On 1/7/25 23:36, Peter Maydell wrote:
On Tue, 7 Jan 2025 at 17:55, Cédric Le Goater <c...@kaod.org> wrote:
Hello,
I would not recommend using qemu_irq_invert() in new code.
I guess in an ideal world we'd implement a QOM object
that encapsulated the the "not gate" logic, similar to
TYPE_OR_IRQ. (Though for TYPE_OR_IRQ we made the mistake
of making it inherit from TYPE_DEVICE, not TYPE_SYSBUS_DEVICE,
so it doesn't get reset properly on system reset and so
the "what happens to the output on reset" is still not
really correct.)
I see how this would work with TYPE_PL181 model but I don't
understand how this could work with TYPE_SYSBUS_SDHCI since
we don't have a gpio line to invert the level. Am I missing
something ?
You have a gpio line, i.e. a qemu_irq, because you're passing
it to qemu_irq_invert(),
This part was proposed by Philippe and is not related to the
proposal.
I am not very familiar with these devices and AFAUI, the
SHDCIState irq being modified has a much broader scope than
a write protection toggle/level. I suppose the irq covers all
kind of transfers.
and you could instead connect it into
a hypothetical TYPE_NOT_IRQ device. qemu_irq are GPIO lines,
the type just has an odd name for historical reasons.
Yes. I was looking for something like this but we don't have
a "card-read-only" GPIO line like in TYPE_PL181. Hence my
confusion.
Thanks,
C.
