Received-SPF: pass (google.com: domain of linux-kernel+bounces-81659-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
From: Lai Jiangshan <jiangshanlai@gmail.com>
To: linux-kernel@vger.kernel.org
Cc: Lai Jiangshan <jiangshan.ljs@antgroup.com>,
	Hou Wenlong <houwenlong.hwl@antgroup.com>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Peter Zijlstra <peterz@infradead.org>,
	Sean Christopherson <seanjc@google.com>,
	Thomas Gleixner <tglx@linutronix.de>,
	Borislav Petkov <bp@alien8.de>,
	Ingo Molnar <mingo@redhat.com>,
	kvm@vger.kernel.org,
	Paolo Bonzini <pbonzini@redhat.com>,
	x86@kernel.org,
	Kees Cook <keescook@chromium.org>,
	Juergen Gross <jgross@suse.com>,
	Dave Hansen <dave.hansen@linux.intel.com>,
	"H. Peter Anvin" <hpa@zytor.com>
Subject: [RFC PATCH 16/73] KVM: x86/PVM: Implement host mmu initialization
Date: Mon, 26 Feb 2024 22:35:33 +0800
Message-Id: <20240226143630.33643-17-jiangshanlai@gmail.com>
In-Reply-To: <20240226143630.33643-1-jiangshanlai@gmail.com>
References: <20240226143630.33643-1-jiangshanlai@gmail.com>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit

From: Lai Jiangshan <jiangshan.ljs@antgroup.com>

For PVM, it utilizes shadow paging as MMU virtualization for the guest.
As the switcher supports guest/host world switches, the host kernel
mapping should be cloned into the guest shadow paging table, similar to
PTI. For simplicity, only the PGD level is cloned. Additionally, the
guest Linux kernel runs in high address space, so PVM will reserve a
kernel virtual area in the host vmalloc area for the guest, also at the
PGD level.

Signed-off-by: Lai Jiangshan <jiangshan.ljs@antgroup.com>
Signed-off-by: Hou Wenlong <houwenlong.hwl@antgroup.com>
---
 arch/x86/kvm/Makefile       |   2 +-
 arch/x86/kvm/pvm/host_mmu.c | 119 ++++++++++++++++++++++++++++++++++++
 arch/x86/kvm/pvm/pvm.h      |  23 +++++++
 3 files changed, 143 insertions(+), 1 deletion(-)
 create mode 100644 arch/x86/kvm/pvm/host_mmu.c
 create mode 100644 arch/x86/kvm/pvm/pvm.h

diff --git a/arch/x86/kvm/Makefile b/arch/x86/kvm/Makefile
index 036458a27d5e..706ccf3eca45 100644
--- a/arch/x86/kvm/Makefile
+++ b/arch/x86/kvm/Makefile
@@ -33,7 +33,7 @@ ifdef CONFIG_HYPERV
 kvm-amd-y		+= svm/svm_onhyperv.o
 endif
 
