Branch data Line data Source code
1 : : /* Debuginfo-over-http server.
2 : : Copyright (C) 2019-2021 Red Hat, Inc.
3 : : This file is part of elfutils.
4 : :
5 : : This file is free software; you can redistribute it and/or modify
6 : : it under the terms of the GNU General Public License as published by
7 : : the Free Software Foundation; either version 3 of the License, or
8 : : (at your option) any later version.
9 : :
10 : : elfutils is distributed in the hope that it will be useful, but
11 : : WITHOUT ANY WARRANTY; without even the implied warranty of
12 : : MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 : : GNU General Public License for more details.
14 : :
15 : : You should have received a copy of the GNU General Public License
16 : : along with this program. If not, see <http://www.gnu.org/licenses/>. */
17 : :
18 : :
19 : : /* cargo-cult from libdwfl linux-kernel-modules.c */
20 : : /* In case we have a bad fts we include this before config.h because it
21 : : can't handle _FILE_OFFSET_BITS.
22 : : Everything we need here is fine if its declarations just come first.
23 : : Also, include sys/types.h before fts. On some systems fts.h is not self
24 : : contained. */
25 : : #ifdef BAD_FTS
26 : : #include <sys/types.h>
27 : : #include <fts.h>
28 : : #endif
29 : :
30 : : #ifdef HAVE_CONFIG_H
31 : : #include "config.h"
32 : : #endif
33 : :
34 : : extern "C" {
35 : : #include "printversion.h"
36 : : }
37 : :
38 : : #include "debuginfod.h"
39 : : #include <dwarf.h>
40 : : #include <system.h>
41 : :
42 : : #include <argp.h>
43 : : #ifdef __GNUC__
44 : : #undef __attribute__ /* glibc bug - rhbz 1763325 */
45 : : #endif
46 : :
47 : : #include <unistd.h>
48 : : #include <stdlib.h>
49 : : #include <libintl.h>
50 : : #include <locale.h>
51 : : #include <pthread.h>
52 : : #include <signal.h>
53 : : #include <sys/stat.h>
54 : : #include <sys/time.h>
55 : : #include <sys/vfs.h>
56 : : #include <unistd.h>
57 : : #include <fcntl.h>
58 : : #include <netdb.h>
59 : :
60 : :
61 : : /* If fts.h is included before config.h, its indirect inclusions may not
62 : : give us the right LFS aliases of these functions, so map them manually. */
63 : : #ifdef BAD_FTS
64 : : #ifdef _FILE_OFFSET_BITS
65 : : #define open open64
66 : : #define fopen fopen64
67 : : #endif
68 : : #else
69 : : #include <sys/types.h>
70 : : #include <fts.h>
71 : : #endif
72 : :
73 : : #include <cstring>
74 : : #include <vector>
75 : : #include <set>
76 : : #include <map>
77 : : #include <string>
78 : : #include <iostream>
79 : : #include <iomanip>
80 : : #include <ostream>
81 : : #include <sstream>
82 : : #include <mutex>
83 : : #include <deque>
84 : : #include <condition_variable>
85 : : #include <thread>
86 : : // #include <regex> // on rhel7 gcc 4.8, not competent
87 : : #include <regex.h>
88 : : // #include <algorithm>
89 : : using namespace std;
90 : :
91 : : #include <gelf.h>
92 : : #include <libdwelf.h>
93 : :
94 : : #include <microhttpd.h>
95 : :
96 : : #if MHD_VERSION >= 0x00097002
97 : : // libmicrohttpd 0.9.71 broke API
98 : : #define MHD_RESULT enum MHD_Result
99 : : #else
100 : : #define MHD_RESULT int
101 : : #endif
102 : :
103 : : #include <curl/curl.h>
104 : : #include <archive.h>
105 : : #include <archive_entry.h>
106 : : #include <sqlite3.h>
107 : :
108 : : #ifdef __linux__
109 : : #include <sys/syscall.h>
110 : : #endif
111 : :
112 : : #ifdef __linux__
113 : : #define tid() syscall(SYS_gettid)
114 : : #else
115 : : #define tid() pthread_self()
116 : : #endif
117 : :
118 : :
119 : : inline bool
120 : 745 : string_endswith(const string& haystack, const string& needle)
121 : : {
122 [ + - ]: 1490 : return (haystack.size() >= needle.size() &&
123 [ + - + + ]: 745 : equal(haystack.end()-needle.size(), haystack.end(),
124 : 745 : needle.begin()));
125 : : }
126 : :
127 : :
128 : : // Roll this identifier for every sqlite schema incompatibility.
129 : : #define BUILDIDS "buildids9"
130 : :
131 : : #if SQLITE_VERSION_NUMBER >= 3008000
132 : : #define WITHOUT_ROWID "without rowid"
133 : : #else
134 : : #define WITHOUT_ROWID ""
135 : : #endif
136 : :
137 : : static const char DEBUGINFOD_SQLITE_DDL[] =
138 : : "pragma foreign_keys = on;\n"
139 : : "pragma synchronous = 0;\n" // disable fsync()s - this cache is disposable across a machine crash
140 : : "pragma journal_mode = wal;\n" // https://sqlite.org/wal.html
141 : : "pragma wal_checkpoint = truncate;\n" // clean out any preexisting wal file
142 : : "pragma journal_size_limit = 0;\n" // limit steady state file (between grooming, which also =truncate's)
143 : : "pragma auto_vacuum = incremental;\n" // https://sqlite.org/pragma.html
144 : : "pragma busy_timeout = 1000;\n" // https://sqlite.org/pragma.html
145 : : // NB: all these are overridable with -D option
146 : :
147 : : // Normalization table for interning file names
148 : : "create table if not exists " BUILDIDS "_files (\n"
149 : : " id integer primary key not null,\n"
150 : : " name text unique not null\n"
151 : : " );\n"
152 : : // Normalization table for interning buildids
153 : : "create table if not exists " BUILDIDS "_buildids (\n"
154 : : " id integer primary key not null,\n"
155 : : " hex text unique not null);\n"
156 : : // Track the completion of scanning of a given file & sourcetype at given time
157 : : "create table if not exists " BUILDIDS "_file_mtime_scanned (\n"
158 : : " mtime integer not null,\n"
159 : : " file integer not null,\n"
160 : : " size integer not null,\n" // in bytes
161 : : " sourcetype text(1) not null\n"
162 : : " check (sourcetype IN ('F', 'R')),\n"
163 : : " foreign key (file) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
164 : : " primary key (file, mtime, sourcetype)\n"
165 : : " ) " WITHOUT_ROWID ";\n"
166 : : "create table if not exists " BUILDIDS "_f_de (\n"
167 : : " buildid integer not null,\n"
168 : : " debuginfo_p integer not null,\n"
169 : : " executable_p integer not null,\n"
170 : : " file integer not null,\n"
171 : : " mtime integer not null,\n"
172 : : " foreign key (file) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
173 : : " foreign key (buildid) references " BUILDIDS "_buildids(id) on update cascade on delete cascade,\n"
174 : : " primary key (buildid, file, mtime)\n"
175 : : " ) " WITHOUT_ROWID ";\n"
176 : : "create table if not exists " BUILDIDS "_f_s (\n"
177 : : " buildid integer not null,\n"
178 : : " artifactsrc integer not null,\n"
179 : : " file integer not null,\n" // NB: not necessarily entered into _mtime_scanned
180 : : " mtime integer not null,\n"
181 : : " foreign key (file) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
182 : : " foreign key (artifactsrc) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
183 : : " foreign key (buildid) references " BUILDIDS "_buildids(id) on update cascade on delete cascade,\n"
184 : : " primary key (buildid, artifactsrc, file, mtime)\n"
185 : : " ) " WITHOUT_ROWID ";\n"
186 : : "create table if not exists " BUILDIDS "_r_de (\n"
187 : : " buildid integer not null,\n"
188 : : " debuginfo_p integer not null,\n"
189 : : " executable_p integer not null,\n"
190 : : " file integer not null,\n"
191 : : " mtime integer not null,\n"
192 : : " content integer not null,\n"
193 : : " foreign key (file) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
194 : : " foreign key (content) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
195 : : " foreign key (buildid) references " BUILDIDS "_buildids(id) on update cascade on delete cascade,\n"
196 : : " primary key (buildid, debuginfo_p, executable_p, file, content, mtime)\n"
197 : : " ) " WITHOUT_ROWID ";\n"
198 : : "create table if not exists " BUILDIDS "_r_sref (\n" // outgoing dwarf sourcefile references from rpm
199 : : " buildid integer not null,\n"
200 : : " artifactsrc integer not null,\n"
201 : : " foreign key (artifactsrc) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
202 : : " foreign key (buildid) references " BUILDIDS "_buildids(id) on update cascade on delete cascade,\n"
203 : : " primary key (buildid, artifactsrc)\n"
204 : : " ) " WITHOUT_ROWID ";\n"
205 : : "create table if not exists " BUILDIDS "_r_sdef (\n" // rpm contents that may satisfy sref
206 : : " file integer not null,\n"
207 : : " mtime integer not null,\n"
208 : : " content integer not null,\n"
209 : : " foreign key (file) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
210 : : " foreign key (content) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
211 : : " primary key (content, file, mtime)\n"
212 : : " ) " WITHOUT_ROWID ";\n"
213 : : // create views to glue together some of the above tables, for webapi D queries
214 : : "create view if not exists " BUILDIDS "_query_d as \n"
215 : : "select\n"
216 : : " b.hex as buildid, n.mtime, 'F' as sourcetype, f0.name as source0, n.mtime as mtime, null as source1\n"
217 : : " from " BUILDIDS "_buildids b, " BUILDIDS "_files f0, " BUILDIDS "_f_de n\n"
218 : : " where b.id = n.buildid and f0.id = n.file and n.debuginfo_p = 1\n"
219 : : "union all select\n"
220 : : " b.hex as buildid, n.mtime, 'R' as sourcetype, f0.name as source0, n.mtime as mtime, f1.name as source1\n"
221 : : " from " BUILDIDS "_buildids b, " BUILDIDS "_files f0, " BUILDIDS "_files f1, " BUILDIDS "_r_de n\n"
222 : : " where b.id = n.buildid and f0.id = n.file and f1.id = n.content and n.debuginfo_p = 1\n"
223 : : ";"
224 : : // ... and for E queries
225 : : "create view if not exists " BUILDIDS "_query_e as \n"
226 : : "select\n"
227 : : " b.hex as buildid, n.mtime, 'F' as sourcetype, f0.name as source0, n.mtime as mtime, null as source1\n"
228 : : " from " BUILDIDS "_buildids b, " BUILDIDS "_files f0, " BUILDIDS "_f_de n\n"
229 : : " where b.id = n.buildid and f0.id = n.file and n.executable_p = 1\n"
230 : : "union all select\n"
231 : : " b.hex as buildid, n.mtime, 'R' as sourcetype, f0.name as source0, n.mtime as mtime, f1.name as source1\n"
232 : : " from " BUILDIDS "_buildids b, " BUILDIDS "_files f0, " BUILDIDS "_files f1, " BUILDIDS "_r_de n\n"
233 : : " where b.id = n.buildid and f0.id = n.file and f1.id = n.content and n.executable_p = 1\n"
234 : : ";"
235 : : // ... and for S queries
236 : : "create view if not exists " BUILDIDS "_query_s as \n"
237 : : "select\n"
238 : : " b.hex as buildid, fs.name as artifactsrc, 'F' as sourcetype, f0.name as source0, n.mtime as mtime, null as source1, null as source0ref\n"
239 : : " from " BUILDIDS "_buildids b, " BUILDIDS "_files f0, " BUILDIDS "_files fs, " BUILDIDS "_f_s n\n"
240 : : " where b.id = n.buildid and f0.id = n.file and fs.id = n.artifactsrc\n"
241 : : "union all select\n"
242 : : " b.hex as buildid, f1.name as artifactsrc, 'R' as sourcetype, f0.name as source0, sd.mtime as mtime, f1.name as source1, fsref.name as source0ref\n"
243 : : " from " BUILDIDS "_buildids b, " BUILDIDS "_files f0, " BUILDIDS "_files f1, " BUILDIDS "_files fsref, "
244 : : " " BUILDIDS "_r_sdef sd, " BUILDIDS "_r_sref sr, " BUILDIDS "_r_de sde\n"
245 : : " where b.id = sr.buildid and f0.id = sd.file and fsref.id = sde.file and f1.id = sd.content\n"
246 : : " and sr.artifactsrc = sd.content and sde.buildid = sr.buildid\n"
247 : : ";"
248 : : // and for startup overview counts
249 : : "drop view if exists " BUILDIDS "_stats;\n"
250 : : "create view if not exists " BUILDIDS "_stats as\n"
251 : : " select 'file d/e' as label,count(*) as quantity from " BUILDIDS "_f_de\n"
252 : : "union all select 'file s',count(*) from " BUILDIDS "_f_s\n"
253 : : "union all select 'archive d/e',count(*) from " BUILDIDS "_r_de\n"
254 : : "union all select 'archive sref',count(*) from " BUILDIDS "_r_sref\n"
255 : : "union all select 'archive sdef',count(*) from " BUILDIDS "_r_sdef\n"
256 : : "union all select 'buildids',count(*) from " BUILDIDS "_buildids\n"
257 : : "union all select 'filenames',count(*) from " BUILDIDS "_files\n"
258 : : "union all select 'files scanned (#)',count(*) from " BUILDIDS "_file_mtime_scanned\n"
259 : : "union all select 'files scanned (mb)',coalesce(sum(size)/1024/1024,0) from " BUILDIDS "_file_mtime_scanned\n"
260 : : #if SQLITE_VERSION_NUMBER >= 3016000
261 : : "union all select 'index db size (mb)',page_count*page_size/1024/1024 as size FROM pragma_page_count(), pragma_page_size()\n"
262 : : #endif
263 : : ";\n"
264 : :
265 : : // schema change history & garbage collection
266 : : //
267 : : // XXX: we could have migration queries here to bring prior-schema
268 : : // data over instead of just dropping it.
269 : : //
270 : : // buildids9: widen the mtime_scanned table
271 : : "" // <<< we are here
272 : : // buildids8: slim the sref table
273 : : "drop table if exists buildids8_f_de;\n"
274 : : "drop table if exists buildids8_f_s;\n"
275 : : "drop table if exists buildids8_r_de;\n"
276 : : "drop table if exists buildids8_r_sref;\n"
277 : : "drop table if exists buildids8_r_sdef;\n"
278 : : "drop table if exists buildids8_file_mtime_scanned;\n"
279 : : "drop table if exists buildids8_files;\n"
280 : : "drop table if exists buildids8_buildids;\n"
281 : : // buildids7: separate _norm table into dense subtype tables
282 : : "drop table if exists buildids7_f_de;\n"
283 : : "drop table if exists buildids7_f_s;\n"
284 : : "drop table if exists buildids7_r_de;\n"
285 : : "drop table if exists buildids7_r_sref;\n"
286 : : "drop table if exists buildids7_r_sdef;\n"
287 : : "drop table if exists buildids7_file_mtime_scanned;\n"
288 : : "drop table if exists buildids7_files;\n"
289 : : "drop table if exists buildids7_buildids;\n"
290 : : // buildids6: drop bolo/rfolo again, represent sources / rpmcontents in main table
291 : : "drop table if exists buildids6_norm;\n"
292 : : "drop table if exists buildids6_files;\n"
293 : : "drop table if exists buildids6_buildids;\n"
294 : : "drop view if exists buildids6;\n"
295 : : // buildids5: redefine srcfile1 column to be '.'-less (for rpms)
296 : : "drop table if exists buildids5_norm;\n"
297 : : "drop table if exists buildids5_files;\n"
298 : : "drop table if exists buildids5_buildids;\n"
299 : : "drop table if exists buildids5_bolo;\n"
300 : : "drop table if exists buildids5_rfolo;\n"
301 : : "drop view if exists buildids5;\n"
302 : : // buildids4: introduce rpmfile RFOLO
303 : : "drop table if exists buildids4_norm;\n"
304 : : "drop table if exists buildids4_files;\n"
305 : : "drop table if exists buildids4_buildids;\n"
306 : : "drop table if exists buildids4_bolo;\n"
307 : : "drop table if exists buildids4_rfolo;\n"
308 : : "drop view if exists buildids4;\n"
309 : : // buildids3*: split out srcfile BOLO
310 : : "drop table if exists buildids3_norm;\n"
311 : : "drop table if exists buildids3_files;\n"
312 : : "drop table if exists buildids3_buildids;\n"
313 : : "drop table if exists buildids3_bolo;\n"
314 : : "drop view if exists buildids3;\n"
315 : : // buildids2: normalized buildid and filenames into interning tables;
316 : : "drop table if exists buildids2_norm;\n"
317 : : "drop table if exists buildids2_files;\n"
318 : : "drop table if exists buildids2_buildids;\n"
319 : : "drop view if exists buildids2;\n"
320 : : // buildids1: made buildid and artifacttype NULLable, to represent cached-negative
321 : : // lookups from sources, e.g. files or rpms that contain no buildid-indexable content
322 : : "drop table if exists buildids1;\n"
323 : : // buildids: original
324 : : "drop table if exists buildids;\n"
325 : : ;
326 : :
327 : : static const char DEBUGINFOD_SQLITE_CLEANUP_DDL[] =
328 : : "pragma wal_checkpoint = truncate;\n" // clean out any preexisting wal file
329 : : ;
330 : :
331 : :
332 : :
333 : :
334 : : /* Name and version of program. */
335 : : /* ARGP_PROGRAM_VERSION_HOOK_DEF = print_version; */ // not this simple for C++
336 : :
337 : : /* Bug report address. */
338 : : ARGP_PROGRAM_BUG_ADDRESS_DEF = PACKAGE_BUGREPORT;
339 : :
340 : : /* Definitions of arguments for argp functions. */
341 : : static const struct argp_option options[] =
342 : : {
343 : : { NULL, 0, NULL, 0, "Scanners:", 1 },
344 : : { "scan-file-dir", 'F', NULL, 0, "Enable ELF/DWARF file scanning.", 0 },
345 : : { "scan-rpm-dir", 'R', NULL, 0, "Enable RPM scanning.", 0 },
346 : : { "scan-deb-dir", 'U', NULL, 0, "Enable DEB scanning.", 0 },
347 : : { "scan-archive", 'Z', "EXT=CMD", 0, "Enable arbitrary archive scanning.", 0 },
348 : : // "source-oci-imageregistry" ...
349 : :
350 : : { NULL, 0, NULL, 0, "Options:", 2 },
351 : : { "logical", 'L', NULL, 0, "Follow symlinks, default=ignore.", 0 },
352 : : { "rescan-time", 't', "SECONDS", 0, "Number of seconds to wait between rescans, 0=disable.", 0 },
353 : : { "groom-time", 'g', "SECONDS", 0, "Number of seconds to wait between database grooming, 0=disable.", 0 },
354 : : { "maxigroom", 'G', NULL, 0, "Run a complete database groom/shrink pass at startup.", 0 },
355 : : { "concurrency", 'c', "NUM", 0, "Limit scanning thread concurrency to NUM.", 0 },
356 : : { "include", 'I', "REGEX", 0, "Include files matching REGEX, default=all.", 0 },
357 : : { "exclude", 'X', "REGEX", 0, "Exclude files matching REGEX, default=none.", 0 },
358 : : { "port", 'p', "NUM", 0, "HTTP port to listen on, default 8002.", 0 },
359 : : { "database", 'd', "FILE", 0, "Path to sqlite database.", 0 },
360 : : { "ddl", 'D', "SQL", 0, "Apply extra sqlite ddl/pragma to connection.", 0 },
361 : : { "verbose", 'v', NULL, 0, "Increase verbosity.", 0 },
362 : : { "regex-groom", 'r', NULL, 0,"Uses regexes from -I and -X arguments to groom the database.",0},
363 : : #define ARGP_KEY_FDCACHE_FDS 0x1001
364 : : { "fdcache-fds", ARGP_KEY_FDCACHE_FDS, "NUM", 0, "Maximum number of archive files to keep in fdcache.", 0 },
365 : : #define ARGP_KEY_FDCACHE_MBS 0x1002
366 : : { "fdcache-mbs", ARGP_KEY_FDCACHE_MBS, "MB", 0, "Maximum total size of archive file fdcache.", 0 },
367 : : #define ARGP_KEY_FDCACHE_PREFETCH 0x1003
368 : : { "fdcache-prefetch", ARGP_KEY_FDCACHE_PREFETCH, "NUM", 0, "Number of archive files to prefetch into fdcache.", 0 },
369 : : #define ARGP_KEY_FDCACHE_MINTMP 0x1004
370 : : { "fdcache-mintmp", ARGP_KEY_FDCACHE_MINTMP, "NUM", 0, "Minimum free space% on tmpdir.", 0 },
371 : : #define ARGP_KEY_FDCACHE_PREFETCH_MBS 0x1005
372 : : { "fdcache-prefetch-mbs", ARGP_KEY_FDCACHE_PREFETCH_MBS, "MB", 0,"Megabytes allocated to the \
373 : : prefetch cache.", 0},
374 : : #define ARGP_KEY_FDCACHE_PREFETCH_FDS 0x1006
375 : : { "fdcache-prefetch-fds", ARGP_KEY_FDCACHE_PREFETCH_FDS, "NUM", 0,"Number of files allocated to the \
376 : : prefetch cache.", 0},
377 : : #define ARGP_KEY_FORWARDED_TTL_LIMIT 0x1007
378 : : {"forwarded-ttl-limit", ARGP_KEY_FORWARDED_TTL_LIMIT, "NUM", 0, "Limit of X-Forwarded-For hops, default 8.", 0},
379 : : #define ARGP_KEY_PASSIVE 0x1008
380 : : { "passive", ARGP_KEY_PASSIVE, NULL, 0, "Do not scan or groom, read-only database.", 0 },
381 : : { NULL, 0, NULL, 0, NULL, 0 },
382 : : };
383 : :
384 : : /* Short description of program. */
385 : : static const char doc[] = "Serve debuginfo-related content across HTTP from files under PATHs.";
386 : :
387 : : /* Strings for arguments in help texts. */
388 : : static const char args_doc[] = "[PATH ...]";
389 : :
390 : : /* Prototype for option handler. */
391 : : static error_t parse_opt (int key, char *arg, struct argp_state *state);
392 : :
393 : : /* Data structure to communicate with argp functions. */
394 : : static struct argp argp =
395 : : {
396 : : options, parse_opt, args_doc, doc, NULL, NULL, NULL
397 : : };
398 : :
399 : :
400 : : static string db_path;
401 : : static sqlite3 *db; // single connection, serialized across all our threads!
402 : : static sqlite3 *dbq; // webapi query-servicing readonly connection, serialized ditto!
403 : : static unsigned verbose;
404 : : static volatile sig_atomic_t interrupted = 0;
405 : : static volatile sig_atomic_t forced_rescan_count = 0;
406 : : static volatile sig_atomic_t sigusr1 = 0;
407 : : static volatile sig_atomic_t forced_groom_count = 0;
408 : : static volatile sig_atomic_t sigusr2 = 0;
409 : : static unsigned http_port = 8002;
410 : : static unsigned rescan_s = 300;
411 : : static unsigned groom_s = 86400;
412 : : static bool maxigroom = false;
413 : : static unsigned concurrency = std::thread::hardware_concurrency() ?: 1;
414 : : static set<string> source_paths;
415 : : static bool scan_files = false;
416 : : static map<string,string> scan_archives;
417 : : static vector<string> extra_ddl;
418 : : static regex_t file_include_regex;
419 : : static regex_t file_exclude_regex;
420 : : static bool regex_groom = false;
421 : : static bool traverse_logical;
422 : : static long fdcache_fds;
423 : : static long fdcache_mbs;
424 : : static long fdcache_prefetch;
425 : : static long fdcache_mintmp;
426 : : static long fdcache_prefetch_mbs;
427 : : static long fdcache_prefetch_fds;
428 : : static unsigned forwarded_ttl_limit = 8;
429 : : static string tmpdir;
430 : : static bool passive_p = false;
431 : :
432 : : static void set_metric(const string& key, double value);
433 : : // static void inc_metric(const string& key);
434 : : static void set_metric(const string& metric,
435 : : const string& lname, const string& lvalue,
436 : : double value);
437 : : static void inc_metric(const string& metric,
438 : : const string& lname, const string& lvalue);
439 : : static void add_metric(const string& metric,
440 : : const string& lname, const string& lvalue,
441 : : double value);
442 : : static void inc_metric(const string& metric,
443 : : const string& lname, const string& lvalue,
444 : : const string& rname, const string& rvalue);
445 : : static void add_metric(const string& metric,
446 : : const string& lname, const string& lvalue,
447 : : const string& rname, const string& rvalue,
448 : : double value);
449 : :
450 : :
451 : : class tmp_inc_metric { // a RAII style wrapper for exception-safe scoped increment & decrement
452 : : string m, n, v;
453 : : public:
454 : 610 : tmp_inc_metric(const string& mname, const string& lname, const string& lvalue):
455 [ + - + - ]: 610 : m(mname), n(lname), v(lvalue)
456 : : {
457 [ + - ]: 610 : add_metric (m, n, v, 1);
458 : 610 : }
459 : 610 : ~tmp_inc_metric()
460 : 610 : {
461 : 610 : add_metric (m, n, v, -1);
462 : 610 : }
463 : : };
464 : :
465 : : class tmp_ms_metric { // a RAII style wrapper for exception-safe scoped timing
466 : : string m, n, v;
467 : : struct timespec ts_start;
468 : : public:
469 : 14862 : tmp_ms_metric(const string& mname, const string& lname, const string& lvalue):
470 [ + - + - ]: 14862 : m(mname), n(lname), v(lvalue)
471 : : {
472 : 14864 : clock_gettime (CLOCK_MONOTONIC, & ts_start);
473 : 14862 : }
474 : 14863 : ~tmp_ms_metric()
475 : 14863 : {
476 : : struct timespec ts_end;
477 : 14863 : clock_gettime (CLOCK_MONOTONIC, & ts_end);
478 : 14864 : double deltas = (ts_end.tv_sec - ts_start.tv_sec)
479 : 14864 : + (ts_end.tv_nsec - ts_start.tv_nsec)/1.e9;
480 : :
481 : 14864 : add_metric (m + "_milliseconds_sum", n, v, (deltas*1000.0));
482 : 14864 : inc_metric (m + "_milliseconds_count", n, v);
483 : 14863 : }
484 : : };
485 : :
486 : :
487 : : /* Handle program arguments. */
488 : : static error_t
489 : 408 : parse_opt (int key, char *arg,
490 : : struct argp_state *state __attribute__ ((unused)))
491 : : {
492 : : int rc;
493 [ + + + + : 408 : switch (key)
+ + + + -
+ + - - +
+ + + + -
+ + + + +
+ + ]
494 : : {
495 : 99 : case 'v': verbose ++; break;
496 : 28 : case 'd':
497 : : /* When using the in-memory database make sure it is shareable,
498 : : so we can open it twice as read/write and read-only. */
499 [ + + ]: 28 : if (strcmp (arg, ":memory:") == 0)
500 : 3 : db_path = "file::memory:?cache=shared";
501 : : else
502 [ + - ]: 25 : db_path = string(arg);
503 : 28 : break;
504 : 28 : case 'p': http_port = (unsigned) atoi(arg);
505 [ + - - + ]: 28 : if (http_port == 0 || http_port > 65535)
506 : 0 : argp_failure(state, 1, EINVAL, "port number");
507 : 28 : break;
508 : 23 : case 'F': scan_files = true; break;
509 : 7 : case 'R':
510 [ + - + - : 7 : scan_archives[".rpm"]="cat"; // libarchive groks rpm natively
+ - ]
511 : 7 : break;
512 : 7 : case 'U':
513 [ + - + - : 7 : scan_archives[".deb"]="(bsdtar -O -x -f - data.tar\\*)<";
+ - ]
514 [ + - + - : 7 : scan_archives[".ddeb"]="(bsdtar -O -x -f - data.tar\\*)<";
+ - ]
515 [ + - + - : 7 : scan_archives[".ipk"]="(bsdtar -O -x -f - data.tar\\*)<";
+ - ]
516 : : // .udeb too?
