brintos

brintos / linux-shallow public Read only

0
0
Text · 13.4 KiB · 3aecf2a Raw
391 lines · plain
1.. SPDX-License-Identifier: GPL-2.02 3=================4KVM-specific MSRs5=================6 7:Author: Glauber Costa <glommer@redhat.com>, Red Hat Inc, 20108 9KVM makes use of some custom MSRs to service some requests.10 11Custom MSRs have a range reserved for them, that goes from120x4b564d00 to 0x4b564dff. There are MSRs outside this area,13but they are deprecated and their use is discouraged.14 15Custom MSR list16---------------17 18The current supported Custom MSR list is:19 20MSR_KVM_WALL_CLOCK_NEW:21	0x4b564d0022 23data:24	4-byte alignment physical address of a memory area which must be25	in guest RAM. This memory is expected to hold a copy of the following26	structure::27 28	 struct pvclock_wall_clock {29		u32   version;30		u32   sec;31		u32   nsec;32	  } __attribute__((__packed__));33 34	whose data will be filled in by the hypervisor. The hypervisor is only35	guaranteed to update this data at the moment of MSR write.36	Users that want to reliably query this information more than once have37	to write more than once to this MSR. Fields have the following meanings:38 39	version:40		guest has to check version before and after grabbing41		time information and check that they are both equal and even.42		An odd version indicates an in-progress update.43 44	sec:45		 number of seconds for wallclock at time of boot.46 47	nsec:48		 number of nanoseconds for wallclock at time of boot.49 50	In order to get the current wallclock time, the system_time from51	MSR_KVM_SYSTEM_TIME_NEW needs to be added.52 53	Note that although MSRs are per-CPU entities, the effect of this54	particular MSR is global.55 56	Availability of this MSR must be checked via bit 3 in 0x4000001 cpuid57	leaf prior to usage.58 59MSR_KVM_SYSTEM_TIME_NEW:60	0x4b564d0161 62data:63	4-byte aligned physical address of a memory area which must be in64	guest RAM, plus an enable bit in bit 0. This memory is expected to hold65	a copy of the following structure::66 67	  struct pvclock_vcpu_time_info {68		u32   version;69		u32   pad0;70		u64   tsc_timestamp;71		u64   system_time;72		u32   tsc_to_system_mul;73		s8    tsc_shift;74		u8    flags;75		u8    pad[2];76	  } __attribute__((__packed__)); /* 32 bytes */77 78	whose data will be filled in by the hypervisor periodically. Only one79	write, or registration, is needed for each VCPU. The interval between80	updates of this structure is arbitrary and implementation-dependent.81	The hypervisor may update this structure at any time it sees fit until82	anything with bit0 == 0 is written to it.83 84	Fields have the following meanings:85 86	version:87		guest has to check version before and after grabbing88		time information and check that they are both equal and even.89		An odd version indicates an in-progress update.90 91	tsc_timestamp:92		the tsc value at the current VCPU at the time93		of the update of this structure. Guests can subtract this value94		from current tsc to derive a notion of elapsed time since the95		structure update.96 97	system_time:98		a host notion of monotonic time, including sleep99		time at the time this structure was last updated. Unit is100		nanoseconds.101 102	tsc_to_system_mul:103		multiplier to be used when converting104		tsc-related quantity to nanoseconds105 106	tsc_shift:107		shift to be used when converting tsc-related108		quantity to nanoseconds. This shift will ensure that109		multiplication with tsc_to_system_mul does not overflow.110		A positive value denotes a left shift, a negative value111		a right shift.112 113		The conversion from tsc to nanoseconds involves an additional114		right shift by 32 bits. With this information, guests can115		derive per-CPU time by doing::116 117			time = (current_tsc - tsc_timestamp)118			if (tsc_shift >= 0)119				time <<= tsc_shift;120			else121				time >>= -tsc_shift;122			time = (time * tsc_to_system_mul) >> 32123			time = time + system_time124 125	flags:126		bits in this field indicate extended capabilities127		coordinated between the guest and the hypervisor. Availability128		of specific flags has to be checked in 0x40000001 cpuid leaf.129		Current flags are:130 131 132		+-----------+--------------+----------------------------------+133		| flag bit  | cpuid bit    | meaning			      |134		+-----------+--------------+----------------------------------+135		|	    |		   | time measures taken across       |136		|    0      |	   24      | multiple cpus are guaranteed to  |137		|	    |		   | be monotonic		      |138		+-----------+--------------+----------------------------------+139		|	    |		   | guest vcpu has been paused by    |140		|    1	    |	  N/A	   | the host			      |141		|	    |		   | See 4.70 in api.txt	      |142		+-----------+--------------+----------------------------------+143 144	