[Qemu-devel] [PULL for-2.10 00/15] Block patches
[PULL 00/19] Block patches
1
The following changes since commit 248b23735645f7cbb503d9be6f5bf825f2a603ab:
1
The following changes since commit 96662996eda78c48aadddd4e76d8615c7eb72d80:
2
2
3
Update version for v2.10.0-rc4 release (2017-08-24 17:34:26 +0100)
3
Merge remote-tracking branch 'remotes/dgilbert/tags/pull-migration-20210513a' into staging (2021-05-14 12:03:47 +0100)
4
4
5
are available in the git repository at:
5
are available in the Git repository at:
6
6
7
git://github.com/stefanha/qemu.git tags/block-pull-request
7
https://github.com/XanClic/qemu.git tags/pull-block-2021-05-14
8
8
9
for you to fetch changes up to 3e4c705212abfe8c9882a00beb2d1466a8a53cec:
9
for you to fetch changes up to c61ebf362d0abf288ce266845519d5a550a1d89f:
10
10
11
qcow2: allocate cluster_cache/cluster_data on demand (2017-08-30 18:02:10 +0100)
11
write-threshold: deal with includes (2021-05-14 16:14:10 +0200)
12
12
13
----------------------------------------------------------------
13
----------------------------------------------------------------
14
Block patches:
15
- drop block/io write notifiers
16
- qemu-iotests enhancements to make debugging easier
17
- rbd parsing fix
18
- HMP qemu-io fix (for iothreads)
19
- mirror job cancel relaxation (do not cancel in-flight requests when a
20
READY mirror job is canceled with force=false)
21
- document qcow2's data_file and data_file_raw features
22
- fix iotest 297 for pylint 2.8
23
- block/copy-on-read refactoring
14
24
15
----------------------------------------------------------------
25
----------------------------------------------------------------
26
Connor Kuehl (3):
27
iotests/231: Update expected deprecation message
28
block/rbd: Add an escape-aware strchr helper
29
Document qemu-img options data_file and data_file_raw
16
30
17
Alberto Garcia (8):
31
Emanuele Giuseppe Esposito (1):
18
throttle: Fix wrong variable name in the header documentation
32
qemu-iotests: fix pylint 2.8 consider-using-with error
19
throttle: Update the throttle_fix_bucket() documentation
20
throttle: Make throttle_is_valid() a bit less verbose
21
throttle: Remove throttle_fix_bucket() / throttle_unfix_bucket()
22
throttle: Make LeakyBucket.avg and LeakyBucket.max integer types
23
throttle: Make burst_length 64bit and add range checks
24
throttle: Test the valid range of config values
25
misc: Remove unused Error variables
26
33
27
Dan Aloni (1):
34
Paolo Bonzini (5):
28
nvme: Fix get/set number of queues feature, again
35
qemu-iotests: do not buffer the test output
36
qemu-iotests: allow passing unittest.main arguments to the test
37
scripts
38
qemu-iotests: move command line and environment handling from
39
TestRunner to TestEnv
40
qemu-iotests: let "check" spawn an arbitrary test command
41
qemu-iotests: fix case of SOCK_DIR already in the environment
29
42
30
Eduardo Habkost (1):
43
Vladimir Sementsov-Ogievskiy (10):
31
oslib-posix: Print errors before aborting on qemu_alloc_stack()
44
monitor: hmp_qemu_io: acquire aio contex, fix crash
45
mirror: stop cancelling in-flight requests on non-force cancel in
46
READY
47
block/copy-on-read: use bdrv_drop_filter() and drop s->active
48
block/write-threshold: don't use write notifiers
49
block: drop write notifiers
50
test-write-threshold: rewrite test_threshold_(not_)trigger tests
51
block/write-threshold: drop extra APIs
52
test-write-threshold: drop extra tests
53
test-write-threshold: drop extra TestStruct structure
54
write-threshold: deal with includes
32
55
33
Fred Rolland (1):
56
docs/tools/qemu-img.rst | 31 +++++++
34
qemu-doc: Add UUID support in initiator name
57
include/block/block_int.h | 15 +---
35
58
include/block/write-threshold.h | 27 ++----
36
Stefan Hajnoczi (4):
59
include/qemu/job.h | 2 +-
37
scripts: add argparse module for Python 2.6 compatibility
60
block.c | 1 -
38
docker.py: Python 2.6 argparse compatibility
61
block/backup.c | 2 +-
39
tests: migration/guestperf Python 2.6 argparse compatibility
62
block/copy-on-read.c | 33 +------
40
qcow2: allocate cluster_cache/cluster_data on demand
63
block/io.c | 11 +--
41
64
block/mirror.c | 6 +-
42
include/qemu/throttle.h | 8 +-
65
block/monitor/block-hmp-cmds.c | 31 ++++---
43
block/qcow.c | 12 +-
66
block/rbd.c | 32 ++++---
44
block/qcow2-cluster.c | 17 +
67
block/write-threshold.c | 91 ++++---------------
45
block/qcow2.c | 20 +-
68
job.c | 2 +-
46
dump.c | 4 +-
69
qemu-io-cmds.c | 8 +-
47
hw/block/nvme.c | 4 +-
70
qemu-io.c | 17 +++-
48
tests/test-throttle.c | 80 +-
71
tests/unit/test-write-threshold.c | 90 ++-----------------
49
util/oslib-posix.c | 2 +
72
tests/qemu-iotests/231 | 4 +
50
util/throttle.c | 86 +-
73
tests/qemu-iotests/231.out | 7 +-
51
COPYING.PYTHON | 270 ++++
74
tests/qemu-iotests/240.out | 8 +-
52
qemu-doc.texi | 5 +-
75
tests/qemu-iotests/245.out | 8 +-
53
scripts/argparse.py | 2406 ++++++++++++++++++++++++++++++++++++
76
tests/qemu-iotests/264 | 2 +-
54
tests/docker/docker.py | 4 +-
77
tests/qemu-iotests/295.out | 6 +-
55
tests/migration/guestperf/shell.py | 8 +-
78
tests/qemu-iotests/296.out | 8 +-
56
14 files changed, 2831 insertions(+), 95 deletions(-)
79
tests/qemu-iotests/check | 19 +++-
57
create mode 100644 COPYING.PYTHON
80
tests/qemu-iotests/iotests.py | 145 +++++++++++++++++-------------
58
create mode 100644 scripts/argparse.py
81
tests/qemu-iotests/pylintrc | 3 +
82
tests/qemu-iotests/testenv.py | 22 ++++-
83
tests/qemu-iotests/testrunner.py | 37 +++-----
84
28 files changed, 289 insertions(+), 379 deletions(-)
59
85
60
--
86
--
61
2.13.5
87
2.31.1
62
88
63
89
diff view generated by jsdifflib
New patch
[PULL 01/19] iotests/231: Update expected deprecation message
1
From: Connor Kuehl <ckuehl@redhat.com>
1
2
3
The deprecation message in the expected output has technically been
4
wrong since the wrong version of a patch was applied to it. Because of
5
this, the test fails. Correct the expected output so that it passes.
6
7
Signed-off-by: Connor Kuehl <ckuehl@redhat.com>
8
Reviewed-by: Max Reitz <mreitz@redhat.com>
9
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
10
Message-Id: <20210421212343.85524-2-ckuehl@redhat.com>
11
Signed-off-by: Max Reitz <mreitz@redhat.com>
12
---
13
tests/qemu-iotests/231.out | 4 +---
14
1 file changed, 1 insertion(+), 3 deletions(-)
15
16
diff --git a/tests/qemu-iotests/231.out b/tests/qemu-iotests/231.out
17
index XXXXXXX..XXXXXXX 100644
18
--- a/tests/qemu-iotests/231.out
19
+++ b/tests/qemu-iotests/231.out
20
@@ -XXX,XX +XXX,XX @@
21
QA output created by 231
22
-qemu-img: RBD options encoded in the filename as keyvalue pairs is deprecated. Future versions may cease to parse these options in the future.
23
+qemu-img: warning: RBD options encoded in the filename as keyvalue pairs is deprecated
24
unable to get monitor info from DNS SRV with service name: ceph-mon
25
-no monitors specified to connect to.
26
qemu-img: Could not open 'json:{'file.driver':'rbd','file.filename':'rbd:rbd/bogus:conf=BOGUS_CONF'}': error connecting: No such file or directory
27
unable to get monitor info from DNS SRV with service name: ceph-mon
28
-no monitors specified to connect to.
29
qemu-img: Could not open 'json:{'file.driver':'rbd','file.pool':'rbd','file.image':'bogus','file.conf':'BOGUS_CONF'}': error connecting: No such file or directory
30
*** done
31
--
32
2.31.1
33
34
diff view generated by jsdifflib
[Qemu-devel] [PULL for-2.10 11/15] scripts: add argparse module for Python 2.6 compatibility
[PULL 02/19] block/rbd: Add an escape-aware strchr helper
1
The minimum Python version supported by QEMU is 2.6. The argparse
1
From: Connor Kuehl <ckuehl@redhat.com>
2
standard library module was only added in Python 2.7. Many scripts
3
would like to use argparse because it supports command-line
4
sub-commands.
5
2
6
This patch adds argparse. See the top of argparse.py for details.
3
Sometimes the parser needs to further split a token it has collected
4
from the token input stream. Right now, it does a cursory check to see
5
if the relevant characters appear in the token to determine if it should
6
break it down further.
7
7
8
Suggested-by: Daniel P. Berrange <berrange@redhat.com>
8
However, qemu_rbd_next_tok() will escape characters as it removes tokens
9
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
9
from the token stream and plain strchr() won't. This can make the
10
Acked-by: John Snow <jsnow@redhat.com>
10
initial strchr() check slightly misleading since it implies
11
Message-id: 20170825155732.15665-2-stefanha@redhat.com
11
qemu_rbd_next_tok() will find the token and split on it, except the
12
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
12
reality is that qemu_rbd_next_tok() will pass over it if it is escaped.
13
14
Use a custom strchr to avoid mixing escaped and unescaped string
15
operations. Furthermore, this code is identical to how
16
qemu_rbd_next_tok() seeks its next token, so incorporate this custom
17
strchr into the body of that function to reduce duplication.
18
19
Reported-by: Han Han <hhan@redhat.com>
20
Fixes: https://bugzilla.redhat.com/1873913
21
Signed-off-by: Connor Kuehl <ckuehl@redhat.com>
22
Message-Id: <20210421212343.85524-3-ckuehl@redhat.com>
23
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
24
Signed-off-by: Max Reitz <mreitz@redhat.com>
13
---
25
---
14
COPYING.PYTHON | 270 ++++++
26
block/rbd.c | 32 +++++++++++++++++++++-----------
15
scripts/argparse.py | 2406 +++++++++++++++++++++++++++++++++++++++++++++++++++
27
tests/qemu-iotests/231 | 4 ++++
16
2 files changed, 2676 insertions(+)
28
tests/qemu-iotests/231.out | 3 +++
17
create mode 100644 COPYING.PYTHON
29
3 files changed, 28 insertions(+), 11 deletions(-)
18
create mode 100644 scripts/argparse.py
19
30
20
diff --git a/COPYING.PYTHON b/COPYING.PYTHON
31
diff --git a/block/rbd.c b/block/rbd.c
21
new file mode 100644
32
index XXXXXXX..XXXXXXX 100644
22
index XXXXXXX..XXXXXXX
33
--- a/block/rbd.c
23
--- /dev/null
34
+++ b/block/rbd.c
24
+++ b/COPYING.PYTHON
35
@@ -XXX,XX +XXX,XX @@ static int qemu_rbd_connect(rados_t *cluster, rados_ioctx_t *io_ctx,
25
@@ -XXX,XX +XXX,XX @@
36
const char *keypairs, const char *secretid,
26
+A. HISTORY OF THE SOFTWARE
37
Error **errp);
27
+==========================
38
39
+static char *qemu_rbd_strchr(char *src, char delim)
40
+{
41
+ char *p;
28
+
42
+
29
+Python was created in the early 1990s by Guido van Rossum at Stichting
43
+ for (p = src; *p; ++p) {
30
+Mathematisch Centrum (CWI, see http://www.cwi.nl) in the Netherlands
44
+ if (*p == delim) {
31
+as a successor of a language called ABC. Guido remains Python's
45
+ return p;
32
+principal author, although it includes many contributions from others.
46
+ }
47
+ if (*p == '\\' && p[1] != '\0') {
48
+ ++p;
49
+ }
50
+ }
33
+
51
+
34
+In 1995, Guido continued his work on Python at the Corporation for
52
+ return NULL;
35
+National Research Initiatives (CNRI, see http://www.cnri.reston.va.us)
53
+}
36
+in Reston, Virginia where he released several versions of the
37
+software.
38
+
39
+In May 2000, Guido and the Python core development team moved to
40
+BeOpen.com to form the BeOpen PythonLabs team. In October of the same
41
+year, the PythonLabs team moved to Digital Creations (now Zope
42
+Corporation, see http://www.zope.com). In 2001, the Python Software
43
+Foundation (PSF, see http://www.python.org/psf/) was formed, a
44
+non-profit organization created specifically to own Python-related
45
+Intellectual Property. Zope Corporation is a sponsoring member of
46
+the PSF.
47
+
48
+All Python releases are Open Source (see http://www.opensource.org for
49
+the Open Source Definition). Historically, most, but not all, Python
50
+releases have also been GPL-compatible; the table below summarizes
51
+the various releases.
52
+
53
+ Release Derived Year Owner GPL-
54
+ from compatible? (1)
55
+
56
+ 0.9.0 thru 1.2 1991-1995 CWI yes
57
+ 1.3 thru 1.5.2 1.2 1995-1999 CNRI yes
58
+ 1.6 1.5.2 2000 CNRI no
59
+ 2.0 1.6 2000 BeOpen.com no
60
+ 1.6.1 1.6 2001 CNRI yes (2)
61
+ 2.1 2.0+1.6.1 2001 PSF no
62
+ 2.0.1 2.0+1.6.1 2001 PSF yes
63
+ 2.1.1 2.1+2.0.1 2001 PSF yes
64
+ 2.2 2.1.1 2001 PSF yes
65
+ 2.1.2 2.1.1 2002 PSF yes
66
+ 2.1.3 2.1.2 2002 PSF yes
67
+ 2.2.1 2.2 2002 PSF yes
68
+ 2.2.2 2.2.1 2002 PSF yes
69
+ 2.2.3 2.2.2 2003 PSF yes
70
+ 2.3 2.2.2 2002-2003 PSF yes
71
+ 2.3.1 2.3 2002-2003 PSF yes
72
+ 2.3.2 2.3.1 2002-2003 PSF yes
73
+ 2.3.3 2.3.2 2002-2003 PSF yes
74
+ 2.3.4 2.3.3 2004 PSF yes
75
+ 2.3.5 2.3.4 2005 PSF yes
76
+ 2.4 2.3 2004 PSF yes
77
+ 2.4.1 2.4 2005 PSF yes
78
+ 2.4.2 2.4.1 2005 PSF yes
79
+ 2.4.3 2.4.2 2006 PSF yes
80
+ 2.5 2.4 2006 PSF yes
81
+ 2.7 2.6 2010 PSF yes
82
+
83
+Footnotes:
84
+
85
+(1) GPL-compatible doesn't mean that we're distributing Python under
86
+ the GPL. All Python licenses, unlike the GPL, let you distribute
87
+ a modified version without making your changes open source. The
88
+ GPL-compatible licenses make it possible to combine Python with
89
+ other software that is released under the GPL; the others don't.
90
+
91
+(2) According to Richard Stallman, 1.6.1 is not GPL-compatible,
92
+ because its license has a choice of law clause. According to
93
+ CNRI, however, Stallman's lawyer has told CNRI's lawyer that 1.6.1
94
+ is "not incompatible" with the GPL.
95
+
96
+Thanks to the many outside volunteers who have worked under Guido's
97
+direction to make these releases possible.
98
+
54
+
99
+
55
+
100
+B. TERMS AND CONDITIONS FOR ACCESSING OR OTHERWISE USING PYTHON
56
static char *qemu_rbd_next_tok(char *src, char delim, char **p)
101
+===============================================================
57
{
58
char *end;
59
60
*p = NULL;
61
62
- for (end = src; *end; ++end) {
63
- if (*end == delim) {
64
- break;
65
- }
66
- if (*end == '\\' && end[1] != '\0') {
67
- end++;
68
- }
69
- }
70
- if (*end == delim) {
71
+ end = qemu_rbd_strchr(src, delim);
72
+ if (end) {
73
*p = end + 1;
74
*end = '\0';
75
}
76
@@ -XXX,XX +XXX,XX @@ static void qemu_rbd_parse_filename(const char *filename, QDict *options,
77
qemu_rbd_unescape(found_str);
78
qdict_put_str(options, "pool", found_str);
79
80
- if (strchr(p, '@')) {
81
+ if (qemu_rbd_strchr(p, '@')) {
82
image_name = qemu_rbd_next_tok(p, '@', &p);
83
84
found_str = qemu_rbd_next_tok(p, ':', &p);
85
@@ -XXX,XX +XXX,XX @@ static void qemu_rbd_parse_filename(const char *filename, QDict *options,
86
image_name = qemu_rbd_next_tok(p, ':', &p);
87
}
88
/* Check for namespace in the image_name */
89
- if (strchr(image_name, '/')) {
90
+ if (qemu_rbd_strchr(image_name, '/')) {
91
found_str = qemu_rbd_next_tok(image_name, '/', &image_name);
92
qemu_rbd_unescape(found_str);
93
qdict_put_str(options, "namespace", found_str);
94
diff --git a/tests/qemu-iotests/231 b/tests/qemu-iotests/231
95
index XXXXXXX..XXXXXXX 100755
96
--- a/tests/qemu-iotests/231
97
+++ b/tests/qemu-iotests/231
98
@@ -XXX,XX +XXX,XX @@ _filter_conf()
99
$QEMU_IMG info "json:{'file.driver':'rbd','file.filename':'rbd:rbd/bogus:conf=${BOGUS_CONF}'}" 2>&1 | _filter_conf
100
$QEMU_IMG info "json:{'file.driver':'rbd','file.pool':'rbd','file.image':'bogus','file.conf':'${BOGUS_CONF}'}" 2>&1 | _filter_conf
101
102
+# Regression test: the qemu-img invocation is expected to fail, but it should
103
+# not seg fault the parser.
104
+$QEMU_IMG create "rbd:rbd/aa\/bb:conf=${BOGUS_CONF}" 1M 2>&1 | _filter_conf
102
+
105
+
103
+PYTHON SOFTWARE FOUNDATION LICENSE VERSION 2
106
# success, all done
104
+--------------------------------------------
107
echo "*** done"
105
+
108
rm -f $seq.full
106
+1. This LICENSE AGREEMENT is between the Python Software Foundation
109
diff --git a/tests/qemu-iotests/231.out b/tests/qemu-iotests/231.out
107
+("PSF"), and the Individual or Organization ("Licensee") accessing and
110
index XXXXXXX..XXXXXXX 100644
108
+otherwise using this software ("Python") in source or binary form and
111
--- a/tests/qemu-iotests/231.out
109
+its associated documentation.
112
+++ b/tests/qemu-iotests/231.out
110
+
113
@@ -XXX,XX +XXX,XX @@ unable to get monitor info from DNS SRV with service name: ceph-mon
111
+2. Subject to the terms and conditions of this License Agreement, PSF
114
qemu-img: Could not open 'json:{'file.driver':'rbd','file.filename':'rbd:rbd/bogus:conf=BOGUS_CONF'}': error connecting: No such file or directory
112
+hereby grants Licensee a nonexclusive, royalty-free, world-wide
115
unable to get monitor info from DNS SRV with service name: ceph-mon
113
+license to reproduce, analyze, test, perform and/or display publicly,
116
qemu-img: Could not open 'json:{'file.driver':'rbd','file.pool':'rbd','file.image':'bogus','file.conf':'BOGUS_CONF'}': error connecting: No such file or directory
114
+prepare derivative works, distribute, and otherwise use Python
117
+Formatting 'rbd:rbd/aa\/bb:conf=BOGUS_CONF', fmt=raw size=1048576
115
+alone or in any derivative version, provided, however, that PSF's
118
+unable to get monitor info from DNS SRV with service name: ceph-mon
116
+License Agreement and PSF's notice of copyright, i.e., "Copyright (c)
119
+qemu-img: rbd:rbd/aa\/bb:conf=BOGUS_CONF: error connecting: No such file or directory
117
+2001, 2002, 2003, 2004, 2005, 2006 Python Software Foundation; All Rights
120
*** done
118
+Reserved" are retained in Python alone or in any derivative version
119
+prepared by Licensee.
120
+
121
+3. In the event Licensee prepares a derivative work that is based on
122
+or incorporates Python or any part thereof, and wants to make
123
+the derivative work available to others as provided herein, then
124
+Licensee hereby agrees to include in any such work a brief summary of
125
+the changes made to Python.
126
+
127
+4. PSF is making Python available to Licensee on an "AS IS"
128
+basis. PSF MAKES NO REPRESENTATIONS OR WARRANTIES, EXPRESS OR
129
+IMPLIED. BY WAY OF EXAMPLE, BUT NOT LIMITATION, PSF MAKES NO AND
130
+DISCLAIMS ANY REPRESENTATION OR WARRANTY OF MERCHANTABILITY OR FITNESS
131
+FOR ANY PARTICULAR PURPOSE OR THAT THE USE OF PYTHON WILL NOT
132
+INFRINGE ANY THIRD PARTY RIGHTS.
133
+
134
+5. PSF SHALL NOT BE LIABLE TO LICENSEE OR ANY OTHER USERS OF PYTHON
135
+FOR ANY INCIDENTAL, SPECIAL, OR CONSEQUENTIAL DAMAGES OR LOSS AS
136
+A RESULT OF MODIFYING, DISTRIBUTING, OR OTHERWISE USING PYTHON,
137
+OR ANY DERIVATIVE THEREOF, EVEN IF ADVISED OF THE POSSIBILITY THEREOF.
138
+
139
+6. This License Agreement will automatically terminate upon a material
140
+breach of its terms and conditions.
141
+
142
+7. Nothing in this License Agreement shall be deemed to create any
143
+relationship of agency, partnership, or joint venture between PSF and
144
+Licensee. This License Agreement does not grant permission to use PSF
145
+trademarks or trade name in a trademark sense to endorse or promote
146
+products or services of Licensee, or any third party.
147
+
148
+8. By copying, installing or otherwise using Python, Licensee
149
+agrees to be bound by the terms and conditions of this License
150
+Agreement.
151
+
152
+
153
+BEOPEN.COM LICENSE AGREEMENT FOR PYTHON 2.0
154
+-------------------------------------------
155
+
156
+BEOPEN PYTHON OPEN SOURCE LICENSE AGREEMENT VERSION 1
157
+
158
+1. This LICENSE AGREEMENT is between BeOpen.com ("BeOpen"), having an
159
+office at 160 Saratoga Avenue, Santa Clara, CA 95051, and the
160
+Individual or Organization ("Licensee") accessing and otherwise using
161
+this software in source or binary form and its associated
162
+documentation ("the Software").
163
+
164
+2. Subject to the terms and conditions of this BeOpen Python License
165
+Agreement, BeOpen hereby grants Licensee a non-exclusive,
166
+royalty-free, world-wide license to reproduce, analyze, test, perform
167
+and/or display publicly, prepare derivative works, distribute, and
168
+otherwise use the Software alone or in any derivative version,
169
+provided, however, that the BeOpen Python License is retained in the
170
+Software, alone or in any derivative version prepared by Licensee.
171
+
172
+3. BeOpen is making the Software available to Licensee on an "AS IS"
173
+basis. BEOPEN MAKES NO REPRESENTATIONS OR WARRANTIES, EXPRESS OR
174
+IMPLIED. BY WAY OF EXAMPLE, BUT NOT LIMITATION, BEOPEN MAKES NO AND
175
+DISCLAIMS ANY REPRESENTATION OR WARRANTY OF MERCHANTABILITY OR FITNESS
176
+FOR ANY PARTICULAR PURPOSE OR THAT THE USE OF THE SOFTWARE WILL NOT
177
+INFRINGE ANY THIRD PARTY RIGHTS.
178
+
179
+4. BEOPEN SHALL NOT BE LIABLE TO LICENSEE OR ANY OTHER USERS OF THE
180
+SOFTWARE FOR ANY INCIDENTAL, SPECIAL, OR CONSEQUENTIAL DAMAGES OR LOSS
181
+AS A RESULT OF USING, MODIFYING OR DISTRIBUTING THE SOFTWARE, OR ANY
182
+DERIVATIVE THEREOF, EVEN IF ADVISED OF THE POSSIBILITY THEREOF.
183
+
184
+5. This License Agreement will automatically terminate upon a material
185
+breach of its terms and conditions.
186
+
187
+6. This License Agreement shall be governed by and interpreted in all
188
+respects by the law of the State of California, excluding conflict of
189
+law provisions. Nothing in this License Agreement shall be deemed to
190
+create any relationship of agency, partnership, or joint venture
191
+between BeOpen and Licensee. This License Agreement does not grant
192
+permission to use BeOpen trademarks or trade names in a trademark
193
+sense to endorse or promote products or services of Licensee, or any
194
+third party. As an exception, the "BeOpen Python" logos available at
195
+http://www.pythonlabs.com/logos.html may be used according to the
196
+permissions granted on that web page.
197
+
198
+7. By copying, installing or otherwise using the software, Licensee
199
+agrees to be bound by the terms and conditions of this License
200
+Agreement.
201
+
202
+
203
+CNRI LICENSE AGREEMENT FOR PYTHON 1.6.1
204
+---------------------------------------
205
+
206
+1. This LICENSE AGREEMENT is between the Corporation for National
207
+Research Initiatives, having an office at 1895 Preston White Drive,
208
+Reston, VA 20191 ("CNRI"), and the Individual or Organization
209
+("Licensee") accessing and otherwise using Python 1.6.1 software in
210
+source or binary form and its associated documentation.
211
+
212
+2. Subject to the terms and conditions of this License Agreement, CNRI
213
+hereby grants Licensee a nonexclusive, royalty-free, world-wide
214
+license to reproduce, analyze, test, perform and/or display publicly,
215
+prepare derivative works, distribute, and otherwise use Python 1.6.1
216
+alone or in any derivative version, provided, however, that CNRI's
217
+License Agreement and CNRI's notice of copyright, i.e., "Copyright (c)
218
+1995-2001 Corporation for National Research Initiatives; All Rights
219
+Reserved" are retained in Python 1.6.1 alone or in any derivative
220
+version prepared by Licensee. Alternately, in lieu of CNRI's License
221
+Agreement, Licensee may substitute the following text (omitting the
222
+quotes): "Python 1.6.1 is made available subject to the terms and
223
+conditions in CNRI's License Agreement. This Agreement together with
224
+Python 1.6.1 may be located on the Internet using the following
225
+unique, persistent identifier (known as a handle): 1895.22/1013. This
226
+Agreement may also be obtained from a proxy server on the Internet
227
+using the following URL: http://hdl.handle.net/1895.22/1013".
228
+
229
+3. In the event Licensee prepares a derivative work that is based on
230
+or incorporates Python 1.6.1 or any part thereof, and wants to make
231
+the derivative work available to others as provided herein, then
232
+Licensee hereby agrees to include in any such work a brief summary of
233
+the changes made to Python 1.6.1.
234
+
235
+4. CNRI is making Python 1.6.1 available to Licensee on an "AS IS"
236
+basis. CNRI MAKES NO REPRESENTATIONS OR WARRANTIES, EXPRESS OR
237
+IMPLIED. BY WAY OF EXAMPLE, BUT NOT LIMITATION, CNRI MAKES NO AND
238
+DISCLAIMS ANY REPRESENTATION OR WARRANTY OF MERCHANTABILITY OR FITNESS
239
+FOR ANY PARTICULAR PURPOSE OR THAT THE USE OF PYTHON 1.6.1 WILL NOT
240
+INFRINGE ANY THIRD PARTY RIGHTS.
241
+
242
+5. CNRI SHALL NOT BE LIABLE TO LICENSEE OR ANY OTHER USERS OF PYTHON
243
+1.6.1 FOR ANY INCIDENTAL, SPECIAL, OR CONSEQUENTIAL DAMAGES OR LOSS AS
244
+A RESULT OF MODIFYING, DISTRIBUTING, OR OTHERWISE USING PYTHON 1.6.1,
245
+OR ANY DERIVATIVE THEREOF, EVEN IF ADVISED OF THE POSSIBILITY THEREOF.
246
+
247
+6. This License Agreement will automatically terminate upon a material
248
+breach of its terms and conditions.
249
+
250
+7. This License Agreement shall be governed by the federal
251
+intellectual property law of the United States, including without
252
+limitation the federal copyright law, and, to the extent such
253
+U.S. federal law does not apply, by the law of the Commonwealth of
254
+Virginia, excluding Virginia's conflict of law provisions.
255
+Notwithstanding the foregoing, with regard to derivative works based
256
+on Python 1.6.1 that incorporate non-separable material that was
257
+previously distributed under the GNU General Public License (GPL), the
258
+law of the Commonwealth of Virginia shall govern this License
259
+Agreement only as to issues arising under or with respect to
260
+Paragraphs 4, 5, and 7 of this License Agreement. Nothing in this
261
+License Agreement shall be deemed to create any relationship of
262
+agency, partnership, or joint venture between CNRI and Licensee. This
263
+License Agreement does not grant permission to use CNRI trademarks or
264
+trade name in a trademark sense to endorse or promote products or
265
+services of Licensee, or any third party.
266
+
267
+8. By clicking on the "ACCEPT" button where indicated, or by copying,
268
+installing or otherwise using Python 1.6.1, Licensee agrees to be
269
+bound by the terms and conditions of this License Agreement.
270
+
271
+ ACCEPT
272
+
273
+
274
+CWI LICENSE AGREEMENT FOR PYTHON 0.9.0 THROUGH 1.2
275
+--------------------------------------------------
276
+
277
+Copyright (c) 1991 - 1995, Stichting Mathematisch Centrum Amsterdam,
278
+The Netherlands. All rights reserved.
