Hyper-V has a feature called Online VHDX Resize: the disk capacity can be
changed by the host when the VM is running. Note: the host doesn't notify
the VM of the capacity change proactively; when the VM sends the next I/O
request to the host, the host will report an error (SCSI_STATUS_CHECK_COND)
with sense data supplied (the sense data indicates a disk capacity change)
and the VM is supposed to check and handle the capacity change properly,
and retry the I/O request.

However, the current I/O response handling in the VM has a bug: the host's
response's status is lost unintentionally, so the upper SCSI layer in the VM
always thinks the I/O request is completed successfully while it's not (this
causes undefined behaviors), and ignores the sense data.

The patch fixes the issue and makes Online VHDX Resize work properly.