517 : 7 : break;
518 : 6 : case 'Z':
519 : : {
520 : 6 : char* extension = strchr(arg, '=');
521 [ - + ]: 6 : if (arg[0] == '\0')
522 : 0 : argp_failure(state, 1, EINVAL, "missing EXT");
523 [ + + ]: 6 : else if (extension)
524 [ + - + - : 3 : scan_archives[string(arg, (extension-arg))]=string(extension+1);
+ - ]
525 : : else
526 [ + - + - : 3 : scan_archives[string(arg)]=string("cat");
+ - ]
527 : : }
528 : 6 : break;
529 : 4 : case 'L':
530 [ - + ]: 4 : if (passive_p)
531 : 0 : argp_failure(state, 1, EINVAL, "-L option inconsistent with passive mode");
532 : 4 : traverse_logical = true;
533 : 4 : break;
534 : 0 : case 'D':
535 [ # # ]: 0 : if (passive_p)
536 : 0 : argp_failure(state, 1, EINVAL, "-D option inconsistent with passive mode");
537 [ # # # # ]: 0 : extra_ddl.push_back(string(arg));
538 : 0 : break;
539 : 20 : case 't':
540 [ - + ]: 20 : if (passive_p)
541 : 0 : argp_failure(state, 1, EINVAL, "-t option inconsistent with passive mode");
542 : 20 : rescan_s = (unsigned) atoi(arg);
543 : 20 : break;
544 : 20 : case 'g':
545 [ - + ]: 20 : if (passive_p)
546 : 0 : argp_failure(state, 1, EINVAL, "-g option inconsistent with passive mode");
547 : 20 : groom_s = (unsigned) atoi(arg);
548 : 20 : break;
549 : 0 : case 'G':
550 [ # # ]: 0 : if (passive_p)
551 : 0 : argp_failure(state, 1, EINVAL, "-G option inconsistent with passive mode");
552 : 0 : maxigroom = true;
553 : 0 : break;
554 : 0 : case 'c':
555 [ # # ]: 0 : if (passive_p)
556 : 0 : argp_failure(state, 1, EINVAL, "-c option inconsistent with passive mode");
557 : 0 : concurrency = (unsigned) atoi(arg);
558 [ # # ]: 0 : if (concurrency < 1) concurrency = 1;
559 : 0 : break;
560 : 1 : case 'I':
561 : : // NB: no problem with unconditional free here - an earlier failed regcomp would exit program
562 [ - + ]: 1 : if (passive_p)
563 : 0 : argp_failure(state, 1, EINVAL, "-I option inconsistent with passive mode");
564 : 1 : regfree (&file_include_regex);
565 : 1 : rc = regcomp (&file_include_regex, arg, REG_EXTENDED|REG_NOSUB);
566 [ - + ]: 1 : if (rc != 0)
567 : 0 : argp_failure(state, 1, EINVAL, "regular expression");
568 : 1 : break;
569 : 1 : case 'X':
570 [ - + ]: 1 : if (passive_p)
571 : 0 : argp_failure(state, 1, EINVAL, "-X option inconsistent with passive mode");
572 : 1 : regfree (&file_exclude_regex);
573 : 1 : rc = regcomp (&file_exclude_regex, arg, REG_EXTENDED|REG_NOSUB);
574 [ - + ]: 1 : if (rc != 0)
575 : 0 : argp_failure(state, 1, EINVAL, "regular expression");
576 : 1 : break;
577 : 1 : case 'r':
578 [ - + ]: 1 : if (passive_p)
579 : 0 : argp_failure(state, 1, EINVAL, "-r option inconsistent with passive mode");
580 : 1 : regex_groom = true;
581 : 1 : break;
582 : 1 : case ARGP_KEY_FDCACHE_FDS:
583 : 1 : fdcache_fds = atol (arg);
584 : 1 : break;
585 : 1 : case ARGP_KEY_FDCACHE_MBS:
586 : 1 : fdcache_mbs = atol (arg);
587 : 1 : break;
588 : 0 : case ARGP_KEY_FDCACHE_PREFETCH:
589 : 0 : fdcache_prefetch = atol (arg);
590 : 0 : break;
591 : 1 : case ARGP_KEY_FDCACHE_MINTMP:
592 : 1 : fdcache_mintmp = atol (arg);
593 [ + - - + ]: 1 : if( fdcache_mintmp > 100 || fdcache_mintmp < 0 )
594 : 0 : argp_failure(state, 1, EINVAL, "fdcache mintmp percent");
595 : 1 : break;
596 : 2 : case ARGP_KEY_FORWARDED_TTL_LIMIT:
597 : 2 : forwarded_ttl_limit = (unsigned) atoi(arg);
598 : 2 : break;
599 : 38 : case ARGP_KEY_ARG:
600 [ + - + - ]: 38 : source_paths.insert(string(arg));
601 : 38 : break;
602 : 1 : case ARGP_KEY_FDCACHE_PREFETCH_FDS:
603 : 1 : fdcache_prefetch_fds = atol(arg);
604 [ - + ]: 1 : if ( fdcache_prefetch_fds < 0)
605 : 0 : argp_failure(state, 1, EINVAL, "fdcache prefetch fds");
606 : 1 : break;
607 : 1 : case ARGP_KEY_FDCACHE_PREFETCH_MBS:
608 : 1 : fdcache_prefetch_mbs = atol(arg);
609 [ - + ]: 1 : if ( fdcache_prefetch_mbs < 0)
610 : 0 : argp_failure(state, 1, EINVAL, "fdcache prefetch mbs");
611 : 1 : break;
612 : 1 : case ARGP_KEY_PASSIVE:
613 : 1 : passive_p = true;
614 : 1 : if (source_paths.size() > 0
615 [ + - ]: 1 : || maxigroom
616 [ + - ]: 1 : || extra_ddl.size() > 0
617 [ + - - + : 2 : || traverse_logical)
- + ]
618 : : // other conflicting options tricky to check
619 : 0 : argp_failure(state, 1, EINVAL, "inconsistent options with passive mode");
620 : 1 : break;
621 : : // case 'h': argp_state_help (state, stderr, ARGP_HELP_LONG|ARGP_HELP_EXIT_OK);
622 : 117 : default: return ARGP_ERR_UNKNOWN;
623 : : }
624 : :
625 : 291 : return 0;
626 : : }
627 : :
628 : :
629 : : ////////////////////////////////////////////////////////////////////////
630 : :
631 : :
632 : : // represent errors that may get reported to an ostream and/or a libmicrohttpd connection
633 : :
634 : : struct reportable_exception
635 : : {
636 : : int code;
637 : : string message;
638 : :
639 : 126 : reportable_exception(int c, const string& m): code(c), message(m) {}
640 : 154 : reportable_exception(const string& m): code(503), message(m) {}
641 : : reportable_exception(): code(503), message() {}
642 : :
643 : : void report(ostream& o) const; // defined under obatched() class below
644 : :
645 : 243 : MHD_RESULT mhd_send_response(MHD_Connection* c) const {
646 : 486 : MHD_Response* r = MHD_create_response_from_buffer (message.size(),
647 : 243 : (void*) message.c_str(),
648 : : MHD_RESPMEM_MUST_COPY);
649 : 243 : MHD_add_response_header (r, "Content-Type", "text/plain");
650 : 243 : MHD_RESULT rc = MHD_queue_response (c, code, r);
651 : 243 : MHD_destroy_response (r);
652 : 243 : return rc;
653 : : }
654 : : };
655 : :
656 : :
657 : : struct sqlite_exception: public reportable_exception
658 : : {
659 : 7 : sqlite_exception(int rc, const string& msg):
660 [ + - + - : 7 : reportable_exception(string("sqlite3 error: ") + msg + ": " + string(sqlite3_errstr(rc) ?: "?")) {
+ - + - +
- + - + -
+ - ]
661 [ + - + - : 7 : inc_metric("error_count","sqlite3",sqlite3_errstr(rc));
+ - + - +
- ]
662 : 7 : }
663 : : };
664 : :
665 : : struct libc_exception: public reportable_exception
666 : : {
667 : 141 : libc_exception(int rc, const string& msg):
668 [ + - + - : 141 : reportable_exception(string("libc error: ") + msg + ": " + string(strerror(rc) ?: "?")) {
+ - + - +
- + - +
- ]
669 [ + - + - : 141 : inc_metric("error_count","libc",strerror(rc));
+ - + - ]
670 : 141 : }
671 : : };
672 : :
673 : :
674 : : struct archive_exception: public reportable_exception
675 : : {
676 : 0 : archive_exception(const string& msg):
677 [ # # # # : 0 : reportable_exception(string("libarchive error: ") + msg) {
# # ]
678 [ # # # # : 0 : inc_metric("error_count","libarchive",msg);
# # ]
679 : 0 : }
680 : 0 : archive_exception(struct archive* a, const string& msg):
681 [ # # # # : 0 : reportable_exception(string("libarchive error: ") + msg + ": " + string(archive_error_string(a) ?: "?")) {
# # # # #
# # # # #
# # ]
682 [ # # # # : 0 : inc_metric("error_count","libarchive",msg + ": " + string(archive_error_string(a) ?: "?"));
# # # # #
# # # # #
# # ]
683 : 0 : }
684 : : };
685 : :
686 : :
687 : : struct elfutils_exception: public reportable_exception
688 : : {
689 : 0 : elfutils_exception(int rc, const string& msg):
690 [ # # # # : 0 : reportable_exception(string("elfutils error: ") + msg + ": " + string(elf_errmsg(rc) ?: "?")) {
# # # # #
# # # # #
# # ]
691 [ # # # # : 0 : inc_metric("error_count","elfutils",elf_errmsg(rc));
# # # # #
# ]
692 : 0 : }
693 : : };
694 : :
695 : :
696 : : ////////////////////////////////////////////////////////////////////////
697 : :
698 : : template <typename Payload>
699 : : class workq
700 : : {
701 : : set<Payload> q; // eliminate duplicates
702 : : mutex mtx;
703 : : condition_variable cv;
704 : : bool dead;
705 : : unsigned idlers; // number of threads busy with wait_idle / done_idle
706 : : unsigned fronters; // number of threads busy with wait_front / done_front
707 : :
708 : : public:
709 : 28 : workq() { dead = false; idlers = 0; fronters = 0; }
710 : 28 : ~workq() {}
711 : :
712 : 346 : void push_back(const Payload& p)
713 : : {
714 [ + - ]: 692 : unique_lock<mutex> lock(mtx);
715 [ + - ]: 346 : q.insert (p);
716 [ + - + - : 346 : set_metric("thread_work_pending","role","scan", q.size());
+ - + - ]
717 : 346 : cv.notify_all();
718 : 346 : }
719 : :
720 : : // kill this workqueue, wake up all idlers / scanners
721 : 28 : void nuke() {
722 [ + - ]: 56 : unique_lock<mutex> lock(mtx);
723 : : // optional: q.clear();
724 : 28 : dead = true;
725 : 28 : cv.notify_all();
726 : 28 : }
727 : :
728 : : // clear the workqueue, when scanning is interrupted with USR2
729 : 0 : void clear() {
730 [ # # ]: 0 : unique_lock<mutex> lock(mtx);
731 : 0 : q.clear();
732 [ # # # # : 0 : set_metric("thread_work_pending","role","scan", q.size());
# # # # ]
733 : : // NB: there may still be some live fronters
734 : 0 : cv.notify_all(); // maybe wake up waiting idlers
735 : 0 : }
736 : :
737 : : // block this scanner thread until there is work to do and no active idler
738 : 442 : bool wait_front (Payload& p)
739 : : {
740 [ + - ]: 884 : unique_lock<mutex> lock(mtx);
741 [ + + + + : 2048 : while (!dead && (q.size() == 0 || idlers > 0))
+ + + + ]
742 : 1606 : cv.wait(lock);
743 [ + + ]: 442 : if (dead)
744 : 96 : return false;
745 : : else
746 : : {
747 [ + - ]: 346 : p = * q.begin();
748 [ + - ]: 346 : q.erase (q.begin());
749 : 346 : fronters ++; // prevent idlers from starting awhile, even if empty q
750 [ + - + - : 346 : set_metric("thread_work_pending","role","scan", q.size());
+ - + - ]
751 : : // NB: don't wake up idlers yet! The consumer is busy
752 : : // processing this element until it calls done_front().
753 : 346 : return true;
754 : : }
755 : : }
756 : :
757 : : // notify waitq that scanner thread is done with that last item
758 : 346 : void done_front ()
759 : : {
760 [ + - ]: 692 : unique_lock<mutex> lock(mtx);
761 : 346 : fronters --;
762 [ + + + + : 346 : if (q.size() == 0 && fronters == 0)
+ + ]
763 : 39 : cv.notify_all(); // maybe wake up waiting idlers
764 : 346 : }
765 : :
766 : : // block this idler thread until there is no work to do
767 : 240 : void wait_idle ()
768 : : {
769 [ + - ]: 240 : unique_lock<mutex> lock(mtx);
770 : 240 : cv.notify_all(); // maybe wake up waiting scanners
771 [ + + + + : 265 : while (!dead && ((q.size() != 0) || fronters > 0))
+ + + + ]
772 : 25 : cv.wait(lock);
773 : 240 : idlers ++;
774 : 240 : }
775 : :
776 : 213 : void done_idle ()
777 : : {
778 [ + - ]: 426 : unique_lock<mutex> lock(mtx);
779 : 213 : idlers --;
780 : 213 : cv.notify_all(); // maybe wake up waiting scanners, but probably not (shutting down)
781 : 213 : }
782 : : };
783 : :
784 : : typedef struct stat stat_t;
785 : : typedef pair<string,stat_t> scan_payload;
786 : 819 : inline bool operator< (const scan_payload& a, const scan_payload& b)
787 : : {
788 : 819 : return a.first < b.first; // don't bother compare the stat fields
789 : : }
790 : : static workq<scan_payload> scanq; // just a single one
791 : : // producer & idler: thread_main_fts_source_paths()
792 : : // consumer: thread_main_scanner()
793 : : // idler: thread_main_groom()
794 : :
795 : :
796 : : ////////////////////////////////////////////////////////////////////////
797 : :
798 : : // Unique set is a thread-safe structure that lends 'ownership' of a value
799 : : // to a thread. Other threads requesting the same thing are made to wait.
800 : : // It's like a semaphore-on-demand.
801 : : template <typename T>
802 : : class unique_set
803 : : {
804 : : private:
805 : : set<T> values;
806 : : mutex mtx;
807 : : condition_variable cv;
808 : : public:
809 : 21 : unique_set() {}
810 : 21 : ~unique_set() {}
811 : :
812 : 338 : void acquire(const T& value)
813 : : {
814 [ + - ]: 676 : unique_lock<mutex> lock(mtx);
815 [ + - + + ]: 617 : while (values.find(value) != values.end())
816 : 279 : cv.wait(lock);
817 [ + - ]: 338 : values.insert(value);
818 : 338 : }
819 : :
820 : 338 : void release(const T& value)
821 : : {
822 [ + - ]: 676 : unique_lock<mutex> lock(mtx);
823 : : // assert (values.find(value) != values.end());
824 [ + - ]: 338 : values.erase(value);
825 : 338 : cv.notify_all();
826 : 338 : }
827 : : };
828 : :
829 : :
830 : : // This is the object that's instantiate to uniquely hold a value in a
831 : : // RAII-pattern way.
832 : : template <typename T>
833 : : class unique_set_reserver
834 : : {
835 : : private:
836 : : unique_set<T>& please_hold;
837 : : T mine;
838 : : public:
839 : 338 : unique_set_reserver(unique_set<T>& t, const T& value):
840 [ + - ]: 338 : please_hold(t), mine(value) { please_hold.acquire(mine); }
841 : 338 : ~unique_set_reserver() { please_hold.release(mine); }
842 : : };
843 : :
844 : :
845 : : ////////////////////////////////////////////////////////////////////////
846 : :
847 : :
848 : : // Print a standard timestamp.
849 : : static ostream&
850 : 5451 : timestamp (ostream &o)
851 : : {
852 : : char datebuf[80];
853 : 5451 : char *now2 = NULL;
854 : 5451 : time_t now_t = time(NULL);
855 : 5447 : struct tm *now = gmtime (&now_t);
856 [ + # ]: 5447 : if (now)
857 : : {
858 : 5449 : (void) strftime (datebuf, sizeof (datebuf), "%c", now);
859 : 5449 : now2 = datebuf;
860 : : }
861 : :
862 : : return o << "[" << (now2 ? now2 : "") << "] "
863 [ + - + - : 5447 : << "(" << getpid () << "/" << tid() << "): ";
+ - + - +
- + - + -
+ - + - ]
864 : : }
865 : :
866 : :
867 : : // A little class that impersonates an ostream to the extent that it can
868 : : // take << streaming operations. It batches up the bits into an internal
869 : : // stringstream until it is destroyed; then flushes to the original ostream.
870 : : // It adds a timestamp
871 : : class obatched
872 : : {
873 : : private:
874 : : ostream& o;
875 : : stringstream stro;
876 : : static mutex lock;
877 : : public:
878 : 5451 : obatched(ostream& oo, bool timestamp_p = true): o(oo)
879 : : {
880 [ + - ]: 5451 : if (timestamp_p)
881 [ + - ]: 5451 : timestamp(stro);
882 : 5450 : }
883 : 5450 : ~obatched()
884 : 5451 : {
885 : 10901 : unique_lock<mutex> do_not_cross_the_streams(obatched::lock);
886 : 5451 : o << stro.str();
887 : 5451 : o.flush();
888 : 5451 : }
889 : : operator ostream& () { return stro; }
890 : 5451 : template <typename T> ostream& operator << (const T& t) { stro << t; return stro; }
891 : : };
892 : : mutex obatched::lock; // just the one, since cout/cerr iostreams are not thread-safe
893 : :
894 : :
895 : 279 : void reportable_exception::report(ostream& o) const {
896 [ + - + - ]: 279 : obatched(o) << message << endl;
897 : 279 : }
898 : :
899 : :
900 : : ////////////////////////////////////////////////////////////////////////
901 : :
902 : :
903 : : // RAII style sqlite prepared-statement holder that matches { } block lifetime
904 : :
905 : : struct sqlite_ps
906 : : {
907 : : private:
908 : : sqlite3* db;
909 : : const string nickname;
910 : : const string sql;
911 : : sqlite3_stmt *pp;
912 : :
913 : : sqlite_ps(const sqlite_ps&); // make uncopyable
914 : : sqlite_ps& operator=(const sqlite_ps &); // make unassignable
915 : :
916 : : public:
917 [ + - ]: 1829 : sqlite_ps (sqlite3* d, const string& n, const string& s): db(d), nickname(n), sql(s) {
918 : : // tmp_ms_metric tick("sqlite3","prep",nickname);
919 [ - + ]: 1829 : if (verbose > 4)
920 [ # # # # : 0 : obatched(clog) << nickname << " prep " << sql << endl;
# # # # #
# ]
921 [ + - ]: 1829 : int rc = sqlite3_prepare_v2 (db, sql.c_str(), -1 /* to \0 */, & this->pp, NULL);
922 [ - + ]: 1829 : if (rc != SQLITE_OK)
923 [ # # # # ]: 0 : throw sqlite_exception(rc, "prepare " + sql);
924 : 1829 : }
925 : :
926 : 7673 : sqlite_ps& reset()
927 : : {
928 [ + - + - : 23018 : tmp_ms_metric tick("sqlite3","reset",nickname);
+ - ]
929 [ + - ]: 7673 : sqlite3_reset(this->pp);
930 : 15345 : return *this;
931 : : }
932 : :
933 : 10128 : sqlite_ps& bind(int parameter, const string& str)
934 : : {
935 [ - + ]: 10128 : if (verbose > 4)
936 [ # # # # : 0 : obatched(clog) << nickname << " bind " << parameter << "=" << str << endl;
# # # # #
# # # ]
937 : 10128 : int rc = sqlite3_bind_text (this->pp, parameter, str.c_str(), -1, SQLITE_TRANSIENT);
938 [ - + ]: 10128 : if (rc != SQLITE_OK)
939 [ # # # # ]: 0 : throw sqlite_exception(rc, "sqlite3 bind");
940 : 10128 : return *this;
941 : : }
942 : :
943 : 3146 : sqlite_ps& bind(int parameter, int64_t value)
944 : : {
945 [ - + ]: 3146 : if (verbose > 4)
946 [ # # # # : 0 : obatched(clog) << nickname << " bind " << parameter << "=" << value << endl;
# # # # #
# # # ]
947 : 3146 : int rc = sqlite3_bind_int64 (this->pp, parameter, value);
948 [ # + ]: 3143 : if (rc != SQLITE_OK)
949 [ - - - - ]: 0 : throw sqlite_exception(rc, "sqlite3 bind");
950 : 3146 : return *this;
951 : : }
952 : :
953 : : sqlite_ps& bind(int parameter)
954 : : {
955 : : if (verbose > 4)
956 : : obatched(clog) << nickname << " bind " << parameter << "=" << "NULL" << endl;
957 : : int rc = sqlite3_bind_null (this->pp, parameter);
958 : : if (rc != SQLITE_OK)
959 : : throw sqlite_exception(rc, "sqlite3 bind");
960 : : return *this;
961 : : }
962 : :
963 : :
964 : 5887 : void step_ok_done() {
965 [ + - + - : 23548 : tmp_ms_metric tick("sqlite3","step_done",nickname);
+ - ]
966 [ + - ]: 5887 : int rc = sqlite3_step (this->pp);
967 [ - + ]: 5887 : if (verbose > 4)
968 [ # # # # : 0 : obatched(clog) << nickname << " step-ok-done(" << sqlite3_errstr(rc) << ") " << sql << endl;
# # # # #
# # # # #
# # ]
969 [ + - + + : 5887 : if (rc != SQLITE_OK && rc != SQLITE_DONE && rc != SQLITE_ROW)
+ + ]
970 [ + - + - ]: 6 : throw sqlite_exception(rc, "sqlite3 step");
971 [ + - ]: 5881 : (void) sqlite3_reset (this->pp);
972 : 5881 : }
973 : :
974 : :
975 : 1303 : int step() {
976 [ + - + - : 3909 : tmp_ms_metric tick("sqlite3","step",nickname);
+ - ]
977 [ + - ]: 1303 : int rc = sqlite3_step (this->pp);
978 [ - + ]: 1304 : if (verbose > 4)
979 [ # # # # : 0 : obatched(clog) << nickname << " step(" << sqlite3_errstr(rc) << ") " << sql << endl;
# # # # #
# # # # #
# # ]
980 : 2608 : return rc;
981 : : }
982 : :
983 : 1829 : ~sqlite_ps () { sqlite3_finalize (this->pp); }
984 : 1718 : operator sqlite3_stmt* () { return this->pp; }
985 : : };
986 : :
987 : :
988 : : ////////////////////////////////////////////////////////////////////////
989 : :
990 : : // RAII style templated autocloser
991 : :
992 : : template <class Payload, class Ignore>
993 : : struct defer_dtor
994 : : {
995 : : public:
996 : : typedef Ignore (*dtor_fn) (Payload);
997 : :
998 : : private:
999 : : Payload p;
1000 : : dtor_fn fn;
1001 : :
1002 : : public:
1003 : 920 : defer_dtor(Payload _p, dtor_fn _fn): p(_p), fn(_fn) {}
1004 : 920 : ~defer_dtor() { (void) (*fn)(p); }
1005 : :
1006 : : private:
1007 : : defer_dtor(const defer_dtor<Payload,Ignore>&); // make uncopyable
1008 : : defer_dtor& operator=(const defer_dtor<Payload,Ignore> &); // make unassignable
1009 : : };
1010 : :
1011 : :
1012 : :
1013 : : ////////////////////////////////////////////////////////////////////////
1014 : :
1015 : :
1016 : : static string
1017 : 1226 : header_censor(const string& str)
1018 : : {
1019 : 1226 : string y;
1020 [ + + ]: 11446 : for (auto&& x : str)
1021 : : {
1022 [ + + + + : 10220 : if (isalnum(x) || x == '/' || x == '.' || x == ',' || x == '_' || x == ':')
+ + + + +
- - + ]
1023 [ + - ]: 10217 : y += x;
1024 : : }
1025 : 1226 : return y;
1026 : : }
1027 : :
1028 : :
1029 : : static string
1030 : 613 : conninfo (struct MHD_Connection * conn)
1031 : : {
1032 : : char hostname[256]; // RFC1035
1033 : : char servname[256];
1034 : 613 : int sts = -1;
1035 : :
1036 [ - + ]: 613 : if (conn == 0)
1037 [ # # ]: 0 : return "internal";
1038 : :
1039 : : /* Look up client address data. */
1040 [ + - ]: 613 : const union MHD_ConnectionInfo *u = MHD_get_connection_info (conn,
1041 : : MHD_CONNECTION_INFO_CLIENT_ADDRESS);
1042 [ + - ]: 613 : struct sockaddr *so = u ? u->client_addr : 0;
1043 : :
1044 [ + - + + ]: 613 : if (so && so->sa_family == AF_INET) {
1045 [ + - ]: 611 : sts = getnameinfo (so, sizeof (struct sockaddr_in), hostname, sizeof (hostname), servname,
1046 : : sizeof (servname), NI_NUMERICHOST | NI_NUMERICSERV);
1047 [ + - + - ]: 2 : } else if (so && so->sa_family == AF_INET6) {
1048 [ + - ]: 2 : sts = getnameinfo (so, sizeof (struct sockaddr_in6), hostname, sizeof (hostname),
1049 : : servname, sizeof (servname), NI_NUMERICHOST | NI_NUMERICSERV);
1050 : : }
1051 [ - + ]: 613 : if (sts != 0) {
1052 : 0 : hostname[0] = servname[0] = '\0';
1053 : : }
1054 : :
1055 : : // extract headers relevant to administration
1056 [ + - + - ]: 613 : const char* user_agent = MHD_lookup_connection_value (conn, MHD_HEADER_KIND, "User-Agent") ?: "";
1057 [ + - + + ]: 613 : const char* x_forwarded_for = MHD_lookup_connection_value (conn, MHD_HEADER_KIND, "X-Forwarded-For") ?: "";
1058 : : // NB: these are untrustworthy, beware if machine-processing log files
1059 : :
1060 [ + - + - : 1839 : return string(hostname) + string(":") + string(servname) +
+ - + - +
- ]
1061 [ + - + - : 2452 : string(" UA:") + header_censor(string(user_agent)) +
+ - + - +
- ]
1062 [ + - + - : 1839 : string(" XFF:") + header_censor(string(x_forwarded_for));
+ - + - +
- ]
1063 : : }
1064 : :
1065 : :
1066 : :
1067 : : ////////////////////////////////////////////////////////////////////////
1068 : :
1069 : :
1070 : : static void
1071 : 115 : add_mhd_last_modified (struct MHD_Response *resp, time_t mtime)
1072 : : {
1073 : 115 : struct tm *now = gmtime (&mtime);
1074 [ + - ]: 115 : if (now != NULL)
1075 : : {
1076 : : char datebuf[80];
1077 : 115 : size_t rc = strftime (datebuf, sizeof (datebuf), "%a, %d %b %Y %T GMT", now);
1078 [ + - + - ]: 115 : if (rc > 0 && rc < sizeof (datebuf))
1079 [ + - ]: 115 : (void) MHD_add_response_header (resp, "Last-Modified", datebuf);
1080 : : }
1081 : :
1082 : 115 : (void) MHD_add_response_header (resp, "Cache-Control", "public");
1083 : 115 : }
1084 : :
1085 : :
1086 : :
1087 : : static struct MHD_Response*
1088 : 32 : handle_buildid_f_match (bool internal_req_t,
1089 : : int64_t b_mtime,
1090 : : const string& b_source0,
1091 : : int *result_fd)
1092 : : {
1093 : : (void) internal_req_t; // ignored
1094 [ + - ]: 32 : int fd = open(b_source0.c_str(), O_RDONLY);
1095 [ - + ]: 32 : if (fd < 0)
1096 [ # # # # : 0 : throw libc_exception (errno, string("open ") + b_source0);
# # ]
1097 : :
1098 : : // NB: use manual close(2) in error case instead of defer_dtor, because
1099 : : // in the normal case, we want to hand the fd over to libmicrohttpd for
1100 : : // file transfer.
