Catangiu, Adrian Costin
2021-04-09 20:20:39 UTC
This RFC is a continuation of a longer kernel patch thread
https://lkml.org/lkml/2021/3/8/677 where we originally thought such
a mechanism belongs. Ultimately, consensus there was that this mechanism
would be better suited in userspace, so systemd was an obvious first choice.
* As GitHub Issue here: https://github.com/systemd/systemd/issues/19269
* An example PoC here: https://github.com/acatangiu/sysgenid-dbus
* Described in this email as follows:
# SysGenID: a system generation id provider
## Background and problem
The System Generation ID feature is required in virtualized or
containerized environments by applications that work with local copies
or caches of world-unique data such as random values, uuids,
monotonically increasing counters, cryptographic nonces, etc.
Such applications can be negatively affected by VM or container
snapshotting when the VM or container is either cloned or returned to
an earlier point in time.
Solving the uniqueness problem strongly enough for cryptographic
purposes requires a mechanism which can deterministically reseed
userspace PRNGs with new entropy at restore time. This mechanism must
also support the high-throughput and low-latency use-cases that led
programmers to pick a userspace PRNG in the first place; be usable by
both application code and libraries; allow transparent retrofitting
behind existing popular PRNG interfaces without changing application
code; it must be efficient, especially on snapshot restore; and be
simple enough for wide adoption.
Introduce a mechanism that standardizes an API for
applications and libraries to be made aware of uniqueness breaking
events such as VM or container snapshotting, and allow them to react
and adapt to such events.
The System Generation ID is meant to help in these scenarios by
providing a monotonically increasing u32 counter that changes each time
the VM or container is restored from a snapshot.
The `sysgenid` service exposes a monotonic incremental System Generation
u32 counter via the DBus `com.RFC.sysgenid` accessible at
`/com/RFC/sysgenid`. It provides asynchronous SysGen
counter update notifications, as well as counter retrieval and
The counter starts from zero when the service is started and
monotonically increments every time the system generation changes.
Userspace applications or libraries can (a)synchronously consume the
system generation counter through the provided DBus interface, to
make any necessary internal adjustments following a system generation
The provided DBus interface operations can be used to build a
system level safe workflow that guest software can follow to protect
itself from negative system snapshot effects.
System generation changes are driven by userspace software through a
dedicated DBus method.
SysGenID alone does not guarantee complete snapshot
safety to applications using it. A certain workflow needs to be
followed at the system level, in order to make the system
snapshot-resilient. Please see the "Snapshot Safety Prerequisites"
## SysGenID DBus interface
- `watcher` - a client using the SysGenID service _watching_ for system generation changes.
- `untracked watcher` - default state for all clients. For a client to be tracked it has
to explicitly opt-in by confirming back to the service the correct _system generation
- `tracked watcher` - a client that is tracked by the service. Such a watcher is considered
`up-to-date` only after confirming back to the service the correct
_system generation counter_.
Once tracked, a client is only _untracked_ when closing its connection to the DBus bus.
- `outdated watcher` - a _tracked_ client that whose tracking has lived through a system
generation change, but has not (yet) confirmed back to the service the correct _system
- `GetSysGenCounter` - returns latest system generation counter.
- `AckWatcherCounter` - marks the client/watcher to be tracked for ACKs, is also
used by the watcher to confirm/ack the correct _sys gen counter_ to the service after
every generation change so the service keeps correct track of it as `outdated` or
Will error if client/watcher confirms/acks the wrong _sys gen counter_.
- `CountOutdatedWatchers` - returns the number of current number of
_outdated tracked watchers_.
A value of `zero` can be interpreted as the system being fully re-adjusted after a
- `TriggerSysGenUpdate` - triggers a generation update (should be a privileged operation).
- `NewSystemGeneration` - system generation change notification, also carries new
_sys gen counter_.
- `SystemReady` - notification sent out when all tracked watchers have _acked_ the new
_sys gen counter_. In other words, when all tracked software has adjusted to the new
The service can keep track of watchers by DBus connections
**Exported read-only file used for memory mappings:**
The service also exports the current _sys gen counter_ through a simple file.
The file contains only 4 bytes of data at offset 0, representing the u32 value
of the system generation counter.
This file is meant to be mapped by other software in the system and be used as
a low-latency generation counter probe mechanism in critical sections.
This mmap() interface is targeted at libraries or code that needs to
check for generation changes in-line, where an event loop is not
available or in cases where DBus calls are too expensive.
