location-header: Only set the location field when the input

Only set headers_out.location when the input also originally did
that.

This still needs a test to prevent regression, but I thought I'd
put this out so people can try this.

Should fix https://github.com/pagespeed/ngx_pagespeed/issues/819
Might help https://github.com/pagespeed/ngx_pagespeed/issues/711

config

@@ -27,8 +27,8 @@ if [ "$mod_pagespeed_dir" = "unset" ] ; then
     echo "   You need to separately download the pagespeed library:"
     echo ""
     echo "     $ cd /path/to/ngx_pagespeed"
-    echo "     $ wget https://dl.google.com/dl/page-speed/psol/1.8.31.2.tar.gz"
-    echo "     $ tar -xzvf 1.8.31.2.tar.gz # expands to psol/"
+    echo "     $ wget https://dl.google.com/dl/page-speed/psol/1.9.32.1.tar.gz"
+    echo "     $ tar -xzvf 1.9.32.1.tar.gz # expands to psol/"
     echo ""
     echo "   Or see the installation instructions:"
     echo "     https://github.com/pagespeed/ngx_pagespeed#how-to-build"
@@ -39,31 +39,6 @@ else
   build_from_source=true
 fi
 
-psol_binary="${PSOL_BINARY:-unset}"
-if [ "$psol_binary" = "unset" ] ; then
-  if $build_from_source ; then
-    psol_binary="\
-        $mod_pagespeed_dir/net/instaweb/automatic/pagespeed_automatic.a"
-  else
-    psol_library_dir="$ngx_addon_dir/psol/lib/$buildtype/$os_name/$arch_name"
-    psol_binary="$psol_library_dir/pagespeed_automatic.a"
-  fi
-fi
-
-echo "mod_pagespeed_dir=$mod_pagespeed_dir"
-echo "build_from_source=$build_from_source"
-
-ngx_feature="psol"
-ngx_feature_name=""
-ngx_feature_run=no
-ngx_feature_incs="
-#include \"net/instaweb/htmlparse/public/html_parse.h\"
-#include \"net/instaweb/htmlparse/public/html_writer_filter.h\"
-#include \"net/instaweb/util/public/string.h\"
-#include \"net/instaweb/util/public/string_writer.h\"
-#include \"net/instaweb/util/public/null_message_handler.h\"
-"
-
 os_name='unknown_os'
 arch_name='unknown_arch'
 uname_os=`uname`
@@ -124,6 +99,31 @@ if [ "$WNO_ERROR" = "YES" ]; then
     CFLAGS="$CFLAGS -Wno-error"
 fi
 
+psol_binary="${PSOL_BINARY:-unset}"
+if [ "$psol_binary" = "unset" ] ; then
+  if $build_from_source ; then
+    psol_binary="\
+        $mod_pagespeed_dir/net/instaweb/automatic/pagespeed_automatic.a"
+  else
+    psol_library_dir="$ngx_addon_dir/psol/lib/$buildtype/$os_name/$arch_name"
+    psol_binary="$psol_library_dir/pagespeed_automatic.a"
+  fi
+fi
+
+echo "mod_pagespeed_dir=$mod_pagespeed_dir"
+echo "build_from_source=$build_from_source"
+
+ngx_feature="psol"
+ngx_feature_name=""
+ngx_feature_run=no
+ngx_feature_incs="
+#include \"pagespeed/kernel/base/string.h\"
+#include \"pagespeed/kernel/base/string_writer.h\"
+#include \"pagespeed/kernel/base/null_message_handler.h\"
+#include \"pagespeed/kernel/html/html_parse.h\"
+#include \"pagespeed/kernel/html/html_writer_filter.h\"
+"
+
 pagespeed_include="\
   $mod_pagespeed_dir \
   $mod_pagespeed_dir/third_party/chromium/src \

src/ngx_base_fetch.cc

@@ -39,7 +39,7 @@ NgxBaseFetch::NgxBaseFetch(ngx_http_request_t* r, int pipe_fd,
       last_buf_sent_(false),
       pipe_fd_(pipe_fd),
       references_(2),
-      handle_error_(true),
+      ipro_lookup_(false),
       preserve_caching_headers_(preserve_caching_headers) {
   if (pthread_mutex_init(&mutex_, NULL)) CHECK(0);
 }
@@ -137,14 +137,20 @@ void NgxBaseFetch::HandleHeadersComplete() {
   int status_code = response_headers()->status_code();
   bool status_ok = (status_code != 0) && (status_code < 400);
 
