Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 438380 - [tracker] software that relies on /dev/root
Summary: [tracker] software that relies on /dev/root
Status: CONFIRMED
Alias: None
Product: Gentoo Linux
Classification: Unclassified
Component: Current packages (show other bugs)
Hardware: All Linux
: Normal normal (vote)
Assignee: Gentoo's Team for Core System packages
URL:
Whiteboard:
Keywords: Tracker
Depends on: 430426 438028 381761 427818 462628
Blocks:
  Show dependency tree
 
Reported: 2012-10-14 20:31 UTC by William Hubbs
Modified: 2023-12-13 01:06 UTC (History)
21 users (show)

See Also:
Package list:
Runtime testing required: ---


Attachments
findroot (findroot,357 bytes, text/plain)
2012-10-14 20:48 UTC, William Hubbs
Details
dev-root-real-root.patch (dev-root-real-root.patch,928 bytes, patch)
2012-10-15 21:37 UTC, William Hubbs
Details | Diff
Script to generate udev rule for creating /dev/root (create_dev_root.sh,382 bytes, patch)
2013-03-17 09:18 UTC, Samuli Suominen (RETIRED)
Details | Diff
Patch for udev-init-scripts (udev-init-scripts-24-dev-root-link.patch,1.13 KB, patch)
2013-03-21 08:00 UTC, Samuli Suominen (RETIRED)
Details | Diff
Fix prev. patch to create /dev/root also for rc_coldplug=NO setups (udev-init-scripts-24-dev-root-link.patch,1.13 KB, patch)
2013-03-21 08:29 UTC, Samuli Suominen (RETIRED)
Details | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description William Hubbs gentoo-dev 2012-10-14 20:31:45 UTC
According to the section on "Locally defined mounts" in [1], the
/dev/root symbolic link is not a requirement or even a recommendation,
so software which is using it  should be fixed to not do so.

This tracker is being opened to track these fixes.

So far, there are three fixes I know of, and I will be adding notes about
the fixes to this tracker as I come across them.

[1] http://www.kernel.org/doc/Documentation/devices.txt
Comment 1 William Hubbs gentoo-dev 2012-10-14 20:41:05 UTC
Here are the fixes that I know of so far:

1. Btrfs file systems do not yet provide a proper device name in
/proc/self/mounts. However, I had a chat with Jonathan Callen. He states
that he is using the kernel git and this has been rectified since 2.6
was released. I allowed him to test a proof-of-concept script I wrote
for case #2 below, and it worked correctly. I will attach that script
here for reference.

2. If software is looping through /proc/self/mounts, it should ignore
the line with "/dev/root" as the device name. It may already be ignoring
the line with "rootfs" as the device name, so this should not be a
difficult fix.

3. If software is attempting to refer to /dev/root directly, this should
not be happening. Instead, it should look up the root file system.
Possibly by looping through /proc/self/mounts keeping case #2 in mind.
Comment 2 William Hubbs gentoo-dev 2012-10-14 20:48:05 UTC
Created attachment 326560 [details]
findroot

This is my proof-of-concept script which shows how to look up a root
file system.

The original script used /proc/mounts, but I changed it here to use
/proc/self/mounts.
Comment 3 William Hubbs gentoo-dev 2012-10-14 21:37:33 UTC
(In reply to comment #1)
> 1. Btrfs file systems do not yet provide a proper device name in
> /proc/self/mounts. However, I had a chat with Jonathan Callen. He states
> that he is using the kernel git and this has been rectified since 2.6

This should be 3.6, sorry about my typo.

I confirmed with Mike Gilbert that this is fixed in linux-3.6.2.
Comment 4 Doug Goldstein (RETIRED) gentoo-dev 2012-10-14 21:56:37 UTC
Pointing to one little comment in the Documentation is a bit bunk IMHO. This has been something exposed to userspace for years and as such is really part of the API and its just yet another example of udev breaking things just for the sake of breaking things. Just put the damn symlink back, its really not hard to create. In some cases the symlink is there from devtmpfs and udev is going out of its way to delete it.
Comment 5 William Hubbs gentoo-dev 2012-10-14 22:39:24 UTC
(In reply to comment #4)
> Pointing to one little comment in the Documentation is a bit bunk IMHO. This
> has been something exposed to userspace for years and as such is really part
> of the API and its just yet another example of udev breaking things just for
> the sake of breaking things. Just put the damn symlink back, its really not
> hard to create. In some cases the symlink is there from devtmpfs and udev is
> going out of its way to delete it.

