Subject: Rehash of NetBSD/sgimips problems - 2-0 Branch build
To: None <port-sgimips@netbsd.org>
From: sgimips NetBSD list <sgimips@mrynet.com>
List: port-sgimips
Date: 05/17/2004 21:32:04
I just built and installed a fresh  build of  the NetBSD-2-0 branch to 
compare it to problems I experience with the Main branch.  The install
is on an SGI O2 R5K machine.

The following has been noted so far:

1) It has the same sshd problems that the Main branch does except that
   the second  invocation of sshd crashes the machine with a hard hang
   followed by a machine reset, rather than any panic or cache
   error messages.    The reboot,  however, indicates the following in
   startup messages:

    savecore: reboot after panic: cache error @ EPC 0x%x ErrCtl 0x%x CacheErr 0x%x
    May 17 19:40:48 mod84 savecore: reboot after panic: cache error @ EPC 0x%x ErrCtl 0x%x CacheErr 0x%x


2) If /bin/csh is the login shell, logging in results in either

    panic: cache error @ EPC 0x8036360c ErrCtl 0x1 CacheErr 0xa00f3db3
    panic: cache error @ EPC 0x8034391c ErrCtl 0x1 CacheErr 0xa00f3dab
    panic: cache error @ EPC 0x8034391c ErrCtl 0x1 CacheErr 0xa00f3dab
    panic: cache error @ EPC 0x8034391c ErrCtl 0x1 CacheErr 0xa00f3dab
    panic: cache error @ EPC 0x8034391c ErrCtl 0x1 CacheErr 0xa00f3dab
    panic: cache error @ EPC 0x8034391c ErrCtl 0x1 CacheErr 0xa00f3dab
    panic: cache error @ EPC 0x8034391c ErrCtl 0x1 CacheErr 0xa00f3dab
    panic: cache error @ EPC 0x8034391c ErrCtl 0x1 CacheErr 0xa00f3dab
    panic: cache error @ EPC 0x8034391c ErrCtl 0x1 CacheErr 0xa00f3dab
    panic: cache error @ EPC 0x8034391c ErrCtl 0x1 CacheErr 0xa00f3dab
    panic: cache error @ EPC 0x8034391c ErrCtl 0x1 CacheErr 0xa00f3dab
    panic: cache error @ EPC 0x8034391c ErrCtl 0x1 CacheErr 0xa00f3dab
    panic: cache error @ EPC 0x8034391c ErrCtl 0x1 CacheErr 0xa00f3dab
    panic: cache error @ EPC 0x8034391c ErrCtl 0x1 CacheErr 0xa00f3da

    ---- OR ----

    Stopped at      0x8033fc98:     jr      ra
                    bdslot: nop
    db> bt
    8033fa84+214 (8ffff000,bf390000,5,d) ra 802999fc sz 0
    8029986c+190 (8ffff000,d,5,33) ra 803692f8 sz 40
    panic: cache error @ EPC 0x803448d8 ErrCtl 0x1 CacheErr 0xa0352363
    Stopped at      0x8033fc98:     jr      ra
                    bdslot: nop
    db> reboot

3)  Executing /bin/csh manually from a shell prompt appears to have 
    no consequences if no command-line arguments are given.  If
    invoked as "csh -l" the above-mentioned cache panics occur.  This
    explains why it can't be used as a login shell as well.

    I'm currently building a new /bin/sh and will put some debugging
    output in the code to try to narrow down the functionality that
    is causing the panic.

To avoid crashes, I've disabled ssh and changed root to use /bin/sh
as its login shell.

Other than the above-mentioned issues, the machine seems otherwise just
as stable as the INDY.

-scott