Add hvmops for setting and getting the suppress #VE bit
v5 cover Add hvmops for setting and getting the suppress #VE bit
As the code stands right now, after DomU has enabled #VE using HVMOP_altp2m_vcpu_enable_notify, all its pages have the #VE suppress bit cleared, generating #VEs for any EPT violation. There is currently no way to change the value of the #VE suppress bit for a page from a domain; it can only be done in Xen internally using ept_set_entry().
Following the discussion from
https://lists.xen.org/archives/html/xen-devel/2017-03/msg01312.html these patches introduce new hvmops for setting and getting this bit and thus
Adrian Pop (3):
x86/altp2m: Add a hvmop for setting the suppress #VE bit x86/altp2m: Add a hvmop for querying the suppress #VE bit
Vlad Ioan Topan (1):
x86/mm: Change default value for suppress #VE in set_mem_access()
tools/libxc/include/xenctrl.h | 4 ++ tools/libxc/xc_altp2m.c | 50 ++++++++++++++++ xen/arch/x86/hvm/hvm.c | 33 ++++++++++ xen/arch/x86/mm/mem_access.c | 103 +++++++++++++++++++++++++++++++- xen/include/public/hvm/hvm_op.h | 13 ++++ xen/include/xen/mem_access.h | 6 ++ 6 files changed, 207 insertions(+), 2 deletions(-)
Signature · 7 lines
2.18.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH v5 1/3 x86/mm: Change default value for suppress #VE in set_mem_access()
From: Vlad Ioan Topan <[email protected]>
The default value for the "suppress #VE" bit set by set_mem_access() currently depends on whether the call is made from the same domain (the bit is set when called from another domain and cleared if called from the same domain). This patch changes that behavior to inherit the old suppress #VE bit value if it is already set and to set it to 1 otherwise, which is safer and more reliable.
--- xen/arch/x86/mm/mem_access.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
xen/arch/x86/mm/mem_access.c +1 -2
@@ -284,8 +284,7 @@ int p2m_set_altp2m_mem_access(struct domain *d, struct p2m_domain *hp2m, } } -return ap2m->set_entry (ap2m, gfn, mfn,PAGE_ORDER_4K , t, a, - current->domain != d); +return ap2m->set_entry (ap2m, gfn, mfn,PAGE_ORDER_4K , t, a,-1 ); } static int set_mem_access(struct domain *d, struct p2m_domain *p2m,
Signature · 7 lines
2.18.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH v5 2/3 x86/altp2m: Add a hvmop for setting the suppress #VE bit
In reply to Adrian PopIntroduce a new hvmop, HVMOP_altp2m_set_suppress_ve, which allows a domain to change the value of the #VE suppress bit for a page.
Add a libxc wrapper for invoking this hvmop.
---
Changes in v5:
- remove the "set_" from struct xen_hvm_altp2m_set_suppress_ve
Changes in v4:
- fix a deadlock:
If p2m_set_suppress_ve() is called on invalid pages the code path wrongly returns without releasing the lock, resulting in a deadlock.
- remove the privileged domain check
Changes in v3:
- fix indentation (Wei Liu)
- use return values other than EINVAL where appropriate (Ian Beulich)
- remove the irrelevant comments from the xen_hvm_altp2m_set_suppress_ve struct (Ian Beulich)
- add comment for the suppress_ve field in the struct above (Ian Beulich)
- remove the typedef and DEFINE_XEN_GUEST_HANDLE for xen_hvm_altp2m_set_suppress_ve (Ian Beulich)
- use XSM_DM_PRIV check instead of domain->is_privileged (Ian Beulich)
Changes in v2:
- check if #VE has been enabled on the target domain (Tamas K Lengyel)
- check if the cpu has the #VE feature
- make the suppress_ve argument boolean (Jan Beulich)
- initialize only local variables that need initializing (Jan Beulich)
- use fewer local variables (Jan Beulich)
- fix indentation (Jan Beulich)
- remove unnecessary braces (Jan Beulich)
- use gfn_lock() instead of p2m_lock() in the non-altp2m case (Jan Beulich)
- merge patch #2 and patch #3 (Jan Beulich)
--- tools/libxc/include/xenctrl.h | 2 ++ tools/libxc/xc_altp2m.c | 24 ++++++++++++++ xen/arch/x86/hvm/hvm.c | 14 +++++++++ xen/arch/x86/mm/mem_access.c | 55 +++++++++++++++++++++++++++++++++ xen/include/public/hvm/hvm_op.h | 11 +++++++ xen/include/xen/mem_access.h | 3 ++ 6 files changed, 109 insertions(+)
tools/libxc/include/xenctrl.h +2 -0
@@ -1952,6 +1952,8 @@ int xc_altp2m_destroy_view(xc_interface *handle, uint32_t domid,/* Switch all vCPUs of the domain to the specified altp2m view */ int xc_altp2m_switch_to_view (xc_interface *handle,uint32_t domid,uint16_t view_id); +int xc_altp2m_set_suppress_ve (xc_interface *handle,uint32_t domid, +uint16_t view_id,xen_pfn_t gfn,bool sve);int xc_altp2m_set_mem_access (xc_interface *handle,uint32_t domid,uint16_t view_id,xen_pfn_t gfn,xenmem_access_t access);
tools/libxc/xc_altp2m.c +24 -0
