @@ -17,12 +17,12 @@ 3.4 \newcommand{\coversheetlogo}{xen.eps} 3.5 3.6 %% Document date 3.7 -\newcommand{\datestring}{24th July 2008} 3.8 +\newcommand{\datestring}{20th October 2008} 3.9 3.10 \newcommand{\releasestatement}{Stable Release} 3.11 3.12 %% Document revision 3.13 -\newcommand{\revstring}{API Revision 1.0.6} 3.14 +\newcommand{\revstring}{API Revision 1.0.7} 3.15 3.16 %% Document authors 3.17 \newcommand{\docauthors}{
4.1 --- a/docs/xen-api/xenapi-datamodel-graph.dot Wed Oct 22 11:38:22 2008 +0900 4.2 +++ b/docs/xen-api/xenapi-datamodel-graph.dot Wed Oct 22 11:46:55 2008 +0900 4.3 @@ -12,9 +12,11 @@ 4.4 digraph "Xen-API Class Diagram" { 4.5 fontname="Verdana"; 4.6 4.7 -node [ shape=box ]; session VM host network VIF PIF SR VDI VBD PBD user XSPolicy ACMPolicy; 4.8 -node [shape=ellipse]; PIF_metrics VIF_metrics VM_metrics VBD_metrics PBD_metrics VM_guest_metrics host_metrics; 4.9 -node [shape=box]; DPCI PPCI host_cpu console VTPM 4.10 +node [ shape=box ]; session VM host network VIF PIF SR VDI VBD PBD user; 4.11 +node [ shape=box ]; XSPolicy ACMPolicy DPCI PPCI host_cpu console VTPM; 4.12 +node [ shape=box ]; DSCSI PSCSI; 4.13 +node [ shape=ellipse ]; VM_metrics VM_guest_metrics host_metrics; 4.14 +node [ shape=ellipse ]; PIF_metrics VIF_metrics VBD_metrics PBD_metrics; 4.15 session -> host [ arrowhead="none" ] 4.16 session -> user [ arrowhead="none" ] 4.17 VM -> VM_metrics [ arrowhead="none" ] 4.18 @@ -41,4 +43,7 @@ XSPolicy -> ACMPolicy [ arrowhead="none" 4.19 DPCI -> VM [ arrowhead="none", arrowtail="crow" ] 4.20 DPCI -> PPCI [ arrowhead="none" ] 4.21 PPCI -> host [ arrowhead="none", arrowtail="crow" ] 4.22 +DSCSI -> VM [ arrowhead="none", arrowtail="crow" ] 4.23 +DSCSI -> PSCSI [ arrowhead="none" ] 4.24 +PSCSI -> host [ arrowhead="none", arrowtail="crow" ] 4.25 }
5.1 --- a/docs/xen-api/xenapi-datamodel.tex Wed Oct 22 11:38:22 2008 +0900 5.2 +++ b/docs/xen-api/xenapi-datamodel.tex Wed Oct 22 11:46:55 2008 +0900 5.3 @@ -46,6 +46,8 @@ Name & Description \\ 5.4 {\tt console} & A console \\ 5.5 {\tt DPCI} & A pass-through PCI device \\ 5.6 {\tt PPCI} & A physical PCI device \\ 5.7 +{\tt DSCSI} & A half-virtualized SCSI device \\ 5.8 +{\tt PSCSI} & A physical SCSI device \\ 5.9 {\tt user} & A user of the system \\ 5.10 {\tt debug} & A basic class for testing \\ 5.11 {\tt XSPolicy} & A class for handling Xen Security Policies \\ 5.12 @@ -74,6 +76,8 @@ VTPM.VM & VM.VTPMs & one-to-many\\ 5.13 console.VM & VM.consoles & one-to-many\\ 5.14 DPCI.VM & VM.DPCIs & one-to-many\\ 5.15 PPCI.host & host.PPCIs & one-to-many\\ 5.16 +DSCSI.VM & VM.DSCSIs & one-to-many\\ 5.17 +PSCSI.host & host.PSCSIs & one-to-many\\ 5.18 host.resident\_VMs & VM.resident\_on & many-to-one\\ 5.19 host.host\_CPUs & host\_cpu.host & many-to-one\\ 5.20 \hline 5.21 @@ -1407,6 +1411,7 @@ Quals & Field & Type & Description \\ 5.22 $\mathit{RO}_\mathit{run}$ & {\tt crash\_dumps} & (crashdump ref) Set & crash dumps associated with this VM \\ 5.23 $\mathit{RO}_\mathit{run}$ & {\tt VTPMs} & (VTPM ref) Set & virtual TPMs \\ 5.24 $\mathit{RO}_\mathit{run}$ & {\tt DPCIs} & (DPCI ref) Set & pass-through PCI devices \\ 5.25 +$\mathit{RO}_\mathit{run}$ & {\tt DSCSIs} & (DSCSI ref) Set & half-virtualized SCSI devices \\ 5.26 $\mathit{RW}$ & {\tt PV/bootloader} & string & name of or path to bootloader \\ 5.27 $\mathit{RW}$ & {\tt PV/kernel} & string & path to the kernel \\ 5.28 $\mathit{RW}$ & {\tt PV/ramdisk} & string & path to the initrd \\ 5.29 @@ -3450,6 +3455,38 @@ value of the field 5.30 \vspace{0.3cm} 5.31 \vspace{0.3cm} 5.32 \vspace{0.3cm} 5.33 +\subsubsection{RPC name:~get\_DSCSIs} 5.34 + 5.35 +{\bf Overview:} 5.36 +Get the DSCSIs field of the given VM. 5.37 + 5.38 + \noindent {\bf Signature:} 5.39 +\begin{verbatim} ((DSCSI ref) Set) get_DSCSIs (session_id s, VM ref self)\end{verbatim} 5.40 + 5.41 + 5.42 +\noindent{\bf Arguments:} 5.43 + 5.44 + 5.45 +\vspace{0.3cm} 5.46 +\begin{tabular}{|c|c|p{7cm}|} 5.47 + \hline 5.48 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.49 +{\tt VM ref } & self & reference to the object \\ \hline 5.50 + 5.51 +\end{tabular} 5.52 + 5.53 +\vspace{0.3cm} 5.54 + 5.55 + \noindent {\bf Return Type:} 5.56 +{\tt 5.57 +(DSCSI ref) Set 5.58 +} 5.59 + 5.60 + 5.61 +value of the field 5.62 +\vspace{0.3cm} 5.63 +\vspace{0.3cm} 5.64 +\vspace{0.3cm} 5.65 \subsubsection{RPC name:~get\_PV\_bootloader} 5.66 5.67 {\bf Overview:} 5.68 @@ -5518,6 +5555,7 @@ Quals & Field & Type & Description \\ 5.69 $\mathit{RW}$ & {\tt crash\_dump\_sr} & SR ref & The SR in which VDIs for crash dumps are created \\ 5.70 $\mathit{RO}_\mathit{run}$ & {\tt PBDs} & (PBD ref) Set & physical blockdevices \\ 5.71 $\mathit{RO}_\mathit{run}$ & {\tt PPCIs} & (PPCI ref) Set & physical PCI devices \\ 5.72 +$\mathit{RO}_\mathit{run}$ & {\tt PSCSIs} & (PSCSI ref) Set & physical SCSI devices \\ 5.73 $\mathit{RO}_\mathit{run}$ & {\tt host\_CPUs} & (host\_cpu ref) Set & The physical CPUs on this host \\ 5.74 $\mathit{RO}_\mathit{run}$ & {\tt metrics} & host\_metrics ref & metrics associated with this host \\ 5.75 \hline 5.76 @@ -6844,6 +6882,38 @@ value of the field 5.77 \vspace{0.3cm} 5.78 \vspace{0.3cm} 5.79 \vspace{0.3cm} 5.80 +\subsubsection{RPC name:~get\_PSCSIs} 5.81 + 5.82 +{\bf Overview:} 5.83 +Get the PSCSIs field of the given host. 5.84 + 5.85 + \noindent {\bf Signature:} 5.86 +\begin{verbatim} ((PSCSI ref) Set) get_PSCSIs (session_id s, host ref self)\end{verbatim} 5.87 + 5.88 + 5.89 +\noindent{\bf Arguments:} 5.90 + 5.91 + 5.92 +\vspace{0.3cm} 5.93 +\begin{tabular}{|c|c|p{7cm}|} 5.94 + \hline 5.95 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.96 +{\tt host ref } & self & reference to the object \\ \hline 5.97 + 5.98 +\end{tabular} 5.99 + 5.100 +\vspace{0.3cm} 5.101 + 5.102 + \noindent {\bf Return Type:} 5.103 +{\tt 5.104 +(PSCSI ref) Set 5.105 +} 5.106 + 5.107 + 5.108 +value of the field 5.109 +\vspace{0.3cm} 5.110 +\vspace{0.3cm} 5.111 +\vspace{0.3cm} 5.112 \subsubsection{RPC name:~get\_host\_CPUs} 5.113 5.114 {\bf Overview:} 5.115 @@ -15723,6 +15793,1096 @@ all fields from the object 5.116 5.117 \vspace{1cm} 5.118 \newpage 5.119 +\section{Class: DSCSI} 5.120 +\subsection{Fields for class: DSCSI} 5.121 +\begin{longtable}{|lllp{0.38\textwidth}|} 5.122 +\hline 5.123 +\multicolumn{1}{|l}{Name} & \multicolumn{3}{l|}{\bf DSCSI} \\ 5.124 +\multicolumn{1}{|l}{Description} & \multicolumn{3}{l|}{\parbox{11cm}{\em A 5.125 +half-virtualized SCSI device.}} \\ 5.126 +\hline 5.127 +Quals & Field & Type & Description \\ 5.128 +\hline 5.129 +$\mathit{RO}_\mathit{run}$ & {\tt uuid} & string & unique identifier/object reference \\ 5.130 +$\mathit{RO}_\mathit{inst}$ & {\tt VM} & VM ref & the virtual machine \\ 5.131 +$\mathit{RO}_\mathit{inst}$ & {\tt PSCSI} & PSCSI ref & the physical SCSI device \\ 5.132 +$\mathit{RO}_\mathit{run}$ & {\tt virtual\_host} & int & the virtual host number \\ 5.133 +$\mathit{RO}_\mathit{run}$ & {\tt virtual\_channel} & int & the virtual channel number \\ 5.134 +$\mathit{RO}_\mathit{run}$ & {\tt virtual\_target} & int & the virtual target number \\ 5.135 +$\mathit{RO}_\mathit{run}$ & {\tt virtual\_lun} & int & the virtual logical unit number \\ 5.136 +$\mathit{RO}_\mathit{inst}$ & {\tt virtual\_HCTL} & string & the virtual HCTL \\ 5.137 +$\mathit{RO}_\mathit{run}$ & {\tt runtime\_properties} & (string $\rightarrow$ string) Map & Device runtime properties \\ 5.138 +\hline 5.139 +\end{longtable} 5.140 +\subsection{RPCs associated with class: DSCSI} 5.141 +\subsubsection{RPC name:~get\_all} 5.142 + 5.143 +{\bf Overview:} 5.144 +Return a list of all the DSCSIs known to the system. 5.145 + 5.146 + \noindent {\bf Signature:} 5.147 +\begin{verbatim} ((DSCSI ref) Set) get_all (session_id s)\end{verbatim} 5.148 + 5.149 + 5.150 +\vspace{0.3cm} 5.151 + 5.152 + \noindent {\bf Return Type:} 5.153 +{\tt 5.154 +(DSCSI ref) Set 5.155 +} 5.156 + 5.157 + 5.158 +references to all objects 5.159 +\vspace{0.3cm} 5.160 +\vspace{0.3cm} 5.161 +\vspace{0.3cm} 5.162 +\subsubsection{RPC name:~get\_uuid} 5.163 + 5.164 +{\bf Overview:} 5.165 +Get the uuid field of the given DSCSI. 5.166 + 5.167 + \noindent {\bf Signature:} 5.168 +\begin{verbatim} string get_uuid (session_id s, DSCSI ref self)\end{verbatim} 5.169 + 5.170 + 5.171 +\noindent{\bf Arguments:} 5.172 + 5.173 + 5.174 +\vspace{0.3cm} 5.175 +\begin{tabular}{|c|c|p{7cm}|} 5.176 + \hline 5.177 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.178 +{\tt DSCSI ref } & self & reference to the object \\ \hline 5.179 + 5.180 +\end{tabular} 5.181 + 5.182 +\vspace{0.3cm} 5.183 + 5.184 + \noindent {\bf Return Type:} 5.185 +{\tt 5.186 +string 5.187 +} 5.188 + 5.189 + 5.190 +value of the field 5.191 +\vspace{0.3cm} 5.192 +\vspace{0.3cm} 5.193 +\vspace{0.3cm} 5.194 +\subsubsection{RPC name:~get\_VM} 5.195 + 5.196 +{\bf Overview:} 5.197 +Get the VM field of the given DSCSI. 5.198 + 5.199 + \noindent {\bf Signature:} 5.200 +\begin{verbatim} (VM ref) get_VM (session_id s, DSCSI ref self)\end{verbatim} 5.201 + 5.202 + 5.203 +\noindent{\bf Arguments:} 5.204 + 5.205 + 5.206 +\vspace{0.3cm} 5.207 +\begin{tabular}{|c|c|p{7cm}|} 5.208 + \hline 5.209 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.210 +{\tt DSCSI ref } & self & reference to the object \\ \hline 5.211 + 5.212 +\end{tabular} 5.213 + 5.214 +\vspace{0.3cm} 5.215 + 5.216 + \noindent {\bf Return Type:} 5.217 +{\tt 5.218 +VM ref 5.219 +} 5.220 + 5.221 + 5.222 +value of the field 5.223 +\vspace{0.3cm} 5.224 +\vspace{0.3cm} 5.225 +\vspace{0.3cm} 5.226 +\subsubsection{RPC name:~get\_PSCSI} 5.227 + 5.228 +{\bf Overview:} 5.229 +Get the PSCSI field of the given DSCSI. 5.230 + 5.231 + \noindent {\bf Signature:} 5.232 +\begin{verbatim} (PSCSI ref) get_PSCSI (session_id s, DSCSI ref self)\end{verbatim} 5.233 + 5.234 + 5.235 +\noindent{\bf Arguments:} 5.236 + 5.237 + 5.238 +\vspace{0.3cm} 5.239 +\begin{tabular}{|c|c|p{7cm}|} 5.240 + \hline 5.241 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.242 +{\tt DSCSI ref } & self & reference to the object \\ \hline 5.243 + 5.244 +\end{tabular} 5.245 + 5.246 +\vspace{0.3cm} 5.247 + 5.248 + \noindent {\bf Return Type:} 5.249 +{\tt 5.250 +PSCSI ref 5.251 +} 5.252 + 5.253 + 5.254 +value of the field 5.255 +\vspace{0.3cm} 5.256 +\vspace{0.3cm} 5.257 +\vspace{0.3cm} 5.258 +\subsubsection{RPC name:~get\_virtual\_host} 5.259 + 5.260 +{\bf Overview:} 5.261 +Get the virtual\_host field of the given DSCSI. 5.262 + 5.263 + \noindent {\bf Signature:} 5.264 +\begin{verbatim} int get_virtual_host (session_id s, DSCSI ref self)\end{verbatim} 5.265 + 5.266 + 5.267 +\noindent{\bf Arguments:} 5.268 + 5.269 + 5.270 +\vspace{0.3cm} 5.271 +\begin{tabular}{|c|c|p{7cm}|} 5.272 + \hline 5.273 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.274 +{\tt DSCSI ref } & self & reference to the object \\ \hline 5.275 + 5.276 +\end{tabular} 5.277 + 5.278 +\vspace{0.3cm} 5.279 + 5.280 + \noindent {\bf Return Type:} 5.281 +{\tt 5.282 +int 5.283 +} 5.284 + 5.285 + 5.286 +value of the field 5.287 +\vspace{0.3cm} 5.288 +\vspace{0.3cm} 5.289 +\vspace{0.3cm} 5.290 +\subsubsection{RPC name:~get\_virtual\_channel} 5.291 + 5.292 +{\bf Overview:} 5.293 +Get the virtual\_channel field of the given DSCSI. 5.294 + 5.295 + \noindent {\bf Signature:} 5.296 +\begin{verbatim} int get_virtual_channel (session_id s, DSCSI ref self)\end{verbatim} 5.297 + 5.298 + 5.299 +\noindent{\bf Arguments:} 5.300 + 5.301 + 5.302 +\vspace{0.3cm} 5.303 +\begin{tabular}{|c|c|p{7cm}|} 5.304 + \hline 5.305 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.306 +{\tt DSCSI ref } & self & reference to the object \\ \hline 5.307 + 5.308 +\end{tabular} 5.309 + 5.310 +\vspace{0.3cm} 5.311 + 5.312 + \noindent {\bf Return Type:} 5.313 +{\tt 5.314 +int 5.315 +} 5.316 + 5.317 + 5.318 +value of the field 5.319 +\vspace{0.3cm} 5.320 +\vspace{0.3cm} 5.321 +\vspace{0.3cm} 5.322 +\subsubsection{RPC name:~get\_virtual\_target} 5.323 + 5.324 +{\bf Overview:} 5.325 +Get the virtual\_target field of the given DSCSI. 5.326 + 5.327 + \noindent {\bf Signature:} 5.328 +\begin{verbatim} int get_virtual_target (session_id s, DSCSI ref self)\end{verbatim} 5.329 + 5.330 + 5.331 +\noindent{\bf Arguments:} 5.332 + 5.333 + 5.334 +\vspace{0.3cm} 5.335 +\begin{tabular}{|c|c|p{7cm}|} 5.336 + \hline 5.337 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.338 +{\tt DSCSI ref } & self & reference to the object \\ \hline 5.339 + 5.340 +\end{tabular} 5.341 + 5.342 +\vspace{0.3cm} 5.343 + 5.344 + \noindent {\bf Return Type:} 5.345 +{\tt 5.346 +int 5.347 +} 5.348 + 5.349 + 5.350 +value of the field 5.351 +\vspace{0.3cm} 5.352 +\vspace{0.3cm} 5.353 +\vspace{0.3cm} 5.354 +\subsubsection{RPC name:~get\_virtual\_lun} 5.355 + 5.356 +{\bf Overview:} 5.357 +Get the virtual\_lun field of the given DSCSI. 5.358 + 5.359 + \noindent {\bf Signature:} 5.360 +\begin{verbatim} int get_virtual_lun (session_id s, DSCSI ref self)\end{verbatim} 5.361 + 5.362 + 5.363 +\noindent{\bf Arguments:} 5.364 + 5.365 + 5.366 +\vspace{0.3cm} 5.367 +\begin{tabular}{|c|c|p{7cm}|} 5.368 + \hline 5.369 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.370 +{\tt DSCSI ref } & self & reference to the object \\ \hline 5.371 + 5.372 +\end{tabular} 5.373 + 5.374 +\vspace{0.3cm} 5.375 + 5.376 + \noindent {\bf Return Type:} 5.377 +{\tt 5.378 +int 5.379 +} 5.380 + 5.381 + 5.382 +value of the field 5.383 +\vspace{0.3cm} 5.384 +\vspace{0.3cm} 5.385 +\vspace{0.3cm} 5.386 +\subsubsection{RPC name:~get\_virtual\_HCTL} 5.387 + 5.388 +{\bf Overview:} 5.389 +Get the virtual\_HCTL field of the given DSCSI. 5.390 + 5.391 + \noindent {\bf Signature:} 5.392 +\begin{verbatim} string get_virtual_HCTL (session_id s, DSCSI ref self)\end{verbatim} 5.393 + 5.394 + 5.395 +\noindent{\bf Arguments:} 5.396 + 5.397 + 5.398 +\vspace{0.3cm} 5.399 +\begin{tabular}{|c|c|p{7cm}|} 5.400 + \hline 5.401 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.402 +{\tt DSCSI ref } & self & reference to the object \\ \hline 5.403 + 5.404 +\end{tabular} 5.405 + 5.406 +\vspace{0.3cm} 5.407 + 5.408 + \noindent {\bf Return Type:} 5.409 +{\tt 5.410 +string 5.411 +} 5.412 + 5.413 + 5.414 +value of the field 5.415 +\vspace{0.3cm} 5.416 +\vspace{0.3cm} 5.417 +\vspace{0.3cm} 5.418 +\subsubsection{RPC name:~get\_runtime\_properties} 5.419 + 5.420 +{\bf Overview:} 5.421 +Get the runtime\_properties field of the given DSCSI. 5.422 + 5.423 + \noindent {\bf Signature:} 5.424 +\begin{verbatim} ((string -> string) Map) get_runtime_properties (session_id s, DSCSI ref self)\end{verbatim} 5.425 + 5.426 + 5.427 +\noindent{\bf Arguments:} 5.428 + 5.429 + 5.430 +\vspace{0.3cm} 5.431 +\begin{tabular}{|c|c|p{7cm}|} 5.432 + \hline 5.433 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.434 +{\tt DSCSI ref } & self & reference to the object \\ \hline 5.435 + 5.436 +\end{tabular} 5.437 + 5.438 +\vspace{0.3cm} 5.439 + 5.440 + \noindent {\bf Return Type:} 5.441 +{\tt 5.442 +(string $\rightarrow$ string) Map 5.443 +} 5.444 + 5.445 + 5.446 +value of the field 5.447 +\vspace{0.3cm} 5.448 +\vspace{0.3cm} 5.449 +\vspace{0.3cm} 5.450 +\subsubsection{RPC name:~create} 5.451 + 5.452 +{\bf Overview:} 5.453 +Create a new DSCSI instance, and return its handle. 5.454 + 5.455 + \noindent {\bf Signature:} 5.456 +\begin{verbatim} (DSCSI ref) create (session_id s, DSCSI record args)\end{verbatim} 5.457 + 5.458 + 5.459 +\noindent{\bf Arguments:} 5.460 + 5.461 + 5.462 +\vspace{0.3cm} 5.463 +\begin{tabular}{|c|c|p{7cm}|} 5.464 + \hline 5.465 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.466 +{\tt DSCSI record } & args & All constructor arguments \\ \hline 5.467 + 5.468 +\end{tabular} 5.469 + 5.470 +\vspace{0.3cm} 5.471 + 5.472 + \noindent {\bf Return Type:} 5.473 +{\tt 5.474 +DSCSI ref 5.475 +} 5.476 + 5.477 + 5.478 +reference to the newly created object 5.479 +\vspace{0.3cm} 5.480 +\vspace{0.3cm} 5.481 +\vspace{0.3cm} 5.482 +\subsubsection{RPC name:~destroy} 5.483 + 5.484 +{\bf Overview:} 5.485 +Destroy the specified DSCSI instance. 5.486 + 5.487 + \noindent {\bf Signature:} 5.488 +\begin{verbatim} void destroy (session_id s, DSCSI ref self)\end{verbatim} 5.489 + 5.490 + 5.491 +\noindent{\bf Arguments:} 5.492 + 5.493 + 5.494 +\vspace{0.3cm} 5.495 +\begin{tabular}{|c|c|p{7cm}|} 5.496 + \hline 5.497 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.498 +{\tt DSCSI ref } & self & reference to the object \\ \hline 5.499 + 5.500 +\end{tabular} 5.501 + 5.502 +\vspace{0.3cm} 5.503 + 5.504 + \noindent {\bf Return Type:} 5.505 +{\tt 5.506 +void 5.507 +} 5.508 + 5.509 + 5.510 +\vspace{0.3cm} 5.511 +\vspace{0.3cm} 5.512 +\vspace{0.3cm} 5.513 +\subsubsection{RPC name:~get\_by\_uuid} 5.514 + 5.515 +{\bf Overview:} 5.516 +Get a reference to the DSCSI instance with the specified UUID. 5.517 + 5.518 + \noindent {\bf Signature:} 5.519 +\begin{verbatim} (DSCSI ref) get_by_uuid (session_id s, string uuid)\end{verbatim} 5.520 + 5.521 + 5.522 +\noindent{\bf Arguments:} 5.523 + 5.524 + 5.525 +\vspace{0.3cm} 5.526 +\begin{tabular}{|c|c|p{7cm}|} 5.527 + \hline 5.528 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.529 +{\tt string } & uuid & UUID of object to return \\ \hline 5.530 + 5.531 +\end{tabular} 5.532 + 5.533 +\vspace{0.3cm} 5.534 + 5.535 + \noindent {\bf Return Type:} 5.536 +{\tt 5.537 +DSCSI ref 5.538 +} 5.539 + 5.540 + 5.541 +reference to the object 5.542 +\vspace{0.3cm} 5.543 +\vspace{0.3cm} 5.544 +\vspace{0.3cm} 5.545 +\subsubsection{RPC name:~get\_record} 5.546 + 5.547 +{\bf Overview:} 5.548 +Get a record containing the current state of the given DSCSI. 5.549 + 5.550 + \noindent {\bf Signature:} 5.551 +\begin{verbatim} (DSCSI record) get_record (session_id s, DSCSI ref self)\end{verbatim} 5.552 + 5.553 + 5.554 +\noindent{\bf Arguments:} 5.555 + 5.556 + 5.557 +\vspace{0.3cm} 5.558 +\begin{tabular}{|c|c|p{7cm}|} 5.559 + \hline 5.560 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.561 +{\tt DSCSI ref } & self & reference to the object \\ \hline 5.562 + 5.563 +\end{tabular} 5.564 + 5.565 +\vspace{0.3cm} 5.566 + 5.567 + \noindent {\bf Return Type:} 5.568 +{\tt 5.569 +DSCSI record 5.570 +} 5.571 + 5.572 + 5.573 +all fields from the object 5.574 +\vspace{0.3cm} 5.575 +\vspace{0.3cm} 5.576 +\vspace{0.3cm} 5.577 + 5.578 +\vspace{1cm} 5.579 +\newpage 5.580 +\section{Class: PSCSI} 5.581 +\subsection{Fields for class: PSCSI} 5.582 +\begin{longtable}{|lllp{0.38\textwidth}|} 5.583 +\hline 5.584 +\multicolumn{1}{|l}{Name} & \multicolumn{3}{l|}{\bf PSCSI} \\ 5.585 +\multicolumn{1}{|l}{Description} & \multicolumn{3}{l|}{\parbox{11cm}{\em A 5.586 +physical SCSI device.}} \\ 5.587 +\hline 5.588 +Quals & Field & Type & Description \\ 5.589 +\hline 5.590 +$\mathit{RO}_\mathit{run}$ & {\tt uuid} & string & unique identifier/object reference \\ 5.591 +$\mathit{RO}_\mathit{run}$ & {\tt host} & host ref & the physical machine to which this PSCSI is connected \\ 5.592 +$\mathit{RO}_\mathit{run}$ & {\tt physical\_host} & int & the physical host number \\ 5.593 +$\mathit{RO}_\mathit{run}$ & {\tt physical\_channel} & int & the physical channel number \\ 5.594 +$\mathit{RO}_\mathit{run}$ & {\tt physical\_target} & int & the physical target number \\ 5.595 +$\mathit{RO}_\mathit{run}$ & {\tt physical\_lun} & int & the physical logical unit number \\ 5.596 +$\mathit{RO}_\mathit{run}$ & {\tt physical\_HCTL} & string & the physical HCTL \\ 5.597 +$\mathit{RO}_\mathit{run}$ & {\tt vendor\_name} & string & the vendor name \\ 5.598 +$\mathit{RO}_\mathit{run}$ & {\tt model} & string & the model \\ 5.599 +$\mathit{RO}_\mathit{run}$ & {\tt type\_id} & int & the SCSI type ID \\ 5.600 +$\mathit{RO}_\mathit{run}$ & {\tt type} & string & the SCSI type \\ 5.601 +$\mathit{RO}_\mathit{run}$ & {\tt dev\_name} & string & the SCSI device name (e.g. sda or st0) \\ 5.602 +$\mathit{RO}_\mathit{run}$ & {\tt sg\_name} & string & the SCSI generic device name (e.g. sg0) \\ 5.603 +$\mathit{RO}_\mathit{run}$ & {\tt revision} & string & the revision \\ 5.604 +$\mathit{RO}_\mathit{run}$ & {\tt scsi\_id} & string & the SCSI ID \\ 5.605 +$\mathit{RO}_\mathit{run}$ & {\tt scsi\_level} & int & the SCSI level \\ 5.606 +\hline 5.607 +\end{longtable} 5.608 +\subsection{RPCs associated with class: PSCSI} 5.609 +\subsubsection{RPC name:~get\_all} 5.610 + 5.611 +{\bf Overview:} 5.612 +Return a list of all the PSCSIs known to the system. 5.613 + 5.614 + \noindent {\bf Signature:} 5.615 +\begin{verbatim} ((PSCSI ref) Set) get_all (session_id s)\end{verbatim} 5.616 + 5.617 + 5.618 +\vspace{0.3cm} 5.619 + 5.620 + \noindent {\bf Return Type:} 5.621 +{\tt 5.622 +(PSCSI ref) Set 5.623 +} 5.624 + 5.625 + 5.626 +references to all objects 5.627 +\vspace{0.3cm} 5.628 +\vspace{0.3cm} 5.629 +\vspace{0.3cm} 5.630 +\subsubsection{RPC name:~get\_uuid} 5.631 + 5.632 +{\bf Overview:} 5.633 +Get the uuid field of the given PSCSI. 5.634 + 5.635 + \noindent {\bf Signature:} 5.636 +\begin{verbatim} string get_uuid (session_id s, PSCSI ref self)\end{verbatim} 5.637 + 5.638 + 5.639 +\noindent{\bf Arguments:} 5.640 + 5.641 + 5.642 +\vspace{0.3cm} 5.643 +\begin{tabular}{|c|c|p{7cm}|} 5.644 + \hline 5.645 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.646 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.647 + 5.648 +\end{tabular} 5.649 + 5.650 +\vspace{0.3cm} 5.651 + 5.652 + \noindent {\bf Return Type:} 5.653 +{\tt 5.654 +string 5.655 +} 5.656 + 5.657 + 5.658 +value of the field 5.659 +\vspace{0.3cm} 5.660 +\vspace{0.3cm} 5.661 +\vspace{0.3cm} 5.662 +\subsubsection{RPC name:~get\_host} 5.663 + 5.664 +{\bf Overview:} 5.665 +Get the host field of the given PSCSI. 5.666 + 5.667 + \noindent {\bf Signature:} 5.668 +\begin{verbatim} (host ref) get_host (session_id s, PSCSI ref self)\end{verbatim} 5.669 + 5.670 + 5.671 +\noindent{\bf Arguments:} 5.672 + 5.673 + 5.674 +\vspace{0.3cm} 5.675 +\begin{tabular}{|c|c|p{7cm}|} 5.676 + \hline 5.677 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.678 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.679 + 5.680 +\end{tabular} 5.681 + 5.682 +\vspace{0.3cm} 5.683 + 5.684 + \noindent {\bf Return Type:} 5.685 +{\tt 5.686 +host ref 5.687 +} 5.688 + 5.689 + 5.690 +value of the field 5.691 +\vspace{0.3cm} 5.692 +\vspace{0.3cm} 5.693 +\vspace{0.3cm} 5.694 +\subsubsection{RPC name:~get\_physical\_host} 5.695 + 5.696 +{\bf Overview:} 5.697 +Get the physical\_host field of the given PSCSI. 5.698 + 5.699 + \noindent {\bf Signature:} 5.700 +\begin{verbatim} int get_physical_host (session_id s, PSCSI ref self)\end{verbatim} 5.701 + 5.702 + 5.703 +\noindent{\bf Arguments:} 5.704 + 5.705 + 5.706 +\vspace{0.3cm} 5.707 +\begin{tabular}{|c|c|p{7cm}|} 5.708 + \hline 5.709 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.710 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.711 + 5.712 +\end{tabular} 5.713 + 5.714 +\vspace{0.3cm} 5.715 + 5.716 + \noindent {\bf Return Type:} 5.717 +{\tt 5.718 +int 5.719 +} 5.720 + 5.721 + 5.722 +value of the field 5.723 +\vspace{0.3cm} 5.724 +\vspace{0.3cm} 5.725 +\vspace{0.3cm} 5.726 +\subsubsection{RPC name:~get\_physical\_channel} 5.727 + 5.728 +{\bf Overview:} 5.729 +Get the physical\_channel field of the given PSCSI. 5.730 + 5.731 + \noindent {\bf Signature:} 5.732 +\begin{verbatim} int get_physical_channel (session_id s, PSCSI ref self)\end{verbatim} 5.733 + 5.734 + 5.735 +\noindent{\bf Arguments:} 5.736 + 5.737 + 5.738 +\vspace{0.3cm} 5.739 +\begin{tabular}{|c|c|p{7cm}|} 5.740 + \hline 5.741 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.742 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.743 + 5.744 +\end{tabular} 5.745 + 5.746 +\vspace{0.3cm} 5.747 + 5.748 + \noindent {\bf Return Type:} 5.749 +{\tt 5.750 +int 5.751 +} 5.752 + 5.753 + 5.754 +value of the field 5.755 +\vspace{0.3cm} 5.756 +\vspace{0.3cm} 5.757 +\vspace{0.3cm} 5.758 +\subsubsection{RPC name:~get\_physical\_target} 5.759 + 5.760 +{\bf Overview:} 5.761 +Get the physical\_target field of the given PSCSI. 5.762 + 5.763 + \noindent {\bf Signature:} 5.764 +\begin{verbatim} int get_physical_target (session_id s, PSCSI ref self)\end{verbatim} 5.765 + 5.766 + 5.767 +\noindent{\bf Arguments:} 5.768 + 5.769 + 5.770 +\vspace{0.3cm} 5.771 +\begin{tabular}{|c|c|p{7cm}|} 5.772 + \hline 5.773 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.774 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.775 + 5.776 +\end{tabular} 5.777 + 5.778 +\vspace{0.3cm} 5.779 + 5.780 + \noindent {\bf Return Type:} 5.781 +{\tt 5.782 +int 5.783 +} 5.784 + 5.785 + 5.786 +value of the field 5.787 +\vspace{0.3cm} 5.788 +\vspace{0.3cm} 5.789 +\vspace{0.3cm} 5.790 +\subsubsection{RPC name:~get\_physical\_lun} 5.791 + 5.792 +{\bf Overview:} 5.793 +Get the physical\_lun field of the given PSCSI. 5.794 + 5.795 + \noindent {\bf Signature:} 5.796 +\begin{verbatim} int get_physical_lun (session_id s, PSCSI ref self)\end{verbatim} 5.797 + 5.798 + 5.799 +\noindent{\bf Arguments:} 5.800 + 5.801 + 5.802 +\vspace{0.3cm} 5.803 +\begin{tabular}{|c|c|p{7cm}|} 5.804 + \hline 5.805 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.806 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.807 + 5.808 +\end{tabular} 5.809 + 5.810 +\vspace{0.3cm} 5.811 + 5.812 + \noindent {\bf Return Type:} 5.813 +{\tt 5.814 +int 5.815 +} 5.816 + 5.817 + 5.818 +value of the field 5.819 +\vspace{0.3cm} 5.820 +\vspace{0.3cm} 5.821 +\vspace{0.3cm} 5.822 +\subsubsection{RPC name:~get\_physical\_HCTL} 5.823 + 5.824 +{\bf Overview:} 5.825 +Get the physical\_HCTL field of the given PSCSI. 5.826 + 5.827 + \noindent {\bf Signature:} 5.828 +\begin{verbatim} string get_physical_HCTL (session_id s, PSCSI ref self)\end{verbatim} 5.829 + 5.830 + 5.831 +\noindent{\bf Arguments:} 5.832 + 5.833 + 5.834 +\vspace{0.3cm} 5.835 +\begin{tabular}{|c|c|p{7cm}|} 5.836 + \hline 5.837 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.838 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.839 + 5.840 +\end{tabular} 5.841 + 5.842 +\vspace{0.3cm} 5.843 + 5.844 + \noindent {\bf Return Type:} 5.845 +{\tt 5.846 +string 5.847 +} 5.848 + 5.849 + 5.850 +value of the field 5.851 +\vspace{0.3cm} 5.852 +\vspace{0.3cm} 5.853 +\vspace{0.3cm} 5.854 +\subsubsection{RPC name:~get\_vendor\_name} 5.855 + 5.856 +{\bf Overview:} 5.857 +Get the vendor\_name field of the given PSCSI. 5.858 + 5.859 + \noindent {\bf Signature:} 5.860 +\begin{verbatim} string get_vendor_name (session_id s, PSCSI ref self)\end{verbatim} 5.861 + 5.862 + 5.863 +\noindent{\bf Arguments:} 5.864 + 5.865 + 5.866 +\vspace{0.3cm} 5.867 +\begin{tabular}{|c|c|p{7cm}|} 5.868 + \hline 5.869 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.870 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.871 + 5.872 +\end{tabular} 5.873 + 5.874 +\vspace{0.3cm} 5.875 + 5.876 + \noindent {\bf Return Type:} 5.877 +{\tt 5.878 +string 5.879 +} 5.880 + 5.881 + 5.882 +value of the field 5.883 +\vspace{0.3cm} 5.884 +\vspace{0.3cm} 5.885 +\vspace{0.3cm} 5.886 +\subsubsection{RPC name:~get\_model} 5.887 + 5.888 +{\bf Overview:} 5.889 +Get the model field of the given PSCSI. 5.890 + 5.891 + \noindent {\bf Signature:} 5.892 +\begin{verbatim} string get_model (session_id s, PSCSI ref self)\end{verbatim} 5.893 + 5.894 + 5.895 +\noindent{\bf Arguments:} 5.896 + 5.897 + 5.898 +\vspace{0.3cm} 5.899 +\begin{tabular}{|c|c|p{7cm}|} 5.900 + \hline 5.901 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.902 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.903 + 5.904 +\end{tabular} 5.905 + 5.906 +\vspace{0.3cm} 5.907 + 5.908 + \noindent {\bf Return Type:} 5.909 +{\tt 5.910 +string 5.911 +} 5.912 + 5.913 + 5.914 +value of the field 5.915 +\vspace{0.3cm} 5.916 +\vspace{0.3cm} 5.917 +\vspace{0.3cm} 5.918 +\subsubsection{RPC name:~get\_type\_id} 5.919 + 5.920 +{\bf Overview:} 5.921 +Get the type\_id field of the given PSCSI. 5.922 + 5.923 + \noindent {\bf Signature:} 5.924 +\begin{verbatim} int get_type_id (session_id s, PSCSI ref self)\end{verbatim} 5.925 + 5.926 + 5.927 +\noindent{\bf Arguments:} 5.928 + 5.929 + 5.930 +\vspace{0.3cm} 5.931 +\begin{tabular}{|c|c|p{7cm}|} 5.932 + \hline 5.933 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.934 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.935 + 5.936 +\end{tabular} 5.937 + 5.938 +\vspace{0.3cm} 5.939 + 5.940 + \noindent {\bf Return Type:} 5.941 +{\tt 5.942 +int 5.943 +} 5.944 + 5.945 + 5.946 +value of the field 5.947 +\vspace{0.3cm} 5.948 +\vspace{0.3cm} 5.949 +\vspace{0.3cm} 5.950 +\subsubsection{RPC name:~get\_type} 5.951 + 5.952 +{\bf Overview:} 5.953 +Get the type field of the given PSCSI. 5.954 + 5.955 + \noindent {\bf Signature:} 5.956 +\begin{verbatim} string get_type (session_id s, PSCSI ref self)\end{verbatim} 5.957 + 5.958 + 5.959 +\noindent{\bf Arguments:} 5.960 + 5.961 + 5.962 +\vspace{0.3cm} 5.963 +\begin{tabular}{|c|c|p{7cm}|} 5.964 + \hline 5.965 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.966 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.967 + 5.968 +\end{tabular} 5.969 + 5.970 +\vspace{0.3cm} 5.971 + 5.972 + \noindent {\bf Return Type:} 5.973 +{\tt 5.974 +string 5.975 +} 5.976 + 5.977 + 5.978 +value of the field 5.979 +\vspace{0.3cm} 5.980 +\vspace{0.3cm} 5.981 +\vspace{0.3cm} 5.982 +\subsubsection{RPC name:~get\_dev\_name} 5.983 + 5.984 +{\bf Overview:} 5.985 +Get the dev\_name field of the given PSCSI. 