Re: [bug report] regression bisected to "block: Make most scsi_req_init() calls implicit"

From: Bart Van Assche
Date: Mon Oct 23 2017 - 17:08:44 EST


On Mon, 2017-10-23 at 09:41 -0600, dann frazier wrote:
> (gdb) list *(sg_io+0x120)
> 0xffff0000084e71a8 is in sg_io (./include/linux/uaccess.h:113).
> 108 static inline unsigned long
> 109 _copy_from_user(void *to, const void __user *from, unsigned long n)
> 110 {
> 111 unsigned long res = n;
> 112 might_fault();
> 113 if (likely(access_ok(VERIFY_READ, from, n))) {
> 114 kasan_check_write(to, n);
> 115 res = raw_copy_from_user(to, from, n);
> 116 }
> 117 if (unlikely(res))

Hello Dann,

Would it be possible to check whether the patch below is sufficient to fix this?

Thanks,

Bart.


diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
index 34e17ee799be..15e11a519801 100644
--- a/drivers/block/virtio_blk.c
+++ b/drivers/block/virtio_blk.c
@@ -597,6 +597,7 @@ static const struct blk_mq_ops virtio_mq_ops = {
.queue_rq = virtio_queue_rq,
.complete = virtblk_request_done,
.init_request = virtblk_init_request,
+ .initialize_rq_fn = scsi_initialize_rq,
.map_queues = virtblk_map_queues,
};

diff --git a/drivers/scsi/sg.c b/drivers/scsi/sg.c
index 0419c2298eab..14096f67eebb 100644
--- a/drivers/scsi/sg.c
+++ b/drivers/scsi/sg.c
@@ -1722,6 +1722,8 @@ sg_start_req(Sg_request *srp, unsigned char *cmd)
}
req = scsi_req(rq);

+ WARN_ON_ONCE(!req->cmd);
+
if (hp->cmd_len > BLK_MAX_CDB)
req->cmd = long_cmdp;
memcpy(req->cmd, cmd, hp->cmd_len);