###############################################################################
#use strict;
-use 5.010;
-use feature 'state'; # included in 5.010
+use v5.10.1;
use warnings;
+use if $^V >= v5.19.11, experimental => 'smartmatch';
use Errno;
use FileHandle;
use Time::Local;
use Cwd;
use File::Basename;
+use Pod::Usage;
+use Getopt::Long;
use FindBin qw'$RealBin';
use lib "$RealBin/lib";
use Exim::Runtest;
+use Exim::Utils qw(uniq numerically);
-use if $ENV{DEBUG} && $ENV{DEBUG} =~ /\bruntest\b/ => ('Smart::Comments' => '####');
+use if $ENV{DEBUG} && scalar($ENV{DEBUG} =~ /\bruntest\b/) => 'Smart::Comments' => '####';
+use if $ENV{DEBUG} && scalar($ENV{DEBUG} =~ /\bruntest\b/) => 'Data::Dumper';
+
+use constant TEST_TOP => 8999;
+use constant TEST_SPECIAL_TOP => 9999;
# Start by initializing some global variables
my $optargs = '';
my $save_output = 0;
my $server_opts = '';
+my $slow = 0;
my $valgrind = 0;
my $have_ipv4 = 1;
my $have_ipv6 = 1;
my $have_largefiles = 0;
-my $test_start = 1;
-my $test_end = $test_top = 8999;
-my $test_special_top = 9999;
my @test_list = ();
-my @test_dirs = ();
# Networks to use for DNS tests. We need to choose some networks that will
# numbers, or handle specific bad conditions in different ways, leading to
# different wording in the error messages, so we cannot compare them.
- s/(TLS error on connection (?:from .* )?\(SSL_\w+\): error:)(.*)/$1 <<detail omitted>>/;
+#XXX This loses any trailing "deliving unencypted to" which is unfortunate
+# but I can't work out how to deal with that.
+ s/(TLS session: \(SSL_\w+\): error:)(.*)(?!: delivering)/$1 <<detail omitted>>/;
+ s/(TLS error on connection from .* \(SSL_\w+\): error:)(.*)/$1 <<detail omitted>>/;
next if /SSL verify error: depth=0 error=certificate not trusted/;
# ======== Maildir things ========
}
next if /^tls_validate_require_cipher child \d+ ended: status=0x0/;
- # We invoke Exim with -D, so we hit this new messag as of Exim 4.73:
+ # We invoke Exim with -D, so we hit this new message as of Exim 4.73:
next if /^macros_trusted overridden to true by whitelisting/;
# We have to omit the localhost ::1 address so that all is well in
# Not all platforms build with DKIM enabled
next if /^PDKIM >> Body data for hash, canonicalized/;
+ # Not all platforms have sendfile support
+ next if /^cannot use sendfile for body: no support$/;
+
+ # Parts of DKIM-specific debug output depend on the time/date
+ next if /^date:\w+,\{SP\}/;
+ next if /^PDKIM \[[^[]+\] (Header hash|b) computed:/;
+
# Not all platforms support TCP Fast Open, and the compile omits the check
if (s/\S+ in hosts_try_fastopen\? no \(option unset\)\n$//)
{
{
# Berkeley DB version differences
next if / Berkeley DB error: /;
+
+ # CHUNKING: exact sizes depend on hostnames in headers
+ s/(=>.* K C="250- \d)\d+ (byte chunk, total \d)\d+/$1nn $2nn/;
}
# ======== All files other than stderr ========
log_failure($log_failed_filename, $testno, $rf);
log_test($log_summary_filename, $testno, 'F') if ($force_continue);
}
- return 1 if /^c$/i;
- last if (/^s$/);
+ return 1 if /^c$/i && $rf !~ /paniclog/ && $rsf !~ /paniclog/;
+ last if (/^[sc]$/);
}
foreach $f ($rf, $rsf)
)($|[ ]=)/x' },
'sys_bindir' =>
- { 'mainlog' => 's%/(usr/)?bin/%SYSBINDIR/%' },
+ { 'mainlog' => 's%/(usr/(local/)?)?bin/%SYSBINDIR/%' },
'sync_check_data' =>
{ 'mainlog' => 's/^(.* SMTP protocol synchronization error .* next input=.{8}).*$/$1<suppressed>/',
# The <SCRIPT> file is open for us to read an optional return code line,
# followed by the command line and any following data lines for stdin. The
# command line can be continued by the use of \. Data lines are not continued
-# in this way. In all lines, the following substutions are made:
+# in this way. In all lines, the following substitutions are made:
#
# DIR => the current directory
# CALLER => the caller of this script
# reference to the subtest number, holding previous value
# reference to the expected return code value
# reference to where to put the command name (for messages)
-# auxilliary information returned from a previous run
+# auxiliary information returned from a previous run
#
-# Returns: 0 the commmand was executed inline, no subprocess was run
+# Returns: 0 the command was executed inline, no subprocess was run
# 1 a non-exim command was run and waited for
# 2 an exim command was run and waited for
# 3 a command was run and not waited for (daemon, server, exim_lock)
# 4 EOF was encountered after an initial return code line
-# Optionally alse a second parameter, a hash-ref, with auxilliary information:
+# Optionally also a second parameter, a hash-ref, with auxiliary information:
# exim_pid: pid of a run process
# munge: name of a post-script results munger
if ($args =~ /\$msg/)
{
- my($listcmd) = "$parm_cwd/eximdir/exim -bp " .
