On Tue, Apr 24, 2012 at 7:13 PM, Stefan Hajnoczi <stefanha@xxxxxxxxx> wrote: > On Tue, Apr 24, 2012 at 8:05 AM, Paolo Bonzini <pbonzini@xxxxxxxxxx> wrote: >> Il 24/04/2012 06:21, ronnie sahlberg ha scritto: >>> Hi Stefan, >>> >>> A little bit off-topic but >>> >>> When you design the proper place and API to plug virt-scsi into an >>> external SCSI parser outside of qemu like the target in the kernel ... >>> >>> It would be very nice if one could also plug virt-scsi into libiscsi >>> and pass the CDBs straight to the remote iSCSI target too. >>> Keep some thoughts on virt-scsi + libiscsi integration. >> >> Yes, that makes a lot of sense. It's a bit harder than scsi-generic but >> we do want to get there. > > Yep. I think previously there was discussion about a libiscsi > SCSIDevice so that guest SCSI commands can be sent to libiscsi LUNs > without going through the QEMU block layer. (Another way to pass > arbitrary SCSI commands to libiscsi is by hooking up .bdrv_aio_ioctl() > with SG_IO scsi-generic compatible code in block/iscsi.c.) bdrv_aio_ioctl() and SG_IO would mean #ifdef __linux__ So maybe it would be better to instead create a new hw/scsi-scsi.c that calls straight into block/iscsi.c ? That would be a lot more work than emulating SG_IO but would work on all platforms. Comments? How important is !linux support ? regards ronnie sahlberg -- To unsubscribe from this list: send the line "unsubscribe target-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
