[snip]
Unfortunately, IDE hardware usually doesn't have an "interrupt pending"
register which we can use in the driver to know that a drive is requesting
service (such a register is present on some IDE interfaces on the Amiga).
This introduces difficulty in correctly handling spurious, yet expected,
"interrupt noise" from within the IDE driver -- we will need the help
of allow_interrupts().
I will try to create a patch for this problem in the next few days.
I have an idea for removing the present IRR code entirely from my deadlock
patch, so this may remove the problem entirely.
Leonard