linux-kernel - Re: [PATCH] kernfs: implement custom llseek method to fix userspace regression

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 for Android: free password hash cracker in your pocket

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

Date:   Tue, 15 Aug 2023 08:48:25 -0700
From:   Dan Williams <dan.j.williams@...el.com>
To:     Valentin Sinitsyn <valesini@...dex-team.ru>,
        Dan Williams <dan.j.williams@...el.com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Tejun Heo <tj@...nel.org>
CC:     Daniel Vetter <daniel.vetter@...ll.ch>,
        Bjorn Helgaas <bhelgaas@...gle.com>,
        <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] kernfs: implement custom llseek method to fix userspace
 regression

Valentin Sinitsyn wrote:
[..]
> > My only concern is whether there are any scenarios where this is not
> > appropriate. I.e. do a bit more work to define a kernfs_ops instance
> > specifically for overriding lseek() in this scenario.
> 
> Not sure I'm getting you here: do you mean something like this?
> 
> struct inode *inode = is_f_mapping_redefined(file) ? file_inode(file) : 
> file->f_mapping->host;

I meant something like the patch below (incomplete, but shows the idea).

> My understanding is file->f_inode should always be non-NULL and point to 
> the inode corresponding the path of the opened file, so it should be 
> safe to call regardless what f_mapping->host is. Do I miss anything?

That matches my understanding and I do not think you missed anything. At
the same time a comment about "PCI resources" is out of place in
fs/kernfs/file.c.

On the rare chance that someone down the line cares about the
difference, a more localized change like this lets this override be done
in generic terms (f_mapping override) without reference to PCI resource
specifics:

diff --git a/fs/sysfs/file.c b/fs/sysfs/file.c
index a12ac0356c69..748804cd889f 100644
--- a/fs/sysfs/file.c
+++ b/fs/sysfs/file.c
@@ -167,6 +167,23 @@ static int sysfs_kf_bin_mmap(struct kernfs_open_file *of,
 	return battr->mmap(of->file, kobj, battr, vma);
 }
 
+static loff_t sysfs_kf_bin_llseek(struct kernfs_open_file *of, loff_t offset,
+				  int whence)
+{
+	struct bin_attribute *battr = of->kn->priv;
+	struct kobject *kobj = of->kn->parent->priv;
+
+	/* when mapping is overridden do not use it to lookup the inode */
+	if (battr->f_mapping) {
+		struct inode *file_inode(of->file);
+
+		return generic_file_llseek_size(of->file, offset, whence,
+						inode->i_sb->s_maxbytes,
+						i_size_read(inode));
+	}
+	return generic_file_llseek(of->file, offset, whence);
+}
+
 static int sysfs_kf_bin_open(struct kernfs_open_file *of)
 {
 	struct bin_attribute *battr = of->kn->priv;
@@ -249,6 +266,7 @@ static const struct kernfs_ops sysfs_bin_kfops_mmap = {
 	.write		= sysfs_kf_bin_write,
 	.mmap		= sysfs_kf_bin_mmap,
 	.open		= sysfs_kf_bin_open,
+	.llseek		= sysfs_kf_bin_llseek,
 };
 
 int sysfs_add_file_mode_ns(struct kernfs_node *parent,
diff --git a/include/linux/kernfs.h b/include/linux/kernfs.h
index 73f5c120def8..9ed535930259 100644
--- a/include/linux/kernfs.h
+++ b/include/linux/kernfs.h
@@ -316,6 +316,7 @@ struct kernfs_ops {
 			 struct poll_table_struct *pt);
 
 	int (*mmap)(struct kernfs_open_file *of, struct vm_area_struct *vma);
+	loff_t (*llseek)(struct kernfs_open_file *of, loff_t offset, int whence);
 };
 
 /*