Add 10 when digit is [a-fA-F] in ConfigMethodRange::InitRanges

[echesakov]

Fix small issue that broke JitStressRange after 6dd6b6c

[AndyAyersMS]

Thanks. That may explain why I was still having trouble using this...

[echesakov]

There is another problem at

runtime/src/coreclr/src/jit/compiler.cpp

Line 1775 in 847f2fa

bRangeAllowStress = fJitStressRange.Contains(info.compMethodHash());

When we execute info.compMethodHash https://github.com/dotnet/runtime/blob/master/src/coreclr/src/jit/compiler.cpp#L5492-L5504 compMethodHashPrivate is not initialized yet and contains either an arbitrary value (in rel) or 0xdddddddd (in dbg).

This is what happens when I set COMPlus_JitStressRange=dddddddd. Basically, all methods are got jit-stressed

root@e489b5beca1a:/opt/code/artifacts/tests/coreclr/Linux.arm64.Checked/JIT/HardwareIntrinsics/General/Vector256_1/Vector256_1_r# COMPlus_DumpJittedMethods=1 COMPlus_JitStress=2 COMPlus_JitStressRange=dddddddd $CORE_ROOT/corerun Vector256_1_r.dll AllBitsSet.UInt16
Compiling    0 System.Runtime.CompilerServices.CastHelpers::StelemRef, IL size = 86, hash=0x1b5e9e9c FullOpts JitStress
Compiling    1 System.Runtime.CompilerServices.CastHelpers::LdelemaRef, IL size = 42, hash=0xd1234a96 Tier-1/FullOpts switched to MinOpts JitStress
Compiling    2 System.SpanHelpers::IndexOf, IL size = 1068, hash=0x576e6ccc FullOpts JitStress
Compiling    3 System.Runtime.Intrinsics.Vector128::Create, IL size = 28, hash=0xabf13899 FullOpts JitStress
Compiling    4 System.Runtime.Intrinsics.Vector128::AsByte, IL size = 7, hash=0x9122c64b Tier-1/FullOpts switched to MinOpts JitStress
Compiling    5 JIT.HardwareIntrinsics.General.Program::Main, IL size = 139, hash=0x8ad3a46d MinOpts JitStress
Compiling    6 JIT.HardwareIntrinsics.General.Program::.cctor, IL size = 2582, hash=0xe03e8610 MinOpts JitStress
Compiling    7 JIT.HardwareIntrinsics.General.Program::GetTestsToRun, IL size = 130, hash=0x38c23919 MinOpts JitStress
Compiling    8 System.Linq.Enumerable::Contains, IL size = 123, hash=0x13ce4c0d FullOpts JitStress
Compiling    9 TestLibrary.TestFramework::BeginTestCase, IL size = 150, hash=0x2feb4048 FullOpts JitStress
Compiling   10 TestLibrary.Generator::.cctor, IL size = 22, hash=0x95396aed Tier-1/FullOpts switched to MinOpts JitStress
Compiling   11 System.SpanHelpers::Contains, IL size = 297, hash=0xabfb7545 FullOpts JitStress
Compiling   12 System.SpanHelpers::IndexOf, IL size = 1001, hash=0xe0a010d4 FullOpts JitStress
Compiling   13 System.Runtime.Intrinsics.Vector128::AsByte, IL size = 7, hash=0x1cf43828 FullOpts JitStress
Compiling   14 System.Runtime.Intrinsics.Vector128::AsUInt64, IL size = 7, hash=0xdf1a5005 FullOpts JitStress
Compiling   15 System.Runtime.Intrinsics.Vector128::ToScalar, IL size = 18, hash=0x1ca6dde8 FullOpts JitStress
Compiling   16 System.SpanHelpers::IndexOfAny, IL size = 423, hash=0x41e07991 FullOpts JitStress
Compiling   17 ILStubClass::IL_STUB_ReversePInvoke, IL size = 52, hash=0x1554399c FullOpts JitStress
Compiling   18 System.SpanHelpers::SequenceCompareTo, IL size = 309, hash=0xbda601aa FullOpts JitStress
Compiling   19 System.Console::WriteLine, IL size = 12, hash=0x450bf92f FullOpts JitStress
Compiling   20 System.Console::.cctor, IL size = 11, hash=0x701237e5 Tier-1/FullOpts switched to MinOpts JitStress
