From: Greg KH <greg@kroah.com>
To: Linux Kernel List <linux-kernel@vger.kernel.org>,
Patrick Mochel <mochel@osdl.org>,
Ivan Kokshaysky <ink@jurassic.park.msu.ru>,
Jeff Garzik <jgarzik@pobox.com>,
Rusty Russell <rusty@rustcorp.com.au>
Subject: Re: PCI driver module unload race?
Date: Mon, 10 Mar 2003 13:44:43 -0800 [thread overview]
Message-ID: <20030310214443.GA13145@kroah.com> (raw)
In-Reply-To: <20030308202101.GA26831@kroah.com>
On Sat, Mar 08, 2003 at 12:21:01PM -0800, Greg KH wrote:
> On Sat, Mar 08, 2003 at 08:09:43PM +0000, Russell King wrote:
> > On Sat, Mar 08, 2003 at 11:12:37AM -0800, Greg KH wrote:
> > > On Sat, Mar 08, 2003 at 10:47:49AM +0000, Russell King wrote:
> > > > Hi,
> > > >
> > > > What prevents the following scenario from happening? It's purely
> > > > theoretical - I haven't seen this occuring.
> > > >
> > > > - Load PCI driver.
> > > >
> > > > - PCI driver registers using pci_module_init(), and adds itself to sysfs.
> > > >
> > > > - Hot-plugin a PCI device which uses this driver. sysfs matches the PCI
> > > > driver, and calls the PCI drivers probe function.
> > >
> > > Ugh, yes you are correct, I can't believe I missed this before.
> > >
> > > How does this patch look?
> >
> > Hrm, I'm wondering whether this should be part of the device model
> > infrastructure. After all, surely every subsystems device driver
> > which could be a module would need this to prevent unload races?
>
> Very good point, I can see myself duplicating this logic for every
> subsystem :)
>
> I'll look into moving this into the driver core later today.
Ok, how about this patch? It boots for me :)
If no one has any complaints, I'll work on moving the USB core to using
this pointer instead of its own in struct usb_driver.
thanks,
greg k-h
# Driver core: add module owner to struct device_driver
diff -Nru a/drivers/base/bus.c b/drivers/base/bus.c
--- a/drivers/base/bus.c Mon Mar 10 13:52:15 2003
+++ b/drivers/base/bus.c Mon Mar 10 13:52:15 2003
@@ -263,14 +263,25 @@
if (dev->bus->match(dev,drv)) {
dev->driver = drv;
if (drv->probe) {
- if ((error = drv->probe(dev))) {
- dev->driver = NULL;
- return error;
+ if (!try_module_get(drv->owner)) {
+ dev_err(dev,
+ "Can't get a module reference for %s\n",
+ drv->name);
+ goto exit;
}
+
+ if ((error = drv->probe(dev)))
+ dev->driver = NULL;
+
+ module_put(drv->owner);
+
+ if (error)
+ goto exit;
}
device_bind_driver(dev);
error = 0;
}
+exit:
return error;
}
@@ -350,8 +361,16 @@
sysfs_remove_link(&drv->kobj,dev->kobj.name);
list_del_init(&dev->driver_list);
devclass_remove_device(dev);
- if (drv->remove)
- drv->remove(dev);
+ if (drv->remove) {
+ if (!try_module_get(drv->owner)) {
+ dev_err(dev,
+ "Can't get a module reference for %s\n",
+ drv->name);
+ } else {
+ drv->remove(dev);
+ module_put(drv->owner);
+ }
+ }
dev->driver = NULL;
}
}
diff -Nru a/drivers/base/power.c b/drivers/base/power.c
--- a/drivers/base/power.c Mon Mar 10 13:52:15 2003
+++ b/drivers/base/power.c Mon Mar 10 13:52:15 2003
@@ -41,10 +41,17 @@
list_for_each(node,&devices_subsys.kset.list) {
struct device * dev = to_dev(node);
if (dev->driver && dev->driver->suspend) {
