[PATCH] Tests: SSI duplicate posted requests test
Maxim Dounin
mdounin at mdounin.ru
Thu Aug 20 22:30:44 UTC 2026
# HG changeset patch
# User Maxim Dounin <mdounin at mdounin.ru>
# Date 1787264866 -10800
# Fri Aug 21 01:27:46 2026 +0300
# Node ID bb3b0c451cd79f055c8c0798d78d4841ec8913f0
# Parent b7be6f57f5a9b9dbb62aa8a6aa3c22ea77cf8004
Tests: SSI duplicate posted requests test.
diff --git a/ssi_duplicate_posted.t b/ssi_duplicate_posted.t
new file mode 100644
--- /dev/null
+++ b/ssi_duplicate_posted.t
@@ -0,0 +1,137 @@
+#!/usr/bin/perl
+
+# (C) Maxim Dounin
+
+# Tests for ssi module, extended tests to catch subrequest double post
+# issue.
+
+###############################################################################
+
+use warnings;
+use strict;
+
+use Test::More;
+use Socket qw/ CRLF /;
+
+BEGIN { use FindBin; chdir($FindBin::Bin); }
+
+use lib 'lib';
+use Test::Nginx;
+
+###############################################################################
+
+select STDERR; $| = 1;
+select STDOUT; $| = 1;
+
+my $t = Test::Nginx->new()->has(qw/http ssi/)->plan(1)
+ ->write_file_expand('nginx.conf', <<'EOF');
+
+%%TEST_GLOBALS%%
+
+daemon off;
+
+events {
+}
+
+http {
+ %%TEST_GLOBALS_HTTP%%
+
+ server {
+ listen 127.0.0.1:8080;
+ server_name localhost;
+
+ location / {
+ ssi on;
+ proxy_pass http://127.0.0.1:8081;
+ }
+
+ location /fi {
+ }
+
+ location /se {
+ }
+ }
+}
+
+EOF
+
+$t->write_file('first.html', 'first');
+$t->write_file('second.html', 'second');
+$t->run_daemon(\&http_daemon);
+$t->run()->waitforsocket('127.0.0.1:' . port(8081));
+
+###############################################################################
+
+# When a subrequest is initiated, its processing is started via posted
+# requests mechanism by the ngx_http_post_request() call, and added to
+# r->postponed to ensure correct output. And if the relevant entry in
+# r->postponed is reached before the first posted request is woken up,
+# the subrequest will end up being posted twice. This is normally safe
+# and mostly equivalent to an additional write event, but wasn't
+# expected to happen when an active subrequest was finalized without any
+# postponed data or buffering.
+#
+# In particular, such a situation was observed with SSI and proxying:
+# when an include was finalized and the main request was woken up,
+# postponed main request data were not flushed as long as the size of
+# busy buffers was below proxy_busy_buffers_size, and when another
+# include arrived from the upstream, it was posted from
+# ngx_http_post_request() and then posted again from the postpone
+# filter. With a static file as an include, this caused "header already
+# sent" alerts.
+
+like(http_get('/'), qr/firstsecond/m, 'ssi');
+
+$t->todo_alerts() unless $t->has_version('1.31.4');
+
+###############################################################################
+
+sub http_daemon {
+ my ($t) = @_;
+
+ my $server = IO::Socket::INET->new(
+ Proto => 'tcp',
+ LocalHost => '127.0.0.1',
+ LocalPort => port(8081),
+ Listen => 5,
+ Reuse => 1
+ )
+ or die "Can't create listening socket: $!\n";
+
+ local $SIG{PIPE} = 'IGNORE';
+
+ while (my $client = $server->accept()) {
+ $client->autoflush(1);
+
+ my $headers = '';
+
+ while (<$client>) {
+ $headers .= $_;
+ last if (/^\x0d?\x0a?$/);
+ }
+
+ next if $headers eq '';
+
+ my $p1 = '<!--#include virtual="/first.html" -->';
+ my $p2 = '<!--#include virtual="/second.html" -->';
+
+ print $client
+ "HTTP/1.1 200 OK" . CRLF .
+ "Connection: close" . CRLF .
+ "Transfer-Encoding: chunked" . CRLF .
+ "Content-Type: text/html" . CRLF .
+ "X-Content-Encoding: gzip" . CRLF .
+ CRLF .
+ sprintf("%x", length $p1) . CRLF .
+ $p1 . CRLF;
+
+ select undef, undef, undef, 0.5;
+
+ print $client
+ sprintf("%x", length $p2) . CRLF .
+ $p2 . CRLF .
+ "0" . CRLF . CRLF;
+ }
+}
+
+###############################################################################
More information about the nginx-devel
mailing list