8298971: Move Console implementation into jdk internal package

[fitzsim]

I have been tasked with fixing:

8354469: Keytool exposes the password in plain text when command is piped using | grep

for 17, 11 and 8, for parity with Oracle's JDK releases.

The affected java.io.Console code has been significantly reorganized in 21 and newer versions, to allow advanced features like different console providers (via the new jdk.console system property) and to support System.console().readPassword() in jshell. The reorganization makes use of class sealing, which is available in 17 but not 11 or 8.

This pull request is my attempt to lay the foundation to be able to smoothly backport JdkConsoleImpl changes to 17, 11 and 8.

I have applied parts of the following fixes:

8298971: Move Console implementation into jdk internal package
8298416: Console should be declared `sealed`
8295803: Console should be usable in jshell and other environments

and I have backported:

8300010: UnsatisfiedLinkError on calling System.console().readPassword() on Windows

so as not to break Windows.

I have tried to make each commit on this branch as small as possible to clearly show what is changing.

Assuming this is applied, the subsequent backports required to fix JDK-8354469:

8366261: Provide utility methods for sun.security.util.Password
8354469: Keytool exposes the password in plain text when command is piped using | grep

apply readily. I have created a branch to show this:

https://github.com/fitzsim/jdk17u-dev/commits/delegate-console-and-backport-keytool-8354469/

I confirmed that on that branch, keytool no longer exposes the password when standard output is redirected.

I did some multi-threaded testing of System.console().format() and System.console().readPassword() to verify the locking in the merge of ProxyConsole.java into Console.java seems correct.

CI testing shows no regressions, though System.console() does not seem to have much tier1 test coverage.

On my Fedora 43 x86-64 machine, make test TEST=all (with the changes rebased to 36f3b1ae7f2cb0acc6409e2b686757756c2fea50) shows roughly the same results before and after the patch.

Before:

==============================
Test summary
==============================
   TEST                                              TOTAL  PASS  FAIL ERROR   
>> jtreg:test/hotspot/jtreg:all                       6282  6186    93     3 <<
>> jtreg:test/jdk:all                                11083 10200   836    47 <<
   jtreg:test/langtools:all                           4257  4257     0     0   
   jtreg:test/jaxp:all                                 467   467     0     0   
   jtreg:test/lib-test:all                              34    34     0     0   
==============================
TEST FAILURE

After:

==============================
Test summary
==============================
   TEST                                              TOTAL  PASS  FAIL ERROR   
>> jtreg:test/hotspot/jtreg:all                       6282  6186    93     3 <<
>> jtreg:test/jdk:all                                11083 10204   829    50 <<
   jtreg:test/langtools:all                           4257  4257     0     0   
   jtreg:test/jaxp:all                                 467   467     0     0   
   jtreg:test/lib-test:all                              34    34     0     0   
==============================
TEST FAILURE

One difference was sun/security/ssl/X509KeyManager/PreferredKey.java which passed before and failed after. The cause was the dummy certificate in ./test/jdk/javax/net/ssl/etc/keystore expiring over the weekend:

# keytool -list -v -keystore ./test/jdk/javax/net/ssl/etc/keystore -storepass passphrase
[...]
Alias name: dummy
[...]
Valid from: Mon May 16 06:06:38 EDT 2016 until: Sat May 16 06:06:38 EDT 2026
[...]

I happened to run the test suite unpatched before the expiration and patched after the expiration. The same test fails for the same reason on jdk21u-dev.

Refreshing the dummy certificate causes the test case to pass again:

keytool -delete -alias dummy -keystore ./test/jdk/javax/net/ssl/etc/keystore -storepass passphrase

keytool -genkeypair -alias dummy -keyalg RSA -keysize 2048 \
  -sigalg SHA256withRSA \
  -dname "CN=dummy.example.com, OU=Dummy, O=Dummy, L=Cupertino, ST=CA, C=US" \
  -validity 3652 -keypass passphrase -keystore ./test/jdk/javax/net/ssl/etc/keystore -storepass passphrase

however ./test/jdk/javax/net/ssl/etc/README mentions that it should be "Version 1" generated by a hacked keytool command, so this is just a workaround. The real fix is to backport JDK-8384815.

The rest of the jtreg results differences were in java/awt and javax/swing, unrelated to System.console().

---

• I confirm that I make this contribution in accordance with the OpenJDK Interim AI Policy.

---

Progress

• Change must be properly reviewed (1 review required, with at least 1 Reviewer)
• Change must not contain extraneous whitespace
• Commit message must refer to an issue
• JDK-8298971 needs maintainer approval

Issue

• JDK-8298971: Move Console implementation into jdk internal package (Bug - P4)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk17u-dev.git pull/4452/head:pull/4452
$ git checkout pull/4452

Update a local copy of the PR:
$ git checkout pull/4452
$ git pull https://git.openjdk.org/jdk17u-dev.git pull/4452/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 4452

View PR using the GUI difftool:
$ git pr show -t 4452

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk17u-dev/pull/4452.diff

Using Webrev

Link to Webrev Comment

[bridgekeeper]

👋 Welcome back fitzsim! A progress list of the required criteria for merging this PR into master will be added to the body of your pull request. There are additional pull request commands available for use with this pull request.

[openjdk]

❗ This change is not yet ready to be integrated.
See the Progress checklist in the description for automated requirements.

[openjdk]

This backport pull request has now been updated with issue from the original commit.

[mlbridge]

Webrevs

• 00: Full (b2399256)

[GoeLin]

Hi @fitzsim
thanks for sharing this! The problem is that the backported changes modify the behaviour of the JDK. They have CSRs and release notes.
Also, it's a lot of code bearing the danger to introduce new issues in 17, which then are only in open 17. The original changes introduced a row of problems already. I'll have a closer look and share what I did so far.

[fitzsim]

Hi @GoeLin , thanks for taking a look at this.

I picked 8298971 only because the title conceptually most closely described what this set of changes is doing. I debated instead just making this a new bug report/pull request. I did try to make the actual changes not introduce any CSRs.

I agree with the amount of code changes being worrying, and 17 (and subsequently 11 and 8) only. I tried to make clear with each commit that the resulting Console.java contents match very closely the contents of ProxyingConsole.java from 21 and onward, for what it's worth.

I am certainly interested in seeing your suggested approach.

[GoeLin]

Hi, you find my backport draft of 8366261 at #4482