| 1 | # SPDX-License-Identifier: GPL-2.0-only |
| 2 | menu "Xen driver support" |
| 3 | depends on XEN |
| 4 | |
| 5 | config XEN_BALLOON |
| 6 | bool "Xen memory balloon driver" |
| 7 | default y |
| 8 | help |
| 9 | The balloon driver allows the Xen domain to request more memory from |
| 10 | the system to expand the domain's memory allocation, or alternatively |
| 11 | return unneeded memory to the system. |
| 12 | |
| 13 | config XEN_BALLOON_MEMORY_HOTPLUG |
| 14 | bool "Memory hotplug support for Xen balloon driver" |
| 15 | depends on XEN_BALLOON && MEMORY_HOTPLUG |
| 16 | default y |
| 17 | help |
| 18 | Memory hotplug support for Xen balloon driver allows expanding memory |
| 19 | available for the system above limit declared at system startup. |
| 20 | It is very useful on critical systems which require long |
| 21 | run without rebooting. |
| 22 | |
| 23 | It's also very useful for non PV domains to obtain unpopulated physical |
| 24 | memory ranges to use in order to map foreign memory or grants. |
| 25 | |
| 26 | Memory could be hotplugged in following steps: |
| 27 | |
| 28 | 1) target domain: ensure that memory auto online policy is in |
| 29 | effect by checking /sys/devices/system/memory/auto_online_blocks |
| 30 | file (should be 'online'). |
| 31 | |
| 32 | 2) control domain: xl mem-max <target-domain> <maxmem> |
| 33 | where <maxmem> is >= requested memory size, |
| 34 | |
| 35 | 3) control domain: xl mem-set <target-domain> <memory> |
| 36 | where <memory> is requested memory size; alternatively memory |
| 37 | could be added by writing proper value to |
| 38 | /sys/devices/system/xen_memory/xen_memory0/target or |
| 39 | /sys/devices/system/xen_memory/xen_memory0/target_kb on the |
| 40 | target domain. |
| 41 | |
| 42 | Alternatively, if memory auto onlining was not requested at step 1 |
| 43 | the newly added memory can be manually onlined in the target domain |
| 44 | by doing the following: |
| 45 | |
| 46 | for i in /sys/devices/system/memory/memory*/state; do \ |
| 47 | [ "`cat "$i"`" = offline ] && echo online > "$i"; done |
| 48 | |
| 49 | or by adding the following line to udev rules: |
| 50 | |
| 51 | SUBSYSTEM=="memory", ACTION=="add", RUN+="/bin/sh -c '[ -f /sys$devpath/state ] && echo online > /sys$devpath/state'" |
| 52 | |
| 53 | config XEN_MEMORY_HOTPLUG_LIMIT |
| 54 | int "Hotplugged memory limit (in GiB) for a PV guest" |
| 55 | default 512 |
| 56 | depends on XEN_HAVE_PVMMU |
| 57 | depends on MEMORY_HOTPLUG |
| 58 | help |
| 59 | Maximum amount of memory (in GiB) that a PV guest can be |
| 60 | expanded to when using memory hotplug. |
| 61 | |
| 62 | A PV guest can have more memory than this limit if is |
| 63 | started with a larger maximum. |
| 64 | |
| 65 | This value is used to allocate enough space in internal |
| 66 | tables needed for physical memory administration. |
| 67 | |
| 68 | config XEN_SCRUB_PAGES_DEFAULT |
| 69 | bool "Scrub pages before returning them to system by default" |
| 70 | depends on XEN_BALLOON |
| 71 | default y |
| 72 | help |
| 73 | Scrub pages before returning them to the system for reuse by |
| 74 | other domains. This makes sure that any confidential data |
| 75 | is not accidentally visible to other domains. It is more |
| 76 | secure, but slightly less efficient. This can be controlled with |
| 77 | xen_scrub_pages=0 parameter and |
| 78 | /sys/devices/system/xen_memory/xen_memory0/scrub_pages. |
| 79 | This option only sets the default value. |
| 80 | |
| 81 | If in doubt, say yes. |
| 82 | |
| 83 | config XEN_DEV_EVTCHN |
| 84 | tristate "Xen /dev/xen/evtchn device" |
| 85 | default y |
| 86 | help |
| 87 | The evtchn driver allows a userspace process to trigger event |
| 88 | channels and to receive notification of an event channel |
| 89 | firing. |
| 90 | If in doubt, say yes. |
