[PATCH V7 0/5] x86/hvm: pkeys, add memory protection-key support

[PATCH V7 0/5] x86/hvm: pkeys, add memory protection-key support

[Huaitong Han]

Changes in v7:
*Add static for pkey_fault.
*Add a comment for page present check and adjust indentation.
*Init pkru_ad and pkru_wd.
*Delete l3e_get_pkey the outer parentheses.
*The first parameter of read_pkru_* use uint32_t type.
*Use EOPNOTSUPP instead of EINVAL as return value on is_pv_vcpu condition.
*Update SDM chapter comments for patch 4.
*Add hvm_vcpu check in sh_gva_to_gfn.
*Rebase in the latest tree for patch 5.
*Add a comment for cpu_has_xsave adjustment.
*Adjust indentation for patch 5.

Changes in v6:
*2 patches merged are not included.
*Don't write XSTATE_PKRU to PV's xcr0.
*Use "if()" instead of "?:" in cpuid handling patch.
*Update read_pkru function.
*Use value 4 instead of CONFIG_PAGING_LEVELS.
*Add George's patch for PFEC_insn_fetch handling.

Changes in v5:
*Add static for opt_pku.
*Update commit message for some patches.
*Add condition 5:the access is to a user page to pkey_fault, and simplify #ifdef
for guest_walk_tables patch.
*Don't write XSTATE_PKRU to PV's xcr0.
*count == 0 is combined in hvm_cpuid function.

Changes in v4:
*Delete gva2gfn patch, and when page is present, PFEC_prot_key is always checked.
*Use RDPKRU instead of xsave_read because RDPKRU does cost less.
*Squash pkeys patch and pkru patch to guest_walk_tables patch.

Changes in v3:
*Get CPUID:ospke depend on guest cpuid instead of host hardware capable,
and Move cpuid patch to the last of patches.
*Move opt_pku to cpu/common.c.
*Use MASK_EXTR for get_pte_pkeys.
*Add quoting for pkru macro, and use static inline pkru_read functions.
*Rebase get_xsave_addr for updated codes, and add uncompressed format support
for xsave state.
*Use fpu_xsave instead of vcpu_save_fpu, and adjust the code style for
leaf_pte_pkeys_check.
*Add parentheses for PFEC_prot_key of gva2gfn funcitons.

Changes in v2:
*Rebase all patches in staging branch
*Disable X86_CR4_PKE on hypervisor, and delete pkru_read/write functions, and
use xsave state read to get pkru value.
*Delete the patch that adds pkeys support for do_page_fault.
*Add pkeys support for gva2gfn so that setting _PAGE_PK_BIT in the return
value can get propagated to the guest correctly.

The protection-key feature provides an additional mechanism by which IA-32e
paging controls access to usermode addresses.

Hardware support for protection keys for user pages is enumerated with CPUID
feature flag CPUID.7.0.ECX[3]:PKU. Software support is CPUID.7.0.ECX[4]:OSPKE
with the setting of CR4.PKE(bit 22).

When CR4.PKE = 1, every linear address is associated with the 4-bit protection
key located in bits 62:59 of the paging-structure entry that mapped the page
containing the linear address. The PKRU register determines, for each
protection key, whether user-mode addresses with that protection key may be
read or written.

The PKRU register (protection key rights for user pages) is a 32-bit register
with the following format: for each i (0 ≤ i ≤ 15), PKRU[2i] is the
access-disable bit for protection key i (ADi); PKRU[2i+1] is the write-disable
bit for protection key i (WDi).

Software can use the RDPKRU and WRPKRU instructions with ECX = 0 to read and
write PKRU. In addition, the PKRU register is XSAVE-managed state and can thus
be read and written by instructions in the XSAVE feature set.

PFEC.PK (bit 5) is defined as protection key violations.

The specification of Protection Keys can be found at SDM (4.6.2, volume 3)
http://www.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-software-developer-manual-325462.pdf.

George Dunlap (1):
  xen/mm: Clean up pfec handling in gva_to_gfn

Huaitong Han (4):
  x86/hvm: pkeys, disable pkeys for guests in non-paging mode
  x86/hvm: pkeys, add pkeys support for guest_walk_tables
  x86/hvm: pkeys, add xstate support for pkeys
  x86/hvm: pkeys, add pkeys support for cpuid handling

