[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Xen-changelog] [xen staging-4.8] x86/pv: Rewrite guest %cr4 handling from scratch



commit 4539dbc05201b852d26d6d9cfaabe16adf8f4b1f
Author:     Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
AuthorDate: Tue Mar 5 15:45:37 2019 +0100
Commit:     Jan Beulich <jbeulich@xxxxxxxx>
CommitDate: Tue Mar 5 15:45:37 2019 +0100

    x86/pv: Rewrite guest %cr4 handling from scratch
    
    The PV cr4 logic is almost impossible to follow, and leaks bits into guest
    context which definitely shouldn't be visible (in particular, VMXE).
    
    The biggest problem however, and source of the complexity, is that it 
derives
    new real and guest cr4 values from the current value in hardware - this is
    context dependent and an inappropriate source of information.
    
    Rewrite the cr4 logic to be invariant of the current value in hardware.
    
    First of all, modify write_ptbase() to always use mmu_cr4_features for IDLE
    and HVM contexts.  mmu_cr4_features *is* the correct value to use, and makes
    the ASSERT() obviously redundant.
    
    For PV guests, curr->arch.pv.ctrlreg[4] remains the guests view of cr4, but
    all logic gets reworked in terms of this and mmu_cr4_features only.
    
    Two masks are introduced; bits which the guest has control over, and bits
    which are forwarded from Xen's settings.  One guest-visible change here is
    that Xen's VMXE setting is no longer visible at all.
    
    pv_make_cr4() follows fairly closely from pv_guest_cr4_to_real_cr4(), but
    deliberately starts with mmu_cr4_features, and only alters the minimal 
subset
    of bits.
    
    The boot-time {compat_,}pv_cr4_mask variables are removed, as they are a
    remnant of the pre-CPUID policy days.  pv_fixup_guest_cr4() gains a related
    derivation from the policy.
    
    Another guest visible change here is that a 32bit PV guest can now flip
    FSGSBASE in its view of CR4.  While the {RD,WR}{FS,GS}BASE instructions are
    unusable outside of a 64bit code segment, the ability to modify FSGSBASE
    matches real hardware behaviour, and avoids the need for any 32bit/64bit
    differences in the logic.
    
    Overall, this patch shouldn't have a practical change in guest behaviour.
    VMXE will disappear from view, and an inquisitive 32bit kernel can now see
    FSGSBASE changing, but this new logic is otherwise bug-compatible with 
before.
    
    This is part of XSA-293.
    
    Signed-off-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
    Reviewed-by: Jan Beulich <jbeulich@xxxxxxxx>
    master commit: b2dd00574a4fc87ca964177f8e752a968c27efb2
    master date: 2019-03-05 13:53:32 +0100
---
 xen/arch/x86/domain.c        | 108 +++++++++++++++++++++++--------------------
 xen/arch/x86/mm.c            |  24 +---------
 xen/arch/x86/traps.c         |   7 +--
 xen/include/asm-x86/domain.h |  22 +++++----
 4 files changed, 78 insertions(+), 83 deletions(-)

diff --git a/xen/arch/x86/domain.c b/xen/arch/x86/domain.c
index 408fc0bc13..6c38d847a3 100644
--- a/xen/arch/x86/domain.c
+++ b/xen/arch/x86/domain.c
@@ -377,6 +377,60 @@ static void release_compat_l4(struct vcpu *v)
     v->arch.guest_table_user = pagetable_null();
 }
 
