C2 crashes with Unschedulable Graph in ChaCha20Poly1305 microbenchmark with ChaCha20 intrinsics and UseAVX=3

There is a work-in-progress RFE for ChaCha20 intrinsics (https://github.com/openjdk/jdk/pull/7702) which appears to be causing an issue with the C2 intermediate representation. This appears to happen under specific conditions:

* The system must be x86_64 and have avx512f support
* This has only been seen on linux, other x86 platforms are unknown at this time.
* -XX:UseAVX=3 must be employed (we have not seen the error without it or with -XX:UseAVX=1 or 2)
* It can be worked around with the inclusion of -XX:+UnlockDiagnosticVMOptions -XX:ArrayOperationPartialInlineSize=0

This so far has only been seen when running test/micro/org/openjdk/bench/javax/crypto/full/CipherBench.java, specifically the ChaCha20Poly1305.decrypt benchmark. This is also without the pending Poly1305 intrinsics which are also in review at this time.

Info for the system used to reproduce this issue:
Linux <hostname-omitted> 4.14.35-2047.515.3.el7uek.x86_64 #2 SMP Thu Jun 30 18:46:19 PDT 2022 x86_64 x86_64 x86_64 GNU/Linux
Intel(R) Xeon(R) Platinum 8167M CPU @ 2.00GHz
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ss ht syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon rep_good nopl xtopology cpuid tsc_known_freq pni pclmulqdq vmx ssse3 fma cx16 pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm abm 3dnowprefetch cpuid_fault invpcid_single pti ssbd ibrs ibpb stibp tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm mpx avx512f avx512dq rdseed adx smap clflushopt clwb avx512cd avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves arat umip pku ospke md_clear


Sample invocation:
<OPEN-REPO-TOP>/build/linux-x64/jdk/bin/java -XX:UseAVX=3 -jar <OPEN-REPO-TOP>/build/linux-x64/images/test/micro/benchmarks.jar ChaCha20Poly1305.decrypt

# JMH version: 1.34
# VM version: JDK 20-internal, OpenJDK 64-Bit Server VM, 20-internal-2022-11-01-1746088.jjnimeh...
# VM invoker: /home/jjnimeh/workspace/jdk-intrin/build/linux-x64/jdk/bin/java
# VM options: -XX:UseAVX=3 -XX:+AlwaysPreTouch
# Blackhole mode: compiler (auto-detected, use -Djmh.blackhole.autoDetect=false to disable)
# Warmup: 3 iterations, 3 s each
# Measurement: 8 iterations, 2 s each
# Timeout: 10 min per iteration
# Threads: 1 thread, will synchronize iterations
# Benchmark mode: Throughput, ops/time
# Benchmark: org.openjdk.bench.javax.crypto.full.CipherBench.ChaCha20Poly1305.decrypt
# Parameters: (dataSize = 256, keyLength = 256, mode = None, padding = NoPadding, permutation = ChaCha20-Poly1305, provider = )

# Run progress: 0.00% complete, ETA 00:10:25
# Fork: 1 of 5
# Warmup Iteration 1: !!! Unschedulable graph !!!
B156 idom=B153 depth=26 321 vmasked_store_evex === 1121 310 315 322 323 [[ 320 473 ]] memory Memory: @byte[int:32]:NotNull:exact[0] *,iid=1079, idx=39;
B59 idom=B52 depth=20 442 addP_rReg_imm === _ 311 311 [[ 443 415 447 473 478 479 ]] byte[int:32]:NotNull:exact[0] *,iid=1079
B144 idom=B143 depth=31 474 vmask_gen === _ 475 477 [[ 473 478 479 470 ]] vectormask[64]:{byte}
Failing node: 473 vmasked_load_evex === _ 321 442 474 [[ 472 ]] vectory[32]:{byte}


The hs_err_pid files from the above run for all four data sizes (256, 1024, 4096, 16384) have been attached to this issue.

It is believed that this may be related to JDK-8252848 or JDK-8266951. Investigations thus far point away from the new stubs themselves.