Remove must-init requirement for GS Cookies

[SingleAccretion]

The cookie is always initialized in the prolog, where no user code is allowed.
Likewise, there should be no GC concerns with it as the prolog is not interruptible.

This change should be validated with some GCStess'ed runs.

The diffs look generally positive as expected, the regressions are because of the differences in the zero-initting strategies (e. g. here it looks like a switch from a longer loop to a shorter loop + two explicit stores).

libraries.pmi.windows.x64.checked.mch

Summary of Code Size diffs:
(Lower is better)

Total bytes of base: 67824
Total bytes of diff: 67374
Total bytes of delta: -450 (-0.66% of base)
    diff is an improvement.

Top file regressions (bytes):
          16 : 212958.dasm (0.65% of base)
           2 : 175944.dasm (0.96% of base)
           2 : 175948.dasm (0.95% of base)
           2 : 175945.dasm (0.95% of base)
           2 : 175947.dasm (0.95% of base)
           2 : 175946.dasm (0.95% of base)
           1 : 203366.dasm (0.57% of base)
           1 : 203361.dasm (0.56% of base)
           1 : 203363.dasm (0.57% of base)
           1 : 203364.dasm (0.53% of base)
           1 : 212535.dasm (0.17% of base)
           1 : 203362.dasm (0.56% of base)
           1 : 83342.dasm (0.17% of base)

Top file improvements (bytes):
         -20 : 212527.dasm (-2.77% of base)
         -19 : 212937.dasm (-2.78% of base)
         -13 : 175648.dasm (-0.98% of base)
         -10 : 203370.dasm (-1.12% of base)
         -10 : 232447.dasm (-2.39% of base)
         -10 : 83292.dasm (-1.22% of base)
         -10 : 232445.dasm (-2.92% of base)
          -9 : 232446.dasm (-2.43% of base)
          -9 : 232448.dasm (-1.94% of base)
          -9 : 203369.dasm (-1.01% of base)
          -8 : 176098.dasm (-1.66% of base)
          -8 : 176067.dasm (-1.65% of base)
          -8 : 212524.dasm (-0.28% of base)
          -8 : 145624.dasm (-1.15% of base)
          -8 : 145625.dasm (-1.15% of base)
          -7 : 212516.dasm (-0.60% of base)
          -7 : 213015.dasm (-0.56% of base)
          -7 : 178539.dasm (-3.52% of base)
          -7 : 213080.dasm (-0.37% of base)
          -6 : 83285.dasm (-2.17% of base)

cc @AndyAyersMS

[EgorBo]

/azp list

[azure-pipelines]

CI/CD Pipelines for this repository:

• coreclr-gc-longrunning
• coreclr-gc-simulator
• runtime-coreclr outerloop
• runtime-coreclr jitstress
• runtime-coreclr jitstressregs
• runtime-coreclr jitstress2-jitstressregs
• runtime-coreclr r2r-extra
• runtime-coreclr jitstress-isas-x86
• runtime-coreclr jitstress-isas-arm
• runtime-coreclr jitstressregs-x86
• runtime-coreclr libraries-jitstressregs
• runtime-coreclr libraries-jitstress2-jitstressregs
• runtime-coreclr r2r
• runtime-coreclr runincontext
• runtime-coreclr crossgen2
• runtime-libraries-coreclr outerloop
• runtime-libraries-coreclr outerloop-windows
• runtime-libraries-coreclr outerloop-linux
• runtime-libraries-coreclr outerloop-osx
• runtime
• runtime-libraries enterprise-linux
• runtime-libraries stress-http
• runtime-libraries stress-ssl
• runtime-dev-innerloop
• runtime-coreclr crossgen2 outerloop
• coreclr-release-outerloop-nightly
• sync-runtime-to-mono
• runtime-coreclr crossgen2-composite
• runtime-jit-experimental
• runtime-coreclr libraries-jitstress
• dotnet-linker-tests
• runtime-coreclr ilasm
• runtime-coreclr clrinterpreter
• runtime-coreclr crossgen2-composite gcstress
• runtime-libraries-mono outerloop
• runtime-staging

[EgorBo]

/azp run runtime-coreclr jitstress

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[AndyAyersMS]

We wanted to run GC stress on this.

Is it not runnable via /azp ? Anyways, started it via the UI.

[AndyAyersMS]

Stress failures all look to be unrelated.

[AndyAyersMS]

Unrelated CI issues were affecting this; going to retry the "runtime" parts.

[AndyAyersMS]

Still puzzled by what CI is doing here. Am going to close/re-open to get a fresh look...

[SingleAccretion]

Looked through all the logs for failures:

1. All Jitstress ones are Assertion failed 'block->bbWeight > BB_ZERO_WEIGHT' #50743.
2. GCStress failures appear to be Comparer_get_Default test is too slow under GCStress=3 #50594 (fixed in Simplify Comparer_get_Default runtime test #50602) and Test failure: Loader/ContextualReflection/ContextualReflection/ContextualReflection.sh #48579 (disabled in Disable TestDynamicAssembly tests #50592).

Due to the fact that this branch is based on top of a commit that does not include the aforementioned fixes, the tests are failing. I will rebase on top of 8e0b54d (so as to not pick up 245cd97 which is believed to be the trigger for #50743) and that will hopefully give us a clear run.

Sorry for not taking a closer look at this sooner :(.

[AndyAyersMS]

Sorry for not taking a closer look at this sooner

No worries.

I am not planning on rerunning any stress legs here -- if the normal CI passes, I'll merge.

[AndyAyersMS]

Going to bounce this once more....

[AndyAyersMS]

Finally!

Thanks again, @SingleAccretion...