5.986 + 5.987 + \noindent {\bf Signature:} 5.988 +\begin{verbatim} string get_dev_name (session_id s, PSCSI ref self)\end{verbatim} 5.989 + 5.990 + 5.991 +\noindent{\bf Arguments:} 5.992 + 5.993 + 5.994 +\vspace{0.3cm} 5.995 +\begin{tabular}{|c|c|p{7cm}|} 5.996 + \hline 5.997 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.998 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.999 + 5.1000 +\end{tabular} 5.1001 + 5.1002 +\vspace{0.3cm} 5.1003 + 5.1004 + \noindent {\bf Return Type:} 5.1005 +{\tt 5.1006 +string 5.1007 +} 5.1008 + 5.1009 + 5.1010 +value of the field 5.1011 +\vspace{0.3cm} 5.1012 +\vspace{0.3cm} 5.1013 +\vspace{0.3cm} 5.1014 +\subsubsection{RPC name:~get\_sg\_name} 5.1015 + 5.1016 +{\bf Overview:} 5.1017 +Get the sg\_name field of the given PSCSI. 5.1018 + 5.1019 + \noindent {\bf Signature:} 5.1020 +\begin{verbatim} string get_sg_name (session_id s, PSCSI ref self)\end{verbatim} 5.1021 + 5.1022 + 5.1023 +\noindent{\bf Arguments:} 5.1024 + 5.1025 + 5.1026 +\vspace{0.3cm} 5.1027 +\begin{tabular}{|c|c|p{7cm}|} 5.1028 + \hline 5.1029 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.1030 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.1031 + 5.1032 +\end{tabular} 5.1033 + 5.1034 +\vspace{0.3cm} 5.1035 + 5.1036 + \noindent {\bf Return Type:} 5.1037 +{\tt 5.1038 +string 5.1039 +} 5.1040 + 5.1041 + 5.1042 +value of the field 5.1043 +\vspace{0.3cm} 5.1044 +\vspace{0.3cm} 5.1045 +\vspace{0.3cm} 5.1046 +\subsubsection{RPC name:~get\_revision} 5.1047 + 5.1048 +{\bf Overview:} 5.1049 +Get the revision field of the given PSCSI. 5.1050 + 5.1051 + \noindent {\bf Signature:} 5.1052 +\begin{verbatim} string get_revision (session_id s, PSCSI ref self)\end{verbatim} 5.1053 + 5.1054 + 5.1055 +\noindent{\bf Arguments:} 5.1056 + 5.1057 + 5.1058 +\vspace{0.3cm} 5.1059 +\begin{tabular}{|c|c|p{7cm}|} 5.1060 + \hline 5.1061 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.1062 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.1063 + 5.1064 +\end{tabular} 5.1065 + 5.1066 +\vspace{0.3cm} 5.1067 + 5.1068 + \noindent {\bf Return Type:} 5.1069 +{\tt 5.1070 +string 5.1071 +} 5.1072 + 5.1073 + 5.1074 +value of the field 5.1075 +\vspace{0.3cm} 5.1076 +\vspace{0.3cm} 5.1077 +\vspace{0.3cm} 5.1078 +\subsubsection{RPC name:~get\_scsi\_id} 5.1079 + 5.1080 +{\bf Overview:} 5.1081 +Get the scsi\_id field of the given PSCSI. 5.1082 + 5.1083 + \noindent {\bf Signature:} 5.1084 +\begin{verbatim} string get_scsi_id (session_id s, PSCSI ref self)\end{verbatim} 5.1085 + 5.1086 + 5.1087 +\noindent{\bf Arguments:} 5.1088 + 5.1089 + 5.1090 +\vspace{0.3cm} 5.1091 +\begin{tabular}{|c|c|p{7cm}|} 5.1092 + \hline 5.1093 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.1094 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.1095 + 5.1096 +\end{tabular} 5.1097 + 5.1098 +\vspace{0.3cm} 5.1099 + 5.1100 + \noindent {\bf Return Type:} 5.1101 +{\tt 5.1102 +string 5.1103 +} 5.1104 + 5.1105 + 5.1106 +value of the field 5.1107 +\vspace{0.3cm} 5.1108 +\vspace{0.3cm} 5.1109 +\vspace{0.3cm} 5.1110 +\subsubsection{RPC name:~get\_scsi\_level} 5.1111 + 5.1112 +{\bf Overview:} 5.1113 +Get the scsi\_level field of the given PSCSI. 5.1114 + 5.1115 + \noindent {\bf Signature:} 5.1116 +\begin{verbatim} int get_scsi_level (session_id s, PSCSI ref self)\end{verbatim} 5.1117 + 5.1118 + 5.1119 +\noindent{\bf Arguments:} 5.1120 + 5.1121 + 5.1122 +\vspace{0.3cm} 5.1123 +\begin{tabular}{|c|c|p{7cm}|} 5.1124 + \hline 5.1125 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.1126 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.1127 + 5.1128 +\end{tabular} 5.1129 + 5.1130 +\vspace{0.3cm} 5.1131 + 5.1132 + \noindent {\bf Return Type:} 5.1133 +{\tt 5.1134 +int 5.1135 +} 5.1136 + 5.1137 + 5.1138 +value of the field 5.1139 +\vspace{0.3cm} 5.1140 +\vspace{0.3cm} 5.1141 +\vspace{0.3cm} 5.1142 +\subsubsection{RPC name:~get\_by\_uuid} 5.1143 + 5.1144 +{\bf Overview:} 5.1145 +Get a reference to the PSCSI instance with the specified UUID. 5.1146 + 5.1147 + \noindent {\bf Signature:} 5.1148 +\begin{verbatim} (PSCSI ref) get_by_uuid (session_id s, string uuid)\end{verbatim} 5.1149 + 5.1150 + 5.1151 +\noindent{\bf Arguments:} 5.1152 + 5.1153 + 5.1154 +\vspace{0.3cm} 5.1155 +\begin{tabular}{|c|c|p{7cm}|} 5.1156 + \hline 5.1157 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.1158 +{\tt string } & uuid & UUID of object to return \\ \hline 5.1159 + 5.1160 +\end{tabular} 5.1161 + 5.1162 +\vspace{0.3cm} 5.1163 + 5.1164 + \noindent {\bf Return Type:} 5.1165 +{\tt 5.1166 +PSCSI ref 5.1167 +} 5.1168 + 5.1169 + 5.1170 +reference to the object 5.1171 +\vspace{0.3cm} 5.1172 +\vspace{0.3cm} 5.1173 +\vspace{0.3cm} 5.1174 +\subsubsection{RPC name:~get\_record} 5.1175 + 5.1176 +{\bf Overview:} 5.1177 +Get a record containing the current state of the given PSCSI. 5.1178 + 5.1179 + \noindent {\bf Signature:} 5.1180 +\begin{verbatim} (PSCSI record) get_record (session_id s, PSCSI ref self)\end{verbatim} 5.1181 + 5.1182 + 5.1183 +\noindent{\bf Arguments:} 5.1184 + 5.1185 + 5.1186 +\vspace{0.3cm} 5.1187 +\begin{tabular}{|c|c|p{7cm}|} 5.1188 + \hline 5.1189 +{\bf type} & {\bf name} & {\bf description} \\ \hline 5.1190 +{\tt PSCSI ref } & self & reference to the object \\ \hline 5.1191 + 5.1192 +\end{tabular} 5.1193 + 5.1194 +\vspace{0.3cm} 5.1195 + 5.1196 + \noindent {\bf Return Type:} 5.1197 +{\tt 5.1198 +PSCSI record 5.1199 +} 5.1200 + 5.1201 + 5.1202 +all fields from the object 5.1203 +\vspace{0.3cm} 5.1204 +\vspace{0.3cm} 5.1205 +\vspace{0.3cm} 5.1206 + 5.1207 +\vspace{1cm} 5.1208 +\newpage 5.1209 \section{Class: user} 5.1210 \subsection{Fields for class: user} 5.1211 \begin{longtable}{|lllp{0.38\textwidth}|}
6.1 --- a/tools/blktap/drivers/block-qcow.c Wed Oct 22 11:38:22 2008 +0900 6.2 +++ b/tools/blktap/drivers/block-qcow.c Wed Oct 22 11:46:55 2008 +0900 6.3 @@ -734,8 +734,8 @@ static int tdqcow_open (struct disk_driv 6.4 6.5 DPRINTF("QCOW: Opening %s\n",name); 6.6 6.7 - o_flags = O_DIRECT | O_LARGEFILE | 6.8 - ((flags == TD_RDONLY) ? O_RDONLY : O_RDWR); 6.9 + /* Since we don't handle O_DIRECT correctly, don't use it */ 6.10 + o_flags = O_LARGEFILE | ((flags == TD_RDONLY) ? O_RDONLY : O_RDWR); 6.11 fd = open(name, o_flags); 6.12 if (fd < 0) { 6.13 DPRINTF("Unable to open %s (%d)\n",name,0 - errno); 6.14 @@ -1385,7 +1385,7 @@ static int tdqcow_get_parent_id(struct d 6.15 filename[len] = '\0'; 6.16 6.17 id->name = strdup(filename); 6.18 - id->drivertype = DISK_TYPE_QCOW; 6.19 + id->drivertype = DISK_TYPE_AIO; 6.20 err = 0; 6.21 out: 6.22 free(buf); 6.23 @@ -1397,17 +1397,15 @@ static int tdqcow_validate_parent(struct 6.24 { 6.25 struct stat stats; 6.26 uint64_t psize, csize; 6.27 - struct tdqcow_state *c = (struct tdqcow_state *)child->private; 6.28 - struct tdqcow_state *p = (struct tdqcow_state *)parent->private; 6.29 6.30 - if (stat(p->name, &stats)) 6.31 + if (stat(parent->name, &stats)) 6.32 return -EINVAL; 6.33 - if (get_filesize(p->name, &psize, &stats)) 6.34 + if (get_filesize(parent->name, &psize, &stats)) 6.35 return -EINVAL; 6.36 6.37 - if (stat(c->name, &stats)) 6.38 + if (stat(child->name, &stats)) 6.39 return -EINVAL; 6.40 - if (get_filesize(c->name, &csize, &stats)) 6.41 + if (get_filesize(child->name, &csize, &stats)) 6.42 return -EINVAL; 6.43 6.44 if (csize != psize)
7.1 --- a/tools/blktap/drivers/block-qcow2.c Wed Oct 22 11:38:22 2008 +0900 7.2 +++ b/tools/blktap/drivers/block-qcow2.c Wed Oct 22 11:46:55 2008 +0900 7.3 @@ -34,6 +34,7 @@ 7.4 #include "tapdisk.h" 7.5 #include "tapaio.h" 7.6 #include "bswap.h" 7.7 +#include "blk.h" 7.8 7.9 #define USE_AIO 7.10 7.11 @@ -1902,6 +1903,42 @@ repeat: 7.12 7.13 #endif 7.14 7.15 +static int get_filesize(char *filename, uint64_t *size, struct stat *st) 7.16 +{ 7.17 + int fd; 7.18 + QCowHeader header; 7.19 + 7.20 + /*Set to the backing file size*/ 7.21 + fd = open(filename, O_RDONLY); 7.22 + if (fd < 0) 7.23 + return -1; 7.24 + if (read(fd, &header, sizeof(header)) < sizeof(header)) { 7.25 + close(fd); 7.26 + return -1; 7.27 + } 7.28 + close(fd); 7.29 + 7.30 + be32_to_cpus(&header.magic); 7.31 + be32_to_cpus(&header.version); 7.32 + be64_to_cpus(&header.size); 7.33 + if (header.magic == QCOW_MAGIC && header.version == QCOW_VERSION) { 7.34 + *size = header.size >> SECTOR_SHIFT; 7.35 + return 0; 7.36 + } 7.37 + 7.38 + if(S_ISBLK(st->st_mode)) { 7.39 + fd = open(filename, O_RDONLY); 7.40 + if (fd < 0) 7.41 + return -1; 7.42 + if (blk_getimagesize(fd, size) != 0) { 7.43 + close(fd); 7.44 + return -1; 7.45 + } 7.46 + close(fd); 7.47 + } else *size = (st->st_size >> SECTOR_SHIFT); 7.48 + return 0; 7.49 +} 7.50 + 7.51 /** 7.52 * @return 7.53 * 0 if parent id successfully retrieved; 7.54 @@ -1916,7 +1953,7 @@ static int qcow_get_parent_id(struct dis 7.55 return TD_NO_PARENT; 7.56 7.57 id->name = strdup(s->backing_file); 7.58 - id->drivertype = DISK_TYPE_QCOW2; 7.59 + id->drivertype = DISK_TYPE_AIO; 7.60 7.61 return 0; 7.62 } 7.63 @@ -1924,15 +1961,22 @@ static int qcow_get_parent_id(struct dis 7.64 static int qcow_validate_parent(struct disk_driver *child, 7.65 struct disk_driver *parent, td_flag_t flags) 7.66 { 7.67 - struct BDRVQcowState *cs = (struct BDRVQcowState*) child->private; 7.68 - struct BDRVQcowState *ps = (struct BDRVQcowState*) parent->private; 7.69 + struct stat stats; 7.70 + uint64_t psize, csize; 7.71 + 7.72 + if (stat(parent->name, &stats)) 7.73 + return -EINVAL; 7.74 + if (get_filesize(parent->name, &psize, &stats)) 7.75 + return -EINVAL; 7.76 7.77 - if (ps->total_sectors != cs->total_sectors) { 7.78 - DPRINTF("qcow_validate_parent(): %#"PRIx64" != %#"PRIx64"\n", 7.79 - ps->total_sectors, cs->total_sectors); 7.80 + if (stat(child->name, &stats)) 7.81 + return -EINVAL; 7.82 + if (get_filesize(child->name, &csize, &stats)) 7.83 return -EINVAL; 7.84 - } 7.85 - 7.86 + 7.87 + if (csize != psize) 7.88 + return -EINVAL; 7.89 + 7.90 return 0; 7.91 } 7.92
8.1 --- a/tools/libxc/xc_domain.c Wed Oct 22 11:38:22 2008 +0900 8.2 +++ b/tools/libxc/xc_domain.c Wed Oct 22 11:46:55 2008 +0900 8.3 @@ -1049,6 +1049,18 @@ int xc_domain_get_machine_address_size(i 8.4 return rc == 0 ? domctl.u.address_size.size : rc; 8.5 } 8.6 8.7 +int xc_domain_suppress_spurious_page_faults(int xc, uint32_t domid) 8.8 +{ 8.9 + DECLARE_DOMCTL; 8.10 + 8.11 + memset(&domctl, 0, sizeof(domctl)); 8.12 + domctl.domain = domid; 8.13 + domctl.cmd = XEN_DOMCTL_suppress_spurious_page_faults; 8.14 + 8.15 + return do_domctl(xc, &domctl); 8.16 + 8.17 +} 8.18 + 8.19 /* 8.20 * Local variables: 8.21 * mode: C
9.1 --- a/tools/libxc/xenctrl.h Wed Oct 22 11:38:22 2008 +0900 9.2 +++ b/tools/libxc/xenctrl.h Wed Oct 22 11:46:55 2008 +0900 9.3 @@ -1103,6 +1103,9 @@ int xc_domain_set_machine_address_size(i 9.4 int xc_domain_get_machine_address_size(int handle, 9.5 uint32_t domid); 9.6 9.7 +int xc_domain_suppress_spurious_page_faults(int handle, 9.8 + uint32_t domid); 9.9 + 9.10 /* Set the target domain */ 9.11 int xc_domain_set_target(int xc_handle, 9.12 uint32_t domid,
10.1 --- a/tools/python/xen/lowlevel/xc/xc.c Wed Oct 22 11:38:22 2008 +0900 10.2 +++ b/tools/python/xen/lowlevel/xc/xc.c Wed Oct 22 11:46:55 2008 +0900 10.3 @@ -859,6 +859,21 @@ static PyObject *pyxc_dom_set_machine_ad 10.4 return zero; 10.5 } 10.6 10.7 +static PyObject *pyxc_dom_suppress_spurious_page_faults(XcObject *self, 10.8 + PyObject *args, 10.9 + PyObject *kwds) 10.10 +{ 10.11 + uint32_t dom; 10.12 + 10.13 + if (!PyArg_ParseTuple(args, "i", &dom)) 10.14 + return NULL; 10.15 + 10.16 + if (xc_domain_suppress_spurious_page_faults(self->xc_handle, dom) != 0) 10.17 + return pyxc_error_to_exception(); 10.18 + 10.19 + Py_INCREF(zero); 10.20 + return zero; 10.21 +} 10.22 #endif /* __i386__ || __x86_64__ */ 10.23 10.24 static PyObject *pyxc_hvm_build(XcObject *self, 10.25 @@ -1911,6 +1926,12 @@ static PyMethodDef pyxc_methods[] = { 10.26 "Set maximum machine address size for this domain.\n" 10.27 " dom [int]: Identifier of domain.\n" 10.28 " width [int]: Maximum machine address width.\n" }, 10.29 + 10.30 + { "domain_suppress_spurious_page_faults", 10.31 + (PyCFunction)pyxc_dom_suppress_spurious_page_faults, 10.32 + METH_VARARGS, "\n" 10.33 + "Do not propagate spurious page faults to this guest.\n" 10.34 + " dom [int]: Identifier of domain.\n" }, 10.35 #endif 10.36 10.37 { NULL, NULL, 0, NULL }
11.1 --- a/tools/python/xen/util/pci.py Wed Oct 22 11:38:22 2008 +0900 11.2 +++ b/tools/python/xen/util/pci.py Wed Oct 22 11:46:55 2008 +0900 11.3 @@ -12,8 +12,8 @@ import re 11.4 import types 11.5 import struct 11.6 import time 11.7 +from xen.util import utils 11.8 11.9 -PROC_MNT_PATH = '/proc/mounts' 11.10 PROC_PCI_PATH = '/proc/bus/pci/devices' 11.11 PROC_PCI_NUM_RESOURCES = 7 11.12 11.13 @@ -97,9 +97,6 @@ MSIX_SIZE_MASK = 0x7ff 11.14 # Global variable to store information from lspci 11.15 lspci_info = None 11.16 11.17 -# Global variable to store the sysfs mount point 11.18 -sysfs_mnt_point = None 11.19 - 11.20 #Calculate PAGE_SHIFT: number of bits to shift an address to get the page number 11.21 PAGE_SIZE = resource.getpagesize() 11.22 PAGE_SHIFT = 0 11.23 @@ -141,20 +138,8 @@ def parse_pci_name(pci_name_string): 11.24 11.25 11.26 def find_sysfs_mnt(): 11.27 - global sysfs_mnt_point 11.28 - if not sysfs_mnt_point is None: 11.29 - return sysfs_mnt_point 11.30 - 11.31 try: 11.32 - mounts_file = open(PROC_MNT_PATH,'r') 11.33 - 11.34 - for line in mounts_file: 11.35 - sline = line.split() 11.36 - if len(sline)<3: 11.37 - continue 11.38 - if sline[2]=='sysfs': 11.39 - sysfs_mnt_point= sline[1] 11.40 - return sysfs_mnt_point 11.41 + return utils.find_sysfs_mount() 11.42 except IOError, (errno, strerr): 11.43 raise PciDeviceParseError(('Failed to locate sysfs mount: %s: %s (%d)'% 11.44 (PROC_PCI_PATH, strerr, errno)))
12.1 --- a/tools/python/xen/util/utils.py Wed Oct 22 11:38:22 2008 +0900 12.2 +++ b/tools/python/xen/util/utils.py Wed Oct 22 11:46:55 2008 +0900 12.3 @@ -48,3 +48,29 @@ def daemonize(prog, args, stdin_tmpfile= 12.4 os.waitpid(pid, 0) 12.5 return daemon_pid 12.6 12.7 +# Global variable to store the sysfs mount point 12.8 +sysfs_mount_point = None 12.9 + 12.10 +PROC_MOUNTS_PATH = '/proc/mounts' 12.11 + 12.12 +def find_sysfs_mount(): 12.13 + global sysfs_mount_point 12.14 + 12.15 + if not sysfs_mount_point is None: 12.16 + return sysfs_mount_point 12.17 + 12.18 + try: 12.19 + mounts_file = open(PROC_MOUNTS_PATH, 'r') 12.20 + 12.21 + for line in mounts_file: 12.22 + sline = line.split() 12.23 + if len(sline) < 3: 12.24 + continue 12.25 + if sline[2] == 'sysfs': 12.26 + sysfs_mount_point= sline[1] 12.27 + return sysfs_mount_point 12.28 + except IOError, (errno, strerr): 12.29 + raise 12.30 + 12.31 + return None 12.32 +
13.1 --- a/tools/python/xen/util/vscsi_util.py Wed Oct 22 11:38:22 2008 +0900 13.2 +++ b/tools/python/xen/util/vscsi_util.py Wed Oct 22 11:46:55 2008 +0900 13.3 @@ -23,32 +23,40 @@ 13.4 """Support for VSCSI Devices. 13.5 """ 13.6 import os 13.7 +import os.path 13.8 import sys 13.9 import re 13.10 import string 13.11 +from xen.util import utils 13.12 13.13 -def _vscsi_hctl_block(name, scsi_devices): 13.14 - """ block-device name is convert into hctl. (e.g., '/dev/sda', 13.15 - '0:0:0:0')""" 13.16 +SYSFS_SCSI_PATH = "/bus/scsi/devices" 13.17 +SYSFS_SCSI_DEV_VENDOR_PATH = '/vendor' 13.18 +SYSFS_SCSI_DEV_MODEL_PATH = '/model' 13.19 +SYSFS_SCSI_DEV_TYPEID_PATH = '/type' 13.20 +SYSFS_SCSI_DEV_REVISION_PATH = '/rev' 13.21 +SYSFS_SCSI_DEV_SCSILEVEL_PATH = '/scsi_level' 13.22 + 13.23 +def _vscsi_get_devname_by(name, scsi_devices): 13.24 + """A device name is gotten by the HCTL. 13.25 + (e.g., '0:0:0:0' to '/dev/sda') 13.26 + """ 13.27 + 13.28 try: 13.29 search = re.compile(r'' + name + '$', re.DOTALL) 13.30 except Exception, e: 13.31 raise VmError("vscsi: invalid expression. " + str(e)) 13.32 - chk = 0 13.33 - for hctl, block, sg, scsi_id in scsi_devices: 13.34 + 13.35 + for hctl, devname, sg, scsi_id in scsi_devices: 13.36 if search.match(hctl): 13.37 - chk = 1 13.38 - break 13.39 + return (hctl, devname) 13.40 13.41 - if chk: 13.42 - return (hctl, block) 13.43 - else: 13.44 - return (None, None) 13.45 + return (None, None) 13.46 13.47 13.48 -def _vscsi_block_scsiid_to_hctl(phyname, scsi_devices): 13.49 - """ block-device name is convert into hctl. (e.g., '/dev/sda', 13.50 - '0:0:0:0')""" 13.51 +def _vscsi_get_hctl_by(phyname, scsi_devices): 13.52 + """An HCTL is gotten by the device name or the scsi_id. 13.53 + (e.g., '/dev/sda' to '0:0:0:0') 13.54 + """ 13.55 13.56 if re.match('/dev/sd[a-z]+([1-9]|1[0-5])?$', phyname): 13.57 # sd driver 13.58 @@ -63,71 +71,148 @@ def _vscsi_block_scsiid_to_hctl(phyname, 13.59 # scsi_id -gu 13.60 name = phyname 13.61 13.62 - chk = 0 13.63 - for hctl, block, sg, scsi_id in scsi_devices: 13.64 - if block == name: 13.65 - chk = 1 13.66 - break 13.67 - elif sg == name: 13.68 - chk = 1 13.69 - break 13.70 - elif scsi_id == name: 13.71 - chk = 1 13.72 - break 13.73 + for hctl, devname, sg, scsi_id in scsi_devices: 13.74 + if name in [devname, sg, scsi_id]: 13.75 + return (hctl, devname) 13.76 13.77 - if chk: 13.78 - return (hctl, block) 13.79 - else: 13.80 - return (None, None) 13.81 + return (None, None) 13.82 13.83 13.84 def vscsi_get_scsidevices(): 13.85 """ get all scsi devices""" 13.86 13.87 - SERCH_SCSI_PATH = "/sys/bus/scsi/devices" 13.88 devices = [] 13.89 + sysfs_mnt = utils.find_sysfs_mount() 13.90 13.91 - for dirpath, dirnames, files in os.walk(SERCH_SCSI_PATH): 13.92 + for dirpath, dirnames, files in os.walk(sysfs_mnt + SYSFS_SCSI_PATH): 13.93 for hctl in dirnames: 13.94 paths = os.path.join(dirpath, hctl) 13.95 - block = "-" 13.96 + devname = None 13.97 + sg = None 13.98 + scsi_id = None 13.99 for f in os.listdir(paths): 13.100 - if re.match('^block', f): 13.101 - os.chdir(os.path.join(paths, f)) 13.102 - block = os.path.basename(os.getcwd()) 13.103 - elif re.match('^tape', f): 13.104 - os.chdir(os.path.join(paths, f)) 13.105 - block = os.path.basename(os.getcwd()) 13.106 - elif re.match('^scsi_changer', f): 13.107 - os.chdir(os.path.join(paths, f)) 13.108 - block = os.path.basename(os.getcwd()) 13.109 - elif re.match('^onstream_tape', f): 13.110 - os.chdir(os.path.join(paths, f)) 13.111 - block = os.path.basename(os.getcwd()) 13.112 + realpath = os.path.realpath(os.path.join(paths, f)) 13.113 + if re.match('^block', f) or \ 13.114 + re.match('^tape', f) or \ 13.115 + re.match('^scsi_changer', f) or \ 13.116 + re.match('^onstream_tape', f): 13.117 + devname = os.path.basename(realpath) 13.118 13.119 if re.match('^scsi_generic', f): 13.120 - os.chdir(os.path.join(paths, f)) 13.121 - sg = os.path.basename(os.getcwd()) 13.122 + sg = os.path.basename(realpath) 13.123 lines = os.popen('/sbin/scsi_id -gu -s /class/scsi_generic/' + sg).read().split() 13.124 - if len(lines) == 0: 13.125 - scsi_id = '-' 13.126 - else: 13.127 + if len(lines): 13.128 scsi_id = lines[0] 13.129 13.130 - devices.append([hctl, block, sg, scsi_id]) 13.131 + devices.append([hctl, devname, sg, scsi_id]) 13.132 13.133 return devices 13.134 13.135 13.136 -def vscsi_search_hctl_and_block(device): 13.137 +def vscsi_get_hctl_and_devname_by(target, scsi_devices = None): 13.138 + if scsi_devices is None: 13.139 + scsi_devices = vscsi_get_scsidevices() 13.140 + 13.141 + if len(target.split(':')) == 4: 13.142 + return _vscsi_get_devname_by(target, scsi_devices) 13.143 + else: 13.144 + return _vscsi_get_hctl_by(target, scsi_devices) 13.145 + 13.146 + 13.147 +def get_scsi_vendor(pHCTL): 13.148 + try: 13.149 + sysfs_mnt = utils.find_sysfs_mount() 13.150 + sysfs_scsi_dev_path = \ 13.151 + os.path.join(sysfs_mnt + SYSFS_SCSI_PATH, pHCTL) 13.152 + scsi_vendor = \ 13.153 + os.popen('cat ' + sysfs_scsi_dev_path + \ 13.154 + SYSFS_SCSI_DEV_VENDOR_PATH).read() 13.155 + return scsi_vendor.splitlines()[0] 13.156 + except: 13.157 + return None 13.158 13.159 - scsi_devices = vscsi_get_scsidevices() 13.160 +def get_scsi_model(pHCTL): 13.161 + try: 13.162 + sysfs_mnt = utils.find_sysfs_mount() 13.163 + sysfs_scsi_dev_path = \ 13.164 + os.path.join(sysfs_mnt + SYSFS_SCSI_PATH, pHCTL) 13.165 + scsi_model = \ 13.166 + os.popen('cat ' + sysfs_scsi_dev_path + \ 13.167 + SYSFS_SCSI_DEV_MODEL_PATH).read() 13.168 + return scsi_model.splitlines()[0] 13.169 + except: 13.170 + return None 13.171 + 13.172 +def get_scsi_typeid(pHCTL): 13.173 + try: 13.174 + sysfs_mnt = utils.find_sysfs_mount() 13.175 + sysfs_scsi_dev_path = \ 13.176 + os.path.join(sysfs_mnt + SYSFS_SCSI_PATH, pHCTL) 13.177 + scsi_typeid = \ 13.178 + os.popen('cat ' + sysfs_scsi_dev_path + \ 13.179 + SYSFS_SCSI_DEV_TYPEID_PATH).read() 13.180 + return int(scsi_typeid.splitlines()[0]) 13.181 + except: 13.182 + return None 13.183 13.184 - tmp = device.split(':') 13.185 - if len(tmp) == 4: 13.186 - (hctl, block) = _vscsi_hctl_block(device, scsi_devices) 13.187 - else: 13.188 - (hctl, block) = _vscsi_block_scsiid_to_hctl(device, scsi_devices) 13.189 +def get_scsi_revision(pHCTL): 13.190 + try: 13.191 + sysfs_mnt = utils.find_sysfs_mount() 13.192 + sysfs_scsi_dev_path = \ 13.193 + os.path.join(sysfs_mnt + SYSFS_SCSI_PATH, pHCTL) 13.194 + scsi_revision = \ 13.195 + os.popen('cat ' + sysfs_scsi_dev_path + \ 13.196 + SYSFS_SCSI_DEV_REVISION_PATH).read() 13.197 + return scsi_revision.splitlines()[0] 13.198 + except: 13.199 + return None 13.200 + 13.201 +def get_scsi_scsilevel(pHCTL): 13.202 + try: 13.203 + sysfs_mnt = utils.find_sysfs_mount() 13.204 + sysfs_scsi_dev_path = \ 13.205 + os.path.join(sysfs_mnt + SYSFS_SCSI_PATH, pHCTL) 13.206 + scsi_scsilevel = \ 13.207 + os.popen('cat ' + sysfs_scsi_dev_path + \ 13.208 + SYSFS_SCSI_DEV_SCSILEVEL_PATH).read() 13.209 + return int(scsi_scsilevel.splitlines()[0]) 13.210 + except: 13.211 + return None 13.212 + 13.213 +def get_all_scsi_devices(): 13.214 + 13.215 + scsi_devs = [] 13.216 13.217 - return (hctl, block) 13.218 + for scsi_info in vscsi_get_scsidevices(): 13.219 + scsi_dev = { 13.220 + 'physical_HCTL': scsi_info[0], 13.221 + 'dev_name': None, 13.222 + 'sg_name': scsi_info[2], 13.223 + 'scsi_id': None 13.224 + } 13.225 + if scsi_info[1] is not None: 13.226 + scsi_dev['dev_name'] = scsi_info[1] 13.227 + if scsi_info[3] is not None: 13.228 + scsi_dev['scsi_id'] = scsi_info[3] 13.229 13.230 + scsi_dev['vendor_name'] = \ 13.231 + get_scsi_vendor(scsi_dev['physical_HCTL']) 13.232 + scsi_dev['model'] = \ 13.233 + get_scsi_model(scsi_dev['physical_HCTL']) 13.234 + scsi_dev['type_id'] = \ 13.235 + get_scsi_typeid(scsi_dev['physical_HCTL']) 13.236 + scsi_dev['revision'] = \ 13.237 + get_scsi_revision(scsi_dev['physical_HCTL']) 13.238 + scsi_dev['scsi_level'] = \ 13.239 + get_scsi_scsilevel(scsi_dev['physical_HCTL']) 13.240 + 13.241 + try: 13.242 + lsscsi_info = os.popen('lsscsi ' + scsi_dev['physical_HCTL']).read().split() 13.243 + scsi_dev['type'] = lsscsi_info[1] 13.244 + except: 13.245 + scsi_dev['type'] = None 13.246 + 13.247 + scsi_devs.append(scsi_dev) 13.248 + 13.249 + return scsi_devs 13.250 +
14.1 --- a/tools/python/xen/xend/XendAPI.py Wed Oct 22 11:38:22 2008 +0900 14.2 +++ b/tools/python/xen/xend/XendAPI.py Wed Oct 22 11:46:55 2008 +0900 14.3 @@ -42,6 +42,8 @@ from XendPIF import XendPIF 14.4 from XendPBD import XendPBD 14.5 from XendPPCI import XendPPCI 14.6 from XendDPCI import XendDPCI 14.7 +from XendPSCSI import XendPSCSI 14.8 +from XendDSCSI import XendDSCSI 14.9 from XendXSPolicy import XendXSPolicy, XendACMPolicy 14.10 14.11 from XendAPIConstants import * 14.12 @@ -480,7 +482,9 @@ classes = { 14.13 'PBD' : valid_object("PBD"), 14.14 'PIF_metrics' : valid_object("PIF_metrics"), 14.15 'PPCI' : valid_object("PPCI"), 14.16 - 'DPCI' : valid_object("DPCI") 14.17 + 'DPCI' : valid_object("DPCI"), 14.18 + 'PSCSI' : valid_object("PSCSI"), 14.19 + 'DSCSI' : valid_object("DSCSI") 14.20 } 14.21 14.22 autoplug_classes = { 14.23 @@ -491,6 +495,8 @@ autoplug_classes = { 14.24 'PIF_metrics' : XendPIFMetrics, 14.25 'PPCI' : XendPPCI, 14.26 'DPCI' : XendDPCI, 14.27 + 'PSCSI' : XendPSCSI, 14.28 + 'DSCSI' : XendDSCSI, 14.29 'XSPolicy' : XendXSPolicy, 14.30 'ACMPolicy' : XendACMPolicy, 14.31 } 14.32 @@ -881,6 +887,7 @@ class XendAPI(object): 14.33 'PBDs', 14.34 'PIFs', 14.35 'PPCIs', 14.36 + 'PSCSIs', 14.37 'host_CPUs', 14.38 'cpu_configuration', 14.39 'metrics', 14.40 @@ -961,6 +968,8 @@ class XendAPI(object): 14.41 return xen_api_success(XendNode.instance().get_PIF_refs()) 14.42 def host_get_PPCIs(self, session, ref): 14.43 return xen_api_success(XendNode.instance().get_PPCI_refs()) 14.44 + def host_get_PSCSIs(self, session, ref): 14.45 + return xen_api_success(XendNode.instance().get_PSCSI_refs()) 14.46 def host_get_host_CPUs(self, session, host_ref): 14.47 return xen_api_success(XendNode.instance().get_host_cpu_refs()) 14.48 def host_get_metrics(self, _, ref): 14.49 @@ -1037,7 +1046,8 @@ class XendAPI(object): 14.50 'logging': {}, 14.51 'PIFs': XendPIF.get_all(), 14.52 'PBDs': XendPBD.get_all(), 14.53 - 'PPCIs': XendPPCI.get_all()} 14.54 + 'PPCIs': XendPPCI.get_all(), 14.55 + 'PSCSIs': XendPSCSI.get_all()} 14.56 return xen_api_success(record) 14.57 14.58 # class methods 14.59 @@ -1158,6 +1168,7 @@ class XendAPI(object): 14.60 'VBDs', 14.61 'VTPMs', 14.62 'DPCIs', 14.63 + 'DSCSIs', 14.64 'tools_version', 14.65 'domid', 14.66 'is_control_domain', 14.67 @@ -1304,6 +1315,10 @@ class XendAPI(object): 14.68 dom = XendDomain.instance().get_vm_by_uuid(vm_ref) 14.69 return xen_api_success(dom.get_dpcis()) 14.70 14.71 + def VM_get_DSCSIs(self, session, vm_ref): 14.72 + dom = XendDomain.instance().get_vm_by_uuid(vm_ref) 14.73 + return xen_api_success(dom.get_dscsis()) 14.74 + 14.75 def VM_get_tools_version(self, session, vm_ref): 14.76 dom = XendDomain.instance().get_vm_by_uuid(vm_ref) 14.77 return dom.get_tools_version() 14.78 @@ -1684,6 +1699,7 @@ class XendAPI(object): 14.79 'VBDs': xeninfo.get_vbds(), 14.80 'VTPMs': xeninfo.get_vtpms(), 14.81 'DPCIs': xeninfo.get_dpcis(), 14.82 + 'DSCSIs': xeninfo.get_dscsis(), 14.83 'PV_bootloader': xeninfo.info.get('PV_bootloader'), 14.84 'PV_kernel': xeninfo.info.get('PV_kernel'), 14.85 'PV_ramdisk': xeninfo.info.get('PV_ramdisk'),
