arm64: Support execute-only permissions with Enhanced PAN

Enhanced Privileged Access Never (EPAN) allows Privileged Access Never
to be used with Execute-only mappings.

Absence of such support was a reason for 24cecc3774 ("arm64: Revert
support for execute-only user mappings"). Thus now it can be revisited
and re-enabled.

Cc: Kees Cook <keescook@chromium.org>
Signed-off-by: Vladimir Murzin <vladimir.murzin@arm.com>
Acked-by: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20210312173811.58284-2-vladimir.murzin@arm.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>

arch/arm64/Kconfig

@@ -1058,6 +1058,9 @@ config SYS_SUPPORTS_HUGETLBFS
 config ARCH_HAS_CACHE_LINE_SIZE
 	def_bool y
 
+config ARCH_HAS_FILTER_PGPROT
+	def_bool y
+
 config ARCH_ENABLE_SPLIT_PMD_PTLOCK
 	def_bool y if PGTABLE_LEVELS > 2
 
@@ -1681,6 +1684,20 @@ config ARM64_MTE
 
 endmenu
 
+menu "ARMv8.7 architectural features"
+
+config ARM64_EPAN
+	bool "Enable support for Enhanced Privileged Access Never (EPAN)"
+	default y
+	depends on ARM64_PAN
+	help
+	 Enhanced Privileged Access Never (EPAN) allows Privileged
+	 Access Never to be used with Execute-only mappings.
+
+	 The feature is detected at runtime, and will remain disabled
+	 if the cpu does not implement the feature.
+endmenu
+
 config ARM64_SVE
 	bool "ARM Scalable Vector Extension support"
 	default y

arch/arm64/include/asm/cpucaps.h

@@ -66,7 +66,8 @@
 #define ARM64_WORKAROUND_1508412		58
 #define ARM64_HAS_LDAPR				59
 #define ARM64_KVM_PROTECTED_MODE		60
+#define ARM64_HAS_EPAN				61
 
-#define ARM64_NCAPS				61
+#define ARM64_NCAPS				62
 
 #endif /* __ASM_CPUCAPS_H */

arch/arm64/include/asm/pgtable-prot.h

@@ -87,12 +87,13 @@ extern bool arm64_use_ng_mappings;
 #define PAGE_SHARED_EXEC	__pgprot(_PAGE_DEFAULT | PTE_USER | PTE_RDONLY | PTE_NG | PTE_PXN | PTE_WRITE)
 #define PAGE_READONLY		__pgprot(_PAGE_DEFAULT | PTE_USER | PTE_RDONLY | PTE_NG | PTE_PXN | PTE_UXN)
 #define PAGE_READONLY_EXEC	__pgprot(_PAGE_DEFAULT | PTE_USER | PTE_RDONLY | PTE_NG | PTE_PXN)
+#define PAGE_EXECONLY		__pgprot(_PAGE_DEFAULT | PTE_RDONLY | PTE_NG | PTE_PXN)
 
 #define __P000  PAGE_NONE
 #define __P001  PAGE_READONLY
 #define __P010  PAGE_READONLY
 #define __P011  PAGE_READONLY
-#define __P100  PAGE_READONLY_EXEC
+#define __P100  PAGE_EXECONLY
 #define __P101  PAGE_READONLY_EXEC
 #define __P110  PAGE_READONLY_EXEC
 #define __P111  PAGE_READONLY_EXEC
@@ -101,7 +102,7 @@ extern bool arm64_use_ng_mappings;
 #define __S001  PAGE_READONLY
 #define __S010  PAGE_SHARED
 #define __S011  PAGE_SHARED
-#define __S100  PAGE_READONLY_EXEC
+#define __S100  PAGE_EXECONLY
 #define __S101  PAGE_READONLY_EXEC
 #define __S110  PAGE_SHARED_EXEC
 #define __S111  PAGE_SHARED_EXEC

arch/arm64/include/asm/pgtable.h

@@ -113,11 +113,12 @@ extern unsigned long empty_zero_page[PAGE_SIZE / sizeof(unsigned long)];
 #define pte_dirty(pte)		(pte_sw_dirty(pte) || pte_hw_dirty(pte))
 
