[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ZNZtdKXslPlKYRWY@nvidia.com>
Date: Fri, 11 Aug 2023 14:18:44 -0300
From: Jason Gunthorpe <jgg@...dia.com>
To: Sean Christopherson <seanjc@...gle.com>
Cc: Yan Zhao <yan.y.zhao@...el.com>, bibo mao <maobibo@...ngson.cn>,
linux-mm@...ck.org, linux-kernel@...r.kernel.org,
kvm@...r.kernel.org, pbonzini@...hat.com, mike.kravetz@...cle.com,
apopple@...dia.com, rppt@...nel.org, akpm@...ux-foundation.org,
kevin.tian@...el.com, david@...hat.com
Subject: Re: [RFC PATCH v2 5/5] KVM: Unmap pages only when it's indeed
protected for NUMA migration
> However, *if* it's ok to invoke MMU notifiers while holding PMD/PTE locks, then
> I think we can achieve what you want without losing batching, and without changing
> secondary MMUs.
No, this is not legal.
> +void change_pmd_range_notify_secondary_mmus(unsigned long addr,
> + struct mmu_notifier_range *range)
> +{
> + if (range->start)
> + return;
> +
> + VM_WARN_ON(addr >= range->end);
> + range->start = addr;
> + mmu_notifier_invalidate_range_start_nonblock(range);
> +}
The 'nonblock' entry point is advisory, if it fails the caller must
not change the mm.
Jason
Powered by blists - more mailing lists