| 91 | |
| 92 | config XEN_BACKEND |
| 93 | bool "Backend driver support" |
| 94 | default XEN_DOM0 |
| 95 | help |
| 96 | Support for backend device drivers that provide I/O services |
| 97 | to other virtual machines. |
| 98 | |
| 99 | config XENFS |
| 100 | tristate "Xen filesystem" |
| 101 | select XEN_PRIVCMD |
| 102 | default y |
| 103 | help |
| 104 | The xen filesystem provides a way for domains to share |
| 105 | information with each other and with the hypervisor. |
| 106 | For example, by reading and writing the "xenbus" file, guests |
| 107 | may pass arbitrary information to the initial domain. |
| 108 | If in doubt, say yes. |
| 109 | |
| 110 | config XEN_COMPAT_XENFS |
| 111 | bool "Create compatibility mount point /proc/xen" |
| 112 | depends on XENFS |
| 113 | default y |
| 114 | help |
| 115 | The old xenstore userspace tools expect to find "xenbus" |
| 116 | under /proc/xen, but "xenbus" is now found at the root of the |
| 117 | xenfs filesystem. Selecting this causes the kernel to create |
| 118 | the compatibility mount point /proc/xen if it is running on |
| 119 | a xen platform. |
| 120 | If in doubt, say yes. |
| 121 | |
| 122 | config XEN_SYS_HYPERVISOR |
| 123 | bool "Create xen entries under /sys/hypervisor" |
| 124 | depends on SYSFS |
| 125 | select SYS_HYPERVISOR |
| 126 | default y |
| 127 | help |
| 128 | Create entries under /sys/hypervisor describing the Xen |
| 129 | hypervisor environment. When running native or in another |
| 130 | virtual environment, /sys/hypervisor will still be present, |
| 131 | but will have no xen contents. |
| 132 | |
| 133 | config XEN_XENBUS_FRONTEND |
| 134 | tristate |
| 135 | |
| 136 | config XEN_GNTDEV |
| 137 | tristate "userspace grant access device driver" |
| 138 | depends on XEN |
| 139 | default m |
| 140 | select MMU_NOTIFIER |
| 141 | select FIND_NORMAL_PAGE |
| 142 | help |
| 143 | Allows userspace processes to use grants. |
| 144 | |
| 145 | config XEN_GNTDEV_DMABUF |
| 146 | bool "Add support for dma-buf grant access device driver extension" |
| 147 | depends on XEN_GNTDEV && XEN_GRANT_DMA_ALLOC |
| 148 | select DMA_SHARED_BUFFER |
| 149 | help |
| 150 | Allows userspace processes and kernel modules to use Xen backed |
| 151 | dma-buf implementation. With this extension grant references to |
| 152 | the pages of an imported dma-buf can be exported for other domain |
| 153 | use and grant references coming from a foreign domain can be |
| 154 | converted into a local dma-buf for local export. |
| 155 | |
| 156 | config XEN_GRANT_DEV_ALLOC |
| 157 | tristate "User-space grant reference allocator driver" |
| 158 | depends on XEN |
| 159 | default m |
| 160 | help |
| 161 | Allows userspace processes to create pages with access granted |
| 162 | to other domains. This can be used to implement frontend drivers |
| 163 | or as part of an inter-domain shared memory channel. |
| 164 | |
| 165 | config XEN_GRANT_DMA_ALLOC |
| 166 | bool "Allow allocating DMA capable buffers with grant reference module" |
| 167 | depends on XEN && HAS_DMA |
| 168 | help |
| 169 | Extends grant table module API to allow allocating DMA capable |
| 170 | buffers and mapping foreign grant references on top of it. |
| 171 | The resulting buffer is similar to one allocated by the balloon |
| 172 | driver in that proper memory reservation is made by |
| 173 | ({increase|decrease}_reservation and VA mappings are updated if |
| 174 | needed). |
| 175 | This is useful for sharing foreign buffers with HW drivers which |
| 176 | cannot work with scattered buffers provided by the balloon driver, |
| 177 | but require DMAable memory instead. |
| 178 | |
| 179 | config SWIOTLB_XEN |
| 180 | def_bool y |
| 181 | depends on ARCH_HAS_DMA_OPS |
| 182 | depends on XEN_PV || ARM || ARM64 |
| 183 | select SWIOTLB |