Availability of this MSR must be checked via bit 3 in 0x4000001 cpuid145	leaf prior to usage.146 147 148MSR_KVM_WALL_CLOCK:149	0x11150 151data and functioning:152	same as MSR_KVM_WALL_CLOCK_NEW. Use that instead.153 154	This MSR falls outside the reserved KVM range and may be removed in the155	future. Its usage is deprecated.156 157	Availability of this MSR must be checked via bit 0 in 0x4000001 cpuid158	leaf prior to usage.159 160MSR_KVM_SYSTEM_TIME:161	0x12162 163data and functioning:164	same as MSR_KVM_SYSTEM_TIME_NEW. Use that instead.165 166	This MSR falls outside the reserved KVM range and may be removed in the167	future. Its usage is deprecated.168 169	Availability of this MSR must be checked via bit 0 in 0x4000001 cpuid170	leaf prior to usage.171 172	The suggested algorithm for detecting kvmclock presence is then::173 174		if (!kvm_para_available())    /* refer to cpuid.txt */175			return NON_PRESENT;176 177		flags = cpuid_eax(0x40000001);178		if (flags & 3) {179			msr_kvm_system_time = MSR_KVM_SYSTEM_TIME_NEW;180			msr_kvm_wall_clock = MSR_KVM_WALL_CLOCK_NEW;181			return PRESENT;182		} else if (flags & 0) {183			msr_kvm_system_time = MSR_KVM_SYSTEM_TIME;184			msr_kvm_wall_clock = MSR_KVM_WALL_CLOCK;185			return PRESENT;186		} else187			return NON_PRESENT;188 189MSR_KVM_ASYNC_PF_EN:190	0x4b564d02191 192data:193	Asynchronous page fault (APF) control MSR.194 195	Bits 63-6 hold 64-byte aligned physical address of a 64 byte memory area196	which must be in guest RAM. This memory is expected to hold the197	following structure::198 199	  struct kvm_vcpu_pv_apf_data {200		/* Used for 'page not present' events delivered via #PF */201		__u32 flags;202 203		/* Used for 'page ready' events delivered via interrupt notification */204		__u32 token;205 206		__u8 pad[56];207	  };208 209	Bits 5-4 of the MSR are reserved and should be zero. Bit 0 is set to 1210	when asynchronous page faults are enabled on the vcpu, 0 when disabled.211	Bit 1 is 1 if asynchronous page faults can be injected when vcpu is in212	cpl == 0. Bit 2 is 1 if asynchronous page faults are delivered to L1 as213	#PF vmexits.  Bit 2 can be set only if KVM_FEATURE_ASYNC_PF_VMEXIT is214	present in CPUID. Bit 3 enables interrupt based delivery of 'page ready'215	events. Bit 3 can only be set if KVM_FEATURE_ASYNC_PF_INT is present in216	CPUID.217 218	'Page not present' events are currently always delivered as synthetic219	#PF exception. During delivery of these events APF CR2 register contains220	a token that will be used to notify the guest when missing page becomes221	available. Also, to make it possible to distinguish between real #PF and222	APF, first 4 bytes of 64 byte memory location ('flags') will be written223	to by the hypervisor at the time of injection. Only first bit of 'flags'224	is currently supported, when set, it indicates that the guest is dealing225	with asynchronous 'page not present' event. If during a page fault APF226	'flags' is '0' it means that this is regular page fault. Guest is227	supposed to clear 'flags' when it is done handling #PF exception so the228	next event can be delivered.229 230	Note, since APF 'page not present' events use the same exception vector231	as regular page fault, guest must reset 'flags' to '0' before it does232	something that can generate normal page fault.233 234	Bytes 4-7 of 64 byte memory location ('token') will be written to by the235	hypervisor at the time of APF 'page ready' event injection. The content236	of these bytes is a token which was previously delivered in CR2 as237	'page not present' event. The event indicates the page is now available.238	Guest is supposed to write '0' to 'token' when it is done handling239	'page ready' event and to write '1' to MSR_KVM_ASYNC_PF_ACK after240	clearing the location; writing to the MSR forces KVM to re-scan its241	queue and deliver the next pending notification.242 243	Note, MSR_KVM_ASYNC_PF_INT MSR specifying the interrupt vector for 'page244	ready' APF delivery needs to be written to before enabling APF mechanism245	in MSR_KVM_ASYNC_PF_EN or interrupt #0 can get injected. The MSR is246	available if KVM_FEATURE_ASYNC_PF_INT is present in CPUID.247 248	Note, previously, 'page ready' events were delivered via the same #PF249	exception as 'page not present' events but this is now deprecated. If250	bit 3 (interrupt based delivery) is not set APF events are not delivered.251 252	If APF is disabled while there are outstanding APFs, they will253	not be delivered.254 255	Currently 'page ready' APF events will be always delivered on the256	same vcpu as 'page not present' event was, but guest should not rely on257	that.258 259MSR_KVM_STEAL_TIME:260	0x4b564d03261 262data:263	64-byte alignment physical address of a memory area which must be264	in guest RAM, plus an enable bit in bit 0. This memory is expected to265	hold a copy