blktests nvme 041,042 leak memory

Maurizio Lombardi mlombard at redhat.com
Fri May 31 04:55:03 PDT 2024


Maybe a better idea:

move device_initialize() to the very beginning of nvme_init_ctrl()
initialize ctrl->instance to a negative value, so we can check in the
.release() method if we have to call ida_free() or not.

Now all the fabrics drivers know that in case of error they have to
execute nvme_put_ctrl().
Could it work?

Maurizio




More information about the Linux-nvme mailing list