 tools/libxc/xc_cpufeature.h       |  3 +++
 tools/libxc/xc_cpuid_x86.c        |  6 +++--
 xen/arch/x86/hvm/hvm.c            | 26 +++++++++++--------
 xen/arch/x86/hvm/vmx/vmx.c        | 11 ++++----
 xen/arch/x86/mm/guest_walk.c      | 54 +++++++++++++++++++++++++++++++++++++++
 xen/arch/x86/mm/hap/guest_walk.c  | 13 +++++++++-
 xen/arch/x86/mm/shadow/multi.c    |  6 +++++
 xen/arch/x86/xstate.c             |  4 +++
 xen/include/asm-x86/guest_pt.h    | 12 +++++++++
 xen/include/asm-x86/hvm/hvm.h     |  2 ++
 xen/include/asm-x86/page.h        |  5 ++++
 xen/include/asm-x86/processor.h   | 40 +++++++++++++++++++++++++++++
 xen/include/asm-x86/x86_64/page.h | 12 +++++++++
 xen/include/asm-x86/xstate.h      |  4 ++-
 14 files changed, 178 insertions(+), 20 deletions(-)

-- 
2.4.3


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

[PATCH V7 1/5] x86/hvm: pkeys, disable pkeys for guests in non-paging mode

[Huaitong Han]

Changes in v7:
no changes.
----

This patch disables pkeys for guest in non-paging mode, However XEN always uses
paging mode to emulate guest non-paging mode, To emulate this behavior, pkeys
needs to be manually disabled when guest switches to non-paging mode.

Signed-off-by: Huaitong Han <huaitong.han@intel.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
---
 xen/arch/x86/hvm/vmx/vmx.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/xen/arch/x86/hvm/vmx/vmx.c b/xen/arch/x86/hvm/vmx/vmx.c
index 04dde83..a0d51cb 100644
--- a/xen/arch/x86/hvm/vmx/vmx.c
+++ b/xen/arch/x86/hvm/vmx/vmx.c
@@ -1368,12 +1368,13 @@ static void vmx_update_guest_cr(struct vcpu *v, unsigned int cr)
         if ( !hvm_paging_enabled(v) )
         {
             /*
-             * SMEP/SMAP is disabled if CPU is in non-paging mode in hardware.
-             * However Xen always uses paging mode to emulate guest non-paging
-             * mode. To emulate this behavior, SMEP/SMAP needs to be manually
-             * disabled when guest VCPU is in non-paging mode.
+             * SMEP/SMAP/PKU is disabled if CPU is in non-paging mode in
+             * hardware. However Xen always uses paging mode to emulate guest
+             * non-paging mode. To emulate this behavior, SMEP/SMAP/PKU needs
+             * to be manually disabled when guest VCPU is in non-paging mode.
              */
-            v->arch.hvm_vcpu.hw_cr[4] &= ~(X86_CR4_SMEP | X86_CR4_SMAP);
+            v->arch.hvm_vcpu.hw_cr[4] &=
+                ~(X86_CR4_SMEP | X86_CR4_SMAP | X86_CR4_PKE);
         }
         __vmwrite(GUEST_CR4, v->arch.hvm_vcpu.hw_cr[4]);
         break;
-- 
2.4.3

[PATCH V7 2/5] x86/hvm: pkeys, add pkeys support for guest_walk_tables

[Huaitong Han]

Changes in v7:
*Add static for pkey_fault.
*Add a comment for page present check and adjust indentation.
*Init pkru_ad and pkru_wd.
*Delete l3e_get_pkey the outer parentheses.
*The first parameter of read_pkru_* use uint32_t type.
----

Protection keys define a new 4-bit protection key field(PKEY) in bits 62:59 of
leaf entries of the page tables.

PKRU register defines 32 bits, there are 16 domains and 2 attribute bits per
domain in pkru, for each i (0 ≤ i ≤ 15), PKRU[2i] is the access-disable bit for
protection key i (ADi); PKRU[2i+1] is the write-disable bit for protection key
i (WDi). PKEY is index to a defined domain.

A fault is considered as a PKU violation if all of the following conditions are
true:
1.CR4_PKE=1.
2.EFER_LMA=1.
3.Page is present with no reserved bit violations.
4.The access is not an instruction fetch.
5.The access is to a user page.
6.PKRU.AD=1
    or The access is a data write and PKRU.WD=1
                and either CR0.WP=1 or it is a user access.

