[PATCH 0/1] [RFC][NOT WORKING] amiga: pcmcia: inb()/inw() on GAYLE_IO space

[Paolo Pisati <p.pisati@gmail.com>]

Hi there,

attached is a driver for the pcmcia side of C='s gayle - it's a WIP, as such
there are many dev_dbg()/pr_err()/pr_dbg() scattered around, you've been warned.

It recognize all pcmcia cards i tried (IOW gayle's attribute space is read and
parsed correctly), socket's interrupt work (IOW you can plug & unplug cards),
but *it seems* any I/O read operation return bogus data (e.g. 0xff or 0xfff).

Here's a dmesg dealing with with a 3Com 3c589 ethernet card:
http://dpaste.com//84TPTPLNV - notice the "hw_addr ff:ff:ff:ff:ff:ff" or
"exiting interrupt, status ffff".

AFAIK inb()/inw() already add the GAYLE_IO offset (
AG_ISA_IO_B and AG_ISA_IO_W in arch/m68k/include/asm/io_mm.h), and i didn't know
if i could read directly (without an ioremap() first), so i tried both cases:

1) ioreamp() sock->phys_io, and subtract the GAYLE_IO offset (see
sock->socket.io_offset).

or 

2) set Amiga base IO port to 0x0 and patch
drivers/pcmcia/rsrc_mgr.c::static_find_io() (see https://dpaste.com//B75LTFF3V)

In both case the 3c589_cs driver request an I/O aperture at 0x0300, large
0x0010, receives it, but all read operations return 0xff or 0xffff.

Any idea? Any guess is welcome since i'm running out of ideas ATM.

Paolo Pisati (1):
  pcmcia: gayle: initial support

 arch/m68k/amiga/pcmcia.c          |   5 +
 arch/m68k/amiga/platform.c        |  29 +++
 arch/m68k/include/asm/amipcmcia.h |  11 ++
 drivers/pcmcia/Kconfig            |   8 +
 drivers/pcmcia/Makefile           |   1 +
 drivers/pcmcia/gayle.c            | 307 ++++++++++++++++++++++++++++++
 6 files changed, 361 insertions(+)
 create mode 100644 drivers/pcmcia/gayle.c

-- 
2.34.1