If you can show me the code in udev that is deleting /dev/root, I will gladly write a patch to fix that.

However, that will not solve the issue for people who aren't using udev.

Also, yes, the kernel does use /dev/root internally, to mount the rootfs if you aren't using an initramfs, but that's the only use I know of in the kernel, and it does not make that device available to user space.
Comment 6 William Hubbs gentoo-dev 2012-10-14 23:23:46 UTC
I found more information. We manufactured /dev/root to get around
bug #175243. It was never an upstream feature.
Comment 7 William Hubbs gentoo-dev 2012-10-14 23:36:27 UTC
My last comment should have said, instead of an upstream feature,
created by the kernel.
Comment 8 William Hubbs gentoo-dev 2012-10-15 12:01:50 UTC
The more I look at this, the issue is really a kernel issue, and I am in
agreement with the fix proposed on bug #175243, comments 11 and 12.
Comment 9 William Hubbs gentoo-dev 2012-10-15 12:44:35 UTC
It looks like the kernel has changed quite a bit in 3.7-rc1, so the patch
mentioned above may not apply.

We need someone to boot with linux-3.7-rc1, without an initramfs, and
let us know if /dev/root is still included in /proc/mounts.
Comment 10 SpanKY gentoo-dev 2012-10-15 20:21:48 UTC
(In reply to comment #1)

ignoring /dev/root is wrong.  most tools that are parsing /proc/self/mounts will want to probe the root mount -- most of the bugs you linked want exactly that.

likewise, copying & pasting the same logic in multiple applications to try and "discover" /dev/root is wrong.  when things change, or bugs are found, the same code needs to be fixed in multiple places.  you'd also have to assume that information was readily available to userspace which it might not be.  especially considering it isn't uncommon for /proc/cmdline and /etc/fstab to contain a UUID= or LABEL= line rather than the actual device.  that'd mean that apps would have to then rely on external applications and/or libs to further parse out that meaning.

(In reply to comment #8)

that's one possibility.  another is to have devtmpfs auto-create the rootfs symlink itself.  either is significantly better than having every single userspace project try and guess at the right answer.
Comment 11 William Hubbs gentoo-dev 2012-10-15 21:35:19 UTC
(In reply to comment #10)
> (In reply to comment #1)
> 
> ignoring /dev/root is wrong.  most tools that are parsing /proc/self/mounts
> will want to probe the root mount -- most of the bugs you linked want
> exactly that.

The problem is that /dev/root does not get placed in /proc/mounts at all if you boot from an initramfs, so it is not reliable.

> likewise, copying & pasting the same logic in multiple applications to try
> and "discover" /dev/root is wrong.  when things change, or bugs are found,
> the same code needs to be fixed in multiple places.  you'd also have to
> assume that information was readily available to userspace which it might
> not be.  especially considering it isn't uncommon for /proc/cmdline and
> /etc/fstab to contain a UUID= or LABEL= line rather than the actual device. 

What happens in /proc/mounts in this situation?

> that'd mean that apps would have to then rely on external applications
> and/or libs to further parse out that meaning.
> 
> (In reply to comment #8)
> 
> that's one possibility.  another is to have devtmpfs auto-create the rootfs
> symlink itself.  either is significantly better than having every single
> userspace project try and guess at the right answer.

I'm about to attach an updated patch from the original bug I referred to in comment #8. This is against the current kernel git, so you will need to test it against 3.7-rc1. The test will be to boot without an initramfs, then see if "/dev/root" shows up in /proc/mounts.
Comment 12 William Hubbs gentoo-dev 2012-10-15 21:37:38 UTC
Created attachment 326652 [details, diff]
dev-root-real-root.patch

Here is the patch; this was originally written by Cardoe.

Please test and make sure that /dev/root doesn't appear in /proc/mounts
after this is applied when you boot without an initramfs.
Comment 13 Naohiro Aota gentoo-dev 2012-10-24 04:55:06 UTC
(In reply to comment #12)
> Created attachment 326652 [details, diff] [details, diff]
> dev-root-real-root.patch
> 
> Here is the patch; this was originally written by Cardoe.
> 
> Please test and make sure that /dev/root doesn't appear in /proc/mounts
> after this is applied when you boot without an initramfs.