Signed-off-by: Huaitong Han <huaitong.han@intel.com>
---
 xen/arch/x86/mm/guest_walk.c      | 54 +++++++++++++++++++++++++++++++++++++++
 xen/arch/x86/mm/hap/guest_walk.c  |  3 +++
 xen/include/asm-x86/guest_pt.h    | 12 +++++++++
 xen/include/asm-x86/hvm/hvm.h     |  2 ++
 xen/include/asm-x86/page.h        |  5 ++++
 xen/include/asm-x86/processor.h   | 40 +++++++++++++++++++++++++++++
 xen/include/asm-x86/x86_64/page.h | 12 +++++++++
 7 files changed, 128 insertions(+)

diff --git a/xen/arch/x86/mm/guest_walk.c b/xen/arch/x86/mm/guest_walk.c
index 18d1acf..5e1111b 100644
--- a/xen/arch/x86/mm/guest_walk.c
+++ b/xen/arch/x86/mm/guest_walk.c
@@ -90,6 +90,54 @@ static uint32_t set_ad_bits(void *guest_p, void *walk_p, int set_dirty)
     return 0;
 }
 
+#if GUEST_PAGING_LEVELS >= 4
+static bool_t pkey_fault(struct vcpu *vcpu, uint32_t pfec,
+        uint32_t pte_flags, uint32_t pte_pkey)
+{
+    uint32_t pkru = 0;
+    bool_t pkru_ad = 0, pkru_wd = 0;
+
+    /* When page isn't present,  PKEY isn't checked. */
+    if ( !(pfec & PFEC_page_present) || is_pv_vcpu(vcpu) )
+        return 0;
+
+    /*
+     * PKU:  additional mechanism by which the paging controls
+     * access to user-mode addresses based on the value in the
+     * PKRU register. A fault is considered as a PKU violation if all
+     * of the following conditions are true:
+     * 1.CR4_PKE=1.
+     * 2.EFER_LMA=1.
+     * 3.Page is present with no reserved bit violations.
+     * 4.The access is not an instruction fetch.
+     * 5.The access is to a user page.
+     * 6.PKRU.AD=1 or
+     *      the access is a data write and PKRU.WD=1 and
+     *          either CR0.WP=1 or it is a user access.
+     */
+    if ( !hvm_pku_enabled(vcpu) ||
+         !hvm_long_mode_enabled(vcpu) ||
+         /* The persent bit is guaranteed by the caller. */
+         (pfec & PFEC_reserved_bit) ||
+         (pfec & PFEC_insn_fetch) ||
+         !(pte_flags & _PAGE_USER) )
+        return 0;
+
+    pkru = read_pkru();
+    if ( unlikely(pkru) )
+    {
+        pkru_ad = read_pkru_ad(pkru, pte_pkey);
+        pkru_wd = read_pkru_wd(pkru, pte_pkey);
+        /* Condition 6 */
+        if ( pkru_ad || (pkru_wd && (pfec & PFEC_write_access) &&
+                    (hvm_wp_enabled(vcpu) || (pfec & PFEC_user_mode))))
+            return 1;
+    }
+
+    return 0;
+}
+#endif
+
 /* Walk the guest pagetables, after the manner of a hardware walker. */
 /* Because the walk is essentially random, it can cause a deadlock 
  * warning in the p2m locking code. Highly unlikely this is an actual
@@ -107,6 +155,7 @@ guest_walk_tables(struct vcpu *v, struct p2m_domain *p2m,
     guest_l3e_t *l3p = NULL;
     guest_l4e_t *l4p;
 #endif
+    unsigned int pkey;
     uint32_t gflags, mflags, iflags, rc = 0;
     bool_t smep = 0, smap = 0;
     bool_t pse1G = 0, pse2M = 0;
@@ -190,6 +239,7 @@ guest_walk_tables(struct vcpu *v, struct p2m_domain *p2m,
         goto out;
     /* Get the l3e and check its flags*/
     gw->l3e = l3p[guest_l3_table_offset(va)];
+    pkey = guest_l3e_get_pkey(gw->l3e);
     gflags = guest_l3e_get_flags(gw->l3e) ^ iflags;
     if ( !(gflags & _PAGE_PRESENT) ) {
         rc |= _PAGE_PRESENT;
@@ -261,6 +311,7 @@ guest_walk_tables(struct vcpu *v, struct p2m_domain *p2m,
 
 #endif /* All levels... */
 
+    pkey = guest_l2e_get_pkey(gw->l2e);
     gflags = guest_l2e_get_flags(gw->l2e) ^ iflags;
     if ( !(gflags & _PAGE_PRESENT) ) {
         rc |= _PAGE_PRESENT;
@@ -324,6 +375,7 @@ guest_walk_tables(struct vcpu *v, struct p2m_domain *p2m,
         if(l1p == NULL)
             goto out;
         gw->l1e = l1p[guest_l1_table_offset(va)];
+        pkey = guest_l1e_get_pkey(gw->l1e);
         gflags = guest_l1e_get_flags(gw->l1e) ^ iflags;
         if ( !(gflags & _PAGE_PRESENT) ) {
             rc |= _PAGE_PRESENT;
@@ -334,6 +386,8 @@ guest_walk_tables(struct vcpu *v, struct p2m_domain *p2m,
 
