Re: [Xen-devel] [PATCH v4] run QEMU as non-root

[Ian Campbell <ian.campbell@xxxxxxxxxx>]

On Mon, 2015-06-29 at 18:06 +0100, Stefano Stabellini wrote:
> On Thu, 25 Jun 2015, Ian Campbell wrote:
> > On Mon, 2015-06-01 at 17:46 +0100, Stefano Stabellini wrote:
> > > Try to use "xen-qemudepriv-$domname" first, then
> > > "xen-qemudepriv-domid$domid", finally "xen-qemudepriv-shared" and root
> > > if everything else fails.
> > > 
> > > The uids need to be manually created by the user or, more likely, by the
> > > xen package maintainer.
> > > 
> > > To actually secure QEMU when running in Dom0, we need at least to
> > > deprivilege the privcmd and xenstore interfaces, this is just the first
> > > step in that direction.
> > > 
> > > Signed-off-by: Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>
> > > 
> > > ---
> > > Changes in v4:
> > > - rename qemu-deprivilege to qemu-deprivilege.txt
> > > - add a note about qemu-deprivilege.txt to INSTALL
> > > - instead of xen-qemudepriv-base + $domid, try xen-qemudepriv-domid$domid
> > > - introduce libxl__dm_runas_helper to make the code nicer
> > > 
> > > Changes in v3:
> > > - clarify doc
> > > - handle errno == ERANGE
> > > ---
> > >  INSTALL                        |    7 ++++++
> > >  docs/misc/qemu-deprivilege.txt |   34 +++++++++++++++++++++++++
> > >  tools/libxl/libxl_dm.c         |   54 
> > > ++++++++++++++++++++++++++++++++++++++++
> > >  tools/libxl/libxl_internal.h   |    4 +++
> > >  4 files changed, 99 insertions(+)
> > >  create mode 100644 docs/misc/qemu-deprivilege.txt
> > > 
> > > diff --git a/INSTALL b/INSTALL
> > > index a0f2e7b..3286a5f 100644
> > > --- a/INSTALL
> > > +++ b/INSTALL
> > > @@ -297,6 +297,13 @@ systemctl enable xendomains.service
> > >  systemctl enable xen-watchdog.service
> > >  
> > > 
> > > +QEMU Deprivilege
> > > +================
> > > +It is recommended to run QEMU as non-root.
> > > +See docs/misc/qemu-deprivilege.txt for an explanation on what you need
> > > +to do at installation time to run QEMU as a regular user.
> > 
> > I'd be tempted to say something like "as a dedicated user" or something,
> > "regular user" sounds like I should be running it as my normal login
> > user or something.
> 
> OK
> 
> 
> > > +
> > > +
> > >  History of options
> > >  ==================
> > >  
> > > diff --git a/docs/misc/qemu-deprivilege.txt 
> > > b/docs/misc/qemu-deprivilege.txt
> > > new file mode 100644
> > > index 0000000..db61aaf
> > > --- /dev/null
> > > +++ b/docs/misc/qemu-deprivilege.txt
> > > @@ -0,0 +1,34 @@
> > > +For security reasons, libxl tries to create QEMU as non-root.
> > > +Libxl looks for the following users in this order:
> > > +
> > > +1) a user named "xen-qemuuser-$domname"
> > > +Where $domname is the name of the domain being created.
> > > +To use this, you just need to create a user with the appropriate name
> > > +for each domain. For example, if your virtual machine is named "windows":
> > > +
> > > +adduser --system xen-qemuuser-windows
> > 
> > I just had an annoying thought: On migration (with xl at least) the
> > receive side appends "-incoming" to the domain name for the duration and
> > then renames ones everything is done, which is suddenly not going to
> > match here :-(
> > 
> > Since this is managed at the xl level I don't see any way libxl could
> > cope without being told separately the domain's "real" name.
> > 
> > I think this reduces the utility of this option quite substantially (or
> > at least presents an enormous caveat), perhaps to the point where we may
> > as well not bother to offer it?
> 
> I think we can detect the "--incoming" and remove it. The only problem
> is that "--incoming" is actually set by xl (xl_cmdimpl.c), not libxl.
> Could we break the abstraction layer just this time and in libxl_dm.c
> have something like:
> 
> 
>      if ((strlen(c_info->name) > strlen("--incoming")) &&
>          (!strcmp(c_info->name + strlen(c_info->name) - strlen("--incoming"), 
> "--incoming"))) {
>          name = libxl__strdup(gc, c_info->name);
>          name[strlen(c_info->name) - strlen("--incoming")] = '\0';
>      } else {
>          name = c_info->name;
>      }
>      user = libxl__sprintf(gc, "%s-%s", LIBXL_QEMU_USER_PREFIX, name);
> 
> 
> This is ugly but it was only meant to demonstrate the idea.

It's not brilliant :-(

At the risk of going off into the weeds, is there any mileage in
exposing some sort of name suffix in the libxl API which can be used by
e.g. xl to handle this stuff? Alternatively some sort of "realname"
field perhaps? Neither very nice, but both better than the layering
violation I think.

> > > +to xen-qemuuser-domid65535. To use this mechanism, you might want to
> > > +create a large number of users at installation time. For example:
> > > +
> > > +for i in '' $(seq 0 65335)
> > 
> > This creates a user with no $i on it too, which I think is a left over
> > from the previous scheme and no longer useful.
>  
> You are right. I'll turn it into the more bash-like:
> 
> for ((i=1; i<65336; i++))
> do

I just meant to remove the unnecessary '' (and adjust the starting point
for seq).

Ian.


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel