New test eliminating libvirt from the equation:
# Create one backing chain and one normal image
$ qemu-img create -f qcow2 /var/lib/libvirt/images/testdisk-nosnap.qcow2 100M
$ qemu-img create -f qcow2 /var/lib/libvirt/images/testdisk-snap-base.qcow2 100M
$ qemu-img create -f qcow2 -b /var/lib/libvirt/images/testdisk-snap-base.qcow2
/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2
# Use a profile similar to those created by libvirt - our guest doesn't do
much, so he doesn't need much
$ cat > /etc/apparmor.d/qemu-testprofile << EOF
# Test profile for bug LP: #1912214
#include <tunables/global>
profile qemu-testprofile flags=(attach_disconnected) {
#include <abstractions/libvirt-qemu>
#include <local/extrarules>
}
EOF
$ touch /etc/apparmor.d/local/extrarules
$ apparmor_parser -r /etc/apparmor.d/qemu-testprofile
# Start qemu into the monitor under control of an apparmor profile
$ aa-exec -p qemu-testprofile -- qemu-system-x86_64 -serial none -monitor stdio
-m 64 -display none
# Attach and detach the disk
(qemu) drive_add 0
if=none,file=/var/lib/libvirt/images/testdisk-nosnap.qcow2,format=qcow2,id=disk1
# This can be detached (e.g. to iterate the test) via
(qemu) drive_del disk1
As-is this fails (expected) since it gets the access to the file blocked:
[58147.954482] audit: type=1400 audit(1611040852.319:86): apparmor="DENIED"
operation="open" profile="qemu-testprofile"
name="/var/lib/libvirt/images/testdisk-nosnap.qcow2" pid=23222
comm="qemu-system-x86" requested_mask="r" denied_mask="r" fsuid=0 ouid=0
Now we can "manually play to be libvirt" and append the rule and reload the
profile.
In another console we can do:
$ echo '"/var/lib/libvirt/images/testdisk-nosnap.qcow2" rwk,' >
/etc/apparmor.d/local/extrarules
$ apparmor_parser -r /etc/apparmor.d/qemu-testprofile
Once that is done the `drive_add` command above will work.
Now if we allow the snapshot file but not the base image in apparmor.
And then attach the snapshot, we expect qemu to parse the snapshot, identify
the backing file and fail to access that.
$ echo '"/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2" rwk,' >
/etc/apparmor.d/local/extrarules
$ apparmor_parser -r /etc/apparmor.d/qemu-testprofile
(qemu) drive_add 0
if=none,file=/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2,format=qcow2,id=disk1
Could not open backing file: Could not open
'/var/lib/libvirt/images/testdisk-snap-base.qcow2': Permission denied
[58777.551226] audit: type=1400 audit(1611041481.906:96): apparmor="DENIED"
operation="open" profile="qemu-testprofile"
name="/var/lib/libvirt/images/testdisk-snap-base.qcow2" pid=23222
comm="qemu-system-x86" requested_mask="r" denied_mask="r" fsuid=0 ouid=0
Perfect - still as expected.
Now if we add both rules we expect this to work.
$ echo '"/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2" rwk,' >
/etc/apparmor.d/local/extrarules
$ echo '"/var/lib/libvirt/images/testdisk-snap-base.qcow2" rwk,' >>
/etc/apparmor.d/local/extrarules
$ apparmor_parser -r /etc/apparmor.d/qemu-testprofile
(qemu) drive_add 0
if=none,file=/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2,format=qcow2,id=disk1
And indeed it does work fine.
This also works fine in the containers ... good, but no help for debugging this.
** Description changed:
Hi,
to be clear I consider it quite likely that the error is on my side, but I'd
appreciate guidance how to continue resolve. I have rubber ducked my setup with
a coworker and talked to jjohansen if the profile would contain an obvious
fault (it did not).
So hereby I'm filing a bug hoping to get some help on this case.
## 1. What happened
I was trying to add a feature to libvirt to support chained qcow files.
That means instead of adding just one path libvirt has to process the chain
of backing files and add all of those. That works already on guest start, but
not on hot-add of devices.
But while I see my code appending the rules I'd expect and issuing the
apparmor_parser calls I'd expect it still fails.
I'd love to get the profile at runtime to ensure things really are loaded as
expected.
But as discussed on IRC that won't work. So I tested and simplified and
wonder for the test below why things fail.
## 2. how to reproduce
I was taking my new code out of the equation - and to do so I was adding
the path that is needed to the local overrides. But it still is blocked.
So - right now - I'm assuming that my code worked in adding the lines,
but the access is still blocked. Therefore let us try to fix this one
first and only then I can debug into what might be failing on the new
code.