279
+
280
+Permission to use, copy, modify, and distribute this software and its
281
+documentation for any purpose and without fee is hereby granted,
282
+provided that the above copyright notice appear in all copies and that
283
+both that copyright notice and this permission notice appear in
284
+supporting documentation, and that the name of Stichting Mathematisch
285
+Centrum or CWI not be used in advertising or publicity pertaining to
286
+distribution of the software without specific, written prior
287
+permission.
288
+
289
+STICHTING MATHEMATISCH CENTRUM DISCLAIMS ALL WARRANTIES WITH REGARD TO
290
+THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
291
+FITNESS, IN NO EVENT SHALL STICHTING MATHEMATISCH CENTRUM BE LIABLE
292
+FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
293
+WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
294
+ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT
295
+OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
296
diff --git a/scripts/argparse.py b/scripts/argparse.py
297
new file mode 100644
298
index XXXXXXX..XXXXXXX
299
--- /dev/null
300
+++ b/scripts/argparse.py
301
@@ -XXX,XX +XXX,XX @@
302
+# This is a local copy of the standard library argparse module taken from PyPI.
303
+# It is licensed under the Python Software Foundation License. This is a
304
+# fallback for Python 2.6 which does not include this module. Python 2.7+ and
305
+# 3+ will never load this module because built-in modules are loaded before
306
+# anything in sys.path.
307
+#
308
+# If your script is not located in the same directory as this file, import it
309
+# like this:
310
+#
311
+# import os
312
+# import sys
313
+# sys.path.append(os.path.join(os.path.dirname(__file__), ..., 'scripts'))
314
+# import argparse
315
+
316
+# Author: Steven J. Bethard <steven.bethard@gmail.com>.
317
+# Maintainer: Thomas Waldmann <tw@waldmann-edv.de>
318
+
319
+"""Command-line parsing library
320
+
321
+This module is an optparse-inspired command-line parsing library that:
322
+
323
+ - handles both optional and positional arguments
324
+ - produces highly informative usage messages
325
+ - supports parsers that dispatch to sub-parsers
326
+
327
+The following is a simple usage example that sums integers from the
328
+command-line and writes the result to a file::
329
+
330
+ parser = argparse.ArgumentParser(
331
+ description='sum the integers at the command line')
332
+ parser.add_argument(
333
+ 'integers', metavar='int', nargs='+', type=int,
334
+ help='an integer to be summed')
335
+ parser.add_argument(
336
+ '--log', default=sys.stdout, type=argparse.FileType('w'),
337
+ help='the file where the sum should be written')
338
+ args = parser.parse_args()
339
+ args.log.write('%s' % sum(args.integers))
340
+ args.log.close()
341
+
342
+The module contains the following public classes:
343
+
344
+ - ArgumentParser -- The main entry point for command-line parsing. As the
345
+ example above shows, the add_argument() method is used to populate
346
+ the parser with actions for optional and positional arguments. Then
347
+ the parse_args() method is invoked to convert the args at the
348
+ command-line into an object with attributes.
349
+
350
+ - ArgumentError -- The exception raised by ArgumentParser objects when
351
+ there are errors with the parser's actions. Errors raised while
352
+ parsing the command-line are caught by ArgumentParser and emitted
353
+ as command-line messages.
354
+
355
+ - FileType -- A factory for defining types of files to be created. As the
356
+ example above shows, instances of FileType are typically passed as
357
+ the type= argument of add_argument() calls.
358
+
359
+ - Action -- The base class for parser actions. Typically actions are
360
+ selected by passing strings like 'store_true' or 'append_const' to
361
+ the action= argument of add_argument(). However, for greater
362
+ customization of ArgumentParser actions, subclasses of Action may
363
+ be defined and passed as the action= argument.
364
+
365
+ - HelpFormatter, RawDescriptionHelpFormatter, RawTextHelpFormatter,
366
+ ArgumentDefaultsHelpFormatter -- Formatter classes which
367
+ may be passed as the formatter_class= argument to the
368
+ ArgumentParser constructor. HelpFormatter is the default,
369
+ RawDescriptionHelpFormatter and RawTextHelpFormatter tell the parser
370
+ not to change the formatting for help text, and
371
+ ArgumentDefaultsHelpFormatter adds information about argument defaults
372
+ to the help.
373
+
374
+All other classes in this module are considered implementation details.
375
+(Also note that HelpFormatter and RawDescriptionHelpFormatter are only
376
+considered public as object names -- the API of the formatter objects is
377
+still considered an implementation detail.)
378
+"""
379
+
380
+__version__ = '1.4.0' # we use our own version number independant of the
381
+ # one in stdlib and we release this on pypi.
382
+
383
+__external_lib__ = True # to make sure the tests really test THIS lib,
384
+ # not the builtin one in Python stdlib
385
+
386
+__all__ = [
387
+ 'ArgumentParser',
388
+ 'ArgumentError',
389
+ 'ArgumentTypeError',
390
+ 'FileType',
391
+ 'HelpFormatter',
392
+ 'ArgumentDefaultsHelpFormatter',
393
+ 'RawDescriptionHelpFormatter',
394
+ 'RawTextHelpFormatter',
395
+ 'Namespace',
396
+ 'Action',
397
+ 'ONE_OR_MORE',
398
+ 'OPTIONAL',
399
+ 'PARSER',
400
+ 'REMAINDER',
401
+ 'SUPPRESS',
402
+ 'ZERO_OR_MORE',
403
+]
404
+
405
+
406
+import copy as _copy
407
+import os as _os
408
+import re as _re
409
+import sys as _sys
410
+import textwrap as _textwrap
411
+
412
+from gettext import gettext as _
413
+
414
+try:
415
+ set
416
+except NameError:
417
+ # for python < 2.4 compatibility (sets module is there since 2.3):
418
+ from sets import Set as set
419
+
420
+try:
421
+ basestring
422
+except NameError:
423
+ basestring = str
424
+
425
+try:
426
+ sorted
427
+except NameError:
428
+ # for python < 2.4 compatibility:
429
+ def sorted(iterable, reverse=False):
430
+ result = list(iterable)
431
+ result.sort()
432
+ if reverse:
433
+ result.reverse()
434
+ return result
435
+
436
+
437
+def _callable(obj):
438
+ return hasattr(obj, '__call__') or hasattr(obj, '__bases__')
439
+
440
+
441
+SUPPRESS = '==SUPPRESS=='
442
+
443
+OPTIONAL = '?'
444
+ZERO_OR_MORE = '*'
445
+ONE_OR_MORE = '+'
446
+PARSER = 'A...'
447
+REMAINDER = '...'
448
+_UNRECOGNIZED_ARGS_ATTR = '_unrecognized_args'
449
+
450
+# =============================
451
+# Utility functions and classes
452
+# =============================
453
+
454
+class _AttributeHolder(object):
455
+ """Abstract base class that provides __repr__.
456
+
457
+ The __repr__ method returns a string in the format::
458
+ ClassName(attr=name, attr=name, ...)
459
+ The attributes are determined either by a class-level attribute,
460
+ '_kwarg_names', or by inspecting the instance __dict__.
461
+ """
462
+
463
+ def __repr__(self):
464
+ type_name = type(self).__name__
465
+ arg_strings = []
466
+ for arg in self._get_args():
467
+ arg_strings.append(repr(arg))
468
+ for name, value in self._get_kwargs():
469
+ arg_strings.append('%s=%r' % (name, value))
470
+ return '%s(%s)' % (type_name, ', '.join(arg_strings))
471
+
472
+ def _get_kwargs(self):
473
+ return sorted(self.__dict__.items())
474
+
475
+ def _get_args(self):
476
+ return []
477
+
478
+
479
+def _ensure_value(namespace, name, value):
480
+ if getattr(namespace, name, None) is None:
481
+ setattr(namespace, name, value)
482
+ return getattr(namespace, name)
483
+
484
+
485
+# ===============
486
+# Formatting Help
487
+# ===============
488
+
489
+class HelpFormatter(object):
490
+ """Formatter for generating usage messages and argument help strings.
491
+
492
+ Only the name of this class is considered a public API. All the methods
493
+ provided by the class are considered an implementation detail.
494
+ """
495
+
496
+ def __init__(self,
497
+ prog,
498
+ indent_increment=2,
499
+ max_help_position=24,
500
+ width=None):
501
+
502
+ # default setting for width
503
+ if width is None:
504
+ try:
505
+ width = int(_os.environ['COLUMNS'])
506
+ except (KeyError, ValueError):
507
+ width = 80
508
+ width -= 2
509
+
510
+ self._prog = prog
511
+ self._indent_increment = indent_increment
512
+ self._max_help_position = max_help_position
513
+ self._width = width
514
+
515
+ self._current_indent = 0
516
+ self._level = 0
517
+ self._action_max_length = 0
518
+
519
+ self._root_section = self._Section(self, None)
520
+ self._current_section = self._root_section
521
+
522
+ self._whitespace_matcher = _re.compile(r'\s+')
523
+ self._long_break_matcher = _re.compile(r'\n\n\n+')
524
+
525
+ # ===============================
526
+ # Section and indentation methods
527
+ # ===============================
528
+ def _indent(self):
529
+ self._current_indent += self._indent_increment
530
+ self._level += 1
531
+
532
+ def _dedent(self):
533
+ self._current_indent -= self._indent_increment
534
+ assert self._current_indent >= 0, 'Indent decreased below 0.'
535
+ self._level -= 1
536
+
537
+ class _Section(object):
538
+
539
+ def __init__(self, formatter, parent, heading=None):
540
+ self.formatter = formatter
541
+ self.parent = parent
542
+ self.heading = heading
543
+ self.items = []
544
+
545
+ def format_help(self):
546
+ # format the indented section
547
+ if self.parent is not None:
548
+ self.formatter._indent()
549
+ join = self.formatter._join_parts
550
+ for func, args in self.items:
551
+ func(*args)
552
+ item_help = join([func(*args) for func, args in self.items])
553
+ if self.parent is not None:
554
+ self.formatter._dedent()
555
+
556
+ # return nothing if the section was empty
557
+ if not item_help:
558
+ return ''
559
+
560
+ # add the heading if the section was non-empty
561
+ if self.heading is not SUPPRESS and self.heading is not None:
562
+ current_indent = self.formatter._current_indent
563
+ heading = '%*s%s:\n' % (current_indent, '', self.heading)
564
+ else:
565
+ heading = ''
566
+
567
+ # join the section-initial newline, the heading and the help
568
+ return join(['\n', heading, item_help, '\n'])
569
+
570
+ def _add_item(self, func, args):
571
+ self._current_section.items.append((func, args))
572
+
573
+ # ========================
574
+ # Message building methods
575
+ # ========================
576
+ def start_section(self, heading):
577
+ self._indent()
578
+ section = self._Section(self, self._current_section, heading)
579
+ self._add_item(section.format_help, [])
580
+ self._current_section = section
581
+
582
+ def end_section(self):
583
+ self._current_section = self._current_section.parent
584
+ self._dedent()
585
+
586
+ def add_text(self, text):
587
+ if text is not SUPPRESS and text is not None:
588
+ self._add_item(self._format_text, [text])
589
+
590
+ def add_usage(self, usage, actions, groups, prefix=None):
591
+ if usage is not SUPPRESS:
592
+ args = usage, actions, groups, prefix
593
+ self._add_item(self._format_usage, args)
594
+
595
+ def add_argument(self, action):
596
+ if action.help is not SUPPRESS:
597
+
598
+ # find all invocations
599
+ get_invocation = self._format_action_invocation
600
+ invocations = [get_invocation(action)]
601
+ for subaction in self._iter_indented_subactions(action):
602
+ invocations.append(get_invocation(subaction))
603
+
604
+ # update the maximum item length
605
+ invocation_length = max([len(s) for s in invocations])
606
+ action_length = invocation_length + self._current_indent
607
+ self._action_max_length = max(self._action_max_length,
608
+ action_length)
609
+
610
+ # add the item to the list
611
+ self._add_item(self._format_action, [action])
612
+
613
+ def add_arguments(self, actions):
614
+ for action in actions:
615
+ self.add_argument(action)
616
+
617
+ # =======================
618
+ # Help-formatting methods
619
+ # =======================
620
+ def format_help(self):
621
+ help = self._root_section.format_help()
622
+ if help:
623
+ help = self._long_break_matcher.sub('\n\n', help)
624
+ help = help.strip('\n') + '\n'
625
+ return help
626
+
627
+ def _join_parts(self, part_strings):
628
+ return ''.join([part
629
+ for part in part_strings
630
+ if part and part is not SUPPRESS])
631
+
632
+ def _format_usage(self, usage, actions, groups, prefix):
633
+ if prefix is None:
634
+ prefix = _('usage: ')
635
+
636
+ # if usage is specified, use that
637
+ if usage is not None:
638
+ usage = usage % dict(prog=self._prog)
639
+
640
+ # if no optionals or positionals are available, usage is just prog
641
+ elif usage is None and not actions:
642
+ usage = '%(prog)s' % dict(prog=self._prog)
643
+
644
+ # if optionals and positionals are available, calculate usage
645
+ elif usage is None:
646
+ prog = '%(prog)s' % dict(prog=self._prog)
647
+
648
+ # split optionals from positionals
649
+ optionals = []
650
+ positionals = []
651
+ for action in actions:
652
+ if action.option_strings:
653
+ optionals.append(action)
654
+ else:
655
+ positionals.append(action)
656
+
657
+ # build full usage string
658
+ format = self._format_actions_usage
659
+ action_usage = format(optionals + positionals, groups)
660
+ usage = ' '.join([s for s in [prog, action_usage] if s])
661
+
662
+ # wrap the usage parts if it's too long
663
+ text_width = self._width - self._current_indent
664
+ if len(prefix) + len(usage) > text_width:
665
+
666
+ # break usage into wrappable parts
667
+ part_regexp = r'\(.*?\)+|\[.*?\]+|\S+'
668
+ opt_usage = format(optionals, groups)
669
+ pos_usage = format(positionals, groups)
670
+ opt_parts = _re.findall(part_regexp, opt_usage)
671
+ pos_parts = _re.findall(part_regexp, pos_usage)
672
+ assert ' '.join(opt_parts) == opt_usage
673
+ assert ' '.join(pos_parts) == pos_usage
674
+
675
+ # helper for wrapping lines
676
+ def get_lines(parts, indent, prefix=None):
677
+ lines = []
678
+ line = []
679
+ if prefix is not None:
680
+ line_len = len(prefix) - 1
681
+ else:
682
+ line_len = len(indent) - 1
683
+ for part in parts:
684
+ if line_len + 1 + len(part) > text_width:
685
+ lines.append(indent + ' '.join(line))
686
+ line = []
687
+ line_len = len(indent) - 1
688
+ line.append(part)
689
+ line_len += len(part) + 1
690
+ if line:
691
+ lines.append(indent + ' '.join(line))
692
+ if prefix is not None:
693
+ lines[0] = lines[0][len(indent):]
694
+ return lines
695
+
696
+ # if prog is short, follow it with optionals or positionals
697
+ if len(prefix) + len(prog) <= 0.75 * text_width:
698
+ indent = ' ' * (len(prefix) + len(prog) + 1)
699
+ if opt_parts:
700
+ lines = get_lines([prog] + opt_parts, indent, prefix)
701
+ lines.extend(get_lines(pos_parts, indent))
702
+ elif pos_parts:
703
+ lines = get_lines([prog] + pos_parts, indent, prefix)
704
+ else:
705
+ lines = [prog]
706
+
707
+ # if prog is long, put it on its own line
708
+ else:
709
+ indent = ' ' * len(prefix)
710
+ parts = opt_parts + pos_parts
711
+ lines = get_lines(parts, indent)
712
+ if len(lines) > 1:
713
+ lines = []
714
+ lines.extend(get_lines(opt_parts, indent))
715
+ lines.extend(get_lines(pos_parts, indent))
716
+ lines = [prog] + lines
717
+
718
+ # join lines into usage
719
+ usage = '\n'.join(lines)
720
+
721
+ # prefix with 'usage:'
722
+ return '%s%s\n\n' % (prefix, usage)
723
+
724
+ def _format_actions_usage(self, actions, groups):
725
+ # find group indices and identify actions in groups
726
+ group_actions = set()
727
+ inserts = {}
728
+ for group in groups:
729
+ try:
730
+ start = actions.index(group._group_actions[0])
731
+ except ValueError:
732
+ continue
733
+ else:
734
+ end = start + len(group._group_actions)
735
+ if actions[start:end] == group._group_actions:
736
+ for action in group._group_actions:
737
+ group_actions.add(action)
738
+ if not group.required:
739
+ if start in inserts:
740
+ inserts[start] += ' ['
741
+ else:
742
+ inserts[start] = '['
743
+ inserts[end] = ']'
744
+ else:
745
+ if start in inserts:
746
+ inserts[start] += ' ('
747
+ else:
748
+ inserts[start] = '('
749
+ inserts[end] = ')'
750
+ for i in range(start + 1, end):
751
+ inserts[i] = '|'
752
+
753
+ # collect all actions format strings
754
+ parts = []
755
+ for i, action in enumerate(actions):
756
+
757
+ # suppressed arguments are marked with None
758
+ # remove | separators for suppressed arguments
759
+ if action.help is SUPPRESS:
760
+ parts.append(None)
761
+ if inserts.get(i) == '|':
762
+ inserts.pop(i)
763
+ elif inserts.get(i + 1) == '|':
764
+ inserts.pop(i + 1)
765
+
766
+ # produce all arg strings
767
+ elif not action.option_strings:
768
+ part = self._format_args(action, action.dest)
769
+
770
+ # if it's in a group, strip the outer []
771
+ if action in group_actions:
772
+ if part[0] == '[' and part[-1] == ']':
773
+ part = part[1:-1]
774
+
775
+ # add the action string to the list
776
+ parts.append(part)
777
+
778
+ # produce the first way to invoke the option in brackets
779
+ else:
780
+ option_string = action.option_strings[0]
781
+
782
+ # if the Optional doesn't take a value, format is:
783
+ # -s or --long
784
+ if action.nargs == 0:
785
+ part = '%s' % option_string
786
+
787
+ # if the Optional takes a value, format is:
788
+ # -s ARGS or --long ARGS
789
+ else:
790
+ default = action.dest.upper()
791
+ args_string = self._format_args(action, default)
792
+ part = '%s %s' % (option_string, args_string)
793
+
794
+ # make it look optional if it's not required or in a group
795
+ if not action.required and action not in group_actions:
796
+ part = '[%s]' % part
797
+
798
+ # add the action string to the list
799
+ parts.append(part)
800
+
801
+ # insert things at the necessary indices
802
+ for i in sorted(inserts, reverse=True):
803
+ parts[i:i] = [inserts[i]]
804
+
805
+ # join all the action items with spaces
806
+ text = ' '.join([item for item in parts if item is not None])
807
+
808
+ # clean up separators for mutually exclusive groups
809
+ open = r'[\[(]'
810
+ close = r'[\])]'
811
+ text = _re.sub(r'(%s) ' % open, r'\1', text)
812
+ text = _re.sub(r' (%s)' % close, r'\1', text)
813
+ text = _re.sub(r'%s *%s' % (open, close), r'', text)
814
+ text = _re.sub(r'\(([^|]*)\)', r'\1', text)
815
+ text = text.strip()
816
+
817
+ # return the text
818
+ return text
819
+
820
+ def _format_text(self, text):
821
+ if '%(prog)' in text:
822
+ text = text % dict(prog=self._prog)
823
+ text_width = self._width - self._current_indent
824
+ indent = ' ' * self._current_indent
825
+ return self._fill_text(text, text_width, indent) + '\n\n'
826
+
827
+ def _format_action(self, action):
828
+ # determine the required width and the entry label
829
+ help_position = min(self._action_max_length + 2,
830
+ self._max_help_position)
831
+ help_width = self._width - help_position
832
+ action_width = help_position - self._current_indent - 2
833
+ action_header = self._format_action_invocation(action)
834
+
835
+ # ho nelp; start on same line and add a final newline
836
+ if not action.help:
837
+ tup = self._current_indent, '', action_header
838
+ action_header = '%*s%s\n' % tup
839
+
840
+ # short action name; start on the same line and pad two spaces
841
+ elif len(action_header) <= action_width:
842
+ tup = self._current_indent, '', action_width, action_header
843
+ action_header = '%*s%-*s ' % tup
844
+ indent_first = 0
845
+
846
+ # long action name; start on the next line
847
+ else:
848
+ tup = self._current_indent, '', action_header
849
+ action_header = '%*s%s\n' % tup
850
+ indent_first = help_position
851
+
852
+ # collect the pieces of the action help
853
+ parts = [action_header]
854
+
855
+ # if there was help for the action, add lines of help text
856
+ if action.help:
857
+ help_text = self._expand_help(action)
858
+ help_lines = self._split_lines(help_text, help_width)
859
+ parts.append('%*s%s\n' % (indent_first, '', help_lines[0]))
860
+ for line in help_lines[1:]:
861
+ parts.append('%*s%s\n' % (help_position, '', line))
862
+
863
+ # or add a newline if the description doesn't end with one
864
+ elif not action_header.endswith('\n'):
865
+ parts.append('\n')
866
+
867
+ # if there are any sub-actions, add their help as well
868
+ for subaction in self._iter_indented_subactions(action):
869
+ parts.append(self._format_action(subaction))
870
+
871
+ # return a single string
872
+ return self._join_parts(parts)
873
+
874
+ def _format_action_invocation(self, action):
875
+ if not action.option_strings:
876
+ metavar, = self._metavar_formatter(action, action.dest)(1)
877
+ return metavar
878
+
879
+ else:
880
+ parts = []
881
+
882
+ # if the Optional doesn't take a value, format is:
883
+ # -s, --long
884
+ if action.nargs == 0:
885
+ parts.extend(action.option_strings)
886
+
887
+ # if the Optional takes a value, format is:
888
+ # -s ARGS, --long ARGS
889
+ else:
890
+ default = action.dest.upper()
891
+ args_string = self._format_args(action, default)
892
+ for option_string in action.option_strings:
893
+ parts.append('%s %s' % (option_string, args_string))
894
+
895
+ return ', '.join(parts)
896
+
897
+ def _metavar_formatter(self, action, default_metavar):
898
+ if action.metavar is not None:
899
+ result = action.metavar
900
+ elif action.choices is not None:
901
+ choice_strs = [str(choice) for choice in action.choices]
902
+ result = '{%s}' % ','.join(choice_strs)
903
+ else:
904
+ result = default_metavar
905
+
906
+ def format(tuple_size):
907
+ if isinstance(result, tuple):
908
+ return result
909
+ else:
910
+ return (result, ) * tuple_size
911
+ return format
912
+
913
+ def _format_args(self, action, default_metavar):
914
+ get_metavar = self._metavar_formatter(action, default_metavar)
915
+ if action.nargs is None:
916
+ result = '%s' % get_metavar(1)
917
+ elif action.nargs == OPTIONAL:
918
+ result = '[%s]' % get_metavar(1)
919
+ elif action.nargs == ZERO_OR_MORE:
920
+ result = '[%s [%s ...]]' % get_metavar(2)
921
+ elif action.nargs == ONE_OR_MORE:
922
+ result = '%s [%s ...]' % get_metavar(2)
923
+ elif action.nargs == REMAINDER:
924
+ result = '...'
925
+ elif action.nargs == PARSER:
926
+ result = '%s ...' % get_metavar(1)
927
+ else:
928
+ formats = ['%s' for _ in range(action.nargs)]
929
+ result = ' '.join(formats) % get_metavar(action.nargs)
930
+ return result
931
+
932
+ def _expand_help(self, action):
933
+ params = dict(vars(action), prog=self._prog)
934
+ for name in list(params):
935
+ if params[name] is SUPPRESS:
936
+ del params[name]
937
+ for name in list(params):
938
+ if hasattr(params[name], '__name__'):
939
+ params[name] = params[name].__name__
940
+ if params.get('choices') is not None:
941
+ choices_str = ', '.join([str(c) for c in params['choices']])
942
+ params['choices'] = choices_str
943
+ return self._get_help_string(action) % params
944
+
945
+ def _iter_indented_subactions(self, action):
946
+ try:
947
+ get_subactions = action._get_subactions
948
+ except AttributeError:
949
+ pass
950
+ else:
951
+ self._indent()
952
+ for subaction in get_subactions():
953
+ yield subaction
954
+ self._dedent()
955
+
956
+ def _split_lines(self, text, width):
957
+ text = self._whitespace_matcher.sub(' ', text).strip()
958
+ return _textwrap.wrap(text, width)
959
+
960
+ def _fill_text(self, text, width, indent):
961
+ text = self._whitespace_matcher.sub(' ', text).strip()
962
+ return _textwrap.fill(text, width, initial_indent=indent,
963
+ subsequent_indent=indent)
964
+
965
+ def _get_help_string(self, action):
966
+ return action.help
967
+
968
+
969
+class RawDescriptionHelpFormatter(HelpFormatter):
970
+ """Help message formatter which retains any formatting in descriptions.
971
+
972
+ Only the name of this class is considered a public API. All the methods
973
+ provided by the class are considered an implementation detail.
974
+ """
975
+
976
+ def _fill_text(self, text, width, indent):
977
+ return ''.join([indent + line for line in text.splitlines(True)])
978
+
979
+
980
+class RawTextHelpFormatter(RawDescriptionHelpFormatter):
981
+ """Help message formatter which retains formatting of all help text.
982
+
983
+ Only the name of this class is considered a public API. All the methods
984
+ provided by the class are considered an implementation detail.
985
+ """
986
+
987
+ def _split_lines(self, text, width):
988
+ return text.splitlines()
989
+
990
+
991
+class ArgumentDefaultsHelpFormatter(HelpFormatter):
992
+ """Help message formatter which adds default values to argument help.
993
+
994
+ Only the name of this class is considered a public API. All the methods
995
+ provided by the class are considered an implementation detail.
996
+ """
997
+
998
+ def _get_help_string(self, action):
999
+ help = action.help
1000
+ if '%(default)' not in action.help:
1001
+ if action.default is not SUPPRESS:
1002
+ defaulting_nargs = [OPTIONAL, ZERO_OR_MORE]
1003
+ if action.option_strings or action.nargs in defaulting_nargs:
1004
+ help += ' (default: %(default)s)'
1005
+ return help
1006
+
1007
+
1008
+# =====================
1009
+# Options and Arguments
1010
+# =====================
1011
+
1012
+def _get_action_name(argument):
1013
+ if argument is None:
1014
+ return None
1015
+ elif argument.option_strings:
1016
+ return '/'.join(argument.option_strings)
1017
+ elif argument.metavar not in (None, SUPPRESS):
1018
+ return argument.metavar
1019
+ elif argument.dest not in (None, SUPPRESS):
1020
+ return argument.dest
1021
+ else:
1022
+ return None
1023
+
1024
+
1025
+class ArgumentError(Exception):
1026
+ """An error from creating or using an argument (optional or positional).
1027
+
1028
+ The string value of this exception is the message, augmented with
1029
+ information about the argument that caused it.
1030
+ """
1031
+
1032
+ def __init__(self, argument, message):
1033
+ self.argument_name = _get_action_name(argument)
1034
+ self.message = message
1035
+
1036
+ def __str__(self):
1037
+ if self.argument_name is None:
1038
+ format = '%(message)s'
1039
+ else:
1040
+ format = 'argument %(argument_name)s: %(message)s'
1041
+ return format % dict(message=self.message,
1042
+ argument_name=self.argument_name)
1043
+
1044
+
1045
+class ArgumentTypeError(Exception):
1046
+ """An error from trying to convert a command line string to a type."""
1047
+ pass
1048
+
1049
+
1050
+# ==============
1051
+# Action classes
1052
+# ==============
1053
+
1054
+class Action(_AttributeHolder):
1055
+ """Information about how to convert command line strings to Python objects.
1056
+
1057
+ Action objects are used by an ArgumentParser to represent the information
1058
+ needed to parse a single argument from one or more strings from the
1059
+ command line. The keyword arguments to the Action constructor are also
1060
+ all attributes of Action instances.
1061
+
1062
+ Keyword Arguments:
1063
+
1064
+ - option_strings -- A list of command-line option strings which
1065
+ should be associated with this action.
1066
+
1067
+ - dest -- The name of the attribute to hold the created object(s)
1068
+
1069
+ - nargs -- The number of command-line arguments that should be
1070
+ consumed. By default, one argument will be consumed and a single
1071
+ value will be produced. Other values include:
1072
+ - N (an integer) consumes N arguments (and produces a list)
1073
+ - '?' consumes zero or one arguments
1074
+ - '*' consumes zero or more arguments (and produces a list)
1075
+ - '+' consumes one or more arguments (and produces a list)
1076
+ Note that the difference between the default and nargs=1 is that
1077
+ with the default, a single value will be produced, while with
1078
+ nargs=1, a list containing a single value will be produced.
1079
+
1080
+ - const -- The value to be produced if the option is specified and the
1081
+ option uses an action that takes no values.
1082
+
1083
+ - default -- The value to be produced if the option is not specified.
1084
+
1085
+ - type -- The type which the command-line arguments should be converted
1086
+ to, should be one of 'string', 'int', 'float', 'complex' or a
1087
+ callable object that accepts a single string argument. If None,
1088
+ 'string' is assumed.
1089
+
1090
+ - choices -- A container of values that should be allowed. If not None,
1091
+ after a command-line argument has been converted to the appropriate
1092
+ type, an exception will be raised if it is not a member of this
1093
+ collection.
1094
+
1095
+ - required -- True if the action must always be specified at the
1096
+ command line. This is only meaningful for optional command-line
1097
+ arguments.
1098
+
1099
+ - help -- The help string describing the argument.
1100
+
1101
+ - metavar -- The name to be used for the option's argument with the
1102
+ help string. If None, the 'dest' value will be used as the name.
