From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail-pj1-f74.google.com (mail-pj1-f74.google.com [209.85.216.74]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id C66DD7E1 for ; Wed, 7 May 2025 00:05:52 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.216.74 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1746576355; cv=none; b=RcQyKa50XPCqSy3uKCokB6JwEp8I4bwbagpAJG6imJL6tGdQlKWocnHhqdeaoGoNv2e05Ygeec6K8RKTkq3FE1FOoRFXXK6TpkKJswDEbXnRj8G2WpbjhTISDjQlDSv0D6a6wNcLQgb1qX1oJq4DD4ZfHiFz21W1fJanC/fv83g= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1746576355; c=relaxed/simple; bh=yaCcykgp16u3flZaJJz5SvxS7btAI9N+qBqGdItQ0JU=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Cc:Content-Type; b=Yicsyr9S7+U/wWeL4u0jCdhDudjcT7DabqOXQki67SmVUC6br3TgSvjqD3slY9KRWPVVkT1aKjlM43ZZ/joeeZeH8WLHCiPr22CwHbJyvPVAa0uu1Tx/vhHp6Z7/2IFSv6CdaSlYsGOKJbl3z1lNTVCD1j1j6i6IyUMQZ1nfBJY= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--seanjc.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=TCndXH77; arc=none smtp.client-ip=209.85.216.74 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--seanjc.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="TCndXH77" Received: by mail-pj1-f74.google.com with SMTP id 98e67ed59e1d1-30a59538b17so3540369a91.3 for ; Tue, 06 May 2025 17:05:52 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1746576352; x=1747181152; darn=lists.linux.dev; h=cc:to:from:subject:message-id:references:mime-version:in-reply-to :date:from:to:cc:subject:date:message-id:reply-to; bh=kf6t6m3UzjOjQqBtIkrrADeHvo58X2zFNTMCoBeF4hc=; b=TCndXH77Sw2tZDFWvh+DEe0LmgAdWxHm+Q4PGMgw6RhKwpxAehI9n8xYwxSTuRdrWs DGUedNKn1+2KnSBmAUdXhhZdIAIy+x1r92XQf9vVeRPA68OsewGTQyL7y0aedL/pH2dN 9KMAKHyEUL6ZwBlwciXqFaRIKUdxRW8v9GaYkwJ+eRwWsYxtu/fhuKj4vvS/iCELw8fE H99BgNjs795nOlR/lnE9VmBSUOL80glyMSYqhCSzgGnWiJHFH5rVaIJRL78TYX37VdVX s+VtywrQORiuiARBBdnCI7rzFsH3MLwAUxDNkj1qCCJtRt/g0qKKNm5uO6/p3p2KFBtB XRbQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1746576352; x=1747181152; h=cc:to:from:subject:message-id:references:mime-version:in-reply-to :date:x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=kf6t6m3UzjOjQqBtIkrrADeHvo58X2zFNTMCoBeF4hc=; b=LNMcCPelZVfrLbaH66Kjhi7O7YwBIPIX4TZu9gdxBL+zb/Ao/aaNkwIA0HGUJs5n5E vxOdjs6K7KTevwaZOqjd56Ih4gP5RJLkENgZX0rWlSNYiSyCbzDwcmHZR2j9ziX+YndP aZ/dX4b/+aOsAEZQGKw+1HsSAw6zZFJ98JxTgjXl5q2b2qgAeiyUsJRsbN8PuptB6Rv3 etQDqZY6MTzJi6TjAgLnzZjvKPbO2SfCI6K1baMR78FFdStnpIZoRpuLVg59/Mga1e/A M+fCpZZy6jdtxna/mT9uRBD0jOZhkOwIa+YH9lKWDwfKDy9gI9kn6btfZknPURIMCKL2 JpnA== X-Forwarded-Encrypted: i=1; AJvYcCWqqhDvDExk9czRUhVri+msXhYYizJKdznljx72D4Iqxu55iDh1NAvh3sKTSXociDAzWiBZKVU=@lists.linux.dev X-Gm-Message-State: AOJu0YzB0gJl96Nh0P6lh/iwHm9l0wm7pVh61r69XJXYTJsYnMK7l4YG SeTkO73RoIKDYMuhow2xMH7+vKRiIdrpZiodGZFvuD5SbpD57v1kUsO1ADHErJyRSptfN2jTdsj GKg== X-Google-Smtp-Source: AGHT+IFMFJ1FHxkmhNJJNI7L8Ht1/2Wzfj8wbIGxb39Uyx2QJZM+bpzpV88A2XUmDg8f6OQLimWa7W0pTy4= X-Received: from pjbnd2.prod.google.com ([2002:a17:90b:4cc2:b0:2f5:63a:4513]) (user=seanjc job=prod-delivery.src-stubby-dispatcher) by 2002:a17:90a:da87:b0:2fe:b907:3b05 with SMTP id 98e67ed59e1d1-30aac29bf2bmr1865902a91.29.1746576352169; Tue, 06 May 2025 17:05:52 -0700 (PDT) Date: Tue, 6 May 2025 17:05:50 -0700 In-Reply-To: <20250109204929.1106563-6-jthoughton@google.com> Precedence: bulk X-Mailing-List: kvmarm@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20250109204929.1106563-1-jthoughton@google.com> <20250109204929.1106563-6-jthoughton@google.com> Message-ID: Subject: Re: [PATCH v2 05/13] KVM: x86/mmu: Add support for KVM_MEM_USERFAULT From: Sean Christopherson To: James Houghton Cc: Paolo Bonzini , Jonathan Corbet , Marc Zyngier , Oliver Upton , Yan Zhao , Nikita Kalyazin , Anish Moorthy , Peter Gonda , Peter Xu , David Matlack , wei.w.wang@intel.com, kvm@vger.kernel.org, linux-doc@vger.kernel.org, linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org, kvmarm@lists.linux.dev Content-Type: text/plain; charset="us-ascii" On Thu, Jan 09, 