linux-kernel - Re: [PATCH] KVM: SVM: Mark SEV launch secret pages as dirty.

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Message-ID: <5ac77c46-88b4-df45-4f02-72adfb096262@redhat.com>
Date:   Wed, 23 Sep 2020 18:59:36 +0200
From:   Paolo Bonzini <pbonzini@...hat.com>
To:     Sean Christopherson <sean.j.christopherson@...el.com>,
        Cfir Cohen <cfir@...gle.com>
Cc:     "kvm @ vger . kernel . org" <kvm@...r.kernel.org>,
        Lendacky Thomas <thomas.lendacky@....com>,
        Singh Brijesh <brijesh.singh@....com>,
        Grimm Jon <Jon.Grimm@....com>,
        David Rientjes <rientjes@...gle.com>,
        Vitaly Kuznetsov <vkuznets@...hat.com>,
        Wanpeng Li <wanpengli@...cent.com>,
        Jim Mattson <jmattson@...gle.com>,
        Joerg Roedel <joro@...tes.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
        "H . Peter Anvin" <hpa@...or.com>, x86@...nel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] KVM: SVM: Mark SEV launch secret pages as dirty.

On 19/09/20 06:55, Sean Christopherson wrote:
> Side topic, while I love the comment (I do, honestly) regarding in-place
> encryption, this is the fourth? instance of the same 4-line comment (6 lines
> if you count the /* and */.  Maybe it's time to do something like
> 
> 	/* LAUNCH_SECRET does in-place encryption, see sev_clflush_pages(). */
> 
> and then have the main comment in sev_clflush_pages().  With the addition of
> X86_FEATURE_SME_COHERENT, there's even a fantastic location for the comment:

Two of the three instances are a bit different though.  What about this
which at least shortens the comment to 2 fewer lines:

diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c
index bb0e89c79a04..7b11546e65ba 100644
--- a/arch/x86/kvm/svm/sev.c
+++ b/arch/x86/kvm/svm/sev.c
@@ -446,10 +446,8 @@ static int sev_launch_update_data(struct kvm *kvm, struct kvm_sev_cmd *argp)
 	}
 
 	/*
-	 * The LAUNCH_UPDATE command will perform in-place encryption of the
-	 * memory content (i.e it will write the same memory region with C=1).
-	 * It's possible that the cache may contain the data with C=0, i.e.,
-	 * unencrypted so invalidate it first.
+	 * Flush before LAUNCH_UPDATE encrypts pages in place, in case the cache
+	 * contains the data that was written unencrypted.
 	 */
 	sev_clflush_pages(inpages, npages);
 
@@ -805,10 +803,8 @@ static int sev_dbg_crypt(struct kvm *kvm, struct kvm_sev_cmd *argp, bool dec)
 		}
 
 		/*
-		 * The DBG_{DE,EN}CRYPT commands will perform {dec,en}cryption of the
-		 * memory content (i.e it will write the same memory region with C=1).
-		 * It's possible that the cache may contain the data with C=0, i.e.,
-		 * unencrypted so invalidate it first.
+		 * Flush before DBG_{DE,EN}CRYPT reads or modifies the pages, flush the
+		 * destination too in case the cache contains its current data.
 		 */
 		sev_clflush_pages(src_p, 1);
 		sev_clflush_pages(dst_p, 1);
@@ -870,10 +866,8 @@ static int sev_launch_secret(struct kvm *kvm, struct kvm_sev_cmd *argp)
 		return PTR_ERR(pages);
 
 	/*
-	 * The LAUNCH_SECRET command will perform in-place encryption of the
-	 * memory content (i.e it will write the same memory region with C=1).
-	 * It's possible that the cache may contain the data with C=0, i.e.,
-	 * unencrypted so invalidate it first.
+	 * Flush before LAUNCH_SECRET encrypts pages in place, in case the cache
+	 * contains the data that was written unencrypted.
 	 */
 	sev_clflush_pages(pages, n);