Port Xen to Hyper-V
v5 cover Port Xen to Hyper-V
Please see individual patches for changes
Wei Liu (8):
x86: introduce CONFIG_GUEST and move code
x86: add missing headers in hypercall.h
x86: drop hypervisor_cpuid_base
x86: introduce hypervisor framework
x86: rename hypervisor_{alloc,free}_unused_page
x86: switch xen guest implementation to use hypervisor framework
x86: be more verbose when running on a hypervisor
x86: introduce CONFIG_HYPERV and detection code
xen/arch/x86/Kconfig | 15 +++-
xen/arch/x86/Makefile | 2 +-
xen/arch/x86/guest/Makefile | 6 +-
xen/arch/x86/guest/hyperv/Makefile | 1 +
xen/arch/x86/guest/hyperv/hyperv.c | 54 +++++++++++++
xen/arch/x86/guest/hypervisor.c | 78 +++++++++++++++++++
xen/arch/x86/guest/xen/Makefile | 4 +
xen/arch/x86/guest/{ => xen}/hypercall_page.S | 0
xen/arch/x86/guest/{ => xen}/pvh-boot.c | 2 +-
xen/arch/x86/guest/{ => xen}/xen.c | 60 +++++++-------
xen/arch/x86/pv/shim.c | 6 +-
xen/arch/x86/setup.c | 7 +-
xen/include/asm-x86/guest.h | 2 +
xen/include/asm-x86/guest/hypercall.h | 4 +
xen/include/asm-x86/guest/hyperv.h | 43 ++++++++++
xen/include/asm-x86/guest/hypervisor.h | 62 +++++++++++++++
xen/include/asm-x86/guest/xen.h | 23 ++----
17 files changed, 312 insertions(+), 57 deletions(-)
create mode 100644 xen/arch/x86/guest/hyperv/Makefile
create mode 100644 xen/arch/x86/guest/hyperv/hyperv.c
create mode 100644 xen/arch/x86/guest/hypervisor.c
create mode 100644 xen/arch/x86/guest/xen/Makefile
rename xen/arch/x86/guest/{ => xen}/hypercall_page.S (100%)
rename xen/arch/x86/guest/{ => xen}/pvh-boot.c (99%)
rename xen/arch/x86/guest/{ => xen}/xen.c (94%)
create mode 100644 xen/include/asm-x86/guest/hyperv.h
create mode 100644 xen/include/asm-x86/guest/hypervisor.hSignature · 7 lines
2.20.1 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH v5 1/8 x86: introduce CONFIG_GUEST and move code
Xen is able to run as a guest on Xen. We plan to make it able to run on Hyper-V as well.
Introduce CONFIG_GUEST which is set to true if either running on Xen or Hyper-V is desired. Restructure code hierarchy for new code to come.
No functional change intended.
---
xen/arch/x86/Kconfig | 7 +++++--
xen/arch/x86/Makefile | 2 +-
xen/arch/x86/guest/Makefile | 5 +----
xen/arch/x86/guest/xen/Makefile | 4 ++++
xen/arch/x86/guest/{ => xen}/hypercall_page.S | 0
xen/arch/x86/guest/{ => xen}/pvh-boot.c | 0
xen/arch/x86/guest/{ => xen}/xen.c | 0
7 files changed, 11 insertions(+), 7 deletions(-)
create mode 100644 xen/arch/x86/guest/xen/Makefile
rename xen/arch/x86/guest/{ => xen}/hypercall_page.S (100%)
rename xen/arch/x86/guest/{ => xen}/pvh-boot.c (100%)
rename xen/arch/x86/guest/{ => xen}/xen.c (100%)xen/arch/x86/Kconfig +5 -2
@@ -161,9 +161,12 @@ config XEN_ALIGN_2M endchoice +config GUEST + bool + config XEN_GUEST - def_bool n - prompt "Xen Guest" + bool "Xen Guest" + select GUEST ---help--- Support for Xen detecting when it is running under Xen.
