[PATCH] SCSI: Add SG_GET_SG_MAX_SEGMENT_SIZE ioctl to sg device

[PATCH] SCSI: Add SG_GET_SG_MAX_SEGMENT_SIZE ioctl to sg device

[Cong Meng]

Add a new ioctl cmd to get the max segment size. It's useful in
virtualization environment for guest to know the queue limits, so
that the guest kicks off legal scsi command to a sg device.

Signed-off-by: Cong Meng <mc@linux.vnet.ibm.com>
---
 drivers/scsi/sg.c |    2 ++
 include/scsi/sg.h |    3 +++
 2 files changed, 5 insertions(+), 0 deletions(-)

diff --git a/drivers/scsi/sg.c b/drivers/scsi/sg.c
index eacd46b..bdcdc9d 100644
--- a/drivers/scsi/sg.c
+++ b/drivers/scsi/sg.c
@@ -913,6 +913,8 @@ sg_ioctl(struct file *filp, unsigned int cmd_in, unsigned long arg)
 		return put_user(val, ip);
 	case SG_GET_SG_TABLESIZE:
 		return put_user(sdp->sg_tablesize, ip);
+	case SG_GET_SG_MAX_SEGMENT_SIZE:
+		return put_user(queue_max_segment_size(sdp->device->request_queue), ip);
 	case SG_SET_RESERVED_SIZE:
 		result = get_user(val, ip);
 		if (result)
diff --git a/include/scsi/sg.h b/include/scsi/sg.h
index a9f3c6f..76d94c1 100644
--- a/include/scsi/sg.h
+++ b/include/scsi/sg.h
@@ -200,6 +200,9 @@ typedef struct sg_req_info { /* used by SG_GET_REQUEST_TABLE ioctl() */
 /* Yields max scatter gather tablesize allowed by current host adapter */
 #define SG_GET_SG_TABLESIZE 0x227F  /* 0 implies can't do scatter gather */
 
+/* Yidld max size of scatter gather element by current host adapter */
+#define SG_GET_SG_MAX_SEGMENT_SIZE 0x2280
+
 #define SG_GET_VERSION_NUM 0x2282 /* Example: version 2.1.34 yields 20134 */
 
 /* Returns -EBUSY if occupied. 3rd argument pointer to int (see next) */
-- 
1.7.7.6

Re: [PATCH] SCSI: Add SG_GET_SG_MAX_SEGMENT_SIZE ioctl to sg device

[James Bottomley]

On Mon, 2012-07-02 at 17:20 +0800, Cong Meng wrote:
> Add a new ioctl cmd to get the max segment size. It's useful in
> virtualization environment for guest to know the queue limits, so
> that the guest kicks off legal scsi command to a sg device.

What's wrong with simply looking in the /sys for this? Plus, if you can
come up with a justifiable reason, this needs to be in block as well so
we can send it to all SCSI devices regardless of what ULD people use.

James

Re: [PATCH] SCSI: Add SG_GET_SG_MAX_SEGMENT_SIZE ioctl to sg device

[Cong Meng]

On Mon, 2012-07-02 at 12:18 +0100, James Bottomley wrote:
> On Mon, 2012-07-02 at 17:20 +0800, Cong Meng wrote:
> > Add a new ioctl cmd to get the max segment size. It's useful in
> > virtualization environment for guest to know the queue limits, so
> > that the guest kicks off legal scsi command to a sg device.
> 
> What's wrong with simply looking in the /sys for this? Plus, if you can

Adding this ioctl is a big convenience for the programs, qemu 
in my case, to which parameter that likes '/dev/sgX' is given. 
With this ioctl, those programs only need to issue an ioctl to 
get the queue limit. Otherwise, 'dev/sgX' is needed to convert 
to /sys path, then open and read it. 
In addition, the potential variation and exceptional situation 
of sys FS, EX mount path, are needed to consider as well.
Is my justification acceptable? 

> come up with a justifiable reason, this needs to be in block as well so
> we can send it to all SCSI devices regardless of what ULD people use.
I can do it.

Thanks.
Cong.
> 
> James
> 
> 

Re: [PATCH] SCSI: Add SG_GET_SG_MAX_SEGMENT_SIZE ioctl to sg device

[James Bottomley]

On Tue, 2012-07-03 at 15:53 +0800, Cong Meng wrote:
> On Mon, 2012-07-02 at 12:18 +0100, James Bottomley wrote:
> > On Mon, 2012-07-02 at 17:20 +0800, Cong Meng wrote:
> > > Add a new ioctl cmd to get the max segment size. It's useful in
> > > virtualization environment for guest to know the queue limits, so
> > > that the guest kicks off legal scsi command to a sg device.
> > 
> > What's wrong with simply looking in the /sys for this? Plus, if you can
> 
> Adding this ioctl is a big convenience for the programs, qemu 
> in my case, to which parameter that likes '/dev/sgX' is given. 
> With this ioctl, those programs only need to issue an ioctl to 
> get the queue limit.

I don't need to know why you need the information, I accept that it's
useful

>  Otherwise, 'dev/sgX' is needed to convert 
> to /sys path, then open and read it. 

Well, yes, what's the problem?

> In addition, the potential variation and exceptional situation 
> of sys FS, EX mount path, are needed to consider as well.
> Is my justification acceptable? 

No. I need to know why you can't get it out of sysfs.  We put a whole
lot of stuff into sysfs to avoid the need to add extra ioctls.  There
need to be a really good reason why /sys doesn't work.

James

> > come up with a justifiable reason, this needs to be in block as well so
> > we can send it to all SCSI devices regardless of what ULD people use.
> I can do it.
> 
> Thanks.
> Cong.
> > 
> > James
> > 
> > 
> 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html