1 _______________________
2 BUILDING THE FSQA SUITE
3 _______________________
6 - cd into the xfstests directory
11 - cd into the xfstests directory
12 - set the ROOT and TOOLROOT env variables for IRIX appropriately
15 ______________________
17 ______________________
19 Preparing system for tests (IRIX and Linux):
21 - compile XFS into your kernel or load XFS modules
22 - install user tools including mkfs.xfs, xfs_db & xfs_bmap
23 - If you wish to run the udf components of the suite install
24 mkfs_udf and udf_db for IRIX and mkudffs for Linux. Also download and
25 build the Philips UDF Verification Software from
26 http://www.extra.research.philips.com/udf/, then copy the udf_test
27 binary to xfstests/src/. If you wish to disable UDF verification test
28 set the environment variable DISABLE_UDF_TEST to 1.
31 - create one or two partitions to use for testing
33 - format as XFS, mount & optionally populate with
35 - one SCRATCH partition (optional)
36 - leave empty and expect this partition to be clobbered
37 by some tests. If this is not provided, many tests will
39 (SCRATCH and TEST must be two DIFFERENT partitions)
41 - for btrfs only: some btrfs test cases will need 3 or more independent
42 SCRATCH disks which should be set using SCRATCH_DEV_POOL (for eg:
43 SCRATCH_DEV_POOL="/dev/sda /dev/sdb /dev/sdc") with which
44 SCRATCH_DEV should be unused by the tester, and for the legacy
45 support SCRATCH_DEV will be set to the first disk of the
46 SCRATCH_DEV_POOL by xfstests script.
48 - setup your environment
49 - setenv TEST_DEV "device containing TEST PARTITION"
50 - setenv TEST_DIR "mount point of TEST PARTITION"
52 - setenv SCRATCH_DEV "device containing SCRATCH PARTITION" OR
53 (btrfs only) setenv SCRATCH_DEV_POOL "to 3 or more SCRATCH disks for
54 testing btrfs raid concepts"
55 - setenv SCRATCH_MNT "mount point for SCRATCH PARTITION"
56 - setenv TAPE_DEV "tape device for testing xfsdump"
57 - setenv RMT_TAPE_DEV "remote tape device for testing xfsdump"
58 - setenv RMT_IRIXTAPE_DEV "remote IRIX tape device for testing xfsdump"
59 - setenv SCRATCH_LOGDEV "device for scratch-fs external log"
60 - setenv SCRATCH_RTDEV "device for scratch-fs realtime data"
61 - setenv TEST_LOGDEV "device for test-fs external log"
62 - setenv TEST_RTDEV "device for test-fs realtime data"
63 - if TEST_LOGDEV and/or TEST_RTDEV, these will always be used.
64 - if SCRATCH_LOGDEV and/or SCRATCH_RTDEV, the USE_EXTERNAL
65 environment variable set to "yes" will enable their use.
66 - setenv DIFF_LENGTH "number of diff lines to print from a failed test",
67 by default 10, set to 0 to print the full diff
68 - setenv FSTYP "the filesystem you want to test", the filesystem
69 type is devised from the TEST_DEV device, but you may want to
70 override it; if unset, the default is 'xfs'
71 - setenv FSSTRESS_AVOID and/or FSX_AVOID, which contain options
72 added to the end of fsstresss and fsx invocations, respectively,
73 in case you wish to exclude certain operational modes from these
76 - or add a case to the switch in common/config assigning
77 these variables based on the hostname of your test
79 - or add these variables to a file called local.config and keep that
80 file in your workarea.
82 - if testing xfsdump, make sure the tape devices have a
83 tape which can be overwritten.
85 - make sure $TEST_DEV is a mounted XFS partition
86 - make sure that $SCRATCH_DEV or $SCRATCH_DEV_POOL contains nothing useful
91 - By default the tests suite will run xfs tests:
92 - ./check '*/001' '*/002' '*/003'
94 - You can explicitly specify NFS, otherwise the filesystem type will be
95 autodetected from $TEST_DEV:
96 ./check -nfs [test(s)]
97 - Groups of tests maybe ran by: ./check -g [group(s)]
98 See the 'group' file for details on groups
99 - for udf tests: ./check -udf [test(s)]
100 Running all the udf tests: ./check -udf -g udf
101 - for running nfs tests: ./check -nfs [test(s)]
102 - To randomize test order: ./check -r [test(s)]
105 The check script tests the return value of each script, and
106 compares the output against the expected output. If the output
107 is not as expected, a diff will be output and an .out.bad file
108 will be produced for the failing test.