# Allow the access that libvirt does not yet allow
$ echo '"/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2" rwk,' >
/etc/apparmor.d/local/abstractions/libvirt-qemu
# prep the disk chain to be hot-added
qemu-img create -f qcow2 /var/lib/libvirt/images/testdisk-snap-base.qcow2 100M
qemu-img create -f qcow2 -b /var/lib/libvirt/images/testdisk-snap-base.qcow2
/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2
cat > hot-add-test.xml << EOF
<disk type='file' device='disk'>
<driver name='qemu' type='qcow2'/>
- <source file='/var/lib/libvirt/images/testdisk-snap-base.qcow2'
index='1'/>
+ <source file='/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2'
index='1'/>
<backingStore type='file' index='2'>
<format type='qcow2'/>
- <source file='/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2'/>
+ <source file='/var/lib/libvirt/images/testdisk-snap-base.qcow2'/>
<backingStore/>
</backingStore>
<target dev='vdc' bus='virtio'/>
</disk>
EOF
# prep a guest to attach to
uvt-simplestreams-libvirt --verbose sync --source
http://cloud-images.ubuntu.com/daily arch=amd64 label=daily release=focal
uvt-kvm create --host-passthrough --password=ubuntu f-test2 release=focal
arch=amd64 label=daily
We have checked the "effective" profile that is loaded. JJohanssen didn't see
an obvious flaw in it. But for reference here:
root@h-libvirt-orig:~# apparmor_parser -p
/etc/apparmor.d/libvirt/libvirt-e033b910-be06-4975-826f-b6fba368d928 |
pastebinit
https://paste.ubuntu.com/p/yxdmMT6638/
# attach the disk
virsh attach-device f-test2 hot-add-test.xml
Libvirt/Qemu works as usual - except (known and expected) does not add a rule
for "/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2". But for that we
have added out local override.
Yet it is denied access.
apparmor="DENIED" operation="open" namespace="root//lxd-h-libvirt-new_
<var-snap-lxd-common-lxd>" profile="libvirt-a2caaf89-0682-464d-92ba-
5295cb5f5128" name="/var/lib/libvirt/images/testdisk-snap-
snapshot.qcow2" pid=2889044 comm="qemu-system-x86" requested_mask="r"
denied_mask="r" fsuid=64055 ouid=64055
## 3. The rule is generally working
To be clear, on the same system if I just open up a new profile and allow
this path to be accessed it works fine. See the working example below.
It must be somewhat that is tied to the way KVM guests get their profile
updates/assigned.
root@h-libvirt-orig:~# cat /etc/apparmor.d/test
#include <tunables/global>
profile test flags=(attach_disconnected) {
#include <abstractions/base>
"/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2" rwk,
"/usr/bin/md5sum" rmix,
}
root@h-libvirt-orig:~# apparmor_parser -r /etc/apparmor.d/test; aa-exec -v -p
test -- md5sum /var/lib/libvirt/images/testdisk-snap-snapshot.qcow2
[6939] aa_change_onexec("test")
[6939] exec md5sum /var/lib/libvirt/images/testdisk-snap-snapshot.qcow2
0f7fc62d270b69ee1b51453fa614d3e4
/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2
root@h-libvirt-orig:~# apparmor_parser -r /etc/apparmor.d/test; aa-exec -v -p
test -- md5sum /var/lib/libvirt/images/testdisk-snap-base.qcow2
[6944] aa_change_onexec("test")
[6944] exec md5sum /var/lib/libvirt/images/testdisk-snap-base.qcow2
md5sum: /var/lib/libvirt/images/testdisk-snap-base.qcow2: Permission denied
## 4. tracking apparmor_parser
I was using (for debug) a wrapper for apparmro_parser but all really
looks as expected
$ cat /usr/sbin/apparmor_parser.wrap
#!/bin/bash
echo "ARGS $@" | /usr/bin/systemd-cat
echo "Content of ${2}:" | /usr/bin/systemd-cat
/usr/bin/ls -laF "${2}" | /usr/bin/systemd-cat
/usr/bin/cat "${2}" | /usr/bin/systemd-cat
echo "Content of ${2}.files:" | /usr/bin/systemd-cat
/usr/bin/ls -laF "${2}.files" | /usr/bin/systemd-cat
/usr/bin/cat "${2}.files" | /usr/bin/systemd-cat
/usr/sbin/apparmor_parser.orig "$@" | /usr/bin/systemd-cat
It is called with "-r filepath" and the content of that filepath contain
the aforementioned rules. So what else might be wrong - why else could
the access be denied despite the rule that should allow it.
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to apparmor in Ubuntu.
https://bugs.launchpad.net/bugs/1912214
Title:
qemu can't access files that are added as rules on hot-add
Status in apparmor package in Ubuntu:
New
Bug description:
Hi,
to be clear I consider it quite likely that the error is on my side, but I'd
appreciate guidance how to continue resolve. I have rubber ducked my setup with
a coworker and talked to jjohansen if the profile would contain an obvious
fault (it did not).
So hereby I'm filing a bug hoping to get some help on this case.
## 1. What happened
I was trying to add a feature to libvirt to support chained qcow files.
That means instead of adding just one path libvirt has to process the chain
of backing files and add all of those. That works already on guest start, but
not on hot-add of devices.
But while I see my code appending the rules I'd expect and issuing the
apparmor_parser calls I'd expect it still fails.
I'd love to get the profile at runtime to ensure things really are loaded as
expected.
But as discussed on IRC that won't work. So I tested and simplified and
wonder for the test below why things fail.