 #if GUEST_PAGING_LEVELS >= 4 /* 64-bit only... */
 set_ad:
+    if ( pkey_fault(v, pfec, gflags, pkey) )
+        rc |= _PAGE_PKEY_BITS;
 #endif
     /* Now re-invert the user-mode requirement for SMEP and SMAP */
     if ( smep || smap )
diff --git a/xen/arch/x86/mm/hap/guest_walk.c b/xen/arch/x86/mm/hap/guest_walk.c
index 11c1b35..49d0328 100644
--- a/xen/arch/x86/mm/hap/guest_walk.c
+++ b/xen/arch/x86/mm/hap/guest_walk.c
@@ -130,6 +130,9 @@ unsigned long hap_p2m_ga_to_gfn(GUEST_PAGING_LEVELS)(
     if ( missing & _PAGE_INVALID_BITS ) 
         pfec[0] |= PFEC_reserved_bit;
 
+    if ( missing & _PAGE_PKEY_BITS )
+        pfec[0] |= PFEC_prot_key;
+
     if ( missing & _PAGE_PAGED )
         pfec[0] = PFEC_page_paged;
 
diff --git a/xen/include/asm-x86/guest_pt.h b/xen/include/asm-x86/guest_pt.h
index 3447973..eb29e62 100644
--- a/xen/include/asm-x86/guest_pt.h
+++ b/xen/include/asm-x86/guest_pt.h
@@ -81,6 +81,11 @@ static inline u32 guest_l1e_get_flags(guest_l1e_t gl1e)
 static inline u32 guest_l2e_get_flags(guest_l2e_t gl2e)
 { return gl2e.l2 & 0xfff; }
 
+static inline u32 guest_l1e_get_pkey(guest_l1e_t gl1e)
+{ return 0; }
+static inline u32 guest_l2e_get_pkey(guest_l2e_t gl2e)
+{ return 0; }
+
 static inline guest_l1e_t guest_l1e_from_gfn(gfn_t gfn, u32 flags)
 { return (guest_l1e_t) { (gfn_x(gfn) << PAGE_SHIFT) | flags }; }
 static inline guest_l2e_t guest_l2e_from_gfn(gfn_t gfn, u32 flags)
@@ -154,6 +159,13 @@ static inline u32 guest_l4e_get_flags(guest_l4e_t gl4e)
 { return l4e_get_flags(gl4e); }
 #endif
 
+static inline u32 guest_l1e_get_pkey(guest_l1e_t gl1e)
+{ return l1e_get_pkey(gl1e); }
+static inline u32 guest_l2e_get_pkey(guest_l2e_t gl2e)
+{ return l2e_get_pkey(gl2e); }
+static inline u32 guest_l3e_get_pkey(guest_l3e_t gl3e)
+{ return l3e_get_pkey(gl3e); }
+
 static inline guest_l1e_t guest_l1e_from_gfn(gfn_t gfn, u32 flags)
 { return l1e_from_pfn(gfn_x(gfn), flags); }
 static inline guest_l2e_t guest_l2e_from_gfn(gfn_t gfn, u32 flags)
diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm-x86/hvm/hvm.h
index a87224b..731dd44 100644
--- a/xen/include/asm-x86/hvm/hvm.h
+++ b/xen/include/asm-x86/hvm/hvm.h
@@ -277,6 +277,8 @@ int hvm_girq_dest_2_vcpu_id(struct domain *d, uint8_t dest, uint8_t dest_mode);
     (hvm_paging_enabled(v) && ((v)->arch.hvm_vcpu.guest_cr[4] & X86_CR4_SMAP))
 #define hvm_nx_enabled(v) \
     (!!((v)->arch.hvm_vcpu.guest_efer & EFER_NX))
+#define hvm_pku_enabled(v) \
+    (hvm_paging_enabled(v) && ((v)->arch.hvm_vcpu.guest_cr[4] & X86_CR4_PKE))
 
 /* Can we use superpages in the HAP p2m table? */
 #define hap_has_1gb (!!(hvm_funcs.hap_capabilities & HVM_HAP_SUPERPAGE_1GB))
diff --git a/xen/include/asm-x86/page.h b/xen/include/asm-x86/page.h
index a095a93..9202f3d 100644
--- a/xen/include/asm-x86/page.h
+++ b/xen/include/asm-x86/page.h
@@ -93,6 +93,11 @@
 #define l3e_get_flags(x)           (get_pte_flags((x).l3))
 #define l4e_get_flags(x)           (get_pte_flags((x).l4))
 
