Branch data Line data Source code
1 : : /* Debuginfo-over-http server.
2 : : Copyright (C) 2019-2024 Red Hat, Inc.
3 : : Copyright (C) 2021, 2022 Mark J. Wielaard <mark@klomp.org>
4 : : This file is part of elfutils.
5 : :
6 : : This file is free software; you can redistribute it and/or modify
7 : : it under the terms of the GNU General Public License as published by
8 : : the Free Software Foundation; either version 3 of the License, or
9 : : (at your option) any later version.
10 : :
11 : : elfutils is distributed in the hope that it will be useful, but
12 : : WITHOUT ANY WARRANTY; without even the implied warranty of
13 : : MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 : : GNU General Public License for more details.
15 : :
16 : : You should have received a copy of the GNU General Public License
17 : : along with this program. If not, see <http://www.gnu.org/licenses/>. */
18 : :
19 : :
20 : : /* cargo-cult from libdwfl linux-kernel-modules.c */
21 : : /* In case we have a bad fts we include this before config.h because it
22 : : can't handle _FILE_OFFSET_BITS.
23 : : Everything we need here is fine if its declarations just come first.
24 : : Also, include sys/types.h before fts. On some systems fts.h is not self
25 : : contained. */
26 : : #ifdef BAD_FTS
27 : : #include <sys/types.h>
28 : : #include <fts.h>
29 : : #endif
30 : :
31 : : #ifdef HAVE_CONFIG_H
32 : : #include "config.h"
33 : : #endif
34 : :
35 : : // #define _GNU_SOURCE
36 : : #ifdef HAVE_SCHED_H
37 : : extern "C" {
38 : : #include <sched.h>
39 : : }
40 : : #endif
41 : : #ifdef HAVE_SYS_RESOURCE_H
42 : : extern "C" {
43 : : #include <sys/resource.h>
44 : : }
45 : : #endif
46 : :
47 : : #ifdef HAVE_EXECINFO_H
48 : : extern "C" {
49 : : #include <execinfo.h>
50 : : }
51 : : #endif
52 : : #ifdef HAVE_MALLOC_H
53 : : extern "C" {
54 : : #include <malloc.h>
55 : : }
56 : : #endif
57 : :
58 : : #include "debuginfod.h"
59 : : #include <dwarf.h>
60 : :
61 : : #include <argp.h>
62 : : #ifdef __GNUC__
63 : : #undef __attribute__ /* glibc bug - rhbz 1763325 */
64 : : #endif
65 : :
66 : : #include <unistd.h>
67 : : #include <stdlib.h>
68 : : #include <locale.h>
69 : : #include <pthread.h>
70 : : #include <signal.h>
71 : : #include <sys/stat.h>
72 : : #include <sys/time.h>
73 : : #include <sys/vfs.h>
74 : : #include <unistd.h>
75 : : #include <fcntl.h>
76 : : #include <netdb.h>
77 : : #include <math.h>
78 : : #include <float.h>
79 : :
80 : :
81 : : /* If fts.h is included before config.h, its indirect inclusions may not
82 : : give us the right LFS aliases of these functions, so map them manually. */
83 : : #ifdef BAD_FTS
84 : : #ifdef _FILE_OFFSET_BITS
85 : : #define open open64
86 : : #define fopen fopen64
87 : : #endif
88 : : #else
89 : : #include <sys/types.h>
90 : : #include <fts.h>
91 : : #endif
92 : :
93 : : #include <cstring>
94 : : #include <vector>
95 : : #include <set>
96 : : #include <unordered_set>
97 : : #include <map>
98 : : #include <string>
99 : : #include <iostream>
100 : : #include <iomanip>
101 : : #include <ostream>
102 : : #include <sstream>
103 : : #include <mutex>
104 : : #include <deque>
105 : : #include <condition_variable>
106 : : #include <exception>
107 : : #include <thread>
108 : : // #include <regex> // on rhel7 gcc 4.8, not competent
109 : : #include <regex.h>
110 : : // #include <algorithm>
111 : : using namespace std;
112 : :
113 : : #include <gelf.h>
114 : : #include <libdwelf.h>
115 : :
116 : : #include <microhttpd.h>
117 : :
118 : : #if MHD_VERSION >= 0x00097002
119 : : // libmicrohttpd 0.9.71 broke API
120 : : #define MHD_RESULT enum MHD_Result
121 : : #else
122 : : #define MHD_RESULT int
123 : : #endif
124 : :
125 : : #ifdef ENABLE_IMA_VERIFICATION
126 : : #include <rpm/rpmlib.h>
127 : : #include <rpm/rpmfi.h>
128 : : #include <rpm/header.h>
129 : : #include <glob.h>
130 : : #endif
131 : :
132 : : #include <curl/curl.h>
133 : : #include <archive.h>
134 : : #include <archive_entry.h>
135 : : #include <sqlite3.h>
136 : :
137 : : #ifdef __linux__
138 : : #include <sys/syscall.h>
139 : : #endif
140 : :
141 : : #ifdef __linux__
142 : : #define tid() syscall(SYS_gettid)
143 : : #else
144 : : #define tid() pthread_self()
145 : : #endif
146 : :
147 : : extern "C" {
148 : : #include "printversion.h"
149 : : #include "system.h"
150 : : }
151 : :
152 : :
153 : : inline bool
154 : 2024 : string_endswith(const string& haystack, const string& needle)
155 : : {
156 [ + - ]: 2024 : return (haystack.size() >= needle.size() &&
157 : 2024 : equal(haystack.end()-needle.size(), haystack.end(),
158 : 2024 : needle.begin()));
159 : : }
160 : :
161 : :
162 : : // Roll this identifier for every sqlite schema incompatibility.
163 : : #define BUILDIDS "buildids10"
164 : :
165 : : #if SQLITE_VERSION_NUMBER >= 3008000
166 : : #define WITHOUT_ROWID "without rowid"
167 : : #else
168 : : #define WITHOUT_ROWID ""
169 : : #endif
170 : :
171 : : static const char DEBUGINFOD_SQLITE_DDL[] =
172 : : "pragma foreign_keys = on;\n"
173 : : "pragma synchronous = 0;\n" // disable fsync()s - this cache is disposable across a machine crash
174 : : "pragma journal_mode = wal;\n" // https://sqlite.org/wal.html
175 : : "pragma wal_checkpoint = truncate;\n" // clean out any preexisting wal file
176 : : "pragma journal_size_limit = 0;\n" // limit steady state file (between grooming, which also =truncate's)
177 : : "pragma auto_vacuum = incremental;\n" // https://sqlite.org/pragma.html
178 : : "pragma busy_timeout = 1000;\n" // https://sqlite.org/pragma.html
179 : : // NB: all these are overridable with -D option
180 : :
181 : : // Normalization table for interning file names
182 : : "create table if not exists " BUILDIDS "_fileparts (\n"
183 : : " id integer primary key not null,\n"
184 : : " name text unique not null\n"
185 : : " );\n"
186 : : "create table if not exists " BUILDIDS "_files (\n"
187 : : " id integer primary key not null,\n"
188 : : " dirname integer not null,\n"
189 : : " basename integer not null,\n"
190 : : " unique (dirname, basename),\n"
191 : : " foreign key (dirname) references " BUILDIDS "_fileparts(id) on delete cascade,\n"
192 : : " foreign key (basename) references " BUILDIDS "_fileparts(id) on delete cascade\n"
193 : : " );\n"
194 : : "create view if not exists " BUILDIDS "_files_v as\n" // a
195 : : " select f.id, n1.name || '/' || n2.name as name\n"
196 : : " from " BUILDIDS "_files f, " BUILDIDS "_fileparts n1, " BUILDIDS "_fileparts n2\n"
197 : : " where f.dirname = n1.id and f.basename = n2.id;\n"
198 : :
199 : : // Normalization table for interning buildids
200 : : "create table if not exists " BUILDIDS "_buildids (\n"
201 : : " id integer primary key not null,\n"
202 : : " hex text unique not null);\n"
203 : : // Track the completion of scanning of a given file & sourcetype at given time
204 : : "create table if not exists " BUILDIDS "_file_mtime_scanned (\n"
205 : : " mtime integer not null,\n"
206 : : " file integer not null,\n"
207 : : " size integer not null,\n" // in bytes
208 : : " sourcetype text(1) not null\n"
209 : : " check (sourcetype IN ('F', 'R')),\n"
210 : : " foreign key (file) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
211 : : " primary key (file, mtime, sourcetype)\n"
212 : : " ) " WITHOUT_ROWID ";\n"
213 : : "create table if not exists " BUILDIDS "_f_de (\n"
214 : : " buildid integer not null,\n"
215 : : " debuginfo_p integer not null,\n"
216 : : " executable_p integer not null,\n"
217 : : " file integer not null,\n"
218 : : " mtime integer not null,\n"
219 : : " foreign key (file) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
220 : : " foreign key (buildid) references " BUILDIDS "_buildids(id) on update cascade on delete cascade,\n"
221 : : " primary key (buildid, file, mtime)\n"
222 : : " ) " WITHOUT_ROWID ";\n"
223 : : // Index for faster delete by file identifier
224 : : "create index if not exists " BUILDIDS "_f_de_idx on " BUILDIDS "_f_de (file, mtime);\n"
225 : : "create table if not exists " BUILDIDS "_f_s (\n"
226 : : " buildid integer not null,\n"
227 : : " artifactsrc integer not null,\n"
228 : : " file integer not null,\n" // NB: not necessarily entered into _mtime_scanned
229 : : " mtime integer not null,\n"
230 : : " foreign key (file) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
231 : : " foreign key (artifactsrc) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
232 : : " foreign key (buildid) references " BUILDIDS "_buildids(id) on update cascade on delete cascade,\n"
233 : : " primary key (buildid, artifactsrc, file, mtime)\n"
234 : : " ) " WITHOUT_ROWID ";\n"
235 : : "create table if not exists " BUILDIDS "_r_de (\n"
236 : : " buildid integer not null,\n"
237 : : " debuginfo_p integer not null,\n"
238 : : " executable_p integer not null,\n"
239 : : " file integer not null,\n"
240 : : " mtime integer not null,\n"
241 : : " content integer not null,\n"
242 : : " foreign key (file) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
243 : : " foreign key (content) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
244 : : " foreign key (buildid) references " BUILDIDS "_buildids(id) on update cascade on delete cascade,\n"
245 : : " primary key (buildid, debuginfo_p, executable_p, file, content, mtime)\n"
246 : : " ) " WITHOUT_ROWID ";\n"
247 : : // Index for faster delete by archive file identifier
248 : : "create index if not exists " BUILDIDS "_r_de_idx on " BUILDIDS "_r_de (file, mtime);\n"
249 : : "create table if not exists " BUILDIDS "_r_sref (\n" // outgoing dwarf sourcefile references from rpm
250 : : " buildid integer not null,\n"
251 : : " artifactsrc integer not null,\n"
252 : : " foreign key (artifactsrc) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
253 : : " foreign key (buildid) references " BUILDIDS "_buildids(id) on update cascade on delete cascade,\n"
254 : : " primary key (buildid, artifactsrc)\n"
255 : : " ) " WITHOUT_ROWID ";\n"
256 : : "create table if not exists " BUILDIDS "_r_sdef (\n" // rpm contents that may satisfy sref
257 : : " file integer not null,\n"
258 : : " mtime integer not null,\n"
259 : : " content integer not null,\n"
260 : : " foreign key (file) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
261 : : " foreign key (content) references " BUILDIDS "_files(id) on update cascade on delete cascade,\n"
262 : : " primary key (content, file, mtime)\n"
263 : : " ) " WITHOUT_ROWID ";\n"
264 : : // create views to glue together some of the above tables, for webapi D queries
265 : : "create view if not exists " BUILDIDS "_query_d as \n"
266 : : "select\n"
267 : : " b.hex as buildid, n.mtime, 'F' as sourcetype, f0.name as source0, n.mtime as mtime, null as source1\n"
268 : : " from " BUILDIDS "_buildids b, " BUILDIDS "_files_v f0, " BUILDIDS "_f_de n\n"
269 : : " where b.id = n.buildid and f0.id = n.file and n.debuginfo_p = 1\n"
270 : : "union all select\n"
271 : : " b.hex as buildid, n.mtime, 'R' as sourcetype, f0.name as source0, n.mtime as mtime, f1.name as source1\n"
272 : : " from " BUILDIDS "_buildids b, " BUILDIDS "_files_v f0, " BUILDIDS "_files_v f1, " BUILDIDS "_r_de n\n"
273 : : " where b.id = n.buildid and f0.id = n.file and f1.id = n.content and n.debuginfo_p = 1\n"
274 : : ";"
275 : : // ... and for E queries
276 : : "create view if not exists " BUILDIDS "_query_e as \n"
277 : : "select\n"
278 : : " b.hex as buildid, n.mtime, 'F' as sourcetype, f0.name as source0, n.mtime as mtime, null as source1\n"
279 : : " from " BUILDIDS "_buildids b, " BUILDIDS "_files_v f0, " BUILDIDS "_f_de n\n"
280 : : " where b.id = n.buildid and f0.id = n.file and n.executable_p = 1\n"
281 : : "union all select\n"
282 : : " b.hex as buildid, n.mtime, 'R' as sourcetype, f0.name as source0, n.mtime as mtime, f1.name as source1\n"
283 : : " from " BUILDIDS "_buildids b, " BUILDIDS "_files_v f0, " BUILDIDS "_files_v f1, " BUILDIDS "_r_de n\n"
284 : : " where b.id = n.buildid and f0.id = n.file and f1.id = n.content and n.executable_p = 1\n"
285 : : ";"
286 : : // ... and for S queries
287 : : "create view if not exists " BUILDIDS "_query_s as \n"
288 : : "select\n"
289 : : " 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"
290 : : " from " BUILDIDS "_buildids b, " BUILDIDS "_files_v f0, " BUILDIDS "_files_v fs, " BUILDIDS "_f_s n\n"
291 : : " where b.id = n.buildid and f0.id = n.file and fs.id = n.artifactsrc\n"
292 : : "union all select\n"
293 : : " 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"
294 : : " from " BUILDIDS "_buildids b, " BUILDIDS "_files_v f0, " BUILDIDS "_files_v f1, " BUILDIDS "_files_v fsref, "
295 : : " " BUILDIDS "_r_sdef sd, " BUILDIDS "_r_sref sr, " BUILDIDS "_r_de sde\n"
296 : : " where b.id = sr.buildid and f0.id = sd.file and fsref.id = sde.file and f1.id = sd.content\n"
297 : : " and sr.artifactsrc = sd.content and sde.buildid = sr.buildid\n"
298 : : ";"
299 : : // and for startup overview counts
300 : : "drop view if exists " BUILDIDS "_stats;\n"
301 : : "create view if not exists " BUILDIDS "_stats as\n"
302 : : " select 'file d/e' as label,count(*) as quantity from " BUILDIDS "_f_de\n"
303 : : "union all select 'file s',count(*) from " BUILDIDS "_f_s\n"
304 : : "union all select 'archive d/e',count(*) from " BUILDIDS "_r_de\n"
305 : : "union all select 'archive sref',count(*) from " BUILDIDS "_r_sref\n"
306 : : "union all select 'archive sdef',count(*) from " BUILDIDS "_r_sdef\n"
307 : : "union all select 'buildids',count(*) from " BUILDIDS "_buildids\n"
308 : : "union all select 'filenames',count(*) from " BUILDIDS "_files\n"
309 : : "union all select 'fileparts',count(*) from " BUILDIDS "_fileparts\n"
310 : : "union all select 'files scanned (#)',count(*) from " BUILDIDS "_file_mtime_scanned\n"
311 : : "union all select 'files scanned (mb)',coalesce(sum(size)/1024/1024,0) from " BUILDIDS "_file_mtime_scanned\n"
312 : : #if SQLITE_VERSION_NUMBER >= 3016000
313 : : "union all select 'index db size (mb)',page_count*page_size/1024/1024 as size FROM pragma_page_count(), pragma_page_size()\n"
314 : : #endif
315 : : ";\n"
316 : :
317 : : // schema change history & garbage collection
318 : : //
319 : : // XXX: we could have migration queries here to bring prior-schema
320 : : // data over instead of just dropping it. But that could incur
321 : : // doubled storage costs.
