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 A677A2EBBA9 for ; Wed, 22 Oct 2025 18:12:52 +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=1761156774; cv=none; b=IOBeQL4LQCDFMZz+dRraXUOCMsfLWCGFHrd3FLDKjc/s96+1gindh90YBkLd3yTFSgF9ygneRZ2+sNZ8SynXYVTcot8DJ8njC/537WPcI6o9a1XGF6IAYmvx8GcJUNzb7r2W2PpuESHBhXDdXS4wmZvsY9jVF9/T97xZe+zAJEg= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1761156774; c=relaxed/simple; bh=UmH1svExxe3696H3/lrjYo2eB/zUc0gH+XKx1eKRj0s=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Cc:Content-Type; b=Bg2RGhaaMXKIa6vKREMNaTxsgDrlZTg4gdA7Q/mV6hwOc35i7ffG2paw/NI5dJ39LrfAZykgI5VWESm0TL/xety+LNQEPXrGbjkRrCMVIzCQjt8WRfRauRqHaAm7FbHAdCeDDCAIUrawp0yXc4nql2PJFFmWFbXlGuyW8x1lEdU= 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=y7QAMZe/; 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="y7QAMZe/" Received: by mail-pj1-f74.google.com with SMTP id 98e67ed59e1d1-32eb864fe90so2978522a91.3 for ; Wed, 22 Oct 2025 11:12:52 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20230601; t=1761156772; x=1761761572; 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=V5F7t24lC7jgleOK3FW+MwSMdfYKm48DJa8i0z/06to=; b=y7QAMZe/QMxl9khd5tm4q1dq2HnNpPIvZur6ltSBDaOstKUNGgVz/nz8h00Qn3Tw1R yT6J/nhDiC/NF7P1xS/qe8fEXJhEdm3f8ds2zwvfWs3/JPoUxd19m1BuBwA6X3HXoncw ALDS72j7x+TEHZK14rgZWbv7LFbaZ7N4g37A0auEnQjuCf3Q654WuTg/96h6TAxlG4tK oC1XoR8DFk+qNDyFhAto+u6tjU7BDzF4A5uQ495H7HHlsnfnN+/mxKCNui5+I/1p8xBC OjVmc14HZB7i/T0LclvERNNpKSl8CubUKz4Mkoc+dtpThkvxRJSUvoKVwvj1kRwf9y0Q B50g== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1761156772; x=1761761572; 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=V5F7t24lC7jgleOK3FW+MwSMdfYKm48DJa8i0z/06to=; b=BAPVQ8+INmrAAhFZ3Q8196b54Q4GQK3ur2ppgmYqQOUrW2yLewesHkhwRpbOVY5CjM v0hyckmr40fVGnohZJTiPnL8y0+y+TIza7BC2rINNe7rLbz0CMAzY7MojPONg2Q6bJP7 cmKPqUGOwCUx2bZNSg3228G3s3W4O+axsBJTIHX/VlalUdz8MKOjWuMB8zOiQjzLAaED BfYcYkMK2nzD9B/iGRi5Pvv9s5rev5doSxzbBV4Xq+T/4E1U/YQxPAXtjIZgyQW+xPjs 7X9V80RWIQg35wSn7pM46/p0muwp8vH3QZxOrRBnuUfQwlH4Oba+CMXxyYpi9XUSmb32 UtQg== X-Forwarded-Encrypted: i=1; AJvYcCW5MkyEd03ziEfOAaYtLWiYxmm4pwipy54uTHXD+gn42ihnrUHzba1BqVR+zwFWIWS3v2iwebs5HoZQ@lists.linux.dev X-Gm-Message-State: AOJu0Yx1VBBjWO7Dq1NwnM6aT0ItSjKU10aHY2p/khh6ylJR0QKenGNJ LLexMOExfXqJ8mO33rsmvaAwOM5yzWdnwDhscvMY6dFLuFVpj6GxLGw5cPZ4wyeiaHOnud9bCxl P7zNFxQ== X-Google-Smtp-Source: AGHT+IFAZr0wpbGx7b8wp6kFow05WwkzX+gl9q/tJGKoWGxIdF2WxHsL5OqcbuH+85tLuHrz8145UQS6wJM= X-Received: from pjbnm19.prod.google.com ([2002:a17:90b:19d3:b0:33d:cdb9:9adf]) (user=seanjc job=prod-delivery.src-stubby-dispatcher) by 2002:a17:90b:3f8d:b0:33b:be31:8193 with SMTP id 98e67ed59e1d1-33bcf85d59dmr30647063a91.6.1761156768896; Wed, 22 Oct 2025 11:12:48 -0700 (PDT) Date: Wed, 22 Oct 2025 11:12:47 -0700 In-Reply-To: Precedence: bulk X-Mailing-List: linux-coco@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20251017003244.186495-1-seanjc@google.com> <20251017003244.186495-5-seanjc@google.com> Message-ID: Subject: Re: [PATCH v3 04/25] KVM: x86/mmu: Add dedicated API to map guest_memfd pfn into TDP MMU From: Sean Christopherson To: Yan Zhao Cc: Marc Zyngier , Oliver Upton , Tianrui Zhao , Bibo Mao , Huacai Chen , Madhavan Srinivasan , Anup Patel , Paul Walmsley , Palmer Dabbelt , Albert Ou , Christian Borntraeger , Janosch Frank , Claudio Imbrenda , Paolo Bonzini , "Kirill A. Shutemov" , linux-arm-kernel@lists.infradead.org, kvmarm@lists.linux.dev, kvm@vger.kernel.org, loongarch@lists.linux.dev, linux-mips@vger.kernel.org, linuxppc-dev@lists.ozlabs.org, kvm-riscv@lists.infradead.org, linux-riscv@lists.infradead.org, x86@kernel.org, linux-coco@lists.linux.dev, linux-kernel@vger.kernel.org, Ira Weiny , Kai Huang , Michael Roth , Vishal Annapurve , Rick Edgecombe , Ackerley Tng , Binbin Wu Content-Type: text/plain; charset="us-ascii" On Wed, Oct 22, 2025, Yan Zhao wrote: > On Tue, Oct 21, 2025 at 09:36:52AM -0700, Sean Christopherson wrote: > > On Tue, Oct 21, 2025, Yan Zhao wrote: > > > On Thu, Oct 16, 2025 at 05:32:22PM -0700, Sean Christopherson wrote: > > > > diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c > > > > index 18d69d48bc55..ba5cca825a7f 100644 > > > > --- a/arch/x86/kvm/mmu/mmu.c > > > > +++ b/arch/x86/kvm/mmu/mmu.c > > > > @@ -5014,6 +5014,65 @@ long kvm_arch_vcpu_pre_fault_memory(struct kvm_vcpu *vcpu, > > > > return min(range->size, end - range->gpa); > > > > } > > > > > > > > +int kvm_tdp_mmu_map_private_pfn(struct kvm_vcpu *vcpu, gfn_t gfn, kvm_pfn_t pfn) > > > > +{ > > > > + struct kvm_page_fault fault = { > > > > + .addr = gfn_to_gpa(gfn), > > > > + .error_code = PFERR_GUEST_FINAL_MASK | PFERR_PRIVATE_ACCESS, > > > > + .prefetch = true, > > > > + .is_tdp = true, > > > > + .nx_huge_page_workaround_enabled = is_nx_huge_page_enabled(vcpu->kvm), > > > > + > > > > + .max_level = PG_LEVEL_4K, > > > > + .req_level = PG_LEVEL_4K, > > > > + .goal_level = PG_LEVEL_4K, > > > > + .is_private = true, > > > > + > > > > + .gfn = gfn, > > > > + .slot = kvm_vcpu_gfn_to_memslot(vcpu, gfn), > > > > + .pfn = pfn, > > > > + .map_writable = true, > > > > + }; > > > > + struct kvm *kvm = vcpu->kvm; > > > > + int r; > > > > + > > > > + lockdep_assert_held(&kvm->slots_lock); > > > Do we need to assert that filemap_invalidate_lock() is held as well? > > > > Hrm, a lockdep assertion would be nice to have, but it's obviously not strictly > > necessary, and I'm not sure it's worth the cost. To safely assert, KVM would need > Not sure. Maybe just add a comment? > But even with kvm_assert_gmem_invalidate_lock_held() and > lockdep_assert_held(&kvm->slots_lock), it seems that > kvm_tdp_mmu_map_private_pfn() still can't guarantee that the pfn is not stale. At some point we have to assume correctness. E.g. one could also argue that holding every locking in the universe still doesn't