If I do
ip tuntap add name tap%d mode tap
then a TAP device with a name like "tap0", "tap1", etc. will be created.
But there's no way for me to find out which name was chosen for the
device created by that command.
Perhaps ip should print the name of tuntap devices after they're
created?
I'd be interested in sending a patch, but I'd need some guidance on how
exactly it should work. Would there be any harm in always printing it?
Should it be behind a flag? Or just when the name contains a `%'?
On Tue, 6 Apr 2021 13:42:40 +0000
Alyssa Ross [off-list ref] wrote:
If I do
ip tuntap add name tap%d mode tap
then a TAP device with a name like "tap0", "tap1", etc. will be created.
But there's no way for me to find out which name was chosen for the
device created by that command.
Use a follow on ip link show or look in sysfs.
Perhaps ip should print the name of tuntap devices after they're
created?
You can already do that with followon command, or use batch to put two commands together.
I'd be interested in sending a patch, but I'd need some guidance on how
exactly it should work. Would there be any harm in always printing it?
Should it be behind a flag? Or just when the name contains a `%'?
Printing the result would make ip tuntap behave differently, which makes
it unique (snowflakes are bad) and would break users scripts.
On Tue, Apr 06, 2021 at 09:22:31AM -0700, Stephen Hemminger wrote:
On Tue, 6 Apr 2021 13:42:40 +0000
Alyssa Ross [off-list ref] wrote:
quoted
If I do
ip tuntap add name tap%d mode tap
then a TAP device with a name like "tap0", "tap1", etc. will be created.
But there's no way for me to find out which name was chosen for the
device created by that command.
Use a follow on ip link show or look in sysfs.
quoted
Perhaps ip should print the name of tuntap devices after they're
created?
You can already do that with followon command, or use batch to put two commands together.
I don't think that's reliable, is it? What happens if I create one
device, but by the time I do ip link show another one has been created?
There's no way to know for sure that the last device was created by the
command I ran.
As I understand it, the only way to avoid a race like that for sure
would be to look at the ifr_name returned by the kernel from TUNSETIFF.