322 : : //
323 : : // buildids10: split the _files table into _parts
324 : : "" // <<< we are here
325 : : // buildids9: widen the mtime_scanned table
326 : : "DROP VIEW IF EXISTS buildids9_stats;\n"
327 : : "DROP INDEX IF EXISTS buildids9_r_de_idx;\n"
328 : : "DROP INDEX IF EXISTS buildids9_f_de_idx;\n"
329 : : "DROP VIEW IF EXISTS buildids9_query_s;\n"
330 : : "DROP VIEW IF EXISTS buildids9_query_e;\n"
331 : : "DROP VIEW IF EXISTS buildids9_query_d;\n"
332 : : "DROP TABLE IF EXISTS buildids9_r_sdef;\n"
333 : : "DROP TABLE IF EXISTS buildids9_r_sref;\n"
334 : : "DROP TABLE IF EXISTS buildids9_r_de;\n"
335 : : "DROP TABLE IF EXISTS buildids9_f_s;\n"
336 : : "DROP TABLE IF EXISTS buildids9_f_de;\n"
337 : : "DROP TABLE IF EXISTS buildids9_file_mtime_scanned;\n"
338 : : "DROP TABLE IF EXISTS buildids9_buildids;\n"
339 : : "DROP TABLE IF EXISTS buildids9_files;\n"
340 : : // buildids8: slim the sref table
341 : : "drop table if exists buildids8_f_de;\n"
342 : : "drop table if exists buildids8_f_s;\n"
343 : : "drop table if exists buildids8_r_de;\n"
344 : : "drop table if exists buildids8_r_sref;\n"
345 : : "drop table if exists buildids8_r_sdef;\n"
346 : : "drop table if exists buildids8_file_mtime_scanned;\n"
347 : : "drop table if exists buildids8_files;\n"
348 : : "drop table if exists buildids8_buildids;\n"
349 : : // buildids7: separate _norm table into dense subtype tables
350 : : "drop table if exists buildids7_f_de;\n"
351 : : "drop table if exists buildids7_f_s;\n"
352 : : "drop table if exists buildids7_r_de;\n"
353 : : "drop table if exists buildids7_r_sref;\n"
354 : : "drop table if exists buildids7_r_sdef;\n"
355 : : "drop table if exists buildids7_file_mtime_scanned;\n"
356 : : "drop table if exists buildids7_files;\n"
357 : : "drop table if exists buildids7_buildids;\n"
358 : : // buildids6: drop bolo/rfolo again, represent sources / rpmcontents in main table
359 : : "drop table if exists buildids6_norm;\n"
360 : : "drop table if exists buildids6_files;\n"
361 : : "drop table if exists buildids6_buildids;\n"
362 : : "drop view if exists buildids6;\n"
363 : : // buildids5: redefine srcfile1 column to be '.'-less (for rpms)
364 : : "drop table if exists buildids5_norm;\n"
365 : : "drop table if exists buildids5_files;\n"
366 : : "drop table if exists buildids5_buildids;\n"
367 : : "drop table if exists buildids5_bolo;\n"
368 : : "drop table if exists buildids5_rfolo;\n"
369 : : "drop view if exists buildids5;\n"
370 : : // buildids4: introduce rpmfile RFOLO
371 : : "drop table if exists buildids4_norm;\n"
372 : : "drop table if exists buildids4_files;\n"
373 : : "drop table if exists buildids4_buildids;\n"
374 : : "drop table if exists buildids4_bolo;\n"
375 : : "drop table if exists buildids4_rfolo;\n"
376 : : "drop view if exists buildids4;\n"
377 : : // buildids3*: split out srcfile BOLO
378 : : "drop table if exists buildids3_norm;\n"
379 : : "drop table if exists buildids3_files;\n"
380 : : "drop table if exists buildids3_buildids;\n"
381 : : "drop table if exists buildids3_bolo;\n"
382 : : "drop view if exists buildids3;\n"
383 : : // buildids2: normalized buildid and filenames into interning tables;
384 : : "drop table if exists buildids2_norm;\n"
385 : : "drop table if exists buildids2_files;\n"
386 : : "drop table if exists buildids2_buildids;\n"
387 : : "drop view if exists buildids2;\n"
388 : : // buildids1: made buildid and artifacttype NULLable, to represent cached-negative
389 : : // lookups from sources, e.g. files or rpms that contain no buildid-indexable content
390 : : "drop table if exists buildids1;\n"
391 : : // buildids: original
392 : : "drop table if exists buildids;\n"
393 : : ;
394 : :
395 : : static const char DEBUGINFOD_SQLITE_CLEANUP_DDL[] =
396 : : "pragma wal_checkpoint = truncate;\n" // clean out any preexisting wal file
397 : : ;
398 : :
399 : :
400 : :
401 : :
402 : : /* Name and version of program. */
403 : : ARGP_PROGRAM_VERSION_HOOK_DEF = print_version;
404 : :
405 : : /* Bug report address. */
406 : : ARGP_PROGRAM_BUG_ADDRESS_DEF = PACKAGE_BUGREPORT;
407 : :
408 : : /* Definitions of arguments for argp functions. */
409 : : static const struct argp_option options[] =
410 : : {
411 : : { NULL, 0, NULL, 0, "Scanners:", 1 },
412 : : { "scan-file-dir", 'F', NULL, 0, "Enable ELF/DWARF file scanning.", 0 },
413 : : { "scan-rpm-dir", 'R', NULL, 0, "Enable RPM scanning.", 0 },
414 : : { "scan-deb-dir", 'U', NULL, 0, "Enable DEB scanning.", 0 },
415 : : { "scan-archive", 'Z', "EXT=CMD", 0, "Enable arbitrary archive scanning.", 0 },
416 : : // "source-oci-imageregistry" ...
417 : :
418 : : { NULL, 0, NULL, 0, "Options:", 2 },
419 : : { "logical", 'L', NULL, 0, "Follow symlinks, default=ignore.", 0 },
420 : : { "rescan-time", 't', "SECONDS", 0, "Number of seconds to wait between rescans, 0=disable.", 0 },
421 : : { "groom-time", 'g', "SECONDS", 0, "Number of seconds to wait between database grooming, 0=disable.", 0 },
422 : : { "maxigroom", 'G', NULL, 0, "Run a complete database groom/shrink pass at startup.", 0 },
423 : : { "concurrency", 'c', "NUM", 0, "Limit scanning thread concurrency to NUM, default=#CPUs.", 0 },
424 : : { "connection-pool", 'C', "NUM", OPTION_ARG_OPTIONAL,
425 : : "Use webapi connection pool with NUM threads, default=unlim.", 0 },
426 : : { "include", 'I', "REGEX", 0, "Include files matching REGEX, default=all.", 0 },
427 : : { "exclude", 'X', "REGEX", 0, "Exclude files matching REGEX, default=none.", 0 },
428 : : { "port", 'p', "NUM", 0, "HTTP port to listen on, default 8002.", 0 },
429 : : { "database", 'd', "FILE", 0, "Path to sqlite database.", 0 },
430 : : { "ddl", 'D', "SQL", 0, "Apply extra sqlite ddl/pragma to connection.", 0 },
431 : : { "verbose", 'v', NULL, 0, "Increase verbosity.", 0 },
432 : : { "regex-groom", 'r', NULL, 0,"Uses regexes from -I and -X arguments to groom the database.",0},
433 : : #define ARGP_KEY_FDCACHE_FDS 0x1001
434 : : { "fdcache-fds", ARGP_KEY_FDCACHE_FDS, "NUM", OPTION_HIDDEN, NULL, 0 },
435 : : #define ARGP_KEY_FDCACHE_MBS 0x1002
436 : : { "fdcache-mbs", ARGP_KEY_FDCACHE_MBS, "MB", 0, "Maximum total size of archive file fdcache.", 0 },
437 : : #define ARGP_KEY_FDCACHE_PREFETCH 0x1003
438 : : { "fdcache-prefetch", ARGP_KEY_FDCACHE_PREFETCH, "NUM", 0, "Number of archive files to prefetch into fdcache.", 0 },
439 : : #define ARGP_KEY_FDCACHE_MINTMP 0x1004
440 : : { "fdcache-mintmp", ARGP_KEY_FDCACHE_MINTMP, "NUM", 0, "Minimum free space% on tmpdir.", 0 },
441 : : #define ARGP_KEY_FDCACHE_PREFETCH_MBS 0x1005
442 : : { "fdcache-prefetch-mbs", ARGP_KEY_FDCACHE_PREFETCH_MBS, "MB", OPTION_HIDDEN, NULL, 0},
443 : : #define ARGP_KEY_FDCACHE_PREFETCH_FDS 0x1006
444 : : { "fdcache-prefetch-fds", ARGP_KEY_FDCACHE_PREFETCH_FDS, "NUM", OPTION_HIDDEN, NULL, 0},
445 : : #define ARGP_KEY_FORWARDED_TTL_LIMIT 0x1007
446 : : {"forwarded-ttl-limit", ARGP_KEY_FORWARDED_TTL_LIMIT, "NUM", 0, "Limit of X-Forwarded-For hops, default 8.", 0},
447 : : #define ARGP_KEY_PASSIVE 0x1008
448 : : { "passive", ARGP_KEY_PASSIVE, NULL, 0, "Do not scan or groom, read-only database.", 0 },
449 : : #define ARGP_KEY_DISABLE_SOURCE_SCAN 0x1009
450 : : { "disable-source-scan", ARGP_KEY_DISABLE_SOURCE_SCAN, NULL, 0, "Do not scan dwarf source info.", 0 },
451 : : #define ARGP_SCAN_CHECKPOINT 0x100A
452 : : { "scan-checkpoint", ARGP_SCAN_CHECKPOINT, "NUM", 0, "Number of files scanned before a WAL checkpoint.", 0 },
453 : : #ifdef ENABLE_IMA_VERIFICATION
454 : : #define ARGP_KEY_KOJI_SIGCACHE 0x100B
455 : : { "koji-sigcache", ARGP_KEY_KOJI_SIGCACHE, NULL, 0, "Do a koji specific mapping of rpm paths to get IMA signatures.", 0 },
456 : : #endif
457 : : { NULL, 0, NULL, 0, NULL, 0 },
458 : : };
459 : :
460 : : /* Short description of program. */
461 : : static const char doc[] = "Serve debuginfo-related content across HTTP from files under PATHs.";
462 : :
463 : : /* Strings for arguments in help texts. */
464 : : static const char args_doc[] = "[PATH ...]";
465 : :
466 : : /* Prototype for option handler. */
467 : : static error_t parse_opt (int key, char *arg, struct argp_state *state);
468 : :
469 : : static unsigned default_concurrency();
470 : :
471 : : /* Data structure to communicate with argp functions. */
472 : : static struct argp argp =
473 : : {
474 : : options, parse_opt, args_doc, doc, NULL, NULL, NULL
475 : : };
476 : :
477 : :
478 : : static string db_path;
479 : : static sqlite3 *db; // single connection, serialized across all our threads!
480 : : static sqlite3 *dbq; // webapi query-servicing readonly connection, serialized ditto!
