xen+tools: Report Interrupt Controller Virtualization capabilities on x86
v3 cover xen+tools: Report Interrupt Controller Virtualization capabilities on x86
Jane Malalane (2):
xen+tools: Report Interrupt Controller Virtualization capabilities on
x86
x86/xen: Allow per-domain usage of hardware virtualized APIC
docs/man/xl.cfg.5.pod.in | 19 +++++++++++++++++ docs/man/xl.conf.5.pod.in | 12 +++++++++++ tools/golang/xenlight/helpers.gen.go | 16 ++++++++++++++ tools/golang/xenlight/types.gen.go | 4 ++++ tools/include/libxl.h | 14 +++++++++++++ tools/libs/light/libxl.c | 3 +++ tools/libs/light/libxl_arch.h | 9 ++++++-- tools/libs/light/libxl_arm.c | 12 +++++++++-- tools/libs/light/libxl_create.c | 22 ++++++++++++-------- tools/libs/light/libxl_types.idl | 4 ++++ tools/libs/light/libxl_x86.c | 39 +++++++++++++++++++++++++++++++++-- tools/ocaml/libs/xc/xenctrl.ml | 7 +++++++ tools/ocaml/libs/xc/xenctrl.mli | 7 +++++++ tools/ocaml/libs/xc/xenctrl_stubs.c | 16 ++++++++++---- tools/xl/xl.c | 8 +++++++ tools/xl/xl.h | 2 ++ tools/xl/xl_info.c | 6 ++++-- tools/xl/xl_parse.c | 16 ++++++++++++++ xen/arch/x86/domain.c | 28 ++++++++++++++++++++++++- xen/arch/x86/hvm/vmx/vmcs.c | 11 ++++++++++ xen/arch/x86/hvm/vmx/vmx.c | 8 +++---- xen/arch/x86/include/asm/domain.h | 3 +++ xen/arch/x86/include/asm/hvm/domain.h | 6 ++++++ xen/arch/x86/sysctl.c | 7 +++++++ xen/arch/x86/traps.c | 8 +++---- xen/include/public/arch-x86/xen.h | 2 ++ xen/include/public/sysctl.h | 11 +++++++++- 27 files changed, 269 insertions(+), 31 deletions(-)
Signature · 1 line
2.11.0
PATCH v3 1/2 xen+tools: Report Interrupt Controller Virtualization capabilities on x86
Add XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapic and XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apic to report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.
For that purpose, also add an arch-specific "capabilities" parameter to struct xen_sysctl_physinfo.
---
CC: Wei Liu <[email protected]>
CC: Anthony PERARD <[email protected]>
CC: Juergen Gross <[email protected]>
CC: Andrew Cooper <[email protected]>
CC: George Dunlap <[email protected]>
CC: Jan Beulich <[email protected]>
CC: Julien Grall <[email protected]>
CC: Stefano Stabellini <[email protected]>
CC: Volodymyr Babchuk <[email protected]>
CC: Bertrand Marquis <[email protected]>
CC: Jun Nakajima <[email protected]>
CC: Kevin Tian <[email protected]>
CC: "Roger Pau Monné" <[email protected]>
v3:
* Define XEN_SYSCTL_PHYSCAP_ARCH_MAX for ABI checking and actually set arch_capbilities, via a call to c_bitmap_to_ocaml_list()
* Have assisted_x2apic_available only depend on cpu_has_vmx_virtualize_x2apic_mode
v2:
* Use one macro LIBXL_HAVE_PHYSINFO_ASSISTED_APIC instead of two
* Pass xcpyshinfo as a pointer in libxl__arch_get_physinfo
* Set assisted_x{2}apic_available to be conditional upon "bsp" and annotate it with __ro_after_init
* Change XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_X{2}APIC to (...)_X86_ASSISTED_X{2}APIC
* Keep XEN_SYSCTL_PHYSCAP_X86_ASSISTED_X{2}APIC contained within sysctl.h
* Fix padding introduced in struct xen_sysctl_physinfo and bump XEN_SYSCTL_INTERFACE_VERSION
--- tools/golang/xenlight/helpers.gen.go | 4 ++++ tools/golang/xenlight/types.gen.go | 2 ++ tools/include/libxl.h | 7 +++++++ tools/libs/light/libxl.c | 3 +++ tools/libs/light/libxl_arch.h | 4 ++++ tools/libs/light/libxl_arm.c | 5 +++++ tools/libs/light/libxl_types.idl | 2 ++ tools/libs/light/libxl_x86.c | 11 +++++++++++ tools/ocaml/libs/xc/xenctrl.ml | 5 +++++ tools/ocaml/libs/xc/xenctrl.mli | 5 +++++ tools/ocaml/libs/xc/xenctrl_stubs.c | 14 +++++++++++--- tools/xl/xl_info.c | 6 ++++-- xen/arch/x86/hvm/vmx/vmcs.c | 7 +++++++ xen/arch/x86/include/asm/domain.h | 3 +++ xen/arch/x86/sysctl.c | 7 +++++++ xen/include/public/sysctl.h | 11 ++++++++++- 16 files changed, 90 insertions(+), 6 deletions(-)
tools/golang/xenlight/helpers.gen.go +4 -0
@@ -3373,6 +3373,8 @@ x.CapVmtrace = bool(xc.cap_vmtrace) x.CapVpmu = bool(xc.cap_vpmu) x.CapGnttabV1 = bool(xc.cap_gnttab_v1) x.CapGnttabV2 = bool(xc.cap_gnttab_v2) +x.CapAssistedXapic = bool(xc.cap_assisted_xapic) +x.CapAssistedX2Apic = bool(xc.cap_assisted_x2apic) return nil}
@@ -3407,6 +3409,8 @@ xc.cap_vmtrace = C.bool(x.CapVmtrace) xc.cap_vpmu = C.bool(x.CapVpmu) xc.cap_gnttab_v1 = C.bool(x.CapGnttabV1) xc.cap_gnttab_v2 = C.bool(x.CapGnttabV2) +xc.cap_assisted_xapic = C.bool(x.CapAssistedXapic) +xc.cap_assisted_x2apic = C.bool(x.CapAssistedX2Apic) return nil }
tools/golang/xenlight/types.gen.go +2 -0
@@ -1014,6 +1014,8 @@ CapVmtrace bool CapVpmu bool CapGnttabV1 bool CapGnttabV2 bool +CapAssistedXapic bool +CapAssistedX2Apic bool } type Connectorinfo struct {
tools/include/libxl.h +7 -0
@@ -528,6 +528,13 @@#define LIBXL_HAVE_MAX_GRANT_VERSION 1/* +* LIBXL_HAVE_PHYSINFO_ASSISTED_APIC indicates that libxl_physinfo has +* cap_assisted_x{2}apic fields, which indicates the availability of x{2}APIC +* hardware assisted virtualization. +*/ +#define LIBXL_HAVE_PHYSINFO_ASSISTED_APIC 1 + +/* * libxl ABI compatibility * * The only guarantee which libxl makes regarding ABI compatibility
tools/libs/light/libxl.c +3 -0
@@ -15,6 +15,7 @@#include "libxl_osdeps.h" #include "libxl_internal.h" +#include "libxl_arch.h" int libxl_ctx_alloc (libxl_ctx **pctx,int version,unsigned flags,xentoollog_logger * lg)
@@ -410,6 +411,8 @@ int libxl_get_physinfo(libxl_ctx *ctx, libxl_physinfo *physinfo) physinfo->cap_gnttab_v2 = !!(xcphysinfo.capabilities & XEN_SYSCTL_PHYSCAP_gnttab_v2); +libxl__arch_get_physinfo (physinfo, &xcphysinfo); +GC_FREE ;return 0 ; }
tools/libs/light/libxl_arch.h +4 -0
@@ -86,6 +86,10 @@ int libxl__arch_extra_memory(libxl__gc *gc, uint64_t *out);_hidden +void libxl__arch_get_physinfo(libxl_physinfo *physinfo, +const xc_physinfo_t *xcphysinfo); + +_hidden void libxl__arch_update_domain_config(libxl__gc *gc,libxl_domain_config *dst,const libxl_domain_config *src);
tools/libs/light/libxl_arm.c +5 -0
@@ -1431,6 +1431,11 @@ int libxl__arch_passthrough_mode_setdefault(libxl__gc *gc,return rc; } +void libxl__arch_get_physinfo (libxl_physinfo *physinfo, +const xc_physinfo_t *xcphysinfo) +{ +} +void libxl__arch_update_domain_config (libxl__gc *gc,libxl_domain_config *dst,const libxl_domain_config *src)
tools/libs/light/libxl_types.idl +2 -0
@@ -1068,6 +1068,8 @@ libxl_physinfo = Struct("physinfo", [ ("cap_vpmu", bool), ("cap_gnttab_v1", bool), ("cap_gnttab_v2", bool), + ("cap_assisted_xapic", bool), + ("cap_assisted_x2apic", bool), ], dir=DIR_OUT) libxl_connectorinfo = Struct("connectorinfo", [
tools/libs/light/libxl_x86.c +11 -0
@@ -866,6 +866,17 @@ int libxl__arch_passthrough_mode_setdefault(libxl__gc *gc,return rc; } +void libxl__arch_get_physinfo (libxl_physinfo *physinfo, +const xc_physinfo_t *xcphysinfo) +{ + physinfo->cap_assisted_xapic = + !!(xcphysinfo->arch_capabilities & +XEN_SYSCTL_PHYSCAP_X86_ASSISTED_XAPIC ); + physinfo->cap_assisted_x2apic = + !!(xcphysinfo->arch_capabilities & +XEN_SYSCTL_PHYSCAP_X86_ASSISTED_X2APIC ); +} +void libxl__arch_update_domain_config (libxl__gc *gc,libxl_domain_config *dst,const libxl_domain_config *src)
tools/ocaml/libs/xc/xenctrl.ml +5 -0
@@ -127,6 +127,10 @@ type physinfo_cap_flag = | CAP_Gnttab_v1 | CAP_Gnttab_v2 +type physinfo_arch_cap_flag = + | CAP_X86_ASSISTED_XAPIC + | CAP_X86_ASSISTED_X2APIC + type physinfo = { threads_per_core : int;
@@ -139,6 +143,7 @@ type physinfo = scrub_pages : nativeint; (* XXX hw_cap *) capabilities : physinfo_cap_flag list; + arch_capabilities : physinfo_arch_cap_flag list; max_nr_cpus : int; }
tools/ocaml/libs/xc/xenctrl.mli +5 -0
@@ -112,6 +112,10 @@ type physinfo_cap_flag = | CAP_Gnttab_v1 | CAP_Gnttab_v2 +type physinfo_arch_cap_flag = + | CAP_X86_ASSISTED_XAPIC + | CAP_X86_ASSISTED_X2APIC + type physinfo = { threads_per_core : int; cores_per_socket : int;
@@ -122,6 +126,7 @@ type physinfo = { free_pages : nativeint; scrub_pages : nativeint; capabilities : physinfo_cap_flag list; + arch_capabilities : physinfo_arch_cap_flag list; max_nr_cpus : int; (** compile-time max possible number of nr_cpus *) } type version = { major : int; minor : int; extra : string; }
tools/ocaml/libs/xc/xenctrl_stubs.c +11 -3
@@ -712,7 +712,7 @@ CAMLprim value stub_xc_send_debug_keys(value xch, value keys) CAMLprim value stub_xc_physinfo(value xch) {CAMLparam1 (xch); -CAMLlocal2 (physinfo, cap_list); +CAMLlocal3 (physinfo, cap_list, arch_cap_list);xc_physinfo_t c_physinfo;int r;
@@ -730,8 +730,15 @@ CAMLprim value stub_xc_physinfo(value xch)/* ! physinfo_cap_flag CAP_ lc */ /* ! XEN_SYSCTL_PHYSCAP_ XEN_SYSCTL_PHYSCAP_MAX max */ (c_physinfo.capabilities); +/* +* arch_capabilities: physinfo_arch_cap_flag list; +*/ + arch_cap_list =c_bitmap_to_ocaml_list +/* ! physinfo_arch_cap_flag CAP_ none */ +/* ! XEN_SYSCTL_PHYSCAP_ XEN_SYSCTL_PHYSCAP_ARCH_MAX max */ + (c_physinfo.arch_capabilities); - physinfo =caml_alloc_tuple (10 ); + physinfo =caml_alloc_tuple (11 );Store_field (physinfo,0 ,Val_int (c_physinfo.threads_per_core));Store_field (physinfo,1 ,Val_int (c_physinfo.cores_per_socket));Store_field (physinfo,2 ,Val_int (c_physinfo.nr_cpus));
@@ -741,7 +748,8 @@ CAMLprim value stub_xc_physinfo(value xch)Store_field (physinfo,6 ,caml_copy_nativeint (c_physinfo.free_pages));Store_field (physinfo,7 ,caml_copy_nativeint (c_physinfo.scrub_pages));Store_field (physinfo,8 , cap_list); -Store_field (physinfo,9 ,Val_int (c_physinfo.max_cpu_id +1 )); +Store_field (physinfo,9 , arch_cap_list); +Store_field (physinfo,10 ,Val_int (c_physinfo.max_cpu_id +1 ));CAMLreturn (physinfo); }
tools/xl/xl_info.c +4 -2
@@ -210,7 +210,7 @@ static void output_physinfo(void) info.hw_cap[4 ], info.hw_cap[5 ], info.hw_cap[6 ], info.hw_cap[7 ] ); - maybe_printf("virt_caps :%s%s%s%s%s%s%s%s%s%s%s\n", + maybe_printf("virt_caps :%s%s%s%s%s%s%s%s%s%s%s%s%s\n", info.cap_pv ? " pv" : "", info.cap_hvm ? " hvm" : "", info.cap_hvm && info.cap_hvm_directio ? " hvm_directio" : "",
@@ -221,7 +221,9 @@ static void output_physinfo(void) info.cap_vmtrace ?" vmtrace" :"" , info.cap_vpmu ?" vpmu" :"" , info.cap_gnttab_v1 ?" gnttab-v1" :"" , - info.cap_gnttab_v2 ?" gnttab-v2" :"" + info.cap_gnttab_v2 ?" gnttab-v2" :"" , + info.cap_assisted_xapic ?" assisted_xapic" :"" , + info.cap_assisted_x2apic ?" assisted_x2apic" :"" ); vinfo =libxl_get_version_info (ctx);
xen/arch/x86/hvm/vmx/vmcs.c +7 -0
@@ -343,6 +343,13 @@ static int vmx_init_vmcs_config(bool bsp) MSR_IA32_VMX_PROCBASED_CTLS2, &mismatch); } +/* Check whether hardware supports accelerated xapic and x2apic. */ +if ( bsp ) + { + assisted_xapic_available = cpu_has_vmx_virtualize_apic_accesses; + assisted_x2apic_available = cpu_has_vmx_virtualize_x2apic_mode; + } +/* The IA32_VMX_EPT_VPID_CAP MSR exists only when EPT or VPID available */ if ( _vmx_secondary_exec_control & (SECONDARY_EXEC_ENABLE_EPT |SECONDARY_EXEC_ENABLE_VPID ) )
xen/arch/x86/include/asm/domain.h +3 -0
@@ -756,6 +756,9 @@ static inline void pv_inject_sw_interrupt(unsigned int vector) : is_pv_32bit_domain(d) ? PV32_VM_ASSIST_MASK \ : PV64_VM_ASSIST_MASK) +extern bool assisted_xapic_available; +extern bool assisted_x2apic_available; +#endif /* __ASM_DOMAIN_H__ */ /*
xen/arch/x86/sysctl.c +7 -0
@@ -69,6 +69,9 @@ struct l3_cache_info {unsigned long size; }; +bool __ro_after_init assisted_xapic_available; +bool __ro_after_init assisted_x2apic_available; +static void l3_cache_get (void *arg) {struct cpuid4_info info;
@@ -135,6 +138,10 @@ void arch_do_physinfo(struct xen_sysctl_physinfo *pi) pi->capabilities |= XEN_SYSCTL_PHYSCAP_hap;if (IS_ENABLED (CONFIG_SHADOW_PAGING ) ) pi->capabilities |= XEN_SYSCTL_PHYSCAP_shadow; +if ( assisted_xapic_available ) + pi->arch_capabilities |=XEN_SYSCTL_PHYSCAP_X86_ASSISTED_XAPIC ; +if ( assisted_x2apic_available ) + pi->arch_capabilities |=XEN_SYSCTL_PHYSCAP_X86_ASSISTED_X2APIC ; } long arch_do_sysctl(
xen/include/public/sysctl.h +10 -1
@@ -35,7 +35,7 @@#include "domctl.h" #include "physdev.h" -#define XEN_SYSCTL_INTERFACE_VERSION 0x00000014 +#define XEN_SYSCTL_INTERFACE_VERSION 0x00000015/* * Read console content from Xen buffer ring.