-  if (status_ok || handle_error_) {
+  if (!ipro_lookup_ || status_ok) {
     // If this is a 404 response we need to count it in the stats.
     if (response_headers()->status_code() == HttpStatus::kNotFound) {
       server_context_->rewrite_stats()->resource_404_count()->Add(1);
     }
   }
 
-  RequestCollection();  // Headers available.
+  // For the IPRO lookup, supress notification of the nginx side here.
+  // If we send both this event and the one from done, nasty stuff will happen
+  // if we loose the race with with the nginx side destructing this base fetch
+  // instance (and thereby clearing the byte and its pending extraneous event.
+  if (!ipro_lookup_) {
+    RequestCollection();  // Headers available.
+  }
 }
 
 bool NgxBaseFetch::HandleFlush(MessageHandler* handler) {

src/ngx_base_fetch.h

@@ -79,7 +79,7 @@ class NgxBaseFetch : public AsyncFetch {
 
   // Called by nginx when it's done with us.
   void Release();
-  void set_handle_error(bool x) { handle_error_ = x; }
+  void set_ipro_lookup(bool x) { ipro_lookup_ = x; }
 
  private:
   virtual bool HandleWrite(const StringPiece& sp, MessageHandler* handler);
@@ -117,7 +117,7 @@ class NgxBaseFetch : public AsyncFetch {
   // decremented once when Done() is called and once when Release() is called.
   int references_;
   pthread_mutex_t mutex_;
-  bool handle_error_;
+  bool ipro_lookup_;
   PreserveCachingHeaders preserve_caching_headers_;
 
   DISALLOW_COPY_AND_ASSIGN(NgxBaseFetch);

src/ngx_pagespeed.cc

@@ -89,6 +89,8 @@ extern ngx_module_t ngx_pagespeed;
 namespace net_instaweb {
 
 const char* kInternalEtagName = "@psol-etag";
+bool factory_init_called = false;
+
 // The process context takes care of proactively initialising
 // a few libraries for us, some of which are not thread-safe
 // when they are initialized lazily.
@@ -206,6 +208,13 @@ ngx_int_t string_piece_to_buffer_chain(
 
 namespace {
 
+// Get the context for this request.  ps_connection_read_handler should already
+// have been called to create it.
+ps_request_ctx_t* ps_get_request_context(ngx_http_request_t* r) {
+  return static_cast<ps_request_ctx_t*>(
+      ngx_http_get_module_ctx(r, ngx_pagespeed));
+}
+
 // Setting headers in nginx is tricky because it's not just a matter of adding
 // them to a list.  You also need to remove them if there's already one there,
 // as well as setting the shortcut pointers (both upper case and lower case).
@@ -286,11 +295,6 @@ void copy_response_headers_from_ngx(const ngx_http_request_t* r,
 
   headers->set_status_code(r->headers_out.status);
 
-  if (r->headers_out.location != NULL) {
-    headers->Add(HttpAttributes::kLocation,
-                 str_to_string_piece(r->headers_out.location->value));
-  }
-
   // Manually copy over the content type because it's not included in
   // request_->headers_out.headers.
   headers->Add(HttpAttributes::kContentType,
@@ -435,7 +439,10 @@ ngx_int_t copy_response_headers_to_ngx(
     } else if (STR_EQ_LITERAL(name, "Last-Modified")) {
       headers_out->last_modified = header;
     } else if (STR_EQ_LITERAL(name, "Location")) {
-      headers_out->location = header;
+      ps_request_ctx_t* ctx = ps_get_request_context(r);
+      if (ctx->location_field_set) {
+        headers_out->location = header;
+      }
     } else if (STR_EQ_LITERAL(name, "Server")) {
       headers_out->server = header;
     } else if (STR_EQ_LITERAL(name, "Content-Length")) {
@@ -610,6 +617,16 @@ char* ps_configure(ngx_conf_t* cf,
                    NgxRewriteOptions** options,
                    MessageHandler* handler,
                    net_instaweb::RewriteOptions::OptionScope option_scope) {
+  ps_main_conf_t* cfg_m = static_cast<ps_main_conf_t*>(
+      ngx_http_conf_get_module_main_conf(cf, ngx_pagespeed));
+
+  if (!factory_init_called) {
+    // Init logging to nginx's default error_log.
+    cfg_m->driver_factory->LoggingInit(cf->cycle->log);
+    cfg_m->driver_factory->Init();
+    factory_init_called = true;
+  }
+
   // args[0] is always "pagespeed"; ignore it.
   ngx_uint_t n_args = cf->args->nelts - 1;
 