@@ -163,6 +163,30 @@ int xc_altp2m_switch_to_view(xc_interface *handle, uint32_t domid,return rc; } +int xc_altp2m_set_suppress_ve (xc_interface *handle,uint32_t domid, +uint16_t view_id,xen_pfn_t gfn,bool sve) +{ +int rc; +DECLARE_HYPERCALL_BUFFER (xen_hvm_altp2m_op_t, arg); + + arg =xc_hypercall_buffer_alloc (handle, arg,sizeof (*arg)); +if ( arg ==NULL ) +return -1 ; + + arg->version =HVMOP_ALTP2M_INTERFACE_VERSION ; + arg->cmd = HVMOP_altp2m_set_suppress_ve; + arg->domain = domid; + arg->u.suppress_ve.view = view_id; + arg->u.suppress_ve.gfn = gfn; + arg->u.suppress_ve.suppress_ve = sve; + + rc =xencall2 (handle->xcall, __HYPERVISOR_hvm_op, HVMOP_altp2m, +HYPERCALL_BUFFER_AS_ARG (arg)); + +xc_hypercall_buffer_free (handle, arg); +return rc; +} +int xc_altp2m_set_mem_access (xc_interface *handle,uint32_t domid,uint16_t view_id,xen_pfn_t gfn,xenmem_access_t access)
xen/arch/x86/hvm/hvm.c +14 -0
@@ -4524,6 +4524,7 @@ static int do_altp2m_op(case HVMOP_altp2m_create_p2m:case HVMOP_altp2m_destroy_p2m:case HVMOP_altp2m_switch_p2m: +case HVMOP_altp2m_set_suppress_ve:case HVMOP_altp2m_set_mem_access:case HVMOP_altp2m_set_mem_access_multi:case HVMOP_altp2m_change_gfn:
@@ -4641,6 +4642,19 @@ static int do_altp2m_op( rc =p2m_switch_domain_altp2m_by_id (d, a.u.view.view);break ; +case HVMOP_altp2m_set_suppress_ve: +if ( a.u.suppress_ve.pad1 || a.u.suppress_ve.pad2 ) + rc = -EINVAL ; +else + { +gfn_t gfn =_gfn (a.u.set_mem_access.gfn); +unsigned int altp2m_idx = a.u.set_mem_access.view; +bool suppress_ve = a.u.suppress_ve.suppress_ve; + + rc =p2m_set_suppress_ve (d, gfn, suppress_ve, altp2m_idx); + } +break ; +case HVMOP_altp2m_set_mem_access:if ( a.u.set_mem_access.pad ) rc = -EINVAL ;
xen/arch/x86/mm/mem_access.c +55 -0
61 lines not on this page · raw message
xen/include/public/hvm/hvm_op.h +11 -0
@@ -38,6 +38,14 @@ struct xen_hvm_param {typedef struct xen_hvm_param xen_hvm_param_t ;DEFINE_XEN_GUEST_HANDLE (xen_hvm_param_t); +struct xen_hvm_altp2m_suppress_ve { +uint16_t view; +uint8_t suppress_ve;/* Boolean type. */ +uint8_t pad1; +uint32_t pad2; +uint64_t gfn; +}; +#if __XEN_INTERFACE_VERSION__ <0x00040900 /* Set the logical level of one of a domain's PCI INTx wires. */
@@ -296,6 +304,8 @@ struct xen_hvm_altp2m_op {#define HVMOP_altp2m_change_gfn 8/* Set access for an array of pages */ #define HVMOP_altp2m_set_mem_access_multi 9 +/* Set the "Suppress #VE" bit on a page */ +#define HVMOP_altp2m_set_suppress_ve 10domid_t domain;uint16_t pad1;uint32_t pad2;
@@ -306,6 +316,7 @@ struct xen_hvm_altp2m_op {struct xen_hvm_altp2m_set_mem_access set_mem_access;struct xen_hvm_altp2m_change_gfn change_gfn;struct xen_hvm_altp2m_set_mem_access_multi set_mem_access_multi; +struct xen_hvm_altp2m_suppress_ve suppress_ve;uint8_t pad[64 ]; } u; };
xen/include/xen/mem_access.h +3 -0
@@ -72,6 +72,9 @@ long p2m_set_mem_access_multi(struct domain *d, uint32_t nr, uint32_t start, uint32_t mask, unsigned int altp2m_idx); +int p2m_set_suppress_ve (struct domain *d,gfn_t gfn,bool suppress_ve, +unsigned int altp2m_idx); +/* * Get access type for a gfn. * If gfn == INVALID_GFN, gets the default access type.
Signature · 7 lines
2.18.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH v5 3/3 x86/altp2m: Add a hvmop for querying the suppress #VE bit
In reply to Adrian Pop--- tools/libxc/include/xenctrl.h | 2 ++ tools/libxc/xc_altp2m.c | 26 +++++++++++++++++++ xen/arch/x86/hvm/hvm.c | 19 ++++++++++++++ xen/arch/x86/mm/mem_access.c | 45 +++++++++++++++++++++++++++++++++ xen/include/public/hvm/hvm_op.h | 2 ++ xen/include/xen/mem_access.h | 3 +++ 6 files changed, 97 insertions(+)
tools/libxc/include/xenctrl.h +2 -0
@@ -1954,6 +1954,8 @@ int xc_altp2m_switch_to_view(xc_interface *handle, uint32_t domid, uint16_t view_id);int xc_altp2m_set_suppress_ve (xc_interface *handle,uint32_t domid,uint16_t view_id,xen_pfn_t gfn,bool sve); +int xc_altp2m_get_suppress_ve (xc_interface *handle,uint32_t domid, +uint16_t view_id,xen_pfn_t gfn,bool *sve);int xc_altp2m_set_mem_access (xc_interface *handle,uint32_t domid,uint16_t view_id,xen_pfn_t gfn,xenmem_access_t access);
tools/libxc/xc_altp2m.c +26 -0
@@ -163,6 +163,32 @@ int xc_altp2m_switch_to_view(xc_interface *handle, uint32_t domid,return rc; } +int xc_altp2m_get_suppress_ve (xc_interface *handle,uint32_t domid, +uint16_t view_id,xen_pfn_t gfn,bool *sve) +{ +int rc; +DECLARE_HYPERCALL_BUFFER (xen_hvm_altp2m_op_t, arg); + + arg =xc_hypercall_buffer_alloc (handle, arg,sizeof (*arg)); +if ( arg ==NULL ) +return -1 ; + + arg->version =HVMOP_ALTP2M_INTERFACE_VERSION ; + arg->cmd = HVMOP_altp2m_get_suppress_ve; + arg->domain = domid; + arg->u.suppress_ve.view = view_id; + arg->u.suppress_ve.gfn = gfn; + + rc =xencall2 (handle->xcall, __HYPERVISOR_hvm_op, HVMOP_altp2m, +HYPERCALL_BUFFER_AS_ARG (arg)); + +if ( !rc ) + *sve = arg->u.suppress_ve.suppress_ve; + +xc_hypercall_buffer_free (handle, arg); +return rc; +} +int xc_altp2m_set_suppress_ve (xc_interface *handle,uint32_t domid,uint16_t view_id,xen_pfn_t gfn,bool sve) {