Fine here. It shows "/dev/vda1" instead of "/dev/root"
Comment 14 Sven 2013-01-20 03:19:51 UTC
I'm having troubles with a missing /dev/root for a while now. (For example, e4defrag refuses to work with a strange error message).
As far as I can see, attachment 326652 [details, diff] is a kernel patch. Are they aware of this upstream? Because I always use kernels straight from kernel.org.
Comment 15 Naohiro Aota gentoo-dev 2013-01-25 00:46:31 UTC
Any comment from kernel team for the above patch?
Comment 16 Mike Pagano gentoo-dev 2013-01-25 16:40:50 UTC
So this patch is for people not using udev? 

I'm ok with including this in 3.7. William, thoughts on submitting upstream?
Comment 17 Mike Pagano gentoo-dev 2013-01-26 00:23:19 UTC
the patch is INSVN
Comment 18 cosmoslx 2013-01-27 01:28:58 UTC
In my case, the workaround script 'findroot' will not find the root.
Because there isn't a '/' mount point without '/dev/root'
my current kernel version is 3.5.7

here is the detail:

# cat /proc/mounts | grep -w /

rootfs / rootfs rw 0 0
/dev/root / ext4 rw,noatime,data=ordered 0 0
Comment 19 Andrew Savchenko gentoo-dev 2013-01-27 10:44:16 UTC
The same here with kernel-3.7.4 and udev-196-r1:

$ grep root /proc/mounts
rootfs / rootfs rw 0 0
/dev/root / ext4 rw,noatime,stripe=32,jqfmt=vfsv1,usrjquota=aquota.user 0 0
/dev/root /home/ftp/distributive/gentoo/portage ext4 rw,noatime,stripe=32,jqfmt=vfsv1,usrjquota=aquota.user 0 0
/dev/root /home/ftp/distributive/gentoo/layman ext4 rw,noatime,stripe=32,jqfmt=vfsv1,usrjquota=aquota.user 0 0

But not /dev/root device!
I have to create it by modifying quota init script in order to quotas to work.

$ grep DEVTMP /boot/config-3.7.4-yoruichi 
CONFIG_DEVTMPFS=y
CONFIG_DEVTMPFS_MOUNT=y
Comment 20 William Hubbs gentoo-dev 2013-01-27 21:10:05 UTC
(In reply to comment #16)
> So this patch is for people not using udev? 
> 
> I'm ok with including this in 3.7. William, thoughts on submitting upstream?

@mpagano:
I think I would rather have this patch in upstream as opposed to having it in our kernel patchset so we are not depending on a gentoo specific fix.

I need to look back at the patch and write a description for it on this bug, but if I do that, can we get it submitted upstream?

As I remember  (I haven't looked at this in a while), The issue I found was that /dev/root should not exist at all in /proc/mounts, but it does if you do not boot with an initramfs.
Comment 21 Mike Pagano gentoo-dev 2013-01-27 21:23:23 UTC
(In reply to comment #20)
> @mpagano:
> I think I would rather have this patch in upstream as opposed to having it
> in our kernel patchset so we are not depending on a gentoo specific fix.
> 
> I need to look back at the patch and write a description for it on this bug,
> but if I do that, can we get it submitted upstream?


I can help you submit it, or submit it, whichever you prefer.
Comment 22 Manfred Knick 2013-03-10 15:47:58 UTC
Re-Visited ? with 3.8.2 ?

Please, c.f. https://bugs.gentoo.org/show_bug.cgi?id=381761#c23 . Thanks.
Comment 23 Samuli Suominen (RETIRED) gentoo-dev 2013-03-17 09:18:08 UTC
Created attachment 342330 [details, diff]
Script to generate udev rule for creating /dev/root

This script can be executed to create a /etc/udev/rules.d/40-root.rules, and then simply issuing `udevadm trigger' will create /dev/root:

ssuominen@null /tmp $ sudo ./create_dev_root.sh 
ssuominen@null /tmp $ cat /etc/udev/rules.d/40-root.rules 
ACTION=="add|change", SUBSYSTEM=="block", ENV{MAJOR}=="8", ENV{MINOR}=="1", SYMLINK+="root"
ssuominen@null /tmp $ sudo -s
null tmp # udevadm trigger
null tmp # ls /dev/root 
/dev/root
null tmp # ls -ld  /dev/root 
lrwxrwxrwx 1 root root 4 Mar 17 11:15 /dev/root -> sda1
Comment 24 Samuli Suominen (RETIRED) gentoo-dev 2013-03-21 08:00:19 UTC
Created attachment 342828 [details, diff]
Patch for udev-init-scripts