+unsigned long pv_fixup_guest_cr4(const struct vcpu *v, unsigned long cr4)
+{
+    unsigned int leaf1_ecx = 0, leaf1_edx = 0;
+    unsigned int leaf7_0_ebx = 0;
+
+    pv_cpuid(1, 0, NULL, NULL, &leaf1_ecx, &leaf1_edx);
+    pv_cpuid(7, 0, NULL, &leaf7_0_ebx, NULL, NULL);
+
+    /* Discard attempts to set guest controllable bits outside of the policy. 
*/
+    cr4 &= ~(((leaf1_edx & cpufeat_mask(X86_FEATURE_TSC))
+              ? 0 : X86_CR4_TSD) |
+             ((leaf1_edx & cpufeat_mask(X86_FEATURE_DE))
+              ? 0 : X86_CR4_DE) |
+             ((leaf7_0_ebx & cpufeat_mask(X86_FEATURE_FSGSBASE))
+              ? 0 : X86_CR4_FSGSBASE) |
+             ((leaf1_ecx & cpufeat_mask(X86_FEATURE_XSAVE))
+              ? 0 : X86_CR4_OSXSAVE));
+
+    /* Masks expected to be disjoint sets. */
+    BUILD_BUG_ON(PV_CR4_GUEST_MASK & PV_CR4_GUEST_VISIBLE_MASK);
+
+    /*
+     * A guest sees the policy subset of its own choice of guest controllable
+     * bits, and a subset of Xen's choice of certain hardware settings.
+     */
+    return ((cr4 & PV_CR4_GUEST_MASK) |
+            (mmu_cr4_features & PV_CR4_GUEST_VISIBLE_MASK));
+}
+
+unsigned long pv_make_cr4(const struct vcpu *v)
+{
+    const struct domain *d = v->domain;
+    unsigned long cr4 = mmu_cr4_features &
+        ~(X86_CR4_PCIDE | X86_CR4_PGE | X86_CR4_TSD);
+
+    /*
+     * PCIDE or PGE depends on the PCID/XPTI settings, but must not both be
+     * set, as it impacts the safety of TLB flushing.
+     */
+    if ( d->arch.pv_domain.pcid )
+        cr4 |= X86_CR4_PCIDE;
+    else if ( !d->arch.pv_domain.xpti )
+        cr4 |= X86_CR4_PGE;
+
+    /*
+     * TSD is needed if either the guest has elected to use it, or Xen is
+     * virtualising the TSC value the guest sees.
+     */
+    if ( d->arch.vtsc || (v->arch.pv_vcpu.ctrlreg[4] & X86_CR4_TSD) )
+        cr4 |= X86_CR4_TSD;
+
+    return cr4;
+}
+
 static void set_domain_xpti(struct domain *d)
 {
     if ( is_pv_32bit_domain(d) )
@@ -564,6 +618,8 @@ int vcpu_initialise(struct vcpu *v)
 
         /* PV guests by default have a 100Hz ticker. */
         v->periodic_period = MILLISECS(10);
+
+        v->arch.pv_vcpu.ctrlreg[4] = pv_fixup_guest_cr4(v, 0);
     }
 
     v->arch.schedule_tail = continue_nonidle_domain;
@@ -576,8 +632,6 @@ int vcpu_initialise(struct vcpu *v)
         v->arch.cr3           = __pa(idle_pg_table);
     }
 
-    v->arch.pv_vcpu.ctrlreg[4] = real_cr4_to_pv_guest_cr4(mmu_cr4_features);
-
     if ( is_pv_32bit_domain(d) )
     {
         if ( (rc = setup_compat_arg_xlat(v)) )
@@ -955,49 +1009,6 @@ int arch_domain_soft_reset(struct domain *d)
     return ret;
 }
 