-kvm-pvm-y 		+= pvm/pvm.o
+kvm-pvm-y 		+= pvm/pvm.o pvm/host_mmu.o
 
 obj-$(CONFIG_KVM)	+= kvm.o
 obj-$(CONFIG_KVM_INTEL)	+= kvm-intel.o
diff --git a/arch/x86/kvm/pvm/host_mmu.c b/arch/x86/kvm/pvm/host_mmu.c
new file mode 100644
index 000000000000..35e97f4f7055
--- /dev/null
+++ b/arch/x86/kvm/pvm/host_mmu.c
@@ -0,0 +1,119 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * PVM host mmu implementation
+ *
+ * Copyright (C) 2020 Ant Group
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2.  See
+ * the COPYING file in the top-level directory.
+ *
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/vmalloc.h>
+
+#include <asm/cpufeature.h>
+#include <asm/vsyscall.h>
+#include <asm/pgtable.h>
+
+#include "mmu.h"
+#include "mmu/spte.h"
+#include "pvm.h"
+
+static struct vm_struct *pvm_va_range_l4;
+
+u32 pml4_index_start;
+u32 pml4_index_end;
+u32 pml5_index_start;
+u32 pml5_index_end;
+
+static int __init guest_address_space_init(void)
+{
+	if (IS_ENABLED(CONFIG_KASAN_VMALLOC)) {
+		pr_warn("CONFIG_KASAN_VMALLOC is not compatible with PVM");
+		return -1;
+	}
+
+	pvm_va_range_l4 = get_vm_area_align(DEFAULT_RANGE_L4_SIZE, PT_L4_SIZE,
+			  VM_ALLOC|VM_NO_GUARD);
+	if (!pvm_va_range_l4)
+		return -1;
+
+	pml4_index_start = __PT_INDEX((u64)pvm_va_range_l4->addr, 4, 9);
+	pml4_index_end = __PT_INDEX((u64)pvm_va_range_l4->addr + (u64)pvm_va_range_l4->size, 4, 9);
+	pml5_index_start = 0x1ff;
+	pml5_index_end = 0x1ff;
+	return 0;
+}
+
+static __init void clone_host_mmu(u64 *spt, u64 *host, int index_start, int index_end)
+{
+	int i;
+
+	for (i = PTRS_PER_PGD/2; i < PTRS_PER_PGD; i++) {
+		/* clone only the range that doesn't belong to guest */
+		if (i >= index_start && i < index_end)
+			continue;
+
+		/* remove userbit from host mmu, which also disable VSYSCALL page */
+		spt[i] = host[i] & ~(_PAGE_USER | SPTE_MMU_PRESENT_MASK);
+	}
+}
+
+u64 *host_mmu_root_pgd;
+u64 *host_mmu_la57_top_p4d;
+
+int __init host_mmu_init(void)
+{
+	u64 *host_pgd;
+
+	if (guest_address_space_init() < 0)
+		return -ENOMEM;
+
+	if (!boot_cpu_has(X86_FEATURE_PTI))
+		host_pgd = (void *)current->mm->pgd;
+	else
+		host_pgd = (void *)kernel_to_user_pgdp(current->mm->pgd);
+
+	host_mmu_root_pgd = (void *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
+
+	if (!host_mmu_root_pgd) {
+		host_mmu_destroy();
+		return -ENOMEM;
+	}
+	if (pgtable_l5_enabled()) {
+		host_mmu_la57_top_p4d = (void *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
+		if (!host_mmu_la57_top_p4d) {
+			host_mmu_destroy();
+			return -ENOMEM;
+		}
+
+		clone_host_mmu(host_mmu_root_pgd, host_pgd, pml5_index_start, pml5_index_end);
+		clone_host_mmu(host_mmu_la57_top_p4d, __va(host_pgd[511] & SPTE_BASE_ADDR_MASK),
+				pml4_index_start, pml4_index_end);
+	} else {
+		clone_host_mmu(host_mmu_root_pgd, host_pgd, pml4_index_start, pml4_index_end);
+	}
+
+	if (pgtable_l5_enabled()) {
+		pr_warn("Supporting for LA57 host is not fully implemented yet.\n");
+		host_mmu_destroy();
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+void host_mmu_destroy(void)
+{
+	if (pvm_va_range_l4)
+		free_vm_area(pvm_va_range_l4);
+	if (host_mmu_root_pgd)
+		free_page((unsigned long)(void *)host_mmu_root_pgd);
+	if (host_mmu_la57_top_p4d)
+		free_page((unsigned long)(void *)host_mmu_la57_top_p4d);
+	pvm_va_range_l4 = NULL;
+	host_mmu_root_pgd = NULL;
+	host_mmu_la57_top_p4d = NULL;
+}
diff --git a/arch/x86/kvm/pvm/pvm.h b/arch/x86/kvm/pvm/pvm.h
new file mode 100644
index 000000000000..7a3732986a6d
--- /dev/null
+++ b/arch/x86/kvm/pvm/pvm.h
@@ -0,0 +1,23 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __KVM_X86_PVM_H
+#define __KVM_X86_PVM_H
+
+#define PT_L4_SHIFT		39
+#define PT_L4_SIZE		(1UL << PT_L4_SHIFT)
+#define DEFAULT_RANGE_L4_SIZE	(32 * PT_L4_SIZE)
+
+#define PT_L5_SHIFT		48
+#define PT_L5_SIZE		(1UL << PT_L5_SHIFT)
+#define DEFAULT_RANGE_L5_SIZE	(32 * PT_L5_SIZE)
+
+extern u32 pml4_index_start;
+extern u32 pml4_index_end;
+extern u32 pml5_index_start;
+extern u32 pml5_index_end;
+
+extern u64 *host_mmu_root_pgd;
+
+void host_mmu_destroy(void);
+int host_mmu_init(void);
+
+#endif /* __KVM_X86_PVM_H */
-- 
2.19.1.6.gb485710b