1 XIVE for sPAPR (pseries machines)
2 =================================
4 The POWER9 processor comes with a new interrupt controller
5 architecture, called XIVE as "eXternal Interrupt Virtualization
6 Engine". It supports a larger number of interrupt sources and offers
7 virtualization features which enables the HW to deliver interrupts
8 directly to virtual processors without hypervisor assistance.
10 A QEMU ``pseries`` machine (which is PAPR compliant) using POWER9
11 processors can run under two interrupt modes:
13 - *Legacy Compatibility Mode*
15 the hypervisor provides identical interfaces and similar
16 functionality to PAPR+ Version 2.7. This is the default mode
18 It is also referred as *XICS* in QEMU.
20 - *XIVE native exploitation mode*
22 the hypervisor provides new interfaces to manage the XIVE control
23 structures, and provides direct control for interrupt management
26 Which interrupt modes can be used by the machine is negotiated with
27 the guest O/S during the Client Architecture Support negotiation
28 sequence. The two modes are mutually exclusive.
30 Both interrupt mode share the same IRQ number space. See below for the
36 QEMU advertises the supported interrupt modes in the device tree
37 property ``ibm,arch-vec-5-platform-support`` in byte 23 and the OS
38 Selection for XIVE is indicated in the ``ibm,architecture-vec-5``
41 The interrupt modes supported by the machine depend on the CPU type
42 (POWER9 is required for XIVE) but also on the machine property
43 ``ic-mode`` which can be set on the command line. It can take the
44 following values: ``xics``, ``xive``, and ``dual`` which is the
45 default mode. ``dual`` means that both modes XICS **and** XIVE are
46 supported and if the guest OS supports XIVE, this mode will be
49 The choosen interrupt mode is activated after a reconfiguration done
55 When the guest starts under KVM, the capabilities of the host kernel
56 and QEMU are also negotiated. Depending on the version of the host
57 kernel, KVM will advertise the XIVE capability to QEMU or not.
59 Nevertheless, the available interrupt modes in the machine should not
60 depend on the XIVE KVM capability of the host. On older kernels
61 without XIVE KVM support, QEMU will use the emulated XIVE device as a
62 fallback and on newer kernels (>=5.2), the KVM XIVE device.
64 As a final refinement, the user can also switch the use of the KVM
65 device with the machine option ``kernel_irqchip``.
71 For guest OSes supporting XIVE, the resulting interrupt modes on host
72 kernels with XIVE KVM support are the following:
74 ============== ============= ============= ================
75 ic-mode kernel_irqchip
76 -------------- ----------------------------------------------
79 ============== ============= ============= ================
80 dual (default) XIVE KVM XIVE emul. XIVE KVM
81 xive XIVE KVM XIVE emul. XIVE KVM
82 xics XICS KVM XICS emul. XICS KVM
83 ============== ============= ============= ================
85 For legacy guest OSes without XIVE support, the resulting interrupt
86 modes are the following:
88 ============== ============= ============= ================
89 ic-mode kernel_irqchip
90 -------------- ----------------------------------------------
93 ============== ============= ============= ================
94 dual (default) XICS KVM XICS emul. XICS KVM
95 xive QEMU error(3) QEMU error(3) QEMU error(3)
96 xics XICS KVM XICS emul. XICS KVM
97 ============== ============= ============= ================
99 (3) QEMU fails at CAS with ``Guest requested unavailable interrupt
100 mode (XICS), either don't set the ic-mode machine property or try
101 ic-mode=xics or ic-mode=dual``
104 No XIVE support in KVM
105 ~~~~~~~~~~~~~~~~~~~~~~
107 For guest OSes supporting XIVE, the resulting interrupt modes on host
108 kernels without XIVE KVM support are the following:
110 ============== ============= ============= ================
111 ic-mode kernel_irqchip
112 -------------- ----------------------------------------------
115 ============== ============= ============= ================
116 dual (default) XIVE emul.(1) XIVE emul. QEMU error (2)
117 xive XIVE emul.(1) XIVE emul. QEMU error (2)
118 xics XICS KVM XICS emul. XICS KVM
119 ============== ============= ============= ================
122 (1) QEMU warns with ``warning: kernel_irqchip requested but unavailable:
123 IRQ_XIVE capability must be present for KVM``
124 (2) QEMU fails with ``kernel_irqchip requested but unavailable:
125 IRQ_XIVE capability must be present for KVM``
128 For legacy guest OSes without XIVE support, the resulting interrupt
129 modes are the following:
131 ============== ============= ============= ================
132 ic-mode kernel_irqchip
133 -------------- ----------------------------------------------
136 ============== ============= ============= ================
137 dual (default) QEMU error(4) XICS emul. QEMU error(4)
138 xive QEMU error(3) QEMU error(3) QEMU error(3)
139 xics XICS KVM XICS emul. XICS KVM
140 ============== ============= ============= ================
142 (3) QEMU fails at CAS with ``Guest requested unavailable interrupt
143 mode (XICS), either don't set the ic-mode machine property or try
144 ic-mode=xics or ic-mode=dual``
145 (4) QEMU/KVM incompatibility due to device destruction in reset. QEMU fails
146 with ``KVM is too old to support ic-mode=dual,kernel-irqchip=on``
149 XIVE Device tree properties
150 ---------------------------
152 The properties for the PAPR interrupt controller node when the *XIVE
153 native exploitation mode* is selected shoud contain:
157 value should be "power-ivpe".
