1 files changed, 66 insertions, 71 deletions

diff --git a/arch/arm64/kernel/syscall.c b/arch/arm64/kernel/syscall.c
index 9a9d98a443fc..9a70d9746b66 100644
--- a/arch/arm64/kernel/syscall.c
+++ b/arch/arm64/kernel/syscall.c
@@ -5,10 +5,11 @@
 #include <linux/errno.h>
 #include <linux/nospec.h>
 #include <linux/ptrace.h>
+#include <linux/randomize_kstack.h>
 #include <linux/syscalls.h>
 
-#include <asm/daifflags.h>
 #include <asm/debug-monitors.h>
+#include <asm/exception.h>
 #include <asm/fpsimd.h>
 #include <asm/syscall.h>
 #include <asm/thread_info.h>
@@ -42,6 +43,8 @@ static void invoke_syscall(struct pt_regs *regs, unsigned int scno,
 {
 	long ret;
 
+	add_random_kstack_offset();
+
 	if (scno < sc_nr) {
 		syscall_fn_t syscall_fn;
 		syscall_fn = syscall_table[array_index_nospec(scno, sc_nr)];
@@ -50,7 +53,20 @@ static void invoke_syscall(struct pt_regs *regs, unsigned int scno,
 		ret = do_ni_syscall(regs, scno);
 	}
 
-	regs->regs[0] = ret;
+	syscall_set_return_value(current, regs, 0, ret);
+
+	/*
+	 * Ultimately, this value will get limited by KSTACK_OFFSET_MAX(),
+	 * but not enough for arm64 stack utilization comfort. To keep
+	 * reasonable stack head room, reduce the maximum offset to 9 bits.
+	 *
+	 * The actual entropy will be further reduced by the compiler when
+	 * applying stack alignment constraints: the AAPCS mandates a
+	 * 16-byte (i.e. 4-bit) aligned SP at function boundaries.
+	 *
+	 * The resulting 5 bits of entropy is seen in SP[8:4].
+	 */
+	choose_random_kstack_offset(get_random_u16() & 0x1FF);
 }
 
 static inline bool has_syscall_work(unsigned long flags)
@@ -58,54 +74,60 @@ static inline bool has_syscall_work(unsigned long flags)
 	return unlikely(flags & _TIF_SYSCALL_WORK);
 }
 
-int syscall_trace_enter(struct pt_regs *regs);
-void syscall_trace_exit(struct pt_regs *regs);
-
-#ifdef CONFIG_ARM64_ERRATUM_1463225
-DECLARE_PER_CPU(int, __in_cortex_a76_erratum_1463225_wa);
-
-static void cortex_a76_erratum_1463225_svc_handler(void)
-{
-	u32 reg, val;
-
-	if (!unlikely(test_thread_flag(TIF_SINGLESTEP)))
-		return;
-
-	if (!unlikely(this_cpu_has_cap(ARM64_WORKAROUND_1463225)))
-		return;
-
-	__this_cpu_write(__in_cortex_a76_erratum_1463225_wa, 1);
-	reg = read_sysreg(mdscr_el1);
-	val = reg | DBG_MDSCR_SS | DBG_MDSCR_KDE;
-	write_sysreg(val, mdscr_el1);
-	asm volatile("msr daifclr, #8");
-	isb();
-
-	/* We will have taken a single-step exception by this point */
-
-	write_sysreg(reg, mdscr_el1);
-	__this_cpu_write(__in_cortex_a76_erratum_1463225_wa, 0);
-}
-#else
-static void cortex_a76_erratum_1463225_svc_handler(void) { }
-#endif /* CONFIG_ARM64_ERRATUM_1463225 */
-
 static void el0_svc_common(struct pt_regs *regs, int scno, int sc_nr,
 			   const syscall_fn_t syscall_table[])
 {
-	unsigned long flags = current_thread_info()->flags;
+	unsigned long flags = read_thread_flags();
 
 	regs->orig_x0 = regs->regs[0];
 	regs->syscallno = scno;
 