xen/arch/x86/hvm/hvm.c +19 -0
@@ -4525,6 +4525,7 @@ static int do_altp2m_op(case HVMOP_altp2m_destroy_p2m:case HVMOP_altp2m_switch_p2m:case HVMOP_altp2m_set_suppress_ve: +case HVMOP_altp2m_get_suppress_ve:case HVMOP_altp2m_set_mem_access:case HVMOP_altp2m_set_mem_access_multi:case HVMOP_altp2m_change_gfn:
@@ -4655,6 +4656,24 @@ static int do_altp2m_op( }break ; +case HVMOP_altp2m_get_suppress_ve: +if ( a.u.suppress_ve.pad1 || a.u.suppress_ve.pad2 ) + rc = -EINVAL ; +else + { +gfn_t gfn =_gfn (a.u.suppress_ve.gfn); +unsigned int altp2m_idx = a.u.suppress_ve.view; +bool suppress_ve; + + rc =p2m_get_suppress_ve (d, gfn, &suppress_ve, altp2m_idx); +if ( !rc ) + { + a.u.suppress_ve.suppress_ve = suppress_ve; + rc =__copy_to_guest (arg, &a,1 ) ? -EFAULT :0 ; + } + } +break ; +case HVMOP_altp2m_set_mem_access:if ( a.u.set_mem_access.pad ) rc = -EINVAL ;
xen/arch/x86/mm/mem_access.c +45 -0
51 lines not on this page · raw message
xen/include/public/hvm/hvm_op.h +2 -0
@@ -306,6 +306,8 @@ struct xen_hvm_altp2m_op {#define HVMOP_altp2m_set_mem_access_multi 9/* Set the "Suppress #VE" bit on a page */ #define HVMOP_altp2m_set_suppress_ve 10 +/* Get the "Suppress #VE" bit of a page */ +#define HVMOP_altp2m_get_suppress_ve 11domid_t domain;uint16_t pad1;uint32_t pad2;
xen/include/xen/mem_access.h +3 -0
@@ -75,6 +75,9 @@ long p2m_set_mem_access_multi(struct domain *d,int p2m_set_suppress_ve (struct domain *d,gfn_t gfn,bool suppress_ve,unsigned int altp2m_idx); +int p2m_get_suppress_ve (struct domain *d,gfn_t gfn,bool *suppress_ve, +unsigned int altp2m_idx); +/* * Get access type for a gfn. * If gfn == INVALID_GFN, gets the default access type.
Signature · 7 lines
2.18.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 3/3] x86/altp2m: Add a hvmop for querying the suppress #VE bit
Quoted 2 lines · Adrian Pop
Signed-off-by Adrian Pop <[email protected]>
Quoted 180 lines
--- tools/libxc/include/xenctrl.h | 2 ++ tools/libxc/xc_altp2m.c | 26 +++++++++++++++++++ xen/arch/x86/hvm/hvm.c | 19 ++++++++++++++ xen/arch/x86/mm/mem_access.c | 45 +++++++++++++++++++++++++++++++++ xen/include/public/hvm/hvm_op.h | 2 ++ xen/include/xen/mem_access.h | 3 +++ 6 files changed, 97 insertions(+)tools/libxc/include/xenctrl.h +2
-0@@ -1954,6 +1954,8 @@ int xc_altp2m_switch_to_view(xc_interface *handle, uint32_t domid, uint16_t view_id);int xc_altp2m_set_suppress_ve (xc_interface *handle,uint32_t domid,uint16_t view_id,xen_pfn_t gfn,bool sve); +int xc_altp2m_get_suppress_ve (xc_interface *handle,uint32_t domid, +uint16_t view_id,xen_pfn_t gfn,bool *sve);int xc_altp2m_set_mem_access (xc_interface *handle,uint32_t domid,uint16_t view_id,xen_pfn_t gfn,xenmem_access_t access);tools/libxc/xc_altp2m.c +26
-0@@ -163,6 +163,32 @@ int xc_altp2m_switch_to_view(xc_interface *handle, uint32_t domid,return rc; } +int xc_altp2m_get_suppress_ve (xc_interface *handle,uint32_t domid, +uint16_t view_id,xen_pfn_t gfn,bool *sve) +{ +int rc; +DECLARE_HYPERCALL_BUFFER (xen_hvm_altp2m_op_t, arg); + + arg =xc_hypercall_buffer_alloc (handle, arg,sizeof (*arg)); +if ( arg ==NULL ) +return -1 ; + + arg->version =HVMOP_ALTP2M_INTERFACE_VERSION ; + arg->cmd = HVMOP_altp2m_get_suppress_ve; + arg->domain = domid; + arg->u.suppress_ve.view = view_id; + arg->u.suppress_ve.gfn = gfn; + + rc =xencall2 (handle->xcall, __HYPERVISOR_hvm_op, HVMOP_altp2m, +HYPERCALL_BUFFER_AS_ARG (arg)); + +if ( !rc ) + *sve = arg->u.suppress_ve.suppress_ve; + +xc_hypercall_buffer_free (handle, arg); +return rc; +} +int xc_altp2m_set_suppress_ve (xc_interface *handle,uint32_t domid,uint16_t view_id,xen_pfn_t gfn,bool sve) {xen/arch/x86/hvm/hvm.c +19
-0@@ -4525,6 +4525,7 @@ static int do_altp2m_op(case HVMOP_altp2m_destroy_p2m:case HVMOP_altp2m_switch_p2m:case HVMOP_altp2m_set_suppress_ve: +case HVMOP_altp2m_get_suppress_ve:case HVMOP_altp2m_set_mem_access:case HVMOP_altp2m_set_mem_access_multi:case HVMOP_altp2m_change_gfn:@@ -4655,6 +4656,24 @@ static int do_altp2m_op( }break ; +case HVMOP_altp2m_get_suppress_ve: +if ( a.u.suppress_ve.pad1 || a.u.suppress_ve.pad2 ) + rc = -EINVAL ; +else + { +gfn_t gfn =_gfn (a.u.suppress_ve.gfn); +unsigned int altp2m_idx = a.u.suppress_ve.view; +bool suppress_ve; + + rc =p2m_get_suppress_ve (d, gfn, &suppress_ve, altp2m_idx); +if ( !rc ) + { + a.u.suppress_ve.suppress_ve = suppress_ve; + rc =__copy_to_guest (arg, &a,1 ) ? -EFAULT :0 ; + } + } +break ; +case HVMOP_altp2m_set_mem_access:if ( a.u.set_mem_access.pad ) rc = -EINVAL ;xen/arch/x86/mm/mem_access.c +45
