[PATCH] mkdumprd: try to get mount options from fstab first

[Baoquan He]

Previously if a target need mount info, the relevant mount options
are got from /proc/mounts by below command:
findmnt -k -f -n -r -o OPTIONS $_dev

This will bring problems. Since /proc/mounts will give out a set
which contains each option. Some options have value specified by
user, some options just have default value if user doesn't specify.
If some mount options are not supported very well, bugs occured.
The more options, the worse.

So in this patch, we try to check fstab to get mount options firstly,
this give user a chance to decide which options they really want.
If they don't give a fstab entry, then we trust all options in
/proc/mounts.

Signed-off-by: Baoquan He 
---
 mkdumprd | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/mkdumprd b/mkdumprd
index 409235b..e9246f2 100644
--- a/mkdumprd
+++ b/mkdumprd
@@ -104,7 +104,12 @@ to_mount() {
     # mount under /sysroot in 2nd kernel, and we umount -R /sysroot before exit
     _target="/sysroot$_target"
     _fstype=$(findmnt -k -f -n -r -o FSTYPE $_dev)
-    _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
+    _options=$(findmnt --fstab -f -n -r -o OPTIONS $_dev)
+    if [ -z "$_options" ]; then
+        _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
+    elif [ "$_options" = "defaults" ]; then
+        _options=""
+    fi
     _options=${_options/#ro/rw} #mount fs target as rw in 2nd kernel
     # "x-initrd.mount" mount failure will trigger isolate emergency service
     # W/o this, systemd won't isolate, thus we won't get to emergency.

[Dave Young]

On 09/04/14 at 04:59pm, Baoquan He wrote:
> Previously if a target need mount info, the relevant mount options
> are got from /proc/mounts by below command:
> findmnt -k -f -n -r -o OPTIONS $_dev
>
> This will bring problems. Since /proc/mounts will give out a set
> which contains each option. Some options have value specified by
> user, some options just have default value if user doesn't specify.
> If some mount options are not supported very well, bugs occured.
> The more options, the worse.
>
> So in this patch, we try to check fstab to get mount options firstly,
> this give user a chance to decide which options they really want.
> If they don't give a fstab entry, then we trust all options in
> /proc/mounts.
> Signed-off-by: Baoquan He
>
> mkdumprd | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/mkdumprd b/mkdumprd
> index 409235b..e9246f2 100644
> --- a/mkdumprd
> +++ b/mkdumprd
> @@ -104,7 +104,12 @@ to_mount() {
>      # mount under /sysroot in 2nd kernel, and we umount -R /sysroot before exit
>      _target="/sysroot$_target"
>      _fstype=$(findmnt -k -f -n -r -o FSTYPE $_dev)
> _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> _options=$(findmnt --fstab -f -n -r -o OPTIONS $_dev)
> if [ -z "$_options" ]; then
>    _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> elif [ "$_options" = "defaults" ]; then
>    _options=""
> fi

defaults should put into options instead of blank string becuase we are passing
fstab lines to dracut..
>  _options=${_options/#ro/rw} #mount fs target as rw in 2nd kernel
>  # "x-initrd.mount" mount failure will trigger isolate emergency service
>  # W/o this, systemd won't isolate, thus we won't get to emergency.
>
> -- 
> 1.8.5.3
>
> kexec mailing list
> [e-mail]
> https://lists.fedoraproject.org/mailman/listinfo/kexec

[Baoquan He]

On 09/04/14 at 05:26pm, Dave Young wrote:
> > --- a/mkdumprd
> > +++ b/mkdumprd
> > @@ -104,7 +104,12 @@ to_mount() {
> >      # mount under /sysroot in 2nd kernel, and we umount -R /sysroot before exit
> >      _target="/sysroot$_target"
> >      _fstype=$(findmnt -k -f -n -r -o FSTYPE $_dev)
> > _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> > _options=$(findmnt --fstab -f -n -r -o OPTIONS $_dev)
> > if [ -z "$_options" ]; then
> >    _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> > elif [ "$_options" = "defaults" ]; then
> >    _options=""
> > fi
>
> defaults should put into options instead of blank string becuase we are passing
> fstab lines to dracut..

