Re: kern/41038: Pentium 1 crashes during boot install kernel

[Martin Husemann <martin%duskware.de@localhost>]

The following reply was made to PR kern/41038; it has been noted by GNATS.

From: Martin Husemann <martin%duskware.de@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: kern/41038: Pentium 1 crashes during boot install kernel
Date: Tue, 21 Apr 2009 09:52:34 +0200

 I worked with Ray and we got a bit of debug output for this:
 
 fpa0 at pci0 dev 17 function 0: FDDI controller
 
 PDQ Descriptor Block = 0xc63da000 (PA = 0x1148000)
     Receive Queue          = 0xc63da000
     Transmit Queue         = 0xc63da800
     Host SMT Queue         = 0xc63db000
     Command Response Queue = 0xc63db280
     Command Request Queue  = 0xc63db300
 PDQ Consumer Block = 0xc63dd000
 PDQ CSRs: BASE = 0xc62d7000
     Port Reset                = 0x0 [0x00000000]
     Host Data                 = 0x4 [0x00000000]
     Port Control              = 0x8 [0x00000000]
     Port Data A               = 0xc [0x00000000]
     Port Data B               = 0x10 [0x00000000]
     Port Status               = 0x14 [0x00000200]
     Host Int Type 0           = 0x18 [0x000000d2]
     Host Int Enable           = 0x1c [0x00000000]
     Type 2 Producer           = 0x20 [0x00000000]
     Command Response Producer = 0x28 [0x00000000]
     Command Request Producer  = 0x2c [0x00000000]
     Host SMT Producer         = 0x30 [0x00000000]
     Unsolicited Producer      = 0x34 [0x00000000]
 PDQ Command Request Buffer = 0xc63db500 (PA=0x1149500)
 PDQ Command Response Buffer = 0xc63db900 (PA=0x1149900)
 PDQ Unsolicit Event Buffer = 0xc63dc000 (PA=0x1147000)
 PDQ Adapter State = DMA Unavailable
 CSR cmd spun 423 times
 CSR cmd spun 422 times
 CSR cmd spun 556 times
 CSR cmd spun 490 times
 uvm_fault(0xc09b0800, 0xc63dd000, 2) -> 0xe
 fatal page fault in supervisor mode
 trap type 6 code b eip c04a4c34 cs 8 eflags 10246 cr2 c63dd000 ilevel 8
 kernel: supervisor trap page fault, code=0
 Stopped in pid 0.1 (system) at  netbsd:pdq_stop+0x260:  movw    $0,0(%eax)
 db{0}> bt
 pdq_stop(c0d5e800,c08d6b57,14,4,34,c63db280,1148800,1149000,1200,c0d5e800) 
 at ne
 tbsd:pdq_stop+0x260
 pdq_initialize(1,c62d7000,c63d9a8c,0,c63d99d4,0,4,0,c5e73bc0,c61c8f10) 
 at netbsd
 :pdq_initialize+0x662
 
pdq_pci_attach(c61c8ef4,c63d99d4,c0ad5a8c,c63d99d4,8,0,c63d99d4,4000,c6282004,0)
  at netbsd:pdq_pci_attach+0x10c
 
 This is pretty strange: PDQ Consumer Block = 0xc63dd000
 is setup in sys/dev/ic/pdq_ifsubr.c:594 (side note: I don't understand the
 #ifdef sparc stuff there).
 
 The driver dies at first write access to pdq->pdq_cbp (i.e. writing a zero
 to 0xc63dd000).
 
 Anyone spot what's wrong with the bus_dmamem_map() call?
 
 Martin