110 Unexpected console messages, crashes and hangs may be considered
111 to be failures but are not necessarily detected by the QA system.
113 __________________________
114 ADDING TO THE FSQA SUITE
115 __________________________
118 Creating new tests scripts:
120 Use the "new" script.
122 Test script environment:
124 When developing a new test script keep the following things in
125 mind. All of the environment variables and shell procedures are
126 available to the script once the "common/rc" file has been
129 1. The tests are run from an arbitrary directory. If you want to
130 do operations on an XFS filesystem (good idea, eh?), then do
131 one of the following:
133 (a) Create directories and files at will in the directory
134 $TEST_DIR ... this is within an XFS filesystem and world
135 writeable. You should cleanup when your test is done,
136 e.g. use a _cleanup shell procedure in the trap ... see
137 001 for an example. If you need to know, the $TEST_DIR
138 directory is within the filesystem on the block device
141 (b) mkfs a new XFS filesystem on $SCRATCH_DEV, and mount this
142 on $SCRATCH_MNT. Call the the _require_scratch function
143 on startup if you require use of the scratch partition.
144 _require_scratch does some checks on $SCRATCH_DEV &
145 $SCRATCH_MNT and makes sure they're unmounted. You should
146 cleanup when your test is done, and in particular unmount
148 Tests can make use of $SCRATCH_LOGDEV and $SCRATCH_RTDEV
149 for testing external log and realtime volumes - however,
150 these tests need to simply "pass" (e.g. cat $seq.out; exit
151 - or default to an internal log) in the common case where
152 these variables are not set.
154 2. You can safely create temporary files that are not part of the
155 filesystem tests (e.g. to catch output, prepare lists of things
156 to do, etc.) in files named $tmp.<anything>. The standard test
157 script framework created by "new" will initialize $tmp and
160 3. By default, tests are run as the same uid as the person
161 executing the control script "check" that runs the test scripts.
163 If you need to be root, add a call to the shell procedure
164 _need_to_be_root ... this will do nothing or exit with an
165 error message depending on your current uid.
167 4. Some other useful shell procedures:
169 _get_fqdn - echo the host's fully qualified
172 _get_pids_by_name - one argument is a process name, and
173 return all of the matching pids on
176 _within_tolerance - fancy numerical "close enough is good
177 enough" filter for deterministic
178 output ... see comments in
179 common/filter for an explanation
181 _filter_date - turn ctime(3) format dates into the
182 string DATE for deterministic
185 _cat_passwd, - dump the content of the password
186 _cat_group or group file (both the local file
187 and the content of the NIS database
188 if it is likely to be present)
190 5. General recommendations, usage conventions, etc.:
191 - When the content of the password or group file is
192 required, get it using the _cat_passwd and _cat_group
193 functions, to ensure NIS information is included if NIS
195 - When calling getfacl in a test, pass the "-n" argument so
196 that numeric rather than symbolic identifiers are used in
201 Each test script has a numerical name, e.g. 007, and an associated
202 verified output, e.g. 007.out.
204 It is important that the verified output is deterministic, and
205 part of the job of the test script is to filter the output to
206 make this so. Examples of the sort of things that need filtering:
213 - variable directory contents
214 - imprecise numbers, especially sizes and times
218 The script "check" may be used to run one or more tests.
220 Test number $seq is deemed to "pass" when:
221 (a) no "core" file is created,
222 (b) the file $seq.notrun is not created,
223 (c) the exit status is 0, and
224 (d) the output matches the verified output.
226 In the "not run" case (b), the $seq.notrun file should contain a
227 short one-line summary of why the test was not run. The standard
228 output is not checked, so this can be used for a more verbose
229 explanation and to provide feedback when the QA test is run
233 To force a non-zero exit status use:
239 won't have the desired effect because of the way the exit trap
242 The recent pass/fail history is maintained in the file "check.log".
243 The elapsed time for the most recent pass for each test is kept