linux-kernel - Re: [PATCH v3 13/20] netfs: Add a netfs inode context

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, 16 Mar 2022 09:06:25 +0000
From:   David Howells <dhowells@...hat.com>
To:     linux-cachefs@...hat.com
Cc:     dhowells@...hat.com, Anna Schumaker <anna.schumaker@...app.com>,
        Steve French <sfrench@...ba.org>,
        Dominique Martinet <asmadeus@...ewreck.org>,
        Jeff Layton <jlayton@...hat.com>,
        David Wysochanski <dwysocha@...hat.com>,
        Ilya Dryomov <idryomov@...il.com>,
        Jeffle Xu <jefflexu@...ux.alibaba.com>,
        Linus Torvalds <torvalds@...ux-foundation.org>,
        linux-afs@...ts.infradead.org, linux-nfs@...r.kernel.org,
        linux-cifs@...r.kernel.org, ceph-devel@...r.kernel.org,
        v9fs-developer@...ts.sourceforge.net,
        linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v3 13/20] netfs: Add a netfs inode context

I found a couple of issues here:

Firstly, netfs_is_cache_enabled() causes the generic/522 xfstest to take 60s
longer.  This can be fixed by:

-       return fscache_cookie_valid(cookie) && fscache_cookie_enabled(cookie);
+       return fscache_cookie_valid(cookie) && cookie->cache_priv &&
+               fscache_cookie_enabled(cookie);

There's no point trying to do caching if there's no cache actually attached.
I wonder if I should actually make this change in fscache_cookie_enabled()
rather than here.

Secondly, the above causes netfs_skip_folio_read() to be skipped a lot more
often than it should, and this hides an incorrect change there.  I made
netfs_skip_folio_read() copy the folio size into a variable to avoid issuing
the calculation multiple times, but I then gave the wrong length when clearing
the tail of the page.  This can be fixed by:

-       zero_user_segments(&folio->page, 0, offset, offset + len, len);
+       zero_user_segments(&folio->page, 0, offset, offset + len, plen);

David