Patch for udev-init-scripts, applies to 24 and 9999, should go into 25
Comment 25 Lars Wendler (Polynomial-C) (RETIRED) gentoo-dev 2013-03-21 08:16:48 UTC
(In reply to comment #24)
> Created attachment 342828 [details, diff] [details, diff]
> Patch for udev-init-scripts
> 
> Patch for udev-init-scripts, applies to 24 and 9999, should go into 25

This patch only creates /dev/root when rc_coldplug=YES is set. With rc_coldplug=NO the whole sequence gets omitted. Please fix.
Comment 26 Samuli Suominen (RETIRED) gentoo-dev 2013-03-21 08:29:56 UTC
Created attachment 342830 [details, diff]
Fix prev. patch to create /dev/root also for rc_coldplug=NO setups

(In reply to comment #25)
> (In reply to comment #24)
> > Created attachment 342828 [details, diff] [details, diff] [details, diff]
> > Patch for udev-init-scripts
> > 
> > Patch for udev-init-scripts, applies to 24 and 9999, should go into 25
> 
> This patch only creates /dev/root when rc_coldplug=YES is set. With
> rc_coldplug=NO the whole sequence gets omitted. Please fix.

This should do it. It again boiled down to what we should do in rc_coldplug=NO setup and whatnot, I actually left it intentionally out when writing the initial patch, however now that you pointed it out... I don't know why... 

Thanks!
Comment 27 William Hubbs gentoo-dev 2013-03-21 12:27:38 UTC
Due to this discussion on lkml, I don't think we should create /dev/root
at all [1]. There is a patch pending, so I want to see what ehappens
with it.

[1] https://lkml.org/lkml/2013/1/31/574
Comment 28 Samuli Suominen (RETIRED) gentoo-dev 2013-03-21 12:30:02 UTC
(In reply to comment #27)
> Due to this discussion on lkml, I don't think we should create /dev/root
> at all [1]. There is a patch pending, so I want to see what ehappens
> with it.
> 
> [1] https://lkml.org/lkml/2013/1/31/574

It makes no difference whatsoever if they take the patch or not, the packages in tree, like quota, are expecting the /dev/root symlink to be there right now and no later kernel changes will fix that
The only way you can avoid adding back the /dev/root symlink is by fixing the applications code, like quota's, no kernel patches involved
Comment 29 Samuli Suominen (RETIRED) gentoo-dev 2013-03-21 12:36:56 UTC
even in the link you referred yourself,
https://lkml.org/lkml/2013/2/4/766
it was noted that the removal of /dev/root breaks some userspaces
and indeed, it does, ours, for quota

i hope we can get new udev-init-scripts with this (Comment #26) and the coldplug fix (bug 461838) in tree soon as possible (today?)
Comment 30 Alexander Bezrukov 2013-03-21 18:40:34 UTC
Even if non-existent /dev/root is not mentioned in /proc/mounts (which would probably fix at least some userspace) it is, IMHO, just convenient to have such a symlink in /dev anyway when boot is from a block device. And for now (where the "/dev/root" line is in /proc/mounts) not only quota is affected.
Comment 31 Samuli Suominen (RETIRED) gentoo-dev 2013-03-21 19:02:12 UTC
(In reply to comment #30)
> Even if non-existent /dev/root is not mentioned in /proc/mounts (which would
> probably fix at least some userspace) it is, IMHO, just convenient to have
> such a symlink in /dev anyway when boot is from a block device. And for now
> (where the "/dev/root" line is in /proc/mounts) not only quota is affected.

Right. Talked this over with WilliamH at IRC and we ended up with:

http://git.overlays.gentoo.org/gitweb/?p=proj/udev-gentoo-scripts.git;a=commit;h=ba463c6841c2cd0ab58aac391bd84613694cb7a9

It will be in udev-init-scripts-25. 

This Tracker will of course stay open until everything has been ported away from /dev/root. The compability symlink support will stay in udev-init-scripts for long as quota is not fixed, sys-boot/grub:2 is not in stable, and the kernel issue decided with.