- pr_debug("suspending device %s\n",dev->name);
- error = dev->driver->suspend(dev,state,level);
- if (error)
- printk(KERN_ERR "%s: suspend returned %d\n",dev->name,error);
+ if (!try_module_get(dev->driver->owner)) {
+ dev_err(dev,
+ "Can't get a module reference for %s\n",
+ dev->driver->name);
+ } else {
+ pr_debug("suspending device %s\n",dev->name);
+ error = dev->driver->suspend(dev,state,level);
+ if (error)
+ printk(KERN_ERR "%s: suspend returned %d\n",dev->name,error);
+ module_put(dev->driver->owner);
+ }
}
}
up_write(&devices_subsys.rwsem);
@@ -67,8 +74,15 @@
list_for_each_prev(node,&devices_subsys.kset.list) {
struct device * dev = to_dev(node);
if (dev->driver && dev->driver->resume) {
- pr_debug("resuming device %s\n",dev->name);
- dev->driver->resume(dev,level);
+ if (!try_module_get(dev->driver->owner)) {
+ dev_err(dev,
+ "Can't get a module reference for %s\n",
+ dev->driver->name);
+ } else {
+ pr_debug("resuming device %s\n",dev->name);
+ dev->driver->resume(dev,level);
+ module_put(dev->driver->owner);
+ }
}
}
up_write(&devices_subsys.rwsem);
@@ -89,8 +103,15 @@
list_for_each(entry,&devices_subsys.kset.list) {
struct device * dev = to_dev(entry);
if (dev->driver && dev->driver->shutdown) {
- pr_debug("shutting down %s\n",dev->name);
- dev->driver->shutdown(dev);
+ if (!try_module_get(dev->driver->owner)) {
+ dev_err(dev,
+ "Can't get a module reference for %s\n",
+ dev->driver->name);
+ } else {
+ pr_debug("shutting down %s\n",dev->name);
+ dev->driver->shutdown(dev);
+ module_put(dev->driver->owner);
+ }
}
}
up_write(&devices_subsys.rwsem);
diff -Nru a/include/linux/device.h b/include/linux/device.h
--- a/include/linux/device.h Mon Mar 10 13:52:15 2003
+++ b/include/linux/device.h Mon Mar 10 13:52:15 2003
@@ -112,6 +112,7 @@
extern void bus_remove_file(struct bus_type *, struct bus_attribute *);
struct device_driver {
+ struct module * owner;
char * name;
struct bus_type * bus;
struct device_class * devclass;
next prev parent reply other threads:[~2003-03-10 21:44 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2003-03-08 10:47 PCI driver module unload race? Russell King
2003-03-08 19:12 ` Greg KH
2003-03-08 19:47 ` Petr Vandrovec
2003-03-08 19:51 ` Greg KH
2003-03-09 2:33 ` Petr Vandrovec
2003-03-08 20:03 ` Russell King
2003-03-08 20:09 ` Russell King
2003-03-08 20:21 ` Greg KH
2003-03-10 21:44 ` Greg KH [this message]
2003-03-10 23:48 ` Oliver Neukum
2003-03-10 23:51 ` Greg KH
2003-03-11 1:04 ` Roman Zippel
2003-03-11 1:15 ` Greg KH
2003-03-11 9:00 ` Oliver Neukum
2003-03-11 15:06 ` Patrick Mochel
2003-03-11 16:07 ` Oliver Neukum
2003-03-16 13:13 ` Rusty Russell
2003-03-11 11:05 ` Roman Zippel
2003-03-11 15:27 ` Patrick Mochel
2003-03-11 20:09 ` Roman Zippel
2003-03-11 19:15 ` Patrick Mochel
2003-03-12 2:28 ` Roman Zippel
2003-03-16 13:05 ` Rusty Russell
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20030310214443.GA13145@kroah.com \
--to=greg@kroah.com \
--cc=ink@jurassic.park.msu.ru \
--cc=jgarzik@pobox.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mochel@osdl.org \
--cc=rusty@rustcorp.com.au \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.