+/* Get pte pkeys (unsigned int). */
+#define l1e_get_pkey(x)           get_pte_pkey((x).l1)
+#define l2e_get_pkey(x)           get_pte_pkey((x).l2)
+#define l3e_get_pkey(x)           get_pte_pkey((x).l3)
+
 /* Construct an empty pte. */
 #define l1e_empty()                ((l1_pgentry_t) { 0 })
 #define l2e_empty()                ((l2_pgentry_t) { 0 })
diff --git a/xen/include/asm-x86/processor.h b/xen/include/asm-x86/processor.h
index 26ba141..9799dd3 100644
--- a/xen/include/asm-x86/processor.h
+++ b/xen/include/asm-x86/processor.h
@@ -374,6 +374,46 @@ static always_inline void clear_in_cr4 (unsigned long mask)
     write_cr4(read_cr4() & ~mask);
 }
 
+static inline unsigned int read_pkru(void)
+{
+    unsigned int pkru;
+    unsigned long cr4 = read_cr4();
+
+    /*
+     * _PAGE_PKEY_BITS have a conflict with _PAGE_GNTTAB used by PV guests,
+     * so that X86_CR4_PKE  is disabled on hypervisor. To use RDPKRU, CR4.PKE
+     * gets temporarily enabled.
+     */
+    write_cr4(cr4 | X86_CR4_PKE);
+    asm volatile (".byte 0x0f,0x01,0xee"
+        : "=a" (pkru) : "c" (0) : "dx");
+    write_cr4(cr4);
+
+    return pkru;
+}
+
+/* Macros for PKRU domain */
+#define PKRU_READ  (0)
+#define PKRU_WRITE (1)
+#define PKRU_ATTRS (2)
+
+/*
+ * PKRU defines 32 bits, there are 16 domains and 2 attribute bits per
+ * domain in pkru, pkeys is index to a defined domain, so the value of
+ * pte_pkeys * PKRU_ATTRS + R/W is offset of a defined domain attribute.
+ */
+static inline bool_t read_pkru_ad(uint32_t pkru, unsigned int pkey)
+{
+    ASSERT(pkey < 16);
+    return (pkru >> (pkey * PKRU_ATTRS + PKRU_READ)) & 1;
+}
+
+static inline bool_t read_pkru_wd(uint32_t pkru, unsigned int pkey)
+{
+    ASSERT(pkey < 16);
+    return (pkru >> (pkey * PKRU_ATTRS + PKRU_WRITE)) & 1;
+}
+
 /*
  *      NSC/Cyrix CPU configuration register indexes
  */
diff --git a/xen/include/asm-x86/x86_64/page.h b/xen/include/asm-x86/x86_64/page.h
index 19ab4d0..86abb94 100644
--- a/xen/include/asm-x86/x86_64/page.h
+++ b/xen/include/asm-x86/x86_64/page.h
@@ -134,6 +134,18 @@ typedef l4_pgentry_t root_pgentry_t;
 #define get_pte_flags(x) (((int)((x) >> 40) & ~0xFFF) | ((int)(x) & 0xFFF))
 #define put_pte_flags(x) (((intpte_t)((x) & ~0xFFF) << 40) | ((x) & 0xFFF))
 
+/*
+ * Protection keys define a new 4-bit protection key field
+ * (PKEY) in bits 62:59 of leaf entries of the page tables.
+ * This corresponds to bit 22:19 of a 24-bit flags.
+ *
+ * Notice: Bit 22 is used by _PAGE_GNTTAB which is visible to PV guests,
+ * so Protection keys must be disabled on PV guests.
+ */
+#define _PAGE_PKEY_BITS  (0x780000)	 /* Protection Keys, 22:19 */
+
+#define get_pte_pkey(x) (MASK_EXTR(get_pte_flags(x), _PAGE_PKEY_BITS))
+
 /* Bit 23 of a 24-bit flag mask. This corresponds to bit 63 of a pte.*/
 #define _PAGE_NX_BIT (1U<<23)
 
-- 
2.4.3


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

[PATCH V7 3/5] x86/hvm: pkeys, add xstate support for pkeys

[Huaitong Han]

Changes in v7:
*Use EOPNOTSUPP instead of EINVAL as return value on is_pv_vcpu condition.
---

The XSAVE feature set can operate on PKRU state only if the feature set is
enabled (CR4.OSXSAVE = 1) and has been configured to manage PKRU state
(XCR0[9] = 1). And XCR0.PKRU is disabled on PV mode without PKU feature
enabled.

