In article <200105051859.OAA02968@doppelganger.waterside.net>,
Rafal Boni <rafal@mediaone.net> wrote:
>In message <200105051717.f45HHTR04267@saruman.ics.muni.cz>, you write: 
>
>-> Rafal Boni wrote:
>-> > I don't think so.  My primary NetBSD box started out as a 486/33 with
>-> > dual ESDI disks hanging off a ISA ESDI card.  And this was a plain ISA
>-> > box,  not a PS/2.
>-> > 
>-> > The disks have long since been replaced (since they could heat my appt.
>-> > and made enough noise that friends mistook the computer for a small jet
>-> > airplane 8-), so I'm not 100% sure if they were ID'd as generic ST506
>-> > disks or as ESDI disks by disklabel, but it's probably worth a second
>-> > look.
>-> 
>-> Hmm, true. Default wd(4) disklabel uses DTYPE_ESDI for disks
>-> not identified as ST506. Perhaps a change to also check for
>-> MCA bus presence would do? The assembler code for that check
>-> would be next ~700 bytes however (slighly modified stand/lib/biosmca.S).
>
>Hmm, that's a hard call... 700 bytes seems like a lot of bloat to add to
>the boot blocks just to figure out if those disks were on an ISA EDSI card
>or an MCA ESDI card.  I'd hate to force the case of having an ESDI boot 
>disk to automatically imply it's hanging off an MCA ESDI controller.

I don't know why you think that biosmca.S generates 700 bytes of code.
Yes, biosmca.o is 704 bytes, but that is an ELF object!
It generates more like 39 bytes of code:

biosmca.o:     file format elf32-i386

Contents of section .text:
 0000 5589e553 51525657 e8fcffff ff66b8a0  U..SQRVW.....f..
 0010 000000b4 c0cd1572 0b678cc1 6789da66  .......r.g..g..f
 0020 89c3eb03 66b30066 e8fcffff ff31c066  ....f..f.....1.f
 0030 89d88b5c 241c894b 008b5c24 20895300  ...\$..K..\$ .S.
 0040 5f5e5a59 5b5dc3                      _^ZY[].         
Contents of section .data:

So it is really a non-issue.

christos