Merge remote-tracking branch 'arm64/for-next/scs' into for-next/core

* arm64/for-next/scs:
  arm64: sdei: Push IS_ENABLED() checks down to callee functions
  arm64: scs: use vmapped IRQ and SDEI shadow stacks
  scs: switch to vmapped shadow stacks

arch/arm64/kernel/Makefile

@@ -58,7 +58,6 @@ obj-$(CONFIG_CRASH_DUMP)		+= crash_dump.o
 obj-$(CONFIG_CRASH_CORE)		+= crash_core.o
 obj-$(CONFIG_ARM_SDE_INTERFACE)		+= sdei.o
 obj-$(CONFIG_ARM64_PTR_AUTH)		+= pointer_auth.o
-obj-$(CONFIG_SHADOW_CALL_STACK)		+= scs.o
 obj-$(CONFIG_ARM64_MTE)			+= mte.o
 
 obj-y					+= vdso/ probes/

arch/arm64/kernel/entry.S

@@ -429,7 +429,7 @@ SYM_CODE_END(__swpan_exit_el0)
 
 #ifdef CONFIG_SHADOW_CALL_STACK
 	/* also switch to the irq shadow stack */
-	adr_this_cpu scs_sp, irq_shadow_call_stack, x26
+	ldr_this_cpu scs_sp, irq_shadow_call_stack_ptr, x26
 #endif
 
 9998:
@@ -1086,9 +1086,9 @@ SYM_CODE_START(__sdei_asm_handler)
 #ifdef CONFIG_SHADOW_CALL_STACK
 	/* Use a separate shadow call stack for normal and critical events */
 	cbnz	w4, 3f
-	adr_this_cpu dst=scs_sp, sym=sdei_shadow_call_stack_normal, tmp=x6
+	ldr_this_cpu dst=scs_sp, sym=sdei_shadow_call_stack_normal_ptr, tmp=x6
 	b	4f
-3:	adr_this_cpu dst=scs_sp, sym=sdei_shadow_call_stack_critical, tmp=x6
+3:	ldr_this_cpu dst=scs_sp, sym=sdei_shadow_call_stack_critical_ptr, tmp=x6
 4:
 #endif
 

arch/arm64/kernel/irq.c

@@ -17,6 +17,7 @@
 #include <linux/init.h>
 #include <linux/irqchip.h>
 #include <linux/kprobes.h>
+#include <linux/scs.h>
 #include <linux/seq_file.h>
 #include <linux/vmalloc.h>
 #include <asm/daifflags.h>
@@ -27,6 +28,25 @@ DEFINE_PER_CPU(struct nmi_ctx, nmi_contexts);
 
 DEFINE_PER_CPU(unsigned long *, irq_stack_ptr);
 
+
+DECLARE_PER_CPU(unsigned long *, irq_shadow_call_stack_ptr);
+
+#ifdef CONFIG_SHADOW_CALL_STACK
+DEFINE_PER_CPU(unsigned long *, irq_shadow_call_stack_ptr);
+#endif
+
+static void init_irq_scs(void)
+{
+	int cpu;
+
+	if (!IS_ENABLED(CONFIG_SHADOW_CALL_STACK))
+		return;
+
+	for_each_possible_cpu(cpu)
+		per_cpu(irq_shadow_call_stack_ptr, cpu) =
+			scs_alloc(cpu_to_node(cpu));
+}
+
 #ifdef CONFIG_VMAP_STACK
 static void init_irq_stacks(void)
 {
@@ -54,6 +74,7 @@ static void init_irq_stacks(void)
 void __init init_IRQ(void)
 {
 	init_irq_stacks();
+	init_irq_scs();
 	irqchip_init();
 	if (!handle_arch_irq)
 		panic("No interrupt controller found.");

arch/arm64/kernel/scs.c

-// SPDX-License-Identifier: GPL-2.0
-/*
- * Shadow Call Stack support.
- *
- * Copyright (C) 2019 Google LLC
- */
-
-#include <linux/percpu.h>
-#include <linux/scs.h>
-
-DEFINE_SCS(irq_shadow_call_stack);
-
-#ifdef CONFIG_ARM_SDE_INTERFACE
-DEFINE_SCS(sdei_shadow_call_stack_normal);
-DEFINE_SCS(sdei_shadow_call_stack_critical);
-#endif

arch/arm64/kernel/sdei.c

@@ -7,6 +7,7 @@
 #include <linux/hardirq.h>
 #include <linux/irqflags.h>
 #include <linux/sched/task_stack.h>
+#include <linux/scs.h>
 #include <linux/uaccess.h>
 
 #include <asm/alternative.h>
@@ -37,6 +38,14 @@ DEFINE_PER_CPU(unsigned long *, sdei_stack_normal_ptr);
 DEFINE_PER_CPU(unsigned long *, sdei_stack_critical_ptr);
 #endif
 