@@ -111,6 +111,13 @@ struct xen_sysctl_tbuf_op {/* Max XEN_SYSCTL_PHYSCAP_* constant. Used for ABI checking. */ #define XEN_SYSCTL_PHYSCAP_MAX XEN_SYSCTL_PHYSCAP_gnttab_v2 +/* The platform supports x{2}apic hardware assisted emulation. */ +#define XEN_SYSCTL_PHYSCAP_X86_ASSISTED_XAPIC (1u << 0) +#define XEN_SYSCTL_PHYSCAP_X86_ASSISTED_X2APIC (1u << 1) + +/* Max XEN_SYSCTL_PHYSCAP_X86{ARM}__* constant. Used for ABI checking. */ +#define XEN_SYSCTL_PHYSCAP_ARCH_MAX XEN_SYSCTL_PHYSCAP_X86_ASSISTED_X2APIC +struct xen_sysctl_physinfo {uint32_t threads_per_core;uint32_t cores_per_socket;
@@ -120,6 +127,8 @@ struct xen_sysctl_physinfo {uint32_t max_node_id;/* Largest possible node ID on this host */ uint32_t cpu_khz;uint32_t capabilities;/* XEN_SYSCTL_PHYSCAP_??? */ +uint32_t arch_capabilities;/* XEN_SYSCTL_PHYSCAP_X86{ARM}_??? */ +uint32_t pad;/* Must be zero. */ uint64_aligned_t total_pages;uint64_aligned_t free_pages;uint64_aligned_t scrub_pages;
Signature · 1 line
2.11.0
Quoted 8 lines · Jane Malalane
Add
XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapicandXEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apicto report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.For that purpose, also add an arch-specific "capabilities" parameter to struct
xen_sysctl_physinfo.
Quoted 8 lines
Suggested-by Andrew Cooper <[email protected]>Signed-off-by Jane Malalane <[email protected]>---
v3:
* DefineXEN_SYSCTL_PHYSCAP_ARCH_MAXfor ABI checking and actually setarch_capbilities, via a call toc_bitmap_to_ocaml_list()
* Haveassisted_x2apic_availableonly depend oncpu_has_vmx_virtualize_x2apic_mode
I understand this was the result from previous discussion, but this needs justifying in the description. Not the least because it differs from when XEN_HVM_CPUID_X2APIC_VIRT would be set as well as from what vmx_vlapic_msr_changed() does. The difference between those two is probably intended (judging from a comment there), but the further difference to what you add isn't obvious.
Which raises another thought: If that hypervisor leaf was part of the HVM feature set, the tool stack could be able to obtain the wanted information without altering sysctl (assuming the conditions to set the respective bits were the same). And I would view it as generally reasonable for there to be a way for tool stacks to know what hypervisor leaves guests are going to get to see (at the maximum and by default).
Quoted 12 lines
xen/include/public/sysctl.h +1
-1@@ -35,7 +35,7 @@#include "domctl.h" #include "physdev.h" -#define XEN_SYSCTL_INTERFACE_VERSION 0x00000014 +#define XEN_SYSCTL_INTERFACE_VERSION 0x00000015/* * Read console content from Xen buffer ring.
xen/include/public/sysctl.h struct xen_sysctl_tbuf_op { · line 119
Quoted 9 lines
xen/include/public/sysctl.h +6
-0@@ -111,6 +111,13 @@ struct xen_sysctl_tbuf_op {/* Max XEN_SYSCTL_PHYSCAP_* constant. Used for ABI checking. */ #define XEN_SYSCTL_PHYSCAP_MAX XEN_SYSCTL_PHYSCAP_gnttab_v2 +/* The platform supports x{2}apic hardware assisted emulation. */ +#define XEN_SYSCTL_PHYSCAP_X86_ASSISTED_XAPIC (1u << 0) +#define XEN_SYSCTL_PHYSCAP_X86_ASSISTED_X2APIC (1u << 1) + +/* Max XEN_SYSCTL_PHYSCAP_X86{ARM}__* constant. Used for ABI checking. */ +#define XEN_SYSCTL_PHYSCAP_ARCH_MAX XEN_SYSCTL_PHYSCAP_X86_ASSISTED_X2APIC
Doesn't this then need to be a per-arch constant? The ABIs would differ unless we required that every bit may only be used for a single purpose. IOW it would want to be named XEN_SYSCTL_PHYSCAP_X86_MAX.
xen/include/public/sysctl.h struct xen_sysctl_physinfo { · line 131
Quoted 6 lines
xen/include/public/sysctl.h +2
-0@@ -120,6 +127,8 @@ struct xen_sysctl_physinfo {uint32_t max_node_id;/* Largest possible node ID on this host */ uint32_t cpu_khz;uint32_t capabilities;/* XEN_SYSCTL_PHYSCAP_??? */ +uint32_t arch_capabilities;/* XEN_SYSCTL_PHYSCAP_X86{ARM}_??? */ +uint32_t pad;/* Must be zero. */
If this was an input field (or could potentially become one), the comment would be applicable. But the whole struct is OUT-only, so either omit the comment or use e.g. "will" or better "reserved" (as people shouldn't make themselves dependent on the field being zero).
Jan
Quoted 20 lines · Jan Beulich
[CAUTION - EXTERNAL EMAIL] DO NOT reply, click links, or open attachments unless you have verified the sender and know the content is safe.
Quoted 8 lines · Jane Malalane
Add
XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapicandXEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apicto report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.For that purpose, also add an arch-specific "capabilities" parameter to struct
xen_sysctl_physinfo.Quoted 8 lines
Suggested-by Andrew Cooper <[email protected]>Signed-off-by Jane Malalane <[email protected]>---
v3:
* DefineXEN_SYSCTL_PHYSCAP_ARCH_MAXfor ABI checking and actually setarch_capbilities, via a call toc_bitmap_to_ocaml_list()
* Haveassisted_x2apic_availableonly depend oncpu_has_vmx_virtualize_x2apic_mode
Quoted 6 lines
I understand this was the result from previous discussion, but this needs justifying in the description. Not the least because it differs from when
XEN_HVM_CPUID_X2APIC_VIRTwould be set as well as from whatvmx_vlapic_msr_changed()does. The difference between those two is probably intended (judging from a comment there), but the further difference to what you add isn't obvious.
Okay, I will make that explicit.
Quoted 7 lines
Which raises another thought: If that hypervisor leaf was part of the HVM feature set, the tool stack could be able to obtain the wanted information without altering sysctl (assuming the conditions to set the respective bits were the same). And I would view it as generally reasonable for there to be a way for tool stacks to know what hypervisor leaves guests are going to get to see (at the maximum and by default).
Like the "cpuid" xtf test allows us to? Makes sense to me. I'm happy to take that up after.
Quoted 16 lines
Quoted 16 lines
xen/include/public/sysctl.h +1
-1@@ -35,7 +35,7 @@#include "domctl.h" #include "physdev.h" -#define XEN_SYSCTL_INTERFACE_VERSION 0x00000014 +#define XEN_SYSCTL_INTERFACE_VERSION 0x00000015/* * Read console content from Xen buffer ring. @@ -111,6 +111,13 @@ struct xen_sysctl_tbuf_op {/* Max XEN_SYSCTL_PHYSCAP_* constant. Used for ABI checking. */ #define XEN_SYSCTL_PHYSCAP_MAX XEN_SYSCTL_PHYSCAP_gnttab_v2
Quoted 10 lines
xen/include/public/sysctl.h struct xen_sysctl_tbuf_op { · line 119
Quoted 6 lines
xen/include/public/sysctl.h +6
-0+/* The platform supports x{2}apic hardware assisted emulation. */ +#define XEN_SYSCTL_PHYSCAP_X86_ASSISTED_XAPIC (1u << 0) +#define XEN_SYSCTL_PHYSCAP_X86_ASSISTED_X2APIC (1u << 1) + +/* Max XEN_SYSCTL_PHYSCAP_X86{ARM}__* constant. Used for ABI checking. */ +#define XEN_SYSCTL_PHYSCAP_ARCH_MAX XEN_SYSCTL_PHYSCAP_X86_ASSISTED_X2APICDoesn't this then need to be a per-arch constant? The ABIs would differ unless we required that every bit may only be used for a single purpose. IOW it would want to be named
XEN_SYSCTL_PHYSCAP_X86_MAX.
Okay.
Quoted 12 lines
xen/include/public/sysctl.h struct xen_sysctl_physinfo { · line 131
Quoted 7 lines
xen/include/public/sysctl.h +2
-0@@ -120,6 +127,8 @@ struct xen_sysctl_physinfo {uint32_t max_node_id;/* Largest possible node ID on this host */ uint32_t cpu_khz;uint32_t capabilities;/* XEN_SYSCTL_PHYSCAP_??? */ +uint32_t arch_capabilities;/* XEN_SYSCTL_PHYSCAP_X86{ARM}_??? */ +uint32_t pad;/* Must be zero. */ If this was an input field (or could potentially become one), the comment would be applicable. But the whole struct is OUT-only, so either omit the comment or use e.g. "will" or better "reserved" (as people shouldn't make themselves dependent on the field being zero).
Will ommit.
Thank you,
Jane.
Quoted 36 lines · Jane Malalane
Quoted 18 lines · Jan Beulich
Quoted 8 lines · Jane Malalane
Add
XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapicandXEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apicto report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.For that purpose, also add an arch-specific "capabilities" parameter to struct
xen_sysctl_physinfo.Quoted 8 lines
Suggested-by Andrew Cooper <[email protected]>Signed-off-by Jane Malalane <[email protected]>---
v3:
* DefineXEN_SYSCTL_PHYSCAP_ARCH_MAXfor ABI checking and actually setarch_capbilities, via a call toc_bitmap_to_ocaml_list()
* Haveassisted_x2apic_availableonly depend oncpu_has_vmx_virtualize_x2apic_modeQuoted 6 lines
I understand this was the result from previous discussion, but this needs justifying in the description. Not the least because it differs from when
XEN_HVM_CPUID_X2APIC_VIRTwould be set as well as from whatvmx_vlapic_msr_changed()does. The difference between those two is probably intended (judging from a comment there), but the further difference to what you add isn't obvious.Okay, I will make that explicit.
Quoted 7 lines
Which raises another thought: If that hypervisor leaf was part of the HVM feature set, the tool stack could be able to obtain the wanted information without altering sysctl (assuming the conditions to set the respective bits were the same). And I would view it as generally reasonable for there to be a way for tool stacks to know what hypervisor leaves guests are going to get to see (at the maximum and by default).