15.1 --- a/tools/python/xen/xend/XendConfig.py Wed Oct 22 11:38:22 2008 +0900 15.2 +++ b/tools/python/xen/xend/XendConfig.py Wed Oct 22 11:46:55 2008 +0900 15.3 @@ -26,6 +26,8 @@ from xen.xend import XendOptions 15.4 from xen.xend import XendAPIStore 15.5 from xen.xend.XendPPCI import XendPPCI 15.6 from xen.xend.XendDPCI import XendDPCI 15.7 +from xen.xend.XendPSCSI import XendPSCSI 15.8 +from xen.xend.XendDSCSI import XendDSCSI 15.9 from xen.xend.XendError import VmError 15.10 from xen.xend.XendDevices import XendDevices 15.11 from xen.xend.PrettyPrint import prettyprintstring 15.12 @@ -210,6 +212,7 @@ XENAPI_CFG_TYPES = { 15.13 'cpuid' : dict, 15.14 'cpuid_check' : dict, 15.15 'machine_address_size': int, 15.16 + 'suppress_spurious_page_faults': bool0, 15.17 } 15.18 15.19 # List of legacy configuration keys that have no equivalent in the 15.20 @@ -781,8 +784,8 @@ class XendConfig(dict): 15.21 log.debug('_sxp_to_xapi(%s)' % scrub_password(sxp_cfg)) 15.22 15.23 # _parse_sxp() below will call device_add() and construct devices. 15.24 - # Some devices (currently only pci) may require VM's uuid, so 15.25 - # setup self['uuid'] beforehand. 15.26 + # Some devices may require VM's uuid, so setup self['uuid'] 15.27 + # beforehand. 15.28 self['uuid'] = sxp.child_value(sxp_cfg, 'uuid', uuid.createString()) 15.29 15.30 cfg = self._parse_sxp(sxp_cfg) 15.31 @@ -1221,29 +1224,28 @@ class XendConfig(dict): 15.32 dev_type = sxp.name(config) 15.33 dev_info = {} 15.34 15.35 - if dev_type == 'pci' or dev_type == 'vscsi': 15.36 + if dev_type == 'pci': 15.37 pci_devs_uuid = sxp.child_value(config, 'uuid', 15.38 uuid.createString()) 15.39 15.40 pci_dict = self.pci_convert_sxp_to_dict(config) 15.41 pci_devs = pci_dict['devs'] 15.42 15.43 - if dev_type != 'vscsi': 15.44 - # create XenAPI DPCI objects. 15.45 - for pci_dev in pci_devs: 15.46 - dpci_uuid = pci_dev.get('uuid') 15.47 - ppci_uuid = XendPPCI.get_by_sbdf(pci_dev['domain'], 15.48 - pci_dev['bus'], 15.49 - pci_dev['slot'], 15.50 - pci_dev['func']) 15.51 - if ppci_uuid is None: 15.52 - continue 15.53 - dpci_record = { 15.54 - 'VM': self['uuid'], 15.55 - 'PPCI': ppci_uuid, 15.56 - 'hotplug_slot': pci_dev.get('vslot', 0) 15.57 - } 15.58 - XendDPCI(dpci_uuid, dpci_record) 15.59 + # create XenAPI DPCI objects. 15.60 + for pci_dev in pci_devs: 15.61 + dpci_uuid = pci_dev.get('uuid') 15.62 + ppci_uuid = XendPPCI.get_by_sbdf(pci_dev['domain'], 15.63 + pci_dev['bus'], 15.64 + pci_dev['slot'], 15.65 + pci_dev['func']) 15.66 + if ppci_uuid is None: 15.67 + continue 15.68 + dpci_record = { 15.69 + 'VM': self['uuid'], 15.70 + 'PPCI': ppci_uuid, 15.71 + 'hotplug_slot': pci_dev.get('vslot', 0) 15.72 + } 15.73 + XendDPCI(dpci_uuid, dpci_record) 15.74 15.75 target['devices'][pci_devs_uuid] = (dev_type, 15.76 {'devs': pci_devs, 15.77 @@ -1253,6 +1255,30 @@ class XendConfig(dict): 15.78 15.79 return pci_devs_uuid 15.80 15.81 + if dev_type == 'vscsi': 15.82 + vscsi_devs_uuid = sxp.child_value(config, 'uuid', 15.83 + uuid.createString()) 15.84 + vscsi_dict = self.vscsi_convert_sxp_to_dict(config) 15.85 + vscsi_devs = vscsi_dict['devs'] 15.86 + 15.87 + # create XenAPI DSCSI objects. 15.88 + for vscsi_dev in vscsi_devs: 15.89 + dscsi_uuid = vscsi_dev.get('uuid') 15.90 + pscsi_uuid = XendPSCSI.get_by_HCTL(vscsi_dev['p-dev']) 15.91 + if pscsi_uuid is None: 15.92 + continue 15.93 + dscsi_record = { 15.94 + 'VM': self['uuid'], 15.95 + 'PSCSI': pscsi_uuid, 15.96 + 'virtual_HCTL': vscsi_dev.get('v-dev') 15.97 + } 15.98 + XendDSCSI(dscsi_uuid, dscsi_record) 15.99 + 15.100 + target['devices'][vscsi_devs_uuid] = \ 15.101 + (dev_type, {'devs': vscsi_devs, 'uuid': vscsi_devs_uuid} ) 15.102 + log.debug("XendConfig: reading device: %s" % vscsi_devs) 15.103 + return vscsi_devs_uuid 15.104 + 15.105 for opt_val in config[1:]: 15.106 try: 15.107 opt, val = opt_val 15.108 @@ -1558,6 +1584,86 @@ class XendConfig(dict): 15.109 15.110 return dev_config 15.111 15.112 + def vscsi_convert_sxp_to_dict(self, dev_sxp): 15.113 + """Convert vscsi device sxp to dict 15.114 + @param dev_sxp: device configuration 15.115 + @type dev_sxp: SXP object (parsed config) 15.116 + @return: dev_config 15.117 + @rtype: dictionary 15.118 + """ 15.119 + # Parsing the device SXP's. In most cases, the SXP looks 15.120 + # like this: 15.121 + # 15.122 + # [device, [vif, [mac, xx:xx:xx:xx:xx:xx], [ip 1.3.4.5]]] 15.123 + # 15.124 + # However, for SCSI devices it looks like this: 15.125 + # 15.126 + # [device, 15.127 + # [vscsi, 15.128 + # [dev, 15.129 + # [devid, 0], [p-devname, sdb], [p-dev, 1:0:0:1], 15.130 + # [v-dev, 0:0:0:0], [state, Initialising] 15.131 + # ], 15.132 + # [dev, 15.133 + # [devid, 0], [p-devname, sdc], [p-dev, 1:0:0:2], 15.134 + # [v-dev, 0:0:0:1], [satet, Initialising] 15.135 + # ] 15.136 + # ], 15.137 + # [vscsi, 15.138 + # [dev, 15.139 + # [devid, 1], [p-devname, sdg], [p-dev, 2:0:0:0], 15.140 + # [v-dev, 1:0:0:0], [state, Initialising] 15.141 + # ], 15.142 + # [dev, 15.143 + # [devid, 1], [p-devname, sdh], [p-dev, 2:0:0:1], 15.144 + # [v-dev, 1:0:0:1], [satet, Initialising] 15.145 + # ] 15.146 + # ] 15.147 + # ] 15.148 + # 15.149 + # It seems the reasoning for this difference is because 15.150 + # vscsiif.py needs all the SCSI device configurations with 15.151 + # same host number at the same time when creating the devices. 15.152 + 15.153 + # For SCSI device hotplug support, the SXP of SCSI devices is 15.154 + # extendend like this: 15.155 + # 15.156 + # [device, 15.157 + # [vscsi, 15.158 + # [dev, 15.159 + # [devid, 0], [p-devname, sdd], [p-dev, 1:0:0:3], 15.160 + # [v-dev, 0:0:0:2], [state, Initialising] 15.161 + # ] 15.162 + # ] 15.163 + # ] 15.164 + # 15.165 + # state 'Initialising' indicates that the device is being attached, 15.166 + # while state 'Closing' indicates that the device is being detached. 15.167 + # 15.168 + # The Dict looks like this: 15.169 + # 15.170 + # { devs: [ {devid: 0, p-devname: sdd, p-dev: 1:0:0:3, 15.171 + # v-dev: 0:0:0:2, state: Initialising} ] } 15.172 + 15.173 + dev_config = {} 15.174 + 15.175 + vscsi_devs = [] 15.176 + for vscsi_dev in sxp.children(dev_sxp, 'dev'): 15.177 + vscsi_dev_info = {} 15.178 + for opt_val in vscsi_dev[1:]: 15.179 + try: 15.180 + opt, val = opt_val 15.181 + vscsi_dev_info[opt] = val 15.182 + except TypeError: 15.183 + pass 15.184 + # append uuid for each vscsi device. 15.185 + vscsi_uuid = vscsi_dev_info.get('uuid', uuid.createString()) 15.186 + vscsi_dev_info['uuid'] = vscsi_uuid 15.187 + vscsi_devs.append(vscsi_dev_info) 15.188 + dev_config['devs'] = vscsi_devs 15.189 + 15.190 + return dev_config 15.191 + 15.192 def console_add(self, protocol, location, other_config = {}): 15.193 dev_uuid = uuid.createString() 15.194 if protocol == 'vt100': 15.195 @@ -1631,7 +1737,7 @@ class XendConfig(dict): 15.196 15.197 dev_type, dev_info = self['devices'][dev_uuid] 15.198 15.199 - if dev_type == 'pci' or dev_type == 'vscsi': # Special case for pci 15.200 + if dev_type == 'pci': # Special case for pci 15.201 pci_dict = self.pci_convert_sxp_to_dict(config) 15.202 pci_devs = pci_dict['devs'] 15.203 15.204 @@ -1639,28 +1745,52 @@ class XendConfig(dict): 15.205 for dpci_uuid in XendDPCI.get_by_VM(self['uuid']): 15.206 XendAPIStore.deregister(dpci_uuid, "DPCI") 15.207 15.208 - if dev_type != 'vscsi': 15.209 - # create XenAPI DPCI objects. 15.210 - for pci_dev in pci_devs: 15.211 - dpci_uuid = pci_dev.get('uuid') 15.212 - ppci_uuid = XendPPCI.get_by_sbdf(pci_dev['domain'], 15.213 - pci_dev['bus'], 15.214 - pci_dev['slot'], 15.215 - pci_dev['func']) 15.216 - if ppci_uuid is None: 15.217 - continue 15.218 - dpci_record = { 15.219 - 'VM': self['uuid'], 15.220 - 'PPCI': ppci_uuid, 15.221 - 'hotplug_slot': pci_dev.get('vslot', 0) 15.222 - } 15.223 - XendDPCI(dpci_uuid, dpci_record) 15.224 + # create XenAPI DPCI objects. 15.225 + for pci_dev in pci_devs: 15.226 + dpci_uuid = pci_dev.get('uuid') 15.227 + ppci_uuid = XendPPCI.get_by_sbdf(pci_dev['domain'], 15.228 + pci_dev['bus'], 15.229 + pci_dev['slot'], 15.230 + pci_dev['func']) 15.231 + if ppci_uuid is None: 15.232 + continue 15.233 + dpci_record = { 15.234 + 'VM': self['uuid'], 15.235 + 'PPCI': ppci_uuid, 15.236 + 'hotplug_slot': pci_dev.get('vslot', 0) 15.237 + } 15.238 + XendDPCI(dpci_uuid, dpci_record) 15.239 15.240 self['devices'][dev_uuid] = (dev_type, 15.241 {'devs': pci_devs, 15.242 'uuid': dev_uuid}) 15.243 return True 15.244 15.245 + if dev_type == 'vscsi': # Special case for vscsi 15.246 + vscsi_dict = self.vscsi_convert_sxp_to_dict(config) 15.247 + vscsi_devs = vscsi_dict['devs'] 15.248 + 15.249 + # destroy existing XenAPI DSCSI objects 15.250 + for dscsi_uuid in XendDSCSI.get_by_VM(self['uuid']): 15.251 + XendAPIStore.deregister(dscsi_uuid, "DSCSI") 15.252 + 15.253 + # create XenAPI DSCSI objects. 15.254 + for vscsi_dev in vscsi_devs: 15.255 + dscsi_uuid = vscsi_dev.get('uuid') 15.256 + pscsi_uuid = XendPSCSI.get_by_HCTL(vscsi_dev['p-dev']) 15.257 + if pscsi_uuid is None: 15.258 + continue 15.259 + dscsi_record = { 15.260 + 'VM': self['uuid'], 15.261 + 'PSCSI': pscsi_uuid, 15.262 + 'virtual_HCTL': vscsi_dev.get('v-dev') 15.263 + } 15.264 + XendDSCSI(dscsi_uuid, dscsi_record) 15.265 + 15.266 + self['devices'][dev_uuid] = \ 15.267 + (dev_type, {'devs': vscsi_devs, 'uuid': dev_uuid} ) 15.268 + return True 15.269 + 15.270 for opt_val in config[1:]: 15.271 try: 15.272 opt, val = opt_val
16.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000 16.2 +++ b/tools/python/xen/xend/XendDSCSI.py Wed Oct 22 11:46:55 2008 +0900 16.3 @@ -0,0 +1,174 @@ 16.4 +#============================================================================ 16.5 +# This library is free software; you can redistribute it and/or 16.6 +# modify it under the terms of version 2.1 of the GNU Lesser General Public 16.7 +# License as published by the Free Software Foundation. 16.8 +# 16.9 +# This library is distributed in the hope that it will be useful, 16.10 +# but WITHOUT ANY WARRANTY; without even the implied warranty of 16.11 +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 16.12 +# Lesser General Public License for more details. 16.13 +# 16.14 +# You should have received a copy of the GNU Lesser General Public 16.15 +# License along with this library; if not, write to the Free Software 16.16 +# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA 16.17 +#============================================================================ 16.18 +# Copyright FUJITSU LIMITED 2008 16.19 +# Masaki Kanno <kanno.masaki@jp.fujitsu.com> 16.20 +#============================================================================ 16.21 + 16.22 +from xen.xend.XendBase import XendBase 16.23 +from xen.xend.XendPSCSI import XendPSCSI 16.24 +from xen.xend import XendAPIStore 16.25 +from xen.xend import sxp 16.26 +from xen.xend import uuid as genuuid 16.27 + 16.28 +import XendDomain, XendNode 16.29 + 16.30 +from XendError import * 16.31 +from XendTask import XendTask 16.32 +from XendLogging import log 16.33 + 16.34 +class XendDSCSI(XendBase): 16.35 + """Representation of a half-virtualized SCSI device.""" 16.36 + 16.37 + def getClass(self): 16.38 + return "DSCSI" 16.39 + 16.40 + def getAttrRO(self): 16.41 + attrRO = ['VM', 16.42 + 'PSCSI', 16.43 + 'virtual_host', 16.44 + 'virtual_channel', 16.45 + 'virtual_target', 16.46 + 'virtual_lun', 16.47 + 'virtual_HCTL', 16.48 + 'runtime_properties'] 16.49 + return XendBase.getAttrRO() + attrRO 16.50 + 16.51 + def getAttrRW(self): 16.52 + attrRW = [] 16.53 + return XendBase.getAttrRW() + attrRW 16.54 + 16.55 + def getAttrInst(self): 16.56 + attrInst = ['VM', 16.57 + 'PSCSI', 16.58 + 'virtual_HCTL'] 16.59 + return XendBase.getAttrInst() + attrInst 16.60 + 16.61 + def getMethods(self): 16.62 + methods = ['destroy'] 16.63 + return XendBase.getMethods() + methods 16.64 + 16.65 + def getFuncs(self): 16.66 + funcs = ['create'] 16.67 + return XendBase.getFuncs() + funcs 16.68 + 16.69 + getClass = classmethod(getClass) 16.70 + getAttrRO = classmethod(getAttrRO) 16.71 + getAttrRW = classmethod(getAttrRW) 16.72 + getAttrInst = classmethod(getAttrInst) 16.73 + getMethods = classmethod(getMethods) 16.74 + getFuncs = classmethod(getFuncs) 16.75 + 16.76 + def create(self, dscsi_struct): 16.77 + 16.78 + # Check if VM is valid 16.79 + xendom = XendDomain.instance() 16.80 + if not xendom.is_valid_vm(dscsi_struct['VM']): 16.81 + raise InvalidHandleError('VM', dscsi_struct['VM']) 16.82 + dom = xendom.get_vm_by_uuid(dscsi_struct['VM']) 16.83 + 16.84 + # Check if PSCSI is valid 16.85 + xennode = XendNode.instance() 16.86 + pscsi_uuid = xennode.get_pscsi_by_uuid(dscsi_struct['PSCSI']) 16.87 + if not pscsi_uuid: 16.88 + raise InvalidHandleError('PSCSI', dscsi_struct['PSCSI']) 16.89 + 16.90 + # Assign PSCSI to VM 16.91 + try: 16.92 + dscsi_ref = XendTask.log_progress(0, 100, \ 16.93 + dom.create_dscsi, \ 16.94 + dscsi_struct) 16.95 + except XendError, e: 16.96 + log.exception("Error in create_dscsi") 16.97 + raise 16.98 + 16.99 + return dscsi_ref 16.100 + 16.101 + create = classmethod(create) 16.102 + 16.103 + def get_by_VM(cls, VM_ref): 16.104 + result = [] 16.105 + for dscsi in XendAPIStore.get_all("DSCSI"): 16.106 + if dscsi.get_VM() == VM_ref: 16.107 + result.append(dscsi.get_uuid()) 16.108 + return result 16.109 + 16.110 + get_by_VM = classmethod(get_by_VM) 16.111 + 16.112 + def __init__(self, uuid, record): 16.113 + XendBase.__init__(self, uuid, record) 16.114 + v_hctl = self.virtual_HCTL.split(':') 16.115 + self.virtual_host = int(v_hctl[0]) 16.116 + self.virtual_channel = int(v_hctl[1]) 16.117 + self.virtual_target = int(v_hctl[2]) 16.118 + self.virtual_lun = int(v_hctl[3]) 16.119 + 16.120 + def get_VM(self): 16.121 + return self.VM 16.122 + 16.123 + def get_PSCSI(self): 16.124 + return self.PSCSI 16.125 + 16.126 + def get_virtual_host(self): 16.127 + return self.virtual_host 16.128 + 16.129 + def get_virtual_channel(self): 16.130 + return self.virtual_channel 16.131 + 16.132 + def get_virtual_target(self): 16.133 + return self.virtual_target 16.134 + 16.135 + def get_virtual_lun(self): 16.136 + return self.virtual_lun 16.137 + 16.138 + def get_virtual_HCTL(self): 16.139 + return self.virtual_HCTL 16.140 + 16.141 + def get_runtime_properties(self): 16.142 + xendom = XendDomain.instance() 16.143 + dominfo = xendom.get_vm_by_uuid(self.VM) 16.144 + 16.145 + try: 16.146 + device_dict = {} 16.147 + for device_sxp in dominfo.getDeviceSxprs('vscsi'): 16.148 + target_dev = None 16.149 + for dev in device_sxp[1][0][1]: 16.150 + vdev = sxp.child_value(dev, 'v-dev') 16.151 + if vdev == self.virtual_HCTL: 16.152 + target_dev = dev 16.153 + break 16.154 + if target_dev is None: 16.155 + continue 16.156 + 16.157 + dev_dict = {} 16.158 + for info in target_dev[1:]: 16.159 + dev_dict[info[0]] = info[1] 16.160 + device_dict['dev'] = dev_dict 16.161 + for info in device_sxp[1][1:]: 16.162 + device_dict[info[0]] = info[1] 16.163 + 16.164 + return device_dict 16.165 + except Exception, exn: 16.166 + log.exception(exn) 16.167 + return {} 16.168 + 16.169 + def destroy(self): 16.170 + xendom = XendDomain.instance() 16.171 + dom = xendom.get_vm_by_uuid(self.get_VM()) 16.172 + if not dom: 16.173 + raise InvalidHandleError("VM", self.get_VM()) 16.174 + XendTask.log_progress(0, 100, \ 16.175 + dom.destroy_dscsi, \ 16.176 + self.get_uuid()) 16.177 +
17.1 --- a/tools/python/xen/xend/XendDomainInfo.py Wed Oct 22 11:38:22 2008 +0900 17.2 +++ b/tools/python/xen/xend/XendDomainInfo.py Wed Oct 22 11:46:55 2008 +0900 17.3 @@ -55,9 +55,11 @@ from xen.xend.XendAPIConstants import * 17.4 17.5 from xen.xend.XendVMMetrics import XendVMMetrics 17.6 17.7 +from xen.xend import XendAPIStore 17.8 from xen.xend.XendPPCI import XendPPCI 17.9 from xen.xend.XendDPCI import XendDPCI 17.10 -from xen.xend import XendAPIStore 17.11 +from xen.xend.XendPSCSI import XendPSCSI 17.12 +from xen.xend.XendDSCSI import XendDSCSI 17.13 17.14 MIGRATE_TIMEOUT = 30.0 17.15 BOOTLOADER_LOOPBACK_DEVICE = '/dev/xvdp' 17.16 @@ -663,6 +665,9 @@ class XendDomainInfo: 17.17 if dev_type == 'pci': 17.18 for dev in dev_config_dict['devs']: 17.19 XendAPIStore.deregister(dev['uuid'], 'DPCI') 17.20 + if dev_type == 'vscsi': 17.21 + for dev in dev_config_dict['devs']: 17.22 + XendAPIStore.deregister(dev['uuid'], 'DSCSI') 17.23 elif dev_type == 'tap': 17.24 self.info['vbd_refs'].remove(dev_uuid) 17.25 else: 17.26 @@ -786,12 +791,11 @@ class XendDomainInfo: 17.27 if dev_class != 'vscsi': 17.28 return False 17.29 17.30 - dev_config = self.info.pci_convert_sxp_to_dict(dev_sxp) 17.31 + dev_config = self.info.vscsi_convert_sxp_to_dict(dev_sxp) 17.32 dev = dev_config['devs'][0] 17.33 - req_devid = sxp.child_value(dev_sxp, 'devid') 17.34 - req_devid = int(req_devid) 17.35 + req_devid = int(dev['devid']) 17.36 existing_dev_info = self._getDeviceInfo_vscsi(req_devid, dev['v-dev']) 17.37 - state = sxp.child_value(dev_sxp, 'state') 17.38 + state = dev['state'] 17.39 17.40 if state == 'Initialising': 17.41 # new create 17.42 @@ -1502,23 +1506,18 @@ class XendDomainInfo: 17.43 return self.info['VCPUs_max'] 17.44 17.45 def setVCpuCount(self, vcpus): 17.46 - if vcpus <= 0: 17.47 - raise XendError('Invalid VCPUs') 17.48 + def vcpus_valid(n): 17.49 + if vcpus <= 0: 17.50 + raise XendError('Zero or less VCPUs is invalid') 17.51 + if self.domid >= 0 and vcpus > self.info['VCPUs_max']: 17.52 + raise XendError('Cannot set vcpus greater than max vcpus on running domain') 17.53 + vcpus_valid(vcpus) 17.54 17.55 self.info['vcpu_avail'] = (1 << vcpus) - 1 17.56 if self.domid >= 0: 17.57 self.storeVm('vcpu_avail', self.info['vcpu_avail']) 17.58 - # update dom differently depending on whether we are adjusting 17.59 - # vcpu number up or down, otherwise _vcpuDomDetails does not 17.60 - # disable the vcpus 17.61 - if self.info['VCPUs_max'] > vcpus: 17.62 - # decreasing 17.63 - self._writeDom(self._vcpuDomDetails()) 17.64 - self.info['VCPUs_live'] = vcpus 17.65 - else: 17.66 - # same or increasing 17.67 - self.info['VCPUs_live'] = vcpus 17.68 - self._writeDom(self._vcpuDomDetails()) 17.69 + self._writeDom(self._vcpuDomDetails()) 17.70 + self.info['VCPUs_live'] = vcpus 17.71 else: 17.72 if self.info['VCPUs_max'] > vcpus: 17.73 # decreasing 17.74 @@ -1528,7 +1527,7 @@ class XendDomainInfo: 17.75 for c in range(self.info['VCPUs_max'], vcpus): 17.76 self.info['cpus'].append(list()) 17.77 self.info['VCPUs_max'] = vcpus 17.78 - xen.xend.XendDomain.instance().managed_config_save(self) 17.79 + xen.xend.XendDomain.instance().managed_config_save(self) 17.80 log.info("Set VCPU count on domain %s to %d", self.info['name_label'], 17.81 vcpus) 17.82 17.83 @@ -2241,6 +2240,10 @@ class XendDomainInfo: 17.84 if self.info.has_key('machine_address_size'): 17.85 log.debug("_initDomain: setting maximum machine address size %d" % self.info['machine_address_size']) 17.86 xc.domain_set_machine_address_size(self.domid, self.info['machine_address_size']) 17.87 + 17.88 + if self.info.has_key('suppress_spurious_page_faults') and self.info['suppress_spurious_page_faults']: 17.89 + log.debug("_initDomain: suppressing spurious page faults") 17.90 + xc.domain_suppress_spurious_page_faults(self.domid) 17.91 17.92 self._createChannels() 17.93 17.94 @@ -3233,6 +3236,9 @@ class XendDomainInfo: 17.95 def get_dpcis(self): 17.96 return XendDPCI.get_by_VM(self.info.get('uuid')) 17.97 17.98 + def get_dscsis(self): 17.99 + return XendDSCSI.get_by_VM(self.info.get('uuid')) 17.100 + 17.101 def create_vbd(self, xenapi_vbd, vdi_image_path): 17.102 """Create a VBD using a VDI from XendStorageRepository. 17.103 17.104 @@ -3413,6 +3419,60 @@ class XendDomainInfo: 17.105 17.106 return dpci_uuid 17.107 17.108 + def create_dscsi(self, xenapi_dscsi): 17.109 + """Create scsi device from the passed struct in Xen API format. 17.110 + 17.111 + @param xenapi_dscsi: DSCSI struct from Xen API 17.112 + @rtype: string 17.113 + @return: UUID 17.114 + """ 17.115 + 17.116 + dscsi_uuid = uuid.createString() 17.117 + 17.118 + # Convert xenapi to sxp 17.119 + pscsi = XendAPIStore.get(xenapi_dscsi.get('PSCSI'), 'PSCSI') 17.120 + devid = int(xenapi_dscsi.get('virtual_HCTL').split(':')[0]) 17.121 + target_vscsi_sxp = \ 17.122 + ['vscsi', 17.123 + ['dev', 17.124 + ['devid', devid], 17.125 + ['p-devname', pscsi.get_dev_name()], 17.126 + ['p-dev', pscsi.get_physical_HCTL()], 17.127 + ['v-dev', xenapi_dscsi.get('virtual_HCTL')], 17.128 + ['state', 'Initialising'], 17.129 + ['uuid', dscsi_uuid] 17.130 + ] 17.131 + ] 17.132 + 17.133 + if self._stateGet() != XEN_API_VM_POWER_STATE_RUNNING: 17.134 + 17.135 + cur_vscsi_sxp = self._getDeviceInfo_vscsi(devid, None) 17.136 + 17.137 + if cur_vscsi_sxp is None: 17.138 + dev_uuid = self.info.device_add('vscsi', cfg_sxp = target_vscsi_sxp) 17.139 + if not dev_uuid: 17.140 + raise XendError('Failed to create device') 17.141 + 17.142 + else: 17.143 + new_vscsi_sxp = ['vscsi'] 17.144 + for existing_dev in sxp.children(cur_vscsi_sxp, 'dev'): 17.145 + new_vscsi_sxp.append(existing_dev) 17.146 + new_vscsi_sxp.append(sxp.child0(target_vscsi_sxp, 'dev')) 17.147 + 17.148 + dev_uuid = sxp.child_value(cur_vscsi_sxp, 'uuid') 17.149 + self.info.device_update(dev_uuid, new_vscsi_sxp) 17.150 + 17.151 + xen.xend.XendDomain.instance().managed_config_save(self) 17.152 + 17.153 + else: 17.154 + try: 17.155 + self.device_configure(target_vscsi_sxp) 17.156 + 17.157 + except Exception, exn: 17.158 + raise XendError('Failed to create device') 17.159 + 17.160 + return dscsi_uuid 17.161 + 17.162 17.163 def destroy_device_by_uuid(self, dev_type, dev_uuid): 17.164 if dev_uuid not in self.info['devices']: 17.165 @@ -3480,6 +3540,41 @@ class XendDomainInfo: 17.166 except Exception, exn: 17.167 raise XendError('Failed to destroy device') 17.168 17.169 + def destroy_dscsi(self, dev_uuid): 17.170 + dscsi = XendAPIStore.get(dev_uuid, 'DSCSI') 17.171 + devid = dscsi.get_virtual_host() 17.172 + vHCTL = dscsi.get_virtual_HCTL() 17.173 + cur_vscsi_sxp = self._getDeviceInfo_vscsi(devid, None) 17.174 + dev_uuid = sxp.child_value(cur_vscsi_sxp, 'uuid') 17.175 + 17.176 + target_dev = None 17.177 + new_vscsi_sxp = ['vscsi'] 17.178 + for dev in sxp.children(cur_vscsi_sxp, 'dev'): 17.179 + if vHCTL == sxp.child_value(dev, 'v-dev'): 17.180 + target_dev = dev 17.181 + else: 17.182 + new_vscsi_sxp.append(dev) 17.183 + 17.184 + if target_dev is None: 17.185 + raise XendError('Failed to destroy device') 17.186 + 17.187 + target_dev.append(['state', 'Closing']) 17.188 + target_vscsi_sxp = ['vscsi', target_dev] 17.189 + 17.190 + if self._stateGet() != XEN_API_VM_POWER_STATE_RUNNING: 17.191 + 17.192 + self.info.device_update(dev_uuid, new_vscsi_sxp) 17.193 + if len(sxp.children(new_vscsi_sxp, 'dev')) == 0: 17.194 + del self.info['devices'][dev_uuid] 17.195 + xen.xend.XendDomain.instance().managed_config_save(self) 17.196 + 17.197 + else: 17.198 + try: 17.199 + self.device_configure(target_vscsi_sxp) 17.200 + 17.201 + except Exception, exn: 17.202 + raise XendError('Failed to destroy device') 17.203 + 17.204 def destroy_xapi_instances(self): 17.205 """Destroy Xen-API instances stored in XendAPIStore. 17.206 """ 17.207 @@ -3504,6 +3599,10 @@ class XendDomainInfo: 17.208 for dpci_uuid in XendDPCI.get_by_VM(self.info.get('uuid')): 17.209 XendAPIStore.deregister(dpci_uuid, "DPCI") 17.210 17.211 + # Destroy DSCSI instances. 17.212 + for dscsi_uuid in XendDSCSI.get_by_VM(self.info.get('uuid')): 17.213 + XendAPIStore.deregister(dscsi_uuid, "DSCSI") 17.214 + 17.215 def has_device(self, dev_class, dev_uuid): 17.216 return (dev_uuid in self.info['%s_refs' % dev_class.lower()]) 17.217
18.1 --- a/tools/python/xen/xend/XendNode.py Wed Oct 22 11:38:22 2008 +0900 18.2 +++ b/tools/python/xen/xend/XendNode.py Wed Oct 22 11:46:55 2008 +0900 18.3 @@ -22,6 +22,7 @@ import xen.lowlevel.xc 18.4 18.5 from xen.util import Brctl 18.6 from xen.util import pci as PciUtil 18.7 +from xen.util import vscsi_util 18.8 from xen.xend import XendAPIStore 18.9 from xen.xend import osdep 18.10 18.11 @@ -38,7 +39,8 @@ from XendNetwork import * 18.12 from XendStateStore import XendStateStore 18.13 from XendMonitor import XendMonitor 18.14 from XendPPCI import XendPPCI 18.15 - 18.16 +from XendPSCSI import XendPSCSI 18.17 + 18.18 class XendNode: 18.19 """XendNode - Represents a Domain 0 Host.""" 18.20 18.21 @@ -53,6 +55,7 @@ class XendNode: 18.22 * network 18.23 * Storage Repository 18.24 * PPCI 18.25 + * PSCSI 18.26 """ 18.27 18.28 self.xc = xen.lowlevel.xc.xc() 18.29 @@ -269,6 +272,24 @@ class XendNode: 18.30 XendPPCI(ppci_uuid, ppci_record) 18.31 18.32 18.33 + # Initialise PSCSIs 18.34 + saved_pscsis = self.state_store.load_state('pscsi') 18.35 + saved_pscsi_table = {} 18.36 + if saved_pscsis: 18.37 + for pscsi_uuid, pscsi_record in saved_pscsis.items(): 18.38 + try: 18.39 + saved_pscsi_table[pscsi_record['scsi_id']] = pscsi_uuid 18.40 + except KeyError: 18.41 + pass 18.42 + 18.43 + for pscsi_record in vscsi_util.get_all_scsi_devices(): 18.44 + if pscsi_record['scsi_id']: 18.45 + # If saved uuid exists, use it. Otherwise create one. 18.46 + pscsi_uuid = saved_pscsi_table.get(pscsi_record['scsi_id'], 18.47 + uuid.createString()) 18.48 + XendPSCSI(pscsi_uuid, pscsi_record) 18.49 + 18.50 + 18.51 ## def network_destroy(self, net_uuid): 18.52 ## del self.networks[net_uuid] 18.53 ## self.save_networks() 18.54 @@ -320,6 +341,15 @@ class XendNode: 18.55 return None 18.56 18.57 18.58 + def get_PSCSI_refs(self): 18.59 + return XendPSCSI.get_all() 18.60 + 18.61 + def get_pscsi_by_uuid(self, pscsi_uuid): 18.62 + if pscsi_uuid in self.get_PSCSI_refs(): 18.63 + return pscsi_uuid 18.64 + return None 18.65 + 18.66 + 18.67 def save(self): 18.68 # save state 18.69 host_record = {self.uuid: {'name_label':self.name, 18.70 @@ -333,6 +363,7 @@ class XendNode: 18.71 self.save_PBDs() 18.72 self.save_SRs() 18.73 self.save_PPCIs() 18.74 + self.save_PSCSIs() 18.75 18.76 def save_PIFs(self): 18.77 pif_records = dict([(pif_uuid, XendAPIStore.get( 18.78 @@ -363,6 +394,12 @@ class XendNode: 18.79 for ppci_uuid in XendPPCI.get_all()]) 18.80 self.state_store.save_state('ppci', ppci_records) 18.81 18.82 + def save_PSCSIs(self): 18.83 + pscsi_records = dict([(pscsi_uuid, XendAPIStore.get( 18.84 + pscsi_uuid, "PSCSI").get_record()) 18.85 + for pscsi_uuid in XendPSCSI.get_all()]) 18.86 + self.state_store.save_state('pscsi', pscsi_records) 18.87 + 18.88 def shutdown(self): 18.89 return 0 18.90
19.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000 19.2 +++ b/tools/python/xen/xend/XendPSCSI.py Wed Oct 22 11:46:55 2008 +0900 19.3 @@ -0,0 +1,143 @@ 19.4 +#============================================================================ 19.5 +# This library is free software; you can redistribute it and/or 19.6 +# modify it under the terms of version 2.1 of the GNU Lesser General Public 19.7 +# License as published by the Free Software Foundation. 19.8 +# 19.9 +# This library is distributed in the hope that it will be useful, 19.10 +# but WITHOUT ANY WARRANTY; without even the implied warranty of 19.11 +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 19.12 +# Lesser General Public License for more details. 19.13 +# 19.14 +# You should have received a copy of the GNU Lesser General Public 19.15 +# License along with this library; if not, write to the Free Software 19.16 +# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA 19.17 +#============================================================================ 19.18 +# Copyright FUJITSU LIMITED 2008 19.19 +# Masaki Kanno <kanno.masaki@jp.fujitsu.com> 19.20 +#============================================================================ 19.21 + 19.22 +from xen.xend.XendBase import XendBase 19.23 +from xen.xend.XendBase import XendAPIStore 19.24 +from xen.xend import uuid as genuuid 19.25 + 19.26 +class XendPSCSI(XendBase): 19.27 + """Representation of a physical SCSI device.""" 19.28 + 19.29 + def getClass(self): 19.30 + return "PSCSI" 19.31 + 19.32 + def getAttrRO(self): 19.33 + attrRO = ['host', 19.34 + 'physical_host', 19.35 + 'physical_channel', 19.36 + 'physical_target', 19.37 + 'physical_lun', 19.38 + 'physical_HCTL', 19.39 + 'vendor_name', 19.40 + 'model', 19.41 + 'type_id', 19.42 + 'type', 19.43 + 'dev_name', 19.44 + 'sg_name', 19.45 + 'revision', 19.46 + 'scsi_id', 19.47 + 'scsi_level'] 19.48 + return XendBase.getAttrRO() + attrRO 19.49 + 19.50 + def getAttrRW(self): 19.51 + attrRW = [] 19.52 + return XendBase.getAttrRW() + attrRW 19.53 + 19.54 + def getAttrInst(self): 19.55 + attrInst = [] 19.56 + return XendBase.getAttrInst() + attrInst 19.57 + 19.58 + def getMethods(self): 19.59 + methods = [] 19.60 + return XendBase.getMethods() + methods 19.61 + 19.62 + def getFuncs(self): 19.63 + funcs = [] 19.64 + return XendBase.getFuncs() + funcs 19.65 + 19.66 + getClass = classmethod(getClass) 19.67 + getAttrRO = classmethod(getAttrRO) 19.68 + getAttrRW = classmethod(getAttrRW) 19.69 + getAttrInst = classmethod(getAttrInst) 19.70 + getMethods = classmethod(getMethods) 19.71 + getFuncs = classmethod(getFuncs) 19.72 + 19.73 + def get_by_HCTL(self, physical_HCTL): 19.74 + for pscsi in XendAPIStore.get_all("PSCSI"): 19.75 + if pscsi.get_physical_HCTL() == physical_HCTL: 19.76 + return pscsi.get_uuid() 19.77 + return None 19.78 + 19.79 + get_by_HCTL = classmethod(get_by_HCTL) 19.80 + 19.81 + def __init__(self, uuid, record): 19.82 + self.physical_HCTL = record['physical_HCTL'] 19.83 + self.vendor_name = record['vendor_name'] 19.84 + self.model = record['model'] 19.85 + self.type_id = record['type_id'] 19.86 + self.type = record['type'] 19.87 + self.dev_name = record['dev_name'] 19.88 + self.sg_name = record['sg_name'] 19.89 + self.revision = record['revision'] 19.90 + self.scsi_id = record['scsi_id'] 19.91 + self.scsi_level = record['scsi_level'] 19.92 + 19.93 + p_hctl = self.physical_HCTL.split(':') 19.94 + self.physical_host = int(p_hctl[0]) 19.95 + self.physical_channel = int(p_hctl[1]) 19.96 + self.physical_target = int(p_hctl[2]) 19.97 + self.physical_lun = int(p_hctl[3]) 19.98 + 19.99 + XendBase.__init__(self, uuid, record) 19.100 + 19.101 + def get_host(self): 19.102 + from xen.xend import XendNode 19.103 + return XendNode.instance().get_uuid() 19.104 + 19.105 + def get_physical_host(self): 19.106 + return self.physical_host 19.107 + 19.108 + def get_physical_channel(self): 19.109 + return self.physical_channel 19.110 + 19.111 + def get_physical_target(self): 19.112 + return self.physical_target 19.113 + 19.114 + def get_physical_lun(self): 19.115 + return self.physical_lun 19.116 + 19.117 + def get_physical_HCTL(self): 19.118 + return self.physical_HCTL 19.119 + 19.120 + def get_vendor_name(self): 19.121 + return self.vendor_name 19.122 + 19.123 + def get_model(self): 19.124 + return self.model 19.125 + 19.126 + def get_type_id(self): 19.127 + return self.type_id 19.128 + 19.129 + def get_type(self): 19.130 + return self.type 19.131 + 19.132 + def get_dev_name(self): 19.133 + return self.dev_name 19.134 + 19.135 + def get_sg_name(self): 19.136 + return self.sg_name 19.137 + 19.138 + def get_revision(self): 19.139 + return self.revision 19.140 + 19.141 + def get_scsi_id(self): 19.142 + return self.scsi_id 19.143 + 19.144 + def get_scsi_level(self): 19.145 + return self.scsi_level 19.146 +
