X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=t%2Frun.perl;h=b90715a6ffe09ce86d04e84ad7d7b1f5b71c604d;hb=a74c0a16c07e4c8a3ae03af13854f190ddb3bd8a;hp=f4e235b7fd3a800fb8c1cc4ed3c1fef81e87da11;hpb=b437cc7f5ad2712d9064c691c5946be80a73eedb;p=public-inbox.git

diff --git a/t/run.perl b/t/run.perl
index f4e235b7..b90715a6 100755
--- a/t/run.perl
+++ b/t/run.perl
@@ -1,5 +1,5 @@
 #!/usr/bin/perl -w
-# Copyright (C) 2019-2021 all contributors <meta@public-inbox.org>
+# Copyright (C) all contributors <meta@public-inbox.org>
 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
 #
 # Parallel test runner which preloads code and reuses worker processes
@@ -10,15 +10,15 @@
 #
 # Usage: $PERL -I lib -w t/run.perl -j4
 # Or via prove(1): prove -lvw t/run.perl :: -j4
-use strict;
-use v5.10.1;
+use v5.12;
 use IO::Handle; # ->autoflush
 use PublicInbox::TestCommon;
 use PublicInbox::Spawn;
 use Getopt::Long qw(:config gnu_getopt no_ignore_case auto_abbrev);
 use Errno qw(EINTR);
 use Fcntl qw(:seek);
-use POSIX qw(_POSIX_PIPE_BUF WNOHANG);
+use POSIX qw(WNOHANG);
+use File::Temp ();
 my $jobs = 1;
 my $repeat = 1;
 $| = 1;
@@ -39,10 +39,24 @@ open my $OLDERR, '>&STDERR' or die "dup STDERR: $!";
 $OLDOUT->autoflush(1);
 $OLDERR->autoflush(1);
 
+my ($run_log, $tmp_rl);
+my $rl = $ENV{TEST_RUN_LOG};
+unless ($rl) {
+	$tmp_rl = File::Temp->new(CLEANUP => 1);
+	$rl = $tmp_rl->filename;
+}
+open $run_log, '+>>', $rl or die "open $rl: $!";
+$run_log->autoflush(1); # one reader, many writers
+
 key2sub($_) for @tests; # precache
 
 my ($for_destroy, $lei_env, $lei_daemon_pid, $owner_pid);
-if (!$ENV{TEST_LEI_DAEMON_PERSIST_DIR} &&
+
+# TEST_LEI_DAEMON_PERSIST is currently broken.  I get ECONNRESET from
+# lei even with high kern.ipc.soacceptqueue=1073741823 or SOMAXCONN, not
+# sure why.  Also, testing our internal inotify usage is unreliable
+# because lei-daemon uses a single inotify FD for all clients.
+if ($ENV{TEST_LEI_DAEMON_PERSIST} && !$ENV{TEST_LEI_DAEMON_PERSIST_DIR} &&
 		(PublicInbox::Spawn->can('recv_cmd4') ||
 			eval { require Socket::MsgHdr })) {
 	$lei_env = {};
@@ -52,6 +66,13 @@ if (!$ENV{TEST_LEI_DAEMON_PERSIST_DIR} &&
 	chomp $lei_daemon_pid;
 	$lei_daemon_pid =~ /\A[0-9]+\z/ or die "no daemon pid: $lei_daemon_pid";
 	kill(0, $lei_daemon_pid) or die "kill $lei_daemon_pid: $!";
+	if (my $t = $ENV{GNU_TAIL}) {
+		system("$t --pid=$lei_daemon_pid -F " .
+			"$lei_env->{XDG_RUNTIME_DIR}/lei/errors.log >&2 &");
+	}
+	if (my $strace_cmd = $ENV{STRACE_CMD}) {
+		system("$strace_cmd -p $lei_daemon_pid &");
+	}
 	$owner_pid = $$;
 }
 
@@ -113,6 +134,7 @@ END { test_status() if (defined($worker_test) && $worker == $$) }
 
 sub run_test ($) {
 	my ($test) = @_;
+	syswrite($run_log, "$$ $test\n");
 	my $log_fh;
 	if ($log_suffix ne '') {
 		my $log = $test;
@@ -144,7 +166,8 @@ my $start_worker = sub {
 	my ($j, $rd, $wr, $todo) = @_;
 	my $pid = fork // DIE "fork: $!";
 	if ($pid == 0) {
-		close $wr if $wr;
+		close $wr;
+		$SIG{USR1} = undef; # undo parent $SIG{USR1}
 		$worker = $$;
 		while (1) {
 			my $r = sysread($rd, my $buf, UINT_SIZE);
@@ -159,7 +182,10 @@ my $start_worker = sub {
 			$tb->reset;
 		}
 		kill 'USR1', $producer if !$eof; # sets $eof in $producer
-		DIE join('', map { "E: $_\n" } @err) if @err;
+		if (@err) { # write to run_log for $sigchld handler
+			syswrite($run_log, "$$ @err\n");
+			DIE join('', map { "E: $_\n" } @err);
+		}
 		exit(0);
 	} else {
 		$pids{$pid} = $j;
@@ -176,15 +202,11 @@ for (my $i = $repeat; $i != 0; $i--) {
 	pipe(my ($rd, $wr)) or DIE "pipe: $!";
 
 	# fill the queue before forking so children can start earlier
-	my $n = (_POSIX_PIPE_BUF / UINT_SIZE);
-	if ($n >= $#todo) {
-		print $wr join('', map { pack('I', $_) } (0..$#todo)) or DIE;
-		undef $wr;
-	} else { # write what we can...
-		$wr->autoflush(1);
-		print $wr join('', map { pack('I', $_) } (0..$n)) or DIE;
-		$n += 1; # and send more ($n..$#todo), later
-	}
+	$wr->autoflush(1);
+	$wr->blocking(0);
+	my $todo_buf = join('', map { pack('I', $_) } (0..$#todo));
+	my $woff = syswrite($wr, $todo_buf) // DIE "syswrite: $!";
+	substr($todo_buf, 0, $woff, '');
 	$eof = undef;
 	local $SIG{USR1} = sub { $eof = 1 };
 	my $sigchld = sub {
@@ -198,7 +220,12 @@ for (my $i = $repeat; $i != 0; $i--) {
 				push @err, "reaped unknown $pid ($?)";
 				next;
 			}
-			push @err, "job[$j] ($?)" if $?;
+			if ($?) {
+				seek($run_log, 0, SEEK_SET);
+				chomp(my @t = grep(/^$pid /, <$run_log>));
+				$t[0] //= "$pid unknown";
+				push @err, "job[$j] ($?) PID=$t[-1]";
+			}
 			# skip_all can exit(0), respawn if needed:
 			if (!$eof) {
 				print $OLDERR "# respawning job[$j]\n";
@@ -211,12 +238,13 @@ for (my $i = $repeat; $i != 0; $i--) {
 	for (my $j = 0; $j < $jobs; $j++) {
 		$start_worker->($j, $rd, $wr, \@todo);
 	}
-	if ($wr) {
+	{
 		local $SIG{CHLD} = $sigchld;
 		# too many tests to fit in the pipe before starting workers,
 		# send the rest now the workers are running
-		print $wr join('', map { pack('I', $_) } ($n..$#todo)) or DIE;
-		undef $wr;
+		$wr->blocking(1);
+		print $wr $todo_buf or DIE;
+		close $wr;
 	}
 
 	$sigchld->(0) while scalar(keys(%pids));
@@ -226,5 +254,8 @@ for (my $i = $repeat; $i != 0; $i--) {
 print $OLDOUT "1..".($repeat * scalar(@tests))."\n" if $repeat >= 0;
 if ($lei_env && $$ == $owner_pid) {
 	my $opt = { 1 => $OLDOUT, 2 => $OLDERR };
+	my $cur_daemon_pid;
+	run_script([qw(lei daemon-pid)], $lei_env, { 1 => \$cur_daemon_pid });
 	run_script([qw(lei daemon-kill)], $lei_env, $opt);
+	DIE "lei daemon restarted\n" if $cur_daemon_pid != $lei_daemon_pid;
 }