Bug 11990

Summary:	If using run level 3, and startx, /usr remounted ro, and unwanted options added to other mountpoint.
Product:	Mageia	Reporter:	Dave Hodgins <davidwhodgins>
Component:	RPM Packages	Assignee:	Chris Denice <eatdirt>
Status:	RESOLVED OLD	QA Contact:
Severity:	normal
Priority:	Normal	CC:	mageia
Version:	3	Keywords:	Triaged
Target Milestone:	---
Hardware:	i586
OS:	Linux
Whiteboard:
Source RPM:	xinit-1.3.2-5.mga3.src.rpm	CVE:
Status comment:

Description Dave Hodgins 2013-12-14 22:18:19 CET

On my old i586 system, if I use run level 3, and startx, mount options change
as follows ...
[dave@hodgins ~]$ diff mount.b4startx.sorted mount.afstartx.sorted 
14c14
< /dev/mapper/bk-usr on /usr type ext4 (rw,relatime,data=ordered)
---
> /dev/mapper/bk-usr on /usr type ext4 (ro,relatime,data=ordered)
19c19
< /dev/mapper/luks91 on /var/mnt/data type ext4 (rw,relatime,nodelalloc,user)
---
> /dev/mapper/luks91 on /var/mnt/data type ext4 (rw,nosuid,nodev,noexec,relatime,nodelalloc,user)

No changes are made to the mount options, if I use run level 5.

I'm guessing about which rpm package to report this against, as I have no
idea what is causing the difference, or where the actual remounting occurs.

The fstab for /usr is
/dev/mapper/bk-usr /usr ext4 acl,commit=0,user_xattr,rw 1 2
(I added the rw, explicitly, when I thought it was dracut mounting it ro,
before I realized it only gets remounted ro when I run startx).

/var/mnt/data is mounted by a script, before running startx with
sudo /bin/mount -v -t ext4 -o defaults,user,exec,dev,suid,relatime,user_xattr /dev/mapper/$MapperName $MountPoint
/usr is dm-0, luks91 is dm-9

# journalctl -ab|grep -e dm-0 -e dm-9
Dec 14 16:13:33 hodgins.homeip.net kernel: EXT4-fs (dm-0): mounted filesystem with ordered data mode. Opts: acl,commit=0,user_xattr
Dec 14 16:13:33 hodgins.homeip.net kernel: EXT4-fs (dm-0): re-mounted. Opts: acl,commit=0,user_xattr
Dec 14 16:14:00 hodgins.homeip.net kernel: EXT4-fs (dm-0): re-mounted. Opts: acl,commit=0,user_xattr,acl,user_xattr,commit=0
Dec 14 16:14:23 hodgins.homeip.net kernel: EXT4-fs (dm-9): mounted filesystem with journalled data mode. Opts: user_xattr
Dec 14 16:14:59 hodgins.homeip.net kernel: EXT4-fs (dm-0): re-mounted. Opts: acl,commit=0,user_xattr,commit=0
Dec 14 16:15:00 hodgins.homeip.net kernel: EXT4-fs (dm-0): re-mounted. Opts: data=ordered,commit=0
Dec 14 16:15:01 hodgins.homeip.net kernel: EXT4-fs (dm-0): re-mounted. Opts: data=ordered,commit=0
Dec 14 16:15:01 hodgins.homeip.net kernel: EXT4-fs (dm-9): re-mounted. Opts: nodelalloc,commit=0

dmesg is not showing any filesystem errors. This happens every time I use
startx with run level 3, and does not happen at all, with run level 5.



Reproducible: 

Steps to Reproduce:

Comment 1 Dave Hodgins 2013-12-14 22:31:48 CET

One possibly interesting thing, changes in /proc/self/mountinfo after running
startx ...
diff mountinfo.b4startx mountinfo.afstartx 
8c8
< 22 21 252:0 / /usr rw,relatime shared:2 - ext4 /dev/mapper/bk-usr rw,data=ordered
---
> 22 21 252:0 / /usr rw,relatime shared:2 - ext4 /dev/mapper/bk-usr ro,data=ordered
33,34c33,34
< 47 39 252:0 /lib/bind /var/lib/named/usr/lib/bind ro,relatime shared:2 - ext4 /dev/mapper/bk-usr rw,data=ordered
< 48 39 252:0 /lib/openssl /var/lib/named/usr/lib/openssl ro,relatime shared:2 - ext4 /dev/mapper/bk-usr rw,data=ordered
---
> 47 39 252:0 /lib/bind /var/lib/named/usr/lib/bind ro,relatime shared:2 - ext4 /dev/mapper/bk-usr ro,data=ordered
> 48 39 252:0 /lib/openssl /var/lib/named/usr/lib/openssl ro,relatime shared:2 - ext4 /dev/mapper/bk-usr ro,data=ordered
36c36
< 88 41 252:9 / /var/mnt/data rw,relatime shared:66 - ext4 /dev/mapper/luks91 rw,nodelalloc
---
> 88 41 252:9 / /var/mnt/data rw,nosuid,nodev,noexec,relatime shared:66 - ext4 /dev/mapper/luks91 rw,nodelalloc

Comment 2 Dave Hodgins 2013-12-14 23:06:34 CET

Another clue. This does not happen with "startx Openbox", but does with either
startx KDE4 or startx GNOME.

Comment 3 Dave Hodgins 2013-12-14 23:54:02 CET

Just fyi, create a new user, and confirmed it also happens with that user.

Comment 4 Dave Hodgins 2013-12-14 23:56:54 CET

typo. Created a new user, and confirmed it also happens with that user.

Manuel Hiebel 2013-12-22 15:13:38 CET

Keywords: (none) => Triaged
CC: (none) => mageia
Assignee: bugsquad => dirteat

Comment 5 Chris Denice 2013-12-22 17:15:30 CET

ok, I'll have a look. I see Colin is in cc, I may call for help at some point... :)

cheers.

Comment 6 Chris Denice 2013-12-29 23:44:03 CET

Hi there,
I cannot reproduce with my mga3, my partitions remain mounted as they are under runlevel 3, even after startx GNOME or KDE (rw, relative, data=ordered).


But I see in your messages that the concernes partitions are under /dev/mapper/; are they encrypted?

cheers,
chris.

Comment 7 Chris Denice 2014-04-21 12:13:30 CEST

I have no chance to help for this bug is noone can confirm or reproduced this bug.
I switched to unconfirmed.

cheers,
chris.

Status: NEW => UNCONFIRMED
Ever confirmed: 1 => 0

Comment 8 Dave Hodgins 2014-06-26 00:16:05 CEST

I've switched that i586 system to Mageia 4, where this doesn't happen.
Closing the bug as old.

Status: UNCONFIRMED => RESOLVED
Resolution: (none) => OLD