-	cortex_a76_erratum_1463225_svc_handler();
-	local_daif_restore(DAIF_PROCCTX);
-	user_exit();
+	/*
+	 * BTI note:
+	 * The architecture does not guarantee that SPSR.BTYPE is zero
+	 * on taking an SVC, so we could return to userspace with a
+	 * non-zero BTYPE after the syscall.
+	 *
+	 * This shouldn't matter except when userspace is explicitly
+	 * doing something stupid, such as setting PROT_BTI on a page
+	 * that lacks conforming BTI/PACIxSP instructions, falling
+	 * through from one executable page to another with differing
+	 * PROT_BTI, or messing with BTYPE via ptrace: in such cases,
+	 * userspace should not be surprised if a SIGILL occurs on
+	 * syscall return.
+	 *
+	 * So, don't touch regs->pstate & PSR_BTYPE_MASK here.
+	 * (Similarly for HVC and SMC elsewhere.)
+	 */
+
+	if (flags & _TIF_MTE_ASYNC_FAULT) {
+		/*
+		 * Process the asynchronous tag check fault before the actual
+		 * syscall. do_notify_resume() will send a signal to userspace
+		 * before the syscall is restarted.
+		 */
+		syscall_set_return_value(current, regs, -ERESTARTNOINTR, 0);
+		return;
+	}
 
 	if (has_syscall_work(flags)) {
-		/* set default errno for user-issued syscall(-1) */
+		/*
+		 * The de-facto standard way to skip a system call using ptrace
+		 * is to set the system call to -1 (NO_SYSCALL) and set x0 to a
+		 * suitable error code for consumption by userspace. However,
+		 * this cannot be distinguished from a user-issued syscall(-1)
+		 * and so we must set x0 to -ENOSYS here in case the tracer doesn't
+		 * issue the skip and we fall into trace_exit with x0 preserved.
+		 *
+		 * This is slightly odd because it also means that if a tracer
+		 * sets the system call number to -1 but does not initialise x0,
+		 * then x0 will be preserved for all system calls apart from a
+		 * user-issued syscall(-1). However, requesting a skip and not
+		 * setting the return value is unlikely to do anything sensible
+		 * anyway.
+		 */
 		if (scno == NO_SYSCALL)
-			regs->regs[0] = -ENOSYS;
+			syscall_set_return_value(current, regs, -ENOSYS, 0);
 		scno = syscall_trace_enter(regs);
 		if (scno == NO_SYSCALL)
 			goto trace_exit;
@@ -119,49 +141,22 @@ static void el0_svc_common(struct pt_regs *regs, int scno, int sc_nr,
 	 * exit regardless, as the old entry assembly did.
 	 */
 	if (!has_syscall_work(flags) && !IS_ENABLED(CONFIG_DEBUG_RSEQ)) {
-		local_daif_mask();
-		flags = current_thread_info()->flags;
-		if (!has_syscall_work(flags)) {
-			/*
-			 * We're off to userspace, where interrupts are
-			 * always enabled after we restore the flags from
-			 * the SPSR.
-			 */
-			trace_hardirqs_on();
+		flags = read_thread_flags();
+		if (!has_syscall_work(flags) && !(flags & _TIF_SINGLESTEP))
 			return;
-		}
-		local_daif_restore(DAIF_PROCCTX);
 	}
 
 trace_exit:
 	syscall_trace_exit(regs);
 }
 
-static inline void sve_user_discard(void)
-{
-	if (!system_supports_sve())
-		return;
-
-	clear_thread_flag(TIF_SVE);
-
-	/*
-	 * task_fpsimd_load() won't be called to update CPACR_EL1 in
-	 * ret_to_user unless TIF_FOREIGN_FPSTATE is still set, which only
-	 * happens if a context switch or kernel_neon_begin() or context
-	 * modification (sigreturn, ptrace) intervenes.
-	 * So, ensure that CPACR_EL1 is already correct for the fast-path case.
-	 */
-	sve_user_disable();
-}
-
-void el0_svc_handler(struct pt_regs *regs)
+void do_el0_svc(struct pt_regs *regs)
 {
-	sve_user_discard();
 	el0_svc_common(regs, regs->regs[8], __NR_syscalls, sys_call_table);
 }
 
 #ifdef CONFIG_COMPAT
-void el0_svc_compat_handler(struct pt_regs *regs)
+void do_el0_svc_compat(struct pt_regs *regs)
 {
 	el0_svc_common(regs, regs->regs[7], __NR_compat_syscalls,
 		       compat_sys_call_table);