20.1 --- a/tools/python/xen/xend/server/vscsiif.py Wed Oct 22 11:38:22 2008 +0900 20.2 +++ b/tools/python/xen/xend/server/vscsiif.py Wed Oct 22 11:46:55 2008 +0900 20.3 @@ -125,10 +125,10 @@ class VSCSIController(DevController): 20.4 state = self.readBackend(devid, devpath + '/state') 20.5 localdevid = self.readBackend(devid, devpath + '/devid') 20.6 dev_dict = {'p-dev': pdev, 20.7 - 'p-devname': pdevname, 20.8 - 'v-dev': pdevname, 20.9 - 'state': state, 20.10 - 'devid': localdevid } 20.11 + 'p-devname': pdevname, 20.12 + 'v-dev': vdev, 20.13 + 'state': state, 20.14 + 'devid': localdevid } 20.15 vscsi_devs.append(dev_dict) 20.16 20.17 config['devs'] = vscsi_devs 20.18 @@ -168,17 +168,17 @@ class VSCSIController(DevController): 20.19 (devid, back, front) = self.getDeviceDetails(config) 20.20 devid = int(devid) 20.21 vscsi_config = config['devs'][0] 20.22 - states = config.get('states', []) 20.23 + state = vscsi_config.get('state', '') 20.24 driver_state = self.readBackend(devid, 'state') 20.25 if str(xenbusState['Connected']) != driver_state: 20.26 raise VmError("Driver status is not connected") 20.27 20.28 uuid = self.readBackend(devid, 'uuid') 20.29 - if states[0] == 'Initialising': 20.30 + if state == 'Initialising': 20.31 back['uuid'] = uuid 20.32 self.writeBackend(devid, back) 20.33 20.34 - elif states[0] == 'Closing': 20.35 + elif state == 'Closing': 20.36 found = False 20.37 devs = self.readBackendList(devid, "vscsi-devs") 20.38 vscsipath = "vscsi-devs/" 20.39 @@ -197,8 +197,8 @@ class VSCSIController(DevController): 20.40 raise VmError("Device %s not connected" % vdev) 20.41 20.42 else: 20.43 - raise XendError('Error configuring device invalid state %s' 20.44 - % state) 20.45 + raise XendError("Error configuring device invalid " 20.46 + "state '%s'" % state) 20.47 20.48 self.writeBackend(devid, 'state', str(xenbusState['Reconfiguring'])) 20.49 return self.readBackend(devid, 'uuid')
21.1 --- a/tools/python/xen/xm/create.dtd Wed Oct 22 11:38:22 2008 +0900 21.2 +++ b/tools/python/xen/xm/create.dtd Wed Oct 22 11:46:55 2008 +0900 21.3 @@ -40,6 +40,7 @@ 21.4 vif*, 21.5 vtpm*, 21.6 pci*, 21.7 + vscsi*, 21.8 console*, 21.9 platform*, 21.10 vcpu_param*, 21.11 @@ -88,6 +89,10 @@ 21.12 func CDATA #REQUIRED 21.13 vslt CDATA #IMPLIED> 21.14 21.15 +<!ELEMENT vscsi EMPTY> 21.16 +<!ATTLIST vscsi p-dev CDATA #REQUIRED 21.17 + v-dev CDATA #REQUIRED> 21.18 + 21.19 <!ELEMENT console (other_config*)> 21.20 <!ATTLIST console protocol (vt100|rfb|rdp) #REQUIRED> 21.21
22.1 --- a/tools/python/xen/xm/create.py Wed Oct 22 11:38:22 2008 +0900 22.2 +++ b/tools/python/xen/xm/create.py Wed Oct 22 11:46:55 2008 +0900 22.3 @@ -583,6 +583,10 @@ gopts.var('machine_address_size', val='B 22.4 fn=set_int, default=None, 22.5 use="""Maximum machine address size""") 22.6 22.7 +gopts.var('suppress_spurious_page_faults', val='yes|no', 22.8 + fn=set_bool, default=None, 22.9 + use="""Do not inject spurious page faults into this guest""") 22.10 + 22.11 def err(msg): 22.12 """Print an error to stderr and exit. 22.13 """ 22.14 @@ -634,6 +638,9 @@ def configure_image(vals): 22.15 if vals.machine_address_size: 22.16 config_image.append(['machine_address_size', vals.machine_address_size]) 22.17 22.18 + if vals.suppress_spurious_page_faults: 22.19 + config_image.append(['suppress_spurious_page_faults', vals.suppress_spurious_page_faults]) 22.20 + 22.21 return config_image 22.22 22.23 def configure_disks(config_devs, vals): 22.24 @@ -696,11 +703,8 @@ def configure_vscsis(config_devs, vals): 22.25 22.26 scsi_devices = vscsi_util.vscsi_get_scsidevices() 22.27 for (p_dev, v_dev, backend) in vals.vscsi: 22.28 - tmp = p_dev.split(':') 22.29 - if len(tmp) == 4: 22.30 - (p_hctl, block) = vscsi_util._vscsi_hctl_block(p_dev, scsi_devices) 22.31 - else: 22.32 - (p_hctl, block) = vscsi_util._vscsi_block_scsiid_to_hctl(p_dev, scsi_devices) 22.33 + (p_hctl, devname) = \ 22.34 + vscsi_util.vscsi_get_hctl_and_devname_by(p_dev, scsi_devices) 22.35 22.36 if p_hctl == None: 22.37 raise ValueError("Cannot find device \"%s\"" % p_dev) 22.38 @@ -716,7 +720,7 @@ def configure_vscsis(config_devs, vals): 22.39 ['state', 'Initialising'], \ 22.40 ['devid', devid], \ 22.41 ['p-dev', p_hctl], \ 22.42 - ['p-devname', block], \ 22.43 + ['p-devname', devname], \ 22.44 ['v-dev', v_dev] ]) 22.45 22.46 if vscsi_lookup_devid(devidlist, devid) == 0: 22.47 @@ -887,7 +891,7 @@ def make_config(vals): 22.48 'restart', 'on_poweroff', 22.49 'on_reboot', 'on_crash', 'vcpus', 'vcpu_avail', 'features', 22.50 'on_xend_start', 'on_xend_stop', 'target', 'cpuid', 22.51 - 'cpuid_check', 'machine_address_size']) 22.52 + 'cpuid_check', 'machine_address_size', 'suppress_spurious_page_faults']) 22.53 22.54 if vals.uuid is not None: 22.55 config.append(['uuid', vals.uuid])
23.1 --- a/tools/python/xen/xm/main.py Wed Oct 22 11:38:22 2008 +0900 23.2 +++ b/tools/python/xen/xm/main.py Wed Oct 22 11:46:55 2008 +0900 23.3 @@ -2235,12 +2235,34 @@ def vscsi_convert_sxp_to_dict(dev_sxp): 23.4 return dev_dict 23.5 23.6 def xm_scsi_list(args): 23.7 - xenapi_unsupported() 23.8 (use_long, params) = arg_check_for_resource_list(args, "scsi-list") 23.9 23.10 dom = params[0] 23.11 23.12 - devs = server.xend.domain.getDeviceSxprs(dom, 'vscsi') 23.13 + devs = [] 23.14 + if serverType == SERVER_XEN_API: 23.15 + 23.16 + dscsi_refs = server.xenapi.VM.get_DSCSIs(get_single_vm(dom)) 23.17 + dscsi_properties = \ 23.18 + map(server.xenapi.DSCSI.get_runtime_properties, dscsi_refs) 23.19 + dscsi_dict = {} 23.20 + for dscsi_property in dscsi_properties: 23.21 + devid = int(dscsi_property['dev']['devid']) 23.22 + try: 23.23 + dscsi_sxp = dscsi_dict[devid] 23.24 + except: 23.25 + dscsi_sxp = [['devs', []]] 23.26 + for key, value in dscsi_property.items(): 23.27 + if key != 'dev': 23.28 + dscsi_sxp.append([key, value]) 23.29 + dev_sxp = ['dev'] 23.30 + dev_sxp.extend(map2sxp(dscsi_property['dev'])) 23.31 + dscsi_sxp[0][1].append(dev_sxp) 23.32 + dscsi_dict[devid] = dscsi_sxp 23.33 + devs = map2sxp(dscsi_dict) 23.34 + 23.35 + else: 23.36 + devs = server.xend.domain.getDeviceSxprs(dom, 'vscsi') 23.37 23.38 if use_long: 23.39 map(PrettyPrint.prettyprint, devs) 23.40 @@ -2464,37 +2486,60 @@ def xm_pci_attach(args): 23.41 else: 23.42 server.xend.domain.device_configure(dom, pci) 23.43 23.44 +def parse_scsi_configuration(p_scsi, v_hctl, state): 23.45 + v = v_hctl.split(':') 23.46 + if len(v) != 4: 23.47 + raise OptionError("Invalid argument: %s" % v_hctl) 23.48 + 23.49 + p_hctl = None 23.50 + devname = None 23.51 + if p_scsi is not None: 23.52 + (p_hctl, devname) = \ 23.53 + vscsi_util.vscsi_get_hctl_and_devname_by(p_scsi) 23.54 + if p_hctl is None: 23.55 + raise OptionError("Cannot find device '%s'" % p_scsi) 23.56 + 23.57 + scsi = ['vscsi'] 23.58 + scsi.append(['dev', \ 23.59 + ['state', state], \ 23.60 + ['devid', int(v[0])], \ 23.61 + ['p-dev', p_hctl], \ 23.62 + ['p-devname', devname], \ 23.63 + ['v-dev', v_hctl] \ 23.64 + ]) 23.65 + 23.66 + return scsi 23.67 + 23.68 def xm_scsi_attach(args): 23.69 - xenapi_unsupported() 23.70 - 23.71 arg_check(args, 'scsi-attach', 3, 4) 23.72 - p_devname = args[1] 23.73 - v_dev = args[2] 23.74 - 23.75 - v_hctl = v_dev.split(':') 23.76 - if len(v_hctl) != 4: 23.77 - raise OptionError("Invalid argument: %s" % v_dev) 23.78 - 23.79 - (p_hctl, block) = vscsi_util.vscsi_search_hctl_and_block(p_devname) 23.80 - 23.81 - if p_hctl == None: 23.82 - raise OptionError("Cannot find device \"%s\"" % p_devname) 23.83 - 23.84 dom = args[0] 23.85 - vscsi = ['vscsi'] 23.86 - vscsi.append(['dev', \ 23.87 - ['state', 'Initialising'], \ 23.88 - ['devid', v_hctl[0]], \ 23.89 - ['p-dev', p_hctl], \ 23.90 - ['p-devname', block], \ 23.91 - ['v-dev', v_dev] ]) 23.92 - 23.93 - if len(args) == 4: 23.94 - vscsi.append(['backend', args[3]]) 23.95 - 23.96 - vscsi.append(['state', 'Initialising']) 23.97 - vscsi.append(['devid', v_hctl[0]]) 23.98 - server.xend.domain.device_configure(dom, vscsi) 23.99 + p_scsi = args[1] 23.100 + v_hctl = args[2] 23.101 + scsi = parse_scsi_configuration(p_scsi, v_hctl, 'Initialising') 23.102 + 23.103 + if serverType == SERVER_XEN_API: 23.104 + 23.105 + scsi_dev = sxp.children(scsi, 'dev')[0] 23.106 + p_hctl = sxp.child_value(scsi_dev, 'p-dev') 23.107 + target_ref = None 23.108 + for pscsi_ref in server.xenapi.PSCSI.get_all(): 23.109 + if p_hctl == server.xenapi.PSCSI.get_physical_HCTL(pscsi_ref): 23.110 + target_ref = pscsi_ref 23.111 + break 23.112 + if target_ref is None: 23.113 + raise OptionError("Cannot find device '%s'" % p_scsi) 23.114 + 23.115 + dscsi_record = { 23.116 + "VM": get_single_vm(dom), 23.117 + "PSCSI": target_ref, 23.118 + "virtual_HCTL": v_hctl 23.119 + } 23.120 + server.xenapi.DSCSI.create(dscsi_record) 23.121 + 23.122 + else: 23.123 + if len(args) == 4: 23.124 + scsi.append(['backend', args[3]]) 23.125 + server.xend.domain.device_configure(dom, scsi) 23.126 23.127 def detach(args, deviceClass): 23.128 rm_cfg = True 23.129 @@ -2587,26 +2632,25 @@ def xm_pci_detach(args): 23.130 server.xend.domain.device_configure(dom, pci) 23.131 23.132 def xm_scsi_detach(args): 23.133 - xenapi_unsupported() 23.134 arg_check(args, 'scsi-detach', 2) 23.135 - 23.136 - v_dev = args[1] 23.137 - v_hctl = v_dev.split(':') 23.138 - if len(v_hctl) != 4: 23.139 - raise OptionError("Invalid argument: %s" % v_dev) 23.140 - 23.141 dom = args[0] 23.142 - vscsi = ['vscsi'] 23.143 - vscsi.append(['dev', \ 23.144 - ['state', 'Closing'], \ 23.145 - ['devid', v_hctl[0]], \ 23.146 - ['p-dev', ''], \ 23.147 - ['p-devname', ''], \ 23.148 - ['v-dev', v_dev] ]) 23.149 - 23.150 - vscsi.append(['state', 'Closing']) 23.151 - vscsi.append(['devid', v_hctl[0]]) 23.152 - server.xend.domain.device_configure(dom, vscsi) 23.153 + v_hctl = args[1] 23.154 + scsi = parse_scsi_configuration(None, v_hctl, 'Closing') 23.155 + 23.156 + if serverType == SERVER_XEN_API: 23.157 + 23.158 + target_ref = None 23.159 + for dscsi_ref in server.xenapi.VM.get_DSCSIs(get_single_vm(dom)): 23.160 + if v_hctl == server.xenapi.DSCSI.get_virtual_HCTL(dscsi_ref): 23.161 + target_ref = dscsi_ref 23.162 + break 23.163 + if target_ref is None: 23.164 + raise OptionError("Device %s not assigned" % v_hctl) 23.165 + 23.166 + server.xenapi.DSCSI.destroy(target_ref) 23.167 + 23.168 + else: 23.169 + server.xend.domain.device_configure(dom, scsi) 23.170 23.171 def xm_vnet_list(args): 23.172 xenapi_unsupported()
24.1 --- a/tools/python/xen/xm/xenapi_create.py Wed Oct 22 11:38:22 2008 +0900 24.2 +++ b/tools/python/xen/xm/xenapi_create.py Wed Oct 22 11:46:55 2008 +0900 24.3 @@ -375,6 +375,12 @@ class xenapi_create: 24.4 24.5 self.create_pcis(vm_ref, pcis) 24.6 24.7 + # Now create scsis 24.8 + 24.9 + scsis = vm.getElementsByTagName("vscsi") 24.10 + 24.11 + self.create_scsis(vm_ref, scsis) 24.12 + 24.13 return vm_ref 24.14 except: 24.15 server.xenapi.VM.destroy(vm_ref) 24.16 @@ -532,6 +538,33 @@ class xenapi_create: 24.17 24.18 return server.xenapi.DPCI.create(dpci_record) 24.19 24.20 + def create_scsis(self, vm_ref, scsis): 24.21 + log(DEBUG, "create_scsis") 24.22 + return map(lambda scsi: self.create_scsi(vm_ref, scsi), scsis) 24.23 + 24.24 + def create_scsi(self, vm_ref, scsi): 24.25 + log(DEBUG, "create_scsi") 24.26 + 24.27 + target_ref = None 24.28 + for pscsi_ref in server.xenapi.PSCSI.get_all(): 24.29 + if scsi.attributes["p-dev"].value == server.xenapi.PSCSI.get_physical_HCTL(pscsi_ref): 24.30 + target_ref = pscsi_ref 24.31 + break 24.32 + if target_ref is None: 24.33 + log(DEBUG, "create_scsi: scsi device not found") 24.34 + return None 24.35 + 24.36 + dscsi_record = { 24.37 + "VM": 24.38 + vm_ref, 24.39 + "PSCSI": 24.40 + target_ref, 24.41 + "virtual_HCTL": 24.42 + scsi.attributes["v-dev"].value 24.43 + } 24.44 + 24.45 + return server.xenapi.DSCSI.create(dscsi_record) 24.46 + 24.47 def get_child_by_name(exp, childname, default = None): 24.48 try: 24.49 return [child for child in sxp.children(exp) 24.50 @@ -563,6 +596,9 @@ class sxp2xml: 24.51 pcis_sxp = map(lambda x: x[1], [device for device in devices 24.52 if device[1][0] == "pci"]) 24.53 24.54 + scsis_sxp = map(lambda x: x[1], [device for device in devices 24.55 + if device[1][0] == "vscsi"]) 24.56 + 24.57 # Create XML Document 24.58 24.59 impl = getDOMImplementation() 24.60 @@ -704,6 +740,12 @@ class sxp2xml: 24.61 24.62 map(vm.appendChild, pcis) 24.63 24.64 + # And now the scsis 24.65 + 24.66 + scsis = self.extract_scsis(scsis_sxp, document) 24.67 + 24.68 + map(vm.appendChild, scsis) 24.69 + 24.70 # Last but not least the consoles... 24.71 24.72 consoles = self.extract_consoles(image, document) 24.73 @@ -894,6 +936,23 @@ class sxp2xml: 24.74 24.75 return pcis 24.76 24.77 + def extract_scsis(self, scsis_sxp, document): 24.78 + 24.79 + scsis = [] 24.80 + 24.81 + for scsi_sxp in scsis_sxp: 24.82 + for dev_sxp in sxp.children(scsi_sxp, "dev"): 24.83 + scsi = document.createElement("vscsi") 24.84 + 24.85 + scsi.attributes["p-dev"] \ 24.86 + = get_child_by_name(dev_sxp, "p-dev") 24.87 + scsi.attributes["v-dev"] \ 24.88 + = get_child_by_name(dev_sxp, "v-dev") 24.89 + 24.90 + scsis.append(scsi) 24.91 + 24.92 + return scsis 24.93 + 24.94 def mk_other_config(self, key, value, document): 24.95 other_config = document.createElement("other_config") 24.96 other_config.attributes["key"] = key
25.1 --- a/tools/xentrace/formats Wed Oct 22 11:38:22 2008 +0900 25.2 +++ b/tools/xentrace/formats Wed Oct 22 11:46:55 2008 +0900 25.3 @@ -116,3 +116,7 @@ 0x0040f00e CPU%(cpu)d %(tsc)d (+%(relt 25.4 0x0040f10e CPU%(cpu)d %(tsc)d (+%(reltsc)8d) shadow_emulate_resync_full [ gfn = 0x%(1)16x ] 25.5 0x0040f00f CPU%(cpu)d %(tsc)d (+%(reltsc)8d) shadow_emulate_resync_only [ gfn = 0x%(1)08x ] 25.6 0x0040f10f CPU%(cpu)d %(tsc)d (+%(reltsc)8d) shadow_emulate_resync_only [ gfn = 0x%(1)16x ] 25.7 + 25.8 +0x00801001 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) cpu_freq_change [ %(1)dMHz -> %(2)dMHz ] 25.9 +0x00802001 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) cpu_idle_entry [ C0 -> C%(1)d ] 25.10 +0x00802002 CPU%(cpu)d %(tsc)d (+%(reltsc)8d) cpu_idle_exit [ C%(1)d -> C0 ]
26.1 --- a/unmodified_drivers/linux-2.6/platform-pci/machine_reboot.c Wed Oct 22 11:38:22 2008 +0900 26.2 +++ b/unmodified_drivers/linux-2.6/platform-pci/machine_reboot.c Wed Oct 22 11:46:55 2008 +0900 26.3 @@ -11,12 +11,6 @@ struct ap_suspend_info { 26.4 atomic_t nr_spinning; 26.5 }; 26.6 26.7 -/* 26.8 - * Use a rwlock to protect the hypercall page from being executed in AP context 26.9 - * while the BSP is re-initializing it after restore. 26.10 - */ 26.11 -static DEFINE_RWLOCK(suspend_lock); 26.12 - 26.13 #ifdef CONFIG_SMP 26.14 26.15 /* 26.16 @@ -33,12 +27,8 @@ static void ap_suspend(void *_info) 26.17 atomic_inc(&info->nr_spinning); 26.18 mb(); 26.19 26.20 - while (info->do_spin) { 26.21 + while (info->do_spin) 26.22 cpu_relax(); 26.23 - read_lock(&suspend_lock); 26.24 - HYPERVISOR_yield(); 26.25 - read_unlock(&suspend_lock); 26.26 - } 26.27 26.28 mb(); 26.29 atomic_dec(&info->nr_spinning); 26.30 @@ -61,9 +51,7 @@ static int bp_suspend(void) 26.31 suspend_cancelled = HYPERVISOR_suspend(0); 26.32 26.33 if (!suspend_cancelled) { 26.34 - write_lock(&suspend_lock); 26.35 platform_pci_resume(); 26.36 - write_unlock(&suspend_lock); 26.37 gnttab_resume(); 26.38 irq_resume(); 26.39 }
27.1 --- a/xen/arch/ia64/vmx/vmmu.c Wed Oct 22 11:38:22 2008 +0900 27.2 +++ b/xen/arch/ia64/vmx/vmmu.c Wed Oct 22 11:46:55 2008 +0900 27.3 @@ -446,7 +446,7 @@ IA64FAULT vmx_vcpu_ptc_ga(VCPU *vcpu, u6 27.4 do { 27.5 cpu = v->processor; 27.6 if (cpu != current->processor) { 27.7 - spin_unlock_wait(&per_cpu(schedule_data, cpu).schedule_lock); 27.8 + spin_barrier(&per_cpu(schedule_data, cpu).schedule_lock); 27.9 /* Flush VHPT on remote processors. */ 27.10 smp_call_function_single(cpu, &ptc_ga_remote_func, 27.11 &args, 0, 1);
28.1 --- a/xen/arch/x86/acpi/cpu_idle.c Wed Oct 22 11:38:22 2008 +0900 28.2 +++ b/xen/arch/x86/acpi/cpu_idle.c Wed Oct 22 11:46:55 2008 +0900 28.3 @@ -40,6 +40,7 @@ 28.4 #include <xen/guest_access.h> 28.5 #include <xen/keyhandler.h> 28.6 #include <xen/cpuidle.h> 28.7 +#include <xen/trace.h> 28.8 #include <asm/cache.h> 28.9 #include <asm/io.h> 28.10 #include <asm/hpet.h> 28.11 @@ -251,6 +252,9 @@ static void acpi_processor_idle(void) 28.12 switch ( cx->type ) 28.13 { 28.14 case ACPI_STATE_C1: 28.15 + /* Trace cpu idle entry */ 28.16 + TRACE_1D(TRC_PM_IDLE_ENTRY, 1); 28.17 + 28.18 /* 28.19 * Invoke C1. 28.20 * Use the appropriate idle routine, the one that would 28.21 @@ -261,6 +265,9 @@ static void acpi_processor_idle(void) 28.22 else 28.23 acpi_safe_halt(); 28.24 28.25 + /* Trace cpu idle exit */ 28.26 + TRACE_1D(TRC_PM_IDLE_EXIT, 1); 28.27 + 28.28 /* 28.29 * TBD: Can't get time duration while in C1, as resumes 28.30 * go to an ISR rather than here. Need to instrument 28.31 @@ -272,12 +279,16 @@ static void acpi_processor_idle(void) 28.32 case ACPI_STATE_C2: 28.33 if ( local_apic_timer_c2_ok ) 28.34 { 28.35 + /* Trace cpu idle entry */ 28.36 + TRACE_1D(TRC_PM_IDLE_ENTRY, 2); 28.37 /* Get start time (ticks) */ 28.38 t1 = inl(pmtmr_ioport); 28.39 /* Invoke C2 */ 28.40 acpi_idle_do_entry(cx); 28.41 /* Get end time (ticks) */ 28.42 t2 = inl(pmtmr_ioport); 28.43 + /* Trace cpu idle exit */ 28.44 + TRACE_1D(TRC_PM_IDLE_EXIT, 2); 28.45 28.46 /* Re-enable interrupts */ 28.47 local_irq_enable(); 28.48 @@ -316,6 +327,8 @@ static void acpi_processor_idle(void) 28.49 ACPI_FLUSH_CPU_CACHE(); 28.50 } 28.51 28.52 + /* Trace cpu idle entry */ 28.53 + TRACE_1D(TRC_PM_IDLE_ENTRY, cx - &power->states[0]); 28.54 /* 28.55 * Before invoking C3, be aware that TSC/APIC timer may be 28.56 * stopped by H/W. Without carefully handling of TSC/APIC stop issues, 28.57 @@ -335,6 +348,8 @@ static void acpi_processor_idle(void) 28.58 28.59 /* recovering TSC */ 28.60 cstate_restore_tsc(); 28.61 + /* Trace cpu idle exit */ 28.62 + TRACE_1D(TRC_PM_IDLE_EXIT, cx - &power->states[0]); 28.63 28.64 if ( power->flags.bm_check && power->flags.bm_control ) 28.65 {
29.1 --- a/xen/arch/x86/cpu/amd.c Wed Oct 22 11:38:22 2008 +0900 29.2 +++ b/xen/arch/x86/cpu/amd.c Wed Oct 22 11:46:55 2008 +0900 29.3 @@ -37,8 +37,8 @@ static unsigned int opt_cpuid_mask_ecx, 29.4 integer_param("cpuid_mask_ecx", opt_cpuid_mask_ecx); 29.5 integer_param("cpuid_mask_edx", opt_cpuid_mask_edx); 29.6 static unsigned int opt_cpuid_mask_ext_ecx, opt_cpuid_mask_ext_edx; 29.7 -integer_param("cpuid_mask_ecx", opt_cpuid_mask_ext_ecx); 29.8 -integer_param("cpuid_mask_edx", opt_cpuid_mask_ext_edx); 29.9 +integer_param("cpuid_mask_ext_ecx", opt_cpuid_mask_ext_ecx); 29.10 +integer_param("cpuid_mask_ext_edx", opt_cpuid_mask_ext_edx); 29.11 29.12 static inline void wrmsr_amd(unsigned int index, unsigned int lo, 29.13 unsigned int hi)
30.1 --- a/xen/arch/x86/domain.c Wed Oct 22 11:38:22 2008 +0900 30.2 +++ b/xen/arch/x86/domain.c Wed Oct 22 11:46:55 2008 +0900 30.3 @@ -575,7 +575,10 @@ int arch_set_info_guest( 30.4 v->arch.guest_context.user_regs.eflags |= 2; 30.5 30.6 if ( is_hvm_vcpu(v) ) 30.7 + { 30.8 + hvm_set_info_guest(v); 30.9 goto out; 30.10 + } 30.11 30.12 /* Only CR0.TS is modifiable by guest or admin. */ 30.13 v->arch.guest_context.ctrlreg[0] &= X86_CR0_TS; 30.14 @@ -1252,11 +1255,11 @@ void context_switch(struct vcpu *prev, s 30.15 flush_tlb_mask(next->vcpu_dirty_cpumask); 30.16 } 30.17 30.18 - local_irq_disable(); 30.19 - 30.20 if ( is_hvm_vcpu(prev) && !list_empty(&prev->arch.hvm_vcpu.tm_list) ) 30.21 pt_save_timer(prev); 30.22 30.23 + local_irq_disable(); 30.24 + 30.25 set_current(next); 30.26 30.27 if ( (per_cpu(curr_vcpu, cpu) == next) || is_idle_vcpu(next) )
31.1 --- a/xen/arch/x86/domctl.c Wed Oct 22 11:38:22 2008 +0900 31.2 +++ b/xen/arch/x86/domctl.c Wed Oct 22 11:46:55 2008 +0900 31.3 @@ -1028,6 +1028,21 @@ long arch_do_domctl( 31.4 } 31.5 break; 31.6 31.7 + case XEN_DOMCTL_suppress_spurious_page_faults: 31.8 + { 31.9 + struct domain *d; 31.10 + 31.11 + ret = -ESRCH; 31.12 + d = rcu_lock_domain_by_id(domctl->domain); 31.13 + if ( d != NULL ) 31.14 + { 31.15 + d->arch.suppress_spurious_page_faults = 1; 31.16 + rcu_unlock_domain(d); 31.17 + ret = 0; 31.18 + } 31.19 + } 31.20 + break; 31.21 + 31.22 default: 31.23 ret = -ENOSYS; 31.24 break;
32.1 --- a/xen/arch/x86/hvm/svm/emulate.c Wed Oct 22 11:38:22 2008 +0900 32.2 +++ b/xen/arch/x86/hvm/svm/emulate.c Wed Oct 22 11:46:55 2008 +0900 32.3 @@ -61,6 +61,34 @@ static unsigned long svm_rip2pointer(str 32.4 return p; 32.5 } 32.6 32.7 +static unsigned long svm_nextrip_insn_length(struct vcpu *v) 32.8 +{ 32.9 + struct vmcb_struct *vmcb = v->arch.hvm_svm.vmcb; 32.10 + 32.11 + if ( !cpu_has_svm_nrips || (vmcb->nextrip <= vmcb->rip) ) 32.12 + return 0; 32.13 + 32.14 +#ifndef NDEBUG 32.15 + switch ( vmcb->exitcode ) 32.16 + { 32.17 + case VMEXIT_CR0_READ... VMEXIT_DR15_WRITE: 32.18 + /* faults due to instruction intercepts */ 32.19 + /* (exitcodes 84-95) are reserved */ 32.20 + case VMEXIT_IDTR_READ ... VMEXIT_TR_WRITE: 32.21 + case VMEXIT_RDTSC ... VMEXIT_MSR: 32.22 + case VMEXIT_VMRUN ... VMEXIT_MWAIT_CONDITIONAL: 32.23 + /* ...and the rest of the #VMEXITs */ 32.24 + case VMEXIT_CR0_SEL_WRITE: 32.25 + case VMEXIT_EXCEPTION_BP: 32.26 + break; 32.27 + default: 32.28 + BUG(); 32.29 + } 32.30 +#endif 32.31 + 32.32 + return vmcb->nextrip - vmcb->rip; 32.33 +} 32.34 + 32.35 /* First byte: Length. Following bytes: Opcode bytes. */ 32.36 #define MAKE_INSTR(nm, ...) static const u8 OPCODE_##nm[] = { __VA_ARGS__ } 32.37 MAKE_INSTR(INVD, 2, 0x0f, 0x08); 32.38 @@ -118,6 +146,9 @@ int __get_instruction_length_from_list(s 32.39 unsigned long fetch_addr; 32.40 unsigned int fetch_len; 32.41 32.42 + if ( (inst_len = svm_nextrip_insn_length(v)) != 0 ) 32.43 + return inst_len; 32.44 + 32.45 /* Fetch up to the next page break; we'll fetch from the next page 32.46 * later if we have to. */ 32.47 fetch_addr = svm_rip2pointer(v);
33.1 --- a/xen/arch/x86/hvm/svm/intr.c Wed Oct 22 11:38:22 2008 +0900 33.2 +++ b/xen/arch/x86/hvm/svm/intr.c Wed Oct 22 11:46:55 2008 +0900 33.3 @@ -100,61 +100,6 @@ static void enable_intr_window(struct vc 33.4 vmcb->general1_intercepts |= GENERAL1_INTERCEPT_VINTR; 33.5 } 33.6 33.7 -extern int vmsi_deliver(struct domain *d, int pirq); 33.8 -static int hvm_pci_msi_assert(struct domain *d, int pirq) 33.9 -{ 33.10 - return vmsi_deliver(d, pirq); 33.11 -} 33.12 - 33.13 -static void svm_dirq_assist(struct vcpu *v) 33.14 -{ 33.15 - unsigned int irq; 33.16 - uint32_t device, intx; 33.17 - struct domain *d = v->domain; 33.18 - struct hvm_irq_dpci *hvm_irq_dpci = d->arch.hvm_domain.irq.dpci; 33.19 - struct dev_intx_gsi_link *digl; 33.20 - 33.21 - if ( !iommu_enabled || (v->vcpu_id != 0) || (hvm_irq_dpci == NULL) ) 33.22 - return; 33.23 - 33.24 - for ( irq = find_first_bit(hvm_irq_dpci->dirq_mask, NR_IRQS); 33.25 - irq < NR_IRQS; 33.26 - irq = find_next_bit(hvm_irq_dpci->dirq_mask, NR_IRQS, irq + 1) ) 33.27 - { 33.28 - if ( !test_and_clear_bit(irq, &hvm_irq_dpci->dirq_mask) ) 33.29 - continue; 33.30 - 33.31 - spin_lock(&d->event_lock); 33.32 - if ( test_bit(_HVM_IRQ_DPCI_MSI, &hvm_irq_dpci->mirq[irq].flags) ) 33.33 - { 33.34 - hvm_pci_msi_assert(d, irq); 33.35 - spin_unlock(&d->event_lock); 33.36 - continue; 33.37 - } 33.38 - 33.39 - stop_timer(&hvm_irq_dpci->hvm_timer[domain_irq_to_vector(d, irq)]); 33.40 - 33.41 - list_for_each_entry ( digl, &hvm_irq_dpci->mirq[irq].digl_list, list ) 33.42 - { 33.43 - device = digl->device; 33.44 - intx = digl->intx; 33.45 - hvm_pci_intx_assert(d, device, intx); 33.46 - hvm_irq_dpci->mirq[irq].pending++; 33.47 - } 33.48 - 33.49 - /* 33.50 - * Set a timer to see if the guest can finish the interrupt or not. For 33.51 - * example, the guest OS may unmask the PIC during boot, before the 33.52 - * guest driver is loaded. hvm_pci_intx_assert() may succeed, but the 33.53 - * guest will never deal with the irq, then the physical interrupt line 33.54 - * will never be deasserted. 33.55 - */ 33.56 - set_timer(&hvm_irq_dpci->hvm_timer[domain_irq_to_vector(d, irq)], 33.57 - NOW() + PT_IRQ_TIME_OUT); 33.58 - spin_unlock(&d->event_lock); 33.59 - } 33.60 -} 33.61 - 33.62 asmlinkage void svm_intr_assist(void) 33.63 { 33.64 struct vcpu *v = current; 33.65 @@ -163,7 +108,7 @@ asmlinkage void svm_intr_assist(void) 33.66 33.67 /* Crank the handle on interrupt state. */ 33.68 pt_update_irq(v); 33.69 - svm_dirq_assist(v); 33.70 + hvm_dirq_assist(v); 33.71 33.72 do { 33.73 intack = hvm_vcpu_has_pending_irq(v);
34.1 --- a/xen/arch/x86/hvm/viridian.c Wed Oct 22 11:38:22 2008 +0900 34.2 +++ b/xen/arch/x86/hvm/viridian.c Wed Oct 22 11:46:55 2008 +0900 34.3 @@ -244,7 +244,6 @@ int rdmsr_viridian_regs(uint32_t idx, ui 34.4 34.5 int viridian_hypercall(struct cpu_user_regs *regs) 34.6 { 34.7 - struct domain *d = current->domain; 34.8 int mode = hvm_guest_x86_mode(current); 34.9 unsigned long input_params_gpa, output_params_gpa; 34.10 uint16_t status = HV_STATUS_SUCCESS; 34.11 @@ -271,7 +270,7 @@ int viridian_hypercall(struct cpu_user_r 34.12 }; 34.13 } output = { 0 }; 34.14 34.15 - ASSERT(is_viridian_domain(d)); 34.16 + ASSERT(is_viridian_domain(current->domain)); 34.17 34.18 switch ( mode ) 34.19 {
35.1 --- a/xen/arch/x86/hvm/vmx/intr.c Wed Oct 22 11:38:22 2008 +0900 35.2 +++ b/xen/arch/x86/hvm/vmx/intr.c Wed Oct 22 11:46:55 2008 +0900 35.3 @@ -103,61 +103,6 @@ static void enable_intr_window(struct vc 35.4 } 35.5 } 35.6 35.7 -extern int vmsi_deliver(struct domain *d, int pirq); 35.8 -static int hvm_pci_msi_assert(struct domain *d, int pirq) 35.9 -{ 35.10 - return vmsi_deliver(d, pirq); 35.11 -} 35.12 - 35.13 -static void vmx_dirq_assist(struct vcpu *v) 35.14 -{ 35.15 - unsigned int irq; 35.16 - uint32_t device, intx; 35.17 - struct domain *d = v->domain; 35.18 - struct hvm_irq_dpci *hvm_irq_dpci = d->arch.hvm_domain.irq.dpci; 35.19 - struct dev_intx_gsi_link *digl; 35.20 - 35.21 - if ( !iommu_enabled || (v->vcpu_id != 0) || (hvm_irq_dpci == NULL) ) 35.22 - return; 35.23 - 35.24 - for ( irq = find_first_bit(hvm_irq_dpci->dirq_mask, NR_IRQS); 35.25 - irq < NR_IRQS; 35.26 - irq = find_next_bit(hvm_irq_dpci->dirq_mask, NR_IRQS, irq + 1) ) 35.27 - { 35.28 - if ( !test_and_clear_bit(irq, &hvm_irq_dpci->dirq_mask) ) 35.29 - continue; 35.30 - 35.31 - spin_lock(&d->event_lock); 35.32 - if ( test_bit(_HVM_IRQ_DPCI_MSI, &hvm_irq_dpci->mirq[irq].flags) ) 35.33 - { 35.34 - hvm_pci_msi_assert(d, irq); 35.35 - spin_unlock(&d->event_lock); 35.36 - continue; 35.37 - } 35.38 - 35.39 - stop_timer(&hvm_irq_dpci->hvm_timer[domain_irq_to_vector(d, irq)]); 35.40 - 35.41 - list_for_each_entry ( digl, &hvm_irq_dpci->mirq[irq].digl_list, list ) 35.42 - { 35.43 - device = digl->device; 35.44 - intx = digl->intx; 35.45 - hvm_pci_intx_assert(d, device, intx); 35.46 - hvm_irq_dpci->mirq[irq].pending++; 35.47 - } 35.48 - 35.49 - /* 35.50 - * Set a timer to see if the guest can finish the interrupt or not. For 35.51 - * example, the guest OS may unmask the PIC during boot, before the 35.52 - * guest driver is loaded. hvm_pci_intx_assert() may succeed, but the 35.53 - * guest will never deal with the irq, then the physical interrupt line 35.54 - * will never be deasserted. 35.55 - */ 35.56 - set_timer(&hvm_irq_dpci->hvm_timer[domain_irq_to_vector(d, irq)], 35.57 - NOW() + PT_IRQ_TIME_OUT); 35.58 - spin_unlock(&d->event_lock); 35.59 - } 35.60 -} 35.61 - 35.62 asmlinkage void vmx_intr_assist(void) 35.63 { 35.64 struct hvm_intack intack; 35.65 @@ -167,7 +112,7 @@ asmlinkage void vmx_intr_assist(void) 35.66 35.67 /* Crank the handle on interrupt state. */ 35.68 pt_update_irq(v); 35.69 - vmx_dirq_assist(v); 35.70 + hvm_dirq_assist(v); 35.71 35.72 do { 35.73 intack = hvm_vcpu_has_pending_irq(v);
36.1 --- a/xen/arch/x86/hvm/vmx/vmx.c Wed Oct 22 11:38:22 2008 +0900 36.2 +++ b/xen/arch/x86/hvm/vmx/vmx.c Wed Oct 22 11:46:55 2008 +0900 36.3 @@ -1184,6 +1184,13 @@ static void vmx_set_uc_mode(struct vcpu 36.4 vpid_sync_all(); 36.5 } 36.6 36.7 +static void vmx_set_info_guest(struct vcpu *v) 36.8 +{ 36.9 + vmx_vmcs_enter(v); 36.10 + __vmwrite(GUEST_DR7, v->arch.guest_context.debugreg[7]); 36.11 + vmx_vmcs_exit(v); 36.12 +} 36.13 + 36.14 static struct hvm_function_table vmx_function_table = { 36.15 .name = "VMX", 36.16 .domain_initialise = vmx_domain_initialise, 36.17 @@ -1214,7 +1221,8 @@ static struct hvm_function_table vmx_fun 36.18 .msr_read_intercept = vmx_msr_read_intercept, 36.19 .msr_write_intercept = vmx_msr_write_intercept, 36.20 .invlpg_intercept = vmx_invlpg_intercept, 36.21 - .set_uc_mode = vmx_set_uc_mode 36.22 + .set_uc_mode = vmx_set_uc_mode, 36.23 + .set_info_guest = vmx_set_info_guest 36.24 }; 36.25 36.26 static unsigned long *vpid_bitmap; 36.27 @@ -2048,8 +2056,12 @@ asmlinkage void vmx_vmexit_handler(struc 36.28 36.29 perfc_incra(vmexits, exit_reason); 36.30 36.31 - if ( exit_reason != EXIT_REASON_EXTERNAL_INTERRUPT ) 36.32 - local_irq_enable(); 36.33 + /* Handle the interrupt we missed before allowing any more in. */ 36.34 + if ( exit_reason == EXIT_REASON_EXTERNAL_INTERRUPT ) 36.35 + vmx_do_extint(regs); 36.36 + 36.37 + /* Now enable interrupts so it's safe to take locks. */ 36.38 + local_irq_enable(); 36.39 36.40 if ( unlikely(exit_reason & VMX_EXIT_REASONS_FAILED_VMENTRY) ) 36.41 return vmx_failed_vmentry(exit_reason, regs); 36.42 @@ -2177,7 +2189,7 @@ asmlinkage void vmx_vmexit_handler(struc 36.43 break; 36.44 } 36.45 case EXIT_REASON_EXTERNAL_INTERRUPT: 36.46 - vmx_do_extint(regs); 36.47 + /* Already handled above. */ 36.48 break; 36.49 case EXIT_REASON_TRIPLE_FAULT: 36.50 hvm_triple_fault();
