Le samedi 16 octobre 2004 Ã 19:44 +1000, Russell Coker a Ãcrit : > Running the latest rawhide I get AVC messages indicating that /bin/udev > (not /sbin/udev) is running in kernel_t during the early stages of system > boot. > > /bin/udev is the file name used in the initrd! $ zcat /boot/initrd-2.6.8-1.624.img | cpio -iv -m -d $ ll -d bin sbin drwxr-xr-x 2 root root 4096 oct 16 11:50 bin lrwxrwxrwx 1 root root 3 oct 16 11:50 sbin -> bin $ ll bin/udev -rwxr-xr-x 1 root root 557808 oct 15 18:05 bin/udev > So it seems that after the SE > Linux policy is loaded (IE after /sbin/init has been run from the main root > fs) there is still a copy of udev from the initrd being run. There are some sleep() in udev. hotplug/udev/... are launched in background. Modprobe/insmod can return even if udev has not finished its job. $ grep insmod init insmod /lib/raid0.ko insmod /lib/jbd.ko insmod /lib/ext3.ko It's how I understand hotplug/udev. > This seems to > be a bug in initrd that could lead to inconsistent behaviour. I'm not sure > how this comes about (and of course apart from SE Linux messages in the > kernel message log all the evidence is gone by the time the system is ready > to login). > > Any suggestions on how to debug this? > > -- > http://www.coker.com.au/selinux/ My NSA Security Enhanced Linux packages > http://www.coker.com.au/bonnie++/ Bonnie++ hard drive benchmark > http://www.coker.com.au/postal/ Postal SMTP/POP benchmark > http://www.coker.com.au/~russell/ My home page >
Attachment:
signature.asc
Description: Ceci est une partie de message =?ISO-8859-1?Q?num=E9riquement?= =?ISO-8859-1?Q?_sign=E9e?=
