Re: June Snapshot of 6.0 woes under qemu

On Saturday 02 July 2005 01:31 am, Andrew Turner wrote:
> Michel Talon wrote:
> >Hello,
> >
> >i have downloaded June snapshot of 6.0 for i386 (and checked md5), and
> >tried to run it on qemu (under FreeBSD-5.4, qemu with or without
> > acceleration, and with various memory sizes). I try to enter the fixit
> > console, and type commands, the console freezes. This is the first time i
> > see that with FreeBSD snapshots, qemu runs fine various Linux distros and
> > Solaris. I suspect there is some strange bug.
>
> I had the same problem yesterday with a custom build install CD. The
> problem is a lock order reversal:
>
> Fixit# ls
> lock order reversal
>  1st 0xc16d26e8 ATA state lock (ATA state lock) _at_
> /usr/src/sys/dev/ata/ata-all.c:297
> 2nd 0xc14611c4 user map (user map) _at_ 
> /usr/src/sys/vm/vm_map.c:2997
> KDB: stack backtrace: 
> kdb_backtrace(0,ffffffff,c0927968,c0928520,c08b464c) at kdb_backtrace+0x29
> witness_checkorder(c14611c4,9,c086a271,bb5) at witness_checkorder+0x564
> _sx_xlock(c14611c4,c086a271,bb5) at _sx_xlock+0x50
> _vm_map_lock_read(c1461180,c086a271,bb5,200dbbb,c) at
> _vm_map_lock_read+0x37 vm_map_lookup(c7f65b7c,c72d4000,2,c7f65b80,c7f65b70)
> at vm_map_lookup+0x28 vm_fault(c1461000,c72d4000,2,0,c16e0000) at
> vm_fault+0x66
> trap_pfault(c7f65c44,0,c72d4800) at trap_pfault+0x137
> trap(8,28,28,c72d4800,c16d2600) at trap+0x33d
> calltrap() at calltrap+0x5
> --- trap 0xc, eip = 0xc04d25cc, esp = 0xc7f65c84, ebp = 0xc7f65ca4 ---
> ata_pio_read(c1825190,800,129,c17c9d00,c17e8400) at ata_pio_read+0x78
> ata_end_transaction(c1825190) at ata_end_transaction+0x8b8
> ata_interrupt(c16d2600) at ata_interrupt+0xdf
> ithread_loop(c16dd800,c7f65d38,c16dd800,c061b418,0) at ithread_loop+0x11c
> fork_exit(c061b418,c16dd800,c7f65d38) at fork_exit+0xa0
> fork_trampoline() at fork_trampoline+0x8
> --- trap 0x1, eip = 0, esp = 0xc7f65d6c, ebp = 0 ---
> panic: vm_fault: fault on nofault entry, addr: c72d4000
> cpuid = 0
> KDB: enter: panic
> [thread pid 26 tid 100010 ]
> Stopped at      kdb_enter+0x2b: nop
> db>
>
> This is known. See:
> http://sources.zabbadoz.net/freebsd/lor.html#101

This is not a LOR.  LORs involving "system map" or "user map" are almost 
always false positives.  The real issue is a panic in ata_pio_read(), 
probably due to a NULL pointer.

-- 
John Baldwin <jhb_at_FreeBSD.org>  <><  http://www.FreeBSD.org/~jhb/
"Power Users Use the Power to Serve"  =  http://www.FreeBSD.org