xen/arch/x86/Makefile +1 -1
@@ -1,7 +1,7 @@ subdir-y += acpi subdir-y += cpu subdir-y += genapic -subdir-$(CONFIG_XEN_GUEST) += guest +subdir-$(CONFIG_GUEST) += guest subdir-$(CONFIG_HVM) += hvm subdir-y += mm subdir-$(CONFIG_XENOPROF) += oprofile
xen/arch/x86/guest/Makefile +1 -4
@@ -1,4 +1 @@ -obj-y += hypercall_page.o -obj-y += xen.o - -obj-bin-$(CONFIG_PVH_GUEST) += pvh-boot.init.o +subdir-$(CONFIG_XEN_GUEST) += xen
xen/arch/x86/guest/xen/Makefile new file +4 -0
@@ -0,0 +1,4 @@ +obj-y += hypercall_page.o +obj-y += xen.o + +obj-bin-$(CONFIG_PVH_GUEST) += pvh-boot.init.o
xen/arch/x86/guest/xen/hypercall_page.S renamed from xen/arch/x86/guest/hypercall_page.S +0 -0
xen/arch/x86/guest/xen/pvh-boot.c renamed from xen/arch/x86/guest/pvh-boot.c +0 -0
xen/arch/x86/guest/xen/xen.c renamed from xen/arch/x86/guest/xen.c +0 -0
Signature · 7 lines
2.20.1 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH v5 2/8 x86: add missing headers in hypercall.h
In reply to Wei LiuInclude asm_defns.h because ASM_CALL_CONSTRAINT is defined there.
Include xen/lib.h because we need ASSERT_UNREACHABLE.
No functional change.
--- xen/include/asm-x86/guest/hypercall.h | 4 ++++ 1 file changed, 4 insertions(+)
xen/include/asm-x86/guest/hypercall.h +4 -0
@@ -23,6 +23,8 @@#include <xen/types.h> +#include <asm/asm_defns.h> +#include <public/xen.h> #include <public/sched.h> #include <public/hvm/hvm_op.h>
@@ -180,6 +182,8 @@ static inline long xen_hypercall_set_evtchn_upcall_vector(#else /* CONFIG_XEN_GUEST */ +#include <xen/lib.h> +#include <public/sched.h> static inline void xen_hypercall_console_write(
Signature · 7 lines
2.20.1 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH v5 3/8 x86: drop hypervisor_cpuid_base
In reply to Wei LiuThe only user is Xen specific code in PV shim. We can therefore export the variable directly.
Move __read_mostly to its standard place while at it.
--- xen/arch/x86/guest/xen/xen.c | 7 +------ xen/arch/x86/pv/shim.c | 2 +- xen/include/asm-x86/guest/xen.h | 2 +- 3 files changed, 3 insertions(+), 8 deletions(-)
xen/arch/x86/guest/xen/xen.c +1 -6
@@ -37,7 +37,7 @@bool __read_mostly xen_guest; -static __read_mostly uint32_t xen_cpuid_base; +uint32_t __read_mostly xen_cpuid_base;extern char hypercall_page[];static struct rangeset *mem;
@@ -297,11 +297,6 @@ int hypervisor_free_unused_page(mfn_t mfn)return rangeset_remove_range (mem,mfn_x (mfn),mfn_x (mfn)); } -uint32_t hypervisor_cpuid_base (void ) -{ -return xen_cpuid_base; -} -static void ap_resume (void *unused) {map_vcpuinfo ();
xen/arch/x86/pv/shim.c +1 -1
@@ -995,7 +995,7 @@ domid_t get_initial_domain_id(void)if ( !pv_shim )return 0 ; -cpuid (hypervisor_cpuid_base () +4 , &eax, &ebx, &ecx, &edx); +cpuid (xen_cpuid_base +4 , &eax, &ebx, &ecx, &edx);return (eax &XEN_HVM_CPUID_DOMID_PRESENT ) ? ecx :1 ; }
xen/include/asm-x86/guest/xen.h +1 -1
@@ -30,13 +30,13 @@extern bool xen_guest;extern bool pv_console; +extern uint32_t xen_cpuid_base;void probe_hypervisor (void );void hypervisor_setup (void );void hypervisor_ap_setup (void );int hypervisor_alloc_unused_page (mfn_t *mfn);int hypervisor_free_unused_page (mfn_t mfn); -uint32_t hypervisor_cpuid_base (void );void hypervisor_resume (void ); DECLARE_PER_CPU(unsigned int, vcpu_id);
Signature · 7 lines
2.20.1 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH v5 4/8 x86: introduce hypervisor framework
In reply to Wei LiuWe will soon implement Hyper-V support for Xen. Add a framework for that.
