get_viotape_info() declares a vio_waitevent on the stack, which
contains a completion, but never initialises the completion.
I have no idea how this ever worked, and on recent kernels it causes
an oops in handle_tape_event() when we access the non-initialised
completion.
Signed-off-by: Michael Ellerman <mich...@ellerman.id.au>
---
arch/powerpc/platforms/iseries/vio.c | 2 ++
1 files changed, 2 insertions(+), 0 deletions(-)
diff --git a/arch/powerpc/platforms/iseries/vio.c
b/arch/powerpc/platforms/iseries/vio.c
index 657b72f..2aa8b56 100644
--- a/arch/powerpc/platforms/iseries/vio.c
+++ b/arch/powerpc/platforms/iseries/vio.c
@@ -474,6 +474,8 @@ static void __init get_viotape_info(struct device_node
*vio_root)
struct vio_waitevent we;
int ret;
+ init_completion(&we.com);
+
ret = viopath_open(viopath_hostLp, viomajorsubtype_tape, 2);
if (ret) {
printk(KERN_WARNING "get_viotape_info: "
--
1.6.3.3
_______________________________________________
Linuxppc-dev mailing list
Linuxppc-dev@lists.ozlabs.org
https://lists.ozlabs.org/listinfo/linuxppc-dev
