This PR adds a new `MAX_COMPILE_THREADS` gradle property to to limit the number
of build-time parallel compilation threads to the smaller of this property and
the number of avialable processors.
Our gradle build spawns up to `NUM_COMPILE_THREADS` compilation tasks in
parallel when building native libraries. By default, `NUM_COMPILE_THREADS` is
set to the value returned by `Runtime.getRuntime().availableProcessors()`. That
value can be overridden on the command line by setting the
`NUM_COMPILE_THREADS` gradle property. We need a way to clamp
`NUM_COMPILE_THREADS` at some maximum value without the caller having to
compute it.
The change in `build.gradle` is simple:
* Define a new `MAX_COMPILE_THREADS` gradle property, which will default to 0
(meaning unlimited) for compatibility with current behavior.
* Set the default value of `NUM_COMPILE_THREADS` as follows:
NUM_COMPILE_THREADS = MAX_COMPILE_THREADS > 0
? Math.min(MAX_COMPILE_THREADS, availableProcessors)
: availableProcessors(
This will allow a test scripts that runs on a number of different platforms to
clamp to a maximum value without having to use jshell or similar to implement
`Math.min(MAX_COMPILE_THREADS, Runtime.getRuntime().availableProcessors())` in
the script.
We have a practical need for this in our headful test systems, several of which
report a higher number of processors than they can effectively support. Without
clamping, the build will either run slower (due to too much swapping) or fail
altogether when building WebKit.
-------------
Commit messages:
- 8374153: Add a MAX_COMPILE_THREADS gradle property to limit number of threads
Changes: https://git.openjdk.org/jfx/pull/2014/files
Webrev: https://webrevs.openjdk.org/?repo=jfx&pr=2014&range=00
Issue: https://bugs.openjdk.org/browse/JDK-8374153
Stats: 23 lines in 2 files changed: 20 ins; 0 del; 3 mod
Patch: https://git.openjdk.org/jfx/pull/2014.diff
Fetch: git fetch https://git.openjdk.org/jfx.git pull/2014/head:pull/2014
PR: https://git.openjdk.org/jfx/pull/2014