This requires moving some of the hypervisor_* functions from xen.h to hypervisor.h.
---
Changes in v5:
1. Make hypervisor_probe return const char *.
Changes in v4:
1. Add ASSERT_UNREACHABLE to stubs.
2. Move __read_mostly.
3. Return hops directly.
4. Drop Paul's review tag.
--- xen/arch/x86/guest/Makefile | 2 + xen/arch/x86/guest/hypervisor.c | 45 +++++++++++++++++++ xen/include/asm-x86/guest.h | 1 + xen/include/asm-x86/guest/hypervisor.h | 62 ++++++++++++++++++++++++++ xen/include/asm-x86/guest/xen.h | 12 ----- 5 files changed, 110 insertions(+), 12 deletions(-) create mode 100644 xen/arch/x86/guest/hypervisor.c create mode 100644 xen/include/asm-x86/guest/hypervisor.h
xen/arch/x86/guest/Makefile +2 -0
@@ -1 +1,3 @@ +obj-y += hypervisor.o + subdir-$(CONFIG_XEN_GUEST) += xen
xen/arch/x86/guest/hypervisor.c new file +45 -0
45 lines not on this page · raw message
xen/include/asm-x86/guest.h +1 -0
@@ -20,6 +20,7 @@#define __X86_GUEST_H__#include <asm/guest/hypercall.h> +#include <asm/guest/hypervisor.h> #include <asm/guest/pvh-boot.h> #include <asm/guest/xen.h> #include <asm/pv/shim.h>
xen/include/asm-x86/guest/hypervisor.h new file +62 -0
62 lines not on this page · raw message
xen/include/asm-x86/guest/xen.h +0 -12
@@ -33,11 +33,8 @@ extern bool pv_console;extern uint32_t xen_cpuid_base;void probe_hypervisor (void ); -void hypervisor_setup (void ); -void hypervisor_ap_setup (void );int hypervisor_alloc_unused_page (mfn_t *mfn);int hypervisor_free_unused_page (mfn_t mfn); -void hypervisor_resume (void ); DECLARE_PER_CPU(unsigned int, vcpu_id); DECLARE_PER_CPU(struct vcpu_info *, vcpu_info);
@@ -49,15 +46,6 @@ DECLARE_PER_CPU(struct vcpu_info *, vcpu_info);static inline void probe_hypervisor (void ) {} -static inline void hypervisor_setup (void ) -{ -ASSERT_UNREACHABLE (); -} -static inline void hypervisor_ap_setup (void ) -{ -ASSERT_UNREACHABLE (); -} -#endif /* CONFIG_XEN_GUEST */ #endif /* __X86_GUEST_XEN_H__ */
Signature · 7 lines
2.20.1 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v5 4/8] x86: introduce hypervisor framework
Quoted 7 lines · Wei Liu
We will soon implement Hyper-V support for Xen. Add a framework for that.
This requires moving some of the hypervisor_* functions from
xen.htohypervisor.h.Signed-off-by Wei Liu <[email protected]>
with one more adjustment (sorry for noticing only now):
Quoted 20 lines
xen/arch/x86/guest/hypervisor.c new file +17
-0@@ -0,0 +1,45 @@ +/****************************************************************************** +* arch/x86/guest/hypervisor.c +* +* Support for detecting and running under a hypervisor. +* +* This program is free software; you can redistribute it and/or modify +* it under the terms of the GNU General Public License as published by +* the Free Software Foundation; either version 2 of the License, or +* (at your option) any later version. +* +* This program is distributed in the hope that it will be useful, +* but WITHOUT ANY WARRANTY; without even the implied warranty of +* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +* GNU General Public License for more details. +* +* You should have received a copy of the GNU General Public License +* along with this program; If not, see <http://www.gnu.org/licenses/>.
xen/arch/x86/guest/hypervisor.c line 27
Quoted 10 lines
xen/arch/x86/guest/hypervisor.c new file +10
-0+* +* Copyright (c) 2019 Microsoft. +*/ +#include <xen/init.h> +#include <xen/types.h> + +#include <asm/cache.h> +#include <asm/guest/hypervisor.h> + +static const struct hypervisor_ops __read_mostly *ops;
The __read_mostly is misplaced - it's an attribute of the variable, not its type, and hence belongs after the * . It just so happens that the compiler is (still) relatively relaxed in what it accepts, but I think at least the gcc manual has a warning towards future more strict behavior.
