Re: PCMCIA broken (was Re: no ndis* at cardbus?)

[christos%zoulas.com@localhost (Christos Zoulas)]

On Feb 11, 10:48pm, ad%netbsd.org@localhost (Andrew Doran) wrote:
-- Subject: Re: PCMCIA broken (was Re: no ndis* at cardbus?)

| On Wed, Feb 11, 2009 at 09:35:44PM +0000, Christos Zoulas wrote:
| > In article <1234379751.836836.989.nullmailer%galant.ukfsn.org@localhost>,
| > Iain Hibbert  <plunky%rya-online.net@localhost> wrote:
| > >On Wed, 11 Feb 2009, Matthias Drochner wrote:
| > >
| > >> As said, CompactFlash works for me. Seems I'm just lucky.
| > >
| > >Although I didn't have 5.0 on my old laptop (Tecra 8100 - dmesg is on
| > >wiki) it worked fine there with a pre-branch -current. It barely boots now
| > >(and I've wiped the disk) but I can set it up for testing with a bunch of
| > >kernels on CD if required.
| > >
| > >> plunky%rya-online.net@localhost said:
| > >> >        RBUS_IO_BASE = 0x4000   # default, does not work
| > >> >        RBUS_IO_BASE = 0x0a00   # works.
| > >>
| > >> Perhaps there is some problem with incomplete decoding?
| > >> The i386 CARDBUS config file suggests 0x600 btw.
| > >
| > >32-bit cards seem to work normally with GENERIC anyway and I tried that
| > >CARDBUS config but found no change with the 16-bit ones.
| > >
| > >I took the 0x0a00 value from PRs kern/32327 and kern/32328 (and, all over
| > >the internet when searching for RBUS_IO_BASE :), which also suggests
| > >reducing RBUS_IO_SIZE and that worked fine for the 16-bit cards but both
| > >32-bit cards paniced the machine in that case (from "cbb0: <rbus> no bus
| > >space")
| > >
| > 
| > Well, I think that 0x4000 used to work before the rbus cleanup that
| > Andy did.  Perhaps a bug sneaked in then?
| 
| I haven't touched rbus in 8+ years.

I meant this one and the switch to using sys/bus.h in rbus.c...
This is around the time it broke for me.

christos

  ------------------------------------------------------------------------
  RCS file: /s/NetBSD/cvsroot/src/sys/arch/x86/include/bus.h,v
  Working file: bus.h
  head: 1.17
  [...]
  total revisions: 45;  selected revisions: 1
  description:
  ----------------------------
  revision 1.14
  date: 2007/09/26 19:48:38;  author: ad;  state: Exp;  lines: +10 -1054
  x86 changes for pcc and LKMs.

  - Replace most inline assembly with proper functions. As a side effect
    this reduces the size of amd64 GENERIC by about 120kB, and i386 by a
    smaller amount. Nearly all of the inlines did something slow, or something
    that does not need to be fast.
  - Make curcpu() and curlwp functions proper, unless __GNUC__ && _KERNEL.
    In that case make them inlines. Makes curlwp LKM and preemption safe.
  - Make bus_space and bus_dma more LKM friendly.
  - Share a few more files between the ports.
  - Other minor changes.