37.1 --- a/xen/arch/x86/irq.c Wed Oct 22 11:38:22 2008 +0900 37.2 +++ b/xen/arch/x86/irq.c Wed Oct 22 11:46:55 2008 +0900 37.3 @@ -510,7 +510,7 @@ int pirq_guest_bind(struct vcpu *v, int 37.4 { 37.5 unsigned int vector; 37.6 irq_desc_t *desc; 37.7 - irq_guest_action_t *action; 37.8 + irq_guest_action_t *action, *newaction = NULL; 37.9 int rc = 0; 37.10 cpumask_t cpumask = CPU_MASK_NONE; 37.11 37.12 @@ -520,7 +520,10 @@ int pirq_guest_bind(struct vcpu *v, int 37.13 retry: 37.14 desc = domain_spin_lock_irq_desc(v->domain, irq, NULL); 37.15 if ( desc == NULL ) 37.16 - return -EINVAL; 37.17 + { 37.18 + rc = -EINVAL; 37.19 + goto out; 37.20 + } 37.21 37.22 action = (irq_guest_action_t *)desc->action; 37.23 vector = desc - irq_desc; 37.24 @@ -533,18 +536,24 @@ int pirq_guest_bind(struct vcpu *v, int 37.25 "Cannot bind IRQ %d to guest. In use by '%s'.\n", 37.26 irq, desc->action->name); 37.27 rc = -EBUSY; 37.28 + goto unlock_out; 37.29 + } 37.30 + 37.31 + if ( newaction == NULL ) 37.32 + { 37.33 + spin_unlock_irq(&desc->lock); 37.34 + if ( (newaction = xmalloc(irq_guest_action_t)) != NULL ) 37.35 + goto retry; 37.36 + gdprintk(XENLOG_INFO, 37.37 + "Cannot bind IRQ %d to guest. Out of memory.\n", 37.38 + irq); 37.39 + rc = -ENOMEM; 37.40 goto out; 37.41 } 37.42 37.43 - action = xmalloc(irq_guest_action_t); 37.44 - if ( (desc->action = (struct irqaction *)action) == NULL ) 37.45 - { 37.46 - gdprintk(XENLOG_INFO, 37.47 - "Cannot bind IRQ %d to guest. Out of memory.\n", 37.48 - irq); 37.49 - rc = -ENOMEM; 37.50 - goto out; 37.51 - } 37.52 + action = newaction; 37.53 + desc->action = (struct irqaction *)action; 37.54 + newaction = NULL; 37.55 37.56 action->nr_guests = 0; 37.57 action->in_flight = 0; 37.58 @@ -568,7 +577,7 @@ int pirq_guest_bind(struct vcpu *v, int 37.59 "Will not share with others.\n", 37.60 irq); 37.61 rc = -EBUSY; 37.62 - goto out; 37.63 + goto unlock_out; 37.64 } 37.65 else if ( action->nr_guests == 0 ) 37.66 { 37.67 @@ -588,17 +597,21 @@ int pirq_guest_bind(struct vcpu *v, int 37.68 gdprintk(XENLOG_INFO, "Cannot bind IRQ %d to guest. " 37.69 "Already at max share.\n", irq); 37.70 rc = -EBUSY; 37.71 - goto out; 37.72 + goto unlock_out; 37.73 } 37.74 37.75 action->guest[action->nr_guests++] = v->domain; 37.76 37.77 + unlock_out: 37.78 + spin_unlock_irq(&desc->lock); 37.79 out: 37.80 - spin_unlock_irq(&desc->lock); 37.81 + if ( newaction != NULL ) 37.82 + xfree(newaction); 37.83 return rc; 37.84 } 37.85 37.86 -static void __pirq_guest_unbind(struct domain *d, int irq, irq_desc_t *desc) 37.87 +static irq_guest_action_t *__pirq_guest_unbind( 37.88 + struct domain *d, int irq, irq_desc_t *desc) 37.89 { 37.90 unsigned int vector; 37.91 irq_guest_action_t *action; 37.92 @@ -644,7 +657,7 @@ static void __pirq_guest_unbind(struct d 37.93 BUG_ON(test_bit(irq, d->pirq_mask)); 37.94 37.95 if ( action->nr_guests != 0 ) 37.96 - return; 37.97 + return NULL; 37.98 37.99 BUG_ON(action->in_flight != 0); 37.100 37.101 @@ -672,15 +685,18 @@ static void __pirq_guest_unbind(struct d 37.102 BUG_ON(!cpus_empty(action->cpu_eoi_map)); 37.103 37.104 desc->action = NULL; 37.105 - xfree(action); 37.106 desc->status &= ~IRQ_GUEST; 37.107 desc->status &= ~IRQ_INPROGRESS; 37.108 kill_timer(&irq_guest_eoi_timer[vector]); 37.109 desc->handler->shutdown(vector); 37.110 + 37.111 + /* Caller frees the old guest descriptor block. */ 37.112 + return action; 37.113 } 37.114 37.115 void pirq_guest_unbind(struct domain *d, int irq) 37.116 { 37.117 + irq_guest_action_t *oldaction = NULL; 37.118 irq_desc_t *desc; 37.119 int vector; 37.120 37.121 @@ -699,16 +715,19 @@ void pirq_guest_unbind(struct domain *d, 37.122 } 37.123 else 37.124 { 37.125 - __pirq_guest_unbind(d, irq, desc); 37.126 + oldaction = __pirq_guest_unbind(d, irq, desc); 37.127 } 37.128 37.129 spin_unlock_irq(&desc->lock); 37.130 + 37.131 + if ( oldaction != NULL ) 37.132 + xfree(oldaction); 37.133 } 37.134 37.135 int pirq_guest_force_unbind(struct domain *d, int irq) 37.136 { 37.137 irq_desc_t *desc; 37.138 - irq_guest_action_t *action; 37.139 + irq_guest_action_t *action, *oldaction = NULL; 37.140 int i, bound = 0; 37.141 37.142 WARN_ON(!spin_is_locked(&d->event_lock)); 37.143 @@ -727,10 +746,14 @@ int pirq_guest_force_unbind(struct domai 37.144 goto out; 37.145 37.146 bound = 1; 37.147 - __pirq_guest_unbind(d, irq, desc); 37.148 + oldaction = __pirq_guest_unbind(d, irq, desc); 37.149 37.150 out: 37.151 spin_unlock_irq(&desc->lock); 37.152 + 37.153 + if ( oldaction != NULL ) 37.154 + xfree(oldaction); 37.155 + 37.156 return bound; 37.157 } 37.158
38.1 --- a/xen/arch/x86/mm/hap/hap.c Wed Oct 22 11:38:22 2008 +0900 38.2 +++ b/xen/arch/x86/mm/hap/hap.c Wed Oct 22 11:46:55 2008 +0900 38.3 @@ -639,9 +639,16 @@ static void 38.4 hap_write_p2m_entry(struct vcpu *v, unsigned long gfn, l1_pgentry_t *p, 38.5 mfn_t table_mfn, l1_pgentry_t new, unsigned int level) 38.6 { 38.7 + uint32_t old_flags; 38.8 + 38.9 hap_lock(v->domain); 38.10 38.11 + old_flags = l1e_get_flags(*p); 38.12 safe_write_pte(p, new); 38.13 + if ( (old_flags & _PAGE_PRESENT) 38.14 + && (level == 1 || (level == 2 && (old_flags & _PAGE_PSE))) ) 38.15 + flush_tlb_mask(v->domain->domain_dirty_cpumask); 38.16 + 38.17 #if CONFIG_PAGING_LEVELS == 3 38.18 /* install P2M in monitor table for PAE Xen */ 38.19 if ( level == 3 )
39.1 --- a/xen/arch/x86/mm/shadow/private.h Wed Oct 22 11:38:22 2008 +0900 39.2 +++ b/xen/arch/x86/mm/shadow/private.h Wed Oct 22 11:46:55 2008 +0900 39.3 @@ -227,32 +227,40 @@ extern void shadow_audit_tables(struct v 39.4 struct shadow_page_info 39.5 { 39.6 union { 39.7 - /* When in use, guest page we're a shadow of */ 39.8 - unsigned long backpointer; 39.9 - /* When free, order of the freelist we're on */ 39.10 - unsigned int order; 39.11 - }; 39.12 - union { 39.13 - /* When in use, next shadow in this hash chain */ 39.14 - struct shadow_page_info *next_shadow; 39.15 - /* When free, TLB flush time when freed */ 39.16 - u32 tlbflush_timestamp; 39.17 - }; 39.18 - struct { 39.19 - unsigned int type:5; /* What kind of shadow is this? */ 39.20 - unsigned int pinned:1; /* Is the shadow pinned? */ 39.21 - unsigned int count:26; /* Reference count */ 39.22 - u32 mbz; /* Must be zero: this is where the owner 39.23 - * field lives in a non-shadow page */ 39.24 - } __attribute__((packed)); 39.25 - union { 39.26 - /* For unused shadow pages, a list of pages of this order; 39.27 - * for pinnable shadows, if pinned, a list of other pinned shadows 39.28 - * (see sh_type_is_pinnable() below for the definition of 39.29 - * "pinnable" shadow types). */ 39.30 - struct list_head list; 39.31 - /* For non-pinnable shadows, a higher entry that points at us */ 39.32 - paddr_t up; 39.33 + /* Ensures that shadow_page_info is same size as page_info. */ 39.34 + struct page_info page_info; 39.35 + 39.36 + struct { 39.37 + union { 39.38 + /* When in use, guest page we're a shadow of */ 39.39 + unsigned long backpointer; 39.40 + /* When free, order of the freelist we're on */ 39.41 + unsigned int order; 39.42 + }; 39.43 + union { 39.44 + /* When in use, next shadow in this hash chain */ 39.45 + struct shadow_page_info *next_shadow; 39.46 + /* When free, TLB flush time when freed */ 39.47 + u32 tlbflush_timestamp; 39.48 + }; 39.49 + struct { 39.50 + unsigned int type:5; /* What kind of shadow is this? */ 39.51 + unsigned int pinned:1; /* Is the shadow pinned? */ 39.52 + unsigned int count:26; /* Reference count */ 39.53 + u32 mbz; /* Must be zero: this is where the 39.54 + * owner field lives in page_info */ 39.55 + } __attribute__((packed)); 39.56 + union { 39.57 + /* For unused shadow pages, a list of pages of this order; for 39.58 + * pinnable shadows, if pinned, a list of other pinned shadows 39.59 + * (see sh_type_is_pinnable() below for the definition of 39.60 + * "pinnable" shadow types). */ 39.61 + struct list_head list; 39.62 + /* For non-pinnable shadows, a higher entry that points 39.63 + * at us. */ 39.64 + paddr_t up; 39.65 + }; 39.66 + }; 39.67 }; 39.68 }; 39.69 39.70 @@ -261,7 +269,8 @@ struct shadow_page_info 39.71 * Also, the mbz field must line up with the owner field of normal 39.72 * pages, so they look properly like anonymous/xen pages. */ 39.73 static inline void shadow_check_page_struct_offsets(void) { 39.74 - BUILD_BUG_ON(sizeof (struct shadow_page_info) > sizeof (struct page_info)); 39.75 + BUILD_BUG_ON(sizeof (struct shadow_page_info) != 39.76 + sizeof (struct page_info)); 39.77 BUILD_BUG_ON(offsetof(struct shadow_page_info, mbz) != 39.78 offsetof(struct page_info, u.inuse._domain)); 39.79 };
40.1 --- a/xen/arch/x86/msi.c Wed Oct 22 11:38:22 2008 +0900 40.2 +++ b/xen/arch/x86/msi.c Wed Oct 22 11:46:55 2008 +0900 40.3 @@ -364,6 +364,7 @@ static struct msi_desc* alloc_msi_entry( 40.4 40.5 INIT_LIST_HEAD(&entry->list); 40.6 entry->dev = NULL; 40.7 + entry->remap_index = -1; 40.8 40.9 return entry; 40.10 }
41.1 --- a/xen/arch/x86/nmi.c Wed Oct 22 11:38:22 2008 +0900 41.2 +++ b/xen/arch/x86/nmi.c Wed Oct 22 11:46:55 2008 +0900 41.3 @@ -72,8 +72,8 @@ int nmi_active; 41.4 #define P6_EVNTSEL_INT (1 << 20) 41.5 #define P6_EVNTSEL_OS (1 << 17) 41.6 #define P6_EVNTSEL_USR (1 << 16) 41.7 -#define P6_EVENT_CPU_CLOCKS_NOT_HALTED 0x79 41.8 -#define P6_NMI_EVENT P6_EVENT_CPU_CLOCKS_NOT_HALTED 41.9 +#define P6_EVENT_CPU_CLOCKS_NOT_HALTED 0x79 41.10 +#define CORE_EVENT_CPU_CLOCKS_NOT_HALTED 0x3c 41.11 41.12 #define P4_ESCR_EVENT_SELECT(N) ((N)<<25) 41.13 #define P4_CCCR_OVF_PMI0 (1<<26) 41.14 @@ -122,10 +122,17 @@ int __init check_nmi_watchdog (void) 41.15 41.16 printk("\n"); 41.17 41.18 - /* now that we know it works we can reduce NMI frequency to 41.19 - something more reasonable; makes a difference in some configs */ 41.20 + /* 41.21 + * Now that we know it works we can reduce NMI frequency to 41.22 + * something more reasonable; makes a difference in some configs. 41.23 + * There's a limit to how slow we can go because writing the perfctr 41.24 + * MSRs only sets the low 32 bits, with the top 8 bits sign-extended 41.25 + * from those, so it's not possible to set up a delay larger than 41.26 + * 2^31 cycles and smaller than (2^40 - 2^31) cycles. 41.27 + * (Intel SDM, section 18.22.2) 41.28 + */ 41.29 if ( nmi_watchdog == NMI_LOCAL_APIC ) 41.30 - nmi_hz = 1; 41.31 + nmi_hz = max(1ul, cpu_khz >> 20); 41.32 41.33 return 0; 41.34 } 41.35 @@ -248,7 +255,7 @@ static void __pminit setup_k7_watchdog(v 41.36 wrmsr(MSR_K7_EVNTSEL0, evntsel, 0); 41.37 } 41.38 41.39 -static void __pminit setup_p6_watchdog(void) 41.40 +static void __pminit setup_p6_watchdog(unsigned counter) 41.41 { 41.42 unsigned int evntsel; 41.43 41.44 @@ -260,7 +267,7 @@ static void __pminit setup_p6_watchdog(v 41.45 evntsel = P6_EVNTSEL_INT 41.46 | P6_EVNTSEL_OS 41.47 | P6_EVNTSEL_USR 41.48 - | P6_NMI_EVENT; 41.49 + | counter; 41.50 41.51 wrmsr(MSR_P6_EVNTSEL0, evntsel, 0); 41.52 write_watchdog_counter("P6_PERFCTR0"); 41.53 @@ -326,7 +333,9 @@ void __pminit setup_apic_nmi_watchdog(vo 41.54 case X86_VENDOR_INTEL: 41.55 switch (boot_cpu_data.x86) { 41.56 case 6: 41.57 - setup_p6_watchdog(); 41.58 + setup_p6_watchdog((boot_cpu_data.x86_model < 14) 41.59 + ? P6_EVENT_CPU_CLOCKS_NOT_HALTED 41.60 + : CORE_EVENT_CPU_CLOCKS_NOT_HALTED); 41.61 break; 41.62 case 15: 41.63 if (!setup_p4_watchdog())
42.1 --- a/xen/arch/x86/platform_hypercall.c Wed Oct 22 11:38:22 2008 +0900 42.2 +++ b/xen/arch/x86/platform_hypercall.c Wed Oct 22 11:46:55 2008 +0900 42.3 @@ -53,15 +53,6 @@ static long cpu_frequency_change_helper( 42.4 return cpu_frequency_change(this_cpu(freq)); 42.5 } 42.6 42.7 -int xenpf_copy_px_states(struct processor_performance *pxpt, 42.8 - struct xen_processor_performance *dom0_px_info) 42.9 -{ 42.10 - if (!pxpt || !dom0_px_info) 42.11 - return -EINVAL; 42.12 - return copy_from_compat(pxpt->states, dom0_px_info->states, 42.13 - dom0_px_info->state_count); 42.14 -} 42.15 - 42.16 ret_t do_platform_op(XEN_GUEST_HANDLE(xen_platform_op_t) u_xenpf_op) 42.17 { 42.18 ret_t ret = 0; 42.19 @@ -372,12 +363,13 @@ ret_t do_platform_op(XEN_GUEST_HANDLE(xe 42.20 switch ( op->u.set_pminfo.type ) 42.21 { 42.22 case XEN_PM_PX: 42.23 - { 42.24 - 42.25 - ret = set_px_pminfo(op->u.set_pminfo.id, 42.26 - &op->u.set_pminfo.perf); 42.27 + if ( !(xen_processor_pmbits & XEN_PROCESSOR_PM_PX) ) 42.28 + { 42.29 + ret = -ENOSYS; 42.30 + break; 42.31 + } 42.32 + ret = set_px_pminfo(op->u.set_pminfo.id, &op->u.set_pminfo.perf); 42.33 break; 42.34 - } 42.35 42.36 case XEN_PM_CX: 42.37 if ( !(xen_processor_pmbits & XEN_PROCESSOR_PM_CX) )
43.1 --- a/xen/arch/x86/smpboot.c Wed Oct 22 11:38:22 2008 +0900 43.2 +++ b/xen/arch/x86/smpboot.c Wed Oct 22 11:46:55 2008 +0900 43.3 @@ -473,13 +473,6 @@ static void construct_percpu_idt(unsigne 43.4 { 43.5 unsigned char idt_load[10]; 43.6 43.7 - /* If IDT table exists since last hotplug, reuse it */ 43.8 - if (!idt_tables[cpu]) { 43.9 - idt_tables[cpu] = xmalloc_array(idt_entry_t, IDT_ENTRIES); 43.10 - memcpy(idt_tables[cpu], idt_table, 43.11 - IDT_ENTRIES*sizeof(idt_entry_t)); 43.12 - } 43.13 - 43.14 *(unsigned short *)(&idt_load[0]) = (IDT_ENTRIES*sizeof(idt_entry_t))-1; 43.15 *(unsigned long *)(&idt_load[2]) = (unsigned long)idt_tables[cpu]; 43.16 __asm__ __volatile__ ( "lidt %0" : "=m" (idt_load) ); 43.17 @@ -908,6 +901,12 @@ static int __devinit do_boot_cpu(int api 43.18 } 43.19 #endif 43.20 43.21 + if (!idt_tables[cpu]) { 43.22 + idt_tables[cpu] = xmalloc_array(idt_entry_t, IDT_ENTRIES); 43.23 + memcpy(idt_tables[cpu], idt_table, 43.24 + IDT_ENTRIES*sizeof(idt_entry_t)); 43.25 + } 43.26 + 43.27 /* 43.28 * This grunge runs the startup process for 43.29 * the targeted processor.
44.1 --- a/xen/arch/x86/traps.c Wed Oct 22 11:38:22 2008 +0900 44.2 +++ b/xen/arch/x86/traps.c Wed Oct 22 11:46:55 2008 +0900 44.3 @@ -710,7 +710,7 @@ static void pv_cpuid(struct cpu_user_reg 44.4 if ( current->domain->domain_id != 0 ) 44.5 { 44.6 if ( !cpuid_hypervisor_leaves(a, &a, &b, &c, &d) ) 44.7 - domain_cpuid(current->domain, a, b, &a, &b, &c, &d); 44.8 + domain_cpuid(current->domain, a, c, &a, &b, &c, &d); 44.9 goto out; 44.10 } 44.11 44.12 @@ -1242,6 +1242,10 @@ asmlinkage void do_page_fault(struct cpu 44.13 regs->error_code, _p(addr)); 44.14 } 44.15 44.16 + if ( unlikely(current->domain->arch.suppress_spurious_page_faults 44.17 + && spurious_page_fault(addr, regs)) ) 44.18 + return; 44.19 + 44.20 propagate_page_fault(addr, regs->error_code); 44.21 } 44.22
45.1 --- a/xen/arch/x86/x86_32/xen.lds.S Wed Oct 22 11:38:22 2008 +0900 45.2 +++ b/xen/arch/x86/x86_32/xen.lds.S Wed Oct 22 11:46:55 2008 +0900 45.3 @@ -26,7 +26,6 @@ SECTIONS 45.4 *(.fixup) 45.5 *(.gnu.warning) 45.6 } :text =0x9090 45.7 - .text.lock : { *(.text.lock) } :text /* out-of-line lock text */ 45.8 45.9 _etext = .; /* End of text section */ 45.10
46.1 --- a/xen/arch/x86/x86_64/Makefile Wed Oct 22 11:38:22 2008 +0900 46.2 +++ b/xen/arch/x86/x86_64/Makefile Wed Oct 22 11:46:55 2008 +0900 46.3 @@ -13,6 +13,7 @@ obj-$(CONFIG_COMPAT) += domain.o 46.4 obj-$(CONFIG_COMPAT) += physdev.o 46.5 obj-$(CONFIG_COMPAT) += platform_hypercall.o 46.6 obj-$(CONFIG_COMPAT) += cpu_idle.o 46.7 +obj-$(CONFIG_COMPAT) += cpufreq.o 46.8 46.9 ifeq ($(CONFIG_COMPAT),y) 46.10 # extra dependencies 46.11 @@ -24,4 +25,5 @@ platform_hypercall.o: ../platform_hyperc 46.12 sysctl.o: ../sysctl.c 46.13 traps.o: compat/traps.c 46.14 cpu_idle.o: ../acpi/cpu_idle.c 46.15 +cpufreq.o: ../../../drivers/cpufreq/cpufreq.c 46.16 endif
47.1 --- a/xen/arch/x86/x86_64/cpu_idle.c Wed Oct 22 11:38:22 2008 +0900 47.2 +++ b/xen/arch/x86/x86_64/cpu_idle.c Wed Oct 22 11:46:55 2008 +0900 47.3 @@ -44,7 +44,7 @@ DEFINE_XEN_GUEST_HANDLE(compat_processor 47.4 xlat_page_current = xlat_page_start; \ 47.5 } while (0) 47.6 47.7 -static void *xlat_malloc(unsigned long *xlat_page_current, size_t size) 47.8 +void *xlat_malloc(unsigned long *xlat_page_current, size_t size) 47.9 { 47.10 void *ret; 47.11
48.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000 48.2 +++ b/xen/arch/x86/x86_64/cpufreq.c Wed Oct 22 11:46:55 2008 +0900 48.3 @@ -0,0 +1,91 @@ 48.4 +/****************************************************************************** 48.5 + * cpufreq.c -- adapt 32b compat guest to 64b hypervisor. 48.6 + * 48.7 + * Copyright (C) 2008, Liu Jinsong <jinsong.liu@intel.com> 48.8 + * 48.9 + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 48.10 + * 48.11 + * This program is free software; you can redistribute it and/or modify 48.12 + * it under the terms of the GNU General Public License as published by 48.13 + * the Free Software Foundation; either version 2 of the License, or (at 48.14 + * your option) any later version. 48.15 + * 48.16 + * This program is distributed in the hope that it will be useful, but 48.17 + * WITHOUT ANY WARRANTY; without even the implied warranty of 48.18 + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 48.19 + * General Public License for more details. 48.20 + * 48.21 + * You should have received a copy of the GNU General Public License along 48.22 + * with this program; if not, write to the Free Software Foundation, Inc., 48.23 + * 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA. 48.24 + * 48.25 + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 48.26 + */ 48.27 +#include <xen/config.h> 48.28 +#include <xen/types.h> 48.29 +#include <xen/xmalloc.h> 48.30 +#include <xen/guest_access.h> 48.31 +#include <compat/platform.h> 48.32 + 48.33 +DEFINE_XEN_GUEST_HANDLE(compat_processor_px_t); 48.34 + 48.35 +#define xlat_page_start ((unsigned long)COMPAT_ARG_XLAT_VIRT_BASE) 48.36 + 48.37 +#define xlat_malloc_init(xlat_page_current) do { \ 48.38 + xlat_page_current = xlat_page_start; \ 48.39 +} while (0) 48.40 + 48.41 +extern void *xlat_malloc(unsigned long *xlat_page_current, size_t size); 48.42 + 48.43 +#define xlat_malloc_array(_p, _t, _c) ((_t *) xlat_malloc(&_p, sizeof(_t) * _c)) 48.44 + 48.45 +extern int 48.46 +set_px_pminfo(uint32_t cpu, struct xen_processor_performance *perf); 48.47 + 48.48 +int 48.49 +compat_set_px_pminfo(uint32_t cpu, struct compat_processor_performance *perf) 48.50 +{ 48.51 + struct xen_processor_performance *xen_perf; 48.52 + unsigned long xlat_page_current; 48.53 + 48.54 + xlat_malloc_init(xlat_page_current); 48.55 + 48.56 + xen_perf = xlat_malloc_array(xlat_page_current, 48.57 + struct xen_processor_performance, 1); 48.58 + if ( unlikely(xen_perf == NULL) ) 48.59 + return -EFAULT; 48.60 + 48.61 +#define XLAT_processor_performance_HNDL_states(_d_, _s_) do { \ 48.62 + xen_processor_px_t *xen_states = NULL; \ 48.63 +\ 48.64 + if ( likely((_s_)->state_count > 0) ) \ 48.65 + { \ 48.66 + XEN_GUEST_HANDLE(compat_processor_px_t) states; \ 48.67 + compat_processor_px_t state; \ 48.68 + int i; \ 48.69 +\ 48.70 + xen_states = xlat_malloc_array(xlat_page_current, \ 48.71 + xen_processor_px_t, (_s_)->state_count); \ 48.72 + if ( unlikely(xen_states == NULL) ) \ 48.73 + return -EFAULT; \ 48.74 +\ 48.75 + if ( unlikely(!compat_handle_okay((_s_)->states, \ 48.76 + (_s_)->state_count)) ) \ 48.77 + return -EFAULT; \ 48.78 + guest_from_compat_handle(states, (_s_)->states); \ 48.79 +\ 48.80 + for ( i = 0; i < _s_->state_count; i++ ) \ 48.81 + { \ 48.82 + if ( unlikely(copy_from_guest_offset(&state, states, i, 1)) ) \ 48.83 + return -EFAULT; \ 48.84 + XLAT_processor_px(&xen_states[i], &state); \ 48.85 + } \ 48.86 + } \ 48.87 +\ 48.88 + set_xen_guest_handle((_d_)->states, xen_states); \ 48.89 +} while (0) 48.90 + XLAT_processor_performance(xen_perf, perf); 48.91 +#undef XLAT_processor_performance_HNDL_states 48.92 + 48.93 + return set_px_pminfo(cpu, xen_perf); 48.94 +}
49.1 --- a/xen/arch/x86/x86_64/mm.c Wed Oct 22 11:38:22 2008 +0900 49.2 +++ b/xen/arch/x86/x86_64/mm.c Wed Oct 22 11:46:55 2008 +0900 49.3 @@ -252,8 +252,6 @@ void __init subarch_init_memory(void) 49.4 BUILD_BUG_ON(offsetof(struct page_info, u.inuse._domain) != 49.5 (offsetof(struct page_info, count_info) + sizeof(u32))); 49.6 BUILD_BUG_ON((offsetof(struct page_info, count_info) & 7) != 0); 49.7 - BUILD_BUG_ON(sizeof(struct page_info) != 49.8 - (32 + BITS_TO_LONGS(NR_CPUS)*sizeof(long))); 49.9 49.10 /* M2P table is mappable read-only by privileged domains. */ 49.11 for ( v = RDWR_MPT_VIRT_START;
50.1 --- a/xen/arch/x86/x86_64/platform_hypercall.c Wed Oct 22 11:38:22 2008 +0900 50.2 +++ b/xen/arch/x86/x86_64/platform_hypercall.c Wed Oct 22 11:46:55 2008 +0900 50.3 @@ -11,14 +11,14 @@ DEFINE_XEN_GUEST_HANDLE(compat_platform_ 50.4 #define xen_platform_op_t compat_platform_op_t 50.5 #define do_platform_op(x) compat_platform_op(_##x) 50.6 50.7 -#define xenpf_copy_px_states compat_xenpf_copy_px_states 50.8 - 50.9 #define xen_processor_px compat_processor_px 50.10 #define xen_processor_px_t compat_processor_px_t 50.11 #define xen_processor_performance compat_processor_performance 50.12 #define xen_processor_performance_t compat_processor_performance_t 50.13 #define xenpf_set_processor_pminfo compat_pf_set_processor_pminfo 50.14 50.15 +#define set_px_pminfo compat_set_px_pminfo 50.16 + 50.17 #define xen_processor_power compat_processor_power 50.18 #define xen_processor_power_t compat_processor_power_t 50.19 #define set_cx_pminfo compat_set_cx_pminfo
51.1 --- a/xen/arch/x86/x86_64/xen.lds.S Wed Oct 22 11:38:22 2008 +0900 51.2 +++ b/xen/arch/x86/x86_64/xen.lds.S Wed Oct 22 11:46:55 2008 +0900 51.3 @@ -24,7 +24,6 @@ SECTIONS 51.4 *(.fixup) 51.5 *(.gnu.warning) 51.6 } :text = 0x9090 51.7 - .text.lock : { *(.text.lock) } :text /* out-of-line lock text */ 51.8 51.9 _etext = .; /* End of text section */ 51.10
52.1 --- a/xen/arch/x86/x86_emulate/x86_emulate.c Wed Oct 22 11:38:22 2008 +0900 52.2 +++ b/xen/arch/x86/x86_emulate/x86_emulate.c Wed Oct 22 11:46:55 2008 +0900 52.3 @@ -236,7 +236,8 @@ static uint8_t twobyte_table[256] = { 52.4 DstReg|SrcMem|ModRM, DstReg|SrcMem|ModRM, 52.5 ByteOp|DstReg|SrcMem|ModRM|Mov, DstReg|SrcMem16|ModRM|Mov, 52.6 /* 0xC0 - 0xC7 */ 52.7 - ByteOp|DstMem|SrcReg|ModRM, DstMem|SrcReg|ModRM, 0, 0, 52.8 + ByteOp|DstMem|SrcReg|ModRM, DstMem|SrcReg|ModRM, 52.9 + 0, DstMem|SrcReg|ModRM|Mov, 52.10 0, 0, 0, ImplicitOps|ModRM, 52.11 /* 0xC8 - 0xCF */ 52.12 ImplicitOps, ImplicitOps, ImplicitOps, ImplicitOps, 52.13 @@ -3910,6 +3911,12 @@ x86_emulate( 52.14 } 52.15 goto add; 52.16 52.17 + case 0xc3: /* movnti */ 52.18 + /* Ignore the non-temporal hint for now. */ 52.19 + generate_exception_if(dst.bytes <= 2, EXC_UD, -1); 52.20 + dst.val = src.val; 52.21 + break; 52.22 + 52.23 case 0xc7: /* Grp9 (cmpxchg8b/cmpxchg16b) */ { 52.24 unsigned long old[2], exp[2], new[2]; 52.25 unsigned int i;
53.1 --- a/xen/common/Makefile Wed Oct 22 11:38:22 2008 +0900 53.2 +++ b/xen/common/Makefile Wed Oct 22 11:46:55 2008 +0900 53.3 @@ -16,6 +16,7 @@ obj-y += sched_sedf.o 53.4 obj-y += schedule.o 53.5 obj-y += shutdown.o 53.6 obj-y += softirq.o 53.7 +obj-y += spinlock.o 53.8 obj-y += stop_machine.o 53.9 obj-y += string.o 53.10 obj-y += symbols.o 53.11 @@ -25,7 +26,7 @@ obj-y += timer.o 53.12 obj-y += trace.o 53.13 obj-y += version.o 53.14 obj-y += vsprintf.o 53.15 -obj-y += xmalloc.o 53.16 +obj-y += xmalloc_tlsf.o 53.17 obj-y += rcupdate.o 53.18 53.19 obj-$(perfc) += perfc.o
54.1 --- a/xen/common/kernel.c Wed Oct 22 11:38:22 2008 +0900 54.2 +++ b/xen/common/kernel.c Wed Oct 22 11:46:55 2008 +0900 54.3 @@ -75,8 +75,7 @@ void cmdline_parse(char *cmdline) 54.4 strlcpy(param->var, optval, param->len); 54.5 break; 54.6 case OPT_UINT: 54.7 - *(unsigned int *)param->var = 54.8 - simple_strtol(optval, (const char **)&optval, 0); 54.9 + *(unsigned int *)param->var = simple_strtol(optval, NULL, 0); 54.10 break; 54.11 case OPT_BOOL: 54.12 case OPT_INVBOOL:
55.1 --- a/xen/common/schedule.c Wed Oct 22 11:38:22 2008 +0900 55.2 +++ b/xen/common/schedule.c Wed Oct 22 11:46:55 2008 +0900 55.3 @@ -455,6 +455,10 @@ static long do_poll(struct sched_poll *s 55.4 goto out; 55.5 #endif 55.6 55.7 + rc = 0; 55.8 + if ( local_events_need_delivery() ) 55.9 + goto out; 55.10 + 55.11 for ( i = 0; i < sched_poll->nr_ports; i++ ) 55.12 { 55.13 rc = -EFAULT;
56.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000 56.2 +++ b/xen/common/spinlock.c Wed Oct 22 11:46:55 2008 +0900 56.3 @@ -0,0 +1,154 @@ 56.4 +#include <xen/config.h> 56.5 +#include <xen/smp.h> 56.6 +#include <xen/spinlock.h> 56.7 + 56.8 +void _spin_lock(spinlock_t *lock) 56.9 +{ 56.10 + _raw_spin_lock(&lock->raw); 56.11 +} 56.12 + 56.13 +void _spin_lock_irq(spinlock_t *lock) 56.14 +{ 56.15 + local_irq_disable(); 56.16 + _raw_spin_lock(&lock->raw); 56.17 +} 56.18 + 56.19 +unsigned long _spin_lock_irqsave(spinlock_t *lock) 56.20 +{ 56.21 + unsigned long flags; 56.22 + local_irq_save(flags); 56.23 + _raw_spin_lock(&lock->raw); 56.24 + return flags; 56.25 +} 56.26 + 56.27 +void _spin_unlock(spinlock_t *lock) 56.28 +{ 56.29 + _raw_spin_unlock(&lock->raw); 56.30 +} 56.31 + 56.32 +void _spin_unlock_irq(spinlock_t *lock) 56.33 +{ 56.34 + _raw_spin_unlock(&lock->raw); 56.35 + local_irq_enable(); 56.36 +} 56.37 + 56.38 +void _spin_unlock_irqrestore(spinlock_t *lock, unsigned long flags) 56.39 +{ 56.40 + _raw_spin_unlock(&lock->raw); 56.41 + local_irq_restore(flags); 56.42 +} 56.43 + 56.44 +int _spin_is_locked(spinlock_t *lock) 56.45 +{ 56.46 + return _raw_spin_is_locked(&lock->raw); 56.47 +} 56.48 + 56.49 +int _spin_trylock(spinlock_t *lock) 56.50 +{ 56.51 + return _raw_spin_trylock(&lock->raw); 56.52 +} 56.53 + 56.54 +void _spin_barrier(spinlock_t *lock) 56.55 +{ 56.56 + do { mb(); } while ( _raw_spin_is_locked(&lock->raw) ); 56.57 + mb(); 56.58 +} 56.59 + 56.60 +void _spin_lock_recursive(spinlock_t *lock) 56.61 +{ 56.62 + int cpu = smp_processor_id(); 56.63 + 56.64 + /* Don't allow overflow of recurse_cpu field. */ 56.65 + BUILD_BUG_ON(NR_CPUS > 0xfffu); 56.66 + 56.67 + if ( likely(lock->recurse_cpu != cpu) ) 56.68 + { 56.69 + spin_lock(lock); 56.70 + lock->recurse_cpu = cpu; 56.71 + } 56.72 + 56.73 + /* We support only fairly shallow recursion, else the counter overflows. */ 56.74 + ASSERT(lock->recurse_cnt < 0xfu); 56.75 + lock->recurse_cnt++; 56.76 +} 56.77 + 56.78 +void _spin_unlock_recursive(spinlock_t *lock) 56.79 +{ 56.80 + if ( likely(--lock->recurse_cnt == 0) ) 56.81 + { 56.82 + lock->recurse_cpu = 0xfffu; 56.83 + spin_unlock(lock); 56.84 + } 56.85 +} 56.86 + 56.87 +void _read_lock(rwlock_t *lock) 56.88 +{ 56.89 + _raw_read_lock(&lock->raw); 56.90 +} 56.91 + 56.92 +void _read_lock_irq(rwlock_t *lock) 56.93 +{ 56.94 + local_irq_disable(); 56.95 + _raw_read_lock(&lock->raw); 56.96 +} 56.97 + 56.98 +unsigned long _read_lock_irqsave(rwlock_t *lock) 56.99 +{ 56.100 + unsigned long flags; 56.101 + local_irq_save(flags); 56.102 + _raw_read_lock(&lock->raw); 56.103 + return flags; 56.104 +} 56.105 + 56.106 +void _read_unlock(rwlock_t *lock) 56.107 +{ 56.108 + _raw_read_unlock(&lock->raw); 56.109 +} 56.110 + 56.111 +void _read_unlock_irq(rwlock_t *lock) 56.112 +{ 56.113 + _raw_read_unlock(&lock->raw); 56.114 + local_irq_enable(); 56.115 +} 56.116 + 56.117 +void _read_unlock_irqrestore(rwlock_t *lock, unsigned long flags) 56.118 +{ 56.119 + _raw_read_unlock(&lock->raw); 56.120 + local_irq_restore(flags); 56.121 +} 56.122 + 56.123 +void _write_lock(rwlock_t *lock) 56.124 +{ 56.125 + _raw_write_lock(&lock->raw); 56.126 +} 56.127 + 56.128 +void _write_lock_irq(rwlock_t *lock) 56.129 +{ 56.130 + local_irq_disable(); 56.131 + _raw_write_lock(&lock->raw); 56.132 +} 56.133 + 56.134 +unsigned long _write_lock_irqsave(rwlock_t *lock) 56.135 +{ 56.136 + unsigned long flags; 56.137 + local_irq_save(flags); 56.138 + _raw_write_lock(&lock->raw); 56.139 + return flags; 56.140 +} 56.141 + 56.142 +void _write_unlock(rwlock_t *lock) 56.143 +{ 56.144 + _raw_write_unlock(&lock->raw); 56.145 +} 56.146 + 56.147 +void _write_unlock_irq(rwlock_t *lock) 56.148 +{ 56.149 + _raw_write_unlock(&lock->raw); 56.150 + local_irq_enable(); 56.151 +} 56.152 + 56.153 +void _write_unlock_irqrestore(rwlock_t *lock, unsigned long flags) 56.154 +{ 56.155 + _raw_write_unlock(&lock->raw); 56.156 + local_irq_restore(flags); 56.157 +}
