RFR: 8312980: C2: "malformed control flow" created during incremental inlining

[Roland Westrelin]

On Mon, 9 Oct 2023 19:27:33 GMT, Daniel D. Daugherty <dcubed at openjdk.org> wrote:

>> Hi Roland, I'm seeing new failures with `compiler/c2/TestVerifyIterativeGVN.java` and `-XX:-TieredCompilation -XX:+AlwaysIncrementalInline`:
>> 
>> 
>> Missed Value optimization:
>> dist dump
>> ---------------------------------------------
>>    1   27  ConI  === 0  [[ 22 22 114 275 145 95 280 136 125 ]]  #int:-1
>>    1  638  CastII  === 629 195  [[ 593 640 244 244 145 275 260 220 247 ]]  #int:>=0:www !jvms: String::checkIndex @ bci:5 (line 4832) StringLatin1::charAt @ bci:3 (line 46) String::charAt @ bci:12 (line 1555) IPAddressUtil::scan @ bci:41 (line 472)
>>    1  142  Region  === 142 283 141  [[ 142 152 143 144 145 ]] #reducible  !jvms: IPAddressUtil::scan @ bci:28 (line 472)
>>    0  145  Phi  === 142 638 27  [[ 149 ]]  #int !jvms: IPAddressUtil::scan @ bci:28 (line 472)
>> Current type:
>> int
>> Optimized type:
>> int:>=-1:www
>> #
>> # A fatal error has been detected by the Java Runtime Environment:
>> #
>> #  Internal Error (/workspace/open/src/hotspot/share/opto/phaseX.cpp:1087), pid=4100110, tid=4100126
>> #  assert(!failure) failed: Missed optimization opportunity in PhaseIterGVN
>> #
>> # JRE version: Java(TM) SE Runtime Environment (22.0) (fastdebug build 22-internal-2023-10-02-1419484.tobias.hartmann.jdk2)
>> # Java VM: Java HotSpot(TM) 64-Bit Server VM (fastdebug 22-internal-2023-10-02-1419484.tobias.hartmann.jdk2, compiled mode, sharing, compressed oops, compressed class ptrs, g1 gc, linux-amd64)
>> # Problematic frame:
>> # V  [libjvm.so+0x152c41e]  PhaseIterGVN::verify_optimize() [clone .part.0]+0x69e
>> #
>> 
>> 
>> Best regards,
>> Tobias
>
> @TobiHartmann and/or @rwestrel  - What's the current status of this PR? We see intermittent
> failures in Mach5 Tier[4-8] due to the issue so I'd love to quiet that down...

@dcubed-ojdk I'm waiting for another review.

-------------

PR Comment: https://git.openjdk.org/jdk/pull/15905#issuecomment-1754498620