- "-DEXIM_PATH=$parm_cwd/eximdir/exim " .
- "-C $parm_cwd/test-config |";
- print ">> Getting queue list from:\n>> $listcmd\n" if ($debug);
- open (QLIST, $listcmd) || tests_exit(-1, "Couldn't run \"exim -bp\": $!\n");
- my(@msglist) = ();
- while (<QLIST>) { push (@msglist, $1) if /^\s*\d+[smhdw]\s+\S+\s+(\S+)/; }
- close(QLIST);
+ my @listcmd = ("$parm_cwd/eximdir/exim", '-bp',
+ "-DEXIM_PATH=$parm_cwd/eximdir/exim",
+ -C => "$parm_cwd/test-config");
+ print ">> Getting queue list from:\n>> @listcmd\n" if $debug;
+ # We need the message ids sorted in ascending order.
+ # Message id is: <timestamp>-<pid>-<fractional-time>. On some systems (*BSD) the
+ # PIDs are randomized, so sorting just the whole PID doesn't work.
+ # We do the Schartz' transformation here (sort on
+ # <timestamp><fractional-time>). Thanks to Kirill Miazine
+ my @msglist =
+ map { $_->[1] } # extract the values
+ sort { $a->[0] cmp $b->[0] } # sort by key
+ map { [join('.' => (split /-/, $_)[0,2]) => $_] } # key (timestamp.fractional-time) => value(message_id)
+ map { /^\s*\d+[smhdw]\s+\S+\s+(\S+)/ } `@listcmd` or tests_exit(-1, "No output from `exim -bp` (@listcmd)\n");
# Done backwards just in case there are more than 9
if ($cmd =~ /\s-DSERVER=server\s/ && $cmd !~ /\s-DNOTDAEMON\s/)
{
- $pidfile = "$parm_cwd/spool/exim-daemon.pid";
if ($debug) { printf ">> daemon: $cmd\n"; }
run_system("sudo mkdir spool/log 2>/dev/null");
run_system("sudo chown $parm_eximuser:$parm_eximgroup spool/log");
while (<SCRIPT>) { $lineno++; last if /^\*{4}\s*$/; } # Ignore any input
# Interlock with daemon startup
- while (! stat("$pidfile") ) { select(undef, undef, undef, 0.3); }
+ for (my $count = 0; ! stat("$pidfile") && $count < 30; $count++ )
+ { select(undef, undef, undef, 0.3); }
return 3; # Don't wait
}
elsif ($cmd =~ /\s-DSERVER=wait:(\d+)\s/)
-##################################################
-# Check for sudo access to root #
-##################################################
-
-print "You need to have sudo access to root to run these tests. Checking ...\n";
-if (system('sudo true >/dev/null') != 0)
- {
- die "** Test for sudo failed: testing abandoned.\n";
- }
-else
- {
- print "Test for sudo OK\n";
- }
-
-
-
##################################################
# See if an Exim binary has been given #
##################################################
# as the path to the binary. If the first argument does not start with a
# '/' but exists in the file system, it's assumed to be the Exim binary.
-($parm_exim, @ARGV) = Exim::Runtest::exim_binary(@ARGV);
-print "Exim binary is $parm_exim\n" if $parm_exim ne '';