-0@@ -550,6 +550,51 @@ out:return rc; } +int p2m_get_suppress_ve (struct domain *d,gfn_t gfn,bool *suppress_ve, +unsigned int altp2m_idx) +{ +struct p2m_domain *host_p2m =p2m_get_hostp2m (d); +struct p2m_domain *ap2m =NULL ; +struct p2m_domain *p2m; +mfn_t mfn; +p2m_access_t a; +p2m_type_t t; + +if ( !cpu_has_vmx_virt_exceptions ) +return -EOPNOTSUPP ; + +/* #VE should be enabled for this vcpu. */ +if (gfn_eq (vcpu_altp2m (current).veinfo_gfn,INVALID_GFN ) ) +return -ENXIO ; + +if ( altp2m_idx >0 ) + { +if ( altp2m_idx >=MAX_ALTP2M || + d->arch.altp2m_eptp[altp2m_idx] ==mfn_x (INVALID_MFN ) ) +return -EINVAL ; + + p2m = ap2m = d->arch.altp2m_p2m[altp2m_idx]; + } +else + p2m = host_p2m; + +gfn_lock (host_p2m, gfn,0 ); + +if ( ap2m ) +p2m_lock (ap2m); + + mfn = p2m->get_entry (p2m, gfn, &t, &a,0 ,NULL , suppress_ve); +if ( !mfn_valid (mfn) ) +return -ESRCH ; + +if ( ap2m ) +p2m_unlock (ap2m); + +gfn_unlock (host_p2m, gfn,0 ); + +return 0 ; +} +/* * Local variables: * mode: C xen/include/public/hvm/hvm_op.h +2
-0@@ -306,6 +306,8 @@ struct xen_hvm_altp2m_op {#define HVMOP_altp2m_set_mem_access_multi 9/* Set the "Suppress #VE" bit on a page */ #define HVMOP_altp2m_set_suppress_ve 10 +/* Get the "Suppress #VE" bit of a page */ +#define HVMOP_altp2m_get_suppress_ve 11domid_t domain;uint16_t pad1;uint32_t pad2;xen/include/xen/mem_access.h +3
-1@@ -75,6 +75,9 @@ long p2m_set_mem_access_multi(struct domain *d,int p2m_set_suppress_ve (struct domain *d,gfn_t gfn,bool suppress_ve,unsigned int altp2m_idx); +int p2m_get_suppress_ve (struct domain *d,gfn_t gfn,bool *suppress_ve, +unsigned int altp2m_idx); +/* * Get access type for a gfn. * If gfn == INVALID_GFN, gets the default access type. --
Quoted 1 line
2.18.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 1/3] x86/mm: Change default value for suppress #VE in set_mem_access()
In reply to Adrian PopQuoted 21 lines · Adrian Pop
From: Vlad Ioan Topan <[email protected]>
The default value for the "suppress #VE" bit set by
set_mem_access()currently depends on whether the call is made from the same domain (the bit is set when called from another domain and cleared if called from the same domain). This patch changes that behavior to inherit the old suppress #VE bit value if it is already set and to set it to 1 otherwise, which is safer and more reliable.Signed-off-by Vlad Ioan Topan <[email protected]>Signed-off-by Adrian Pop <[email protected]>--- xen/arch/x86/mm/mem_access.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)xen/arch/x86/mm/mem_access.c +0
-0
xen/arch/x86/mm/mem_access.c int p2m_set_altp2m_mem_access(struct domain *d, struct p2m_domain *hp2m, · line 288
Quoted 5 lines
xen/arch/x86/mm/mem_access.c +0
-2@@ -284,8 +284,7 @@ int p2m_set_altp2m_mem_access(struct domain *d, struct p2m_domain *hp2m, } } -return ap2m->set_entry (ap2m, gfn, mfn,PAGE_ORDER_4K , t, a, - current->domain != d);
Could you add a comment here saying "inherit the old suppress #VE bit value if it is already set and to set it to 1 otherwise" to explain the meaning of the magic -1 value?
Thanks,
Tamas
Quoted 6 lines
xen/arch/x86/mm/mem_access.c +1
-1+return ap2m->set_entry (ap2m, gfn, mfn,PAGE_ORDER_4K , t, a,-1 ); } static int set_mem_access(struct domain *d, struct p2m_domain *p2m, --2.18.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 3/3] x86/altp2m: Add a hvmop for querying the suppress #VE bit
In reply to Tamas K LengyelQuoted 5 lines · Tamas K Lengyel
Quoted 2 lines · Adrian Pop
Signed-off-by Adrian Pop <[email protected]>Acked-by Tamas K Lengyel <[email protected]>
Thanks!
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 1/3] x86/mm: Change default value for suppress #VE in set_mem_access()
In reply to Tamas K LengyelQuoted 25 lines · Tamas K Lengyel
Quoted 24 lines · Adrian Pop
From: Vlad Ioan Topan <[email protected]>
The default value for the "suppress #VE" bit set by
set_mem_access()currently depends on whether the call is made from the same domain (the bit is set when called from another domain and cleared if called from the same domain). This patch changes that behavior to inherit the old suppress #VE bit value if it is already set and to set it to 1 otherwise, which is safer and more reliable.Signed-off-by Vlad Ioan Topan <[email protected]>Signed-off-by Adrian Pop <[email protected]>--- xen/arch/x86/mm/mem_access.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)xen/arch/x86/mm/mem_access.c +0
-0@@ -284,8 +284,7 @@ int p2m_set_altp2m_mem_access(struct domain *d, struct p2m_domain *hp2m, } }
Quoted 6 lines
xen/arch/x86/mm/mem_access.c int p2m_set_altp2m_mem_access(struct domain *d, struct p2m_domain *hp2m, · line 288
Quoted 2 lines
xen/arch/x86/mm/mem_access.c +0
-2-return ap2m->set_entry (ap2m, gfn, mfn,PAGE_ORDER_4K , t, a, - current->domain != d);Could you add a comment here saying "inherit the old suppress #VE bit value if it is already set and to set it to 1 otherwise" to explain the meaning of the magic -1 value?
Sure. Thank you!
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 3/3] x86/altp2m: Add a hvmop for querying the suppress #VE bit
In reply to Adrian PopQuoted 1 line · Adrian Pop
Signed-off-by Adrian Pop <[email protected]>
FAOD I'm expecting you to address Tamas' comment in patch 1 and resend.
Wei.
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 3/3] x86/altp2m: Add a hvmop for querying the suppress #VE bit
Quoted 6 lines · Wei Liu
Quoted 1 line · Adrian Pop
Signed-off-by Adrian Pop <[email protected]>Acked-by Wei Liu <[email protected]>FAOD I'm expecting you to address Tamas' comment in patch 1 and resend.
Hello Wei,
That has already been addressed in V6, and the patch is already in:
http://xenbits.xen.org/gitweb/?p=xen.git;a=commitdiff;h=eea4ec2b66dad87ec745778ab9f00e12ef0f2760
Thanks,
Razvan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 3/3] x86/altp2m: Add a hvmop for querying the suppress #VE bit
Quoted 6 lines · Razvan Cojocaru
Quoted 5 lines · Wei Liu
Quoted 1 line · Adrian Pop
Signed-off-by Adrian Pop <[email protected]>Acked-by Wei Liu <[email protected]>
Quoted 7 lines
Quoted 1 line
FAOD I'm expecting you to address Tamas' comment in patch 1 and resend.
Hello Wei,
That has already been addressed in V6, and the patch is already in:
http://xenbits.xen.org/gitweb/?p=xen.git;a=commitdiff;h=eea4ec2b66dad87ec745778ab9f00e12ef0f2760
Yep. Saw that yesterday and I committed the last two patches of v6.
Wei.
Quoted 4 lines
Thanks,
Razvan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 3/3] x86/altp2m: Add a hvmop for querying the suppress #VE bit
Quoted 9 lines · Wei Liu
Quoted 8 lines · Razvan Cojocaru
Quoted 6 lines · Wei Liu
Quoted 1 line · Adrian Pop
Signed-off-by Adrian Pop <[email protected]>Acked-by Wei Liu <[email protected]>FAOD I'm expecting you to address Tamas' comment in patch 1 and resend.
Quoted 7 lines
Quoted 5 lines
Hello Wei,
That has already been addressed in V6, and the patch is already in:
http://xenbits.xen.org/gitweb/?p=xen.git;a=commitdiff;h=eea4ec2b66dad87ec745778ab9f00e12ef0f2760
Yep. Saw that yesterday and I committed the last two patches of v6.
Actually those patches will need acks from George, too. I will wait until EOD to revert them. Sorry, my memory is completely wrong regarding the maintainership of altp2m code.
Wei.
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 2/3] x86/altp2m: Add a hvmop for setting the suppress #VE bit
In reply to Adrian PopQuoted 8 lines · Adrian Pop
Introduce a new hvmop,
HVMOP_altp2m_set_suppress_ve, which allows a domain to change the value of the #VE suppress bit for a page.Add a libxc wrapper for invoking this hvmop.
Signed-off-by Adrian Pop <[email protected]>Acked-by Wei Liu <[email protected]>Acked-by Tamas K Lengyel <[email protected]>
Sorry I haven't had a chance to weigh in on this one yet. A couple fo comments...
xen/arch/x86/mm/mem_access.c void arch_p2m_set_access_required(struct domain *d, bool access_required) · line 502
Quoted 13 lines
xen/arch/x86/mm/mem_access.c +5
-0@@ -495,6 +495,61 @@ void arch_p2m_set_access_required(struct domain *d, bool access_required) } } +/* +* Set/clear the #VE suppress bit for a page. Only available on VMX. +*/ +int p2m_set_suppress_ve (struct domain *d,gfn_t gfn,bool suppress_ve, +unsigned int altp2m_idx)
This should clearly be in p2m.c, and...
xen/arch/x86/mm/mem_access.c void arch_p2m_set_access_required(struct domain *d, bool access_required) · line 513
Quoted 11 lines
xen/arch/x86/mm/mem_access.c +11
-0+{ +struct p2m_domain *host_p2m =p2m_get_hostp2m (d); +struct p2m_domain *ap2m =NULL ; +struct p2m_domain *p2m; +mfn_t mfn; +p2m_access_t a; +p2m_type_t t; +int rc; + +if ( !cpu_has_vmx_virt_exceptions ) +return -EOPNOTSUPP ;
We should avoid Intel-specific checks in common code.
In fact, this is wrong, because you can choose to run a guest in shadow mode even on a system with virt exceptions -- in which case you'd trigger the ASSERT() in p2m-pt.c:p2m_pt_set_entry().
Probably what should happen is that we should move the vmx check into p2m-ept.c:p2m_ept_set_entry(), and replace the ASSERT(sve = 0) in p2m-pt.c:p2m_pt_set_entry() with "if ( sve != 0 ) return -ENOTSUPP;".
Although what should probably *really* happen is that HVMOP_altp2m_vcpu_enable_notify should fail with -EOPNOTSUPP instead of silently succeeding.
Everything else looks good.
I realize this has been checked in already; no need to revert if you can send a follow-up patch in the next couple of (business) days.
Thanks,
-George
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 2/3] x86/altp2m: Add a hvmop for setting the suppress #VE bit
Quoted 52 lines · George Dunlap
Quoted 8 lines · Adrian Pop
Introduce a new hvmop,
HVMOP_altp2m_set_suppress_ve, which allows a domain to change the value of the #VE suppress bit for a page.Add a libxc wrapper for invoking this hvmop.