481 : : static unsigned verbose;
482 : : static volatile sig_atomic_t interrupted = 0;
483 : : static volatile sig_atomic_t forced_rescan_count = 0;
484 : : static volatile sig_atomic_t sigusr1 = 0;
485 : : static volatile sig_atomic_t forced_groom_count = 0;
486 : : static volatile sig_atomic_t sigusr2 = 0;
487 : : static unsigned http_port = 8002;
488 : : static unsigned rescan_s = 300;
489 : : static unsigned groom_s = 86400;
490 : : static bool maxigroom = false;
491 : : static unsigned concurrency = default_concurrency();
492 : : static int connection_pool = 0;
493 : : static set<string> source_paths;
494 : : static bool scan_files = false;
495 : : static map<string,string> scan_archives;
496 : : static vector<string> extra_ddl;
497 : : static regex_t file_include_regex;
498 : : static regex_t file_exclude_regex;
499 : : static bool regex_groom = false;
500 : : static bool traverse_logical;
501 : : static long fdcache_mbs;
502 : : static long fdcache_prefetch;
503 : : static long fdcache_mintmp;
504 : : static unsigned forwarded_ttl_limit = 8;
505 : : static bool scan_source_info = true;
506 : : static string tmpdir;
507 : : static bool passive_p = false;
508 : : static long scan_checkpoint = 256;
509 : : #ifdef ENABLE_IMA_VERIFICATION
510 : : static bool requires_koji_sigcache_mapping = false;
511 : : #endif
512 : :
513 : : static void set_metric(const string& key, double value);
514 : : static void inc_metric(const string& key);
515 : : static void add_metric(const string& metric,
516 : : double value);
517 : : static void set_metric(const string& metric,
518 : : const string& lname, const string& lvalue,
519 : : double value);
520 : : static void inc_metric(const string& metric,
521 : : const string& lname, const string& lvalue);
522 : : static void add_metric(const string& metric,
523 : : const string& lname, const string& lvalue,
524 : : double value);
525 : : static void inc_metric(const string& metric,
526 : : const string& lname, const string& lvalue,
527 : : const string& rname, const string& rvalue);
528 : : static void add_metric(const string& metric,
529 : : const string& lname, const string& lvalue,
530 : : const string& rname, const string& rvalue,
531 : : double value);
532 : :
533 : :
534 : : class tmp_inc_metric { // a RAII style wrapper for exception-safe scoped increment & decrement
535 : : string m, n, v;
536 : : public:
537 : 2621 : tmp_inc_metric(const string& mname, const string& lname, const string& lvalue):
538 [ + - + - ]: 2621 : m(mname), n(lname), v(lvalue)
539 : : {
540 [ + - ]: 2621 : add_metric (m, n, v, 1);
541 [ - - - - : 2621 : }
- - ]
542 : 2621 : ~tmp_inc_metric()
543 : : {
544 : 2621 : add_metric (m, n, v, -1);
545 [ - + - + : 2621 : }
- + ]
546 : : };
547 : :
548 : : class tmp_ms_metric { // a RAII style wrapper for exception-safe scoped timing
549 : : string m, n, v;
550 : : struct timespec ts_start;
551 : : public:
552 : 329423 : tmp_ms_metric(const string& mname, const string& lname, const string& lvalue):
553 [ + - + - ]: 329423 : m(mname), n(lname), v(lvalue)
554 : : {
555 : 329448 : clock_gettime (CLOCK_MONOTONIC, & ts_start);
556 [ - - - - ]: 329687 : }
557 : 329788 : ~tmp_ms_metric()
558 : : {
559 : 329788 : struct timespec ts_end;
560 : 329788 : clock_gettime (CLOCK_MONOTONIC, & ts_end);
561 : 329772 : double deltas = (ts_end.tv_sec - ts_start.tv_sec)
562 : 329772 : + (ts_end.tv_nsec - ts_start.tv_nsec)/1.e9;
563 : :
564 [ + - ]: 329772 : add_metric (m + "_milliseconds_sum", n, v, (deltas*1000.0));
565 [ + + ]: 329806 : inc_metric (m + "_milliseconds_count", n, v);
566 [ + + - + : 611895 : }
- + ]
567 : : };
568 : :
569 : :
570 : : /* Handle program arguments. */
571 : : static error_t
572 : 1148 : parse_opt (int key, char *arg,
573 : : struct argp_state *state __attribute__ ((unused)))
574 : : {
575 : 1148 : int rc;
576 [ + + + + : 1148 : switch (key)
+ + + + +
- + + - -
+ + + + +
+ + + + +
- + + ]
577 : : {
578 : 286 : case 'v': verbose ++; break;
579 : 74 : case 'd':
580 : : /* When using the in-memory database make sure it is shareable,
581 : : so we can open it twice as read/write and read-only. */
582 [ + + ]: 74 : if (strcmp (arg, ":memory:") == 0)
583 : 1162 : db_path = "file::memory:?cache=shared";
584 : : else
585 [ + - ]: 120 : db_path = string(arg);
586 : : break;
587 : 74 : case 'p': http_port = (unsigned) atoi(arg);
588 [ + - ]: 74 : if (http_port == 0 || http_port > 65535)
589 : 0 : argp_failure(state, 1, EINVAL, "port number");
590 : : break;
591 : 50 : case 'F': scan_files = true; break;
592 : 22 : case 'R':
593 [ + - + - : 22 : scan_archives[".rpm"]="cat"; // libarchive groks rpm natively
- + ]
594 : 22 : break;
595 : 14 : case 'U':
596 [ + - + - : 14 : scan_archives[".deb"]="(bsdtar -O -x -f - data.tar\\*)<";
- + ]
597 [ + - + - : 14 : scan_archives[".ddeb"]="(bsdtar -O -x -f - data.tar\\*)<";
- + ]
598 [ + - + - : 14 : scan_archives[".ipk"]="(bsdtar -O -x -f - data.tar\\*)<";
- + ]
599 : : // .udeb too?
600 : 14 : break;
601 : 38 : case 'Z':
602 : 38 : {
603 [ - + ]: 38 : char* extension = strchr(arg, '=');
604 [ - + ]: 38 : if (arg[0] == '\0')
605 : 0 : argp_failure(state, 1, EINVAL, "missing EXT");
606 [ + + ]: 38 : else if (extension)
607 [ + - + - : 20 : scan_archives[string(arg, (extension-arg))]=string(extension+1);
- + - + -
- ]
608 : : else
609 [ + - + - : 18 : scan_archives[string(arg)]=string("cat");
- + - + -
- ]
610 : : }
611 : : break;
612 : 8 : case 'L':
613 [ - + ]: 8 : if (passive_p)
614 : 0 : argp_failure(state, 1, EINVAL, "-L option inconsistent with passive mode");
615 : 8 : traverse_logical = true;
616 : 8 : break;
617 : 0 : case 'D':
618 [ # # ]: 0 : if (passive_p)
619 : 0 : argp_failure(state, 1, EINVAL, "-D option inconsistent with passive mode");
620 [ # # ]: 0 : extra_ddl.push_back(string(arg));
621 : 0 : break;
622 : 58 : case 't':
623 [ - + ]: 58 : if (passive_p)
624 : 0 : argp_failure(state, 1, EINVAL, "-t option inconsistent with passive mode");
625 : 58 : rescan_s = (unsigned) atoi(arg);
626 : 58 : break;
627 : 58 : case 'g':
628 [ - + ]: 58 : if (passive_p)
629 : 0 : argp_failure(state, 1, EINVAL, "-g option inconsistent with passive mode");
630 : 58 : groom_s = (unsigned) atoi(arg);
631 : 58 : break;
632 : 0 : case 'G':
633 [ # # ]: 0 : if (passive_p)
634 : 0 : argp_failure(state, 1, EINVAL, "-G option inconsistent with passive mode");
635 : 0 : maxigroom = true;
636 : 0 : break;
637 : 0 : case 'c':
638 [ # # ]: 0 : if (passive_p)
639 : 0 : argp_failure(state, 1, EINVAL, "-c option inconsistent with passive mode");
640 : 0 : concurrency = (unsigned) atoi(arg);
641 [ # # ]: 0 : if (concurrency < 1) concurrency = 1;
642 : : break;
643 : 6 : case 'C':
644 [ + + ]: 6 : if (arg)
645 : : {
646 : 4 : connection_pool = atoi(arg);
647 [ + - ]: 4 : if (connection_pool < 2)
648 : 0 : argp_failure(state, 1, EINVAL, "-C NUM minimum 2");
649 : : }
650 : : break;
651 : 4 : case 'I':
652 : : // NB: no problem with unconditional free here - an earlier failed regcomp would exit program
653 [ - + ]: 4 : if (passive_p)
654 : 0 : argp_failure(state, 1, EINVAL, "-I option inconsistent with passive mode");
655 : 4 : regfree (&file_include_regex);
656 : 4 : rc = regcomp (&file_include_regex, arg, REG_EXTENDED|REG_NOSUB);
657 [ + - ]: 4 : if (rc != 0)
658 : 0 : argp_failure(state, 1, EINVAL, "regular expression");
659 : : break;
660 : 6 : case 'X':
661 [ - + ]: 6 : if (passive_p)
662 : 0 : argp_failure(state, 1, EINVAL, "-X option inconsistent with passive mode");
663 : 6 : regfree (&file_exclude_regex);
664 : 6 : rc = regcomp (&file_exclude_regex, arg, REG_EXTENDED|REG_NOSUB);
665 [ + - ]: 6 : if (rc != 0)
666 : 0 : argp_failure(state, 1, EINVAL, "regular expression");
667 : : break;
668 : 4 : case 'r':
669 [ - + ]: 4 : if (passive_p)
670 : 0 : argp_failure(state, 1, EINVAL, "-r option inconsistent with passive mode");
671 : 4 : regex_groom = true;
672 : 4 : break;
673 : : case ARGP_KEY_FDCACHE_FDS:
674 : : // deprecated
675 : : break;
676 : 4 : case ARGP_KEY_FDCACHE_MBS:
677 : 4 : fdcache_mbs = atol (arg);
678 : 4 : break;
679 : 2 : case ARGP_KEY_FDCACHE_PREFETCH:
680 : 2 : fdcache_prefetch = atol (arg);
681 : 2 : break;
682 : 2 : case ARGP_KEY_FDCACHE_MINTMP:
683 : 2 : fdcache_mintmp = atol (arg);
684 [ + - ]: 2 : if( fdcache_mintmp > 100 || fdcache_mintmp < 0 )
685 : 0 : argp_failure(state, 1, EINVAL, "fdcache mintmp percent");
686 : : break;
687 : 4 : case ARGP_KEY_FORWARDED_TTL_LIMIT:
688 : 4 : forwarded_ttl_limit = (unsigned) atoi(arg);
689 : 4 : break;
690 : 102 : case ARGP_KEY_ARG:
691 [ + - ]: 102 : source_paths.insert(string(arg));
692 : 102 : break;
693 : : case ARGP_KEY_FDCACHE_PREFETCH_FDS:
694 : : // deprecated
695 : : break;
696 : : case ARGP_KEY_FDCACHE_PREFETCH_MBS:
697 : : // deprecated
698 : : break;
699 : 2 : case ARGP_KEY_PASSIVE:
700 : 2 : passive_p = true;
701 [ + - ]: 2 : if (source_paths.size() > 0
702 [ + - ]: 2 : || maxigroom
703 [ + - ]: 2 : || extra_ddl.size() > 0
704 [ + - + - ]: 4 : || traverse_logical)
705 : : // other conflicting options tricky to check
706 : 0 : argp_failure(state, 1, EINVAL, "inconsistent options with passive mode");
707 : : break;
708 : 0 : case ARGP_KEY_DISABLE_SOURCE_SCAN:
709 : 0 : scan_source_info = false;
710 : 0 : break;
711 : 2 : case ARGP_SCAN_CHECKPOINT:
712 : 2 : scan_checkpoint = atol (arg);
713 [ + - ]: 2 : if (scan_checkpoint < 0)
714 : 0 : argp_failure(state, 1, EINVAL, "scan checkpoint");
715 : : break;
716 : : #ifdef ENABLE_IMA_VERIFICATION
717 : : case ARGP_KEY_KOJI_SIGCACHE:
718 : : requires_koji_sigcache_mapping = true;
719 : : break;
720 : : #endif
721 : : // case 'h': argp_state_help (state, stderr, ARGP_HELP_LONG|ARGP_HELP_EXIT_OK);
722 : : default: return ARGP_ERR_UNKNOWN;
723 : : }
724 : :
725 : : return 0;
726 : : }
727 : :
728 : :
729 : : ////////////////////////////////////////////////////////////////////////
730 : :
731 : :
732 : : static void add_mhd_response_header (struct MHD_Response *r,
733 : : const char *h, const char *v);
734 : :
735 : : // represent errors that may get reported to an ostream and/or a libmicrohttpd connection
736 : :
737 : 8 : struct reportable_exception
738 : : {
739 : : int code;
740 : : string message;
741 : :
742 [ - - + - : 86 : reportable_exception(int c, const string& m): code(c), message(m) {}
- - + - +
- ]
743 [ - - - - : 604 : reportable_exception(const string& m): code(503), message(m) {}
- - - - +
- - - - -
+ - - - -
- - - + -
- - ]
744 : : reportable_exception(): code(503), message() {}
745 : :
746 : : void report(ostream& o) const; // defined under obatched() class below
747 : :
748 : 634 : MHD_RESULT mhd_send_response(MHD_Connection* c) const {
749 : 1268 : MHD_Response* r = MHD_create_response_from_buffer (message.size(),
750 : 634 : (void*) message.c_str(),
751 : : MHD_RESPMEM_MUST_COPY);
752 : 634 : add_mhd_response_header (r, "Content-Type", "text/plain");
753 : 634 : MHD_RESULT rc = MHD_queue_response (c, code, r);
754 : 634 : MHD_destroy_response (r);
755 : 634 : return rc;
756 : : }
757 : : };
758 : :
759 : :
760 : : struct sqlite_exception: public reportable_exception
761 : : {
762 : 0 : sqlite_exception(int rc, const string& msg):
763 [ # # # # : 0 : reportable_exception(string("sqlite3 error: ") + msg + ": " + string(sqlite3_errstr(rc) ?: "?")) {
# # # # #
# # # # #
# # # # #
# ]
764 [ # # # # : 0 : inc_metric("error_count","sqlite3",sqlite3_errstr(rc));
# # # # #
# # # # #
# # # # ]
765 [ # # ]: 0 : }
766 : : };
767 : :
768 [ + - - - ]: 4 : struct libc_exception: public reportable_exception
769 : : {
770 : 594 : libc_exception(int rc, const string& msg):
771 [ - + + - : 2376 : reportable_exception(string("libc error: ") + msg + ": " + string(strerror(rc) ?: "?")) {
+ - + - +
- - + - +
- + + - -
- ]
772 [ + - + - : 1188 : inc_metric("error_count","libc",strerror(rc));
+ - + - -
+ + - - -
- - ]
773 [ - - ]: 594 : }
774 : : };
775 : :
776 : :
777 : : struct archive_exception: public reportable_exception
778 : : {
779 : 0 : archive_exception(const string& msg):
780 [ # # # # : 0 : reportable_exception(string("libarchive error: ") + msg) {
# # ]
781 [ # # # # : 0 : inc_metric("error_count","libarchive",msg);
# # # # #
# ]
782 [ # # ]: 0 : }
783 : 0 : archive_exception(struct archive* a, const string& msg):
784 [ # # # # : 0 : reportable_exception(string("libarchive error: ") + msg + ": " + string(archive_error_string(a) ?: "?")) {
# # # # #
# # # # #
# # # # #
# ]
785 [ # # # # : 0 : inc_metric("error_count","libarchive",msg + ": " + string(archive_error_string(a) ?: "?"));
# # # # #
# # # # #
# # # # #
# # # # #
# # # # #
# ]
786 [ # # ]: 0 : }
787 : : };
788 : :
789 : :
790 : : struct elfutils_exception: public reportable_exception
791 : : {
792 : 0 : elfutils_exception(int rc, const string& msg):
793 [ # # # # : 0 : reportable_exception(string("elfutils error: ") + msg + ": " + string(elf_errmsg(rc) ?: "?")) {
# # # # #
# # # # #
# # # # #
# ]
794 [ # # # # : 0 : inc_metric("error_count","elfutils",elf_errmsg(rc));
# # # # #
# # # # #
# # # # ]
795 [ # # ]: 0 : }
796 : : };
797 : :
798 : :
799 : : ////////////////////////////////////////////////////////////////////////
800 : :
801 : : template <typename Payload>
802 : : class workq
803 : : {
804 : : unordered_set<Payload> q; // eliminate duplicates
805 : : mutex mtx;
806 : : condition_variable cv;
807 : : bool dead;
808 : : unsigned idlers; // number of threads busy with wait_idle / done_idle
809 : : unsigned fronters; // number of threads busy with wait_front / done_front
810 : :
811 : : public:
812 : 74 : workq() { dead = false; idlers = 0; fronters = 0; }
813 : 74 : ~workq() {}
814 : :
815 : 1186 : void push_back(const Payload& p)
816 : : {
817 : 1186 : unique_lock<mutex> lock(mtx);
818 [ + - ]: 1186 : q.insert (p);
819 [ + - + - : 2372 : set_metric("thread_work_pending","role","scan", q.size());
+ - + - -
+ - + - -
- - ]
820 : 1186 : cv.notify_all();
821 : 1186 : }
822 : :
823 : : // kill this workqueue, wake up all idlers / scanners
824 : 74 : void nuke() {
825 : 74 : unique_lock<mutex> lock(mtx);
826 : : // optional: q.clear();
827 : 74 : dead = true;
828 : 74 : cv.notify_all();
829 : 74 : }
830 : :
831 : : // clear the workqueue, when scanning is interrupted with USR2
832 : 0 : void clear() {
833 : 0 : unique_lock<mutex> lock(mtx);
834 : 0 : q.clear();
835 [ # # # # : 0 : set_metric("thread_work_pending","role","scan", q.size());
# # # # #
# # # # #
# # ]
836 : : // NB: there may still be some live fronters
837 : 0 : cv.notify_all(); // maybe wake up waiting idlers
838 : 0 : }
839 : :
840 : : // block this scanner thread until there is work to do and no active idler
841 : 1450 : bool wait_front (Payload& p)
842 : : {
843 : 1450 : unique_lock<mutex> lock(mtx);
844 [ + + + + : 4776 : while (!dead && (q.size() == 0 || idlers > 0))
+ + ]
845 [ + - ]: 3326 : cv.wait(lock);
846 [ + + ]: 1450 : if (dead)
847 : : return false;
848 : : else
849 : : {
850 [ + - ]: 1186 : p = * q.begin();
851 : 1186 : q.erase (q.begin());
852 : 1186 : fronters ++; // prevent idlers from starting awhile, even if empty q
853 [ + - + - : 2372 : set_metric("thread_work_pending","role","scan", q.size());
+ - + - -
+ - + - -
- - - - ]
854 : : // NB: don't wake up idlers yet! The consumer is busy
855 : : // processing this element until it calls done_front().
