You are not logged in.
Done
Very well, now may a solution present itself
Offline
It needs someone affected to locate the causal commit. I provided a very simplistic first step in post #13 that or
bisecting bugs with git to locate the causal commit. Then the commit can be reported upstream / reverted from the arch build.
Offline
I didn't have time yet to follow your instructions, but I will and I will report my progress here
Offline
Install the devtools package then:
...
Then install the packages with pacman -U.
So I finally had the time to walk through this (thanks @loqs) and confirmed that I'm able to boot without the delay having followed those steps.
[2019-11-23T12:08:03-0500] [PACMAN] Running 'pacman -U systemd-243.138-1-x86_64.pkg.tar.xz systemd-libs-243.138-1-x86_64.pkg.tar.xz systemd-resolvconf-243.138-1-x86_64.pkg.tar.xz systemd-sysvcompat-243.138-1-x86_64.pkg.tar.xz'
[2019-11-23T12:08:04-0500] [ALPM] transaction started
[2019-11-23T12:08:05-0500] [ALPM] downgraded systemd-libs (243.162-2 -> 243.138-1)
[2019-11-23T12:08:05-0500] [ALPM] downgraded systemd (243.162-2 -> 243.138-1)
[2019-11-23T12:08:05-0500] [ALPM] installed systemd-resolvconf (243.138-1)
[2019-11-23T12:08:05-0500] [ALPM] downgraded systemd-sysvcompat (243.162-2 -> 243.138-1)
[2019-11-23T12:08:05-0500] [ALPM] transaction completed
Last edited by CarbonChauvinist (2019-11-23 17:23:25)
"the wind-blown way, wanna win? don't play"
Offline
60 commits eliminated as the cause 24 to go
What do you think the next check should be? 5db454b8031c58a743cc4ee3d5d1dd01dcff17e8 adds 5 commits labeled udev back in.
That would still excluded two udev commits b9a619bb67387aa87e5fa916516eef4ced81f625 and aa1fc791c7a10402b420c4e94deed5ca2c33ae57
Edit:
https://github.com/systemd/systemd-stab … 243-stable
Last edited by loqs (2019-11-23 17:29:30)
Offline
I'm going to try 5db454b next I guess. I assume just follow the same steps you previously outlined, just swapping out the desired commit.
Edit: -- Tested and yes, delay is present with this commit.
Edit #2 -- Tested one commit back, b774282 and delay is NOT present. So does this mean it appears to be 5db454b is the cause?
Last edited by CarbonChauvinist (2019-11-23 18:43:03)
"the wind-blown way, wanna win? don't play"
Offline
Looks like it. As a cross check you can move the commit back to e51d9bf9e5ac5a6618c175cd9b5cfdc6733cd5d1 and add 5db454b8031c58a743cc4ee3d5d1dd01dcff17e8 to the reverts.
Offline
Yes, adding 5db454b to the reverts array when building e51d9bf works just fine, I can boot without the hang.
"the wind-blown way, wanna win? don't play"
Offline
Thank you for all the work tracking down the cause.
Could you please open a bug report on the arch bug tracker so the commit can be reverted and upstream so the issue can be fixed.
Offline
Thank you for all the work tracking down the cause.
Could you please open a bug report on the arch bug tracker so the commit can be reverted and upstream so the issue can be fixed.
Yes, thank you @CarbonChauvinist. Sorry I couldn't help more. Three kids eat up your free time
Offline
Offline
Hi, I'm affected by the same issue that appeared just after upgrading systemd to version 243.162-2.
Thank you @CarbonChauvinist for having opened the bug report on GitHub.
After having read this comment, I tried to reset a configuration of systemd for a silent boot mode that I did in the past by following this guide on ArchWiki.
In my situation, those settings now cause the delay during the boot.
My previous configuration of HOOKS in /etc/mkinitcpio was the following:
HOOKS=(base systemd fsck autodetect modconf block filesystems keyboard)
Simply, by restoring the original configuration, I didn't notice the 30 seconds delay.
HOOKS=(base udev autodetect modconf block filesystems keyboard fsck)
About "systemd-fsck-root.service" and "systemd-fsck@.service", at the time I kept the original ones in "/usr/lib/systemd/system/", copied them in "/etc/systemd/system/" and modified. Now I just deleted them.
After a couple of reboots/shutdowns, the delay seems to be disappeared.
I hope this could be helpful to solve the problem.
