Hi, The Visual LVM is a GUI LVM manager, it need to access system files and block devices for enumerate storage devices and LVM information.
So I request the interface as reject information.
Thanks
Hi, The Visual LVM is a GUI LVM manager, it need to access system files and block devices for enumerate storage devices and LVM information.
So I request the interface as reject information.
Thanks
Hey @weLees,
Which specific system files and/or directories this snap needs? I see the latest revision has:
plugs:
enum-devices:
interface: system-files
read:
- /sys/devices
- /dev
- /proc
- /proc/partitions
etc-lvm-backup:
interface: system-files
read:
- /etc/lvm/backup
- /etc
- /etc/visual_lvm
- /dev
write:
- /etc/lvm/backup
- /etc
- /etc/visual_lvm
- /dev/mapper/control
- /proc/sys/vm/drop_caches
- /run/lock/lvm
- /dev
null-device:
interface: system-files
write:
- /dev/nul
But some of those do not look good to me. For example, most of /proc
can be achieved by plugging system-observe iface (https://github.com/snapcore/snapd/blob/61b66aaa1beb861f308ea7dd75c2f0c042bf29a3/interfaces/builtin/system_observe.go#L60) and hardware-observe should be enough for files in /sys
pertaining to hardware https://github.com/snapcore/snapd/blob/master/interfaces/builtin/hardware_observe.go#L48
Then you are requesting write
access to the entire /etc
and /dev
directories. But also listing specific files inside those as well. Do you really need such wide accesses? Also, write
implies read
so just keeping write
is enough.
I am +1 for block-devices
(as I read the previous request which contains further explanation)
@weLees ping - did you have the chance to try the suggestions above? In general the kinds of paths which you are requesting like /dev, /proc etc are not appropriate for system-files
so instead you should try and find existing interfaces that provide support for these, and as suggested by @emitorino, snappy-debug
should be able to help identify these.
Yes we are trying to access via other path, it’s difficult.
We have tried best to remove interfaces, and the new submitted version is the result.
It is the least interfaces version.
I believe the discussion has now been moved to this other topic. I am updating our review queue to refer to such topic.