Statically built yenta does not work anymore (in FC5)

[Pete Zaitcev]

Hi, guys:

I think I've read about this somewhere, but cannot find it anymore.
Also, the "static vs. dynamic" as related to bridge types confuses
search engines...

This happened with Fedora kernel 2.6.13-1.1483_FC5 originally. I have
poked with stock 2.6.13-rc6 to verify that it's not Fedora specific.
If I build yenta_socket into kernel statically, the probing does not
happen. The dmesg looks like this:

ACPI: PCI Interrupt 0000:00:0c.0[A] -> Link [LNKB] -> GSI 9 (level, low) -> IRQ
9
Yenta: CardBus bridge found at 0000:00:0c.0 [104d:8082]
Yenta: ISA IRQ mask 0x0cb8, PCI irq 9
Socket status: 30000006
cs: : pcmcia_register_socket(0xc0398600)
cs: pcmcia_socket0: parse_events: events 00000080

In Yenta is loaded as a module, it looks like this (lesser debug level):

ACPI: PCI Interrupt 0000:00:0c.0[A] -> Link [LNKB] -> GSI 9 (level, low) -> IRQ 9
Yenta: CardBus bridge found at 0000:00:0c.0 [104d:8082]
Yenta: ISA IRQ mask 0x0cb8, PCI irq 9
Socket status: 30000006
cs: IO port probe 0x3d4-0x4ff: excluding 0x4cc-0x4d3
cs: IO port probe 0x3c0-0x3d2: clean.
cs: IO port probe 0x100-0x3af: excluding 0x170-0x177 0x370-0x37f 0x398-0x39f
cs: IO port probe 0xc00-0xcff: clean.
cs: IO port probe 0x800-0x8ff: clean.
cs: IO port probe 0xa00-0xaff: clean.

What is the theory of operations here? What sequence of events is
supposed to happen between pcmcia_register_socket and the do_io_probe()?

In the old PCMCIA, cardmgr would invoke the "adjust" ioctl, but we
got rid of that. So... I need to know the expected sequence of events
to understand where the chain breaks down.

Fedora Core Rawhide builds yenta_socket statically for some reason.
But even if DaveJ changes that, I'd like to know the theory behind this.

-- Pete