| 1 |
On Wed, 24 Sep 2014 11:25:34 +0000 (UTC) |
| 2 |
Duncan <1i5t5.duncan@×××.net> wrote: |
| 3 |
|
| 4 |
> |
| 5 |
> Have you tried the kernel's own devtmpfs? How well does it work compared |
| 6 |
> to a static dev, etc? |
| 7 |
> |
| 8 |
|
| 9 |
No, I have not tried devtmpfs. |
| 10 |
|
| 11 |
The reason for devtmpfs is to allow faster boots by not having udev |
| 12 |
need to parse the sysfs hierarchy to discover devices. Thus devtmpfs, |
| 13 |
although it can be used independently of udev, is really intended to |
| 14 |
assist udev. |
| 15 |
|
| 16 |
A static /dev tree is good enough for me at this point. I always build |
| 17 |
my own machines (even laptops) and I know exactly what hardware I have |
| 18 |
and what device nodes to create. For plug-in or USB hardware, I can parse |
| 19 |
sysfs with my own code as easily as udev can. |
| 20 |
|
| 21 |
This method may seem strange and even regressive and stubborn to many |
| 22 |
Linux users. All I can say in response is that one has to be a little |
| 23 |
bit fanatical to even use Linux, and I am probably more than just a little |
| 24 |
bit fanatical. |
| 25 |
|
| 26 |
But I certainly appreciate the suggestion and I will keep it mind |
| 27 |
when eventually I am forced, kicking and screaming, to accept udev, |
| 28 |
systemd, etc., etc., etc. |
Archives