Bug#928981: fai-client: regression introduced with bugfix for #925247: leaving /run/udev behind itself

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

Bug#928981: fai-client: regression introduced with bugfix for #925247: leaving /run/udev behind itself

Michael Prokop-5
Package: fai-client
Version: 5.8.4
Severity: important

Hi,

with the following change I noticed a regression in FAI's behavior
of task_updatebase(), which leaves /run/udev behind itself:

| commit a03e71b12a544460ba78c97f67623a461b3ecc54
| Author: Thomas Lange <[hidden email]>
| Date:   Mon Mar 25 21:22:25 2019 +0100
|
|     mount /run/udev into /target, closes: #925247
|
| diff --git a/lib/updatebase b/lib/updatebase
| index 5ed845d1..fc1d1e4d 100755
| --- a/lib/updatebase
| +++ b/lib/updatebase
| @@ -13,6 +13,8 @@ if [ "$FAI_ACTION" = "install" -o "$FAI_ACTION" = "dirinstall" ]; then
|      if [ -f /etc/init.d/udev ]; then
|        mount --bind /dev $FAI_ROOT/dev
|        mount --make-private $FAI_ROOT/dev
| +      mkdir -p $target/run/udev
| +      mount --bind /run/udev $target/run/udev
|      fi
|      mount -t devpts devpts $FAI_ROOT/dev/pts
|      mount --make-private $FAI_ROOT/dev/pts

I noticed the bug in grml-live's fai dirinstall usage, leaving
/run/udev inside the chroot behind:

| rm: cannot remove '/srv/jenkins/jobs/grml64-small_sid/workspace/grml_chroot/run/udev': Device or resource busy

regards
-mika-