I think we can stop talking in this Tracker now again. Sorry everyone for the noise.
Comment 32 William Hubbs gentoo-dev 2013-03-21 21:33:40 UTC
I think the kernel issue is not really relevant as long as the user
space components are fixed.
Comment 33 cilly 2013-03-28 09:40:18 UTC
quota depends on it:

warnquota: Cannot stat() mounted device /dev/root: No such file or directory

sys-fs/quota-3.17-r1

x86
Comment 34 Samuli Suominen (RETIRED) gentoo-dev 2013-03-28 09:51:11 UTC
(In reply to comment #33)
> quota depends on it:
> 
> warnquota: Cannot stat() mounted device /dev/root: No such file or directory
> 
> sys-fs/quota-3.17-r1
> 
> x86

no kidding? that's why this bug is linked to bug 438028 and that's why /dev/root was restored with udev-init-scripts-25 (except for btrfs where it can't work correctly)

please refrain from commenting to this tracker, if you have new packages failing, file a new bug and make it block this one (except there is one already for quota)
Comment 35 cilly 2013-07-16 07:54:13 UTC
Anyhow, I stepped into the same problem and somehow the issue is not solved for me and quota is still complaining about missing /dev/root since udev does not create it.

Workaround:

1. add the following two lines to a file with a number prior to 61, i.e. 
/etc/udev/rules.d/10-local.rules:

# root for quota in /dev/root for udev >172 only
KERNEL=="sda3", SUBSYSTEM=="block", SYMLINK+="root"

(In my case root is on sda3, so make changes appropriate to your filesystem!)

2. make sure, udev does not remove it again and add the following two lines to the file:
/etc/udev/rules.d/61-dev-root-link.rules 

# root for quota in /dev/root for udev >172 only
ACTION!="remove", SUBSYSTEM=="block", ENV{MAJOR}=="", ENV{MINOR}=="", SYMLINK+="root"

(Make sure, the KERNEL and ACTION lines are in a single line! Lines beginning with # are comments and may be omitted.)

Hope this helps!

(using udev-init-scripts-26, udev-204)
Comment 36 Sven 2013-07-16 08:19:04 UTC
(In reply to Samuli Suominen from comment #34)
> no kidding? that's why this bug is linked to bug 438028 and that's why
> /dev/root was restored with udev-init-scripts-25 (except for btrfs where it
> can't work correctly)

I have udev-init-scripts-26 and /dev/root is not created.
Comment 37 cilly 2013-07-16 08:42:44 UTC
(In reply to Sven from comment #36)
 
> I have udev-init-scripts-26 and /dev/root is not created.

Same here, but there is a workaround, see my comment:

https://bugs.gentoo.org/show_bug.cgi?id=438380#c35
Comment 38 Tom Wijsman (TomWij) (RETIRED) gentoo-dev 2013-08-17 12:12:09 UTC
(In reply to cilly from comment #35)
> Anyhow, I stepped into the same problem and somehow the issue is not solved
> for me and quota is still complaining about missing /dev/root since udev
> does not create it.

This bug is not the place to report about quota, please use bug #438028 instead.
Comment 39 Samuli Suominen (RETIRED) gentoo-dev 2013-08-17 12:42:49 UTC
It's not very complex system that creates the symlink. You can look inside of the /lib/udev/dev-root-link.sh for the udevadm command it uses to get the required info.

This has been working for me since 25 without issues on multiple boxes :-/

# ls -ld /dev/root
lrwxrwxrwx 1 root root 4 Aug 11 17:50 /dev/root -> sda1
ssuominen@null ~/gentoo-x86/profiles $ ls -ld /dev/{root,sda1}
lrwxrwxrwx 1 root root    4 Aug 11 17:50 /dev/root -> sda1
brw-rw---- 1 root disk 8, 1 Aug 11 17:50 /dev/sda1

# cat /run/udev/rules.d/61-dev-root-link.rules 
ACTION=="add|change", SUBSYSTEM=="block", ENV{MAJOR}=="8", ENV{MINOR}=="1", SYMLINK+="root"

# rc-update show |grep udev
                 udev |                       sysinit
           udev-mount |                       sysinit

# grep dev-root /etc/init.d/udev
/etc/init.d/udev:		/lib/udev/dev-root-link.sh

# ls -l /lib/udev/dev-root-link.sh

[ebuild   R    ] sys-fs/udev-206-r3  USE="acl firmware-loader gudev introspection kmod openrc -doc (-selinux) -static-libs" ABI_X86="32 (64) (-x32)" 0 kB
[ebuild   R    ]  sys-fs/udev-init-scripts-26  0 kB
Comment 40 Samuli Suominen (RETIRED) gentoo-dev 2013-08-17 12:45:26 UTC
(In reply to Samuli Suominen from comment #39)
> # cat /run/udev/rules.d/61-dev-root-link.rules 
> ACTION=="add|change", SUBSYSTEM=="block", ENV{MAJOR}=="8", ENV{MINOR}=="1",
> SYMLINK+="root"

These come from the udevadm command of:

$ udevadm info --export --export-prefix=ROOT_ --device-id-of-file=/
ROOT_MAJOR=8
ROOT_MINOR=1

So if the udevadm command isn't itself failing, and the filesystem is not btrfs, then I can't really explain why it wouldn't create the link.