Yes, you are right. Will post v2.
>

[PATCH v2] mkdumprd: try to get mount options from fstab first

[Baoquan He]

Previously if a target need mount info, the relevant mount options
are got from /proc/mounts by below command:
findmnt -k -f -n -r -o OPTIONS $_dev

This will bring problems. Since /proc/mounts will give out a set
which contains each option. Some options have value specified by
user, some options just have default value if user doesn't specify.
If some mount options are not supported very well, bugs occured.
The more options, the worse.

So in this patch, we try to check fstab to get mount options firstly,
this give user a chance to decide which options they really want.
If they don't give a fstab entry, then we trust all options in
/proc/mounts.

Signed-off-by: Baoquan He 
---
 mkdumprd | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/mkdumprd b/mkdumprd
index 409235b..a30d9ca 100644
--- a/mkdumprd
+++ b/mkdumprd
@@ -104,7 +104,8 @@ to_mount() {
     # mount under /sysroot in 2nd kernel, and we umount -R /sysroot before exit
     _target="/sysroot$_target"
     _fstype=$(findmnt -k -f -n -r -o FSTYPE $_dev)
-    _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
+    _options=$(findmnt --fstab -f -n -r -o OPTIONS $_dev)
+    [ -z "$_options" ] && _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
     _options=${_options/#ro/rw} #mount fs target as rw in 2nd kernel
     # "x-initrd.mount" mount failure will trigger isolate emergency service
     # W/o this, systemd won't isolate, thus we won't get to emergency.

[PATCH v2] mkdumprd: try to get mount options from fstab first

[Vivek Goyal]

On Thu, Sep 04, 2014 at 05:38:17PM +0800, Baoquan He wrote:
> Previously if a target need mount info, the relevant mount options
> are got from /proc/mounts by below command:
> findmnt -k -f -n -r -o OPTIONS $_dev
>
> This will bring problems. Since /proc/mounts will give out a set
> which contains each option. Some options have value specified by
> user, some options just have default value if user doesn't specify.
> If some mount options are not supported very well, bugs occured.
> The more options, the worse.
>
> So in this patch, we try to check fstab to get mount options firstly,
> this give user a chance to decide which options they really want.
> If they don't give a fstab entry, then we trust all options in
> /proc/mounts.
> Signed-off-by: Baoquan He
>
> mkdumprd | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/mkdumprd b/mkdumprd
> index 409235b..a30d9ca 100644
> --- a/mkdumprd
> +++ b/mkdumprd
> @@ -104,7 +104,8 @@ to_mount() {
>      # mount under /sysroot in 2nd kernel, and we umount -R /sysroot before exit
>      _target="/sysroot$_target"
>      _fstype=$(findmnt -k -f -n -r -o FSTYPE $_dev)
> _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> _options=$(findmnt --fstab -f -n -r -o OPTIONS $_dev)
> [ -z "$_options" ] && _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)

Ok, so kernel recognizes -o defaults?

Secondly, question of x-systemd.* options still remains.

Thanks
Vivek

[PATCH v2] mkdumprd: try to get mount options from fstab first

[Baoquan He]

On 09/15/14 at 11:09am, Vivek Goyal wrote:
> On Thu, Sep 04, 2014 at 05:38:17PM +0800, Baoquan He wrote:
> > Previously if a target need mount info, the relevant mount options
> > are got from /proc/mounts by below command:
> > findmnt -k -f -n -r -o OPTIONS $_dev
> >
> > This will bring problems. Since /proc/mounts will give out a set
> > which contains each option. Some options have value specified by
> > user, some options just have default value if user doesn't specify.
> > If some mount options are not supported very well, bugs occured.
> > The more options, the worse.
> >
> > So in this patch, we try to check fstab to get mount options firstly,
> > this give user a chance to decide which options they really want.
> > If they don't give a fstab entry, then we trust all options in
> > /proc/mounts.
> > Signed-off-by: Baoquan He
> >
> > mkdumprd | 3 ++-
> >  1 file changed, 2 insertions(+), 1 deletion(-)
> >
> > diff --git a/mkdumprd b/mkdumprd
> > index 409235b..a30d9ca 100644
> > --- a/mkdumprd
> > +++ b/mkdumprd
> > @@ -104,7 +104,8 @@ to_mount() {
> >      # mount under /sysroot in 2nd kernel, and we umount -R /sysroot before exit
> >      _target="/sysroot$_target"
> >      _fstype=$(findmnt -k -f -n -r -o FSTYPE $_dev)
> > _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> > _options=$(findmnt --fstab -f -n -r -o OPTIONS $_dev)
> > [ -z "$_options" ] && _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
>
> Ok, so kernel recognizes -o defaults?
>
> Secondly, question of x-systemd.* options still remains.

