kvm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Paolo Bonzini <pbonzini@redhat.com>
To: linux-kernel@vger.kernel.org, kvm@vger.kernel.org
Cc: mlevitsk@redhat.com, seanjc@google.com, stable@vger.kernel.org,
	Felix Wilhelm <fwilhelm@google.com>
Subject: [PATCH 07/13] KVM: SEV-ES: go over the sev_pio_data buffer in multiple passes if needed
Date: Fri, 22 Oct 2021 11:36:10 -0400	[thread overview]
Message-ID: <20211022153616.1722429-8-pbonzini@redhat.com> (raw)
In-Reply-To: <20211022153616.1722429-1-pbonzini@redhat.com>

The PIO scratch buffer is larger than a single page, and therefore
it is not possible to copy it in a single step to vcpu->arch/pio_data.
Bound each call to emulator_pio_in/out to a single page; keep
track of how many I/O operations are left in vcpu->arch.sev_pio_count,
so that the operation can be restarted in the complete_userspace_io
callback.

For OUT, this means that the previous kvm_sev_es_outs implementation
becomes an iterator of the loop, and we can consume the sev_pio_data
buffer before leaving to userspace.

For IN, instead, consuming the buffer and decreasing sev_pio_count
is always done in the complete_userspace_io callback, because that
is when the memcpy is done into sev_pio_data.

Cc: stable@vger.kernel.org
Fixes: 7ed9abfe8e9f ("KVM: SVM: Support string IO operations for an SEV-ES guest")
Reported-by: Felix Wilhelm <fwilhelm@google.com>
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 arch/x86/include/asm/kvm_host.h |  1 +
 arch/x86/kvm/x86.c              | 72 +++++++++++++++++++++++++--------
 2 files changed, 57 insertions(+), 16 deletions(-)

diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index 6bed6c416c6c..5a0298aa56ba 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -703,6 +703,7 @@ struct kvm_vcpu_arch {
 	struct kvm_pio_request pio;
 	void *pio_data;
 	void *sev_pio_data;
+	unsigned sev_pio_count;
 
 	u8 event_exit_inst_len;
 
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 23e772412134..b26647a5ea22 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -12386,38 +12386,77 @@ int kvm_sev_es_mmio_read(struct kvm_vcpu *vcpu, gpa_t gpa, unsigned int bytes,
 EXPORT_SYMBOL_GPL(kvm_sev_es_mmio_read);
 
 static int kvm_sev_es_outs(struct kvm_vcpu *vcpu, unsigned int size,
-			   unsigned int port, unsigned int count)
+			   unsigned int port);
+
+static int complete_sev_es_emulated_outs(struct kvm_vcpu *vcpu)
 {
-	int ret = emulator_pio_out(vcpu, size, port,
-				   vcpu->arch.sev_pio_data, count);
+	int size = vcpu->arch.pio.size;
+	int port = vcpu->arch.pio.port;
+
+	vcpu->arch.pio.count = 0;
+	if (vcpu->arch.sev_pio_count)
+		return kvm_sev_es_outs(vcpu, size, port);
+	return 1;
+}
+
+static int kvm_sev_es_outs(struct kvm_vcpu *vcpu, unsigned int size,
+			   unsigned int port)
+{
+	for (;;) {
+		unsigned int count =
+			min_t(unsigned int, PAGE_SIZE / size, vcpu->arch.sev_pio_count);
+		int ret = emulator_pio_out(vcpu, size, port, vcpu->arch.sev_pio_data, count);
+
+		/* memcpy done already by emulator_pio_out.  */
+		vcpu->arch.sev_pio_count -= count;
+		vcpu->arch.sev_pio_data += count * vcpu->arch.pio.size;
+		if (!ret)
+			break;
 
-	if (ret) {
 		/* Emulation done by the kernel.  */
-		return ret;
+		if (!vcpu->arch.sev_pio_count)
+			return 1;
 	}
 
-	vcpu->arch.pio.count = 0;
+	vcpu->arch.complete_userspace_io = complete_sev_es_emulated_outs;
 	return 0;
 }
 
+static int kvm_sev_es_ins(struct kvm_vcpu *vcpu, unsigned int size,
+			  unsigned int port);
+
+static void advance_sev_es_emulated_ins(struct kvm_vcpu *vcpu)
+{
+	unsigned count = vcpu->arch.pio.count;
+	complete_emulator_pio_in(vcpu, vcpu->arch.sev_pio_data);
+	vcpu->arch.sev_pio_count -= count;
+	vcpu->arch.sev_pio_data += count * vcpu->arch.pio.size;
+}
+
 static int complete_sev_es_emulated_ins(struct kvm_vcpu *vcpu)
 {
-	memcpy(vcpu->arch.sev_pio_data, vcpu->arch.pio_data,
-	       vcpu->arch.pio.count * vcpu->arch.pio.size);
-	vcpu->arch.pio.count = 0;
+	int size = vcpu->arch.pio.size;
+	int port = vcpu->arch.pio.port;
 
