Bug 2228450

Description of problem:
When the installer is executed in rescue mode and the existing system is mounted by anaconda in read-write mode, the logs stored in /mnt/sysroot/var/log/anaconda are rewritten and SELinux context of some files may be changed. This happens right after anaconda mounts the detected system.

In other words, %post scripts from https://github.com/rhinstaller/anaconda/tree/rhel-8/data/post-scripts are executed even in rescue mode, which is not expected.

Version-Release number of selected component (if applicable):
RHEL-8.8 GA

How reproducible:
Always

Steps to Reproduce:
1. Install a RHEL system
2. Create a file in /root with an unexpected SELinux context:
# touch /root/my_file
# chcon -t httpd_sys_content_t /root/my_file
3. Start the installer in rescue mode ("inst.rescue" on the kernel command line)
4. Let anaconda find and mount the existing partitions
5. Check logs in /mnt/sysroot/var/log/anaconda and SELinux context of /mnt/sysroot/root/my_file

Actual results:
/mnt/sysroot/var/log/anaconda contains logs from the current rescue mode
The SELinux context of /mnt/sysroot/root/my_file is reset to the default

Expected results:
No changes made in the system being rescued.

Additional info:
A workaround to avoid rewriting /var/log/anaconda is to append "inst.nosave=all" on the kernel command line when starting rescue mode.