@@ -660,8 +677,6 @@ char* ps_configure(ngx_conf_t* cf,
     // directive yet.  That happens below in ParseAndSetOptions().
   }
 
-  ps_main_conf_t* cfg_m = static_cast<ps_main_conf_t*>(
-      ngx_http_cycle_get_module_main_conf(cf->cycle, ngx_pagespeed));
   if (*options == NULL) {
     *options = new NgxRewriteOptions(
         cfg_m->driver_factory->thread_system());
@@ -730,7 +745,16 @@ void ps_cleanup_srv_conf(void* data) {
   // from being executed
 
   if (!factory_deleted && cfg_s->server_context != NULL) {
-    delete cfg_s->server_context->factory();
+    NgxRewriteDriverFactory* factory = dynamic_cast<NgxRewriteDriverFactory*>(
+      cfg_s->server_context->factory());
+
+    if (!factory_init_called) {
+      factory->LoggingInit(ngx_cycle->log);
+      factory->Init();
+      factory_init_called = true;
+    }
+
+    delete factory;
     factory_deleted = true;
   }
   if (cfg_s->proxy_fetch_factory != NULL) {
@@ -800,7 +824,7 @@ void* ps_create_main_conf(ngx_conf_t* cf) {
       new SystemThreadSystem(),
       "" /* hostname, not used */,
       -1 /* port, not used */);
-  cfg_m->driver_factory->Init();
+  factory_init_called = false;
   ps_set_conf_cleanup_handler(cf, ps_cleanup_main_conf, cfg_m);
   return cfg_m;
 }
@@ -1084,13 +1108,6 @@ GoogleString ps_determine_url(ngx_http_request_t* r) {
                 host, port_string, str_to_string_piece(r->unparsed_uri));
 }
 
-// Get the context for this request.  ps_connection_read_handler should already
-// have been called to create it.
-ps_request_ctx_t* ps_get_request_context(ngx_http_request_t* r) {
-  return static_cast<ps_request_ctx_t*>(
-      ngx_http_get_module_ctx(r, ngx_pagespeed));
-}
-
 void ps_release_base_fetch(ps_request_ctx_t* ctx);
 
 // we are still at pagespeed phase
@@ -1178,6 +1195,12 @@ ngx_int_t ps_base_fetch_handler(ngx_http_request_t* r) {
                 STR_CASE_EQ_LITERAL(header->key, "Last-Modified") ||
                 STR_CASE_EQ_LITERAL(header->key, "Expires"))))) {
           header->hash = 0;
+          if (STR_CASE_EQ_LITERAL(header->key, "Location")) {
+            // There's a possible issue with the location header, where setting
+            // the hash to 0 is not enough. See:
+            // https://github.com/nginx/nginx/blob/master/src/http/ngx_http_header_filter_module.c#L314
+            r->headers_out.location = NULL;
+          }
         }
       }
     } else {
@@ -1280,6 +1303,15 @@ void ps_connection_read_handler(ngx_event_t* ev) {
     rc = read(c->fd, chr, 256);
   } while (rc > 0 || (rc == -1 && errno == EINTR));  // Retry on EINTR.
 
+  if (r->connection->error) {
+    ngx_log_error(NGX_LOG_DEBUG, ngx_cycle->log, 0,
+                  "pagespeed [%p] request already finalized", r);
+    ctx->pagespeed_connection = NULL;
+    ngx_close_connection(c);
+    ngx_http_finalize_request(r, NGX_ERROR);
+    return;
+  }
+
   if (rc == -1 && errno != EAGAIN && errno != EWOULDBLOCK) {
     ctx->pagespeed_connection = NULL;
     ngx_close_connection(c);
@@ -1844,12 +1876,20 @@ ngx_int_t ps_resource_handler(ngx_http_request_t* r,
     CHECK(ctx == NULL);
     ctx = new ps_request_ctx_t();
 
