Date: Tue, 13 May 2008 22:09:31 +0300 From: Andriy Gapon <avg@icyb.net.ua> To: Kostik Belousov <kostikbel@gmail.com> Cc: freebsd-hackers@freebsd.org Subject: Re: devctl (alike?) for devfs Message-ID: <4829E76B.8010204@icyb.net.ua> In-Reply-To: <4829E658.3000605@icyb.net.ua> References: <480E4269.2090604@icyb.net.ua> <480FBAB9.1000904@icyb.net.ua> <48103F36.6060707@icyb.net.ua> <200804240811.26183.jhb@freebsd.org> <4810FD1E.70602@icyb.net.ua> <20080425095009.GD18958@deviant.kiev.zoral.com.ua> <4811E6BC.4060306@icyb.net.ua> <20080425143646.GF18958@deviant.kiev.zoral.com.ua> <48275C0C.2040601@icyb.net.ua> <20080511214833.GB18958@deviant.kiev.zoral.com.ua> <4829E658.3000605@icyb.net.ua>
next in thread | previous in thread | raw e-mail | index | archive | help
This is a multi-part message in MIME format.
--------------000802000206000606030404
Content-Type: text/plain; charset=KOI8-U; format=flowed
Content-Transfer-Encoding: 7bit
on 13/05/2008 22:04 Andriy Gapon said the following:
> on 12/05/2008 00:48 Kostik Belousov said the following:
>> No, we do not have a leak, but we have somewhat non-obvious behaviour.
>>
>> The cdev structure is freed only after the last reference to cdev is
>> gone. Typical holder of the reference is the devfs vnode. In the normal
>> usage, the vnode is present until both the device is destroyed _and_
>> devfs_populate_loop() run is performed. This function actually reclaim
>> the vnodes for destroyed devices, that causes last reference to cdev to
>> be dropped and memory freed.
>>
>> The populate loop is called syncronously from the upper levels. The
>> easiest method to trigger it is to do ls /dev, since it is called from
>> the devfs_lookupx().
>
> Thank you for the explanation! ls did trigger DESTROY notifications.
> But this arbitrary delay between device entry going away and
> notification about about it is a little bit "uncool".
>
> So I re-wrote the patch to post notifications about deleted devices with
> si_refcount > 0 in a fashion similar to how si_refcount=0 devices are
> freed. I put those cdev-s onto a special list (re-/ab-using si_siblings
> LIST link field) and bump their si_refcount to prevent them from getting
> destroyed before the notification is sent (it would need si_name).
> Then, in dev_unlock_and_free() I send notifications and call dev_rel on
> the devices.
>
> I am not entirely satisfied with the code:
> 1. I don't like the way I move LIST elements from one head to the other,
> this should be a macro in queue.h
> 2. I am not sure about the names I picked
> 3. I slightly don't like a fact that parent-child destroy notifications
> are sent in reverse order because of my simplistic LIST usage.
>
Oops! I attached the older patch.
New patch is here.
--
Andriy Gapon
--------------000802000206000606030404
Content-Type: text/plain;
name="devfs-notify.diff"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="devfs-notify.diff"
diff --git a/sys/kern/kern_conf.c b/sys/kern/kern_conf.c
index 1db25f8..7106182 100644
--- a/sys/kern/kern_conf.c
+++ b/sys/kern/kern_conf.c
@@ -30,6 +30,7 @@
#include <sys/param.h>
#include <sys/kernel.h>
#include <sys/systm.h>
+#include <sys/bus.h>
#include <sys/bio.h>
#include <sys/lock.h>
#include <sys/mutex.h>
@@ -63,6 +64,8 @@ static struct cdev_priv_list cdevp_free_list =
TAILQ_HEAD_INITIALIZER(cdevp_free_list);
static SLIST_HEAD(free_cdevsw, cdevsw) cdevsw_gt_post_list =
SLIST_HEAD_INITIALIZER();
+static LIST_HEAD(cdev_notif, cdev) cdev_notif_list =
+ LIST_HEAD_INITIALIZER(cdev_notif_list);
void
dev_lock(void)
@@ -82,8 +85,10 @@ dev_unlock_and_free(void)
{
struct cdev_priv_list cdp_free;
struct free_cdevsw csw_free;
+ struct cdev_notif cd_notif;
struct cdev_priv *cdp;
struct cdevsw *csw;
+ struct cdev *cd;
mtx_assert(&devmtx, MA_OWNED);
@@ -95,10 +100,23 @@ dev_unlock_and_free(void)
TAILQ_CONCAT(&cdp_free, &cdevp_free_list, cdp_list);
csw_free = cdevsw_gt_post_list;
SLIST_INIT(&cdevsw_gt_post_list);
+ /* XXX How to do the following nicely? */
+ cd_notif = cdev_notif_list;
+ if (!LIST_EMPTY(&cd_notif))
+ LIST_FIRST(&cd_notif)->si_siblings.le_prev = &LIST_FIRST(&cd_notif);
+ LIST_INIT(&cdev_notif_list);
mtx_unlock(&devmtx);
+ while ((cd = LIST_FIRST(&cd_notif)) != NULL) {
+ if (!cold)
+ devctl_notify("DEVFS", cd->si_name, "DESTROY", NULL);
+ LIST_REMOVE(cd, si_siblings);
+ dev_rel(cd);
+ }
while ((cdp = TAILQ_FIRST(&cdp_free)) != NULL) {
+ if (!cold)
+ devctl_notify("DEVFS", cdp->cdp_c.si_name, "DESTROY", NULL);
TAILQ_REMOVE(&cdp_free, cdp, cdp_list);
devfs_free(&cdp->cdp_c);
}
@@ -706,6 +724,10 @@ make_dev_credv(int flags, struct cdevsw *devsw, int minornr,
devfs_create(dev);
clean_unrhdrl(devfs_inos);
dev_unlock_and_free();
+
+ if (!cold)
+ devctl_notify("DEVFS", dev->si_name, "CREATE", NULL);
+
return (dev);
}
@@ -794,6 +816,10 @@ make_dev_alias(struct cdev *pdev, const char *fmt, ...)
clean_unrhdrl(devfs_inos);
dev_unlock();
dev_depends(pdev, dev);
+
+ if (!cold)
+ devctl_notify("DEVFS", dev->si_name, "CREATE", NULL);
+
return (dev);
}
@@ -861,6 +887,9 @@ destroy_devl(struct cdev *dev)
if (dev->si_refcount > 0) {
LIST_INSERT_HEAD(&dead_cdevsw.d_devs, dev, si_list);
+ /* (re-/ab-)use si_siblings for destroy notification list */
+ LIST_INSERT_HEAD(&cdev_notif_list, dev, si_siblings);
+ dev_refl(dev); /* Avoid race with dev_rel() */
} else {
dev_free_devlocked(dev);
}
--------------000802000206000606030404--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4829E76B.8010204>