arm64: simplify sysreg manipulation

A while back we added {read,write}_sysreg accessors to handle accesses
to system registers, without the usual boilerplate asm volatile,
temporary variable, etc.

This patch makes use of these across arm64 to make code shorter and
clearer. For sequences with a trailing ISB, the existing isb() macro is
also used so that asm blocks can be removed entirely.

A few uses of inline assembly for msr/mrs are left as-is. Those
manipulating sp_el0 for the current thread_info value have special
clobber requiremends.
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>

arch/arm64/include/asm/hw_breakpoint.h

@@ -18,6 +18,7 @@
 
 #include <asm/cputype.h>
 #include <asm/cpufeature.h>
+#include <asm/sysreg.h>
 #include <asm/virt.h>
 
 #ifdef __KERNEL__
@@ -98,18 +99,18 @@ static inline void decode_ctrl_reg(u32 reg,
 #define AARCH64_DBG_REG_WCR	(AARCH64_DBG_REG_WVR + ARM_MAX_WRP)
 
 /* Debug register names. */
-#define AARCH64_DBG_REG_NAME_BVR	"bvr"
-#define AARCH64_DBG_REG_NAME_BCR	"bcr"
-#define AARCH64_DBG_REG_NAME_WVR	"wvr"
-#define AARCH64_DBG_REG_NAME_WCR	"wcr"
+#define AARCH64_DBG_REG_NAME_BVR	bvr
+#define AARCH64_DBG_REG_NAME_BCR	bcr
+#define AARCH64_DBG_REG_NAME_WVR	wvr
+#define AARCH64_DBG_REG_NAME_WCR	wcr
 
 /* Accessor macros for the debug registers. */
 #define AARCH64_DBG_READ(N, REG, VAL) do {\
-	asm volatile("mrs %0, dbg" REG #N "_el1" : "=r" (VAL));\
+	VAL = read_sysreg(dbg##REG##N##_el1);\
 } while (0)
 
 #define AARCH64_DBG_WRITE(N, REG, VAL) do {\
-	asm volatile("msr dbg" REG #N "_el1, %0" :: "r" (VAL));\
+	write_sysreg(VAL, dbg##REG##N##_el1);\
 } while (0)
 
 struct task_struct;

arch/arm64/include/asm/mmu_context.h

@@ -27,16 +27,14 @@
 #include <asm-generic/mm_hooks.h>
 #include <asm/cputype.h>
 #include <asm/pgtable.h>
+#include <asm/sysreg.h>
 #include <asm/tlbflush.h>
 
 #ifdef CONFIG_PID_IN_CONTEXTIDR
 static inline void contextidr_thread_switch(struct task_struct *next)
 {
-	asm(
-	"	msr	contextidr_el1, %0\n"
-	"	isb"
-	:
-	: "r" (task_pid_nr(next)));
+	write_sysreg(task_pid_nr(next), contextidr_el1);
+	isb();
 }
 #else
 static inline void contextidr_thread_switch(struct task_struct *next)
@@ -51,11 +49,8 @@ static inline void cpu_set_reserved_ttbr0(void)
 {
 	unsigned long ttbr = virt_to_phys(empty_zero_page);
 
-	asm(
-	"	msr	ttbr0_el1, %0			// set TTBR0\n"
-	"	isb"
-	:
-	: "r" (ttbr));
+	write_sysreg(ttbr, ttbr0_el1);
+	isb();
 }
 
 /*
@@ -81,13 +76,11 @@ static inline void __cpu_set_tcr_t0sz(unsigned long t0sz)
 	if (!__cpu_uses_extended_idmap())
 		return;
 
-	asm volatile (
-	"	mrs	%0, tcr_el1	;"
-	"	bfi	%0, %1, %2, %3	;"
-	"	msr	tcr_el1, %0	;"
-	"	isb"
-	: "=&r" (tcr)
-	: "r"(t0sz), "I"(TCR_T0SZ_OFFSET), "I"(TCR_TxSZ_WIDTH));
+	tcr = read_sysreg(tcr_el1);
+	tcr &= ~TCR_T0SZ_MASK;
+	tcr |= t0sz << TCR_T0SZ_OFFSET;
+	write_sysreg(tcr, tcr_el1);
+	isb();
 }
 
 #define cpu_set_default_tcr_t0sz()	__cpu_set_tcr_t0sz(TCR_T0SZ(VA_BITS))

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

@@ -208,6 +208,7 @@
 #define TCR_T1SZ(x)		((UL(64) - (x)) << TCR_T1SZ_OFFSET)
 #define TCR_TxSZ(x)		(TCR_T0SZ(x) | TCR_T1SZ(x))
 #define TCR_TxSZ_WIDTH		6
+#define TCR_T0SZ_MASK		(((UL(1) << TCR_TxSZ_WIDTH) - 1) << TCR_T0SZ_OFFSET)
 
 #define TCR_IRGN0_SHIFT		8
 #define TCR_IRGN0_MASK		(UL(3) << TCR_IRGN0_SHIFT)

arch/arm64/include/asm/sysreg.h

@@ -253,16 +253,6 @@ asm(
 "	.endm\n"
 );
 
-static inline void config_sctlr_el1(u32 clear, u32 set)
-{
-	u32 val;
-
-	asm volatile("mrs %0, sctlr_el1" : "=r" (val));
-	val &= ~clear;
-	val |= set;
-	asm volatile("msr sctlr_el1, %0" : : "r" (val));
-}
-
 /*
  * Unlike read_cpuid, calls to read_sysreg are never expected to be
  * optimized away or replaced with synthetic values.
@@ -283,6 +273,16 @@ static inline void config_sctlr_el1(u32 clear, u32 set)
 		     : : "rZ" (__val));				\
 } while (0)
 