+    ctx->base_fetch = NULL;
+    ctx->pagespeed_connection = NULL;
     ctx->r = r;
-    ctx->write_pending = false;
     ctx->html_rewrite = false;
     ctx->in_place = false;
-    ctx->pagespeed_connection = NULL;
+    ctx->write_pending = false;
+    ctx->fetch_done = false;
     ctx->preserve_caching_headers = kDontPreserveHeaders;
+    ctx->proxy_fetch = NULL;
+    ctx->inflater_ = NULL;
+    ctx->driver = NULL;
+    ctx->recorder = NULL;
+    ctx->ipro_response_headers = NULL;
+    ctx->location_field_set = false;
 
     // See build_context_for_request() in mod_instaweb.cc
     // TODO(jefftk): Is this the right place to be modifying caching headers for
@@ -2023,7 +2063,7 @@ ngx_int_t ps_resource_handler(ngx_http_request_t* r,
         url_string.c_str());
 
     ctx->in_place = true;
-    ctx->base_fetch->set_handle_error(false);
+    ctx->base_fetch->set_ipro_lookup(true);
     ctx->driver->FetchInPlaceResource(
         url, false /* proxy_mode */, ctx->base_fetch);
 
@@ -2263,6 +2303,8 @@ ngx_int_t ps_html_rewrite_header_filter(ngx_http_request_t* r) {
   }
 
   ps_strip_html_headers(r);
+  // See https://github.com/pagespeed/ngx_pagespeed/issues/819
+  ctx->location_field_set = r->headers_out.location != NULL;
 
   // TODO(jefftk): is this thread safe?
   copy_response_headers_from_ngx(r, ctx->base_fetch->response_headers());
@@ -3074,10 +3116,15 @@ ngx_int_t ps_init_module(ngx_cycle_t* cycle) {
           "UseNativeFetcher is on, please configure a resolver.");
       return NGX_ERROR;
     }
-
+    // Update logging to the configured error_log in the http{} block.
     cfg_m->driver_factory->LoggingInit(cycle->log);
     cfg_m->driver_factory->RootInit();
   } else {
+    if (!factory_init_called) {
+      cfg_m->driver_factory->LoggingInit(cycle->log);
+      cfg_m->driver_factory->Init();
+      factory_init_called = true;
+    }
     delete cfg_m->driver_factory;
     cfg_m->driver_factory = NULL;
   }
@@ -3114,11 +3161,11 @@ ngx_int_t ps_init_child_process(ngx_cycle_t* cycle) {
     // Some server{} blocks may not have a ServerContext in that case we must
     // not instantiate a ProxyFetchFactory.
     if (cfg_s->server_context != NULL) {
-      cfg_s->proxy_fetch_factory = new ProxyFetchFactory(cfg_s->server_context);
       ngx_http_core_loc_conf_t* clcf = static_cast<ngx_http_core_loc_conf_t*>(
           cscfp[s]->ctx->loc_conf[ngx_http_core_module.ctx_index]);
       cfg_m->driver_factory->SetServerContextMessageHandler(
           cfg_s->server_context, clcf->error_log);
+      cfg_s->proxy_fetch_factory = new ProxyFetchFactory(cfg_s->server_context);
     }
   }
 

src/ngx_pagespeed.h

@@ -107,6 +107,7 @@ typedef struct {
   // We need to remember the URL here as well since we may modify what NGX
   // gets by stripping our special query params and honoring X-Forwarded-Proto.
   GoogleString url_string;
+  bool location_field_set;
 } ps_request_ctx_t;
 
 

src/ngx_rewrite_options.cc

@@ -498,6 +498,10 @@ NgxRewriteOptions* NgxRewriteOptions::Clone() const {
   return options;
 }
 
+void NgxRewriteOptions::Merge(const RewriteOptions& src) {
+  SystemRewriteOptions::Merge(src);
+}
+
 const NgxRewriteOptions* NgxRewriteOptions::DynamicCast(
     const RewriteOptions* instance) {
   return dynamic_cast<const NgxRewriteOptions*>(instance);

src/ngx_rewrite_options.h

@@ -140,6 +140,7 @@ class NgxRewriteOptions : public SystemRewriteOptions {
 
   // Make an identical copy of these options and return it.
   virtual NgxRewriteOptions* Clone() const;
+  virtual void Merge(const RewriteOptions& src);
 
   // Returns a suitably down cast version of 'instance' if it is an instance
   // of this class, NULL if not.