In such cases, logic can be added in-line with the sensitive code to check the
counter and trigger on-demand/just-in-time readjustments when changes are
detected on the memory mapped file.
Users of this interface that plan to lazily adjust most likely don't need to
also use the DBus interface, since tracking or waiting on them doesn't make sense.
### Service interface DBus XML specification
<arg name="watcher_counter" type="u" direction="in"/>
<arg name="sysgen_counter" type="u" direction="out"/>
<arg name="outdated_watchers" type="u" direction="out"/>
<arg name="sysgen_counter" type="u" direction="out"/>
<arg name="min_gen" type="u" direction="in"/>
<arg name="sysgen_counter" type="u"/>
<arg name="xml_data" type="s" direction="out"/>
## Snapshot Safety Prerequisites and Example
If VM, container or other system-level snapshots happen asynchronously,
at arbitrary times during an active workload there is no practical way
to ensure that in-flight local copies or caches of world-unique data
such as random values, secrets, UUIDs, etc are properly scrubbed and
The challenge stems from the fact that the categorization of data as
snapshot-sensitive is only known to the software working with it, and
this software has no logical control over the moment in time when an
external system snapshot occurs.
Let's take an OpenSSL session token for example. Even if the library
code is made 100% snapshot-safe, meaning the library guarantees that
the session token is unique (any snapshot that happened during the
library call did not duplicate or leak the token), the token is still
vulnerable to snapshot events while it transits the various layers of
the library caller, then the various layers of the OS before leaving
To catch a secret while it's in-flight, we'd have to validate system
generation at every layer, every step of the way. Even if that would
be deemed the right solution, it would be a long road and a whole
universe to patch before we get there.
Bottom line is we don't have a way to track all of these in-flight
secrets and dynamically scrub them from existence with snapshot
events happening arbitrarily.
### Simplifying assumption - safety prerequisite
**Control the snapshot flow**, disallow snapshots coming at arbitrary
moments in the workload lifetime.
Use a system-level overseer entity that quiesces the system before
snapshot, and post-snapshot-resume oversees that software components
have readjusted to new environment, to the new generation. Only after,
will the overseer un-quiesce the system and allow active workloads.
Software components can choose whether they want to be tracked and
waited on by the overseer by using the marking themselves as tracked
The sysgenid service standardizes the API for system software to
find out about needing to readjust and at the same time provides a
mechanism for the overseer entity to wait for everyone to be done, the
system to have readjusted, so it can un-quiesce.
### Example snapshot-safe workflow
1) Before taking a snapshot, quiesce the VM/container/system. Exactly
how this is achieved is very workload-specific, but the general
description is to get all software to an expected state where their
event loops dry up and they are effectively quiesced.
2) Take snapshot.
3) Resume the VM/container/system from said snapshot.
4) Overseer will trigger generation bump using
5) Software components which have the DBus `NewGeneration` signal in
their event loops are notified of the generation change.
They do their specific internal adjustments. Some may have chosen to
be tracked and waited on by the overseer, others might choose to do
their adjustments out of band and not block the overseer.
Tracked ones *must* signal when they are done/ready by confirming the
new sys gen counter using the `AckWatcherCounter` DBus method.
6) Overseer will block and wait for all tracked watchers by waiting on
the `SystemReady` DBus signal. Once all tracked watchers are done
in step 5, the signal is sent by `sysgenid` service and overseer will
know that the system has readjusted and is ready for active workload.
7) Overseer un-quiesces system.
8) There is a class of software, usually libraries, most notably PRNGs
or SSLs, that don't fit the event-loop model and also have strict
latency requirements. These can take advantage of the
_exported read-only file used for memory mappings_. They can map the
file and check sys gen counter value in-line with the critical section
and can do so with low latency. When they are called after un-quiesce,
they can just-in-time adjust based on the updated mapped value.
For a well-designed service stack, these libraries should not be
called while system is quiesced. When workload is resumed by the
overseer, on the first call into these libs, they will safely JIT
Users of this lazy on-demand readjustment model should not use the
DBus interface or at least not enable watcher tracking since doing so
would introduce a logical deadlock:
lazy adjustments happen only after un-quiesce, but un-quiesce is
blocked until all tracked watchers are up-to-date.
Amazon Development Center (Romania) S.R.L. registered office: 27A Sf. Lazar Street, UBC5, floor 2, Iasi, Iasi County, 700045, Romania. Registered in Romania. Registration number J22/2621/2005.