ensure the pfn is fresh, because theoretically guest_memfd could violate the locking scheme. Aha! And to further harden and document this code, this API can be gated on CONFIG_KVM_GUEST_MEMFD=y, as pointed out by the amazing-as-always test bot: https://lore.kernel.org/all/202510221928.ikBXHGCf-lkp@intel.com We could go a step further and gate it on CONFIG_KVM_INTEL_TDX=y, but I don't like that idea as I think it'd would be a net negative in terms of documenation, compared to checking CONFIG_KVM_GUEST_MEMFD. And in general I don't want to set a precedent of ifdef-ing common x86 based on what vendor code _currently_ needs an API. > e.g., if hypothetically those locks were released and re-acquired after getting > the pfn. > > > to first assert that the file refcount is elevated, e.g. to guard against > > guest_memfd _really_ screwing up and not grabbing a reference to the underlying > > file. > > > > E.g. it'd have to be something like this: > > > > diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c > > index 94d7f32a03b6..5d46b2ac0292 100644 > > --- a/arch/x86/kvm/mmu/mmu.c > > +++ b/arch/x86/kvm/mmu/mmu.c > > @@ -5014,6 +5014,18 @@ long kvm_arch_vcpu_pre_fault_memory(struct kvm_vcpu *vcpu, > > return min(range->size, end - range->gpa); > > } > > > > +static void kvm_assert_gmem_invalidate_lock_held(struct kvm_memory_slot *slot) > > +{ > > +#ifdef CONFIG_PROVE_LOCKING > > + if (WARN_ON_ONCE(!kvm_slot_has_gmem(slot)) || > > + WARN_ON_ONCE(!slot->gmem.file) || > > + WARN_ON_ONCE(!file_count(slot->gmem.file))) > > + return; > > + > > + lockdep_assert_held(file_inode(&slot->gmem.file)->i_mapping->invalidate_lock)); > lockdep_assert_held(&file_inode(slot->gmem.file)->i_mapping->invalidate_lock); > > +#endif > > +} > > + > > int kvm_tdp_mmu_map_private_pfn(struct kvm_vcpu *vcpu, gfn_t gfn, kvm_pfn_t pfn) > > { > > struct kvm_page_fault fault = { > > @@ -5038,6 +5050,8 @@ int kvm_tdp_mmu_map_private_pfn(struct kvm_vcpu *vcpu, gfn_t gfn, kvm_pfn_t pfn) > > > > lockdep_assert_held(&kvm->slots_lock); > > > > + kvm_assert_gmem_invalidate_lock_held(fault.slot); > > + > > if (KVM_BUG_ON(!tdp_mmu_enabled, kvm)) > > return -EIO; > > -- > > > > Which I suppose isn't that terrible? > Is it good if we test is_page_fault_stale()? e.g., No, because it can only get false positives, e.g. if an mmu_notifier invalidation on shared, non-guest_memfd memory. Though a sanity check would be nice to have; I believe we can simply do: diff --git a/arch/x86/kvm/mmu/tdp_mmu.c b/arch/x86/kvm/mmu/tdp_mmu.c index c5734ca5c17d..440fd8f80397 100644 --- a/arch/x86/kvm/mmu/tdp_mmu.c +++ b/arch/x86/kvm/mmu/tdp_mmu.c @@ -1273,6 +1273,8 @@ int kvm_tdp_mmu_map(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault) struct kvm_mmu_page *sp; int ret = RET_PF_RETRY; + KVM_MMU_WARN_ON(!root || root->role.invalid); + kvm_mmu_hugepage_adjust(vcpu, fault); trace_kvm_mmu_spte_requested(fault);