-/*
- * These are the masks of CR4 bits (subject to hardware availability) which a
- * PV guest may not legitimiately attempt to modify.
- */
-static unsigned long __read_mostly pv_cr4_mask, compat_pv_cr4_mask;
-
-static int __init init_pv_cr4_masks(void)
-{
-    unsigned long common_mask = ~X86_CR4_TSD;
-
-    /*
-     * All PV guests may attempt to modify TSD, DE and OSXSAVE.
-     */
-    if ( cpu_has_de )
-        common_mask &= ~X86_CR4_DE;
-    if ( cpu_has_xsave )
-        common_mask &= ~X86_CR4_OSXSAVE;
-
-    pv_cr4_mask = compat_pv_cr4_mask = common_mask;
-
-    /*
-     * 64bit PV guests may attempt to modify FSGSBASE.
-     */
-    if ( cpu_has_fsgsbase )
-        pv_cr4_mask &= ~X86_CR4_FSGSBASE;
-
-    return 0;
-}
-__initcall(init_pv_cr4_masks);
-
-unsigned long pv_guest_cr4_fixup(const struct vcpu *v, unsigned long guest_cr4)
-{
-    unsigned long hv_cr4 = real_cr4_to_pv_guest_cr4(read_cr4());
-    unsigned long mask = is_pv_32bit_vcpu(v) ? compat_pv_cr4_mask : 
pv_cr4_mask;
-
-    if ( (guest_cr4 & mask) != (hv_cr4 & mask) )
-        printk(XENLOG_G_WARNING
-               "d%d attempted to change %pv's CR4 flags %08lx -> %08lx\n",
-               current->domain->domain_id, v, hv_cr4, guest_cr4);
-
-    return (hv_cr4 & mask) | (guest_cr4 & ~mask);
-}
-
 #define xen_vcpu_guest_context vcpu_guest_context
 #define fpu_ctxt fpu_ctxt.x
 CHECK_FIELD_(struct, vcpu_guest_context, fpu_ctxt);
@@ -1011,7 +1022,7 @@ int arch_set_info_guest(
     struct domain *d = v->domain;
     unsigned long cr3_gfn;
     struct page_info *cr3_page;
-    unsigned long flags, cr4;
+    unsigned long flags;
     unsigned int i;
     int rc = 0, compat;
 
@@ -1228,9 +1239,8 @@ int arch_set_info_guest(
     v->arch.pv_vcpu.ctrlreg[0] &= X86_CR0_TS;
     v->arch.pv_vcpu.ctrlreg[0] |= read_cr0() & ~X86_CR0_TS;
 
-    cr4 = v->arch.pv_vcpu.ctrlreg[4];
-    v->arch.pv_vcpu.ctrlreg[4] = cr4 ? pv_guest_cr4_fixup(v, cr4) :
-        real_cr4_to_pv_guest_cr4(mmu_cr4_features);
+    v->arch.pv_vcpu.ctrlreg[4] =
+        pv_fixup_guest_cr4(v, v->arch.pv_vcpu.ctrlreg[4]);
 
     memset(v->arch.debugreg, 0, sizeof(v->arch.debugreg));
     for ( i = 0; i < 8; i++ )
diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c
index 683a846145..ae8a9e7f60 100644
--- a/xen/arch/x86/mm.c
+++ b/xen/arch/x86/mm.c
@@ -517,33 +517,13 @@ void make_cr3(struct vcpu *v, unsigned long mfn)
         v->arch.cr3 |= get_pcid_bits(v, false);
 }
 
-unsigned long pv_guest_cr4_to_real_cr4(const struct vcpu *v)
-{
-    const struct domain *d = v->domain;
-    unsigned long cr4;
-
-    cr4 = v->arch.pv_vcpu.ctrlreg[4] & ~X86_CR4_DE;
-    cr4 |= mmu_cr4_features & (X86_CR4_PSE | X86_CR4_SMEP | X86_CR4_SMAP |
-                               X86_CR4_OSXSAVE | X86_CR4_FSGSBASE);
-
-    if ( d->arch.pv_domain.pcid )
-        cr4 |= X86_CR4_PCIDE;
-    else if ( !d->arch.pv_domain.xpti )
-        cr4 |= X86_CR4_PGE;
-
-    cr4 |= d->arch.vtsc ? X86_CR4_TSD : 0;
-
-    return cr4;
-}
-
 void write_ptbase(struct vcpu *v)
 {
     struct cpu_info *cpu_info = get_cpu_info();
     unsigned long new_cr4;
 
     new_cr4 = (is_pv_vcpu(v) && !is_idle_vcpu(v))
-              ? pv_guest_cr4_to_real_cr4(v)
-              : ((read_cr4() & ~(X86_CR4_PCIDE | X86_CR4_TSD)) | X86_CR4_PGE);
+              ? pv_make_cr4(v) : mmu_cr4_features;
 
     if ( is_pv_vcpu(v) && v->domain->arch.pv_domain.xpti )
     {
@@ -562,8 +542,6 @@ void write_ptbase(struct vcpu *v)
         switch_cr3_cr4(v->arch.cr3, new_cr4);
         cpu_info->pv_cr3 = 0;
     }
-
-    ASSERT(is_pv_vcpu(v) || read_cr4() == mmu_cr4_features);
 }
 