Jan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v5 4/8] x86: introduce hypervisor framework
Quoted 10 lines · Jan Beulich
Quoted 7 lines · Wei Liu
We will soon implement Hyper-V support for Xen. Add a framework for that.
This requires moving some of the hypervisor_* functions from
xen.htohypervisor.h.Signed-off-by Wei Liu <[email protected]>Acked-by Jan Beulich <[email protected]>
[...]
Quoted 9 lines
xen/arch/x86/guest/hypervisor.c line 27
Quoted 3 lines
xen/arch/x86/guest/hypervisor.c new file +3
-0+#include <asm/guest/hypervisor.h> + +static const struct hypervisor_ops __read_mostly *ops;The
__read_mostlyis misplaced - it's an attribute of the variable, not its type, and hence belongs after the * . It just so happens that the compiler is (still) relatively relaxed in what it accepts, but I think at least the gcc manual has a warning towards future more strict behavior.
Fixed.
Wei.
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH v5 5/8 x86: rename hypervisor_{alloc, free}_unused_page
In reply to Wei LiuThey are used in Xen code only.
No functional change.
--- xen/arch/x86/guest/xen/xen.c | 6 +++--- xen/arch/x86/pv/shim.c | 4 ++-- xen/include/asm-x86/guest/xen.h | 4 ++-- 3 files changed, 7 insertions(+), 7 deletions(-)
xen/arch/x86/guest/xen/xen.c +3 -3
@@ -93,7 +93,7 @@ static void map_shared_info(void)unsigned int i;unsigned long rc; -if (hypervisor_alloc_unused_page (&mfn) ) +if (xg_alloc_unused_page (&mfn) )panic ("unable to reserve shared info memory page\n" ); xatp.gpfn =mfn_x (mfn);
@@ -280,7 +280,7 @@ void hypervisor_ap_setup(void)init_evtchn (); } -int hypervisor_alloc_unused_page (mfn_t *mfn) +int xg_alloc_unused_page (mfn_t *mfn) {unsigned long m;int rc;
@@ -292,7 +292,7 @@ int hypervisor_alloc_unused_page(mfn_t *mfn)return rc; } -int hypervisor_free_unused_page (mfn_t mfn) +int xg_free_unused_page (mfn_t mfn) {return rangeset_remove_range (mem,mfn_x (mfn),mfn_x (mfn)); }
xen/arch/x86/pv/shim.c +2 -2
@@ -742,7 +742,7 @@ static long pv_shim_grant_table_op(unsigned int cmd, };mfn_t mfn; - rc =hypervisor_alloc_unused_page (&mfn); + rc =xg_alloc_unused_page (&mfn);if ( rc ) { gprintk(XENLOG_ERR,
@@ -754,7 +754,7 @@ static long pv_shim_grant_table_op(unsigned int cmd, rc =xen_hypercall_memory_op (XENMEM_add_to_physmap, &xatp);if ( rc ) { -hypervisor_free_unused_page (mfn); +xg_free_unused_page (mfn);break ; }
xen/include/asm-x86/guest/xen.h +2 -2
@@ -33,8 +33,8 @@ extern bool pv_console;extern uint32_t xen_cpuid_base;void probe_hypervisor (void ); -int hypervisor_alloc_unused_page (mfn_t *mfn); -int hypervisor_free_unused_page (mfn_t mfn); +int xg_alloc_unused_page (mfn_t *mfn); +int xg_free_unused_page (mfn_t mfn); DECLARE_PER_CPU(unsigned int, vcpu_id); DECLARE_PER_CPU(struct vcpu_info *, vcpu_info);
Signature · 7 lines
2.20.1 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH v5 6/8 x86: switch xen guest implementation to use hypervisor framework
In reply to Wei Liu---
Changes in v5:
1. xen_probe -> xg_probe
2. Drop xen_ prefix for private functions
--- xen/arch/x86/guest/hypervisor.c | 27 +++++++++++++++++- xen/arch/x86/guest/xen/pvh-boot.c | 2 +- xen/arch/x86/guest/xen/xen.c | 47 ++++++++++++++++++------------- xen/arch/x86/setup.c | 2 +- xen/include/asm-x86/guest/xen.h | 5 ++-- 5 files changed, 59 insertions(+), 24 deletions(-)
xen/arch/x86/guest/hypervisor.c +26 -1
