xtensa: support coprocessors on SMP

Current coprocessor support on xtensa only works correctly on
uniprocessor configurations. Make it work on SMP too and keep it lazy.

Make coprocessor_owner array per-CPU and move it to struct exc_table for
easy access from the fast_coprocessor exception handler. Allow task to
have live coprocessors only on single CPU, record this CPU number in the
struct thread_info::cp_owner_cpu. Change struct thread_info::cpenable
meaning to be 'coprocessors live on cp_owner_cpu'.
Introduce C-level coprocessor exception handler that flushes and
releases live coprocessors of the task taking 'coprocessor disabled'
exception and call it from the fast_coprocessor handler when the task
has live coprocessors on other CPU.
Make coprocessor_flush_all and coprocessor_release_all work correctly
when called from any CPU by sending IPI to the cp_owner_cpu. Add
function coprocessor_flush_release_all to do flush followed by release
atomically. Add function local_coprocessors_flush_release_all to flush
and release all coprocessors on the local CPU and use it to flush
coprocessor contexts from the CPU that goes offline.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>

1 files changed, 87 insertions, 25 deletions

diff --git a/arch/xtensa/kernel/process.c b/arch/xtensa/kernel/process.c
index e8bfbca5f001..7e38292dd07a 100644
--- a/arch/xtensa/kernel/process.c
+++ b/arch/xtensa/kernel/process.c
@@ -47,6 +47,7 @@
 #include <asm/asm-offsets.h>
 #include <asm/regs.h>
 #include <asm/hw_breakpoint.h>
+#include <asm/traps.h>
 
 extern void ret_from_fork(void);
 extern void ret_from_kernel_thread(void);
@@ -63,52 +64,114 @@ EXPORT_SYMBOL(__stack_chk_guard);
 
 #if XTENSA_HAVE_COPROCESSORS
 
-void coprocessor_release_all(struct thread_info *ti)
+void local_coprocessors_flush_release_all(void)
 {
-	unsigned long cpenable;
-	int i;
+	struct thread_info **coprocessor_owner;
+	struct thread_info *unique_owner[XCHAL_CP_MAX];
+	int n = 0;
+	int i, j;
 
-	/* Make sure we don't switch tasks during this operation. */
+	coprocessor_owner = this_cpu_ptr(&exc_table)->coprocessor_owner;
+	xtensa_set_sr(XCHAL_CP_MASK, cpenable);
 
-	preempt_disable();
+	for (i = 0; i < XCHAL_CP_MAX; i++) {
+		struct thread_info *ti = coprocessor_owner[i];
 
-	/* Walk through all cp owners and release it for the requested one. */
+		if (ti) {
+			coprocessor_flush(ti, i);
 
-	cpenable = ti->cpenable;
+			for (j = 0; j < n; j++)
+				if (unique_owner[j] == ti)
+					break;
+			if (j == n)
+				unique_owner[n++] = ti;
 
-	for (i = 0; i < XCHAL_CP_MAX; i++) {
-		if (coprocessor_owner[i] == ti) {
-			coprocessor_owner[i] = 0;
-			cpenable &= ~(1 << i);
+			coprocessor_owner[i] = NULL;
 		}
 	}
+	for (i = 0; i < n; i++) {
+		/* pairs with memw (1) in fast_coprocessor and memw in switch_to */
+		smp_wmb();
+		unique_owner[i]->cpenable = 0;
+	}
+	xtensa_set_sr(0, cpenable);
+}
 
-	ti->cpenable = cpenable;
+static void local_coprocessor_release_all(void *info)
+{
+	struct thread_info *ti = info;
+	struct thread_info **coprocessor_owner;
+	int i;
+
+	coprocessor_owner = this_cpu_ptr(&exc_table)->coprocessor_owner;
+
+	/* Walk through all cp owners and release it for the requested one. */
+
+	for (i = 0; i < XCHAL_CP_MAX; i++) {
+		if (coprocessor_owner[i] == ti)
+			coprocessor_owner[i] = NULL;
+	}
+	/* pairs with memw (1) in fast_coprocessor and memw in switch_to */
+	smp_wmb();
+	ti->cpenable = 0;
 	if (ti == current_thread_info())
 		xtensa_set_sr(0, cpenable);
+}
 
-	preempt_enable();
+void coprocessor_release_all(struct thread_info *ti)
+{
+	if (ti->cpenable) {
+		/* pairs with memw (2) in fast_coprocessor */
+		smp_rmb();
+		smp_call_function_single(ti->cp_owner_cpu,
+					 local_coprocessor_release_all,
+					 ti, true);
+	}
 }
 
-void coprocessor_flush_all(struct thread_info *ti)
+static void local_coprocessor_flush_all(void *info)
 {
-	unsigned long cpenable, old_cpenable;
+	struct thread_info *ti = info;
+	struct thread_info **coprocessor_owner;
+	unsigned long old_cpenable;
 	int i;
 
-	preempt_disable();
-
-	old_cpenable = xtensa_get_sr(cpenable);
-	cpenable = ti->cpenable;
-	xtensa_set_sr(cpenable, cpenable);
+	coprocessor_owner = this_cpu_ptr(&exc_table)->coprocessor_owner;
+	old_cpenable = xtensa_xsr(ti->cpenable, cpenable);
 
 	for (i = 0; i < XCHAL_CP_MAX; i++) {
-		if ((cpenable & 1) != 0 && coprocessor_owner[i] == ti)
+		if (coprocessor_owner[i] == ti)
 			coprocessor_flush(ti, i);
-		cpenable >>= 1;
 	}
 	xtensa_set_sr(old_cpenable, cpenable);
+}
 
-	preempt_enable();
+void coprocessor_flush_all(struct thread_info *ti)
+{
+	if (ti->cpenable) {
+		/* pairs with memw (2) in fast_coprocessor */
+		smp_rmb();
+		smp_call_function_single(ti->cp_owner_cpu,
+					 local_coprocessor_flush_all,
+					 ti, true);
+	}
+}
+
+static void local_coprocessor_flush_release_all(void *info)
+{
+	local_coprocessor_flush_all(info);
+	local_coprocessor_release_all(info);
+}
+
+void coprocessor_flush_release_all(struct thread_info *ti)
+{
+	if (ti->cpenable) {
+		/* pairs with memw (2) in fast_coprocessor */
+		smp_rmb();
+		smp_call_function_single(ti->cp_owner_cpu,
+					 local_coprocessor_flush_release_all,
+					 ti, true);
+	}
 }
 
 #endif
@@ -140,8 +203,7 @@ void flush_thread(void)
 {
 #if XTENSA_HAVE_COPROCESSORS
 	struct thread_info *ti = current_thread_info();
-	coprocessor_flush_all(ti);
-	coprocessor_release_all(ti);
+	coprocessor_flush_release_all(ti);
 #endif
 	flush_ptrace_hw_breakpoint(current);
 }