 /*
diff --git a/xen/arch/x86/traps.c b/xen/arch/x86/traps.c
index 77f786cef6..a8aebe5405 100644
--- a/xen/arch/x86/traps.c
+++ b/xen/arch/x86/traps.c
@@ -966,7 +966,8 @@ static void _domain_cpuid(const struct domain *currd,
                           unsigned int *eax, unsigned int *ebx,
                           unsigned int *ecx, unsigned int *edx)
 {
-    if ( !is_control_domain(currd) && !is_hardware_domain(currd) )
+    if ( !is_control_domain(currd) && !is_hardware_domain(currd) &&
+         !is_idle_domain(currd) )
         domain_cpuid(currd, leaf, subleaf, eax, ebx, ecx, edx);
     else
         cpuid_count(leaf, subleaf, eax, ebx, ecx, edx);
@@ -2372,8 +2373,8 @@ static int priv_op_write_cr(unsigned int reg, unsigned 
long val,
     }
 
     case 4: /* Write CR4 */
-        curr->arch.pv_vcpu.ctrlreg[4] = pv_guest_cr4_fixup(curr, val);
-        write_cr4(pv_guest_cr4_to_real_cr4(curr));
+        curr->arch.pv_vcpu.ctrlreg[4] = pv_fixup_guest_cr4(curr, val);
+        write_cr4(pv_make_cr4(curr));
         ctxt_switch_levelling(curr);
         return X86EMUL_OKAY;
     }
diff --git a/xen/include/asm-x86/domain.h b/xen/include/asm-x86/domain.h
index e7abaadcbc..01372838ec 100644
--- a/xen/include/asm-x86/domain.h
+++ b/xen/include/asm-x86/domain.h
@@ -642,16 +642,22 @@ bool_t update_secondary_system_time(struct vcpu *,
 void vcpu_show_execution_state(struct vcpu *);
 void vcpu_show_registers(const struct vcpu *);
 
-/* Clean up CR4 bits that are not under guest control. */
-unsigned long pv_guest_cr4_fixup(const struct vcpu *, unsigned long guest_cr4);
+/*
+ * Bits which a PV guest can toggle in its view of cr4.  Some are loaded into
+ * hardware, while some are fully emulated.
+ */
+#define PV_CR4_GUEST_MASK \
+    (X86_CR4_TSD | X86_CR4_DE | X86_CR4_FSGSBASE | X86_CR4_OSXSAVE)
+
+/* Bits which a PV guest may observe from the real hardware settings. */
+#define PV_CR4_GUEST_VISIBLE_MASK \
+    (X86_CR4_PAE | X86_CR4_MCE | X86_CR4_OSFXSR | X86_CR4_OSXMMEXCPT)
 
-/* Convert between guest-visible and real CR4 values. */
-unsigned long pv_guest_cr4_to_real_cr4(const struct vcpu *v);
+/* Given a new cr4 value, construct the resulting guest-visible cr4 value. */
+unsigned long pv_fixup_guest_cr4(const struct vcpu *v, unsigned long cr4);
 
-#define real_cr4_to_pv_guest_cr4(c)                         \
-    ((c) & ~(X86_CR4_PGE | X86_CR4_PSE | X86_CR4_TSD |      \
-             X86_CR4_OSXSAVE | X86_CR4_SMEP |               \
-             X86_CR4_FSGSBASE | X86_CR4_SMAP | X86_CR4_PCIDE))
+/* Create a cr4 value to load into hardware, based on vcpu settings. */
+unsigned long pv_make_cr4(const struct vcpu *v);
 
 void domain_cpuid(const struct domain *d,
                   unsigned int  input,
--
generated by git-patchbot for /home/xen/git/xen.git#staging-4.8

_______________________________________________
Xen-changelog mailing list
Xen-changelog@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/xen-changelog

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.