856 : 1186 : return true;
857 : : }
858 : 1450 : }
859 : :
860 : : // notify waitq that scanner thread is done with that last item
861 : 1186 : void done_front ()
862 : : {
863 : 1186 : unique_lock<mutex> lock(mtx);
864 : 1186 : fronters --;
865 [ + + + + ]: 1186 : if (q.size() == 0 && fronters == 0)
866 : 95 : cv.notify_all(); // maybe wake up waiting idlers
867 : 1186 : }
868 : :
869 : : // block this idler thread until there is no work to do
870 : 574 : void wait_idle ()
871 : : {
872 : 574 : unique_lock<mutex> lock(mtx);
873 : 574 : cv.notify_all(); // maybe wake up waiting scanners
874 [ + + + + : 629 : while (!dead && ((q.size() != 0) || fronters > 0))
+ + ]
875 [ + - ]: 55 : cv.wait(lock);
876 [ + - ]: 574 : idlers ++;
877 : 574 : }
878 : :
879 : 502 : void done_idle ()
880 : : {
881 : 502 : unique_lock<mutex> lock(mtx);
882 : 502 : idlers --;
883 : 502 : cv.notify_all(); // maybe wake up waiting scanners, but probably not (shutting down)
884 : 502 : }
885 : : };
886 : :
887 : : typedef struct stat stat_t;
888 : : typedef pair<string,stat_t> scan_payload;
889 : : inline bool operator< (const scan_payload& a, const scan_payload& b)
890 : : {
891 : : return a.first < b.first; // don't bother compare the stat fields
892 : : }
893 : :
894 : : namespace std { // https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56480
895 : : template<> struct hash<::scan_payload>
896 : : {
897 : 4866 : std::size_t operator() (const ::scan_payload& p) const noexcept
898 : : {
899 [ + + + + ]: 4866 : return hash<string>()(p.first);
900 : : }
901 : : };
902 : : template<> struct equal_to<::scan_payload>
903 : : {
904 : 471 : std::size_t operator() (const ::scan_payload& a, const ::scan_payload& b) const noexcept
905 : : {
906 [ - + - - ]: 471 : return a.first == b.first;
907 : : }
908 : : };
909 : : }
910 : :
911 : : static workq<scan_payload> scanq; // just a single one
912 : : // producer & idler: thread_main_fts_source_paths()
913 : : // consumer: thread_main_scanner()
914 : : // idler: thread_main_groom()
915 : :
916 : :
917 : : ////////////////////////////////////////////////////////////////////////
918 : :
919 : : // Unique set is a thread-safe structure that lends 'ownership' of a value
920 : : // to a thread. Other threads requesting the same thing are made to wait.
921 : : // It's like a semaphore-on-demand.
922 : : template <typename T>
923 : : class unique_set
924 : : {
925 : : private:
926 : : set<T> values;
927 : : mutex mtx;
928 : : condition_variable cv;
929 : : public:
930 : 57 : unique_set() {}
931 : 57 : ~unique_set() {}
932 : :
933 : 1909 : void acquire(const T& value)
934 : : {
935 : 1909 : unique_lock<mutex> lock(mtx);
936 [ + + ]: 2307 : while (values.find(value) != values.end())
937 [ + - ]: 398 : cv.wait(lock);
938 [ + - ]: 1909 : values.insert(value);
939 : 1909 : }
940 : :
941 : 1909 : void release(const T& value)
942 : : {
943 : 1909 : unique_lock<mutex> lock(mtx);
944 : : // assert (values.find(value) != values.end());
945 : 1909 : values.erase(value);
946 : 1909 : cv.notify_all();
947 : 1909 : }
948 : : };
949 : :
950 : :
951 : : // This is the object that's instantiate to uniquely hold a value in a
952 : : // RAII-pattern way.
953 : : template <typename T>
954 : : class unique_set_reserver
955 : : {
956 : : private:
957 : : unique_set<T>& please_hold;
958 : : T mine;
959 : : public:
960 : 1909 : unique_set_reserver(unique_set<T>& t, const T& value):
961 [ + - - - ]: 1909 : please_hold(t), mine(value) { please_hold.acquire(mine); }
962 [ + - ]: 1909 : ~unique_set_reserver() { please_hold.release(mine); }
963 : : };
964 : :
965 : :
966 : : ////////////////////////////////////////////////////////////////////////
967 : :
968 : : // periodic_barrier is a concurrency control object that lets N threads
969 : : // periodically (based on counter value) agree to wait at a barrier,
970 : : // let one of them carry out some work, then be set free
971 : :
972 : : class periodic_barrier
973 : : {
974 : : private:
975 : : unsigned period; // number of count() reports to trigger barrier activation
976 : : unsigned threads; // number of threads participating
977 : : mutex mtx; // protects all the following fields
978 : : unsigned counter; // count of count() reports in the current generation
979 : : unsigned generation; // barrier activation generation
980 : : unsigned waiting; // number of threads waiting for barrier
981 : : bool dead; // bring out your
982 : : condition_variable cv;
983 : : public:
984 : 66 : periodic_barrier(unsigned t, unsigned p):
985 : 66 : period(p), threads(t), counter(0), generation(0), waiting(0), dead(false) { }
986 : : virtual ~periodic_barrier() {}
987 : :
988 : : virtual void periodic_barrier_work() noexcept = 0;
989 : 66 : void nuke() {
990 : 66 : unique_lock<mutex> lock(mtx);
991 : 66 : dead = true;
992 : 66 : cv.notify_all();
993 : 66 : }
994 : :
995 : 1450 : void count()
996 : : {
997 : 1450 : unique_lock<mutex> lock(mtx);
998 : 1450 : unsigned prev_generation = this->generation;
999 [ + + ]: 1450 : if (counter < period-1) // normal case: counter just freely running
1000 : : {
1001 : 1334 : counter ++;
1002 : 1334 : return;
1003 : : }
1004 [ + + ]: 116 : else if (counter == period-1) // we're the doer
1005 : : {
1006 : 30 : counter = period; // entering barrier holding phase
1007 : 30 : cv.notify_all();
1008 [ + + + + ]: 138 : while (waiting < threads-1 && !dead)
1009 [ + - ]: 78 : cv.wait(lock);
1010 : : // all other threads are now stuck in the barrier
1011 : 30 : this->periodic_barrier_work(); // NB: we're holding the mutex the whole time
1012 : : // reset for next barrier, releasing other waiters
1013 : 30 : counter = 0;
1014 : 30 : generation ++;
1015 : 30 : cv.notify_all();
1016 : 30 : return;
1017 : : }
1018 [ + - ]: 86 : else if (counter == period) // we're a waiter, in holding phase
1019 : : {
1020 : 86 : waiting ++;
1021 : 86 : cv.notify_all();
1022 [ + + + - : 320 : while (counter == period && generation == prev_generation && !dead)
+ + ]
1023 [ + - ]: 148 : cv.wait(lock);
1024 : 86 : waiting --;
1025 : 86 : return;
1026 : : }
1027 : 1450 : }
1028 : : };
1029 : :
1030 : :
1031 : :
1032 : : ////////////////////////////////////////////////////////////////////////
1033 : :
1034 : :
1035 : : // Print a standard timestamp.
1036 : : static ostream&
1037 : 42183 : timestamp (ostream &o)
1038 : : {
1039 : 42183 : char datebuf[80];
1040 : 42183 : char *now2 = NULL;
1041 : 42183 : time_t now_t = time(NULL);
1042 : 42183 : struct tm now;
1043 : 42183 : struct tm *nowp = gmtime_r (&now_t, &now);
1044 [ + - ]: 42184 : if (nowp)
1045 : : {
1046 : 42184 : (void) strftime (datebuf, sizeof (datebuf), "%c", nowp);
1047 : 42184 : now2 = datebuf;
1048 : : }
1049 : :
1050 : 42184 : return o << "[" << (now2 ? now2 : "") << "] "
1051 [ - + ]: 42184 : << "(" << getpid () << "/" << tid() << "): ";
1052 : : }
1053 : :
1054 : :
1055 : : // A little class that impersonates an ostream to the extent that it can
1056 : : // take << streaming operations. It batches up the bits into an internal
1057 : : // stringstream until it is destroyed; then flushes to the original ostream.
1058 : : // It adds a timestamp
1059 : : class obatched
1060 : : {
1061 : : private:
1062 : : ostream& o;
1063 : : stringstream stro;
1064 : : static mutex lock;
1065 : : public:
1066 : 42183 : obatched(ostream& oo, bool timestamp_p = true): o(oo)
1067 : : {
1068 [ + - ]: 42184 : if (timestamp_p)
1069 [ + - ]: 42184 : timestamp(stro);
1070 : 42184 : }
1071 : 42182 : ~obatched()
1072 : : {
1073 : 42182 : unique_lock<mutex> do_not_cross_the_streams(obatched::lock);
1074 [ + - ]: 42184 : o << stro.str();
1075 : 42184 : o.flush();
1076 : 42184 : }
1077 : : operator ostream& () { return stro; }
1078 [ - - + - : 30270 : template <typename T> ostream& operator << (const T& t) { stro << t; return stro; }
+ - + - +
- + - + -
- - - - -
- + - + -
+ - + - +
- + - + -
+ - + - +
- + - + -
- - + - +
- + - + -
+ - + - +
- + - - -
+ - + - +
- + - + -
+ - + - +
- + - + -
+ - + - -
- - - - -
+ - - - +
- - - - -
- - + - -
- + - - -
+ - - - -
- + - + -
+ - + - +
- + - - -
- - ]
1079 : : };
1080 : : mutex obatched::lock; // just the one, since cout/cerr iostreams are not thread-safe
1081 : :
1082 : :
1083 : 696 : void reportable_exception::report(ostream& o) const {
1084 [ + - + - ]: 696 : obatched(o) << message << endl;
1085 : 696 : }
1086 : :
1087 : :
1088 : : ////////////////////////////////////////////////////////////////////////
1089 : :
1090 : :
1091 : : // RAII style sqlite prepared-statement holder that matches { } block lifetime
1092 : :
1093 : : struct sqlite_ps
1094 : : {
1095 : : private:
1096 : : sqlite3* db;
1097 : : const string nickname;
1098 : : const string sql;
1099 : : sqlite3_stmt *pp;
1100 : :
1101 : : sqlite_ps(const sqlite_ps&); // make uncopyable
1102 : : sqlite_ps& operator=(const sqlite_ps &); // make unassignable
1103 : :
1104 : : public:
1105 [ + - - - ]: 7075 : sqlite_ps (sqlite3* d, const string& n, const string& s): db(d), nickname(n), sql(s) {
1106 : : // tmp_ms_metric tick("sqlite3","prep",nickname);
1107 [ + + ]: 7075 : if (verbose > 4)
1108 [ + - + - : 348 : obatched(clog) << nickname << " prep " << sql << endl;
+ - + - +
- - - ]
1109 [ + - ]: 7075 : int rc = sqlite3_prepare_v2 (db, sql.c_str(), -1 /* to \0 */, & this->pp, NULL);
1110 [ - + ]: 7075 : if (rc != SQLITE_OK)
1111 [ # # # # ]: 0 : throw sqlite_exception(rc, "prepare " + sql);
1112 : 7075 : }
1113 : :
1114 : 181351 : sqlite_ps& reset()
1115 : : {
1116 [ + - + - : 362678 : tmp_ms_metric tick("sqlite3","reset",nickname);
- + - - ]
1117 [ + - ]: 181327 : sqlite3_reset(this->pp);
1118 : 181349 : return *this;
1119 : 181346 : }
1120 : :
1121 : 208326 : sqlite_ps& bind(int parameter, const string& str)
1122 : : {
1123 [ + + ]: 208326 : if (verbose > 4)
1124 [ + - + - : 1418 : obatched(clog) << nickname << " bind " << parameter << "=" << str << endl;
+ - + - +
- + - ]
1125 : 208326 : int rc = sqlite3_bind_text (this->pp, parameter, str.c_str(), -1, SQLITE_TRANSIENT);
1126 [ - + ]: 208318 : if (rc != SQLITE_OK)
1127 [ # # # # ]: 0 : throw sqlite_exception(rc, "sqlite3 bind");
1128 : 208318 : return *this;
1129 : : }
1130 : :
1131 : 53124 : sqlite_ps& bind(int parameter, int64_t value)
1132 : : {
1133 [ + + ]: 53124 : if (verbose > 4)
1134 [ + - + - : 636 : obatched(clog) << nickname << " bind " << parameter << "=" << value << endl;
+ - + - +
- + - ]
1135 : 53124 : int rc = sqlite3_bind_int64 (this->pp, parameter, value);
1136 [ - + ]: 53123 : if (rc != SQLITE_OK)
1137 [ # # # # ]: 0 : throw sqlite_exception(rc, "sqlite3 bind");
1138 : 53123 : return *this;
1139 : : }
1140 : :
1141 : : sqlite_ps& bind(int parameter)
1142 : : {
1143 : : if (verbose > 4)
1144 : : obatched(clog) << nickname << " bind " << parameter << "=" << "NULL" << endl;
1145 : : int rc = sqlite3_bind_null (this->pp, parameter);
1146 : : if (rc != SQLITE_OK)
1147 : : throw sqlite_exception(rc, "sqlite3 bind");
1148 : : return *this;
1149 : : }
1150 : :
1151 : :
1152 : 112126 : void step_ok_done() {
1153 [ + - + - : 224225 : tmp_ms_metric tick("sqlite3","step_done",nickname);
- + - - ]
1154 [ + - ]: 112099 : int rc = sqlite3_step (this->pp);
1155 [ + + ]: 112140 : if (verbose > 4)
1156 [ + - + - : 832 : obatched(clog) << nickname << " step-ok-done(" << sqlite3_errstr(rc) << ") " << sql << endl;
+ - + - +
- + - + -
+ - ]
1157 [ + + - + ]: 112140 : if (rc != SQLITE_OK && rc != SQLITE_DONE && rc != SQLITE_ROW)
1158 [ # # # # ]: 0 : throw sqlite_exception(rc, "sqlite3 step");
1159 [ + - ]: 112140 : (void) sqlite3_reset (this->pp);
1160 : 112140 : }
1161 : :
1162 : :
1163 : 36310 : int step() {
1164 [ + - + - : 72620 : tmp_ms_metric tick("sqlite3","step",nickname);
- + - - ]
1165 [ + - ]: 36310 : int rc = sqlite3_step (this->pp);
1166 [ + + ]: 36311 : if (verbose > 4)
1167 [ + - + - : 378 : obatched(clog) << nickname << " step(" << sqlite3_errstr(rc) << ") " << sql << endl;
+ - + - +
- + - + -
+ - ]
1168 : 36311 : return rc;
1169 : 36311 : }
1170 : :
1171 [ + + + + ]: 14069 : ~sqlite_ps () { sqlite3_finalize (this->pp); }
1172 [ + - + - : 4143 : operator sqlite3_stmt* () { return this->pp; }
+ - ]
1173 : : };
1174 : :
1175 : :
1176 : : ////////////////////////////////////////////////////////////////////////
1177 : :
1178 : :
1179 : : struct sqlite_checkpoint_pb: public periodic_barrier
1180 : : {
1181 : : // NB: don't use sqlite_ps since it can throw exceptions during ctor etc.
