1 _______________________
2 BUILDING THE FSQA SUITE
3 _______________________
6 - cd into the xfstests directory and run make.
9 - cd into the xfstests directory
10 - set the ROOT and TOOLROOT env variables for IRIX appropriately
13 ______________________
15 ______________________
17 Preparing system for tests (IRIX and Linux):
19 - compile XFS into your kernel or load XFS modules
20 - install user tools including mkfs.xfs, xfs_db & xfs_bmap
21 - If you wish to run the udf components of the suite install
22 mkfs_udf and udf_db for IRIX and mkudffs for Linux. Also download and
23 build the Philips UDF Verification Software from
24 http://www.extra.research.philips.com/udf/, then copy the udf_test
25 binary to xfstests/src/. If you wish to disable UDF verification test
26 set the environment variable DISABLE_UDF_TEST to 1.
29 - create one or two partitions to use for testing
31 - format as XFS, mount & optionally populate with
33 - one SCRATCH partition (optional)
34 - leave empty and expect this partition to be clobbered
35 by some tests. If this is not provided, many tests will
37 (SCRATCH and TEST must be two DIFFERENT partitions)
39 - for btrfs only: some btrfs test cases will need 3 or more independent
40 SCRATCH disks which should be set using SCRATCH_DEV_POOL (for eg:
41 SCRATCH_DEV_POOL="/dev/sda /dev/sdb /dev/sdc") with which
42 SCRATCH_DEV should be unused by the tester, and for the legacy
43 support SCRATCH_DEV will be set to the first disk of the
44 SCRATCH_DEV_POOL by xfstests script.
46 - setup your environment
47 - setenv TEST_DEV "device containing TEST PARTITION"
48 - setenv TEST_DIR "mount point of TEST PARTITION"
50 - setenv SCRATCH_DEV "device containing SCRATCH PARTITION" OR
51 (btrfs only) setenv SCRATCH_DEV_POOL "to 3 or more SCRATCH disks for
52 testing btrfs raid concepts"
53 - setenv SCRATCH_MNT "mount point for SCRATCH PARTITION"
54 - setenv TAPE_DEV "tape device for testing xfsdump"
55 - setenv RMT_TAPE_DEV "remote tape device for testing xfsdump"
56 - setenv RMT_IRIXTAPE_DEV "remote IRIX tape device for testing xfsdump"
57 - setenv SCRATCH_LOGDEV "device for scratch-fs external log"
58 - setenv SCRATCH_RTDEV "device for scratch-fs realtime data"
59 - setenv TEST_LOGDEV "device for test-fs external log"
60 - setenv TEST_RTDEV "device for test-fs realtime data"
61 - if TEST_LOGDEV and/or TEST_RTDEV, these will always be used.
62 - if SCRATCH_LOGDEV and/or SCRATCH_RTDEV, the USE_EXTERNAL
63 environment variable set to "yes" will enable their use.
64 - or add a case to the switch in common.config assigning
65 these variables based on the hostname of your test
67 - or add these variables to a file called local.config and keep that
68 file in your workarea.
70 - if testing xfsdump, make sure the tape devices have a
71 tape which can be overwritten.
73 - make sure $TEST_DEV is a mounted XFS partition
74 - make sure that $SCRATCH_DEV or $SCRATCH_DEV_POOL contains nothing useful
79 - By default the tests suite will run xfs tests:
80 - ./check 001 002 003 ... or you can explicitly run a filesystem:
81 ./check -xfs [test(s)]
82 - You can run a range of tests: ./check 067-078
83 - Groups of tests maybe ran by: ./check -g [group(s)]
84 See the 'group' file for details on groups
85 - for udf tests: ./check -udf [test(s)]
86 Running all the udf tests: ./check -udf -g udf
87 - for running nfs tests: ./check -nfs [test(s)]
88 - To randomize test order: ./check -r [test(s)]
91 The check script tests the return value of each script, and
92 compares the output against the expected output. If the output
93 is not as expected, a diff will be output and an .out.bad file
94 will be produced for the failing test.
96 Unexpected console messages, crashes and hangs may be considered
97 to be failures but are not necessarily detected by the QA system.
99 __________________________
100 ADDING TO THE FSQA SUITE
101 __________________________
104 Creating new tests scripts:
106 Use the "new" script.
108 Test script environment:
110 When developing a new test script keep the following things in
111 mind. All of the environment variables and shell procedures are
112 available to the script once the "common.rc" file has been
115 1. The tests are run from an arbitrary directory. If you want to
116 do operations on an XFS filesystem (good idea, eh?), then do
117 one of the following:
119 (a) Create directories and files at will in the directory
120 $TEST_DIR ... this is within an XFS filesystem and world
121 writeable. You should cleanup when your test is done,
122 e.g. use a _cleanup shell procedure in the trap ... see
123 001 for an example. If you need to know, the $TEST_DIR
124 directory is within the filesystem on the block device
127 (b) mkfs a new XFS filesystem on $SCRATCH_DEV, and mount this
128 on $SCRATCH_MNT. Call the the _require_scratch function
129 on startup if you require use of the scratch partition.
130 _require_scratch does some checks on $SCRATCH_DEV &
131 $SCRATCH_MNT and makes sure they're unmounted. You should
132 cleanup when your test is done, and in particular unmount
134 Tests can make use of $SCRATCH_LOGDEV and $SCRATCH_RTDEV
135 for testing external log and realtime volumes - however,
136 these tests need to simply "pass" (e.g. cat $seq.out; exit
137 - or default to an internal log) in the common case where
138 these variables are not set.
140 2. You can safely create temporary files that are not part of the
141 filesystem tests (e.g. to catch output, prepare lists of things
142 to do, etc.) in files named $tmp.<anything>. The standard test
143 script framework created by "new" will initialize $tmp and
146 3. By default, tests are run as the same uid as the person
147 executing the control script "check" that runs the test scripts.
149 If you need to be root, add a call to the shell procedure
150 _need_to_be_root ... this will do nothing or exit with an
151 error message depending on your current uid.
153 4. Some other useful shell procedures:
155 _get_fqdn - echo the host's fully qualified
158 _get_pids_by_name - one argument is a process name, and
159 return all of the matching pids on
162 _within_tolerance - fancy numerical "close enough is good
163 enough" filter for deterministic
164 output ... see comments in
165 common.filter for an explanation
167 _filter_date - turn ctime(3) format dates into the
168 string DATE for deterministic
171 _cat_passwd, - dump the content of the password
172 _cat_group or group file (both the local file
173 and the content of the NIS database
174 if it is likely to be present)
176 4. General recommendations, usage conventions, etc.:
177 - When the content of the password or group file is
178 required, get it using the _cat_passwd and _cat_group
179 functions, to ensure NIS information is included if NIS
181 - When calling getfacl in a test, pass the "-n" argument so
182 that numeric rather than symbolic identifiers are used in
187 Each test script has a numerical name, e.g. 007, and an associated
188 verified output, e.g. 007.out.
190 It is important that the verified output is deterministic, and
191 part of the job of the test script is to filter the output to
192 make this so. Examples of the sort of things that need filtering:
199 - variable directory contents
200 - imprecise numbers, especially sizes and times
202 Use the "remake" script to recreate the verified output for one
207 The script "check" may be used to run one or more tests.
209 Test number $seq is deemed to "pass" when:
210 (a) no "core" file is created,
211 (b) the file $seq.notrun is not created,
212 (c) the exit status is 0, and
213 (d) the output matches the verified output.
215 In the "not run" case (b), the $seq.notrun file should contain a
216 short one-line summary of why the test was not run. The standard
217 output is not checked, so this can be used for a more verbose
218 explanation and to provide feedback when the QA test is run
222 To force a non-zero exit status use:
228 won't have the desired effect because of the way the exit trap
231 The recent pass/fail history is maintained in the file "check.log".
232 The elapsed time for the most recent pass for each test is kept