1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
|
From 70ad9c5fdeac4814050080c87e06d44292ecf868 Mon Sep 17 00:00:00 2001
From: Andrew Cooper <andrew.cooper3@citrix.com>
Date: Tue, 2 Apr 2024 16:18:05 +0200
Subject: [PATCH 61/67] x86/cpu-policy: Fix visibility of HTT/CMP_LEGACY in max
policies
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
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@citrix.com>
Reviewed-by: Roger Pau Monné <roger.pau@citrix.com>
master commit: e2d8a652251660c3252d92b442e1a9c5d6e6a1e9
master date: 2024-03-01 20:14:19 +0000
---
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 ed64d56294..24acd12ce2 100644
--- a/xen/arch/x86/cpu-policy.c
+++ b/xen/arch/x86/cpu-policy.c
@@ -458,6 +458,16 @@ 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)
@@ -512,6 +522,18 @@ 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
@@ -844,13 +866,6 @@ 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.
--
2.44.0
|
GitWeb