1182 : 66 : sqlite_checkpoint_pb(unsigned t, unsigned p):
1183 : 132 : periodic_barrier(t, p) { }
1184 : :
1185 : 30 : void periodic_barrier_work() noexcept
1186 : : {
1187 : 30 : (void) sqlite3_exec (db, "pragma wal_checkpoint(truncate);", NULL, NULL, NULL);
1188 : 30 : }
1189 : : };
1190 : :
1191 : : static periodic_barrier* scan_barrier = 0; // initialized in main()
1192 : :
1193 : :
1194 : : ////////////////////////////////////////////////////////////////////////
1195 : :
1196 : : // RAII style templated autocloser
1197 : :
1198 : : template <class Payload, class Ignore>
1199 : : struct defer_dtor
1200 : : {
1201 : : public:
1202 : : typedef Ignore (*dtor_fn) (Payload);
1203 : :
1204 : : private:
1205 : : Payload p;
1206 : : dtor_fn fn;
1207 : :
1208 : : public:
1209 : 3322 : defer_dtor(Payload _p, dtor_fn _fn): p(_p), fn(_fn) {}
1210 : 714 : ~defer_dtor() { (void) (*fn)(p); }
1211 : :
1212 : : private:
1213 : : defer_dtor(const defer_dtor<Payload,Ignore>&); // make uncopyable
1214 : : defer_dtor& operator=(const defer_dtor<Payload,Ignore> &); // make unassignable
1215 : : };
1216 : :
1217 : :
1218 : :
1219 : : ////////////////////////////////////////////////////////////////////////
1220 : :
1221 : :
1222 : : static string
1223 : 5254 : header_censor(const string& str)
1224 : : {
1225 : 5254 : string y;
1226 [ + + ]: 55857 : for (auto&& x : str)
1227 : : {
1228 [ + + ]: 50603 : if (isalnum(x) || x == '/' || x == '.' || x == ',' || x == '_' || x == ':')
1229 [ + - ]: 101200 : y += x;
1230 : : }
1231 : 5254 : return y;
1232 : 0 : }
1233 : :
1234 : :
1235 : : static string
1236 : 2627 : conninfo (struct MHD_Connection * conn)
1237 : : {
1238 : 2627 : char hostname[256]; // RFC1035
1239 : 2627 : char servname[256];
1240 : 2627 : int sts = -1;
1241 : :
1242 [ - + ]: 2627 : if (conn == 0)
1243 : 0 : return "internal";
1244 : :
1245 : : /* Look up client address data. */
1246 : 2627 : const union MHD_ConnectionInfo *u = MHD_get_connection_info (conn,
1247 : : MHD_CONNECTION_INFO_CLIENT_ADDRESS);
1248 [ + - ]: 2627 : struct sockaddr *so = u ? u->client_addr : 0;
1249 : :
1250 [ + - - + ]: 2627 : if (so && so->sa_family == AF_INET) {
1251 : 0 : sts = getnameinfo (so, sizeof (struct sockaddr_in),
1252 : : hostname, sizeof (hostname),
1253 : : servname, sizeof (servname),
1254 : : NI_NUMERICHOST | NI_NUMERICSERV);
1255 [ + - ]: 2627 : } else if (so && so->sa_family == AF_INET6) {
1256 : 2627 : struct sockaddr_in6* addr6 = (struct sockaddr_in6*) so;
1257 [ + - + - : 2627 : if (IN6_IS_ADDR_V4MAPPED(&addr6->sin6_addr)) {
- + ]
1258 : 2627 : struct sockaddr_in addr4;
1259 : 2627 : memset (&addr4, 0, sizeof(addr4));
1260 : 2627 : addr4.sin_family = AF_INET;
1261 : 2627 : addr4.sin_port = addr6->sin6_port;
1262 : 2627 : memcpy (&addr4.sin_addr.s_addr, addr6->sin6_addr.s6_addr+12, sizeof(addr4.sin_addr.s_addr));
1263 : 2627 : sts = getnameinfo ((struct sockaddr*) &addr4, sizeof (addr4),
1264 : : hostname, sizeof (hostname),
1265 : : servname, sizeof (servname),
1266 : : NI_NUMERICHOST | NI_NUMERICSERV);
1267 : : } else {
1268 : 0 : sts = getnameinfo (so, sizeof (struct sockaddr_in6),
1269 : : hostname, sizeof (hostname),
1270 : : servname, sizeof (servname),
1271 : : NI_NUMERICHOST | NI_NUMERICSERV);
1272 : : }
1273 : : }
1274 : :
1275 [ - + ]: 2627 : if (sts != 0) {
1276 : 0 : hostname[0] = servname[0] = '\0';
1277 : : }
1278 : :
1279 : : // extract headers relevant to administration
1280 [ - + ]: 2627 : const char* user_agent = MHD_lookup_connection_value (conn, MHD_HEADER_KIND, "User-Agent") ?: "";
1281 [ + + ]: 2627 : const char* x_forwarded_for = MHD_lookup_connection_value (conn, MHD_HEADER_KIND, "X-Forwarded-For") ?: "";
1282 : : // NB: these are untrustworthy, beware if machine-processing log files
1283 : :
1284 [ + - + - : 7881 : return string(hostname) + string(":") + string(servname) +
+ - + - +
- + - - +
- + - + -
+ - + - +
- - - - -
- ]
1285 [ + - + - : 11335 : string(" UA:") + header_censor(string(user_agent)) +
+ - + - +
- - + + +
- + + + -
+ - - -
- ]
1286 [ + - + - : 7893 : string(" XFF:") + header_censor(string(x_forwarded_for));
+ - + + +
+ - - ]
1287 : : }
1288 : :
1289 : :
1290 : :
1291 : : ////////////////////////////////////////////////////////////////////////
1292 : :
1293 : : /* Wrapper for MHD_add_response_header that logs an error if we
1294 : : couldn't add the specified header. */
1295 : : static void
1296 : 8687 : add_mhd_response_header (struct MHD_Response *r,
1297 : : const char *h, const char *v)
1298 : : {
1299 [ - + ]: 8687 : if (MHD_add_response_header (r, h, v) == MHD_NO)
1300 [ # # # # : 0 : obatched(clog) << "Error: couldn't add '" << h << "' header" << endl;
# # ]
1301 : 8687 : }
1302 : :
1303 : : static void
1304 : 1323 : add_mhd_last_modified (struct MHD_Response *resp, time_t mtime)
1305 : : {
1306 : 1323 : struct tm now;
1307 : 1323 : struct tm *nowp = gmtime_r (&mtime, &now);
1308 [ + - ]: 1323 : if (nowp != NULL)
1309 : : {
1310 : 1323 : char datebuf[80];
1311 : 1323 : size_t rc = strftime (datebuf, sizeof (datebuf), "%a, %d %b %Y %T GMT",
1312 : : nowp);
1313 [ + - ]: 1323 : if (rc > 0 && rc < sizeof (datebuf))
1314 : 1323 : add_mhd_response_header (resp, "Last-Modified", datebuf);
1315 : : }
1316 : :
1317 : 1323 : add_mhd_response_header (resp, "Cache-Control", "public");
1318 : 1323 : }
1319 : :
1320 : : // quote all questionable characters of str for safe passage through a sh -c expansion.
1321 : : static string
1322 : 56 : shell_escape(const string& str)
1323 : : {
1324 : 56 : string y;
1325 [ + + ]: 7308 : for (auto&& x : str)
1326 : : {
1327 [ + + + + ]: 7252 : if (! isalnum(x) && x != '/')
1328 [ + - ]: 574 : y += "\\";
1329 [ + - ]: 14504 : y += x;
1330 : : }
1331 : 56 : return y;
1332 : 0 : }
1333 : :
1334 : :
1335 : : // PR25548: Perform POSIX / RFC3986 style path canonicalization on the input string.
1336 : : //
1337 : : // Namely:
1338 : : // // -> /
1339 : : // /foo/../ -> /
1340 : : // /./ -> /
1341 : : //
1342 : : // This mapping is done on dwarf-side source path names, which may
1343 : : // include these constructs, so we can deal with debuginfod clients
1344 : : // that accidentally canonicalize the paths.
1345 : : //
1346 : : // realpath(3) is close but not quite right, because it also resolves
1347 : : // symbolic links. Symlinks at the debuginfod server have nothing to
1348 : : // do with the build-time symlinks, thus they must not be considered.
1349 : : //
1350 : : // see also curl Curl_dedotdotify() aka RFC3986, which we mostly follow here
1351 : : // see also libc __realpath()
1352 : : // see also llvm llvm::sys::path::remove_dots()
1353 : : static string
1354 : 15425 : canon_pathname (const string& input)
1355 : : {
1356 : 15425 : string i = input; // 5.2.4 (1)
1357 : 15425 : string o;
1358 : :
1359 : 147238 : while (i.size() != 0)
1360 : : {
1361 : : // 5.2.4 (2) A
1362 [ + - - + : 263626 : if (i.substr(0,3) == "../")
- + ]
1363 [ # # # # ]: 0 : i = i.substr(3);
1364 [ + - - + : 263626 : else if(i.substr(0,2) == "./")
- + ]
1365 [ # # # # ]: 0 : i = i.substr(2);
1366 : :
1367 : : // 5.2.4 (2) B
1368 [ + - - + : 263626 : else if (i.substr(0,3) == "/./")
+ + ]
1369 [ + - + + ]: 2826 : i = i.substr(2);
1370 [ - + ]: 130165 : else if (i == "/.")
1371 [ # # ]: 0 : i = ""; // no need to handle "/." complete-path-segment case; we're dealing with file names
1372 : :
1373 : : // 5.2.4 (2) C
1374 [ + - - + : 260330 : else if (i.substr(0,4) == "/../") {
+ + ]
1375 [ + - + + ]: 2068 : i = i.substr(3);
1376 : 2068 : string::size_type sl = o.rfind("/");
1377 [ + - ]: 2068 : if (sl != string::npos)
1378 [ + - + - ]: 4136 : o = o.substr(0, sl);
1379 : : else
1380 [ # # ]: 0 : o = "";
1381 [ - + ]: 128097 : } else if (i == "/..")
1382 [ # # ]: 0 : i = ""; // no need to handle "/.." complete-path-segment case; we're dealing with file names
1383 : :
1384 : : // 5.2.4 (2) D
1385 : : // no need to handle these cases; we're dealing with file names
1386 [ - + ]: 128097 : else if (i == ".")
1387 [ # # ]: 0 : i = "";
1388 [ - + ]: 128097 : else if (i == "..")
1389 [ # # ]: 0 : i = "";
1390 : :
1391 : : // POSIX special: map // to /
1392 [ + - - + : 256194 : else if (i.substr(0,2) == "//")
+ + ]
1393 [ + - + + ]: 152 : i = i.substr(1);
1394 : :
1395 : : // 5.2.4 (2) E
1396 : : else {
1397 [ - + ]: 127961 : string::size_type next_slash = i.find("/", (i[0]=='/' ? 1 : 0)); // skip first slash
1398 [ + - + + ]: 255922 : o += i.substr(0, next_slash);
1399 [ + + ]: 127961 : if (next_slash == string::npos)
1400 [ + - + + ]: 162663 : i = "";
1401 : : else
1402 [ + - + + : 212691 : i = i.substr(next_slash);
- - ]
1403 : : }
1404 : : }
1405 : :
1406 [ + - ]: 15425 : return o;
1407 : 15425 : }
1408 : :
1409 : :
1410 : : // Estimate available free space for a given filesystem via statfs(2).
