[PATCH 3.10 079/268] sg_write()/bsg_write() is not fit to be called under KERNEL_DS

[Willy Tarreau]

From: Al Viro <viro@xxxxxxxxxxxxxxxxxx>

commit 128394eff343fc6d2f32172f03e24829539c5835 upstream.

Both damn things interpret userland pointers embedded into the payload;
worse, they are actually traversing those.  Leaving aside the bad
API design, this is very much _not_ safe to call with KERNEL_DS.
Bail out early if that happens.

Signed-off-by: Al Viro <viro@xxxxxxxxxxxxxxxxxx>
Signed-off-by: Willy Tarreau <w@xxxxxx>
---
 block/bsg.c       | 3 +++
 drivers/scsi/sg.c | 3 +++
 2 files changed, 6 insertions(+)

diff --git a/block/bsg.c b/block/bsg.c
index 420a5a9..76801e5 100644
--- a/block/bsg.c
+++ b/block/bsg.c
@@ -675,6 +675,9 @@ bsg_write(struct file *file, const char __user *buf, size_t count, loff_t *ppos)
 
 	dprintk("%s: write %Zd bytes\n", bd->name, count);
 
+	if (unlikely(segment_eq(get_fs(), KERNEL_DS)))
+		return -EINVAL;
+
 	bsg_set_block(bd, file);
 
 	bytes_written = 0;
diff --git a/drivers/scsi/sg.c b/drivers/scsi/sg.c
index 1f65e32..291791a 100644
--- a/drivers/scsi/sg.c
+++ b/drivers/scsi/sg.c
@@ -568,6 +568,9 @@ sg_write(struct file *filp, const char __user *buf, size_t count, loff_t * ppos)
 	sg_io_hdr_t *hp;
 	unsigned char cmnd[MAX_COMMAND_SIZE];
 
+	if (unlikely(segment_eq(get_fs(), KERNEL_DS)))
+		return -EINVAL;
+
 	if ((!(sfp = (Sg_fd *) filp->private_data)) || (!(sdp = sfp->parentdp)))
 		return -ENXIO;
 	SCSI_LOG_TIMEOUT(3, printk("sg_write: %s, count=%d\n",
-- 
2.8.0.rc2.1.gbe9624a