Signed-off-by Adrian Pop <[email protected]>Acked-by Wei Liu <[email protected]>Acked-by Tamas K Lengyel <[email protected]>Sorry I haven't had a chance to weigh in on this one yet. A couple fo comments...
xen/arch/x86/mm/mem_access.c void arch_p2m_set_access_required(struct domain *d, bool access_required) · line 502
Quoted 13 lines
xen/arch/x86/mm/mem_access.c +5
-0@@ -495,6 +495,61 @@ void arch_p2m_set_access_required(struct domain *d, bool access_required) } } +/* +* Set/clear the #VE suppress bit for a page. Only available on VMX. +*/ +int p2m_set_suppress_ve (struct domain *d,gfn_t gfn,bool suppress_ve, +unsigned int altp2m_idx)This should clearly be in
p2m.c, and...xen/arch/x86/mm/mem_access.c void arch_p2m_set_access_required(struct domain *d, bool access_required) · line 513
Quoted 11 lines
xen/arch/x86/mm/mem_access.c +11
-0+{ +struct p2m_domain *host_p2m =p2m_get_hostp2m (d); +struct p2m_domain *ap2m =NULL ; +struct p2m_domain *p2m; +mfn_t mfn; +p2m_access_t a; +p2m_type_t t; +int rc; + +if ( !cpu_has_vmx_virt_exceptions ) +return -EOPNOTSUPP ;We should avoid Intel-specific checks in common code.
In fact, this is wrong, because you can choose to run a guest in shadow mode even on a system with virt exceptions -- in which case you'd trigger the
ASSERT()in p2m-pt.c:p2m_pt_set_entry().Probably what should happen is that we should move the vmx check into p2m-ept.c:p2m_ept_set_entry(), and replace the ASSERT(sve = 0) in p2m-pt.c:p2m_pt_set_entry() with "if ( sve != 0 ) return
-ENOTSUPP;".
Quoted 8 lines
Although what should probably *really* happen is that
HVMOP_altp2m_vcpu_enable_notifyshould fail with-EOPNOTSUPPinstead of silently succeeding.Everything else looks good.
I realize this has been checked in already; no need to revert if you can send a follow-up patch in the next couple of (business) days.
Thanks for the review! I'll submit a follow up patch as soon as possible.
Thanks,
Razvan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 3/3] x86/altp2m: Add a hvmop for querying the suppress #VE bit
In reply to Adrian PopQuoted 111 lines · Adrian Pop
Signed-off-by Adrian Pop <[email protected]>--- tools/libxc/include/xenctrl.h | 2 ++ tools/libxc/xc_altp2m.c | 26 +++++++++++++++++++ xen/arch/x86/hvm/hvm.c | 19 ++++++++++++++ xen/arch/x86/mm/mem_access.c | 45 +++++++++++++++++++++++++++++++++ xen/include/public/hvm/hvm_op.h | 2 ++ xen/include/xen/mem_access.h | 3 +++ 6 files changed, 97 insertions(+)tools/libxc/include/xenctrl.h +2
-0@@ -1954,6 +1954,8 @@ int xc_altp2m_switch_to_view(xc_interface *handle, uint32_t domid, uint16_t view_id);int xc_altp2m_set_suppress_ve (xc_interface *handle,uint32_t domid,uint16_t view_id,xen_pfn_t gfn,bool sve); +int xc_altp2m_get_suppress_ve (xc_interface *handle,uint32_t domid, +uint16_t view_id,xen_pfn_t gfn,bool *sve);int xc_altp2m_set_mem_access (xc_interface *handle,uint32_t domid,uint16_t view_id,xen_pfn_t gfn,xenmem_access_t access);tools/libxc/xc_altp2m.c +26
-0@@ -163,6 +163,32 @@ int xc_altp2m_switch_to_view(xc_interface *handle, uint32_t domid,return rc; } +int xc_altp2m_get_suppress_ve (xc_interface *handle,uint32_t domid, +uint16_t view_id,xen_pfn_t gfn,bool *sve) +{ +int rc; +DECLARE_HYPERCALL_BUFFER (xen_hvm_altp2m_op_t, arg); + + arg =xc_hypercall_buffer_alloc (handle, arg,sizeof (*arg)); +if ( arg ==NULL ) +return -1 ; + + arg->version =HVMOP_ALTP2M_INTERFACE_VERSION ; + arg->cmd = HVMOP_altp2m_get_suppress_ve; + arg->domain = domid; + arg->u.suppress_ve.view = view_id; + arg->u.suppress_ve.gfn = gfn; + + rc =xencall2 (handle->xcall, __HYPERVISOR_hvm_op, HVMOP_altp2m, +HYPERCALL_BUFFER_AS_ARG (arg)); + +if ( !rc ) + *sve = arg->u.suppress_ve.suppress_ve; + +xc_hypercall_buffer_free (handle, arg); +return rc; +} +int xc_altp2m_set_suppress_ve (xc_interface *handle,uint32_t domid,uint16_t view_id,xen_pfn_t gfn,bool sve) {xen/arch/x86/hvm/hvm.c +19
-0@@ -4525,6 +4525,7 @@ static int do_altp2m_op(case HVMOP_altp2m_destroy_p2m:case HVMOP_altp2m_switch_p2m:case HVMOP_altp2m_set_suppress_ve: +case HVMOP_altp2m_get_suppress_ve:case HVMOP_altp2m_set_mem_access:case HVMOP_altp2m_set_mem_access_multi:case HVMOP_altp2m_change_gfn:@@ -4655,6 +4656,24 @@ static int do_altp2m_op( }break ; +case HVMOP_altp2m_get_suppress_ve: +if ( a.u.suppress_ve.pad1 || a.u.suppress_ve.pad2 ) + rc = -EINVAL ; +else + { +gfn_t gfn =_gfn (a.u.suppress_ve.gfn); +unsigned int altp2m_idx = a.u.suppress_ve.view; +bool suppress_ve; + + rc =p2m_get_suppress_ve (d, gfn, &suppress_ve, altp2m_idx); +if ( !rc ) + { + a.u.suppress_ve.suppress_ve = suppress_ve; + rc =__copy_to_guest (arg, &a,1 ) ? -EFAULT :0 ; + } + } +break ; +case HVMOP_altp2m_set_mem_access:if ( a.u.set_mem_access.pad ) rc = -EINVAL ;xen/arch/x86/mm/mem_access.c +6
-0@@ -550,6 +550,51 @@ out:return rc; } +int p2m_get_suppress_ve (struct domain *d,gfn_t gfn,bool *suppress_ve, +unsigned int altp2m_idx) +{ +struct p2m_domain *host_p2m =p2m_get_hostp2m (d); +struct p2m_domain *ap2m =NULL ; +struct p2m_domain *p2m;
xen/arch/x86/mm/mem_access.c out: · line 568
Quoted 10 lines
xen/arch/x86/mm/mem_access.c +10
-0+mfn_t mfn; +p2m_access_t a; +p2m_type_t t; + +if ( !cpu_has_vmx_virt_exceptions ) +return -EOPNOTSUPP ; + +/* #VE should be enabled for this vcpu. */ +if (gfn_eq (vcpu_altp2m (current).veinfo_gfn,INVALID_GFN ) ) +return -ENXIO ;
Basically the same comments as for 2/3: Move to p2m.c, and get rid of the vmx-ism.
