1 |
walt wrote:
|
2 |
> On 05/14/2015 10:56 PM, Raffaele BELARDI wrote: |
3 |
>> I have an amd64 system with an old 3.3.x kernel. Recently (I think after |
4 |
>> last udev update to 217) the boot process became very slow due to "udev |
5 |
>> waiting for uevents to populate /dev". After a minute or so udev prints |
6 |
>> something about a lazy device (a TV tuner) then the boot continues. |
7 |
>> |
8 |
>> Yesterday I tried to update the kernel to 4.0.2... |
9 |
> |
10 |
> That kernel version is marked ~amd64 (unstable/testing). Are you accepting |
11 |
> the ~amd64 keyword in your make.conf? If you ask for help in this group you |
12 |
> should include that info as part of your question. You'll get better answers |
13 |
> if you do. |
14 |
> |
15 |
> Just as important in the last year or so is whether you are using systemd or |
16 |
> openrc at boot time. |
17 |
> |
18 |
> Life never gets simpler over time :( |
19 |
> |
20 |
|
21 |
:-[ You're right, that was not a very good request from my side.
|
22 |
|
23 |
System is ~amd64, openrc, udev 219. Rebuilding the kernel with lots of
|
24 |
changes the boot completed at least once. Now it still hangs in the same
|
25 |
stage while waiting the NVIDIA video driver instead of the TV tuner
|
26 |
driver. But pressing CTRL-C it continues without loading the faulty
|
27 |
driver and at least I'm able to continue debugging.
|
28 |
|
29 |
So the udev hang is in reality an NVIDIA driver problem, so that's what
|
30 |
I'm working on now.
|
31 |
|
32 |
Anyway I do have the impression that udev keeps a state somewhere,
|
33 |
because more than once it happened that on first boot with new kernel it
|
34 |
worked fine, starting from the second boot it had the 'waiting for
|
35 |
uevents' problem. I tried to delete /etc/udev/hwdb.bin but it didn't change.
|
36 |
|
37 |
raffaele |