161 value should be "ibm,power-ivpe".
165 contains the base address and size of the thread interrupt
166 managnement areas (TIMA), for the User level and for the Guest OS
167 level. Only the Guest OS level is taken into account today.
169 - ``ibm,xive-eq-sizes``
171 the size of the event queues. One cell per size supported, contains
172 log2 of size, in ascending order.
174 - ``ibm,xive-lisn-ranges``
176 the IRQ interrupt number ranges assigned to the guest for the IPIs.
178 The root node also exports :
180 - ``ibm,plat-res-int-priorities``
182 contains a list of priorities that the hypervisor has reserved for
188 IRQ Number space of the ``pseries`` machine is 8K wide and is the same
189 for both interrupt mode. The different ranges are defined as follow :
191 - ``0x0000 .. 0x0FFF`` 4K CPU IPIs (only used under XIVE)
192 - ``0x1000 .. 0x1000`` 1 EPOW
193 - ``0x1001 .. 0x1001`` 1 HOTPLUG
194 - ``0x1002 .. 0x10FF`` unused
195 - ``0x1100 .. 0x11FF`` 256 VIO devices
196 - ``0x1200 .. 0x127F`` 32x4 LSIs for PHB devices
197 - ``0x1280 .. 0x12FF`` unused
198 - ``0x1300 .. 0x1FFF`` PHB MSIs (dynamically allocated)
203 The state of the XIVE interrupt controller can be queried through the
204 monitor commands ``info pic``. The output comes in two parts.
206 First, the state of the thread interrupt context registers is dumped
212 CPU[0000]: QW NSR CPPR IPB LSMFB ACK# INC AGE PIPR W2
213 CPU[0000]: USER 00 00 00 00 00 00 00 00 00000000
214 CPU[0000]: OS 00 ff 00 00 ff 00 ff ff 80000400
215 CPU[0000]: POOL 00 00 00 00 00 00 00 00 00000000
216 CPU[0000]: PHYS 00 00 00 00 00 00 00 ff 00000000
219 In the case of a ``pseries`` machine, QEMU acts as the hypervisor and only
220 the O/S and USER register rings make sense. ``W2`` contains the vCPU CAM
221 line which is set to the VP identifier.
223 Then comes the routing information which aggregates the EAS and the
229 LISN PQ EISN CPU/PRIO EQ
230 00000000 MSI -- 00000010 0/6 380/16384 @1fe3e0000 ^1 [ 80000010 ... ]
231 00000001 MSI -- 00000010 1/6 305/16384 @1fc230000 ^1 [ 80000010 ... ]
232 00000002 MSI -- 00000010 2/6 220/16384 @1fc2f0000 ^1 [ 80000010 ... ]
233 00000003 MSI -- 00000010 3/6 201/16384 @1fc390000 ^1 [ 80000010 ... ]
234 00000004 MSI -Q M 00000000
235 00000005 MSI -Q M 00000000
236 00000006 MSI -Q M 00000000
237 00000007 MSI -Q M 00000000
238 00001000 MSI -- 00000012 0/6 380/16384 @1fe3e0000 ^1 [ 80000010 ... ]
239 00001001 MSI -- 00000013 0/6 380/16384 @1fe3e0000 ^1 [ 80000010 ... ]
240 00001100 MSI -- 00000100 1/6 305/16384 @1fc230000 ^1 [ 80000010 ... ]
241 00001101 MSI -Q M 00000000
242 00001200 LSI -Q M 00000000
243 00001201 LSI -Q M 00000000
244 00001202 LSI -Q M 00000000
245 00001203 LSI -Q M 00000000
246 00001300 MSI -- 00000102 1/6 305/16384 @1fc230000 ^1 [ 80000010 ... ]
247 00001301 MSI -- 00000103 2/6 220/16384 @1fc2f0000 ^1 [ 80000010 ... ]
248 00001302 MSI -- 00000104 3/6 201/16384 @1fc390000 ^1 [ 80000010 ... ]
250 The source information and configuration:
252 - The ``LISN`` column outputs the interrupt number of the source in
253 range ``[ 0x0 ... 0x1FFF ]`` and its type : ``MSI`` or ``LSI``
254 - The ``PQ`` column reflects the state of the PQ bits of the source :
256 - ``--`` source is ready to take events
257 - ``P-`` an event was sent and an EOI is PENDING
258 - ``PQ`` an event was QUEUED
259 - ``-Q`` source is OFF
261 a ``M`` indicates that source is *MASKED* at the EAS level,
263 The targeting configuration :
265 - The ``EISN`` column is the event data that will be queued in the event
267 - The ``CPU/PRIO`` column is the tuple defining the CPU number and
268 priority queue serving the source.
269 - The ``EQ`` column outputs :
271 - the current index of the event queue/ the max number of entries
272 - the O/S event queue address
274 - the last entries that were pushed in the event queue.