Anothre idea is to get rid of these checks altogether -- returning 'false' when the feature isn't supported or enabled shouldn't be a big deal. But I don't feel strongly enough about it to argue either way.
-George
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 2/3] x86/altp2m: Add a hvmop for setting the suppress #VE bit
In reply to George DunlapQuoted 21 lines · George Dunlap
xen/arch/x86/mm/mem_access.c void arch_p2m_set_access_required(struct domain *d, bool access_required) · line 502
Quoted 2 lines
xen/arch/x86/mm/mem_access.c +2
-0+int p2m_set_suppress_ve (struct domain *d,gfn_t gfn,bool suppress_ve, +unsigned int altp2m_idx)This should clearly be in
p2m.c, and...xen/arch/x86/mm/mem_access.c void arch_p2m_set_access_required(struct domain *d, bool access_required) · line 513
Quoted 11 lines
xen/arch/x86/mm/mem_access.c +11
-0+{ +struct p2m_domain *host_p2m =p2m_get_hostp2m (d); +struct p2m_domain *ap2m =NULL ; +struct p2m_domain *p2m; +mfn_t mfn; +p2m_access_t a; +p2m_type_t t; +int rc; + +if ( !cpu_has_vmx_virt_exceptions ) +return -EOPNOTSUPP ;We should avoid Intel-specific checks in common code.
In fact, this is wrong, because you can choose to run a guest in shadow mode even on a system with virt exceptions -- in which case you'd trigger the
ASSERT()in p2m-pt.c:p2m_pt_set_entry().
Quoted 7 lines
Probably what should happen is that we should move the vmx check into p2m-ept.c:p2m_ept_set_entry(), and replace the ASSERT(sve = 0) in p2m-pt.c:p2m_pt_set_entry() with "if ( sve != 0 ) return
-ENOTSUPP;".Although what should probably *really* happen is that
HVMOP_altp2m_vcpu_enable_notifyshould fail with-EOPNOTSUPPinstead of silently succeeding.
Do you mean HVMOP_altp2m_set_suppress_ve here, or am I misunderstanding your comment? I'm happy to do the exact modifications you've requested above but I'm afraid I don't follow the HVMOP_altp2m_vcpu_enable_notify comment.
Thanks,
Razvan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 3/3] x86/altp2m: Add a hvmop for querying the suppress #VE bit
In reply to George DunlapQuoted 27 lines · George Dunlap
Quoted 12 lines · Adrian Pop
Signed-off-by Adrian Pop <[email protected]>+++
b/xen/arch/x86/mm/mem_access.cxen/arch/x86/mm/mem_access.c +6
-0@@ -550,6 +550,51 @@ out:return rc; } +int p2m_get_suppress_ve (struct domain *d,gfn_t gfn,bool *suppress_ve, +unsigned int altp2m_idx) +{ +struct p2m_domain *host_p2m =p2m_get_hostp2m (d); +struct p2m_domain *ap2m =NULL ; +struct p2m_domain *p2m;xen/arch/x86/mm/mem_access.c out: · line 568
Quoted 10 lines
xen/arch/x86/mm/mem_access.c +10
-0+mfn_t mfn; +p2m_access_t a; +p2m_type_t t; + +if ( !cpu_has_vmx_virt_exceptions ) +return -EOPNOTSUPP ; + +/* #VE should be enabled for this vcpu. */ +if (gfn_eq (vcpu_altp2m (current).veinfo_gfn,INVALID_GFN ) ) +return -ENXIO ;Basically the same comments as for 2/3: Move to
p2m.c, and get rid of the vmx-ism.
Quoted 3 lines
Anothre idea is to get rid of these checks altogether -- returning 'false' when the feature isn't supported or enabled shouldn't be a big deal. But I don't feel strongly enough about it to argue either way.
I've moved the functions to p2m.c and started moving the checks when it occured to me that it might be cha to move them to ept_get_entry() / ept_set_entry() in p2m-ept.c.
ept_get_entry() uses a bool_t *sve pointer, so I could check if the pointer is not NULL, and if it's not perform the checks then, and so require the caller to always pass a NULL pointer in non-#VE cases. Still, ept_get_entry() returns a mfn_t, so all can do there is return INVALID_MFN, which is not very helpful.
ept_set_entry() is more complicated: it takes "int sve" as a parameter (which makes sense because it can also be -1) - but then really there's no special value which should trigger a check. Any value could be a legitimate attempt to set that bit - although 0 (clearing it) might be viewed as a special case, like it is in the NPT code.
I could indeed remove all the checks and allow setting and getting that bit regardless of whether #VE is enabled and / or available - if that's acceptable (though I believe someone has required the checks at some point in Adrian's work with the patches).
Thanks,
Razvan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 3/3] x86/altp2m: Add a hvmop for querying the suppress #VE bit
Quoted 28 lines · Razvan Cojocaru
Quoted 27 lines · George Dunlap
Quoted 12 lines · Adrian Pop
Signed-off-by Adrian Pop <[email protected]>+++
b/xen/arch/x86/mm/mem_access.cxen/arch/x86/mm/mem_access.c +6
-0@@ -550,6 +550,51 @@ out:return rc; } +int p2m_get_suppress_ve (struct domain *d,gfn_t gfn,bool *suppress_ve, +unsigned int altp2m_idx) +{ +struct p2m_domain *host_p2m =p2m_get_hostp2m (d); +struct p2m_domain *ap2m =NULL ; +struct p2m_domain *p2m;xen/arch/x86/mm/mem_access.c out: · line 568
Quoted 10 lines
xen/arch/x86/mm/mem_access.c +10
-0+mfn_t mfn; +p2m_access_t a; +p2m_type_t t; + +if ( !cpu_has_vmx_virt_exceptions ) +return -EOPNOTSUPP ; + +/* #VE should be enabled for this vcpu. */ +if (gfn_eq (vcpu_altp2m (current).veinfo_gfn,INVALID_GFN ) ) +return -ENXIO ;Basically the same comments as for 2/3: Move to
p2m.c, and get rid of the vmx-ism.