I just put the information of fstab in 1st kernel into fstab in 2nd
kernel. They should handle fstab in the same mechanism. I just tested
the -o defaults, it works. There should not be difference between 1st
kernel and 2nd kernel.
> Thanks
> Vivek

[PATCH v2] mkdumprd: try to get mount options from fstab first

[Vivek Goyal]

On Tue, Sep 16, 2014 at 03:06:25PM +0800, Baoquan He wrote:
> On 09/15/14 at 11:09am, Vivek Goyal wrote:
> > On Thu, Sep 04, 2014 at 05:38:17PM +0800, Baoquan He wrote:
> > > Previously if a target need mount info, the relevant mount options
> > > are got from /proc/mounts by below command:
> > > findmnt -k -f -n -r -o OPTIONS $_dev
> > >
> > > This will bring problems. Since /proc/mounts will give out a set
> > > which contains each option. Some options have value specified by
> > > user, some options just have default value if user doesn't specify.
> > > If some mount options are not supported very well, bugs occured.
> > > The more options, the worse.
> > >
> > > So in this patch, we try to check fstab to get mount options firstly,
> > > this give user a chance to decide which options they really want.
> > > If they don't give a fstab entry, then we trust all options in
> > > /proc/mounts.
> > > Signed-off-by: Baoquan He
> > >
> > > mkdumprd | 3 ++-
> > >  1 file changed, 2 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/mkdumprd b/mkdumprd
> > > index 409235b..a30d9ca 100644
> > > --- a/mkdumprd
> > > +++ b/mkdumprd
> > > @@ -104,7 +104,8 @@ to_mount() {
> > >      # mount under /sysroot in 2nd kernel, and we umount -R /sysroot before exit
> > >      _target="/sysroot$_target"
> > >      _fstype=$(findmnt -k -f -n -r -o FSTYPE $_dev)
> > > _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> > > _options=$(findmnt --fstab -f -n -r -o OPTIONS $_dev)
> > > [ -z "$_options" ] && _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> >
> > Ok, so kernel recognizes -o defaults?
> >
> > Secondly, question of x-systemd.* options still remains.
>
> I just put the information of fstab in 1st kernel into fstab in 2nd
> kernel. They should handle fstab in the same mechanism. I just tested
> the -o defaults, it works. There should not be difference between 1st
> kernel and 2nd kernel.

what about x-systemd* options?

I think dracut puts them into fstab files and respective mount units
are created and that's why we don't see the failure.

Otherwise I got a feeling that if you pass x-systemd* directly to mount
command, it might fail.

Thanks
Vivek

[PATCH v2] mkdumprd: try to get mount options from fstab first

[Baoquan He]

On 09/16/14 at 09:08am, Vivek Goyal wrote:
> On Tue, Sep 16, 2014 at 03:06:25PM +0800, Baoquan He wrote:
> > > @@ -104,7 +104,8 @@ to_mount() {
> > > >  # mount under /sysroot in 2nd kernel, and we umount -R /sysroot before exit
> > > >  _target="/sysroot$_target"
> > > >  _fstype=$(findmnt -k -f -n -r -o FSTYPE $_dev)
> > > > _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> > > > _options=$(findmnt --fstab -f -n -r -o OPTIONS $_dev)
> > > > [ -z "$_options" ] && _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> > >
> > > Ok, so kernel recognizes -o defaults?
> > >
> > > Secondly, question of x-systemd.* options still remains.
> >
> > I just put the information of fstab in 1st kernel into fstab in 2nd
> > kernel. They should handle fstab in the same mechanism. I just tested
> > the -o defaults, it works. There should not be difference between 1st
> > kernel and 2nd kernel.
>
> what about x-systemd* options?
>
> I think dracut puts them into fstab files and respective mount units
> are created and that's why we don't see the failure.
>
> Otherwise I got a feeling that if you pass x-systemd* directly to mount
> command, it might fail.

