Reply-To: Sean Christopherson <seanjc@google.com>
Date:   Tue,  7 Feb 2023 00:21:53 +0000
Mime-Version: 1.0
Message-ID: <20230207002156.521736-1-seanjc@google.com>
Subject: [PATCH v2 0/3] KVM: SVM: Fix GATag bug for >256 vCPUs
From:   Sean Christopherson <seanjc@google.com>
To:     Sean Christopherson <seanjc@google.com>,
        Paolo Bonzini <pbonzini@redhat.com>
Cc:     kvm@vger.kernel.org, linux-kernel@vger.kernel.org,
        Alejandro Jimenez <alejandro.j.jimenez@oracle.com>,
        Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Fix a bug in KVM's use of the GATag where it unintentionally drops a bit
from vCPU IDs greater than 255 and as a result wakes the wrong vCPU.

Suravee and/or Alejandro, can you give this proper testing?  It's compile
tested only at this point.  I'll do basic testing before officially
applying, but AFAIK I don't have access to x2AVIC hardware, nor do I have
a ready-to-go configuration to properly exercise this code.

Thanks!

Sean Christopherson (2):
  KVM: SVM: Fix a benign off-by-one bug in AVIC physical table mask
  KVM: SVM: WARN if GATag generation drops VM or vCPU ID information

Suravee Suthikulpanit (1):
  KVM: SVM: Modify AVIC GATag to support max number of 512 vCPUs

 arch/x86/include/asm/svm.h | 12 +++++++-----
 arch/x86/kvm/svm/avic.c    | 37 ++++++++++++++++++++++++++++---------
 2 files changed, 35 insertions(+), 14 deletions(-)


base-commit: 32e69f232db4ca11f26e5961daeff93906ce232f
-- 
2.39.1.519.gcb327c4b5f-goog