No. That's an old (but unfixed) issue. Same general topic though - ref. counting.
The problem is that Al Viro removed a chunk of reference counting
code from i2c-proc in kernel 2.5.18;
and we are trying to figure out why and how to get the feature back in a
way that is satisfactory.
phil at netroedge.com wrote:
>
> I'm a little confused by the thread, so excuse me if I'm commenting on
> a dead issue:
>
> I think the issue with i2c-proc incrementing it's counter was that it
> was possible to cause an 'oops' when modules were inserted and removed
> in particular ways. I.e., something like this caused an oops:
>
> insmod i2c-proc
> insmod eeprom
> cd /proc/sys/dev/sensors/eeprom-0-50
> cat *
> rmmod i2c-proc
> cat *
>
> oops!
>
> Does this help?
>
> Phil
>
> On Fri, Sep 06, 2002 at 01:11:44PM -0400, Mark Studebaker wrote:
> > Don't know - anybody?
> >
> > Alexander Viro wrote:
> > >
> > > On Thu, 5 Sep 2002, Mark D. Studebaker wrote:
> > >
> > > > removed from i2c-proc.c. As the comments say, the code prevents module
> > > > unloading
> > > > if the /proc files are in use.
> > >
> > > So does setting ->owner on them.
> > >
> > > > Why is this a bad thing?
> > >
> > > Because module has no business messing with its own reference count.
> > >
> > > > Is there a better way to do it?
> > >
> > > See above.
> >
> > Does Al mean we already do the right thing in i2c-dev.c:89
> > static struct file_operations i2cdev_fops {
> > owner: THIS_MODULE,
> >
> > and i2c-dev.c:495 int __init i2c_dev_init
> > ...
> > if (register_chrdev(I2C_MAJOR, "i2c", &i2cdev_fops))
> >
> > If so, we need to find a method to verify his statement.
> > Oh, Linus applied the reverse patch for this to 2.5.34.
> > Albert
> > --
> > Albert Cranford Deerfield Beach FL USA
> > ac9410 at bellsouth.net
> >
> > ===============================================================================================================
> > Date: Thu, 05 Sep 2002 22:51:55 -0400
> > From: "Mark D. Studebaker" <mds at paradyne.com>
> > Organization: Paradyne Networks, Largo FL
> > X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.4.18 i586)
> > X-Accept-Language: en
> > To: viro at math.psu.edu
> > CC: Albert Cranford <ac9410 at attbi.com>
> > Subject: Re: i2c-proc correction
> >
> > Al,
> >
> > Albert Cranford and I work on the lm_sensors project.
> >
> > in May 2002 you removed what was called
> > "s390 procfs abuse" lines with i2c_fill_inode procedure in
> > drivers/i2c/i2c-proc.c -
> > kernel 2.5.18 -
> >
> > URL
> >
> > http://linux.bkbits.net:8080/linux-2.5/diffs/drivers/i2c/i2c-proc.c at 1.3?nav=index.html|src/.|src/drivers|src/drivers/i2c|hist/drivers/i2c/i2c-proc.c
> >
> > We inadvertently put the lines back into kernel 2.5.32 with our recent
> > patch. Sorry.
> > We are submitting a patch to Linus (below) to take them back out in
> > kernel 2.5.34.
> >
> > However, I don't understand what 's390 procfs abuse' is or why these
> > lines were
> > removed from i2c-proc.c. As the comments say, the code prevents module
> > unloading
> > if the /proc files are in use. Why is this a bad thing? Is there a
> > better way to do it?
> >
> > Thank you very much for your help.
> >
> > mds
> >
> >
> >
> > > --- linux-2.5.33/drivers/i2c/i2c-proc.c.orig 2002-09-04 09:04:30.000000000 -0400
> > > +++ linux/drivers/i2c/i2c-proc.c 2002-09-04 09:08:59.000000000 -0400
> > > @@ -60,7 +60,6 @@
> > > static struct ctl_table_header *i2c_entries[SENSORS_ENTRY_MAX];
> > >
> > > static struct i2c_client *i2c_clients[SENSORS_ENTRY_MAX];
> > > -static unsigned short i2c_inodes[SENSORS_ENTRY_MAX];
> > >
> > > static ctl_table sysctl_table[] = {
> > > {CTL_DEV, "dev", NULL, 0, 0555},
> > > @@ -189,8 +188,6 @@
> > > return id;
> > > }
> > > #endif /* DEBUG */
> > > - i2c_inodes[id - 256] =
> > > - new_header->ctl_table->child->child->de->low_ino;
> > > new_header->ctl_table->child->child->de->owner = controlling_mod;
> > >
> > > return id;
> > > @@ -213,49 +210,6 @@
> > > }
> > > }
> > >
> > > -/* Monitor access for /proc/sys/dev/sensors; make unloading i2c-proc.o
> > > - impossible if some process still uses it or some file in it */
> > > -void i2c_fill_inode(struct inode *inode, int fill)
> > > -{
> > > - if (fill)
> > > - MOD_INC_USE_COUNT;
> > > - else
> > > - MOD_DEC_USE_COUNT;
> > > -}
> > > -
> > > -/* Monitor access for /proc/sys/dev/sensors/ directories; make unloading
> > > - the corresponding module impossible if some process still uses it or
> > > - some file in it */
> > > -void i2c_dir_fill_inode(struct inode *inode, int fill)
> > > -{
> > > - int i;
> > > - struct i2c_client *client;
> > > -
> > > -#ifdef DEBUG
> > > - if (!inode) {
> > > - printk(KERN_ERR "i2c-proc.o: Warning: inode NULL in fill_inode()\n");
> > > - return;
> > > - }
> > > -#endif /* def DEBUG */
> > > -
> > > - for (i = 0; i < SENSORS_ENTRY_MAX; i++)
> > > - if (i2c_clients[i]
> > > - && (i2c_inodes[i] == inode->i_ino)) break;
> > > -#ifdef DEBUG
> > > - if (i == SENSORS_ENTRY_MAX) {
> > > - printk
> > > - (KERN_ERR "i2c-proc.o: Warning: inode (%ld) not found in fill_inode()\n",
> > > - inode->i_ino);
> > > - return;
> > > - }
> > > -#endif /* def DEBUG */
> > > - client = i2c_clients[i];
> > > - if (fill)
> > > - client->driver->inc_use(client);
> > > - else
> > > - client->driver->dec_use(client);
> > > -}
> > > -
> > > int i2c_proc_chips(ctl_table * ctl, int write, struct file *filp,
> > > void *buffer, size_t * lenp)
> > > {
> > >
>
> --
> Philip Edelbrock -- IS Manager -- Edge Design, Corvallis, OR
> phil at netroedge.com -- http://www.netroedge.com/~phil
> PGP F16: 01 D2 FD 01 B5 46 F4 F0 3A 8B 9D 7E 14 7F FB 7A