+DECLARE_PER_CPU(unsigned long *, sdei_shadow_call_stack_normal_ptr);
+DECLARE_PER_CPU(unsigned long *, sdei_shadow_call_stack_critical_ptr);
+
+#ifdef CONFIG_SHADOW_CALL_STACK
+DEFINE_PER_CPU(unsigned long *, sdei_shadow_call_stack_normal_ptr);
+DEFINE_PER_CPU(unsigned long *, sdei_shadow_call_stack_critical_ptr);
+#endif
+
 static void _free_sdei_stack(unsigned long * __percpu *ptr, int cpu)
 {
 	unsigned long *p;
@@ -52,6 +61,9 @@ static void free_sdei_stacks(void)
 {
 	int cpu;
 
+	if (!IS_ENABLED(CONFIG_VMAP_STACK))
+		return;
+
 	for_each_possible_cpu(cpu) {
 		_free_sdei_stack(&sdei_stack_normal_ptr, cpu);
 		_free_sdei_stack(&sdei_stack_critical_ptr, cpu);
@@ -75,6 +87,9 @@ static int init_sdei_stacks(void)
 	int cpu;
 	int err = 0;
 
+	if (!IS_ENABLED(CONFIG_VMAP_STACK))
+		return 0;
+
 	for_each_possible_cpu(cpu) {
 		err = _init_sdei_stack(&sdei_stack_normal_ptr, cpu);
 		if (err)
@@ -90,6 +105,62 @@ static int init_sdei_stacks(void)
 	return err;
 }
 
+static void _free_sdei_scs(unsigned long * __percpu *ptr, int cpu)
+{
+	void *s;
+
+	s = per_cpu(*ptr, cpu);
+	if (s) {
+		per_cpu(*ptr, cpu) = NULL;
+		scs_free(s);
+	}
+}
+
+static void free_sdei_scs(void)
+{
+	int cpu;
+
+	for_each_possible_cpu(cpu) {
+		_free_sdei_scs(&sdei_shadow_call_stack_normal_ptr, cpu);
+		_free_sdei_scs(&sdei_shadow_call_stack_critical_ptr, cpu);
+	}
+}
+
+static int _init_sdei_scs(unsigned long * __percpu *ptr, int cpu)
+{
+	void *s;
+
+	s = scs_alloc(cpu_to_node(cpu));
+	if (!s)
+		return -ENOMEM;
+	per_cpu(*ptr, cpu) = s;
+
+	return 0;
+}
+
+static int init_sdei_scs(void)
+{
+	int cpu;
+	int err = 0;
+
+	if (!IS_ENABLED(CONFIG_SHADOW_CALL_STACK))
+		return 0;
+
+	for_each_possible_cpu(cpu) {
+		err = _init_sdei_scs(&sdei_shadow_call_stack_normal_ptr, cpu);
+		if (err)
+			break;
+		err = _init_sdei_scs(&sdei_shadow_call_stack_critical_ptr, cpu);
+		if (err)
+			break;
+	}
+
+	if (err)
+		free_sdei_scs();
+
+	return err;
+}
+
 static bool on_sdei_normal_stack(unsigned long sp, struct stack_info *info)
 {
 	unsigned long low = (unsigned long)raw_cpu_read(sdei_stack_normal_ptr);
@@ -130,13 +201,14 @@ unsigned long sdei_arch_get_entry_point(int conduit)
 	 */
 	if (is_hyp_mode_available() && !is_kernel_in_hyp_mode()) {
 		pr_err("Not supported on this hardware/boot configuration\n");
-		return 0;
+		goto out_err;
 	}
 
-	if (IS_ENABLED(CONFIG_VMAP_STACK)) {
-		if (init_sdei_stacks())
-			return 0;
-	}
+	if (init_sdei_stacks())
+		goto out_err;
+
+	if (init_sdei_scs())
+		goto out_err_free_stacks;
 
 	sdei_exit_mode = (conduit == SMCCC_CONDUIT_HVC) ? SDEI_EXIT_HVC : SDEI_EXIT_SMC;
 
@@ -151,6 +223,10 @@ unsigned long sdei_arch_get_entry_point(int conduit)
 #endif /* CONFIG_UNMAP_KERNEL_AT_EL0 */
 		return (unsigned long)__sdei_asm_handler;
 
+out_err_free_stacks:
+	free_sdei_stacks();
+out_err:
+	return 0;
 }
 
 /*

include/linux/scs.h

@@ -15,24 +15,18 @@
 
 #ifdef CONFIG_SHADOW_CALL_STACK
 
-/*
- * In testing, 1 KiB shadow stack size (i.e. 128 stack frames on a 64-bit
- * architecture) provided ~40% safety margin on stack usage while keeping
- * memory allocation overhead reasonable.
- */
-#define SCS_SIZE		SZ_1K
+#define SCS_ORDER		0
+#define SCS_SIZE		(PAGE_SIZE << SCS_ORDER)
 #define GFP_SCS			(GFP_KERNEL | __GFP_ZERO)
 
 /* An illegal pointer value to mark the end of the shadow stack. */
 #define SCS_END_MAGIC		(0x5f6UL + POISON_POINTER_DELTA)
 
-/* Allocate a static per-CPU shadow stack */
-#define DEFINE_SCS(name)						\
-	DEFINE_PER_CPU(unsigned long [SCS_SIZE/sizeof(long)], name)	\
-
 #define task_scs(tsk)		(task_thread_info(tsk)->scs_base)
 #define task_scs_sp(tsk)	(task_thread_info(tsk)->scs_sp)
 
+void *scs_alloc(int node);
+void scs_free(void *s);
 void scs_init(void);
 int scs_prepare(struct task_struct *tsk, int node);
 void scs_release(struct task_struct *tsk);
@@ -61,6 +55,8 @@ static inline bool task_scs_end_corrupted(struct task_struct *tsk)
 
 #else /* CONFIG_SHADOW_CALL_STACK */
 
+static inline void *scs_alloc(int node) { return NULL; }
+static inline void scs_free(void *s) {}
 static inline void scs_init(void) {}
 static inline void scs_task_reset(struct task_struct *tsk) {}
 static inline int scs_prepare(struct task_struct *tsk, int node) { return 0; }

kernel/scs.c

@@ -5,26 +5,49 @@
  * Copyright (C) 2019 Google LLC
  */
 
+#include <linux/cpuhotplug.h>
 #include <linux/kasan.h>
 #include <linux/mm.h>
 #include <linux/scs.h>
-#include <linux/slab.h>
+#include <linux/vmalloc.h>
 #include <linux/vmstat.h>
 
-static struct kmem_cache *scs_cache;
-
 static void __scs_account(void *s, int account)
 {
-	struct page *scs_page = virt_to_page(s);
+	struct page *scs_page = vmalloc_to_page(s);
 
 	mod_node_page_state(page_pgdat(scs_page), NR_KERNEL_SCS_KB,
 			    account * (SCS_SIZE / SZ_1K));
 }
 
-static void *scs_alloc(int node)
+/* Matches NR_CACHED_STACKS for VMAP_STACK */
+#define NR_CACHED_SCS 2
+static DEFINE_PER_CPU(void *, scs_cache[NR_CACHED_SCS]);
+
+static void *__scs_alloc(int node)
 {
-	void *s = kmem_cache_alloc_node(scs_cache, GFP_SCS, node);
+	int i;
+	void *s;
+
+	for (i = 0; i < NR_CACHED_SCS; i++) {
+		s = this_cpu_xchg(scs_cache[i], NULL);
+		if (s) {
+			kasan_unpoison_vmalloc(s, SCS_SIZE);
+			memset(s, 0, SCS_SIZE);
+			return s;
+		}
+	}
+
+	return __vmalloc_node_range(SCS_SIZE, 1, VMALLOC_START, VMALLOC_END,
+				    GFP_SCS, PAGE_KERNEL, 0, node,
+				    __builtin_return_address(0));
+}
 
+void *scs_alloc(int node)
+{
+	void *s;
+
+	s = __scs_alloc(node);
 	if (!s)
 		return NULL;
 
@@ -34,21 +57,47 @@ static void *scs_alloc(int node)
 	 * Poison the allocation to catch unintentional accesses to
 	 * the shadow stack when KASAN is enabled.
 	 */
-	kasan_poison_object_data(scs_cache, s);
+	kasan_poison_vmalloc(s, SCS_SIZE);
 	__scs_account(s, 1);
 	return s;
 }
 
-static void scs_free(void *s)
+void scs_free(void *s)
 {
+	int i;
+
 	__scs_account(s, -1);
-	kasan_unpoison_object_data(scs_cache, s);
-	kmem_cache_free(scs_cache, s);
+
+	/*
+	 * We cannot sleep as this can be called in interrupt context,
+	 * so use this_cpu_cmpxchg to update the cache, and vfree_atomic
+	 * to free the stack.
+	 */
+
+	for (i = 0; i < NR_CACHED_SCS; i++)
+		if (this_cpu_cmpxchg(scs_cache[i], 0, s) == NULL)
+			return;
+
+	vfree_atomic(s);
+}
+
+static int scs_cleanup(unsigned int cpu)
+{
+	int i;
+	void **cache = per_cpu_ptr(scs_cache, cpu);
+
+	for (i = 0; i < NR_CACHED_SCS; i++) {
+		vfree(cache[i]);
+		cache[i] = NULL;
+	}
+
+	return 0;
 }
 
 void __init scs_init(void)
 {
-	scs_cache = kmem_cache_create("scs_cache", SCS_SIZE, 0, 0, NULL);
+	cpuhp_setup_state(CPUHP_BP_PREPARE_DYN, "scs:scs_cache", NULL,
+			  scs_cleanup);
 }
 
 int scs_prepare(struct task_struct *tsk, int node)