virtualbox status on 8.0-STABLE i386

[Mikolaj Golub]

Hi,

Recently I have updated my 8.0-STABLE i386 system and have learnt that
virtualbox begins to crash my box with the error

panic: vm_fault: fault on nofault entry, addr: c1608000

(kgdb) bt
#0  doadump () at pcpu.h:246
#1  0xc04ec379 in db_fncall (dummy1=-1064468854, dummy2=0, dummy3=-1, dummy4=0xe865d5bc "пуeХ")
    at /usr/src/sys/ddb/db_command.c:548
#2  0xc04ec7af in db_command (last_cmdp=0xc0e04c9c, cmd_table=0x0, dopager=0)
    at /usr/src/sys/ddb/db_command.c:445
#3  0xc04ec864 in db_command_script (command=0xc0e05bc4 "call doadump")
    at /usr/src/sys/ddb/db_command.c:516
#4  0xc04f09a0 in db_script_exec (scriptname=0xe865d6c8 "kdb.enter.panic", warnifnotfound=Variable "warnifnotfound" is not available.
)
    at /usr/src/sys/ddb/db_script.c:302
#5  0xc04f0a87 in db_script_kdbenter (eventname=0xc0cc248d "panic") at /usr/src/sys/ddb/db_script.c:324
#6  0xc04ee768 in db_trap (type=3, code=0) at /usr/src/sys/ddb/db_main.c:228
#7  0xc08d7d06 in kdb_trap (type=3, code=0, tf=0xe865d804) at /usr/src/sys/kern/subr_kdb.c:535
#8  0xc0beb39b in trap (frame=0xe865d804) at /usr/src/sys/i386/i386/trap.c:690
#9  0xc0bccd0b in calltrap () at /usr/src/sys/i386/i386/exception.s:165
#10 0xc08d7e8a in kdb_enter (why=0xc0cc248d "panic", msg=0xc0cc248d "panic") at cpufunc.h:71
#11 0xc08a88b6 in panic (fmt=0xc0cecbc4 "vm_fault: fault on nofault entry, addr: %lx")
    at /usr/src/sys/kern/kern_shutdown.c:562
#12 0xc0b0c3d7 in vm_fault (map=0xc1990000, vaddr=3244326912, fault_type=Variable "fault_type" is not available.
)
    at /usr/src/sys/vm/vm_fault.c:283
#13 0xc0bea7d6 in trap_pfault (frame=0xe865dac0, usermode=0, eva=3244330720)
    at /usr/src/sys/i386/i386/trap.c:840
#14 0xc0beb225 in trap (frame=0xe865dac0) at /usr/src/sys/i386/i386/trap.c:533
#15 0xc0bccd0b in calltrap () at /usr/src/sys/i386/i386/exception.s:165
#16 0xc12beed0 in rtR0MemObjNativeGetPagePhysAddr (pMem=0xc5ed3110, iPage=0) at pmap.h:300
#17 0xc12ac354 in SUPR0LockMem (pSession=0xc5c61c10, pvR3=695959552, cPages=1, paPages=0xc5f83668)
    at SUPDrv.c:2307
#18 0xc12ac8cb in supdrvIOCtl (uIOCtl=536892942, pDevExt=0xc12c9ac0, pSession=0xc5c61c10, 
    pReqHdr=0xc5f83650) at SUPDrv.c:1245
#19 0xc12b0c3a in VBoxDrvFreeBSDIOCtl (pDev=0xc665d800, ulCmd=536892942, 
    pvData=0xe865dd00 "ю8 )\003╬кюq\002", fFile=3, pTd=0xc69556f0)
    at /usr/ports/emulators/virtualbox-ose-kmod/work/VirtualBox-3.1.2_OSE/out/freebsd.x86/debug/bin/src/vboxdrv/freebsd/SUPDrv-freebsd.c:505
#20 0xc0829658 in devfs_ioctl_f (fp=0xc670fa80, com=536892942, data=0xe865dd00, cred=0xc6bbeb00, 
    td=0xc69556f0) at /usr/src/sys/fs/devfs/devfs_vnops.c:659
#21 0xc08eec8d in kern_ioctl (td=0xc69556f0, fd=7, com=536892942, data=0xe865dd00 "ю8 )\003╬кюq\002")
    at file.h:262
#22 0xc08eee14 in ioctl (td=0xc69556f0, uap=0xe865dcf8) at /usr/src/sys/kern/sys_generic.c:678
#23 0xc0beaad0 in syscall (frame=0xe865dd38) at /usr/src/sys/i386/i386/trap.c:1111
#24 0xc0bccda0 in Xint0x80_syscall () at /usr/src/sys/i386/i386/exception.s:261
#25 0x00000033 in ?? ()
Previous frame inner to this frame (corrupt stack?)
(kgdb) fr 16
#16 0xc12beed0 in rtR0MemObjNativeGetPagePhysAddr (pMem=0xc5ed3110, iPage=0) at pmap.h:300
300                     pa = (pa & PG_FRAME) | (va & PAGE_MASK);
(kgdb) list
295                      * access the PTE because it would use the new PDE.  It is,
296                      * however, safe to use the old PDE because the page table
297                      * page is preserved by the promotion.
298                      */
299                     pa = KPTmap[i386_btop(va)];
300                     pa = (pa & PG_FRAME) | (va & PAGE_MASK);
301             }
302             return (pa);
303     }
304

There were some changes in this part recently (r203182):

http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/i386/include/pmap.h.diff?r1=1.140.2.2;r2=1.140.2.3;only_with_tag=RELENG_8

So I removed KPTmap[i386_btop(va)] with *vtopte(va) and have working
virtualbox again, but I suppose this is rather the problem with virualbox and
not with the kernel code.

In February Alexander Eichner posted the patch to freebsd-emulation@ (thread
with the subject "Patch to fix VirtualBox with recent kernel versions"):

http://lists.freebsd.org/pipermail/freebsd-emulation/2010-February/007434.html

But it does not fix my panics. The patch adds additional handling in
rtR0MemObjNativeGetPagePhysAddr() for the case 
pMem.enmType == RTR0MEMOBJTYPE_MAPPING, while I am observing the panics
for pMem.enmType == RTR0MEMOBJTYPE_LOCK:

(kgdb) fr 17
#17 0xc12ac354 in SUPR0LockMem (pSession=0xc5c61c10, pvR3=695959552, cPages=1, paPages=0xc5f83668)
    at SUPDrv.c:2307
2307                paPages[iPage] = RTR0MemObjGetPagePhysAddr(Mem.MemObj, iPage);
(kgdb) p Mem.MemObj.enmType
$1 = RTR0MEMOBJTYPE_LOCK

So, it looks like some additional handling should be added for this case...

-- 
Mikolaj Golub