Ensure flags stored to raw relo record is 16 bits

[dsouzai]

The _flags field of the TR_RelocationRecordBinaryTemplate struct is 16 bits wide. This was to ensure that the relocation infrastructure could store more than 4 bits of Relocation Flags (the other 4 bits are reserved for the Cross Platform Flags).

The code in TR_RelocationRecord::setReloFlags did some validation to ensure that the bits did not overlap; however, it did so using a uint8_t which could lose some bits due to a shift. This PR fixes this by ensuring a uint16_t is used.

This PR also changes how the flags are set; existing flags in the raw relocation record are now preserved.

[dsouzai]

@jdmpapin could you please review?

[jdmpapin]

existing flags in the raw relocation record are now preserved.

The main change in this regard seems to be that setReloFlags() used to preserve only the "cross-platform" flags, but now it preserves all of them. I think it would be good for the commit message to mention the reason for this part of the change, e.g. maybe callers have been (incorrectly) expecting it to work that way?

[dsouzai]

I think it would be good for the commit message to mention the reason for this part of the change, e.g. maybe callers have been (incorrectly) expecting it to work that way?

Yeah I can update the commit message. As far as I can tell, nothing really depends on the relo flags not being preserved; however, because setReloFlags and setFlags were not consistent, I figured it's better to just make them consistent just to strengthen the conceptual integrity.

[dsouzai]

Updated commit message.

[jdmpapin]

Jenkins test sanity.functional+aot all jdk17

[dsouzai]

Build failures due to #17527 (review)

[babsingh]

#17986 should fix the build errors. I will relaunch the PR builds once #17986 is merged.

[dsouzai]

Realized that this PR actually does fix a bug (introduced by #17621), namely that the needsFullSizeRuntimeAssumption and methodTracingEnabled flags do get lost in the shift; added the bug label.

[babsingh]

Relaunching PR builds since #17986 has been merged:

Jenkins test sanity.functional+aot all jdk17

[dsouzai]

x86 failure is due to an infra issue:

 Cannot contact ub16x64j91: java.lang.InterruptedException

aix failure is due to #17396:

[2023-08-19T01:20:06.457Z]  [ERR] Exception in thread "main" java.awt.AWTError: Can't connect to X11 window server using 'unix:0' as the value of the DISPLAY variable.
[2023-08-19T01:20:06.457Z]  [ERR] 	at java.desktop/sun.awt.X11GraphicsEnvironment.initDisplay(Native Method)
[2023-08-19T01:20:06.457Z]  [ERR] 	at java.desktop/sun.awt.X11GraphicsEnvironment$1.run(X11GraphicsEnvironment.java:105)

[jdmpapin]

Jenkins test sanity+aot xlinux,aix jdk17

[dsouzai]

Jenkins test sanity.functional+aot xlinux,aix jdk17

[dsouzai]

aix failure due to the same issue

[2023-08-21T22:57:02.933Z]  [OUT] fontmanger LIB: libfontmanager.so
[2023-08-21T22:57:02.933Z]  [ERR] Exception in thread "main" java.awt.AWTError: Can't connect to X11 window server using 'unix:0' as the value of the DISPLAY variable.
[2023-08-21T22:57:02.933Z]  [ERR] 	at java.desktop/sun.awt.X11GraphicsEnvironment.initDisplay(Native Method)

I've always had the suspicion, but now I'm starting to really think that issue is related to AOT and not infra. At any rate, this PR should be good for merging.