1103
+ """
1104
+
1105
+ def __init__(self,
1106
+ option_strings,
1107
+ dest,
1108
+ nargs=None,
1109
+ const=None,
1110
+ default=None,
1111
+ type=None,
1112
+ choices=None,
1113
+ required=False,
1114
+ help=None,
1115
+ metavar=None):
1116
+ self.option_strings = option_strings
1117
+ self.dest = dest
1118
+ self.nargs = nargs
1119
+ self.const = const
1120
+ self.default = default
1121
+ self.type = type
1122
+ self.choices = choices
1123
+ self.required = required
1124
+ self.help = help
1125
+ self.metavar = metavar
1126
+
1127
+ def _get_kwargs(self):
1128
+ names = [
1129
+ 'option_strings',
1130
+ 'dest',
1131
+ 'nargs',
1132
+ 'const',
1133
+ 'default',
1134
+ 'type',
1135
+ 'choices',
1136
+ 'help',
1137
+ 'metavar',
1138
+ ]
1139
+ return [(name, getattr(self, name)) for name in names]
1140
+
1141
+ def __call__(self, parser, namespace, values, option_string=None):
1142
+ raise NotImplementedError(_('.__call__() not defined'))
1143
+
1144
+
1145
+class _StoreAction(Action):
1146
+
1147
+ def __init__(self,
1148
+ option_strings,
1149
+ dest,
1150
+ nargs=None,
1151
+ const=None,
1152
+ default=None,
1153
+ type=None,
1154
+ choices=None,
1155
+ required=False,
1156
+ help=None,
1157
+ metavar=None):
1158
+ if nargs == 0:
1159
+ raise ValueError('nargs for store actions must be > 0; if you '
1160
+ 'have nothing to store, actions such as store '
1161
+ 'true or store const may be more appropriate')
1162
+ if const is not None and nargs != OPTIONAL:
1163
+ raise ValueError('nargs must be %r to supply const' % OPTIONAL)
1164
+ super(_StoreAction, self).__init__(
1165
+ option_strings=option_strings,
1166
+ dest=dest,
1167
+ nargs=nargs,
1168
+ const=const,
1169
+ default=default,
1170
+ type=type,
1171
+ choices=choices,
1172
+ required=required,
1173
+ help=help,
1174
+ metavar=metavar)
1175
+
1176
+ def __call__(self, parser, namespace, values, option_string=None):
1177
+ setattr(namespace, self.dest, values)
1178
+
1179
+
1180
+class _StoreConstAction(Action):
1181
+
1182
+ def __init__(self,
1183
+ option_strings,
1184
+ dest,
1185
+ const,
1186
+ default=None,
1187
+ required=False,
1188
+ help=None,
1189
+ metavar=None):
1190
+ super(_StoreConstAction, self).__init__(
1191
+ option_strings=option_strings,
1192
+ dest=dest,
1193
+ nargs=0,
1194
+ const=const,
1195
+ default=default,
1196
+ required=required,
1197
+ help=help)
1198
+
1199
+ def __call__(self, parser, namespace, values, option_string=None):
1200
+ setattr(namespace, self.dest, self.const)
1201
+
1202
+
1203
+class _StoreTrueAction(_StoreConstAction):
1204
+
1205
+ def __init__(self,
1206
+ option_strings,
1207
+ dest,
1208
+ default=False,
1209
+ required=False,
1210
+ help=None):
1211
+ super(_StoreTrueAction, self).__init__(
1212
+ option_strings=option_strings,
1213
+ dest=dest,
1214
+ const=True,
1215
+ default=default,
1216
+ required=required,
1217
+ help=help)
1218
+
1219
+
1220
+class _StoreFalseAction(_StoreConstAction):
1221
+
1222
+ def __init__(self,
1223
+ option_strings,
1224
+ dest,
1225
+ default=True,
1226
+ required=False,
1227
+ help=None):
1228
+ super(_StoreFalseAction, self).__init__(
1229
+ option_strings=option_strings,
1230
+ dest=dest,
1231
+ const=False,
1232
+ default=default,
1233
+ required=required,
1234
+ help=help)
1235
+
1236
+
1237
+class _AppendAction(Action):
1238
+
1239
+ def __init__(self,
1240
+ option_strings,
1241
+ dest,
1242
+ nargs=None,
1243
+ const=None,
1244
+ default=None,
1245
+ type=None,
1246
+ choices=None,
1247
+ required=False,
1248
+ help=None,
1249
+ metavar=None):
1250
+ if nargs == 0:
1251
+ raise ValueError('nargs for append actions must be > 0; if arg '
1252
+ 'strings are not supplying the value to append, '
1253
+ 'the append const action may be more appropriate')
1254
+ if const is not None and nargs != OPTIONAL:
1255
+ raise ValueError('nargs must be %r to supply const' % OPTIONAL)
1256
+ super(_AppendAction, self).__init__(
1257
+ option_strings=option_strings,
1258
+ dest=dest,
1259
+ nargs=nargs,
1260
+ const=const,
1261
+ default=default,
1262
+ type=type,
1263
+ choices=choices,
1264
+ required=required,
1265
+ help=help,
1266
+ metavar=metavar)
1267
+
1268
+ def __call__(self, parser, namespace, values, option_string=None):
1269
+ items = _copy.copy(_ensure_value(namespace, self.dest, []))
1270
+ items.append(values)
1271
+ setattr(namespace, self.dest, items)
1272
+
1273
+
1274
+class _AppendConstAction(Action):
1275
+
1276
+ def __init__(self,
1277
+ option_strings,
1278
+ dest,
1279
+ const,
1280
+ default=None,
1281
+ required=False,
1282
+ help=None,
1283
+ metavar=None):
1284
+ super(_AppendConstAction, self).__init__(
1285
+ option_strings=option_strings,
1286
+ dest=dest,
1287
+ nargs=0,
1288
+ const=const,
1289
+ default=default,
1290
+ required=required,
1291
+ help=help,
1292
+ metavar=metavar)
1293
+
1294
+ def __call__(self, parser, namespace, values, option_string=None):
1295
+ items = _copy.copy(_ensure_value(namespace, self.dest, []))
1296
+ items.append(self.const)
1297
+ setattr(namespace, self.dest, items)
1298
+
1299
+
1300
+class _CountAction(Action):
1301
+
1302
+ def __init__(self,
1303
+ option_strings,
1304
+ dest,
1305
+ default=None,
1306
+ required=False,
1307
+ help=None):
1308
+ super(_CountAction, self).__init__(
1309
+ option_strings=option_strings,
1310
+ dest=dest,
1311
+ nargs=0,
1312
+ default=default,
1313
+ required=required,
1314
+ help=help)
1315
+
1316
+ def __call__(self, parser, namespace, values, option_string=None):
1317
+ new_count = _ensure_value(namespace, self.dest, 0) + 1
1318
+ setattr(namespace, self.dest, new_count)
1319
+
1320
+
1321
+class _HelpAction(Action):
1322
+
1323
+ def __init__(self,
1324
+ option_strings,
1325
+ dest=SUPPRESS,
1326
+ default=SUPPRESS,
1327
+ help=None):
1328
+ super(_HelpAction, self).__init__(
1329
+ option_strings=option_strings,
1330
+ dest=dest,
1331
+ default=default,
1332
+ nargs=0,
1333
+ help=help)
1334
+
1335
+ def __call__(self, parser, namespace, values, option_string=None):
1336
+ parser.print_help()
1337
+ parser.exit()
1338
+
1339
+
1340
+class _VersionAction(Action):
1341
+
1342
+ def __init__(self,
1343
+ option_strings,
1344
+ version=None,
1345
+ dest=SUPPRESS,
1346
+ default=SUPPRESS,
1347
+ help="show program's version number and exit"):
1348
+ super(_VersionAction, self).__init__(
1349
+ option_strings=option_strings,
1350
+ dest=dest,
1351
+ default=default,
1352
+ nargs=0,
1353
+ help=help)
1354
+ self.version = version
1355
+
1356
+ def __call__(self, parser, namespace, values, option_string=None):
1357
+ version = self.version
1358
+ if version is None:
1359
+ version = parser.version
1360
+ formatter = parser._get_formatter()
1361
+ formatter.add_text(version)
1362
+ parser.exit(message=formatter.format_help())
1363
+
1364
+
1365
+class _SubParsersAction(Action):
1366
+
1367
+ class _ChoicesPseudoAction(Action):
1368
+
1369
+ def __init__(self, name, aliases, help):
1370
+ metavar = dest = name
1371
+ if aliases:
1372
+ metavar += ' (%s)' % ', '.join(aliases)
1373
+ sup = super(_SubParsersAction._ChoicesPseudoAction, self)
1374
+ sup.__init__(option_strings=[], dest=dest, help=help,
1375
+ metavar=metavar)
1376
+
1377
+ def __init__(self,
1378
+ option_strings,
1379
+ prog,
1380
+ parser_class,
1381
+ dest=SUPPRESS,
1382
+ help=None,
1383
+ metavar=None):
1384
+
1385
+ self._prog_prefix = prog
1386
+ self._parser_class = parser_class
1387
+ self._name_parser_map = {}
1388
+ self._choices_actions = []
1389
+
1390
+ super(_SubParsersAction, self).__init__(
1391
+ option_strings=option_strings,
1392
+ dest=dest,
1393
+ nargs=PARSER,
1394
+ choices=self._name_parser_map,
1395
+ help=help,
1396
+ metavar=metavar)
1397
+
1398
+ def add_parser(self, name, **kwargs):
1399
+ # set prog from the existing prefix
1400
+ if kwargs.get('prog') is None:
1401
+ kwargs['prog'] = '%s %s' % (self._prog_prefix, name)
1402
+
1403
+ aliases = kwargs.pop('aliases', ())
1404
+
1405
+ # create a pseudo-action to hold the choice help
1406
+ if 'help' in kwargs:
1407
+ help = kwargs.pop('help')
1408
+ choice_action = self._ChoicesPseudoAction(name, aliases, help)
1409
+ self._choices_actions.append(choice_action)
1410
+
1411
+ # create the parser and add it to the map
1412
+ parser = self._parser_class(**kwargs)
1413
+ self._name_parser_map[name] = parser
1414
+
1415
+ # make parser available under aliases also
1416
+ for alias in aliases:
1417
+ self._name_parser_map[alias] = parser
1418
+
1419
+ return parser
1420
+
1421
+ def _get_subactions(self):
1422
+ return self._choices_actions
1423
+
1424
+ def __call__(self, parser, namespace, values, option_string=None):
1425
+ parser_name = values[0]
1426
+ arg_strings = values[1:]
1427
+
1428
+ # set the parser name if requested
1429
+ if self.dest is not SUPPRESS:
1430
+ setattr(namespace, self.dest, parser_name)
1431
+
1432
+ # select the parser
1433
+ try:
1434
+ parser = self._name_parser_map[parser_name]
1435
+ except KeyError:
1436
+ tup = parser_name, ', '.join(self._name_parser_map)
1437
+ msg = _('unknown parser %r (choices: %s)' % tup)
1438
+ raise ArgumentError(self, msg)
1439
+
1440
+ # parse all the remaining options into the namespace
1441
+ # store any unrecognized options on the object, so that the top
1442
+ # level parser can decide what to do with them
1443
+ namespace, arg_strings = parser.parse_known_args(arg_strings, namespace)
1444
+ if arg_strings:
1445
+ vars(namespace).setdefault(_UNRECOGNIZED_ARGS_ATTR, [])
1446
+ getattr(namespace, _UNRECOGNIZED_ARGS_ATTR).extend(arg_strings)
1447
+
1448
+
1449
+# ==============
1450
+# Type classes
1451
+# ==============
1452
+
1453
+class FileType(object):
1454
+ """Factory for creating file object types
1455
+
1456
+ Instances of FileType are typically passed as type= arguments to the
1457
+ ArgumentParser add_argument() method.
1458
+
1459
+ Keyword Arguments:
1460
+ - mode -- A string indicating how the file is to be opened. Accepts the
1461
+ same values as the builtin open() function.
1462
+ - bufsize -- The file's desired buffer size. Accepts the same values as
1463
+ the builtin open() function.
1464
+ """
1465
+
1466
+ def __init__(self, mode='r', bufsize=None):
1467
+ self._mode = mode
1468
+ self._bufsize = bufsize
1469
+
1470
+ def __call__(self, string):
1471
+ # the special argument "-" means sys.std{in,out}
1472
+ if string == '-':
1473
+ if 'r' in self._mode:
1474
+ return _sys.stdin
1475
+ elif 'w' in self._mode:
1476
+ return _sys.stdout
1477
+ else:
1478
+ msg = _('argument "-" with mode %r' % self._mode)
1479
+ raise ValueError(msg)
1480
+
1481
+ try:
1482
+ # all other arguments are used as file names
1483
+ if self._bufsize:
1484
+ return open(string, self._mode, self._bufsize)
1485
+ else:
1486
+ return open(string, self._mode)
1487
+ except IOError:
1488
+ err = _sys.exc_info()[1]
1489
+ message = _("can't open '%s': %s")
1490
+ raise ArgumentTypeError(message % (string, err))
1491
+
1492
+ def __repr__(self):
1493
+ args = [self._mode, self._bufsize]
1494
+ args_str = ', '.join([repr(arg) for arg in args if arg is not None])
1495
+ return '%s(%s)' % (type(self).__name__, args_str)
1496
+
1497
+# ===========================
1498
+# Optional and Positional Parsing
1499
+# ===========================
1500
+
1501
+class Namespace(_AttributeHolder):
1502
+ """Simple object for storing attributes.
1503
+
1504
+ Implements equality by attribute names and values, and provides a simple
1505
+ string representation.
1506
+ """
1507
+
1508
+ def __init__(self, **kwargs):
1509
+ for name in kwargs:
1510
+ setattr(self, name, kwargs[name])
1511
+
1512
+ __hash__ = None
1513
+
1514
+ def __eq__(self, other):
1515
+ return vars(self) == vars(other)
1516
+
1517
+ def __ne__(self, other):
1518
+ return not (self == other)
1519
+
1520
+ def __contains__(self, key):
1521
+ return key in self.__dict__
1522
+
1523
+
1524
+class _ActionsContainer(object):
1525
+
1526
+ def __init__(self,
1527
+ description,
1528
+ prefix_chars,
1529
+ argument_default,
1530
+ conflict_handler):
1531
+ super(_ActionsContainer, self).__init__()
1532
+
1533
+ self.description = description
1534
+ self.argument_default = argument_default
1535
+ self.prefix_chars = prefix_chars
1536
+ self.conflict_handler = conflict_handler
1537
+
1538
+ # set up registries
1539
+ self._registries = {}
1540
+
1541
+ # register actions
1542
+ self.register('action', None, _StoreAction)
1543
+ self.register('action', 'store', _StoreAction)
1544
+ self.register('action', 'store_const', _StoreConstAction)
1545
+ self.register('action', 'store_true', _StoreTrueAction)
1546
+ self.register('action', 'store_false', _StoreFalseAction)
1547
+ self.register('action', 'append', _AppendAction)
1548
+ self.register('action', 'append_const', _AppendConstAction)
1549
+ self.register('action', 'count', _CountAction)
1550
+ self.register('action', 'help', _HelpAction)
1551
+ self.register('action', 'version', _VersionAction)
1552
+ self.register('action', 'parsers', _SubParsersAction)
1553
+
1554
+ # raise an exception if the conflict handler is invalid
1555
+ self._get_handler()
1556
+
1557
+ # action storage
1558
+ self._actions = []
1559
+ self._option_string_actions = {}
1560
+
1561
+ # groups
1562
+ self._action_groups = []
1563
+ self._mutually_exclusive_groups = []
1564
+
1565
+ # defaults storage
1566
+ self._defaults = {}
1567
+
1568
+ # determines whether an "option" looks like a negative number
1569
+ self._negative_number_matcher = _re.compile(r'^-\d+$|^-\d*\.\d+$')
1570
+
1571
+ # whether or not there are any optionals that look like negative
1572
+ # numbers -- uses a list so it can be shared and edited
1573
+ self._has_negative_number_optionals = []
1574
+
1575
+ # ====================
1576
+ # Registration methods
1577
+ # ====================
1578
+ def register(self, registry_name, value, object):
1579
+ registry = self._registries.setdefault(registry_name, {})
1580
+ registry[value] = object
1581
+
1582
+ def _registry_get(self, registry_name, value, default=None):
1583
+ return self._registries[registry_name].get(value, default)
1584
+
1585
+ # ==================================
1586
+ # Namespace default accessor methods
1587
+ # ==================================
1588
+ def set_defaults(self, **kwargs):
1589
+ self._defaults.update(kwargs)
1590
+
1591
+ # if these defaults match any existing arguments, replace
1592
+ # the previous default on the object with the new one
1593
+ for action in self._actions:
1594
+ if action.dest in kwargs:
1595
+ action.default = kwargs[action.dest]
1596
+
1597
+ def get_default(self, dest):
1598
+ for action in self._actions:
1599
+ if action.dest == dest and action.default is not None:
1600
+ return action.default
1601
+ return self._defaults.get(dest, None)
1602
+
1603
+
1604
+ # =======================
1605
+ # Adding argument actions
1606
+ # =======================
1607
+ def add_argument(self, *args, **kwargs):
1608
+ """
1609
+ add_argument(dest, ..., name=value, ...)
1610
+ add_argument(option_string, option_string, ..., name=value, ...)
1611
+ """
1612
+
1613
+ # if no positional args are supplied or only one is supplied and
1614
+ # it doesn't look like an option string, parse a positional
1615
+ # argument
1616
+ chars = self.prefix_chars
1617
+ if not args or len(args) == 1 and args[0][0] not in chars:
1618
+ if args and 'dest' in kwargs:
1619
+ raise ValueError('dest supplied twice for positional argument')
1620
+ kwargs = self._get_positional_kwargs(*args, **kwargs)
1621
+
1622
+ # otherwise, we're adding an optional argument
1623
+ else:
1624
+ kwargs = self._get_optional_kwargs(*args, **kwargs)
1625
+
1626
+ # if no default was supplied, use the parser-level default
1627
+ if 'default' not in kwargs:
1628
+ dest = kwargs['dest']
1629
+ if dest in self._defaults:
1630
+ kwargs['default'] = self._defaults[dest]
1631
+ elif self.argument_default is not None:
1632
+ kwargs['default'] = self.argument_default
1633
+
1634
+ # create the action object, and add it to the parser
1635
+ action_class = self._pop_action_class(kwargs)
1636
+ if not _callable(action_class):
1637
+ raise ValueError('unknown action "%s"' % action_class)
1638
+ action = action_class(**kwargs)
1639
+
1640
+ # raise an error if the action type is not callable
1641
+ type_func = self._registry_get('type', action.type, action.type)
1642
+ if not _callable(type_func):
1643
+ raise ValueError('%r is not callable' % type_func)
1644
+
1645
+ return self._add_action(action)
1646
+
1647
+ def add_argument_group(self, *args, **kwargs):
1648
+ group = _ArgumentGroup(self, *args, **kwargs)
1649
+ self._action_groups.append(group)
1650
+ return group
1651
+
1652
+ def add_mutually_exclusive_group(self, **kwargs):
1653
+ group = _MutuallyExclusiveGroup(self, **kwargs)
1654
+ self._mutually_exclusive_groups.append(group)
1655
+ return group
1656
+
1657
+ def _add_action(self, action):
1658
+ # resolve any conflicts
1659
+ self._check_conflict(action)
1660
+
1661
+ # add to actions list
1662
+ self._actions.append(action)
1663
+ action.container = self
1664
+
1665
+ # index the action by any option strings it has
1666
+ for option_string in action.option_strings:
1667
+ self._option_string_actions[option_string] = action
1668
+
1669
+ # set the flag if any option strings look like negative numbers
1670
+ for option_string in action.option_strings:
1671
+ if self._negative_number_matcher.match(option_string):
1672
+ if not self._has_negative_number_optionals:
1673
+ self._has_negative_number_optionals.append(True)
1674
+
1675
+ # return the created action
1676
+ return action
1677
+
1678
+ def _remove_action(self, action):
1679
+ self._actions.remove(action)
1680
+
1681
+ def _add_container_actions(self, container):
1682
+ # collect groups by titles
1683
+ title_group_map = {}
1684
+ for group in self._action_groups:
1685
+ if group.title in title_group_map:
1686
+ msg = _('cannot merge actions - two groups are named %r')
1687
+ raise ValueError(msg % (group.title))
1688
+ title_group_map[group.title] = group
1689
+
1690
+ # map each action to its group
1691
+ group_map = {}
1692
+ for group in container._action_groups:
1693
+
1694
+ # if a group with the title exists, use that, otherwise
1695
+ # create a new group matching the container's group
1696
+ if group.title not in title_group_map:
1697
+ title_group_map[group.title] = self.add_argument_group(
1698
+ title=group.title,
1699
+ description=group.description,
1700
+ conflict_handler=group.conflict_handler)
1701
+
1702
+ # map the actions to their new group
1703
+ for action in group._group_actions:
1704
+ group_map[action] = title_group_map[group.title]
1705
+
1706
+ # add container's mutually exclusive groups
1707
+ # NOTE: if add_mutually_exclusive_group ever gains title= and
1708
+ # description= then this code will need to be expanded as above
1709
+ for group in container._mutually_exclusive_groups:
1710
+ mutex_group = self.add_mutually_exclusive_group(
1711
+ required=group.required)
1712
+
1713
+ # map the actions to their new mutex group
1714
+ for action in group._group_actions:
1715
+ group_map[action] = mutex_group
1716
+
1717
+ # add all actions to this container or their group
1718
+ for action in container._actions:
1719
+ group_map.get(action, self)._add_action(action)
1720
+
1721
+ def _get_positional_kwargs(self, dest, **kwargs):
1722
+ # make sure required is not specified
1723
+ if 'required' in kwargs:
1724
+ msg = _("'required' is an invalid argument for positionals")
1725
+ raise TypeError(msg)
1726
+
1727
+ # mark positional arguments as required if at least one is
1728
+ # always required
1729
+ if kwargs.get('nargs') not in [OPTIONAL, ZERO_OR_MORE]:
1730
+ kwargs['required'] = True
1731
+ if kwargs.get('nargs') == ZERO_OR_MORE and 'default' not in kwargs:
1732
+ kwargs['required'] = True
1733
+
1734
+ # return the keyword arguments with no option strings
1735
+ return dict(kwargs, dest=dest, option_strings=[])
1736
+
1737
+ def _get_optional_kwargs(self, *args, **kwargs):
1738
+ # determine short and long option strings
1739
+ option_strings = []
1740
+ long_option_strings = []
1741
+ for option_string in args:
1742
+ # error on strings that don't start with an appropriate prefix
1743
+ if not option_string[0] in self.prefix_chars:
1744
+ msg = _('invalid option string %r: '
1745
+ 'must start with a character %r')
1746
+ tup = option_string, self.prefix_chars
1747
+ raise ValueError(msg % tup)
1748
+
1749
+ # strings starting with two prefix characters are long options
1750
+ option_strings.append(option_string)
1751
+ if option_string[0] in self.prefix_chars:
1752
+ if len(option_string) > 1:
1753
+ if option_string[1] in self.prefix_chars:
1754
+ long_option_strings.append(option_string)
1755
+
1756
+ # infer destination, '--foo-bar' -> 'foo_bar' and '-x' -> 'x'
1757
+ dest = kwargs.pop('dest', None)
1758
+ if dest is None:
1759
+ if long_option_strings:
1760
+ dest_option_string = long_option_strings[0]
1761
+ else:
1762
+ dest_option_string = option_strings[0]
1763
+ dest = dest_option_string.lstrip(self.prefix_chars)
1764
+ if not dest:
1765
+ msg = _('dest= is required for options like %r')
1766
+ raise ValueError(msg % option_string)
1767
+ dest = dest.replace('-', '_')
1768
+
1769
+ # return the updated keyword arguments
1770
+ return dict(kwargs, dest=dest, option_strings=option_strings)
1771
+
1772
+ def _pop_action_class(self, kwargs, default=None):
1773
+ action = kwargs.pop('action', default)
1774
+ return self._registry_get('action', action, action)
1775
+
1776
+ def _get_handler(self):
1777
+ # determine function from conflict handler string
1778
+ handler_func_name = '_handle_conflict_%s' % self.conflict_handler
1779
+ try:
1780
+ return getattr(self, handler_func_name)
1781
+ except AttributeError:
1782
+ msg = _('invalid conflict_resolution value: %r')
1783
+ raise ValueError(msg % self.conflict_handler)
1784
+
1785
+ def _check_conflict(self, action):
1786
+
1787
+ # find all options that conflict with this option
1788
+ confl_optionals = []
1789
+ for option_string in action.option_strings:
1790
+ if option_string in self._option_string_actions:
1791
+ confl_optional = self._option_string_actions[option_string]
1792
+ confl_optionals.append((option_string, confl_optional))
1793
+
1794
+ # resolve any conflicts
1795
+ if confl_optionals:
1796
+ conflict_handler = self._get_handler()
1797
+ conflict_handler(action, confl_optionals)
1798
+
1799
+ def _handle_conflict_error(self, action, conflicting_actions):
1800
+ message = _('conflicting option string(s): %s')
1801
+ conflict_string = ', '.join([option_string
1802
+ for option_string, action
1803
+ in conflicting_actions])
1804
+ raise ArgumentError(action, message % conflict_string)
1805
+
1806
+ def _handle_conflict_resolve(self, action, conflicting_actions):
1807
+
1808
+ # remove all conflicting options
1809
+ for option_string, action in conflicting_actions:
1810
+
1811
+ # remove the conflicting option
1812
+ action.option_strings.remove(option_string)
1813
+ self._option_string_actions.pop(option_string, None)
1814
+
1815
+ # if the option now has no option string, remove it from the
1816
+ # container holding it
1817
+ if not action.option_strings:
1818
+ action.container._remove_action(action)
1819
+
1820
+
1821
+class _ArgumentGroup(_ActionsContainer):
1822
+
1823
+ def __init__(self, container, title=None, description=None, **kwargs):
1824
+ # add any missing keyword arguments by checking the container
1825
+ update = kwargs.setdefault
1826
+ update('conflict_handler', container.conflict_handler)
1827
+ update('prefix_chars', container.prefix_chars)
1828
+ update('argument_default', container.argument_default)
1829
+ super_init = super(_ArgumentGroup, self).__init__
1830
+ super_init(description=description, **kwargs)
1831
+
1832
+ # group attributes
1833
+ self.title = title
1834
+ self._group_actions = []
1835
+
1836
+ # share most attributes with the container
1837
+ self._registries = container._registries
1838
+ self._actions = container._actions
1839
+ self._option_string_actions = container._option_string_actions
1840
+ self._defaults = container._defaults
1841
+ self._has_negative_number_optionals = \
1842
+ container._has_negative_number_optionals
1843
+
1844
+ def _add_action(self, action):
1845
+ action = super(_ArgumentGroup, self)._add_action(action)
1846
+ self._group_actions.append(action)
1847
+ return action
1848
+
1849
+ def _remove_action(self, action):
1850
+ super(_ArgumentGroup, self)._remove_action(action)
1851
+ self._group_actions.remove(action)
1852
+
1853
+
1854
+class _MutuallyExclusiveGroup(_ArgumentGroup):
1855
+
1856
+ def __init__(self, container, required=False):
1857
+ super(_MutuallyExclusiveGroup, self).__init__(container)
1858
+ self.required = required
1859
+ self._container = container
1860
+
1861
+ def _add_action(self, action):
1862
+ if action.required:
1863
+ msg = _('mutually exclusive arguments must be optional')
1864
+ raise ValueError(msg)
1865
+ action = self._container._add_action(action)
1866
+ self._group_actions.append(action)
1867
+ return action
1868
+
1869
+ def _remove_action(self, action):
1870
+ self._container._remove_action(action)
1871
+ self._group_actions.remove(action)
1872
+
1873
+
1874
+class ArgumentParser(_AttributeHolder, _ActionsContainer):
1875
+ """Object for parsing command line strings into Python objects.
1876
+
1877
+ Keyword Arguments:
1878
+ - prog -- The name of the program (default: sys.argv[0])
1879
+ - usage -- A usage message (default: auto-generated from arguments)
1880
+ - description -- A description of what the program does
1881
+ - epilog -- Text following the argument descriptions
1882
+ - parents -- Parsers whose arguments should be copied into this one
1883
+ - formatter_class -- HelpFormatter class for printing help messages
1884
+ - prefix_chars -- Characters that prefix optional arguments
1885
+ - fromfile_prefix_chars -- Characters that prefix files containing
1886
+ additional arguments
1887
+ - argument_default -- The default value for all arguments
1888
+ - conflict_handler -- String indicating how to handle conflicts
1889
+ - add_help -- Add a -h/-help option
1890
+ """
1891
+
1892
+ def __init__(self,
1893
+ prog=None,
1894
+ usage=None,
1895
+ description=None,
1896
+ epilog=None,
1897
+ version=None,
1898
+ parents=[],
1899
+ formatter_class=HelpFormatter,
1900
+ prefix_chars='-',
1901
+ fromfile_prefix_chars=None,
1902
+ argument_default=None,
1903
+ conflict_handler='error',
1904
+ add_help=True):
1905
+
1906
+ if version is not None:
1907
+ import warnings
1908
+ warnings.warn(
1909
+ """The "version" argument to ArgumentParser is deprecated. """
1910
+ """Please use """
1911
+ """"add_argument(..., action='version', version="N", ...)" """
1912
+ """instead""", DeprecationWarning)
1913
+
1914
+ superinit = super(ArgumentParser, self).__init__
1915
+ superinit(description=description,
1916
+ prefix_chars=prefix_chars,
1917
+ argument_default=argument_default,
1918
+ conflict_handler=conflict_handler)
1919
+
1920
+ # default setting for prog
1921
+ if prog is None:
1922
+ prog = _os.path.basename(_sys.argv[0])
1923
+
1924
+ self.prog = prog
1925
+ self.usage = usage
1926
+ self.epilog = epilog
1927
+ self.version = version
1928
+ self.formatter_class = formatter_class
1929
+ self.fromfile_prefix_chars = fromfile_prefix_chars
1930
+ self.add_help = add_help
1931
+
1932
+ add_group = self.add_argument_group
1933
+ self._positionals = add_group(_('positional arguments'))
1934
+ self._optionals = add_group(_('optional arguments'))
1935
+ self._subparsers = None
1936
+
1937
+ # register types
1938
+ def identity(string):
1939
+ return string
1940
+ self.register('type', None, identity)
1941
+
1942
+ # add help and version arguments if necessary
1943
+ # (using explicit default to override global argument_default)
1944
+ if '-' in prefix_chars:
1945
+ default_prefix = '-'
1946
+ else:
1947
+ default_prefix = prefix_chars[0]
1948
+ if self.add_help:
1949
+ self.add_argument(
1950
+ default_prefix+'h', default_prefix*2+'help',
1951
+ action='help', default=SUPPRESS,
1952
+ help=_('show this help message and exit'))
1953
+ if self.version:
1954
+ self.add_argument(
1955
+ default_prefix+'v', default_prefix*2+'version',
1956
+ action='version', default=SUPPRESS,
1957
+ version=self.version,
1958
+ help=_("show program's version number and exit"))
1959
+
1960
+ # add parent arguments and defaults
1961
+ for parent in parents:
1962
+ self._add_container_actions(parent)
1963
+ try:
1964
+ defaults = parent._defaults
1965
+ except AttributeError:
1966
+ pass
1967
+ else:
1968
+ self._defaults.update(defaults)
1969
+
1970
+ # =======================
1971
+ # Pretty __repr__ methods
1972
+ # =======================
1973
+ def _get_kwargs(self):
1974
+ names = [
1975
+ 'prog',
1976
+ 'usage',
1977
+ 'description',
1978
+ 'version',
1979
+ 'formatter_class',
1980
+ 'conflict_handler',
1981
+ 'add_help',
1982
+ ]
1983
+ return [(name, getattr(self, name)) for name in names]
1984
+
1985
+ # ==================================
1986
+ # Optional/Positional adding methods
1987
+ # ==================================
1988
+ def add_subparsers(self, **kwargs):
1989
+ if self._subparsers is not None:
1990
+ self.error(_('cannot have multiple subparser arguments'))
1991
+
1992
+ # add the parser class to the arguments if it's not present
1993
+ kwargs.setdefault('parser_class', type(self))
1994
+
1995
+ if 'title' in kwargs or 'description' in kwargs:
1996
+ title = _(kwargs.pop('title', 'subcommands'))
1997
+ description = _(kwargs.pop('description', None))
1998
+ self._subparsers = self.add_argument_group(title, description)
1999
+ else:
2000
+ self._subparsers = self._positionals
2001
+
2002
+ # prog defaults to the usage message of this parser, skipping
2003
+ # optional arguments and with no "usage:" prefix
2004
+ if kwargs.get('prog') is None:
2005
+ formatter = self._get_formatter()
2006
+ positionals = self._get_positional_actions()
2007
+ groups = self._mutually_exclusive_groups
2008
+ formatter.add_usage(self.usage, positionals, groups, '')
2009
+ kwargs['prog'] = formatter.format_help().strip()
2010
+
2011
+ # create the parsers action and add it to the positionals list
2012
+ parsers_class = self._pop_action_class(kwargs, 'parsers')
2013
+ action = parsers_class(option_strings=[], **kwargs)
2014
+ self._subparsers._add_action(action)
2015
+
2016
+ # return the created parsers action
2017
+ return action
2018
+
2019
+ def _add_action(self, action):
2020
+ if action.option_strings:
2021
+ self._optionals._add_action(action)
2022
+ else:
2023
+ self._positionals._add_action(action)
2024
+ return action
2025
+
2026
+ def _get_optional_actions(self):
2027
+ return [action
2028
+ for action in self._actions
2029
+ if action.option_strings]
2030
+
2031
+ def _get_positional_actions(self):
2032
+ return [action
2033
+ for action in self._actions
2034
+ if not action.option_strings]
2035
+
2036
+ # =====================================
2037
+ # Command line argument parsing methods
2038
+ # =====================================
2039
+ def parse_args(self, args=None, namespace=None):
2040
+ args, argv = self.parse_known_args(args, namespace)
2041
+ if argv:
2042
+ msg = _('unrecognized arguments: %s')
2043
+ self.error(msg % ' '.join(argv))
2044
+ return args
2045
+
2046
+ def parse_known_args(self, args=None, namespace=None):
2047
+ # args default to the system args
2048
+ if args is None:
2049
+ args = _sys.argv[1:]
2050
+
2051
+ # default Namespace built from parser defaults
2052
+ if namespace is None:
2053
+ namespace = Namespace()
2054
+
2055
+ # add any action defaults that aren't present
2056
+ for action in self._actions:
2057
+ if action.dest is not SUPPRESS:
2058
+ if not hasattr(namespace, action.dest):
2059
+ if action.default is not SUPPRESS:
2060
+ setattr(namespace, action.dest, action.default)
2061
+
2062
+ # add any parser defaults that aren't present
2063
+ for dest in self._defaults:
2064
+ if not hasattr(namespace, dest):
2065
+ setattr(namespace, dest, self._defaults[dest])
2066
+
2067
+ # parse the arguments and exit if there are any errors
2068
+ try:
2069
+ namespace, args = self._parse_known_args(args, namespace)
2070
+ if hasattr(namespace, _UNRECOGNIZED_ARGS_ATTR):
2071
+ args.extend(getattr(namespace, _UNRECOGNIZED_ARGS_ATTR))
2072
+ delattr(namespace, _UNRECOGNIZED_ARGS_ATTR)
2073
+ return namespace, args
2074
+ except ArgumentError:
2075
+ err = _sys.exc_info()[1]
2076
+ self.error(str(err))
2077
+
2078
+ def _parse_known_args(self, arg_strings, namespace):
2079
+ # replace arg strings that are file references
2080
+ if self.fromfile_prefix_chars is not None:
2081
+ arg_strings = self._read_args_from_files(arg_strings)
2082
+
2083
+ # map all mutually exclusive arguments to the other arguments
2084
+ # they can't occur with
2085
+ action_conflicts = {}
2086
+ for mutex_group in self._mutually_exclusive_groups:
2087
+ group_actions = mutex_group._group_actions
2088
+ for i, mutex_action in enumerate(mutex_group._group_actions):
2089
+ conflicts = action_conflicts.setdefault(mutex_action, [])
2090
+ conflicts.extend(group_actions[:i])
2091
+ conflicts.extend(group_actions[i + 1:])
2092
+
2093
+ # find all option indices, and determine the arg_string_pattern
2094
+ # which has an 'O' if there is an option at an index,
2095
+ # an 'A' if there is an argument, or a '-' if there is a '--'
2096
+ option_string_indices = {}
2097
+ arg_string_pattern_parts = []
2098
+ arg_strings_iter = iter(arg_strings)
2099
+ for i, arg_string in enumerate(arg_strings_iter):
2100
+
2101
+ # all args after -- are non-options
2102
+ if arg_string == '--':
2103
+ arg_string_pattern_parts.append('-')
2104
+ for arg_string in arg_strings_iter:
2105
+ arg_string_pattern_parts.append('A')
2106
+
2107
+ # otherwise, add the arg to the arg strings
2108
+ # and note the index if it was an option
2109
+ else:
2110
+ option_tuple = self._parse_optional(arg_string)
2111
+ if option_tuple is None:
2112
+ pattern = 'A'
2113
+ else:
2114
+ option_string_indices[i] = option_tuple
2115
+ pattern = 'O'
2116
+ arg_string_pattern_parts.append(pattern)
2117
+
2118
+ # join the pieces together to form the pattern
2119
+ arg_strings_pattern = ''.join(arg_string_pattern_parts)
2120
+
2121
+ # converts arg strings to the appropriate and then takes the action
2122
+ seen_actions = set()
2123
+ seen_non_default_actions = set()
2124
+
2125
+ def take_action(action, argument_strings, option_string=None):
2126
+ seen_actions.add(action)
2127
+ argument_values = self._get_values(action, argument_strings)
2128
+
2129
+ # error if this argument is not allowed with other previously
2130
+ # seen arguments, assuming that actions that use the default
2131
+ # value don't really count as "present"
2132
+ if argument_values is not action.default:
2133
+ seen_non_default_actions.add(action)
2134
+ for conflict_action in action_conflicts.get(action, []):
2135
+ if conflict_action in seen_non_default_actions:
2136
+ msg = _('not allowed with argument %s')
2137
+ action_name = _get_action_name(conflict_action)
2138
+ raise ArgumentError(action, msg % action_name)
2139
+
2140
+ # take the action if we didn't receive a SUPPRESS value
2141
+ # (e.g. from a default)
2142
+ if argument_values is not SUPPRESS:
2143
+ action(self, namespace, argument_values, option_string)
2144
+
2145
+ # function to convert arg_strings into an optional action
2146
+ def consume_optional(start_index):
2147
+
2148
+ # get the optional identified at this index
2149
+ option_tuple = option_string_indices[start_index]
2150
+ action, option_string, explicit_arg = option_tuple
2151
+
2152
+ # identify additional optionals in the same arg string
2153
+ # (e.g. -xyz is the same as -x -y -z if no args are required)
2154
+ match_argument = self._match_argument
2155
+ action_tuples = []
2156
+ while True:
2157
+
2158
+ # if we found no optional action, skip it
2159
+ if action is None:
2160
+ extras.append(arg_strings[start_index])
2161
+ return start_index + 1
2162
+
2163
+ # if there is an explicit argument, try to match the
2164
+ # optional's string arguments to only this
2165
+ if explicit_arg is not None:
2166
+ arg_count = match_argument(action, 'A')
2167
+
2168
+ # if the action is a single-dash option and takes no
2169
+ # arguments, try to parse more single-dash options out
2170
+ # of the tail of the option string
2171
+ chars = self.prefix_chars
2172
+ if arg_count == 0 and option_string[1] not in chars:
2173
+ action_tuples.append((action, [], option_string))
2174
+ char = option_string[0]
2175
+ option_string = char + explicit_arg[0]
2176
+ new_explicit_arg = explicit_arg[1:] or None
2177
+ optionals_map = self._option_string_actions
2178
+ if option_string in optionals_map:
2179
+ action = optionals_map[option_string]
2180
+ explicit_arg = new_explicit_arg
2181
+ else:
2182
+ msg = _('ignored explicit argument %r')
2183
+ raise ArgumentError(action, msg % explicit_arg)
2184
+
2185
+ # if the action expect exactly one argument, we've
2186
+ # successfully matched the option; exit the loop
2187
+ elif arg_count == 1:
2188
+ stop = start_index + 1
2189
+ args = [explicit_arg]
2190
+ action_tuples.append((action, args, option_string))
2191
+ break
2192
+
2193
+ # error if a double-dash option did not use the
2194
+ # explicit argument
2195
+ else:
2196
+ msg = _('ignored explicit argument %r')
2197
+ raise ArgumentError(action, msg % explicit_arg)
2198
+
2199
+ # if there is no explicit argument, try to match the
2200
+ # optional's string arguments with the following strings
2201
+ # if successful, exit the loop
2202
+ else:
2203
+ start = start_index + 1
2204
+ selected_patterns = arg_strings_pattern[start:]
2205
+ arg_count = match_argument(action, selected_patterns)
2206
+ stop = start + arg_count
2207
+ args = arg_strings[start:stop]
2208
+ action_tuples.append((action, args, option_string))
2209
+ break
2210
+
2211
+ # add the Optional to the list and return the index at which
2212
+ # the Optional's string args stopped
2213
+ assert action_tuples
2214
+ for action, args, option_string in action_tuples:
2215
+ take_action(action, args, option_string)
2216
+ return stop
2217
+
2218
+ # the list of Positionals left to be parsed; this is modified
2219
+ # by consume_positionals()
2220
+ positionals = self._get_positional_actions()
2221
+
2222
+ # function to convert arg_strings into positional actions
2223
+ def consume_positionals(start_index):
2224
+ # match as many Positionals as possible
2225
+ match_partial = self._match_arguments_partial
2226
+ selected_pattern = arg_strings_pattern[start_index:]
2227
+ arg_counts = match_partial(positionals, selected_pattern)
2228
+
2229
+ # slice off the appropriate arg strings for each Positional
2230
+ # and add the Positional and its args to the list
2231
+ for action, arg_count in zip(positionals, arg_counts):
2232
+ args = arg_strings[start_index: start_index + arg_count]
2233
+ start_index += arg_count
2234
+ take_action(action, args)
2235
+
2236
+ # slice off the Positionals that we just parsed and return the
2237
+ # index at which the Positionals' string args stopped
2238
+ positionals[:] = positionals[len(arg_counts):]
2239
+ return start_index
2240
+
2241
+ # consume Positionals and Optionals alternately, until we have
2242
+ # passed the last option string
2243
+ extras = []
2244
+ start_index = 0
2245
+ if option_string_indices:
2246
+ max_option_string_index = max(option_string_indices)
2247
+ else:
2248
+ max_option_string_index = -1
2249
+ while start_index <= max_option_string_index:
2250
+
2251
+ # consume any Positionals preceding the next option
2252
+ next_option_string_index = min([
2253
+ index
2254
+ for index in option_string_indices
2255
+ if index >= start_index])
2256
+ if start_index != next_option_string_index:
2257
+ positionals_end_index = consume_positionals(start_index)
2258
+
2259
+ # only try to parse the next optional if we didn't consume
2260
+ # the option string during the positionals parsing
2261
+ if positionals_end_index > start_index:
2262
+ start_index = positionals_end_index
2263
+ continue
2264
+ else:
2265
+ start_index = positionals_end_index
2266
+
2267
+ # if we consumed all the positionals we could and we're not
2268
+ # at the index of an option string, there were extra arguments
2269
+ if start_index not in option_string_indices:
2270
+ strings = arg_strings[start_index:next_option_string_index]
2271
+ extras.extend(strings)
2272
+ start_index = next_option_string_index
2273
+
2274
+ # consume the next optional and any arguments for it
2275
+ start_index = consume_optional(start_index)
2276
+
2277
+ # consume any positionals following the last Optional
2278
+ stop_index = consume_positionals(start_index)
2279
+
2280
+ # if we didn't consume all the argument strings, there were extras
2281
+ extras.extend(arg_strings[stop_index:])
2282
+
2283
+ # if we didn't use all the Positional objects, there were too few
2284
+ # arg strings supplied.
2285
+ if positionals:
2286
+ self.error(_('too few arguments'))
2287
+
2288
+ # make sure all required actions were present, and convert defaults.
2289
+ for action in self._actions:
2290
+ if action not in seen_actions:
2291
+ if action.required:
2292
+ name = _get_action_name(action)
2293
+ self.error(_('argument %s is required') % name)
2294
+ else:
2295
+ # Convert action default now instead of doing it before
2296
+ # parsing arguments to avoid calling convert functions
2297
+ # twice (which may fail) if the argument was given, but
2298
+ # only if it was defined already in the namespace
2299
+ if (action.default is not None and
2300
+ isinstance(action.default, basestring) and
2301
+ hasattr(namespace, action.dest) and
2302
+ action.default is getattr(namespace, action.dest)):
2303
+ setattr(namespace, action.dest,
2304
+ self._get_value(action, action.default))
2305
+
2306
+ # make sure all required groups had one option present
2307
+ for group in self._mutually_exclusive_groups:
2308
+ if group.required:
2309
+ for action in group._group_actions:
2310
+ if action in seen_non_default_actions:
2311
+ break
2312
+
2313
+ # if no actions were used, report the error
2314
+ else:
2315
+ names = [_get_action_name(action)
2316
+ for action in group._group_actions
2317
+ if action.help is not SUPPRESS]
2318
+ msg = _('one of the arguments %s is required')
2319
+ self.error(msg % ' '.join(names))
2320
+
2321
+ # return the updated namespace and the extra arguments
2322
+ return namespace, extras
2323
+
2324
+ def _read_args_from_files(self, arg_strings):
2325
+ # expand arguments referencing files
2326
+ new_arg_strings = []
2327
+ for arg_string in arg_strings:
2328
+
2329
+ # for regular arguments, just add them back into the list
2330
+ if arg_string[0] not in self.fromfile_prefix_chars:
2331
+ new_arg_strings.append(arg_string)
2332
+
2333
+ # replace arguments referencing files with the file content
2334
+ else:
2335
+ try:
2336
+ args_file = open(arg_string[1:])
2337
+ try:
2338
+ arg_strings = []
2339
+ for arg_line in args_file.read().splitlines():
2340
+ for arg in self.convert_arg_line_to_args(arg_line):
2341
+ arg_strings.append(arg)
2342
+ arg_strings = self._read_args_from_files(arg_strings)
2343
+ new_arg_strings.extend(arg_strings)
2344
+ finally:
2345
+ args_file.close()
2346
+ except IOError:
2347
+ err = _sys.exc_info()[1]
2348
+ self.error(str(err))
2349
+
2350
+ # return the modified argument list
2351
+ return new_arg_strings
2352
+
2353
+ def convert_arg_line_to_args(self, arg_line):
2354
+ return [arg_line]
2355
+
2356
+ def _match_argument(self, action, arg_strings_pattern):
2357
+ # match the pattern for this action to the arg strings
2358
+ nargs_pattern = self._get_nargs_pattern(action)
2359
+ match = _re.match(nargs_pattern, arg_strings_pattern)
2360
+
2361
+ # raise an exception if we weren't able to find a match
2362
+ if match is None:
2363
+ nargs_errors = {
2364
+ None: _('expected one argument'),
2365
+ OPTIONAL: _('expected at most one argument'),
2366
+ ONE_OR_MORE: _('expected at least one argument'),
2367
+ }
2368
+ default = _('expected %s argument(s)') % action.nargs
2369
+ msg = nargs_errors.get(action.nargs, default)
2370
+ raise ArgumentError(action, msg)
2371
+
2372
+ # return the number of arguments matched
2373
+ return len(match.group(1))
2374
+
2375
+ def _match_arguments_partial(self, actions, arg_strings_pattern):
2376
+ # progressively shorten the actions list by slicing off the
2377
+ # final actions until we find a match
2378
+ result = []
2379
+ for i in range(len(actions), 0, -1):
2380
+ actions_slice = actions[:i]
2381
+ pattern = ''.join([self._get_nargs_pattern(action)
2382
+ for action in actions_slice])
2383
+ match = _re.match(pattern, arg_strings_pattern)
2384
+ if match is not None:
2385
+ result.extend([len(string) for string in match.groups()])
2386
+ break
2387
+
2388
+ # return the list of arg string counts
2389
+ return result
2390
+
2391
+ def _parse_optional(self, arg_string):
2392
+ # if it's an empty string, it was meant to be a positional
2393
+ if not arg_string:
2394
+ return None
2395
+
2396
+ # if it doesn't start with a prefix, it was meant to be positional
2397
+ if not arg_string[0] in self.prefix_chars:
2398
+ return None
2399
+
2400
+ # if the option string is present in the parser, return the action
2401
+ if arg_string in self._option_string_actions:
2402
+ action = self._option_string_actions[arg_string]
2403
+ return action, arg_string, None
2404
+
2405
+ # if it's just a single character, it was meant to be positional
2406
+ if len(arg_string) == 1:
2407
+ return None
2408
+
2409
+ # if the option string before the "=" is present, return the action
2410
+ if '=' in arg_string:
2411
+ option_string, explicit_arg = arg_string.split('=', 1)
2412
+ if option_string in self._option_string_actions:
2413
+ action = self._option_string_actions[option_string]
2414
+ return action, option_string, explicit_arg
2415
+
2416
+ # search through all possible prefixes of the option string
2417
+ # and all actions in the parser for possible interpretations
2418
+ option_tuples = self._get_option_tuples(arg_string)
2419
+
2420
+ # if multiple actions match, the option string was ambiguous
2421
+ if len(option_tuples) > 1:
2422
+ options = ', '.join([option_string
2423
+ for action, option_string, explicit_arg in option_tuples])
2424
+ tup = arg_string, options
2425
+ self.error(_('ambiguous option: %s could match %s') % tup)
2426
+
2427
+ # if exactly one action matched, this segmentation is good,
2428
+ # so return the parsed action
2429
+ elif len(option_tuples) == 1:
2430
+ option_tuple, = option_tuples
2431
+ return option_tuple
2432
+
2433
+ # if it was not found as an option, but it looks like a negative
2434
+ # number, it was meant to be positional
2435
+ # unless there are negative-number-like options
2436
+ if self._negative_number_matcher.match(arg_string):
2437
+ if not self._has_negative_number_optionals:
2438
+ return None
2439
+
2440
+ # if it contains a space, it was meant to be a positional
2441
+ if ' ' in arg_string:
2442
+ return None
2443
+
2444
+ # it was meant to be an optional but there is no such option
2445
+ # in this parser (though it might be a valid option in a subparser)
2446
+ return None, arg_string, None
2447
+
2448
+ def _get_option_tuples(self, option_string):
2449
+ result = []
2450
+
2451
+ # option strings starting with two prefix characters are only
2452
+ # split at the '='
2453
+ chars = self.prefix_chars
2454
+ if option_string[0] in chars and option_string[1] in chars:
2455
+ if '=' in option_string:
2456
+ option_prefix, explicit_arg = option_string.split('=', 1)
2457
+ else:
2458
+ option_prefix = option_string
2459
+ explicit_arg = None
2460
+ for option_string in self._option_string_actions:
2461
+ if option_string.startswith(option_prefix):
2462
+ action = self._option_string_actions[option_string]
2463
+ tup = action, option_string, explicit_arg
2464
+ result.append(tup)
2465
+
2466
+ # single character options can be concatenated with their arguments
2467
+ # but multiple character options always have to have their argument
2468
+ # separate
2469
+ elif option_string[0] in chars and option_string[1] not in chars:
2470
+ option_prefix = option_string
2471
+ explicit_arg = None
2472
+ short_option_prefix = option_string[:2]
2473
+ short_explicit_arg = option_string[2:]
2474
+
2475
+ for option_string in self._option_string_actions:
2476
+ if option_string == short_option_prefix:
2477
+ action = self._option_string_actions[option_string]
2478
+ tup = action, option_string, short_explicit_arg
2479
+ result.append(tup)
2480
+ elif option_string.startswith(option_prefix):
2481
+ action = self._option_string_actions[option_string]
2482
+ tup = action, option_string, explicit_arg
2483
+ result.append(tup)
2484
+
2485
+ # shouldn't ever get here
2486
+ else:
2487
+ self.error(_('unexpected option string: %s') % option_string)
2488
+
2489
+ # return the collected option tuples
2490
+ return result
2491
+
2492
+ def _get_nargs_pattern(self, action):
2493
+ # in all examples below, we have to allow for '--' args
2494
+ # which are represented as '-' in the pattern
2495
+ nargs = action.nargs
2496
+
2497
+ # the default (None) is assumed to be a single argument
2498
+ if nargs is None:
2499
+ nargs_pattern = '(-*A-*)'
2500
+
2501
+ # allow zero or one arguments
2502
+ elif nargs == OPTIONAL:
2503
+ nargs_pattern = '(-*A?-*)'
2504
+
2505
+ # allow zero or more arguments
2506
+ elif nargs == ZERO_OR_MORE:
2507
+ nargs_pattern = '(-*[A-]*)'
2508
+
2509
+ # allow one or more arguments
2510
+ elif nargs == ONE_OR_MORE:
2511
+ nargs_pattern = '(-*A[A-]*)'
2512
+
2513
+ # allow any number of options or arguments
2514
+ elif nargs == REMAINDER:
2515
+ nargs_pattern = '([-AO]*)'
2516
+
2517
+ # allow one argument followed by any number of options or arguments
2518
+ elif nargs == PARSER:
2519
+ nargs_pattern = '(-*A[-AO]*)'
2520
+
2521
+ # all others should be integers
2522
+ else:
2523
+ nargs_pattern = '(-*%s-*)' % '-*'.join('A' * nargs)
2524
+
2525
+ # if this is an optional action, -- is not allowed
2526
+ if action.option_strings:
2527
+ nargs_pattern = nargs_pattern.replace('-*', '')
2528
+ nargs_pattern = nargs_pattern.replace('-', '')
2529
+
2530
+ # return the pattern
2531
+ return nargs_pattern
2532
+
2533
+ # ========================
2534
+ # Value conversion methods
2535
+ # ========================
2536
+ def _get_values(self, action, arg_strings):
2537
+ # for everything but PARSER args, strip out '--'
2538
+ if action.nargs not in [PARSER, REMAINDER]:
2539
+ arg_strings = [s for s in arg_strings if s != '--']
2540
+
2541
+ # optional argument produces a default when not present
2542
+ if not arg_strings and action.nargs == OPTIONAL:
2543
+ if action.option_strings:
2544
+ value = action.const
2545
+ else:
2546
+ value = action.default
2547
+ if isinstance(value, basestring):
2548
+ value = self._get_value(action, value)
2549
+ self._check_value(action, value)
2550
+
2551
+ # when nargs='*' on a positional, if there were no command-line
2552
+ # args, use the default if it is anything other than None
2553
+ elif (not arg_strings and action.nargs == ZERO_OR_MORE and
2554
+ not action.option_strings):
2555
+ if action.default is not None:
2556
+ value = action.default
2557
+ else:
2558
+ value = arg_strings
2559
+ self._check_value(action, value)
2560
+
2561
+ # single argument or optional argument produces a single value
2562
+ elif len(arg_strings) == 1 and action.nargs in [None, OPTIONAL]:
2563
+ arg_string, = arg_strings
2564
+ value = self._get_value(action, arg_string)
2565
+ self._check_value(action, value)
2566
+
2567
+ # REMAINDER arguments convert all values, checking none
2568
+ elif action.nargs == REMAINDER:
2569
+ value = [self._get_value(action, v) for v in arg_strings]
2570
+
2571
+ # PARSER arguments convert all values, but check only the first
2572
+ elif action.nargs == PARSER:
2573
+ value = [self._get_value(action, v) for v in arg_strings]
2574
+ self._check_value(action, value[0])
2575
+
2576
+ # all other types of nargs produce a list
2577
+ else:
2578
+ value = [self._get_value(action, v) for v in arg_strings]
2579
+ for v in value:
2580
+ self._check_value(action, v)
2581
+
2582
+ # return the converted value
2583
+ return value
2584
+
2585
+ def _get_value(self, action, arg_string):
2586
+ type_func = self._registry_get('type', action.type, action.type)
2587
+ if not _callable(type_func):
2588
+ msg = _('%r is not callable')
2589
+ raise ArgumentError(action, msg % type_func)
2590
+
2591
+ # convert the value to the appropriate type
2592
+ try:
2593
+ result = type_func(arg_string)
2594
+
2595
+ # ArgumentTypeErrors indicate errors
2596
+ except ArgumentTypeError:
2597
+ name = getattr(action.type, '__name__', repr(action.type))
2598
+ msg = str(_sys.exc_info()[1])
2599
+ raise ArgumentError(action, msg)
2600
+
2601
+ # TypeErrors or ValueErrors also indicate errors
2602
+ except (TypeError, ValueError):
2603
+ name = getattr(action.type, '__name__', repr(action.type))
2604
+ msg = _('invalid %s value: %r')
2605
+ raise ArgumentError(action, msg % (name, arg_string))
2606
+
2607
+ # return the converted value
2608
+ return result
2609
+
2610
+ def _check_value(self, action, value):
2611
+ # converted value must be one of the choices (if specified)
2612
+ if action.choices is not None and value not in action.choices:
2613
+ tup = value, ', '.join(map(repr, action.choices))
2614
+ msg = _('invalid choice: %r (choose from %s)') % tup
2615
+ raise ArgumentError(action, msg)
2616
+
2617
+ # =======================
2618
+ # Help-formatting methods
2619
+ # =======================
2620
+ def format_usage(self):
2621
+ formatter = self._get_formatter()
2622
+ formatter.add_usage(self.usage, self._actions,
2623
+ self._mutually_exclusive_groups)
2624
+ return formatter.format_help()
2625
+
2626
+ def format_help(self):
2627
+ formatter = self._get_formatter()
2628
+
2629
+ # usage
2630
+ formatter.add_usage(self.usage, self._actions,
2631
+ self._mutually_exclusive_groups)
2632
+
2633
+ # description
2634
+ formatter.add_text(self.description)
2635
+
2636
+ # positionals, optionals and user-defined groups
2637
+ for action_group in self._action_groups:
2638
+ formatter.start_section(action_group.title)
2639
+ formatter.add_text(action_group.description)
2640
+ formatter.add_arguments(action_group._group_actions)
2641
+ formatter.end_section()
2642
+
2643
+ # epilog
2644
+ formatter.add_text(self.epilog)
2645
+
2646
+ # determine help from format above
2647
+ return formatter.format_help()
2648
+
2649
+ def format_version(self):
2650
+ import warnings
2651
+ warnings.warn(
2652
+ 'The format_version method is deprecated -- the "version" '
2653
+ 'argument to ArgumentParser is no longer supported.',
2654
+ DeprecationWarning)
2655
+ formatter = self._get_formatter()
2656
+ formatter.add_text(self.version)
2657
+ return formatter.format_help()
2658
+
2659
+ def _get_formatter(self):
2660
+ return self.formatter_class(prog=self.prog)
2661
+
2662
+ # =====================
2663
+ # Help-printing methods
2664
+ # =====================
2665
+ def print_usage(self, file=None):
2666
+ if file is None:
2667
+ file = _sys.stdout
2668
+ self._print_message(self.format_usage(), file)
2669
+
2670
+ def print_help(self, file=None):
2671
+ if file is None:
2672
+ file = _sys.stdout
2673
+ self._print_message(self.format_help(), file)
2674
+
2675
+ def print_version(self, file=None):
2676
+ import warnings
2677
+ warnings.warn(
2678
+ 'The print_version method is deprecated -- the "version" '
2679
+ 'argument to ArgumentParser is no longer supported.',
2680
+ DeprecationWarning)
2681
+ self._print_message(self.format_version(), file)
2682
+
2683
+ def _print_message(self, message, file=None):
2684
+ if message:
2685
+ if file is None:
2686
+ file = _sys.stderr
2687
+ file.write(message)
2688
+
2689
+ # ===============
2690
+ # Exiting methods
2691
+ # ===============
2692
+ def exit(self, status=0, message=None):
2693
+ if message:
2694
+ self._print_message(message, _sys.stderr)
2695
+ _sys.exit(status)
2696
+
2697
+ def error(self, message):
2698
+ """error(message: string)
2699
+
2700
+ Prints a usage message incorporating the message to stderr and
2701
+ exits.
2702
+
2703
+ If you override this in a subclass, it should not return -- it
2704
+ should either exit or raise an exception.
2705
+ """
2706
+ self.print_usage(_sys.stderr)
2707
+ self.exit(2, _('%s: error: %s\n') % (self.prog, message))
2708
--
121
--
2709
2.13.5
122
2.31.1
2710
123
2711
124
diff view generated by jsdifflib
New patch
[PULL 03/19] monitor: hmp_qemu_io: acquire aio contex, fix crash
1
1
From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2
3
Max reported the following bug:
4
5
$ ./qemu-img create -f raw src.img 1G
6
$ ./qemu-img create -f raw dst.img 1G
7
8
$ (echo '
9
{"execute":"qmp_capabilities"}
10
{"execute":"blockdev-mirror",
11
"arguments":{"job-id":"mirror",
12
"device":"source",
13
"target":"target",
14
"sync":"full",
15
"filter-node-name":"mirror-top"}}
16
'; sleep 3; echo '
17
{"execute":"human-monitor-command",
18
"arguments":{"command-line":
19
"qemu-io mirror-top \"write 0 1G\""}}') \
20
| x86_64-softmmu/qemu-system-x86_64 \
21
-qmp stdio \
22
-blockdev file,node-name=source,filename=src.img \
23
-blockdev file,node-name=target,filename=dst.img \
24
-object iothread,id=iothr0 \
25
-device virtio-blk,drive=source,iothread=iothr0
26
27
crashes:
28
29
0 raise () at /usr/lib/libc.so.6
30
1 abort () at /usr/lib/libc.so.6
31
2 error_exit
32
(err=<optimized out>,
33
msg=msg@entry=0x55fbb1634790 <__func__.27> "qemu_mutex_unlock_impl")
34
at ../util/qemu-thread-posix.c:37
35
3 qemu_mutex_unlock_impl
36
(mutex=mutex@entry=0x55fbb25ab6e0,
37
file=file@entry=0x55fbb1636957 "../util/async.c",
38
line=line@entry=650)
39
at ../util/qemu-thread-posix.c:109
40
4 aio_context_release (ctx=ctx@entry=0x55fbb25ab680) at ../util/async.c:650
41
5 bdrv_do_drained_begin
42
(bs=bs@entry=0x55fbb3a87000, recursive=recursive@entry=false,
43
parent=parent@entry=0x0,
44
ignore_bds_parents=ignore_bds_parents@entry=false,
45
poll=poll@entry=true) at ../block/io.c:441
46
6 bdrv_do_drained_begin
47
(poll=true, ignore_bds_parents=false, parent=0x0, recursive=false,
48
bs=0x55fbb3a87000) at ../block/io.c:448
49
7 blk_drain (blk=0x55fbb26c5a00) at ../block/block-backend.c:1718
50
8 blk_unref (blk=0x55fbb26c5a00) at ../block/block-backend.c:498
51
9 blk_unref (blk=0x55fbb26c5a00) at ../block/block-backend.c:491
52
10 hmp_qemu_io (mon=0x7fffaf3fc7d0, qdict=<optimized out>)
53
at ../block/monitor/block-hmp-cmds.c:628
54
55
man pthread_mutex_unlock
56
...
57
EPERM The mutex type is PTHREAD_MUTEX_ERRORCHECK or
58
PTHREAD_MUTEX_RECURSIVE, or the mutex is a robust mutex, and the
59
current thread does not own the mutex.
60
61
So, thread doesn't own the mutex. And we have iothread here.
62
63
Next, note that AIO_WAIT_WHILE() documents that ctx must be acquired
64
exactly once by caller. But where is it acquired in the call stack?
65
Seems nowhere.
66
67
qemuio_command do acquire aio context.. But we need context acquired
68
around blk_unref() as well and actually around blk_insert_bs() too.
69
70
Let's refactor qemuio_command so that it doesn't acquire aio context
71
but callers do that instead. This way we can cleanly acquire aio
72
context in hmp_qemu_io() around all three calls.
73
74
Reported-by: Max Reitz <mreitz@redhat.com>
75
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
76
Message-Id: <20210423134233.51495-1-vsementsov@virtuozzo.com>
77
[mreitz: Fixed comment]
78
Signed-off-by: Max Reitz <mreitz@redhat.com>
79
---
80
block/monitor/block-hmp-cmds.c | 31 +++++++++++++++++++++----------
81
qemu-io-cmds.c | 8 ++++----
82
qemu-io.c | 17 +++++++++++++++--
83
3 files changed, 40 insertions(+), 16 deletions(-)
84
85
diff --git a/block/monitor/block-hmp-cmds.c b/block/monitor/block-hmp-cmds.c
86
index XXXXXXX..XXXXXXX 100644
87
--- a/block/monitor/block-hmp-cmds.c
88
+++ b/block/monitor/block-hmp-cmds.c
89
@@ -XXX,XX +XXX,XX @@ void hmp_eject(Monitor *mon, const QDict *qdict)
90
91
void hmp_qemu_io(Monitor *mon, const QDict *qdict)
92
{
93
- BlockBackend *blk;
94
+ BlockBackend *blk = NULL;
95
+ BlockDriverState *bs = NULL;
96
BlockBackend *local_blk = NULL;
97
+ AioContext *ctx = NULL;
98
bool qdev = qdict_get_try_bool(qdict, "qdev", false);
99
const char *device = qdict_get_str(qdict, "device");
100
const char *command = qdict_get_str(qdict, "command");
101
@@ -XXX,XX +XXX,XX @@ void hmp_qemu_io(Monitor *mon, const QDict *qdict)
102
} else {
103
blk = blk_by_name(device);
104
if (!blk) {
105
- BlockDriverState *bs = bdrv_lookup_bs(NULL, device, &err);
106
- if (bs) {
107
- blk = local_blk = blk_new(bdrv_get_aio_context(bs),
108
- 0, BLK_PERM_ALL);
109
- ret = blk_insert_bs(blk, bs, &err);
110
- if (ret < 0) {
111
- goto fail;
112
- }
113
- } else {
114
+ bs = bdrv_lookup_bs(NULL, device, &err);
115
+ if (!bs) {
116
goto fail;
117
}
118
}
119
}
120
121
+ ctx = blk ? blk_get_aio_context(blk) : bdrv_get_aio_context(bs);
122
+ aio_context_acquire(ctx);
123
+
124
+ if (bs) {
125
+ blk = local_blk = blk_new(bdrv_get_aio_context(bs), 0, BLK_PERM_ALL);
126
+ ret = blk_insert_bs(blk, bs, &err);
127
+ if (ret < 0) {
128
+ goto fail;
129
+ }
130
+ }
131
+
132
/*
133
* Notably absent: Proper permission management. This is sad, but it seems
134
* almost impossible to achieve without changing the semantics and thereby
135
@@ -XXX,XX +XXX,XX @@ void hmp_qemu_io(Monitor *mon, const QDict *qdict)
136
137
fail:
138
blk_unref(local_blk);
139
+
140
+ if (ctx) {
141
+ aio_context_release(ctx);
142
+ }
143
+
144
hmp_handle_error(mon, err);
145
}
146
147
diff --git a/qemu-io-cmds.c b/qemu-io-cmds.c
148
index XXXXXXX..XXXXXXX 100644
149
--- a/qemu-io-cmds.c
150
+++ b/qemu-io-cmds.c
151
@@ -XXX,XX +XXX,XX @@ static const cmdinfo_t help_cmd = {
152
.oneline = "help for one or all commands",
153
};
154
155
+/*
156
+ * Called with aio context of blk acquired. Or with qemu_get_aio_context()
157
+ * context acquired if blk is NULL.
158
+ */
159
int qemuio_command(BlockBackend *blk, const char *cmd)
160
{
161
- AioContext *ctx;
162
char *input;
163
const cmdinfo_t *ct;
164
char **v;
165
@@ -XXX,XX +XXX,XX @@ int qemuio_command(BlockBackend *blk, const char *cmd)
166
if (c) {
167
ct = find_command(v[0]);
168
if (ct) {
169
- ctx = blk ? blk_get_aio_context(blk) : qemu_get_aio_context();
170
- aio_context_acquire(ctx);
171
ret = command(blk, ct, c, v);
172
- aio_context_release(ctx);
173
} else {
174
fprintf(stderr, "command \"%s\" not found\n", v[0]);
175
ret = -EINVAL;
176
diff --git a/qemu-io.c b/qemu-io.c
177
index XXXXXXX..XXXXXXX 100644
178
--- a/qemu-io.c
179
+++ b/qemu-io.c
180
@@ -XXX,XX +XXX,XX @@ static void prep_fetchline(void *opaque)
181
*fetchable= 1;
182
}
183
184
+static int do_qemuio_command(const char *cmd)
185
+{
186
+ int ret;
187
+ AioContext *ctx =
188
+ qemuio_blk ? blk_get_aio_context(qemuio_blk) : qemu_get_aio_context();
189
+
190
+ aio_context_acquire(ctx);
191
+ ret = qemuio_command(qemuio_blk, cmd);
192
+ aio_context_release(ctx);
193
+
194
+ return ret;
195
+}
196
+
197
static int command_loop(void)
198
{
199
int i, fetchable = 0, prompted = 0;
200
@@ -XXX,XX +XXX,XX @@ static int command_loop(void)
201
char *input;
202
203
for (i = 0; !quit_qemu_io && i < ncmdline; i++) {
204
- ret = qemuio_command(qemuio_blk, cmdline[i]);
205
+ ret = do_qemuio_command(cmdline[i]);
206
if (ret < 0) {
207
last_error = ret;
208
}
209
@@ -XXX,XX +XXX,XX @@ static int command_loop(void)
210
if (input == NULL) {
211
break;
212
}
213
- ret = qemuio_command(qemuio_blk, input);
214
+ ret = do_qemuio_command(input);
215
g_free(input);
216
217
if (ret < 0) {
218
--
219
2.31.1
220
221
diff view generated by jsdifflib
[Qemu-devel] [PULL for-2.10 06/15] throttle: Make LeakyBucket.avg and LeakyBucket.max integer types
[PULL 04/19] mirror: stop cancelling in-flight requests on non-force cancel in READY
1
From: Alberto Garcia <berto@igalia.com>
1
From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2
2
3
Both the throttling limits set with the throttling.iops-* and
3
If mirror is READY than cancel operation is not discarding the whole
4
throttling.bps-* options and their QMP equivalents defined in the
4
result of the operation, but instead it's a documented way get a
5
BlockIOThrottle struct are integer values.
5
point-in-time snapshot of source disk.
6
6
7
Those limits are also reported in the BlockDeviceInfo struct and they
7
So, we should not cancel any requests if mirror is READ and
8
are integers there as well.
8
force=false. Let's fix that case.
9
9
10
Therefore there's no reason to store them internally as double and do
10
Note, that bug that we have before this commit is not critical, as the
11
the conversion everytime we're setting or querying them, so this patch
11
only .bdrv_cancel_in_flight implementation is nbd_cancel_in_flight()
12
uses uint64_t for those types. Let's also use an unsigned type because
12
and it cancels only requests waiting for reconnection, so it should be
13
we don't allow negative values anyway.
13
rare case.
14
14
15
LeakyBucket.level and LeakyBucket.burst_level do however remain double
15
Fixes: 521ff8b779b11c394dbdc43f02e158dd99df308a
16
because their value changes depending on the fraction of time elapsed
16
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
17
since the previous I/O operation.
17
Message-Id: <20210421075858.40197-1-vsementsov@virtuozzo.com>
18
Signed-off-by: Max Reitz <mreitz@redhat.com>
19
---
20
include/block/block_int.h | 2 +-
21
include/qemu/job.h | 2 +-
22
block/backup.c | 2 +-
23
block/mirror.c | 6 ++++--
24
job.c | 2 +-
25
tests/qemu-iotests/264 | 2 +-
26
6 files changed, 9 insertions(+), 7 deletions(-)
18
27
19
Signed-off-by: Alberto Garcia <berto@igalia.com>
28
diff --git a/include/block/block_int.h b/include/block/block_int.h
20
Message-id: f29b840422767b5be2c41c2dfdbbbf6c5f8fedf8.1503580370.git.berto@igalia.com
21
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
22
---
23
include/qemu/throttle.h | 4 ++--
24
tests/test-throttle.c | 3 ++-
25
util/throttle.c | 7 +++----
26
3 files changed, 7 insertions(+), 7 deletions(-)
27
28
diff --git a/include/qemu/throttle.h b/include/qemu/throttle.h
29
index XXXXXXX..XXXXXXX 100644
29
index XXXXXXX..XXXXXXX 100644
30
--- a/include/qemu/throttle.h
30
--- a/include/block/block_int.h
31
+++ b/include/qemu/throttle.h
31
+++ b/include/block/block_int.h
32
@@ -XXX,XX +XXX,XX @@ typedef enum {
32
@@ -XXX,XX +XXX,XX @@ struct BlockDriver {
33
*/
33
* of in-flight requests, so don't waste the time if possible.
34
34
*
35
typedef struct LeakyBucket {
35
* One example usage is to avoid waiting for an nbd target node reconnect
36
- double avg; /* average goal in units per second */
36
- * timeout during job-cancel.
37
- double max; /* leaky bucket max burst in units */
37
+ * timeout during job-cancel with force=true.
38
+ uint64_t avg; /* average goal in units per second */
38
*/
39
+ uint64_t max; /* leaky bucket max burst in units */
39
void (*bdrv_cancel_in_flight)(BlockDriverState *bs);
40
double level; /* bucket level in units */
40
41
double burst_level; /* bucket level in units (for computing bursts) */
41
diff --git a/include/qemu/job.h b/include/qemu/job.h
42
unsigned burst_length; /* max length of the burst period, in seconds */
43
diff --git a/tests/test-throttle.c b/tests/test-throttle.c
44
index XXXXXXX..XXXXXXX 100644
42
index XXXXXXX..XXXXXXX 100644
45
--- a/tests/test-throttle.c
43
--- a/include/qemu/job.h
46
+++ b/tests/test-throttle.c
44
+++ b/include/qemu/job.h
47
@@ -XXX,XX +XXX,XX @@ static void test_enabled(void)
45
@@ -XXX,XX +XXX,XX @@ struct JobDriver {
48
for (i = 0; i < BUCKETS_COUNT; i++) {
46
/**
49
throttle_config_init(&cfg);
47
* If the callback is not NULL, it will be invoked in job_cancel_async
50
set_cfg_value(false, i, 150);
48
*/
51
+ g_assert(throttle_is_valid(&cfg, NULL));
49
- void (*cancel)(Job *job);
52
g_assert(throttle_enabled(&cfg));
50
+ void (*cancel)(Job *job, bool force);
53
}
51
54
52
55
for (i = 0; i < BUCKETS_COUNT; i++) {
53
/** Called when the job is freed */
56
throttle_config_init(&cfg);
54
diff --git a/block/backup.c b/block/backup.c
57
set_cfg_value(false, i, -150);
55
index XXXXXXX..XXXXXXX 100644
58
- g_assert(!throttle_enabled(&cfg));
56
--- a/block/backup.c
59
+ g_assert(!throttle_is_valid(&cfg, NULL));
57
+++ b/block/backup.c
58
@@ -XXX,XX +XXX,XX @@ static void coroutine_fn backup_set_speed(BlockJob *job, int64_t speed)
60
}
59
}
61
}
60
}
62
61
63
diff --git a/util/throttle.c b/util/throttle.c
62
-static void backup_cancel(Job *job)
63
+static void backup_cancel(Job *job, bool force)
64
{
65
BackupBlockJob *s = container_of(job, BackupBlockJob, common.job);
66
67
diff --git a/block/mirror.c b/block/mirror.c
64
index XXXXXXX..XXXXXXX 100644
68
index XXXXXXX..XXXXXXX 100644
65
--- a/util/throttle.c
69
--- a/block/mirror.c
66
+++ b/util/throttle.c
70
+++ b/block/mirror.c
67
@@ -XXX,XX +XXX,XX @@ int64_t throttle_compute_wait(LeakyBucket *bkt)
71
@@ -XXX,XX +XXX,XX @@ static bool mirror_drained_poll(BlockJob *job)
68
/* If bkt->max is 0 we still want to allow short bursts of I/O
72
return !!s->in_flight;
69
* from the guest, otherwise every other request will be throttled
73
}
70
* and performance will suffer considerably. */
74
71
- bucket_size = bkt->avg / 10;
75
-static void mirror_cancel(Job *job)
72
+ bucket_size = (double) bkt->avg / 10;
76
+static void mirror_cancel(Job *job, bool force)
73
burst_bucket_size = 0;
77
{
74
} else {
78
MirrorBlockJob *s = container_of(job, MirrorBlockJob, common.job);
75
/* If we have a burst limit then we have to wait until all I/O
79
BlockDriverState *target = blk_bs(s->target);
76
* at burst rate has finished before throttling to bkt->avg */
80
77
bucket_size = bkt->max * bkt->burst_length;
81
- bdrv_cancel_in_flight(target);
78
- burst_bucket_size = bkt->max / 10;
82
+ if (force || !job_is_ready(job)) {
79
+ burst_bucket_size = (double) bkt->max / 10;
83
+ bdrv_cancel_in_flight(target);
84
+ }
85
}
86
87
static const BlockJobDriver mirror_job_driver = {
88
diff --git a/job.c b/job.c
89
index XXXXXXX..XXXXXXX 100644
90
--- a/job.c
91
+++ b/job.c
92
@@ -XXX,XX +XXX,XX @@ static int job_finalize_single(Job *job)
93
static void job_cancel_async(Job *job, bool force)
94
{
95
if (job->driver->cancel) {
96
- job->driver->cancel(job);
97
+ job->driver->cancel(job, force);
80
}
98
}
81
99
if (job->user_paused) {
82
/* If the main bucket is full then we have to wait */
100
/* Do not call job_enter here, the caller will handle it. */
83
@@ -XXX,XX +XXX,XX @@ bool throttle_is_valid(ThrottleConfig *cfg, Error **errp)
101
diff --git a/tests/qemu-iotests/264 b/tests/qemu-iotests/264
84
102
index XXXXXXX..XXXXXXX 100755
85
for (i = 0; i < BUCKETS_COUNT; i++) {
103
--- a/tests/qemu-iotests/264
86
LeakyBucket *bkt = &cfg->buckets[i];
104
+++ b/tests/qemu-iotests/264
87
- if (bkt->avg < 0 || bkt->max < 0 ||
105
@@ -XXX,XX +XXX,XX @@ class TestNbdReconnect(iotests.QMPTestCase):
88
- bkt->avg > THROTTLE_VALUE_MAX || bkt->max > THROTTLE_VALUE_MAX) {
106
self.assert_qmp(result, 'return', {})
89
+ if (bkt->avg > THROTTLE_VALUE_MAX || bkt->max > THROTTLE_VALUE_MAX) {
107
90
error_setg(errp, "bps/iops/max values must be within [0, %lld]",
108
def cancel_job(self):
91
THROTTLE_VALUE_MAX);
109
- result = self.vm.qmp('block-job-cancel', device='drive0')
92
return false;
110
+ result = self.vm.qmp('block-job-cancel', device='drive0', force=True)
111
self.assert_qmp(result, 'return', {})
112
113
start_t = time.time()
93
--
114
--
94
2.13.5
115
2.31.1
95
116
96
117
diff view generated by jsdifflib
New patch
[PULL 05/19] qemu-iotests: do not buffer the test output
1
1
From: Paolo Bonzini <pbonzini@redhat.com>
2
3
Instead of buffering the test output into a StringIO, patch it on
4
the fly by wrapping sys.stdout's write method. This can be
5
done unconditionally, even if using -d, which makes execute_unittest
6
a bit simpler.
7
8
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
9
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
10
Tested-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
11
Message-Id: <20210323181928.311862-2-pbonzini@redhat.com>
12
Message-Id: <20210503110110.476887-2-pbonzini@redhat.com>
13
Signed-off-by: Max Reitz <mreitz@redhat.com>
14
---
15
tests/qemu-iotests/240.out | 8 ++--
16
tests/qemu-iotests/245.out | 8 ++--
17
tests/qemu-iotests/295.out | 6 +--
18
tests/qemu-iotests/296.out | 8 ++--
19
tests/qemu-iotests/iotests.py | 70 ++++++++++++++++++++---------------
20
5 files changed, 56 insertions(+), 44 deletions(-)
21
22
diff --git a/tests/qemu-iotests/240.out b/tests/qemu-iotests/240.out
23
index XXXXXXX..XXXXXXX 100644
24
--- a/tests/qemu-iotests/240.out
25
+++ b/tests/qemu-iotests/240.out
26
@@ -XXX,XX +XXX,XX @@
27
{"return": {}}
28
{"execute": "blockdev-del", "arguments": {"node-name": "hd0"}}
29
{"return": {}}
30
-==Attach two SCSI disks using the same block device and the same iothread==
31
+.==Attach two SCSI disks using the same block device and the same iothread==
32
{"execute": "blockdev-add", "arguments": {"driver": "null-co", "node-name": "hd0", "read-only": true, "read-zeroes": true}}
33
{"return": {}}
34
{"execute": "object-add", "arguments": {"id": "iothread0", "qom-type": "iothread"}}
35
@@ -XXX,XX +XXX,XX @@
36
{"return": {}}
37
{"execute": "blockdev-del", "arguments": {"node-name": "hd0"}}
38
{"return": {}}
39
-==Attach two SCSI disks using the same block device but different iothreads==
40
+.==Attach two SCSI disks using the same block device but different iothreads==
41
{"execute": "blockdev-add", "arguments": {"driver": "null-co", "node-name": "hd0", "read-only": true, "read-zeroes": true}}
42
{"return": {}}
43
{"execute": "object-add", "arguments": {"id": "iothread0", "qom-type": "iothread"}}
44
@@ -XXX,XX +XXX,XX @@
45
{"return": {}}
46
{"execute": "blockdev-del", "arguments": {"node-name": "hd0"}}
47
{"return": {}}
48
-==Attach a SCSI disks using the same block device as a NBD server==
49
+.==Attach a SCSI disks using the same block device as a NBD server==
50
{"execute": "blockdev-add", "arguments": {"driver": "null-co", "node-name": "hd0", "read-only": true, "read-zeroes": true}}
51
{"return": {}}
52
{"execute": "nbd-server-start", "arguments": {"addr": {"data": {"path": "SOCK_DIR/PID-nbd.sock"}, "type": "unix"}}}
53
@@ -XXX,XX +XXX,XX @@
54
{"return": {}}
55
{"execute": "device_add", "arguments": {"drive": "hd0", "driver": "scsi-hd", "id": "scsi-hd0"}}
56
{"return": {}}
57
-....
58
+.
59
----------------------------------------------------------------------
60
Ran 4 tests
61
62
diff --git a/tests/qemu-iotests/245.out b/tests/qemu-iotests/245.out
63
index XXXXXXX..XXXXXXX 100644
64
--- a/tests/qemu-iotests/245.out
65
+++ b/tests/qemu-iotests/245.out
66
@@ -XXX,XX +XXX,XX @@
67
-{"execute": "job-finalize", "arguments": {"id": "commit0"}}
68
+..{"execute": "job-finalize", "arguments": {"id": "commit0"}}
69
{"return": {}}
70
{"data": {"id": "commit0", "type": "commit"}, "event": "BLOCK_JOB_PENDING", "timestamp": {"microseconds": "USECS", "seconds": "SECS"}}
71
{"data": {"device": "commit0", "len": 3145728, "offset": 3145728, "speed": 0, "type": "commit"}, "event": "BLOCK_JOB_COMPLETED", "timestamp": {"microseconds": "USECS", "seconds": "SECS"}}
72
-{"execute": "job-finalize", "arguments": {"id": "stream0"}}
73
+...{"execute": "job-finalize", "arguments": {"id": "stream0"}}
74
{"return": {}}
75
{"data": {"id": "stream0", "type": "stream"}, "event": "BLOCK_JOB_PENDING", "timestamp": {"microseconds": "USECS", "seconds": "SECS"}}
76
{"data": {"device": "stream0", "len": 3145728, "offset": 3145728, "speed": 0, "type": "stream"}, "event": "BLOCK_JOB_COMPLETED", "timestamp": {"microseconds": "USECS", "seconds": "SECS"}}
77
-{"execute": "job-finalize", "arguments": {"id": "stream0"}}
78
+.{"execute": "job-finalize", "arguments": {"id": "stream0"}}
79
{"return": {}}
80
{"data": {"id": "stream0", "type": "stream"}, "event": "BLOCK_JOB_PENDING", "timestamp": {"microseconds": "USECS", "seconds": "SECS"}}
81
{"data": {"device": "stream0", "len": 3145728, "offset": 3145728, "speed": 0, "type": "stream"}, "event": "BLOCK_JOB_COMPLETED", "timestamp": {"microseconds": "USECS", "seconds": "SECS"}}
82
-.....................
83
+...............
84
----------------------------------------------------------------------
85
Ran 21 tests
86
87
diff --git a/tests/qemu-iotests/295.out b/tests/qemu-iotests/295.out
88
index XXXXXXX..XXXXXXX 100644
89
--- a/tests/qemu-iotests/295.out
90
+++ b/tests/qemu-iotests/295.out
91
@@ -XXX,XX +XXX,XX @@
92
{"return": {}}
93
{"execute": "job-dismiss", "arguments": {"id": "job_erase_key"}}
94
{"return": {}}
95
-{"execute": "job-dismiss", "arguments": {"id": "job_add_key"}}
96
+.{"execute": "job-dismiss", "arguments": {"id": "job_add_key"}}
97
{"return": {}}
98
{"execute": "job-dismiss", "arguments": {"id": "job_erase_key"}}
99
{"return": {}}
100
@@ -XXX,XX +XXX,XX @@ Job failed: Invalid password, cannot unlock any keyslot
101
{"return": {}}
102
{"execute": "job-dismiss", "arguments": {"id": "job_add_key"}}
103
{"return": {}}
104
-{"execute": "job-dismiss", "arguments": {"id": "job_add_key"}}
105
+.{"execute": "job-dismiss", "arguments": {"id": "job_add_key"}}
106
{"return": {}}
107
{"execute": "job-dismiss", "arguments": {"id": "job_add_key"}}
108
{"return": {}}
109
@@ -XXX,XX +XXX,XX @@ Job failed: All the active keyslots match the (old) password that was given and
110
{"return": {}}
111
{"execute": "job-dismiss", "arguments": {"id": "job_erase_key"}}
112
{"return": {}}
113
-...
114
+.
115
----------------------------------------------------------------------
116
Ran 3 tests
117
118
diff --git a/tests/qemu-iotests/296.out b/tests/qemu-iotests/296.out
119
index XXXXXXX..XXXXXXX 100644
120
--- a/tests/qemu-iotests/296.out
121
+++ b/tests/qemu-iotests/296.out
122
@@ -XXX,XX +XXX,XX @@ Job failed: Failed to get shared "consistent read" lock
123
qemu-img: Failed to get shared "consistent read" lock
124
Is another process using the image [TEST_DIR/test.img]?
125
126
-Formatting 'TEST_DIR/test.img', fmt=luks size=1048576 key-secret=keysec0 iter-time=10
127
+.Formatting 'TEST_DIR/test.img', fmt=luks size=1048576 key-secret=keysec0 iter-time=10
128
129
Job failed: Block node is read-only
130
{"execute": "job-dismiss", "arguments": {"id": "job0"}}
131
@@ -XXX,XX +XXX,XX @@ Job failed: Failed to get shared "consistent read" lock
132
{"return": {}}
133
{"execute": "job-dismiss", "arguments": {"id": "job0"}}
134
{"return": {}}
135
-Formatting 'TEST_DIR/test.img', fmt=luks size=1048576 key-secret=keysec0 iter-time=10
136
+.Formatting 'TEST_DIR/test.img', fmt=luks size=1048576 key-secret=keysec0 iter-time=10
137
138
{"return": {}}
139
{"error": {"class": "GenericError", "desc": "Failed to get \"write\" lock"}}
140
-Formatting 'TEST_DIR/test.img', fmt=luks size=1048576 key-secret=keysec0 iter-time=10
141
+.Formatting 'TEST_DIR/test.img', fmt=luks size=1048576 key-secret=keysec0 iter-time=10
142
143
{"return": {}}
144
{"return": {}}
145
-....
146
+.
147
----------------------------------------------------------------------
148
Ran 4 tests
149
150
diff --git a/tests/qemu-iotests/iotests.py b/tests/qemu-iotests/iotests.py
151
index XXXXXXX..XXXXXXX 100644
152
--- a/tests/qemu-iotests/iotests.py
153
+++ b/tests/qemu-iotests/iotests.py
154
@@ -XXX,XX +XXX,XX @@
155
import bz2
156
from collections import OrderedDict
157
import faulthandler
158
-import io
159
import json
160
import logging
161
import os
162
@@ -XXX,XX +XXX,XX @@
163
import sys
164
import time
165
from typing import (Any, Callable, Dict, Iterable,
166
- List, Optional, Sequence, Tuple, TypeVar)
167
+ List, Optional, Sequence, TextIO, Tuple, Type, TypeVar)
168
import unittest
169
170
from contextlib import contextmanager
171
@@ -XXX,XX +XXX,XX @@ def func_wrapper(*args, **kwargs):
172
return func(*args, **kwargs)
173
return func_wrapper
174
175
+# We need to filter out the time taken from the output so that
176
+# qemu-iotest can reliably diff the results against master output,
177
+# and hide skipped tests from the reference output.
178
+
179
+class ReproducibleTestResult(unittest.TextTestResult):
180
+ def addSkip(self, test, reason):
181
+ # Same as TextTestResult, but print dot instead of "s"
182
+ unittest.TestResult.addSkip(self, test, reason)
183
+ if self.showAll:
184
+ self.stream.writeln("skipped {0!r}".format(reason))
185
+ elif self.dots:
186
+ self.stream.write(".")
187
+ self.stream.flush()
188
+
189
+class ReproducibleStreamWrapper:
190
+ def __init__(self, stream: TextIO):
191
+ self.stream = stream
192
+
193
+ def __getattr__(self, attr):
194
+ if attr in ('stream', '__getstate__'):
195
+ raise AttributeError(attr)
196
+ return getattr(self.stream, attr)
197
+
198
+ def write(self, arg=None):
199
+ arg = re.sub(r'Ran (\d+) tests? in [\d.]+s', r'Ran \1 tests', arg)
200
+ arg = re.sub(r' \(skipped=\d+\)', r'', arg)
201
+ self.stream.write(arg)
202
+
203
+class ReproducibleTestRunner(unittest.TextTestRunner):
204
+ def __init__(self, stream: Optional[TextIO] = None,
205
+ resultclass: Type[unittest.TestResult] = ReproducibleTestResult,
206
+ **kwargs: Any) -> None:
207
+ rstream = ReproducibleStreamWrapper(stream or sys.stdout)
208
+ super().__init__(stream=rstream, # type: ignore
209
+ descriptions=True,
210
+ resultclass=resultclass,
211
+ **kwargs)
212
+
213
def execute_unittest(debug=False):
214
"""Executes unittests within the calling module."""
