Re: devfs: BKL *not* taken while opening devices

Richard Gooch (rgooch@ras.ucalgary.ca)
Fri, 10 May 2002 18:45:23 -0600


Russell King writes:
> Hi,
>
> Nice simple bug report.
>
> Kernel 2.5.8... Devfs devfs_open() bypasses the normal chrdev_open and
> blkdev_open functions, and misses out taking the BKL. 2.5.10 is the
> same.
>
> Certainly the tty layer (and probably many of the other devices as well)
> require the BKL to be taken before calling the open method.

Yes, I'll be submitting a patch to fix this.

> Also, the following looks wrong:
>
> if ( S_ISBLK (inode->i_mode) )
> {
> file->f_op = &def_blk_fops;
> if (df->ops) inode->i_bdev->bd_op = df->ops;
> }
> else file->f_op = fops_get ( (struct file_operations *) df->ops );
> if (file->f_op)
> err = file->f_op->open ? (*file->f_op->open) (inode, file) : 0;
> else
> {
> /* Fallback to legacy scheme */
> if ( S_ISCHR (inode->i_mode) ) err = chrdev_open (inode, file);
> else err = -ENODEV;
> }
> if (err < 0) return err;
>
> We can return without releasing the file operations after fops_get(),
> thereby effectively locking modules in memory.

How is this different from the chrdev_open() logic? It seems to me
this is taken care of in fput(), so no need to worry.

Regards,

Richard....
Permanent: rgooch@atnf.csiro.au
Current: rgooch@ras.ucalgary.ca
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/