From: Max Reitz <mreitz@redhat.com>
To: Stefan Hajnoczi <stefanha@gmail.com>
Cc: Kevin Wolf <kwolf@redhat.com>,
qemu-devel@nongnu.org, qemu-block@nongnu.org
Subject: Re: [PATCH 00/18] block: Allow exporting BDSs via FUSE
Date: Fri, 20 Dec 2019 11:30:33 +0100 [thread overview]
Message-ID: <a372e742-182a-ec98-d297-ed8d43e10734@redhat.com> (raw)
In-Reply-To: <20191220100846.GC1635864@stefanha-x1.localdomain>
[-- Attachment #1.1: Type: text/plain, Size: 8895 bytes --]
On 20.12.19 11:08, Stefan Hajnoczi wrote:
> On Thu, Dec 19, 2019 at 03:38:00PM +0100, Max Reitz wrote:
>> Preamble: This series is based on a combination of my (current) block
>> branch and “iotests: Minor fixes”. I’ve pushed it here:
>>
>> https://git.xanclic.moe/XanClic/qemu fuse-exports-v1
>>
>> (The base is on fuse-exports-v1-base.)
>>
>>
>> Hi,
>>
>> Ever since I found out that you can mount FUSE filesystems on regular
>> files (not just directories), I had the idea of adding FUSE block
>> exports to qemu where you can export block nodes as raw images. The
>> best thing is that you’d be able to mount an image on itself, so
>> whatever format it may be in, qemu lets it appear as a raw image (and
>> you can then use regular tools like dd on it).
>>
>> I started with some concept of a qemu-blkfuse daemon (similar to
>> qemu-nbd), but never sent patches, for two reasons: (1) Performance was
>> not good, (2) it didn’t seem right, for some reason.
>>
>> Now Kevin is proposing a storage daemon for multiple export types like
>> NBD, and he also mentioned FUSE (while knowing of my previous attempts).
>> Now it does seem right to add FUSE exports to qemu, but only in the form
>> of some module with a proper QAPI/QMP binding.
>>
>> Performance is still quite bad, but who cares. We can always improve
>> it, if the need arises.
>>
>>
>> This series does the following:
>>
>> First, add the FUSE export module (block/fuse.c) that implements the
>> basic file access functions. (Note that you need libfuse 3.8.0 or later
>> for SEEK_HOLE/SEEK_DATA.)
>>
>> Second, it allows using FUSE exports as a protocol in the iotests and
>> makes many iotests work with it. (The file node is exported by a
>> background qemu instance to $SOCK_DIR.)
>> Note that I only ran raw and qcow2 on it; I’m sure other formats
>> currently have some failing tests.
>>
>> This gives us a lot of coverage for, well, not free (it does take ten
>> patches), but for cheap; but there are still some more specialized
>> things we want to test, so third and last, this series adds an iotest
>> dedicated to FUSE exports.
>>
>>
>> Final rather important notice: I didn’t really run the iotests with this
>> yet. I wanted to, but they appear rather broken on current master,
>> actually. I’m not yet sure whether that’s because something in my setup
>> broke in the last two weeks, or because there’s quite something broken
>> in master (it does look like there are a couple things broken in master
>> currently).
>>
>>
>> Max Reitz (18):
>> configure: Detect libfuse
>> fuse: Allow exporting BDSs via FUSE
>> fuse: Implement standard FUSE operations
>> fuse: Add fuse-export-remove
>> fuse: Allow growable exports
>> fuse: (Partially) implement fallocate()
>> fuse: Implement hole detection through lseek
>> iotests: Do not needlessly filter _make_test_img
>> iotests: Do not pipe _make_test_img
>> iotests: Use convert -n in some cases
>> iotests: Avoid renaming images
>> iotests: Derive image names from $TEST_IMG
>> iotests/091: Use _cleanup_qemu instad of "wait"
>> iotests: Restrict some Python tests to file
>> iotests: Let _make_test_img guess $TEST_IMG_FILE
>> iotests: Allow testing FUSE exports
>> iotests: Enable fuse for many tests
>> iotests/281: Add test for FUSE exports
>>
>> block.c | 4 +
>> block/Makefile.objs | 3 +
>> block/fuse.c | 668 +++++++++++++++++++++++++++++++
>> configure | 68 ++++
>> include/block/fuse.h | 24 ++
>> qapi/block.json | 42 ++
>> tests/qemu-iotests/013 | 9 +-
>> tests/qemu-iotests/013.out | 3 +-
>> tests/qemu-iotests/018 | 5 +-
>> tests/qemu-iotests/018.out | 1 +
>> tests/qemu-iotests/020 | 2 +-
>> tests/qemu-iotests/025 | 2 +-
>> tests/qemu-iotests/026 | 2 +-
>> tests/qemu-iotests/028 | 16 +-
>> tests/qemu-iotests/028.out | 3 +
>> tests/qemu-iotests/031 | 2 +-
>> tests/qemu-iotests/034 | 2 +-
>> tests/qemu-iotests/036 | 2 +-
>> tests/qemu-iotests/037 | 2 +-
>> tests/qemu-iotests/038 | 2 +-
>> tests/qemu-iotests/039 | 2 +-
>> tests/qemu-iotests/046 | 7 +-
>> tests/qemu-iotests/046.out | 2 +-
>> tests/qemu-iotests/050 | 2 +-
>> tests/qemu-iotests/054 | 2 +-
>> tests/qemu-iotests/060 | 2 +-
>> tests/qemu-iotests/071 | 21 +-
>> tests/qemu-iotests/072 | 5 +-
>> tests/qemu-iotests/072.out | 1 +
>> tests/qemu-iotests/079 | 2 +-
>> tests/qemu-iotests/080 | 2 +-
>> tests/qemu-iotests/089 | 5 +-
>> tests/qemu-iotests/089.out | 1 +
>> tests/qemu-iotests/090 | 2 +-
>> tests/qemu-iotests/091 | 5 +-
>> tests/qemu-iotests/095 | 2 +-
>> tests/qemu-iotests/097 | 2 +-
>> tests/qemu-iotests/098 | 2 +-
>> tests/qemu-iotests/102 | 2 +-
>> tests/qemu-iotests/103 | 2 +-
>> tests/qemu-iotests/106 | 2 +-
>> tests/qemu-iotests/107 | 2 +-
>> tests/qemu-iotests/108 | 2 +-
>> tests/qemu-iotests/111 | 2 +-
>> tests/qemu-iotests/112 | 2 +-
>> tests/qemu-iotests/115 | 2 +-
>> tests/qemu-iotests/117 | 2 +-
>> tests/qemu-iotests/120 | 2 +-
>> tests/qemu-iotests/121 | 2 +-
>> tests/qemu-iotests/127 | 2 +-
>> tests/qemu-iotests/133 | 2 +-
>> tests/qemu-iotests/137 | 2 +-
>> tests/qemu-iotests/138 | 2 +-
>> tests/qemu-iotests/140 | 2 +-
>> tests/qemu-iotests/154 | 2 +-
>> tests/qemu-iotests/161 | 14 +-
>> tests/qemu-iotests/171 | 2 +-
>> tests/qemu-iotests/174 | 10 +-
>> tests/qemu-iotests/175 | 8 +-
>> tests/qemu-iotests/176 | 2 +-
>> tests/qemu-iotests/177 | 2 +-
>> tests/qemu-iotests/179 | 2 +-
>> tests/qemu-iotests/183 | 2 +-
>> tests/qemu-iotests/186 | 2 +-
>> tests/qemu-iotests/187 | 2 +-
>> tests/qemu-iotests/191 | 2 +-
>> tests/qemu-iotests/195 | 2 +-
>> tests/qemu-iotests/200 | 5 +-
>> tests/qemu-iotests/200.out | 4 +-
>> tests/qemu-iotests/204 | 2 +-
>> tests/qemu-iotests/206 | 1 +
>> tests/qemu-iotests/214 | 2 +-
>> tests/qemu-iotests/217 | 2 +-
>> tests/qemu-iotests/220 | 2 +-
>> tests/qemu-iotests/221 | 2 +-
>> tests/qemu-iotests/229 | 5 +-
>> tests/qemu-iotests/229.out | 6 +-
>> tests/qemu-iotests/242 | 1 +
>> tests/qemu-iotests/247 | 2 +-
>> tests/qemu-iotests/249 | 8 +-
>> tests/qemu-iotests/250 | 2 +-
>> tests/qemu-iotests/252 | 2 +-
>> tests/qemu-iotests/265 | 2 +-
>> tests/qemu-iotests/268 | 2 +-
>> tests/qemu-iotests/272 | 2 +-
>> tests/qemu-iotests/273 | 2 +-
>> tests/qemu-iotests/279 | 2 +-
>> tests/qemu-iotests/281 | 328 +++++++++++++++
>> tests/qemu-iotests/281.out | 92 +++++
>> tests/qemu-iotests/check | 6 +
>> tests/qemu-iotests/common.filter | 5 +-
>> tests/qemu-iotests/common.rc | 190 ++++++++-
>> tests/qemu-iotests/group | 1 +
>> 93 files changed, 1571 insertions(+), 120 deletions(-)
>> create mode 100644 block/fuse.c
>> create mode 100644 include/block/fuse.h
>> create mode 100755 tests/qemu-iotests/281
>> create mode 100644 tests/qemu-iotests/281.out
>
> A lot of qemu-iotests changes but the actual FUSE code is small and well
> worth it for the flexibility and convenience that this feature brings.
> Nice!
Good that you like it! :-)
> Please send a follow-up patch that adds a qemu(1) -blockdev
> 'Driver-specific options for "fuse"' documentation section.
What exactly do you mean? This is not a block driver, so it doesn’t
work as part of -blockdev. Currently, it can only be used through QMP
(fuse-export-add/fuse-export-remove).
I placed it into block/ because that just seemed like the least bad
place to me (apart from creating a new top-level directory like nbd has)
– and also because we already have quite some few non-driver files in
block/ (io.c, the jobs (where some got drivers only rather recently),
accounting.c, ...).
Max
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]
next prev parent reply other threads:[~2019-12-20 10:31 UTC|newest]
Thread overview: 37+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-12-19 14:38 [PATCH 00/18] block: Allow exporting BDSs via FUSE Max Reitz
2019-12-19 14:38 ` [PATCH 01/18] configure: Detect libfuse Max Reitz
2019-12-19 14:38 ` [PATCH 02/18] fuse: Allow exporting BDSs via FUSE Max Reitz
2019-12-20 10:26 ` Kevin Wolf
2019-12-20 10:48 ` Max Reitz
2019-12-20 11:24 ` Kevin Wolf
2019-12-20 12:09 ` Max Reitz
2019-12-20 12:48 ` Markus Armbruster
2019-12-20 12:58 ` Kevin Wolf
2019-12-20 13:25 ` Markus Armbruster
2019-12-20 21:18 ` Eric Blake
2019-12-20 12:49 ` Markus Armbruster
2019-12-20 13:02 ` Kevin Wolf
2019-12-20 21:15 ` Eric Blake
2020-01-06 12:00 ` Max Reitz
2019-12-19 14:38 ` [PATCH 03/18] fuse: Implement standard FUSE operations Max Reitz
2019-12-19 14:38 ` [PATCH 04/18] fuse: Add fuse-export-remove Max Reitz
2019-12-19 14:38 ` [PATCH 05/18] fuse: Allow growable exports Max Reitz
2019-12-19 14:38 ` [PATCH 06/18] fuse: (Partially) implement fallocate() Max Reitz
2019-12-19 14:38 ` [PATCH 07/18] fuse: Implement hole detection through lseek Max Reitz
2019-12-19 14:38 ` [PATCH 08/18] iotests: Do not needlessly filter _make_test_img Max Reitz
2019-12-19 14:38 ` [PATCH 09/18] iotests: Do not pipe _make_test_img Max Reitz
2019-12-19 14:38 ` [PATCH 10/18] iotests: Use convert -n in some cases Max Reitz
2019-12-19 14:38 ` [PATCH 11/18] iotests: Avoid renaming images Max Reitz
2019-12-19 14:38 ` [PATCH 12/18] iotests: Derive image names from $TEST_IMG Max Reitz
2019-12-19 14:38 ` [PATCH 13/18] iotests/091: Use _cleanup_qemu instad of "wait" Max Reitz
2019-12-19 14:38 ` [PATCH 14/18] iotests: Restrict some Python tests to file Max Reitz
2019-12-19 14:38 ` [PATCH 15/18] iotests: Let _make_test_img guess $TEST_IMG_FILE Max Reitz
2019-12-19 14:38 ` [PATCH 16/18] iotests: Allow testing FUSE exports Max Reitz
2019-12-19 14:38 ` [PATCH 17/18] iotests: Enable fuse for many tests Max Reitz
2019-12-19 14:38 ` [PATCH 18/18] iotests/281: Add test for FUSE exports Max Reitz
2019-12-19 19:05 ` [PATCH 00/18] block: Allow exporting BDSs via FUSE Max Reitz
2019-12-20 10:08 ` Stefan Hajnoczi
2019-12-20 10:30 ` Max Reitz [this message]
2019-12-20 12:50 ` Kevin Wolf
2019-12-20 21:20 ` Eric Blake
2020-01-02 11:22 ` Stefan Hajnoczi
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=a372e742-182a-ec98-d297-ed8d43e10734@redhat.com \
--to=mreitz@redhat.com \
--cc=kwolf@redhat.com \
--cc=qemu-block@nongnu.org \
--cc=qemu-devel@nongnu.org \
--cc=stefanha@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).