Quoted 1 line
Like the "cpuid" xtf test allows us to?
I don't think I understand the question. That xtf test is concerned about checking the CPUID output it gets to see itself. It doesn't care about what other guests might get to see, nor the maximum and default.
Quoted 1 line
Makes sense to me. I'm happy to take that up after.
"After" what?
Jan
Quoted 41 lines · Jan Beulich
[CAUTION - EXTERNAL EMAIL] DO NOT reply, click links, or open attachments unless you have verified the sender and know the content is safe.
Quoted 36 lines · Jane Malalane
Quoted 18 lines · Jan Beulich
Quoted 8 lines · Jane Malalane
Add
XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapicandXEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apicto report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.For that purpose, also add an arch-specific "capabilities" parameter to struct
xen_sysctl_physinfo.Quoted 8 lines
Suggested-by Andrew Cooper <[email protected]>Signed-off-by Jane Malalane <[email protected]>---
v3:
* DefineXEN_SYSCTL_PHYSCAP_ARCH_MAXfor ABI checking and actually setarch_capbilities, via a call toc_bitmap_to_ocaml_list()
* Haveassisted_x2apic_availableonly depend oncpu_has_vmx_virtualize_x2apic_modeQuoted 6 lines
I understand this was the result from previous discussion, but this needs justifying in the description. Not the least because it differs from when
XEN_HVM_CPUID_X2APIC_VIRTwould be set as well as from whatvmx_vlapic_msr_changed()does. The difference between those two is probably intended (judging from a comment there), but the further difference to what you add isn't obvious.Okay, I will make that explicit.
Quoted 7 lines
Which raises another thought: If that hypervisor leaf was part of the HVM feature set, the tool stack could be able to obtain the wanted information without altering sysctl (assuming the conditions to set the respective bits were the same). And I would view it as generally reasonable for there to be a way for tool stacks to know what hypervisor leaves guests are going to get to see (at the maximum and by default).
Quoted 1 line
Like the "cpuid" xtf test allows us to?
Quoted 7 lines
I don't think I understand the question. That xtf test is concerned about checking the CPUID output it gets to see itself. It doesn't care about what other guests might get to see, nor the maximum and default.
Quoted 1 line
Makes sense to me. I'm happy to take that up after.
"After" what?
So I meant to say that I could add the Xen CPUID leaves (40000x...) to the policy so that toolstacks could know what hypervisor leaves guests are going to see - in a future patch, as this wouldn't just expose XEN_HVM_CPUID_APIC_ACCESS_VIRT and XEN_HVM_CPUID_X2APIC_VIRT (0x40000x04) but other features too.
But, at the same time, w.r.t. this patch in particular, using XEN_HVM_CPUID_APIC_ACCESS_VIRT and XEN_HVM_CPUID_X2APIC_VIRT to detect assisted APIC gives us less flexibility to add more fine grained controls in the future.
Thanks,
Jane.
Quoted 46 lines · Jane Malalane
Quoted 45 lines · Jan Beulich
[CAUTION - EXTERNAL EMAIL] DO NOT reply, click links, or open attachments unless you have verified the sender and know the content is safe.
Quoted 36 lines · Jane Malalane
Quoted 18 lines · Jan Beulich
Quoted 8 lines · Jane Malalane
Add
XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapicandXEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apicto report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.For that purpose, also add an arch-specific "capabilities" parameter to struct
xen_sysctl_physinfo.Quoted 8 lines
Suggested-by Andrew Cooper <[email protected]>Signed-off-by Jane Malalane <[email protected]>---
v3:
* DefineXEN_SYSCTL_PHYSCAP_ARCH_MAXfor ABI checking and actually setarch_capbilities, via a call toc_bitmap_to_ocaml_list()
* Haveassisted_x2apic_availableonly depend oncpu_has_vmx_virtualize_x2apic_modeQuoted 6 lines
I understand this was the result from previous discussion, but this needs justifying in the description. Not the least because it differs from when
XEN_HVM_CPUID_X2APIC_VIRTwould be set as well as from whatvmx_vlapic_msr_changed()does. The difference between those two is probably intended (judging from a comment there), but the further difference to what you add isn't obvious.Okay, I will make that explicit.
Quoted 7 lines
Which raises another thought: If that hypervisor leaf was part of the HVM feature set, the tool stack could be able to obtain the wanted information without altering sysctl (assuming the conditions to set the respective bits were the same). And I would view it as generally reasonable for there to be a way for tool stacks to know what hypervisor leaves guests are going to get to see (at the maximum and by default).
Quoted 1 line
Like the "cpuid" xtf test allows us to?
I don't think I understand the question. That xtf test is concerned about checking the CPUID output it gets to see itself. It doesn't care about what other guests might get to see, nor the maximum and default.
Quoted 8 lines
Quoted 3 lines
Quoted 1 line
Makes sense to me. I'm happy to take that up after.
"After" what?
So I meant to say that I could add the Xen CPUID leaves (40000x...) to the policy so that toolstacks could know what hypervisor leaves guests are going to see - in a future patch, as this wouldn't just expose
XEN_HVM_CPUID_APIC_ACCESS_VIRTandXEN_HVM_CPUID_X2APIC_VIRT(0x40000x04) but other features too.
But doing this in a future patch (i.e. subsequent to this one) would mean to first introduce the sysctl just to then rip it out again. Hence my desire to consider the alternative before we settle on the sysctl.
Quoted 4 lines
But, at the same time, w.r.t. this patch in particular, using
XEN_HVM_CPUID_APIC_ACCESS_VIRTandXEN_HVM_CPUID_X2APIC_VIRTto detect assisted APIC gives us less flexibility to add more fine grained controls in the future.
I'm afraid I can't follow: All I'm talking about is how to expose the same kind of information to the tool stack. I don't see how the mechanism chosen would limit flexibility going forward.
Jan
Quoted 41 lines · Jan Beulich
[CAUTION - EXTERNAL EMAIL] DO NOT reply, click links, or open attachments unless you have verified the sender and know the content is safe.
Quoted 36 lines · Jane Malalane
Quoted 18 lines · Jan Beulich
Quoted 8 lines · Jane Malalane
Add
XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapicandXEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apicto report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.For that purpose, also add an arch-specific "capabilities" parameter to struct
xen_sysctl_physinfo.Quoted 8 lines
Suggested-by Andrew Cooper <[email protected]>Signed-off-by Jane Malalane <[email protected]>---
v3:
* DefineXEN_SYSCTL_PHYSCAP_ARCH_MAXfor ABI checking and actually setarch_capbilities, via a call toc_bitmap_to_ocaml_list()
* Haveassisted_x2apic_availableonly depend oncpu_has_vmx_virtualize_x2apic_modeQuoted 6 lines
I understand this was the result from previous discussion, but this needs justifying in the description. Not the least because it differs from when
XEN_HVM_CPUID_X2APIC_VIRTwould be set as well as from whatvmx_vlapic_msr_changed()does. The difference between those two is probably intended (judging from a comment there), but the further difference to what you add isn't obvious.Okay, I will make that explicit.
Quoted 7 lines
Which raises another thought: If that hypervisor leaf was part of the HVM feature set, the tool stack could be able to obtain the wanted information without altering sysctl (assuming the conditions to set the respective bits were the same). And I would view it as generally reasonable for there to be a way for tool stacks to know what hypervisor leaves guests are going to get to see (at the maximum and by default).
Quoted 1 line
Like the "cpuid" xtf test allows us to?
Quoted 7 lines
I don't think I understand the question. That xtf test is concerned about checking the CPUID output it gets to see itself. It doesn't care about what other guests might get to see, nor the maximum and default.
Quoted 1 line
Makes sense to me. I'm happy to take that up after.
"After" what?
So I meant to say that I could add the Xen CPUID leaves (40000x...) to the policy so that toolstacks could know what hypervisor leaves guests are going to see - in a future patch, as this wouldn't just expose XEN_HVM_CPUID_APIC_ACCESS_VIRT and XEN_HVM_CPUID_X2APIC_VIRT (0x40000x04) but other features too.
But, at the same time, w.r.t. this patch in particular, using XEN_HVM_CPUID_APIC_ACCESS_VIRT and XEN_HVM_CPUID_X2APIC_VIRT to detect assisted APIC gives us less flexibility to add more fine grained controls in the future.
Thanks,
Jane.
Quoted 25 lines · Jan Beulich
Quoted 8 lines · Jane Malalane
Add
XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapicandXEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apicto report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.For that purpose, also add an arch-specific "capabilities" parameter to struct
xen_sysctl_physinfo.Quoted 8 lines
Suggested-by Andrew Cooper <[email protected]>Signed-off-by Jane Malalane <[email protected]>---
v3:
* DefineXEN_SYSCTL_PHYSCAP_ARCH_MAXfor ABI checking and actually setarch_capbilities, via a call toc_bitmap_to_ocaml_list()
* Haveassisted_x2apic_availableonly depend oncpu_has_vmx_virtualize_x2apic_modeI understand this was the result from previous discussion, but this needs justifying in the description. Not the least because it differs from when
XEN_HVM_CPUID_X2APIC_VIRTwould be set as well as from whatvmx_vlapic_msr_changed()does. The difference between those two is probably intended (judging from a comment there), but the further difference to what you add isn't obvious.
Quoted 7 lines
Which raises another thought: If that hypervisor leaf was part of the HVM feature set, the tool stack could be able to obtain the wanted information without altering sysctl (assuming the conditions to set the respective bits were the same). And I would view it as generally reasonable for there to be a way for tool stacks to know what hypervisor leaves guests are going to get to see (at the maximum and by default).
I'm not sure using CPUID would be appropriate for this. Those fields are supposed to be used by a guest to decide whether it should prefer the x{2}APIC over PV alternatives for certain operations (ie: IPIs for example), but the level of control we can provide with the sysctl is more fine grained.
The current proposal is limited to the exposure and control of the usage of APIC virtualization, but we could also expose availability and per-domain enablement of APIC register virtualization and posted interrupts.
Thanks, Roger.
Quoted 40 lines · Roger Pau Monné
Quoted 25 lines · Jan Beulich
Quoted 8 lines · Jane Malalane
Add
XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapicandXEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apicto report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.For that purpose, also add an arch-specific "capabilities" parameter to struct
xen_sysctl_physinfo.Quoted 8 lines
Suggested-by Andrew Cooper <[email protected]>Signed-off-by Jane Malalane <[email protected]>---
v3:
* DefineXEN_SYSCTL_PHYSCAP_ARCH_MAXfor ABI checking and actually setarch_capbilities, via a call toc_bitmap_to_ocaml_list()
* Haveassisted_x2apic_availableonly depend oncpu_has_vmx_virtualize_x2apic_modeI understand this was the result from previous discussion, but this needs justifying in the description. Not the least because it differs from when
XEN_HVM_CPUID_X2APIC_VIRTwould be set as well as from whatvmx_vlapic_msr_changed()does. The difference between those two is probably intended (judging from a comment there), but the further difference to what you add isn't obvious.Quoted 7 lines
Which raises another thought: If that hypervisor leaf was part of the HVM feature set, the tool stack could be able to obtain the wanted information without altering sysctl (assuming the conditions to set the respective bits were the same). And I would view it as generally reasonable for there to be a way for tool stacks to know what hypervisor leaves guests are going to get to see (at the maximum and by default).
I'm not sure using CPUID would be appropriate for this. Those fields are supposed to be used by a guest to decide whether it should prefer the x{2}APIC over PV alternatives for certain operations (ie: IPIs for example), but the level of control we can provide with the sysctl is more fine grained.
Quoted 4 lines
The current proposal is limited to the exposure and control of the usage of APIC virtualization, but we could also expose availability and per-domain enablement of APIC register virtualization and posted interrupts.
But then I would still like to avoid duplication of information exposure and expose through the featureset what can be exposed there and limit sysctl to what cannot be expressed otherwise.
Jan
Quoted 41 lines · Jan Beulich
Quoted 40 lines · Roger Pau Monné
Quoted 25 lines · Jan Beulich
Quoted 8 lines · Jane Malalane
Add
XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapicandXEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apicto report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.For that purpose, also add an arch-specific "capabilities" parameter to struct
xen_sysctl_physinfo.Quoted 8 lines
Suggested-by Andrew Cooper <[email protected]>Signed-off-by Jane Malalane <[email protected]>---
v3:
* DefineXEN_SYSCTL_PHYSCAP_ARCH_MAXfor ABI checking and actually setarch_capbilities, via a call toc_bitmap_to_ocaml_list()
* Haveassisted_x2apic_availableonly depend oncpu_has_vmx_virtualize_x2apic_modeI understand this was the result from previous discussion, but this needs justifying in the description. Not the least because it differs from when
XEN_HVM_CPUID_X2APIC_VIRTwould be set as well as from whatvmx_vlapic_msr_changed()does. The difference between those two is probably intended (judging from a comment there), but the further difference to what you add isn't obvious.Quoted 7 lines
Which raises another thought: If that hypervisor leaf was part of the HVM feature set, the tool stack could be able to obtain the wanted information without altering sysctl (assuming the conditions to set the respective bits were the same). And I would view it as generally reasonable for there to be a way for tool stacks to know what hypervisor leaves guests are going to get to see (at the maximum and by default).
I'm not sure using CPUID would be appropriate for this. Those fields are supposed to be used by a guest to decide whether it should prefer the x{2}APIC over PV alternatives for certain operations (ie: IPIs for example), but the level of control we can provide with the sysctl is more fine grained.
