Add reencryption by frederic-hoerni · Pull Request #569 · canonical/secboot

frederic-hoerni · 2026-09-08T16:35:58Z

This adds the feature of reencryption of active LUKS2 containers.

Main changes:

API functions:
- ReencryptionForActiveVolume
- Initialize
- Resume
- Status
Named tokens now may have 2 keyslots, which happens during reencryption.
OpenRead sets a shared lock on the LUKS2 container so that reencryption cannot happen while a StorageContainerReader is open.
Introduction of secboot-tool, that is a tool to test and illustrates high-level functions of secboot (mostly reencryption at the moment).

tlaurion · 2026-09-17T13:35:11Z

Would be awesome if this go cryptsetup api was in its own repository. u-root and other projects would benefit of this as well

frederic-hoerni · 2026-09-23T07:10:34Z

Would be awesome if this go cryptsetup api was in its own repository

@tlaurion, we have no plan for that at the moment.

frederic-hoerni · 2026-09-25T15:15:56Z

Please rebase on top of master (and expect conflicts).

ruifm

Looking great!

ruifm

LGTM

This adds support for reencrypting active LUKS2 containers. Main changes: - API functions: * ReencryptionForActiveVolume * Initialize * Resume * Status - Named LUKS2 tokens now may have 2 keyslots, which happens during reencryption.

OpenRead now sets a shared lock on the LUKS2 container so that reencryption cannot happen while a StorageContainerReader is open.

valentindavid · 2026-10-06T11:07:26Z

+)
+
+const UsageActivate = `
+usage: secboot-tool activate [<options>] <device> <active-name> <unlock-key-hex>


Is this a debugging tool? Is there a plan to ship that?

I do not think we want any key in command line. Unless everything was running in pid namespaces, this is public information.

This is a debugging tool. We do not plan to ship it for now.

valentindavid

Very quick pass. Just some questions and remark for now.

valentindavid · 2026-10-06T11:17:41Z

 			"--token-replace", "/dev/null"); err == nil {
 			features |= FeatureTokenReplace
 		}
+		if _, err := cryptsetupCmd(nil, "--test-args", "reencrypt", "--keys-from-stdin-sizes", "1,2",


Just a note while reviewing. The feature is https://gitlab.com/cryptsetup/cryptsetup/-/merge_requests/958

I suppose we should wait for this to land upstream before we ship that in snapd.

I think it does not prevent merging this PR.
If cryptsetup is lacking the feature, then the following code:

reencryption, err := secboot.ReencryptionForActiveVolume(activeName)

will return this error: luks2 backend is missing the reencryption feature.

valentindavid · 2026-10-06T11:20:37Z

+		"--key-file", "-",
+		"--batch-mode",
+		"--resume-only",
+		"--progress-frequency", "1",


I wonder if it is something we will want to configure in the future.

I don't think there is a need for a higher frequency. If we later find a need for a lower frequency we will discuss it then.

It is also worth noting that this parameter is somewhat approximate. With a setting of 1 second, the actually observed period between 2 reports is between 1 and 3 seconds.

valentindavid · 2026-10-06T11:21:53Z

+}
+
+// ReadCryptsetupStatus returns the cryptsetup status of an active dm volume
+func ReadCryptsetupStatus(activeName string) (*CryptsetupStatus, error) {


I am a bit afraid about parsing that. it feels like it might change in the future.

Should we add a json output upstream?

cryptsetup status has a --debug-json option that does not produce any json. Attempts to upstream might be unpredictable.

Besides, don't we have integration tests that would detect incompatible changes?

valentindavid · 2026-10-06T11:30:49Z

+
+// KeyslotNamesSortedById returns the token names sorted
+// by their keyslot identifier (in ascending order)
+// Assumption: there is only one keyslot by token


What is the reason to sort them? I wonder if we could use something different that is a bit more safe without this assumption.

This is based on how cryptsetup is organized. They look at all possible slots and unlock the active ones in this order.

valentindavid · 2026-10-06T11:32:17Z

+		case "online-reencrypt-v3":
+			return true
+		default:
+			continue


What if it is online-reencrypt-v4? We return false here.

Ok we can accept these as well. v4 is not used at the moment, but I found that cryptsetup has a test that considers online-reencrypt-v999 correct.

valentindavid · 2026-10-06T11:35:16Z

+
+func Logf(level int, format string, v ...any) {
+	if level <= logLevelPolicy {
+		fmt.Fprintf(os.Stderr, format, v...)


I think we need to be able to initialized the output from snapd.

I added SetOuput.

I meant "SetOutput":

func SetOutput(newOutput io.Writer)` { ... }

valentindavid · 2026-10-06T11:37:18Z

+	if len(i.keyslotIds) == 1 {
+		return i.keyslotIds[0]
+	} else {
+		return luks2.AnySlot


Not sure if this the right return for len(i.keyslotIds) > 1. Do we verify somewhere else that we can have only 0 or 1?