Compiling   21 System.Console::<get_Out>g__EnsureInitialized|25_0, IL size = 63, hash=0x062f85e9 FullOpts JitStress
Compiling   22 System.ConsolePal::OpenStandardOutput, IL size = 43, hash=0x173bd4aa FullOpts JitStress
Compiling   23 <>c::.cctor, IL size = 11, hash=0x7a92f338 Tier-1/FullOpts switched to MinOpts JitStress
Compiling   24 <>c::.ctor, IL size = 7, hash=0xcce2e927 FullOpts JitStress
Compiling   25 <>c::<OpenStandardOutput>b__9_0, IL size = 11, hash=0x8d7e91a1 FullOpts JitStress
Compiling   26 FileDescriptors::.cctor, IL size = 34, hash=0x29a8927f Tier-1/FullOpts switched to MinOpts JitStress
Compiling   27 FileDescriptors::CreateFileHandle, IL size = 13, hash=0x52e72907 FullOpts JitStress
Compiling   28 ILStubClass::IL_STUB_PInvoke, IL size = 79, hash=0x9ca0dbaa FullOpts JitStress
Compiling   29 System.Console::CreateOutputWriter, IL size = 50, hash=0x8abcf282 FullOpts JitStress
Compiling   30 System.Console::get_OutputEncoding, IL size = 72, hash=0x2b0acc16 FullOpts JitStress
Compiling   31 System.ConsolePal::GetConsoleEncoding, IL size = 23, hash=0x63b461bc FullOpts JitStress
Compiling   32 System.Text.EncodingHelper::GetEncodingFromCharset, IL size = 25, hash=0x2d07865b FullOpts JitStress
Compiling   33 System.Text.EncodingHelper::GetCharset, IL size = 120, hash=0x5fb34eab FullOpts JitStress
Compiling   34 System.Text.EncodingHelper::.cctor, IL size = 36, hash=0xb28e5a04 Tier-1/FullOpts switched to MinOpts JitStress
Compiling   35 System.Text.EncodingExtensions::RemovePreamble, IL size = 25, hash=0x02a7f0d3 FullOpts JitStress
Compiling   36 System.IO.ConsoleStream::get_CanWrite, IL size = 7, hash=0xfcc48126 FullOpts JitStress
Compiling   37 System.IO.ConsoleStream::get_CanSeek, IL size = 2, hash=0xc1997943 FullOpts JitStress
Compiling   38 UnixConsoleStream::Flush, IL size = 26, hash=0x39be8ac7 FullOpts JitStress
Compiling   38 UnixConsoleStream::Flush, IL size = 26, hash=0x39be8ac7 MinOpts JitStress
Compiling   39 System.IO.ConsoleStream::Flush, IL size = 15, hash=0x0b5b804e FullOpts JitStress
Compiling   40 UnixConsoleStream::Write, IL size = 25, hash=0x59ed5039 FullOpts JitStress
Compiling   41 System.IO.ConsoleStream::ValidateWrite, IL size = 83, hash=0x489d9676 FullOpts JitStress

[echesakov]

cc @dotnet/jit-contrib in case if someone tries to use JitStressRange

[echesakov]

I think the second issue was introduced by #38103 and could be fixed by moving the following lines https://github.com/dotnet/runtime/blob/master/src/coreclr/src/jit/compiler.cpp#L5290-L5292 from Compiler::compCompile to Compiler::compInit to make info.compMethodHashPrivate zero-ed before a call to info.compMethodHash().

@CarolEidt Does this sound reasonable?

[AntonLapounov]

I do not think the "Check for overflow" immediately after the fix or m_badChar handling are correct either.

[briansull]

LGTM

[echesakov]

I do not think the "Check for overflow" immediately after the fix or m_badChar handling are correct either.

@AntonLapounov Yeah, I think you are right but I would like to fix JitStressRange first since it doesn't work even with a valid string now. We should look at overflow check and bad character logic.

[echesakov]

@dotnet/jit-contrib I just pushed another change required for fixing the JitStressRange as per #39754 (comment). Can someone please sign off one more time?