linux-kernel - Re: [PATCH] bcachefs: chardev: make bch_chardev

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Wed, 6 Mar 2024 21:01:59 +0800
From: Hongbo Li <lihongbo22@...wei.com>
To: "Ricardo B. Marliere" <ricardo@...liere.net>
CC: Kent Overstreet <kent.overstreet@...ux.dev>, Brian Foster
	<bfoster@...hat.com>, <linux-bcachefs@...r.kernel.org>,
	<linux-kernel@...r.kernel.org>, Greg Kroah-Hartman
	<gregkh@...uxfoundation.org>
Subject: Re: [PATCH] bcachefs: chardev: make bch_chardev_class constant



On 2024/3/6 19:50, Ricardo B. Marliere wrote:
> On  6 Mar 09:23, Hongbo Li wrote:
>> On 2024/3/6 4:21, Ricardo B. Marliere wrote:
>>> Since commit 43a7206b0963 ("driver core: class: make class_register() take
>>> a const *"), the driver core allows for struct class to be in read-only
>>> memory, so move the bch_chardev_class structure to be declared at build
>>> time placing it into read-only memory, instead of having to be dynamically
>>> allocated at boot time.
>>>
>>> Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
>>> Suggested-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
>>> Signed-off-by: Ricardo B. Marliere <ricardo@...liere.net>
>>> ---
>>>    fs/bcachefs/chardev.c | 23 ++++++++++++-----------
>>>    1 file changed, 12 insertions(+), 11 deletions(-)
>>>
>>> diff --git a/fs/bcachefs/chardev.c b/fs/bcachefs/chardev.c
>>> index 226b39c17667..af587453fd3d 100644
>>> --- a/fs/bcachefs/chardev.c
>>> +++ b/fs/bcachefs/chardev.c
>>> @@ -940,7 +940,9 @@ static const struct file_operations bch_chardev_fops = {
>>>    };
>>>    
>>>    static int bch_chardev_major;
>>> -static struct class *bch_chardev_class;
>>> +static const struct class bch_chardev_class = {
>>> +	.name = "bcachefs",
>>> +};
>>>    static struct device *bch_chardev;
>>>    
>>>    void bch2_fs_chardev_exit(struct bch_fs *c)
>>> @@ -957,7 +959,7 @@ int bch2_fs_chardev_init(struct bch_fs *c)
>>>    	if (c->minor < 0)
>>>    		return c->minor;
>>>    
>>> -	c->chardev = device_create(bch_chardev_class, NULL,
>>> +	c->chardev = device_create(&bch_chardev_class, NULL,
>>>    				   MKDEV(bch_chardev_major, c->minor), c,
>>>    				   "bcachefs%u-ctl", c->minor);
>>>    	if (IS_ERR(c->chardev))
>>> @@ -968,26 +970,25 @@ int bch2_fs_chardev_init(struct bch_fs *c)
>>>    
>>>    void bch2_chardev_exit(void)
>>>    {
>>> -	if (!IS_ERR_OR_NULL(bch_chardev_class))
>>> -		device_destroy(bch_chardev_class,
>>> -			       MKDEV(bch_chardev_major, U8_MAX));
>>> -	if (!IS_ERR_OR_NULL(bch_chardev_class))
>>> -		class_destroy(bch_chardev_class);
>>> +	device_destroy(&bch_chardev_class, MKDEV(bch_chardev_major, U8_MAX));
>>> +	class_unregister(&bch_chardev_class);
>>>    	if (bch_chardev_major > 0)
>>>    		unregister_chrdev(bch_chardev_major, "bcachefs");
>>>    }
>>>    
>>>    int __init bch2_chardev_init(void)
>>>    {
>>> +	int ret;
>>> +
>>>    	bch_chardev_major = register_chrdev(0, "bcachefs-ctl", &bch_chardev_fops);
>>>    	if (bch_chardev_major < 0)
>>>    		return bch_chardev_major;
>>>    
>>> -	bch_chardev_class = class_create("bcachefs");
>>> -	if (IS_ERR(bch_chardev_class))
>>> -		return PTR_ERR(bch_chardev_class);
>>> +	ret = class_register(&bch_chardev_class);
>>> +	if (ret)
>> Here, I think you should call class_unregister to relase resource which
>> allocated before. And the same thing shoud be done in other exception exit.
> 
> Hi Hongbo,
> 
> Thank you for the feedback. Did you mean that bch_chardev_major should
> be unregistered if the class_register() call fails? Because if it does,
> there is no need to call class_unregister().
> 
>>> +		return ret;
>>>    
>>> -	bch_chardev = device_create(bch_chardev_class, NULL,
>>> +	bch_chardev = device_create(&bch_chardev_class, NULL,
>>>    				    MKDEV(bch_chardev_major, U8_MAX),
>>>    				    NULL, "bcachefs-ctl");
>>>    	if (IS_ERR(bch_chardev))
>> like here..
> 
> Can you please elaborate?
> 
Yes, this I mean. The example like this:

```c
    major = register_chrdev(0, xx, xx);
    class = class_create("bcachefs");
    if (IS_ERR(class))
         goto out_chrdev:

    ...
out_class:
     class_unregister(xxx);
out_chrdev:
     unregister_chrdev(xx, xx);
     ...
```

Although this was not part of your original intention, it is indeed a flaw.


> Best regards,
> -	Ricardo.
> 
> 
>>>
>>> ---
>>> base-commit: 90d35da658da8cff0d4ecbb5113f5fac9d00eb72
>>> change-id: 20240305-bcachefs-27a4bb8b9f4f
>>>
>>> Best regards,
>