new scsi-generic abstraction, use SG_IO (Christoph Hellwig)
authoraliguori <aliguori@c046a42c-6fe2-441c-8c8c-71466251a162>
Sat, 28 Mar 2009 17:28:41 +0000 (17:28 +0000)
committeraliguori <aliguori@c046a42c-6fe2-441c-8c8c-71466251a162>
Sat, 28 Mar 2009 17:28:41 +0000 (17:28 +0000)
commit221f715d90ec5fec569a19119887445c037bca86
tree01d6f162df77773cf09c6193c6dc84219f297370
parent64a7fde8e85fbadb0dadee6ed1c293cd86f5fb29
new scsi-generic abstraction, use SG_IO (Christoph Hellwig)

Okay, I started looking into how to handle scsi-generic I/O in the
new world order.

I think the best is to use the SG_IO ioctl instead of the read/write
interface as that allows us to support scsi passthrough on disk/cdrom
devices, too.  See Hannes patch on the kvm list from August for an
example.

Now that we always do ioctls we don't need another abstraction than
bdrv_ioctl for the synchronous requests for now, and for asynchronous
requests I've added a aio_ioctl abstraction keeping it simple.

Long-term we might want to move the ops to a higher-level abstraction
and let the low-level code fill out the request header, but I'm lazy
enough to leave that to the people trying to support scsi-passthrough
on a non-Linux OS.

Tested lightly by issuing various sg_ commands from sg3-utils in a guest
to a host CDROM device.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@6895 c046a42c-6fe2-441c-8c8c-71466251a162
block-raw-posix.c
block.c
block.h
block_int.h
hw/scsi-generic.c
posix-aio-compat.c
posix-aio-compat.h