Quoted 8 lines
Quoted 4 lines
The current proposal is limited to the exposure and control of the usage of APIC virtualization, but we could also expose availability and per-domain enablement of APIC register virtualization and posted interrupts.
But then I would still like to avoid duplication of information exposure and expose through the featureset what can be exposed there and limit sysctl to what cannot be expressed otherwise.
So you would rather prefer to expose this information in a synthetic CPUID leaf?
I assume the duplication of information will depend on what we end up exposing with the sysctl interface, whether it's just support for SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES and SECONDARY_EXEC_VIRTUALIZE_X2APIC_MODE or there's more to it.
Thanks, Roger.
Quoted 47 lines · Roger Pau Monné
Quoted 46 lines · Jan Beulich
Quoted 44 lines · Roger Pau Monné
Quoted 25 lines · Jan Beulich
Quoted 8 lines · Jane Malalane
Add
XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapicandXEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apicto report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.For that purpose, also add an arch-specific "capabilities" parameter to struct
xen_sysctl_physinfo.Quoted 8 lines
Suggested-by Andrew Cooper <[email protected]>Signed-off-by Jane Malalane <[email protected]>---
v3:
* DefineXEN_SYSCTL_PHYSCAP_ARCH_MAXfor ABI checking and actually setarch_capbilities, via a call toc_bitmap_to_ocaml_list()
* Haveassisted_x2apic_availableonly depend oncpu_has_vmx_virtualize_x2apic_modeI understand this was the result from previous discussion, but this needs justifying in the description. Not the least because it differs from when
XEN_HVM_CPUID_X2APIC_VIRTwould be set as well as from whatvmx_vlapic_msr_changed()does. The difference between those two is probably intended (judging from a comment there), but the further difference to what you add isn't obvious.Quoted 7 lines
Which raises another thought: If that hypervisor leaf was part of the HVM feature set, the tool stack could be able to obtain the wanted information without altering sysctl (assuming the conditions to set the respective bits were the same). And I would view it as generally reasonable for there to be a way for tool stacks to know what hypervisor leaves guests are going to get to see (at the maximum and by default).
I'm not sure using CPUID would be appropriate for this. Those fields are supposed to be used by a guest to decide whether it should prefer the x{2}APIC over PV alternatives for certain operations (ie: IPIs for example), but the level of control we can provide with the sysctl is more fine grained.
The current proposal is limited to the exposure and control of the usage of APIC virtualization, but we could also expose availability and per-domain enablement of APIC register virtualization and posted interrupts.
Quoted 6 lines
Quoted 3 lines
But then I would still like to avoid duplication of information exposure and expose through the featureset what can be exposed there and limit sysctl to what cannot be expressed otherwise.
So you would rather prefer to expose this information in a synthetic CPUID leaf?
Depends on what you mean by "synthetic leaf". We already have a leaf. What I'm suggesting to consider to the give that hypervisor leaf a representation in the featureset.
Jan
Quoted 52 lines · Jan Beulich
Quoted 51 lines · Roger Pau Monné
Quoted 49 lines · Jan Beulich
Quoted 40 lines · Roger Pau Monné
Quoted 25 lines · Jan Beulich
Quoted 8 lines · Jane Malalane
Add
XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapicandXEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apicto report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.For that purpose, also add an arch-specific "capabilities" parameter to struct
xen_sysctl_physinfo.Quoted 8 lines
Suggested-by Andrew Cooper <[email protected]>Signed-off-by Jane Malalane <[email protected]>---
v3:
* DefineXEN_SYSCTL_PHYSCAP_ARCH_MAXfor ABI checking and actually setarch_capbilities, via a call toc_bitmap_to_ocaml_list()
* Haveassisted_x2apic_availableonly depend oncpu_has_vmx_virtualize_x2apic_modeI understand this was the result from previous discussion, but this needs justifying in the description. Not the least because it differs from when
XEN_HVM_CPUID_X2APIC_VIRTwould be set as well as from whatvmx_vlapic_msr_changed()does. The difference between those two is probably intended (judging from a comment there), but the further difference to what you add isn't obvious.Quoted 7 lines
Which raises another thought: If that hypervisor leaf was part of the HVM feature set, the tool stack could be able to obtain the wanted information without altering sysctl (assuming the conditions to set the respective bits were the same). And I would view it as generally reasonable for there to be a way for tool stacks to know what hypervisor leaves guests are going to get to see (at the maximum and by default).
I'm not sure using CPUID would be appropriate for this. Those fields are supposed to be used by a guest to decide whether it should prefer the x{2}APIC over PV alternatives for certain operations (ie: IPIs for example), but the level of control we can provide with the sysctl is more fine grained.
Quoted 4 lines
The current proposal is limited to the exposure and control of the usage of APIC virtualization, but we could also expose availability and per-domain enablement of APIC register virtualization and posted interrupts.
But then I would still like to avoid duplication of information exposure and expose through the featureset what can be exposed there and limit sysctl to what cannot be expressed otherwise.
Quoted 6 lines
Quoted 2 lines
So you would rather prefer to expose this information in a synthetic CPUID leaf?
Depends on what you mean by "synthetic leaf". We already have a leaf. What I'm suggesting to consider to the give that hypervisor leaf a representation in the featureset.
Hm, but then we won't be able to expose more fine grained controls,
ie: separate between basic APIC virtualization support, APIC register virtualization and interrupt virtualization. We would need to keep the meaning of XEN_HVM_CPUID_APIC_ACCESS_VIRT / XEN_HVM_CPUID_X2APIC_VIRT (and exposing more fine grained features to guests make no sense).
Thanks, Roger.
Quoted 60 lines · Roger Pau Monné
Quoted 52 lines · Jan Beulich
Quoted 51 lines · Roger Pau Monné
Quoted 49 lines · Jan Beulich
Quoted 40 lines · Roger Pau Monné
Quoted 25 lines · Jan Beulich
Quoted 8 lines · Jane Malalane
Add
XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapicandXEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apicto report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.For that purpose, also add an arch-specific "capabilities" parameter to struct
xen_sysctl_physinfo.Quoted 8 lines
Suggested-by Andrew Cooper <[email protected]>Signed-off-by Jane Malalane <[email protected]>---
v3:
* DefineXEN_SYSCTL_PHYSCAP_ARCH_MAXfor ABI checking and actually setarch_capbilities, via a call toc_bitmap_to_ocaml_list()
* Haveassisted_x2apic_availableonly depend oncpu_has_vmx_virtualize_x2apic_modeI understand this was the result from previous discussion, but this needs justifying in the description. Not the least because it differs from when
XEN_HVM_CPUID_X2APIC_VIRTwould be set as well as from whatvmx_vlapic_msr_changed()does. The difference between those two is probably intended (judging from a comment there), but the further difference to what you add isn't obvious.Quoted 7 lines
Which raises another thought: If that hypervisor leaf was part of the HVM feature set, the tool stack could be able to obtain the wanted information without altering sysctl (assuming the conditions to set the respective bits were the same). And I would view it as generally reasonable for there to be a way for tool stacks to know what hypervisor leaves guests are going to get to see (at the maximum and by default).
I'm not sure using CPUID would be appropriate for this. Those fields are supposed to be used by a guest to decide whether it should prefer the x{2}APIC over PV alternatives for certain operations (ie: IPIs for example), but the level of control we can provide with the sysctl is more fine grained.
Quoted 4 lines
The current proposal is limited to the exposure and control of the usage of APIC virtualization, but we could also expose availability and per-domain enablement of APIC register virtualization and posted interrupts.
But then I would still like to avoid duplication of information exposure and expose through the featureset what can be exposed there and limit sysctl to what cannot be expressed otherwise.
Quoted 6 lines
Quoted 2 lines
So you would rather prefer to expose this information in a synthetic CPUID leaf?
Depends on what you mean by "synthetic leaf". We already have a leaf. What I'm suggesting to consider to the give that hypervisor leaf a representation in the featureset.
Quoted 5 lines
Hm, but then we won't be able to expose more fine grained controls,
ie: separate between basic APIC virtualization support, APIC register virtualization and interrupt virtualization. We would need to keep the meaning ofXEN_HVM_CPUID_APIC_ACCESS_VIRT/XEN_HVM_CPUID_X2APIC_VIRT(and exposing more fine grained features to guests make no sense).
I did say before that once (if ever) finer grained controls are wanted, a sysctl like suggested would indeed look to be the way to report the capability. But we aren't at that point.
Jan
Quoted 67 lines · Jan Beulich
Quoted 60 lines · Roger Pau Monné
Quoted 52 lines · Jan Beulich
Quoted 51 lines · Roger Pau Monné
Quoted 49 lines · Jan Beulich
Quoted 40 lines · Roger Pau Monné
Quoted 25 lines · Jan Beulich
Quoted 8 lines · Jane Malalane
Add
XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapicandXEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apicto report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.For that purpose, also add an arch-specific "capabilities" parameter to struct
xen_sysctl_physinfo.Quoted 8 lines
Suggested-by Andrew Cooper <[email protected]>Signed-off-by Jane Malalane <[email protected]>---
v3:
* DefineXEN_SYSCTL_PHYSCAP_ARCH_MAXfor ABI checking and actually setarch_capbilities, via a call toc_bitmap_to_ocaml_list()
* Haveassisted_x2apic_availableonly depend oncpu_has_vmx_virtualize_x2apic_modeI understand this was the result from previous discussion, but this needs justifying in the description. Not the least because it differs from when
XEN_HVM_CPUID_X2APIC_VIRTwould be set as well as from whatvmx_vlapic_msr_changed()does. The difference between those two is probably intended (judging from a comment there), but the further difference to what you add isn't obvious.Quoted 7 lines
Which raises another thought: If that hypervisor leaf was part of the HVM feature set, the tool stack could be able to obtain the wanted information without altering sysctl (assuming the conditions to set the respective bits were the same). And I would view it as generally reasonable for there to be a way for tool stacks to know what hypervisor leaves guests are going to get to see (at the maximum and by default).
I'm not sure using CPUID would be appropriate for this. Those fields are supposed to be used by a guest to decide whether it should prefer the x{2}APIC over PV alternatives for certain operations (ie: IPIs for example), but the level of control we can provide with the sysctl is more fine grained.
Quoted 4 lines
The current proposal is limited to the exposure and control of the usage of APIC virtualization, but we could also expose availability and per-domain enablement of APIC register virtualization and posted interrupts.
But then I would still like to avoid duplication of information exposure and expose through the featureset what can be exposed there and limit sysctl to what cannot be expressed otherwise.
Quoted 6 lines
Quoted 2 lines
So you would rather prefer to expose this information in a synthetic CPUID leaf?
Depends on what you mean by "synthetic leaf". We already have a leaf. What I'm suggesting to consider to the give that hypervisor leaf a representation in the featureset.
Quoted 5 lines
Hm, but then we won't be able to expose more fine grained controls,
ie: separate between basic APIC virtualization support, APIC register virtualization and interrupt virtualization. We would need to keep the meaning ofXEN_HVM_CPUID_APIC_ACCESS_VIRT/XEN_HVM_CPUID_X2APIC_VIRT(and exposing more fine grained features to guests make no sense).
Quoted 3 lines
I did say before that once (if ever) finer grained controls are wanted, a sysctl like suggested would indeed look to be the way to report the capability. But we aren't at that point.
So we would expose everything in the 0x40000000 range, and caller would figure out the position of the Xen leaves doing a signature check until the Xen leaf is found?
Would we represent the max policy as having Viridian enabled (so Xen leaves starting at 0x40000100)?
I would agree with this if the hypervisor leaves where already part of the managed CPUID data, but the work required to expose the leaves in the policy/featureset doesn't seem trivial. Making those leaves part of the policy will likely be done at some point, and then we can decide to drop the sysctl bits.
Thanks, Roger.
Quoted 68 lines · Roger Pau Monné
Quoted 67 lines · Jan Beulich
Quoted 65 lines · Roger Pau Monné
Quoted 52 lines · Jan Beulich
Quoted 51 lines · Roger Pau Monné
Quoted 49 lines · Jan Beulich
Quoted 40 lines · Roger Pau Monné
Quoted 25 lines · Jan Beulich
Quoted 8 lines · Jane Malalane
Add
XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapicandXEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apicto report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.For that purpose, also add an arch-specific "capabilities" parameter to struct
xen_sysctl_physinfo.Quoted 8 lines
Suggested-by Andrew Cooper <[email protected]>Signed-off-by Jane Malalane <[email protected]>---
v3:
* DefineXEN_SYSCTL_PHYSCAP_ARCH_MAXfor ABI checking and actually setarch_capbilities, via a call toc_bitmap_to_ocaml_list()
* Haveassisted_x2apic_availableonly depend oncpu_has_vmx_virtualize_x2apic_modeI understand this was the result from previous discussion, but this needs justifying in the description. Not the least because it differs from when
XEN_HVM_CPUID_X2APIC_VIRTwould be set as well as from whatvmx_vlapic_msr_changed()does. The difference between those two is probably intended (judging from a comment there), but the further difference to what you add isn't obvious.Quoted 7 lines
Which raises another thought: If that hypervisor leaf was part of the HVM feature set, the tool stack could be able to obtain the wanted information without altering sysctl (assuming the conditions to set the respective bits were the same). And I would view it as generally reasonable for there to be a way for tool stacks to know what hypervisor leaves guests are going to get to see (at the maximum and by default).
I'm not sure using CPUID would be appropriate for this. Those fields are supposed to be used by a guest to decide whether it should prefer the x{2}APIC over PV alternatives for certain operations (ie: IPIs for example), but the level of control we can provide with the sysctl is more fine grained.
Quoted 4 lines
The current proposal is limited to the exposure and control of the usage of APIC virtualization, but we could also expose availability and per-domain enablement of APIC register virtualization and posted interrupts.
