Received-SPF: pass (google.com: domain of linux-kernel+bounces-81718-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
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 69/73] x86/pvm: Implement mmu related PVOPS
Date: Mon, 26 Feb 2024 22:36:26 +0800
Message-Id: <20240226143630.33643-70-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>

CR2 is passed directly in the event entry, allowing it to be read
directly in PVOPS. Additionally, write_cr3() for context switch needs to
notify the hypervisor in its PVOPS. For performance reasons, TLB-related
PVOPS utilize hypercalls.

Signed-off-by: Lai Jiangshan <jiangshan.ljs@antgroup.com>
Signed-off-by: Hou Wenlong <houwenlong.hwl@antgroup.com>
---
 arch/x86/kernel/pvm.c | 56 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 56 insertions(+)

diff --git a/arch/x86/kernel/pvm.c b/arch/x86/kernel/pvm.c
index b4522947374d..1dc2c0fb7daa 100644
--- a/arch/x86/kernel/pvm.c
+++ b/arch/x86/kernel/pvm.c
@@ -21,6 +21,7 @@
 #include <asm/traps.h>
 
 DEFINE_PER_CPU_PAGE_ALIGNED(struct pvm_vcpu_struct, pvm_vcpu_struct);
+static DEFINE_PER_CPU(unsigned long, pvm_guest_cr3);
 
 unsigned long pvm_range_start __initdata;
 unsigned long pvm_range_end __initdata;
@@ -153,6 +154,52 @@ static noinstr void pvm_safe_halt(void)
 	pvm_hypercall0(PVM_HC_IRQ_HALT);
 }
 
+static noinstr unsigned long pvm_read_cr2(void)
+{
+	return this_cpu_read(pvm_vcpu_struct.cr2);
+}
+
+static noinstr void pvm_write_cr2(unsigned long cr2)
+{
+	native_write_cr2(cr2);
+	this_cpu_write(pvm_vcpu_struct.cr2, cr2);
+}
+
+static unsigned long pvm_read_cr3(void)
+{
+	return this_cpu_read(pvm_guest_cr3);
+}
+
+static unsigned long pvm_user_pgd(unsigned long pgd)
+{
+	return pgd | BIT(PTI_PGTABLE_SWITCH_BIT) | BIT(X86_CR3_PTI_PCID_USER_BIT);
+}
+
+static void pvm_write_cr3(unsigned long val)
+{
+	/* Convert CR3_NO_FLUSH bit to hypercall flags. */
+	unsigned long flags = ~val >> 63;
+	unsigned long pgd = val & ~X86_CR3_PCID_NOFLUSH;
+
+	this_cpu_write(pvm_guest_cr3, pgd);
+	pvm_hypercall3(PVM_HC_LOAD_PGTBL, flags, pgd, pvm_user_pgd(pgd));
+}
+
+static void pvm_flush_tlb_user(void)
+{
+	pvm_hypercall0(PVM_HC_TLB_FLUSH_CURRENT);
+}
+
+static void pvm_flush_tlb_kernel(void)
+{
+	pvm_hypercall0(PVM_HC_TLB_FLUSH);
+}
+
+static void pvm_flush_tlb_one_user(unsigned long addr)
+{
+	pvm_hypercall1(PVM_HC_TLB_INVLPG, addr);
+}
+
 void __init pvm_early_event(struct pt_regs *regs)
 {
 	int vector = regs->orig_ax >> 32;
@@ -397,6 +444,15 @@ void __init pvm_early_setup(void)
 	pv_ops.irq.irq_enable = __PV_IS_CALLEE_SAVE(pvm_irq_enable);
 	pv_ops.irq.safe_halt = pvm_safe_halt;
 
+	this_cpu_write(pvm_guest_cr3, __native_read_cr3());
+	pv_ops.mmu.read_cr2 = __PV_IS_CALLEE_SAVE(pvm_read_cr2);
+	pv_ops.mmu.write_cr2 = pvm_write_cr2;
+	pv_ops.mmu.read_cr3 = pvm_read_cr3;
+	pv_ops.mmu.write_cr3 = pvm_write_cr3;
+	pv_ops.mmu.flush_tlb_user = pvm_flush_tlb_user;
+	pv_ops.mmu.flush_tlb_kernel = pvm_flush_tlb_kernel;
+	pv_ops.mmu.flush_tlb_one_user = pvm_flush_tlb_one_user;
+
 	wrmsrl(MSR_PVM_VCPU_STRUCT, __pa(this_cpu_ptr(&pvm_vcpu_struct)));
 	wrmsrl(MSR_PVM_EVENT_ENTRY, (unsigned long)(void *)pvm_early_kernel_event_entry - 256);
 	wrmsrl(MSR_PVM_SUPERVISOR_REDZONE, PVM_SUPERVISOR_REDZONE_SIZE);
-- 
2.19.1.6.gb485710b