@@ -22,18 +22,43 @@#include <xen/types.h> #include <asm/cache.h> -#include <asm/guest/hypervisor.h> +#include <asm/guest.h> static const struct hypervisor_ops __read_mostly *ops;const char *__init hypervisor_probe (void ) { +if ( !cpu_has_hypervisor ) +return NULL ; + +if ( ops ) +return ops->name; + + ops =xg_probe ();if ( ops )return ops->name;return NULL ; } +void __init hypervisor_setup (void ) +{ +if ( ops && ops->setup ) + ops->setup (); +} + +void hypervisor_ap_setup (void ) +{ +if ( ops && ops->ap_setup ) + ops->ap_setup (); +} + +void hypervisor_resume (void ) +{ +if ( ops && ops->resume ) + ops->resume (); +} +/* * Local variables: * mode: C
xen/arch/x86/guest/xen/pvh-boot.c +1 -1
@@ -103,7 +103,7 @@ void __init pvh_init(multiboot_info_t **mbi, module_t **mod) {convert_pvh_info (mbi, mod); -probe_hypervisor (); +hypervisor_probe ();ASSERT (xen_guest);get_memory_map ();
xen/arch/x86/guest/xen/xen.c +28 -19
77 lines not on this page · raw message
xen/arch/x86/setup.c +1 -1
@@ -763,7 +763,7 @@ void __init noreturn __start_xen(unsigned long mbi_p)* allocing any xenheap structures wanted in lower memory. */ kexec_early_calculations (); -probe_hypervisor (); +hypervisor_probe ();parse_video_info ();
xen/include/asm-x86/guest/xen.h +3 -2
@@ -23,6 +23,7 @@#include <asm/e820.h> #include <asm/fixmap.h> +#include <asm/guest/hypervisor.h> #define XEN_shared_info ((struct shared_info *)fix_to_virt(FIX_XEN_SHARED_INFO))
@@ -32,7 +33,7 @@ extern bool xen_guest;extern bool pv_console;extern uint32_t xen_cpuid_base; -void probe_hypervisor (void ); +const struct hypervisor_ops *xg_probe (void );int xg_alloc_unused_page (mfn_t *mfn);int xg_free_unused_page (mfn_t mfn);
@@ -44,7 +45,7 @@ DECLARE_PER_CPU(struct vcpu_info *, vcpu_info);#define xen_guest 0#define pv_console 0 -static inline void probe_hypervisor (void ) {} +static inline const struct hypervisor_ops *xg_probe (void ) {return NULL ; }#endif /* CONFIG_XEN_GUEST */ #endif /* __X86_GUEST_XEN_H__ */
Signature · 7 lines
2.20.1 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v5 6/8] x86: switch xen guest implementation to use hypervisor framework
Quoted 1 line · Wei Liu
Signed-off-by Wei Liu <[email protected]>
again with one more remark:
xen/arch/x86/guest/xen/xen.c void hypervisor_resume(void) · line 313
Quoted 5 lines
xen/arch/x86/guest/xen/xen.c +1
-0@@ -326,6 +310,31 @@ void hypervisor_resume(void)pv_console_init (); } +static const struct hypervisor_ops xg_ops = {
Along with other static variable not having an xg_ prefix, this one could lose its one, too. But I'm not going to make this a requirement.
Jan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v5 6/8] x86: switch xen guest implementation to use hypervisor framework
Quoted 10 lines · Jan Beulich
Quoted 1 line · Wei Liu
Signed-off-by Wei Liu <[email protected]>Acked-by Jan Beulich <[email protected]>again with one more remark:
Quoted 4 lines
xen/arch/x86/guest/xen/xen.c +0
-0@@ -326,6 +310,31 @@ void hypervisor_resume(void)pv_console_init (); }
Quoted 5 lines
xen/arch/x86/guest/xen/xen.c void hypervisor_resume(void) · line 313
Quoted 1 line
xen/arch/x86/guest/xen/xen.c +1
-0+static const struct hypervisor_ops xg_ops = {Along with other static variable not having an
xg_prefix, this one could lose its one, too. But I'm not going to make this a requirement.
I will drop the xg_ prefix locally per your comment here.
Wei.