But then I would still like to avoid duplication of information exposure and expose through the featureset what can be exposed there and limit sysctl to what cannot be expressed otherwise.
Quoted 6 lines
Quoted 2 lines
So you would rather prefer to expose this information in a synthetic CPUID leaf?
Depends on what you mean by "synthetic leaf". We already have a leaf. What I'm suggesting to consider to the give that hypervisor leaf a representation in the featureset.
Hm, but then we won't be able to expose more fine grained controls,
ie: separate between basic APIC virtualization support, APIC register virtualization and interrupt virtualization. We would need to keep the meaning ofXEN_HVM_CPUID_APIC_ACCESS_VIRT/XEN_HVM_CPUID_X2APIC_VIRT(and exposing more fine grained features to guests make no sense).
Quoted 7 lines
Quoted 3 lines
I did say before that once (if ever) finer grained controls are wanted, a sysctl like suggested would indeed look to be the way to report the capability. But we aren't at that point.
So we would expose everything in the
0x40000000range, and caller would figure out the position of the Xen leaves doing a signature check until the Xen leaf is found?
The leaf number doesn't matter. The FEATURESET_* constants are part of the ABI (just that the names we give them aren't in the public headers). There would simply be a new FEATURESET_x4a.
Quoted 5 lines
I would agree with this if the hypervisor leaves where already part of the managed CPUID data, but the work required to expose the leaves in the policy/featureset doesn't seem trivial. Making those leaves part of the policy will likely be done at some point, and then we can decide to drop the sysctl bits.
I may well be underestimating the amount of work involved. I wanted to put this up as a possible alternative. Seeing that it's not liked, I'm not going to insist to further pursue this.
Jan
Quoted 16 lines · Jane Malalane
Add
XEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_xapicandXEN_SYSCTL_PHYSCAP_ARCH_ASSISTED_x2apicto report accelerated xapic and x2apic, on x86 hardware.
No such features are currently implemented on AMD hardware.For that purpose, also add an arch-specific "capabilities" parameter to struct
xen_sysctl_physinfo.Suggested-by Andrew Cooper <[email protected]>Signed-off-by Jane Malalane <[email protected]>---
tools/include/libxl.h +0
-0
tools/include/libxl.h line 532
Quoted 5 lines
tools/include/libxl.h +2
-0@@ -528,6 +528,13 @@#define LIBXL_HAVE_MAX_GRANT_VERSION 1/* +* LIBXL_HAVE_PHYSINFO_ASSISTED_APIC indicates that libxl_physinfo has +* cap_assisted_x{2}apic fields, which indicates the availability of x{2}APIC
I think I'd rather have both cap_assisted_xapic and cap_assisted_x2apic spelled out in the comment as that would allow to grep for both string.
tools/include/libxl.h line 535
Quoted 3 lines
tools/include/libxl.h +3
-0+* hardware assisted virtualization. +*/ +#define LIBXL_HAVE_PHYSINFO_ASSISTED_APIC 1
Otherwise, tools/ side looks good.
Thanks,
Signature · 1 line
Anthony PERARD
PATCH v3 2/2 x86/xen: Allow per-domain usage of hardware virtualized APIC
In reply to Jane MalalaneIntroduce a new per-domain creation x86 specific flag to select whether hardware assisted virtualization should be used for x{2}APIC.
A per-domain option is added to xl in order to select the usage of x{2}APIC hardware assisted vitualization, as well as a global configuration option.
Having all APIC interaction exit to Xen for emulation is slow and can induce much overhead. Hardware can speed up x{2}APIC by decoding the APIC access and providing a VM exit with a more specific exit reason than a regular EPT fault or by altogether avoiding a VM exit.
On the other hand, being able to disable x{2}APIC hardware assisted vitualization can be useful for testing and debugging purposes.
Note: vmx_install_vlapic_mapping doesn't require modifications regardless of whether the guest has virtualize_apic_accesses enabled or not, i.e., setting the the APIC_ACCESS_ADDR VMCS field is fine so long as virtualize_apic_accesses is supported by the CPU.
---
CC: Wei Liu <[email protected]>
CC: Anthony PERARD <[email protected]>
CC: Juergen Gross <[email protected]>
CC: Andrew Cooper <[email protected]>
CC: George Dunlap <[email protected]>
CC: Jan Beulich <[email protected]>
CC: Julien Grall <[email protected]>
CC: Stefano Stabellini <[email protected]>
CC: Christian Lindig <[email protected]>
CC: David Scott <[email protected]>
CC: Volodymyr Babchuk <[email protected]>
CC: "Roger Pau Monné" <[email protected]>
v3:
* Change info in xl.cfg to better express reality and fix capitalization of x{2}apic
* Move "physinfo" variable definition to the beggining of libxl__domain_build_info_setdefault()
* Reposition brackets in if statement to match libxl coding style
* Shorten logic in libxl__arch_domain_build_info_setdefault()
* Correct dprintk message in arch_sanitise_domain_config()
* Make appropriate changes in vmx_vlapic_msr_changed() and cpuid_hypervisor_leaves() for amended "assisted_x2apic" bit
* Remove unneeded parantheses
v2:
* Add a LIBXL_HAVE_ASSISTED_APIC macro
* Pass xcpyshinfo as a pointer in libxl__arch_get_physinfo
* Add a return statement in now "int" libxl__arch_domain_build_info_setdefault()
* Preserve libxl__arch_domain_build_info_setdefault 's location in libxl_create.c
* Correct x{2}apic default setting logic in libxl__arch_domain_prepare_config()
* Correct logic for parsing assisted_x{2}apic host/guest options in xl_parse.c and initialize them to -1 in xl.c
* Use guest options directly in vmx_vlapic_msr_changed
* Fix indentation of bool assisted_x{2}apic in struct hvm_domain
* Add a change in xenctrl_stubs.c to pass xenctrl ABI checks
--- docs/man/xl.cfg.5.pod.in | 19 +++++++++++++++++++ docs/man/xl.conf.5.pod.in | 12 ++++++++++++ tools/golang/xenlight/helpers.gen.go | 12 ++++++++++++ tools/golang/xenlight/types.gen.go | 2 ++ tools/include/libxl.h | 7 +++++++ tools/libs/light/libxl_arch.h | 5 +++-- tools/libs/light/libxl_arm.c | 7 +++++-- tools/libs/light/libxl_create.c | 22 +++++++++++++--------- tools/libs/light/libxl_types.idl | 2 ++ tools/libs/light/libxl_x86.c | 28 ++++++++++++++++++++++++++-- tools/ocaml/libs/xc/xenctrl.ml | 2 ++ tools/ocaml/libs/xc/xenctrl.mli | 2 ++ tools/ocaml/libs/xc/xenctrl_stubs.c | 2 +- tools/xl/xl.c | 8 ++++++++ tools/xl/xl.h | 2 ++ tools/xl/xl_parse.c | 16 ++++++++++++++++ xen/arch/x86/domain.c | 28 +++++++++++++++++++++++++++- xen/arch/x86/hvm/vmx/vmcs.c | 4 ++++ xen/arch/x86/hvm/vmx/vmx.c | 8 ++++---- xen/arch/x86/include/asm/hvm/domain.h | 6 ++++++ xen/arch/x86/traps.c | 8 ++++---- xen/include/public/arch-x86/xen.h | 2 ++ 22 files changed, 179 insertions(+), 25 deletions(-)
docs/man/xl.cfg.5.pod.in +19 -0
@@ -1862,6 +1862,25 @@ firmware tables when using certain older guest Operating Systems. These tables have been superseded by newer constructs within the ACPI tables. +=item B<assisted_xapic=BOOLEAN> + +B<(x86 only)> Enables or disables hardware assisted virtualization for +xAPIC. With this option enabled, a memory-mapped APIC access will be +decoded by hardware and either issue a more specific VM exit than just +an EPT fault, or altogether avoid a VM exit. Notice full +virtualization for xAPIC can only be achieved if hardware supports +“APIC-register virtualization” and “virtual-interrupt delivery”. The +default is settable via L<xl.conf(5)>. + +=item B<assisted_x2apic=BOOLEAN> + +B<(x86 only)> Enables or disables hardware assisted virtualization for +x2APIC. With this option enabled, an MSR-Based APIC access will +either issue a VM exit or altogether avoid one. Notice full +virtualization for x2APIC can only be achieved if hardware supports +“APIC-register virtualization” and “virtual-interrupt delivery”. The +default is settable via L<xl.conf(5)>. + =item B<nx=BOOLEAN> B<(x86 only)> Hides or exposes the No-eXecute capability. This allows a guest
docs/man/xl.conf.5.pod.in +12 -0
@@ -107,6 +107,18 @@ Sets the default value for the C<max_grant_version> domain config value. Default: maximum grant version supported by the hypervisor. +=item B<assisted_xapic=BOOLEAN> + +If enabled, domains will use xAPIC hardware assisted virtualization by default. + +Default: enabled if supported. + +=item B<assisted_x2apic=BOOLEAN> + +If enabled, domains will use x2APIC hardware assisted virtualization by default. + +Default: enabled if supported. + =item B<vif.default.script="PATH"> Configures the default hotplug script used by virtual network devices.
tools/golang/xenlight/helpers.gen.go +12 -0
@@ -1120,6 +1120,12 @@ x.ArchArm.Vuart = VuartType(xc.arch_arm.vuart) if err := x.ArchX86.MsrRelaxed.fromC(&xc.arch_x86.msr_relaxed);err != nil { return fmt.Errorf("converting field ArchX86.MsrRelaxed: %v", err) } +if err := x.ArchX86.AssistedXapic.fromC(&xc.arch_x86.assisted_xapic);err != nil { +return fmt.Errorf("converting field ArchX86.AssistedXapic: %v", err) +} +if err := x.ArchX86.AssistedX2Apic.fromC(&xc.arch_x86.assisted_x2apic);err != nil { +return fmt.Errorf("converting field ArchX86.AssistedX2Apic: %v", err) +} x.Altp2M = Altp2MMode(xc.altp2m) x.VmtraceBufKb = int(xc.vmtrace_buf_kb) if err := x.Vpmu.fromC(&xc.vpmu);err != nil {
@@ -1605,6 +1611,12 @@ xc.arch_arm.vuart = C.libxl_vuart_type(x.ArchArm.Vuart) if err := x.ArchX86.MsrRelaxed.toC(&xc.arch_x86.msr_relaxed); err != nil { return fmt.Errorf("converting field ArchX86.MsrRelaxed: %v", err) } +if err := x.ArchX86.AssistedXapic.toC(&xc.arch_x86.assisted_xapic); err != nil { +return fmt.Errorf("converting field ArchX86.AssistedXapic: %v", err) +} +if err := x.ArchX86.AssistedX2Apic.toC(&xc.arch_x86.assisted_x2apic); err != nil { +return fmt.Errorf("converting field ArchX86.AssistedX2Apic: %v", err) +} xc.altp2m = C.libxl_altp2m_mode(x.Altp2M) xc.vmtrace_buf_kb = C.int(x.VmtraceBufKb) if err := x.Vpmu.toC(&xc.vpmu); err != nil {
tools/golang/xenlight/types.gen.go +2 -0
@@ -520,6 +520,8 @@ Vuart VuartType } ArchX86 struct { MsrRelaxed Defbool +AssistedXapic Defbool +AssistedX2Apic Defbool } Altp2M Altp2MMode VmtraceBufKb int
tools/include/libxl.h +7 -0
@@ -535,6 +535,13 @@#define LIBXL_HAVE_PHYSINFO_ASSISTED_APIC 1/* +* LIBXL_HAVE_ASSISTED_APIC indicates that libxl_domain_build_info has +* assisted_x{2}apic fields, for enabling hardware assisted virtualization for +* x{2}apic per domain. +*/ +#define LIBXL_HAVE_ASSISTED_APIC 1 + +/* * libxl ABI compatibility * * The only guarantee which libxl makes regarding ABI compatibility
tools/libs/light/libxl_arch.h +3 -2
@@ -71,8 +71,9 @@ void libxl__arch_domain_create_info_setdefault(libxl__gc *gc, libxl_domain_create_info *c_info);_hidden -void libxl__arch_domain_build_info_setdefault(libxl__gc *gc, -libxl_domain_build_info *b_info); +int libxl__arch_domain_build_info_setdefault(libxl__gc *gc, +libxl_domain_build_info *b_info, +const libxl_physinfo *physinfo); _hidden int libxl__arch_passthrough_mode_setdefault(libxl__gc *gc,
tools/libs/light/libxl_arm.c +5 -2
@@ -1384,8 +1384,9 @@ void libxl__arch_domain_create_info_setdefault(libxl__gc *gc, } } -void libxl__arch_domain_build_info_setdefault (libxl__gc *gc, -libxl_domain_build_info *b_info) +int libxl__arch_domain_build_info_setdefault (libxl__gc *gc, +libxl_domain_build_info *b_info, +const libxl_physinfo *physinfo) {/* ACPI is disabled by default */ libxl_defbool_setdefault (&b_info->acpi, false);
@@ -1399,6 +1400,8 @@ void libxl__arch_domain_build_info_setdefault(libxl__gc *gc,memset (&b_info->u,'\0' ,sizeof (b_info->u)); b_info->type =LIBXL_DOMAIN_TYPE_INVALID ;libxl_domain_build_info_init_type (b_info,LIBXL_DOMAIN_TYPE_PVH ); + +return 0 ; } int libxl__arch_passthrough_mode_setdefault(libxl__gc *gc,
tools/libs/light/libxl_create.c +13 -9
@@ -75,6 +75,7 @@ int libxl__domain_build_info_setdefault(libxl__gc *gc, libxl_domain_build_info *b_info) {int i, rc; +libxl_physinfo info; if (b_info->type != LIBXL_DOMAIN_TYPE_HVM && b_info->type != LIBXL_DOMAIN_TYPE_PV &&
@@ -264,7 +265,18 @@ int libxl__domain_build_info_setdefault(libxl__gc *gc,if (!b_info->event_channels) b_info->event_channels =1023 ; -libxl__arch_domain_build_info_setdefault (gc, b_info); + rc =libxl_get_physinfo (CTX , &info); +if (rc) { +LOG (ERROR ,"failed to get hypervisor info" ); +return rc; + } + + rc =libxl__arch_domain_build_info_setdefault (gc, b_info, &info); +if (rc) { +LOG (ERROR ,"unable to set domain arch build info defaults" ); +return rc; + } +libxl_defbool_setdefault (&b_info->dm_restrict, false);if (b_info->iommu_memkb ==LIBXL_MEMKB_DEFAULT )
@@ -457,14 +469,6 @@ int libxl__domain_build_info_setdefault(libxl__gc *gc, }if (b_info->max_grant_version ==LIBXL_MAX_GRANT_DEFAULT ) { -libxl_physinfo info; - - rc =libxl_get_physinfo (CTX , &info); -if (rc) { -LOG (ERROR ,"failed to get hypervisor info" ); -return rc; - } -if (info.cap_gnttab_v2) b_info->max_grant_version =2 ;else if (info.cap_gnttab_v1)
tools/libs/light/libxl_types.idl +2 -0
@@ -648,6 +648,8 @@ libxl_domain_build_info = Struct("domain_build_info",[ ("vuart", libxl_vuart_type), ])), ("arch_x86", Struct(None, [("msr_relaxed", libxl_defbool), + ("assisted_xapic", libxl_defbool), + ("assisted_x2apic", libxl_defbool), ])), # Alternate p2m is not bound to any architecture or guest type, as it is # supported by x86 HVM and ARM support is planned.
