Re: Keyboard Problem with V5.1 on an E4000

[Julian Coleman <jdc%coris.org.uk@localhost>]

Hi,

> so, i ported the fhc and central drivers, and the clock and zs
> attachments for them.  however, it crashes when attaching the
> central device.  not sure what is up there yet, it crashes in
> a very strange way.
> 
> myself and others will keep looking.

Thanks for porting fhc and cnetral.  I tried this out on an E3500 with a
serial console.  Console output is:

  mainbus0 (root): SUNW,Ultra-Enterprise (5-slot Sun Enterprise E3500): hostid 
809341f3
  cpu0 at mainbus0: SUNW,UltraSPARC-II @ 400 MHz, UPA id 6
  cpu0: 16K instruction (32 b/l), 16K data (32 b/l), 8192K external (64 b/l)
    ...
  cpu7 at mainbus0: SUNW,UltraSPARC-II @ 400 MHz, UPA id 19
  cpu7: 16K instruction (32 b/l), 16K data (32 b/l), 8192K external (64 b/l)
  central0 at mainbus0
  fhc0 at central0 board 1: SUNW,501-2511
  clock0 at fhc0: mk48t59
  zs0 at fhc0
  zstty0 at zs0 channel 0 (console i/o)
  zstty1 at zs0 channel 1
  
  Fatal Reset
    
  @(#) Ultra Enterprise 3.2 Version 30 created 2002/10/25 14:03
  3,0>FATAL ERROR
  3,0>    At time of error: System software was running.
  3,0>    Diagnosis: Board 1, software, any system board
  3,0>Log Date: Aug  8 12:11:34 GMT 2011
  3,0>
  3,0>RESET INFO for IO Type 5   board in slot 1
  3,0>    AC ESR 00000000.00000011 IIS UPA_A_ERR
  3,0>    DC[0] 00
  3,0>    DC[1] 00
  3,0>    DC[2] 00
  3,0>    DC[3] 00
  3,0>    DC[4] 00
  3,0>    DC[5] 00
  3,0>    DC[6] 00
  3,0>    DC[7] 00
  3,0>    FHC  CSR 00040000 LOC_FATAL
  3,0>    FHC RCSR 02000000  FATAL

The reset is at the point where we enable interrupts on the serial port (line
473 of sys/arch/sparc64/dev/zs.c [1]).  According to the "Sun Enterprise xx00
Problem Solving Manual", section 4.13 [2], the reset is caused by "an
interrupt being sent to an unmapped or non-accepting destination".

This machine has an I/O board in slot 1 and CPU/memory boards in slots 3, 5,
7 and 9.  It looks like the zs interrupt was sent from the I/O board, even
though the serial ports are on the clock board in slot 0.  If I comment out
the interrupt enable, the machine will boot (but it's not possible to use
the console ;-).  Devices on the I/O board (SCSI, ethernet) appear to be
working.

So, is there some interrupt mapping that we are missing when we attach either
the CPU's or the boards (fhc or central) on a multi-board machine?

Thanks,

J

[1] http://opengrok.netbsd.org/source/xref/src/sys/arch/sparc64/dev/zs.c#473
[2] 
http://www.sunshack.org/data/sh/2.0/infoserver.central/data/910/910-4188/pdf/910-4188-11.pdf
-- 
  My other computer also runs NetBSD    /        Sailing at Newbiggin
        http://www.netbsd.org/        /   http://www.newbigginsailingclub.org/