Fix build breaks with gcc 11.3.1

[dsouzai]

• Fix documentation to ensure OMR_GC_MODRON_SCAVENGER is enabled during the build
• Add -O3 to omrtrace component to prevent warning about not compiling with -O when _FORTIFY_SOURCE is specified.

[dsouzai]

@keithc-ca @babsingh could you please review?

[babsingh]

We should be able to build OMR without OMR_GC_MODRON_SCAVENGER. It should not be a mandatory requirement. So, the below error should be fixed by either consistently using the same ifdef macros or enabling dependent ifdef macros correctly. MarkingScheme.hpp:292 uses OMR_GC_CONCURRENT_SCAVENGER whereas GCExtensionsBase.hpp uses OMR_GC_MODRON_SCAVENGER for isScavengerBackOutFlagRaised. fyi @dmitripivkine @amicic @LinHu2016

/home/harpreet/omr/gc/base/MarkingScheme.hpp:292:67: error: 'class MM_GCExtensionsBase' has no member named 'isScavengerBackOutFlagRaised'; did you mean 'isScavengerEnabled'?
   if (_extensions->isConcurrentScavengerEnabled() && _extensions->isScavengerBackOutFlagRaised()) {
                                                                   ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
                                                                   isScavengerEnabled

[dmitripivkine]

We should be able to build OMR without OMR_GC_MODRON_SCAVENGER. It should not be a mandatory requirement. So, the below error should be fixed by either consistently using the same ifdef macros or enabling dependent ifdef macros correctly. MarkingScheme.hpp:292 uses OMR_GC_CONCURRENT_SCAVENGER whereas GCExtensionsBase.hpp uses OMR_GC_MODRON_SCAVENGER for isScavengerBackOutFlagRaised. fyi @dmitripivkine @amicic @LinHu2016

/home/harpreet/omr/gc/base/MarkingScheme.hpp:292:67: error: 'class MM_GCExtensionsBase' has no member named 'isScavengerBackOutFlagRaised'; did you mean 'isScavengerEnabled'?
   if (_extensions->isConcurrentScavengerEnabled() && _extensions->isScavengerBackOutFlagRaised()) {
                                                                   ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
                                                                   isScavengerEnabled

Yes, if should be fixed eventually. However I have suspicion the condition check was missed at time of split code for OMR many years ago. I think there are more missed checks like this in the code (not only Scavenger but another GC features related). So OMR code requires inspection/cleaning project. Also when it completes it required testing support - compiling code with disabled features on regular basis. Otherwise new errors are going to be introduced unavoidably. Currently we have no resources to address this, so, this is another issue for backlog

[babsingh]

LGTM. GC fix is a good temporary workaround. -O3 is only set when _FORTIFY_SOURCE is enabled on Linux. @keithc-ca Do these change look good to you?

[babsingh]

jenkins build all

To make sure, the -O3 change works fine on all Linux platforms.

[babsingh]

https://ci.eclipse.org/omr/job/PullRequest-win_x86-64/3469/console

@AdamBrousseau Noticed a new infra issue on Windows, which is unrelated to this PR. Is this a known infra issue?

[AdamBrousseau]

jenkins build win

[AdamBrousseau]

We were testing a new machine. Back to normal now.

[dsouzai]

riscv build failed because of:

11:23:11  20: [----------] 18 tests from RWMutex
13:05:04  Cancelling nested steps due to timeout
13:05:04  Sending interrupt signal to process
13:05:24  After 20s process did not stop

It's not the first time there's been a hang in RWMutex (see #277 (comment)) though that was all the way in 2016 on OSX, so kind of a stretch to link these two.

Will run the test again to see if it's reproducable.

[dsouzai]

jenkins build riscv64

[dsouzai]

Don't see the RWMutex hang this time around; riscv64 failure now because of #6704

[babsingh]

The failures seen in the PR builds are unrelated to the changes in this PR: #6704 and #6706. Thus, merging.