tools/libs/light/libxl_x86.c +26 -2
@@ -23,6 +23,14 @@ int libxl__arch_domain_prepare_config(libxl__gc *gc,if (libxl_defbool_val (d_config->b_info.arch_x86.msr_relaxed)) config->arch.misc_flags |=XEN_X86_MSR_RELAXED ; +if (d_config->c_info.type !=LIBXL_DOMAIN_TYPE_PV ) + { +if (libxl_defbool_val (d_config->b_info.arch_x86.assisted_xapic)) + config->arch.misc_flags |=XEN_X86_ASSISTED_XAPIC ; + +if (libxl_defbool_val (d_config->b_info.arch_x86.assisted_x2apic)) + config->arch.misc_flags |=XEN_X86_ASSISTED_X2APIC ; + }return 0 ; }
@@ -819,11 +827,27 @@ void libxl__arch_domain_create_info_setdefault(libxl__gc *gc, { } -void libxl__arch_domain_build_info_setdefault (libxl__gc *gc, -libxl_domain_build_info *b_info) +int libxl__arch_domain_build_info_setdefault (libxl__gc *gc, +libxl_domain_build_info *b_info, +const libxl_physinfo *physinfo) {libxl_defbool_setdefault (&b_info->acpi, true);libxl_defbool_setdefault (&b_info->arch_x86.msr_relaxed, false); + +if (b_info->type !=LIBXL_DOMAIN_TYPE_PV ) { +libxl_defbool_setdefault (&b_info->arch_x86.assisted_xapic, + physinfo->cap_assisted_xapic); +libxl_defbool_setdefault (&b_info->arch_x86.assisted_x2apic, + physinfo->cap_assisted_x2apic); + } + +else if (!libxl_defbool_is_default (b_info->arch_x86.assisted_xapic) || + !libxl_defbool_is_default (b_info->arch_x86.assisted_x2apic)) { +LOG (ERROR ,"Interrupt Controller Virtualization not supported for PV" ); +return ERROR_INVAL ; + } + +return 0 ; } int libxl__arch_passthrough_mode_setdefault(libxl__gc *gc,
tools/ocaml/libs/xc/xenctrl.ml +2 -0
@@ -50,6 +50,8 @@ type x86_arch_emulation_flags = type x86_arch_misc_flags = | X86_MSR_RELAXED + | X86_ASSISTED_XAPIC + | X86_ASSISTED_X2APIC type xen_x86_arch_domainconfig = {
tools/ocaml/libs/xc/xenctrl.mli +2 -0
@@ -44,6 +44,8 @@ type x86_arch_emulation_flags = type x86_arch_misc_flags = | X86_MSR_RELAXED + | X86_ASSISTED_XAPIC + | X86_ASSISTED_X2APIC type xen_x86_arch_domainconfig = { emulation_flags: x86_arch_emulation_flags list;
tools/ocaml/libs/xc/xenctrl_stubs.c +1 -1
@@ -239,7 +239,7 @@ CAMLprim value stub_xc_domain_create(value xch, value wanted_domid, value config cfg.arch.misc_flags =ocaml_list_to_c_bitmap /* ! x86_arch_misc_flags X86_ none */ -/* ! XEN_X86_ XEN_X86_MSR_RELAXED all */ +/* ! XEN_X86_ XEN_X86_ASSISTED_X2APIC max */ (VAL_MISC_FLAGS );#undef VAL_MISC_FLAGS
tools/xl/xl.c +8 -0
@@ -57,6 +57,8 @@ int max_grant_frames = -1;int max_maptrack_frames =-1 ;int max_grant_version =LIBXL_MAX_GRANT_DEFAULT ;libxl_domid domid_policy =INVALID_DOMID ; +int assisted_xapic =-1 ; +int assisted_x2apic =-1 ;xentoollog_level minmsglevel = minmsglevel_default;
@@ -201,6 +203,12 @@ static void parse_global_config(const char *configfile,if (!xlu_cfg_get_long (config,"claim_mode" , &l,0 )) claim_mode = l; +if (!xlu_cfg_get_long (config,"assisted_xapic" , &l,0 )) + assisted_xapic = l; + +if (!xlu_cfg_get_long (config,"assisted_x2apic" , &l,0 )) + assisted_x2apic = l; +xlu_cfg_replace_string (config,"remus.default.netbufscript" , &default_remus_netbufscript,0 ); xlu_cfg_replace_string (config, "colo.default.proxyscript",
tools/xl/xl.h +2 -0
@@ -286,6 +286,8 @@ extern libxl_bitmap global_vm_affinity_mask;extern libxl_bitmap global_hvm_affinity_mask;extern libxl_bitmap global_pv_affinity_mask;extern libxl_domid domid_policy; +extern int assisted_xapic; +extern int assisted_x2apic;enum output_format {OUTPUT_FORMAT_JSON ,
tools/xl/xl_parse.c +16 -0
@@ -1681,6 +1681,22 @@ void parse_config_data(const char *config_source,xlu_cfg_get_defbool (config,"vpt_align" , &b_info->u.hvm.vpt_align,0 );xlu_cfg_get_defbool (config,"apic" , &b_info->apic,0 ); + e =xlu_cfg_get_long (config,"assisted_xapic" , &l ,0 ); +if ((e ==ESRCH && assisted_xapic !=-1 ))/* use global default if present */ +libxl_defbool_set (&b_info->arch_x86.assisted_xapic, assisted_xapic); +else if (!e) +libxl_defbool_set (&b_info->arch_x86.assisted_xapic, l); +else +exit (1 ); + + e =xlu_cfg_get_long (config,"assisted_x2apic" , &l,0 ); +if ((e ==ESRCH && assisted_x2apic !=-1 ))/* use global default if present */ +libxl_defbool_set (&b_info->arch_x86.assisted_x2apic, assisted_x2apic); +else if (!e) +libxl_defbool_set (&b_info->arch_x86.assisted_x2apic, l); +else +exit (1 ); +switch (xlu_cfg_get_list (config,"viridian" , &viridian, &num_viridian,1 )) {
xen/arch/x86/domain.c +27 -1
@@ -619,6 +619,8 @@ int arch_sanitise_domain_config(struct xen_domctl_createdomain *config)bool hvm = config->flags & XEN_DOMCTL_CDF_hvm;bool hap = config->flags & XEN_DOMCTL_CDF_hap;bool nested_virt = config->flags & XEN_DOMCTL_CDF_nested_virt; +bool assisted_xapic = config->arch.misc_flags &XEN_X86_ASSISTED_XAPIC ; +bool assisted_x2apic = config->arch.misc_flags &XEN_X86_ASSISTED_X2APIC ;unsigned int max_vcpus;if ( hvm ? !hvm_enabled : !IS_ENABLED (CONFIG_PV ) )
@@ -685,13 +687,31 @@ int arch_sanitise_domain_config(struct xen_domctl_createdomain *config) } } -if ( config->arch.misc_flags & ~XEN_X86_MSR_RELAXED ) +if ( config->arch.misc_flags & ~(XEN_X86_MSR_RELAXED | +XEN_X86_ASSISTED_XAPIC | +XEN_X86_ASSISTED_X2APIC ) ) {dprintk (XENLOG_INFO ,"Invalid arch misc flags %#x\n" , config->arch.misc_flags);return -EINVAL ; } +if ( (assisted_xapic || assisted_x2apic) && !hvm ) + { +dprintk (XENLOG_INFO , +"Interrupt Controller Virtualization not supported for PV\n" ); +return -EINVAL ; + } + +if ( (assisted_xapic && !assisted_xapic_available) || + (assisted_x2apic && !assisted_x2apic_available) ) + { +dprintk (XENLOG_INFO , +"Hardware assisted x%sAPIC requested but not available\n" , + assisted_xapic && !assisted_xapic_available ?"" :"2" ); +return -EINVAL ; + } +return 0 ; }
@@ -864,6 +884,12 @@ int arch_domain_create(struct domain *d, d->arch.msr_relaxed = config->arch.misc_flags &XEN_X86_MSR_RELAXED ; + d->arch.hvm.assisted_xapic = + config->arch.misc_flags &XEN_X86_ASSISTED_XAPIC ; + + d->arch.hvm.assisted_x2apic = + config->arch.misc_flags &XEN_X86_ASSISTED_X2APIC ; +return 0 ; fail:
xen/arch/x86/hvm/vmx/vmcs.c +4 -0
@@ -1155,6 +1155,10 @@ static int construct_vmcs(struct vcpu *v)__vmwrite (PLE_WINDOW , ple_window); } +if ( !v->domain->arch.hvm.assisted_xapic ) + v->arch.hvm.vmx.secondary_exec_control &= + ~SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES ; +if ( cpu_has_vmx_secondary_exec_control )__vmwrite (SECONDARY_VM_EXEC_CONTROL , v->arch.hvm.vmx.secondary_exec_control);
xen/arch/x86/hvm/vmx/vmx.c +4 -4
@@ -3333,15 +3333,15 @@ static void vmx_install_vlapic_mapping(struct vcpu *v)void vmx_vlapic_msr_changed (struct vcpu *v) { -int virtualize_x2apic_mode; +bool virtualize_x2apic_mode;struct vlapic *vlapic =vcpu_vlapic (v);unsigned int msr; virtualize_x2apic_mode = ( (cpu_has_vmx_apic_reg_virt || cpu_has_vmx_virtual_intr_delivery) && - cpu_has_vmx_virtualize_x2apic_mode ); + v->domain->arch.hvm.assisted_x2apic ); -if ( !cpu_has_vmx_virtualize_apic_accesses && +if ( !v->domain->arch.hvm.assisted_xapic && !virtualize_x2apic_mode )return ;
@@ -3373,7 +3373,7 @@ void vmx_vlapic_msr_changed(struct vcpu *v)vmx_clear_msr_intercept (v,MSR_X2APIC_SELF ,VMX_MSR_W ); } } -else +else if ( v->domain->arch.hvm.assisted_xapic ) v->arch.hvm.vmx.secondary_exec_control |=SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES ; }
xen/arch/x86/include/asm/hvm/domain.h +6 -0
@@ -117,6 +117,12 @@ struct hvm_domain {bool is_s3_suspended; +/* xAPIC hardware assisted virtualization. */ +bool assisted_xapic; + +/* x2APIC hardware assisted virtualization. */ +bool assisted_x2apic; +/* hypervisor intercepted msix table */ struct list_head msixtbl_list;
xen/arch/x86/traps.c +4 -4
@@ -1115,7 +1115,8 @@ void cpuid_hypervisor_leaves(const struct vcpu *v, uint32_t leaf,if ( !is_hvm_domain (d) || subleaf !=0 )break ; -if ( cpu_has_vmx_apic_reg_virt ) +if ( cpu_has_vmx_apic_reg_virt && + v->domain->arch.hvm.assisted_xapic ) res->a |=XEN_HVM_CPUID_APIC_ACCESS_VIRT ;/*
@@ -1124,9 +1125,8 @@ void cpuid_hypervisor_leaves(const struct vcpu *v, uint32_t leaf,* and wrmsr in the guest will run without VMEXITs (see * vmx_vlapic_msr_changed()). */ -if ( cpu_has_vmx_virtualize_x2apic_mode && - cpu_has_vmx_apic_reg_virt && - cpu_has_vmx_virtual_intr_delivery ) +if ( cpu_has_vmx_apic_reg_virt && cpu_has_vmx_virtual_intr_delivery && + v->domain->arch.hvm.assisted_x2apic ) res->a |=XEN_HVM_CPUID_X2APIC_VIRT ;/*
xen/include/public/arch-x86/xen.h +2 -0
@@ -317,6 +317,8 @@ struct xen_arch_domainconfig {* doesn't allow the guest to read or write to the underlying MSR. */ #define XEN_X86_MSR_RELAXED (1u << 0) +#define XEN_X86_ASSISTED_XAPIC (1u << 1) +#define XEN_X86_ASSISTED_X2APIC (1u << 2)uint32_t misc_flags; };
Signature · 1 line
2.11.0
Re: [PATCH v3 2/2] x86/xen: Allow per-domain usage of hardware virtualized APIC
Quoted 5 lines · Jane Malalane
xen/arch/x86/hvm/vmx/vmx.c +0
-0@@ -3333,15 +3333,15 @@ static void vmx_install_vlapic_mapping(struct vcpu *v)void vmx_vlapic_msr_changed (struct vcpu *v)
xen/arch/x86/hvm/vmx/vmx.c static void vmx_install_vlapic_mapping(struct vcpu *v) · line 3342
Quoted 10 lines
xen/arch/x86/hvm/vmx/vmx.c +2
-2{ -int virtualize_x2apic_mode; +bool virtualize_x2apic_mode;struct vlapic *vlapic =vcpu_vlapic (v);unsigned int msr; virtualize_x2apic_mode = ( (cpu_has_vmx_apic_reg_virt || cpu_has_vmx_virtual_intr_delivery) && - cpu_has_vmx_virtualize_x2apic_mode ); + v->domain->arch.hvm.assisted_x2apic );
Following from my comment on patch 1, I'd expect this to become a simple assignment of v->domain->arch.hvm.assisted_x2apic (at which point the local variable could go away), just like ...