-
-
##################################################
# Sort out options and which tests are to be run #
# options are passed on to Exim calls within the tests. Typically, this is used
# to turn on Exim debugging while setting up a test.
-while (@ARGV > 0 && $ARGV[0] =~ /^-/)
- {
- my($arg) = shift @ARGV;
- if ($optargs eq '')
- {
- if ($arg eq "-DEBUG") { $debug = 1; $cr = "\n"; next; }
- if ($arg eq "-DIFF") { $cf = "diff -u"; next; }
- if ($arg eq "-CONTINUE"){$force_continue = 1;
- $more = "cat";
- next; }
- if ($arg eq "-UPDATE") { $force_update = 1; next; }
- if ($arg eq "-NOIPV4") { $have_ipv4 = 0; next; }
- if ($arg eq "-NOIPV6") { $have_ipv6 = 0; next; }
- if ($arg eq "-KEEP") { $save_output = 1; next; }
- if ($arg eq "-VALGRIND") { $valgrind = 1; next; }
- if ($arg =~ /^-FLAVOU?R$/) { $flavour = shift; next; }
- }
- $optargs .= " $arg";
- }
+Getopt::Long::Configure qw(no_getopt_compat);
+GetOptions(
+ 'debug' => sub { $debug = 1; $cr = "\n" },
+ 'diff' => sub { $cf = 'diff -u' },
+ 'continue' => sub { $force_continue = 1; $more = 'cat' },
+ 'update' => \$force_update,
+ 'ipv4!' => \$have_ipv4,
+ 'ipv6!' => \$have_ipv6,
+ 'keep' => \$save_output,
+ 'slow' => \$slow,
+ 'valgrind' => \$valgrind,
+ 'range=s{2}' => \my @range_wanted,
+ 'test=i@' => \my @tests_wanted,
+ 'flavor|flavour=s' => $flavour,
+ 'help' => sub { pod2usage(-exit => 0) },
+ 'man' => sub {
+ pod2usage(
+ -exit => 0,
+ -verbose => 2,
+ -noperldoc => system('perldoc -V 2>/dev/null 1>&2')
+ );
+ },
+) or pod2usage;
+
+($parm_exim, @ARGV) = Exim::Runtest::exim_binary(@ARGV);
+print "Exim binary is `$parm_exim'\n" if defined $parm_exim;
-# Any subsequent arguments are a range of test numbers.
-if (@ARGV > 0)
+my @wanted = sort numerically uniq
+ @tests_wanted ? @tests_wanted : (),
+ @range_wanted ? $range_wanted[0] .. $range_wanted[1] : (),
+ @ARGV ? @ARGV == 1 ? $ARGV[0] :
+ $ARGV[1] eq '+' ? $ARGV[0]..($ARGV[0] >= 9000 ? TEST_SPECIAL_TOP : TEST_TOP) :
+ 0+$ARGV[0]..0+$ARGV[1] # add 0 to cope with test numbers starting with zero
+ : ();
+@wanted = 1..TEST_TOP if not @wanted;
+
+##################################################
+# Check for sudo access to root #
+##################################################
+
+print "You need to have sudo access to root to run these tests. Checking ...\n";
+if (system('sudo true >/dev/null') != 0)
+ {
+ die "** Test for sudo failed: testing abandoned.\n";
+ }
+else
{
- $test_end = $test_start = $ARGV[0];
- $test_end = $ARGV[1] if (@ARGV > 1);
- $test_end = ($test_start >= 9000)? $test_special_top : $test_top
- if $test_end eq "+";
- die "** Test numbers out of order\n" if ($test_end < $test_start);
+ print "Test for sudo OK\n";
}
+
+
##################################################
# Make the command's directory current #
##################################################
# If $parm_exim is still empty, ask the caller
-if ($parm_exim eq '')
+if (not $parm_exim)
{
print "** Did not find an Exim binary to test\n";
for ($i = 0; $i < 5; $i++)
# This test for an active SpamAssassin is courtesy of John Jetmore.
# The tests are hard coded to localhost:783, so no point in making
# this test flexible like the clamav test until the test scripts are
- # changed. spamd doesn't have the nice PING/PONG protoccol that
+ # changed. spamd doesn't have the nice PING/PONG protocol that
# clamd does, but it does respond to errors in an informative manner,
# so use that.
print " OK\n";
}
+tests_exit(-1, "Failed to unlink $log_summary_filename: $!")
+ if not unlink($log_summary_filename) and -e $log_summary_filename;
##################################################
# Create a list of available tests #
# because the current binary does not support the right facilities, and also
# those that are outside the numerical range selected.
-print "\nTest range is $test_start to $test_end (flavour $flavour)\n";
+printf "\nWill run %d tests between %d and %d for flavour %s\n",
+ scalar(@wanted), $wanted[0], $wanted[-1], $flavour;
+
print "Omitting \${dlfunc expansion tests (loadable module not present)\n"
if $dlfunc_deleted;
print "Omitting dbm tests (unable to copy exim_dbmbuild)\n"
if $dbm_build_deleted;
-opendir(DIR, "scripts") || tests_exit(-1, "Failed to opendir(\"scripts\"): $!");
-@test_dirs = sort readdir(DIR);
-closedir(DIR);
-
-# Remove . and .. and CVS from the list.
-for ($i = 0; $i < @test_dirs; $i++)
- {
- my($d) = $test_dirs[$i];
- if ($d eq "." || $d eq ".." || $d eq "CVS")
- {
- splice @test_dirs, $i, 1;
- $i--;
- }
- }
+my @test_dirs = grep { not /^CVS$/ } map { basename $_ } glob 'scripts/*'
+ or die tests_exit(-1, "Failed to find test scripts in 'scripts/*`: $!");
# Scan for relevant tests
-
-tests_exit(-1, "Failed to unlink $log_summary_filename")
- if (-e $log_summary_filename && !unlink($log_summary_filename));
-for ($i = 0; $i < @test_dirs; $i++)
+# HS12: Needs to be reworked.
+DIR: for (my $i = 0; $i < @test_dirs; $i++)
{
my($testdir) = $test_dirs[$i];
my($wantthis) = 1;
# Skip this directory if the first test is equal or greater than the first
# test in the next directory.
- next if ($i < @test_dirs - 1) &&
- ($test_start >= substr($test_dirs[$i+1], 0, 4));
+ next DIR if ($i < @test_dirs - 1) &&
+ ($wanted[0] >= substr($test_dirs[$i+1], 0, 4));
# No need to carry on if the end test is less than the first test in this
# subdirectory.
- last if $test_end < substr($testdir, 0, 4);
+ last DIR if $wanted[-1] < substr($testdir, 0, 4);
# Check requirements, if any.
- if (open(REQUIRES, "scripts/$testdir/REQUIRES"))
+ if (open(my $requires, "scripts/$testdir/REQUIRES"))
{
- while (<REQUIRES>)
+ while (<$requires>)
{
next if /^\s*$/;
s/\s+$//;
tests_exit(-1, "Unknown line in \"scripts/$testdir/REQUIRES\": \"$_\"");
}
}
- close(REQUIRES);
}
else
{
# We want the tests from this subdirectory, provided they are in the
# range that was selected.
- opendir(SUBDIR, "scripts/$testdir") ||
- tests_exit(-1, "Failed to opendir(\"scripts/$testdir\"): $!");
- @testlist = sort readdir(SUBDIR);
- close(SUBDIR);
+ @testlist = grep { $_ ~~ @wanted } grep { /^\d+(?:\.\d+)?$/ } map { basename $_ } glob "scripts/$testdir/*";
+ tests_exit(-1, "Failed to read test scripts from `scripts/$testdir/*': $!")
+ if not @testlist;
foreach $test (@testlist)
{
- next if ($test !~ /^\d{4}(?:\.\d+)?$/);
- if (!$wantthis || $test < $test_start || $test > $test_end)
+ if (!$wantthis)
{
log_test($log_summary_filename, $test, '.');
}
}
}
-print ">>Test List: @test_list\n", if $debug;
+print ">>Test List:\n", join "\n", @test_list, '' if $debug;
##################################################
if (not $force_continue) {
# runtest needs to interact if we're not in continue
# mode. It does so by communicate to /dev/tty
- open(T, "/dev/tty") or tests_exit(-1, "Failed to open /dev/tty: $!");
+ open(T, '<', '/dev/tty') or tests_exit(-1, "Failed to open /dev/tty: $!");
+ print "\nPress RETURN to run the tests: ";
+ <T>;
}
-print "\nPress RETURN to run the tests: ";
-$_ = $force_continue ? "c" : <T>;
-print "\n";
-
-$lasttestdir = '';
-
foreach $test (@test_list)
{
- local($lineno) = 0;
- local($commandno) = 0;
- local($subtestno) = 0;
+ state $lasttestdir = '';
+
+ local $lineno = 0;
+ local $commandno = 0;
+ local $subtestno = 0;
+ local $sortlog = 0;
+
(local $testno = $test) =~ s|.*/||;
- local($sortlog) = 0;
- my($gnutls) = 0;
- my($docheck) = 1;
- my($thistestdir) = substr($test, 0, -5);
+ # Leaving traces in the process table and in the environment
+ # gives us a chance to identify hanging processes (exim daemons)
+ local $0 = "[runtest $testno]";
+ local $ENV{EXIM_TEST_NUMBER} = $testno;
+
+ my $gnutls = 0;
+ my $docheck = 1;
+ my $thistestdir = substr($test, 0, -5);
$dynamic_socket->close() if $dynamic_socket;
$gnutls = 0;
if (-s "scripts/$thistestdir/REQUIRES")
{
- my($indent) = '';
+ my $indent = '';
print "\n>>> The following tests require: ";
- open(IN, "scripts/$thistestdir/REQUIRES") ||
- tests_exit(-1, "Failed to open scripts/$thistestdir/REQUIRES: $1");
- while (<IN>)
+ open(my $requires, '<', "scripts/$thistestdir/REQUIRES") ||
+ tests_exit(-1, "Failed to open scripts/$thistestdir/REQUIRES: $!");
+ while (<$requires>)
{
$gnutls = 1 if /^support GnuTLS/;
print $indent, $_;
$indent = ">>> ";
}
- close(IN);
}
+ $lasttestdir = $thistestdir;
}
- $lasttestdir = $thistestdir;
# Remove any debris in the spool directory and the test-mail directory
# and also the files for collecting stdout and stderr. Then put back
my($rc, $run_extra) = run_command($testno, \$subtestno, \$expectrc, \$commandname, $TEST_STATE);
my($cmdrc) = $?;
- $0 = "[runtest $testno]";
-
if ($debug) {
print ">> rc=$rc cmdrc=$cmdrc\n";
if (defined $run_extra) {
if ($? != 0)
{
if (($? & 0xff) == 0)
- { printf("Server return code %d", $?/256); }
+ { printf("Server return code %d for test %d starting line %d", $?/256,
+ $testno, $subtest_startline); }
elsif (($? & 0xff00) == 0)
{ printf("Server killed by signal %d", $? & 255); }
else
if ($docheck)
{
+ sleep 1 if $slow;
my $rc = check_output($TEST_STATE->{munge});
log_test($log_summary_filename, $testno, 'P') if ($rc == 0);
if ($rc < 2)
# Exit from the test script #
##################################################
-tests_exit(-1, "No runnable tests selected") if @test_list == 0;
+tests_exit(-1, "No runnable tests selected") if not @test_list;
tests_exit(0);
+__END__
+
+=head1 NAME
+
+ runtest - run the exim testsuite
+
+=head1 SYNOPSIS
+
+ runtest [exim-path] [options] [test0 [test1]]
+
+=head1 DESCRIPTION
+
+B<runtest> runs the Exim testsuite.
+
+=head1 OPTIONS
+
+For legacy reasons the options are not case sensitive.
+
+=over
+
+=item B<--continue>
+
+Do not stop for user interaction or on errors. (default: off)
+
+=item B<--debug>
+
+This option enables the output of debug information when running the
+various test commands. (default: off)
+
+=item B<--diff>
+
+Use C<diff -u> for comparing the expected output with the produced
+output. (default: use a built-in routine)
+
+=item B<--flavor>|B<--flavour> I<flavour>
+
+Override the expected results for results for a specific (OS) flavour.
+(default: unused)
+
+=item B<--[no]ipv4>
+
+Skip IPv4 related setup and tests (default: use ipv4)
+
+=item B<--[no]ipv6>
+
+Skip IPv6 related setup and tests (default: use ipv6)
+
+=item B<--keep>
+
+Keep the various output files produced during a test run. (default: don't keep)
+
+=item B<--range> I<n0> I<n1>
+
+Run tests between (including) I<n0> and I<n1>. A "+" may be used to specify the "last
+test available".
+
+=item B<--slow>
+
+Insert some delays to compensate for a slow host system. (default: off)
+
+=item B<--test> I<n>
+
+Run the specified test. This option may used multiple times.
+
+=item B<--update>
+
+Automatically update the recorded (expected) data on mismatch. (default: off)
+
+=item B<--valgrind>
+
+Start Exim wrapped by I<valgrind>. (default: don't use valgrind)
+
+=back
+
+=cut
+
+
# End of runtest script
git://git.exim.org / exim.git / blobdiff