Signed-off-by: Huaitong Han <huaitong.han@intel.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
---
 xen/arch/x86/xstate.c        | 4 ++++
 xen/include/asm-x86/xstate.h | 4 +++-
 2 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/xen/arch/x86/xstate.c b/xen/arch/x86/xstate.c
index 4e87ab3..50d9e48 100644
--- a/xen/arch/x86/xstate.c
+++ b/xen/arch/x86/xstate.c
@@ -579,6 +579,10 @@ int handle_xsetbv(u32 index, u64 new_bv)
     if ( (new_bv & ~xfeature_mask) || !valid_xcr0(new_bv) )
         return -EINVAL;
 
+    /* XCR0.PKRU is disabled on PV mode. */
+    if ( is_pv_vcpu(curr) && (new_bv & XSTATE_PKRU) )
+        return -EOPNOTSUPP;
+
     if ( !set_xcr0(new_bv) )
         return -EFAULT;
 
diff --git a/xen/include/asm-x86/xstate.h b/xen/include/asm-x86/xstate.h
index 12d939b..f7c41ba 100644
--- a/xen/include/asm-x86/xstate.h
+++ b/xen/include/asm-x86/xstate.h
@@ -34,13 +34,15 @@
 #define XSTATE_OPMASK  (1ULL << 5)
 #define XSTATE_ZMM     (1ULL << 6)
 #define XSTATE_HI_ZMM  (1ULL << 7)
+#define XSTATE_PKRU    (1ULL << 9)
 #define XSTATE_LWP     (1ULL << 62) /* AMD lightweight profiling */
 #define XSTATE_FP_SSE  (XSTATE_FP | XSTATE_SSE)
 #define XCNTXT_MASK    (XSTATE_FP | XSTATE_SSE | XSTATE_YMM | XSTATE_OPMASK | \
                         XSTATE_ZMM | XSTATE_HI_ZMM | XSTATE_NONLAZY)
 
 #define XSTATE_ALL     (~(1ULL << 63))
-#define XSTATE_NONLAZY (XSTATE_LWP | XSTATE_BNDREGS | XSTATE_BNDCSR)
+#define XSTATE_NONLAZY (XSTATE_LWP | XSTATE_BNDREGS | XSTATE_BNDCSR | \
+                        XSTATE_PKRU)
 #define XSTATE_LAZY    (XSTATE_ALL & ~XSTATE_NONLAZY)
 #define XSTATE_COMPACTION_ENABLED  (1ULL << 63)
 
-- 
2.4.3

[PATCH V7 4/5] xen/mm: Clean up pfec handling in gva_to_gfn

[Huaitong Han]

From: George Dunlap <george.dunlap@citrix.com>

Changes in v7:
*Update SDM chapter comments.
*Add hvm_vcpu check in sh_gva_to_gfn.
---

At the moment, the pfec argument to gva_to_gfn has two functions:

* To inform guest_walk what kind of access is happenind

* As a value to pass back into the guest in the event of a fault.

Unfortunately this is not quite treated consistently: the hvm_fetch_*
function will "pre-clear" the PFEC_insn_fetch flag before calling
gva_to_gfn; meaning guest_walk doesn't actually know whether a given
access is an instruction fetch or not.  This works now, but will cause
issues when pkeys are introduced, since guest_walk will need to know
whether an access is an instruction fetch even if it doesn't return
PFEC_insn_fetch.

Fix this by making a clean separation for in and out functionalities
of the pfec argument:

1. Always pass in the access type to gva_to_gfn

2. Filter out inappropriate access flags before returning from gva_to_gfn.

(The PFEC_insn_fetch flag should only be passed to the guest if either NX or
SMEP is enabled.  See Intel 64 Developer's Manual, Volume 3, Chapter Paging,
PAGE-FAULT EXCEPTIONS)

Signed-off-by: George Dunlap <george.dunlap@citrix.com>
Signed-off-by: Huaitong Han <huaitong.han@intel.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
---
 xen/arch/x86/hvm/hvm.c           |  8 ++------
 xen/arch/x86/mm/hap/guest_walk.c | 10 +++++++++-
 xen/arch/x86/mm/shadow/multi.c   |  6 ++++++
 3 files changed, 17 insertions(+), 7 deletions(-)

diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
index 674feea..5ec2ae1 100644
--- a/xen/arch/x86/hvm/hvm.c
+++ b/xen/arch/x86/hvm/hvm.c
@@ -4438,11 +4438,9 @@ enum hvm_copy_result hvm_copy_from_guest_virt(
 enum hvm_copy_result hvm_fetch_from_guest_virt(
     void *buf, unsigned long vaddr, int size, uint32_t pfec)
 {
-    if ( hvm_nx_enabled(current) || hvm_smep_enabled(current) )
-        pfec |= PFEC_insn_fetch;
     return __hvm_copy(buf, vaddr, size,
                       HVMCOPY_from_guest | HVMCOPY_fault | HVMCOPY_virt,
-                      PFEC_page_present | pfec);
+                      PFEC_page_present | PFEC_insn_fetch | pfec);
 }
 
