MIME-Version: 1.0
References: <20211205133039.GD33002@xsang-OptiPlex-9020> <56b9d000-8743-52cb-4f10-4d3fa2b30f29@redhat.com>
In-Reply-To: <56b9d000-8743-52cb-4f10-4d3fa2b30f29@redhat.com>
From:   David Matlack <dmatlack@google.com>
Date:   Mon, 6 Dec 2021 09:19:52 -0800
Message-ID: <CALzav=dn-Oe1v9qTp=ag92Kn96JOb3AX9JJA4P5VcLksV8-vLw@mail.gmail.com>
Subject: Re: [KVM] d3750a0923: WARNING:possible_circular_locking_dependency_detected
To:     Paolo Bonzini <pbonzini@redhat.com>
Cc:     kernel test robot <oliver.sang@intel.com>,
        0day robot <lkp@intel.com>,
        LKML <linux-kernel@vger.kernel.org>, lkp@lists.01.org,
        kvm@vger.kernel.org, Ben Gardon <bgardon@google.com>,
        Joerg Roedel <joro@8bytes.org>,
        Jim Mattson <jmattson@google.com>,
        Wanpeng Li <wanpengli@tencent.com>,
        Vitaly Kuznetsov <vkuznets@redhat.com>,
        Sean Christopherson <seanjc@google.com>,
        Janis Schoetterl-Glausch <scgl@linux.vnet.ibm.com>,
        Junaid Shahid <junaids@google.com>,
        Oliver Upton <oupton@google.com>,
        Harish Barathvajasankar <hbarath@google.com>,
        Peter Xu <peterx@redhat.com>, Peter Shier <pshier@google.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Sun, Dec 5, 2021 at 10:55 PM Paolo Bonzini <pbonzini@redhat.com> wrote:
>
> On 12/5/21 14:30, kernel test robot wrote:
> >
> > Chain exists of:
> >   fs_reclaim --> mmu_notifier_invalidate_range_start --> &(kvm)->mmu_lock
> >
> >  Possible unsafe locking scenario:
> >
> >        CPU0                    CPU1
> >        ----                    ----
> >   lock(&(kvm)->mmu_lock);
> >                                lock(mmu_notifier_invalidate_range_start);
> >                                lock(&(kvm)->mmu_lock);
> >   lock(fs_reclaim);
> >
>
> David, this is yours; basically, kvm_mmu_topup_memory_cache must be
> called outside the mmu_lock.

Ah, I see. kvm_arch_mmu_enable_log_dirty_pt_masked is called with
mmu_lock already held. I'll make sure to address this in v1. In theory
this should just go away when I switch away from using split_caches to
Sean's suggestion of allocating under the mmu_lock with reclaim
disabled.