| 184 | |
| 185 | config XEN_PCI_STUB |
| 186 | bool |
| 187 | |
| 188 | config XEN_PCIDEV_STUB |
| 189 | tristate "Xen PCI-device stub driver" |
| 190 | depends on PCI && !X86 && XEN |
| 191 | depends on XEN_BACKEND |
| 192 | select XEN_PCI_STUB |
| 193 | default m |
| 194 | help |
| 195 | The PCI device stub driver provides limited version of the PCI |
| 196 | device backend driver without para-virtualized support for guests. |
| 197 | If you select this to be a module, you will need to make sure no |
| 198 | other driver has bound to the device(s) you want to make visible to |
| 199 | other guests. |
| 200 | |
| 201 | The "hide" parameter (only applicable if backend driver is compiled |
| 202 | into the kernel) allows you to bind the PCI devices to this module |
| 203 | from the default device drivers. The argument is the list of PCI BDFs: |
| 204 | xen-pciback.hide=(03:00.0)(04:00.0) |
| 205 | |
| 206 | If in doubt, say m. |
| 207 | |
| 208 | config XEN_PCIDEV_BACKEND |
| 209 | tristate "Xen PCI-device backend driver" |
| 210 | depends on PCI && X86 && XEN |
| 211 | depends on XEN_BACKEND |
| 212 | select XEN_PCI_STUB |
| 213 | default m |
| 214 | help |
| 215 | The PCI device backend driver allows the kernel to export arbitrary |
| 216 | PCI devices to other guests. If you select this to be a module, you |
| 217 | will need to make sure no other driver has bound to the device(s) |
| 218 | you want to make visible to other guests. |
| 219 | |
| 220 | The parameter "passthrough" allows you specify how you want the PCI |
| 221 | devices to appear in the guest. You can choose the default (0) where |
| 222 | PCI topology starts at 00.00.0, or (1) for passthrough if you want |
| 223 | the PCI devices topology appear the same as in the host. |
| 224 | |
| 225 | The "hide" parameter (only applicable if backend driver is compiled |
| 226 | into the kernel) allows you to bind the PCI devices to this module |
| 227 | from the default device drivers. The argument is the list of PCI BDFs: |
| 228 | xen-pciback.hide=(03:00.0)(04:00.0) |
| 229 | |
| 230 | If in doubt, say m. |
| 231 | |
| 232 | config XEN_PVCALLS_FRONTEND |
| 233 | tristate "XEN PV Calls frontend driver" |
| 234 | depends on INET && XEN |
| 235 | select XEN_XENBUS_FRONTEND |
| 236 | help |
| 237 | Experimental frontend for the Xen PV Calls protocol |
| 238 | (https://xenbits.xen.org/docs/unstable/misc/pvcalls.html). It |
| 239 | sends a small set of POSIX calls to the backend, which |
| 240 | implements them. |
| 241 | |
| 242 | config XEN_PVCALLS_BACKEND |
| 243 | tristate "XEN PV Calls backend driver" |
| 244 | depends on INET && XEN && XEN_BACKEND |
| 245 | help |
| 246 | Experimental backend for the Xen PV Calls protocol |
| 247 | (https://xenbits.xen.org/docs/unstable/misc/pvcalls.html). It |
| 248 | allows PV Calls frontends to send POSIX calls to the backend, |
| 249 | which implements them. |
| 250 | |
| 251 | If in doubt, say n. |
| 252 | |
| 253 | config XEN_SCSI_BACKEND |
| 254 | tristate "XEN SCSI backend driver" |
| 255 | depends on XEN && XEN_BACKEND && TARGET_CORE |
| 256 | help |
| 257 | The SCSI backend driver allows the kernel to export its SCSI Devices |
| 258 | to other guests via a high-performance shared-memory interface. |
| 259 | Only needed for systems running as XEN driver domains (e.g. Dom0) and |
| 260 | if guests need generic access to SCSI devices. |
| 261 | |
| 262 | config XEN_PRIVCMD |
| 263 | tristate "Xen hypercall passthrough driver" |
| 264 | depends on XEN |
| 265 | default m |
| 266 | help |
| 267 | The hypercall passthrough driver allows privileged user programs to |
| 268 | perform Xen hypercalls. This driver is normally required for systems |
| 269 | running as Dom0 to perform privileged operations, but in some |
