Re: [PATCH v3 1/3] debugfs: Add write support to debugfs_create_str()

[Georgi Djakov]

Hi Greg,

Thanks for the comments!

On 12.08.23 13:40, Greg KH wrote:
> On Mon, Aug 07, 2023 at 07:29:12AM -0700, Mike Tipton wrote:
> > Currently, debugfs_create_str() only supports reading strings from
> > debugfs. Add support for writing them as well.
> >
> > Based on original implementation by Peter Zijlstra [0]. Write support
> > was present in the initial patch version, but dropped in v2 due to lack
> > of users. We have a user now, so reintroduce it.
> >
> > [0] https://lore.kernel.org/all/YF3Hv5zXb%2F6lauzs@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/
> >
> > Signed-off-by: Mike Tipton <quic_mdtipton@xxxxxxxxxxx>
> > ---
> >   fs/debugfs/file.c | 48 +++++++++++++++++++++++++++++++++++++++++++++--
> >   1 file changed, 46 insertions(+), 2 deletions(-)
> >
> > diff --git a/fs/debugfs/file.c b/fs/debugfs/file.c
> > index b7711888dd17..87b3753aa4b1 100644
> > --- a/fs/debugfs/file.c
> > +++ b/fs/debugfs/file.c
> > @@ -904,8 +904,52 @@ EXPORT_SYMBOL_GPL(debugfs_create_str);
> >   static ssize_t debugfs_write_file_str(struct file *file, const char __user *user_buf,
> >   				      size_t count, loff_t *ppos)
> >   {
> > -	/* This is really only for read-only strings */
> > -	return -EINVAL;
> > +	struct dentry *dentry = F_DENTRY(file);
> > +	char *old, *new = NULL;
> > +	int pos = *ppos;
> > +	int r;
> > +
> > +	r = debugfs_file_get(dentry);
> > +	if (unlikely(r))
> > +		return r;
> > +
> > +	old = *(char **)file->private_data;
> > +
> > +	/* only allow strict concatenation */
> > +	r = -EINVAL;
> > +	if (pos && pos != strlen(old))
> > +		goto error;
> > +
> > +	r = -E2BIG;
> > +	if (pos + count + 1 > PAGE_SIZE)
> > +		goto error;
> > +
> > +	r = -ENOMEM;
> > +	new = kmalloc(pos + count + 1, GFP_KERNEL);
> > +	if (!new)
> > +		goto error;
> > +
> > +	if (pos)
> > +		memcpy(new, old, pos);
> > +
> > +	r = -EFAULT;
> > +	if (copy_from_user(new + pos, user_buf, count))
> > +		goto error;
> > +
> > +	new[pos + count] = '\0';
> > +	strim(new);
> > +
> > +	rcu_assign_pointer(*(char **)file->private_data, new);
> > +	synchronize_rcu();
> > +	kfree(old);
> > +
> > +	debugfs_file_put(dentry);
> > +	return count;
> > +
> > +error:
> > +	kfree(new);
> > +	debugfs_file_put(dentry);
> > +	return r;
> >   }
>
> So you just added write support for ALL debugfs files that use the
> string interface, what did you just allow to break?

Not really. According to the existing code, the write support for strings
is enabled only when the file is created with +w permissions. For read-only
files, we use fops_str_ro, which is the case for all existing string files:

$ git grep -w debugfs_create_str | egrep -v "fs/debugfs/file.c|include/linux/debugfs.h"
drivers/firmware/arm_scmi/driver.c:	debugfs_create_str("instance_name", 0400, top_dentry,
drivers/firmware/arm_scmi/driver.c:	debugfs_create_str("type", 0400, trans, (char **)&dbg->type);
drivers/opp/debugfs.c:	debugfs_create_str("of_name", S_IRUGO, d, (char **)&opp->of_name);

For fops_str_ro, the .write function is not implemented, so nothing should break?

> I recommend just using your own debugfs file function instead, as this
> could cause bad problems, right?

Agree, and that should be exactly what this patch does.

> Are you sure that all string calls can
> handle the variable be freed underneath it like this call will allow to
> happen?

Looks fine, at least for this patch-set.

Thanks,
Georgi