1101 : :
1102 : : struct stat s;
1103 : 32 : int rc = fstat(fd, &s);
1104 [ - + ]: 32 : if (rc < 0)
1105 : : {
1106 [ # # ]: 0 : close(fd);
1107 [ # # # # : 0 : throw libc_exception (errno, string("fstat ") + b_source0);
# # ]
1108 : : }
1109 : :
1110 [ - + ]: 32 : if ((int64_t) s.st_mtime != b_mtime)
1111 : : {
1112 [ # # ]: 0 : if (verbose)
1113 [ # # # # : 0 : obatched(clog) << "mtime mismatch for " << b_source0 << endl;
# # # # ]
1114 [ # # ]: 0 : close(fd);
1115 : 0 : return 0;
1116 : : }
1117 : :
1118 [ + - + - : 32 : inc_metric ("http_responses_total","result","file");
+ - + - ]
1119 [ + - ]: 32 : struct MHD_Response* r = MHD_create_response_from_fd ((uint64_t) s.st_size, fd);
1120 [ - + ]: 32 : if (r == 0)
1121 : : {
1122 [ # # ]: 0 : if (verbose)
1123 [ # # # # : 0 : obatched(clog) << "cannot create fd-response for " << b_source0 << endl;
# # # # ]
1124 [ # # ]: 0 : close(fd);
1125 : : }
1126 : : else
1127 : : {
1128 [ + - ]: 32 : MHD_add_response_header (r, "Content-Type", "application/octet-stream");
1129 [ + - ]: 64 : std::string file = b_source0.substr(b_source0.find_last_of("/")+1, b_source0.length());
1130 [ + - + - ]: 32 : MHD_add_response_header (r, "X-DEBUGINFOD-SIZE", to_string(s.st_size).c_str() );
1131 [ + - ]: 32 : MHD_add_response_header (r, "X-DEBUGINFOD-FILE", file.c_str() );
1132 [ + - ]: 32 : add_mhd_last_modified (r, s.st_mtime);
1133 [ + - ]: 32 : if (verbose > 1)
1134 [ + - + - : 32 : obatched(clog) << "serving file " << b_source0 << endl;
+ - + - ]
1135 : : /* libmicrohttpd will close it. */
1136 [ + - ]: 32 : if (result_fd)
1137 : 32 : *result_fd = fd;
1138 : : }
1139 : :
1140 : 32 : return r;
1141 : : }
1142 : :
1143 : :
1144 : : // quote all questionable characters of str for safe passage through a sh -c expansion.
1145 : : static string
1146 : 17 : shell_escape(const string& str)
1147 : : {
1148 : 17 : string y;
1149 [ + + ]: 1378 : for (auto&& x : str)
1150 : : {
1151 [ + + + + ]: 1361 : if (! isalnum(x) && x != '/')
1152 [ + - ]: 138 : y += "\\";
1153 [ + - ]: 1361 : y += x;
1154 : : }
1155 : 17 : return y;
1156 : : }
1157 : :
1158 : :
1159 : : // PR25548: Perform POSIX / RFC3986 style path canonicalization on the input string.
1160 : : //
1161 : : // Namely:
1162 : : // // -> /
1163 : : // /foo/../ -> /
1164 : : // /./ -> /
1165 : : //
1166 : : // This mapping is done on dwarf-side source path names, which may
1167 : : // include these constructs, so we can deal with debuginfod clients
1168 : : // that accidentally canonicalize the paths.
1169 : : //
1170 : : // realpath(3) is close but not quite right, because it also resolves
1171 : : // symbolic links. Symlinks at the debuginfod server have nothing to
1172 : : // do with the build-time symlinks, thus they must not be considered.
1173 : : //
1174 : : // see also curl Curl_dedotdotify() aka RFC3986, which we mostly follow here
1175 : : // see also libc __realpath()
1176 : : // see also llvm llvm::sys::path::remove_dots()
1177 : : static string
1178 : 1600 : canon_pathname (const string& input)
1179 : : {
1180 [ + - ]: 3200 : string i = input; // 5.2.4 (1)
1181 : 1600 : string o;
1182 : :
1183 [ + + ]: 10236 : while (i.size() != 0)
1184 : : {
1185 : : // 5.2.4 (2) A
1186 [ + - - + ]: 8636 : if (i.substr(0,3) == "../")
1187 [ # # ]: 0 : i = i.substr(3);
1188 [ + - - + ]: 8636 : else if(i.substr(0,2) == "./")
1189 [ # # ]: 0 : i = i.substr(2);
1190 : :
1191 : : // 5.2.4 (2) B
1192 [ + - + + ]: 8636 : else if (i.substr(0,3) == "/./")
1193 [ + - ]: 173 : i = i.substr(2);
1194 [ - + ]: 8463 : else if (i == "/.")
1195 [ # # ]: 0 : i = ""; // no need to handle "/." complete-path-segment case; we're dealing with file names
1196 : :
1197 : : // 5.2.4 (2) C
1198 [ + - + + ]: 8463 : else if (i.substr(0,4) == "/../") {
1199 [ + - ]: 251 : i = i.substr(3);
1200 [ + - ]: 251 : string::size_type sl = o.rfind("/");
1201 [ + - ]: 251 : if (sl != string::npos)
1202 [ + - ]: 251 : o = o.substr(0, sl);
1203 : : else
1204 [ # # ]: 0 : o = "";
1205 [ - + ]: 8212 : } else if (i == "/..")
1206 [ # # ]: 0 : i = ""; // no need to handle "/.." complete-path-segment case; we're dealing with file names
1207 : :
1208 : : // 5.2.4 (2) D
1209 : : // no need to handle these cases; we're dealing with file names
1210 [ - + ]: 8212 : else if (i == ".")
1211 [ # # ]: 0 : i = "";
1212 [ - + ]: 8212 : else if (i == "..")
1213 [ # # ]: 0 : i = "";
1214 : :
1215 : : // POSIX special: map // to /
1216 [ + - + + ]: 8212 : else if (i.substr(0,2) == "//")
1217 [ + - ]: 56 : i = i.substr(1);
1218 : :
1219 : : // 5.2.4 (2) E
1220 : : else {
1221 [ + - + - ]: 8156 : string::size_type next_slash = i.find("/", (i[0]=='/' ? 1 : 0)); // skip first slash
1222 [ + - + - ]: 8156 : o += i.substr(0, next_slash);
1223 [ + + ]: 8156 : if (next_slash == string::npos)
1224 [ + - ]: 1600 : i = "";
1225 : : else
1226 [ + - ]: 6556 : i = i.substr(next_slash);
1227 : : }
1228 : : }
1229 : :
1230 : 3200 : return o;
1231 : : }
1232 : :
1233 : :
1234 : : // Estimate available free space for a given filesystem via statfs(2).
1235 : : // Return true if the free fraction is known to be smaller than the
1236 : : // given minimum percentage. Also update a related metric.
1237 : 872 : bool statfs_free_enough_p(const string& path, const string& label, long minfree = 0)
1238 : : {
1239 : : struct statfs sfs;
1240 : 872 : int rc = statfs(path.c_str(), &sfs);
1241 [ + + ]: 872 : if (rc == 0)
1242 : : {
1243 : 871 : double s = (double) sfs.f_bavail / (double) sfs.f_blocks;
1244 [ + - + - : 871 : set_metric("filesys_free_ratio","purpose",label, s);
+ - ]
1245 : 871 : return ((s * 100.0) < minfree);
1246 : : }
1247 : 1 : return false;
1248 : : }
1249 : :
1250 : :
1251 : :
1252 : : // A map-like class that owns a cache of file descriptors (indexed by
1253 : : // file / content names).
1254 : : //
1255 : : // If only it could use fd's instead of file names ... but we can't
1256 : : // dup(2) to create independent descriptors for the same unlinked
1257 : : // files, so would have to use some goofy linux /proc/self/fd/%d
1258 : : // hack such as the following
1259 : :
1260 : : #if 0
1261 : : int superdup(int fd)
1262 : : {
1263 : : #ifdef __linux__
1264 : : char *fdpath = NULL;
1265 : : int rc = asprintf(& fdpath, "/proc/self/fd/%d", fd);
1266 : : int newfd;
1267 : : if (rc >= 0)
1268 : : newfd = open(fdpath, O_RDONLY);
1269 : : else
1270 : : newfd = -1;
1271 : : free (fdpath);
1272 : : return newfd;
1273 : : #else
1274 : : return -1;
1275 : : #endif
1276 : : }
1277 : : #endif
1278 : :
1279 : : class libarchive_fdcache
1280 : : {
1281 : : private:
1282 : : mutex fdcache_lock;
1283 : :
1284 : : struct fdcache_entry
1285 : : {
1286 : : string archive;
1287 : : string entry;
1288 : : string fd;
1289 : : double fd_size_mb; // slightly rounded up megabytes
1290 : : };
1291 : : deque<fdcache_entry> lru; // @head: most recently used
1292 : : long max_fds;
1293 : : deque<fdcache_entry> prefetch; // prefetched
1294 : : long max_mbs;
1295 : : long max_prefetch_mbs;
1296 : : long max_prefetch_fds;
1297 : :
1298 : : public:
1299 : 233 : void set_metrics()
1300 : : {
1301 : 233 : double fdcache_mb = 0.0;
1302 : 233 : double prefetch_mb = 0.0;
1303 [ + + ]: 1456 : for (auto i = lru.begin(); i < lru.end(); i++)
1304 : 1223 : fdcache_mb += i->fd_size_mb;
1305 [ + + ]: 266 : for (auto j = prefetch.begin(); j < prefetch.end(); j++)
1306 : 33 : prefetch_mb += j->fd_size_mb;
1307 [ + - + - ]: 233 : set_metric("fdcache_bytes", fdcache_mb*1024.0*1024.0);
1308 [ + - + - ]: 233 : set_metric("fdcache_count", lru.size());
1309 [ + - + - ]: 233 : set_metric("fdcache_prefetch_bytes", prefetch_mb*1024.0*1024.0);
1310 [ + - + - ]: 233 : set_metric("fdcache_prefetch_count", prefetch.size());
1311 : 233 : }
1312 : :
1313 : 74 : void intern(const string& a, const string& b, string fd, off_t sz, bool front_p)
1314 : : {
1315 : : {
1316 [ + - ]: 148 : unique_lock<mutex> lock(fdcache_lock);
1317 : : // nuke preexisting copy
1318 [ + + ]: 567 : for (auto i = lru.begin(); i < lru.end(); i++)
1319 : : {
1320 [ + + - + : 493 : if (i->archive == a && i->entry == b)
- + ]
1321 : : {
1322 : 0 : unlink (i->fd.c_str());
1323 [ # # ]: 0 : lru.erase(i);
1324 [ # # # # : 0 : inc_metric("fdcache_op_count","op","dequeue");
# # # # ]
1325 : 0 : break; // must not continue iterating
1326 : : }
1327 : : }
1328 : : // nuke preexisting copy in prefetch
1329 [ + + ]: 92 : for (auto i = prefetch.begin(); i < prefetch.end(); i++)
1330 : : {
1331 [ + + - + : 18 : if (i->archive == a && i->entry == b)
- + ]
1332 : : {
1333 : 0 : unlink (i->fd.c_str());
1334 [ # # ]: 0 : prefetch.erase(i);
1335 [ # # # # : 0 : inc_metric("fdcache_op_count","op","prefetch_dequeue");
# # # # ]
1336 : 0 : break; // must not continue iterating
1337 : : }
1338 : : }
1339 : 74 : double mb = (sz+65535)/1048576.0; // round up to 64K block
1340 [ + - + - : 148 : fdcache_entry n = { a, b, fd, mb };
+ - ]
1341 [ + + ]: 74 : if (front_p)
1342 : : {
1343 [ + - + - : 59 : inc_metric("fdcache_op_count","op","enqueue");
+ - + - ]
1344 [ + - ]: 59 : lru.push_front(n);
1345 : : }
1346 : : else
1347 : : {
1348 [ + - + - : 15 : inc_metric("fdcache_op_count","op","prefetch_enqueue");
+ - + - ]
1349 [ + - ]: 15 : prefetch.push_front(n);
1350 : : }
1351 [ + + ]: 74 : if (verbose > 3)
1352 [ + - + - ]: 118 : obatched(clog) << "fdcache interned a=" << a << " b=" << b
1353 [ + - + - : 59 : << " fd=" << fd << " mb=" << mb << " front=" << front_p << endl;
+ - + - +
- + - + -
+ - + - +
- ]
1354 : : }
1355 : 74 : set_metrics();
1356 : :
1357 : : // NB: we age the cache at lookup time too
1358 [ + - + - : 74 : if (statfs_free_enough_p(tmpdir, "tmpdir", fdcache_mintmp))
- + ]
1359 : : {
1360 [ # # # # : 0 : inc_metric("fdcache_op_count","op","emerg-flush");
# # # # ]
1361 [ # # # # ]: 0 : obatched(clog) << "fdcache emergency flush for filling tmpdir" << endl;
1362 : 0 : this->limit(0, 0, 0, 0); // emergency flush
1363 : : }
1364 [ + + ]: 74 : else if (front_p)
1365 : 59 : this->limit(max_fds, max_mbs, max_prefetch_fds, max_prefetch_mbs); // age cache if required
1366 : 74 : }
1367 : :
1368 : 81 : int lookup(const string& a, const string& b)
1369 : : {
1370 : 81 : int fd = -1;
1371 : : {
1372 [ + - ]: 162 : unique_lock<mutex> lock(fdcache_lock);
1373 [ + + ]: 577 : for (auto i = lru.begin(); i < lru.end(); i++)
1374 : : {
1375 [ + + + + : 516 : if (i->archive == a && i->entry == b)
+ + ]
1376 : : { // found it; move it to head of lru
1377 [ + - ]: 40 : fdcache_entry n = *i;
1378 [ + - ]: 20 : lru.erase(i); // invalidates i, so no more iteration!
1379 [ + - ]: 20 : lru.push_front(n);
1380 [ + - + - : 20 : inc_metric("fdcache_op_count","op","requeue_front");
+ - + - ]
1381 [ + - ]: 20 : fd = open(n.fd.c_str(), O_RDONLY);
1382 : 20 : break;
1383 : : }
1384 : : }
1385 : : // Iterate through prefetch while fd == -1 to ensure that no duplication between lru and
1386 : : // prefetch occurs.
1387 [ + + + + : 90 : for ( auto i = prefetch.begin(); fd == -1 && i < prefetch.end(); ++i)
+ + ]
1388 : : {
1389 [ + + + - : 11 : if (i->archive == a && i->entry == b)
+ + ]
1390 : : { // found it; take the entry from the prefetch deque to the lru deque, since it has now been accessed.
1391 [ + - ]: 4 : fdcache_entry n = *i;
1392 [ + - ]: 2 : prefetch.erase(i);
1393 [ + - ]: 2 : lru.push_front(n);
1394 [ + - + - : 2 : inc_metric("fdcache_op_count","op","prefetch_access");
+ - + - ]
1395 [ + - ]: 2 : fd = open(n.fd.c_str(), O_RDONLY);
1396 : 2 : break;
1397 : : }
1398 : : }
1399 : : }
1400 : :
1401 [ + - + - : 81 : if (statfs_free_enough_p(tmpdir, "tmpdir", fdcache_mintmp))
- + ]
1402 : : {
1403 [ # # # # : 0 : inc_metric("fdcache_op_count","op","emerg-flush");
# # # # ]
1404 [ # # # # ]: 0 : obatched(clog) << "fdcache emergency flush for filling tmpdir" << endl;
1405 : 0 : this->limit(0, 0, 0, 0); // emergency flush
1406 : : }
1407 [ + + ]: 81 : else if (fd >= 0)
1408 : 22 : this->limit(max_fds, max_mbs, max_prefetch_fds, max_prefetch_mbs); // age cache if required
1409 : :
1410 : 81 : return fd;
1411 : : }
1412 : :
1413 : 84 : int probe(const string& a, const string& b) // just a cache residency check - don't modify LRU state, don't open
1414 : : {
1415 [ + - ]: 168 : unique_lock<mutex> lock(fdcache_lock);
1416 [ + + ]: 605 : for (auto i = lru.begin(); i < lru.end(); i++)
1417 : : {
1418 [ + + + + : 531 : if (i->archive == a && i->entry == b)
+ + ]
1419 : : {
1420 [ + - + - : 10 : inc_metric("fdcache_op_count","op","probe_hit");
+ - + - ]
1421 : 10 : return true;
1422 : : }
1423 : : }
1424 [ + + ]: 92 : for (auto i = prefetch.begin(); i < prefetch.end(); i++)
1425 : : {
1426 [ + + - + : 18 : if (i->archive == a && i->entry == b)
- + ]
1427 : : {
1428 [ # # # # : 0 : inc_metric("fdcache_op_count","op","prefetch_probe_hit");
# # # # ]
1429 : 0 : return true;
1430 : : }
1431 : : }
1432 [ + - + - : 74 : inc_metric("fdcache_op_count","op","probe_miss");
+ - + - ]
1433 : 74 : return false;
1434 : : }
1435 : :
1436 : 0 : void clear(const string& a, const string& b)
1437 : : {
1438 [ # # ]: 0 : unique_lock<mutex> lock(fdcache_lock);
1439 [ # # ]: 0 : for (auto i = lru.begin(); i < lru.end(); i++)
1440 : : {
1441 [ # # # # : 0 : if (i->archive == a && i->entry == b)
# # ]
1442 : : { // found it; erase it from lru
1443 [ # # ]: 0 : fdcache_entry n = *i;
1444 [ # # ]: 0 : lru.erase(i); // invalidates i, so no more iteration!
1445 [ # # # # : 0 : inc_metric("fdcache_op_count","op","clear");
# # # # ]
1446 : 0 : unlink (n.fd.c_str());
1447 [ # # ]: 0 : set_metrics();
1448 : 0 : return;
1449 : : }
1450 : : }
1451 [ # # ]: 0 : for (auto i = prefetch.begin(); i < prefetch.end(); i++)
1452 : : {
1453 [ # # # # : 0 : if (i->archive == a && i->entry == b)
# # ]
1454 : : { // found it; erase it from lru
1455 [ # # ]: 0 : fdcache_entry n = *i;
1456 [ # # ]: 0 : prefetch.erase(i); // invalidates i, so no more iteration!
1457 [ # # # # : 0 : inc_metric("fdcache_op_count","op","prefetch_clear");
# # # # ]
1458 : 0 : unlink (n.fd.c_str());
1459 [ # # ]: 0 : set_metrics();
1460 : 0 : return;
1461 : : }
1462 : : }
1463 : : }
1464 : :
1465 : 187 : void limit(long maxfds, long maxmbs, long maxprefetchfds, long maxprefetchmbs , bool metrics_p = true)
1466 : : {
1467 [ + + + + : 187 : if (verbose > 3 && (this->max_fds != maxfds || this->max_mbs != maxmbs))
- + ]
1468 [ + - + - : 60 : obatched(clog) << "fdcache limited to maxfds=" << maxfds << " maxmbs=" << maxmbs << endl;
+ - + - +
- + - ]
1469 : :
1470 [ + - ]: 374 : unique_lock<mutex> lock(fdcache_lock);
1471 : 187 : this->max_fds = maxfds;
1472 : 187 : this->max_mbs = maxmbs;
1473 : 187 : this->max_prefetch_fds = maxprefetchfds;
1474 : 187 : this->max_prefetch_mbs = maxprefetchmbs;
1475 : 187 : long total_fd = 0;
1476 : 187 : double total_mb = 0.0;
1477 [ + + ]: 858 : for (auto i = lru.begin(); i < lru.end(); i++)
1478 : : {
1479 : : // accumulate totals from most recently used one going backward
1480 : 681 : total_fd ++;
1481 : 681 : total_mb += i->fd_size_mb;
1482 [ + + - + ]: 681 : if (total_fd > this->max_fds || total_mb > this->max_mbs)
1483 : : {
1484 : : // found the cut here point!
1485 : :
1486 [ + + ]: 71 : for (auto j = i; j < lru.end(); j++) // close all the fds from here on in
1487 : : {
1488 [ + + ]: 61 : if (verbose > 3)
1489 [ + - + - ]: 110 : obatched(clog) << "fdcache evicted a=" << j->archive << " b=" << j->entry
1490 [ + - + - : 55 : << " fd=" << j->fd << " mb=" << j->fd_size_mb << endl;
+ - + - +
- + - + -
+ - ]
1491 [ + + ]: 61 : if (metrics_p)
1492 [ + - + - : 22 : inc_metric("fdcache_op_count","op","evict");
+ - + - ]
1493 : 61 : unlink (j->fd.c_str());
1494 : : }
1495 : :
1496 [ + - ]: 10 : lru.erase(i, lru.end()); // erase the nodes generally
1497 : 10 : break;
1498 : : }
1499 : : }
1500 : 187 : total_fd = 0;
1501 : 187 : total_mb = 0.0;
1502 [ + + ]: 187 : for(auto i = prefetch.begin(); i < prefetch.end(); i++){
1503 : : // accumulate totals from most recently used one going backward
1504 : 7 : total_fd ++;
1505 : 7 : total_mb += i->fd_size_mb;
1506 [ - + - - ]: 7 : if (total_fd > this->max_prefetch_fds || total_mb > this->max_prefetch_mbs)
1507 : : {
1508 : : // found the cut here point!
1509 [ + + ]: 20 : for (auto j = i; j < prefetch.end(); j++) // close all the fds from here on in
1510 : : {
1511 [ + + ]: 13 : if (verbose > 3)
1512 [ + - + - ]: 8 : obatched(clog) << "fdcache evicted from prefetch a=" << j->archive << " b=" << j->entry
1513 [ + - + - : 4 : << " fd=" << j->fd << " mb=" << j->fd_size_mb << endl;
+ - + - +
- + - + -
+ - ]
1514 [ + + ]: 13 : if (metrics_p)
1515 [ + - + - : 12 : inc_metric("fdcache_op_count","op","prefetch_evict");
+ - + - ]
1516 : 13 : unlink (j->fd.c_str());
1517 : : }
1518 : :
1519 [ + - ]: 7 : prefetch.erase(i, prefetch.end()); // erase the nodes generally
1520 : 7 : break;
1521 : : }
1522 : : }
1523 [ + + + - ]: 187 : if (metrics_p) set_metrics();
1524 : 187 : }
1525 : :
1526 : :
1527 : 28 : ~libarchive_fdcache()
1528 : 28 : {
1529 : : // unlink any fdcache entries in $TMPDIR
1530 : : // don't update metrics; those globals may be already destroyed
1531 : 28 : limit(0, 0, 0, 0, false);
1532 : 28 : }
1533 : : };
1534 : : static libarchive_fdcache fdcache;
1535 : :
1536 : :
1537 : : // For security/portability reasons, many distro-package archives have
1538 : : // a "./" in front of path names; others have nothing, others have
1539 : : // "/". Canonicalize them all to a single leading "/", with the
1540 : : // assumption that this matches the dwarf-derived file names too.
1541 : 328 : string canonicalized_archive_entry_pathname(struct archive_entry *e)
1542 : : {
1543 [ + - + - ]: 656 : string fn = archive_entry_pathname(e);
1544 [ - + ]: 328 : if (fn.size() == 0)
1545 : 0 : return fn;
1546 [ + - - + ]: 328 : if (fn[0] == '/')
1547 : 0 : return fn;
1548 [ + - + + ]: 328 : if (fn[0] == '.')
1549 [ + - ]: 273 : return fn.substr(1);
1550 : : else
1551 [ + - + - ]: 108 : return string("/")+fn;
1552 : : }
1553 : :
1554 : :
1555 : :
1556 : : static struct MHD_Response*
1557 : 110 : handle_buildid_r_match (bool internal_req_p,
1558 : : int64_t b_mtime,
1559 : : const string& b_source0,
1560 : : const string& b_source1,
1561 : : int *result_fd)
1562 : : {
1563 : : struct stat fs;
1564 : 110 : int rc = stat (b_source0.c_str(), &fs);
1565 [ + + ]: 110 : if (rc != 0)
1566 [ + - + - : 29 : throw libc_exception (errno, string("stat ") + b_source0);
+ - ]
1567 : :
1568 [ - + ]: 81 : if ((int64_t) fs.st_mtime != b_mtime)
1569 : : {
1570 [ # # ]: 0 : if (verbose)
1571 [ # # # # : 0 : obatched(clog) << "mtime mismatch for " << b_source0 << endl;
# # # # ]
1572 : 0 : return 0;
1573 : : }
1574 : :
1575 : : // check for a match in the fdcache first
1576 [ + - ]: 81 : int fd = fdcache.lookup(b_source0, b_source1);
1577 [ + + ]: 81 : while (fd >= 0) // got one!; NB: this is really an if() with a possible branch out to the end
1578 : : {
1579 : 22 : rc = fstat(fd, &fs);
1580 [ - + ]: 22 : if (rc < 0) // disappeared?
1581 : : {
1582 [ # # ]: 0 : if (verbose)
1583 [ # # # # : 0 : obatched(clog) << "cannot fstat fdcache " << b_source0 << endl;
# # # # ]
1584 [ # # ]: 0 : close(fd);
1585 [ # # ]: 0 : fdcache.clear(b_source0, b_source1);
1586 : 0 : break; // branch out of if "loop", to try new libarchive fetch attempt
1587 : : }
1588 : :
1589 [ + - ]: 22 : struct MHD_Response* r = MHD_create_response_from_fd (fs.st_size, fd);
1590 [ - + ]: 22 : if (r == 0)
1591 : : {
1592 [ # # ]: 0 : if (verbose)
1593 [ # # # # : 0 : obatched(clog) << "cannot create fd-response for " << b_source0 << endl;
# # # # ]
1594 [ # # ]: 0 : close(fd);
1595 : 0 : break; // branch out of if "loop", to try new libarchive fetch attempt
1596 : : }
1597 : :
1598 [ + - + - : 22 : inc_metric ("http_responses_total","result","archive fdcache");
+ - + - ]
1599 : :
1600 [ + - ]: 22 : MHD_add_response_header (r, "Content-Type", "application/octet-stream");
1601 [ + - + - ]: 22 : MHD_add_response_header (r, "X-DEBUGINFOD-SIZE", to_string(fs.st_size).c_str());
1602 [ + - ]: 22 : MHD_add_response_header (r, "X-DEBUGINFOD-ARCHIVE", b_source0.c_str());
1603 [ + - ]: 22 : MHD_add_response_header (r, "X-DEBUGINFOD-FILE", b_source1.c_str());
1604 [ + - ]: 22 : add_mhd_last_modified (r, fs.st_mtime);
1605 [ + - ]: 22 : if (verbose > 1)
1606 [ + - + - : 22 : obatched(clog) << "serving fdcache archive " << b_source0 << " file " << b_source1 << endl;
+ - + - +
- + - ]
1607 : : /* libmicrohttpd will close it. */
1608 [ + - ]: 22 : if (result_fd)
1609 : 22 : *result_fd = fd;
1610 : 22 : return r;
1611 : : // NB: see, we never go around the 'loop' more than once
1612 : : }
1613 : :
1614 : : // no match ... grumble, must process the archive
1615 [ + - ]: 118 : string archive_decoder = "/dev/null";
1616 [ + - ]: 118 : string archive_extension = "";
1617 [ + + ]: 133 : for (auto&& arch : scan_archives)
1618 [ + - + + ]: 74 : if (string_endswith(b_source0, arch.first))
1619 : : {
1620 [ + - ]: 59 : archive_extension = arch.first;
1621 [ + - ]: 59 : archive_decoder = arch.second;
1622 : : }
1623 : : FILE* fp;
1624 : : defer_dtor<FILE*,int>::dtor_fn dfn;
1625 [ + + ]: 59 : if (archive_decoder != "cat")
1626 : : {
1627 [ + - + - : 21 : string popen_cmd = archive_decoder + " " + shell_escape(b_source0);
+ - ]
1628 [ + - ]: 7 : fp = popen (popen_cmd.c_str(), "r"); // "e" O_CLOEXEC?
1629 : 7 : dfn = pclose;
1630 [ - + ]: 7 : if (fp == NULL)
1631 [ # # # # : 0 : throw libc_exception (errno, string("popen ") + popen_cmd);
# # ]
1632 : : }
1633 : : else
1634 : : {
1635 [ + - ]: 52 : fp = fopen (b_source0.c_str(), "r");
1636 : 52 : dfn = fclose;
1637 [ - + ]: 52 : if (fp == NULL)
1638 [ # # # # : 0 : throw libc_exception (errno, string("fopen ") + b_source0);
# # ]
1639 : : }
1640 : 118 : defer_dtor<FILE*,int> fp_closer (fp, dfn);
1641 : :
1642 : : struct archive *a;
1643 [ + - ]: 59 : a = archive_read_new();
1644 [ - + ]: 59 : if (a == NULL)
1645 [ # # # # ]: 0 : throw archive_exception("cannot create archive reader");
1646 : 59 : defer_dtor<struct archive*,int> archive_closer (a, archive_read_free);
1647 : :
1648 [ + - ]: 59 : rc = archive_read_support_format_all(a);
1649 [ - + ]: 59 : if (rc != ARCHIVE_OK)
1650 [ # # # # ]: 0 : throw archive_exception(a, "cannot select all format");
1651 [ + - ]: 59 : rc = archive_read_support_filter_all(a);
1652 [ - + ]: 59 : if (rc != ARCHIVE_OK)
1653 [ # # # # ]: 0 : throw archive_exception(a, "cannot select all filters");
1654 : :
1655 [ + - ]: 59 : rc = archive_read_open_FILE (a, fp);
1656 [ - + ]: 59 : if (rc != ARCHIVE_OK)
1657 [ # # # # ]: 0 : throw archive_exception(a, "cannot open archive from pipe");
1658 : :
1659 : : // archive traversal is in three stages, no, four stages:
1660 : : // 1) skip entries whose names do not match the requested one
1661 : : // 2) extract the matching entry name (set r = result)
1662 : : // 3) extract some number of prefetched entries (just into fdcache)
1663 : : // 4) abort any further processing
1664 : 59 : struct MHD_Response* r = 0; // will set in stage 2
1665 [ + + ]: 59 : unsigned prefetch_count =
1666 : : internal_req_p ? 0 : fdcache_prefetch; // will decrement in stage 3
1667 : :
1668 [ + + + + ]: 602 : while(r == 0 || prefetch_count > 0) // stage 1, 2, or 3
1669 : : {
1670 [ - + ]: 594 : if (interrupted)
1671 : 51 : break;
1672 : :
1673 : : struct archive_entry *e;
1674 [ + - ]: 594 : rc = archive_read_next_header (a, &e);
1675 [ + + ]: 594 : if (rc != ARCHIVE_OK)
1676 : 51 : break;
1677 : :
1678 [ + - + + ]: 543 : if (! S_ISREG(archive_entry_mode (e))) // skip non-files completely
1679 : 543 : continue;
1680 : :
1681 [ + - ]: 111 : string fn = canonicalized_archive_entry_pathname (e);
1682 [ + + + + : 111 : if ((r == 0) && (fn != b_source1)) // stage 1
+ + ]
1683 : 27 : continue;
1684 : :
1685 [ + - + + ]: 84 : if (fdcache.probe (b_source0, fn)) // skip if already interned
1686 : 10 : continue;
1687 : :
1688 : : // extract this file to a temporary file
1689 : 74 : char* tmppath = NULL;
1690 : 74 : rc = asprintf (&tmppath, "%s/debuginfod.XXXXXX", tmpdir.c_str());
1691 [ - + ]: 74 : if (rc < 0)
1692 [ # # # # ]: 0 : throw libc_exception (ENOMEM, "cannot allocate tmppath");
1693 : 74 : defer_dtor<void*,void> tmmpath_freer (tmppath, free);
1694 [ + - ]: 74 : fd = mkstemp (tmppath);
1695 [ - + ]: 74 : if (fd < 0)
1696 [ # # # # ]: 0 : throw libc_exception (errno, "cannot create temporary file");
1697 : : // NB: don't unlink (tmppath), as fdcache will take charge of it.
1698 : :
1699 : : // NB: this can take many uninterruptible seconds for a huge file
1700 [ + - ]: 74 : rc = archive_read_data_into_fd (a, fd);
1701 [ - + ]: 74 : if (rc != ARCHIVE_OK) // e.g. ENOSPC!
1702 : : {
1703 [ # # ]: 0 : close (fd);
1704 : 0 : unlink (tmppath);
1705 [ # # # # ]: 0 : throw archive_exception(a, "cannot extract file");
1706 : : }
1707 : :
1708 : : // Set the mtime so the fdcache file mtimes, even prefetched ones,
1709 : : // propagate to future webapi clients.
1710 : : struct timeval tvs[2];
1711 [ + - ]: 74 : tvs[0].tv_sec = tvs[1].tv_sec = archive_entry_mtime(e);
1712 : 74 : tvs[0].tv_usec = tvs[1].tv_usec = 0;
1713 : 74 : (void) futimes (fd, tvs); /* best effort */
1714 : :
1715 [ + + ]: 74 : if (r != 0) // stage 3
1716 : : {
1717 : : // NB: now we know we have a complete reusable file; make fdcache
1718 : : // responsible for unlinking it later.
1719 [ + - + - ]: 15 : fdcache.intern(b_source0, fn,
1720 : : tmppath, archive_entry_size(e),
1721 [ + - ]: 30 : false); // prefetched ones go to the prefetch cache
1722 : 15 : prefetch_count --;
1723 [ + - ]: 15 : close (fd); // we're not saving this fd to make a mhd-response from!
1724 : 15 : continue;
1725 : : }
1726 : :
1727 : : // NB: now we know we have a complete reusable file; make fdcache
1728 : : // responsible for unlinking it later.
1729 [ + - + - ]: 59 : fdcache.intern(b_source0, b_source1,
1730 : : tmppath, archive_entry_size(e),
1731 [ + - ]: 118 : true); // requested ones go to the front of lru
1732 : :
1733 [ + - + - : 59 : inc_metric ("http_responses_total","result",archive_extension + " archive");
+ - + - ]
1734 [ + - + - ]: 59 : r = MHD_create_response_from_fd (archive_entry_size(e), fd);
1735 [ - + ]: 59 : if (r == 0)
1736 : : {
1737 [ # # ]: 0 : if (verbose)
1738 [ # # # # : 0 : obatched(clog) << "cannot create fd-response for " << b_source0 << endl;
# # # # ]
1739 [ # # ]: 0 : close(fd);
1740 : 0 : break; // assume no chance of better luck around another iteration; no other copies of same file
1741 : : }
1742 : : else
1743 : : {
1744 [ + - ]: 59 : MHD_add_response_header (r, "Content-Type", "application/octet-stream");
1745 [ + - ]: 118 : std::string file = b_source1.substr(b_source1.find_last_of("/")+1, b_source1.length());
1746 [ + - + - ]: 59 : MHD_add_response_header (r, "X-DEBUGINFOD-SIZE", to_string(fs.st_size).c_str());
1747 [ + - ]: 59 : MHD_add_response_header (r, "X-DEBUGINFOD-ARCHIVE", b_source0.c_str());
1748 [ + - ]: 59 : MHD_add_response_header (r, "X-DEBUGINFOD-FILE", file.c_str());
1749 : :
1750 [ + - + - ]: 59 : add_mhd_last_modified (r, archive_entry_mtime(e));
1751 [ + - ]: 59 : if (verbose > 1)
1752 [ + - + - : 59 : obatched(clog) << "serving archive " << b_source0 << " file " << b_source1 << endl;
+ - + - +
- + - ]
1753 : : /* libmicrohttpd will close it. */
1754 [ + - ]: 59 : if (result_fd)
1755 : 59 : *result_fd = fd;
1756 : 59 : continue;
1757 : : }
1758 : : }
1759 : :
1760 : : // XXX: rpm/file not found: delete this R entry?
1761 : 59 : return r;
1762 : : }
1763 : :
1764 : :
1765 : : static struct MHD_Response*
1766 : 142 : handle_buildid_match (bool internal_req_p,
1767 : : int64_t b_mtime,
1768 : : const string& b_stype,
1769 : : const string& b_source0,
1770 : : const string& b_source1,
1771 : : int *result_fd)
1772 : : {
1773 : : try
1774 : : {
1775 [ + + ]: 142 : if (b_stype == "F")
1776 [ + - ]: 32 : return handle_buildid_f_match(internal_req_p, b_mtime, b_source0, result_fd);
1777 [ + - ]: 110 : else if (b_stype == "R")
1778 [ + + ]: 110 : return handle_buildid_r_match(internal_req_p, b_mtime, b_source0, b_source1, result_fd);
1779 : : }
1780 [ - + ]: 58 : catch (const reportable_exception &e)
1781 : : {
1782 [ + - ]: 29 : e.report(clog);
1783 : : // Report but swallow libc etc. errors here; let the caller
1784 : : // iterate to other matches of the content.
1785 : : }
1786 : :
1787 : 29 : return 0;
1788 : : }
1789 : :
1790 : :
1791 : : static int
1792 : 348 : debuginfod_find_progress (debuginfod_client *, long a, long b)
1793 : : {
1794 [ - + ]: 348 : if (verbose > 4)
1795 [ # # # # : 0 : obatched(clog) << "federated debuginfod progress=" << a << "/" << b << endl;
# # # # #
# ]
1796 : :
1797 : 348 : return interrupted;
1798 : : }
1799 : :
1800 : :
1801 : : // a little lru pool of debuginfod_client*s for reuse between query threads
1802 : :
1803 : : mutex dc_pool_lock;
1804 : : deque<debuginfod_client*> dc_pool;
1805 : :
1806 : 237 : debuginfod_client* debuginfod_pool_begin()
1807 : : {
1808 [ + - ]: 474 : unique_lock<mutex> lock(dc_pool_lock);
1809 [ + + ]: 237 : if (dc_pool.size() > 0)
1810 : : {
1811 [ + - + - : 224 : inc_metric("dc_pool_op_count","op","begin-reuse");
+ - + - ]
1812 : 224 : debuginfod_client *c = dc_pool.front();
1813 : 224 : dc_pool.pop_front();
1814 : 224 : return c;
1815 : : }
1816 [ + - + - : 13 : inc_metric("dc_pool_op_count","op","begin-new");
+ - + - ]
1817 [ + - ]: 13 : return debuginfod_begin();
1818 : : }
1819 : :
1820 : :
1821 : 25 : void debuginfod_pool_groom()
1822 : : {
1823 [ + - ]: 50 : unique_lock<mutex> lock(dc_pool_lock);
1824 [ + + ]: 26 : while (dc_pool.size() > 0)
1825 : : {
1826 [ + - + - : 1 : inc_metric("dc_pool_op_count","op","end");
+ - + - ]
1827 [ + - ]: 1 : debuginfod_end(dc_pool.front());
1828 : 1 : dc_pool.pop_front();
1829 : : }
1830 : 25 : }
1831 : :
1832 : :
1833 : 235 : void debuginfod_pool_end(debuginfod_client* c)
1834 : : {
1835 [ + - ]: 470 : unique_lock<mutex> lock(dc_pool_lock);
1836 [ + - + - : 235 : inc_metric("dc_pool_op_count","op","end-save");
+ - + - ]
1837 [ + - ]: 235 : dc_pool.push_front(c); // accelerate reuse, vs. push_back
1838 : 235 : }
1839 : :
1840 : :
1841 : : static struct MHD_Response*
1842 : 354 : handle_buildid (MHD_Connection* conn,
1843 : : const string& buildid /* unsafe */,
1844 : : string& artifacttype /* unsafe, cleanse on exception/return */,
1845 : : const string& suffix /* unsafe */,
1846 : : int *result_fd)
1847 : : {
1848 : : // validate artifacttype
1849 : 708 : string atype_code;
1850 [ + + + - ]: 354 : if (artifacttype == "debuginfo") atype_code = "D";
1851 [ + + + - ]: 254 : else if (artifacttype == "executable") atype_code = "E";
1852 [ + + + - ]: 24 : else if (artifacttype == "source") atype_code = "S";
1853 : : else {
1854 [ + - ]: 2 : artifacttype = "invalid"; // PR28242 ensure http_resposes metrics don't propagate unclean user data
1855 [ + - + - ]: 2 : throw reportable_exception("invalid artifacttype");
1856 : : }
1857 : :
1858 [ + - + - : 352 : inc_metric("http_requests_total", "type", artifacttype);
+ - ]
1859 : :
1860 [ + + - + : 352 : if (atype_code == "S" && suffix == "")
- + ]
1861 [ # # # # ]: 0 : throw reportable_exception("invalid source suffix");
1862 : :
1863 : : // validate buildid
1864 : 352 : if ((buildid.size() < 2) || // not empty
1865 [ + + + - : 703 : (buildid.size() % 2) || // even number
- + + + ]
1866 : 351 : (buildid.find_first_not_of("0123456789abcdef") != string::npos)) // pure tasty lowercase hex
1867 [ + - + - ]: 1 : throw reportable_exception("invalid buildid");
1868 : :
1869 [ + - ]: 351 : if (verbose > 1)
1870 [ + - + - ]: 702 : obatched(clog) << "searching for buildid=" << buildid << " artifacttype=" << artifacttype
1871 [ + - + - : 351 : << " suffix=" << suffix << endl;
+ - + - +
- + - ]
1872 : :
1873 : : // If invoked from the scanner threads, use the scanners' read-write
1874 : : // connection. Otherwise use the web query threads' read-only connection.
1875 [ + + ]: 351 : sqlite3 *thisdb = (conn == 0) ? db : dbq;
1876 : :
1877 : 351 : sqlite_ps *pp = 0;
1878 : :
1879 [ + + ]: 351 : if (atype_code == "D")
1880 : : {
1881 : 100 : pp = new sqlite_ps (thisdb, "mhd-query-d",
1882 : : "select mtime, sourcetype, source0, source1 from " BUILDIDS "_query_d where buildid = ? "
1883 [ + - + - : 100 : "order by mtime desc");
+ - + - ]
1884 [ + - ]: 100 : pp->reset();
1885 [ + - ]: 100 : pp->bind(1, buildid);
1886 : : }
1887 [ + + ]: 251 : else if (atype_code == "E")
1888 : : {
1889 : 229 : pp = new sqlite_ps (thisdb, "mhd-query-e",
1890 : : "select mtime, sourcetype, source0, source1 from " BUILDIDS "_query_e where buildid = ? "
1891 [ + - + - : 229 : "order by mtime desc");
+ - + - ]
1892 [ + - ]: 229 : pp->reset();
1893 [ + - ]: 229 : pp->bind(1, buildid);
1894 : : }
1895 [ + - ]: 22 : else if (atype_code == "S")
1896 : : {
1897 : : // PR25548
1898 : : // Incoming source queries may come in with either dwarf-level OR canonicalized paths.
1899 : : // We let the query pass with either one.
1900 : :
1901 : 22 : pp = new sqlite_ps (thisdb, "mhd-query-s",
1902 : : "select mtime, sourcetype, source0, source1 from " BUILDIDS "_query_s where buildid = ? and artifactsrc in (?,?) "
1903 [ + - + - : 22 : "order by sharedprefix(source0,source0ref) desc, mtime desc");
+ - + - ]
1904 [ + - ]: 22 : pp->reset();
1905 [ + - ]: 22 : pp->bind(1, buildid);
1906 : : // NB: we don't store the non-canonicalized path names any more, but old databases
1907 : : // might have them (and no canon ones), so we keep searching for both.
1908 [ + - ]: 22 : pp->bind(2, suffix);
1909 [ + - + - ]: 22 : pp->bind(3, canon_pathname(suffix));
1910 : : }
1911 : 702 : unique_ptr<sqlite_ps> ps_closer(pp); // release pp if exception or return
1912 : :
1913 : : // consume all the rows
1914 : : while (1)
1915 : : {
1916 [ + - ]: 380 : int rc = pp->step();
1917 [ + + ]: 380 : if (rc == SQLITE_DONE) break;
1918 [ + + ]: 143 : if (rc != SQLITE_ROW)
1919 [ + - + - ]: 1 : throw sqlite_exception(rc, "step");
1920 : :
1921 [ + - ]: 142 : int64_t b_mtime = sqlite3_column_int64 (*pp, 0);
1922 [ + - + - : 142 : string b_stype = string((const char*) sqlite3_column_text (*pp, 1) ?: ""); /* by DDL may not be NULL */
+ - ]
1923 [ + - + - : 142 : string b_source0 = string((const char*) sqlite3_column_text (*pp, 2) ?: ""); /* may be NULL */
+ - ]
1924 [ + - + + : 142 : string b_source1 = string((const char*) sqlite3_column_text (*pp, 3) ?: ""); /* may be NULL */
+ - ]
1925 : :
1926 [ + - ]: 142 : if (verbose > 1)
1927 [ + - + - : 284 : obatched(clog) << "found mtime=" << b_mtime << " stype=" << b_stype
+ - ]
1928 [ + - + - : 142 : << " source0=" << b_source0 << " source1=" << b_source1 << endl;
+ - + - +
- + - +
- ]
1929 : :
1930 : : // Try accessing the located match.
1931 : : // XXX: in case of multiple matches, attempt them in parallel?
1932 [ + - ]: 142 : auto r = handle_buildid_match (conn ? false : true,
1933 : : b_mtime, b_stype, b_source0, b_source1, result_fd);
1934 [ + + ]: 142 : if (r)
1935 : 113 : return r;
1936 : 29 : }
1937 [ + - ]: 237 : pp->reset();
1938 : :
1939 : : // We couldn't find it in the database. Last ditch effort
1940 : : // is to defer to other debuginfo servers.
1941 : :
1942 : 237 : int fd = -1;
1943 [ + - ]: 237 : debuginfod_client *client = debuginfod_pool_begin ();
1944 [ + - ]: 237 : if (client != NULL)
1945 : : {
1946 [ + - ]: 237 : debuginfod_set_progressfn (client, & debuginfod_find_progress);
1947 : :
1948 [ + - ]: 237 : if (conn)
1949 : : {
1950 : : // Transcribe incoming User-Agent:
1951 [ + - + - : 474 : string ua = MHD_lookup_connection_value (conn, MHD_HEADER_KIND, "User-Agent") ?: "";
+ - ]
1952 [ + - + - ]: 711 : string ua_complete = string("User-Agent: ") + ua;
1953 [ + - ]: 237 : debuginfod_add_http_header (client, ua_complete.c_str());
1954 : :
1955 : : // Compute larger XFF:, for avoiding info loss during
1956 : : // federation, and for future cyclicity detection.
1957 [ + - + + : 474 : string xff = MHD_lookup_connection_value (conn, MHD_HEADER_KIND, "X-Forwarded-For") ?: "";
+ - ]
1958 [ + + ]: 237 : if (xff != "")
1959 [ + - + - ]: 9 : xff += string(", "); // comma separated list
1960 : :
1961 : 237 : unsigned int xff_count = 0;
1962 [ + + ]: 357 : for (auto&& i : xff){
1963 [ + + ]: 120 : if (i == ',') xff_count++;
1964 : : }
1965 : :
1966 : : // if X-Forwarded-For: exceeds N hops,
1967 : : // do not delegate a local lookup miss to upstream debuginfods.
1968 [ + + ]: 237 : if (xff_count >= forwarded_ttl_limit)
1969 [ + - ]: 2 : throw reportable_exception(MHD_HTTP_NOT_FOUND, "not found, --forwared-ttl-limit reached \
1970 [ + - ]: 4 : and will not query the upstream servers");
1971 : :
1972 : : // Compute the client's numeric IP address only - so can't merge with conninfo()
1973 [ + - ]: 235 : const union MHD_ConnectionInfo *u = MHD_get_connection_info (conn,
1974 : : MHD_CONNECTION_INFO_CLIENT_ADDRESS);
1975 [ + - ]: 235 : struct sockaddr *so = u ? u->client_addr : 0;
1976 : 235 : char hostname[256] = ""; // RFC1035
1977 [ + - + - ]: 235 : if (so && so->sa_family == AF_INET)
1978 [ + - ]: 235 : (void) getnameinfo (so, sizeof (struct sockaddr_in), hostname, sizeof (hostname), NULL, 0,
1979 : : NI_NUMERICHOST);
1980 [ # # # # ]: 0 : else if (so && so->sa_family == AF_INET6)
1981 [ # # ]: 0 : (void) getnameinfo (so, sizeof (struct sockaddr_in6), hostname, sizeof (hostname), NULL, 0,
1982 : : NI_NUMERICHOST);
1983 : :
1984 [ + - + - : 940 : string xff_complete = string("X-Forwarded-For: ")+xff+string(hostname);
+ - + - ]
1985 [ + - ]: 235 : debuginfod_add_http_header (client, xff_complete.c_str());
1986 : : }
1987 : :
1988 [ + + ]: 235 : if (artifacttype == "debuginfo")
1989 [ + - ]: 33 : fd = debuginfod_find_debuginfo (client,
1990 : 33 : (const unsigned char*) buildid.c_str(),
1991 : : 0, NULL);
1992 [ + + ]: 202 : else if (artifacttype == "executable")
1993 [ + - ]: 201 : fd = debuginfod_find_executable (client,
1994 : 201 : (const unsigned char*) buildid.c_str(),
1995 : : 0, NULL);
1996 [ + - ]: 1 : else if (artifacttype == "source")
1997 [ + - ]: 1 : fd = debuginfod_find_source (client,
1998 : 1 : (const unsigned char*) buildid.c_str(),
1999 : : 0, suffix.c_str(), NULL);
2000 : : }
2001 : : else
2002 : 0 : fd = -errno; /* Set by debuginfod_begin. */
2003 [ + - ]: 235 : debuginfod_pool_end (client);
2004 : :
2005 [ + + ]: 235 : if (fd >= 0)
2006 : : {
2007 [ + - + - : 2 : inc_metric ("http_responses_total","result","upstream");
+ - + - ]
2008 : : struct stat s;
2009 : 2 : int rc = fstat (fd, &s);
2010 [ + - ]: 2 : if (rc == 0)
2011 : : {
2012 [ + - ]: 2 : auto r = MHD_create_response_from_fd ((uint64_t) s.st_size, fd);
2013 [ + - ]: 2 : if (r)
2014 : : {
2015 [ + - ]: 2 : MHD_add_response_header (r, "Content-Type", "application/octet-stream");
2016 [ + - ]: 2 : add_mhd_last_modified (r, s.st_mtime);
2017 [ + - ]: 2 : if (verbose > 1)
2018 [ + - + - : 2 : obatched(clog) << "serving file from upstream debuginfod/cache" << endl;
+ - ]
2019 [ + - ]: 2 : if (result_fd)
2020 : 2 : *result_fd = fd;
2021 : 2 : return r; // NB: don't close fd; libmicrohttpd will
2022 : : }
2023 : : }
2024 [ # # ]: 0 : close (fd);
2025 : : }
2026 : : else
2027 [ + + + ]: 233 : switch(fd)
2028 : : {
2029 : 117 : case -ENOSYS:
2030 : 117 : break;
2031 : 6 : case -ENOENT:
2032 : 6 : break;
2033 : 110 : default: // some more tricky error
2034 [ + - + - ]: 110 : throw libc_exception(-fd, "upstream debuginfod query failed");
2035 : : }
2036 : :
2037 [ + - + - ]: 123 : throw reportable_exception(MHD_HTTP_NOT_FOUND, "not found");
2038 : : }
2039 : :
2040 : :
2041 : : ////////////////////////////////////////////////////////////////////////
2042 : :
2043 : : static map<string,double> metrics; // arbitrary data for /metrics query
2044 : : // NB: store int64_t since all our metrics are integers; prometheus accepts double
2045 : : static mutex metrics_lock;
2046 : : // NB: these objects get released during the process exit via global dtors
2047 : : // do not call them from within other global dtors
2048 : :
2049 : : // utility function for assembling prometheus-compatible
2050 : : // name="escaped-value" strings
2051 : : // https://prometheus.io/docs/instrumenting/exposition_formats/
2052 : : static string
2053 : 47589 : metric_label(const string& name, const string& value)
2054 : : {
2055 : 47589 : string x = name + "=\"";
2056 [ + + ]: 644122 : for (auto&& c : value)
2057 [ - - - + ]: 597376 : switch(c)
2058 : : {
2059 [ # # ]: 0 : case '\\': x += "\\\\"; break;
2060 [ # # ]: 0 : case '\"': x += "\\\""; break;
2061 [ # # ]: 0 : case '\n': x += "\\n"; break;
2062 [ + - ]: 597376 : default: x += c; break;
2063 : : }
2064 [ + - ]: 47511 : x += "\"";
2065 : 47600 : return x;
2066 : : }
2067 : :
2068 : :
2069 : : // add prometheus-format metric name + label tuple (if any) + value
2070 : :
2071 : : static void
2072 : 988 : set_metric(const string& metric, double value)
2073 : : {
2074 [ + - ]: 988 : unique_lock<mutex> lock(metrics_lock);
2075 [ + - ]: 988 : metrics[metric] = value;
2076 : 988 : }
2077 : : #if 0 /* unused */
2078 : : static void
2079 : : inc_metric(const string& metric)
2080 : : {
2081 : : unique_lock<mutex> lock(metrics_lock);
2082 : : metrics[metric] ++;
2083 : : }
2084 : : #endif
2085 : : static void
2086 : 2043 : set_metric(const string& metric,
2087 : : const string& lname, const string& lvalue,
2088 : : double value)
2089 : : {
2090 [ + - + - : 6129 : string key = (metric + "{" + metric_label(lname, lvalue) + "}");
+ - + - ]
2091 [ + - ]: 2044 : unique_lock<mutex> lock(metrics_lock);
2092 [ + - ]: 2044 : metrics[key] = value;
2093 : 2044 : }
2094 : :
2095 : : static void
2096 : 19566 : inc_metric(const string& metric,
2097 : : const string& lname, const string& lvalue)
2098 : : {
2099 [ + - + - : 58694 : string key = (metric + "{" + metric_label(lname, lvalue) + "}");
+ - + - ]
2100 [ + - ]: 19566 : unique_lock<mutex> lock(metrics_lock);
2101 [ + - ]: 19566 : metrics[key] ++;
2102 : 19565 : }
2103 : : static void
2104 : 18631 : add_metric(const string& metric,
2105 : : const string& lname, const string& lvalue,
2106 : : double value)
2107 : : {
2108 [ + - + - : 55885 : string key = (metric + "{" + metric_label(lname, lvalue) + "}");
+ - + - ]
2109 [ + - ]: 18631 : unique_lock<mutex> lock(metrics_lock);
2110 [ + - ]: 18635 : metrics[key] += value;
2111 : 18634 : }
2112 : : #if 0
2113 : : static void
2114 : : add_metric(const string& metric,
2115 : : double value)
2116 : : {
2117 : : unique_lock<mutex> lock(metrics_lock);
2118 : : metrics[metric] += value;
2119 : : }
2120 : : #endif
2121 : :
2122 : :
2123 : : // and more for higher arity labels if needed
2124 : :
2125 : : static void
2126 : 1839 : inc_metric(const string& metric,
2127 : : const string& lname, const string& lvalue,
2128 : : const string& rname, const string& rvalue)
2129 : : {
2130 [ + - ]: 1839 : string key = (metric + "{"
2131 [ + - + - : 3678 : + metric_label(lname, lvalue) + ","
+ - ]
2132 [ + - + - : 5517 : + metric_label(rname, rvalue) + "}");
+ - ]
2133 [ + - ]: 1839 : unique_lock<mutex> lock(metrics_lock);
2134 [ + - ]: 1839 : metrics[key] ++;
2135 : 1839 : }
2136 : : static void
2137 : 1839 : add_metric(const string& metric,
2138 : : const string& lname, const string& lvalue,
2139 : : const string& rname, const string& rvalue,
2140 : : double value)
2141 : : {
2142 [ + - ]: 1839 : string key = (metric + "{"
2143 [ + - + - : 3678 : + metric_label(lname, lvalue) + ","
+ - ]
2144 [ + - + - : 5517 : + metric_label(rname, rvalue) + "}");
+ - ]
2145 [ + - ]: 1839 : unique_lock<mutex> lock(metrics_lock);
2146 [ + - ]: 1839 : metrics[key] += value;
2147 : 1839 : }
2148 : :
2149 : : static struct MHD_Response*
2150 : 272 : handle_metrics (off_t* size)
2151 : : {
2152 [ + - ]: 544 : stringstream o;
2153 : : {
2154 [ + - ]: 544 : unique_lock<mutex> lock(metrics_lock);
2155 [ + + ]: 23515 : for (auto&& i : metrics)
2156 : 23243 : o << i.first
2157 : : << " "
2158 [ + - + - : 23243 : << std::setprecision(std::numeric_limits<double>::digits10 + 1)
+ - ]
2159 [ + - ]: 23243 : << i.second
2160 [ + - ]: 23243 : << endl;
2161 : : }
2162 [ + - ]: 272 : const string& os = o.str();
2163 [ + - ]: 544 : MHD_Response* r = MHD_create_response_from_buffer (os.size(),
2164 : 272 : (void*) os.c_str(),
2165 : : MHD_RESPMEM_MUST_COPY);
2166 : 272 : *size = os.size();
2167 [ + - ]: 272 : MHD_add_response_header (r, "Content-Type", "text/plain");
2168 : 544 : return r;
2169 : : }
2170 : :
2171 : : static struct MHD_Response*
2172 : 0 : handle_root (off_t* size)
2173 : : {
2174 [ # # # # : 0 : static string version = "debuginfod (" + string (PACKAGE_NAME) + ") "
# # ]
2175 [ # # # # : 0 : + string (PACKAGE_VERSION);
# # # # #
# ]
2176 : 0 : MHD_Response* r = MHD_create_response_from_buffer (version.size (),
2177 : 0 : (void *) version.c_str (),
2178 : : MHD_RESPMEM_PERSISTENT);
2179 : 0 : *size = version.size ();
2180 : 0 : MHD_add_response_header (r, "Content-Type", "text/plain");
2181 : 0 : return r;
2182 : : }
2183 : :
2184 : :
2185 : : ////////////////////////////////////////////////////////////////////////
2186 : :
2187 : :
2188 : : /* libmicrohttpd callback */
2189 : : static MHD_RESULT
2190 : 1225 : handler_cb (void * /*cls*/,
2191 : : struct MHD_Connection *connection,
2192 : : const char *url,
2193 : : const char *method,
2194 : : const char * /*version*/,
2195 : : const char * /*upload_data*/,
2196 : : size_t * /*upload_data_size*/,
2197 : : void ** ptr)
2198 : : {
2199 : 1225 : struct MHD_Response *r = NULL;
2200 [ + - ]: 2450 : string url_copy = url;
2201 : :
2202 : : /* libmicrohttpd always makes (at least) two callbacks: once just
2203 : : past the headers, and one after the request body is finished
2204 : : being received. If we process things early (first callback) and
2205 : : queue a response, libmicrohttpd would suppress http keep-alive
2206 : : (via connection->read_closed = true). */
2207 : : static int aptr; /* just some random object to use as a flag */
2208 [ + + ]: 1225 : if (&aptr != *ptr)
2209 : : {
2210 : : /* do never respond on first call */
2211 : 612 : *ptr = &aptr;
2212 : 612 : return MHD_YES;
2213 : : }
2214 : 613 : *ptr = NULL; /* reset when done */
2215 : :
2216 [ + - ]: 613 : const char *maxsize_string = MHD_lookup_connection_value(connection, MHD_HEADER_KIND, "X-DEBUGINFOD-MAXSIZE");
2217 : 611 : long maxsize = 0;
2218 [ + + + - ]: 611 : if (maxsize_string != NULL && maxsize_string[0] != '\0')
2219 : 1 : maxsize = atol(maxsize_string);
2220 : : else
2221 : 610 : maxsize = 0;
2222 : :
2223 : : #if MHD_VERSION >= 0x00097002
2224 : : enum MHD_Result rc;
2225 : : #else
2226 : : int rc = MHD_NO; // mhd
2227 : : #endif
2228 : 611 : int http_code = 500;
2229 : 611 : off_t http_size = -1;
2230 : : struct timespec ts_start, ts_end;
2231 : 611 : clock_gettime (CLOCK_MONOTONIC, &ts_start);
2232 : 611 : double afteryou = 0.0;
2233 : 1224 : string artifacttype, suffix;
2234 : :
2235 : : try
2236 : : {
2237 [ + - - + ]: 613 : if (string(method) != "GET")
2238 [ # # # # ]: 0 : throw reportable_exception(400, "we support GET only");
2239 : :
2240 : : /* Start decoding the URL. */
2241 : 611 : size_t slash1 = url_copy.find('/', 1);
2242 [ + - ]: 1226 : string url1 = url_copy.substr(0, slash1); // ok even if slash1 not found
2243 : :
2244 [ + + + - : 612 : if (slash1 != string::npos && url1 == "/buildid")
+ + ]
2245 : : {
2246 : : // PR27863: block this thread awhile if another thread is already busy
2247 : : // fetching the exact same thing. This is better for Everyone.
2248 : : // The latecomer says "... after you!" and waits.
2249 [ + - + - : 335 : add_metric ("thread_busy", "role", "http-buildid-after-you", 1);
+ - + - ]
2250 : : #ifdef HAVE_PTHREAD_SETNAME_NP
2251 : 338 : (void) pthread_setname_np (pthread_self(), "mhd-buildid-after-you");
2252 : : #endif
2253 : : struct timespec tsay_start, tsay_end;
2254 : 338 : clock_gettime (CLOCK_MONOTONIC, &tsay_start);
2255 [ + + + - ]: 338 : static unique_set<string> busy_urls;
2256 [ + - ]: 676 : unique_set_reserver<string> after_you(busy_urls, url_copy);
2257 : 338 : clock_gettime (CLOCK_MONOTONIC, &tsay_end);
2258 : 338 : afteryou = (tsay_end.tv_sec - tsay_start.tv_sec) + (tsay_end.tv_nsec - tsay_start.tv_nsec)/1.e9;
2259 [ + - + - : 338 : add_metric ("thread_busy", "role", "http-buildid-after-you", -1);
+ - + - ]
2260 : :
2261 [ + - + - : 1690 : tmp_inc_metric m ("thread_busy", "role", "http-buildid");
+ - + - ]
2262 : : #ifdef HAVE_PTHREAD_SETNAME_NP
2263 : 338 : (void) pthread_setname_np (pthread_self(), "mhd-buildid");
2264 : : #endif
2265 : 338 : size_t slash2 = url_copy.find('/', slash1+1);
2266 [ - + ]: 338 : if (slash2 == string::npos)
2267 [ # # # # ]: 0 : throw reportable_exception("/buildid/ webapi error, need buildid");
2268 : :
2269 [ + - ]: 676 : string buildid = url_copy.substr(slash1+1, slash2-slash1-1);
2270 : :
2271 : 338 : size_t slash3 = url_copy.find('/', slash2+1);
2272 : :
2273 [ + + ]: 338 : if (slash3 == string::npos)
2274 : : {
2275 [ + - ]: 316 : artifacttype = url_copy.substr(slash2+1);
2276 [ + - ]: 316 : suffix = "";
2277 : : }
2278 : : else
2279 : : {
2280 [ + - ]: 22 : artifacttype = url_copy.substr(slash2+1, slash3-slash2-1);
2281 [ + - ]: 22 : suffix = url_copy.substr(slash3); // include the slash in the suffix
2282 : : }
2283 : :
2284 : : // get the resulting fd so we can report its size
2285 : : int fd;
2286 [ + + ]: 338 : r = handle_buildid(connection, buildid, artifacttype, suffix, &fd);
2287 [ + - ]: 99 : if (r)
2288 : : {
2289 : : struct stat fs;
2290 [ + - ]: 99 : if (fstat(fd, &fs) == 0)
2291 : 99 : http_size = fs.st_size;
2292 : : // libmicrohttpd will close (fd);
2293 : : }
2294 : : }
2295 [ + + ]: 275 : else if (url1 == "/metrics")
2296 : : {
2297 [ + - + - : 1088 : tmp_inc_metric m ("thread_busy", "role", "http-metrics");
+ - + - ]
2298 [ + - ]: 272 : artifacttype = "metrics";
2299 [ + - + - : 272 : inc_metric("http_requests_total", "type", artifacttype);
+ - ]
2300 [ + - ]: 272 : r = handle_metrics(& http_size);
2301 : : }
2302 [ - + ]: 3 : else if (url1 == "/")
2303 : : {
2304 [ # # ]: 0 : artifacttype = "/";
2305 [ # # # # : 0 : inc_metric("http_requests_total", "type", artifacttype);
# # ]
2306 [ # # ]: 0 : r = handle_root(& http_size);
2307 : : }
2308 : : else
2309 [ + - + - : 3 : throw reportable_exception("webapi error, unrecognized '" + url1 + "'");
+ - ]
2310 : :
2311 [ - + ]: 371 : if (r == 0)
2312 [ # # # # ]: 0 : throw reportable_exception("internal error, missing response");
2313 : :
2314 [ + + + - ]: 371 : if (maxsize > 0 && http_size > maxsize)
2315 : : {
2316 [ + - ]: 1 : MHD_destroy_response(r);
2317 [ + - + - : 1 : throw reportable_exception(406, "File too large, max size=" + std::to_string(maxsize));
+ - ]
2318 : : }
2319 : :
2320 [ + - ]: 370 : rc = MHD_queue_response (connection, MHD_HTTP_OK, r);
2321 : 370 : http_code = MHD_HTTP_OK;
2322 [ + - ]: 370 : MHD_destroy_response (r);
2323 : : }
2324 [ - + ]: 243 : catch (const reportable_exception& e)
2325 : : {
2326 [ + - + - : 243 : inc_metric("http_responses_total","result","error");
+ - + - ]
2327 [ + - ]: 243 : e.report(clog);
2328 : 243 : http_code = e.code;
2329 : 243 : http_size = e.message.size();
2330 [ + - ]: 243 : rc = e.mhd_send_response (connection);
2331 : : }
2332 : :
2333 : 613 : clock_gettime (CLOCK_MONOTONIC, &ts_end);
2334 : 613 : double deltas = (ts_end.tv_sec - ts_start.tv_sec) + (ts_end.tv_nsec - ts_start.tv_nsec)/1.e9;
2335 : : // afteryou: delay waiting for other client's identical query to complete
2336 : : // deltas: total latency, including afteryou waiting
2337 [ + - + - : 1226 : obatched(clog) << conninfo(connection)
+ - ]
2338 : : << ' ' << method << ' ' << url
2339 [ + - + - : 613 : << ' ' << http_code << ' ' << http_size
+ - + - +
- + - +
- ]
2340 [ + - + - : 613 : << ' ' << (int)(afteryou*1000) << '+' << (int)((deltas-afteryou)*1000) << "ms"
+ - + - +
- + - ]
2341 [ + - ]: 613 : << endl;
2342 : :
2343 : : // related prometheus metrics
2344 [ + - ]: 613 : string http_code_str = to_string(http_code);
2345 [ + - ]: 613 : add_metric("http_responses_transfer_bytes_sum",
2346 [ + - + - : 1226 : "code", http_code_str, "type", artifacttype, http_size);
+ - ]
2347 [ + - ]: 613 : inc_metric("http_responses_transfer_bytes_count",
2348 [ + - + - : 1226 : "code", http_code_str, "type", artifacttype);
+ - ]
2349 : :
2350 [ + - ]: 613 : add_metric("http_responses_duration_milliseconds_sum",
2351 [ + - + - : 1226 : "code", http_code_str, "type", artifacttype, deltas*1000); // prometheus prefers _seconds and floating point
+ - ]
2352 [ + - ]: 613 : inc_metric("http_responses_duration_milliseconds_count",
2353 [ + - + - : 1226 : "code", http_code_str, "type", artifacttype);
+ - ]
2354 : :
2355 [ + - ]: 613 : add_metric("http_responses_after_you_milliseconds_sum",
2356 [ + - + - : 1226 : "code", http_code_str, "type", artifacttype, afteryou*1000);
+ - ]
2357 [ + - ]: 613 : inc_metric("http_responses_after_you_milliseconds_count",
2358 [ + - + - : 1226 : "code", http_code_str, "type", artifacttype);
+ - ]
2359 : :
2360 : 613 : return rc;
2361 : : }
2362 : :
2363 : :
2364 : : ////////////////////////////////////////////////////////////////////////
2365 : : // borrowed originally from src/nm.c get_local_names()
2366 : :
2367 : : static void
2368 : 78 : dwarf_extract_source_paths (Elf *elf, set<string>& debug_sourcefiles)
2369 : : noexcept // no exceptions - so we can simplify the altdbg resource release at end
2370 : : {
2371 : 78 : Dwarf* dbg = dwarf_begin_elf (elf, DWARF_C_READ, NULL);
2372 [ - + ]: 78 : if (dbg == NULL)
2373 : 0 : return;
2374 : :
2375 : 78 : Dwarf* altdbg = NULL;
2376 : 78 : int altdbg_fd = -1;
2377 : :
2378 : : // DWZ handling: if we have an unsatisfied debug-alt-link, add an
2379 : : // empty string into the outgoing sourcefiles set, so the caller
2380 : : // should know that our data is incomplete.
2381 : : const char *alt_name_p;
2382 : : const void *alt_build_id; // elfutils-owned memory
2383 : 78 : ssize_t sz = dwelf_dwarf_gnu_debugaltlink (dbg, &alt_name_p, &alt_build_id);
2384 [ + + ]: 78 : if (sz > 0) // got one!
2385 : : {
2386 : 32 : string buildid;
2387 : 16 : unsigned char* build_id_bytes = (unsigned char*) alt_build_id;
2388 [ + + ]: 336 : for (ssize_t idx=0; idx<sz; idx++)
2389 : : {
2390 : 320 : buildid += "0123456789abcdef"[build_id_bytes[idx] >> 4];
2391 : 320 : buildid += "0123456789abcdef"[build_id_bytes[idx] & 0xf];
2392 : : }
2393 : :
2394 [ + - ]: 16 : if (verbose > 3)
2395 : 16 : obatched(clog) << "Need altdebug buildid=" << buildid << endl;
2396 : :
2397 : : // but is it unsatisfied the normal elfutils ways?
2398 : 16 : Dwarf* alt = dwarf_getalt (dbg);
2399 [ + - ]: 16 : if (alt == NULL)
2400 : : {
2401 : : // Yup, unsatisfied the normal way. Maybe we can satisfy it
2402 : : // from our own debuginfod database.
2403 : : int alt_fd;
2404 : 16 : struct MHD_Response *r = 0;
2405 : : try
2406 : : {
2407 [ + - ]: 32 : string artifacttype = "debuginfo";
2408 [ + - + - ]: 16 : r = handle_buildid (0, buildid, artifacttype, "", &alt_fd);
2409 : : }
2410 [ - - ]: 0 : catch (const reportable_exception& e)
2411 : : {
2412 : : // swallow exceptions
2413 : : }
2414 : :
2415 : : // NB: this is not actually recursive! This invokes the web-query
2416 : : // path, which cannot get back into the scan code paths.
2417 [ + - ]: 16 : if (r)
2418 : : {
2419 : : // Found it!
2420 : 16 : altdbg_fd = dup(alt_fd); // ok if this fails, downstream failures ok
2421 : 16 : alt = altdbg = dwarf_begin (altdbg_fd, DWARF_C_READ);
2422 : : // NB: must close this dwarf and this fd at the bottom of the function!
2423 : 16 : MHD_destroy_response (r); // will close alt_fd
2424 [ + - ]: 16 : if (alt)
2425 : 16 : dwarf_setalt (dbg, alt);
2426 : : }
2427 : : }
2428 : : else
2429 : : {
2430 : : // NB: dwarf_setalt(alt) inappropriate - already done!
2431 : : // NB: altdbg will stay 0 so nothing tries to redundantly dealloc.
2432 : : }
2433 : :
2434 [ + - ]: 16 : if (alt)
2435 : : {
2436 [ + - ]: 16 : if (verbose > 3)
2437 : 16 : obatched(clog) << "Resolved altdebug buildid=" << buildid << endl;
2438 : : }
2439 : : else // (alt == NULL) - signal possible presence of poor debuginfo
2440 : : {
2441 : 0 : debug_sourcefiles.insert("");
2442 [ # # ]: 0 : if (verbose > 3)
2443 : 0 : obatched(clog) << "Unresolved altdebug buildid=" << buildid << endl;
2444 : : }
2445 : : }
2446 : :
2447 : 78 : Dwarf_Off offset = 0;
2448 : : Dwarf_Off old_offset;
2449 : : size_t hsize;
2450 : :
2451 [ + + ]: 1068 : while (dwarf_nextcu (dbg, old_offset = offset, &offset, &hsize, NULL, NULL, NULL) == 0)
2452 : : {
2453 : : Dwarf_Die cudie_mem;
2454 : 990 : Dwarf_Die *cudie = dwarf_offdie (dbg, old_offset + hsize, &cudie_mem);
2455 : :
2456 [ - + ]: 989 : if (cudie == NULL)
2457 : 8 : continue;
2458 [ + + ]: 989 : if (dwarf_tag (cudie) != DW_TAG_compile_unit)
2459 : 8 : continue;
2460 : :
2461 [ + - ]: 981 : const char *cuname = dwarf_diename(cudie) ?: "unknown";
2462 : :
2463 : : Dwarf_Files *files;
2464 : : size_t nfiles;
2465 [ - + ]: 981 : if (dwarf_getsrcfiles (cudie, &files, &nfiles) != 0)
2466 : 0 : continue;
2467 : :
2468 : : // extract DW_AT_comp_dir to resolve relative file names
2469 : 982 : const char *comp_dir = "";
2470 : : const char *const *dirs;
2471 : : size_t ndirs;
2472 [ + - + - ]: 1964 : if (dwarf_getsrcdirs (files, &dirs, &ndirs) == 0 &&
2473 [ + - ]: 982 : dirs[0] != NULL)
2474 : 982 : comp_dir = dirs[0];
2475 [ - + ]: 982 : if (comp_dir == NULL)
2476 : 0 : comp_dir = "";
2477 : :
2478 [ + + ]: 982 : if (verbose > 3)
2479 : 114 : obatched(clog) << "searching for sources for cu=" << cuname << " comp_dir=" << comp_dir
2480 : 57 : << " #files=" << nfiles << " #dirs=" << ndirs << endl;
2481 : :
2482 [ - + - - ]: 982 : if (comp_dir[0] == '\0' && cuname[0] != '/')
2483 : : {
2484 : : // This is a common symptom for dwz-compressed debug files,
2485 : : // where the altdebug file cannot be resolved.
2486 [ # # ]: 0 : if (verbose > 3)
2487 : 0 : obatched(clog) << "skipping cu=" << cuname << " due to empty comp_dir" << endl;
2488 : 0 : continue;
2489 : : }
2490 : :
2491 [ + + ]: 15266 : for (size_t f = 1; f < nfiles; f++)
2492 : : {
2493 : 14284 : const char *hat = dwarf_filesrc (files, f, NULL, NULL);
2494 [ - + ]: 14284 : if (hat == NULL)
2495 : 0 : continue;
2496 : :
2497 [ - + ]: 14284 : if (string(hat) == "<built-in>") // gcc intrinsics, don't bother record
2498 : 0 : continue;
2499 : :
2500 : 14284 : string waldo;
2501 [ + + ]: 14284 : if (hat[0] == '/') // absolute
2502 : 8814 : waldo = (string (hat));
2503 [ + - ]: 5470 : else if (comp_dir[0] != '\0') // comp_dir relative
2504 : 5470 : waldo = (string (comp_dir) + string("/") + string (hat));
2505 : : else
2506 : : {
2507 [ # # ]: 0 : if (verbose > 3)
2508 : 0 : obatched(clog) << "skipping hat=" << hat << " due to empty comp_dir" << endl;
2509 : 0 : continue;
2510 : : }
2511 : :
2512 : : // NB: this is the 'waldo' that a dbginfo client will have
2513 : : // to supply for us to give them the file The comp_dir
2514 : : // prefixing is a definite complication. Otherwise we'd
2515 : : // have to return a setof comp_dirs (one per CU!) with
2516 : : // corresponding filesrc[] names, instead of one absolute
2517 : : // resoved set. Maybe we'll have to do that anyway. XXX
2518 : :
2519 [ - + ]: 14284 : if (verbose > 4)
2520 : 0 : obatched(clog) << waldo
2521 [ # # ]: 0 : << (debug_sourcefiles.find(waldo)==debug_sourcefiles.end() ? " new" : " dup") << endl;
2522 : :
2523 : 14284 : debug_sourcefiles.insert (waldo);
2524 : : }
2525 : : }
2526 : :
2527 : 78 : dwarf_end(dbg);
2528 [ + + ]: 78 : if (altdbg)
2529 : 16 : dwarf_end(altdbg);
2530 [ + + ]: 78 : if (altdbg_fd >= 0)
2531 : 16 : close(altdbg_fd);
2532 : : }
2533 : :
2534 : :
2535 : :
2536 : : static void
2537 : 368 : elf_classify (int fd, bool &executable_p, bool &debuginfo_p, string &buildid, set<string>& debug_sourcefiles)
2538 : : {
2539 : 368 : Elf *elf = elf_begin (fd, ELF_C_READ_MMAP_PRIVATE, NULL);
2540 [ - + ]: 372 : if (elf == NULL)
2541 : 0 : return;
2542 : :
2543 : : try // catch our types of errors and clean up the Elf* object
2544 : : {
2545 [ + - + + ]: 372 : if (elf_kind (elf) != ELF_K_ELF)
2546 : : {
2547 [ + - ]: 221 : elf_end (elf);
2548 : 222 : return;
2549 : : }
2550 : :
2551 : : GElf_Ehdr ehdr_storage;
2552 [ + - ]: 151 : GElf_Ehdr *ehdr = gelf_getehdr (elf, &ehdr_storage);
2553 [ - + ]: 151 : if (ehdr == NULL)
2554 : : {
2555 [ # # ]: 0 : elf_end (elf);
2556 : 0 : return;
2557 : : }
2558 : 151 : auto elf_type = ehdr->e_type;
2559 : :
2560 : : const void *build_id; // elfutils-owned memory
2561 [ + - ]: 151 : ssize_t sz = dwelf_elf_gnu_build_id (elf, & build_id);
2562 [ - + ]: 151 : if (sz <= 0)
2563 : : {
2564 : : // It's not a diagnostic-worthy error for an elf file to lack build-id.
2565 : : // It might just be very old.
