Re: [PATCH v3] scsi: target: fix recursive locking in __configfs_open_file()

[Prithvi]

On Thu, Feb 05, 2026 at 07:26:44PM +0000, Al Viro wrote:
> On Thu, Feb 05, 2026 at 09:56:24PM +0530, Prithvi Tambewagh wrote:
> 
> > +	r = kern_path(db_root_stage, LOOKUP_FOLLOW, &path);
> > +	if (r) {
> >  		pr_err("db_root: cannot open: %s\n", db_root_stage);
> >  		goto unlock;
> >  	}
> > -	if (!S_ISDIR(file_inode(fp)->i_mode)) {
> > -		filp_close(fp, NULL);
> > +	if (!d_is_dir(path.dentry)) {
> > +		path_put(&path);
> >  		pr_err("db_root: not a directory: %s\n", db_root_stage);
> > +		r = -ENOTDIR;
> >  		goto unlock;
> >  	}
> > -	filp_close(fp, NULL);
> > +	path_put(&path);
> 
> Just pass it LOOKUP_FOLLOW | LOOKUP_DIRECTORY and be done with the manual
> "is it a directory" tests in any form...

Hello Al,

I sincerely apologise for the delayed response. I was testing the change you 
suggested, however, whenever I tried testing my patch against the latest 
commit where syzbot reported this bug (commit 3a8660878839faadb4f1a6dd72c3179c1df56787
of upstream repository) it gave me a build failure immediately without any
debug log, just the message:

syzbot tried to test the proposed patch but the build/boot failed:

failed to run ["make" "KERNELVERSION=syzkaller" "KERNELRELEASE=syzkaller" "LOCALVERSION=-syzkaller" "-j" "48" "ARCH=x86_64" "bzImage"]: exit status 2

The issue seems to occur multiple times when a patch is tested against the 
latest commit where syzbot reported the issue while it doesn't occur on that 
latest commit of the upstream repository.

However, testing the change on the latest commit of upstream reprository 
(commit 72c395024dac5e215136cbff793455f065603b06) gives a positive result 
that the reproducer doesn't trigger any issue.

Reference: https://lore.kernel.org/all/6767d8ea.050a0220.226966.0021.GAE@xxxxxxxxxx/T/#mbf32aeb54c4cae609d3b6176ad8dcd99bfc51ad2

IIUC, since the reported failure appears to be unrelated to the change and is working
successfully on latest commit of upstream, I wanted to confirm if v4 based on 
these findings is acceptable.

What do you think?

Thank you,
Prithvi