Quoted 2 lines
Jan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH v5 7/8 x86: be more verbose when running on a hypervisor
In reply to Wei LiuAlso replace reference to xen_guest.
---
Changes in v5:
1. Cache and use hypervisor name instead
--- xen/arch/x86/setup.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-)
xen/arch/x86/setup.c +5 -2
@@ -700,6 +700,7 @@ void __init noreturn __start_xen(unsigned long mbi_p) .max_grant_frames = -1, .max_maptrack_frames = -1, }; +const char *hypervisor_name;/* Critical region without IDT or TSS. Any fault is deadly! */
@@ -763,7 +764,7 @@ void __init noreturn __start_xen(unsigned long mbi_p)* allocing any xenheap structures wanted in lower memory. */ kexec_early_calculations (); -hypervisor_probe (); + hypervisor_name =hypervisor_probe ();parse_video_info ();
@@ -788,6 +789,8 @@ void __init noreturn __start_xen(unsigned long mbi_p)printk ("Command line: %s\n" , cmdline);printk ("Xen image load base address: %#lx\n" , xen_phys_start); +if ( hypervisor_name ) +printk ("Running on %s\n" , hypervisor_name);#ifdef CONFIG_VIDEO printk ("Video information:\n" );
@@ -1569,7 +1572,7 @@ void __init noreturn __start_xen(unsigned long mbi_p) max_cpus = nr_cpu_ids; } -if ( xen_guest ) +if ( hypervisor_name )hypervisor_setup ();/* Low mappings were only needed for some BIOS table parsing. */
Signature · 7 lines
2.20.1 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v5 7/8] x86: be more verbose when running on a hypervisor
Quoted 3 lines · Wei Liu
Also replace reference to
xen_guest.Signed-off-by Wei Liu <[email protected]>
However, ...
Quoted 11 lines
xen/arch/x86/setup.c +1
-0@@ -700,6 +700,7 @@ void __init noreturn __start_xen(unsigned long mbi_p) .max_grant_frames = -1, .max_maptrack_frames = -1, }; +const char *hypervisor_name;/* Critical region without IDT or TSS. Any fault is deadly! */
xen/arch/x86/setup.c void __init noreturn __start_xen(unsigned long mbi_p) · line 767
Quoted 5 lines
xen/arch/x86/setup.c +1
-1@@ -763,7 +764,7 @@ void __init noreturn __start_xen(unsigned long mbi_p)* allocing any xenheap structures wanted in lower memory. */ kexec_early_calculations (); -hypervisor_probe (); + hypervisor_name =hypervisor_probe ();
... you no longer calling this function multiple time, why does patch 4 still put in a respective guard?
Jan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v5 7/8] x86: be more verbose when running on a hypervisor
Quoted 6 lines · Jan Beulich
Quoted 3 lines · Wei Liu
Also replace reference to
xen_guest.Signed-off-by Wei Liu <[email protected]>Acked-by Jan Beulich <[email protected]>
Thanks.
Quoted 17 lines
However, ...
Quoted 14 lines
xen/arch/x86/setup.c +1
-0@@ -700,6 +700,7 @@ void __init noreturn __start_xen(unsigned long mbi_p) .max_grant_frames = -1, .max_maptrack_frames = -1, }; +const char *hypervisor_name;/* Critical region without IDT or TSS. Any fault is deadly! */ @@ -763,7 +764,7 @@ void __init noreturn __start_xen(unsigned long mbi_p)* allocing any xenheap structures wanted in lower memory. */ kexec_early_calculations ();
Quoted 5 lines
xen/arch/x86/setup.c void __init noreturn __start_xen(unsigned long mbi_p) · line 767
Quoted 2 lines
xen/arch/x86/setup.c +1
-1-hypervisor_probe (); + hypervisor_name =hypervisor_probe ();... you no longer calling this function multiple time, why does patch 4 still put in a respective guard?
Remnant from previous iterations.
I can submit a follow-up patch to drop that -- do really want to invalidate all the reviews and acks I got so far.
Wei.
Quoted 2 lines
Jan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v5 7/8] x86: be more verbose when running on a hypervisor
Quoted 33 lines · Wei Liu
Quoted 6 lines · Jan Beulich
Quoted 3 lines · Wei Liu
Also replace reference to
xen_guest.Signed-off-by Wei Liu <[email protected]>Acked-by Jan Beulich <[email protected]>Thanks.