Offline
git diff of PKGBUILD after applying proposed fix from upstream.
diff --git a/trunk/PKGBUILD b/trunk/PKGBUILD
index cd3c133..1fe28aa 100644
--- a/trunk/PKGBUILD
+++ b/trunk/PKGBUILD
@@ -39,7 +39,8 @@ source=(# fragment is latest tag for source verification, final merge in prepare
'30-systemd-sysctl.hook'
'30-systemd-tmpfiles.hook'
'30-systemd-udev-reload.hook'
- '30-systemd-update.hook')
+ '30-systemd-update.hook'
+ https://github.com/systemd/systemd/commit/030f4571670537c76355c5d923468c9a61aa77e9.patch)
sha512sums=('SKIP'
'SKIP'
'9348683829190628e25b7b3300fd880c426d555bde330d5fc5150a9a54b3ad9d4d1f2e69ea1dc6d6f086693dacc53c5af30f1fa7ad9b479791fd77bcdafa430e'
@@ -59,12 +60,14 @@ sha512sums=('SKIP'
'da783e3bfc6469b92dee4064a13e2b427520d3d96b57c95a4e07aaca3e844d95210a8b16122b022080f5452d65096f274dd1c1467725bbdb2e40ef304b78774a'
'577e33a1c50b4b41157a67f64162b035dd0c4a541e19cee55a100048bdb50cb2c82852741b1372989a0fe4c4782ba477522747fcc81d72aed99b3db512a86447'
'e4a9d7607fe93daf1d45270971c8d8455c4bfc2c0bea8bcad05aeb89847edee23cd1a41073a72042622acf417018fe254f5bfc137604fe2c71292680bf67a1c2'
- '209b01b044877cc986757fa4009a92ea98f480306c2530075d153203c3cd2b3afccab6aacc1453dee8857991e04270572f1700310705d7a0f4d5bed27fab8c67')
+ '209b01b044877cc986757fa4009a92ea98f480306c2530075d153203c3cd2b3afccab6aacc1453dee8857991e04270572f1700310705d7a0f4d5bed27fab8c67'
+ '4538d076f7f49a13346b5325f35be427472c93bfa3fa05a494e0d73c8919986fa9dbc7b2b5f1d317ca356840227f37f64f92b1f89947da740f09d171bbd00aec')
_backports=(
)
_reverts=(
+ '5db454b8031c58a743cc4ee3d5d1dd01dcff17e8'
)
prepare() {
@@ -88,6 +91,9 @@ prepare() {
# Replace cdrom/dialout/tape groups with optical/uucp/storage
patch -Np1 -i ../0001-Use-Arch-Linux-device-access-groups.patch
+
+ # Fix shutdown hang caused by recent udev change
+ patch -p1 -i ../030f4571670537c76355c5d923468c9a61aa77e9.patch
}
pkgver() {
Offline
@mcaliandro - IIUC, you've bypassed the issue by not using a systemd init when you switched from systemd to udev hooks.
@loqs - there's been some movement on the upstream bug report, specifically I've been asked to:
please add 030f457 as backport or patch (keeping the revert of 5db454b in place), and try again
Thanks to your help I'm able to build with the revert of 5db454b -- can you please clarify as to how to add the backport? I guess I could just add the referenced 030f457 commit to the backports array? But I'm not sure if that's the right step as it appears to be different than say the commits that were listed in the systemd-stable repo that you had me testing originally?
--Edit: I see while I was typing up this post you've already beaten me to it @loqs, gave the answer before I even asked the question - new record! Thanks. I'll test this later tonight and reply back.
--Edit 2: Adding 030f457 and keeping the revert of 5db454b in place appears to address the reported issue. I can confirm that I'm no longer experiencing the delay when booting. Thanks again to @loqs for the invaluable help. Not sure what the next steps are -- maybe I should update the arch bug report?
Last edited by CarbonChauvinist (2019-11-26 21:19:23)
"the wind-blown way, wanna win? don't play"
Offline
This has been fixed for me in systemd-244-1, just pulled down an update today.
Offline
From what I can tell this should be fixed in the newest systemd packages just released (244-1) to core?
244 stable contains the 403e4b4 commit that fixes this issue. I'd been using a patched version of 243.162 to address, but I'll be able to confirm once I'm home and on the box that was presenting the issue.
Edit -- ninja'd
Edit #2 -- For anyone interested in what the actual cause was, check here
Last edited by CarbonChauvinist (2019-12-04 15:59:01)
"the wind-blown way, wanna win? don't play"
Offline
Confirm systemd 244-1 fixed my issue, too.
Eenie meenie, chili beanie, the spirits are about to speak -- Bullwinkle J. Moose
It's a big club...and you ain't in it -- George Carlin
Registered Linux user #149839
perl -e 'print$i=pack(c5,(41*2),sqrt(7056),(unpack(c,H)-2),oct(115),10); '
Offline
Me too, issue fixed.
Thanks a lot to all those who contributed to reporting and fixing the bug
Offline