Hello Sulev-Madis,

What's your target platform ?

I recently discovered that GPIO interrupts are not supported in "GPIO Bus" driver on x86 platforms because of lack of INTRNG support (simply not implemented). I found a 6-year old diff D26407 from avg@ to add this functionally which stuck on review because of some unsolved issues. I renewed avg's patch so it could be applied to -current and posted for review (D57968) again, but it did get much traction too. So, if you are interested in getting GPIO interrupts on x86/amd64 - no luck till gpiobus driver is "fixed". No ideas about arm and others.

---
Regards,
Ruslan

On 2026-08-16 04:09, Sulev-Madis Silber wrote:
i thought i saw message somewhere here but i lost it

google isn't helping much either


<https://github.com/cyclaero/shutdd>

<https://reviews.freebsd.org/D52102>


but what's current state of gpio interrupts?

i need simple interface, like a util that i can run which then starts
printing events on stdout which i can then pipe into shell or perl
that starts doing actions on this

could even devd?

currently there's nothing in base nor ports. everyone needing this
function would need to read manpages, googling and so on. and then
come up with c code

i once did that too

but i don't want to forge hammer each time i want to bang a nail into
wall! i just want to grab hammer, and bang, nail is in

that needs to be solved somehow

io output control is bit problematic too

but i was able to take old gonzoua wrappers for python/ruby/perl. take
perl part of it. it still works. it seems like bad test code from 2013

for output, it would also be good to run something that accepts things
from stdin

so it would be like "gpio-input | program | gpio-output"

or it would fork those as children inside

anything that could make my life easier

people in linux have probably understood how hard is to write code so
they exposed all of this as virtual "files" or devices

unsure what could be done here

this is 12 year old issue actually

i probably, with a ton of cursing, can come up with something and then
publish it so noone else would ask it again

however there must be many much better c coders here?

possibly i would want it in base but port will do. considering
maintaining it too. but i find c so hard

seems like speed is not really an issue here, considering i did

<http://ketas.si.pri.ee/misc/ws2812/ws2812.html>

which does realtime animations with perl/ssh/spi util (which i
patched) piped together

it took massive effort from me

so i'm looking ways to minimize this effort for people who want to script on hw

Reply via email to