1411 : : // Return true if the free fraction is known to be smaller than the
1412 : : // given minimum percentage. Also update a related metric.
1413 : 2652 : bool statfs_free_enough_p(const string& path, const string& label, long minfree = 0)
1414 : : {
1415 : 2652 : struct statfs sfs;
1416 : 2652 : int rc = statfs(path.c_str(), &sfs);
1417 [ + + ]: 2652 : if (rc == 0)
1418 : : {
1419 : 2602 : double s = (double) sfs.f_bavail / (double) sfs.f_blocks;
1420 [ + - + - : 5204 : set_metric("filesys_free_ratio","purpose",label, s);
- + - - ]
1421 : 2602 : return ((s * 100.0) < minfree);
1422 : : }
1423 : : return false;
1424 : : }
1425 : :
1426 : :
1427 : :
1428 : : // A map-like class that owns a cache of file descriptors (indexed by
1429 : : // file / content names).
1430 : : //
1431 : : // If only it could use fd's instead of file names ... but we can't
1432 : : // dup(2) to create independent descriptors for the same unlinked
1433 : : // files, so would have to use some goofy linux /proc/self/fd/%d
1434 : : // hack such as the following
1435 : :
1436 : : #if 0
1437 : : int superdup(int fd)
1438 : : {
1439 : : #ifdef __linux__
1440 : : char *fdpath = NULL;
1441 : : int rc = asprintf(& fdpath, "/proc/self/fd/%d", fd);
1442 : : int newfd;
1443 : : if (rc >= 0)
1444 : : newfd = open(fdpath, O_RDONLY);
1445 : : else
1446 : : newfd = -1;
1447 : : free (fdpath);
1448 : : return newfd;
1449 : : #else
1450 : : return -1;
1451 : : #endif
1452 : : }
1453 : : #endif
1454 : :
1455 : : class libarchive_fdcache
1456 : : {
1457 : : private:
1458 : : mutex fdcache_lock;
1459 : :
1460 : : typedef pair<string,string> key; // archive, entry
1461 [ + - ]: 208 : struct fdcache_entry
1462 : : {
1463 : : string fd; // file name (probably in $TMPDIR), not an actual open fd (EMFILE)
1464 : : double fd_size_mb; // slightly rounded up megabytes
1465 : : time_t freshness; // when was this entry created or requested last
1466 : : unsigned request_count; // how many requests were made; or 0=prefetch only
1467 : : double latency; // how many seconds it took to extract the file
1468 : : };
1469 : :
1470 : : map<key,fdcache_entry> entries; // optimized for lookup
1471 : : time_t last_cleaning;
1472 : : long max_mbs;
1473 : :
1474 : : public:
1475 : 282 : void set_metrics()
1476 : : {
1477 : 282 : double fdcache_mb = 0.0;
1478 : 282 : double prefetch_mb = 0.0;
1479 : 282 : unsigned fdcache_count = 0;
1480 : 282 : unsigned prefetch_count = 0;
1481 [ + + ]: 1730 : for (auto &i : entries) {
1482 [ + + ]: 1448 : if (i.second.request_count) {
1483 : 1354 : fdcache_mb += i.second.fd_size_mb;
1484 : 1354 : fdcache_count ++;
1485 : : } else {
1486 : 94 : prefetch_mb += i.second.fd_size_mb;
1487 : 94 : prefetch_count ++;
1488 : : }
1489 : : }
1490 [ + - ]: 282 : set_metric("fdcache_bytes", fdcache_mb*1024.0*1024.0);
1491 [ + - ]: 282 : set_metric("fdcache_count", fdcache_count);
1492 [ + - ]: 282 : set_metric("fdcache_prefetch_bytes", prefetch_mb*1024.0*1024.0);
1493 [ + - ]: 282 : set_metric("fdcache_prefetch_count", prefetch_count);
1494 : 282 : }
1495 : :
1496 : 208 : void intern(const string& a, const string& b, string fd, off_t sz,
1497 : : bool requested_p, double lat)
1498 : : {
1499 : 208 : {
1500 : 208 : unique_lock<mutex> lock(fdcache_lock);
1501 : 208 : time_t now = time(NULL);
1502 : : // there is a chance it's already in here, just wasn't found last time
1503 : : // if so, there's nothing to do but count our luck
1504 [ + - ]: 208 : auto i = entries.find(make_pair(a,b));
1505 [ - + ]: 208 : if (i != entries.end())
1506 : : {
1507 [ # # # # : 0 : inc_metric("fdcache_op_count","op","redundant_intern");
# # # # #
# # # # #
# # ]
1508 [ # # ]: 0 : if (requested_p) i->second.request_count ++; // repeat prefetch doesn't count
1509 : 0 : i->second.freshness = now;
1510 : : // We need to nuke the temp file, since interning passes
1511 : : // responsibility over the path to this structure. It is
1512 : : // possible that the caller still has an fd open, but that's
1513 : : // OK.
1514 : 0 : unlink (fd.c_str());
1515 : 0 : return;
1516 : : }
1517 : 208 : double mb = (sz+65535)/1048576.0; // round up to 64K block
1518 : 208 : fdcache_entry n = { .fd=fd, .fd_size_mb=mb,
1519 : 208 : .freshness=now, .request_count = requested_p?1U:0U,
1520 [ + - + + ]: 208 : .latency=lat};
1521 [ + - + - : 208 : entries.insert(make_pair(make_pair(a,b),n));
+ - ]
1522 : :
1523 [ + + ]: 208 : if (requested_p)
1524 [ + - + - : 320 : inc_metric("fdcache_op_count","op","enqueue");
+ - + - -
+ - + - -
- - ]
1525 : : else
1526 [ + - + - : 144 : inc_metric("fdcache_op_count","op","prefetch_enqueue");
+ - + - -
+ + - - -
- - ]
1527 : :
1528 [ + + ]: 208 : if (verbose > 3)
1529 [ + - + - : 534 : obatched(clog) << "fdcache interned a=" << a << " b=" << b
- - ]
1530 [ + - + - : 178 : << " fd=" << fd << " mb=" << mb << " front=" << requested_p
+ - + - +
- + - + -
+ - ]
1531 [ + - + - : 178 : << " latency=" << lat << endl;
+ - ]
1532 : :
1533 [ + - ]: 208 : set_metrics();
1534 : 208 : }
1535 : :
1536 : : // NB: we age the cache at lookup time too
1537 [ + - - + : 208 : if (statfs_free_enough_p(tmpdir, "tmpdir", fdcache_mintmp))
- + ]
1538 : : {
1539 [ # # # # : 0 : inc_metric("fdcache_op_count","op","emerg-flush");
# # # # #
# # # #
# ]
1540 [ # # ]: 0 : obatched(clog) << "fdcache emergency flush for filling tmpdir" << endl;
1541 : 0 : this->limit(0); // emergency flush
1542 : : }
1543 : : else // age cache normally
1544 : 208 : this->limit(max_mbs);
1545 : : }
1546 : :
1547 : 740 : int lookup(const string& a, const string& b)
1548 : : {
1549 : 740 : int fd = -1;
1550 : 740 : {
1551 : 740 : unique_lock<mutex> lock(fdcache_lock);
1552 [ + - ]: 740 : auto i = entries.find(make_pair(a,b));
1553 [ + + ]: 740 : if (i != entries.end())
1554 : : {
1555 [ + + ]: 576 : if (i->second.request_count == 0) // was a prefetch!
1556 : : {
1557 [ + - + - ]: 16 : inc_metric("fdcache_prefetch_saved_milliseconds_count");
1558 [ + - + - ]: 32 : add_metric("fdcache_prefetch_saved_milliseconds_sum", i->second.latency*1000.);
1559 : : }
1560 : 576 : i->second.request_count ++;
1561 : 576 : i->second.freshness = time(NULL);
1562 : : // brag about our success
1563 [ + - + - : 1152 : inc_metric("fdcache_op_count","op","prefetch_access"); // backward compat
+ - + - -
+ - + - -
- - ]
1564 [ + - + - ]: 576 : inc_metric("fdcache_saved_milliseconds_count");
1565 [ + - + - ]: 576 : add_metric("fdcache_saved_milliseconds_sum", i->second.latency*1000.);
1566 [ + - ]: 740 : fd = open(i->second.fd.c_str(), O_RDONLY);
1567 : : }
1568 : 0 : }
1569 : :
1570 [ + + ]: 740 : if (fd >= 0)
1571 [ + - + - : 1152 : inc_metric("fdcache_op_count","op","lookup_hit");
+ - - + -
+ - - -
- ]
1572 : : else
1573 [ + - + - : 328 : inc_metric("fdcache_op_count","op","lookup_miss");
+ - - + -
+ - - -
- ]
1574 : :
1575 : : // NB: no need to age the cache after just a lookup
1576 : :
1577 : 740 : return fd;
1578 : : }
1579 : :
1580 : 226 : int probe(const string& a, const string& b) // just a cache residency check - don't modify state, don't open
1581 : : {
1582 : 226 : unique_lock<mutex> lock(fdcache_lock);
1583 [ + - ]: 226 : auto i = entries.find(make_pair(a,b));
1584 [ + + ]: 226 : if (i != entries.end()) {
1585 [ + - + - : 52 : inc_metric("fdcache_op_count","op","probe_hit");
+ - + - -
+ - + - -
- - ]
1586 : 26 : return true;
1587 : : } else {
1588 [ + - + - : 400 : inc_metric("fdcache_op_count","op","probe_miss");
+ - + - -
+ - + - -
- - ]
1589 : 200 : return false;
1590 : : }
1591 : 226 : }
1592 : :
1593 : 0 : void clear(const string& a, const string& b)
1594 : : {
1595 : 0 : unique_lock<mutex> lock(fdcache_lock);
1596 [ # # ]: 0 : auto i = entries.find(make_pair(a,b));
1597 [ # # ]: 0 : if (i != entries.end()) {
1598 [ # # # # : 0 : inc_metric("fdcache_op_count","op",
# # # # #
# # # # #
# # # # ]
1599 [ # # ]: 0 : i->second.request_count > 0 ? "clear" : "prefetch_clear");
1600 : 0 : unlink (i->second.fd.c_str());
1601 : 0 : entries.erase(i);
1602 [ # # ]: 0 : set_metrics();
1603 : 0 : return;
1604 : : }
1605 : 0 : }
1606 : :
1607 : 356 : void limit(long maxmbs, bool metrics_p = true)
1608 : : {
1609 : 356 : time_t now = time(NULL);
1610 : :
1611 : : // avoid overly frequent limit operations
1612 [ + + + + ]: 356 : if (maxmbs > 0 && (now - this->last_cleaning) < 10) // probably not worth parametrizing
1613 : 208 : return;
1614 : 148 : this->last_cleaning = now;
1615 : :
1616 [ + + + - ]: 148 : if (verbose > 3 && (this->max_mbs != maxmbs))
1617 [ + - + - ]: 192 : obatched(clog) << "fdcache limited to maxmbs=" << maxmbs << endl;
1618 : :
1619 : 148 : unique_lock<mutex> lock(fdcache_lock);
1620 : :
1621 : 148 : this->max_mbs = maxmbs;
1622 : 148 : double total_mb = 0.0;
1623 : :
1624 : 148 : map<double, pair<string,string>> sorted_entries;
1625 [ + + ]: 356 : for (auto &i: entries)
1626 : : {
1627 : 208 : total_mb += i.second.fd_size_mb;
1628 : :
1629 : : // need a scalar quantity that combines these inputs in a sensible way:
1630 : : //
1631 : : // 1) freshness of this entry (last time it was accessed)
1632 : : // 2) size of this entry
1633 : : // 3) number of times it has been accessed (or if just prefetched with 0 accesses)
1634 : : // 4) latency it required to extract
1635 : : //
1636 : : // The lower the "score", the earlier garbage collection will
1637 : : // nuke it, so to prioritize entries for preservation, the
1638 : : // score should be higher, and vice versa.
1639 : 208 : time_t factor_1_freshness = (now - i.second.freshness); // seconds
1640 : 208 : double factor_2_size = i.second.fd_size_mb; // megabytes
1641 : 208 : unsigned factor_3_accesscount = i.second.request_count; // units
1642 : 208 : double factor_4_latency = i.second.latency; // seconds
1643 : :
1644 : : #if 0
1645 : : double score = - factor_1_freshness; // simple LRU
1646 : : #endif
1647 : :
1648 [ + + ]: 208 : double score = 0.
1649 : 208 : - log1p(factor_1_freshness) // penalize old file
1650 : 208 : - log1p(factor_2_size) // penalize large file
1651 : 208 : + factor_4_latency * factor_3_accesscount; // reward slow + repeatedly read files
1652 : :
1653 [ + + ]: 208 : if (verbose > 4)
1654 [ + - ]: 64 : obatched(clog) << "fdcache scored score=" << score
1655 [ + - + - ]: 64 : << " a=" << i.first.first << " b=" << i.first.second
1656 [ + - + - : 96 : << " f1=" << factor_1_freshness << " f2=" << factor_2_size
+ - + - +
- + - +
- ]
1657 [ + - + - : 32 : << " f3=" << factor_3_accesscount << " f4=" << factor_4_latency
+ - + - +
- ]
1658 : 32 : << endl;
1659 : :
1660 [ + - + - ]: 416 : sorted_entries.insert(make_pair(score, i.first));
1661 : : }
1662 : :
1663 : 148 : unsigned cleaned = 0;
1664 : 148 : unsigned entries_original = entries.size();
1665 : 148 : double cleaned_score_min = DBL_MAX;
1666 : 148 : double cleaned_score_max = DBL_MIN;
1667 : :
1668 : : // drop as many entries[] as needed to bring total mb down to the threshold
1669 [ + + ]: 356 : for (auto &i: sorted_entries) // in increasing score order!
