Received-SPF: pass (google.com: domain of linux-kernel+bounces-81703-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
From: Lai Jiangshan <jiangshanlai@gmail.com>
To: linux-kernel@vger.kernel.org
Cc: Hou Wenlong <houwenlong.hwl@antgroup.com>,
	Lai Jiangshan <jiangshan.ljs@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>,
	David Woodhouse <dwmw@amazon.co.uk>,
	Brian Gerst <brgerst@gmail.com>,
	Josh Poimboeuf <jpoimboe@kernel.org>,
	Thomas Garnier <thgarnie@chromium.org>,
	Ard Biesheuvel <ardb@kernel.org>,
	Tom Lendacky <thomas.lendacky@amd.com>
Subject: [RFC PATCH 55/73] x86/pvm: Relocate kernel image to specific virtual address range
Date: Mon, 26 Feb 2024 22:36:12 +0800
Message-Id: <20240226143630.33643-56-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: Hou Wenlong <houwenlong.hwl@antgroup.com>

For a PVM guest, it is only allowed to run in the specific virtual
address range provided by the hypervisor. Therefore, the PVM guest needs
to be a PIE kernel and perform relocation during the booting process.
Additionally, for a compressed kernel image, kaslr needs to be disabled;
otherwise, it will fail to boot.

Signed-off-by: Hou Wenlong <houwenlong.hwl@antgroup.com>
Signed-off-by: Lai Jiangshan <jiangshan.ljs@antgroup.com>
---
 arch/x86/Kconfig                  |  3 ++-
 arch/x86/kernel/head64_identity.c | 27 +++++++++++++++++++++++++++
 arch/x86/kernel/head_64.S         | 13 +++++++++++++
 arch/x86/kernel/pvm.c             |  5 ++++-
 4 files changed, 46 insertions(+), 2 deletions(-)

diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 2ccc8a27e081..1b4bea3db53d 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -853,7 +853,8 @@ config KVM_GUEST
 
 config PVM_GUEST
 	bool "PVM Guest support"
-	depends on X86_64 && KVM_GUEST
+	depends on X86_64 && KVM_GUEST && X86_PIE
+	select RELOCATABLE_UNCOMPRESSED_KERNEL
 	default n
 	help
 	  This option enables the kernel to run as a PVM guest under the PVM
diff --git a/arch/x86/kernel/head64_identity.c b/arch/x86/kernel/head64_identity.c
index 4548ad615ecf..4e6a073d9e6c 100644
--- a/arch/x86/kernel/head64_identity.c
+++ b/arch/x86/kernel/head64_identity.c
@@ -20,6 +20,7 @@
 #include <asm/trapnr.h>
 #include <asm/sev.h>
 #include <asm/init.h>
+#include <asm/pvm_para.h>
 
 extern pmd_t early_dynamic_pgts[EARLY_DYNAMIC_PAGE_TABLES][PTRS_PER_PMD];
 extern unsigned int next_early_pgt;
@@ -385,3 +386,29 @@ void __head __relocate_kernel(unsigned long physbase, unsigned long virtbase)
 	}
 }
 #endif
+
+#ifdef CONFIG_PVM_GUEST
+extern unsigned long pvm_range_start;
+extern unsigned long pvm_range_end;
+
+static void __head detect_pvm_range(void)
+{
+	unsigned long msr_val;
+	unsigned long pml4_index_start, pml4_index_end;
+
+	msr_val = __rdmsr(MSR_PVM_LINEAR_ADDRESS_RANGE);
+	pml4_index_start = msr_val & 0x1ff;
+	pml4_index_end = (msr_val >> 16) & 0x1ff;
+	pvm_range_start = (0x1fffe00 | pml4_index_start) * P4D_SIZE;
+	pvm_range_end = (0x1fffe00 | pml4_index_end) * P4D_SIZE;
+}
+
+void __head pvm_relocate_kernel(unsigned long physbase)
+{
+	if (!pvm_detect())
+		return;
+
+	detect_pvm_range();
+	__relocate_kernel(physbase, pvm_range_end - (2UL << 30));
+}
+#endif
diff --git a/arch/x86/kernel/head_64.S b/arch/x86/kernel/head_64.S
index b8278f05bbd0..1d931bab4393 100644
--- a/arch/x86/kernel/head_64.S
+++ b/arch/x86/kernel/head_64.S
@@ -91,6 +91,19 @@ SYM_CODE_START_NOALIGN(startup_64)
 	movq	%rdx, PER_CPU_VAR(this_cpu_off)
 #endif
 
+#ifdef CONFIG_PVM_GUEST
+	leaq	_text(%rip), %rdi
+	call	pvm_relocate_kernel
+#ifdef CONFIG_SMP
+	/* Fill __per_cpu_offset[0] again, because it got relocated. */
+	movabs	$__per_cpu_load, %rdx
+	movabs	$__per_cpu_start, %rax
+	subq	%rax, %rdx
+	movq	%rdx, __per_cpu_offset(%rip)
+	movq	%rdx, PER_CPU_VAR(this_cpu_off)
+#endif
+#endif
+
 	call	startup_64_setup_env
 
 	/* Now switch to __KERNEL_CS so IRET works reliably */
diff --git a/arch/x86/kernel/pvm.c b/arch/x86/kernel/pvm.c
index 2d27044eaf25..fc82c71b305b 100644
--- a/arch/x86/kernel/pvm.c
+++ b/arch/x86/kernel/pvm.c
@@ -13,9 +13,12 @@
 #include <asm/cpufeature.h>
 #include <asm/pvm_para.h>
 
+unsigned long pvm_range_start __initdata;
+unsigned long pvm_range_end __initdata;
+
 void __init pvm_early_setup(void)
 {
-	if (!pvm_detect())
+	if (!pvm_range_end)
 		return;
 
 	setup_force_cpu_cap(X86_FEATURE_KVM_PVM_GUEST);
-- 
2.19.1.6.gb485710b