xen/arch/x86/hvm/vmx/vmx.c static void vmx_install_vlapic_mapping(struct vcpu *v) · line 3346
Quoted 4 lines
xen/arch/x86/hvm/vmx/vmx.c +1
-1-if ( !cpu_has_vmx_virtualize_apic_accesses && +if ( !v->domain->arch.hvm.assisted_xapic && !virtualize_x2apic_mode )return ;
... here.
xen/arch/x86/hvm/vmx/vmx.c void cpuid_hypervisor_leaves(const struct vcpu *v, uint32_t leaf, · line 1130
Quoted 10 lines
xen/arch/x86/hvm/vmx/vmx.c +2
-3@@ -1124,9 +1125,8 @@ void cpuid_hypervisor_leaves(const struct vcpu *v, uint32_t leaf,* and wrmsr in the guest will run without VMEXITs (see * vmx_vlapic_msr_changed()). */ -if ( cpu_has_vmx_virtualize_x2apic_mode && - cpu_has_vmx_apic_reg_virt && - cpu_has_vmx_virtual_intr_delivery ) +if ( cpu_has_vmx_apic_reg_virt && cpu_has_vmx_virtual_intr_delivery && + v->domain->arch.hvm.assisted_x2apic ) res->a |=XEN_HVM_CPUID_X2APIC_VIRT ;
While within the 80 cols limit, I think it would help readability if you kept it at one sub-condition per line.
Jan
Re: [PATCH v3 2/2] x86/xen: Allow per-domain usage of hardware virtualized APIC
Quoted 23 lines · Jan Beulich
[CAUTION - EXTERNAL EMAIL] DO NOT reply, click links, or open attachments unless you have verified the sender and know the content is safe.
Quoted 5 lines · Jane Malalane
xen/arch/x86/hvm/vmx/vmx.c +0
-0@@ -3333,15 +3333,15 @@ static void vmx_install_vlapic_mapping(struct vcpu *v)void vmx_vlapic_msr_changed (struct vcpu *v)xen/arch/x86/hvm/vmx/vmx.c static void vmx_install_vlapic_mapping(struct vcpu *v) · line 3342
Quoted 10 lines
xen/arch/x86/hvm/vmx/vmx.c +2
-2{ -int virtualize_x2apic_mode; +bool virtualize_x2apic_mode;struct vlapic *vlapic =vcpu_vlapic (v);unsigned int msr; virtualize_x2apic_mode = ( (cpu_has_vmx_apic_reg_virt || cpu_has_vmx_virtual_intr_delivery) && - cpu_has_vmx_virtualize_x2apic_mode ); + v->domain->arch.hvm.assisted_x2apic );Following from my comment on patch 1, I'd expect this to become a simple assignment of
v->domain->arch.hvm.assisted_x2apic(at which point the local variable could go away), just like ...
Quoted 6 lines
xen/arch/x86/hvm/vmx/vmx.c static void vmx_install_vlapic_mapping(struct vcpu *v) · line 3346
Quoted 4 lines
xen/arch/x86/hvm/vmx/vmx.c +1
-1-if ( !cpu_has_vmx_virtualize_apic_accesses && +if ( !v->domain->arch.hvm.assisted_xapic && !virtualize_x2apic_mode )return ;... here.
Previosuly we discussed setting v->domain->arch.hvm.assisted_xapic equal to only cpu_has_vmx_virtualize_x2apic_mode, that's why I have those additional checks as at least apic_reg_virt or virtual_intr_delivery are needed for the subsequent parts of this function. I might be misunderstanding your question.
Unless you mean that we should fallback to having v->domain->arch.hvm.assisted_xapic depend on those other features...?
Quoted 14 lines
xen/arch/x86/hvm/vmx/vmx.c void cpuid_hypervisor_leaves(const struct vcpu *v, uint32_t leaf, · line 1130
Quoted 11 lines
xen/arch/x86/hvm/vmx/vmx.c +2
-3@@ -1124,9 +1125,8 @@ void cpuid_hypervisor_leaves(const struct vcpu *v, uint32_t leaf,* and wrmsr in the guest will run without VMEXITs (see * vmx_vlapic_msr_changed()). */ -if ( cpu_has_vmx_virtualize_x2apic_mode && - cpu_has_vmx_apic_reg_virt && - cpu_has_vmx_virtual_intr_delivery ) +if ( cpu_has_vmx_apic_reg_virt && cpu_has_vmx_virtual_intr_delivery && + v->domain->arch.hvm.assisted_x2apic ) res->a |=XEN_HVM_CPUID_X2APIC_VIRT ;While within the 80 cols limit, I think it would help readability if you kept it at one sub-condition per line.
Sure.
Thank you,
Jane.
Re: [PATCH v3 2/2] x86/xen: Allow per-domain usage of hardware virtualized APIC
Quoted 29 lines · Jane Malalane
Quoted 28 lines · Jan Beulich
[CAUTION - EXTERNAL EMAIL] DO NOT reply, click links, or open attachments unless you have verified the sender and know the content is safe.
Quoted 5 lines · Jane Malalane
xen/arch/x86/hvm/vmx/vmx.c +0
-0@@ -3333,15 +3333,15 @@ static void vmx_install_vlapic_mapping(struct vcpu *v)void vmx_vlapic_msr_changed (struct vcpu *v)xen/arch/x86/hvm/vmx/vmx.c static void vmx_install_vlapic_mapping(struct vcpu *v) · line 3342
Quoted 10 lines
xen/arch/x86/hvm/vmx/vmx.c +2
-2{ -int virtualize_x2apic_mode; +bool virtualize_x2apic_mode;struct vlapic *vlapic =vcpu_vlapic (v);unsigned int msr; virtualize_x2apic_mode = ( (cpu_has_vmx_apic_reg_virt || cpu_has_vmx_virtual_intr_delivery) && - cpu_has_vmx_virtualize_x2apic_mode ); + v->domain->arch.hvm.assisted_x2apic );Following from my comment on patch 1, I'd expect this to become a simple assignment of
v->domain->arch.hvm.assisted_x2apic(at which point the local variable could go away), just like ...Quoted 4 lines
xen/arch/x86/hvm/vmx/vmx.c +1
-1-if ( !cpu_has_vmx_virtualize_apic_accesses && +if ( !v->domain->arch.hvm.assisted_xapic && !virtualize_x2apic_mode )return ;
Quoted 6 lines
Quoted 1 line
... here.
Previosuly we discussed setting
v->domain->arch.hvm.assisted_xapicequal to onlycpu_has_vmx_virtualize_x2apic_mode, that's why I have those additional checks as at leastapic_reg_virtorvirtual_intr_deliveryare needed for the subsequent parts of this function. I might be misunderstanding your question.
My expectation would have been that assisted_x2apic_available is assigned what is (in context above) assigned to virtualize_x2apic_mode (in patch 1). Anything deviating from this needs, I think, explaining there.
Quoted 2 lines
Unless you mean that we should fallback to having
v->domain->arch.hvm.assisted_xapicdepend on those other features...?
No, xapic is fine afaic.
Jan
Re: [PATCH v3 2/2] x86/xen: Allow per-domain usage of hardware virtualized APIC
Quoted 37 lines · Jan Beulich
Quoted 29 lines · Jane Malalane
Quoted 28 lines · Jan Beulich
[CAUTION - EXTERNAL EMAIL] DO NOT reply, click links, or open attachments unless you have verified the sender and know the content is safe.
Quoted 5 lines · Jane Malalane
xen/arch/x86/hvm/vmx/vmx.c +0
-0@@ -3333,15 +3333,15 @@ static void vmx_install_vlapic_mapping(struct vcpu *v)void vmx_vlapic_msr_changed (struct vcpu *v)xen/arch/x86/hvm/vmx/vmx.c static void vmx_install_vlapic_mapping(struct vcpu *v) · line 3342
Quoted 10 lines
xen/arch/x86/hvm/vmx/vmx.c +2
-2{ -int virtualize_x2apic_mode; +bool virtualize_x2apic_mode;struct vlapic *vlapic =vcpu_vlapic (v);unsigned int msr; virtualize_x2apic_mode = ( (cpu_has_vmx_apic_reg_virt || cpu_has_vmx_virtual_intr_delivery) && - cpu_has_vmx_virtualize_x2apic_mode ); + v->domain->arch.hvm.assisted_x2apic );Following from my comment on patch 1, I'd expect this to become a simple assignment of
v->domain->arch.hvm.assisted_x2apic(at which point the local variable could go away), just like ...Quoted 4 lines
xen/arch/x86/hvm/vmx/vmx.c +1
-1-if ( !cpu_has_vmx_virtualize_apic_accesses && +if ( !v->domain->arch.hvm.assisted_xapic && !virtualize_x2apic_mode )return ;Quoted 6 lines
Quoted 1 line
... here.
Previosuly we discussed setting
v->domain->arch.hvm.assisted_xapicequal to onlycpu_has_vmx_virtualize_x2apic_mode, that's why I have those additional checks as at leastapic_reg_virtorvirtual_intr_deliveryare needed for the subsequent parts of this function. I might be misunderstanding your question.
Quoted 3 lines
My expectation would have been that
assisted_x2apic_availableis assigned what is (in context above) assigned tovirtualize_x2apic_mode(in patch 1). Anything deviating from this needs, I think, explaining there.
Oh, sorry, I kept thinking you meant cpu_has_... instead of the local variable and it was just all confusing me. Would it help if I didn't use a local variable at all, or changed the name then. This is really just a check that is done before executing the code below in the function.
Thanks,
Jane.
Re: [PATCH v3 2/2] x86/xen: Allow per-domain usage of hardware virtualized APIC
In reply to Jan BeulichQuoted 12 lines · Jan Beulich
Quoted 11 lines · Jane Malalane
xen/arch/x86/hvm/vmx/vmx.c +1
-1@@ -3333,15 +3333,15 @@ static void vmx_install_vlapic_mapping(struct vcpu *v)void vmx_vlapic_msr_changed (struct vcpu *v) { -int virtualize_x2apic_mode; +bool virtualize_x2apic_mode;struct vlapic *vlapic =vcpu_vlapic (v);unsigned int msr;
Quoted 8 lines
xen/arch/x86/hvm/vmx/vmx.c static void vmx_install_vlapic_mapping(struct vcpu *v) · line 3342
Quoted 4 lines
xen/arch/x86/hvm/vmx/vmx.c +1
-1virtualize_x2apic_mode = ( (cpu_has_vmx_apic_reg_virt || cpu_has_vmx_virtual_intr_delivery) && - cpu_has_vmx_virtualize_x2apic_mode ); + v->domain->arch.hvm.assisted_x2apic );Following from my comment on patch 1, I'd expect this to become a simple assignment of
v->domain->arch.hvm.assisted_x2apic(at which point the local variable could go away), just like ...
I think we want to keep assisted_x{2}apic mapped to SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES and SECONDARY_EXEC_VIRTUALIZE_X2APIC_MODE respectively, so that in the future we could add further controls for SECONDARY_EXEC_APIC_REGISTER_VIRT and interrupt delivery.
Thanks, Roger.
Re: [PATCH v3 2/2] x86/xen: Allow per-domain usage of hardware virtualized APIC
Quoted 22 lines · Roger Pau Monné
Quoted 12 lines · Jan Beulich
Quoted 11 lines · Jane Malalane
xen/arch/x86/hvm/vmx/vmx.c +1
-1@@ -3333,15 +3333,15 @@ static void vmx_install_vlapic_mapping(struct vcpu *v)void vmx_vlapic_msr_changed (struct vcpu *v) { -int virtualize_x2apic_mode; +bool virtualize_x2apic_mode;struct vlapic *vlapic =vcpu_vlapic (v);unsigned int msr;Quoted 8 lines
xen/arch/x86/hvm/vmx/vmx.c static void vmx_install_vlapic_mapping(struct vcpu *v) · line 3342
Quoted 4 lines
xen/arch/x86/hvm/vmx/vmx.c +1
-1virtualize_x2apic_mode = ( (cpu_has_vmx_apic_reg_virt || cpu_has_vmx_virtual_intr_delivery) && - cpu_has_vmx_virtualize_x2apic_mode ); + v->domain->arch.hvm.assisted_x2apic );Following from my comment on patch 1, I'd expect this to become a simple assignment of
v->domain->arch.hvm.assisted_x2apic(at which point the local variable could go away), just like ...
