linux-kernel - [PATCH 3/4] KVM: x86: Add a VM stat exposing when KVM PIT is set to reinject mode

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20240429155738.990025-4-alejandro.j.jimenez@oracle.com>
Date: Mon, 29 Apr 2024 15:57:37 +0000
From: Alejandro Jimenez <alejandro.j.jimenez@...cle.com>
To: kvm@...r.kernel.org
Cc: seanjc@...gle.com, pbonzini@...hat.com, linux-kernel@...r.kernel.org,
        suravee.suthikulpanit@....com, vashegde@....com, mlevitsk@...hat.com,
        joao.m.martins@...cle.com, boris.ostrovsky@...cle.com,
        mark.kanda@...cle.com, alejandro.j.jimenez@...cle.com
Subject: [PATCH 3/4] KVM: x86: Add a VM stat exposing when KVM PIT is set to reinject mode

Add a stat to query when PIT is in reinject mode, which can have a large
performance impact due to disabling SVM AVIC.
When using in-kernel irqchip, QEMU and KVM default to creating a PIT in
reinject mode, since this is necessary for old guest operating systems that
use the PIT for timing. Unfortunately, reinject mode relies on EOI
interception and so SVM AVIC must be inhibited when the PIT is set up using
this mode.

Signed-off-by: Alejandro Jimenez <alejandro.j.jimenez@...cle.com>
---
 arch/x86/include/asm/kvm_host.h | 1 +
 arch/x86/kvm/i8254.c            | 2 ++
 arch/x86/kvm/x86.c              | 3 ++-
 3 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index f3b40cfebec4..e7e3213cefae 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -1535,6 +1535,7 @@ struct kvm_vm_stat {
 	u64 max_mmu_page_hash_collisions;
 	u64 max_mmu_rmap_size;
 	u64 synic_auto_eoi_used;
+	u64 pit_reinject_mode;
 };
 
 struct kvm_vcpu_stat {
diff --git a/arch/x86/kvm/i8254.c b/arch/x86/kvm/i8254.c
index cd57a517d04a..44e593e909a1 100644
--- a/arch/x86/kvm/i8254.c
+++ b/arch/x86/kvm/i8254.c
@@ -316,6 +316,8 @@ void kvm_pit_set_reinject(struct kvm_pit *pit, bool reinject)
 		kvm_unregister_irq_mask_notifier(kvm, 0, &pit->mask_notifier);
 	}
 
+	kvm->stat.pit_reinject_mode = reinject;
+
 	atomic_set(&ps->reinject, reinject);
 }
 
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 27e339133068..03cb933920cb 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -257,7 +257,8 @@ const struct _kvm_stats_desc kvm_vm_stats_desc[] = {
 	STATS_DESC_ICOUNTER(VM, nx_lpage_splits),
 	STATS_DESC_PCOUNTER(VM, max_mmu_rmap_size),
 	STATS_DESC_PCOUNTER(VM, max_mmu_page_hash_collisions),
-	STATS_DESC_IBOOLEAN(VM, synic_auto_eoi_used)
+	STATS_DESC_IBOOLEAN(VM, synic_auto_eoi_used),
+	STATS_DESC_IBOOLEAN(VM, pit_reinject_mode)
 };
 
 const struct kvm_stats_header kvm_vm_stats_header = {
-- 
2.39.3