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

[xen master] x86/cpu-policy: Fix visibility of HTT/CMP_LEGACY in max policies



commit e2d8a652251660c3252d92b442e1a9c5d6e6a1e9
Author:     Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
AuthorDate: Wed Feb 28 20:22:24 2024 +0000
Commit:     Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
CommitDate: Fri Mar 1 20:14:19 2024 +0000

    x86/cpu-policy: Fix visibility of HTT/CMP_LEGACY in max policies
    
    The block in recalculate_cpuid_policy() predates the proper split between
    default and max policies, and was a "slightly max for a toolstack which 
knows
    about it" capability.  It didn't get transformed properly in Xen 4.14.
    
    Because Xen will accept a VM with HTT/CMP_LEGACY seen, they should be 
visible
    in the max polices.  Keep the default policy matching host settings.
    
    This manifested as an incorrectly-rejected migration across XenServer's Xen
    4.13 -> 4.17 upgrade, as Xapi is slowly growing the logic to check a VM
    against the target max policy.
    
    Signed-off-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
    Reviewed-by: Roger Pau Monné <roger.pau@xxxxxxxxxx>
---
 xen/arch/x86/cpu-policy.c | 29 ++++++++++++++++++++++-------
 1 file changed, 22 insertions(+), 7 deletions(-)

diff --git a/xen/arch/x86/cpu-policy.c b/xen/arch/x86/cpu-policy.c
index a4327abd13..609db6946f 100644
--- a/xen/arch/x86/cpu-policy.c
+++ b/xen/arch/x86/cpu-policy.c
@@ -454,6 +454,16 @@ static void __init 
guest_common_max_feature_adjustments(uint32_t *fs)
              raw_cpu_policy.feat.clwb )
             __set_bit(X86_FEATURE_CLWB, fs);
     }
+
+    /*
+     * Topology information inside the guest is entirely at the toolstack's
+     * discretion, and bears no relationship to the host we're running on.
+     *
+     * HTT identifies p->basic.lppp as valid
+     * CMP_LEGACY identifies p->extd.nc as valid
+     */
+    __set_bit(X86_FEATURE_HTT, fs);
+    __set_bit(X86_FEATURE_CMP_LEGACY, fs);
 }
 
 static void __init guest_common_default_feature_adjustments(uint32_t *fs)
@@ -490,6 +500,18 @@ static void __init 
guest_common_default_feature_adjustments(uint32_t *fs)
             __clear_bit(X86_FEATURE_CLWB, fs);
     }
 
+    /*
+     * Topology information is at the toolstack's discretion so these are
+     * unconditionally set in max, but pick a default which matches the host.
+     */
+    __clear_bit(X86_FEATURE_HTT, fs);
+    if ( cpu_has_htt )
+        __set_bit(X86_FEATURE_HTT, fs);
+
+    __clear_bit(X86_FEATURE_CMP_LEGACY, fs);
+    if ( cpu_has_cmp_legacy )
+        __set_bit(X86_FEATURE_CMP_LEGACY, fs);
+
     /*
      * On certain hardware, speculative or errata workarounds can result in
      * TSX being placed in "force-abort" mode, where it doesn't actually
@@ -837,13 +859,6 @@ void recalculate_cpuid_policy(struct domain *d)
         }
     }
 
-    /*
-     * Allow the toolstack to set HTT and CMP_LEGACY.  These bits
-     * affect how to interpret topology information in other cpuid leaves.
-     */
-    __set_bit(X86_FEATURE_HTT, max_fs);
-    __set_bit(X86_FEATURE_CMP_LEGACY, max_fs);
-
     /*
      * 32bit PV domains can't use any Long Mode features, and cannot use
      * SYSCALL on non-AMD hardware.
--
generated by git-patchbot for /home/xen/git/xen.git#master



 


Rackspace

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