Re: kernel panic with greater that 8 GB of memory

From: Steve Kargl (sgk_at_troutmask.apl.washington.edu)
Date: 11/30/04

  • Next message: Scott Long: "Re: boot menu"
    Date: Tue, 30 Nov 2004 10:35:55 -0800
    To: Scott Long <scottl@freebsd.org>
    
    

    On Tue, Nov 30, 2004 at 06:04:56AM -0700, Scott Long wrote:
    > Steve Kargl wrote:
    > >On Mon, Nov 29, 2004 at 01:13:41PM -0800, Steve Kargl wrote:
    > >
    > >>kmem_suballoc: bad status return of 3
    > >>panic: kmem_suballoc
    > >>cpuid = 0
    > >>KDB: stack backtrace:
    > >>kdb_backtrace() at kdb_backtrace+0x37
    > >>panic() at painc+0x1d1
    > >>kmem_suballoc() at kmem_suballoc+0x93
    > >>kmeminit() at kmeminit+0x16d
    > >>mi_startuup() at mi_startup+0xb6
    > >>btext90 at btext+0x2c
    > >>KDB: enter: panic
    > >>[thread pid 0 tid 0]
    > >>
    > >>This is a Tyan K8S Pro motherboard with 16 GB of PC2700
    > >>Cosair ECC memory. The BIOS was flashed to the latest
    > >>available from Tyan. The scheduler is 4BSD. There are
    > >>no ATA devices in the system and no ATA subsystem in
    > >>the kernel. This a SCSI only box. World and kernel
    > >>were rebuilt today with CFLAGS=-O -pipe and
    > >>COPTFLAGS=-O -pipe.
    > >>
    > >>If I set hw.physmem="#G" n /boot/loader.conf where # is
    > >>less than or equal to 8 the box boots up (see dmesg below).
    > >>If # > 8, then we have the above panic.
    > >>
    > >>Any ideas where to look for the a fix?
    > >>
    > >
    > >(dmesg snipped)
    > >(CC'd to -current because the may impact any system with >8 GB of ram)
    > >
    > >As a data point, I've downloaded, burned, and booted the netbsd/amd64
    > >ISO image. Netbsd's dmesg shows that netbsd correctly finds 16 GB
    > >of memory on the system.
    > >
    >
    > I fixed this on i386 a few months ago, and I applied a similar fix to
    > amd64 on the advice of Peter Wemm. Go beat him up on this one =-)
    >

    Well, I have more data and Peter advice may have been correct.
    >From the beastie menu :-O, I dropped to the loader prompt. If
    I do

    set hint.acpi.0.disabled="1"
    set hw.phymem="6G"

    The system boots up fine (although I thought AMD64 required acpi).
    Next, if I do

    set hint.acpi.0.disabled="1"
    set hw.phymem="16G"

    The system starts to boot and dmesg reports the 16 GB of memory, but
    I rapidly get this panic.

    Fatal trap 12: page fault while in kernel mode
    cpuid = 0; apic id = 00
    fault virtual address = 0xffffffff80af6a80
    fault code = supervisor read, page not present
    instruction pointer = 0x8:0xffffffff802a5814
    stack pointer = 0x10:0xffffffff80728a30
    code segment = base 0x0, limit 0xfffff, type 0x1b
                          = DPL 0, pres 1, long 1, def32 0, gran 1
    processor eflags = resume, IOPL = 0
    current process = 0 (swapper)
    [thread pid 0 tid 0]
    Stopped at isitmychild+0x4: decl %eax

    db> trace
    isitmychild() at isitmychild+0x4
    isitmydescendant() at isitmydescendant+0x16
    witness_checkorder() at witness_checkorder+0x44a
    _mtx_lock_flags() at _mtx_lock_flags+0x7a
    lim_hold() at lim_hold+0x29
    fork1() at fork1+0xaa0
    kthread_create() at kthread_create+0xa3
    ktrace_init() at ktrace_init+0xe4
    mi_startup() at mi_startup+0xb6

    So, it looks like ACPI on this Tyan motherboard is restricting
    the memory to 8G.

    -- 
    Steve
    _______________________________________________
    freebsd-current@freebsd.org mailing list
    http://lists.freebsd.org/mailman/listinfo/freebsd-current
    To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"
    

  • Next message: Scott Long: "Re: boot menu"