57.1 --- a/xen/common/timer.c Wed Oct 22 11:38:22 2008 +0900 57.2 +++ b/xen/common/timer.c Wed Oct 22 11:46:55 2008 +0900 57.3 @@ -114,34 +114,19 @@ static int remove_from_heap(struct timer 57.4 57.5 57.6 /* Add new entry @t to @heap. Return TRUE if new top of heap. */ 57.7 -static int add_to_heap(struct timer ***pheap, struct timer *t) 57.8 +static int add_to_heap(struct timer **heap, struct timer *t) 57.9 { 57.10 - struct timer **heap = *pheap; 57.11 int sz = GET_HEAP_SIZE(heap); 57.12 57.13 - /* Copy the heap if it is full. */ 57.14 + /* Fail if the heap is full. */ 57.15 if ( unlikely(sz == GET_HEAP_LIMIT(heap)) ) 57.16 - { 57.17 - /* old_limit == (2^n)-1; new_limit == (2^(n+4))-1 */ 57.18 - int old_limit = GET_HEAP_LIMIT(heap); 57.19 - int new_limit = ((old_limit + 1) << 4) - 1; 57.20 - if ( in_irq() ) 57.21 - goto out; 57.22 - heap = xmalloc_array(struct timer *, new_limit + 1); 57.23 - if ( heap == NULL ) 57.24 - goto out; 57.25 - memcpy(heap, *pheap, (old_limit + 1) * sizeof(*heap)); 57.26 - SET_HEAP_LIMIT(heap, new_limit); 57.27 - if ( old_limit != 0 ) 57.28 - xfree(*pheap); 57.29 - *pheap = heap; 57.30 - } 57.31 + return 0; 57.32 57.33 SET_HEAP_SIZE(heap, ++sz); 57.34 heap[sz] = t; 57.35 t->heap_offset = sz; 57.36 up_heap(heap, sz); 57.37 - out: 57.38 + 57.39 return (t->heap_offset == 1); 57.40 } 57.41 57.42 @@ -210,7 +195,7 @@ static int add_entry(struct timers *time 57.43 /* Try to add to heap. t->heap_offset indicates whether we succeed. */ 57.44 t->heap_offset = 0; 57.45 t->status = TIMER_STATUS_in_heap; 57.46 - rc = add_to_heap(&timers->heap, t); 57.47 + rc = add_to_heap(timers->heap, t); 57.48 if ( t->heap_offset != 0 ) 57.49 return rc; 57.50 57.51 @@ -368,6 +353,27 @@ static void timer_softirq_action(void) 57.52 void *data; 57.53 57.54 ts = &this_cpu(timers); 57.55 + heap = ts->heap; 57.56 + 57.57 + /* If we are using overflow linked list, try to allocate a larger heap. */ 57.58 + if ( unlikely(ts->list != NULL) ) 57.59 + { 57.60 + /* old_limit == (2^n)-1; new_limit == (2^(n+4))-1 */ 57.61 + int old_limit = GET_HEAP_LIMIT(heap); 57.62 + int new_limit = ((old_limit + 1) << 4) - 1; 57.63 + struct timer **newheap = xmalloc_array(struct timer *, new_limit + 1); 57.64 + if ( newheap != NULL ) 57.65 + { 57.66 + spin_lock_irq(&ts->lock); 57.67 + memcpy(newheap, heap, (old_limit + 1) * sizeof(*heap)); 57.68 + SET_HEAP_LIMIT(newheap, new_limit); 57.69 + ts->heap = newheap; 57.70 + spin_unlock_irq(&ts->lock); 57.71 + if ( old_limit != 0 ) 57.72 + xfree(heap); 57.73 + heap = newheap; 57.74 + } 57.75 + } 57.76 57.77 spin_lock_irq(&ts->lock); 57.78 57.79 @@ -380,9 +386,8 @@ static void timer_softirq_action(void) 57.80 t->status = TIMER_STATUS_inactive; 57.81 add_entry(ts, t); 57.82 } 57.83 - 57.84 - heap = ts->heap; 57.85 - now = NOW(); 57.86 + 57.87 + now = NOW(); 57.88 57.89 while ( (GET_HEAP_SIZE(heap) != 0) && 57.90 ((t = heap[1])->expires < (now + TIMER_SLOP)) ) 57.91 @@ -397,9 +402,6 @@ static void timer_softirq_action(void) 57.92 spin_unlock_irq(&ts->lock); 57.93 (*fn)(data); 57.94 spin_lock_irq(&ts->lock); 57.95 - 57.96 - /* Heap may have grown while the lock was released. */ 57.97 - heap = ts->heap; 57.98 } 57.99 57.100 deadline = GET_HEAP_SIZE(heap) ? heap[1]->expires : 0;
58.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000 58.2 +++ b/xen/common/xmalloc_tlsf.c Wed Oct 22 11:46:55 2008 +0900 58.3 @@ -0,0 +1,599 @@ 58.4 +/* 58.5 + * Two Levels Segregate Fit memory allocator (TLSF) 58.6 + * Version 2.3.2 58.7 + * 58.8 + * Written by Miguel Masmano Tello <mimastel@doctor.upv.es> 58.9 + * 58.10 + * Thanks to Ismael Ripoll for his suggestions and reviews 58.11 + * 58.12 + * Copyright (C) 2007, 2006, 2005, 2004 58.13 + * 58.14 + * This code is released using a dual license strategy: GPL/LGPL 58.15 + * You can choose the licence that better fits your requirements. 58.16 + * 58.17 + * Released under the terms of the GNU General Public License Version 2.0 58.18 + * Released under the terms of the GNU Lesser General Public License 58.19 + * Version 2.1 58.20 + * 58.21 + * This is kernel port of TLSF allocator. 58.22 + * Original code can be found at: http://rtportal.upv.es/rtmalloc/ 58.23 + * Adapted for Linux by Nitin Gupta (nitingupta910@gmail.com) 58.24 + * (http://code.google.com/p/compcache/source/browse/trunk/sub-projects 58.25 + * /allocators/tlsf-kmod r229 dated Aug 27, 2008 58.26 + * Adapted for Xen by Dan Magenheimer (dan.magenheimer@oracle.com) 58.27 + */ 58.28 + 58.29 +#include <xen/config.h> 58.30 +#include <xen/irq.h> 58.31 +#include <xen/mm.h> 58.32 +#include <asm/time.h> 58.33 + 58.34 +#define MAX_POOL_NAME_LEN 16 58.35 + 58.36 +/* Some IMPORTANT TLSF parameters */ 58.37 +#define MEM_ALIGN (sizeof(void *) * 2) 58.38 +#define MEM_ALIGN_MASK (~(MEM_ALIGN - 1)) 58.39 + 58.40 +#define MAX_FLI (30) 58.41 +#define MAX_LOG2_SLI (5) 58.42 +#define MAX_SLI (1 << MAX_LOG2_SLI) 58.43 + 58.44 +#define FLI_OFFSET (6) 58.45 +/* tlsf structure just will manage blocks bigger than 128 bytes */ 58.46 +#define SMALL_BLOCK (128) 58.47 +#define REAL_FLI (MAX_FLI - FLI_OFFSET) 58.48 +#define MIN_BLOCK_SIZE (sizeof(struct free_ptr)) 58.49 +#define BHDR_OVERHEAD (sizeof(struct bhdr) - MIN_BLOCK_SIZE) 58.50 + 58.51 +#define PTR_MASK (sizeof(void *) - 1) 58.52 +#define BLOCK_SIZE_MASK (0xFFFFFFFF - PTR_MASK) 58.53 + 58.54 +#define GET_NEXT_BLOCK(addr, r) ((struct bhdr *) \ 58.55 + ((char *)(addr) + (r))) 58.56 +#define ROUNDUP_SIZE(r) (((r) + MEM_ALIGN - 1) & MEM_ALIGN_MASK) 58.57 +#define ROUNDDOWN_SIZE(r) ((r) & MEM_ALIGN_MASK) 58.58 +#define ROUNDUP_PAGE(r) (((r) + PAGE_SIZE - 1) & PAGE_MASK) 58.59 + 58.60 +#define BLOCK_STATE (0x1) 58.61 +#define PREV_STATE (0x2) 58.62 + 58.63 +/* bit 0 of the block size */ 58.64 +#define FREE_BLOCK (0x1) 58.65 +#define USED_BLOCK (0x0) 58.66 + 58.67 +/* bit 1 of the block size */ 58.68 +#define PREV_FREE (0x2) 58.69 +#define PREV_USED (0x0) 58.70 + 58.71 +static spinlock_t pool_list_lock; 58.72 +static struct list_head pool_list_head; 58.73 + 58.74 +struct free_ptr { 58.75 + struct bhdr *prev; 58.76 + struct bhdr *next; 58.77 +}; 58.78 + 58.79 +struct bhdr { 58.80 + /* All blocks in a region are linked in order of physical address */ 58.81 + struct bhdr *prev_hdr; 58.82 + /* 58.83 + * The size is stored in bytes 58.84 + * bit 0: block is free, if set 58.85 + * bit 1: previous block is free, if set 58.86 + */ 58.87 + u32 size; 58.88 + /* Free blocks in individual freelists are linked */ 58.89 + union { 58.90 + struct free_ptr free_ptr; 58.91 + u8 buffer[sizeof(struct free_ptr)]; 58.92 + } ptr; 58.93 +}; 58.94 + 58.95 +struct xmem_pool { 58.96 + /* First level bitmap (REAL_FLI bits) */ 58.97 + u32 fl_bitmap; 58.98 + 58.99 + /* Second level bitmap */ 58.100 + u32 sl_bitmap[REAL_FLI]; 58.101 + 58.102 + /* Free lists */ 58.103 + struct bhdr *matrix[REAL_FLI][MAX_SLI]; 58.104 + 58.105 + spinlock_t lock; 58.106 + 58.107 + unsigned long init_size; 58.108 + unsigned long max_size; 58.109 + unsigned long grow_size; 58.110 + 58.111 + /* Basic stats */ 58.112 + unsigned long used_size; 58.113 + unsigned long num_regions; 58.114 + 58.115 + /* User provided functions for expanding/shrinking pool */ 58.116 + xmem_pool_get_memory *get_mem; 58.117 + xmem_pool_put_memory *put_mem; 58.118 + 58.119 + struct list_head list; 58.120 + 58.121 + void *init_region; 58.122 + char name[MAX_POOL_NAME_LEN]; 58.123 +}; 58.124 + 58.125 +/* 58.126 + * Helping functions 58.127 + */ 58.128 + 58.129 +/** 58.130 + * Returns indexes (fl, sl) of the list used to serve request of size r 58.131 + */ 58.132 +static inline void MAPPING_SEARCH(unsigned long *r, int *fl, int *sl) 58.133 +{ 58.134 + int t; 58.135 + 58.136 + if ( *r < SMALL_BLOCK ) 58.137 + { 58.138 + *fl = 0; 58.139 + *sl = *r / (SMALL_BLOCK / MAX_SLI); 58.140 + } 58.141 + else 58.142 + { 58.143 + t = (1 << (fls(*r) - 1 - MAX_LOG2_SLI)) - 1; 58.144 + *r = *r + t; 58.145 + *fl = fls(*r) - 1; 58.146 + *sl = (*r >> (*fl - MAX_LOG2_SLI)) - MAX_SLI; 58.147 + *fl -= FLI_OFFSET; 58.148 + /*if ((*fl -= FLI_OFFSET) < 0) // FL will be always >0! 58.149 + *fl = *sl = 0; 58.150 + */ 58.151 + *r &= ~t; 58.152 + } 58.153 +} 58.154 + 58.155 +/** 58.156 + * Returns indexes (fl, sl) which is used as starting point to search 58.157 + * for a block of size r. It also rounds up requested size(r) to the 58.158 + * next list. 58.159 + */ 58.160 +static inline void MAPPING_INSERT(unsigned long r, int *fl, int *sl) 58.161 +{ 58.162 + if ( r < SMALL_BLOCK ) 58.163 + { 58.164 + *fl = 0; 58.165 + *sl = r / (SMALL_BLOCK / MAX_SLI); 58.166 + } 58.167 + else 58.168 + { 58.169 + *fl = fls(r) - 1; 58.170 + *sl = (r >> (*fl - MAX_LOG2_SLI)) - MAX_SLI; 58.171 + *fl -= FLI_OFFSET; 58.172 + } 58.173 +} 58.174 + 58.175 +/** 58.176 + * Returns first block from a list that hold blocks larger than or 58.177 + * equal to the one pointed by the indexes (fl, sl) 58.178 + */ 58.179 +static inline struct bhdr *FIND_SUITABLE_BLOCK(struct xmem_pool *p, int *fl, 58.180 + int *sl) 58.181 +{ 58.182 + u32 tmp = p->sl_bitmap[*fl] & (~0 << *sl); 58.183 + struct bhdr *b = NULL; 58.184 + 58.185 + if ( tmp ) 58.186 + { 58.187 + *sl = ffs(tmp) - 1; 58.188 + b = p->matrix[*fl][*sl]; 58.189 + } 58.190 + else 58.191 + { 58.192 + *fl = ffs(p->fl_bitmap & (~0 << (*fl + 1))) - 1; 58.193 + if ( likely(*fl > 0) ) 58.194 + { 58.195 + *sl = ffs(p->sl_bitmap[*fl]) - 1; 58.196 + b = p->matrix[*fl][*sl]; 58.197 + } 58.198 + } 58.199 + 58.200 + return b; 58.201 +} 58.202 + 58.203 +/** 58.204 + * Remove first free block(b) from free list with indexes (fl, sl). 58.205 + */ 58.206 +static inline void EXTRACT_BLOCK_HDR(struct bhdr *b, struct xmem_pool *p, int fl, 58.207 + int sl) 58.208 +{ 58.209 + p->matrix[fl][sl] = b->ptr.free_ptr.next; 58.210 + if ( p->matrix[fl][sl] ) 58.211 + { 58.212 + p->matrix[fl][sl]->ptr.free_ptr.prev = NULL; 58.213 + } 58.214 + else 58.215 + { 58.216 + clear_bit(sl, &p->sl_bitmap[fl]); 58.217 + if ( !p->sl_bitmap[fl] ) 58.218 + clear_bit(fl, &p->fl_bitmap); 58.219 + } 58.220 + b->ptr.free_ptr = (struct free_ptr) {NULL, NULL}; 58.221 +} 58.222 + 58.223 +/** 58.224 + * Removes block(b) from free list with indexes (fl, sl) 58.225 + */ 58.226 +static inline void EXTRACT_BLOCK(struct bhdr *b, struct xmem_pool *p, int fl, 58.227 + int sl) 58.228 +{ 58.229 + if ( b->ptr.free_ptr.next ) 58.230 + b->ptr.free_ptr.next->ptr.free_ptr.prev = 58.231 + b->ptr.free_ptr.prev; 58.232 + if ( b->ptr.free_ptr.prev ) 58.233 + b->ptr.free_ptr.prev->ptr.free_ptr.next = 58.234 + b->ptr.free_ptr.next; 58.235 + if ( p->matrix[fl][sl] == b ) 58.236 + { 58.237 + p->matrix[fl][sl] = b->ptr.free_ptr.next; 58.238 + if ( !p->matrix[fl][sl] ) 58.239 + { 58.240 + clear_bit(sl, &p->sl_bitmap[fl]); 58.241 + if ( !p->sl_bitmap[fl] ) 58.242 + clear_bit (fl, &p->fl_bitmap); 58.243 + } 58.244 + } 58.245 + b->ptr.free_ptr = (struct free_ptr) {NULL, NULL}; 58.246 +} 58.247 + 58.248 +/** 58.249 + * Insert block(b) in free list with indexes (fl, sl) 58.250 + */ 58.251 +static inline void INSERT_BLOCK(struct bhdr *b, struct xmem_pool *p, int fl, int sl) 58.252 +{ 58.253 + b->ptr.free_ptr = (struct free_ptr) {NULL, p->matrix[fl][sl]}; 58.254 + if ( p->matrix[fl][sl] ) 58.255 + p->matrix[fl][sl]->ptr.free_ptr.prev = b; 58.256 + p->matrix[fl][sl] = b; 58.257 + set_bit(sl, &p->sl_bitmap[fl]); 58.258 + set_bit(fl, &p->fl_bitmap); 58.259 +} 58.260 + 58.261 +/** 58.262 + * Region is a virtually contiguous memory region and Pool is 58.263 + * collection of such regions 58.264 + */ 58.265 +static inline void ADD_REGION(void *region, unsigned long region_size, 58.266 + struct xmem_pool *pool) 58.267 +{ 58.268 + int fl, sl; 58.269 + struct bhdr *b, *lb; 58.270 + 58.271 + b = (struct bhdr *)(region); 58.272 + b->prev_hdr = NULL; 58.273 + b->size = ROUNDDOWN_SIZE(region_size - 2 * BHDR_OVERHEAD) 58.274 + | FREE_BLOCK | PREV_USED; 58.275 + MAPPING_INSERT(b->size & BLOCK_SIZE_MASK, &fl, &sl); 58.276 + INSERT_BLOCK(b, pool, fl, sl); 58.277 + /* The sentinel block: allows us to know when we're in the last block */ 58.278 + lb = GET_NEXT_BLOCK(b->ptr.buffer, b->size & BLOCK_SIZE_MASK); 58.279 + lb->prev_hdr = b; 58.280 + lb->size = 0 | USED_BLOCK | PREV_FREE; 58.281 + pool->used_size += BHDR_OVERHEAD; /* only sentinel block is "used" */ 58.282 + pool->num_regions++; 58.283 +} 58.284 + 58.285 +/* 58.286 + * TLSF pool-based allocator start. 58.287 + */ 58.288 + 58.289 +struct xmem_pool *xmem_pool_create( 58.290 + const char *name, 58.291 + xmem_pool_get_memory get_mem, 58.292 + xmem_pool_put_memory put_mem, 58.293 + unsigned long init_size, 58.294 + unsigned long max_size, 58.295 + unsigned long grow_size) 58.296 +{ 58.297 + struct xmem_pool *pool; 58.298 + void *region; 58.299 + int pool_bytes, pool_order; 58.300 + 58.301 + BUG_ON(max_size && (max_size < init_size)); 58.302 + 58.303 + pool_bytes = ROUNDUP_SIZE(sizeof(*pool)); 58.304 + pool_order = get_order_from_bytes(pool_bytes); 58.305 + 58.306 + pool = (void *)alloc_xenheap_pages(pool_order); 58.307 + if ( pool == NULL ) 58.308 + return NULL; 58.309 + memset(pool, 0, pool_bytes); 58.310 + 58.311 + /* Round to next page boundary */ 58.312 + init_size = ROUNDUP_PAGE(init_size); 58.313 + max_size = ROUNDUP_PAGE(max_size); 58.314 + grow_size = ROUNDUP_PAGE(grow_size); 58.315 + 58.316 + /* pool global overhead not included in used size */ 58.317 + pool->used_size = 0; 58.318 + 58.319 + pool->init_size = init_size; 58.320 + pool->max_size = max_size; 58.321 + pool->grow_size = grow_size; 58.322 + pool->get_mem = get_mem; 58.323 + pool->put_mem = put_mem; 58.324 + strlcpy(pool->name, name, sizeof(pool->name)); 58.325 + region = get_mem(init_size); 58.326 + if ( region == NULL ) 58.327 + goto out_region; 58.328 + ADD_REGION(region, init_size, pool); 58.329 + pool->init_region = region; 58.330 + 58.331 + spin_lock_init(&pool->lock); 58.332 + 58.333 + spin_lock(&pool_list_lock); 58.334 + list_add_tail(&pool->list, &pool_list_head); 58.335 + spin_unlock(&pool_list_lock); 58.336 + 58.337 + return pool; 58.338 + 58.339 + out_region: 58.340 + free_xenheap_pages(pool, pool_order); 58.341 + return NULL; 58.342 +} 58.343 + 58.344 +unsigned long xmem_pool_get_used_size(struct xmem_pool *pool) 58.345 +{ 58.346 + return pool->used_size; 58.347 +} 58.348 + 58.349 +unsigned long xmem_pool_get_total_size(struct xmem_pool *pool) 58.350 +{ 58.351 + unsigned long total; 58.352 + total = ROUNDUP_SIZE(sizeof(*pool)) 58.353 + + pool->init_size 58.354 + + (pool->num_regions - 1) * pool->grow_size; 58.355 + return total; 58.356 +} 58.357 + 58.358 +void xmem_pool_destroy(struct xmem_pool *pool) 58.359 +{ 58.360 + if ( pool == NULL ) 58.361 + return; 58.362 + 58.363 + /* User is destroying without ever allocating from this pool */ 58.364 + if ( xmem_pool_get_used_size(pool) == BHDR_OVERHEAD ) 58.365 + { 58.366 + pool->put_mem(pool->init_region); 58.367 + pool->used_size -= BHDR_OVERHEAD; 58.368 + } 58.369 + 58.370 + /* Check for memory leaks in this pool */ 58.371 + if ( xmem_pool_get_used_size(pool) ) 58.372 + printk("memory leak in pool: %s (%p). " 58.373 + "%lu bytes still in use.\n", 58.374 + pool->name, pool, xmem_pool_get_used_size(pool)); 58.375 + 58.376 + spin_lock(&pool_list_lock); 58.377 + list_del_init(&pool->list); 58.378 + spin_unlock(&pool_list_lock); 58.379 + pool->put_mem(pool); 58.380 +} 58.381 + 58.382 +void *xmem_pool_alloc(unsigned long size, struct xmem_pool *pool) 58.383 +{ 58.384 + struct bhdr *b, *b2, *next_b, *region; 58.385 + int fl, sl; 58.386 + unsigned long tmp_size; 58.387 + 58.388 + size = (size < MIN_BLOCK_SIZE) ? MIN_BLOCK_SIZE : ROUNDUP_SIZE(size); 58.389 + /* Rounding up the requested size and calculating fl and sl */ 58.390 + 58.391 + spin_lock(&pool->lock); 58.392 + retry_find: 58.393 + MAPPING_SEARCH(&size, &fl, &sl); 58.394 + 58.395 + /* Searching a free block */ 58.396 + if ( !(b = FIND_SUITABLE_BLOCK(pool, &fl, &sl)) ) 58.397 + { 58.398 + /* Not found */ 58.399 + if ( size > (pool->grow_size - 2 * BHDR_OVERHEAD) ) 58.400 + goto out_locked; 58.401 + if ( pool->max_size && (pool->init_size + 58.402 + pool->num_regions * pool->grow_size 58.403 + > pool->max_size) ) 58.404 + goto out_locked; 58.405 + spin_unlock(&pool->lock); 58.406 + if ( (region = pool->get_mem(pool->grow_size)) == NULL ) 58.407 + goto out; 58.408 + spin_lock(&pool->lock); 58.409 + ADD_REGION(region, pool->grow_size, pool); 58.410 + goto retry_find; 58.411 + } 58.412 + EXTRACT_BLOCK_HDR(b, pool, fl, sl); 58.413 + 58.414 + /*-- found: */ 58.415 + next_b = GET_NEXT_BLOCK(b->ptr.buffer, b->size & BLOCK_SIZE_MASK); 58.416 + /* Should the block be split? */ 58.417 + tmp_size = (b->size & BLOCK_SIZE_MASK) - size; 58.418 + if ( tmp_size >= sizeof(struct bhdr) ) 58.419 + { 58.420 + tmp_size -= BHDR_OVERHEAD; 58.421 + b2 = GET_NEXT_BLOCK(b->ptr.buffer, size); 58.422 + 58.423 + b2->size = tmp_size | FREE_BLOCK | PREV_USED; 58.424 + b2->prev_hdr = b; 58.425 + 58.426 + next_b->prev_hdr = b2; 58.427 + 58.428 + MAPPING_INSERT(tmp_size, &fl, &sl); 58.429 + INSERT_BLOCK(b2, pool, fl, sl); 58.430 + 58.431 + b->size = size | (b->size & PREV_STATE); 58.432 + } 58.433 + else 58.434 + { 58.435 + next_b->size &= (~PREV_FREE); 58.436 + b->size &= (~FREE_BLOCK); /* Now it's used */ 58.437 + } 58.438 + 58.439 + pool->used_size += (b->size & BLOCK_SIZE_MASK) + BHDR_OVERHEAD; 58.440 + 58.441 + spin_unlock(&pool->lock); 58.442 + return (void *)b->ptr.buffer; 58.443 + 58.444 + /* Failed alloc */ 58.445 + out_locked: 58.446 + spin_unlock(&pool->lock); 58.447 + 58.448 + out: 58.449 + return NULL; 58.450 +} 58.451 + 58.452 +void xmem_pool_free(void *ptr, struct xmem_pool *pool) 58.453 +{ 58.454 + struct bhdr *b, *tmp_b; 58.455 + int fl = 0, sl = 0; 58.456 + 58.457 + if ( unlikely(ptr == NULL) ) 58.458 + return; 58.459 + 58.460 + b = (struct bhdr *)((char *) ptr - BHDR_OVERHEAD); 58.461 + 58.462 + spin_lock(&pool->lock); 58.463 + b->size |= FREE_BLOCK; 58.464 + pool->used_size -= (b->size & BLOCK_SIZE_MASK) + BHDR_OVERHEAD; 58.465 + b->ptr.free_ptr = (struct free_ptr) { NULL, NULL}; 58.466 + tmp_b = GET_NEXT_BLOCK(b->ptr.buffer, b->size & BLOCK_SIZE_MASK); 58.467 + if ( tmp_b->size & FREE_BLOCK ) 58.468 + { 58.469 + MAPPING_INSERT(tmp_b->size & BLOCK_SIZE_MASK, &fl, &sl); 58.470 + EXTRACT_BLOCK(tmp_b, pool, fl, sl); 58.471 + b->size += (tmp_b->size & BLOCK_SIZE_MASK) + BHDR_OVERHEAD; 58.472 + } 58.473 + if ( b->size & PREV_FREE ) 58.474 + { 58.475 + tmp_b = b->prev_hdr; 58.476 + MAPPING_INSERT(tmp_b->size & BLOCK_SIZE_MASK, &fl, &sl); 58.477 + EXTRACT_BLOCK(tmp_b, pool, fl, sl); 58.478 + tmp_b->size += (b->size & BLOCK_SIZE_MASK) + BHDR_OVERHEAD; 58.479 + b = tmp_b; 58.480 + } 58.481 + tmp_b = GET_NEXT_BLOCK(b->ptr.buffer, b->size & BLOCK_SIZE_MASK); 58.482 + tmp_b->prev_hdr = b; 58.483 + 58.484 + MAPPING_INSERT(b->size & BLOCK_SIZE_MASK, &fl, &sl); 58.485 + 58.486 + if ( (b->prev_hdr == NULL) && ((tmp_b->size & BLOCK_SIZE_MASK) == 0) ) 58.487 + { 58.488 + pool->put_mem(b); 58.489 + pool->num_regions--; 58.490 + pool->used_size -= BHDR_OVERHEAD; /* sentinel block header */ 58.491 + goto out; 58.492 + } 58.493 + 58.494 + INSERT_BLOCK(b, pool, fl, sl); 58.495 + 58.496 + tmp_b->size |= PREV_FREE; 58.497 + tmp_b->prev_hdr = b; 58.498 + out: 58.499 + spin_unlock(&pool->lock); 58.500 +} 58.501 + 58.502 +/* 58.503 + * Glue for xmalloc(). 58.504 + */ 58.505 + 58.506 +static struct xmem_pool *xenpool; 58.507 + 58.508 +static void *xmalloc_pool_get(unsigned long size) 58.509 +{ 58.510 + ASSERT(size == PAGE_SIZE); 58.511 + return alloc_xenheap_pages(0); 58.512 +} 58.513 + 58.514 +static void xmalloc_pool_put(void *p) 58.515 +{ 58.516 + free_xenheap_pages(p,0); 58.517 +} 58.518 + 58.519 +static void *xmalloc_whole_pages(unsigned long size) 58.520 +{ 58.521 + struct bhdr *b; 58.522 + unsigned int pageorder = get_order_from_bytes(size + BHDR_OVERHEAD); 58.523 + 58.524 + b = alloc_xenheap_pages(pageorder); 58.525 + if ( b == NULL ) 58.526 + return NULL; 58.527 + 58.528 + b->size = (1 << (pageorder + PAGE_SHIFT)); 58.529 + return (void *)b->ptr.buffer; 58.530 +} 58.531 + 58.532 +static void tlsf_init(void) 58.533 +{ 58.534 + INIT_LIST_HEAD(&pool_list_head); 58.535 + spin_lock_init(&pool_list_lock); 58.536 + xenpool = xmem_pool_create( 58.537 + "xmalloc", xmalloc_pool_get, xmalloc_pool_put, 58.538 + PAGE_SIZE, 0, PAGE_SIZE); 58.539 + BUG_ON(!xenpool); 58.540 +} 58.541 + 58.542 +/* 58.543 + * xmalloc() 58.544 + */ 58.545 + 58.546 +void *_xmalloc(unsigned long size, unsigned long align) 58.547 +{ 58.548 + void *p; 58.549 + u32 pad; 58.550 + 58.551 + ASSERT(!in_irq()); 58.552 + 58.553 + ASSERT((align & (align - 1)) == 0); 58.554 + if ( align < MEM_ALIGN ) 58.555 + align = MEM_ALIGN; 58.556 + size += align - MEM_ALIGN; 58.557 + 58.558 + if ( !xenpool ) 58.559 + tlsf_init(); 58.560 + 58.561 + if ( size >= (PAGE_SIZE - (2*BHDR_OVERHEAD)) ) 58.562 + p = xmalloc_whole_pages(size); 58.563 + else 58.564 + p = xmem_pool_alloc(size, xenpool); 58.565 + 58.566 + /* Add alignment padding. */ 58.567 + if ( (pad = -(long)p & (align - 1)) != 0 ) 58.568 + { 58.569 + char *q = (char *)p + pad; 58.570 + struct bhdr *b = (struct bhdr *)(q - BHDR_OVERHEAD); 58.571 + ASSERT(q > (char *)p); 58.572 + b->size = pad | 1; 58.573 + p = q; 58.574 + } 58.575 + 58.576 + ASSERT(((unsigned long)p & (align - 1)) == 0); 58.577 + return p; 58.578 +} 58.579 + 58.580 +void xfree(void *p) 58.581 +{ 58.582 + struct bhdr *b; 58.583 + 58.584 + ASSERT(!in_irq()); 58.585 + 58.586 + if ( p == NULL ) 58.587 + return; 58.588 + 58.589 + /* Strip alignment padding. */ 58.590 + b = (struct bhdr *)((char *) p - BHDR_OVERHEAD); 58.591 + if ( b->size & 1 ) 58.592 + { 58.593 + p = (char *)p - (b->size & ~1u); 58.594 + b = (struct bhdr *)((char *)p - BHDR_OVERHEAD); 58.595 + ASSERT(!(b->size & 1)); 58.596 + } 58.597 + 58.598 + if ( b->size >= (PAGE_SIZE - (2*BHDR_OVERHEAD)) ) 58.599 + free_xenheap_pages((void *)b, get_order_from_bytes(b->size)); 58.600 + else 58.601 + xmem_pool_free(p, xenpool); 58.602 +}
59.1 --- a/xen/drivers/Makefile Wed Oct 22 11:38:22 2008 +0900 59.2 +++ b/xen/drivers/Makefile Wed Oct 22 11:46:55 2008 +0900 59.3 @@ -1,6 +1,6 @@ 59.4 subdir-y += char 59.5 subdir-y += cpufreq 59.6 subdir-y += pci 59.7 -subdir-$(x86) += passthrough 59.8 +subdir-y += passthrough 59.9 subdir-$(HAS_ACPI) += acpi 59.10 subdir-$(HAS_VGA) += video
60.1 --- a/xen/drivers/acpi/pmstat.c Wed Oct 22 11:38:22 2008 +0900 60.2 +++ b/xen/drivers/acpi/pmstat.c Wed Oct 22 11:46:55 2008 +0900 60.3 @@ -52,7 +52,7 @@ int do_get_pm_info(struct xen_sysctl_get 60.4 int ret = 0; 60.5 const struct processor_pminfo *pmpt; 60.6 60.7 - if ( (op->cpuid >= NR_CPUS) || !cpu_online(op->cpuid) ) 60.8 + if ( !op || (op->cpuid >= NR_CPUS) || !cpu_online(op->cpuid) ) 60.9 return -EINVAL; 60.10 pmpt = processor_pminfo[op->cpuid]; 60.11 60.12 @@ -87,7 +87,7 @@ int do_get_pm_info(struct xen_sysctl_get 60.13 uint64_t tmp_idle_ns; 60.14 struct pm_px *pxpt = cpufreq_statistic_data[op->cpuid]; 60.15 60.16 - if ( !pxpt ) 60.17 + if ( !pxpt || !pxpt->u.pt || !pxpt->u.trans_pt ) 60.18 return -ENODATA; 60.19 60.20 total_idle_ns = get_cpu_idle_time(op->cpuid);
61.1 --- a/xen/drivers/char/ns16550.c Wed Oct 22 11:38:22 2008 +0900 61.2 +++ b/xen/drivers/char/ns16550.c Wed Oct 22 11:46:55 2008 +0900 61.3 @@ -18,17 +18,19 @@ 61.4 #include <asm/io.h> 61.5 61.6 /* 61.7 - * Configure serial port with a string <baud>,DPS,<io-base>,<irq>. 61.8 + * Configure serial port with a string: 61.9 + * <baud>[/<clock_hz>][,DPS[,<io-base>[,<irq>]]]. 61.10 * The tail of the string can be omitted if platform defaults are sufficient. 61.11 * If the baud rate is pre-configured, perhaps by a bootloader, then 'auto' 61.12 - * can be specified in place of a numeric baud rate. 61.13 + * can be specified in place of a numeric baud rate. Polled mode is specified 61.14 + * by requesting irq 0. 61.15 */ 61.16 static char opt_com1[30] = "", opt_com2[30] = ""; 61.17 string_param("com1", opt_com1); 61.18 string_param("com2", opt_com2); 61.19 61.20 static struct ns16550 { 61.21 - int baud, data_bits, parity, stop_bits, irq; 61.22 + int baud, clock_hz, data_bits, parity, stop_bits, irq; 61.23 unsigned long io_base; /* I/O port or memory-mapped I/O address. */ 61.24 char *remapped_io_base; /* Remapped virtual address of mmap I/O. */ 61.25 /* UART with IRQ line: interrupt-driven I/O. */ 61.26 @@ -192,7 +194,7 @@ static void __devinit ns16550_init_preir 61.27 if ( uart->baud != BAUD_AUTO ) 61.28 { 61.29 /* Baud rate specified: program it into the divisor latch. */ 61.30 - divisor = UART_CLOCK_HZ / (uart->baud * 16); 61.31 + divisor = uart->clock_hz / (uart->baud << 4); 61.32 ns_write_reg(uart, DLL, (char)divisor); 61.33 ns_write_reg(uart, DLM, (char)(divisor >> 8)); 61.34 } 61.35 @@ -201,7 +203,7 @@ static void __devinit ns16550_init_preir 61.36 /* Baud rate already set: read it out from the divisor latch. */ 61.37 divisor = ns_read_reg(uart, DLL); 61.38 divisor |= ns_read_reg(uart, DLM) << 8; 61.39 - uart->baud = UART_CLOCK_HZ / (divisor * 16); 61.40 + uart->baud = uart->clock_hz / (divisor << 4); 61.41 } 61.42 ns_write_reg(uart, LCR, lcr); 61.43 61.44 @@ -355,6 +357,12 @@ static void __init ns16550_parse_port_co 61.45 else if ( (baud = simple_strtoul(conf, &conf, 10)) != 0 ) 61.46 uart->baud = baud; 61.47 61.48 + if ( *conf == '/') 61.49 + { 61.50 + conf++; 61.51 + uart->clock_hz = simple_strtoul(conf, &conf, 0) << 4; 61.52 + } 61.53 + 61.54 if ( *conf != ',' ) 61.55 goto config_parsed; 61.56 conf++; 61.57 @@ -408,6 +416,7 @@ void __init ns16550_init(int index, stru 61.58 uart->baud = (defaults->baud ? : 61.59 console_has((index == 0) ? "com1" : "com2") 61.60 ? BAUD_AUTO : 0); 61.61 + uart->clock_hz = UART_CLOCK_HZ; 61.62 uart->data_bits = defaults->data_bits; 61.63 uart->parity = parse_parity_char(defaults->parity); 61.64 uart->stop_bits = defaults->stop_bits;
62.1 --- a/xen/drivers/cpufreq/cpufreq.c Wed Oct 22 11:38:22 2008 +0900 62.2 +++ b/xen/drivers/cpufreq/cpufreq.c Wed Oct 22 11:46:55 2008 +0900 62.3 @@ -34,6 +34,7 @@ 62.4 #include <xen/sched.h> 62.5 #include <xen/timer.h> 62.6 #include <xen/xmalloc.h> 62.7 +#include <xen/guest_access.h> 62.8 #include <xen/domain.h> 62.9 #include <asm/bug.h> 62.10 #include <asm/io.h> 62.11 @@ -185,10 +186,18 @@ int cpufreq_del_cpu(unsigned int cpu) 62.12 return 0; 62.13 } 62.14 62.15 +static void print_PCT(struct xen_pct_register *ptr) 62.16 +{ 62.17 + printk(KERN_INFO "\t_PCT: descriptor=%d, length=%d, space_id=%d, " 62.18 + "bit_width=%d, bit_offset=%d, reserved=%d, address=%"PRId64"\n", 62.19 + ptr->descriptor, ptr->length, ptr->space_id, ptr->bit_width, 62.20 + ptr->bit_offset, ptr->reserved, ptr->address); 62.21 +} 62.22 + 62.23 static void print_PSS(struct xen_processor_px *ptr, int count) 62.24 { 62.25 int i; 62.26 - printk(KERN_INFO "\t_PSS:\n"); 62.27 + printk(KERN_INFO "\t_PSS: state_count=%d\n", count); 62.28 for (i=0; i<count; i++){ 62.29 printk(KERN_INFO "\tState%d: %"PRId64"MHz %"PRId64"mW %"PRId64"us " 62.30 "%"PRId64"us 0x%"PRIx64" 0x%"PRIx64"\n", 62.31 @@ -211,20 +220,19 @@ static void print_PSD( struct xen_psd_pa 62.32 ptr->num_processors); 62.33 } 62.34 62.35 +static void print_PPC(unsigned int platform_limit) 62.36 +{ 62.37 + printk(KERN_INFO "\t_PPC: %d\n", platform_limit); 62.38 +} 62.39 + 62.40 int set_px_pminfo(uint32_t acpi_id, struct xen_processor_performance *dom0_px_info) 62.41 { 62.42 int ret=0, cpuid; 62.43 struct processor_pminfo *pmpt; 62.44 struct processor_performance *pxpt; 62.45 62.46 - if ( !(xen_processor_pmbits & XEN_PROCESSOR_PM_PX) ) 62.47 - { 62.48 - ret = -ENOSYS; 62.49 - goto out; 62.50 - } 62.51 - 62.52 cpuid = get_cpu_id(acpi_id); 62.53 - if ( cpuid < 0 ) 62.54 + if ( cpuid < 0 || !dom0_px_info) 62.55 { 62.56 ret = -EINVAL; 62.57 goto out; 62.58 @@ -256,6 +264,8 @@ int set_px_pminfo(uint32_t acpi_id, stru 62.59 memcpy ((void *)&pxpt->status_register, 62.60 (void *)&dom0_px_info->status_register, 62.61 sizeof(struct xen_pct_register)); 62.62 + print_PCT(&pxpt->control_register); 62.63 + print_PCT(&pxpt->status_register); 62.64 } 62.65 if ( dom0_px_info->flags & XEN_PX_PSS ) 62.66 { 62.67 @@ -265,12 +275,8 @@ int set_px_pminfo(uint32_t acpi_id, stru 62.68 ret = -ENOMEM; 62.69 goto out; 62.70 } 62.71 - if ( xenpf_copy_px_states(pxpt, dom0_px_info) ) 62.72 - { 62.73 - xfree(pxpt->states); 62.74 - ret = -EFAULT; 62.75 - goto out; 62.76 - } 62.77 + copy_from_guest(pxpt->states, dom0_px_info->states, 62.78 + dom0_px_info->state_count); 62.79 pxpt->state_count = dom0_px_info->state_count; 62.80 print_PSS(pxpt->states,pxpt->state_count); 62.81 } 62.82 @@ -285,6 +291,7 @@ int set_px_pminfo(uint32_t acpi_id, stru 62.83 if ( dom0_px_info->flags & XEN_PX_PPC ) 62.84 { 62.85 pxpt->platform_limit = dom0_px_info->platform_limit; 62.86 + print_PPC(pxpt->platform_limit); 62.87 62.88 if ( pxpt->init == XEN_PX_INIT ) 62.89 {
63.1 --- a/xen/drivers/cpufreq/utility.c Wed Oct 22 11:38:22 2008 +0900 63.2 +++ b/xen/drivers/cpufreq/utility.c Wed Oct 22 11:46:55 2008 +0900 63.3 @@ -27,6 +27,7 @@ 63.4 #include <xen/types.h> 63.5 #include <xen/sched.h> 63.6 #include <xen/timer.h> 63.7 +#include <xen/trace.h> 63.8 #include <asm/config.h> 63.9 #include <acpi/cpufreq/cpufreq.h> 63.10 #include <public/sysctl.h> 63.11 @@ -72,27 +73,30 @@ int cpufreq_statistic_init(unsigned int 63.12 struct pm_px *pxpt = cpufreq_statistic_data[cpuid]; 63.13 const struct processor_pminfo *pmpt = processor_pminfo[cpuid]; 63.14 63.15 - count = pmpt->perf.state_count; 63.16 - 63.17 if ( !pmpt ) 63.18 return -EINVAL; 63.19 63.20 + if ( pxpt ) 63.21 + return 0; 63.22 + 63.23 + count = pmpt->perf.state_count; 63.24 + 63.25 + pxpt = xmalloc(struct pm_px); 63.26 if ( !pxpt ) 63.27 - { 63.28 - pxpt = xmalloc(struct pm_px); 63.29 - if ( !pxpt ) 63.30 - return -ENOMEM; 63.31 - memset(pxpt, 0, sizeof(*pxpt)); 63.32 - cpufreq_statistic_data[cpuid] = pxpt; 63.33 - } 63.34 + return -ENOMEM; 63.35 + memset(pxpt, 0, sizeof(*pxpt)); 63.36 + cpufreq_statistic_data[cpuid] = pxpt; 63.37 63.38 pxpt->u.trans_pt = xmalloc_array(uint64_t, count * count); 63.39 - if (!pxpt->u.trans_pt) 63.40 + if (!pxpt->u.trans_pt) { 63.41 + xfree(pxpt); 63.42 return -ENOMEM; 63.43 + } 63.44 63.45 pxpt->u.pt = xmalloc_array(struct pm_px_val, count); 63.46 if (!pxpt->u.pt) { 63.47 xfree(pxpt->u.trans_pt); 63.48 + xfree(pxpt); 63.49 return -ENOMEM; 63.50 } 63.51 63.52 @@ -119,7 +123,8 @@ void cpufreq_statistic_exit(unsigned int 63.53 return; 63.54 xfree(pxpt->u.trans_pt); 63.55 xfree(pxpt->u.pt); 63.56 - memset(pxpt, 0, sizeof(struct pm_px)); 63.57 + xfree(pxpt); 63.58 + cpufreq_statistic_data[cpuid] = NULL; 63.59 } 63.60 63.61 void cpufreq_statistic_reset(unsigned int cpuid) 63.62 @@ -128,7 +133,7 @@ void cpufreq_statistic_reset(unsigned in 63.63 struct pm_px *pxpt = cpufreq_statistic_data[cpuid]; 63.64 const struct processor_pminfo *pmpt = processor_pminfo[cpuid]; 63.65 63.66 - if ( !pxpt || !pmpt ) 63.67 + if ( !pmpt || !pxpt || !pxpt->u.pt || !pxpt->u.trans_pt ) 63.68 return; 63.69 63.70 count = pmpt->perf.state_count; 63.71 @@ -293,7 +298,13 @@ int __cpufreq_driver_target(struct cpufr 63.72 int retval = -EINVAL; 63.73 63.74 if (cpu_online(policy->cpu) && cpufreq_driver->target) 63.75 + { 63.76 + unsigned int prev_freq = policy->cur; 63.77 + 63.78 retval = cpufreq_driver->target(policy, target_freq, relation); 63.79 + if ( retval == 0 ) 63.80 + TRACE_2D(TRC_PM_FREQ_CHANGE, prev_freq/1000, policy->cur/1000); 63.81 + } 63.82 63.83 return retval; 63.84 }
