-----Original Message-----
From: Greg KH [mailto:greg@kroah.com]
Sent: Wednesday, February 23, 2011 4:27 PM
To: Haiyang Zhang
Cc: Hank Janssen; KY Srinivasan; Abhishek Kane (Mindtree Consulting PVT LTD);
gregkh@suse.de; linux-kernel@vger.kernel.org; devel@linuxdriverproject.org;
virtualization@lists.osdl.org
Subject: Re: [PATCH 1/4] staging: hv: Fix the code depending on struct
vmbus_driver_context data order

On Wed, Feb 23, 2011 at 12:19:55PM -0800, Haiyang Zhang wrote:

quoted

The patch fixed the code depending on the exact order of fields in the
struct vmbus_driver_context, so the unused field drv_ctx can be removed,
and drv_obj doesn't have to be the second field in this structure.

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Hank Janssen <redacted>

---
 drivers/staging/hv/blkvsc_drv.c  |    2 ++
 drivers/staging/hv/netvsc_drv.c  |    2 ++
 drivers/staging/hv/storvsc_drv.c |    2 ++
 drivers/staging/hv/vmbus.h       |    2 +-
 drivers/staging/hv/vmbus_drv.c   |   14 +-------------
 5 files changed, 8 insertions(+), 14 deletions(-)

diff --git a/drivers/staging/hv/blkvsc_drv.c b/drivers/staging/hv/blkvsc_drv.c
index 36a0adb..293ab8e 100644
--- a/drivers/staging/hv/blkvsc_drv.c
+++ b/drivers/staging/hv/blkvsc_drv.c

@@ -177,6 +177,8 @@ static int blkvsc_drv_init(int (*drv_init)(struct hv_driver

*drv))

quoted

 	struct driver_context *drv_ctx = &g_blkvsc_drv.drv_ctx;
 	int ret;

+	drv_ctx->hv_drv = &storvsc_drv_obj->base;
+
 	storvsc_drv_obj->ring_buffer_size = blkvsc_ringbuffer_size;

 	/* Callback to client driver to complete the initialization */

diff --git a/drivers/staging/hv/netvsc_drv.c b/drivers/staging/hv/netvsc_drv.c
index 03f9740..364b6c7 100644
--- a/drivers/staging/hv/netvsc_drv.c
+++ b/drivers/staging/hv/netvsc_drv.c

@@ -500,6 +500,8 @@ static int netvsc_drv_init(int (*drv_init)(struct hv_driver

*drv))

quoted

 	struct driver_context *drv_ctx = &g_netvsc_drv.drv_ctx;
 	int ret;

+	drv_ctx->hv_drv = &net_drv_obj->base;
+
 	net_drv_obj->ring_buf_size = ring_size * PAGE_SIZE;
 	net_drv_obj->recv_cb = netvsc_recv_callback;
 	net_drv_obj->link_status_change = netvsc_linkstatus_callback;

diff --git a/drivers/staging/hv/storvsc_drv.c b/drivers/staging/hv/storvsc_drv.c
index a8427ff..33acee5 100644
--- a/drivers/staging/hv/storvsc_drv.c
+++ b/drivers/staging/hv/storvsc_drv.c

@@ -140,6 +140,8 @@ static int storvsc_drv_init(int (*drv_init)(struct hv_driver

*drv))

quoted

 	struct storvsc_driver_object *storvsc_drv_obj = &g_storvsc_drv.drv_obj;
 	struct driver_context *drv_ctx = &g_storvsc_drv.drv_ctx;

+	drv_ctx->hv_drv = &storvsc_drv_obj->base;
+
 	storvsc_drv_obj->ring_buffer_size = storvsc_ringbuffer_size;

 	/* Callback to client driver to complete the initialization */

diff --git a/drivers/staging/hv/vmbus.h b/drivers/staging/hv/vmbus.h
index 42f2adb..fd9d00f 100644
--- a/drivers/staging/hv/vmbus.h
+++ b/drivers/staging/hv/vmbus.h

@@ -30,8 +30,8 @@

 struct driver_context {
 	struct hv_guid class_id;
-
 	struct device_driver driver;
+	struct hv_driver *hv_drv;

If you have a pointer to hv_driver, why do you need a full 'struct
device_driver' here?  That sounds really wrong.

Actually, having 'struct device_driver' within a structure called
"driver_context" seems wrong, this should be what 'struct hv_driver'
really is, right?

We could certainly use better names to describe the layering here: 
Think of driver_context as a representation of a generic hyperV device 
driver. So, this structure will embed the generic struct device_driver.
The pointer to hv_drv allows for further specialization based on 
the device type. For instance the block driver has its own hv_driver while
the network driver has its own instance of hv_driver. I am not defending
this layering, and I agree we could name these abstractions to be more 
intuitive and also considerably improve the layering.

 

Regards,

K. Y

So no, I can't take this patch, as it isn't solving the root problem
here.

thanks,

greg k-h