2566 [ # # ]: 0 : elf_end (elf);
2567 : 0 : return;
2568 : : }
2569 : :
2570 : : // build_id is a raw byte array; convert to hexadecimal *lowercase*
2571 : 151 : unsigned char* build_id_bytes = (unsigned char*) build_id;
2572 [ + + ]: 3143 : for (ssize_t idx=0; idx<sz; idx++)
2573 : : {
2574 [ + - ]: 2992 : buildid += "0123456789abcdef"[build_id_bytes[idx] >> 4];
2575 [ + - ]: 2987 : buildid += "0123456789abcdef"[build_id_bytes[idx] & 0xf];
2576 : : }
2577 : :
2578 : : // now decide whether it's an executable - namely, any allocatable section has
2579 : : // PROGBITS;
2580 [ + + + + ]: 151 : if (elf_type == ET_EXEC || elf_type == ET_DYN)
2581 : : {
2582 : : size_t shnum;
2583 [ + - ]: 131 : int rc = elf_getshdrnum (elf, &shnum);
2584 [ - + ]: 133 : if (rc < 0)
2585 [ # # # # ]: 0 : throw elfutils_exception(rc, "getshdrnum");
2586 : :
2587 : 133 : executable_p = false;
2588 [ + + ]: 2486 : for (size_t sc = 0; sc < shnum; sc++)
2589 : : {
2590 [ + - ]: 2427 : Elf_Scn *scn = elf_getscn (elf, sc);
2591 [ - + ]: 2432 : if (scn == NULL)
2592 : 0 : continue;
2593 : :
2594 : : GElf_Shdr shdr_mem;
2595 [ + - ]: 2432 : GElf_Shdr *shdr = gelf_getshdr (scn, &shdr_mem);
2596 [ - + ]: 2423 : if (shdr == NULL)
2597 : 0 : continue;
2598 : :
2599 : : // allocated (loadable / vm-addr-assigned) section with available content?
2600 [ + + + + ]: 2423 : if ((shdr->sh_type == SHT_PROGBITS) && (shdr->sh_flags & SHF_ALLOC))
2601 : : {
2602 [ - + ]: 70 : if (verbose > 4)
2603 [ # # # # : 0 : obatched(clog) << "executable due to SHF_ALLOC SHT_PROGBITS sc=" << sc << endl;
# # # # ]
2604 : 70 : executable_p = true;
2605 : 70 : break; // no need to keep looking for others
2606 : : }
2607 : : } // iterate over sections
2608 : : } // executable_p classification
2609 : :
2610 : : // now decide whether it's a debuginfo - namely, if it has any .debug* or .zdebug* sections
2611 : : // logic mostly stolen from fweimer@redhat.com's elfclassify drafts
2612 : : size_t shstrndx;
2613 [ + - ]: 149 : int rc = elf_getshdrstrndx (elf, &shstrndx);
2614 [ - + ]: 151 : if (rc < 0)
2615 [ # # # # ]: 0 : throw elfutils_exception(rc, "getshdrstrndx");
2616 : :
2617 : 151 : Elf_Scn *scn = NULL;
2618 : 151 : bool symtab_p = false;
2619 : 151 : bool bits_alloc_p = false;
2620 : : while (true)
2621 : : {
2622 [ + - ]: 3958 : scn = elf_nextscn (elf, scn);
2623 [ + + ]: 3966 : if (scn == NULL)
2624 : 73 : break;
2625 : : GElf_Shdr shdr_storage;
2626 [ + - ]: 3893 : GElf_Shdr *shdr = gelf_getshdr (scn, &shdr_storage);
2627 [ - + ]: 3893 : if (shdr == NULL)
2628 : 0 : break;
2629 [ + - ]: 3893 : const char *section_name = elf_strptr (elf, shstrndx, shdr->sh_name);
2630 [ - + ]: 3891 : if (section_name == NULL)
2631 : 0 : break;
2632 [ + + + + : 7703 : if (startswith (section_name, ".debug_line") ||
+ + ]
2633 : 3815 : startswith (section_name, ".zdebug_line"))
2634 : : {
2635 : 78 : debuginfo_p = true;
2636 : 78 : dwarf_extract_source_paths (elf, debug_sourcefiles);
2637 : 78 : break; // expecting only one .*debug_line, so no need to look for others
2638 : : }
2639 [ + + - + : 7371 : else if (startswith (section_name, ".debug_") ||
+ + ]
2640 : 3558 : startswith (section_name, ".zdebug_"))
2641 : : {
2642 : 246 : debuginfo_p = true;
2643 : : // NB: don't break; need to parse .debug_line for sources
2644 : : }
2645 [ + + ]: 3561 : else if (shdr->sh_type == SHT_SYMTAB)
2646 : : {
2647 : 11 : symtab_p = true;
2648 : : }
2649 [ + + ]: 3550 : else if (shdr->sh_type != SHT_NOBITS
2650 [ + + ]: 2073 : && shdr->sh_type != SHT_NOTE
2651 [ + + ]: 1750 : && (shdr->sh_flags & SHF_ALLOC) != 0)
2652 : : {
2653 : 1495 : bits_alloc_p = true;
2654 : : }
2655 : 3807 : }
2656 : :
2657 : : // For more expansive elf/split-debuginfo classification, we
2658 : : // want to identify as debuginfo "strip -s"-produced files
2659 : : // without .debug_info* (like libicudata), but we don't want to
2660 : : // identify "strip -g" executables (with .symtab left there).
2661 [ + + - + ]: 151 : if (symtab_p && !bits_alloc_p)
2662 : 0 : debuginfo_p = true;
2663 : : }
2664 [ # # ]: 0 : catch (const reportable_exception& e)
2665 : : {
2666 [ # # ]: 0 : e.report(clog);
2667 : : }
2668 : 151 : elf_end (elf);
2669 : : }
2670 : :
2671 : :
2672 : : static void
2673 : 317 : scan_source_file (const string& rps, const stat_t& st,
2674 : : sqlite_ps& ps_upsert_buildids,
2675 : : sqlite_ps& ps_upsert_files,
2676 : : sqlite_ps& ps_upsert_de,
2677 : : sqlite_ps& ps_upsert_s,
2678 : : sqlite_ps& ps_query,
2679 : : sqlite_ps& ps_scan_done,
2680 : : unsigned& fts_cached,
2681 : : unsigned& fts_executable,
2682 : : unsigned& fts_debuginfo,
2683 : : unsigned& fts_sourcefiles)
2684 : : {
2685 : : /* See if we know of it already. */
2686 : : int rc = ps_query
2687 [ + - ]: 317 : .reset()
2688 [ + - ]: 319 : .bind(1, rps)
2689 [ + - ]: 319 : .bind(2, st.st_mtime)
2690 [ + - ]: 318 : .step();
2691 [ + - ]: 319 : ps_query.reset();
2692 [ + + ]: 319 : if (rc == SQLITE_ROW) // i.e., a result, as opposed to DONE (no results)
2693 : : // no need to recheck a file/version we already know
2694 : : // specifically, no need to elf-begin a file we already determined is non-elf
2695 : : // (so is stored with buildid=NULL)
2696 : : {
2697 : 162 : fts_cached++;
2698 : 162 : return;
2699 : : }
2700 : :
2701 : 157 : bool executable_p = false, debuginfo_p = false; // E and/or D
2702 : 314 : string buildid;
2703 : 314 : set<string> sourcefiles;
2704 : :
2705 [ + - ]: 157 : int fd = open (rps.c_str(), O_RDONLY);
2706 : : try
2707 : : {
2708 [ + + ]: 149 : if (fd >= 0)
2709 [ + - ]: 148 : elf_classify (fd, executable_p, debuginfo_p, buildid, sourcefiles);
2710 : : else
2711 [ + - + - : 1 : throw libc_exception(errno, string("open ") + rps);
+ - ]
2712 : 156 : add_metric ("scanned_bytes_total","source","file",
2713 [ + - + - : 156 : st.st_size);
+ - + - ]
2714 [ + - + - : 156 : inc_metric ("scanned_files_total","source","file");
+ - + - ]
2715 : : }
2716 : : // NB: we catch exceptions here too, so that we can
2717 : : // cache the corrupt-elf case (!executable_p &&
2718 : : // !debuginfo_p) just below, just as if we had an
2719 : : // EPERM error from open(2).
2720 [ - + ]: 2 : catch (const reportable_exception& e)
2721 : : {
2722 [ + - ]: 1 : e.report(clog);
2723 : : }
2724 : :
2725 [ + + ]: 157 : if (fd >= 0)
2726 [ + - ]: 156 : close (fd);
2727 : :
2728 : : // register this file name in the interning table
2729 : : ps_upsert_files
2730 [ + - ]: 157 : .reset()
2731 [ + - ]: 157 : .bind(1, rps)
2732 [ + - ]: 157 : .step_ok_done();
2733 : :
2734 [ + + ]: 157 : if (buildid == "")
2735 : : {
2736 : : // no point storing an elf file without buildid
2737 : 128 : executable_p = false;
2738 : 128 : debuginfo_p = false;
2739 : : }
2740 : : else
2741 : : {
2742 : : // register this build-id in the interning table
2743 : : ps_upsert_buildids
2744 [ + - ]: 29 : .reset()
2745 [ + - ]: 29 : .bind(1, buildid)
2746 [ + - ]: 29 : .step_ok_done();
2747 : : }
2748 : :
2749 [ + + ]: 157 : if (executable_p)
2750 : 18 : fts_executable ++;
2751 [ + + ]: 157 : if (debuginfo_p)
2752 : 18 : fts_debuginfo ++;
2753 [ + + + + ]: 157 : if (executable_p || debuginfo_p)
2754 : : {
2755 : : ps_upsert_de
2756 [ + - ]: 29 : .reset()
2757 [ + - ]: 29 : .bind(1, buildid)
2758 [ + + + - ]: 29 : .bind(2, debuginfo_p ? 1 : 0)
2759 [ + + + - ]: 29 : .bind(3, executable_p ? 1 : 0)
2760 [ + - ]: 29 : .bind(4, rps)
2761 [ + - ]: 29 : .bind(5, st.st_mtime)
2762 [ + - ]: 29 : .step_ok_done();
2763 : : }
2764 [ + + ]: 157 : if (executable_p)
2765 [ + - + - : 18 : inc_metric("found_executable_total","source","files");
+ - + - ]
2766 [ + + ]: 157 : if (debuginfo_p)
2767 [ + - + - : 18 : inc_metric("found_debuginfo_total","source","files");
+ - + - ]
2768 : :
2769 [ + + + - : 157 : if (sourcefiles.size() && buildid != "")
+ + ]
2770 : : {
2771 : 18 : fts_sourcefiles += sourcefiles.size();
2772 : :
2773 [ + + ]: 1479 : for (auto&& dwarfsrc : sourcefiles)
2774 : : {
2775 : 1461 : char *srp = realpath(dwarfsrc.c_str(), NULL);
2776 [ + + ]: 1461 : if (srp == NULL) // also if DWZ unresolved dwarfsrc=""
2777 : 18 : continue; // unresolvable files are not a serious problem
2778 : : // throw libc_exception(errno, "fts/file realpath " + srcpath);
2779 [ + - ]: 1443 : string srps = string(srp);
2780 : 1443 : free (srp);
2781 : :
2782 : : struct stat sfs;
2783 : 1443 : rc = stat(srps.c_str(), &sfs);
2784 [ - + ]: 1443 : if (rc != 0)
2785 : 0 : continue;
2786 : :
2787 [ + - ]: 1443 : if (verbose > 2)
2788 [ + - + - ]: 2886 : obatched(clog) << "recorded buildid=" << buildid << " file=" << srps
2789 [ + - + - : 1443 : << " mtime=" << sfs.st_mtime
+ - + - +
- ]
2790 [ + - + - : 1443 : << " as source " << dwarfsrc << endl;
+ - ]
2791 : :
2792 : : ps_upsert_files
2793 [ + - ]: 1443 : .reset()
2794 [ + - ]: 1443 : .bind(1, srps)
2795 [ + - ]: 1443 : .step_ok_done();
2796 : :
2797 : : // PR25548: store canonicalized dwarfsrc path
2798 [ + - ]: 2886 : string dwarfsrc_canon = canon_pathname (dwarfsrc);
2799 [ + + ]: 1443 : if (dwarfsrc_canon != dwarfsrc)
2800 : : {
2801 [ + + ]: 272 : if (verbose > 3)
2802 [ + - + - : 5 : obatched(clog) << "canonicalized src=" << dwarfsrc << " alias=" << dwarfsrc_canon << endl;
+ - + - +
- + - ]
2803 : : }
2804 : :
2805 : : ps_upsert_files
2806 [ + - ]: 1443 : .reset()
2807 [ + - ]: 1443 : .bind(1, dwarfsrc_canon)
2808 [ + - ]: 1443 : .step_ok_done();
2809 : :
2810 : : ps_upsert_s
2811 [ + - ]: 1443 : .reset()
2812 [ + - ]: 1443 : .bind(1, buildid)
2813 [ + - ]: 1443 : .bind(2, dwarfsrc_canon)
2814 [ + - ]: 1443 : .bind(3, srps)
2815 [ + - ]: 1443 : .bind(4, sfs.st_mtime)
2816 [ + - ]: 1443 : .step_ok_done();
2817 : :
2818 [ + - + - : 1443 : inc_metric("found_sourcerefs_total","source","files");
+ - + - ]
2819 : : }
2820 : : }
2821 : :
2822 : : ps_scan_done
2823 [ + - ]: 157 : .reset()
2824 [ + - ]: 157 : .bind(1, rps)
2825 [ + - ]: 157 : .bind(2, st.st_mtime)
2826 [ + - ]: 157 : .bind(3, st.st_size)
2827 [ + - ]: 157 : .step_ok_done();
2828 : :
2829 [ + - ]: 157 : if (verbose > 2)
2830 [ + - + - ]: 314 : obatched(clog) << "recorded buildid=" << buildid << " file=" << rps
2831 [ + - + - : 157 : << " mtime=" << st.st_mtime << " atype="
+ - + - +
- ]
2832 : : << (executable_p ? "E" : "")
2833 [ + - + + : 157 : << (debuginfo_p ? "D" : "") << endl;
+ - + + +
- + - ]
2834 : : }
2835 : :
2836 : :
2837 : :
2838 : :
2839 : :
2840 : : // Analyze given archive file of given age; record buildids / exec/debuginfo-ness of its
2841 : : // constituent files with given upsert statements.
2842 : : static void
2843 : 124 : archive_classify (const string& rps, string& archive_extension,
2844 : : sqlite_ps& ps_upsert_buildids, sqlite_ps& ps_upsert_files,
2845 : : sqlite_ps& ps_upsert_de, sqlite_ps& ps_upsert_sref, sqlite_ps& ps_upsert_sdef,
2846 : : time_t mtime,
2847 : : unsigned& fts_executable, unsigned& fts_debuginfo, unsigned& fts_sref, unsigned& fts_sdef,
2848 : : bool& fts_sref_complete_p)
2849 : : {
2850 [ + - ]: 248 : string archive_decoder = "/dev/null";
2851 [ + + ]: 310 : for (auto&& arch : scan_archives)
2852 [ + - + + ]: 186 : if (string_endswith(rps, arch.first))
2853 : : {
2854 [ + - ]: 124 : archive_extension = arch.first;
2855 [ + - ]: 124 : archive_decoder = arch.second;
2856 : : }
2857 : :
2858 : : FILE* fp;
2859 : : defer_dtor<FILE*,int>::dtor_fn dfn;
2860 [ + + ]: 124 : if (archive_decoder != "cat")
2861 : : {
2862 [ + - + - : 30 : string popen_cmd = archive_decoder + " " + shell_escape(rps);
+ - ]
2863 [ + - ]: 10 : fp = popen (popen_cmd.c_str(), "r"); // "e" O_CLOEXEC?
2864 : 10 : dfn = pclose;
2865 [ - + ]: 10 : if (fp == NULL)
2866 [ # # # # : 0 : throw libc_exception (errno, string("popen ") + popen_cmd);
# # ]
2867 : : }
2868 : : else
2869 : : {
2870 [ + - ]: 114 : fp = fopen (rps.c_str(), "r");
2871 : 114 : dfn = fclose;
2872 [ - + ]: 114 : if (fp == NULL)
2873 [ # # # # : 0 : throw libc_exception (errno, string("fopen ") + rps);
# # ]
2874 : : }
2875 : 248 : defer_dtor<FILE*,int> fp_closer (fp, dfn);
2876 : :
2877 : : struct archive *a;
2878 [ + - ]: 124 : a = archive_read_new();
2879 [ - + ]: 124 : if (a == NULL)
2880 [ # # # # ]: 0 : throw archive_exception("cannot create archive reader");
2881 : 248 : defer_dtor<struct archive*,int> archive_closer (a, archive_read_free);
2882 : :
2883 [ + - ]: 124 : int rc = archive_read_support_format_all(a);
2884 [ - + ]: 123 : if (rc != ARCHIVE_OK)
2885 [ # # # # ]: 0 : throw archive_exception(a, "cannot select all formats");
2886 [ + - ]: 123 : rc = archive_read_support_filter_all(a);
2887 [ - + ]: 124 : if (rc != ARCHIVE_OK)
2888 [ # # # # ]: 0 : throw archive_exception(a, "cannot select all filters");
2889 : :
2890 [ + - ]: 124 : rc = archive_read_open_FILE (a, fp);
2891 [ - + ]: 124 : if (rc != ARCHIVE_OK)
2892 [ # # # # ]: 0 : throw archive_exception(a, "cannot open archive from pipe");
2893 : :
2894 [ + + ]: 124 : if (verbose > 3)
2895 [ + - + - : 116 : obatched(clog) << "libarchive scanning " << rps << endl;
+ - + - ]
2896 : :
2897 : : while(1) // parse archive entries
2898 : : {
2899 [ - + ]: 855 : if (interrupted)
2900 : 0 : break;
2901 : :
2902 : : try
2903 : : {
2904 : : struct archive_entry *e;
2905 [ + - ]: 855 : rc = archive_read_next_header (a, &e);
2906 [ + + ]: 854 : if (rc != ARCHIVE_OK)
2907 : 124 : break;
2908 : :
2909 [ + - + + ]: 730 : if (! S_ISREG(archive_entry_mode (e))) // skip non-files completely
2910 : 514 : continue;
2911 : :
2912 [ + - ]: 434 : string fn = canonicalized_archive_entry_pathname (e);
2913 : :
2914 [ + + ]: 217 : if (verbose > 3)
2915 [ + - + - : 197 : obatched(clog) << "libarchive checking " << fn << endl;
+ - + - ]
2916 : :
2917 : : // extract this file to a temporary file
2918 : 217 : char* tmppath = NULL;
2919 : 217 : rc = asprintf (&tmppath, "%s/debuginfod.XXXXXX", tmpdir.c_str());
2920 [ - + ]: 217 : if (rc < 0)
2921 [ # # # # ]: 0 : throw libc_exception (ENOMEM, "cannot allocate tmppath");
2922 : 434 : defer_dtor<void*,void> tmmpath_freer (tmppath, free);
2923 [ + - ]: 217 : int fd = mkstemp (tmppath);
2924 [ - + ]: 217 : if (fd < 0)
2925 [ # # # # ]: 0 : throw libc_exception (errno, "cannot create temporary file");
2926 : 217 : unlink (tmppath); // unlink now so OS will release the file as soon as we close the fd
2927 : 434 : defer_dtor<int,int> minifd_closer (fd, close);
2928 : :
2929 [ + - ]: 217 : rc = archive_read_data_into_fd (a, fd);
2930 [ - + ]: 217 : if (rc != ARCHIVE_OK)
2931 [ # # # # ]: 0 : throw archive_exception(a, "cannot extract file");
2932 : :
2933 : : // finally ... time to run elf_classify on this bad boy and update the database
2934 : 217 : bool executable_p = false, debuginfo_p = false;
2935 : 434 : string buildid;
2936 : 434 : set<string> sourcefiles;
2937 [ + - ]: 216 : elf_classify (fd, executable_p, debuginfo_p, buildid, sourcefiles);
2938 : : // NB: might throw
2939 : :
2940 [ + + ]: 217 : if (buildid != "") // intern buildid
2941 : : {
2942 : : ps_upsert_buildids
2943 [ + - ]: 122 : .reset()
2944 [ + - ]: 122 : .bind(1, buildid)
2945 [ + - ]: 122 : .step_ok_done();
2946 : : }
2947 : :
2948 : : ps_upsert_files // register this rpm constituent file name in interning table
2949 [ + - ]: 217 : .reset()
2950 [ + - ]: 217 : .bind(1, fn)
2951 [ + - ]: 217 : .step_ok_done();
2952 : :
2953 [ + + ]: 217 : if (sourcefiles.size() > 0) // sref records needed
2954 : : {
2955 : : // NB: we intern each source file once. Once raw, as it
2956 : : // appears in the DWARF file list coming back from
2957 : : // elf_classify() - because it'll end up in the
2958 : : // _norm.artifactsrc column. We don't also put another
2959 : : // version with a '.' at the front, even though that's
2960 : : // how rpm/cpio packs names, because we hide that from
2961 : : // the database for storage efficiency.
2962 : :
2963 [ + + ]: 186 : for (auto&& s : sourcefiles)
2964 : : {
2965 [ - + ]: 135 : if (s == "")
2966 : : {
2967 : 0 : fts_sref_complete_p = false;
2968 : 0 : continue;
2969 : : }
2970 : :
2971 : : // PR25548: store canonicalized source path
2972 : 135 : const string& dwarfsrc = s;
2973 [ + - ]: 136 : string dwarfsrc_canon = canon_pathname (dwarfsrc);
2974 [ + + ]: 135 : if (dwarfsrc_canon != dwarfsrc)
2975 : : {
2976 [ + - ]: 10 : if (verbose > 3)
2977 [ + - + - : 10 : obatched(clog) << "canonicalized src=" << dwarfsrc << " alias=" << dwarfsrc_canon << endl;
+ - + - +
- + - ]
2978 : : }
2979 : :
2980 : : ps_upsert_files
2981 [ + - ]: 135 : .reset()
2982 [ + - ]: 135 : .bind(1, dwarfsrc_canon)
2983 [ + - ]: 135 : .step_ok_done();
2984 : :
2985 : : ps_upsert_sref
2986 [ + - ]: 135 : .reset()
2987 [ + - ]: 135 : .bind(1, buildid)
2988 [ + - ]: 135 : .bind(2, dwarfsrc_canon)
2989 [ + + ]: 135 : .step_ok_done();
2990 : :
2991 : 134 : fts_sref ++;
2992 : : }
2993 : : }
2994 : :
2995 [ + + ]: 216 : if (executable_p)
2996 : 52 : fts_executable ++;
2997 [ + + ]: 216 : if (debuginfo_p)
2998 : 69 : fts_debuginfo ++;
2999 : :
3000 [ + + + + ]: 216 : if (executable_p || debuginfo_p)
3001 : : {
3002 : : ps_upsert_de
3003 [ + - ]: 121 : .reset()
3004 [ + - ]: 121 : .bind(1, buildid)
3005 [ + + + - ]: 121 : .bind(2, debuginfo_p ? 1 : 0)
3006 [ + + + - ]: 121 : .bind(3, executable_p ? 1 : 0)
3007 [ + - ]: 121 : .bind(4, rps)
3008 [ + - ]: 121 : .bind(5, mtime)
3009 [ + - ]: 121 : .bind(6, fn)
3010 [ + + ]: 121 : .step_ok_done();
3011 : : }
3012 : : else // potential source - sdef record
3013 : : {
3014 : 95 : fts_sdef ++;
3015 : : ps_upsert_sdef
3016 [ + - ]: 95 : .reset()
3017 [ + - ]: 95 : .bind(1, rps)
3018 [ + - ]: 95 : .bind(2, mtime)
3019 [ + - ]: 95 : .bind(3, fn)
3020 [ + + ]: 95 : .step_ok_done();
3021 : : }
3022 : :
3023 [ + - + + : 212 : if ((verbose > 2) && (executable_p || debuginfo_p))
+ + ]
3024 [ + - + - ]: 240 : obatched(clog) << "recorded buildid=" << buildid << " rpm=" << rps << " file=" << fn
3025 [ + - + - : 120 : << " mtime=" << mtime << " atype="
+ - + - +
- + - +
- ]
3026 : : << (executable_p ? "E" : "")
3027 : : << (debuginfo_p ? "D" : "")
3028 [ + - + + : 120 : << " sourcefiles=" << sourcefiles.size() << endl;
+ - + + +
- + - + -
+ - ]
3029 : :
3030 : : }
3031 [ - + ]: 10 : catch (const reportable_exception& e)
3032 : : {
3033 [ + - ]: 5 : e.report(clog);
3034 : : }
3035 : 731 : }
3036 : 124 : }
3037 : :
3038 : :
3039 : :
3040 : : // scan for archive files such as .rpm
3041 : : static void
3042 : 254 : scan_archive_file (const string& rps, const stat_t& st,
3043 : : sqlite_ps& ps_upsert_buildids,
3044 : : sqlite_ps& ps_upsert_files,
3045 : : sqlite_ps& ps_upsert_de,
3046 : : sqlite_ps& ps_upsert_sref,
3047 : : sqlite_ps& ps_upsert_sdef,
3048 : : sqlite_ps& ps_query,
3049 : : sqlite_ps& ps_scan_done,
3050 : : unsigned& fts_cached,
3051 : : unsigned& fts_executable,
3052 : : unsigned& fts_debuginfo,
3053 : : unsigned& fts_sref,
3054 : : unsigned& fts_sdef)
3055 : : {
3056 : : /* See if we know of it already. */
3057 : : int rc = ps_query
3058 [ + - ]: 254 : .reset()
3059 [ + - ]: 254 : .bind(1, rps)
3060 [ + - ]: 254 : .bind(2, st.st_mtime)
3061 [ + - ]: 254 : .step();
3062 [ + - ]: 254 : ps_query.reset();
3063 [ + + ]: 254 : if (rc == SQLITE_ROW) // i.e., a result, as opposed to DONE (no results)
3064 : : // no need to recheck a file/version we already know
3065 : : // specifically, no need to parse this archive again, since we already have
3066 : : // it as a D or E or S record,
3067 : : // (so is stored with buildid=NULL)
3068 : : {
3069 : 130 : fts_cached ++;
3070 : 130 : return;
3071 : : }
3072 : :
3073 : : // intern the archive file name
3074 : : ps_upsert_files
3075 [ + - ]: 124 : .reset()
3076 [ + - ]: 124 : .bind(1, rps)
3077 [ + - ]: 124 : .step_ok_done();
3078 : :
3079 : : // extract the archive contents
3080 : 124 : unsigned my_fts_executable = 0, my_fts_debuginfo = 0, my_fts_sref = 0, my_fts_sdef = 0;
3081 : 124 : bool my_fts_sref_complete_p = true;
3082 : : try
3083 : : {
3084 : 248 : string archive_extension;
3085 : 124 : archive_classify (rps, archive_extension,
3086 : : ps_upsert_buildids, ps_upsert_files,
3087 : : ps_upsert_de, ps_upsert_sref, ps_upsert_sdef, // dalt
3088 [ + - ]: 124 : st.st_mtime,
3089 : : my_fts_executable, my_fts_debuginfo, my_fts_sref, my_fts_sdef,
3090 : : my_fts_sref_complete_p);
3091 [ + - ]: 248 : add_metric ("scanned_bytes_total","source",archive_extension + " archive",
3092 [ + - + - : 372 : st.st_size);
+ - ]
3093 [ + - + - : 124 : inc_metric ("scanned_files_total","source",archive_extension + " archive");
+ - + - ]
3094 [ + - + - ]: 124 : add_metric("found_debuginfo_total","source",archive_extension + " archive",
3095 [ + - + - ]: 248 : my_fts_debuginfo);
3096 [ + - + - ]: 124 : add_metric("found_executable_total","source",archive_extension + " archive",
3097 [ + - + - ]: 248 : my_fts_executable);
3098 [ + - + - ]: 124 : add_metric("found_sourcerefs_total","source",archive_extension + " archive",
3099 [ + - + - ]: 248 : my_fts_sref);
3100 : : }
3101 [ - - ]: 0 : catch (const reportable_exception& e)
3102 : : {
3103 [ - - ]: 0 : e.report(clog);
3104 : : }
3105 : :
3106 [ + - ]: 124 : if (verbose > 2)
3107 [ + - + - ]: 248 : obatched(clog) << "scanned archive=" << rps
3108 [ + - + - : 124 : << " mtime=" << st.st_mtime
+ - ]
3109 [ + - ]: 124 : << " executables=" << my_fts_executable
3110 [ + - + - ]: 124 : << " debuginfos=" << my_fts_debuginfo
3111 [ + - + - ]: 124 : << " srefs=" << my_fts_sref
3112 [ + - + - ]: 124 : << " sdefs=" << my_fts_sdef
3113 [ + - + - ]: 124 : << endl;
3114 : :
3115 : 124 : fts_executable += my_fts_executable;
3116 : 124 : fts_debuginfo += my_fts_debuginfo;
3117 : 124 : fts_sref += my_fts_sref;
3118 : 124 : fts_sdef += my_fts_sdef;
3119 : :
3120 [ + - ]: 124 : if (my_fts_sref_complete_p) // leave incomplete?