Quoted 7 lines
Quoted 3 lines
Anothre idea is to get rid of these checks altogether -- returning 'false' when the feature isn't supported or enabled shouldn't be a big deal. But I don't feel strongly enough about it to argue either way.
I've moved the functions to
p2m.cand started moving the checks when it occured to me that it might be cha to move them toept_get_entry()/ept_set_entry()inp2m-ept.c.
Sorry, wrote "challenging" here, my touchpad got in the way and removed a part of the word.
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 2/3] x86/altp2m: Add a hvmop for setting the suppress #VE bit
In reply to a message not in the archiveSorry, looks like this may not have gone through.
-G
Quoted 45 lines · George Dunlap
Quoted 27 lines · Razvan Cojocaru
Quoted 25 lines · George Dunlap
xen/arch/x86/mm/mem_access.c void arch_p2m_set_access_required(struct domain *d, bool access_required) · line 502
Quoted 2 lines
xen/arch/x86/mm/mem_access.c +2
-0+int p2m_set_suppress_ve (struct domain *d,gfn_t gfn,bool suppress_ve, +unsigned int altp2m_idx)This should clearly be in
p2m.c, and...xen/arch/x86/mm/mem_access.c void arch_p2m_set_access_required(struct domain *d, bool access_required) · line 513
Quoted 11 lines
xen/arch/x86/mm/mem_access.c +11
-0+{ +struct p2m_domain *host_p2m =p2m_get_hostp2m (d); +struct p2m_domain *ap2m =NULL ; +struct p2m_domain *p2m; +mfn_t mfn; +p2m_access_t a; +p2m_type_t t; +int rc; + +if ( !cpu_has_vmx_virt_exceptions ) +return -EOPNOTSUPP ;We should avoid Intel-specific checks in common code.
In fact, this is wrong, because you can choose to run a guest in shadow mode even on a system with virt exceptions -- in which case you'd trigger the
ASSERT()in p2m-pt.c:p2m_pt_set_entry().Probably what should happen is that we should move the vmx check into p2m-ept.c:p2m_ept_set_entry(), and replace the ASSERT(sve = 0) in p2m-pt.c:p2m_pt_set_entry() with "if ( sve != 0 ) return
-ENOTSUPP;".Quoted 8 lines
Quoted 3 lines
Although what should probably *really* happen is that
HVMOP_altp2m_vcpu_enable_notifyshould fail with-EOPNOTSUPPinstead of silently succeeding.Do you mean
HVMOP_altp2m_set_suppress_vehere, or am I misunderstanding your comment? I'm happy to do the exact modifications you've requested above but I'm afraid I don't follow theHVMOP_altp2m_vcpu_enable_notifycomment.At the moment, it appears that you can call
HVMOP_altp2m_vcpu_enable_notifyto set a vcpu'sveinfo_gfneven if the hardware or the p2m mode doesn't support #VE. It would be better if callers got the-EOPNOTSUPPerror at that point, rather than havingvcpu_enable_notifysucceed and then havingset_suppress_vefail.
Quoted 5 lines
To be clear, I'm not requiring that as a clean-up, I'm just saying that such a change would be an improvement (which you may consider doing if you wish).
-George
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH v5 2/3] x86/altp2m: Add a hvmop for setting the suppress #VE bit
Quoted 51 lines · George Dunlap
Sorry, looks like this may not have gone through.
-GQuoted 48 lines · George Dunlap
Quoted 27 lines · Razvan Cojocaru
Quoted 25 lines · George Dunlap
xen/arch/x86/mm/mem_access.c void arch_p2m_set_access_required(struct domain *d, bool access_required) · line 502
Quoted 2 lines
xen/arch/x86/mm/mem_access.c +2
-0+int p2m_set_suppress_ve (struct domain *d,gfn_t gfn,bool suppress_ve, +unsigned int altp2m_idx)This should clearly be in
p2m.c, and...xen/arch/x86/mm/mem_access.c void arch_p2m_set_access_required(struct domain *d, bool access_required) · line 513
Quoted 11 lines
xen/arch/x86/mm/mem_access.c +11
-0+{ +struct p2m_domain *host_p2m =p2m_get_hostp2m (d); +struct p2m_domain *ap2m =NULL ; +struct p2m_domain *p2m; +mfn_t mfn; +p2m_access_t a; +p2m_type_t t; +int rc; + +if ( !cpu_has_vmx_virt_exceptions ) +return -EOPNOTSUPP ;We should avoid Intel-specific checks in common code.
In fact, this is wrong, because you can choose to run a guest in shadow mode even on a system with virt exceptions -- in which case you'd trigger the
ASSERT()in p2m-pt.c:p2m_pt_set_entry().Probably what should happen is that we should move the vmx check into p2m-ept.c:p2m_ept_set_entry(), and replace the ASSERT(sve = 0) in p2m-pt.c:p2m_pt_set_entry() with "if ( sve != 0 ) return
-ENOTSUPP;".Quoted 8 lines
Quoted 3 lines
Although what should probably *really* happen is that
HVMOP_altp2m_vcpu_enable_notifyshould fail with-EOPNOTSUPPinstead of silently succeeding.Do you mean
HVMOP_altp2m_set_suppress_vehere, or am I misunderstanding your comment? I'm happy to do the exact modifications you've requested above but I'm afraid I don't follow theHVMOP_altp2m_vcpu_enable_notifycomment.At the moment, it appears that you can call
HVMOP_altp2m_vcpu_enable_notifyto set a vcpu'sveinfo_gfneven if the hardware or the p2m mode doesn't support #VE. It would be better if callers got the-EOPNOTSUPPerror at that point, rather than havingvcpu_enable_notifysucceed and then havingset_suppress_vefail.To be clear, I'm not requiring that as a clean-up, I'm just saying that such a change would be an improvement (which you may consider doing if you wish).
Undestood, thank you for the clarification! We'll follow up with a patch on that.
I have in the meantime sent out a patch with the requested altp2m cleanup (code moved to p2m.c and checks moved to p2m-ept.c. I haven't seen a reply to my question about whether this is what you meant (perhaps it got lost as well?), so I hope I've understood your request correctly.
Thanks,
Razvan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel