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  PHC 
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Date:   Fri, 21 Aug 2020 21:20:54 -0700
From:   John Hubbard <>
To:     Andrew Morton <>
CC:     Alexander Viro <>,
        Christoph Hellwig <>,
        Ilya Dryomov <>,
        Jens Axboe <>, Jeff Layton <>,
        <>, <>,
        <>, <>,
        <>, LKML <>,
        John Hubbard <>
Subject: [PATCH 0/5] bio: Direct IO: convert to pin_user_pages_fast()


This converts the Direct IO block/bio layer over to use FOLL_PIN pages
(those acquired via pin_user_pages*()). This effectively converts
several file systems (ext4, for example) that use the common Direct IO
routines. See "Remaining work", below for a bit more detail there.

Quite a few approaches have been considered over the years. This one is
inspired by Christoph Hellwig's July, 2019 observation that there are
only 5 ITER_ types, and we can simplify handling of them for Direct IO
[1]. After working through how bio submission and completion works, I
became convinced that this is the simplest and cleanest approach to

Not content to let well enough alone, I then continued on to the
unthinkable: adding a new flag to struct bio, whose "short int" flags
field was full, thuse triggering an expansion of the field from 16, to
32 bits. This allows for a nice assertion in bio_release_pages(), that
the bio page release mechanism matches the page acquisition mechanism.
This is especially welcome for a change that affects a lot of callers
and could really make a mess if there is a bug somewhere.

I'm unable to spot any performance implications, either theoretically or
via (rather light) performance testing, from enlarging bio.bi_flags, but
I suspect that there are maybe still valid reasons for having such a
tiny bio.bi_flags field. I just have no idea what they are. (Hardware
that knows the size of a bio? No, because there would be obvious
build-time assertions, and comments about such a constraint.) Anyway, I
can drop that patch if it seems like too much cost for too little

And finally, as long as we're all staring at the iter_iov code, I'm
including a nice easy ceph patch, that removes one more caller of

Design notes ============

This whole approach depends on certain concepts:

1) Each struct bio instance must not mix different types of pages:
FOLL_PIN and non-FOLL_PIN pages. (By FOLL_PIN I'm referring to pages
that were acquired and pinned via pin_user_page*() routines.)
Fortunately, this is already an enforced constraint for bio's, as
evidenced by the existence and use of BIO_NO_PAGE_REF.

2) Christoph Hellwig's July, 2019 observation that there are
only 5 ITER_ types, and we can simplify handling of them for Direct IO
[1]. Accordingly, this series implements the following pseudocode:

Direct IO behavior:


    ITER_KVEC:    // already elevated page refcount, leave alone
    ITER_BVEC:    // already elevated page refcount, leave alone
    ITER_PIPE:    // just, no :)
    ITER_DISCARD: // discard
        return -EFAULT or -ENVALID;

...which works for callers that already have sorted out which case they
are in. Such as, Direct IO in the block/bio layers.

Now, this does leave ITER_KVEC and ITER_BVEC unconverted, but on the
other hand, it's not clear that these are actually affected in the real
world, by the get_user_pages()+filesystem interaction problems of [2].
If it turns out to matter, then those can be handled too, but it's just
more refactoring and surgery to do so.


Performance: no obvious regressions from running fio (direct=1: Direct
IO) on both SSD and NVMe drives.

Functionality: selected non-destructive bare metal xfstests on xfs,
ext4, btrfs, orangefs filesystems, plus LTP tests.

Note that I have only a single x86 64-bit test machine, though.

Remaining work

Non-converted call sites for iter_iov_get_pages*() at the
moment include: net, crypto, cifs, ceph, vhost, fuse, nfs/direct,

About-to-be-converted sites (in a subsequent patch) are: Direct IO for
filesystems that use the generic read/write functions.


[2] "Explicit pinning of user-space pages":

John Hubbard (5):
  iov_iter: introduce iov_iter_pin_user_pages*() routines
  mm/gup: introduce pin_user_page()
  bio: convert get_user_pages_fast() --> pin_user_pages_fast()
  bio: introduce BIO_FOLL_PIN flag
  fs/ceph: use pipe_get_pages_alloc() for pipe

 block/bio.c               | 29 +++++++------
 block/blk-map.c           |  7 +--
 fs/ceph/file.c            |  3 +-
 fs/direct-io.c            | 30 ++++++-------
 fs/iomap/direct-io.c      |  2 +-
 include/linux/blk_types.h |  5 ++-
 include/linux/mm.h        |  2 +
 include/linux/uio.h       |  9 +++-
 lib/iov_iter.c            | 91 +++++++++++++++++++++++++++++++++++++--
 mm/gup.c                  | 30 +++++++++++++
 10 files changed, 169 insertions(+), 39 deletions(-)


Powered by blists - more mailing lists