215
216
verbosity = 2 if debug else 1
217
-
218
- if debug:
219
- output = sys.stdout
220
- else:
221
- # We need to filter out the time taken from the output so that
222
- # qemu-iotest can reliably diff the results against master output.
223
- output = io.StringIO()
224
-
225
- runner = unittest.TextTestRunner(stream=output, descriptions=True,
226
- verbosity=verbosity)
227
- try:
228
- # unittest.main() will use sys.exit(); so expect a SystemExit
229
- # exception
230
- unittest.main(testRunner=runner)
231
- finally:
232
- # We need to filter out the time taken from the output so that
233
- # qemu-iotest can reliably diff the results against master output.
234
- if not debug:
235
- out = output.getvalue()
236
- out = re.sub(r'Ran (\d+) tests? in [\d.]+s', r'Ran \1 tests', out)
237
-
238
- # Hide skipped tests from the reference output
239
- out = re.sub(r'OK \(skipped=\d+\)', 'OK', out)
240
- out_first_line, out_rest = out.split('\n', 1)
241
- out = out_first_line.replace('s', '.') + '\n' + out_rest
242
-
243
- sys.stderr.write(out)
244
+ runner = ReproducibleTestRunner(verbosity=verbosity)
245
+ unittest.main(testRunner=runner)
246
247
def execute_setup_common(supported_fmts: Sequence[str] = (),
248
supported_platforms: Sequence[str] = (),
249
--
250
2.31.1
251
252
diff view generated by jsdifflib
New patch
[PULL 06/19] qemu-iotests: allow passing unittest.main arguments to the test scripts
1
From: Paolo Bonzini <pbonzini@redhat.com>
1
2
3
Python test scripts that use unittest consist of multiple tests.
4
unittest.main allows selecting which tests to run, but currently this
5
is not possible because the iotests wrapper ignores sys.argv.
6
7
unittest.main command line options also allow the user to pick the
8
desired options for verbosity, failfast mode, etc. While "-d" is
9
currently translated to "-v", it also enables extra debug output,
10
and other options are not available at all.
11
12
These command line options only work if the unittest.main testRunner
13
argument is a type, rather than a TestRunner instance. Therefore, pass
14
the class name and "verbosity" argument to unittest.main, and adjust for
15
the different default warnings between TextTestRunner and unittest.main.
16
17
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
18
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
19
Tested-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
20
Message-Id: <20210323181928.311862-3-pbonzini@redhat.com>
21
Message-Id: <20210503110110.476887-3-pbonzini@redhat.com>
22
Signed-off-by: Max Reitz <mreitz@redhat.com>
23
---
24
tests/qemu-iotests/iotests.py | 14 +++++++++-----
25
1 file changed, 9 insertions(+), 5 deletions(-)
26
27
diff --git a/tests/qemu-iotests/iotests.py b/tests/qemu-iotests/iotests.py
28
index XXXXXXX..XXXXXXX 100644
29
--- a/tests/qemu-iotests/iotests.py
30
+++ b/tests/qemu-iotests/iotests.py
31
@@ -XXX,XX +XXX,XX @@ def __init__(self, stream: Optional[TextIO] = None,
32
resultclass=resultclass,
33
**kwargs)
34
35
-def execute_unittest(debug=False):
36
+def execute_unittest(argv: List[str], debug: bool = False) -> None:
37
"""Executes unittests within the calling module."""
38
39
- verbosity = 2 if debug else 1
40
- runner = ReproducibleTestRunner(verbosity=verbosity)
41
- unittest.main(testRunner=runner)
42
+ # Some tests have warnings, especially ResourceWarnings for unclosed
43
+ # files and sockets. Ignore them for now to ensure reproducibility of
44
+ # the test output.
45
+ unittest.main(argv=argv,
46
+ testRunner=ReproducibleTestRunner,
47
+ verbosity=2 if debug else 1,
48
+ warnings=None if sys.warnoptions else 'ignore')
49
50
def execute_setup_common(supported_fmts: Sequence[str] = (),
51
supported_platforms: Sequence[str] = (),
52
@@ -XXX,XX +XXX,XX @@ def execute_test(*args, test_function=None, **kwargs):
53
54
debug = execute_setup_common(*args, **kwargs)
55
if not test_function:
56
- execute_unittest(debug)
57
+ execute_unittest(sys.argv, debug)
58
else:
59
test_function()
60
61
--
62
2.31.1
63
64
diff view generated by jsdifflib
[Qemu-devel] [PULL for-2.10 14/15] qemu-doc: Add UUID support in initiator name
[PULL 07/19] qemu-iotests: move command line and environment handling from TestRunner to TestEnv
1
From: Fred Rolland <rollandf@gmail.com>
1
From: Paolo Bonzini <pbonzini@redhat.com>
2
2
3
Update doc with the usage of UUID for initiator name.
3
In the next patch, "check" will learn how to execute a test script without
4
going through TestRunner. To enable this, keep only the text output
5
and subprocess handling in the TestRunner; move into TestEnv the logic
6
to prepare for running a subprocess.
4
7
5
Related-To: https://bugzilla.redhat.com/1006468
8
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
6
Signed-off-by: Fred Rolland <frolland@redhat.com>
9
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
7
Message-id: 20170823084830.30500-1-frolland@redhat.com
10
Tested-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
8
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
11
Message-Id: <20210323181928.311862-4-pbonzini@redhat.com>
12
Message-Id: <20210503110110.476887-4-pbonzini@redhat.com>
13
Signed-off-by: Max Reitz <mreitz@redhat.com>
9
---
14
---
10
qemu-doc.texi | 5 +++--
15
tests/qemu-iotests/testenv.py | 17 ++++++++++++++++-
11
1 file changed, 3 insertions(+), 2 deletions(-)
16
tests/qemu-iotests/testrunner.py | 14 +-------------
17
2 files changed, 17 insertions(+), 14 deletions(-)
12
18
13
diff --git a/qemu-doc.texi b/qemu-doc.texi
19
diff --git a/tests/qemu-iotests/testenv.py b/tests/qemu-iotests/testenv.py
14
index XXXXXXX..XXXXXXX 100644
20
index XXXXXXX..XXXXXXX 100644
15
--- a/qemu-doc.texi
21
--- a/tests/qemu-iotests/testenv.py
16
+++ b/qemu-doc.texi
22
+++ b/tests/qemu-iotests/testenv.py
17
@@ -XXX,XX +XXX,XX @@ in a configuration file provided via '-readconfig' or directly on the
23
@@ -XXX,XX +XXX,XX @@
18
command line.
24
import random
19
25
import subprocess
20
If the initiator-name is not specified qemu will use a default name
26
import glob
21
-of 'iqn.2008-11.org.linux-kvm[:<name>'] where <name> is the name of the
27
-from typing import Dict, Any, Optional, ContextManager
22
+of 'iqn.2008-11.org.linux-kvm[:<uuid>'] where <uuid> is the UUID of the
28
+from typing import List, Dict, Any, Optional, ContextManager
23
+virtual machine. If the UUID is not specified qemu will use
29
24
+'iqn.2008-11.org.linux-kvm[:<name>'] where <name> is the name of the
30
25
virtual machine.
31
def isxfile(path: str) -> bool:
26
32
@@ -XXX,XX +XXX,XX @@ class TestEnv(ContextManager['TestEnv']):
33
'CACHEMODE_IS_DEFAULT', 'IMGFMT_GENERIC', 'IMGOPTSSYNTAX',
34
'IMGKEYSECRET', 'QEMU_DEFAULT_MACHINE', 'MALLOC_PERTURB_']
35
36
+ def prepare_subprocess(self, args: List[str]) -> Dict[str, str]:
37
+ if self.debug:
38
+ args.append('-d')
39
+
40
+ with open(args[0], encoding="utf-8") as f:
41
+ try:
42
+ if f.readline().rstrip() == '#!/usr/bin/env python3':
43
+ args.insert(0, self.python)
44
+ except UnicodeDecodeError: # binary test? for future.
45
+ pass
46
+
47
+ os_env = os.environ.copy()
48
+ os_env.update(self.get_env())
49
+ return os_env
50
+
51
def get_env(self) -> Dict[str, str]:
52
env = {}
53
for v in self.env_variables:
54
diff --git a/tests/qemu-iotests/testrunner.py b/tests/qemu-iotests/testrunner.py
55
index XXXXXXX..XXXXXXX 100644
56
--- a/tests/qemu-iotests/testrunner.py
57
+++ b/tests/qemu-iotests/testrunner.py
58
@@ -XXX,XX +XXX,XX @@ class TestRunner(ContextManager['TestRunner']):
59
def __init__(self, env: TestEnv, makecheck: bool = False,
60
color: str = 'auto') -> None:
61
self.env = env
62
- self.test_run_env = self.env.get_env()
63
self.makecheck = makecheck
64
self.last_elapsed = LastElapsedTime('.last-elapsed-cache', env)
65
66
@@ -XXX,XX +XXX,XX @@ def do_run_test(self, test: str) -> TestResult:
67
silent_unlink(p)
68
69
args = [str(f_test.resolve())]
70
- if self.env.debug:
71
- args.append('-d')
27
-
72
-
28
@example
73
- with f_test.open(encoding="utf-8") as f:
29
Setting a specific initiator name to use when logging in to the target
74
- try:
30
-iscsi initiator-name=iqn.qemu.test:my-initiator
75
- if f.readline().rstrip() == '#!/usr/bin/env python3':
76
- args.insert(0, self.env.python)
77
- except UnicodeDecodeError: # binary test? for future.
78
- pass
79
-
80
- env = os.environ.copy()
81
- env.update(self.test_run_env)
82
+ env = self.env.prepare_subprocess(args)
83
84
t0 = time.time()
85
with f_bad.open('w', encoding="utf-8") as f:
31
--
86
--
32
2.13.5
87
2.31.1
33
88
34
89
diff view generated by jsdifflib
[Qemu-devel] [PULL for-2.10 12/15] docker.py: Python 2.6 argparse compatibility
[PULL 08/19] qemu-iotests: let "check" spawn an arbitrary test command
1
Add the scripts/ directory to sys.path so Python 2.6 will be able to
1
From: Paolo Bonzini <pbonzini@redhat.com>
2
import argparse.
3
2
4
Cc: Fam Zheng <famz@redhat.com>
3
Right now there is no easy way for "check" to print a reproducer command.
5
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
4
Because such a reproducer command line would be huge, we can instead teach
6
Acked-by: John Snow <jsnow@redhat.com>
5
check to start a command of our choice. This can be for example a Python
7
Acked-by: Fam Zheng <famz@redhat.com>
6
unit test with arguments to only run a specific subtest.
8
Message-id: 20170825155732.15665-3-stefanha@redhat.com
7
9
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8
Move the trailing empty line to print_env(), since it always looks better
9
and one caller was not adding it.
10
11
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
12
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
13
Tested-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
14
Message-Id: <20210323181928.311862-5-pbonzini@redhat.com>
15
Message-Id: <20210503110110.476887-5-pbonzini@redhat.com>
16
Signed-off-by: Max Reitz <mreitz@redhat.com>
10
---
17
---
11
tests/docker/docker.py | 4 +++-
18
tests/qemu-iotests/check | 19 ++++++++++++++++++-
12
1 file changed, 3 insertions(+), 1 deletion(-)
19
tests/qemu-iotests/testenv.py | 3 ++-
20
tests/qemu-iotests/testrunner.py | 1 -
21
3 files changed, 20 insertions(+), 3 deletions(-)
13
22
14
diff --git a/tests/docker/docker.py b/tests/docker/docker.py
23
diff --git a/tests/qemu-iotests/check b/tests/qemu-iotests/check
15
index XXXXXXX..XXXXXXX 100755
24
index XXXXXXX..XXXXXXX 100755
16
--- a/tests/docker/docker.py
25
--- a/tests/qemu-iotests/check
17
+++ b/tests/docker/docker.py
26
+++ b/tests/qemu-iotests/check
18
@@ -XXX,XX +XXX,XX @@
27
@@ -XXX,XX +XXX,XX @@
19
20
import os
28
import os
21
import sys
29
import sys
22
+sys.path.append(os.path.join(os.path.dirname(__file__),
30
import argparse
23
+ '..', '..', 'scripts'))
31
+import shutil
24
+import argparse
32
+from pathlib import Path
25
import subprocess
33
+
26
import json
34
from findtests import TestFinder
27
import hashlib
35
from testenv import TestEnv
28
import atexit
36
from testrunner import TestRunner
29
import uuid
37
@@ -XXX,XX +XXX,XX @@ def make_argparser() -> argparse.ArgumentParser:
30
-import argparse
38
'rerun failed ./check command, starting from the '
31
import tempfile
39
'middle of the process.')
32
import re
40
g_sel.add_argument('tests', metavar='TEST_FILES', nargs='*',
33
import signal
41
- help='tests to run')
42
+ help='tests to run, or "--" followed by a command')
43
44
return p
45
46
@@ -XXX,XX +XXX,XX @@ if __name__ == '__main__':
47
imgopts=args.imgopts, misalign=args.misalign,
48
debug=args.debug, valgrind=args.valgrind)
49
50
+ if len(sys.argv) > 1 and sys.argv[-len(args.tests)-1] == '--':
51
+ if not args.tests:
52
+ sys.exit("missing command after '--'")
53
+ cmd = args.tests
54
+ env.print_env()
55
+ exec_pathstr = shutil.which(cmd[0])
56
+ if exec_pathstr is None:
57
+ sys.exit('command not found: ' + cmd[0])
58
+ exec_path = Path(exec_pathstr).resolve()
59
+ cmd[0] = str(exec_path)
60
+ full_env = env.prepare_subprocess(cmd)
61
+ os.chdir(exec_path.parent)
62
+ os.execve(cmd[0], cmd, full_env)
63
+
64
testfinder = TestFinder(test_dir=env.source_iotests)
65
66
groups = args.groups.split(',') if args.groups else None
67
diff --git a/tests/qemu-iotests/testenv.py b/tests/qemu-iotests/testenv.py
68
index XXXXXXX..XXXXXXX 100644
69
--- a/tests/qemu-iotests/testenv.py
70
+++ b/tests/qemu-iotests/testenv.py
71
@@ -XXX,XX +XXX,XX @@ def print_env(self) -> None:
72
PLATFORM -- {platform}
73
TEST_DIR -- {TEST_DIR}
74
SOCK_DIR -- {SOCK_DIR}
75
-SOCKET_SCM_HELPER -- {SOCKET_SCM_HELPER}"""
76
+SOCKET_SCM_HELPER -- {SOCKET_SCM_HELPER}
77
+"""
78
79
args = collections.defaultdict(str, self.get_env())
80
81
diff --git a/tests/qemu-iotests/testrunner.py b/tests/qemu-iotests/testrunner.py
82
index XXXXXXX..XXXXXXX 100644
83
--- a/tests/qemu-iotests/testrunner.py
84
+++ b/tests/qemu-iotests/testrunner.py
85
@@ -XXX,XX +XXX,XX @@ def run_tests(self, tests: List[str]) -> bool:
86
87
if not self.makecheck:
88
self.env.print_env()
89
- print()
90
91
test_field_width = max(len(os.path.basename(t)) for t in tests) + 2
92
34
--
93
--
35
2.13.5
94
2.31.1
36
95
37
96
diff view generated by jsdifflib
[Qemu-devel] [PULL for-2.10 02/15] throttle: Fix wrong variable name in the header documentation
[PULL 09/19] qemu-iotests: fix case of SOCK_DIR already in the environment
1
From: Alberto Garcia <berto@igalia.com>
1
From: Paolo Bonzini <pbonzini@redhat.com>
2
2
3
The level of the burst bucket is stored in bkt.burst_level, not
3
Due to a typo, in this case the SOCK_DIR was not being created.
4
bkt.burst_length.
5
4
6
Signed-off-by: Alberto Garcia <berto@igalia.com>
5
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
7
Reviewed-by: Manos Pitsidianakis <el13635@mail.ntua.gr>
6
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8
Message-id: 49aab2711d02f285567f3b3b13a113847af33812.1503580370.git.berto@igalia.com
7
Tested-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
9
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8
Message-Id: <20210323181928.311862-6-pbonzini@redhat.com>
9
Message-Id: <20210503110110.476887-6-pbonzini@redhat.com>
10
Signed-off-by: Max Reitz <mreitz@redhat.com>
10
---
11
---
11
include/qemu/throttle.h | 2 +-
12
tests/qemu-iotests/testenv.py | 2 +-
12
1 file changed, 1 insertion(+), 1 deletion(-)
13
1 file changed, 1 insertion(+), 1 deletion(-)
13
14
14
diff --git a/include/qemu/throttle.h b/include/qemu/throttle.h
15
diff --git a/tests/qemu-iotests/testenv.py b/tests/qemu-iotests/testenv.py
15
index XXXXXXX..XXXXXXX 100644
16
index XXXXXXX..XXXXXXX 100644
16
--- a/include/qemu/throttle.h
17
--- a/tests/qemu-iotests/testenv.py
17
+++ b/include/qemu/throttle.h
18
+++ b/tests/qemu-iotests/testenv.py
18
@@ -XXX,XX +XXX,XX @@ typedef enum {
19
@@ -XXX,XX +XXX,XX @@ def init_directories(self) -> None:
19
* - The bkt.avg rate does not apply until the bucket is full,
20
try:
20
* allowing the user to do bursts until then. The I/O limit during
21
self.sock_dir = os.environ['SOCK_DIR']
21
* bursts is bkt.max. To enforce this limit we keep an additional
22
self.tmp_sock_dir = False
22
- * bucket in bkt.burst_length that leaks at a rate of bkt.max units
23
- Path(self.test_dir).mkdir(parents=True, exist_ok=True)
23
+ * bucket in bkt.burst_level that leaks at a rate of bkt.max units
24
+ Path(self.sock_dir).mkdir(parents=True, exist_ok=True)
24
* per second.
25
except KeyError:
25
*
26
self.sock_dir = tempfile.mkdtemp()
26
* - Because of all of the above, the user can perform I/O at a
27
self.tmp_sock_dir = True
27
--
28
--
28
2.13.5
29
2.31.1
29
30
30
31
diff view generated by jsdifflib
[Qemu-devel] [PULL for-2.10 13/15] tests: migration/guestperf Python 2.6 argparse compatibility
[PULL 10/19] Document qemu-img options data_file and data_file_raw
1
Add the scripts/ directory to sys.path so Python 2.6 will be able to
1
From: Connor Kuehl <ckuehl@redhat.com>
2
import argparse.
3
2
4
Cc: Daniel P. Berrange <berrange@redhat.com>
3
The contents of this patch were initially developed and posted by Han
5
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
4
Han[1], however, it appears the original patch was not applied. Since
6
Acked-by: John Snow <jsnow@redhat.com>
5
then, the relevant documentation has been moved and adapted to a new
7
Acked-by: Fam Zheng <famz@redhat.com>
6
format.
8
Message-id: 20170825155732.15665-4-stefanha@redhat.com
7
9
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8
I've taken most of the original wording and tweaked it according to
9
some of the feedback from the original patch submission. I've also
10
adapted it to restructured text, which is the format the documentation
11
currently uses.
12
13
[1] https://lists.nongnu.org/archive/html/qemu-block/2019-10/msg01253.html
14
15
Fixes: https://bugzilla.redhat.com/1763105
16
Signed-off-by: Han Han <hhan@redhat.com>
17
Suggested-by: Max Reitz <mreitz@redhat.com>
18
[ Max: provided description of data_file_raw behavior ]
19
Signed-off-by: Connor Kuehl <ckuehl@redhat.com>
20
Message-Id: <20210505195512.391128-1-ckuehl@redhat.com>
21
Signed-off-by: Max Reitz <mreitz@redhat.com>
10
---
22
---
11
tests/migration/guestperf/shell.py | 8 +++++---
23
docs/tools/qemu-img.rst | 31 +++++++++++++++++++++++++++++++
12
1 file changed, 5 insertions(+), 3 deletions(-)
24
1 file changed, 31 insertions(+)
13
25
14
diff --git a/tests/migration/guestperf/shell.py b/tests/migration/guestperf/shell.py
26
diff --git a/docs/tools/qemu-img.rst b/docs/tools/qemu-img.rst
15
index XXXXXXX..XXXXXXX 100644
27
index XXXXXXX..XXXXXXX 100644
16
--- a/tests/migration/guestperf/shell.py
28
--- a/docs/tools/qemu-img.rst
17
+++ b/tests/migration/guestperf/shell.py
29
+++ b/docs/tools/qemu-img.rst
18
@@ -XXX,XX +XXX,XX @@
30
@@ -XXX,XX +XXX,XX @@ Supported image file formats:
19
#
31
issue ``lsattr filename`` to check if the NOCOW flag is set or not
20
32
(Capital 'C' is NOCOW flag).
21
33
22
-import argparse
34
+ ``data_file``
23
-import fnmatch
35
+ Filename where all guest data will be stored. If this option is used,
24
import os
36
+ the qcow2 file will only contain the image's metadata.
25
import os.path
37
+
26
-import platform
38
+ Note: Data loss will occur if the given filename already exists when
27
import sys
39
+ using this option with ``qemu-img create`` since ``qemu-img`` will create
28
+sys.path.append(os.path.join(os.path.dirname(__file__),
40
+ the data file anew, overwriting the file's original contents. To simply
29
+ '..', '..', '..', 'scripts'))
41
+ update the reference to point to the given pre-existing file, use
30
+import argparse
42
+ ``qemu-img amend``.
31
+import fnmatch
43
+
32
+import platform
44
+ ``data_file_raw``
33
45
+ If this option is set to ``on``, QEMU will always keep the external data
34
from guestperf.hardware import Hardware
46
+ file consistent as a standalone read-only raw image.
35
from guestperf.engine import Engine
47
+
48
+ It does this by forwarding all write accesses to the qcow2 file through to
49
+ the raw data file, including their offsets. Therefore, data that is visible
50
+ on the qcow2 node (i.e., to the guest) at some offset is visible at the same
51
+ offset in the raw data file. This results in a read-only raw image. Writes
52
+ that bypass the qcow2 metadata may corrupt the qcow2 metadata because the
53
+ out-of-band writes may result in the metadata falling out of sync with the
54
+ raw image.
55
+
56
+ If this option is ``off``, QEMU will use the data file to store data in an
57
+ arbitrary manner. The file’s content will not make sense without the
58
+ accompanying qcow2 metadata. Where data is written will have no relation to
59
+ its offset as seen by the guest, and some writes (specifically zero writes)
60
+ may not be forwarded to the data file at all, but will only be handled by
61
+ modifying qcow2 metadata.
62
+
63
+ This option can only be enabled if ``data_file`` is set.
64
+
65
``Other``
66
67
QEMU also supports various other image file formats for
36
--
68
--
37
2.13.5
69
2.31.1
38
70
39
71
diff view generated by jsdifflib
[Qemu-devel] [PULL for-2.10 15/15] qcow2: allocate cluster_cache/cluster_data on demand
[PULL 11/19] block/copy-on-read: use bdrv_drop_filter() and drop s->active
1
Most qcow2 files are uncompressed so it is wasteful to allocate (32 + 1)
1
From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2
* cluster_size + 512 bytes upfront. Allocate s->cluster_cache and
3
s->cluster_data when the first read operation is performance on a
4
compressed cluster.
5
2
6
The buffers are freed in .bdrv_close(). .bdrv_open() no longer has any
3
Now, after huge update of block graph permission update algorithm, we
7
code paths that can allocate these buffers, so remove the free functions
4
don't need this workaround with active state of the filter. Drop it and
8
in the error code path.
5
use new smart bdrv_drop_filter() function.
9
6
10
This patch can result in significant memory savings when many qcow2
7
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
11
disks are attached or backing file chains are long:
8
Message-Id: <20210506194143.394141-1-vsementsov@virtuozzo.com>
9
Signed-off-by: Max Reitz <mreitz@redhat.com>
10
---
11
block/copy-on-read.c | 33 +--------------------------------
12
1 file changed, 1 insertion(+), 32 deletions(-)
12
13
13
Before 12.81% (1,023,193,088B)
14
diff --git a/block/copy-on-read.c b/block/copy-on-read.c
14
After 5.36% (393,893,888B)
15
16
Reported-by: Alexey Kardashevskiy <aik@ozlabs.ru>
17
Tested-by: Alexey Kardashevskiy <aik@ozlabs.ru>
18
Reviewed-by: Eric Blake <eblake@redhat.com>
19
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
20
Message-id: 20170821135530.32344-1-stefanha@redhat.com
21
Cc: Kevin Wolf <kwolf@redhat.com>
22
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
23
---
24
block/qcow2-cluster.c | 17 +++++++++++++++++
25
block/qcow2.c | 12 ------------
26
2 files changed, 17 insertions(+), 12 deletions(-)
27
28
diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
29
index XXXXXXX..XXXXXXX 100644
15
index XXXXXXX..XXXXXXX 100644
30
--- a/block/qcow2-cluster.c
16
--- a/block/copy-on-read.c
31
+++ b/block/qcow2-cluster.c
17
+++ b/block/copy-on-read.c
32
@@ -XXX,XX +XXX,XX @@ int qcow2_decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset)
18
@@ -XXX,XX +XXX,XX @@
33
nb_csectors = ((cluster_offset >> s->csize_shift) & s->csize_mask) + 1;
19
34
sector_offset = coffset & 511;
20
35
csize = nb_csectors * 512 - sector_offset;
21
typedef struct BDRVStateCOR {
36
+
22
- bool active;
37
+ /* Allocate buffers on first decompress operation, most images are
23
BlockDriverState *bottom_bs;
38
+ * uncompressed and the memory overhead can be avoided. The buffers
24
bool chain_frozen;
39
+ * are freed in .bdrv_close().
25
} BDRVStateCOR;
40
+ */
26
@@ -XXX,XX +XXX,XX @@ static int cor_open(BlockDriverState *bs, QDict *options, int flags,
41
+ if (!s->cluster_data) {
27
*/
42
+ /* one more sector for decompressed data alignment */
28
bdrv_ref(bottom_bs);
43
+ s->cluster_data = qemu_try_blockalign(bs->file->bs,
44
+ QCOW_MAX_CRYPT_CLUSTERS * s->cluster_size + 512);
45
+ if (!s->cluster_data) {
46
+ return -ENOMEM;
47
+ }
48
+ }
49
+ if (!s->cluster_cache) {
50
+ s->cluster_cache = g_malloc(s->cluster_size);
51
+ }
52
+
53
BLKDBG_EVENT(bs->file, BLKDBG_READ_COMPRESSED);
54
ret = bdrv_read(bs->file, coffset >> 9, s->cluster_data,
55
nb_csectors);
56
diff --git a/block/qcow2.c b/block/qcow2.c
57
index XXXXXXX..XXXXXXX 100644
58
--- a/block/qcow2.c
59
+++ b/block/qcow2.c
60
@@ -XXX,XX +XXX,XX @@ static int qcow2_do_open(BlockDriverState *bs, QDict *options, int flags,
61
goto fail;
62
}
29
}
63
30
- state->active = true;
64
- s->cluster_cache = g_malloc(s->cluster_size);
31
state->bottom_bs = bottom_bs;
65
- /* one more sector for decompressed data alignment */
32
66
- s->cluster_data = qemu_try_blockalign(bs->file->bs, QCOW_MAX_CRYPT_CLUSTERS
33
/*
67
- * s->cluster_size + 512);
34
@@ -XXX,XX +XXX,XX @@ static void cor_child_perm(BlockDriverState *bs, BdrvChild *c,
68
- if (s->cluster_data == NULL) {
35
uint64_t perm, uint64_t shared,
69
- error_setg(errp, "Could not allocate temporary cluster buffer");
36
uint64_t *nperm, uint64_t *nshared)
70
- ret = -ENOMEM;
37
{
71
- goto fail;
38
- BDRVStateCOR *s = bs->opaque;
39
-
40
- if (!s->active) {
41
- /*
42
- * While the filter is being removed
43
- */
44
- *nperm = 0;
45
- *nshared = BLK_PERM_ALL;
46
- return;
72
- }
47
- }
73
-
48
-
74
s->cluster_cache_offset = -1;
49
*nperm = perm & PERM_PASSTHROUGH;
75
s->flags = flags;
50
*nshared = (shared & PERM_PASSTHROUGH) | PERM_UNCHANGED;
76
51
77
@@ -XXX,XX +XXX,XX @@ static int qcow2_do_open(BlockDriverState *bs, QDict *options, int flags,
52
@@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_copy_on_read = {
78
if (s->refcount_block_cache) {
53
79
qcow2_cache_destroy(bs, s->refcount_block_cache);
54
void bdrv_cor_filter_drop(BlockDriverState *cor_filter_bs)
55
{
56
- BdrvChild *child;
57
- BlockDriverState *bs;
58
BDRVStateCOR *s = cor_filter_bs->opaque;
59
60
- child = bdrv_filter_child(cor_filter_bs);
61
- if (!child) {
62
- return;
63
- }
64
- bs = child->bs;
65
-
66
- /* Retain the BDS until we complete the graph change. */
67
- bdrv_ref(bs);
68
- /* Hold a guest back from writing while permissions are being reset. */
69
- bdrv_drained_begin(bs);
70
- /* Drop permissions before the graph change. */
71
- s->active = false;
72
/* unfreeze, as otherwise bdrv_replace_node() will fail */
73
if (s->chain_frozen) {
74
s->chain_frozen = false;
75
bdrv_unfreeze_backing_chain(cor_filter_bs, s->bottom_bs);
80
}
76
}
81
- g_free(s->cluster_cache);
77
- bdrv_child_refresh_perms(cor_filter_bs, child, &error_abort);
82
- qemu_vfree(s->cluster_data);
78
- bdrv_replace_node(cor_filter_bs, bs, &error_abort);
83
qcrypto_block_free(s->crypto);
79
-
84
qapi_free_QCryptoBlockOpenOptions(s->crypto_opts);
80
- bdrv_drained_end(bs);
85
return ret;
81
- bdrv_unref(bs);
82
+ bdrv_drop_filter(cor_filter_bs, &error_abort);
83
bdrv_unref(cor_filter_bs);
84
}
85
86
--
86
--
87
2.13.5
87
2.31.1
88
88
89
89
diff view generated by jsdifflib
[Qemu-devel] [PULL for-2.10 07/15] throttle: Make burst_length 64bit and add range checks
[PULL 12/19] qemu-iotests: fix pylint 2.8 consider-using-with error
1
From: Alberto Garcia <berto@igalia.com>
1
From: Emanuele Giuseppe Esposito <eesposit@redhat.com>
2
2
3
LeakyBucket.burst_length is defined as an unsigned integer but the
3
pylint 2.8 introduces consider-using-with error, suggesting
4
code never checks for overflows and it only makes sure that the value
4
to use the 'with' block statement when possible.
