Module Name: src
Committed By: brad
Date: Fri Nov 10 13:17:17 UTC 2023
Modified Files:
src/sys/dev/gpio: gpioirq.c
Log Message:
For /dev/ reads against gpioirq(4) implement the following:
o O_NONBLOCK on reads
o Add a d_poll function and associated sel[init|notify|record|destroy]
calls to the driver so that select(2) and poll(2) work as expected.
With these in place async use cases work against /dev/gpioirqN
To generate a diff of this commit:
cvs rdiff -u -r1.2 -r1.3 src/sys/dev/gpio/gpioirq.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
Modified files:
Index: src/sys/dev/gpio/gpioirq.c
diff -u src/sys/dev/gpio/gpioirq.c:1.2 src/sys/dev/gpio/gpioirq.c:1.3
--- src/sys/dev/gpio/gpioirq.c:1.2 Mon Nov 6 00:35:05 2023
+++ src/sys/dev/gpio/gpioirq.c Fri Nov 10 13:17:17 2023
@@ -1,4 +1,4 @@
-/* $NetBSD: gpioirq.c,v 1.2 2023/11/06 00:35:05 brad Exp $ */
+/* $NetBSD: gpioirq.c,v 1.3 2023/11/10 13:17:17 brad Exp $ */
/*
* Copyright (c) 2016, 2023 Brad Spencer <b...@anduin.eldar.org>
@@ -17,7 +17,7 @@
*/
#include <sys/cdefs.h>
-__KERNEL_RCSID(0, "$NetBSD: gpioirq.c,v 1.2 2023/11/06 00:35:05 brad Exp $");
+__KERNEL_RCSID(0, "$NetBSD: gpioirq.c,v 1.3 2023/11/10 13:17:17 brad Exp $");
/*
* GPIO driver that uses interrupts and can send that fact to userland.
@@ -34,6 +34,9 @@ __KERNEL_RCSID(0, "$NetBSD: gpioirq.c,v
#include <sys/pool.h>
#include <sys/kmem.h>
#include <sys/condvar.h>
+#include <sys/vnode.h>
+#include <sys/select.h>
+#include <sys/poll.h>
#include <dev/gpio/gpiovar.h>
@@ -67,6 +70,7 @@ struct gpioirq_softc {
kcondvar_t sc_cond_dying;
pool_cache_t sc_readpool;
char *sc_readpoolname;
+ struct selinfo sc_rsel;
SIMPLEQ_HEAD(,gpioirq_read_q) sc_read_queue;
};
@@ -97,6 +101,7 @@ extern struct cfdriver gpioirq_cd;
static dev_type_open(gpioirq_open);
static dev_type_read(gpioirq_read);
static dev_type_close(gpioirq_close);
+static dev_type_poll(gpioirq_poll);
const struct cdevsw gpioirq_cdevsw = {
.d_open = gpioirq_open,
.d_close = gpioirq_close,
@@ -105,7 +110,7 @@ const struct cdevsw gpioirq_cdevsw = {
.d_ioctl = noioctl,
.d_stop = nostop,
.d_tty = notty,
- .d_poll = nopoll,
+ .d_poll = gpioirq_poll,
.d_mmap = nommap,
.d_kqfilter = nokqfilter,
.d_discard = nodiscard,
@@ -189,6 +194,7 @@ gpioirq_attach(device_t parent, device_t
sc->sc_readpool = pool_cache_init(sizeof(struct gpioirq_read_q),0,0,0,sc->sc_readpoolname,NULL,IPL_VM,NULL,NULL,NULL);
pool_cache_sethiwat(sc->sc_readpool,100);
SIMPLEQ_INIT(&sc->sc_read_queue);
+ selinit(&sc->sc_rsel);
for(int apin = 0; apin < sc->sc_npins; apin++) {
if (!gpio_intr_str(sc->sc_gpio, &sc->sc_map, apin, irqmode,
@@ -263,6 +269,7 @@ gpioirq_intr(void *arg)
q->parentunit = is->i_parentunit;
q->theval = val;
SIMPLEQ_INSERT_TAIL(&sc->sc_read_queue,q,read_q);
+ selnotify(&sc->sc_rsel, POLLIN|POLLRDNORM, NOTE_SUBMIT);
cv_signal(&sc->sc_condreadready);
} else {
aprint_error("Could not allocate memory for read pool\n");
@@ -304,6 +311,10 @@ gpioirq_read(dev_t dev, struct uio *uio,
if (!sc)
return (ENXIO);
+ if (sc->sc_dying) {
+ return EIO;
+ }
+
while (uio->uio_resid > 0) {
any = 0;
error = 0;
@@ -316,7 +327,11 @@ gpioirq_read(dev_t dev, struct uio *uio,
any = 1;
break;
} else {
- error = cv_wait_sig(&sc->sc_condreadready,&sc->sc_read_mutex);
+ if (flags & IO_NDELAY) {
+ error = EWOULDBLOCK;
+ } else {
+ error = cv_wait_sig(&sc->sc_condreadready,&sc->sc_read_mutex);
+ }
if (sc->sc_dying)
error = EIO;
if (error == 0)
@@ -358,6 +373,10 @@ gpioirq_close(dev_t dev, int flags, int
sc = device_lookup_private(&gpioirq_cd, minor(dev));
+ if (sc->sc_dying) {
+ return(0);
+ }
+
mutex_enter(&sc->sc_lock);
while ((q = SIMPLEQ_FIRST(&sc->sc_read_queue)) != NULL) {
SIMPLEQ_REMOVE_HEAD(&sc->sc_read_queue, read_q);
@@ -369,6 +388,31 @@ gpioirq_close(dev_t dev, int flags, int
return(0);
}
+static int
+gpioirq_poll(dev_t dev, int events, struct lwp *l)
+{
+ struct gpioirq_softc *sc;
+ int revents = 0;
+
+ sc = device_lookup_private(&gpioirq_cd, minor(dev));
+
+ mutex_enter(&sc->sc_read_mutex);
+ if (sc->sc_dying) {
+ mutex_exit(&sc->sc_read_mutex);
+ return POLLHUP;
+ }
+
+ if ((events & (POLLIN | POLLRDNORM)) != 0) {
+ if (!SIMPLEQ_EMPTY(&sc->sc_read_queue))
+ revents |= events & (POLLIN | POLLRDNORM);
+ else
+ selrecord(l, &sc->sc_rsel);
+ }
+
+ mutex_exit(&sc->sc_read_mutex);
+ return revents;
+}
+
int
gpioirq_detach(device_t self, int flags)
{
@@ -413,6 +457,7 @@ gpioirq_detach(device_t self, int flags)
mutex_destroy(&sc->sc_read_mutex);
mutex_destroy(&sc->sc_lock);
+ seldestroy(&sc->sc_rsel);
return (0);
}
