These bits were intruced in Linux-3.3, so don't refer to them.
Rather, err out on all unimplemented bits.  This is a different
patch from what David suggested.
Reported-by: Juan David Gonzalez Cobas <dcobas@cern.ch>
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

This is not increasing the version number, because I plan to merge
before splitting off the fmc project.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

The code was checking the first LM32 instruction, but it used wrong
endianness.

This fixes the check, but the fix uncovered a bug: the "wrc=1"
parameter was only loading the first card. This fixes the bug
as well but not overwriting the user-provided parameter.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

When handling the interrupt, mask the pending bits with the not-empty
mask, otherwise an unexpected trigger-ready status may overflow the
for loop and eventually oops to panic.

Meanwhile, use a symbolic name for the mask and be more aggressive in
disabling (disable all bits: ho harm done for unimplemented interrupt
sources)
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

This is not really a problem if we have edge-triggered interrupts as
input to the Gennum, but will lock the computer on unload when
experimenting with level-triggered interrupts.

We haven't been able to run Message Signalled Interrupt
reliably, for a quirck in the Gennum chip and the kernel
insisting to configure it according to the standard. So this
adds "use_msi" as a module parameter, disabled by default.
See documentation for details.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>