Re: kern/58539: AVX-512 support incomplete/broken

["Thomas Klausner via gnats" <gnats-admin%NetBSD.org@localhost>]

The following reply was made to PR toolchain/58539; it has been noted by GNATS.

From: Thomas Klausner <wiz%netbsd.org@localhost>
To: Taylor R Campbell <riastradh%netbsd.org@localhost>
Cc: PHO <pho%cielonegro.org@localhost>, gnats-bugs%netbsd.org@localhost
Subject: Re: kern/58539: AVX-512 support incomplete/broken
Date: Thu, 25 Dec 2025 00:02:47 +0100

 Hi!
 
 My main development machine is running NetBSD 11.99.4/x86_64 now.
 I have no environment for testing 9/10/11.
 
 On Wed, Dec 24, 2025 at 07:45:50PM +0100, Taylor R Campbell wrote:
 > If you boot a kernel from the tip of netbsd-9, netbsd-10, netbsd-11,
 > or HEAD on an affected machine:
 > 
 > 1. Is this still reproducible, when you re-enable the upstream AVX512
 >    logic?
 
 I tried moving away
 lang/ghc98/patches/patch-libraries_text_cbits_measure__off.c and the
 build did succeed now!
 
 > 2. Can you share the output of `cpuctl identify 0'?
 