2025, James Houghton wrote: > Adhering to the requirements of KVM Userfault: > > 1. Zap all sptes for the memslot when KVM_MEM_USERFAULT is toggled on > with kvm_arch_flush_shadow_memslot(). > 2. Only all PAGE_SIZE sptes when KVM_MEM_USERFAULT is enabled (for both > normal/GUP memory and guest_memfd memory). > 3. Reconstruct huge mappings when KVM_MEM_USERFAULT is toggled off with > kvm_mmu_recover_huge_pages(). This is the behavior when dirty logging > is disabled; remain consistent with it. > > With the new logic in kvm_mmu_slot_apply_flags(), I've simplified the > two dirty-logging-toggle checks into one, and I have dropped the > WARN_ON() that was there. > > Signed-off-by: James Houghton > --- > arch/x86/kvm/Kconfig | 1 + > arch/x86/kvm/mmu/mmu.c | 27 +++++++++++++++++++++---- > arch/x86/kvm/mmu/mmu_internal.h | 20 +++++++++++++++--- > arch/x86/kvm/x86.c | 36 ++++++++++++++++++++++++--------- > include/linux/kvm_host.h | 5 ++++- > 5 files changed, 71 insertions(+), 18 deletions(-) > > diff --git a/arch/x86/kvm/Kconfig b/arch/x86/kvm/Kconfig > index ea2c4f21c1ca..286c6825cd1c 100644 > --- a/arch/x86/kvm/Kconfig > +++ b/arch/x86/kvm/Kconfig > @@ -47,6 +47,7 @@ config KVM_X86 > select KVM_GENERIC_PRE_FAULT_MEMORY > select KVM_GENERIC_PRIVATE_MEM if KVM_SW_PROTECTED_VM > select KVM_WERROR if WERROR > + select HAVE_KVM_USERFAULT > > config KVM > tristate "Kernel-based Virtual Machine (KVM) support" > diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c > index 2401606db260..5cab2785b97f 100644 > --- a/arch/x86/kvm/mmu/mmu.c > +++ b/arch/x86/kvm/mmu/mmu.c > @@ -4280,14 +4280,19 @@ static inline u8 kvm_max_level_for_order(int order) > return PG_LEVEL_4K; > } > > -static u8 kvm_max_private_mapping_level(struct kvm *kvm, kvm_pfn_t pfn, > - u8 max_level, int gmem_order) > +static u8 kvm_max_private_mapping_level(struct kvm *kvm, > + struct kvm_memory_slot *slot, > + kvm_pfn_t pfn, u8 max_level, > + int gmem_order) > { > u8 req_max_level; > > if (max_level == PG_LEVEL_4K) > return PG_LEVEL_4K; > > + if (kvm_memslot_userfault(slot)) Unless I'm missing something, this can go in kvm_mmu_hugepage_adjust(): diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c index a4439e9e0726..49eb6b9b268c 100644 --- a/arch/x86/kvm/mmu/mmu.c +++ b/arch/x86/kvm/mmu/mmu.c @@ -3304,7 +3304,7 @@ void kvm_mmu_hugepage_adjust(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault if (is_error_noslot_pfn(fault->pfn)) return; - if (kvm_slot_dirty_track_enabled(slot)) + if (kvm_slot_dirty_track_enabled(slot) || kvm_is_userfault_memslot(slot)) return; > static inline int kvm_mmu_do_page_fault(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa, > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c > index 1b04092ec76a..2abb425a6514 100644 > --- a/arch/x86/kvm/x86.c > +++ b/arch/x86/kvm/x86.c > @@ -13053,12 +13053,36 @@ static void kvm_mmu_slot_apply_flags(struct kvm *kvm, > u32 new_flags = new ? new->flags : 0; > bool log_dirty_pages = new_flags & KVM_MEM_LOG_DIRTY_PAGES; > > + /* > + * When toggling KVM Userfault on, zap all sptes so that userfault-ness > + * will be respected at refault time. All new faults will only install > + * small sptes. Therefore, when toggling it off, recover hugepages. > + * > + * For MOVE and DELETE, there will be nothing to do, as the old > + * mappings will have already been deleted by > + * kvm_arch_flush_shadow_memslot(). > + * > + * For CREATE, no mappings will have been created yet. > + */ Eh, trim this down and the reference the comment below to explain why FLAGS_ONLY is the only case that needs to be handled. > + if ((old_flags ^ new_flags) & KVM_MEM_USERFAULT && > + (change == KVM_MR_FLAGS_ONLY)) { > + if (old_flags & KVM_MEM_USERFAULT) > + kvm_mmu_recover_huge_pages(kvm, new); > + else > + kvm_arch_flush_shadow_memslot(kvm, old); The call to kvm_arch_flush_shadow_memslot() should definitely go in common code. The fancy recovery logic is arch specific, but blasting the memslot when userfault is toggled on is not.