I checked systemd.mount manpage, it's saying that mount units may be
configured via unit files or /etc/fstab, and /etc/fstab is preferred.
Meanwhile if x-systemd.xxx is specified in systemd.mount files, it will
specify the specific behavior of systemd. That means x-systemd.xxx is
options for systemd.mount, not for /etc/fstab. So if get entry from
/etc/fstab via "findmnt --fstab xxx" command, no worry needed for
x-systemd.xxx. And x-systemd.xxx is not options of fstab, no need to
filter them.

Thanks
Baoquan
> Thanks
> Vivek

[PATCH v2] mkdumprd: try to get mount options from fstab first

[Vivek Goyal]

On Wed, Sep 17, 2014 at 06:46:49PM +0800, Baoquan He wrote:
> On 09/16/14 at 09:08am, Vivek Goyal wrote:
> > On Tue, Sep 16, 2014 at 03:06:25PM +0800, Baoquan He wrote:
> > > > @@ -104,7 +104,8 @@ to_mount() {
> > > > >  # mount under /sysroot in 2nd kernel, and we umount -R /sysroot before exit
> > > > >  _target="/sysroot$_target"
> > > > >  _fstype=$(findmnt -k -f -n -r -o FSTYPE $_dev)
> > > > > _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> > > > > _options=$(findmnt --fstab -f -n -r -o OPTIONS $_dev)
> > > > > [ -z "$_options" ] && _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> > > >
> > > > Ok, so kernel recognizes -o defaults?
> > > >
> > > > Secondly, question of x-systemd.* options still remains.
> > >
> > > I just put the information of fstab in 1st kernel into fstab in 2nd
> > > kernel. They should handle fstab in the same mechanism. I just tested
> > > the -o defaults, it works. There should not be difference between 1st
> > > kernel and 2nd kernel.
> >
> > what about x-systemd* options?
> >
> > I think dracut puts them into fstab files and respective mount units
> > are created and that's why we don't see the failure.
> >
> > Otherwise I got a feeling that if you pass x-systemd* directly to mount
> > command, it might fail.
>
> I checked systemd.mount manpage, it's saying that mount units may be
> configured via unit files or /etc/fstab, and /etc/fstab is preferred.
> Meanwhile if x-systemd.xxx is specified in systemd.mount files, it will
> specify the specific behavior of systemd. That means x-systemd.xxx is
> options for systemd.mount, not for /etc/fstab. So if get entry from
> /etc/fstab via "findmnt --fstab xxx" command, no worry needed for
> x-systemd.xxx. And x-systemd.xxx is not options of fstab, no need to
> filter them.

Bao,

Yes x-systemd* are options for systemd. But they can be specified in
/etc/fstab file. Only my system they are present.

So question is, who removes these options from command line when actual
"mount" is called.

Thanks
Vivek

[PATCH v2] mkdumprd: try to get mount options from fstab first

[Baoquan He]

On 09/17/14 at 08:49am, Vivek Goyal wrote:
> On Wed, Sep 17, 2014 at 06:46:49PM +0800, Baoquan He wrote:
> Bao,
>
> Yes x-systemd* are options for systemd. But they can be specified in
> /etc/fstab file. Only my system they are present.
>
> So question is, who removes these options from command line when actual
> "mount" is called.

I added x-systemd.device-timeout=1 into the nfs mount entry of
/etc/fstab, kdump can load and dump successfully. That means this can be
recognized by mount and dracut doesn't filter it out specifically.