3121 : : ps_scan_done
3122 [ + - ]: 124 : .reset()
3123 [ + - ]: 124 : .bind(1, rps)
3124 [ + - ]: 124 : .bind(2, st.st_mtime)
3125 [ + - ]: 124 : .bind(3, st.st_size)
3126 [ + - ]: 124 : .step_ok_done();
3127 : : }
3128 : :
3129 : :
3130 : :
3131 : : ////////////////////////////////////////////////////////////////////////
3132 : :
3133 : :
3134 : :
3135 : : // The thread that consumes file names off of the scanq. We hold
3136 : : // the persistent sqlite_ps's at this level and delegate file/archive
3137 : : // scanning to other functions.
3138 : : static void*
3139 : 96 : thread_main_scanner (void* arg)
3140 : : {
3141 : : (void) arg;
3142 : :
3143 : : // all the prepared statements fit to use, the _f_ set:
3144 [ + - + - : 384 : sqlite_ps ps_f_upsert_buildids (db, "file-buildids-intern", "insert or ignore into " BUILDIDS "_buildids VALUES (NULL, ?);");
+ - ]
3145 [ + - + - : 384 : sqlite_ps ps_f_upsert_files (db, "file-files-intern", "insert or ignore into " BUILDIDS "_files VALUES (NULL, ?);");
+ - ]
3146 : : sqlite_ps ps_f_upsert_de (db, "file-de-upsert",
3147 : : "insert or ignore into " BUILDIDS "_f_de "
3148 : : "(buildid, debuginfo_p, executable_p, file, mtime) "
3149 : : "values ((select id from " BUILDIDS "_buildids where hex = ?),"
3150 : : " ?,?,"
3151 [ + - + - : 384 : " (select id from " BUILDIDS "_files where name = ?), ?);");
+ - ]
3152 : : sqlite_ps ps_f_upsert_s (db, "file-s-upsert",
3153 : : "insert or ignore into " BUILDIDS "_f_s "
3154 : : "(buildid, artifactsrc, file, mtime) "
3155 : : "values ((select id from " BUILDIDS "_buildids where hex = ?),"
3156 : : " (select id from " BUILDIDS "_files where name = ?),"
3157 : : " (select id from " BUILDIDS "_files where name = ?),"
3158 [ + - + - : 384 : " ?);");
+ - ]
3159 : : sqlite_ps ps_f_query (db, "file-negativehit-find",
3160 : : "select 1 from " BUILDIDS "_file_mtime_scanned where sourcetype = 'F' "
3161 [ + - + - : 384 : "and file = (select id from " BUILDIDS "_files where name = ?) and mtime = ?;");
+ - ]
3162 : : sqlite_ps ps_f_scan_done (db, "file-scanned",
3163 : : "insert or ignore into " BUILDIDS "_file_mtime_scanned (sourcetype, file, mtime, size)"
3164 [ + - + - : 384 : "values ('F', (select id from " BUILDIDS "_files where name = ?), ?, ?);");
+ - ]
3165 : :
3166 : : // and now for the _r_ set
3167 [ + - + - : 384 : sqlite_ps ps_r_upsert_buildids (db, "rpm-buildid-intern", "insert or ignore into " BUILDIDS "_buildids VALUES (NULL, ?);");
+ - ]
3168 [ + - + - : 384 : sqlite_ps ps_r_upsert_files (db, "rpm-file-intern", "insert or ignore into " BUILDIDS "_files VALUES (NULL, ?);");
+ - ]
3169 : : sqlite_ps ps_r_upsert_de (db, "rpm-de-insert",
3170 : : "insert or ignore into " BUILDIDS "_r_de (buildid, debuginfo_p, executable_p, file, mtime, content) values ("
3171 : : "(select id from " BUILDIDS "_buildids where hex = ?), ?, ?, "
3172 : : "(select id from " BUILDIDS "_files where name = ?), ?, "
3173 [ + - + - : 384 : "(select id from " BUILDIDS "_files where name = ?));");
+ - ]
3174 : : sqlite_ps ps_r_upsert_sref (db, "rpm-sref-insert",
3175 : : "insert or ignore into " BUILDIDS "_r_sref (buildid, artifactsrc) values ("
3176 : : "(select id from " BUILDIDS "_buildids where hex = ?), "
3177 [ + - + - : 384 : "(select id from " BUILDIDS "_files where name = ?));");
+ - ]
3178 : : sqlite_ps ps_r_upsert_sdef (db, "rpm-sdef-insert",
3179 : : "insert or ignore into " BUILDIDS "_r_sdef (file, mtime, content) values ("
3180 : : "(select id from " BUILDIDS "_files where name = ?), ?,"
3181 [ + - + - : 384 : "(select id from " BUILDIDS "_files where name = ?));");
+ - ]
3182 : : sqlite_ps ps_r_query (db, "rpm-negativehit-query",
3183 : : "select 1 from " BUILDIDS "_file_mtime_scanned where "
3184 [ + - + - : 384 : "sourcetype = 'R' and file = (select id from " BUILDIDS "_files where name = ?) and mtime = ?;");
+ - ]
3185 : : sqlite_ps ps_r_scan_done (db, "rpm-scanned",
3186 : : "insert or ignore into " BUILDIDS "_file_mtime_scanned (sourcetype, file, mtime, size)"
3187 [ + - + - : 288 : "values ('R', (select id from " BUILDIDS "_files where name = ?), ?, ?);");
+ - ]
3188 : :
3189 : :
3190 : 96 : unsigned fts_cached = 0, fts_executable = 0, fts_debuginfo = 0, fts_sourcefiles = 0;
3191 : 96 : unsigned fts_sref = 0, fts_sdef = 0;
3192 : :
3193 [ + - + - : 96 : add_metric("thread_count", "role", "scan", 1);
+ - + - ]
3194 [ + - + - : 96 : add_metric("thread_busy", "role", "scan", 1);
+ - + - ]
3195 [ + + ]: 538 : while (! interrupted)
3196 : : {
3197 : 442 : scan_payload p;
3198 : :
3199 [ + - + - : 442 : add_metric("thread_busy", "role", "scan", -1);
+ - + - ]
3200 [ + - ]: 442 : bool gotone = scanq.wait_front(p);
3201 [ + - + - : 441 : add_metric("thread_busy", "role", "scan", 1);
+ - + - ]
3202 : :
3203 [ + + ]: 442 : if (! gotone) continue; // go back to waiting
3204 : :
3205 : : try
3206 : : {
3207 : 346 : bool scan_archive = false;
3208 [ + + ]: 831 : for (auto&& arch : scan_archives)
3209 [ + - + + ]: 485 : if (string_endswith(p.first, arch.first))
3210 : 254 : scan_archive = true;
3211 : :
3212 [ + + ]: 346 : if (scan_archive)
3213 [ + - ]: 254 : scan_archive_file (p.first, p.second,
3214 : : ps_r_upsert_buildids,
3215 : : ps_r_upsert_files,
3216 : : ps_r_upsert_de,
3217 : : ps_r_upsert_sref,
3218 : : ps_r_upsert_sdef,
3219 : : ps_r_query,
3220 : : ps_r_scan_done,
3221 : : fts_cached,
3222 : : fts_executable,
3223 : : fts_debuginfo,
3224 : : fts_sref,
3225 : : fts_sdef);
3226 : :
3227 [ + + ]: 346 : if (scan_files) // NB: maybe "else if" ?
3228 [ + - ]: 319 : scan_source_file (p.first, p.second,
3229 : : ps_f_upsert_buildids,
3230 : : ps_f_upsert_files,
3231 : : ps_f_upsert_de,
3232 : : ps_f_upsert_s,
3233 : : ps_f_query,
3234 : : ps_f_scan_done,
3235 : : fts_cached, fts_executable, fts_debuginfo, fts_sourcefiles);
3236 : : }
3237 [ - - ]: 0 : catch (const reportable_exception& e)
3238 : : {
3239 [ - - ]: 0 : e.report(cerr);
3240 : : }
3241 : :
3242 [ + - ]: 346 : scanq.done_front(); // let idlers run
3243 : :
3244 [ + + + + : 346 : if (fts_cached || fts_executable || fts_debuginfo || fts_sourcefiles || fts_sref || fts_sdef)
+ + + - +
- ]
3245 : : {} // NB: not just if a successful scan - we might have encountered -ENOSPC & failed
3246 [ + - + - ]: 346 : (void) statfs_free_enough_p(db_path, "database"); // report sqlite filesystem size
3247 [ + - + - ]: 346 : (void) statfs_free_enough_p(tmpdir, "tmpdir"); // this too, in case of fdcache/tmpfile usage
3248 : :
3249 : : // finished a scanning step -- not a "loop", because we just
3250 : : // consume the traversal loop's work, whenever
3251 [ + - + - : 346 : inc_metric("thread_work_total","role","scan");
+ - + - ]
3252 : : }
3253 : :
3254 : :
3255 [ + - + - : 96 : add_metric("thread_busy", "role", "scan", -1);
+ - + - ]
3256 : 192 : return 0;
3257 : : }
3258 : :
3259 : :
3260 : :
3261 : : // The thread that traverses all the source_paths and enqueues all the
3262 : : // matching files into the file/archive scan queue.
3263 : : static void
3264 : 47 : scan_source_paths()
3265 : : {
3266 : : // NB: fedora 31 glibc/fts(3) crashes inside fts_read() on empty
3267 : : // path list.
3268 [ + + ]: 47 : if (source_paths.empty())
3269 : 1 : return;
3270 : :
3271 : : // Turn the source_paths into an fts(3)-compatible char**. Since
3272 : : // source_paths[] does not change after argv processing, the
3273 : : // c_str()'s are safe to keep around awile.
3274 : 92 : vector<const char *> sps;
3275 [ + + ]: 128 : for (auto&& sp: source_paths)
3276 [ + - ]: 82 : sps.push_back(sp.c_str());
3277 [ + - ]: 46 : sps.push_back(NULL);
3278 : :
3279 [ + + + - ]: 46 : FTS *fts = fts_open ((char * const *)sps.data(),
3280 : : (traverse_logical ? FTS_LOGICAL : FTS_PHYSICAL|FTS_XDEV)
3281 : : | FTS_NOCHDIR /* multithreaded */,
3282 : : NULL);
3283 [ - + ]: 46 : if (fts == NULL)
3284 [ # # # # ]: 0 : throw libc_exception(errno, "cannot fts_open");
3285 : 46 : defer_dtor<FTS*,int> fts_cleanup (fts, fts_close);
3286 : :
3287 : : struct timespec ts_start, ts_end;
3288 : 46 : clock_gettime (CLOCK_MONOTONIC, &ts_start);
3289 : 46 : unsigned fts_scanned = 0, fts_regex = 0;
3290 : :
3291 : : FTSENT *f;
3292 [ + - + + ]: 699 : while ((f = fts_read (fts)) != NULL)
3293 : : {
3294 [ - + ]: 653 : if (interrupted) break;
3295 : :
3296 [ - + ]: 653 : if (sigusr2 != forced_groom_count) // stop early if groom triggered
3297 : : {
3298 [ # # ]: 0 : scanq.clear(); // clear previously issued work for scanner threads
3299 : 0 : break;
3300 : : }
3301 : :
3302 : 653 : fts_scanned ++;
3303 : :
3304 [ + - ]: 653 : if (verbose > 2)
3305 [ + - + - : 653 : obatched(clog) << "fts traversing " << f->fts_path << endl;
+ - + - ]
3306 : :
3307 [ + + + + : 653 : switch (f->fts_info)
+ ]
3308 : : {
3309 : 346 : case FTS_F:
3310 : : {
3311 : : /* Found a file. Convert it to an absolute path, so
3312 : : the buildid database does not have relative path
3313 : : names that are unresolvable from a subsequent run
3314 : : in a different cwd. */
3315 : 346 : char *rp = realpath(f->fts_path, NULL);
3316 [ - + ]: 346 : if (rp == NULL)
3317 : 0 : continue; // ignore dangling symlink or such
3318 [ + - ]: 692 : string rps = string(rp);
3319 : 346 : free (rp);
3320 : :
3321 [ + - ]: 346 : bool ri = !regexec (&file_include_regex, rps.c_str(), 0, 0, 0);
3322 [ + - ]: 346 : bool rx = !regexec (&file_exclude_regex, rps.c_str(), 0, 0, 0);
3323 [ + - - + ]: 346 : if (!ri || rx)
3324 : : {
3325 [ # # ]: 0 : if (verbose > 3)
3326 [ # # # # ]: 0 : obatched(clog) << "fts skipped by regex "
3327 [ # # # # : 0 : << (!ri ? "I" : "") << (rx ? "X" : "") << endl;
# # # # #
# ]
3328 : 0 : fts_regex ++;
3329 [ # # ]: 0 : if (!ri)
3330 [ # # # # : 0 : inc_metric("traversed_total","type","file-skipped-I");
# # # # ]
3331 [ # # ]: 0 : if (rx)
3332 [ # # # # : 0 : inc_metric("traversed_total","type","file-skipped-X");
# # # # ]
3333 : : }
3334 : : else
3335 : : {
3336 [ + - + - ]: 346 : scanq.push_back (make_pair(rps, *f->fts_statp));
3337 [ + - + - : 346 : inc_metric("traversed_total","type","file");
+ - + - ]
3338 : : }
3339 : : }
3340 : 346 : break;
3341 : :
3342 : 1 : case FTS_ERR:
3343 : : case FTS_NS:
3344 : : // report on some types of errors because they may reflect fixable misconfiguration
3345 : : {
3346 [ + - + - : 4 : auto x = libc_exception(f->fts_errno, string("fts traversal ") + string(f->fts_path));
+ - + - ]
3347 [ + - ]: 1 : x.report(cerr);
3348 : : }
3349 [ + - + - : 1 : inc_metric("traversed_total","type","error");
+ - + - ]
3350 : 1 : break;
3351 : :
3352 : 6 : case FTS_SL: // ignore, but count because debuginfod -L would traverse these
3353 [ + - + - : 6 : inc_metric("traversed_total","type","symlink");
+ - + - ]
3354 : 6 : break;
3355 : :
3356 : 150 : case FTS_D: // ignore
3357 [ + - + - : 150 : inc_metric("traversed_total","type","directory");
+ - + - ]
3358 : 150 : break;
3359 : :
3360 : 150 : default: // ignore
3361 [ + - + - : 150 : inc_metric("traversed_total","type","other");
+ - + - ]
3362 : 150 : break;
3363 : : }
3364 : : }
3365 : 46 : clock_gettime (CLOCK_MONOTONIC, &ts_end);
3366 : 46 : double deltas = (ts_end.tv_sec - ts_start.tv_sec) + (ts_end.tv_nsec - ts_start.tv_nsec)/1.e9;
3367 : :
3368 [ + - + - : 92 : obatched(clog) << "fts traversed source paths in " << deltas << "s, scanned=" << fts_scanned
+ - + - +
- ]
3369 [ + - + - : 46 : << ", regex-skipped=" << fts_regex << endl;
+ - ]
3370 : : }
3371 : :
3372 : :
3373 : : static void*
3374 : 24 : thread_main_fts_source_paths (void* arg)
3375 : : {
3376 : : (void) arg; // ignore; we operate on global data
3377 : :
3378 [ + - + - : 24 : set_metric("thread_tid", "role","traverse", tid());
+ - + - ]
3379 [ + - + - : 24 : add_metric("thread_count", "role", "traverse", 1);
+ - + - ]
3380 : :
3381 : 24 : time_t last_rescan = 0;
3382 : :
3383 [ + - ]: 119 : while (! interrupted)
3384 : : {
3385 : 119 : sleep (1);
3386 : 119 : scanq.wait_idle(); // don't start a new traversal while scanners haven't finished the job
3387 : 119 : scanq.done_idle(); // release the hounds
3388 [ + + ]: 119 : if (interrupted) break;
3389 : :
3390 : 95 : time_t now = time(NULL);
3391 : 95 : bool rescan_now = false;
3392 [ + + ]: 95 : if (last_rescan == 0) // at least one initial rescan is documented even for -t0
3393 : 22 : rescan_now = true;
3394 [ + + + + ]: 95 : if (rescan_s > 0 && (long)now > (long)(last_rescan + rescan_s))
3395 : 4 : rescan_now = true;
3396 [ + + ]: 95 : if (sigusr1 != forced_rescan_count)
3397 : : {
3398 : 27 : forced_rescan_count = sigusr1;
3399 : 27 : rescan_now = true;
3400 : : }
3401 [ + + ]: 95 : if (rescan_now)
3402 : : {
3403 [ + - + - : 47 : set_metric("thread_busy", "role","traverse", 1);
+ - + - ]
3404 : : try
3405 : : {
3406 [ + - ]: 47 : scan_source_paths();
3407 : : }
3408 [ - - ]: 0 : catch (const reportable_exception& e)
3409 : : {
3410 [ - - ]: 0 : e.report(cerr);
3411 : : }
3412 : 47 : last_rescan = time(NULL); // NB: now was before scanning
3413 : : // finished a traversal loop
3414 [ + - + - : 47 : inc_metric("thread_work_total", "role","traverse");
+ - + - ]
3415 [ + - + - : 47 : set_metric("thread_busy", "role","traverse", 0);
+ - + - ]
3416 : : }
3417 : : }
3418 : :
3419 : 24 : return 0;
3420 : : }
3421 : :
3422 : :
3423 : :
3424 : : ////////////////////////////////////////////////////////////////////////
3425 : :
3426 : : static void
3427 : 25 : database_stats_report()
3428 : : {
3429 : : sqlite_ps ps_query (db, "database-overview",
3430 [ + - + - : 100 : "select label,quantity from " BUILDIDS "_stats");
+ - ]
3431 : :
3432 [ + - + - : 25 : obatched(clog) << "database record counts:" << endl;
+ - ]
3433 : : while (1)
3434 : : {
3435 [ - + ]: 275 : if (interrupted) break;
3436 [ - + ]: 275 : if (sigusr1 != forced_rescan_count) // stop early if scan triggered
3437 : 0 : break;
3438 : :
3439 [ + - ]: 275 : int rc = ps_query.step();
3440 [ + + ]: 275 : if (rc == SQLITE_DONE) break;
3441 [ - + ]: 250 : if (rc != SQLITE_ROW)
3442 [ # # # # ]: 0 : throw sqlite_exception(rc, "step");
3443 : :
3444 [ + - ]: 500 : obatched(clog)
3445 [ + - + - ]: 500 : << right << setw(20) << ((const char*) sqlite3_column_text(ps_query, 0) ?: (const char*) "NULL")
3446 : : << " "
3447 [ + - + - : 500 : << (sqlite3_column_text(ps_query, 1) ?: (const unsigned char*) "NULL")
+ - + - +
- + - +
- ]
3448 [ + - ]: 250 : << endl;
3449 : :
3450 [ + - + - ]: 250 : set_metric("groom", "statistic",
3451 [ + - ]: 250 : ((const char*) sqlite3_column_text(ps_query, 0) ?: (const char*) "NULL"),
3452 [ + - + - : 750 : (sqlite3_column_double(ps_query, 1)));
+ - + - ]
3453 : 250 : }
3454 : 25 : }
3455 : :
3456 : :
3457 : : // Do a round of database grooming that might take many minutes to run.
3458 : 26 : void groom()
3459 : : {
3460 [ + - + - : 26 : obatched(clog) << "grooming database" << endl;
+ - ]
3461 : :
3462 : : struct timespec ts_start, ts_end;
3463 : 26 : clock_gettime (CLOCK_MONOTONIC, &ts_start);
3464 : :
3465 : : // scan for files that have disappeared
3466 : : sqlite_ps files (db, "check old files",
3467 : : "select distinct s.mtime, s.file, f.name from "
3468 : : BUILDIDS "_file_mtime_scanned s, " BUILDIDS "_files f "
3469 [ + - + - : 79 : "where f.id = s.file");
+ - ]
3470 : : // NB: Because _ftime_mtime_scanned can contain both F and
3471 : : // R records for the same file, this query would return duplicates if the
3472 : : // DISTINCT qualifier were not there.
3473 [ + - ]: 26 : files.reset();
3474 : :
3475 : : // DECISION TIME - we enumerate stale fileids/mtimes
3476 [ + - ]: 27 : deque<pair<int64_t,int64_t> > stale_fileid_mtime;
3477 : :
3478 : 26 : time_t time_start = time(NULL);
3479 : : while(1)
3480 : : {
3481 : : // PR28514: limit grooming iteration to O(rescan time), to avoid
3482 : : // slow filesystem tests over many files locking out rescans for
3483 : : // too long.
3484 [ + + - + : 76 : if (rescan_s > 0 && (long)time(NULL) > (long)(time_start + rescan_s))
- + ]
3485 : : {
3486 [ # # # # : 0 : inc_metric("groomed_total", "decision", "aborted");
# # # # ]
3487 : 0 : break;
3488 : : }
3489 : :
3490 [ - + ]: 76 : if (interrupted) break;
3491 : :
3492 [ + - ]: 76 : int rc = files.step();
3493 [ + + ]: 76 : if (rc != SQLITE_ROW)
3494 : 26 : break;
3495 : :
3496 [ + - ]: 50 : int64_t mtime = sqlite3_column_int64 (files, 0);
3497 [ + - ]: 50 : int64_t fileid = sqlite3_column_int64 (files, 1);
3498 [ + - + - ]: 50 : const char* filename = ((const char*) sqlite3_column_text (files, 2) ?: "");
3499 : : struct stat s;
3500 [ + - ]: 50 : bool reg_include = !regexec (&file_include_regex, filename, 0, 0, 0);
3501 [ + - ]: 50 : bool reg_exclude = !regexec (&file_exclude_regex, filename, 0, 0, 0);
3502 : :
3503 : 50 : rc = stat(filename, &s);
3504 [ - + - - : 50 : if ( (regex_groom && reg_exclude && !reg_include) || rc < 0 || (mtime != (int64_t) s.st_mtime) )
- - + + -
+ ]
3505 : : {
3506 [ + - ]: 4 : if (verbose > 2)
3507 [ + - + - : 4 : obatched(clog) << "groom: stale file=" << filename << " mtime=" << mtime << endl;
+ - + - +
- + - ]
3508 [ + - + - ]: 4 : stale_fileid_mtime.push_back(make_pair(fileid,mtime));
3509 [ + - + - : 4 : inc_metric("groomed_total", "decision", "stale");
+ - + - ]
3510 [ + - + - : 4 : set_metric("thread_work_pending","role","groom", stale_fileid_mtime.size());
+ - + - ]
3511 : : }
3512 : : else
3513 [ + - + - : 46 : inc_metric("groomed_total", "decision", "fresh");
+ - + - ]
3514 : :
3515 [ - + ]: 50 : if (sigusr1 != forced_rescan_count) // stop early if scan triggered
3516 : 0 : break;
3517 : 50 : }
3518 [ + - ]: 26 : files.reset();
3519 : :
3520 : : // ACTION TIME
3521 : :
3522 : : // Now that we know which file/mtime tuples are stale, actually do
3523 : : // the deletion from the database. Doing this during the SELECT
3524 : : // iteration above results in undefined behaviour in sqlite, as per
3525 : : // https://www.sqlite.org/isolation.html
3526 : :
3527 : : // We could shuffle stale_fileid_mtime[] here. It'd let aborted
3528 : : // sequences of nuke operations resume at random locations, instead
3529 : : // of just starting over. But it doesn't matter much either way,
3530 : : // as long as we make progress.
3531 : :
3532 [ + - + - : 79 : sqlite_ps files_del_f_de (db, "nuke f_de", "delete from " BUILDIDS "_f_de where file = ? and mtime = ?");
+ - ]
3533 [ + - + - : 79 : sqlite_ps files_del_r_de (db, "nuke r_de", "delete from " BUILDIDS "_r_de where file = ? and mtime = ?");
+ - ]
3534 : : sqlite_ps files_del_scan (db, "nuke f_m_s", "delete from " BUILDIDS "_file_mtime_scanned "
3535 [ + - + - : 79 : "where file = ? and mtime = ?");
+ - ]
3536 : :
3537 [ + + ]: 30 : while (! stale_fileid_mtime.empty())
3538 : : {
3539 : 4 : auto stale = stale_fileid_mtime.front();
3540 : 4 : stale_fileid_mtime.pop_front();
3541 [ + - + - : 4 : set_metric("thread_work_pending","role","groom", stale_fileid_mtime.size());
+ - + - ]
3542 : :
3543 : : // PR28514: limit grooming iteration to O(rescan time), to avoid
3544 : : // slow nuke_* queries over many files locking out rescans for too
3545 : : // long. We iterate over the files in random() sequence to avoid
3546 : : // partial checks going over the same set.
3547 [ - + - - : 4 : if (rescan_s > 0 && (long)time(NULL) > (long)(time_start + rescan_s))
- + ]
3548 : : {
3549 [ # # # # : 0 : inc_metric("groomed_total", "action", "aborted");
# # # # ]
3550 : 0 : break;
3551 : : }
3552 : :
3553 [ - + ]: 4 : if (interrupted) break;
3554 : :
3555 : 4 : int64_t fileid = stale.first;
3556 : 4 : int64_t mtime = stale.second;
3557 [ + - + - : 4 : files_del_f_de.reset().bind(1,fileid).bind(2,mtime).step_ok_done();
+ - + - ]
3558 [ + - + - : 4 : files_del_r_de.reset().bind(1,fileid).bind(2,mtime).step_ok_done();
+ - + - ]
3559 [ + - + - : 4 : files_del_scan.reset().bind(1,fileid).bind(2,mtime).step_ok_done();
+ - + - ]
3560 [ + - + - : 4 : inc_metric("groomed_total", "action", "cleaned");
+ - + - ]
3561 : :
3562 [ - + ]: 4 : if (sigusr1 != forced_rescan_count) // stop early if scan triggered
3563 : 0 : break;
3564 : : }
3565 : 26 : stale_fileid_mtime.clear(); // no need for this any longer
3566 [ + - + - : 26 : set_metric("thread_work_pending","role","groom", stale_fileid_mtime.size());
+ - + - ]
3567 : :
3568 : : // delete buildids with no references in _r_de or _f_de tables;
3569 : : // cascades to _r_sref & _f_s records
3570 : : sqlite_ps buildids_del (db, "nuke orphan buildids",
3571 : : "delete from " BUILDIDS "_buildids "
3572 : : "where not exists (select 1 from " BUILDIDS "_f_de d where " BUILDIDS "_buildids.id = d.buildid) "
3573 [ + - + - : 79 : "and not exists (select 1 from " BUILDIDS "_r_de d where " BUILDIDS "_buildids.id = d.buildid)");
+ - ]
3574 [ + - + + ]: 26 : buildids_del.reset().step_ok_done();
3575 : :
3576 [ - + ]: 25 : if (interrupted) return;
3577 : :
3578 : : // NB: "vacuum" is too heavy for even daily runs: it rewrites the entire db, so is done as maxigroom -G
3579 [ + - + - : 100 : sqlite_ps g1 (db, "incremental vacuum", "pragma incremental_vacuum");
+ - ]
3580 [ + - + - ]: 25 : g1.reset().step_ok_done();
3581 [ + - + - : 100 : sqlite_ps g2 (db, "optimize", "pragma optimize");
+ - ]
3582 [ + - + - ]: 25 : g2.reset().step_ok_done();
3583 [ + - + - : 75 : sqlite_ps g3 (db, "wal checkpoint", "pragma wal_checkpoint=truncate");
+ - ]
3584 [ + - + - ]: 25 : g3.reset().step_ok_done();
3585 : :
3586 [ + - ]: 25 : database_stats_report();
3587 : :
3588 [ + - + - ]: 25 : (void) statfs_free_enough_p(db_path, "database"); // report sqlite filesystem size
3589 : :
3590 [ + - ]: 25 : sqlite3_db_release_memory(db); // shrink the process if possible
3591 [ + - ]: 25 : sqlite3_db_release_memory(dbq); // ... for both connections
3592 [ + - ]: 25 : debuginfod_pool_groom(); // and release any debuginfod_client objects we've been holding onto
3593 : :
3594 [ + - ]: 25 : fdcache.limit(0,0,0,0); // release the fdcache contents
3595 [ + - ]: 25 : fdcache.limit(fdcache_fds, fdcache_mbs, fdcache_prefetch_fds, fdcache_prefetch_mbs); // restore status quo parameters
3596 : :
3597 : 25 : clock_gettime (CLOCK_MONOTONIC, &ts_end);
3598 : 25 : double deltas = (ts_end.tv_sec - ts_start.tv_sec) + (ts_end.tv_nsec - ts_start.tv_nsec)/1.e9;
3599 : :
3600 [ + - + - : 25 : obatched(clog) << "groomed database in " << deltas << "s" << endl;
+ - + - +
- ]
3601 : : }
3602 : :
3603 : :
3604 : : static void*
3605 : 27 : thread_main_groom (void* /*arg*/)
3606 : : {
3607 [ + - + - : 27 : set_metric("thread_tid", "role", "groom", tid());
+ - + - ]
3608 [ + - + - : 27 : add_metric("thread_count", "role", "groom", 1);
+ - + - ]
3609 : :
3610 : 27 : time_t last_groom = 0;
3611 : :
3612 : : while (1)
3613 : : {
3614 : 121 : sleep (1);
3615 : 121 : scanq.wait_idle(); // PR25394: block scanners during grooming!
3616 [ + + ]: 121 : if (interrupted) break;
3617 : :
3618 : 94 : time_t now = time(NULL);
3619 : 94 : bool groom_now = false;
3620 [ + + ]: 94 : if (last_groom == 0) // at least one initial groom is documented even for -g0
3621 : 23 : groom_now = true;
3622 [ + + + + ]: 94 : if (groom_s > 0 && (long)now > (long)(last_groom + groom_s))
3623 : 4 : groom_now = true;
3624 [ + + ]: 94 : if (sigusr2 != forced_groom_count)
3625 : : {
3626 : 3 : forced_groom_count = sigusr2;
3627 : 3 : groom_now = true;
3628 : : }
3629 [ + + ]: 94 : if (groom_now)
3630 : : {
3631 [ + - + - : 26 : set_metric("thread_busy", "role", "groom", 1);
+ - + - ]
3632 : : try
3633 : : {
3634 [ + + ]: 26 : groom ();
3635 : : }
3636 [ - + ]: 1 : catch (const sqlite_exception& e)
3637 : : {
3638 [ + - + - : 1 : obatched(cerr) << e.message << endl;
+ - ]
3639 : : }
3640 : 26 : last_groom = time(NULL); // NB: now was before grooming
3641 : : // finished a grooming loop
3642 [ + - + - : 26 : inc_metric("thread_work_total", "role", "groom");
+ - + - ]
3643 [ + - + - : 26 : set_metric("thread_busy", "role", "groom", 0);
+ - + - ]
3644 : : }
3645 : :
3646 : 94 : scanq.done_idle();
3647 : 94 : }
3648 : :
3649 : 27 : return 0;
3650 : : }
3651 : :
3652 : :
3653 : : ////////////////////////////////////////////////////////////////////////
3654 : :
3655 : :
3656 : : static void
3657 : 28 : signal_handler (int /* sig */)
3658 : : {
3659 : 28 : interrupted ++;
3660 : :
3661 [ + + ]: 28 : if (db)
3662 : 27 : sqlite3_interrupt (db);
3663 [ + - ]: 28 : if (dbq)
3664 : 28 : sqlite3_interrupt (dbq);
3665 : :
3666 : : // NB: don't do anything else in here
3667 : 28 : }
3668 : :
3669 : : static void
3670 : 27 : sigusr1_handler (int /* sig */)
3671 : : {
3672 : 27 : sigusr1 ++;
3673 : : // NB: don't do anything else in here
3674 : 27 : }
3675 : :
3676 : : static void
3677 : 3 : sigusr2_handler (int /* sig */)
3678 : : {
3679 : 3 : sigusr2 ++;
3680 : : // NB: don't do anything else in here
3681 : 3 : }
3682 : :
3683 : :
3684 : :
3685 : :
3686 : :
3687 : : // A user-defined sqlite function, to score the sharedness of the
3688 : : // prefix of two strings. This is used to compare candidate debuginfo
3689 : : // / source-rpm names, so that the closest match
3690 : : // (directory-topology-wise closest) is found. This is important in
3691 : : // case the same sref (source file name) is in many -debuginfo or
3692 : : // -debugsource RPMs, such as when multiple versions/releases of the
3693 : : // same package are in the database.
3694 : :
3695 : 131 : static void sqlite3_sharedprefix_fn (sqlite3_context* c, int argc, sqlite3_value** argv)
3696 : : {
3697 [ - + ]: 131 : if (argc != 2)
3698 : 0 : sqlite3_result_error(c, "expect 2 string arguments", -1);
3699 [ + - + + ]: 262 : else if ((sqlite3_value_type(argv[0]) != SQLITE_TEXT) ||
3700 [ + + ]: 131 : (sqlite3_value_type(argv[1]) != SQLITE_TEXT))
3701 : 3 : sqlite3_result_null(c);
3702 : : else
3703 : : {
3704 : 128 : const unsigned char* a = sqlite3_value_text (argv[0]);
3705 : 128 : const unsigned char* b = sqlite3_value_text (argv[1]);
3706 : 128 : int i = 0;
3707 [ + + ]: 10314 : while (*a++ == *b++)
3708 : 10186 : i++;
3709 : 128 : sqlite3_result_int (c, i);
3710 : : }
3711 : 131 : }
3712 : :
3713 : :
3714 : : int
3715 : 28 : main (int argc, char *argv[])
3716 : : {
3717 : 28 : (void) setlocale (LC_ALL, "");
3718 : 28 : (void) bindtextdomain (PACKAGE_TARNAME, LOCALEDIR);
3719 : 28 : (void) textdomain (PACKAGE_TARNAME);
3720 : :
3721 : : /* Tell the library which version we are expecting. */
3722 [ + - ]: 28 : elf_version (EV_CURRENT);
3723 : :
3724 [ - + + - ]: 28 : tmpdir = string(getenv("TMPDIR") ?: "/tmp");
3725 : :
3726 : : /* Set computed default values. */
3727 [ + - + - : 28 : db_path = string(getenv("HOME") ?: "/") + string("/.debuginfod.sqlite"); /* XDG? */
+ - + - ]
3728 [ + - ]: 28 : int rc = regcomp (& file_include_regex, ".*", REG_EXTENDED|REG_NOSUB); // match everything
3729 [ - + ]: 28 : if (rc != 0)
3730 : : error (EXIT_FAILURE, 0, "regcomp failure: %d", rc);
3731 [ + - ]: 28 : rc = regcomp (& file_exclude_regex, "^$", REG_EXTENDED|REG_NOSUB); // match nothing
3732 [ - + ]: 28 : if (rc != 0)
3733 : : error (EXIT_FAILURE, 0, "regcomp failure: %d", rc);
3734 : :
3735 : : // default parameters for fdcache are computed from system stats
3736 : : struct statfs sfs;
3737 : 28 : rc = statfs(tmpdir.c_str(), &sfs);
3738 [ - + ]: 28 : if (rc < 0)
3739 : 0 : fdcache_mbs = 1024; // 1 gigabyte
3740 : : else
3741 : 28 : fdcache_mbs = sfs.f_bavail * sfs.f_bsize / 1024 / 1024 / 4; // 25% of free space
3742 : 28 : fdcache_mintmp = 25; // emergency flush at 25% remaining (75% full)
3743 : 28 : fdcache_prefetch = 64; // guesstimate storage is this much less costly than re-decompression
3744 : 28 : fdcache_fds = (concurrency + fdcache_prefetch) * 2;
3745 : :
3746 : : /* Parse and process arguments. */
3747 : : int remaining;
3748 : 28 : argp_program_version_hook = print_version; // this works
3749 [ + - ]: 28 : (void) argp_parse (&argp, argc, argv, ARGP_IN_ORDER, &remaining, NULL);
3750 [ - + ]: 28 : if (remaining != argc)
3751 : 0 : error (EXIT_FAILURE, 0,
3752 : 0 : "unexpected argument: %s", argv[remaining]);
3753 : :
3754 [ + + + + : 28 : if (scan_archives.size()==0 && !scan_files && source_paths.size()>0)
- + - + ]
3755 [ # # # # : 0 : obatched(clog) << "warning: without -F -R -U -Z, ignoring PATHs" << endl;
# # ]
3756 : :
3757 [ + - ]: 28 : fdcache.limit(fdcache_fds, fdcache_mbs, fdcache_prefetch_fds, fdcache_prefetch_mbs);
3758 : :
3759 : 28 : (void) signal (SIGPIPE, SIG_IGN); // microhttpd can generate it incidentally, ignore
3760 : 28 : (void) signal (SIGINT, signal_handler); // ^C
3761 : 28 : (void) signal (SIGHUP, signal_handler); // EOF
3762 : 28 : (void) signal (SIGTERM, signal_handler); // systemd
3763 : 28 : (void) signal (SIGUSR1, sigusr1_handler); // end-user
3764 : 28 : (void) signal (SIGUSR2, sigusr2_handler); // end-user
3765 : :
3766 : : /* Get database ready. */
3767 [ + + ]: 28 : if (! passive_p)
3768 : : {
3769 [ + - ]: 27 : rc = sqlite3_open_v2 (db_path.c_str(), &db, (SQLITE_OPEN_READWRITE
3770 : : |SQLITE_OPEN_URI
3771 : : |SQLITE_OPEN_PRIVATECACHE
3772 : : |SQLITE_OPEN_CREATE
3773 : : |SQLITE_OPEN_FULLMUTEX), /* thread-safe */
3774 : : NULL);
3775 [ - + ]: 27 : if (rc == SQLITE_CORRUPT)
3776 : : {
3777 : 0 : (void) unlink (db_path.c_str());
3778 [ # # ]: 0 : error (EXIT_FAILURE, 0,
3779 : : "cannot open %s, deleted database: %s", db_path.c_str(), sqlite3_errmsg(db));
3780 : : }
3781 [ - + ]: 27 : else if (rc)
3782 : : {
3783 [ # # ]: 0 : error (EXIT_FAILURE, 0,
3784 : : "cannot open %s, consider deleting database: %s", db_path.c_str(), sqlite3_errmsg(db));
3785 : : }
3786 : : }
3787 : :
3788 : : // open the readonly query variant
3789 : : // NB: PRIVATECACHE allows web queries to operate in parallel with
3790 : : // much other grooming/scanning operation.
3791 [ + - ]: 28 : rc = sqlite3_open_v2 (db_path.c_str(), &dbq, (SQLITE_OPEN_READONLY
3792 : : |SQLITE_OPEN_URI
3793 : : |SQLITE_OPEN_PRIVATECACHE
3794 : : |SQLITE_OPEN_FULLMUTEX), /* thread-safe */
3795 : : NULL);
3796 [ - + ]: 28 : if (rc)
3797 : : {
3798 [ # # ]: 0 : error (EXIT_FAILURE, 0,
3799 : : "cannot open %s, consider deleting database: %s", db_path.c_str(), sqlite3_errmsg(dbq));
3800 : : }
3801 : :
3802 : :
3803 [ + - + - ]: 56 : obatched(clog) << "opened database " << db_path
3804 [ + - + + : 28 : << (db?" rw":"") << (dbq?" ro":"") << endl;
+ - + - +
- + - ]
3805 [ + - + - : 28 : obatched(clog) << "sqlite version " << sqlite3_version << endl;
+ - + - ]
3806 [ + - + - : 28 : obatched(clog) << "service mode " << (passive_p ? "passive":"active") << endl;
+ + + - +
- ]
3807 : :
3808 : : // add special string-prefix-similarity function used in rpm sref/sdef resolution
3809 [ + - ]: 28 : rc = sqlite3_create_function(dbq, "sharedprefix", 2, SQLITE_UTF8, NULL,
3810 : : & sqlite3_sharedprefix_fn, NULL, NULL);
3811 [ - + ]: 28 : if (rc != SQLITE_OK)
3812 [ # # ]: 0 : error (EXIT_FAILURE, 0,
3813 : : "cannot create sharedprefix function: %s", sqlite3_errmsg(dbq));
3814 : :
3815 [ + + ]: 28 : if (! passive_p)
3816 : : {
3817 [ + + ]: 27 : if (verbose > 3)
3818 [ + - + - : 14 : obatched(clog) << "ddl: " << DEBUGINFOD_SQLITE_DDL << endl;
+ - + - ]
3819 [ + - ]: 27 : rc = sqlite3_exec (db, DEBUGINFOD_SQLITE_DDL, NULL, NULL, NULL);
3820 [ - + ]: 27 : if (rc != SQLITE_OK)
3821 : : {
3822 [ # # ]: 0 : error (EXIT_FAILURE, 0,
3823 : : "cannot run database schema ddl: %s", sqlite3_errmsg(db));
3824 : : }
3825 : : }
3826 : :
3827 : : // Start httpd server threads. Separate pool for IPv4 and IPv6, in
3828 : : // case the host only has one protocol stack.
3829 [ + - ]: 28 : MHD_Daemon *d4 = MHD_start_daemon (MHD_USE_THREAD_PER_CONNECTION
3830 : : #if MHD_VERSION >= 0x00095300
3831 : : | MHD_USE_INTERNAL_POLLING_THREAD
3832 : : #else
3833 : : | MHD_USE_SELECT_INTERNALLY
3834 : : #endif
3835 : : | MHD_USE_DEBUG, /* report errors to stderr */
3836 : : http_port,
3837 : : NULL, NULL, /* default accept policy */
3838 : : handler_cb, NULL, /* handler callback */
3839 : : MHD_OPTION_END);
3840 [ + - ]: 28 : MHD_Daemon *d6 = MHD_start_daemon (MHD_USE_THREAD_PER_CONNECTION
3841 : : #if MHD_VERSION >= 0x00095300
3842 : : | MHD_USE_INTERNAL_POLLING_THREAD
3843 : : #else
3844 : : | MHD_USE_SELECT_INTERNALLY
3845 : : #endif
3846 : : | MHD_USE_IPv6
3847 : : | MHD_USE_DEBUG, /* report errors to stderr */
3848 : : http_port,
3849 : : NULL, NULL, /* default accept policy */
3850 : : handler_cb, NULL, /* handler callback */
3851 : : MHD_OPTION_END);
3852 : :
3853 [ - + - - ]: 28 : if (d4 == NULL && d6 == NULL) // neither ipv4 nor ipv6? boo
3854 : : {
3855 : 0 : sqlite3 *database = db;
3856 : 0 : sqlite3 *databaseq = dbq;
3857 : 0 : db = dbq = 0; // for signal_handler not to freak
3858 [ # # ]: 0 : sqlite3_close (databaseq);
3859 [ # # ]: 0 : sqlite3_close (database);
3860 : 0 : error (EXIT_FAILURE, 0, "cannot start http server at port %d", http_port);
3861 : : }
3862 : :
3863 [ + - + - ]: 56 : obatched(clog) << "started http server on "
3864 : : << (d4 != NULL ? "IPv4 " : "")
3865 : : << (d6 != NULL ? "IPv6 " : "")
3866 [ + - + - : 28 : << "port=" << http_port << endl;
+ - + - +
- + - +
- ]
3867 : :
3868 : : // add maxigroom sql if -G given
3869 [ - + ]: 28 : if (maxigroom)
3870 : : {
3871 [ # # # # : 0 : obatched(clog) << "maxigrooming database, please wait." << endl;
# # ]
3872 [ # # # # ]: 0 : extra_ddl.push_back("create index if not exists " BUILDIDS "_r_sref_arc on " BUILDIDS "_r_sref(artifactsrc);");
3873 [ # # # # ]: 0 : extra_ddl.push_back("delete from " BUILDIDS "_r_sdef where not exists (select 1 from " BUILDIDS "_r_sref b where " BUILDIDS "_r_sdef.content = b.artifactsrc);");
3874 [ # # # # ]: 0 : extra_ddl.push_back("drop index if exists " BUILDIDS "_r_sref_arc;");
3875 : :
3876 : : // NB: we don't maxigroom the _files interning table. It'd require a temp index on all the
3877 : : // tables that have file foreign-keys, which is a lot.
3878 : :
3879 : : // NB: with =delete, may take up 3x disk space total during vacuum process
3880 : : // vs. =off (only 2x but may corrupt database if program dies mid-vacuum)
3881 : : // vs. =wal (>3x observed, but safe)
3882 [ # # # # ]: 0 : extra_ddl.push_back("pragma journal_mode=delete;");
3883 [ # # # # ]: 0 : extra_ddl.push_back("vacuum;");
3884 [ # # # # ]: 0 : extra_ddl.push_back("pragma journal_mode=wal;");
3885 : : }
3886 : :
3887 : : // run extra -D sql if given
3888 [ + + ]: 28 : if (! passive_p)
3889 [ - + ]: 27 : for (auto&& i: extra_ddl)
3890 : : {
3891 [ # # ]: 0 : if (verbose > 1)
3892 [ # # # # : 0 : obatched(clog) << "extra ddl:\n" << i << endl;
# # # # ]
3893 [ # # ]: 0 : rc = sqlite3_exec (db, i.c_str(), NULL, NULL, NULL);
3894 [ # # # # : 0 : if (rc != SQLITE_OK && rc != SQLITE_DONE && rc != SQLITE_ROW)
# # ]
3895 [ # # ]: 0 : error (0, 0,
3896 : : "warning: cannot run database extra ddl %s: %s", i.c_str(), sqlite3_errmsg(db));
3897 : :
3898 [ # # ]: 0 : if (maxigroom)
3899 [ # # # # : 0 : obatched(clog) << "maxigroomed database" << endl;
# # ]
3900 : : }
3901 : :
3902 [ + + ]: 28 : if (! passive_p)
3903 [ + - + - : 27 : obatched(clog) << "search concurrency " << concurrency << endl;
+ - + - ]
3904 [ + + ]: 28 : if (! passive_p)
3905 [ + - + - : 27 : obatched(clog) << "rescan time " << rescan_s << endl;
+ - + - ]
3906 [ + - + - : 28 : obatched(clog) << "fdcache fds " << fdcache_fds << endl;
+ - + - ]
3907 [ + - + - : 28 : obatched(clog) << "fdcache mbs " << fdcache_mbs << endl;
+ - + - ]
3908 [ + - + - : 28 : obatched(clog) << "fdcache prefetch " << fdcache_prefetch << endl;
+ - + - ]
3909 [ + - + - : 28 : obatched(clog) << "fdcache tmpdir " << tmpdir << endl;
+ - + - ]
3910 [ + - + - : 28 : obatched(clog) << "fdcache tmpdir min% " << fdcache_mintmp << endl;
+ - + - ]
3911 [ + + ]: 28 : if (! passive_p)
3912 [ + - + - : 27 : obatched(clog) << "groom time " << groom_s << endl;
+ - + - ]
3913 [ + - + - : 28 : obatched(clog) << "prefetch fds " << fdcache_prefetch_fds << endl;
+ - + - ]
3914 [ + - + - : 28 : obatched(clog) << "prefetch mbs " << fdcache_prefetch_mbs << endl;
+ - + - ]
3915 [ + - + - : 28 : obatched(clog) << "forwarded ttl limit " << forwarded_ttl_limit << endl;
+ - + - ]
3916 : :
3917 [ + + ]: 28 : if (scan_archives.size()>0)
3918 : : {
3919 [ + - ]: 32 : obatched ob(clog);
3920 [ + - ]: 16 : auto& o = ob << "accepting archive types ";
3921 [ + + ]: 50 : for (auto&& arch : scan_archives)
3922 [ + - + - : 34 : o << arch.first << "(" << arch.second << ") ";
+ - + - ]
3923 [ + - ]: 16 : o << endl;
3924 : : }
3925 : 28 : const char* du = getenv(DEBUGINFOD_URLS_ENV_VAR);
3926 [ + + + + ]: 28 : if (du && du[0] != '\0') // set to non-empty string?
3927 [ + - + - : 7 : obatched(clog) << "upstream debuginfod servers: " << du << endl;
+ - + - ]
3928 : :
3929 : 28 : vector<pthread_t> all_threads;
3930 : :
3931 [ + + ]: 28 : if (! passive_p)
3932 : : {
3933 : : pthread_t pt;
3934 : 27 : rc = pthread_create (& pt, NULL, thread_main_groom, NULL);
3935 [ - + ]: 27 : if (rc)
3936 : : error (EXIT_FAILURE, rc, "cannot spawn thread to groom database\n");
3937 : : else
3938 : : {
3939 : : #ifdef HAVE_PTHREAD_SETNAME_NP
3940 : 27 : (void) pthread_setname_np (pt, "groom");
3941 : : #endif
3942 [ + - ]: 27 : all_threads.push_back(pt);
3943 : : }
3944 : :
3945 [ + + + + : 27 : if (scan_files || scan_archives.size() > 0)
+ + ]
3946 : : {
3947 : 24 : rc = pthread_create (& pt, NULL, thread_main_fts_source_paths, NULL);
3948 [ - + ]: 24 : if (rc)
3949 : : error (EXIT_FAILURE, rc, "cannot spawn thread to traverse source paths\n");
3950 : : #ifdef HAVE_PTHREAD_SETNAME_NP
3951 : 24 : (void) pthread_setname_np (pt, "traverse");
3952 : : #endif
3953 [ + - ]: 24 : all_threads.push_back(pt);
3954 : :
3955 [ + + ]: 120 : for (unsigned i=0; i<concurrency; i++)
3956 : : {
3957 : 96 : rc = pthread_create (& pt, NULL, thread_main_scanner, NULL);
3958 [ - + ]: 96 : if (rc)
3959 : : error (EXIT_FAILURE, rc, "cannot spawn thread to scan source files / archives\n");
3960 : : #ifdef HAVE_PTHREAD_SETNAME_NP
3961 : 96 : (void) pthread_setname_np (pt, "scan");
3962 : : #endif
3963 [ + - ]: 96 : all_threads.push_back(pt);
3964 : : }
3965 : : }
3966 : : }
3967 : :
3968 : : /* Trivial main loop! */
3969 [ + - + - ]: 28 : set_metric("ready", 1);
3970 [ + + ]: 86 : while (! interrupted)
3971 [ + - ]: 58 : pause ();
3972 [ + - ]: 28 : scanq.nuke(); // wake up any remaining scanq-related threads, let them die
3973 [ + - + - ]: 28 : set_metric("ready", 0);
3974 : :
3975 [ + - ]: 28 : if (verbose)
3976 [ + - + - : 28 : obatched(clog) << "stopping" << endl;
+ - ]
3977 : :
3978 : : /* Join all our threads. */
3979 [ + + ]: 175 : for (auto&& it : all_threads)
3980 [ + - ]: 147 : pthread_join (it, NULL);
3981 : :
3982 : : /* Stop all the web service threads. */
3983 [ + - + - ]: 28 : if (d4) MHD_stop_daemon (d4);
3984 [ + - + - ]: 28 : if (d6) MHD_stop_daemon (d6);
3985 : :
3986 [ + + ]: 28 : if (! passive_p)
3987 : : {
3988 : : /* With all threads known dead, we can clean up the global resources. */
3989 [ + - ]: 27 : rc = sqlite3_exec (db, DEBUGINFOD_SQLITE_CLEANUP_DDL, NULL, NULL, NULL);
3990 [ - + ]: 27 : if (rc != SQLITE_OK)
3991 : : {
3992 [ # # ]: 0 : error (0, 0,
3993 : : "warning: cannot run database cleanup ddl: %s", sqlite3_errmsg(db));
3994 : : }
3995 : : }
3996 : :
3997 : : // NB: no problem with unconditional free here - an earlier failed regcomp would exit program
3998 [ + - ]: 28 : (void) regfree (& file_include_regex);
3999 [ + - ]: 28 : (void) regfree (& file_exclude_regex);
4000 : :
4001 : 28 : sqlite3 *database = db;
4002 : 28 : sqlite3 *databaseq = dbq;
4003 : 28 : db = dbq = 0; // for signal_handler not to freak
4004 [ + - ]: 28 : (void) sqlite3_close (databaseq);
4005 [ + + ]: 28 : if (! passive_p)
4006 [ + - ]: 27 : (void) sqlite3_close (database);
4007 : :
4008 : 28 : return 0;
4009 : : }
|
