Hi D,
I'm not sure what exactly your goal is.
Prevent the guest OS from realizing, that the HDD is of type
"virtio"? In that case I'd look at how the hard drive is specified to
qemu-system-x86_64. Or first of all, how the controller is specified.
At the level of command-line args to qemu-system-x86_64, I'd focus on
the options -machine. I haven't found an option to specify the IDE
controller model to emulate - therefore, I'd suggest trying
-machine pc
or
-machine q35
see also
-machine help
Since I am using RedHat family, I focus at q35 machine type (pc-i440fx is no longer supported) and virtio-blk/virtio-scsi device.
take windows 10 as an example, it will identify the disk as "ssd"or "hard disk" when installed on physical hardware.
It will identify qemu virtual disk as "thin provisioned disk"(with newer virio-blk/virtio-scsi) driver. (old virtio driver will make windows think the disk is a traditional hard disk).
Windows use NTFS and sometimes it needs defragment. the default weekly disk optimization task will do "trim+ sometimes defragment" to ssd. and do "defragment only" to hard disk.
I think "thin provisioned disk" is like "ssd" under windows.
Or, do you specifically want to prevent qemu from using the
discard/trim feature?
Why?
It is generally useful.
In bare metal SSD hardware, it saves the wear on your flash chips.
In a Qemu VM, it allows Qemu to save storage space in the filesystem
underlying your image file by using a so called "sparse" image file.
Which in turn can save space in backups.
trim is useful for ssd and sparse image file and I am using it. but for hard disk I don't want extra layers to slow it down. so I use thick-provisiond lvm device for vm disk backend.
trim is bad for the situation. in fact my vm may have both ssd backend and hard disk backend. I don't care much at linux. but they need different optimization method under windows. I can manually setup defrag job for it (although some windows versions didn't allow defrag for thin device) . it is best if I can set it correctly at hypervisor level so the upper layer os can handle it automatically.
Thanks a lot for your help!