 #define pte_valid(pte)		(!!(pte_val(pte) & PTE_VALID))
+/*
+ * Execute-only user mappings do not have the PTE_USER bit set. All valid
+ * kernel mappings have the PTE_UXN bit set.
+ */
 #define pte_valid_not_user(pte) \
-	((pte_val(pte) & (PTE_VALID | PTE_USER)) == PTE_VALID)
+	((pte_val(pte) & (PTE_VALID | PTE_USER | PTE_UXN)) == (PTE_VALID | PTE_UXN))
-#define pte_valid_user(pte) \
-	((pte_val(pte) & (PTE_VALID | PTE_USER)) == (PTE_VALID | PTE_USER))
-
 /*
  * Could the pte be present in the TLB? We must check mm_tlb_flush_pending
  * so that we don't erroneously return false for pages that have been
@@ -130,12 +131,14 @@ extern unsigned long empty_zero_page[PAGE_SIZE / sizeof(unsigned long)];
 	(mm_tlb_flush_pending(mm) ? pte_present(pte) : pte_valid(pte))
 
 /*
- * p??_access_permitted() is true for valid user mappings (subject to the
+ * p??_access_permitted() is true for valid user mappings (PTE_USER
- * write permission check). PROT_NONE mappings do not have the PTE_VALID bit
+ * bit set, subject to the write permission check). For execute-only
- * set.
+ * mappings, like PROT_EXEC with EPAN (both PTE_USER and PTE_UXN bits
+ * not set) must return false. PROT_NONE mappings do not have the
+ * PTE_VALID bit set.
  */
 #define pte_access_permitted(pte, write) \
-	(pte_valid_user(pte) && (!(write) || pte_write(pte)))
+	(((pte_val(pte) & (PTE_VALID | PTE_USER)) == (PTE_VALID | PTE_USER)) && (!(write) || pte_write(pte)))
 #define pmd_access_permitted(pmd, write) \
 	(pte_access_permitted(pmd_pte(pmd), (write)))
 #define pud_access_permitted(pud, write) \
@@ -995,6 +998,18 @@ static inline bool arch_wants_old_prefaulted_pte(void)
 }
 #define arch_wants_old_prefaulted_pte	arch_wants_old_prefaulted_pte
 
+static inline pgprot_t arch_filter_pgprot(pgprot_t prot)
+{
+	if (cpus_have_const_cap(ARM64_HAS_EPAN))
+		return prot;
+
+	if (pgprot_val(prot) != pgprot_val(PAGE_EXECONLY))
+		return prot;
+
+	return PAGE_READONLY_EXEC;
+}
+
+
 #endif /* !__ASSEMBLY__ */
 
 #endif /* __ASM_PGTABLE_H */

arch/arm64/include/asm/sysreg.h

@@ -597,6 +597,7 @@
 	(SCTLR_EL2_RES1 | ENDIAN_SET_EL2)
 
 /* SCTLR_EL1 specific flags. */
+#define SCTLR_EL1_EPAN		(BIT(57))
 #define SCTLR_EL1_ATA0		(BIT(42))
 
 #define SCTLR_EL1_TCF0_SHIFT	38
@@ -637,7 +638,7 @@
 	 SCTLR_EL1_SED  | SCTLR_ELx_I    | SCTLR_EL1_DZE  | SCTLR_EL1_UCT   | \
 	 SCTLR_EL1_NTWE | SCTLR_ELx_IESB | SCTLR_EL1_SPAN | SCTLR_ELx_ITFSB | \
 	 SCTLR_ELx_ATA  | SCTLR_EL1_ATA0 | ENDIAN_SET_EL1 | SCTLR_EL1_UCI   | \
-	 SCTLR_EL1_RES1)
+	 SCTLR_EL1_EPAN | SCTLR_EL1_RES1)
 