 enum hvm_copy_result hvm_copy_to_guest_virt_nofault(
@@ -4464,11 +4462,9 @@ enum hvm_copy_result hvm_copy_from_guest_virt_nofault(
 enum hvm_copy_result hvm_fetch_from_guest_virt_nofault(
     void *buf, unsigned long vaddr, int size, uint32_t pfec)
 {
-    if ( hvm_nx_enabled(current) || hvm_smep_enabled(current) )
-        pfec |= PFEC_insn_fetch;
     return __hvm_copy(buf, vaddr, size,
                       HVMCOPY_from_guest | HVMCOPY_no_fault | HVMCOPY_virt,
-                      PFEC_page_present | pfec);
+                      PFEC_page_present | PFEC_insn_fetch | pfec);
 }
 
 unsigned long copy_to_user_hvm(void *to, const void *from, unsigned int len)
diff --git a/xen/arch/x86/mm/hap/guest_walk.c b/xen/arch/x86/mm/hap/guest_walk.c
index 49d0328..d2716f9 100644
--- a/xen/arch/x86/mm/hap/guest_walk.c
+++ b/xen/arch/x86/mm/hap/guest_walk.c
@@ -82,7 +82,7 @@ unsigned long hap_p2m_ga_to_gfn(GUEST_PAGING_LEVELS)(
     if ( !top_page )
     {
         pfec[0] &= ~PFEC_page_present;
-        return INVALID_GFN;
+        goto out_tweak_pfec;
     }
     top_mfn = _mfn(page_to_mfn(top_page));
 
@@ -139,6 +139,14 @@ unsigned long hap_p2m_ga_to_gfn(GUEST_PAGING_LEVELS)(
     if ( missing & _PAGE_SHARED )
         pfec[0] = PFEC_page_shared;
 
+ out_tweak_pfec:
+    /*
+     * SDM Intel 64 Volume 3, Chapter Paging, PAGE-FAULT EXCEPTIONS:
+     * The PFEC_insn_fetch flag is set only when NX or SMEP are enabled.
+     */
+    if ( !hvm_nx_enabled(v) && !hvm_smep_enabled(v) )
+        pfec[0] &= ~PFEC_insn_fetch;
+
     return INVALID_GFN;
 }
 
diff --git a/xen/arch/x86/mm/shadow/multi.c b/xen/arch/x86/mm/shadow/multi.c
index 162c06f..d42597c 100644
--- a/xen/arch/x86/mm/shadow/multi.c
+++ b/xen/arch/x86/mm/shadow/multi.c
@@ -3669,6 +3669,12 @@ sh_gva_to_gfn(struct vcpu *v, struct p2m_domain *p2m,
             pfec[0] &= ~PFEC_page_present;
         if ( missing & _PAGE_INVALID_BITS )
             pfec[0] |= PFEC_reserved_bit;
+        /*
+         * SDM Intel 64 Volume 3, Chapter Paging, PAGE-FAULT EXCEPTIONS:
+         * The PFEC_insn_fetch flag is set only when NX or SMEP are enabled.
+         */
+        if ( is_hvm_vcpu(v) && !hvm_nx_enabled(v) && !hvm_smep_enabled(v) )
+            pfec[0] &= ~PFEC_insn_fetch;
         return INVALID_GFN;
     }
     gfn = guest_walk_to_gfn(&gw);
-- 
2.4.3

[PATCH V7 5/5] x86/hvm: pkeys, add pkeys support for cpuid handling

[Huaitong Han]

Changes in v7:
*Rebase in the latest tree.
*Add a comment for cpu_has_xsave adjustment.
*Adjust indentation.
---

This patch adds pkeys support for cpuid handing.

Pkeys hardware support is CPUID.7.0.ECX[3]:PKU. software support is
CPUID.7.0.ECX[4]:OSPKE and it reflects the support setting of CR4.PKE.

X86_FEATURE_OSXSAVE depends on guest X86_FEATURE_XSAVE, but cpu_has_xsave
function reflects hypervisor X86_FEATURE_XSAVE, it is fixed too.