| 270 | disaggregated Xen setups this driver might be needed for other |
| 271 | domains, too. |
| 272 | |
| 273 | config XEN_PRIVCMD_EVENTFD |
| 274 | bool "Xen Ioeventfd and irqfd support" |
| 275 | depends on XEN_PRIVCMD && XEN_VIRTIO && EVENTFD |
| 276 | help |
| 277 | Using the ioeventfd / irqfd mechanism a virtio backend running in a |
| 278 | daemon can speed up interrupt delivery from / to a guest. |
| 279 | |
| 280 | config XEN_ACPI_PROCESSOR |
| 281 | tristate "Xen ACPI processor" |
| 282 | depends on XEN && XEN_DOM0 && X86 && ACPI_PROCESSOR && CPU_FREQ |
| 283 | default m |
| 284 | help |
| 285 | This ACPI processor uploads Power Management information to the Xen |
| 286 | hypervisor. |
| 287 | |
| 288 | To do that the driver parses the Power Management data and uploads |
| 289 | said information to the Xen hypervisor. Then the Xen hypervisor can |
| 290 | select the proper Cx and Pxx states. It also registers itself as the |
| 291 | SMM so that other drivers (such as ACPI cpufreq scaling driver) will |
| 292 | not load. |
| 293 | |
| 294 | To compile this driver as a module, choose M here: the module will be |
| 295 | called xen_acpi_processor If you do not know what to choose, select |
| 296 | M here. If the CPUFREQ drivers are built in, select Y here. |
| 297 | |
| 298 | config XEN_MCE_LOG |
| 299 | bool "Xen platform mcelog" |
| 300 | depends on XEN_PV_DOM0 && X86_MCE |
| 301 | help |
| 302 | Allow kernel fetching MCE error from Xen platform and |
| 303 | converting it into Linux mcelog format for mcelog tools |
| 304 | |
| 305 | config XEN_HAVE_PVMMU |
| 306 | bool |
| 307 | |
| 308 | config XEN_EFI |
| 309 | def_bool y |
| 310 | depends on (ARM || ARM64 || X86_64) && EFI |
| 311 | |
| 312 | config XEN_ACPI |
| 313 | def_bool y |
| 314 | depends on X86 && ACPI |
| 315 | |
| 316 | config XEN_SYMS |
| 317 | bool "Xen symbols" |
| 318 | depends on X86 && XEN_DOM0 && XENFS |
| 319 | default y if KALLSYMS |
| 320 | help |
| 321 | Exports hypervisor symbols (along with their types and addresses) via |
| 322 | /proc/xen/xensyms file, similar to /proc/kallsyms |
| 323 | |
| 324 | config XEN_HAVE_VPMU |
| 325 | bool |
| 326 | |
| 327 | config XEN_FRONT_PGDIR_SHBUF |
| 328 | tristate |
| 329 | |
| 330 | config XEN_UNPOPULATED_ALLOC |
| 331 | bool "Use unpopulated memory ranges for guest mappings" |
| 332 | depends on ZONE_DEVICE |
| 333 | default XEN_BACKEND || XEN_GNTDEV || XEN_DOM0 |
| 334 | help |
| 335 | Use unpopulated memory ranges in order to create mappings for guest |
| 336 | memory regions, including grant maps and foreign pages. This avoids |
| 337 | having to balloon out RAM regions in order to obtain physical memory |
| 338 | space to create such mappings. |
| 339 | |
| 340 | config XEN_GRANT_DMA_IOMMU |
| 341 | bool |
| 342 | select IOMMU_API |
| 343 | |
| 344 | config XEN_GRANT_DMA_OPS |
| 345 | bool |
| 346 | |
| 347 | config XEN_VIRTIO |
| 348 | bool "Xen virtio support" |
| 349 | depends on ARCH_HAS_DMA_OPS |
| 350 | depends on VIRTIO |
| 351 | select XEN_GRANT_DMA_OPS |
| 352 | select XEN_GRANT_DMA_IOMMU if OF |
| 353 | help |
| 354 | Enable virtio support for running as Xen guest. Depending on the |
| 355 | guest type this will require special support on the backend side |
| 356 | (qemu or kernel, depending on the virtio device types used). |
| 357 | |
| 358 | If in doubt, say n. |
| 359 | |
| 360 | config XEN_VIRTIO_FORCE_GRANT |
| 361 | bool "Require Xen virtio support to use grants" |
| 362 | depends on XEN_VIRTIO |
| 363 | help |
| 364 | Require virtio for Xen guests to use grant mappings. |
| 365 | This will avoid the need to give the backend the right to map all |
| 366 | of the guest memory. This will need support on the backend side |
| 367 | (e.g. qemu or kernel, depending on the virtio device types used). |
| 368 | |
| 369 | endmenu |