xtensa: handle coprocessor exceptions in kernel mode

In order to let drivers use xtensa coprocessors on behalf of the calling
process the kernel must handle coprocessor exceptions from the kernel
mode the same way as from the user mode.

This is not sufficient to allow using coprocessors transparently in IRQ
or softirq context. Should such users exist they must be aware of the
context and do the right thing, e.g. preserve the coprocessor state and
resore it after use.

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

1 files changed, 1 insertions, 1 deletions

diff --git a/arch/xtensa/kernel/traps.c b/arch/xtensa/kernel/traps.c
index 95903f25e523..62c497605128 100644
--- a/arch/xtensa/kernel/traps.c
+++ b/arch/xtensa/kernel/traps.c
@@ -69,7 +69,7 @@ static void do_debug(struct pt_regs *regs);
 #define USER		0x02
 
 #define COPROCESSOR(x)							\
-{ EXCCAUSE_COPROCESSOR ## x ## _DISABLED, USER, fast_coprocessor }
+{ EXCCAUSE_COPROCESSOR ## x ## _DISABLED, USER|KRNL, fast_coprocessor }
 
 typedef struct {
 	int cause;