 /* MAIR_ELx memory attributes (used by Linux) */
 #define MAIR_ATTR_DEVICE_nGnRnE		UL(0x00)

arch/arm64/kernel/cpufeature.c

@@ -1821,6 +1821,18 @@ static const struct arm64_cpu_capabilities arm64_features[] = {
 		.cpu_enable = cpu_enable_pan,
 	},
 #endif /* CONFIG_ARM64_PAN */
+#ifdef CONFIG_ARM64_EPAN
+	{
+		.desc = "Enhanced Privileged Access Never",
+		.capability = ARM64_HAS_EPAN,
+		.type = ARM64_CPUCAP_SYSTEM_FEATURE,
+		.matches = has_cpuid_feature,
+		.sys_reg = SYS_ID_AA64MMFR1_EL1,
+		.field_pos = ID_AA64MMFR1_PAN_SHIFT,
+		.sign = FTR_UNSIGNED,
+		.min_field_value = 3,
+	},
+#endif /* CONFIG_ARM64_EPAN */
 #ifdef CONFIG_ARM64_LSE_ATOMICS
 	{
 		.desc = "LSE atomic instructions",

arch/arm64/mm/fault.c

@@ -527,7 +527,7 @@ static int __kprobes do_page_fault(unsigned long far, unsigned int esr,
 	const struct fault_info *inf;
 	struct mm_struct *mm = current->mm;
 	vm_fault_t fault;
-	unsigned long vm_flags = VM_ACCESS_FLAGS;
+	unsigned long vm_flags;
 	unsigned int mm_flags = FAULT_FLAG_DEFAULT;
 	unsigned long addr = untagged_addr(far);
 
@@ -544,12 +544,28 @@ static int __kprobes do_page_fault(unsigned long far, unsigned int esr,
 	if (user_mode(regs))
 		mm_flags |= FAULT_FLAG_USER;
 
+	/*
+	 * vm_flags tells us what bits we must have in vma->vm_flags
+	 * for the fault to be benign, __do_page_fault() would check
+	 * vma->vm_flags & vm_flags and returns an error if the
+	 * intersection is empty
+	 */
 	if (is_el0_instruction_abort(esr)) {
+		/* It was exec fault */
 		vm_flags = VM_EXEC;
 		mm_flags |= FAULT_FLAG_INSTRUCTION;
 	} else if (is_write_abort(esr)) {
+		/* It was write fault */
 		vm_flags = VM_WRITE;
 		mm_flags |= FAULT_FLAG_WRITE;
+	} else {
+		/* It was read fault */
+		vm_flags = VM_READ;
+		/* Write implies read */
+		vm_flags |= VM_WRITE;
+		/* If EPAN is absent then exec implies read */
+		if (!cpus_have_const_cap(ARM64_HAS_EPAN))
+			vm_flags |= VM_EXEC;
 	}
 
 	if (is_ttbr0_addr(addr) && is_el1_permission_fault(addr, esr, regs)) {

mm/mmap.c

@@ -93,6 +93,12 @@ static void unmap_region(struct mm_struct *mm,
  * MAP_PRIVATE	r: (no) no	r: (yes) yes	r: (no) yes	r: (no) yes
  *		w: (no) no	w: (no) no	w: (copy) copy	w: (no) no
  *		x: (no) no	x: (no) yes	x: (no) yes	x: (yes) yes
+ *
+ * On arm64, PROT_EXEC has the following behaviour for both MAP_SHARED and
+ * MAP_PRIVATE (with Enhanced PAN supported):
+ *								r: (no) no
+ *								w: (no) no
+ *								x: (yes) yes
  */
 pgprot_t protection_map[16] __ro_after_init = {
 	__P000, __P001, __P010, __P011, __P100, __P101, __P110, __P111,