+	advance_sev_es_emulated_ins(vcpu);
+	if (vcpu->arch.sev_pio_count)
+		return kvm_sev_es_ins(vcpu, size, port);
 	return 1;
 }
 
 static int kvm_sev_es_ins(struct kvm_vcpu *vcpu, unsigned int size,
-			  unsigned int port, unsigned int count)
+			  unsigned int port)
 {
-	int ret = emulator_pio_in(vcpu, size, port,
-				  vcpu->arch.sev_pio_data, count);
+	for (;;) {
+		unsigned int count =
+			min_t(unsigned int, PAGE_SIZE / size, vcpu->arch.sev_pio_count);
+		if (!__emulator_pio_in(vcpu, size, port, count))
+			break;
 
-	if (ret) {
 		/* Emulation done by the kernel.  */
-		return ret;
+		advance_sev_es_emulated_ins(vcpu);
+		if (!vcpu->arch.sev_pio_count)
+			return 1;
 	}
 
 	vcpu->arch.complete_userspace_io = complete_sev_es_emulated_ins;
@@ -12429,8 +12468,9 @@ int kvm_sev_es_string_io(struct kvm_vcpu *vcpu, unsigned int size,
 			 int in)
 {
 	vcpu->arch.sev_pio_data = data;
-	return in ? kvm_sev_es_ins(vcpu, size, port, count)
-		  : kvm_sev_es_outs(vcpu, size, port, count);
+	vcpu->arch.sev_pio_count = count;
+	return in ? kvm_sev_es_ins(vcpu, size, port)
+		  : kvm_sev_es_outs(vcpu, size, port);
 }
 EXPORT_SYMBOL_GPL(kvm_sev_es_string_io);
 
-- 
2.27.0



  parent reply	other threads:[~2021-10-22 15:36 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-10-22 15:36 [PATCH v2 00/13] fixes and cleanups for string I/O emulation Paolo Bonzini
2021-10-22 15:36 ` [PATCH 01/13] KVM: SEV-ES: rename guest_ins_data to sev_pio_data Paolo Bonzini
2021-10-22 15:36 ` [PATCH 02/13] KVM: x86: leave vcpu->arch.pio.count alone in emulator_pio_in_out Paolo Bonzini
2021-10-22 15:36 ` [PATCH 03/13] KVM: SEV-ES: clean up kvm_sev_es_ins/outs Paolo Bonzini
2021-10-22 15:36 ` [PATCH 04/13] KVM: x86: split the two parts of emulator_pio_in Paolo Bonzini
2021-10-22 15:36 ` [PATCH 05/13] KVM: x86: remove unnecessary arguments from complete_emulator_pio_in Paolo Bonzini
2021-10-22 15:36 ` [PATCH 06/13] KVM: SEV-ES: keep INS functions together Paolo Bonzini
2021-10-22 15:36 ` Paolo Bonzini [this message]
2021-10-22 15:36 ` [PATCH 08/13] KVM: x86: inline kernel_pio into its sole caller Paolo Bonzini
2021-10-26 13:55   ` Maxim Levitsky
2021-10-22 15:36 ` [PATCH 09/13] KVM: x86: move all vcpu->arch.pio* setup in emulator_pio_in_out Paolo Bonzini
2021-10-26 13:56   ` Maxim Levitsky
2021-10-22 15:36 ` [PATCH 10/13] KVM: x86: wean in-kernel PIO from vcpu->arch.pio* Paolo Bonzini
2021-10-26 13:56   ` Maxim Levitsky
2021-10-22 15:36 ` [PATCH 11/13] KVM: x86: wean fast IN from emulator_pio_in Paolo Bonzini
2021-10-26 13:56   ` Maxim Levitsky
2021-10-22 15:36 ` [PATCH 12/13] KVM: x86: de-underscorify __emulator_pio_in Paolo Bonzini
2021-10-26 13:56   ` Maxim Levitsky
2021-10-22 15:36 ` [PATCH 13/13] KVM: SEV-ES: reuse advance_sev_es_emulated_ins for OUT too Paolo Bonzini
2021-10-26 13:57   ` Maxim Levitsky

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20211022153616.1722429-8-pbonzini@redhat.com \
    --to=pbonzini@redhat.com \
    --cc=fwilhelm@google.com \
    --cc=kvm@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mlevitsk@redhat.com \
    --cc=seanjc@google.com \
    --cc=stable@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).