kvm: arm64: Set up hyp percpu data for nVHE

Add hyp percpu section to linker script and rename the corresponding ELF
sections of hyp/nvhe object files. This moves all nVHE-specific percpu
variables to the new hyp percpu section.

Allocate sufficient amount of memory for all percpu hyp regions at global KVM
init time and create corresponding hyp mappings.

The base addresses of hyp percpu regions are kept in a dynamically allocated
array in the kernel.

Add NULL checks in PMU event-reset code as it may run before KVM memory is
initialized.
Signed-off-by: NDavid Brazdil <dbrazdil@google.com>
Signed-off-by: NMarc Zyngier <maz@kernel.org>
Acked-by: NWill Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20200922204910.7265-10-dbrazdil@google.com

arch/arm64/include/asm/kvm_asm.h

@@ -66,8 +66,19 @@
 #define CHOOSE_VHE_SYM(sym)	sym
 #define CHOOSE_NVHE_SYM(sym)	kvm_nvhe_sym(sym)
 
-#define this_cpu_ptr_nvhe_sym(sym)	this_cpu_ptr(&kvm_nvhe_sym(sym))
-#define per_cpu_ptr_nvhe_sym(sym, cpu)	per_cpu_ptr(&kvm_nvhe_sym(sym), cpu)
+/*
+ * Compute pointer to a symbol defined in nVHE percpu region.
+ * Returns NULL if percpu memory has not been allocated yet.
+ */
+#define this_cpu_ptr_nvhe_sym(sym)	per_cpu_ptr_nvhe_sym(sym, smp_processor_id())
+#define per_cpu_ptr_nvhe_sym(sym, cpu)						\
+	({									\
+		unsigned long base, off;					\
+		base = kvm_arm_hyp_percpu_base[cpu];				\
+		off = (unsigned long)&CHOOSE_NVHE_SYM(sym) -			\
+		      (unsigned long)&CHOOSE_NVHE_SYM(__per_cpu_start);		\
+		base ? (typeof(CHOOSE_NVHE_SYM(sym))*)(base + off) : NULL;	\
+	})
 
 #ifndef __KVM_NVHE_HYPERVISOR__
 /*
@@ -117,6 +128,10 @@ DECLARE_KVM_HYP_SYM(__kvm_hyp_vector);
 #define __kvm_hyp_init		CHOOSE_NVHE_SYM(__kvm_hyp_init)
 #define __kvm_hyp_vector	CHOOSE_HYP_SYM(__kvm_hyp_vector)
 
+extern unsigned long kvm_arm_hyp_percpu_base[NR_CPUS];
+DECLARE_KVM_NVHE_SYM(__per_cpu_start);
+DECLARE_KVM_NVHE_SYM(__per_cpu_end);
+
 #ifdef CONFIG_KVM_INDIRECT_VECTORS
 extern atomic_t arm64_el2_vector_last_slot;
 DECLARE_KVM_HYP_SYM(__bp_harden_hyp_vecs);

arch/arm64/kernel/vmlinux.lds.S

@@ -28,8 +28,15 @@ jiffies = jiffies_64;
 	__start___kvm_ex_table = .;				\
 	*(__kvm_ex_table)					\
 	__stop___kvm_ex_table = .;
+
+#define HYPERVISOR_PERCPU_SECTION				\
+	. = ALIGN(PAGE_SIZE);					\
+	HYP_SECTION_NAME(.data..percpu) : {			\
+		*(HYP_SECTION_NAME(.data..percpu))		\
+	}
 #else /* CONFIG_KVM */
 #define HYPERVISOR_EXTABLE
+#define HYPERVISOR_PERCPU_SECTION
 #endif
 
 #define HYPERVISOR_TEXT					\
@@ -195,6 +202,7 @@ SECTIONS
 	}
 
 	PERCPU_SECTION(L1_CACHE_BYTES)
+	HYPERVISOR_PERCPU_SECTION
 
 	.rela.dyn : ALIGN(8) {
 		*(.rela .rela*)

arch/arm64/kvm/arm.c

@@ -47,6 +47,7 @@ __asm__(".arch_extension	virt");
 #endif
 
 static DEFINE_PER_CPU(unsigned long, kvm_arm_hyp_stack_page);
+unsigned long kvm_arm_hyp_percpu_base[NR_CPUS];
 
 /* The VMID used in the VTTBR */
 static atomic64_t kvm_vmid_gen = ATOMIC64_INIT(1);
@@ -1255,6 +1256,19 @@ long kvm_arch_vm_ioctl(struct file *filp,
 	}
 }
 