## 2. how to reproduce
I was taking my new code out of the equation - and to do so I was
adding the path that is needed to the local overrides. But it still is
blocked. So - right now - I'm assuming that my code worked in adding
the lines, but the access is still blocked. Therefore let us try to
fix this one first and only then I can debug into what might be
failing on the new code.
# Allow the access that libvirt does not yet allow
$ echo '"/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2" rwk,' >
/etc/apparmor.d/local/abstractions/libvirt-qemu
# prep the disk chain to be hot-added
qemu-img create -f qcow2 /var/lib/libvirt/images/testdisk-snap-base.qcow2 100M
qemu-img create -f qcow2 -b /var/lib/libvirt/images/testdisk-snap-base.qcow2
/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2
cat > hot-add-test.xml << EOF
<disk type='file' device='disk'>
<driver name='qemu' type='qcow2'/>
<source file='/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2'
index='1'/>
<backingStore type='file' index='2'>
<format type='qcow2'/>
<source file='/var/lib/libvirt/images/testdisk-snap-base.qcow2'/>
<backingStore/>
</backingStore>
<target dev='vdc' bus='virtio'/>
</disk>
EOF
# prep a guest to attach to
uvt-simplestreams-libvirt --verbose sync --source
http://cloud-images.ubuntu.com/daily arch=amd64 label=daily release=focal
uvt-kvm create --host-passthrough --password=ubuntu f-test2 release=focal
arch=amd64 label=daily
We have checked the "effective" profile that is loaded. JJohanssen didn't see
an obvious flaw in it. But for reference here:
root@h-libvirt-orig:~# apparmor_parser -p
/etc/apparmor.d/libvirt/libvirt-e033b910-be06-4975-826f-b6fba368d928 |
pastebinit
https://paste.ubuntu.com/p/yxdmMT6638/
# attach the disk
virsh attach-device f-test2 hot-add-test.xml
Libvirt/Qemu works as usual - except (known and expected) does not add a rule
for "/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2". But for that we
have added out local override.
Yet it is denied access.
apparmor="DENIED" operation="open" namespace="root//lxd-h-libvirt-new_
<var-snap-lxd-common-lxd>" profile="libvirt-a2caaf89-0682-464d-92ba-
5295cb5f5128" name="/var/lib/libvirt/images/testdisk-snap-
snapshot.qcow2" pid=2889044 comm="qemu-system-x86" requested_mask="r"
denied_mask="r" fsuid=64055 ouid=64055
## 3. The rule is generally working
To be clear, on the same system if I just open up a new profile and allow
this path to be accessed it works fine. See the working example below.
It must be somewhat that is tied to the way KVM guests get their profile
updates/assigned.
root@h-libvirt-orig:~# cat /etc/apparmor.d/test
#include <tunables/global>
profile test flags=(attach_disconnected) {
#include <abstractions/base>
"/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2" rwk,
"/usr/bin/md5sum" rmix,
}
root@h-libvirt-orig:~# apparmor_parser -r /etc/apparmor.d/test; aa-exec -v -p
test -- md5sum /var/lib/libvirt/images/testdisk-snap-snapshot.qcow2
[6939] aa_change_onexec("test")
[6939] exec md5sum /var/lib/libvirt/images/testdisk-snap-snapshot.qcow2
0f7fc62d270b69ee1b51453fa614d3e4
/var/lib/libvirt/images/testdisk-snap-snapshot.qcow2
root@h-libvirt-orig:~# apparmor_parser -r /etc/apparmor.d/test; aa-exec -v -p
test -- md5sum /var/lib/libvirt/images/testdisk-snap-base.qcow2
[6944] aa_change_onexec("test")
[6944] exec md5sum /var/lib/libvirt/images/testdisk-snap-base.qcow2
md5sum: /var/lib/libvirt/images/testdisk-snap-base.qcow2: Permission denied
## 4. tracking apparmor_parser
I was using (for debug) a wrapper for apparmro_parser but all really
looks as expected
$ cat /usr/sbin/apparmor_parser.wrap
#!/bin/bash
echo "ARGS $@" | /usr/bin/systemd-cat
echo "Content of ${2}:" | /usr/bin/systemd-cat
/usr/bin/ls -laF "${2}" | /usr/bin/systemd-cat
/usr/bin/cat "${2}" | /usr/bin/systemd-cat
echo "Content of ${2}.files:" | /usr/bin/systemd-cat
/usr/bin/ls -laF "${2}.files" | /usr/bin/systemd-cat
/usr/bin/cat "${2}.files" | /usr/bin/systemd-cat
/usr/sbin/apparmor_parser.orig "$@" | /usr/bin/systemd-cat
It is called with "-r filepath" and the content of that filepath
contain the aforementioned rules. So what else might be wrong - why
else could the access be denied despite the rule that should allow it.
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/apparmor/+bug/1912214/+subscriptions
--
Mailing list: https://launchpad.net/~touch-packages
Post to : touch-packages@lists.launchpad.net
Unsubscribe : https://launchpad.net/~touch-packages
More help : https://help.launchpad.net/ListHelp
