On Tue, Mar 02, 2021 at 08:04:46AM +0100, Philippe Mathieu-Daudé wrote: > On 3/1/21 7:13 PM, Stefan Hajnoczi wrote: > > On Mon, Mar 01, 2021 at 12:53:29PM +0100, Philippe Mathieu-Daudé wrote: > > FWIW this API isn't perfect because the file could be reopened with QMP > > and the existing mmap would remain in place. > > Can you show me a QMP example or point me at the command?
x-blockdev-change and other commands can reopen or reconfigure the BlockDriverState graph - the mmap user would not be aware of this. For example, block_set_io_throttle won't take effect if the guest has the device mmapped. > This shouldn't happen with the pflash. It's not possible to say that because pflash has a DEFINE_PROP_DRIVE("drive") property. The storage is backed by a --drive/--blockdev and the user could send any QMP command that operates on drives :(. Users probably won't but there is nothing stopping them. The block layer has a permission system (BLK_PERM_*). Maybe it's possible to use it to lock a BDS while mmap is active? Stefan
signature.asc
Description: PGP signature