of the following structure::266 267	  struct kvm_steal_time {268		__u64 steal;269		__u32 version;270		__u32 flags;271		__u8  preempted;272		__u8  u8_pad[3];273		__u32 pad[11];274	  }275 276	whose data will be filled in by the hypervisor periodically. Only one277	write, or registration, is needed for each VCPU. The interval between278	updates of this structure is arbitrary and implementation-dependent.279	The hypervisor may update this structure at any time it sees fit until280	anything with bit0 == 0 is written to it. Guest is required to make sure281	this structure is initialized to zero.282 283	Fields have the following meanings:284 285	version:286		a sequence counter. In other words, guest has to check287		this field before and after grabbing time information and make288		sure they are both equal and even. An odd version indicates an289		in-progress update.290 291	flags:292		At this point, always zero. May be used to indicate293		changes in this structure in the future.294 295	steal:296		the amount of time in which this vCPU did not run, in297		nanoseconds. Time during which the vcpu is idle, will not be298		reported as steal time.299 300	preempted:301		indicate the vCPU who owns this struct is running or302		not. Non-zero values mean the vCPU has been preempted. Zero303		means the vCPU is not preempted. NOTE, it is always zero if the304		the hypervisor doesn't support this field.305 306MSR_KVM_EOI_EN:307	0x4b564d04308 309data:310	Bit 0 is 1 when PV end of interrupt is enabled on the vcpu; 0311	when disabled.  Bit 1 is reserved and must be zero.  When PV end of312	interrupt is enabled (bit 0 set), bits 63-2 hold a 4-byte aligned313	physical address of a 4 byte memory area which must be in guest RAM and314	must be zeroed.315 316	The first, least significant bit of 4 byte memory location will be317	written to by the hypervisor, typically at the time of interrupt318	injection.  Value of 1 means that guest can skip writing EOI to the apic319	(using MSR or MMIO write); instead, it is sufficient to signal320	EOI by clearing the bit in guest memory - this location will321	later be polled by the hypervisor.322	Value of 0 means that the EOI write is required.323 324	It is always safe for the guest to ignore the optimization and perform325	the APIC EOI write anyway.326 327	Hypervisor is guaranteed to only modify this least328	significant bit while in the current VCPU context, this means that329	guest does not need to use either lock prefix or memory ordering330	primitives to synchronise with the hypervisor.331 332	However, hypervisor can set and clear this memory bit at any time:333	therefore to make sure hypervisor does not interrupt the334	guest and clear the least significant bit in the memory area335	in the window between guest testing it to detect336	whether it can skip EOI apic write and between guest337	clearing it to signal EOI to the hypervisor,338	guest must both read the least significant bit in the memory area and339	clear it using a single CPU instruction, such as test and clear, or340	compare and exchange.341 342MSR_KVM_POLL_CONTROL:343	0x4b564d05344 345	Control host-side polling.346 347data:348	Bit 0 enables (1) or disables (0) host-side HLT polling logic.349 350	KVM guests can request the host not to poll on HLT, for example if351	they are performing polling themselves.352 353MSR_KVM_ASYNC_PF_INT:354	0x4b564d06355 356data:357	Second asynchronous page fault (APF) control MSR.358 359	Bits 0-7: APIC vector for delivery of 'page ready' APF events.360	Bits 8-63: Reserved361 362	Interrupt vector for asynchnonous 'page ready' notifications delivery.363	The vector has to be set up before asynchronous page fault mechanism364	is enabled in MSR_KVM_ASYNC_PF_EN.  The MSR is only available if365	KVM_FEATURE_ASYNC_PF_INT is present in CPUID.366 367MSR_KVM_ASYNC_PF_ACK:368	0x4b564d07369 370data:371	Asynchronous page fault (APF) acknowledgment.372 373	When the guest is done processing 'page ready' APF event and 'token'374	field in 'struct kvm_vcpu_pv_apf_data' is cleared it is supposed to375	write '1' to bit 0 of the MSR, this causes the host to re-scan its queue376	and check if there are more notifications pending. The MSR is available377	if KVM_FEATURE_ASYNC_PF_INT is present in CPUID.378 379MSR_KVM_MIGRATION_CONTROL:380        0x4b564d08381 382data:383        This MSR is available if KVM_FEATURE_MIGRATION_CONTROL is present in384        CPUID.  Bit 0 represents whether live migration of the guest is allowed.385 386        When a guest is started, bit 0 will be 0 if the guest has encrypted387        memory and 1 if the guest does not have encrypted memory.  If the388        guest is communicating page encryption status to the host using the389        ``KVM_HC_MAP_GPA_RANGE`` hypercall, it can set bit 0 in this MSR to390        allow live migration of the guest.391