Tomasz Wlostowski authored 12 years ago and Alessandro Rubini committed 12 years ago

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>

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 redoes part of
   0dcf4ed3

 kernel/spec-fmc: only enable irq1 from fpga
that I reverted. With this patch I only accept the two interrupt
lines from the FPGA (even if only one is being used),
ignoring the test points.

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 reverts commit 0dcf4ed3.

After this I started having irq problems (missing events) so I'm not sure
it was the right thing to do. Try reverting as a quick solution for me.

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 period pulse on a channel was designed as a generic "mirror one channel
to another with a predefined delay".

However, on different boards self-stamping has different delays, so
if we are mirroring to ourselves, add the delay to the previous setpoint
instead of using the timestamp we got, so the 8ns difference doesn't
accumulate over time.

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

The Gennum manual doesn't say what is the miminum glitch for it to
detect the edge, but this appears to work. I can't be sure, though.

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

There's the risk of getting spurious interrupts if the test points are
not connected as I use them, and the same may happen to the unused irq0
pin from the fpga

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>