[Unsupported config]Bug check during boot with bare metal SATA

[pocketprobe]

After the release of V9.2-1 I wanted to see if adding the device IDs for the SATA controllers in AHCI mode would cause the machine to crash during boot still. After seeing the addition of VirtIO drivers for KVM/Virtualbox or at least the foundations of... I was inspired. The following transcript is from a modified install ISO. Similar behavior was found from a rebuilt boot memory disk and additions to user_config.dat.

If there is interest in more info, or working on this I'd be happy to provide whatever nessissary.

Code: Select all

VSI OpenVMS Boot Manager: V9.2-1, Build 122

  ENABLED:  Symmetric Multi-Processing
  ENABLED:  Crash Dump Processing
  ENABLED:  Console output to Legacy COM 1 Port

  Checking Required Processor Features:
  PASSED

  BOOT MANAGER DEVICE:  DMM0
  DEFAULT BOOT DEVICE:  DMM0

  AUTOACTION: BOOT DMM0 0 00000000  (Countdown = 5)

  CONNECT A REMOTE TERMINAL SESSION NOW.

Press <ESC> to stop automatic action or any other key to skip countdown and continue.

5...4...3...2...1...

Booting...
%VMS_BOOTMGR-I-INSTALL, Booting an OpenVMS Installation Kit...

%%%%%%%%%%% VSI OpenVMS (tm) x86-64 %%%%%%%%%%%
%%%%%%%%

 %%%
_______________________________________________

      GRAPHICAL OUTPUT HAS BEEN SUSPENDED
      USE A TERMINAL UTILITY FOR ACCESS
_______________________________________________

VSI Primary Kernel SYSBOOT May 16 2023 13:21:17


        VMS Software, Inc. OpenVMS (TM) x86_64 Operating System, V9.2-1  
                    Copyright 2023 VMS Software, Inc.

  MDS Mitigation active, variant haswell(HASWELL/BROADWELL)

%SMP-I-CPUTRN, CPU #9 has joined the active set.
%SMP-I-CPUTRN, CPU #8 has joined the active set.
%SMP-I-CPUTRN, CPU #2 has joined the active set.
%SMP-I-CPUTRN, CPU #3 has joined the active set.
%SMP-I-CPUTRN, CPU #1 has joined the active set.
%SMP-I-CPUTRN, CPU #6 has joined the active set.
%SMP-I-CPUTRN, CPU #12 has joined the active set.
%SMP-I-CPUTRN, CPU #4 has joined the active set.
%SMP-I-CPUTRN, CPU #10 has joined the active set.
%SMP-I-CPUTRN, CPU #7 has joined the active set.
%SMP-I-CPUTRN, CPU #13 has joined the active set.
%SMP-I-CPUTRN, CPU #5 has joined the active set.
%SMP-I-CPUTRN, CPU #11 has joined the active set.



    Installing required known files...

    Configuring devices...

MMG$PAGEFAULT IPLHI:
  Fault_VA:  00000000000004AC
  Err Code:  0000000000000000
  MMF_Flags: 0000000000000000
  Intstk_p:  FFFFFFFFA48D3C80
  RIP:       FFFF830064C5E92A
  Int_ps:    0000000000000800

*** Page Fault IPL Too High ***



VSI Dump Kernel SYSBOOT May 16 2023 13:21:17


**** OpenVMS x86_64 Operating System V9.2-1   - BUGCHECK ****

** Bugcheck code = 0000035C: PGFIPLHI, Pagefault with IPL too high
** Crash Time:            16-JUN-2023 00:31:28.20
** Crash CPU: 00000000    Primary CPU: 00000000
** Highest CPU number:    0000000D
** Active CPUs:           00000000.00003FFF
** Current Process:       "STARTUP"
** Current PSB ID:        00000001
** Image Name:            DMM1:[SYS0.SYSCOMMON.][SYSEXE]SYSMAN.EXE;1

** Error logs not dumped, system disk is write locked.
** No useable system memory dump file found.

Restarting the system...

[pjacobi]

The only bare metal SATA testing on X86 was on the discontinued Atom project. The driver is picky about supported SATA controllers. It only supports:

0x3A22 ICH10
0x2829 ICH8
0x2922 ICH9
0x07E0 VMware
0x5AE3 Atom E9000

The driver is set offline for unknown devices. I think setting FIBRE_SCSI_RSV1 to -1 in SYSBOOT enables messages from the driver.


Paul A. Jacobi
VMS Software, Inc.