5
is not 0.
6
5
7
In practice this means that the user can set something like
6
Modify all subprocess.Popen call to use the 'with' statement,
8
throttling.iops-total-max-length=4294967300 despite being larger than
7
except the one in __init__ of QemuIoInteractive class, since
9
UINT_MAX and the final value after casting to unsigned int will be 4.
8
it is assigned to a class field and used in other methods.
10
9
11
This patch changes the data type to uint64_t. This does not increase
10
Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
12
the storage size of LeakyBucket, and allows us to assign the value
11
Message-Id: <20210510190449.65948-1-eesposit@redhat.com>
13
directly from qemu_opt_get_number() or BlockIOThrottle and then do the
12
[mreitz: Disable bad-option-value warning in the iotests' pylintrc, so
14
checks directly in throttle_is_valid().
13
that disabling consider-using-with in QemuIoInteractive will
14
not produce a warning in pre-2.8 pylint versions]
15
Signed-off-by: Max Reitz <mreitz@redhat.com>
16
---
17
tests/qemu-iotests/iotests.py | 65 ++++++++++++++++----------------
18
tests/qemu-iotests/pylintrc | 3 ++
19
tests/qemu-iotests/testrunner.py | 22 +++++------
20
3 files changed, 47 insertions(+), 43 deletions(-)
15
21
16
The value of burst_length does not have a specific upper limit,
22
diff --git a/tests/qemu-iotests/iotests.py b/tests/qemu-iotests/iotests.py
17
but since the bucket size is defined by max * burst_length we have
18
to prevent overflows. Instead of going for UINT64_MAX or something
19
similar this patch reuses THROTTLE_VALUE_MAX, which allows I/O bursts
20
of 1 GiB/s for 10 days in a row.
21
22
Signed-off-by: Alberto Garcia <berto@igalia.com>
23
Message-id: 1b2e3049803f71cafb2e1fa1be4fb47147a0d398.1503580370.git.berto@igalia.com
24
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
25
---
26
include/qemu/throttle.h | 2 +-
27
util/throttle.c | 5 +++++
28
2 files changed, 6 insertions(+), 1 deletion(-)
29
30
diff --git a/include/qemu/throttle.h b/include/qemu/throttle.h
31
index XXXXXXX..XXXXXXX 100644
23
index XXXXXXX..XXXXXXX 100644
32
--- a/include/qemu/throttle.h
24
--- a/tests/qemu-iotests/iotests.py
33
+++ b/include/qemu/throttle.h
25
+++ b/tests/qemu-iotests/iotests.py
34
@@ -XXX,XX +XXX,XX @@ typedef struct LeakyBucket {
26
@@ -XXX,XX +XXX,XX @@ def qemu_tool_pipe_and_status(tool: str, args: Sequence[str],
35
uint64_t max; /* leaky bucket max burst in units */
27
Run a tool and return both its output and its exit code
36
double level; /* bucket level in units */
28
"""
37
double burst_level; /* bucket level in units (for computing bursts) */
29
stderr = subprocess.STDOUT if connect_stderr else None
38
- unsigned burst_length; /* max length of the burst period, in seconds */
30
- subp = subprocess.Popen(args,
39
+ uint64_t burst_length; /* max length of the burst period, in seconds */
31
- stdout=subprocess.PIPE,
40
} LeakyBucket;
32
- stderr=stderr,
41
33
- universal_newlines=True)
42
/* The following structure is used to configure a ThrottleState
34
- output = subp.communicate()[0]
43
diff --git a/util/throttle.c b/util/throttle.c
35
- if subp.returncode < 0:
36
- cmd = ' '.join(args)
37
- sys.stderr.write(f'{tool} received signal {-subp.returncode}: {cmd}\n')
38
- return (output, subp.returncode)
39
+ with subprocess.Popen(args, stdout=subprocess.PIPE,
40
+ stderr=stderr, universal_newlines=True) as subp:
41
+ output = subp.communicate()[0]
42
+ if subp.returncode < 0:
43
+ cmd = ' '.join(args)
44
+ sys.stderr.write(f'{tool} received signal \
45
+ {-subp.returncode}: {cmd}\n')
46
+ return (output, subp.returncode)
47
48
def qemu_img_pipe_and_status(*args: str) -> Tuple[str, int]:
49
"""
50
@@ -XXX,XX +XXX,XX @@ def qemu_io_silent_check(*args):
51
class QemuIoInteractive:
52
def __init__(self, *args):
53
self.args = qemu_io_args_no_fmt + list(args)
54
+ # We need to keep the Popen objext around, and not
55
+ # close it immediately. Therefore, disable the pylint check:
56
+ # pylint: disable=consider-using-with
57
self._p = subprocess.Popen(self.args, stdin=subprocess.PIPE,
58
stdout=subprocess.PIPE,
59
stderr=subprocess.STDOUT,
60
@@ -XXX,XX +XXX,XX @@ def qemu_nbd_popen(*args):
61
cmd.extend(args)
62
63
log('Start NBD server')
64
- p = subprocess.Popen(cmd)
65
- try:
66
- while not os.path.exists(pid_file):
67
- if p.poll() is not None:
68
- raise RuntimeError(
69
- "qemu-nbd terminated with exit code {}: {}"
70
- .format(p.returncode, ' '.join(cmd)))
71
-
72
- time.sleep(0.01)
73
- yield
74
- finally:
75
- if os.path.exists(pid_file):
76
- os.remove(pid_file)
77
- log('Kill NBD server')
78
- p.kill()
79
- p.wait()
80
+ with subprocess.Popen(cmd) as p:
81
+ try:
82
+ while not os.path.exists(pid_file):
83
+ if p.poll() is not None:
84
+ raise RuntimeError(
85
+ "qemu-nbd terminated with exit code {}: {}"
86
+ .format(p.returncode, ' '.join(cmd)))
87
+
88
+ time.sleep(0.01)
89
+ yield
90
+ finally:
91
+ if os.path.exists(pid_file):
92
+ os.remove(pid_file)
93
+ log('Kill NBD server')
94
+ p.kill()
95
+ p.wait()
96
97
def compare_images(img1, img2, fmt1=imgfmt, fmt2=imgfmt):
98
'''Return True if two image files are identical'''
99
@@ -XXX,XX +XXX,XX @@ def compare_images(img1, img2, fmt1=imgfmt, fmt2=imgfmt):
100
101
def create_image(name, size):
102
'''Create a fully-allocated raw image with sector markers'''
103
- file = open(name, 'wb')
104
- i = 0
105
- while i < size:
106
- sector = struct.pack('>l504xl', i // 512, i // 512)
107
- file.write(sector)
108
- i = i + 512
109
- file.close()
110
+ with open(name, 'wb') as file:
111
+ i = 0
112
+ while i < size:
113
+ sector = struct.pack('>l504xl', i // 512, i // 512)
114
+ file.write(sector)
115
+ i = i + 512
116
117
def image_size(img):
118
'''Return image's virtual size'''
119
diff --git a/tests/qemu-iotests/pylintrc b/tests/qemu-iotests/pylintrc
44
index XXXXXXX..XXXXXXX 100644
120
index XXXXXXX..XXXXXXX 100644
45
--- a/util/throttle.c
121
--- a/tests/qemu-iotests/pylintrc
46
+++ b/util/throttle.c
122
+++ b/tests/qemu-iotests/pylintrc
47
@@ -XXX,XX +XXX,XX @@ bool throttle_is_valid(ThrottleConfig *cfg, Error **errp)
123
@@ -XXX,XX +XXX,XX @@ disable=invalid-name,
48
return false;
124
too-many-public-methods,
49
}
125
# pylint warns about Optional[] etc. as unsubscriptable in 3.9
50
126
unsubscriptable-object,
51
+ if (bkt->max && bkt->burst_length > THROTTLE_VALUE_MAX / bkt->max) {
127
+ # Sometimes we need to disable a newly introduced pylint warning.
52
+ error_setg(errp, "burst length too high for this burst rate");
128
+ # Doing so should not produce a warning in older versions of pylint.
53
+ return false;
129
+ bad-option-value,
54
+ }
130
# These are temporary, and should be removed:
55
+
131
missing-docstring,
56
if (bkt->max && !bkt->avg) {
132
too-many-return-statements,
57
error_setg(errp, "bps_max/iops_max require corresponding"
133
diff --git a/tests/qemu-iotests/testrunner.py b/tests/qemu-iotests/testrunner.py
58
" bps/iops values");
134
index XXXXXXX..XXXXXXX 100644
135
--- a/tests/qemu-iotests/testrunner.py
136
+++ b/tests/qemu-iotests/testrunner.py
137
@@ -XXX,XX +XXX,XX @@ def do_run_test(self, test: str) -> TestResult:
138
139
t0 = time.time()
140
with f_bad.open('w', encoding="utf-8") as f:
141
- proc = subprocess.Popen(args, cwd=str(f_test.parent), env=env,
142
- stdout=f, stderr=subprocess.STDOUT)
143
- try:
144
- proc.wait()
145
- except KeyboardInterrupt:
146
- proc.terminate()
147
- proc.wait()
148
- return TestResult(status='not run',
149
- description='Interrupted by user',
150
- interrupted=True)
151
- ret = proc.returncode
152
+ with subprocess.Popen(args, cwd=str(f_test.parent), env=env,
153
+ stdout=f, stderr=subprocess.STDOUT) as proc:
154
+ try:
155
+ proc.wait()
156
+ except KeyboardInterrupt:
157
+ proc.terminate()
158
+ proc.wait()
159
+ return TestResult(status='not run',
160
+ description='Interrupted by user',
161
+ interrupted=True)
162
+ ret = proc.returncode
163
164
elapsed = round(time.time() - t0, 1)
165
59
--
166
--
60
2.13.5
167
2.31.1
61
168
62
169
diff view generated by jsdifflib
[Qemu-devel] [PULL for-2.10 05/15] throttle: Remove throttle_fix_bucket() / throttle_unfix_bucket()
[PULL 13/19] block/write-threshold: don't use write notifiers
1
From: Alberto Garcia <berto@igalia.com>
1
From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2
2
3
The throttling code can change internally the value of bkt->max if it
3
write-notifiers are used only for write-threshold. New code for such
4
hasn't been set by the user. The problem with this is that if we want
4
purpose should create filters.
5
to retrieve the original value we have to undo this change first. This
6
is ugly and unnecessary: this patch removes the throttle_fix_bucket()
7
and throttle_unfix_bucket() functions completely and moves the logic
8
to throttle_compute_wait().
9
5
10
Signed-off-by: Alberto Garcia <berto@igalia.com>
6
Let's better special-case write-threshold and drop write notifiers at
11
Reviewed-by: Manos Pitsidianakis <el13635@mail.ntua.gr>
7
all. (Actually, write-threshold is special-cased anyway, as the only
12
Message-id: 5b0b9e1ac6eb208d709eddc7b09e7669a523bff3.1503580370.git.berto@igalia.com
8
user of write-notifiers)
13
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
9
10
So, create a new direct interface for bdrv_co_write_req_prepare() and
11
drop all write-notifier related logic from write-threshold.c.
12
13
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
14
Reviewed-by: Max Reitz <mreitz@redhat.com>
15
Message-Id: <20210506090621.11848-2-vsementsov@virtuozzo.com>
16
Reviewed-by: Eric Blake <eblake@redhat.com>
17
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
18
[mreitz: Adjusted comment as per Eric's suggestion]
19
Signed-off-by: Max Reitz <mreitz@redhat.com>
14
---
20
---
15
util/throttle.c | 62 +++++++++++++++++++++------------------------------------
21
include/block/block_int.h | 1 -
16
1 file changed, 23 insertions(+), 39 deletions(-)
22
include/block/write-threshold.h | 9 +++++
23
block/io.c | 5 ++-
24
block/write-threshold.c | 70 +++++++--------------------------
25
4 files changed, 27 insertions(+), 58 deletions(-)
17
26
18
diff --git a/util/throttle.c b/util/throttle.c
27
diff --git a/include/block/block_int.h b/include/block/block_int.h
19
index XXXXXXX..XXXXXXX 100644
28
index XXXXXXX..XXXXXXX 100644
20
--- a/util/throttle.c
29
--- a/include/block/block_int.h
21
+++ b/util/throttle.c
30
+++ b/include/block/block_int.h
22
@@ -XXX,XX +XXX,XX @@ static int64_t throttle_do_compute_wait(double limit, double extra)
31
@@ -XXX,XX +XXX,XX @@ struct BlockDriverState {
23
int64_t throttle_compute_wait(LeakyBucket *bkt)
32
24
{
33
/* threshold limit for writes, in bytes. "High water mark". */
25
double extra; /* the number of extra units blocking the io */
34
uint64_t write_threshold_offset;
26
+ double bucket_size; /* I/O before throttling to bkt->avg */
35
- NotifierWithReturn write_threshold_notifier;
27
+ double burst_bucket_size; /* Before throttling to bkt->max */
36
28
37
/* Writing to the list requires the BQL _and_ the dirty_bitmap_mutex.
29
if (!bkt->avg) {
38
* Reading from the list can be done with either the BQL or the
39
diff --git a/include/block/write-threshold.h b/include/block/write-threshold.h
40
index XXXXXXX..XXXXXXX 100644
41
--- a/include/block/write-threshold.h
42
+++ b/include/block/write-threshold.h
43
@@ -XXX,XX +XXX,XX @@ bool bdrv_write_threshold_is_set(const BlockDriverState *bs);
44
uint64_t bdrv_write_threshold_exceeded(const BlockDriverState *bs,
45
const BdrvTrackedRequest *req);
46
47
+/*
48
+ * bdrv_write_threshold_check_write
49
+ *
50
+ * Check whether the specified request exceeds the write threshold.
51
+ * If so, send a corresponding event and disable write threshold checking.
52
+ */
53
+void bdrv_write_threshold_check_write(BlockDriverState *bs, int64_t offset,
54
+ int64_t bytes);
55
+
56
#endif
57
diff --git a/block/io.c b/block/io.c
58
index XXXXXXX..XXXXXXX 100644
59
--- a/block/io.c
60
+++ b/block/io.c
61
@@ -XXX,XX +XXX,XX @@
62
#include "block/blockjob_int.h"
63
#include "block/block_int.h"
64
#include "block/coroutines.h"
65
+#include "block/write-threshold.h"
66
#include "qemu/cutils.h"
67
#include "qapi/error.h"
68
#include "qemu/error-report.h"
69
@@ -XXX,XX +XXX,XX @@ bdrv_co_write_req_prepare(BdrvChild *child, int64_t offset, int64_t bytes,
70
} else {
71
assert(child->perm & BLK_PERM_WRITE);
72
}
73
- return notifier_with_return_list_notify(&bs->before_write_notifiers,
74
- req);
75
+ bdrv_write_threshold_check_write(bs, offset, bytes);
76
+ return 0;
77
case BDRV_TRACKED_TRUNCATE:
78
assert(child->perm & BLK_PERM_RESIZE);
30
return 0;
79
return 0;
31
}
80
diff --git a/block/write-threshold.c b/block/write-threshold.c
32
81
index XXXXXXX..XXXXXXX 100644
33
- /* If the bucket is full then we have to wait */
82
--- a/block/write-threshold.c
34
- extra = bkt->level - bkt->max * bkt->burst_length;
83
+++ b/block/write-threshold.c
35
+ if (!bkt->max) {
84
@@ -XXX,XX +XXX,XX @@ bool bdrv_write_threshold_is_set(const BlockDriverState *bs)
36
+ /* If bkt->max is 0 we still want to allow short bursts of I/O
85
return bs->write_threshold_offset > 0;
37
+ * from the guest, otherwise every other request will be throttled
38
+ * and performance will suffer considerably. */
39
+ bucket_size = bkt->avg / 10;
40
+ burst_bucket_size = 0;
41
+ } else {
42
+ /* If we have a burst limit then we have to wait until all I/O
43
+ * at burst rate has finished before throttling to bkt->avg */
44
+ bucket_size = bkt->max * bkt->burst_length;
45
+ burst_bucket_size = bkt->max / 10;
46
+ }
47
+
48
+ /* If the main bucket is full then we have to wait */
49
+ extra = bkt->level - bucket_size;
50
if (extra > 0) {
51
return throttle_do_compute_wait(bkt->avg, extra);
52
}
53
54
- /* If the bucket is not full yet we have to make sure that we
55
- * fulfill the goal of bkt->max units per second. */
56
+ /* If the main bucket is not full yet we still have to check the
57
+ * burst bucket in order to enforce the burst limit */
58
if (bkt->burst_length > 1) {
59
- /* We use 1/10 of the max value to smooth the throttling.
60
- * See throttle_fix_bucket() for more details. */
61
- extra = bkt->burst_level - bkt->max / 10;
62
+ extra = bkt->burst_level - burst_bucket_size;
63
if (extra > 0) {
64
return throttle_do_compute_wait(bkt->max, extra);
65
}
66
@@ -XXX,XX +XXX,XX @@ bool throttle_is_valid(ThrottleConfig *cfg, Error **errp)
67
return true;
68
}
86
}
69
87
70
-/* fix bucket parameters */
88
-static void write_threshold_disable(BlockDriverState *bs)
71
-static void throttle_fix_bucket(LeakyBucket *bkt)
72
-{
89
-{
73
- double min;
90
- if (bdrv_write_threshold_is_set(bs)) {
74
-
91
- notifier_with_return_remove(&bs->write_threshold_notifier);
75
- /* zero bucket level */
92
- bs->write_threshold_offset = 0;
76
- bkt->level = bkt->burst_level = 0;
77
-
78
- /* If bkt->max is 0 we still want to allow short bursts of I/O
79
- * from the guest, otherwise every other request will be throttled
80
- * and performance will suffer considerably. */
81
- min = bkt->avg / 10;
82
- if (bkt->avg && !bkt->max) {
83
- bkt->max = min;
84
- }
93
- }
85
-}
94
-}
86
-
95
-
87
-/* undo internal bucket parameter changes (see throttle_fix_bucket()) */
96
uint64_t bdrv_write_threshold_exceeded(const BlockDriverState *bs,
88
-static void throttle_unfix_bucket(LeakyBucket *bkt)
97
const BdrvTrackedRequest *req)
98
{
99
@@ -XXX,XX +XXX,XX @@ uint64_t bdrv_write_threshold_exceeded(const BlockDriverState *bs,
100
return 0;
101
}
102
103
-static int coroutine_fn before_write_notify(NotifierWithReturn *notifier,
104
- void *opaque)
89
-{
105
-{
90
- if (bkt->max < bkt->avg) {
106
- BdrvTrackedRequest *req = opaque;
91
- bkt->max = 0;
107
- BlockDriverState *bs = req->bs;
108
- uint64_t amount = 0;
109
-
110
- amount = bdrv_write_threshold_exceeded(bs, req);
111
- if (amount > 0) {
112
- qapi_event_send_block_write_threshold(
113
- bs->node_name,
114
- amount,
115
- bs->write_threshold_offset);
116
-
117
- /* autodisable to avoid flooding the monitor */
118
- write_threshold_disable(bs);
92
- }
119
- }
120
-
121
- return 0; /* should always let other notifiers run */
93
-}
122
-}
94
-
123
-
95
/* Used to configure the throttle
124
-static void write_threshold_register_notifier(BlockDriverState *bs)
96
*
125
-{
97
* @ts: the throttle state we are working on
126
- bs->write_threshold_notifier.notify = before_write_notify;
98
@@ -XXX,XX +XXX,XX @@ void throttle_config(ThrottleState *ts,
127
- bdrv_add_before_write_notifier(bs, &bs->write_threshold_notifier);
99
128
-}
100
ts->cfg = *cfg;
129
-
101
130
-static void write_threshold_update(BlockDriverState *bs,
102
+ /* Zero bucket level */
131
- int64_t threshold_bytes)
103
for (i = 0; i < BUCKETS_COUNT; i++) {
132
-{
104
- throttle_fix_bucket(&ts->cfg.buckets[i]);
133
- bs->write_threshold_offset = threshold_bytes;
105
+ ts->cfg.buckets[i].level = 0;
134
-}
106
+ ts->cfg.buckets[i].burst_level = 0;
135
-
107
}
136
void bdrv_write_threshold_set(BlockDriverState *bs, uint64_t threshold_bytes)
108
109
ts->previous_leak = qemu_clock_get_ns(clock_type);
110
@@ -XXX,XX +XXX,XX @@ void throttle_config(ThrottleState *ts,
111
*/
112
void throttle_get_config(ThrottleState *ts, ThrottleConfig *cfg)
113
{
137
{
114
- int i;
138
- if (bdrv_write_threshold_is_set(bs)) {
115
-
139
- if (threshold_bytes > 0) {
116
*cfg = ts->cfg;
140
- write_threshold_update(bs, threshold_bytes);
117
-
141
- } else {
118
- for (i = 0; i < BUCKETS_COUNT; i++) {
142
- write_threshold_disable(bs);
119
- throttle_unfix_bucket(&cfg->buckets[i]);
143
- }
144
- } else {
145
- if (threshold_bytes > 0) {
146
- /* avoid multiple registration */
147
- write_threshold_register_notifier(bs);
148
- write_threshold_update(bs, threshold_bytes);
149
- }
150
- /* discard bogus disable request */
120
- }
151
- }
152
+ bs->write_threshold_offset = threshold_bytes;
121
}
153
}
122
154
123
155
void qmp_block_set_write_threshold(const char *node_name,
156
@@ -XXX,XX +XXX,XX @@ void qmp_block_set_write_threshold(const char *node_name,
157
158
aio_context_release(aio_context);
159
}
160
+
161
+void bdrv_write_threshold_check_write(BlockDriverState *bs, int64_t offset,
162
+ int64_t bytes)
163
+{
164
+ int64_t end = offset + bytes;
165
+ uint64_t wtr = bs->write_threshold_offset;
166
+
167
+ if (wtr > 0 && end > wtr) {
168
+ qapi_event_send_block_write_threshold(bs->node_name, end - wtr, wtr);
169
+
170
+ /* autodisable to avoid flooding the monitor */
171
+ bdrv_write_threshold_set(bs, 0);
172
+ }
173
+}
124
--
174
--
125
2.13.5
175
2.31.1
126
176
127
177
diff view generated by jsdifflib
[Qemu-devel] [PULL for-2.10 04/15] throttle: Make throttle_is_valid() a bit less verbose
[PULL 14/19] block: drop write notifiers
1
From: Alberto Garcia <berto@igalia.com>
1
From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2
2
3
Use a pointer to the bucket instead of repeating cfg->buckets[i] all
3
They are unused now.
4
the time. This makes the code more concise and will help us expand the
5
checks later and save a few line breaks.
6
4
7
Signed-off-by: Alberto Garcia <berto@igalia.com>
5
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
8
Message-id: 763ffc40a26b17d54cf93f5a999e4656049fcf0c.1503580370.git.berto@igalia.com
6
Reviewed-by: Max Reitz <mreitz@redhat.com>
9
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
7
Message-Id: <20210506090621.11848-3-vsementsov@virtuozzo.com>
8
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
9
Signed-off-by: Max Reitz <mreitz@redhat.com>
10
---
10
---
11
util/throttle.c | 15 +++++++--------
11
include/block/block_int.h | 12 ------------
12
1 file changed, 7 insertions(+), 8 deletions(-)
12
block.c | 1 -
13
block/io.c | 6 ------
14
3 files changed, 19 deletions(-)
13
15
14
diff --git a/util/throttle.c b/util/throttle.c
16
diff --git a/include/block/block_int.h b/include/block/block_int.h
15
index XXXXXXX..XXXXXXX 100644
17
index XXXXXXX..XXXXXXX 100644
16
--- a/util/throttle.c
18
--- a/include/block/block_int.h
17
+++ b/util/throttle.c
19
+++ b/include/block/block_int.h
18
@@ -XXX,XX +XXX,XX @@ bool throttle_is_valid(ThrottleConfig *cfg, Error **errp)
20
@@ -XXX,XX +XXX,XX @@ struct BlockDriverState {
21
*/
22
int64_t total_sectors;
23
24
- /* Callback before write request is processed */
25
- NotifierWithReturnList before_write_notifiers;
26
-
27
/* threshold limit for writes, in bytes. "High water mark". */
28
uint64_t write_threshold_offset;
29
30
@@ -XXX,XX +XXX,XX @@ void bdrv_parse_filename_strip_prefix(const char *filename, const char *prefix,
31
bool bdrv_backing_overridden(BlockDriverState *bs);
32
33
34
-/**
35
- * bdrv_add_before_write_notifier:
36
- *
37
- * Register a callback that is invoked before write requests are processed but
38
- * after any throttling or waiting for overlapping requests.
39
- */
40
-void bdrv_add_before_write_notifier(BlockDriverState *bs,
41
- NotifierWithReturn *notifier);
42
-
43
/**
44
* bdrv_add_aio_context_notifier:
45
*
46
diff --git a/block.c b/block.c
47
index XXXXXXX..XXXXXXX 100644
48
--- a/block.c
49
+++ b/block.c
50
@@ -XXX,XX +XXX,XX @@ BlockDriverState *bdrv_new(void)
51
for (i = 0; i < BLOCK_OP_TYPE_MAX; i++) {
52
QLIST_INIT(&bs->op_blockers[i]);
19
}
53
}
20
54
- notifier_with_return_list_init(&bs->before_write_notifiers);
21
for (i = 0; i < BUCKETS_COUNT; i++) {
55
qemu_co_mutex_init(&bs->reqs_lock);
22
- if (cfg->buckets[i].avg < 0 ||
56
qemu_mutex_init(&bs->dirty_bitmap_mutex);
23
- cfg->buckets[i].max < 0 ||
57
bs->refcnt = 1;
24
- cfg->buckets[i].avg > THROTTLE_VALUE_MAX ||
58
diff --git a/block/io.c b/block/io.c
25
- cfg->buckets[i].max > THROTTLE_VALUE_MAX) {
59
index XXXXXXX..XXXXXXX 100644
26
+ LeakyBucket *bkt = &cfg->buckets[i];
60
--- a/block/io.c
27
+ if (bkt->avg < 0 || bkt->max < 0 ||
61
+++ b/block/io.c
28
+ bkt->avg > THROTTLE_VALUE_MAX || bkt->max > THROTTLE_VALUE_MAX) {
62
@@ -XXX,XX +XXX,XX @@ bool bdrv_qiov_is_aligned(BlockDriverState *bs, QEMUIOVector *qiov)
29
error_setg(errp, "bps/iops/max values must be within [0, %lld]",
63
return true;
30
THROTTLE_VALUE_MAX);
64
}
31
return false;
65
32
}
66
-void bdrv_add_before_write_notifier(BlockDriverState *bs,
33
67
- NotifierWithReturn *notifier)
34
- if (!cfg->buckets[i].burst_length) {
68
-{
35
+ if (!bkt->burst_length) {
69
- notifier_with_return_list_add(&bs->before_write_notifiers, notifier);
36
error_setg(errp, "the burst length cannot be 0");
70
-}
37
return false;
71
-
38
}
72
void bdrv_io_plug(BlockDriverState *bs)
39
73
{
40
- if (cfg->buckets[i].burst_length > 1 && !cfg->buckets[i].max) {
74
BdrvChild *child;
41
+ if (bkt->burst_length > 1 && !bkt->max) {
42
error_setg(errp, "burst length set without burst rate");
43
return false;
44
}
45
46
- if (cfg->buckets[i].max && !cfg->buckets[i].avg) {
47
+ if (bkt->max && !bkt->avg) {
48
error_setg(errp, "bps_max/iops_max require corresponding"
49
" bps/iops values");
50
return false;
51
}
52
53
- if (cfg->buckets[i].max && cfg->buckets[i].max < cfg->buckets[i].avg) {
54
+ if (bkt->max && bkt->max < bkt->avg) {
55
error_setg(errp, "bps_max/iops_max cannot be lower than bps/iops");
56
return false;
57
}
58
--
75
--
59
2.13.5
76
2.31.1
60
77
61
78
diff view generated by jsdifflib
[Qemu-devel] [PULL for-2.10 10/15] misc: Remove unused Error variables
[PULL 15/19] test-write-threshold: rewrite test_threshold_(not_)trigger tests
1
From: Alberto Garcia <berto@igalia.com>
1
From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2
2
3
There's a few cases which we're passing an Error pointer to a function
3
These tests use bdrv_write_threshold_exceeded() API, which is used only
4
only to discard it immediately afterwards without checking it. In
4
for test (since pre-previous commit). Better is testing real API, which
5
these cases we can simply remove the variable and pass NULL instead.
5
is used in block.c as well.
6
6
7
Signed-off-by: Alberto Garcia <berto@igalia.com>
7
So, let's call bdrv_write_threshold_check_write(), and check is
8
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
8
bs->write_threshold_offset cleared or not (it's cleared iff threshold
9
Reviewed-by: Eric Blake <eblake@redhat.com>
9
triggered).
10
Message-id: 20170829120836.16091-1-berto@igalia.com
10
11
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
11
Also we get rid of BdrvTrackedRequest use here. Note, that paranoiac
12
bdrv_check_request() calls were added in 8b1170012b1 to protect
13
BdrvTrackedRequest. Drop them now.
