[Qemu-devel] [PATCH] Add iSCSI support for QEMU

[Ronnie Sahlberg <ronniesahlberg@gmail.com>]

List,

Please find a patch that adds built-in iSCSI support to QEMU when built and linked against the multiplatform iscsi initiator library at  git://github.com/sahlberg/libiscsi.git


I have implemented all suggested changed from Stefans review, except
* Fail the build if qemu blocksize is not multiple of 512.
Maybe QEMU will never use such, but if it does I dont think a blockdriver that can not support such should fail the build. Hence it just refuses to use the driver at runtime instead.

* I did not implement a synchronous bdrv_flush since I understand that Stefans patch will be merged shortly since it is required for other backend drivers anyway.


In some situations, using a built-in iscsi inititator has benefits against mounting the LUNs on the host.

* Privacy: The iSCSI LUNs are private to the guest and are not visible either to the host, nor to any processes running on the host.
* Ease of managment : If you have very many guests and very many, thousands of, iSCSI LUNs. It is inconvenient to have to expose all LUNs to the underlying host.
* No root requirement. Since the iSCSI LUNs are not mounted as devices on the host, ordinary users can set up and use iSCSI resources without the need for root privilege on the host to map the devices to local scsi devices.
* Since the network I/O for the iSCSI LUN is now done by the QEMU process instead of the kernel, each VM's iSCSI I/O traffic can be insolated & controlled via the cgroups network contorller / tc network classifier.
* Portable across OS. Each OS has different tools for setting up iSCSI usage. A native driver lets QEMU users setup iSCSI in the same way regardless of the level of OS support for iSCSI.

Please merge this patch to master or explain how I should change the patch so that it becomes acceptable for inclusion into QEMU.


regards
ronnie sahlberg