+static unsigned long nvhe_percpu_size(void)
+{
+	return (unsigned long)CHOOSE_NVHE_SYM(__per_cpu_end) -
+		(unsigned long)CHOOSE_NVHE_SYM(__per_cpu_start);
+}
+
+static unsigned long nvhe_percpu_order(void)
+{
+	unsigned long size = nvhe_percpu_size();
+
+	return size ? get_order(size) : 0;
+}
+
 static void cpu_init_hyp_mode(void)
 {
 	phys_addr_t pgd_ptr;
@@ -1270,8 +1284,8 @@ static void cpu_init_hyp_mode(void)
 	 * kernel's mapping to the linear mapping, and store it in tpidr_el2
 	 * so that we can use adr_l to access per-cpu variables in EL2.
 	 */
-	tpidr_el2 = ((unsigned long)this_cpu_ptr(&kvm_host_data) -
-		     (unsigned long)kvm_ksym_ref(&kvm_host_data));
+	tpidr_el2 = (unsigned long)this_cpu_ptr_nvhe_sym(__per_cpu_start) -
+		    (unsigned long)kvm_ksym_ref(CHOOSE_NVHE_SYM(__per_cpu_start));
 
 	pgd_ptr = kvm_mmu_get_httbr();
 	hyp_stack_ptr = __this_cpu_read(kvm_arm_hyp_stack_page) + PAGE_SIZE;
@@ -1461,8 +1475,10 @@ static void teardown_hyp_mode(void)
 	int cpu;
 
 	free_hyp_pgds();
-	for_each_possible_cpu(cpu)
+	for_each_possible_cpu(cpu) {
 		free_page(per_cpu(kvm_arm_hyp_stack_page, cpu));
+		free_pages(kvm_arm_hyp_percpu_base[cpu], nvhe_percpu_order());
+	}
 }
 
 /**
@@ -1495,6 +1511,24 @@ static int init_hyp_mode(void)
 		per_cpu(kvm_arm_hyp_stack_page, cpu) = stack_page;
 	}
 
+	/*
+	 * Allocate and initialize pages for Hypervisor-mode percpu regions.
+	 */
+	for_each_possible_cpu(cpu) {
+		struct page *page;
+		void *page_addr;
+
+		page = alloc_pages(GFP_KERNEL, nvhe_percpu_order());
+		if (!page) {
+			err = -ENOMEM;
+			goto out_err;
+		}
+
+		page_addr = page_address(page);
+		memcpy(page_addr, CHOOSE_NVHE_SYM(__per_cpu_start), nvhe_percpu_size());
+		kvm_arm_hyp_percpu_base[cpu] = (unsigned long)page_addr;
+	}
+
 	/*
 	 * Map the Hyp-code called directly from the host
 	 */
@@ -1539,6 +1573,21 @@ static int init_hyp_mode(void)
 		}
 	}
 
+	/*
+	 * Map Hyp percpu pages
+	 */
+	for_each_possible_cpu(cpu) {
+		char *percpu_begin = (char *)kvm_arm_hyp_percpu_base[cpu];
+		char *percpu_end = percpu_begin + nvhe_percpu_size();
+
+		err = create_hyp_mappings(percpu_begin, percpu_end, PAGE_HYP);
+
+		if (err) {
+			kvm_err("Cannot map hyp percpu region\n");
+			goto out_err;
+		}
+	}
+
 	for_each_possible_cpu(cpu) {
 		kvm_host_data_t *cpu_data;
 

arch/arm64/kvm/hyp/nvhe/hyp.lds.S

@@ -7,7 +7,13 @@
  */
 
 #include <asm/hyp_image.h>
+#include <asm-generic/vmlinux.lds.h>
+#include <asm/cache.h>
+#include <asm/memory.h>
 
 SECTIONS {
 	HYP_SECTION(.text)
+	HYP_SECTION_NAME(.data..percpu) : {
+		PERCPU_INPUT(L1_CACHE_BYTES)
+	}
 }

arch/arm64/kvm/pmu.c

@@ -33,7 +33,7 @@ void kvm_set_pmu_events(u32 set, struct perf_event_attr *attr)
 {
 	struct kvm_host_data *ctx = this_cpu_ptr_hyp_sym(kvm_host_data);
 
-	if (!kvm_pmu_switch_needed(attr))
+	if (!ctx || !kvm_pmu_switch_needed(attr))
 		return;
 
 	if (!attr->exclude_host)
@@ -49,6 +49,9 @@ void kvm_clr_pmu_events(u32 clr)
 {
 	struct kvm_host_data *ctx = this_cpu_ptr_hyp_sym(kvm_host_data);
 
+	if (!ctx)
+		return;
+
 	ctx->pmu_events.events_host &= ~clr;
 	ctx->pmu_events.events_guest &= ~clr;
 }