Quoted 17 lines
However, ...
Quoted 14 lines
xen/arch/x86/setup.c +1
-0@@ -700,6 +700,7 @@ void __init noreturn __start_xen(unsigned long mbi_p) .max_grant_frames = -1, .max_maptrack_frames = -1, }; +const char *hypervisor_name;/* Critical region without IDT or TSS. Any fault is deadly! */ @@ -763,7 +764,7 @@ void __init noreturn __start_xen(unsigned long mbi_p)* allocing any xenheap structures wanted in lower memory. */ kexec_early_calculations ();Quoted 5 lines
xen/arch/x86/setup.c void __init noreturn __start_xen(unsigned long mbi_p) · line 767
Quoted 2 lines
xen/arch/x86/setup.c +1
-1-hypervisor_probe (); + hypervisor_name =hypervisor_probe ();... you no longer calling this function multiple time, why does patch 4 still put in a respective guard?
Quoted 4 lines
Remnant from previous iterations.
I can submit a follow-up patch to drop that -- do really want to invalidate all the reviews and acks I got so far.
According to my records patch 4 had no acks except mine, which you could keep with this change (in fact I was thinking of making it dependent upon the dropping of this leftover). Subsequent patches may only need re-basing, which doesn't imply dropping of any acks.
Jan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v5 7/8] x86: be more verbose when running on a hypervisor
Quoted 36 lines · Jan Beulich
Quoted 35 lines · Wei Liu
Quoted 6 lines · Jan Beulich
Quoted 3 lines · Wei Liu
Also replace reference to
xen_guest.Signed-off-by Wei Liu <[email protected]>Acked-by Jan Beulich <[email protected]>Thanks.
Quoted 17 lines
However, ...
Quoted 14 lines
xen/arch/x86/setup.c +1
-0@@ -700,6 +700,7 @@ void __init noreturn __start_xen(unsigned long mbi_p) .max_grant_frames = -1, .max_maptrack_frames = -1, }; +const char *hypervisor_name;/* Critical region without IDT or TSS. Any fault is deadly! */ @@ -763,7 +764,7 @@ void __init noreturn __start_xen(unsigned long mbi_p)* allocing any xenheap structures wanted in lower memory. */ kexec_early_calculations ();Quoted 5 lines
xen/arch/x86/setup.c void __init noreturn __start_xen(unsigned long mbi_p) · line 767
Quoted 2 lines
xen/arch/x86/setup.c +1
-1-hypervisor_probe (); + hypervisor_name =hypervisor_probe ();... you no longer calling this function multiple time, why does patch 4 still put in a respective guard?
Remnant from previous iterations.
Quoted 7 lines
Quoted 2 lines
I can submit a follow-up patch to drop that -- do really want to invalidate all the reviews and acks I got so far.
According to my records patch 4 had no acks except mine, which you could keep with this change (in fact I was thinking of making it dependent upon the dropping of this leftover). Subsequent patches may only need re-basing, which doesn't imply dropping of any acks.
OK. In that case, I will drop it locally. If that causes any substantial changes, I will post another version; otherwise I will just keep all the tags and push this series soon-ish.
How does that sound to you?
Wei.
Quoted 2 lines
Jan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v5 7/8] x86: be more verbose when running on a hypervisor
Quoted 45 lines · Wei Liu
Quoted 36 lines · Jan Beulich
Quoted 35 lines · Wei Liu
Quoted 6 lines · Jan Beulich
Quoted 3 lines · Wei Liu
Also replace reference to
xen_guest.Signed-off-by Wei Liu <[email protected]>Acked-by Jan Beulich <[email protected]>Thanks.
Quoted 17 lines
However, ...
Quoted 14 lines
xen/arch/x86/setup.c +1
-0@@ -700,6 +700,7 @@ void __init noreturn __start_xen(unsigned long mbi_p) .max_grant_frames = -1, .max_maptrack_frames = -1, }; +const char *hypervisor_name;/* Critical region without IDT or TSS. Any fault is deadly! */ @@ -763,7 +764,7 @@ void __init noreturn __start_xen(unsigned long mbi_p)* allocing any xenheap structures wanted in lower memory. */ kexec_early_calculations ();Quoted 5 lines
xen/arch/x86/setup.c void __init noreturn __start_xen(unsigned long mbi_p) · line 767
Quoted 2 lines
xen/arch/x86/setup.c +1
-1-hypervisor_probe (); + hypervisor_name =hypervisor_probe ();... you no longer calling this function multiple time, why does patch 4 still put in a respective guard?
