Re: [PATCHSET v6 4/8] fuse: allow servers to use iomap for better file IO performance

This is an OpenPGP/MIME signed message (RFC 4880 and 3156)
--------------oNoKUhM0fhMW0yPJIdrVMC00
Content-Type: multipart/mixed; boundary="------------XJKrQABjOfTNuc2vdQJeIFbC";
 protected-headers="v1"
From: Demi Marie Obenour <[email protected]>
To: [email protected]
Cc: [email protected], [email protected], [email protected],
 [email protected], [email protected], [email protected],
 [email protected], [email protected],
 [email protected]
Message-ID: <[email protected]>
Subject: Re: [PATCHSET v6 4/8] fuse: allow servers to use iomap for better
 file IO performance
References: <176169810144.1424854.11439355400009006946.stgit@frogsfrogsfrogs>
In-Reply-To: <176169810144.1424854.11439355400009006946.stgit@frogsfrogsfrogs>

--------------XJKrQABjOfTNuc2vdQJeIFbC
Content-Type: multipart/mixed; boundary="------------1YwruzpyHib5np6aKkR2Xltz"

--------------1YwruzpyHib5np6aKkR2Xltz
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

> By keeping the I/O path mostly within the kernel, we can dramatically
> increase the speed of disk-based filesystems.

ZFS, BTRFS, and bcachefs all support compression, checksumming,
and RAID.  ZFS and bcachefs also support encryption, and f2fs and
ext4 support fscrypt.

Will this patchset be able to improve FUSE implementations of these
filesystems?  I'd rather not be in the situation where one can have
a FUSE filesystem that is fast, but only if it doesn't support modern
data integrity or security features.

I'm not a filesystem developer, but here are some ideas (that you
can take or leave):

1. Keep the compression, checksumming, and/or encryption in-kernel,
   and have userspace tell the kernel what algorithm and/or encryption
   key to use.  These algorithms are generally well-known and secure
   against malicious input.  It might be necessary to make an extra
   data copy, but ideally that copy could just stay within the
   CPU caches.

2. Somehow integrate with the blk-crypto framework.  This has the
   advantage that it supports inline encryption hardware, which
   I suspect is needed for this to be usable on mobile devices.
   After all, the keys on these systems are often not even visible
   to the kernel, let alone to userspace.

3. Figure out a way to make a userspace data path fast enough.
   To prevent data corruption by unprivileged users of the FS,
   it's necessary to make a copy before checksumming, compression,
   or authenticated encryption.  If this copy is done in the kernel,
   the server doesn't have to perform its own copy.  By using large
   ring buffers, it might be possible to amortize the context switch
   cost away.

   Authenticated encryption also needs a copy in the *other* direction:
   if the (untrusted) client can see unauthenticated plaintext, it's
   a security vulnerability.  That needs another copy from server
   buffers to client buffers, and the kernel can do that as well.

4. Make context switches much faster.  L4-style IPC is incredibly fast,
   at least if one doesn't have to worry about Spectre.  Unfortunately,
   nowadays one *does* need to worry about Spectre.

Obviously, none of these will be as fast as doing DMA directly to user
buffers.  However, all of these features (except for encryption using
inline encryption hardware) come at a performance penalty already.
I just don't want a FUSE server to have to pay a much larger penalty
than a kernel filesystem would.

I'm CCing the bcachefs, BTRFS, and ZFS-on-Linux mailing lists.
--=20
Sincerely,
Demi Marie Obenour (she/her/hers)

--------------1YwruzpyHib5np6aKkR2Xltz
Content-Type: application/pgp-keys; name="OpenPGP_0xB288B55FFF9C22C1.asc"
Content-Disposition: attachment; filename="OpenPGP_0xB288B55FFF9C22C1.asc"
Content-Description: OpenPGP public key
Content-Transfer-Encoding: quoted-printable

-----BEGIN PGP PUBLIC KEY BLOCK-----
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3D=3D
=3Dx94R
-----END PGP PUBLIC KEY BLOCK-----

--------------1YwruzpyHib5np6aKkR2Xltz--

--------------XJKrQABjOfTNuc2vdQJeIFbC--

--------------oNoKUhM0fhMW0yPJIdrVMC00
Content-Type: application/pgp-signature; name="OpenPGP_signature.asc"
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename="OpenPGP_signature.asc"

-----BEGIN PGP SIGNATURE-----
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=8SWs
-----END PGP SIGNATURE-----

--------------oNoKUhM0fhMW0yPJIdrVMC00--