Anyway, please file a new bug for udev-init-scripts with the required information if this is failing for you with current udev-init-scripts. Thanks.
Comment 41 Chris Brannon (RETIRED) gentoo-dev 2014-02-16 03:14:50 UTC
With both vanilla-sources and hardened-sources, if you do not boot
with an initramfs, then /proc/mounts does not contain an entry for your
root partition, but it does contain /dev/root.
This breaks extlinux if the /dev/root symlink does not exist.
Ordinarily, I don't think extlinux would have any trouble.
The problem for extlinux is that there isn't an entry for the physical device.

This isn't a problem for gentoo-sources, since it has a patch to replace
/dev/root with the real root device in /proc/mounts.
So I'm pretty convinced that the presence of /dev/root in /proc/mounts
is a kernel bug.

Here's my /proc/mounts on a system with hardened-sources.
WilliamH confirmed this for vanilla-sources.

rootfs / rootfs rw 0 0
/dev/root / ext3 rw,noatime,errors=continue,user_xattr,barrier=1,data=ordered 0 0
# There should be an entry for /dev/sda3 here, but there isn't.
devtmpfs /dev devtmpfs rw,relatime,size=251088k,nr_inodes=62772,mode=755 0 0
proc /proc proc rw,nosuid,nodev,noexec,relatime 0 0
tmpfs /run tmpfs rw,nosuid,nodev,relatime,size=50320k,mode=755 0 0
mqueue /dev/mqueue mqueue rw,nosuid,nodev,noexec,relatime 0 0
devpts /dev/pts devpts rw,nosuid,noexec,relatime,gid=5,mode=620 0 0
shm /dev/shm tmpfs rw,nosuid,nodev,noexec,relatime 0 0
sysfs /sys sysfs rw,nosuid,nodev,noexec,relatime 0 0
debugfs /sys/kernel/debug debugfs rw,nosuid,nodev,noexec,relatime 0 0
cgroup_root /sys/fs/cgroup tmpfs rw,nosuid,nodev,noexec,relatime,size=10240k,mode=755 0 0
openrc /sys/fs/cgroup/openrc cgroup rw,nosuid,nodev,noexec,relatime,release_agent=/lib64/rc/sh/cgroup-release-agent.sh,name=openrc 0 0
cpuset /sys/fs/cgroup/cpuset cgroup rw,nosuid,nodev,noexec,relatime,cpuset 0 0
cpu /sys/fs/cgroup/cpu cgroup rw,nosuid,nodev,noexec,relatime,cpu 0 0
cpuacct /sys/fs/cgroup/cpuacct cgroup rw,nosuid,nodev,noexec,relatime,cpuacct 0 0
freezer /sys/fs/cgroup/freezer cgroup rw,nosuid,nodev,noexec,relatime,freezer 0 0
binfmt_misc /proc/sys/fs/binfmt_misc binfmt_misc rw,nosuid,nodev,noexec,relatime 0 0
Comment 42 Nick Bowler 2015-01-15 15:04:04 UTC
So I don't understand.  Is /dev/root supposed to be created on Gentoo or not?

I filed bug 536658 to report that the symlink is NOT being created, but it was closed as a dup of bug 438028 (issue in quota).
Comment 43 Milos Ivanovic 2016-10-03 08:40:56 UTC
William,

What happened with the patch you proposed at https://lkml.org/lkml/2013/1/31/574? There hasn't been any discussion since early 2013. I just installed a vanilla kernel and, as has been stated, /dev/root is present in /proc/mounts which is undesired.

I'd like to help get this in mainline unless there's a major reason against it.
Comment 44 Milos Ivanovic 2016-10-03 08:43:25 UTC
For additional reference, the discussion can be read at https://patchwork.kernel.org/patch/2076031/ and seems to end without any formal resolution.