Remnant from previous iterations.
Quoted 7 lines
Quoted 2 lines
I can submit a follow-up patch to drop that -- do really want to invalidate all the reviews and acks I got so far.
According to my records patch 4 had no acks except mine, which you could keep with this change (in fact I was thinking of making it dependent upon the dropping of this leftover). Subsequent patches may only need re-basing, which doesn't imply dropping of any acks.
Quoted 5 lines
OK. In that case, I will drop it locally. If that causes any substantial changes, I will post another version; otherwise I will just keep all the tags and push this series soon-ish.
How does that sound to you?
And it turns out it is indeed trivial. Dropping that hunk in patch 4 only requires a minor fixup to patch 6.
Wei.
Quoted 5 lines
Wei.
Quoted 2 lines
Jan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH v5 8/8 x86: introduce CONFIG_HYPERV and detection code
In reply to Wei LiuWe use the same code structure as we did for Xen.
As starters, detect Hyper-V in probe routine. More complex functionalities will be added later.
--- xen/arch/x86/Kconfig | 8 +++++ xen/arch/x86/guest/Makefile | 1 + xen/arch/x86/guest/hyperv/Makefile | 1 + xen/arch/x86/guest/hyperv/hyperv.c | 54 ++++++++++++++++++++++++++++++ xen/arch/x86/guest/hypervisor.c | 8 +++++ xen/include/asm-x86/guest.h | 1 + xen/include/asm-x86/guest/hyperv.h | 43 ++++++++++++++++++++++++ 7 files changed, 116 insertions(+) create mode 100644 xen/arch/x86/guest/hyperv/Makefile create mode 100644 xen/arch/x86/guest/hyperv/hyperv.c create mode 100644 xen/include/asm-x86/guest/hyperv.h
xen/arch/x86/Kconfig +8 -0
@@ -164,6 +164,14 @@ endchoice config GUEST bool +config HYPERV_GUEST + bool "Hyper-V Guest" + select GUEST + ---help--- + Support for Xen detecting when it is running under Hyper-V. + + If unsure, say N. + config XEN_GUEST bool "Xen Guest" select GUEST
xen/arch/x86/guest/Makefile +1 -0
@@ -1,3 +1,4 @@ obj-y += hypervisor.o +subdir-$(CONFIG_HYPERV_GUEST) += hyperv subdir-$(CONFIG_XEN_GUEST) += xen
xen/arch/x86/guest/hyperv/Makefile new file +1 -0
@@ -0,0 +1 @@ +obj-y += hyperv.o
xen/arch/x86/guest/hyperv/hyperv.c new file +54 -0
54 lines not on this page · raw message
xen/arch/x86/guest/hypervisor.c +8 -0
@@ -38,6 +38,14 @@ const char *__init hypervisor_probe(void)if ( ops )return ops->name; +/* +* Detection of Hyper-V must come after Xen to avoid false positive due +* to viridian support +*/ + ops =hyperv_probe (); +if ( ops ) +return ops->name; +return NULL ; }
xen/include/asm-x86/guest.h +1 -0
@@ -20,6 +20,7 @@#define __X86_GUEST_H__#include <asm/guest/hypercall.h> +#include <asm/guest/hyperv.h> #include <asm/guest/hypervisor.h> #include <asm/guest/pvh-boot.h> #include <asm/guest/xen.h>
xen/include/asm-x86/guest/hyperv.h new file +43 -0
43 lines not on this page · raw message
Signature · 7 lines
2.20.1 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH v5 8/8] x86: introduce CONFIG_HYPERV and detection code
On Sat, Nov 30, 2019 at 11:57:37AM +0000, Wei Liu wrote:
[...]
xen/arch/x86/guest/hyperv/hyperv.c line 25
Quoted 6 lines
+ */
xen/arch/x86/guest/hyperv/hyperv.c new file +5
-0+#include <xen/init.h> + +#include <asm/guest.h> + +static const struct hypervisor_ops hyperv_ops = {
Since xg_ops has lost its xg_ prefix, I also take the liberty to drop the hyperv_ prefix here to make things more consistent.
Wei.
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel