Re: [PATCH] block: Mark virtio_blk with __refdata in virtio_blk.c

[Rakib Mullick <rakib.mullick@gmail.com>]

On 7/17/09, Tejun Heo <tj@kernel.org> wrote:
> Hello,
>
> Generally looks good to me.
>
>
>  > +/* We were warned by false positive warning, so __refdata comes into rescue. */
>  > +static struct virtio_driver __refdata virtio_blk = {
>
>
> But it would be nicer if the comment contains a bit more of detail.
>  Can you please beef it up a bit?  After that I'll queue it in my temp
>  tree and send to Linus' way.
>
 Ok. Assuming that changelog is ok. I'm just sending the patch here
with a beefed up comment. Hopefully everyone will like it. If not or
if requires resend please notice.

Thanks,

----
Signed-off-by: Rakib Mullick <rakib.mullick@gmail.com>

--- linus/drivers/block/virtio_blk.c	2009-07-17 21:07:40.000000000 +0600
+++ rakib/drivers/block/virtio_blk.c	2009-07-17 21:14:36.000000000 +0600
@@ -424,7 +424,13 @@ static unsigned int features[] = {
 	VIRTIO_BLK_F_SCSI, VIRTIO_BLK_F_IDENTIFY
 };

-static struct virtio_driver virtio_blk = {
+/*
+ * virtio_blk messes up by simultaneously refering to a __devinit and
+ * a __devexit function. Which causes a false positive warning. So we
+ * use __refdata to avoid this warning.
+ */
+
+static struct virtio_driver __refdata virtio_blk = {
 	.feature_table = features,
 	.feature_table_size = ARRAY_SIZE(features),
 	.driver.name =	KBUILD_MODNAME,