revert: KVM: x86: Manually calculate reserved bits when loading PDPTRS

Problematic on old Intel CPUs:
https://forum.proxmox.com/threads/pve-kernel-5-0-21-4-pve-cause-debian-guests-to-reboot-loop-on-older-intel-cpus.59377/page-3#post-276413
https://bugzilla.kernel.org/show_bug.cgi?id=205441#c1

Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
This commit is contained in:
Thomas Lamprecht 2019-11-11 11:15:41 +01:00
parent f665e167a1
commit 03505c4ba3

View File

@ -0,0 +1,50 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Thomas Lamprecht <t.lamprecht@proxmox.com>
Date: Mon, 11 Nov 2019 11:12:32 +0100
Subject: [PATCH] Revert "KVM: x86: Manually calculate reserved bits when
loading PDPTRS"
This was back-ported from the 5.4 merge window, on kernels <5.3 it
makes issues with old intel CPUs (like more than ~ 10 years ago
released ones).
Reverse bisect does not show a clear commit which would fix that, so
revert this for now.
This reverts commit 208007519a7385a57b0c0a3c180142a521594876.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
---
arch/x86/kvm/x86.c | 11 +++--------
1 file changed, 3 insertions(+), 8 deletions(-)
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index f832fa5011cd..fd00cfdfb6c3 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -672,14 +672,8 @@ static int kvm_read_nested_guest_page(struct kvm_vcpu *vcpu, gfn_t gfn,
data, offset, len, access);
}
-static inline u64 pdptr_rsvd_bits(struct kvm_vcpu *vcpu)
-{
- return rsvd_bits(cpuid_maxphyaddr(vcpu), 63) | rsvd_bits(5, 8) |
- rsvd_bits(1, 2);
-}
-
/*
- * Load the pae pdptrs. Return 1 if they are all valid, 0 otherwise.
+ * Load the pae pdptrs. Return true is they are all valid.
*/
int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3)
{
@@ -698,7 +692,8 @@ int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3)
}
for (i = 0; i < ARRAY_SIZE(pdpte); ++i) {
if ((pdpte[i] & PT_PRESENT_MASK) &&
- (pdpte[i] & pdptr_rsvd_bits(vcpu))) {
+ (pdpte[i] &
+ vcpu->arch.mmu->guest_rsvd_check.rsvd_bits_mask[0][2])) {
ret = 0;
goto out;
}