64.1 --- a/xen/drivers/passthrough/Makefile Wed Oct 22 11:38:22 2008 +0900 64.2 +++ b/xen/drivers/passthrough/Makefile Wed Oct 22 11:46:55 2008 +0900 64.3 @@ -1,4 +1,5 @@ 64.4 subdir-$(x86) += vtd 64.5 +subdir-$(ia64) += vtd 64.6 subdir-$(x86) += amd 64.7 64.8 obj-y += iommu.o
65.1 --- a/xen/drivers/passthrough/io.c Wed Oct 22 11:38:22 2008 +0900 65.2 +++ b/xen/drivers/passthrough/io.c Wed Oct 22 11:46:55 2008 +0900 65.3 @@ -20,6 +20,9 @@ 65.4 65.5 #include <xen/event.h> 65.6 #include <xen/iommu.h> 65.7 +#include <asm/hvm/irq.h> 65.8 +#include <asm/hvm/iommu.h> 65.9 +#include <xen/hvm/irq.h> 65.10 65.11 static void pt_irq_time_out(void *data) 65.12 { 65.13 @@ -245,6 +248,7 @@ int hvm_do_IRQ_dpci(struct domain *d, un 65.14 return 1; 65.15 } 65.16 65.17 +#ifdef SUPPORT_MSI_REMAPPING 65.18 void hvm_dpci_msi_eoi(struct domain *d, int vector) 65.19 { 65.20 struct hvm_irq_dpci *hvm_irq_dpci = d->arch.hvm_domain.irq.dpci; 65.21 @@ -278,6 +282,63 @@ void hvm_dpci_msi_eoi(struct domain *d, 65.22 spin_unlock(&d->event_lock); 65.23 } 65.24 65.25 +extern int vmsi_deliver(struct domain *d, int pirq); 65.26 +static int hvm_pci_msi_assert(struct domain *d, int pirq) 65.27 +{ 65.28 + return vmsi_deliver(d, pirq); 65.29 +} 65.30 +#endif 65.31 + 65.32 +void hvm_dirq_assist(struct vcpu *v) 65.33 +{ 65.34 + unsigned int irq; 65.35 + uint32_t device, intx; 65.36 + struct domain *d = v->domain; 65.37 + struct hvm_irq_dpci *hvm_irq_dpci = d->arch.hvm_domain.irq.dpci; 65.38 + struct dev_intx_gsi_link *digl; 65.39 + 65.40 + if ( !iommu_enabled || (v->vcpu_id != 0) || (hvm_irq_dpci == NULL) ) 65.41 + return; 65.42 + 65.43 + for ( irq = find_first_bit(hvm_irq_dpci->dirq_mask, NR_IRQS); 65.44 + irq < NR_IRQS; 65.45 + irq = find_next_bit(hvm_irq_dpci->dirq_mask, NR_IRQS, irq + 1) ) 65.46 + { 65.47 + if ( !test_and_clear_bit(irq, &hvm_irq_dpci->dirq_mask) ) 65.48 + continue; 65.49 + 65.50 + spin_lock(&d->event_lock); 65.51 +#ifdef SUPPORT_MSI_REMAPPING 65.52 + if ( test_bit(_HVM_IRQ_DPCI_MSI, &hvm_irq_dpci->mirq[irq].flags) ) 65.53 + { 65.54 + hvm_pci_msi_assert(d, irq); 65.55 + spin_unlock(&d->event_lock); 65.56 + continue; 65.57 + } 65.58 +#endif 65.59 + stop_timer(&hvm_irq_dpci->hvm_timer[domain_irq_to_vector(d, irq)]); 65.60 + 65.61 + list_for_each_entry ( digl, &hvm_irq_dpci->mirq[irq].digl_list, list ) 65.62 + { 65.63 + device = digl->device; 65.64 + intx = digl->intx; 65.65 + hvm_pci_intx_assert(d, device, intx); 65.66 + hvm_irq_dpci->mirq[irq].pending++; 65.67 + } 65.68 + 65.69 + /* 65.70 + * Set a timer to see if the guest can finish the interrupt or not. For 65.71 + * example, the guest OS may unmask the PIC during boot, before the 65.72 + * guest driver is loaded. hvm_pci_intx_assert() may succeed, but the 65.73 + * guest will never deal with the irq, then the physical interrupt line 65.74 + * will never be deasserted. 65.75 + */ 65.76 + set_timer(&hvm_irq_dpci->hvm_timer[domain_irq_to_vector(d, irq)], 65.77 + NOW() + PT_IRQ_TIME_OUT); 65.78 + spin_unlock(&d->event_lock); 65.79 + } 65.80 +} 65.81 + 65.82 void hvm_dpci_eoi(struct domain *d, unsigned int guest_gsi, 65.83 union vioapic_redir_entry *ent) 65.84 {
66.1 --- a/xen/drivers/passthrough/iommu.c Wed Oct 22 11:38:22 2008 +0900 66.2 +++ b/xen/drivers/passthrough/iommu.c Wed Oct 22 11:46:55 2008 +0900 66.3 @@ -19,8 +19,6 @@ 66.4 #include <xen/paging.h> 66.5 #include <xen/guest_access.h> 66.6 66.7 -extern struct iommu_ops intel_iommu_ops; 66.8 -extern struct iommu_ops amd_iommu_ops; 66.9 static void parse_iommu_param(char *s); 66.10 static int iommu_populate_page_table(struct domain *d); 66.11 int intel_vtd_setup(void);
67.1 --- a/xen/drivers/passthrough/pci.c Wed Oct 22 11:38:22 2008 +0900 67.2 +++ b/xen/drivers/passthrough/pci.c Wed Oct 22 11:46:55 2008 +0900 67.3 @@ -21,6 +21,8 @@ 67.4 #include <xen/list.h> 67.5 #include <xen/prefetch.h> 67.6 #include <xen/iommu.h> 67.7 +#include <asm/hvm/iommu.h> 67.8 +#include <asm/hvm/irq.h> 67.9 #include <xen/delay.h> 67.10 #include <xen/keyhandler.h> 67.11 67.12 @@ -207,6 +209,7 @@ void pci_release_devices(struct domain * 67.13 } 67.14 } 67.15 67.16 +#ifdef SUPPORT_MSI_REMAPPING 67.17 static void dump_pci_devices(unsigned char ch) 67.18 { 67.19 struct pci_dev *pdev; 67.20 @@ -236,7 +239,7 @@ static int __init setup_dump_pcidevs(voi 67.21 return 0; 67.22 } 67.23 __initcall(setup_dump_pcidevs); 67.24 - 67.25 +#endif 67.26 67.27 67.28 /*
68.1 --- a/xen/drivers/passthrough/vtd/Makefile Wed Oct 22 11:38:22 2008 +0900 68.2 +++ b/xen/drivers/passthrough/vtd/Makefile Wed Oct 22 11:46:55 2008 +0900 68.3 @@ -1,4 +1,5 @@ 68.4 subdir-$(x86) += x86 68.5 +subdir-$(ia64) += ia64 68.6 68.7 obj-y += iommu.o 68.8 obj-y += dmar.o
69.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000 69.2 +++ b/xen/drivers/passthrough/vtd/ia64/Makefile Wed Oct 22 11:46:55 2008 +0900 69.3 @@ -0,0 +1,1 @@ 69.4 +obj-y += vtd.o
70.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000 70.2 +++ b/xen/drivers/passthrough/vtd/ia64/vtd.c Wed Oct 22 11:46:55 2008 +0900 70.3 @@ -0,0 +1,112 @@ 70.4 +/* 70.5 + * Copyright (c) 2008, Intel Corporation. 70.6 + * 70.7 + * This program is free software; you can redistribute it and/or modify it 70.8 + * under the terms and conditions of the GNU General Public License, 70.9 + * version 2, as published by the Free Software Foundation. 70.10 + * 70.11 + * This program is distributed in the hope it will be useful, but WITHOUT 70.12 + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or 70.13 + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for 70.14 + * more details. 70.15 + * 70.16 + * You should have received a copy of the GNU General Public License along with 70.17 + * this program; if not, write to the Free Software Foundation, Inc., 59 Temple 70.18 + * Place - Suite 330, Boston, MA 02111-1307 USA. 70.19 + * 70.20 + * Copyright (C) Allen Kay <allen.m.kay@intel.com> 70.21 + * Copyright (C) Weidong Han <weidong.han@intel.com> 70.22 + */ 70.23 + 70.24 +#include <xen/sched.h> 70.25 +#include <xen/domain_page.h> 70.26 +#include <xen/iommu.h> 70.27 +#include <asm/xensystem.h> 70.28 +#include <asm/sal.h> 70.29 +#include "../iommu.h" 70.30 +#include "../dmar.h" 70.31 +#include "../vtd.h" 70.32 + 70.33 + 70.34 +int vector_irq[NR_VECTORS] __read_mostly = { [0 ... NR_VECTORS - 1] = -1}; 70.35 +/* irq_vectors is indexed by the sum of all RTEs in all I/O APICs. */ 70.36 +u8 irq_vector[NR_IRQ_VECTORS] __read_mostly; 70.37 + 70.38 +void *map_vtd_domain_page(u64 maddr) 70.39 +{ 70.40 + return (void *)((u64)map_domain_page(maddr >> PAGE_SHIFT) | 70.41 + (maddr & (PAGE_SIZE - PAGE_SIZE_4K))); 70.42 +} 70.43 + 70.44 +void unmap_vtd_domain_page(void *va) 70.45 +{ 70.46 + unmap_domain_page(va); 70.47 +} 70.48 + 70.49 +/* Allocate page table, return its machine address */ 70.50 +u64 alloc_pgtable_maddr(void) 70.51 +{ 70.52 + struct page_info *pg; 70.53 + u64 *vaddr; 70.54 + 70.55 + pg = alloc_domheap_page(NULL, 0); 70.56 + vaddr = map_domain_page(page_to_mfn(pg)); 70.57 + if ( !vaddr ) 70.58 + return 0; 70.59 + memset(vaddr, 0, PAGE_SIZE); 70.60 + 70.61 + iommu_flush_cache_page(vaddr); 70.62 + unmap_domain_page(vaddr); 70.63 + 70.64 + return page_to_maddr(pg); 70.65 +} 70.66 + 70.67 +void free_pgtable_maddr(u64 maddr) 70.68 +{ 70.69 + if ( maddr != 0 ) 70.70 + free_domheap_page(maddr_to_page(maddr)); 70.71 +} 70.72 + 70.73 +unsigned int get_cache_line_size(void) 70.74 +{ 70.75 + return L1_CACHE_BYTES; 70.76 +} 70.77 + 70.78 +void cacheline_flush(char * addr) 70.79 +{ 70.80 + ia64_fc(addr); 70.81 + ia64_sync_i(); 70.82 + ia64_srlz_i(); 70.83 +} 70.84 + 70.85 +void flush_all_cache() 70.86 +{ 70.87 + ia64_sal_cache_flush(3); 70.88 +} 70.89 + 70.90 +void * map_to_nocache_virt(int nr_iommus, u64 maddr) 70.91 +{ 70.92 + return (void *) ( maddr + __IA64_UNCACHED_OFFSET); 70.93 +} 70.94 + 70.95 +struct hvm_irq_dpci *domain_get_irq_dpci(struct domain *domain) 70.96 +{ 70.97 + if ( !domain ) 70.98 + return NULL; 70.99 + 70.100 + return domain->arch.hvm_domain.irq.dpci; 70.101 +} 70.102 + 70.103 +int domain_set_irq_dpci(struct domain *domain, struct hvm_irq_dpci *dpci) 70.104 +{ 70.105 + if ( !domain || !dpci ) 70.106 + return 0; 70.107 + 70.108 + domain->arch.hvm_domain.irq.dpci = dpci; 70.109 + return 1; 70.110 +} 70.111 + 70.112 +void hvm_dpci_isairq_eoi(struct domain *d, unsigned int isairq) 70.113 +{ 70.114 + /* dummy */ 70.115 +}
71.1 --- a/xen/drivers/passthrough/vtd/intremap.c Wed Oct 22 11:38:22 2008 +0900 71.2 +++ b/xen/drivers/passthrough/vtd/intremap.c Wed Oct 22 11:46:55 2008 +0900 71.3 @@ -21,6 +21,7 @@ 71.4 #include <xen/irq.h> 71.5 #include <xen/sched.h> 71.6 #include <xen/iommu.h> 71.7 +#include <asm/hvm/iommu.h> 71.8 #include <xen/time.h> 71.9 #include <xen/pci.h> 71.10 #include <xen/pci_regs.h> 71.11 @@ -128,7 +129,13 @@ static int ioapic_rte_to_remap_entry(str 71.12 memcpy(&new_ire, iremap_entry, sizeof(struct iremap_entry)); 71.13 71.14 if ( rte_upper ) 71.15 + { 71.16 +#if defined(__i386__) || defined(__x86_64__) 71.17 new_ire.lo.dst = (value >> 24) << 8; 71.18 +#else /* __ia64__ */ 71.19 + new_ire.lo.dst = value >> 16; 71.20 +#endif 71.21 + } 71.22 else 71.23 { 71.24 *(((u32 *)&new_rte) + 0) = value; 71.25 @@ -179,7 +186,7 @@ unsigned int io_apic_read_remap_rte( 71.26 struct IO_xAPIC_route_entry old_rte = { 0 }; 71.27 struct IO_APIC_route_remap_entry *remap_rte; 71.28 int rte_upper = (reg & 1) ? 1 : 0; 71.29 - struct iommu *iommu = ioapic_to_iommu(mp_ioapics[apic].mpc_apicid); 71.30 + struct iommu *iommu = ioapic_to_iommu(IO_APIC_ID(apic)); 71.31 struct ir_ctrl *ir_ctrl = iommu_ir_ctrl(iommu); 71.32 71.33 if ( !iommu || !ir_ctrl || ir_ctrl->iremap_maddr == 0 || 71.34 @@ -224,7 +231,7 @@ void io_apic_write_remap_rte( 71.35 struct IO_xAPIC_route_entry old_rte = { 0 }; 71.36 struct IO_APIC_route_remap_entry *remap_rte; 71.37 unsigned int rte_upper = (reg & 1) ? 1 : 0; 71.38 - struct iommu *iommu = ioapic_to_iommu(mp_ioapics[apic].mpc_apicid); 71.39 + struct iommu *iommu = ioapic_to_iommu(IO_APIC_ID(apic)); 71.40 struct ir_ctrl *ir_ctrl = iommu_ir_ctrl(iommu); 71.41 int saved_mask; 71.42 71.43 @@ -253,7 +260,7 @@ void io_apic_write_remap_rte( 71.44 *(IO_APIC_BASE(apic)+4) = *(((int *)&old_rte)+0); 71.45 remap_rte->mask = saved_mask; 71.46 71.47 - if ( ioapic_rte_to_remap_entry(iommu, mp_ioapics[apic].mpc_apicid, 71.48 + if ( ioapic_rte_to_remap_entry(iommu, IO_APIC_ID(apic), 71.49 &old_rte, rte_upper, value) ) 71.50 { 71.51 *IO_APIC_BASE(apic) = rte_upper ? (reg + 1) : reg; 71.52 @@ -328,7 +335,8 @@ static int remap_entry_to_msi_msg( 71.53 } 71.54 71.55 static int msi_msg_to_remap_entry( 71.56 - struct iommu *iommu, struct pci_dev *pdev, struct msi_msg *msg) 71.57 + struct iommu *iommu, struct pci_dev *pdev, 71.58 + struct msi_desc *msi_desc, struct msi_msg *msg) 71.59 { 71.60 struct iremap_entry *iremap_entry = NULL, *iremap_entries; 71.61 struct iremap_entry new_ire; 71.62 @@ -336,32 +344,18 @@ static int msi_msg_to_remap_entry( 71.63 unsigned int index; 71.64 unsigned long flags; 71.65 struct ir_ctrl *ir_ctrl = iommu_ir_ctrl(iommu); 71.66 - int i = 0; 71.67 71.68 remap_rte = (struct msi_msg_remap_entry *) msg; 71.69 spin_lock_irqsave(&ir_ctrl->iremap_lock, flags); 71.70 71.71 - iremap_entries = 71.72 - (struct iremap_entry *)map_vtd_domain_page(ir_ctrl->iremap_maddr); 71.73 - 71.74 - /* If the entry for a PCI device has been there, use the old entry, 71.75 - * Or, assign a new entry for it. 71.76 - */ 71.77 - for ( i = 0; i <= ir_ctrl->iremap_index; i++ ) 71.78 + if ( msi_desc->remap_index < 0 ) 71.79 { 71.80 - iremap_entry = &iremap_entries[i]; 71.81 - if ( iremap_entry->hi.sid == 71.82 - ((pdev->bus << 8) | pdev->devfn) ) 71.83 - break; 71.84 - } 71.85 - 71.86 - if ( i > ir_ctrl->iremap_index ) 71.87 - { 71.88 - ir_ctrl->iremap_index++; 71.89 + ir_ctrl->iremap_index++; 71.90 index = ir_ctrl->iremap_index; 71.91 + msi_desc->remap_index = index; 71.92 } 71.93 else 71.94 - index = i; 71.95 + index = msi_desc->remap_index; 71.96 71.97 if ( index > IREMAP_ENTRY_NR - 1 ) 71.98 { 71.99 @@ -369,11 +363,13 @@ static int msi_msg_to_remap_entry( 71.100 "%s: intremap index (%d) is larger than" 71.101 " the maximum index (%ld)!\n", 71.102 __func__, index, IREMAP_ENTRY_NR - 1); 71.103 - unmap_vtd_domain_page(iremap_entries); 71.104 + msi_desc->remap_index = -1; 71.105 spin_unlock_irqrestore(&ir_ctrl->iremap_lock, flags); 71.106 return -EFAULT; 71.107 } 71.108 71.109 + iremap_entries = 71.110 + (struct iremap_entry *)map_vtd_domain_page(ir_ctrl->iremap_maddr); 71.111 iremap_entry = &iremap_entries[index]; 71.112 memcpy(&new_ire, iremap_entry, sizeof(struct iremap_entry)); 71.113 71.114 @@ -450,7 +446,7 @@ void msi_msg_write_remap_rte( 71.115 if ( !iommu || !ir_ctrl || ir_ctrl->iremap_maddr == 0 ) 71.116 return; 71.117 71.118 - msi_msg_to_remap_entry(iommu, pdev, msg); 71.119 + msi_msg_to_remap_entry(iommu, pdev, msi_desc, msg); 71.120 } 71.121 #elif defined(__ia64__) 71.122 void msi_msg_read_remap_rte( 71.123 @@ -482,7 +478,7 @@ int intremap_setup(struct iommu *iommu) 71.124 { 71.125 dprintk(XENLOG_WARNING VTDPREFIX, 71.126 "Cannot allocate memory for ir_ctrl->iremap_maddr\n"); 71.127 - return -ENODEV; 71.128 + return -ENOMEM; 71.129 } 71.130 ir_ctrl->iremap_index = -1; 71.131 } 71.132 @@ -490,10 +486,10 @@ int intremap_setup(struct iommu *iommu) 71.133 #if defined(ENABLED_EXTENDED_INTERRUPT_SUPPORT) 71.134 /* set extended interrupt mode bit */ 71.135 ir_ctrl->iremap_maddr |= 71.136 - ecap_ext_intr(iommu->ecap) ? (1 << IRTA_REG_EIMI_SHIFT) : 0; 71.137 + ecap_ext_intr(iommu->ecap) ? (1 << IRTA_REG_EIME_SHIFT) : 0; 71.138 #endif 71.139 - /* size field = 256 entries per 4K page = 8 - 1 */ 71.140 - ir_ctrl->iremap_maddr |= 7; 71.141 + /* set size of the interrupt remapping table */ 71.142 + ir_ctrl->iremap_maddr |= IRTA_REG_TABLE_SIZE; 71.143 dmar_writeq(iommu->reg, DMAR_IRTA_REG, ir_ctrl->iremap_maddr); 71.144 71.145 /* set SIRTP */
72.1 --- a/xen/drivers/passthrough/vtd/iommu.c Wed Oct 22 11:38:22 2008 +0900 72.2 +++ b/xen/drivers/passthrough/vtd/iommu.c Wed Oct 22 11:46:55 2008 +0900 72.3 @@ -24,6 +24,7 @@ 72.4 #include <xen/xmalloc.h> 72.5 #include <xen/domain_page.h> 72.6 #include <xen/iommu.h> 72.7 +#include <asm/hvm/iommu.h> 72.8 #include <xen/numa.h> 72.9 #include <xen/time.h> 72.10 #include <xen/pci.h> 72.11 @@ -218,10 +219,10 @@ static u64 addr_to_dma_page_maddr(struct 72.12 if ( !alloc ) 72.13 break; 72.14 maddr = alloc_pgtable_maddr(); 72.15 + if ( !maddr ) 72.16 + break; 72.17 dma_set_pte_addr(*pte, maddr); 72.18 vaddr = map_vtd_domain_page(maddr); 72.19 - if ( !vaddr ) 72.20 - break; 72.21 72.22 /* 72.23 * high level table always sets r/w, last level 72.24 @@ -234,8 +235,6 @@ static u64 addr_to_dma_page_maddr(struct 72.25 else 72.26 { 72.27 vaddr = map_vtd_domain_page(pte->val); 72.28 - if ( !vaddr ) 72.29 - break; 72.30 } 72.31 72.32 if ( level == 2 ) 72.33 @@ -569,26 +568,6 @@ static void dma_pte_clear_one(struct dom 72.34 unmap_vtd_domain_page(page); 72.35 } 72.36 72.37 -/* clear last level pte, a tlb flush should be followed */ 72.38 -static void dma_pte_clear_range(struct domain *domain, u64 start, u64 end) 72.39 -{ 72.40 - struct hvm_iommu *hd = domain_hvm_iommu(domain); 72.41 - int addr_width = agaw_to_width(hd->agaw); 72.42 - 72.43 - start &= (((u64)1) << addr_width) - 1; 72.44 - end &= (((u64)1) << addr_width) - 1; 72.45 - /* in case it's partial page */ 72.46 - start = PAGE_ALIGN_4K(start); 72.47 - end &= PAGE_MASK_4K; 72.48 - 72.49 - /* we don't need lock here, nobody else touches the iova range */ 72.50 - while ( start < end ) 72.51 - { 72.52 - dma_pte_clear_one(domain, start); 72.53 - start += PAGE_SIZE_4K; 72.54 - } 72.55 -} 72.56 - 72.57 static void iommu_free_pagetable(u64 pt_maddr, int level) 72.58 { 72.59 int i; 72.60 @@ -877,6 +856,7 @@ static void dma_msi_data_init(struct iom 72.61 spin_unlock_irqrestore(&iommu->register_lock, flags); 72.62 } 72.63 72.64 +#ifdef SUPPORT_MSI_REMAPPING 72.65 static void dma_msi_addr_init(struct iommu *iommu, int phy_cpu) 72.66 { 72.67 u64 msi_address; 72.68 @@ -893,6 +873,12 @@ static void dma_msi_addr_init(struct iom 72.69 dmar_writel(iommu->reg, DMAR_FEUADDR_REG, (u32)(msi_address >> 32)); 72.70 spin_unlock_irqrestore(&iommu->register_lock, flags); 72.71 } 72.72 +#else 72.73 +static void dma_msi_addr_init(struct iommu *iommu, int phy_cpu) 72.74 +{ 72.75 + /* ia64: TODO */ 72.76 +} 72.77 +#endif 72.78 72.79 static void dma_msi_set_affinity(unsigned int vector, cpumask_t dest) 72.80 { 72.81 @@ -1024,7 +1010,7 @@ static int intel_iommu_domain_init(struc 72.82 { 72.83 struct hvm_iommu *hd = domain_hvm_iommu(d); 72.84 struct iommu *iommu = NULL; 72.85 - u64 i; 72.86 + u64 i, j, tmp; 72.87 struct acpi_drhd_unit *drhd; 72.88 72.89 drhd = list_entry(acpi_drhd_units.next, typeof(*drhd), list); 72.90 @@ -1043,11 +1029,13 @@ static int intel_iommu_domain_init(struc 72.91 */ 72.92 for ( i = 0; i < max_page; i++ ) 72.93 { 72.94 - if ( xen_in_range(i << PAGE_SHIFT_4K, (i + 1) << PAGE_SHIFT_4K) || 72.95 - tboot_in_range(i << PAGE_SHIFT_4K, (i + 1) << PAGE_SHIFT_4K) ) 72.96 + if ( xen_in_range(i << PAGE_SHIFT, (i + 1) << PAGE_SHIFT) || 72.97 + tboot_in_range(i << PAGE_SHIFT, (i + 1) << PAGE_SHIFT) ) 72.98 continue; 72.99 72.100 - iommu_map_page(d, i, i); 72.101 + tmp = 1 << (PAGE_SHIFT - PAGE_SHIFT_4K); 72.102 + for ( j = 0; j < tmp; j++ ) 72.103 + iommu_map_page(d, (i*tmp+j), (i*tmp+j)); 72.104 } 72.105 72.106 setup_dom0_devices(d); 72.107 @@ -1511,75 +1499,26 @@ int intel_iommu_unmap_page(struct domain 72.108 return 0; 72.109 } 72.110 72.111 -int iommu_page_mapping(struct domain *domain, paddr_t iova, 72.112 - paddr_t hpa, size_t size, int prot) 72.113 -{ 72.114 - struct hvm_iommu *hd = domain_hvm_iommu(domain); 72.115 - struct acpi_drhd_unit *drhd; 72.116 - struct iommu *iommu; 72.117 - u64 start_pfn, end_pfn; 72.118 - struct dma_pte *page = NULL, *pte = NULL; 72.119 - int index; 72.120 - u64 pg_maddr; 72.121 - 72.122 - if ( (prot & (DMA_PTE_READ|DMA_PTE_WRITE)) == 0 ) 72.123 - return -EINVAL; 72.124 - 72.125 - iova = (iova >> PAGE_SHIFT_4K) << PAGE_SHIFT_4K; 72.126 - start_pfn = hpa >> PAGE_SHIFT_4K; 72.127 - end_pfn = (PAGE_ALIGN_4K(hpa + size)) >> PAGE_SHIFT_4K; 72.128 - index = 0; 72.129 - while ( start_pfn < end_pfn ) 72.130 - { 72.131 - pg_maddr = addr_to_dma_page_maddr(domain, iova + PAGE_SIZE_4K*index, 1); 72.132 - if ( pg_maddr == 0 ) 72.133 - return -ENOMEM; 72.134 - page = (struct dma_pte *)map_vtd_domain_page(pg_maddr); 72.135 - pte = page + (start_pfn & LEVEL_MASK); 72.136 - dma_set_pte_addr(*pte, (paddr_t)start_pfn << PAGE_SHIFT_4K); 72.137 - dma_set_pte_prot(*pte, prot); 72.138 - iommu_flush_cache_entry(pte); 72.139 - unmap_vtd_domain_page(page); 72.140 - start_pfn++; 72.141 - index++; 72.142 - } 72.143 - 72.144 - if ( index > 0 ) 72.145 - { 72.146 - for_each_drhd_unit ( drhd ) 72.147 - { 72.148 - iommu = drhd->iommu; 72.149 - if ( test_bit(iommu->index, &hd->iommu_bitmap) ) 72.150 - if ( iommu_flush_iotlb_psi(iommu, domain_iommu_domid(domain), 72.151 - iova, index, 1)) 72.152 - iommu_flush_write_buffer(iommu); 72.153 - } 72.154 - } 72.155 - 72.156 - return 0; 72.157 -} 72.158 - 72.159 -int iommu_page_unmapping(struct domain *domain, paddr_t addr, size_t size) 72.160 -{ 72.161 - dma_pte_clear_range(domain, addr, addr + size); 72.162 - 72.163 - return 0; 72.164 -} 72.165 - 72.166 static int iommu_prepare_rmrr_dev(struct domain *d, 72.167 struct acpi_rmrr_unit *rmrr, 72.168 u8 bus, u8 devfn) 72.169 { 72.170 - u64 size; 72.171 - int ret; 72.172 + int ret = 0; 72.173 + u64 base, end; 72.174 + unsigned long base_pfn, end_pfn; 72.175 72.176 - /* page table init */ 72.177 - size = rmrr->end_address - rmrr->base_address + 1; 72.178 - ret = iommu_page_mapping(d, rmrr->base_address, 72.179 - rmrr->base_address, size, 72.180 - DMA_PTE_READ|DMA_PTE_WRITE); 72.181 - if ( ret ) 72.182 - return ret; 72.183 + ASSERT(rmrr->base_address < rmrr->end_address); 72.184 + 72.185 + base = rmrr->base_address & PAGE_MASK_4K; 72.186 + base_pfn = base >> PAGE_SHIFT_4K; 72.187 + end = PAGE_ALIGN_4K(rmrr->end_address); 72.188 + end_pfn = end >> PAGE_SHIFT_4K; 72.189 + 72.190 + while ( base_pfn < end_pfn ) 72.191 + { 72.192 + intel_iommu_map_page(d, base_pfn, base_pfn); 72.193 + base_pfn++; 72.194 + } 72.195 72.196 if ( domain_context_mapped(bus, devfn) == 0 ) 72.197 ret = domain_context_mapping(d, bus, devfn);
73.1 --- a/xen/drivers/passthrough/vtd/qinval.c Wed Oct 22 11:38:22 2008 +0900 73.2 +++ b/xen/drivers/passthrough/vtd/qinval.c Wed Oct 22 11:46:55 2008 +0900 73.3 @@ -428,7 +428,11 @@ int qinval_setup(struct iommu *iommu) 73.4 { 73.5 qi_ctrl->qinval_maddr = alloc_pgtable_maddr(); 73.6 if ( qi_ctrl->qinval_maddr == 0 ) 73.7 - panic("Cannot allocate memory for qi_ctrl->qinval_maddr\n"); 73.8 + { 73.9 + dprintk(XENLOG_WARNING VTDPREFIX, 73.10 + "Cannot allocate memory for qi_ctrl->qinval_maddr\n"); 73.11 + return -ENOMEM; 73.12 + } 73.13 flush->context = flush_context_qi; 73.14 flush->iotlb = flush_iotlb_qi; 73.15 }
74.1 --- a/xen/drivers/passthrough/vtd/utils.c Wed Oct 22 11:38:22 2008 +0900 74.2 +++ b/xen/drivers/passthrough/vtd/utils.c Wed Oct 22 11:46:55 2008 +0900 74.3 @@ -204,6 +204,7 @@ void print_vtd_entries(struct iommu *iom 74.4 74.5 void dump_iommu_info(unsigned char key) 74.6 { 74.7 +#if defined(__i386__) || defined(__x86_64__) 74.8 struct acpi_drhd_unit *drhd; 74.9 struct iommu *iommu; 74.10 int i; 74.11 @@ -305,6 +306,10 @@ void dump_iommu_info(unsigned char key) 74.12 } 74.13 } 74.14 } 74.15 +#else 74.16 + printk("%s: not implemnted on IA64 for now.\n", __func__); 74.17 + /* ia64: TODO */ 74.18 +#endif 74.19 } 74.20 74.21 /*
75.1 --- a/xen/drivers/passthrough/vtd/x86/vtd.c Wed Oct 22 11:38:22 2008 +0900 75.2 +++ b/xen/drivers/passthrough/vtd/x86/vtd.c Wed Oct 22 11:46:55 2008 +0900 75.3 @@ -41,17 +41,19 @@ u64 alloc_pgtable_maddr(void) 75.4 { 75.5 struct page_info *pg; 75.6 u64 *vaddr; 75.7 + unsigned long mfn; 75.8 75.9 pg = alloc_domheap_page(NULL, 0); 75.10 - vaddr = map_domain_page(page_to_mfn(pg)); 75.11 - if ( !vaddr ) 75.12 + if ( !pg ) 75.13 return 0; 75.14 + mfn = page_to_mfn(pg); 75.15 + vaddr = map_domain_page(mfn); 75.16 memset(vaddr, 0, PAGE_SIZE); 75.17 75.18 iommu_flush_cache_page(vaddr); 75.19 unmap_domain_page(vaddr); 75.20 75.21 - return page_to_maddr(pg); 75.22 + return (u64)mfn << PAGE_SHIFT_4K; 75.23 } 75.24 75.25 void free_pgtable_maddr(u64 maddr)
76.1 --- a/xen/include/acpi/cpufreq/processor_perf.h Wed Oct 22 11:38:22 2008 +0900 76.2 +++ b/xen/include/acpi/cpufreq/processor_perf.h Wed Oct 22 11:46:55 2008 +0900 76.3 @@ -60,8 +60,5 @@ struct pm_px { 76.4 76.5 extern struct pm_px *cpufreq_statistic_data[NR_CPUS]; 76.6 76.7 -int xenpf_copy_px_states(struct processor_performance *pxpt, 76.8 - struct xen_processor_performance *dom0_px_info); 76.9 - 76.10 int cpufreq_cpu_init(unsigned int cpuid); 76.11 #endif /* __XEN_PROCESSOR_PM_H__ */
77.1 --- a/xen/include/asm-ia64/linux-xen/asm/spinlock.h Wed Oct 22 11:38:22 2008 +0900 77.2 +++ b/xen/include/asm-ia64/linux-xen/asm/spinlock.h Wed Oct 22 11:46:55 2008 +0900 77.3 @@ -27,25 +27,16 @@ typedef struct { 77.4 #ifdef DEBUG_SPINLOCK 77.5 void *locker; 77.6 #endif 77.7 -#ifdef XEN 77.8 - unsigned char recurse_cpu; 77.9 - unsigned char recurse_cnt; 77.10 -#endif 77.11 -} spinlock_t; 77.12 +} raw_spinlock_t; 77.13 77.14 #ifdef XEN 77.15 #ifdef DEBUG_SPINLOCK 77.16 -#define SPIN_LOCK_UNLOCKED /*(spinlock_t)*/ { 0, NULL, -1, 0 } 77.17 +#define _RAW_SPIN_LOCK_UNLOCKED /*(raw_spinlock_t)*/ { 0, NULL } 77.18 #else 77.19 -#define SPIN_LOCK_UNLOCKED /*(spinlock_t)*/ { 0, -1, 0 } 77.20 +#define _RAW_SPIN_LOCK_UNLOCKED /*(raw_spinlock_t)*/ { 0 } 77.21 #endif 77.22 -static inline void spin_lock_init(spinlock_t *lock) 77.23 -{ 77.24 - *lock = ((spinlock_t)SPIN_LOCK_UNLOCKED); 77.25 -} 77.26 #else 77.27 -#define SPIN_LOCK_UNLOCKED /*(spinlock_t)*/ { 0 } 77.28 -#define spin_lock_init(x) ((x)->lock = 0) 77.29 +#define _RAW_SPIN_LOCK_UNLOCKED /*(raw_spinlock_t)*/ { 0 } 77.30 #endif 77.31 77.32 #ifdef ASM_SUPPORTED 77.33 @@ -59,7 +50,7 @@ static inline void spin_lock_init(spinlo 77.34 #define IA64_SPINLOCK_CLOBBERS "ar.ccv", "ar.pfs", "p14", "p15", "r27", "r28", "r29", "r30", "b6", "memory" 77.35 77.36 static inline void 77.37 -_raw_spin_lock_flags (spinlock_t *lock, unsigned long flags) 77.38 +_raw_spin_lock_flags (raw_spinlock_t *lock, unsigned long flags) 77.39 { 77.40 register volatile unsigned int *ptr asm ("r31") = &lock->lock; 77.41 77.42 @@ -136,10 +127,9 @@ do { \ 77.43 } while (0) 77.44 #endif /* !ASM_SUPPORTED */ 77.45 77.46 -#define spin_is_locked(x) ((x)->lock != 0) 77.47 -#define _raw_spin_unlock(x) do { barrier(); ((spinlock_t *) x)->lock = 0; } while (0) 77.48 +#define _raw_spin_is_locked(x) ((x)->lock != 0) 77.49 +#define _raw_spin_unlock(x) do { barrier(); (x)->lock = 0; } while (0) 77.50 #define _raw_spin_trylock(x) (cmpxchg_acq(&(x)->lock, 0, 1) == 0) 77.51 -#define spin_unlock_wait(x) do { barrier(); } while ((x)->lock) 77.52 77.53 typedef struct { 77.54 volatile unsigned int read_counter : 31; 77.55 @@ -147,16 +137,12 @@ typedef struct { 77.56 #ifdef CONFIG_PREEMPT 77.57 unsigned int break_lock; 77.58 #endif 77.59 -} rwlock_t; 77.60 -#define RW_LOCK_UNLOCKED /*(rwlock_t)*/ { 0, 0 } 77.61 - 77.62 -#define rwlock_init(x) do { *(x) = (rwlock_t) RW_LOCK_UNLOCKED; } while(0) 77.63 -#define read_can_lock(rw) (*(volatile int *)(rw) >= 0) 77.64 -#define write_can_lock(rw) (*(volatile int *)(rw) == 0) 77.65 +} raw_rwlock_t; 77.66 +#define _RAW_RW_LOCK_UNLOCKED /*(raw_rwlock_t)*/ { 0, 0 } 77.67 77.68 #define _raw_read_lock(rw) \ 77.69 do { \ 77.70 - rwlock_t *__read_lock_ptr = (rw); \ 77.71 + raw_rwlock_t *__read_lock_ptr = (rw); \ 77.72 \ 77.73 while (unlikely(ia64_fetchadd(1, (int *) __read_lock_ptr, acq) < 0)) { \ 77.74 ia64_fetchadd(-1, (int *) __read_lock_ptr, rel); \ 77.75 @@ -167,7 +153,7 @@ do { \ 77.76 77.77 #define _raw_read_unlock(rw) \ 77.78 do { \ 77.79 - rwlock_t *__read_lock_ptr = (rw); \ 77.80 + raw_rwlock_t *__read_lock_ptr = (rw); \ 77.81 ia64_fetchadd(-1, (int *) __read_lock_ptr, rel); \ 77.82 } while (0) 77.83 77.84 @@ -230,7 +216,4 @@ do { \ 77.85 clear_bit(31, (x)); \ 77.86 }) 77.87 77.88 -#ifdef XEN 77.89 -#include <asm/xenspinlock.h> 77.90 -#endif 77.91 #endif /* _ASM_IA64_SPINLOCK_H */
78.1 --- a/xen/include/asm-ia64/xenspinlock.h Wed Oct 22 11:38:22 2008 +0900 78.2 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 78.3 @@ -1,30 +0,0 @@ 78.4 -#ifndef _ASM_IA64_XENSPINLOCK_H 78.5 -#define _ASM_IA64_XENSPINLOCK_H 78.6 - 78.7 -/* 78.8 - * spin_[un]lock_recursive(): Use these forms when the lock can (safely!) be 78.9 - * reentered recursively on the same CPU. All critical regions that may form 78.10 - * part of a recursively-nested set must be protected by these forms. If there 78.11 - * are any critical regions that cannot form part of such a set, they can use 78.12 - * standard spin_[un]lock(). 78.13 - */ 78.14 -#define _raw_spin_lock_recursive(_lock) \ 78.15 - do { \ 78.16 - int cpu = smp_processor_id(); \ 78.17 - if ( likely((_lock)->recurse_cpu != cpu) ) \ 78.18 - { \ 78.19 - spin_lock(_lock); \ 78.20 - (_lock)->recurse_cpu = cpu; \ 78.21 - } \ 78.22 - (_lock)->recurse_cnt++; \ 78.23 - } while ( 0 ) 78.24 - 78.25 -#define _raw_spin_unlock_recursive(_lock) \ 78.26 - do { \ 78.27 - if ( likely(--(_lock)->recurse_cnt == 0) ) \ 78.28 - { \ 78.29 - (_lock)->recurse_cpu = -1; \ 78.30 - spin_unlock(_lock); \ 78.31 - } \ 78.32 - } while ( 0 ) 78.33 -#endif /* _ASM_IA64_XENSPINLOCK_H */
79.1 --- a/xen/include/asm-x86/domain.h Wed Oct 22 11:38:22 2008 +0900 79.2 +++ b/xen/include/asm-x86/domain.h Wed Oct 22 11:46:55 2008 +0900 79.3 @@ -250,6 +250,8 @@ struct arch_domain 79.4 bool_t is_32bit_pv; 79.5 /* Is shared-info page in 32-bit format? */ 79.6 bool_t has_32bit_shinfo; 79.7 + /* Domain cannot handle spurious page faults? */ 79.8 + bool_t suppress_spurious_page_faults; 79.9 79.10 /* Continuable domain_relinquish_resources(). */ 79.11 enum {
