From: Francesco Chemolli <kinkie@squid-cache.org>
Date: Fri, 28 May 2021 18:19:54 +0000 (+0000)
Subject: Refactor Via stats to replace hash_table with std::unordered_map (#829)
X-Git-Tag: SQUID_6_0_1~329
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=cf1f23eea5cf4bc3053f6153ac0b96c415ad6033;p=thirdparty%2Fsquid.git

Refactor Via stats to replace hash_table with std::unordered_map (#829)
---

diff --git a/src/client_side_request.cc b/src/client_side_request.cc
index bd1c3e4170..6b94783201 100644
--- a/src/client_side_request.cc
+++ b/src/client_side_request.cc
@@ -51,6 +51,7 @@
 #include "proxyp/Header.h"
 #include "redirect.h"
 #include "rfc1738.h"
+#include "sbuf/StringConvert.h"
 #include "SquidConfig.h"
 #include "SquidTime.h"
 #include "Store.h"
@@ -1111,7 +1112,7 @@ clientInterpretRequestHeaders(ClientHttpRequest * http)
         }
 
 #if USE_FORW_VIA_DB
-        fvdbCountVia(s.termedBuf());
+        fvdbCountVia(StringToSBuf(s));
 
 #endif
 
diff --git a/src/log/access_log.cc b/src/log/access_log.cc
index 6a1621a8af..00b1efcb04 100644
--- a/src/log/access_log.cc
+++ b/src/log/access_log.cc
@@ -11,9 +11,11 @@
 #include "squid.h"
 #include "AccessLogEntry.h"
 #include "acl/Checklist.h"
+#include "sbuf/Algorithms.h"
 #if USE_ADAPTATION
 #include "adaptation/Config.h"
 #endif
+#include "base/PackableStream.h"
 #include "CachePeer.h"
 #include "error/Detail.h"
 #include "errorpage.h"
@@ -30,6 +32,7 @@
 #include "MemBuf.h"
 #include "mgr/Registration.h"
 #include "rfc1738.h"
+#include "sbuf/SBuf.h"
 #include "SquidConfig.h"
 #include "SquidTime.h"
 #include "Store.h"
@@ -39,6 +42,8 @@
 #include "eui/Eui64.h"
 #endif
 
+#include <unordered_map>
+
 #if HEADERS_LOG
 static Logfile *headerslog = NULL;
 #endif
@@ -56,7 +61,14 @@ typedef struct {
     hash_link hash;
     int n;
 } fvdb_entry;
-static hash_table *via_table = NULL;
+
+using HeaderValueCountsElement = std::pair<SBuf, uint64_t>;
+/// counts the number of header field value occurrences
+using HeaderValueCounts = std::unordered_map<SBuf, uint64_t, std::hash<SBuf>, std::equal_to<SBuf>, PoolingAllocator<HeaderValueCountsElement> >;
+
+/// counts the number of HTTP Via header field value occurrences
+static HeaderValueCounts TheViaCounts;
+
 static hash_table *forw_table = NULL;
 static void fvdbInit();
 static void fvdbDumpTable(StoreEntry * e, hash_table * hash);
@@ -443,7 +455,6 @@ accessLogInit(void)
 static void
 fvdbInit(void)
 {
-    via_table = hash_create((HASHCMP *) strcmp, 977, hash4);
     forw_table = hash_create((HASHCMP *) strcmp, 977, hash4);
 }
 
@@ -475,9 +486,9 @@ fvdbCount(hash_table * hash, const char *key)
 }
 
 void
-fvdbCountVia(const char *key)
+fvdbCountVia(const SBuf &headerValue)
 {
-    fvdbCount(via_table, key);
+    ++TheViaCounts[headerValue];
 }
 
 void
@@ -503,10 +514,19 @@ fvdbDumpTable(StoreEntry * e, hash_table * hash)
     }
 }
 
+static void
+fvdbDumpCounts(StoreEntry &e, const HeaderValueCounts &counts)
+{
+    PackableStream os(e);
+    for (const auto &i : counts)
+        os << std::setw(9) << i.second << ' ' << i.first << "\n";
+}
+
 static void
 fvdbDumpVia(StoreEntry * e)
 {
-    fvdbDumpTable(e, via_table);
+    assert(e);
+    fvdbDumpCounts(*e, TheViaCounts);
 }
 
 static void
@@ -527,9 +547,7 @@ fvdbFreeEntry(void *data)
 static void
 fvdbClear(void)
 {
-    hashFreeItems(via_table, fvdbFreeEntry);
-    hashFreeMemory(via_table);
-    via_table = hash_create((HASHCMP *) strcmp, 977, hash4);
+    TheViaCounts.clear();
     hashFreeItems(forw_table, fvdbFreeEntry);
     hashFreeMemory(forw_table);
     forw_table = hash_create((HASHCMP *) strcmp, 977, hash4);
diff --git a/src/log/access_log.h b/src/log/access_log.h
index 51f42cf443..4f97386503 100644
--- a/src/log/access_log.h
+++ b/src/log/access_log.h
@@ -12,8 +12,12 @@
 #define SQUID_LOG_ACCESS_LOG_H_
 
 #include "LogTags.h"
+#include "sbuf/forward.h"
+
+/// count occurrences of the given Via header value
+/// XXX: this function preserves all counted values until the next log rotation
+void fvdbCountVia(const SBuf &);
 
-void fvdbCountVia(const char *key);
 void fvdbCountForw(const char *key);
 
 #if HEADERS_LOG
diff --git a/src/tests/stub_liblog.cc b/src/tests/stub_liblog.cc
index a07b333484..acc3623837 100644
--- a/src/tests/stub_liblog.cc
+++ b/src/tests/stub_liblog.cc
@@ -30,7 +30,7 @@ void accessLogInit(void) STUB
 const char *accessLogTime(time_t) STUB_RETVAL(nullptr)
 
 #include "log/access_log.h"
-void fvdbCountVia(const char *) STUB
+void fvdbCountVia(const SBuf &) STUB
 void fvdbCountForw(const char *) STUB
 #if HEADERS_LOG
 void headersLog(int, int, const HttpRequestMethod &, void *) STUB