 cpu0: highest basic info 00000010
 cpu0: highest extended info 80000028
 cpu0: "AMD Ryzen 9 7950X 16-Core Processor            "
 cpu0: AMD Family 19h (686-class), 4491.52 MHz
 cpu0: family 0x19 model 0x61 stepping 0x2 (id 0xa60f12)
 cpu0: features 0x178bfbff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE>
 cpu0: features 0x178bfbff<MCA,CMOV,PAT,PSE36,CLFSH,MMX,FXSR,SSE,SSE2,HTT>
 cpu0: features1 0x7ed8320b<SSE3,PCLMULQDQ,MONITOR,SSSE3,FMA,CX16,SSE41,SSE42>
 cpu0: features1 0x7ed8320b<MOVBE,POPCNT,AES,XSAVE,OSXSAVE,AVX,F16C,RDRAND>
 cpu0: features2 0x2fd3fbff<SYSCALL/SYSRET,NOX,MMXX,MMX,FXSR,FFXSR,P1GB,RDTSCP>
 cpu0: features2 0x2fd3fbff<LONG>
 cpu0: features3 0x75c237ff<LAHF,CMPLEGACY,SVM,EAPIC,ALTMOVCR0,LZCNT,SSE4A>
 cpu0: features3 0x75c237ff<MISALIGNSSE,3DNOWPREFETCH,OSVW,IBS,SKINIT,WDT,TCE>
 cpu0: features3 0x75c237ff<TopoExt,PCExtC,PCExtNB,DBExt,L2IPERFC,MWAITX>
 cpu0: features3 0x75c237ff<AddrMaskExt>
 cpu0: features5 0xf1bf97a9<FSGSBASE,BMI1,AVX2,SMEP,BMI2,ERMS,INVPCID,QM,PQE>
 cpu0: features5 0xf1bf97a9<AVX512F,AVX512DQ,RDSEED,ADX,SMAP,AVX512_IFMA>
 cpu0: features5 0xf1bf97a9<CLFLUSHOPT,CLWB,AVX512CD,SHA,AVX512BW,AVX512VL>
 cpu0: features6 0x405fce<AVX512_VBMI,UMIP,PKU,AVX512_VBMI2,CET_SS,GFNI,VAES>
 cpu0: features6 0x405fce<VPCLMULQDQ,AVX512_VNNI,AVX512_BITALG,AVX512_VPOPCNTDQ>
 cpu0: features6 0x405fce<MAWAU=0,RDPID>
 cpu0: xsave features 0x2e7<x87,SSE,AVX,Opmask,ZMM_Hi256,Hi16_ZMM,PKRU>
 cpu0: xsave instructions 0xf<XSAVEOPT,XSAVEC,XGETBV,XSAVES>
 cpu0: xsave area size: current 2432, maximum 2440, xgetbv enabled
 cpu0: enabled xsave 0xe7<x87,SSE,AVX,Opmask,ZMM_Hi256,Hi16_ZMM>
 cpu0: I-cache: 32KB 64B/line 8-way, D-cache: 32KB 64B/line 8-way
 cpu0: L2 cache: 1MB 64B/line 8-way
 cpu0: L3 cache: 32MB 64B/line 16-way
 cpu0: ITLB: 64 4KB entries fully associative, 64 2MB entries fully associative, 64 1GB entries fully associative
 cpu0: DTLB: 72 4KB entries fully associative, 72 2MB entries fully associative, 72 1GB entries fully associative
 cpu0: L2 ITLB: 512 4KB entries 4-way, 512 2MB entries 2-way
 cpu0: L2 DTLB: 3072 4KB entries 8-way, 3072 2MB entries 6-way, 64 1GB entries fully associative
 cpu0: Initial APIC ID 0
 cpu0: Cluster/Package ID 0
 cpu0: Core ID 0
 cpu0: SMT ID 0
 cpu0: MONITOR/MWAIT extensions 0x3<EMX,IBE>
 cpu0: monitor-line size 64
 cpu0: C0 substates 1
 cpu0: C1 substates 1
 cpu0: DSPM-eax 0x4<ARAT>
 cpu0: DSPM-ecx 0x1<HWF,NTDC=0>
 cpu0: SEF highest subleaf 00000001
 cpu0: SEF-subleaf1-eax 0x20<AVX512_BF16>
 cpu0: Power Management features: 0x6799<TS,TTP,HTC,HWP,ITSC,CPB,EffFreq,CONNSTBY,RAPL>
 cpu0: AMD Extended features 0x791ef257<CLZERO,IRPERF,XSAVEERPTR,RDPRU,MBE>
 cpu0: AMD Extended features 0x791ef257<WBNOINVD,IBPB,INT_WBINVD,IBRS,STIBP>
 cpu0: AMD Extended features 0x791ef257<STIBP_ALWAYSON,PREFER_IBRS>
 cpu0: AMD Extended features 0x791ef257<IBRS_SAMEMODE,EFER_LSMSLE_UN,SSBD,CPPC>
 cpu0: AMD Extended features 0x791ef257<PSFD,BTC_NO,IBPB_RET>
 cpu0: AMD Extended features2 0x62fcf<NoNestedDataBp>
 cpu0: AMD Extended features2 0x62fcf<FsGsKernelGsBaseNonSerializing>
 cpu0: AMD Extended features2 0x62fcf<LfenceAlwaysSerialize,SmmPgCfgLock>
 cpu0: AMD Extended features2 0x62fcf<NullSelectClearsBase,UpperAddressIgnore>
 cpu0: AMD Extended features2 0x62fcf<AutomaticIBRS,NoSmmCtlMSR,FSRS,FSRC>
 cpu0: AMD Extended features2 0x62fcf<PrefetchCtlMSR,CpuidUserDis,EPSF>
 cpu0: RAS features 0x3b<OVFL_RECOV,SUCCOR,MCAX>
 cpu0: SVM Rev. 1
 cpu0: SVM NASID 32768
 cpu0: SVM features 0x1ebfbcff<NP,LbrVirt,SVML,NRIPS,TSCRate,VMCBCleanBits>
 cpu0: SVM features 0x1ebfbcff<FlushByASID,DecodeAssist,PauseFilter,B11>
 cpu0: SVM features 0x1ebfbcff<PFThreshold,AVIC,V_VMSAVE_VMLOAD,VGIF,GMET>
 cpu0: SVM features 0x1ebfbcff<x2AVIC,SSSCHECK,SPEC_CTRL,ROGPT>
 cpu0: SVM features 0x1ebfbcff<HOST_MCE_OVERRIDE,VNMI,IBSVIRT>
 cpu0: SVM features 0x1ebfbcff<ExtLvtOffsetFaultChg,VmcbAddrChkChg>
 cpu0: IBS features 0xbff<IBSFFV,FetchSam,OpSam,RdWrOpCnt,OpCnt,BrnTrgt>
 cpu0: IBS features 0xbff<OpCntExt,RipInvalidChk,OpBrnFuse,IbsFetchCtlExtd>
 cpu0: IBS features 0xbff<Zen4IbsExtensions>
 cpu0: Encrypted Memory features 0x1<SME>
 cpu0: Perfmon: 0x7<PerfMonV2,LbrStack,LbrAndPmcFreeze>
 cpu0: Perfmon: counters: Core 6, Northbridge 16, UMC 8
 cpu0: Perfmon: LBR Stack 16 entries
 cpu0: Hetero workload class: 0 classes
 cpu0: UCode version: 0xa60120c
 
 
 > 3. Can you try running the attached program and tell me what it prints?
 > 
 > 	$ make avx512ftest DBG=-g\ -O2\ -Wall\ -Werror\ -mavx512f
 > 	$ ./avx512ftest
 
 Sure.
 
 33
 00
 00
 00
 32
 00
 00
 00
 3d
 00
 00
 00
 3c
 00
 00
 00
 3f
 00
 00
 00
 3e
 00
 00
 00
 39
 00
 00
 00
 38
 00
 00
 00
 3b
 00
 00
 00
 3a
 00
 00
 00
 05
 00
 00
 00
 04
 00
 00
 00
 07
 00
 00
 00
 06
 00
 00
 00
 01
 00
 00
 00
 00
 00
 00
 00
 
 Cheers,
  Thomas