1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
|
#!/bin/sh
# Use Failmalloc to test behaviour in the face of out-of-memory conditions.
# The test runs a binary multiple times while configuring Failmalloc to fail a
# different malloc() call each time, while looking for abnormal program exits
# due to segfaults. See https://www.nongnu.org/failmalloc/
#
# Ideally, it would ensure that the test binary returns an error code on each
# failure, but this often doesn't happen. This is a problem that should be
# rectified, but the API doesn't allow returning an error code in many
# functions that could encounter a problem. The issue could be solved in more
# cases with more judicious use of log calls with EXIF_LOG_CODE_NO_MEMORY
# codes.
#
# Copyright (C) 2018-2021 Dan Fandrich <dan@coneharvesters.com>, et. al.
# SPDX-License-Identifier: LGPL-2.0-or-later
srcdir="${srcdir:-.}"
VERBOSE=
if [ "$1" = "-v" ] ; then
VERBOSE=1
fi
if [ x"$FAILMALLOC_PATH" = x ]; then
echo "libfailmalloc is not available"
exit 77
fi
BINARY_PREFIX=./
if [ -e .libs/lt-test-value ]; then
# If libtool is in use, the normal "binary" is actually a shell script which
# would be interfered with by libfailmalloc. Instead, use the special lt-
# binary which should work properly.
BINARY_PREFIX=".libs/lt-"
fi
# Usage: failmalloc_binary_test #iterations binary <optional arguments>
# FIXME: auto-determine #iterations by comparing the output of each run
# with the output of a normal run, and exiting when that happens.
failmalloc_binary_test () {
binary="$BINARY_PREFIX$2"
iterations="$1"
shift
shift
echo Checking "$binary" for "$iterations" iterations
for n in $(seq "$iterations"); do
test "$VERBOSE" = 1 && { echo "$n"; set -x; }
FAILMALLOC_INTERVAL="$n" LD_PRELOAD="${LD_PRELOAD:+$LD_PRELOAD:}$FAILMALLOC_PATH" "$binary" "$@" >/dev/null
s=$?
test "$VERBOSE" = 1 && set +x;
if test "$s" -ge 125; then
# Such status codes only happen due to termination due to a signal
# like SIGSEGV or ASAN errors (ignoring a couple that the shell
# itself produces).
echo "Abnormal binary exit status $s at malloc #$n on $binary"
echo FAILURE
exit 1
fi
done
}
# Make ASAN errors return a high number to differentiate them from regular test
# errors (which are ignored). This only does something if ASAN was configured
# in the build.
export ASAN_OPTIONS="exitcode=125${ASAN_OPTIONS:+:$ASAN_OPTIONS}"
# The number of iterations is determined empirically to be about twice as
# high as the maximum number of mallocs performed by the test program in order
# to avoid lowering code coverage in the case of future code changes that cause
# more allocations.
failmalloc_binary_test 700 test-mem
failmalloc_binary_test 500 test-value
for f in ${srcdir}/testdata/*jpg; do
echo "Testing `basename "$f"`"
failmalloc_binary_test 600 test-parse$EXEEXT "$f"
done
# N.B. adding the following binaries doesn't actually increase code coverage:
# test-extract -o /dev/null
# test-gps
# test-mnote
# test-parse --swap-byte-order
echo PASSED
|