Below is the fstab of my kvm guest:

[root@localhost ~]# cat /etc/fstab

#
# /etc/fstab
# Created by anaconda on Fri Mar 21 06:21:58 2014
#
# Accessible filesystems, by reference, are maintained under '/dev/disk'
# See man pages fstab(5), findfs(8), mount(8) and/or blkid(8) for more
# info
#
/dev/mapper/fedora-root /                       ext4    defaults
1 1
UUID=a3ab560e-e001-403a-b470-974ee5c1bf53 /boot                   ext4
defaults        1 2
/dev/mapper/fedora-swap swap                    swap    defaults
0 0
192.168.100.227:/var/crash/nfs  /mnt    nfs
defaults,x-systemd.device-timeout=1        0 0
> Thanks
> Vivek

[PATCH v2] mkdumprd: try to get mount options from fstab first

[Vivek Goyal]

On Thu, Sep 04, 2014 at 05:38:17PM +0800, Baoquan He wrote:
> Previously if a target need mount info, the relevant mount options
> are got from /proc/mounts by below command:
> findmnt -k -f -n -r -o OPTIONS $_dev
>
> This will bring problems. Since /proc/mounts will give out a set
> which contains each option. Some options have value specified by
> user, some options just have default value if user doesn't specify.
> If some mount options are not supported very well, bugs occured.
> The more options, the worse.
>
> So in this patch, we try to check fstab to get mount options firstly,
> this give user a chance to decide which options they really want.
> If they don't give a fstab entry, then we trust all options in
> /proc/mounts.
>
> Signed-off-by: Baoquan He

Acked-by: Vivek Goyal

Vivek
> mkdumprd | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/mkdumprd b/mkdumprd
> index 409235b..a30d9ca 100644
> --- a/mkdumprd
> +++ b/mkdumprd
> @@ -104,7 +104,8 @@ to_mount() {
>      # mount under /sysroot in 2nd kernel, and we umount -R /sysroot before exit
>      _target="/sysroot$_target"
>      _fstype=$(findmnt -k -f -n -r -o FSTYPE $_dev)
> _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> _options=$(findmnt --fstab -f -n -r -o OPTIONS $_dev)
> [ -z "$_options" ] && _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> _options=${_options/#ro/rw} #mount fs target as rw in 2nd kernel
> # "x-initrd.mount" mount failure will trigger isolate emergency service
> # W/o this, systemd won't isolate, thus we won't get to emergency.
>
> -- 
> 1.8.5.3
>
> kexec mailing list
> [e-mail]
> https://lists.fedoraproject.org/mailman/listinfo/kexec

[Vivek Goyal]

On Thu, Sep 04, 2014 at 04:59:29PM +0800, Baoquan He wrote:
> Previously if a target need mount info, the relevant mount options
> are got from /proc/mounts by below command:
> findmnt -k -f -n -r -o OPTIONS $_dev
>
> This will bring problems. Since /proc/mounts will give out a set
> which contains each option. Some options have value specified by
> user, some options just have default value if user doesn't specify.
> If some mount options are not supported very well, bugs occured.
> The more options, the worse.
>
> So in this patch, we try to check fstab to get mount options firstly,
> this give user a chance to decide which options they really want.
> If they don't give a fstab entry, then we trust all options in
> /proc/mounts.
> Signed-off-by: Baoquan He
>
> mkdumprd | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/mkdumprd b/mkdumprd
> index 409235b..e9246f2 100644
> --- a/mkdumprd
> +++ b/mkdumprd
> @@ -104,7 +104,12 @@ to_mount() {
>      # mount under /sysroot in 2nd kernel, and we umount -R /sysroot before exit
>      _target="/sysroot$_target"
>      _fstype=$(findmnt -k -f -n -r -o FSTYPE $_dev)
> _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> _options=$(findmnt --fstab -f -n -r -o OPTIONS $_dev)
> if [ -z "$_options" ]; then
>    _options=$(findmnt -k -f -n -r -o OPTIONS $_dev)
> elif [ "$_options" = "defaults" ]; then
>    _options=""
> fi

Bao, why do we set options to null in case of "defaults"?

man page says following.

defaults
                     use  default  options: rw, suid, dev, exec, auto, nouser,
                     and async.

So who comes up with defult values of these paratmers?

Secondly, what happens to parameters like.

x-systemd.device-timeout=0

I think kernel does not uderstand these options. Will we not get errors
during mounting if we don't filter out these options?

Thanks
Vivek