Quoted 5 lines
I think we want to keep assisted_x{2}apic mapped to
SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSESandSECONDARY_EXEC_VIRTUALIZE_X2APIC_MODErespectively, so that in the future we could add further controls forSECONDARY_EXEC_APIC_REGISTER_VIRTand interrupt delivery.
If we want to be able to control more (most?) VMX sub-features, it would seem to me as if this would better be modeled accordingly right away. At that point there would likely need to be VMX and SVM specific controls rather than general HVM ones. Plus then it might make sense to match bit assignments in our interface with that in the VT-x spec.
Jan
Re: [PATCH v3 2/2] x86/xen: Allow per-domain usage of hardware virtualized APIC
Quoted 29 lines · Jan Beulich
Quoted 22 lines · Roger Pau Monné
Quoted 12 lines · Jan Beulich
Quoted 11 lines · Jane Malalane
xen/arch/x86/hvm/vmx/vmx.c +1
-1@@ -3333,15 +3333,15 @@ static void vmx_install_vlapic_mapping(struct vcpu *v)void vmx_vlapic_msr_changed (struct vcpu *v) { -int virtualize_x2apic_mode; +bool virtualize_x2apic_mode;struct vlapic *vlapic =vcpu_vlapic (v);unsigned int msr;Quoted 8 lines
xen/arch/x86/hvm/vmx/vmx.c static void vmx_install_vlapic_mapping(struct vcpu *v) · line 3342
Quoted 4 lines
xen/arch/x86/hvm/vmx/vmx.c +1
-1virtualize_x2apic_mode = ( (cpu_has_vmx_apic_reg_virt || cpu_has_vmx_virtual_intr_delivery) && - cpu_has_vmx_virtualize_x2apic_mode ); + v->domain->arch.hvm.assisted_x2apic );Following from my comment on patch 1, I'd expect this to become a simple assignment of
v->domain->arch.hvm.assisted_x2apic(at which point the local variable could go away), just like ...Quoted 5 lines
I think we want to keep assisted_x{2}apic mapped to
SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSESandSECONDARY_EXEC_VIRTUALIZE_X2APIC_MODErespectively, so that in the future we could add further controls forSECONDARY_EXEC_APIC_REGISTER_VIRTand interrupt delivery.
Quoted 4 lines
If we want to be able to control more (most?) VMX sub-features, it would seem to me as if this would better be modeled accordingly right away. At that point there would likely need to be VMX and SVM specific controls rather than general HVM ones.
I would have to check the AMD interface for hardware APIC virtualization support, I'm not sure how different the control values are there.
Quoted 3 lines
Plus then it might
make sense to match bit assignments in our interface with that in the VT-x spec.
That could work for things in secondary_exec_control, but posted interrupts are controlled in pin based exec control, so we would need to expose two different fields? Not sure it's worth the extra effort to match bit positions with the spec (or maybe I'm not understanding this correctly).
Are you suggesting a (VMX) generic interface where the hypervisor exposes the raw vmx_secondary_exec_control and possibly vmx_pin_based_exec_control and let the toolstack play with it, setting in the VMCS what it gets back from the toolstack?
That would imply quite a rework of the code in order to detect enabled features based on domain specific VMX fields (instead of using the global vmx_{secondary,pin_based}_exec_control variables)
Thanks, Roger.
Re: [PATCH v3 2/2] x86/xen: Allow per-domain usage of hardware virtualized APIC
Quoted 49 lines · Roger Pau Monné
Quoted 29 lines · Jan Beulich
Quoted 22 lines · Roger Pau Monné
Quoted 12 lines · Jan Beulich
Quoted 11 lines · Jane Malalane
xen/arch/x86/hvm/vmx/vmx.c +1
-1@@ -3333,15 +3333,15 @@ static void vmx_install_vlapic_mapping(struct vcpu *v)void vmx_vlapic_msr_changed (struct vcpu *v) { -int virtualize_x2apic_mode; +bool virtualize_x2apic_mode;struct vlapic *vlapic =vcpu_vlapic (v);unsigned int msr;Quoted 8 lines
xen/arch/x86/hvm/vmx/vmx.c static void vmx_install_vlapic_mapping(struct vcpu *v) · line 3342
Quoted 4 lines
xen/arch/x86/hvm/vmx/vmx.c +1
-1virtualize_x2apic_mode = ( (cpu_has_vmx_apic_reg_virt || cpu_has_vmx_virtual_intr_delivery) && - cpu_has_vmx_virtualize_x2apic_mode ); + v->domain->arch.hvm.assisted_x2apic );Following from my comment on patch 1, I'd expect this to become a simple assignment of
v->domain->arch.hvm.assisted_x2apic(at which point the local variable could go away), just like ...Quoted 5 lines
I think we want to keep assisted_x{2}apic mapped to
SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSESandSECONDARY_EXEC_VIRTUALIZE_X2APIC_MODErespectively, so that in the future we could add further controls forSECONDARY_EXEC_APIC_REGISTER_VIRTand interrupt delivery.Quoted 4 lines
If we want to be able to control more (most?) VMX sub-features, it would seem to me as if this would better be modeled accordingly right away. At that point there would likely need to be VMX and SVM specific controls rather than general HVM ones.
I would have to check the AMD interface for hardware APIC virtualization support, I'm not sure how different the control values are there.
Quoted 3 lines
Plus then it might
make sense to match bit assignments in our interface with that in the VT-x spec.That could work for things in
secondary_exec_control, but posted interrupts are controlled in pin based exec control, so we would need to expose two different fields? Not sure it's worth the extra effort to match bit positions with the spec (or maybe I'm not understanding this correctly).
Quoted 4 lines
Are you suggesting a (VMX) generic interface where the hypervisor exposes the raw
vmx_secondary_exec_controland possiblyvmx_pin_based_exec_controland let the toolstack play with it, setting in the VMCS what it gets back from the toolstack?
Not necessarily all of them, but on a case by case basis. But _where_ a control bit would appear (if supported) would be well known up front. This wouldn't be very different from VMX'es "supports the 1-setting of" information provided via MSRs. The hypervisor would indicate which of the bits can be controlled on a per-guest basis.
Jan
Re: [PATCH v3 2/2] x86/xen: Allow per-domain usage of hardware virtualized APIC
In reply to Jane Malalanetools/include/libxl.h line 539
Quoted 10 lines · Jane Malalane
tools/include/libxl.h +2
-0@@ -535,6 +535,13 @@#define LIBXL_HAVE_PHYSINFO_ASSISTED_APIC 1/* +* LIBXL_HAVE_ASSISTED_APIC indicates that libxl_domain_build_info has +* assisted_x{2}apic fields, for enabling hardware assisted virtualization for
Could you spell out both "assisted_xapic and assisted_x2apic" as that would allow for grep to find both string.
tools/libs/light/libxl_arm.c void libxl__arch_domain_build_info_setdefault(libxl__gc *gc,
Quoted 22 lines
tools/include/libxl.h +5
-0+* x{2}apic per domain. +*/ +#define LIBXL_HAVE_ASSISTED_APIC 1 + +/* tools/libs/light/libxl_arm.c +3
-2@@ -1384,8 +1384,9 @@ void libxl__arch_domain_create_info_setdefault(libxl__gc *gc, } } -void libxl__arch_domain_build_info_setdefault (libxl__gc *gc, -libxl_domain_build_info *b_info) +int libxl__arch_domain_build_info_setdefault (libxl__gc *gc, +libxl_domain_build_info *b_info, +const libxl_physinfo *physinfo) {/* ACPI is disabled by default */ libxl_defbool_setdefault (&b_info->acpi, false);@@ -1399,6 +1400,8 @@ void libxl__arch_domain_build_info_setdefault(libxl__gc *gc,
There is another return in this function, which want to return 0 rather than void.
Quoted 23 lines
tools/libs/light/libxl_arm.c +2
-0memset (&b_info->u,'\0' ,sizeof (b_info->u)); b_info->type =LIBXL_DOMAIN_TYPE_INVALID ;libxl_domain_build_info_init_type (b_info,LIBXL_DOMAIN_TYPE_PVH ); + +return 0 ; } int libxl__arch_passthrough_mode_setdefault(libxl__gc *gc,tools/libs/light/libxl_x86.c +3
-2@@ -819,11 +827,27 @@ void libxl__arch_domain_create_info_setdefault(libxl__gc *gc, { } -void libxl__arch_domain_build_info_setdefault (libxl__gc *gc, -libxl_domain_build_info *b_info) +int libxl__arch_domain_build_info_setdefault (libxl__gc *gc, +libxl_domain_build_info *b_info, +const libxl_physinfo *physinfo) {libxl_defbool_setdefault (&b_info->acpi, true);
tools/libs/light/libxl_x86.c void libxl__arch_domain_create_info_setdefault(libxl__gc *gc, · line 844
Quoted 10 lines
tools/libs/light/libxl_x86.c +9
-0libxl_defbool_setdefault (&b_info->arch_x86.msr_relaxed, false); + +if (b_info->type !=LIBXL_DOMAIN_TYPE_PV ) { +libxl_defbool_setdefault (&b_info->arch_x86.assisted_xapic, + physinfo->cap_assisted_xapic); +libxl_defbool_setdefault (&b_info->arch_x86.assisted_x2apic, + physinfo->cap_assisted_x2apic); + } + + else if (!libxl_defbool_is_default(b_info->arch_x86.assisted_xapic) ||
This "else" needs to be on the same line as the "}" 2 lines above.
tools/libs/light/libxl_x86.c void libxl__arch_domain_create_info_setdefault(libxl__gc *gc, · line 850
Quoted 6 lines
tools/libs/light/libxl_x86.c +6
-0+ !libxl_defbool_is_default(b_info->arch_x86.assisted_x2apic)) { +LOG (ERROR ,"Interrupt Controller Virtualization not supported for PV" ); +return ERROR_INVAL ; + } + +return 0 ;
Thanks,
Signature · 1 line
Anthony PERARD
Re: [PATCH v3 2/2] x86/xen: Allow per-domain usage of hardware virtualized APIC
Quoted 8 lines · Anthony PERARD
Quoted 7 lines · Jane Malalane
tools/include/libxl.h +0
-0@@ -535,6 +535,13 @@#define LIBXL_HAVE_PHYSINFO_ASSISTED_APIC 1
Quoted 6 lines
tools/include/libxl.h line 539
Quoted 3 lines
tools/include/libxl.h +2
-0/* +* LIBXL_HAVE_ASSISTED_APIC indicates that libxl_domain_build_info has +* assisted_x{2}apic fields, for enabling hardware assisted virtualization for Could you spell out both "
assisted_xapicandassisted_x2apic" as that would allow for grep to find both string.
Will do (for both cases).
Quoted 54 lines
tools/libs/light/libxl_arm.c void libxl__arch_domain_build_info_setdefault(libxl__gc *gc,
Quoted 23 lines
tools/include/libxl.h +5
-0+* x{2}apic per domain. +*/ +#define LIBXL_HAVE_ASSISTED_APIC 1 + +/* tools/libs/light/libxl_arm.c +3
-2@@ -1384,8 +1384,9 @@ void libxl__arch_domain_create_info_setdefault(libxl__gc *gc, } } -void libxl__arch_domain_build_info_setdefault (libxl__gc *gc, -libxl_domain_build_info *b_info) +int libxl__arch_domain_build_info_setdefault (libxl__gc *gc, +libxl_domain_build_info *b_info, +const libxl_physinfo *physinfo) {/* ACPI is disabled by default */ libxl_defbool_setdefault (&b_info->acpi, false);@@ -1399,6 +1400,8 @@ void libxl__arch_domain_build_info_setdefault(libxl__gc *gc,There is another return in this function, which want to return 0 rather than void.
Quoted 27 lines
tools/libs/light/libxl_arm.c +2
-0memset (&b_info->u,'\0' ,sizeof (b_info->u)); b_info->type =LIBXL_DOMAIN_TYPE_INVALID ;libxl_domain_build_info_init_type (b_info,LIBXL_DOMAIN_TYPE_PVH ); + +return 0 ; } int libxl__arch_passthrough_mode_setdefault(libxl__gc *gc,tools/libs/light/libxl_x86.c +6
-2@@ -819,11 +827,27 @@ void libxl__arch_domain_create_info_setdefault(libxl__gc *gc, { } -void libxl__arch_domain_build_info_setdefault (libxl__gc *gc, -libxl_domain_build_info *b_info) +int libxl__arch_domain_build_info_setdefault (libxl__gc *gc, +libxl_domain_build_info *b_info, +const libxl_physinfo *physinfo) {libxl_defbool_setdefault (&b_info->acpi, true);libxl_defbool_setdefault (&b_info->arch_x86.msr_relaxed, false); + +if (b_info->type !=LIBXL_DOMAIN_TYPE_PV ) { + libxl_defbool_setdefault(&b_info->arch_x86.assisted_xapic,
Quoted 8 lines
tools/libs/light/libxl_x86.c void libxl__arch_domain_create_info_setdefault(libxl__gc *gc, · line 844
Quoted 6 lines
tools/libs/light/libxl_x86.c +6
-0+ physinfo->cap_assisted_xapic); +libxl_defbool_setdefault (&b_info->arch_x86.assisted_x2apic, + physinfo->cap_assisted_x2apic); + } + + else if (!libxl_defbool_is_default(b_info->arch_x86.assisted_xapic) ||This "else" needs to be on the same line as the "}" 2 lines above.
Okay.
Thank you,
Jane.