14
15
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
16
Reviewed-by: Max Reitz <mreitz@redhat.com>
17
Message-Id: <20210506090621.11848-4-vsementsov@virtuozzo.com>
18
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
19
Signed-off-by: Max Reitz <mreitz@redhat.com>
12
---
20
---
13
block/qcow.c | 12 +++---------
21
tests/unit/test-write-threshold.c | 22 ++++------------------
14
block/qcow2.c | 8 ++------
22
1 file changed, 4 insertions(+), 18 deletions(-)
15
dump.c | 4 +---
16
3 files changed, 6 insertions(+), 18 deletions(-)
17
23
18
diff --git a/block/qcow.c b/block/qcow.c
24
diff --git a/tests/unit/test-write-threshold.c b/tests/unit/test-write-threshold.c
19
index XXXXXXX..XXXXXXX 100644
25
index XXXXXXX..XXXXXXX 100644
20
--- a/block/qcow.c
26
--- a/tests/unit/test-write-threshold.c
21
+++ b/block/qcow.c
27
+++ b/tests/unit/test-write-threshold.c
22
@@ -XXX,XX +XXX,XX @@ static uint64_t get_cluster_offset(BlockDriverState *bs,
28
@@ -XXX,XX +XXX,XX @@ static void test_threshold_multi_set_get(void)
23
start_sect = (offset & ~(s->cluster_size - 1)) >> 9;
29
24
for(i = 0; i < s->cluster_sectors; i++) {
30
static void test_threshold_not_trigger(void)
25
if (i < n_start || i >= n_end) {
31
{
26
- Error *err = NULL;
32
- uint64_t amount = 0;
27
memset(s->cluster_data, 0x00, 512);
33
uint64_t threshold = 4 * 1024 * 1024;
28
if (qcrypto_block_encrypt(s->crypto, start_sect + i,
34
BlockDriverState bs;
29
s->cluster_data,
35
- BdrvTrackedRequest req;
30
BDRV_SECTOR_SIZE,
36
31
- &err) < 0) {
37
memset(&bs, 0, sizeof(bs));
32
- error_free(err);
38
- memset(&req, 0, sizeof(req));
33
+ NULL) < 0) {
39
- req.offset = 1024;
34
errno = EIO;
40
- req.bytes = 1024;
35
return -1;
41
-
36
}
42
- bdrv_check_request(req.offset, req.bytes, &error_abort);
37
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_readv(BlockDriverState *bs, int64_t sector_num,
43
38
QEMUIOVector hd_qiov;
44
bdrv_write_threshold_set(&bs, threshold);
39
uint8_t *buf;
45
- amount = bdrv_write_threshold_exceeded(&bs, &req);
40
void *orig_buf;
46
- g_assert_cmpuint(amount, ==, 0);
41
- Error *err = NULL;
47
+ bdrv_write_threshold_check_write(&bs, 1024, 1024);
42
48
+ g_assert_cmpuint(bdrv_write_threshold_get(&bs), ==, threshold);
43
if (qiov->niov > 1) {
44
buf = orig_buf = qemu_try_blockalign(bs, qiov->size);
45
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_readv(BlockDriverState *bs, int64_t sector_num,
46
if (bs->encrypted) {
47
assert(s->crypto);
48
if (qcrypto_block_decrypt(s->crypto, sector_num, buf,
49
- n * BDRV_SECTOR_SIZE, &err) < 0) {
50
+ n * BDRV_SECTOR_SIZE, NULL) < 0) {
51
goto fail;
52
}
53
}
54
@@ -XXX,XX +XXX,XX @@ done:
55
return ret;
56
57
fail:
58
- error_free(err);
59
ret = -EIO;
60
goto done;
61
}
49
}
62
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow_co_writev(BlockDriverState *bs, int64_t sector_num,
50
63
break;
51
64
}
52
static void test_threshold_trigger(void)
65
if (bs->encrypted) {
66
- Error *err = NULL;
67
assert(s->crypto);
68
if (qcrypto_block_encrypt(s->crypto, sector_num, buf,
69
- n * BDRV_SECTOR_SIZE, &err) < 0) {
70
- error_free(err);
71
+ n * BDRV_SECTOR_SIZE, NULL) < 0) {
72
ret = -EIO;
73
break;
74
}
75
diff --git a/block/qcow2.c b/block/qcow2.c
76
index XXXXXXX..XXXXXXX 100644
77
--- a/block/qcow2.c
78
+++ b/block/qcow2.c
79
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv(BlockDriverState *bs, uint64_t offset,
80
assert(s->crypto);
81
assert((offset & (BDRV_SECTOR_SIZE - 1)) == 0);
82
assert((cur_bytes & (BDRV_SECTOR_SIZE - 1)) == 0);
83
- Error *err = NULL;
84
if (qcrypto_block_decrypt(s->crypto,
85
(s->crypt_physical_offset ?
86
cluster_offset + offset_in_cluster :
87
offset) >> BDRV_SECTOR_BITS,
88
cluster_data,
89
cur_bytes,
90
- &err) < 0) {
91
- error_free(err);
92
+ NULL) < 0) {
93
ret = -EIO;
94
goto fail;
95
}
96
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwritev(BlockDriverState *bs, uint64_t offset,
97
qemu_iovec_concat(&hd_qiov, qiov, bytes_done, cur_bytes);
98
99
if (bs->encrypted) {
100
- Error *err = NULL;
101
assert(s->crypto);
102
if (!cluster_data) {
103
cluster_data = qemu_try_blockalign(bs->file->bs,
104
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwritev(BlockDriverState *bs, uint64_t offset,
105
cluster_offset + offset_in_cluster :
106
offset) >> BDRV_SECTOR_BITS,
107
cluster_data,
108
- cur_bytes, &err) < 0) {
109
- error_free(err);
110
+ cur_bytes, NULL) < 0) {
111
ret = -EIO;
112
goto fail;
113
}
114
diff --git a/dump.c b/dump.c
115
index XXXXXXX..XXXXXXX 100644
116
--- a/dump.c
117
+++ b/dump.c
118
@@ -XXX,XX +XXX,XX @@ static void dump_process(DumpState *s, Error **errp)
119
120
static void *dump_thread(void *data)
121
{
53
{
122
- Error *err = NULL;
54
- uint64_t amount = 0;
123
DumpState *s = (DumpState *)data;
55
uint64_t threshold = 4 * 1024 * 1024;
124
- dump_process(s, &err);
56
BlockDriverState bs;
125
- error_free(err);
57
- BdrvTrackedRequest req;
126
+ dump_process(s, NULL);
58
127
return NULL;
59
memset(&bs, 0, sizeof(bs));
60
- memset(&req, 0, sizeof(req));
61
- req.offset = (4 * 1024 * 1024) - 1024;
62
- req.bytes = 2 * 1024;
63
-
64
- bdrv_check_request(req.offset, req.bytes, &error_abort);
65
66
bdrv_write_threshold_set(&bs, threshold);
67
- amount = bdrv_write_threshold_exceeded(&bs, &req);
68
- g_assert_cmpuint(amount, >=, 1024);
69
+ bdrv_write_threshold_check_write(&bs, threshold - 1024, 2 * 1024);
70
+ g_assert_cmpuint(bdrv_write_threshold_get(&bs), ==, 0);
128
}
71
}
129
72
73
typedef struct TestStruct {
130
--
74
--
131
2.13.5
75
2.31.1
132
76
133
77
diff view generated by jsdifflib
[Qemu-devel] [PULL for-2.10 03/15] throttle: Update the throttle_fix_bucket() documentation
[PULL 16/19] block/write-threshold: drop extra APIs
1
From: Alberto Garcia <berto@igalia.com>
1
From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2
2
3
The way the throttling algorithm works is that requests start being
3
bdrv_write_threshold_exceeded() is unused.
4
throttled once the bucket level exceeds the burst limit. When we get
5
there the bucket leaks at the level set by the user (bkt->avg), and
6
that leak rate is what prevents guest I/O from exceeding the desired
7
limit.
8
4
9
If we don't allow bursts (i.e. bkt->max == 0) then we can start
5
bdrv_write_threshold_is_set() is used only to double check the value of
10
throttling requests immediately. The problem with keeping the
6
bs->write_threshold_offset in tests. No real sense in it (both tests do
11
threshold at 0 is that it only allows one request at a time, and as
7
check real value with help of bdrv_write_threshold_get())
12
soon as there's a bit of I/O from the guest every other request will
13
be throttled and performance will suffer considerably. That can even
14
make the guest unable to reach the throttle limit if that limit is
15
high enough, and that happens regardless of the block scheduler used
16
by the guest.
17
8
18
Increasing that threshold gives flexibility to the guest, allowing it
9
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
19
to perform short bursts of I/O before being throttled. Increasing the
10
Reviewed-by: Max Reitz <mreitz@redhat.com>
20
threshold too much does not make a difference in the long run (because
11
Message-Id: <20210506090621.11848-5-vsementsov@virtuozzo.com>
21
it's the leak rate what defines the actual throughput) but it does
12
Reviewed-by: Eric Blake <eblake@redhat.com>
22
allow the guest to perform longer initial bursts and exceed the
13
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
23
throttle limit for a short while.
14
[mreitz: Adjusted commit message as per Eric's suggestion]
15
Signed-off-by: Max Reitz <mreitz@redhat.com>
16
---
17
include/block/write-threshold.h | 24 ------------------------
18
block/write-threshold.c | 19 -------------------
19
tests/unit/test-write-threshold.c | 4 ----
20
3 files changed, 47 deletions(-)
24
21
25
A burst value of bkt->avg / 10 allows the guest to perform 100ms'
22
diff --git a/include/block/write-threshold.h b/include/block/write-threshold.h
26
worth of I/O at the target rate without being throttled.
27
28
Signed-off-by: Alberto Garcia <berto@igalia.com>
29
Message-id: 31aae6645f0d1fbf3860fb2b528b757236f0c0a7.1503580370.git.berto@igalia.com
30
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
31
---
32
util/throttle.c | 11 +++--------
33
1 file changed, 3 insertions(+), 8 deletions(-)
34
35
diff --git a/util/throttle.c b/util/throttle.c
36
index XXXXXXX..XXXXXXX 100644
23
index XXXXXXX..XXXXXXX 100644
37
--- a/util/throttle.c
24
--- a/include/block/write-threshold.h
38
+++ b/util/throttle.c
25
+++ b/include/block/write-threshold.h
39
@@ -XXX,XX +XXX,XX @@ static void throttle_fix_bucket(LeakyBucket *bkt)
26
@@ -XXX,XX +XXX,XX @@ void bdrv_write_threshold_set(BlockDriverState *bs, uint64_t threshold_bytes);
40
/* zero bucket level */
27
*/
41
bkt->level = bkt->burst_level = 0;
28
uint64_t bdrv_write_threshold_get(const BlockDriverState *bs);
42
29
43
- /* The following is done to cope with the Linux CFQ block scheduler
30
-/*
44
- * which regroup reads and writes by block of 100ms in the guest.
31
- * bdrv_write_threshold_is_set
45
- * When they are two process one making reads and one making writes cfq
32
- *
46
- * make a pattern looking like the following:
33
- * Tell if a write threshold is set for a given BDS.
47
- * WWWWWWWWWWWRRRRRRRRRRRRRRWWWWWWWWWWWWWwRRRRRRRRRRRRRRRRR
34
- */
48
- * Having a max burst value of 100ms of the average will help smooth the
35
-bool bdrv_write_threshold_is_set(const BlockDriverState *bs);
49
- * throttling
36
-
50
- */
37
-/*
51
+ /* If bkt->max is 0 we still want to allow short bursts of I/O
38
- * bdrv_write_threshold_exceeded
52
+ * from the guest, otherwise every other request will be throttled
39
- *
53
+ * and performance will suffer considerably. */
40
- * Return the extent of a write request that exceeded the threshold,
54
min = bkt->avg / 10;
41
- * or zero if the request is below the threshold.
55
if (bkt->avg && !bkt->max) {
42
- * Return zero also if the threshold was not set.
56
bkt->max = min;
43
- *
44
- * NOTE: here we assume the following holds for each request this code
45
- * deals with:
46
- *
47
- * assert((req->offset + req->bytes) <= UINT64_MAX)
48
- *
49
- * Please not there is *not* an actual C assert().
50
- */
51
-uint64_t bdrv_write_threshold_exceeded(const BlockDriverState *bs,
52
- const BdrvTrackedRequest *req);
53
-
54
/*
55
* bdrv_write_threshold_check_write
56
*
57
diff --git a/block/write-threshold.c b/block/write-threshold.c
58
index XXXXXXX..XXXXXXX 100644
59
--- a/block/write-threshold.c
60
+++ b/block/write-threshold.c
61
@@ -XXX,XX +XXX,XX @@ uint64_t bdrv_write_threshold_get(const BlockDriverState *bs)
62
return bs->write_threshold_offset;
63
}
64
65
-bool bdrv_write_threshold_is_set(const BlockDriverState *bs)
66
-{
67
- return bs->write_threshold_offset > 0;
68
-}
69
-
70
-uint64_t bdrv_write_threshold_exceeded(const BlockDriverState *bs,
71
- const BdrvTrackedRequest *req)
72
-{
73
- if (bdrv_write_threshold_is_set(bs)) {
74
- if (req->offset > bs->write_threshold_offset) {
75
- return (req->offset - bs->write_threshold_offset) + req->bytes;
76
- }
77
- if ((req->offset + req->bytes) > bs->write_threshold_offset) {
78
- return (req->offset + req->bytes) - bs->write_threshold_offset;
79
- }
80
- }
81
- return 0;
82
-}
83
-
84
void bdrv_write_threshold_set(BlockDriverState *bs, uint64_t threshold_bytes)
85
{
86
bs->write_threshold_offset = threshold_bytes;
87
diff --git a/tests/unit/test-write-threshold.c b/tests/unit/test-write-threshold.c
88
index XXXXXXX..XXXXXXX 100644
89
--- a/tests/unit/test-write-threshold.c
90
+++ b/tests/unit/test-write-threshold.c
91
@@ -XXX,XX +XXX,XX @@ static void test_threshold_not_set_on_init(void)
92
BlockDriverState bs;
93
memset(&bs, 0, sizeof(bs));
94
95
- g_assert(!bdrv_write_threshold_is_set(&bs));
96
-
97
res = bdrv_write_threshold_get(&bs);
98
g_assert_cmpint(res, ==, 0);
99
}
100
@@ -XXX,XX +XXX,XX @@ static void test_threshold_set_get(void)
101
102
bdrv_write_threshold_set(&bs, threshold);
103
104
- g_assert(bdrv_write_threshold_is_set(&bs));
105
-
106
res = bdrv_write_threshold_get(&bs);
107
g_assert_cmpint(res, ==, threshold);
108
}
57
--
109
--
58
2.13.5
110
2.31.1
59
111
60
112
diff view generated by jsdifflib
[Qemu-devel] [PULL for-2.10 01/15] nvme: Fix get/set number of queues feature, again
[PULL 17/19] test-write-threshold: drop extra tests
1
From: Dan Aloni <dan@kernelim.com>
1
From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2
2
3
The number of queues that should be return by the admin command should:
3
Testing set/get of one 64bit variable doesn't seem necessary. We have a
4
lot of such variables. Also remaining tests do test set/get anyway.
4
5
5
1) Only mention the number of non-admin queues.
6
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
6
2) It is zero-based, meaning that '0 == one non-admin queue',
7
Reviewed-by: Max Reitz <mreitz@redhat.com>
7
'1 == two non-admin queues', and so forth.
8
Message-Id: <20210506090621.11848-7-vsementsov@virtuozzo.com>
9
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
10
Signed-off-by: Max Reitz <mreitz@redhat.com>
11
---
12
tests/unit/test-write-threshold.c | 43 -------------------------------
13
1 file changed, 43 deletions(-)
8
14
9
Because our `num_queues` means the number of queues _plus_ the admin
15
diff --git a/tests/unit/test-write-threshold.c b/tests/unit/test-write-threshold.c
10
queue, then the right calculation for the number returned from the admin
11
command is `num_queues - 2`, combining the two requirements mentioned.
12
13
The issue was discovered by reducing num_queues from 64 to 8 and running
14
a Linux VM with an SMP parameter larger than that (e.g. 22). It tries to
15
utilize all queues, and therefore fails with an invalid queue number
16
when trying to queue I/Os on the last queue.
17
18
Signed-off-by: Dan Aloni <dan@kernelim.com>
19
CC: Alex Friedman <alex@e8storage.com>
20
CC: Keith Busch <keith.busch@intel.com>
21
CC: Stefan Hajnoczi <stefanha@redhat.com>
22
Reviewed-by: Keith Busch <keith.busch@intel.com>
23
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
24
---
25
hw/block/nvme.c | 4 ++--
26
1 file changed, 2 insertions(+), 2 deletions(-)
27
28
diff --git a/hw/block/nvme.c b/hw/block/nvme.c
29
index XXXXXXX..XXXXXXX 100644
16
index XXXXXXX..XXXXXXX 100644
30
--- a/hw/block/nvme.c
17
--- a/tests/unit/test-write-threshold.c
31
+++ b/hw/block/nvme.c
18
+++ b/tests/unit/test-write-threshold.c
32
@@ -XXX,XX +XXX,XX @@ static uint16_t nvme_get_feature(NvmeCtrl *n, NvmeCmd *cmd, NvmeRequest *req)
19
@@ -XXX,XX +XXX,XX @@
33
result = blk_enable_write_cache(n->conf.blk);
20
#include "block/write-threshold.h"
34
break;
21
35
case NVME_NUMBER_OF_QUEUES:
22
36
- result = cpu_to_le32((n->num_queues - 1) | ((n->num_queues - 1) << 16));
23
-static void test_threshold_not_set_on_init(void)
37
+ result = cpu_to_le32((n->num_queues - 2) | ((n->num_queues - 2) << 16));
24
-{
38
break;
25
- uint64_t res;
39
default:
26
- BlockDriverState bs;
40
return NVME_INVALID_FIELD | NVME_DNR;
27
- memset(&bs, 0, sizeof(bs));
41
@@ -XXX,XX +XXX,XX @@ static uint16_t nvme_set_feature(NvmeCtrl *n, NvmeCmd *cmd, NvmeRequest *req)
28
-
42
break;
29
- res = bdrv_write_threshold_get(&bs);
43
case NVME_NUMBER_OF_QUEUES:
30
- g_assert_cmpint(res, ==, 0);
44
req->cqe.result =
31
-}
45
- cpu_to_le32((n->num_queues - 1) | ((n->num_queues - 1) << 16));
32
-
46
+ cpu_to_le32((n->num_queues - 2) | ((n->num_queues - 2) << 16));
33
-static void test_threshold_set_get(void)
47
break;
34
-{
48
default:
35
- uint64_t threshold = 4 * 1024 * 1024;
49
return NVME_INVALID_FIELD | NVME_DNR;
36
- uint64_t res;
37
- BlockDriverState bs;
38
- memset(&bs, 0, sizeof(bs));
39
-
40
- bdrv_write_threshold_set(&bs, threshold);
41
-
42
- res = bdrv_write_threshold_get(&bs);
43
- g_assert_cmpint(res, ==, threshold);
44
-}
45
-
46
-static void test_threshold_multi_set_get(void)
47
-{
48
- uint64_t threshold1 = 4 * 1024 * 1024;
49
- uint64_t threshold2 = 15 * 1024 * 1024;
50
- uint64_t res;
51
- BlockDriverState bs;
52
- memset(&bs, 0, sizeof(bs));
53
-
54
- bdrv_write_threshold_set(&bs, threshold1);
55
- bdrv_write_threshold_set(&bs, threshold2);
56
- res = bdrv_write_threshold_get(&bs);
57
- g_assert_cmpint(res, ==, threshold2);
58
-}
59
-
60
static void test_threshold_not_trigger(void)
61
{
62
uint64_t threshold = 4 * 1024 * 1024;
63
@@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv)
64
{
65
size_t i;
66
TestStruct tests[] = {
67
- { "/write-threshold/not-set-on-init",
68
- test_threshold_not_set_on_init },
69
- { "/write-threshold/set-get",
70
- test_threshold_set_get },
71
- { "/write-threshold/multi-set-get",
72
- test_threshold_multi_set_get },
73
{ "/write-threshold/not-trigger",
74
test_threshold_not_trigger },
75
{ "/write-threshold/trigger",
50
--
76
--
51
2.13.5
77
2.31.1
52
78
53
79
diff view generated by jsdifflib
[Qemu-devel] [PULL for-2.10 08/15] throttle: Test the valid range of config values
[PULL 18/19] test-write-threshold: drop extra TestStruct structure
1
From: Alberto Garcia <berto@igalia.com>
1
From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2
2
3
Signed-off-by: Alberto Garcia <berto@igalia.com>
3
We don't need this extra logic: it doesn't make code simpler.
4
Message-id: a57dd6274e1b6dc9c28769fec4c7ea543be5c5e3.1503580370.git.berto@igalia.com
4
5
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
6
Reviewed-by: Max Reitz <mreitz@redhat.com>
7
Message-Id: <20210506090621.11848-8-vsementsov@virtuozzo.com>
8
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
9
Signed-off-by: Max Reitz <mreitz@redhat.com>
6
---
10
---
7
tests/test-throttle.c | 77 +++++++++++++++++++++++++++++++++++++++++++++++++++
11
tests/unit/test-write-threshold.c | 20 +++-----------------
8
1 file changed, 77 insertions(+)
12
1 file changed, 3 insertions(+), 17 deletions(-)
9
13
10
diff --git a/tests/test-throttle.c b/tests/test-throttle.c
14
diff --git a/tests/unit/test-write-threshold.c b/tests/unit/test-write-threshold.c
11
index XXXXXXX..XXXXXXX 100644
15
index XXXXXXX..XXXXXXX 100644
12
--- a/tests/test-throttle.c
16
--- a/tests/unit/test-write-threshold.c
13
+++ b/tests/test-throttle.c
17
+++ b/tests/unit/test-write-threshold.c
14
@@ -XXX,XX +XXX,XX @@ static void test_is_valid(void)
18
@@ -XXX,XX +XXX,XX @@ static void test_threshold_trigger(void)
15
test_is_valid_for_value(1, true);
19
g_assert_cmpuint(bdrv_write_threshold_get(&bs), ==, 0);
16
}
20
}
17
21
18
+static void test_ranges(void)
22
-typedef struct TestStruct {
19
+{
23
- const char *name;
20
+ int i;
24
- void (*func)(void);
25
-} TestStruct;
26
-
27
28
int main(int argc, char **argv)
29
{
30
- size_t i;
31
- TestStruct tests[] = {
32
- { "/write-threshold/not-trigger",
33
- test_threshold_not_trigger },
34
- { "/write-threshold/trigger",
35
- test_threshold_trigger },
36
- { NULL, NULL }
37
- };
38
-
39
g_test_init(&argc, &argv, NULL);
40
- for (i = 0; tests[i].name != NULL; i++) {
41
- g_test_add_func(tests[i].name, tests[i].func);
42
- }
43
+ g_test_add_func("/write-threshold/not-trigger", test_threshold_not_trigger);
44
+ g_test_add_func("/write-threshold/trigger", test_threshold_trigger);
21
+
45
+
22
+ for (i = 0; i < BUCKETS_COUNT; i++) {
46
return g_test_run();
23
+ LeakyBucket *b = &cfg.buckets[i];
47
}
24
+ throttle_config_init(&cfg);
25
+
26
+ /* avg = 0 means throttling is disabled, but the config is valid */
27
+ b->avg = 0;
28
+ g_assert(throttle_is_valid(&cfg, NULL));
29
+ g_assert(!throttle_enabled(&cfg));
30
+
31
+ /* These are valid configurations (values <= THROTTLE_VALUE_MAX) */
32
+ b->avg = 1;
33
+ g_assert(throttle_is_valid(&cfg, NULL));
34
+
35
+ b->avg = THROTTLE_VALUE_MAX;
36
+ g_assert(throttle_is_valid(&cfg, NULL));
37
+
38
+ b->avg = THROTTLE_VALUE_MAX;
39
+ b->max = THROTTLE_VALUE_MAX;
40
+ g_assert(throttle_is_valid(&cfg, NULL));
41
+
42
+ /* Values over THROTTLE_VALUE_MAX are not allowed */
43
+ b->avg = THROTTLE_VALUE_MAX + 1;
44
+ g_assert(!throttle_is_valid(&cfg, NULL));
45
+
46
+ b->avg = THROTTLE_VALUE_MAX;
47
+ b->max = THROTTLE_VALUE_MAX + 1;
48
+ g_assert(!throttle_is_valid(&cfg, NULL));
49
+
50
+ /* burst_length must be between 1 and THROTTLE_VALUE_MAX */
51
+ b->avg = 1;
52
+ b->max = 1;
53
+ b->burst_length = 0;
54
+ g_assert(!throttle_is_valid(&cfg, NULL));
55
+
56
+ b->avg = 1;
57
+ b->max = 1;
58
+ b->burst_length = 1;
59
+ g_assert(throttle_is_valid(&cfg, NULL));
60
+
61
+ b->avg = 1;
62
+ b->max = 1;
63
+ b->burst_length = THROTTLE_VALUE_MAX;
64
+ g_assert(throttle_is_valid(&cfg, NULL));
65
+
66
+ b->avg = 1;
67
+ b->max = 1;
68
+ b->burst_length = THROTTLE_VALUE_MAX + 1;
69
+ g_assert(!throttle_is_valid(&cfg, NULL));
70
+
71
+ /* burst_length * max cannot exceed THROTTLE_VALUE_MAX */
72
+ b->avg = 1;
73
+ b->max = 2;
74
+ b->burst_length = THROTTLE_VALUE_MAX / 2;
75
+ g_assert(throttle_is_valid(&cfg, NULL));
76
+
77
+ b->avg = 1;
78
+ b->max = 3;
79
+ b->burst_length = THROTTLE_VALUE_MAX / 2;
80
+ g_assert(!throttle_is_valid(&cfg, NULL));
81
+
82
+ b->avg = 1;
83
+ b->max = THROTTLE_VALUE_MAX;
84
+ b->burst_length = 1;
85
+ g_assert(throttle_is_valid(&cfg, NULL));
86
+
87
+ b->avg = 1;
88
+ b->max = THROTTLE_VALUE_MAX;
89
+ b->burst_length = 2;
90
+ g_assert(!throttle_is_valid(&cfg, NULL));
91
+ }
92
+}
93
+
94
static void test_max_is_missing_limit(void)
95
{
96
int i;
97
@@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv)
98
g_test_add_func("/throttle/config/enabled", test_enabled);
99
g_test_add_func("/throttle/config/conflicting", test_conflicting_config);
100
g_test_add_func("/throttle/config/is_valid", test_is_valid);
101
+ g_test_add_func("/throttle/config/ranges", test_ranges);
102
g_test_add_func("/throttle/config/max", test_max_is_missing_limit);
103
g_test_add_func("/throttle/config/iops_size",
104
test_iops_size_is_missing_limit);
105
--
48
--
106
2.13.5
49
2.31.1
107
50
108
51
diff view generated by jsdifflib
[Qemu-devel] [PULL for-2.10 09/15] oslib-posix: Print errors before aborting on qemu_alloc_stack()
[PULL 19/19] write-threshold: deal with includes
1
From: Eduardo Habkost <ehabkost@redhat.com>
1
From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
2
2
3
If QEMU is running on a system that's out of memory and mmap()
3
"qemu/typedefs.h" is enough for include/block/write-threshold.h header
4
fails, QEMU aborts with no error message at all, making it hard
4
with forward declaration of BlockDriverState. Also drop extra includes
5
to debug the reason for the failure.
5
from block/write-threshold.c and tests/unit/test-write-threshold.c
6
6
7
Add perror() calls that will print error information before
7
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
8
aborting.
8
Message-Id: <20210506090621.11848-9-vsementsov@virtuozzo.com>
9
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
10
Signed-off-by: Max Reitz <mreitz@redhat.com>
11
---
12
include/block/write-threshold.h | 2 +-
13
block/write-threshold.c | 2 --
14
tests/unit/test-write-threshold.c | 1 -
15
3 files changed, 1 insertion(+), 4 deletions(-)
9
16
10
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
17
diff --git a/include/block/write-threshold.h b/include/block/write-threshold.h
11
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
12
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
13
Message-id: 20170829212053.6003-1-ehabkost@redhat.com
14
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
15
---
16
util/oslib-posix.c | 2 ++
17
1 file changed, 2 insertions(+)
18
19
diff --git a/util/oslib-posix.c b/util/oslib-posix.c
20
index XXXXXXX..XXXXXXX 100644
18
index XXXXXXX..XXXXXXX 100644
21
--- a/util/oslib-posix.c
19
--- a/include/block/write-threshold.h
22
+++ b/util/oslib-posix.c
20
+++ b/include/block/write-threshold.h
23
@@ -XXX,XX +XXX,XX @@ void *qemu_alloc_stack(size_t *sz)
21
@@ -XXX,XX +XXX,XX @@
24
ptr = mmap(NULL, *sz, PROT_READ | PROT_WRITE,
22
#ifndef BLOCK_WRITE_THRESHOLD_H
25
MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
23
#define BLOCK_WRITE_THRESHOLD_H
26
if (ptr == MAP_FAILED) {
24
27
+ perror("failed to allocate memory for stack");
25
-#include "block/block_int.h"
28
abort();
26
+#include "qemu/typedefs.h"
29
}
27
30
28
/*
31
@@ -XXX,XX +XXX,XX @@ void *qemu_alloc_stack(size_t *sz)
29
* bdrv_write_threshold_set:
32
guardpage = ptr;
30
diff --git a/block/write-threshold.c b/block/write-threshold.c
33
#endif
31
index XXXXXXX..XXXXXXX 100644
34
if (mprotect(guardpage, pagesz, PROT_NONE) != 0) {
32
--- a/block/write-threshold.c
35
+ perror("failed to set up stack guard page");
33
+++ b/block/write-threshold.c
36
abort();
34
@@ -XXX,XX +XXX,XX @@
37
}
35
36
#include "qemu/osdep.h"
37
#include "block/block_int.h"
38
-#include "qemu/coroutine.h"
39
#include "block/write-threshold.h"
40
-#include "qemu/notify.h"
41
#include "qapi/error.h"
42
#include "qapi/qapi-commands-block-core.h"
43
#include "qapi/qapi-events-block-core.h"
44
diff --git a/tests/unit/test-write-threshold.c b/tests/unit/test-write-threshold.c
45
index XXXXXXX..XXXXXXX 100644
46
--- a/tests/unit/test-write-threshold.c
47
+++ b/tests/unit/test-write-threshold.c
48
@@ -XXX,XX +XXX,XX @@
49
*/
50
51
#include "qemu/osdep.h"
52
-#include "qapi/error.h"
53
#include "block/block_int.h"
54
#include "block/write-threshold.h"
38
55
39
--
56
--
40
2.13.5
57
2.31.1
41
58
42
59
diff view generated by jsdifflib

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.