We can have 2 keyslots here. In that case, we cannot decide to choose one over the other. So we return Any.

I added a description with this in mind.

valentindavid · 2026-10-06T11:45:34Z

+	streamLines := func(pipe io.Reader, outputDone chan<- struct{}) {
+		scanner := bufio.NewScanner(pipe)
+		for scanner.Scan() {
+			rawBytes := scanner.Bytes() // should be in JSON format


So this is newline delimited json sequence? Go's json does not support json sequence for rfc7464?

It is a newline delimited json sequence. But not rfc7464 as there is no ASCII Record Separator (0x1E).
Moreover we may also catch non-json errors here.

This lets the caller to change the default stderr log output.

frederic-hoerni force-pushed the feature/luks-reencryption branch 2 times, most recently from 3815939 to 731117d Compare September 17, 2026 08:54

frederic-hoerni changed the title ~~Add reencryption API (MVP)~~ Add reencryption Sep 17, 2026

frederic-hoerni force-pushed the feature/luks-reencryption branch from 0ba1781 to d6ced40 Compare September 22, 2026 16:25

frederic-hoerni force-pushed the feature/luks-reencryption branch from 81a124a to a95d46c Compare September 23, 2026 08:32

frederic-hoerni marked this pull request as ready for review September 24, 2026 06:34

frederic-hoerni requested a review from valentindavid September 24, 2026 06:34

frederic-hoerni self-assigned this Sep 24, 2026

frederic-hoerni requested a review from pedronis September 24, 2026 06:35

frederic-hoerni commented Sep 24, 2026

View reviewed changes

Comment thread internal/luksview/tokens_test.go

Comment thread internal/luks2/cryptsetup.go Outdated

Comment thread luks2/backend.go

frederic-hoerni commented Sep 24, 2026

View reviewed changes

Comment thread keydata_luks.go Outdated

ruifm reviewed Sep 25, 2026

View reviewed changes

Comment thread internal/luks2/cryptsetup.go Outdated

Comment thread luks2/reencrypt.go

Comment thread reencrypt.go

Comment thread luks2/luks2_test.go

Comment thread luks2/reencrypt.go

Comment thread luks2/reencrypt.go

frederic-hoerni commented Sep 28, 2026

View reviewed changes

Comment thread luks2/reader.go Outdated

frederic-hoerni commented Sep 28, 2026

View reviewed changes

Comment thread internal/luks2/cryptsetup.go

frederic-hoerni force-pushed the feature/luks-reencryption branch from 5f920db to d95f727 Compare September 30, 2026 07:22

ruifm approved these changes Sep 30, 2026

View reviewed changes

frederic-hoerni force-pushed the feature/luks-reencryption branch from f4029ad to 3176726 Compare September 30, 2026 12:44

log: add minimal logging framework

d53339d

frederic-hoerni force-pushed the feature/luks-reencryption branch from 3176726 to 41dcfa7 Compare September 30, 2026 13:58

frederic-hoerni added 6 commits October 1, 2026 11:32

cmd: add reencrypt and secboot-tool

60a712b

Makefile: handle <pkg> targets

86e6c9b

Mark deprecated ActivateVolume*, DeactivateVolume

0b5a786

Add reencryption

455fbd0

This adds support for reencrypting active LUKS2 containers. Main changes: - API functions: * ReencryptionForActiveVolume * Initialize * Resume * Status - Named LUKS2 tokens now may have 2 keyslots, which happens during reencryption.

Add README-reencrypt

54581b7

luks2: acquire a shared lock in OpenRead

6bff423

OpenRead now sets a shared lock on the LUKS2 container so that reencryption cannot happen while a StorageContainerReader is open.

frederic-hoerni force-pushed the feature/luks-reencryption branch from 97850e3 to 6bff423 Compare October 1, 2026 09:47

valentindavid reviewed Oct 6, 2026

View reviewed changes

frederic-hoerni added 3 commits October 8, 2026 15:13

log: add SetOutput

8be967f

This lets the caller to change the default stderr log output.

luksview: more flexibility in reencryption detection

93d9565

luks2: add description for PreferredKeyslotId

a94f97e

Conversation

frederic-hoerni commented Sep 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tlaurion commented Sep 17, 2026

Uh oh!

frederic-hoerni commented Sep 23, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

frederic-hoerni commented Sep 25, 2026

Uh oh!

ruifm left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ruifm left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

valentindavid left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

frederic-hoerni Oct 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

frederic-hoerni Oct 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

frederic-hoerni commented Sep 8, 2026 •

edited

Loading

frederic-hoerni Oct 8, 2026 •

edited

Loading

frederic-hoerni Oct 8, 2026 •

edited

Loading