Signed-off-by: Huaitong Han <huaitong.han@intel.com>
---
 tools/libxc/xc_cpufeature.h |  3 +++
 tools/libxc/xc_cpuid_x86.c  |  6 ++++--
 xen/arch/x86/hvm/hvm.c      | 18 +++++++++++++-----
 3 files changed, 20 insertions(+), 7 deletions(-)

diff --git a/tools/libxc/xc_cpufeature.h b/tools/libxc/xc_cpufeature.h
index ee53679..866cf0b 100644
--- a/tools/libxc/xc_cpufeature.h
+++ b/tools/libxc/xc_cpufeature.h
@@ -144,4 +144,7 @@
 #define X86_FEATURE_CLFLUSHOPT  23 /* CLFLUSHOPT instruction */
 #define X86_FEATURE_CLWB        24 /* CLWB instruction */
 
+/* Intel-defined CPU features, CPUID level 0x00000007:0 (ecx) */
+#define X86_FEATURE_PKU     3
+
 #endif /* __LIBXC_CPUFEATURE_H */
diff --git a/tools/libxc/xc_cpuid_x86.c b/tools/libxc/xc_cpuid_x86.c
index c142595..5408dd0 100644
--- a/tools/libxc/xc_cpuid_x86.c
+++ b/tools/libxc/xc_cpuid_x86.c
@@ -430,9 +430,11 @@ static void xc_cpuid_hvm_policy(xc_interface *xch,
                         bitmaskof(X86_FEATURE_PCOMMIT) |
                         bitmaskof(X86_FEATURE_CLWB) |
                         bitmaskof(X86_FEATURE_CLFLUSHOPT));
+            regs[2] &= bitmaskof(X86_FEATURE_PKU);
         } else
-            regs[1] = 0;
-        regs[0] = regs[2] = regs[3] = 0;
+            regs[1] = regs[2] = 0;
+
+        regs[0] = regs[3] = 0;
         break;
 
     case 0x0000000d:
diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
index 5ec2ae1..1389173 100644
--- a/xen/arch/x86/hvm/hvm.c
+++ b/xen/arch/x86/hvm/hvm.c
@@ -4572,7 +4572,7 @@ void hvm_cpuid(unsigned int input, unsigned int *eax, unsigned int *ebx,
             __clear_bit(X86_FEATURE_APIC & 31, edx);
 
         /* Fix up OSXSAVE. */
-        if ( cpu_has_xsave )
+        if ( *ecx & cpufeat_mask(X86_FEATURE_XSAVE) )
             *ecx |= (v->arch.hvm_vcpu.guest_cr[4] & X86_CR4_OSXSAVE) ?
                      cpufeat_mask(X86_FEATURE_OSXSAVE) : 0;
 
@@ -4593,16 +4593,24 @@ void hvm_cpuid(unsigned int input, unsigned int *eax, unsigned int *ebx,
             if ( !cpu_has_smap )
                 *ebx &= ~cpufeat_mask(X86_FEATURE_SMAP);
 
-            /* Don't expose MPX to hvm when VMX support is not available */
+            /* Don't expose MPX to hvm when VMX support is not available. */
             if ( !(vmx_vmexit_control & VM_EXIT_CLEAR_BNDCFGS) ||
                  !(vmx_vmentry_control & VM_ENTRY_LOAD_BNDCFGS) )
                 *ebx &= ~cpufeat_mask(X86_FEATURE_MPX);
 
-            /* Don't expose INVPCID to non-hap hvm. */
             if ( !hap_enabled(d) )
-                *ebx &= ~cpufeat_mask(X86_FEATURE_INVPCID);
+            {
+                 /* Don't expose INVPCID to non-hap hvm. */
+                 *ebx &= ~cpufeat_mask(X86_FEATURE_INVPCID);
+                 /* X86_FEATURE_PKU is not yet implemented for shadow paging. */
+                 *ecx &= ~cpufeat_mask(X86_FEATURE_PKU);
+            }
+
+            if ( (*ecx & cpufeat_mask(X86_FEATURE_PKU)) &&
+                 (v->arch.hvm_vcpu.guest_cr[4] & X86_CR4_PKE) )
+                *ecx |= cpufeat_mask(X86_FEATURE_OSPKE);
 
-            /* Don't expose PCOMMIT to hvm when VMX support is not available */
+            /* Don't expose PCOMMIT to hvm when VMX support is not available. */
             if ( !cpu_has_vmx_pcommit )
                 *ebx &= ~cpufeat_mask(X86_FEATURE_PCOMMIT);
         }
-- 
2.4.3

Re: [PATCH V7 1/5] x86/hvm: pkeys, disable pkeys for guests in non-paging mode

[Jan Beulich]

>>> On 27.01.16 at 09:30, <huaitong.han@intel.com> wrote:
> Changes in v7:
> no changes.
> ----
> 
> This patch disables pkeys for guest in non-paging mode, However XEN always 
> uses
> paging mode to emulate guest non-paging mode, To emulate this behavior, pkeys
> needs to be manually disabled when guest switches to non-paging mode.
> 
> Signed-off-by: Huaitong Han <huaitong.han@intel.com>
> Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
> ---

The "Changes in" belongs here, the way it's done now will require
extra work while committing. Hence this needs to be resent in
proper shape.

Jan