X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=plocate.cpp;h=e08c019b84d58f7b74054533010cfb61adfe4c16;hb=a9d9d88d7105a9273eac65b520757f86a421eee2;hp=2f3129326662bd52512f9ed50ba011acf627627f;hpb=4ecef43ea601ab99ad2f2cf7715dbc52e8f3c775;p=plocate

diff --git a/plocate.cpp b/plocate.cpp
index 2f31293..e08c019 100644
--- a/plocate.cpp
+++ b/plocate.cpp
@@ -1,6 +1,8 @@
+#include "access_rx_cache.h"
 #include "db.h"
 #include "dprintf.h"
 #include "io_uring_engine.h"
+#include "needle.h"
 #include "parse_trigrams.h"
 #include "serializer.h"
 #include "turbopfor.h"
@@ -13,20 +15,18 @@
 #include <condition_variable>
 #include <deque>
 #include <fcntl.h>
-#include <fnmatch.h>
 #include <functional>
 #include <getopt.h>
 #include <inttypes.h>
 #include <iterator>
 #include <limits>
 #include <locale.h>
-#include <map>
 #include <memory>
 #include <mutex>
 #include <regex.h>
+#include <stdint.h>
 #include <stdio.h>
 #include <stdlib.h>
-#include <stdint.h>
 #include <string.h>
 #include <string>
 #include <string_view>
@@ -49,6 +49,7 @@ bool ignore_case = false;
 bool only_count = false;
 bool print_nul = false;
 bool use_debug = false;
+bool flush_cache = false;
 bool patterns_are_regex = false;
 bool use_extended_regex = false;
 int64_t limit_matches = numeric_limits<int64_t>::max();
@@ -57,110 +58,6 @@ int64_t limit_left = numeric_limits<int64_t>::max();
 steady_clock::time_point start;
 ZSTD_DDict *ddict = nullptr;
 
-regex_t compile_regex(const string &needle);
-
-struct Needle {
-	enum { STRSTR,
-	       REGEX,
-	       GLOB } type;
-	string str;  // Filled in no matter what.
-	regex_t re;  // For REGEX.
-};
-
-bool matches(const Needle &needle, const char *haystack)
-{
-	if (needle.type == Needle::STRSTR) {
-		return strstr(haystack, needle.str.c_str()) != nullptr;
-	} else if (needle.type == Needle::GLOB) {
-		int flags = ignore_case ? FNM_CASEFOLD : 0;
-		return fnmatch(needle.str.c_str(), haystack, flags) == 0;
-	} else {
-		assert(needle.type == Needle::REGEX);
-		return regexec(&needle.re, haystack, /*nmatch=*/0, /*pmatch=*/nullptr, /*flags=*/0) == 0;
-	}
-}
-
-class AccessRXCache {
-public:
-	AccessRXCache(IOUringEngine *engine)
-		: engine(engine) {}
-	void check_access(const char *filename, bool allow_async, function<void(bool)> cb);
-
-private:
-	unordered_map<string, bool> cache;
-	struct PendingStat {
-		string filename;
-		function<void(bool)> cb;
-	};
-	map<string, vector<PendingStat>> pending_stats;
-	IOUringEngine *engine;
-	mutex mu;
-};
-
-void AccessRXCache::check_access(const char *filename, bool allow_async, function<void(bool)> cb)
-{
-	lock_guard<mutex> lock(mu);
-	if (engine == nullptr || !engine->get_supports_stat()) {
-		allow_async = false;
-	}
-
-	for (const char *end = strchr(filename + 1, '/'); end != nullptr; end = strchr(end + 1, '/')) {
-		string parent_path(filename, end - filename);  // string_view from C++20.
-		auto cache_it = cache.find(parent_path);
-		if (cache_it != cache.end()) {
-			// Found in the cache.
-			if (!cache_it->second) {
-				cb(false);
-				return;
-			}
-			continue;
-		}
-
-		if (!allow_async) {
-			bool ok = access(parent_path.c_str(), R_OK | X_OK) == 0;
-			cache.emplace(parent_path, ok);
-			if (!ok) {
-				cb(false);
-				return;
-			}
-			continue;
-		}
-
-		// We want to call access(), but it could block on I/O. io_uring doesn't support
-		// access(), but we can do a dummy asynchonous statx() to populate the kernel's cache,
-		// which nearly always makes the next access() instantaneous.
-
-		// See if there's already a pending stat that matches this,
-		// or is a subdirectory.
-		auto it = pending_stats.lower_bound(parent_path);
-		if (it != pending_stats.end() && it->first.size() >= parent_path.size() &&
-		    it->first.compare(0, parent_path.size(), parent_path) == 0) {
-			it->second.emplace_back(PendingStat{ filename, move(cb) });
-		} else {
-			it = pending_stats.emplace(filename, vector<PendingStat>{}).first;
-			engine->submit_stat(filename, [this, it, filename{ strdup(filename) }, cb{ move(cb) }] {
-				// The stat returned, so now do the actual access() calls.
-				// All of them should be in cache, so don't fire off new statx()
-				// calls during that check.
-				check_access(filename, /*allow_async=*/false, move(cb));
-				free(filename);
-
-				// Call all others that waited for the same stat() to finish.
-				// They may fire off new stat() calls if needed.
-				vector<PendingStat> pending = move(it->second);
-				pending_stats.erase(it);
-				for (PendingStat &ps : pending) {
-					check_access(ps.filename.c_str(), /*allow_async=*/true, move(ps.cb));
-				}
-			});
-		}
-		return;  // The rest will happen in async context.
-	}
-
-	// Passed all checks.
-	cb(true);
-}
-
 class Corpus {
 public:
 	Corpus(int fd, IOUringEngine *engine);
@@ -184,8 +81,7 @@ public:
 Corpus::Corpus(int fd, IOUringEngine *engine)
 	: fd(fd), engine(engine)
 {
-	// Enable to test cold-cache behavior (except for access()).
-	if (false) {
+	if (flush_cache) {
 		off_t len = lseek(fd, 0, SEEK_END);
 		if (len == -1) {
 			perror("lseek");
@@ -404,7 +300,7 @@ uint64_t scan_all_docids(const vector<Needle> &needles, int fd, const Corpus &co
 	uint32_t num_blocks = corpus.get_num_filename_blocks();
 	unique_ptr<uint64_t[]> offsets(new uint64_t[num_blocks + 1]);
 	complete_pread(fd, offsets.get(), (num_blocks + 1) * sizeof(uint64_t), corpus.offset_for_block(0));
-	atomic<uint64_t> matched{0};
+	atomic<uint64_t> matched{ 0 };
 
 	mutex mu;
 	condition_variable queue_added, queue_removed;
@@ -737,41 +633,6 @@ void do_search_file(const vector<Needle> &needles, const char *filename)
 	}
 }
 
-string unescape_glob_to_plain_string(const string &needle)
-{
-	string unescaped;
-	for (size_t i = 0; i < needle.size(); i += read_unigram(needle, i).second) {
-		uint32_t ch = read_unigram(needle, i).first;
-		assert(ch != WILDCARD_UNIGRAM);
-		if (ch == PREMATURE_END_UNIGRAM) {
-			fprintf(stderr, "Pattern '%s' ended prematurely\n", needle.c_str());
-			exit(1);
-		}
-		unescaped.push_back(ch);
-	}
-	return unescaped;
-}
-
-regex_t compile_regex(const string &needle)
-{
-	regex_t re;
-	int flags = REG_NOSUB;
-	if (ignore_case) {
-		flags |= REG_ICASE;
-	}
-	if (use_extended_regex) {
-		flags |= REG_EXTENDED;
-	}
-	int err = regcomp(&re, needle.c_str(), flags);
-	if (err != 0) {
-		char errbuf[256];
-		regerror(err, &re, errbuf, sizeof(errbuf));
-		fprintf(stderr, "Error when compiling regex '%s': %s\n", needle.c_str(), errbuf);
-		exit(1);
-	}
-	return re;
-}
-
 void usage()
 {
 	printf(
@@ -802,6 +663,7 @@ void version()
 int main(int argc, char **argv)
 {
 	constexpr int EXTENDED_REGEX = 1000;
+	constexpr int FLUSH_CACHE = 1001;
 	static const struct option long_options[] = {
 		{ "help", no_argument, 0, 'h' },
 		{ "count", no_argument, 0, 'c' },
@@ -813,6 +675,8 @@ int main(int argc, char **argv)
 		{ "regexp", no_argument, 0, 'r' },
 		{ "regex", no_argument, 0, EXTENDED_REGEX },
 		{ "debug", no_argument, 0, 'D' },  // Not documented.
+		// Enable to test cold-cache behavior (except for access()). Not documented.
+		{ "flush-cache", no_argument, 0, FLUSH_CACHE },
 		{ 0, 0, 0, 0 }
 	};
 
@@ -857,6 +721,9 @@ int main(int argc, char **argv)
 		case 'D':
 			use_debug = true;
 			break;
+		case FLUSH_CACHE:
+			flush_cache = true;
+			break;
 		case 'V':
 			version();
 			break;
@@ -865,10 +732,12 @@ int main(int argc, char **argv)
 		}
 	}
 
-	if (use_debug) {
+	if (use_debug || flush_cache) {
 		// Debug information would leak information about which files exist,
 		// so drop setgid before we open the file; one would either need to run
-		// as root, or use a locally-built file.
+		// as root, or use a locally-built file. Doing the same thing for
+		// flush_cache is mostly paranoia, in an attempt to prevent random users
+		// from making plocate slow for everyone else.
 		if (setgid(getgid()) != 0) {
 			perror("setgid");
 			exit(EXIT_FAILURE);