From: Vsevolod Stakhov <vsevolod@rspamd.com>
Date: Sun, 23 Oct 2022 17:37:34 +0000 (+0100)
Subject: [Project] Synchronize hyperscan caches via the main process
X-Git-Tag: 3.4~28
X-Git-Url: https://source.dussan.org/?a=commitdiff_plain;h=63c6d3080d531e7eb4bf32ac985408dbd0513c6b;p=rspamd.git

[Project] Synchronize hyperscan caches via the main process
---

diff --git a/src/libserver/hyperscan_tools.cxx b/src/libserver/hyperscan_tools.cxx
index 4f3ac013a..3383915c3 100644
--- a/src/libserver/hyperscan_tools.cxx
+++ b/src/libserver/hyperscan_tools.cxx
@@ -31,6 +31,7 @@
 #include <unistd.h> /* for unlink */
 #include <optional>
 #include "unix-std.h"
+#include "rspamd_control.h"
 
 #define msg_info_hyperscan(...)   rspamd_default_log_function (G_LOG_LEVEL_INFO, \
         "hyperscan", "", \
@@ -66,7 +67,8 @@ private:
 
 	virtual ~hs_known_files_cache() {
 		// Cleanup cache dir
-		if (rspamd_current_worker != nullptr && rspamd_worker_is_primary_controller(rspamd_current_worker)) {
+		/* We clean dir merely if we are running from the main process */
+		if (rspamd_current_worker == nullptr) {
 			auto cleanup_dir = [&](std::string_view dir) -> void {
 				for (const auto &ext : cache_extensions) {
 					glob_t globbuf;
@@ -423,6 +425,23 @@ void
 rspamd_hyperscan_notice_known(const char *fname)
 {
 	rspamd::util::hs_known_files_cache::get().add_cached_file(fname);
+
+	if (rspamd_current_worker != nullptr) {
+		/* Also notify main process */
+		struct rspamd_srv_command notice_cmd;
+
+		if (strlen(fname) >= sizeof(notice_cmd.cmd.hyperscan_cache_file.path)) {
+			msg_err("internal error: length of the filename %d ('%s') is larger than control buffer path: %d",
+				(int)strlen(fname), fname, (int)sizeof(notice_cmd.cmd.hyperscan_cache_file.path));
+		}
+		else {
+			notice_cmd.type = RSPAMD_NOTICE_HYPERSCAN_CACHE;
+			rspamd_srv_send_command(rspamd_current_worker,
+				rspamd_current_worker->srv->event_loop, &notice_cmd, -1,
+				nullptr,
+				nullptr);
+		}
+	}
 }
 
 #endif // WITH_HYPERSCAN
\ No newline at end of file
diff --git a/src/libserver/rspamd_control.c b/src/libserver/rspamd_control.c
index 199efa948..63999ab6f 100644
--- a/src/libserver/rspamd_control.c
+++ b/src/libserver/rspamd_control.c
@@ -27,6 +27,10 @@
 #include <sys/resource.h>
 #endif
 
+#ifdef WITH_HYPERSCAN
+#include "hyperscan_tools.h"
+#endif
+
 static ev_tstamp io_timeout = 30.0;
 static ev_tstamp worker_io_timeout = 0.5;
 
@@ -1014,6 +1018,12 @@ rspamd_srv_handler (EV_P_ ev_io *w, int revents)
 			case RSPAMD_SRV_HEALTH:
 				rspamd_fill_health_reply (srv, &rdata->rep);
 				break;
+			case RSPAMD_NOTICE_HYPERSCAN_CACHE:
+#ifdef WITH_HYPERSCAN
+				rspamd_hyperscan_notice_known(cmd.cmd.hyperscan_cache_file.path);
+#endif
+				rdata->rep.reply.hyperscan_cache_file.unused = 0;
+				break;
 			default:
 				msg_err ("unknown command type: %d", cmd.type);
 				break;
diff --git a/src/libserver/rspamd_control.h b/src/libserver/rspamd_control.h
index 1c0f593e3..dff48927d 100644
--- a/src/libserver/rspamd_control.h
+++ b/src/libserver/rspamd_control.h
@@ -47,12 +47,13 @@ enum rspamd_srv_type {
 	RSPAMD_SRV_ON_FORK,
 	RSPAMD_SRV_HEARTBEAT,
 	RSPAMD_SRV_HEALTH,
+	RSPAMD_NOTICE_HYPERSCAN_CACHE,
 };
 
 enum rspamd_log_pipe_type {
 	RSPAMD_LOG_PIPE_SYMBOLS = 0,
 };
-#define CONTROL_PATHLEN MIN(PATH_MAX, PIPE_BUF - sizeof(int) * 2)
+#define CONTROL_PATHLEN MIN(PATH_MAX, PIPE_BUF - sizeof(int) * 2 - sizeof(gint64))
 struct rspamd_control_command {
 	enum rspamd_control_type type;
 	union {
@@ -174,6 +175,10 @@ struct rspamd_srv_command {
 		struct {
 			guint status;
 		} health;
+		/* Used when a worker loads a valid hyperscan file */
+		struct {
+			char path[CONTROL_PATHLEN];
+		} hyperscan_cache_file;
 	} cmd;
 };
 
@@ -205,6 +210,9 @@ struct rspamd_srv_reply {
 			guint scanners_count;
 			guint workers_hb_lost;
 		} health;
+		struct {
+			int unused;
+		} hyperscan_cache_file;
 	} reply;
 };