1670 : : {
1671 [ - + ]: 208 : if (this->max_mbs > 0 // if this is not a "clear entire table"
1672 [ # # ]: 0 : && total_mb < this->max_mbs) // we've cleared enough to meet threshold
1673 : : break; // stop clearing
1674 : :
1675 [ - + ]: 208 : auto j = entries.find(i.second);
1676 [ - + ]: 208 : if (j == entries.end())
1677 : 0 : continue; // should not happen
1678 : :
1679 [ + + ]: 208 : if (cleaned == 0)
1680 : 32 : cleaned_score_min = i.first;
1681 : 208 : cleaned++;
1682 : 208 : cleaned_score_max = i.first;
1683 : :
1684 [ + + ]: 208 : if (verbose > 3)
1685 [ + - + - ]: 534 : obatched(clog) << "fdcache evicted score=" << i.first
1686 [ + - + - ]: 356 : << " a=" << i.second.first << " b=" << i.second.second
1687 [ + - + - : 534 : << " fd=" << j->second.fd << " mb=" << j->second.fd_size_mb
+ - + - +
- + - ]
1688 [ + - + - : 178 : << " rq=" << j->second.request_count << " lat=" << j->second.latency
+ - + - ]
1689 [ + - + - : 178 : << " fr=" << (now - j->second.freshness)
+ - ]
1690 : 178 : << endl;
1691 [ - + ]: 208 : if (metrics_p)
1692 [ # # # # : 0 : inc_metric("fdcache_op_count","op","evict");
# # # # #
# # # # #
# # ]
1693 : :
1694 : 208 : total_mb -= j->second.fd_size_mb;
1695 : 208 : unlink (j->second.fd.c_str());
1696 : 208 : entries.erase(j);
1697 : : }
1698 : :
1699 [ + + ]: 148 : if (metrics_p)
1700 [ + - + - : 148 : inc_metric("fdcache_op_count","op","evict_cycle");
+ - + - -
+ - + - -
- - ]
1701 : :
1702 [ + - + + ]: 148 : if (verbose > 1 && cleaned > 0)
1703 : : {
1704 [ + - + - : 96 : obatched(clog) << "fdcache evicted num=" << cleaned << " of=" << entries_original
+ - + - ]
1705 [ + - + - : 32 : << " min=" << cleaned_score_min << " max=" << cleaned_score_max
+ - + - +
- ]
1706 : 32 : << endl;
1707 : : }
1708 : :
1709 [ + + + - ]: 148 : if (metrics_p) set_metrics();
1710 : 148 : }
1711 : :
1712 : :
1713 : 74 : ~libarchive_fdcache()
1714 : : {
1715 : : // unlink any fdcache entries in $TMPDIR
1716 : : // don't update metrics; those globals may be already destroyed
1717 : 74 : limit(0, false);
1718 : 74 : }
1719 : : };
1720 : : static libarchive_fdcache fdcache;
1721 : :
1722 : : /* Search ELF_FD for an ELF/DWARF section with name SECTION.
1723 : : If found copy the section to a temporary file and return
1724 : : its file descriptor, otherwise return -1.
1725 : :
1726 : : The temporary file's mtime will be set to PARENT_MTIME.
1727 : : B_SOURCE should be a description of the parent file suitable
1728 : : for printing to the log. */
1729 : :
1730 : : static int
1731 : 12 : extract_section (int elf_fd, int64_t parent_mtime,
1732 : : const string& b_source, const string& section,
1733 : : const timespec& extract_begin)
1734 : : {
1735 : : /* Search the fdcache. */
1736 : 12 : struct stat fs;
1737 : 12 : int fd = fdcache.lookup (b_source, section);
1738 [ - + ]: 12 : if (fd >= 0)
1739 : : {
1740 [ # # ]: 0 : if (fstat (fd, &fs) != 0)
1741 : : {
1742 [ # # ]: 0 : if (verbose)
1743 [ # # ]: 0 : obatched (clog) << "cannot fstate fdcache "
1744 [ # # # # : 0 : << b_source << " " << section << endl;
# # ]
1745 : 0 : close (fd);
1746 : 0 : return -1;
1747 : : }
1748 [ # # ]: 0 : if ((int64_t) fs.st_mtime != parent_mtime)
1749 : : {
1750 [ # # ]: 0 : if (verbose)
1751 [ # # ]: 0 : obatched(clog) << "mtime mismatch for "
1752 [ # # # # : 0 : << b_source << " " << section << endl;
# # ]
1753 : 0 : close (fd);
1754 : 0 : return -1;
1755 : : }
1756 : : /* Success. */
1757 : : return fd;
1758 : : }
1759 : :
1760 : 12 : Elf *elf = elf_begin (elf_fd, ELF_C_READ_MMAP_PRIVATE, NULL);
1761 [ - + ]: 12 : if (elf == NULL)
1762 : : return -1;
1763 : :
1764 : : /* Try to find the section and copy the contents into a separate file. */
1765 : 12 : try
1766 : : {
1767 : 12 : size_t shstrndx;
1768 [ + - ]: 12 : int rc = elf_getshdrstrndx (elf, &shstrndx);
1769 [ - + ]: 12 : if (rc < 0)
1770 [ # # # # ]: 0 : throw elfutils_exception (rc, "getshdrstrndx");
1771 : :
1772 : : Elf_Scn *scn = NULL;
1773 : 432 : while (true)
1774 : : {
1775 [ + - ]: 222 : scn = elf_nextscn (elf, scn);
1776 [ + - ]: 222 : if (scn == NULL)
1777 : : break;
1778 : 222 : GElf_Shdr shdr_storage;
1779 [ + - ]: 222 : GElf_Shdr *shdr = gelf_getshdr (scn, &shdr_storage);
1780 [ + - ]: 222 : if (shdr == NULL)
1781 : : break;
1782 : :
1783 [ + - ]: 222 : const char *scn_name = elf_strptr (elf, shstrndx, shdr->sh_name);
1784 [ + - ]: 222 : if (scn_name == NULL)
1785 : : break;
1786 [ + + ]: 222 : if (scn_name == section)
1787 : : {
1788 : 12 : Elf_Data *data = NULL;
1789 : :
1790 : : /* We found the desired section. */
1791 [ + - ]: 12 : data = elf_rawdata (scn, NULL);
1792 [ - + ]: 12 : if (data == NULL)
1793 [ # # # # : 0 : throw elfutils_exception (elf_errno (), "elfraw_data");
# # ]
1794 [ + + ]: 12 : if (data->d_buf == NULL)
1795 : : {
1796 [ + - + - ]: 8 : obatched(clog) << "section " << section
1797 [ + - + - ]: 4 : << " is empty" << endl;
1798 : 4 : break;
1799 : : }
1800 : :
1801 : : /* Create temporary file containing the section. */
1802 : 8 : char *tmppath = NULL;
1803 : 8 : rc = asprintf (&tmppath, "%s/debuginfod-section.XXXXXX", tmpdir.c_str());
1804 [ - + ]: 8 : if (rc < 0)
1805 [ # # # # ]: 0 : throw libc_exception (ENOMEM, "cannot allocate tmppath");
1806 : 8 : defer_dtor<void*,void> tmmpath_freer (tmppath, free);
1807 [ + - ]: 8 : fd = mkstemp (tmppath);
1808 [ - + ]: 8 : if (fd < 0)
1809 [ # # # # ]: 0 : throw libc_exception (errno, "cannot create temporary file");
1810 : :
1811 [ + - ]: 8 : ssize_t res = write_retry (fd, data->d_buf, data->d_size);
1812 [ + - - + ]: 8 : if (res < 0 || (size_t) res != data->d_size) {
1813 [ # # ]: 0 : close (fd);
1814 : 0 : unlink (tmppath);
1815 [ # # # # ]: 0 : throw libc_exception (errno, "cannot write to temporary file");
1816 : : }
1817 : :
1818 : : /* Set mtime to be the same as the parent file's mtime. */
1819 : 8 : struct timespec tvs[2];
1820 [ - + ]: 8 : if (fstat (elf_fd, &fs) != 0) {
1821 [ # # ]: 0 : close (fd);
1822 : 0 : unlink (tmppath);
1823 [ # # # # ]: 0 : throw libc_exception (errno, "cannot fstat file");
1824 : : }
1825 : :
1826 : 8 : tvs[0].tv_sec = 0;
1827 : 8 : tvs[0].tv_nsec = UTIME_OMIT;
1828 : 8 : tvs[1] = fs.st_mtim;
1829 : 8 : (void) futimens (fd, tvs);
1830 : :
1831 : 8 : struct timespec extract_end;
1832 : 8 : clock_gettime (CLOCK_MONOTONIC, &extract_end);
1833 : 8 : double extract_time = (extract_end.tv_sec - extract_begin.tv_sec)
1834 : 8 : + (extract_end.tv_nsec - extract_begin.tv_nsec)/1.e9;
1835 : :
1836 : : /* Add to fdcache. */
1837 [ + - + - ]: 8 : fdcache.intern (b_source, section, tmppath, data->d_size, true, extract_time);
1838 : 8 : break;
1839 : 12 : }
1840 : 210 : }
1841 : : }
1842 [ - - ]: 0 : catch (const reportable_exception &e)
1843 : : {
1844 [ - - ]: 0 : e.report (clog);
1845 [ - - ]: 0 : close (fd);
1846 : 0 : fd = -1;
1847 : 0 : }
1848 : :
1849 : 12 : elf_end (elf);
1850 : : return fd;
1851 : : }
1852 : :
1853 : : static struct MHD_Response*
1854 : 595 : handle_buildid_f_match (bool internal_req_t,
1855 : : int64_t b_mtime,
1856 : : const string& b_source0,
1857 : : const string& section,
1858 : : int *result_fd)
1859 : : {
1860 : 595 : (void) internal_req_t; // ignored
1861 : :
1862 : 595 : struct timespec extract_begin;
1863 : 595 : clock_gettime (CLOCK_MONOTONIC, &extract_begin);
1864 : :
1865 : 595 : int fd = open(b_source0.c_str(), O_RDONLY);
1866 [ - + ]: 595 : if (fd < 0)
1867 [ # # # # : 0 : throw libc_exception (errno, string("open ") + b_source0);
# # # # ]
1868 : :
1869 : : // NB: use manual close(2) in error case instead of defer_dtor, because
1870 : : // in the normal case, we want to hand the fd over to libmicrohttpd for
1871 : : // file transfer.
1872 : :
1873 : 595 : struct stat s;
1874 : 595 : int rc = fstat(fd, &s);
1875 [ - + ]: 595 : if (rc < 0)
1876 : : {
1877 : 0 : close(fd);
1878 [ # # # # : 0 : throw libc_exception (errno, string("fstat ") + b_source0);
# # # # ]
1879 : : }
1880 : :
1881 [ - + ]: 595 : if ((int64_t) s.st_mtime != b_mtime)
1882 : : {
1883 [ # # ]: 0 : if (verbose)
1884 [ # # # # ]: 0 : obatched(clog) << "mtime mismatch for " << b_source0 << endl;
1885 : 0 : close(fd);
1886 : 0 : return 0;
1887 : : }
1888 : :
1889 [ + + ]: 595 : if (!section.empty ())
1890 : : {
1891 : 6 : int scn_fd = extract_section (fd, s.st_mtime, b_source0, section, extract_begin);
1892 : 6 : close (fd);
1893 : :
1894 [ + + ]: 6 : if (scn_fd >= 0)
1895 : 4 : fd = scn_fd;
1896 : : else
1897 : : {
1898 [ + - ]: 2 : if (verbose)
1899 [ + - ]: 6 : obatched (clog) << "cannot find section " << section
1900 [ + - + - : 2 : << " for " << b_source0 << endl;
+ - ]
1901 : 2 : return 0;
1902 : : }
1903 : :
1904 : 4 : rc = fstat(fd, &s);
1905 [ - + ]: 4 : if (rc < 0)
1906 : : {
1907 : 0 : close (fd);
1908 [ # # # # : 0 : throw libc_exception (errno, string ("fstat ") + b_source0
# # # # #
# # # #
# ]
1909 [ # # # # : 0 : + string (" ") + section);
# # # # #
# ]
1910 : : }
1911 : : }
1912 : :
1913 : 593 : struct MHD_Response* r = MHD_create_response_from_fd ((uint64_t) s.st_size, fd);
1914 [ + - + - : 1186 : inc_metric ("http_responses_total","result","file");
+ - - + -
+ - - -
- ]
1915 [ - + ]: 593 : if (r == 0)
1916 : : {
1917 [ # # ]: 0 : if (verbose)
1918 [ # # ]: 0 : obatched(clog) << "cannot create fd-response for " << b_source0
1919 [ # # # # : 0 : << " section=" << section << endl;
# # ]
1920 : 0 : close(fd);
1921 : : }
1922 : : else
1923 : : {
1924 : 593 : add_mhd_response_header (r, "Content-Type", "application/octet-stream");
1925 [ + - ]: 593 : add_mhd_response_header (r, "X-DEBUGINFOD-SIZE",
1926 : 593 : to_string(s.st_size).c_str());
1927 : 593 : add_mhd_response_header (r, "X-DEBUGINFOD-FILE", b_source0.c_str());
1928 : 593 : add_mhd_last_modified (r, s.st_mtime);
1929 [ + - ]: 593 : if (verbose > 1)
1930 [ + - + - : 1186 : obatched(clog) << "serving file " << b_source0 << " section=" << section << endl;
+ - + - ]
1931 : : /* libmicrohttpd will close it. */
1932 [ - + ]: 593 : if (result_fd)
1933 : 593 : *result_fd = fd;
1934 : : }
1935 : :
1936 : : return r;
1937 : : }
1938 : :
1939 : : // For security/portability reasons, many distro-package archives have
1940 : : // a "./" in front of path names; others have nothing, others have
1941 : : // "/". Canonicalize them all to a single leading "/", with the
1942 : : // assumption that this matches the dwarf-derived file names too.
1943 : 974 : string canonicalized_archive_entry_pathname(struct archive_entry *e)
1944 : : {
1945 : 974 : string fn = archive_entry_pathname(e);
1946 [ - + ]: 974 : if (fn.size() == 0)
1947 : 0 : return fn;
1948 [ - + ]: 974 : if (fn[0] == '/')
1949 : 0 : return fn;
1950 [ + + ]: 974 : if (fn[0] == '.')
