On Wed, 20 Nov 2024 14:26:27 GMT, Severin Gehwolf <[email protected]> wrote:
> Please review this simple patch which adds a new external system property
> `jdk.patched` when the runtime has been patched with the `--patch-module`
> switch. This is useful for two reasons: 1) it allows one to determine at
> run-time whether or not `--patch-module` has been used (by querying the said
> property) 2) allows tools, such as `jlink` doing the same without needing to
> expose internal properties the JVM sets on initialization.
>
> Testing:
>
> - [x] GHA (includes new test)
> - [x] New jtreg test
>
> Thoughts?
src/hotspot/share/runtime/arguments.cpp line 2272:
> 2270: if (res == JNI_OK) {
> 2271: // Add jdk.patched system property when processing of args was
> OK
> 2272: PropertyList_unique_add(&_system_properties, "jdk.patched",
> "true",
Hello Severin, in addition, should we explicitly set this system property to
`false` (or remove the system property) when `--patch-module` isn't used? I
think that would then prevent applications from passing `-Djdk.patched=true`
when the runtime hasn't been patched.
While at it, should there be additional guards to prevent applications from
changing the value of this system property by doing
`System.setProperty("jdk.patched")` at runtime?
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/22277#discussion_r1851376171
