periph/conn

WaitForEdge Fails on Rasperry Pi Kernel 6.6.16-v8+

Closed this issue · 2 comments

Describe the bug
Using the specified kernel, a call go WaitForEdge immediately unblocks and returns.

To Reproduce
Steps to reproduce the behavior:

  1. Run program
package main

import (
        "log"
        "periph.io/x/conn/v3/driver/driverreg"
        "periph.io/x/conn/v3/gpio"
        "periph.io/x/conn/v3/gpio/gpioreg"
        "periph.io/x/host/v3"
)

func main() {
    _, err:=host.Init()
    if err!=nil {
        panic(err)
    }
    _, err = driverreg.Init()
    if err!=nil {
        panic(err)
    }
    pin := gpioreg.ByName("GPIO19")
    pin.In(gpio.PullUp, gpio.BothEdges)
    for  {
        edgeReceived := pin.WaitForEdge(-1)
        if edgeReceived {
            log.Print("Edge Received., State=", pin.Read())
        } else {
            log.Print("No Edge Received")
        }
    }
}
  1. Run it.
  2. Note the output continuously scrolls "No Edge Received".

Expected behavior
The program should have blocked until the button was pressed.

Platform (please complete the following information):

  • OS: Raspbian Bookworm
  • Board: Pi 4 B 4GB
  • Kernel: 6.6.16-v8+

Additional context
The program has been working for several years. After a kernel update, it stopped working. Experimentation shows that reading the state of the button does work.

I've searched around, and found some references that this might be related to deprecation of /sys/class/gpio?

FWIW, I tried python and RPi.GPIO and it sees the button press as well, but attempt to add event for edge detection fails.

I tried running my test program under strace and got:

...
futex(0x200d00, FUTEX_WAIT_PRIVATE, 0, NULL) = 0
futex(0x200d00, FUTEX_WAIT_PRIVATE, 0, NULL) = -1 EAGAIN (Resource temporarily unavailable)
nanosleep({tv_sec=0, tv_nsec=3000}, NULL) = 0
futex(0x4000100148, FUTEX_WAKE_PRIVATE, 1) = 1
futex(0x4000100148, FUTEX_WAKE_PRIVATE, 1) = 1
newfstatat(AT_FDCWD, "/sys/bus/platform/drivers/raspberrypi-firmware", {st_mode=S_IFDIR|0755, st_size=0, ...}, 0) = 0
openat(AT_FDCWD, "/proc/device-tree/system/linux,revision", O_RDONLY|O_CLOEXEC) = 7
fcntl(7, F_GETFL)                       = 0x20000 (flags O_RDONLY|O_LARGEFILE)
fcntl(7, F_SETFL, O_RDONLY|O_NONBLOCK|O_LARGEFILE) = 0
epoll_ctl(4, EPOLL_CTL_ADD, 7, {events=EPOLLIN|EPOLLOUT|EPOLLRDHUP|EPOLLET, data={u32=223870981, u64=35813851309539333}}) = 0
fstat(7, {st_mode=S_IFREG|0444, st_size=4, ...}) = 0
read(7, "\0\3001\22", 512)              = 4
read(7, "", 508)                        = 0
epoll_ctl(4, EPOLL_CTL_DEL, 7, 0x4000064c80) = 0
close(7)                                = 0
futex(0x4000080148, FUTEX_WAKE_PRIVATE, 1) = 1
futex(0x4000100148, FUTEX_WAKE_PRIVATE, 1) = 1
openat(AT_FDCWD, "/etc/localtime", O_RDONLY) = 7
read(7, "TZif2\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\6\0\0\0\6\0\0\0\0"..., 4096) = 2460
read(7, "", 4096)                       = 0
close(7)                                = 0
write(2, "2024/02/12 12:44:38 No Edge Rece"..., 372024/02/12 12:44:38 No Edge Received
) = 37
futex(0x4000100148, FUTEX_WAKE_PRIVATE, 1) = 1
futex(0x200d00, FUTEX_WAIT_PRIVATE, 0, NULL) = 0
2024/02/12 12:44:39 No Edge Received

Dang, sorry to hear. I fear someone will have to do the migration. Since the driver is in host, not in conn, it's tracked at periph/host#39.

Closing this because periph/host#59 resolves it.