80.1 --- a/xen/include/asm-x86/hvm/hvm.h Wed Oct 22 11:38:22 2008 +0900 80.2 +++ b/xen/include/asm-x86/hvm/hvm.h Wed Oct 22 11:46:55 2008 +0900 80.3 @@ -128,6 +128,7 @@ struct hvm_function_table { 80.4 int (*msr_write_intercept)(struct cpu_user_regs *regs); 80.5 void (*invlpg_intercept)(unsigned long vaddr); 80.6 void (*set_uc_mode)(struct vcpu *v); 80.7 + void (*set_info_guest)(struct vcpu *v); 80.8 }; 80.9 80.10 extern struct hvm_function_table hvm_funcs; 80.11 @@ -314,4 +315,10 @@ int hvm_virtual_to_linear_addr( 80.12 unsigned int addr_size, 80.13 unsigned long *linear_addr); 80.14 80.15 +static inline void hvm_set_info_guest(struct vcpu *v) 80.16 +{ 80.17 + if ( hvm_funcs.set_info_guest ) 80.18 + return hvm_funcs.set_info_guest(v); 80.19 +} 80.20 + 80.21 #endif /* __ASM_X86_HVM_HVM_H__ */
81.1 --- a/xen/include/asm-x86/hvm/irq.h Wed Oct 22 11:38:22 2008 +0900 81.2 +++ b/xen/include/asm-x86/hvm/irq.h Wed Oct 22 11:46:55 2008 +0900 81.3 @@ -22,62 +22,11 @@ 81.4 #ifndef __ASM_X86_HVM_IRQ_H__ 81.5 #define __ASM_X86_HVM_IRQ_H__ 81.6 81.7 -#include <xen/types.h> 81.8 -#include <xen/spinlock.h> 81.9 -#include <asm/irq.h> 81.10 #include <asm/pirq.h> 81.11 +#include <xen/hvm/irq.h> 81.12 #include <asm/hvm/hvm.h> 81.13 #include <asm/hvm/vpic.h> 81.14 #include <asm/hvm/vioapic.h> 81.15 -#include <public/hvm/save.h> 81.16 - 81.17 -struct dev_intx_gsi_link { 81.18 - struct list_head list; 81.19 - uint8_t device; 81.20 - uint8_t intx; 81.21 - uint8_t gsi; 81.22 - uint8_t link; 81.23 -}; 81.24 - 81.25 -#define _HVM_IRQ_DPCI_MSI 0x1 81.26 - 81.27 -struct hvm_gmsi_info { 81.28 - uint32_t gvec; 81.29 - uint32_t gflags; 81.30 -}; 81.31 - 81.32 -struct hvm_mirq_dpci_mapping { 81.33 - uint32_t flags; 81.34 - int pending; 81.35 - struct list_head digl_list; 81.36 - struct domain *dom; 81.37 - struct hvm_gmsi_info gmsi; 81.38 -}; 81.39 - 81.40 -struct hvm_girq_dpci_mapping { 81.41 - uint8_t valid; 81.42 - uint8_t device; 81.43 - uint8_t intx; 81.44 - uint8_t machine_gsi; 81.45 -}; 81.46 - 81.47 -#define NR_ISAIRQS 16 81.48 -#define NR_LINK 4 81.49 -/* Protected by domain's event_lock */ 81.50 -struct hvm_irq_dpci { 81.51 - /* Machine IRQ to guest device/intx mapping. */ 81.52 - DECLARE_BITMAP(mapping, NR_PIRQS); 81.53 - struct hvm_mirq_dpci_mapping mirq[NR_IRQS]; 81.54 - /* Guest IRQ to guest device/intx mapping. */ 81.55 - struct hvm_girq_dpci_mapping girq[NR_IRQS]; 81.56 - uint8_t msi_gvec_pirq[NR_VECTORS]; 81.57 - DECLARE_BITMAP(dirq_mask, NR_IRQS); 81.58 - /* Record of mapped ISA IRQs */ 81.59 - DECLARE_BITMAP(isairq_map, NR_ISAIRQS); 81.60 - /* Record of mapped Links */ 81.61 - uint8_t link_cnt[NR_LINK]; 81.62 - struct timer hvm_timer[NR_IRQS]; 81.63 -}; 81.64 81.65 struct hvm_irq { 81.66 /* 81.67 @@ -149,27 +98,16 @@ struct hvm_irq { 81.68 81.69 #define hvm_isa_irq_to_gsi(isa_irq) ((isa_irq) ? : 2) 81.70 81.71 -/* Modify state of a PCI INTx wire. */ 81.72 -void hvm_pci_intx_assert( 81.73 - struct domain *d, unsigned int device, unsigned int intx); 81.74 -void hvm_pci_intx_deassert( 81.75 - struct domain *d, unsigned int device, unsigned int intx); 81.76 - 81.77 -/* Modify state of an ISA device's IRQ wire. */ 81.78 -void hvm_isa_irq_assert( 81.79 - struct domain *d, unsigned int isa_irq); 81.80 -void hvm_isa_irq_deassert( 81.81 - struct domain *d, unsigned int isa_irq); 81.82 - 81.83 -void hvm_set_pci_link_route(struct domain *d, u8 link, u8 isa_irq); 81.84 - 81.85 -void hvm_maybe_deassert_evtchn_irq(void); 81.86 -void hvm_assert_evtchn_irq(struct vcpu *v); 81.87 -void hvm_set_callback_via(struct domain *d, uint64_t via); 81.88 - 81.89 /* Check/Acknowledge next pending interrupt. */ 81.90 struct hvm_intack hvm_vcpu_has_pending_irq(struct vcpu *v); 81.91 struct hvm_intack hvm_vcpu_ack_pending_irq(struct vcpu *v, 81.92 struct hvm_intack intack); 81.93 81.94 +/* 81.95 + * Currently IA64 Xen doesn't support MSI. So for x86, we define this macro 81.96 + * to control the conditional compilation of some MSI-related functions. 81.97 + * This macro will be removed once IA64 has MSI support. 81.98 + */ 81.99 +#define SUPPORT_MSI_REMAPPING 1 81.100 + 81.101 #endif /* __ASM_X86_HVM_IRQ_H__ */
82.1 --- a/xen/include/asm-x86/hvm/svm/vmcb.h Wed Oct 22 11:38:22 2008 +0900 82.2 +++ b/xen/include/asm-x86/hvm/svm/vmcb.h Wed Oct 22 11:46:55 2008 +0900 82.3 @@ -393,7 +393,9 @@ struct vmcb_struct { 82.4 eventinj_t eventinj; /* offset 0xA8 */ 82.5 u64 h_cr3; /* offset 0xB0 */ 82.6 lbrctrl_t lbr_control; /* offset 0xB8 */ 82.7 - u64 res09[104]; /* offset 0xC0 pad to save area */ 82.8 + u64 res09; /* offset 0xC0 */ 82.9 + u64 nextrip; /* offset 0xC8 */ 82.10 + u64 res10a[102]; /* offset 0xD0 pad to save area */ 82.11 82.12 svm_segment_register_t es; /* offset 1024 */ 82.13 svm_segment_register_t cs;
83.1 --- a/xen/include/asm-x86/io_apic.h Wed Oct 22 11:38:22 2008 +0900 83.2 +++ b/xen/include/asm-x86/io_apic.h Wed Oct 22 11:46:55 2008 +0900 83.3 @@ -20,6 +20,8 @@ 83.4 ((volatile int *)(__fix_to_virt(FIX_IO_APIC_BASE_0 + idx) \ 83.5 + (mp_ioapics[idx].mpc_apicaddr & ~PAGE_MASK))) 83.6 83.7 +#define IO_APIC_ID(idx) (mp_ioapics[idx].mpc_apicid) 83.8 + 83.9 /* 83.10 * The structure of the IO-APIC: 83.11 */
84.1 --- a/xen/include/asm-x86/msi.h Wed Oct 22 11:38:22 2008 +0900 84.2 +++ b/xen/include/asm-x86/msi.h Wed Oct 22 11:46:55 2008 +0900 84.3 @@ -90,10 +90,11 @@ struct msi_desc { 84.4 84.5 void __iomem *mask_base; 84.6 struct pci_dev *dev; 84.7 - int vector; 84.8 + int vector; 84.9 84.10 - /* Last set MSI message */ 84.11 - struct msi_msg msg; 84.12 + struct msi_msg msg; /* Last set MSI message */ 84.13 + 84.14 + int remap_index; /* index in interrupt remapping table */ 84.15 }; 84.16 84.17 /*
85.1 --- a/xen/include/asm-x86/rwlock.h Wed Oct 22 11:38:22 2008 +0900 85.2 +++ b/xen/include/asm-x86/rwlock.h Wed Oct 22 11:46:55 2008 +0900 85.3 @@ -22,25 +22,19 @@ 85.4 85.5 #define __build_read_lock_ptr(rw, helper) \ 85.6 asm volatile(LOCK "subl $1,(%0)\n\t" \ 85.7 - "js 2f\n" \ 85.8 + "jns 1f\n\t" \ 85.9 + "call " helper "\n\t" \ 85.10 "1:\n" \ 85.11 - ".section .text.lock,\"ax\"\n" \ 85.12 - "2:\tcall " helper "\n\t" \ 85.13 - "jmp 1b\n" \ 85.14 - ".previous" \ 85.15 ::"a" (rw) : "memory") 85.16 85.17 #define __build_read_lock_const(rw, helper) \ 85.18 asm volatile(LOCK "subl $1,%0\n\t" \ 85.19 - "js 2f\n" \ 85.20 - "1:\n" \ 85.21 - ".section .text.lock,\"ax\"\n" \ 85.22 - "2:\tpush %%"__OP"ax\n\t" \ 85.23 + "jns 1f\n\t" \ 85.24 + "push %%"__OP"ax\n\t" \ 85.25 "lea %0,%%"__OP"ax\n\t" \ 85.26 "call " helper "\n\t" \ 85.27 "pop %%"__OP"ax\n\t" \ 85.28 - "jmp 1b\n" \ 85.29 - ".previous" \ 85.30 + "1:\n" \ 85.31 :"=m" (*(volatile int *)rw) : : "memory") 85.32 85.33 #define __build_read_lock(rw, helper) do { \ 85.34 @@ -52,25 +46,19 @@ 85.35 85.36 #define __build_write_lock_ptr(rw, helper) \ 85.37 asm volatile(LOCK "subl $" RW_LOCK_BIAS_STR ",(%0)\n\t" \ 85.38 - "jnz 2f\n" \ 85.39 + "jz 1f\n\t" \ 85.40 + "call " helper "\n\t" \ 85.41 "1:\n" \ 85.42 - ".section .text.lock,\"ax\"\n" \ 85.43 - "2:\tcall " helper "\n\t" \ 85.44 - "jmp 1b\n" \ 85.45 - ".previous" \ 85.46 ::"a" (rw) : "memory") 85.47 85.48 #define __build_write_lock_const(rw, helper) \ 85.49 asm volatile(LOCK "subl $" RW_LOCK_BIAS_STR ",(%0)\n\t" \ 85.50 - "jnz 2f\n" \ 85.51 - "1:\n" \ 85.52 - ".section .text.lock,\"ax\"\n" \ 85.53 - "2:\tpush %%"__OP"ax\n\t" \ 85.54 + "jz 1f\n\t" \ 85.55 + "push %%"__OP"ax\n\t" \ 85.56 "lea %0,%%"__OP"ax\n\t" \ 85.57 "call " helper "\n\t" \ 85.58 "pop %%"__OP"ax\n\t" \ 85.59 - "jmp 1b\n" \ 85.60 - ".previous" \ 85.61 + "1:\n" \ 85.62 :"=m" (*(volatile int *)rw) : : "memory") 85.63 85.64 #define __build_write_lock(rw, helper) do { \
86.1 --- a/xen/include/asm-x86/spinlock.h Wed Oct 22 11:38:22 2008 +0900 86.2 +++ b/xen/include/asm-x86/spinlock.h Wed Oct 22 11:46:55 2008 +0900 86.3 @@ -8,104 +8,71 @@ 86.4 86.5 typedef struct { 86.6 volatile s16 lock; 86.7 - s8 recurse_cpu; 86.8 - u8 recurse_cnt; 86.9 -} spinlock_t; 86.10 +} raw_spinlock_t; 86.11 86.12 -#define SPIN_LOCK_UNLOCKED /*(spinlock_t)*/ { 1, -1, 0 } 86.13 +#define _RAW_SPIN_LOCK_UNLOCKED /*(raw_spinlock_t)*/ { 1 } 86.14 86.15 -#define spin_lock_init(x) do { *(x) = (spinlock_t) SPIN_LOCK_UNLOCKED; } while(0) 86.16 -#define spin_is_locked(x) (*(volatile char *)(&(x)->lock) <= 0) 86.17 +#define _raw_spin_is_locked(x) \ 86.18 + (*(volatile char *)(&(x)->lock) <= 0) 86.19 86.20 -static inline void _raw_spin_lock(spinlock_t *lock) 86.21 +static always_inline void _raw_spin_lock(raw_spinlock_t *lock) 86.22 { 86.23 - __asm__ __volatile__ ( 86.24 - "1: lock; decb %0 \n" 86.25 - " js 2f \n" 86.26 - ".section .text.lock,\"ax\"\n" 86.27 + asm volatile ( 86.28 + "1: lock; decw %0 \n" 86.29 + " jns 3f \n" 86.30 "2: rep; nop \n" 86.31 - " cmpb $0,%0 \n" 86.32 + " cmpw $0,%0 \n" 86.33 " jle 2b \n" 86.34 " jmp 1b \n" 86.35 - ".previous" 86.36 - : "=m" (lock->lock) : : "memory" ); 86.37 -} 86.38 - 86.39 -static inline void _raw_spin_unlock(spinlock_t *lock) 86.40 -{ 86.41 - ASSERT(spin_is_locked(lock)); 86.42 - __asm__ __volatile__ ( 86.43 - "movb $1,%0" 86.44 + "3:" 86.45 : "=m" (lock->lock) : : "memory" ); 86.46 } 86.47 86.48 -static inline int _raw_spin_trylock(spinlock_t *lock) 86.49 +static always_inline void _raw_spin_unlock(raw_spinlock_t *lock) 86.50 { 86.51 - char oldval; 86.52 - __asm__ __volatile__( 86.53 - "xchgb %b0,%1" 86.54 - :"=q" (oldval), "=m" (lock->lock) 86.55 - :"0" (0) : "memory"); 86.56 - return oldval > 0; 86.57 + ASSERT(_raw_spin_is_locked(lock)); 86.58 + asm volatile ( 86.59 + "movw $1,%0" 86.60 + : "=m" (lock->lock) : : "memory" ); 86.61 } 86.62 86.63 -/* 86.64 - * spin_[un]lock_recursive(): Use these forms when the lock can (safely!) be 86.65 - * reentered recursively on the same CPU. All critical regions that may form 86.66 - * part of a recursively-nested set must be protected by these forms. If there 86.67 - * are any critical regions that cannot form part of such a set, they can use 86.68 - * standard spin_[un]lock(). 86.69 - */ 86.70 -#define _raw_spin_lock_recursive(_lock) \ 86.71 - do { \ 86.72 - int cpu = smp_processor_id(); \ 86.73 - if ( likely((_lock)->recurse_cpu != cpu) ) \ 86.74 - { \ 86.75 - spin_lock(_lock); \ 86.76 - (_lock)->recurse_cpu = cpu; \ 86.77 - } \ 86.78 - (_lock)->recurse_cnt++; \ 86.79 - } while ( 0 ) 86.80 - 86.81 -#define _raw_spin_unlock_recursive(_lock) \ 86.82 - do { \ 86.83 - if ( likely(--(_lock)->recurse_cnt == 0) ) \ 86.84 - { \ 86.85 - (_lock)->recurse_cpu = -1; \ 86.86 - spin_unlock(_lock); \ 86.87 - } \ 86.88 - } while ( 0 ) 86.89 - 86.90 +static always_inline int _raw_spin_trylock(raw_spinlock_t *lock) 86.91 +{ 86.92 + s16 oldval; 86.93 + asm volatile ( 86.94 + "xchgw %w0,%1" 86.95 + :"=r" (oldval), "=m" (lock->lock) 86.96 + :"0" (0) : "memory" ); 86.97 + return (oldval > 0); 86.98 +} 86.99 86.100 typedef struct { 86.101 volatile unsigned int lock; 86.102 -} rwlock_t; 86.103 +} raw_rwlock_t; 86.104 86.105 -#define RW_LOCK_UNLOCKED /*(rwlock_t)*/ { RW_LOCK_BIAS } 86.106 - 86.107 -#define rwlock_init(x) do { *(x) = (rwlock_t) RW_LOCK_UNLOCKED; } while(0) 86.108 +#define _RAW_RW_LOCK_UNLOCKED /*(raw_rwlock_t)*/ { RW_LOCK_BIAS } 86.109 86.110 /* 86.111 * On x86, we implement read-write locks as a 32-bit counter 86.112 * with the high bit (sign) being the "contended" bit. 86.113 */ 86.114 -static inline void _raw_read_lock(rwlock_t *rw) 86.115 +static always_inline void _raw_read_lock(raw_rwlock_t *rw) 86.116 { 86.117 __build_read_lock(rw, "__read_lock_failed"); 86.118 } 86.119 86.120 -static inline void _raw_write_lock(rwlock_t *rw) 86.121 +static always_inline void _raw_write_lock(raw_rwlock_t *rw) 86.122 { 86.123 __build_write_lock(rw, "__write_lock_failed"); 86.124 } 86.125 86.126 -#define _raw_read_unlock(rw) \ 86.127 - __asm__ __volatile__ ( \ 86.128 - "lock ; incl %0" : \ 86.129 +#define _raw_read_unlock(rw) \ 86.130 + asm volatile ( \ 86.131 + "lock ; incl %0" : \ 86.132 "=m" ((rw)->lock) : : "memory" ) 86.133 -#define _raw_write_unlock(rw) \ 86.134 - __asm__ __volatile__ ( \ 86.135 - "lock ; addl $" RW_LOCK_BIAS_STR ",%0" : \ 86.136 +#define _raw_write_unlock(rw) \ 86.137 + asm volatile ( \ 86.138 + "lock ; addl $" RW_LOCK_BIAS_STR ",%0" : \ 86.139 "=m" ((rw)->lock) : : "memory" ) 86.140 86.141 #endif /* __ASM_SPINLOCK_H */
87.1 --- a/xen/include/public/domctl.h Wed Oct 22 11:38:22 2008 +0900 87.2 +++ b/xen/include/public/domctl.h Wed Oct 22 11:46:55 2008 +0900 87.3 @@ -614,6 +614,10 @@ DEFINE_XEN_GUEST_HANDLE(xen_domctl_subsc 87.4 #define XEN_DOMCTL_set_machine_address_size 51 87.5 #define XEN_DOMCTL_get_machine_address_size 52 87.6 87.7 +/* 87.8 + * Do not inject spurious page faults into this domain. 87.9 + */ 87.10 +#define XEN_DOMCTL_suppress_spurious_page_faults 53 87.11 87.12 struct xen_domctl { 87.13 uint32_t cmd;
88.1 --- a/xen/include/public/trace.h Wed Oct 22 11:38:22 2008 +0900 88.2 +++ b/xen/include/public/trace.h Wed Oct 22 11:46:55 2008 +0900 88.3 @@ -38,6 +38,7 @@ 88.4 #define TRC_MEM 0x0010f000 /* Xen memory trace */ 88.5 #define TRC_PV 0x0020f000 /* Xen PV traces */ 88.6 #define TRC_SHADOW 0x0040f000 /* Xen shadow tracing */ 88.7 +#define TRC_PM 0x0080f000 /* Xen power management trace */ 88.8 #define TRC_ALL 0x0ffff000 88.9 #define TRC_HD_TO_EVENT(x) ((x)&0x0fffffff) 88.10 #define TRC_HD_CYCLE_FLAG (1UL<<31) 88.11 @@ -146,6 +147,15 @@ 88.12 #define TRC_HVM_LMSW (TRC_HVM_HANDLER + 0x19) 88.13 #define TRC_HVM_LMSW64 (TRC_HVM_HANDLER + TRC_64_FLAG + 0x19) 88.14 88.15 +/* trace subclasses for power management */ 88.16 +#define TRC_PM_FREQ 0x00801000 /* xen cpu freq events */ 88.17 +#define TRC_PM_IDLE 0x00802000 /* xen cpu idle events */ 88.18 + 88.19 +/* trace events for per class */ 88.20 +#define TRC_PM_FREQ_CHANGE (TRC_PM_FREQ + 0x01) 88.21 +#define TRC_PM_IDLE_ENTRY (TRC_PM_IDLE + 0x01) 88.22 +#define TRC_PM_IDLE_EXIT (TRC_PM_IDLE + 0x02) 88.23 + 88.24 /* This structure represents a single trace buffer record. */ 88.25 struct t_rec { 88.26 uint32_t event:28;
89.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000 89.2 +++ b/xen/include/xen/hvm/irq.h Wed Oct 22 11:46:55 2008 +0900 89.3 @@ -0,0 +1,99 @@ 89.4 +/****************************************************************************** 89.5 + * irq.h 89.6 + * 89.7 + * Interrupt distribution and delivery logic. 89.8 + * 89.9 + * Copyright (c) 2006, K A Fraser, XenSource Inc. 89.10 + * 89.11 + * This program is free software; you can redistribute it and/or modify it 89.12 + * under the terms and conditions of the GNU General Public License, 89.13 + * version 2, as published by the Free Software Foundation. 89.14 + * 89.15 + * This program is distributed in the hope it will be useful, but WITHOUT 89.16 + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or 89.17 + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for 89.18 + * more details. 89.19 + * 89.20 + * You should have received a copy of the GNU General Public License along with 89.21 + * this program; if not, write to the Free Software Foundation, Inc., 59 Temple 89.22 + * Place - Suite 330, Boston, MA 02111-1307 USA. 89.23 + */ 89.24 + 89.25 +#ifndef __XEN_HVM_IRQ_H__ 89.26 +#define __XEN_HVM_IRQ_H__ 89.27 + 89.28 +#include <xen/types.h> 89.29 +#include <xen/spinlock.h> 89.30 +#include <asm/irq.h> 89.31 +#include <public/hvm/save.h> 89.32 + 89.33 +struct dev_intx_gsi_link { 89.34 + struct list_head list; 89.35 + uint8_t device; 89.36 + uint8_t intx; 89.37 + uint8_t gsi; 89.38 + uint8_t link; 89.39 +}; 89.40 + 89.41 +#define _HVM_IRQ_DPCI_MSI 0x1 89.42 + 89.43 +struct hvm_gmsi_info { 89.44 + uint32_t gvec; 89.45 + uint32_t gflags; 89.46 +}; 89.47 + 89.48 +struct hvm_mirq_dpci_mapping { 89.49 + uint32_t flags; 89.50 + int pending; 89.51 + struct list_head digl_list; 89.52 + struct domain *dom; 89.53 + struct hvm_gmsi_info gmsi; 89.54 +}; 89.55 + 89.56 +struct hvm_girq_dpci_mapping { 89.57 + uint8_t valid; 89.58 + uint8_t device; 89.59 + uint8_t intx; 89.60 + uint8_t machine_gsi; 89.61 +}; 89.62 + 89.63 +#define NR_ISAIRQS 16 89.64 +#define NR_LINK 4 89.65 + 89.66 +/* Protected by domain's event_lock */ 89.67 +struct hvm_irq_dpci { 89.68 + /* Machine IRQ to guest device/intx mapping. */ 89.69 + DECLARE_BITMAP(mapping, NR_PIRQS); 89.70 + struct hvm_mirq_dpci_mapping mirq[NR_IRQS]; 89.71 + /* Guest IRQ to guest device/intx mapping. */ 89.72 + struct hvm_girq_dpci_mapping girq[NR_IRQS]; 89.73 + uint8_t msi_gvec_pirq[NR_VECTORS]; 89.74 + DECLARE_BITMAP(dirq_mask, NR_IRQS); 89.75 + /* Record of mapped ISA IRQs */ 89.76 + DECLARE_BITMAP(isairq_map, NR_ISAIRQS); 89.77 + /* Record of mapped Links */ 89.78 + uint8_t link_cnt[NR_LINK]; 89.79 + struct timer hvm_timer[NR_IRQS]; 89.80 +}; 89.81 + 89.82 +/* Modify state of a PCI INTx wire. */ 89.83 +void hvm_pci_intx_assert( 89.84 + struct domain *d, unsigned int device, unsigned int intx); 89.85 +void hvm_pci_intx_deassert( 89.86 + struct domain *d, unsigned int device, unsigned int intx); 89.87 + 89.88 +/* Modify state of an ISA device's IRQ wire. */ 89.89 +void hvm_isa_irq_assert( 89.90 + struct domain *d, unsigned int isa_irq); 89.91 +void hvm_isa_irq_deassert( 89.92 + struct domain *d, unsigned int isa_irq); 89.93 + 89.94 +void hvm_set_pci_link_route(struct domain *d, u8 link, u8 isa_irq); 89.95 + 89.96 +void hvm_maybe_deassert_evtchn_irq(void); 89.97 +void hvm_assert_evtchn_irq(struct vcpu *v); 89.98 +void hvm_set_callback_via(struct domain *d, uint64_t via); 89.99 + 89.100 +void hvm_dirq_assist(struct vcpu *v); 89.101 + 89.102 +#endif /* __XEN_HVM_IRQ_H__ */
90.1 --- a/xen/include/xen/spinlock.h Wed Oct 22 11:38:22 2008 +0900 90.2 +++ b/xen/include/xen/spinlock.h Wed Oct 22 11:46:55 2008 +0900 90.3 @@ -3,93 +3,95 @@ 90.4 90.5 #include <xen/config.h> 90.6 #include <asm/system.h> 90.7 - 90.8 -#define spin_lock_irqsave(lock, flags) \ 90.9 - do { local_irq_save(flags); spin_lock(lock); } while ( 0 ) 90.10 -#define spin_lock_irq(lock) \ 90.11 - do { local_irq_disable(); spin_lock(lock); } while ( 0 ) 90.12 - 90.13 -#define read_lock_irqsave(lock, flags) \ 90.14 - do { local_irq_save(flags); read_lock(lock); } while ( 0 ) 90.15 -#define read_lock_irq(lock) \ 90.16 - do { local_irq_disable(); read_lock(lock); } while ( 0 ) 90.17 - 90.18 -#define write_lock_irqsave(lock, flags) \ 90.19 - do { local_irq_save(flags); write_lock(lock); } while ( 0 ) 90.20 -#define write_lock_irq(lock) \ 90.21 - do { local_irq_disable(); write_lock(lock); } while ( 0 ) 90.22 - 90.23 -#define spin_unlock_irqrestore(lock, flags) \ 90.24 - do { spin_unlock(lock); local_irq_restore(flags); } while ( 0 ) 90.25 -#define spin_unlock_irq(lock) \ 90.26 - do { spin_unlock(lock); local_irq_enable(); } while ( 0 ) 90.27 - 90.28 -#define read_unlock_irqrestore(lock, flags) \ 90.29 - do { read_unlock(lock); local_irq_restore(flags); } while ( 0 ) 90.30 -#define read_unlock_irq(lock) \ 90.31 - do { read_unlock(lock); local_irq_enable(); } while ( 0 ) 90.32 - 90.33 -#define write_unlock_irqrestore(lock, flags) \ 90.34 - do { write_unlock(lock); local_irq_restore(flags); } while ( 0 ) 90.35 -#define write_unlock_irq(lock) \ 90.36 - do { write_unlock(lock); local_irq_enable(); } while ( 0 ) 90.37 - 90.38 -#ifdef CONFIG_SMP 90.39 - 90.40 #include <asm/spinlock.h> 90.41 90.42 -#else 90.43 +typedef struct { 90.44 + raw_spinlock_t raw; 90.45 + u16 recurse_cpu:12; 90.46 + u16 recurse_cnt:4; 90.47 +} spinlock_t; 90.48 + 90.49 +#define SPIN_LOCK_UNLOCKED { _RAW_SPIN_LOCK_UNLOCKED, 0xfffu, 0 } 90.50 +#define DEFINE_SPINLOCK(l) spinlock_t l = SPIN_LOCK_UNLOCKED 90.51 +#define spin_lock_init(l) (*(l) = (spinlock_t)SPIN_LOCK_UNLOCKED) 90.52 + 90.53 +typedef struct { 90.54 + raw_rwlock_t raw; 90.55 +} rwlock_t; 90.56 90.57 -#if (__GNUC__ > 2) 90.58 -typedef struct { } spinlock_t; 90.59 -#define SPIN_LOCK_UNLOCKED /*(spinlock_t)*/ { } 90.60 -#else 90.61 -typedef struct { int gcc_is_buggy; } spinlock_t; 90.62 -#define SPIN_LOCK_UNLOCKED /*(spinlock_t)*/ { 0 } 90.63 -#endif 90.64 +#define RW_LOCK_UNLOCKED { _RAW_RW_LOCK_UNLOCKED } 90.65 +#define DEFINE_RWLOCK(l) rwlock_t l = RW_LOCK_UNLOCKED 90.66 +#define rwlock_init(l) (*(l) = (rwlock_t)RW_LOCK_UNLOCKED) 90.67 + 90.68 +void _spin_lock(spinlock_t *lock); 90.69 +void _spin_lock_irq(spinlock_t *lock); 90.70 +unsigned long _spin_lock_irqsave(spinlock_t *lock); 90.71 90.72 -#define spin_lock_init(lock) do { } while(0) 90.73 -#define spin_is_locked(lock) (0) 90.74 -#define _raw_spin_lock(lock) (void)(lock) 90.75 -#define _raw_spin_trylock(lock) ({1; }) 90.76 -#define _raw_spin_unlock(lock) do { } while(0) 90.77 -#define _raw_spin_lock_recursive(lock) do { } while(0) 90.78 -#define _raw_spin_unlock_recursive(lock) do { } while(0) 90.79 +void _spin_unlock(spinlock_t *lock); 90.80 +void _spin_unlock_irq(spinlock_t *lock); 90.81 +void _spin_unlock_irqrestore(spinlock_t *lock, unsigned long flags); 90.82 + 90.83 +int _spin_is_locked(spinlock_t *lock); 90.84 +int _spin_trylock(spinlock_t *lock); 90.85 +void _spin_barrier(spinlock_t *lock); 90.86 90.87 -#if (__GNUC__ > 2) 90.88 -typedef struct { } rwlock_t; 90.89 -#define RW_LOCK_UNLOCKED /*(rwlock_t)*/ { } 90.90 -#else 90.91 -typedef struct { int gcc_is_buggy; } rwlock_t; 90.92 -#define RW_LOCK_UNLOCKED /*(rwlock_t)*/ { 0 } 90.93 -#endif 90.94 +void _spin_lock_recursive(spinlock_t *lock); 90.95 +void _spin_unlock_recursive(spinlock_t *lock); 90.96 + 90.97 +void _read_lock(rwlock_t *lock); 90.98 +void _read_lock_irq(rwlock_t *lock); 90.99 +unsigned long _read_lock_irqsave(rwlock_t *lock); 90.100 + 90.101 +void _read_unlock(rwlock_t *lock); 90.102 +void _read_unlock_irq(rwlock_t *lock); 90.103 +void _read_unlock_irqrestore(rwlock_t *lock, unsigned long flags); 90.104 + 90.105 +void _write_lock(rwlock_t *lock); 90.106 +void _write_lock_irq(rwlock_t *lock); 90.107 +unsigned long _write_lock_irqsave(rwlock_t *lock); 90.108 90.109 -#define rwlock_init(lock) do { } while(0) 90.110 -#define _raw_read_lock(lock) (void)(lock) /* Not "unused variable". */ 90.111 -#define _raw_read_unlock(lock) do { } while(0) 90.112 -#define _raw_write_lock(lock) (void)(lock) /* Not "unused variable". */ 90.113 -#define _raw_write_unlock(lock) do { } while(0) 90.114 - 90.115 -#endif 90.116 +void _write_unlock(rwlock_t *lock); 90.117 +void _write_unlock_irq(rwlock_t *lock); 90.118 +void _write_unlock_irqrestore(rwlock_t *lock, unsigned long flags); 90.119 90.120 -#define spin_lock(_lock) _raw_spin_lock(_lock) 90.121 -#define spin_trylock(_lock) _raw_spin_trylock(_lock) 90.122 -#define spin_unlock(_lock) _raw_spin_unlock(_lock) 90.123 -#define spin_lock_recursive(_lock) _raw_spin_lock_recursive(_lock) 90.124 -#define spin_unlock_recursive(_lock) _raw_spin_unlock_recursive(_lock) 90.125 -#define read_lock(_lock) _raw_read_lock(_lock) 90.126 -#define read_unlock(_lock) _raw_read_unlock(_lock) 90.127 -#define write_lock(_lock) _raw_write_lock(_lock) 90.128 -#define write_unlock(_lock) _raw_write_unlock(_lock) 90.129 +#define spin_lock(l) _spin_lock(l) 90.130 +#define spin_lock_irq(l) _spin_lock_irq(l) 90.131 +#define spin_lock_irqsave(l, f) ((f) = _spin_lock_irqsave(l)) 90.132 + 90.133 +#define spin_unlock(l) _spin_unlock(l) 90.134 +#define spin_unlock_irq(l) _spin_unlock_irq(l) 90.135 +#define spin_unlock_irqrestore(l, f) _spin_unlock_irqrestore(l, f) 90.136 + 90.137 +#define spin_is_locked(l) _raw_spin_is_locked(&(l)->raw) 90.138 +#define spin_trylock(l) _spin_trylock(l) 90.139 90.140 /* Ensure a lock is quiescent between two critical operations. */ 90.141 -static inline void spin_barrier(spinlock_t *lock) 90.142 -{ 90.143 - do { mb(); } while ( spin_is_locked(lock) ); 90.144 - mb(); 90.145 -} 90.146 +#define spin_barrier(l) _spin_barrier(l) 90.147 + 90.148 +/* 90.149 + * spin_[un]lock_recursive(): Use these forms when the lock can (safely!) be 90.150 + * reentered recursively on the same CPU. All critical regions that may form 90.151 + * part of a recursively-nested set must be protected by these forms. If there 90.152 + * are any critical regions that cannot form part of such a set, they can use 90.153 + * standard spin_[un]lock(). 90.154 + */ 90.155 +#define spin_lock_recursive(l) _spin_lock_recursive(l) 90.156 +#define spin_unlock_recursive(l) _spin_unlock_recursive(l) 90.157 90.158 -#define DEFINE_SPINLOCK(x) spinlock_t x = SPIN_LOCK_UNLOCKED 90.159 -#define DEFINE_RWLOCK(x) rwlock_t x = RW_LOCK_UNLOCKED 90.160 +#define read_lock(l) _read_lock(l) 90.161 +#define read_lock_irq(l) _read_lock_irq(l) 90.162 +#define read_lock_irqsave(l, f) ((f) = _read_lock_irqsave(l)) 90.163 + 90.164 +#define read_unlock(l) _read_unlock(l) 90.165 +#define read_unlock_irq(l) _read_unlock_irq(l) 90.166 +#define read_unlock_irqrestore(l, f) _read_unlock_irqrestore(l, f) 90.167 + 90.168 +#define write_lock(l) _write_lock(l) 90.169 +#define write_lock_irq(l) _write_lock_irq(l) 90.170 +#define write_lock_irqsave(l, f) ((f) = _write_lock_irqsave(l)) 90.171 + 90.172 +#define write_unlock(l) _write_unlock(l) 90.173 +#define write_unlock_irq(l) _write_unlock_irq(l) 90.174 +#define write_unlock_irqrestore(l, f) _write_unlock_irqrestore(l, f) 90.175 90.176 #endif /* __SPINLOCK_H__ */
91.1 --- a/xen/include/xen/xmalloc.h Wed Oct 22 11:38:22 2008 +0900 91.2 +++ b/xen/include/xen/xmalloc.h Wed Oct 22 11:46:55 2008 +0900 91.3 @@ -2,11 +2,16 @@ 91.4 #ifndef __XMALLOC_H__ 91.5 #define __XMALLOC_H__ 91.6 91.7 +/* 91.8 + * Xen malloc/free-style interface. 91.9 + */ 91.10 + 91.11 /* Allocate space for typed object. */ 91.12 #define xmalloc(_type) ((_type *)_xmalloc(sizeof(_type), __alignof__(_type))) 91.13 91.14 /* Allocate space for array of typed objects. */ 91.15 -#define xmalloc_array(_type, _num) ((_type *)_xmalloc_array(sizeof(_type), __alignof__(_type), _num)) 91.16 +#define xmalloc_array(_type, _num) \ 91.17 + ((_type *)_xmalloc_array(sizeof(_type), __alignof__(_type), _num)) 91.18 91.19 /* Allocate untyped storage. */ 91.20 #define xmalloc_bytes(_bytes) (_xmalloc(_bytes, SMP_CACHE_BYTES)) 91.21 @@ -15,8 +20,9 @@ 91.22 extern void xfree(void *); 91.23 91.24 /* Underlying functions */ 91.25 -extern void *_xmalloc(size_t size, size_t align); 91.26 -static inline void *_xmalloc_array(size_t size, size_t align, size_t num) 91.27 +extern void *_xmalloc(unsigned long size, unsigned long align); 91.28 +static inline void *_xmalloc_array( 91.29 + unsigned long size, unsigned long align, unsigned long num) 91.30 { 91.31 /* Check for overflow. */ 91.32 if (size && num > UINT_MAX / size) 91.33 @@ -24,4 +30,73 @@ static inline void *_xmalloc_array(size_ 91.34 return _xmalloc(size * num, align); 91.35 } 91.36 91.37 +/* 91.38 + * Pooled allocator interface. 91.39 + */ 91.40 + 91.41 +struct xmem_pool; 91.42 + 91.43 +typedef void *(xmem_pool_get_memory)(unsigned long bytes); 91.44 +typedef void (xmem_pool_put_memory)(void *ptr); 91.45 + 91.46 +/** 91.47 + * xmem_pool_create - create dynamic memory pool 91.48 + * @name: name of the pool 91.49 + * @get_mem: callback function used to expand pool 91.50 + * @put_mem: callback function used to shrink pool 91.51 + * @init_size: inital pool size (in bytes) 91.52 + * @max_size: maximum pool size (in bytes) - set this as 0 for no limit 91.53 + * @grow_size: amount of memory (in bytes) added to pool whenever required 91.54 + * 91.55 + * All size values are rounded up to next page boundary. 91.56 + */ 91.57 +struct xmem_pool *xmem_pool_create( 91.58 + const char *name, 91.59 + xmem_pool_get_memory get_mem, 91.60 + xmem_pool_put_memory put_mem, 91.61 + unsigned long init_size, 91.62 + unsigned long max_size, 91.63 + unsigned long grow_size); 91.64 + 91.65 +/** 91.66 + * xmem_pool_destroy - cleanup given pool 91.67 + * @mem_pool: Pool to be destroyed 91.68 + * 91.69 + * Data structures associated with pool are freed. 91.70 + * All memory allocated from pool must be freed before 91.71 + * destorying it. 91.72 + */ 91.73 +void xmem_pool_destroy(struct xmem_pool *pool); 91.74 + 91.75 +/** 91.76 + * xmem_pool_alloc - allocate memory from given pool 91.77 + * @size: no. of bytes 91.78 + * @mem_pool: pool to allocate from 91.79 + */ 91.80 +void *xmem_pool_alloc(unsigned long size, struct xmem_pool *pool); 91.81 + 91.82 +/** 91.83 + * xmem_pool_free - free memory from given pool 91.84 + * @ptr: address of memory to be freed 91.85 + * @mem_pool: pool to free from 91.86 + */ 91.87 +void xmem_pool_free(void *ptr, struct xmem_pool *pool); 91.88 + 91.89 +/** 91.90 + * xmem_pool_get_used_size - get memory currently used by given pool 91.91 + * 91.92 + * Used memory includes stored data + metadata + internal fragmentation 91.93 + */ 91.94 +unsigned long xmem_pool_get_used_size(struct xmem_pool *pool); 91.95 + 91.96 +/** 91.97 + * xmem_pool_get_total_size - get total memory currently allocated for pool 91.98 + * 91.99 + * This is the total memory currently allocated for this pool which includes 91.100 + * used size + free size. 91.101 + * 91.102 + * (Total - Used) is good indicator of memory efficiency of allocator. 91.103 + */ 91.104 +unsigned long xmem_pool_get_total_size(struct xmem_pool *pool); 91.105 + 91.106 #endif /* __XMALLOC_H__ */
92.1 --- a/xen/include/xlat.lst Wed Oct 22 11:38:22 2008 +0900 92.2 +++ b/xen/include/xlat.lst Wed Oct 22 11:46:55 2008 +0900 92.3 @@ -55,3 +55,7 @@ 92.4 ! processor_cx platform.h 92.5 ! processor_flags platform.h 92.6 ! processor_power platform.h 92.7 +! pct_register platform.h 92.8 +! processor_px platform.h 92.9 +! psd_package platform.h 92.10 +! processor_performance platform.h