+static inline void config_sctlr_el1(u32 clear, u32 set)
+{
+	u32 val;
+
+	val = read_sysreg(sctlr_el1);
+	val &= ~clear;
+	val |= set;
+	write_sysreg(val, sctlr_el1);
+}
+
 #endif
 
 #endif	/* __ASM_SYSREG_H */

arch/arm64/include/asm/thread_info.h

@@ -75,6 +75,9 @@ static inline struct thread_info *current_thread_info(void) __attribute_const__;
 
 /*
  * struct thread_info can be accessed directly via sp_el0.
+ *
+ * We don't use read_sysreg() as we want the compiler to cache the value where
+ * possible.
  */
 static inline struct thread_info *current_thread_info(void)
 {

arch/arm64/kernel/cacheinfo.c

@@ -39,7 +39,7 @@ static inline enum cache_type get_cache_type(int level)
 
 	if (level > MAX_CACHE_LEVEL)
 		return CACHE_TYPE_NOCACHE;
-	asm volatile ("mrs     %x0, clidr_el1" : "=r" (clidr));
+	clidr = read_sysreg(clidr_el1);
 	return CLIDR_CTYPE(clidr, level);
 }
 
@@ -55,11 +55,9 @@ u64 __attribute_const__ cache_get_ccsidr(u64 csselr)
 
 	WARN_ON(preemptible());
 
-	/* Put value into CSSELR */
-	asm volatile("msr csselr_el1, %x0" : : "r" (csselr));
+	write_sysreg(csselr, csselr_el1);
 	isb();
-	/* Read result out of CCSIDR */
-	asm volatile("mrs %x0, ccsidr_el1" : "=r" (ccsidr));
+	ccsidr = read_sysreg(ccsidr_el1);
 
 	return ccsidr;
 }

arch/arm64/kernel/debug-monitors.c

@@ -46,16 +46,14 @@ static void mdscr_write(u32 mdscr)
 {
 	unsigned long flags;
 	local_dbg_save(flags);
-	asm volatile("msr mdscr_el1, %0" :: "r" (mdscr));
+	write_sysreg(mdscr, mdscr_el1);
 	local_dbg_restore(flags);
 }
 NOKPROBE_SYMBOL(mdscr_write);
 
 static u32 mdscr_read(void)
 {
-	u32 mdscr;
-	asm volatile("mrs %0, mdscr_el1" : "=r" (mdscr));
-	return mdscr;
+	return read_sysreg(mdscr_el1);
 }
 NOKPROBE_SYMBOL(mdscr_read);
 
@@ -134,7 +132,7 @@ NOKPROBE_SYMBOL(disable_debug_monitors);
  */
 static int clear_os_lock(unsigned int cpu)
 {
-	asm volatile("msr oslar_el1, %0" : : "r" (0));
+	write_sysreg(0, oslar_el1);
 	isb();
 	return 0;
 }

arch/arm64/kernel/process.c

@@ -202,7 +202,7 @@ void show_regs(struct pt_regs * regs)
 
 static void tls_thread_flush(void)
 {
-	asm ("msr tpidr_el0, xzr");
+	write_sysreg(0, tpidr_el0);
 
 	if (is_compat_task()) {
 		current->thread.tp_value = 0;
@@ -213,7 +213,7 @@ static void tls_thread_flush(void)
 		 * with a stale shadow state during context switch.
 		 */
 		barrier();
-		asm ("msr tpidrro_el0, xzr");
+		write_sysreg(0, tpidrro_el0);
 	}
 }
 
@@ -253,7 +253,7 @@ int copy_thread(unsigned long clone_flags, unsigned long stack_start,
 		 * Read the current TLS pointer from tpidr_el0 as it may be
 		 * out-of-sync with the saved value.
 		 */
-		asm("mrs %0, tpidr_el0" : "=r" (*task_user_tls(p)));
+		*task_user_tls(p) = read_sysreg(tpidr_el0);
 
 		if (stack_start) {
 			if (is_compat_thread(task_thread_info(p)))
@@ -289,17 +289,15 @@ static void tls_thread_switch(struct task_struct *next)
 {
 	unsigned long tpidr, tpidrro;
 
-	asm("mrs %0, tpidr_el0" : "=r" (tpidr));
+	tpidr = read_sysreg(tpidr_el0);
 	*task_user_tls(current) = tpidr;
 
 	tpidr = *task_user_tls(next);
 	tpidrro = is_compat_thread(task_thread_info(next)) ?
 		  next->thread.tp_value : 0;
 
-	asm(
-	"	msr	tpidr_el0, %0\n"
-	"	msr	tpidrro_el0, %1"
-	: : "r" (tpidr), "r" (tpidrro));
+	write_sysreg(tpidr, tpidr_el0);
+	write_sysreg(tpidrro, tpidrro_el0);
 }
 
 /* Restore the UAO state depending on next's addr_limit */

arch/arm64/kernel/sys_compat.c

@@ -94,7 +94,7 @@ long compat_arm_syscall(struct pt_regs *regs)
 		 * See comment in tls_thread_flush.
 		 */
 		barrier();
-		asm ("msr tpidrro_el0, %0" : : "r" (regs->regs[0]));
+		write_sysreg(regs->regs[0], tpidrro_el0);
 		return 0;
 
 	default: