Re: [Qemu-devel] OS/2 PPC booting making some progress

[Tero Kaarlela <turo.kaarlela@co.inet.fi>]

J. Mayer wrote:

>>   2. on debug 2. What is this unaffected IO port 838 it tries to read & 
>>write ?
>>    
>>
>
>This port is not documented in the PREP specification. You should check
>in the Linux kernel to see if this port seems important. Or you may take
>a look to Motorola PowerStack documentation: those boards are well
>documented and OS/2 is supposed to know how to boot on it.
>  
>

  Hmmm I found following info from OS/2 stanza files(files describing 
hardware to bootloader:

    rdf.stz

  planar_ide_0:
        dev_class       = IDE
        adap_name       = "IDE Disk Controller"
        setting         = "io:ide_int_reg:838"  <=
        setting         = "io:ide_ctrl:3f6-3f7"
        setting         = "io:ide_cmd:1f0-1f7"
        setting         = "irq:ide_irq:d"
        setting         = "arb:ide_arb:6"
        setting         = "intrtn:ide_rtn:0"

So port 0838 is ide_int_reg

and from ppc_info.stz:

#Delmar/SouthWind
6050:
        name            = "IBM Delmar / South Wind PowerPC Platform"
        pci_config_space        = 0x80800000
        bus             = isa:4
        bus             = pci:4
        built_in        = planar_kbd
        built_in        = planar_mouse
        built_in        = planar_floppy
        built_in        = ppc_planar_parallel
        built_in        = COM1
        built_in        = COM2
        built_in        = CMOS_type1
        built_in        = planar_ide_0
        built_in        = planar_ide_1

So my guess is that OS/2 ppc reads machine configuration from these 
files instead of residual data.

>  
>
>>   3. Might that unsupported ioport prevent booting?
>>    
>>
>
>Can be possible...
>Have to check the code that uses that I/O port...
>
>[...]
>
>  
>
 Hmmm... i succeeded to make it go further with following options:

   boot-device ide_cdrom2
   boot-fs cdrom
then it works and loads boot.cfg from cd-image
shows all lines from boot.cfg (and I have option to uncomment/change all 
these)
loads kernel & device drivers into memory(as defined in boot.cfg)
but after boot.cfg has been completely read and real boot should start 
it gives me invalid/unsupported opcodes again.

Tero