linux-kernel - Re: [PATCH 3/4] jffs2: pass the correct prototype to read_cache

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] [day] [month] [year] [list]

Message-ID: <20190618202721.GD17978@ZenIV.linux.org.uk>
Date:   Tue, 18 Jun 2019 21:27:21 +0100
From:   Al Viro <viro@...iv.linux.org.uk>
To:     Christoph Hellwig <hch@....de>
Cc:     Andrew Morton <akpm@...ux-foundation.org>,
        Sami Tolvanen <samitolvanen@...gle.com>,
        Kees Cook <keescook@...omium.org>,
        Nick Desaulniers <ndesaulniers@...gle.com>,
        linux-mtd@...ts.infradead.org, linux-nfs@...r.kernel.org,
        linux-mm@...ck.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 3/4] jffs2: pass the correct prototype to read_cache_page

On Mon, May 20, 2019 at 07:57:30AM +0200, Christoph Hellwig wrote:
> Fix the callback jffs2 passes to read_cache_page to actually have the
> proper type expected.  Casting around function pointers can easily
> hide typing bugs, and defeats control flow protection.

FWIW, this
unsigned char *jffs2_gc_fetch_page(struct jffs2_sb_info *c,
                                   struct jffs2_inode_info *f,
                                   unsigned long offset,
                                   unsigned long *priv)
{
        struct inode *inode = OFNI_EDONI_2SFFJ(f);
        struct page *pg;

        pg = read_cache_page(inode->i_mapping, offset >> PAGE_SHIFT,
                             (void *)jffs2_do_readpage_unlock, inode);
        if (IS_ERR(pg))
                return (void *)pg;

        *priv = (unsigned long)pg;
        return kmap(pg);
}
looks like crap.  And so does this:
void jffs2_gc_release_page(struct jffs2_sb_info *c,
                           unsigned char *ptr,
                           unsigned long *priv)
{
        struct page *pg = (void *)*priv;

        kunmap(pg);
        put_page(pg);
}

	First of all, there's only one caller for each of those, and both
are direct calls.  So passing struct page * around that way is ridiculous.
What's more, there is no reason not to do kmap() in caller (i.e. in
jffs2_garbage_collect_dnode()).  That way jffs2_gc_fetch_page() would
simply be return read_cache_page(....), and in the caller we'd have

        struct page *pg;
        unsigned char *pg_ptr;
...
        mutex_unlock(&f->sem);
        pg = jffs2_gc_fetch_page(c, f, start);
        if (IS_ERR(pg)) {
		mutex_lock(&f->sem);
                pr_warn("read_cache_page() returned error: %ld\n", PTR_ERR(pg));
                return PTR_ERR(pg);
        }
	pg_ptr = kmap(pg);
	mutex_lock(&f->sem);
...
	kunmap(pg);
	put_page(pg);

and that's it, preserving the current locking and with saner types...