1951 [ + - ]: 776 : return fn.substr(1);
1952 : : else
1953 [ + - + - : 396 : return string("/")+fn;
- - ]
1954 : 974 : }
1955 : :
1956 : :
1957 : :
1958 : : static struct MHD_Response*
1959 : 786 : handle_buildid_r_match (bool internal_req_p,
1960 : : int64_t b_mtime,
1961 : : const string& b_source0,
1962 : : const string& b_source1,
1963 : : const string& section,
1964 : : int *result_fd)
1965 : : {
1966 : 786 : struct timespec extract_begin;
1967 : 786 : clock_gettime (CLOCK_MONOTONIC, &extract_begin);
1968 : :
1969 : 786 : struct stat fs;
1970 : 786 : int rc = stat (b_source0.c_str(), &fs);
1971 [ + + ]: 786 : if (rc != 0)
1972 [ + - + - : 116 : throw libc_exception (errno, string("stat ") + b_source0);
+ - - + ]
1973 : :
1974 [ - + ]: 728 : if ((int64_t) fs.st_mtime != b_mtime)
1975 : : {
1976 [ # # ]: 0 : if (verbose)
1977 [ # # # # ]: 0 : obatched(clog) << "mtime mismatch for " << b_source0 << endl;
1978 : 0 : return 0;
1979 : : }
1980 : :
1981 : : // Extract the IMA per-file signature (if it exists)
1982 : 728 : string ima_sig = "";
1983 : : #ifdef ENABLE_IMA_VERIFICATION
1984 : : do
1985 : : {
1986 : : FD_t rpm_fd;
1987 : : if(!(rpm_fd = Fopen(b_source0.c_str(), "r.ufdio"))) // read, uncompressed, rpm/rpmio.h
1988 : : {
1989 : : if (verbose) obatched(clog) << "There was an error while opening " << b_source0 << endl;
1990 : : break; // Exit IMA extraction
1991 : : }
1992 : :
1993 : : Header rpm_hdr;
1994 : : if(RPMRC_FAIL == rpmReadPackageFile(NULL, rpm_fd, b_source0.c_str(), &rpm_hdr))
1995 : : {
1996 : : if (verbose) obatched(clog) << "There was an error while reading the header of " << b_source0 << endl;
1997 : : Fclose(rpm_fd);
1998 : : break; // Exit IMA extraction
1999 : : }
2000 : :
2001 : : // Fill sig_tag_data with an alloc'd copy of the array of IMA signatures (if they exist)
2002 : : struct rpmtd_s sig_tag_data;
2003 : : rpmtdReset(&sig_tag_data);
2004 : : do{ /* A do-while so we can break out of the koji sigcache checking on failure */
2005 : : if(requires_koji_sigcache_mapping)
2006 : : {
2007 : : /* NB: Koji builds result in a directory structure like the following
2008 : : - PACKAGE/VERSION/RELEASE
2009 : : - ARCH1
2010 : : - foo.rpm // The rpm known by debuginfod
2011 : : - ...
2012 : : - ARCHN
2013 : : - data
2014 : : - signed // Periodically purged (and not scanned by debuginfod)
2015 : : - sigcache
2016 : : - ARCH1
2017 : : - foo.rpm.sig // An empty rpm header
2018 : : - ...
2019 : : - ARCHN
2020 : : - PACKAGE_KEYID1
2021 : : - ARCH1
2022 : : - foo.rpm.sig // The header of the signed rpm. This is the file we need to extract the IMA signatures
2023 : : - ...
2024 : : - ARCHN
2025 : : - ...
2026 : : - PACKAGE_KEYIDn
2027 : :
2028 : : We therefore need to do a mapping:
2029 : :
2030 : : P/V/R/A/N-V-R.A.rpm ->
2031 : : P/V/R/data/sigcache/KEYID/A/N-V-R.A.rpm.sig
2032 : :
2033 : : There are 2 key insights here
2034 : :
2035 : : 1. We need to go 2 directories down from sigcache to get to the
2036 : : rpm header. So to distinguish ARCH1/foo.rpm.sig and
2037 : : PACKAGE_KEYID1/ARCH1/foo.rpm.sig we can look 2 directories down
2038 : :
2039 : : 2. It's safe to assume that the user will have all of the
2040 : : required verification certs. So we can pick from any of the
2041 : : PACKAGE_KEYID* directories. For simplicity we choose first we
2042 : : match against
2043 : :
2044 : : See: https://pagure.io/koji/issue/3670
2045 : : */
2046 : :
2047 : : // Do the mapping from b_source0 to the koji path for the signed rpm header
2048 : : string signed_rpm_path = b_source0;
2049 : : size_t insert_pos = string::npos;
2050 : : for(int i = 0; i < 2; i++) insert_pos = signed_rpm_path.rfind("/", insert_pos) - 1;
2051 : : string globbed_path = signed_rpm_path.insert(insert_pos + 1, "/data/sigcache/*").append(".sig"); // The globbed path we're seeking
2052 : : glob_t pglob;
2053 : : int grc;
2054 : : if(0 != (grc = glob(globbed_path.c_str(), GLOB_NOSORT, NULL, &pglob)))
2055 : : {
2056 : : // Break out, but only report real errors
2057 : : if (verbose && grc != GLOB_NOMATCH) obatched(clog) << "There was an error (" << strerror(errno) << ") globbing " << globbed_path << endl;
2058 : : break; // Exit koji sigcache check
2059 : : }
2060 : : signed_rpm_path = pglob.gl_pathv[0]; // See insight 2 above
2061 : : globfree(&pglob);
2062 : :
2063 : : if (verbose > 2) obatched(clog) << "attempting IMA signature extraction from koji header " << signed_rpm_path << endl;
2064 : :
2065 : : FD_t sig_rpm_fd;
2066 : : if(NULL == (sig_rpm_fd = Fopen(signed_rpm_path.c_str(), "r")))
2067 : : {
2068 : : if (verbose) obatched(clog) << "There was an error while opening " << signed_rpm_path << endl;
2069 : : break; // Exit koji sigcache check
2070 : : }
2071 : :
2072 : : Header sig_hdr = headerRead(sig_rpm_fd, HEADER_MAGIC_YES /* Validate magic too */ );
2073 : : if (!sig_hdr || 1 != headerGet(sig_hdr, RPMSIGTAG_FILESIGNATURES, &sig_tag_data, HEADERGET_ALLOC))
2074 : : {
2075 : : if (verbose) obatched(clog) << "Unable to extract RPMSIGTAG_FILESIGNATURES from " << signed_rpm_path << endl;
2076 : : }
2077 : : headerFree(sig_hdr); // We can free here since sig_tag_data has an alloc'd copy of the data
2078 : : Fclose(sig_rpm_fd);
2079 : : }
2080 : : }while(false);
2081 : :
2082 : : if(0 == sig_tag_data.count)
2083 : : {
2084 : : // In the general case (or a fallback from the koji sigcache mapping not finding signatures)
2085 : : // we can just (try) extract the signatures from the rpm header
2086 : : if (1 != headerGet(rpm_hdr, RPMTAG_FILESIGNATURES, &sig_tag_data, HEADERGET_ALLOC))
2087 : : {
2088 : : if (verbose) obatched(clog) << "Unable to extract RPMTAG_FILESIGNATURES from " << b_source0 << endl;
2089 : : }
2090 : : }
2091 : : // Search the array for the signature coresponding to b_source1
2092 : : int idx = -1;
2093 : : char *sig = NULL;
2094 : : rpmfi hdr_fi = rpmfiNew(NULL, rpm_hdr, RPMTAG_BASENAMES, RPMFI_FLAGS_QUERY);
2095 : : do
2096 : : {
2097 : : sig = (char*)rpmtdNextString(&sig_tag_data);
2098 : : idx = rpmfiNext(hdr_fi);
2099 : : }
2100 : : while (idx != -1 && 0 != strcmp(b_source1.c_str(), rpmfiFN(hdr_fi)));
2101 : : rpmfiFree(hdr_fi);
2102 : :
2103 : : if(sig && 0 != strlen(sig) && idx != -1)
2104 : : {
2105 : : if (verbose > 2) obatched(clog) << "Found IMA signature for " << b_source1 << ":\n" << sig << endl;
2106 : : ima_sig = sig;
2107 : : inc_metric("http_responses_total","extra","ima-sigs-extracted");
2108 : : }
2109 : : else
2110 : : {
2111 : : if (verbose > 2) obatched(clog) << "Could not find IMA signature for " << b_source1 << endl;
2112 : : }
2113 : :
2114 : : rpmtdFreeData (&sig_tag_data);
2115 : : headerFree(rpm_hdr);
2116 : : Fclose(rpm_fd);
2117 : : } while(false);
2118 : : #endif
2119 : :
2120 : : // check for a match in the fdcache first
2121 [ + - ]: 728 : int fd = fdcache.lookup(b_source0, b_source1);
2122 [ + + ]: 728 : while (fd >= 0) // got one!; NB: this is really an if() with a possible branch out to the end
2123 : : {
2124 : 576 : rc = fstat(fd, &fs);
2125 [ - + ]: 576 : if (rc < 0) // disappeared?
2126 : : {
2127 [ # # ]: 0 : if (verbose)
2128 [ # # # # : 0 : obatched(clog) << "cannot fstat fdcache " << b_source0 << endl;
# # ]
2129 [ # # ]: 0 : close(fd);
2130 [ # # ]: 0 : fdcache.clear(b_source0, b_source1);
2131 : : break; // branch out of if "loop", to try new libarchive fetch attempt
2132 : : }
2133 : :
2134 [ + + ]: 576 : if (!section.empty ())
2135 : : {
2136 [ + - + - ]: 2 : int scn_fd = extract_section (fd, fs.st_mtime,
2137 [ + - + - : 4 : b_source0 + ":" + b_source1,
- + ]
2138 : : section, extract_begin);
2139 [ + - ]: 2 : close (fd);
2140 [ - + ]: 2 : if (scn_fd >= 0)
2141 : 0 : fd = scn_fd;
2142 : : else
2143 : : {
2144 [ + - ]: 2 : if (verbose)
2145 [ + - + - ]: 6 : obatched (clog) << "cannot find section " << section
2146 : : << " for archive " << b_source0
2147 [ + - + - : 2 : << " file " << b_source1 << endl;
+ - + - +
- ]
2148 : 2 : return 0;
2149 : : }
2150 : :
2151 : 0 : rc = fstat(fd, &fs);
2152 [ # # ]: 0 : if (rc < 0)
2153 : : {
2154 [ # # ]: 0 : close (fd);
2155 [ # # # # ]: 0 : throw libc_exception (errno,
2156 [ # # # # : 0 : string ("fstat archive ") + b_source0 + string (" file ") + b_source1
# # # # #
# # # # #
# # # # #
# # # #
# ]
2157 [ # # # # : 0 : + string (" section ") + section);
# # # # #
# ]
2158 : : }
2159 : : }
2160 : :
2161 [ + - ]: 574 : struct MHD_Response* r = MHD_create_response_from_fd (fs.st_size, fd);
2162 [ - + ]: 574 : if (r == 0)
2163 : : {
2164 [ # # ]: 0 : if (verbose)
2165 [ # # # # : 0 : obatched(clog) << "cannot create fd-response for " << b_source0 << endl;
# # ]
2166 [ # # ]: 0 : close(fd);
2167 : : break; // branch out of if "loop", to try new libarchive fetch attempt
2168 : : }
2169 : :
2170 [ + - + - : 1148 : inc_metric ("http_responses_total","result","archive fdcache");
+ - + - -
+ - + - -
- - ]
2171 : :
2172 [ + - ]: 574 : add_mhd_response_header (r, "Content-Type", "application/octet-stream");
2173 [ + - ]: 574 : add_mhd_response_header (r, "X-DEBUGINFOD-SIZE",
2174 [ + - ]: 574 : to_string(fs.st_size).c_str());
2175 [ + - ]: 574 : add_mhd_response_header (r, "X-DEBUGINFOD-ARCHIVE", b_source0.c_str());
2176 [ + - ]: 574 : add_mhd_response_header (r, "X-DEBUGINFOD-FILE", b_source1.c_str());
2177 [ - + - - ]: 574 : if(!ima_sig.empty()) add_mhd_response_header(r, "X-DEBUGINFOD-IMASIGNATURE", ima_sig.c_str());
2178 [ + - ]: 574 : add_mhd_last_modified (r, fs.st_mtime);
2179 [ + - ]: 574 : if (verbose > 1)
2180 [ + - + - ]: 1722 : obatched(clog) << "serving fdcache archive " << b_source0
2181 : : << " file " << b_source1
2182 : : << " section=" << section
2183 [ + - + - : 574 : << " IMA signature=" << ima_sig << endl;
+ - + - +
- + - +
- ]
2184 : : /* libmicrohttpd will close it. */
2185 [ + - ]: 574 : if (result_fd)
2186 : 574 : *result_fd = fd;
2187 : : return r;
2188 : : // NB: see, we never go around the 'loop' more than once
2189 : : }
2190 : :
2191 : : // no match ... grumble, must process the archive
2192 [ + - - - ]: 152 : string archive_decoder = "/dev/null";
2193 [ + - - - ]: 152 : string archive_extension = "";
2194 [ + + ]: 342 : for (auto&& arch : scan_archives)
2195 [ + + ]: 190 : if (string_endswith(b_source0, arch.first))
2196 : : {
2197 [ + - ]: 152 : archive_extension = arch.first;
2198 [ + - ]: 342 : archive_decoder = arch.second;
2199 : : }
2200 : 152 : FILE* fp;
2201 : :
2202 : 152 : defer_dtor<FILE*,int>::dtor_fn dfn;
2203 [ + + ]: 152 : if (archive_decoder != "cat")
2204 : : {
2205 [ + - + - : 48 : string popen_cmd = archive_decoder + " " + shell_escape(b_source0);
+ - - + -
- - - ]
2206 [ + - ]: 24 : fp = popen (popen_cmd.c_str(), "r"); // "e" O_CLOEXEC?
2207 : 24 : dfn = pclose;
2208 [ - + ]: 24 : if (fp == NULL)
2209 [ # # # # : 0 : throw libc_exception (errno, string("popen ") + popen_cmd);
# # # # ]
2210 : 24 : }
2211 : : else
2212 : : {
2213 [ + - ]: 128 : fp = fopen (b_source0.c_str(), "r");
2214 : 128 : dfn = fclose;
2215 [ - + ]: 128 : if (fp == NULL)
2216 [ # # # # : 0 : throw libc_exception (errno, string("fopen ") + b_source0);
# # # # ]
2217 : : }
2218 : 152 : defer_dtor<FILE*,int> fp_closer (fp, dfn);
2219 : :
2220 : 152 : struct archive *a;
2221 [ + - ]: 152 : a = archive_read_new();
2222 [ - + ]: 152 : if (a == NULL)
2223 [ # # # # ]: 0 : throw archive_exception("cannot create archive reader");
2224 : 152 : defer_dtor<struct archive*,int> archive_closer (a, archive_read_free);
2225 : :
2226 [ + - ]: 152 : rc = archive_read_support_format_all(a);
2227 [ - + ]: 152 : if (rc != ARCHIVE_OK)
2228 [ # # # # ]: 0 : throw archive_exception(a, "cannot select all format");
2229 [ + - ]: 152 : rc = archive_read_support_filter_all(a);
2230 [ - + ]: 152 : if (rc != ARCHIVE_OK)
2231 [ # # # # ]: 0 : throw archive_exception(a, "cannot select all filters");
2232 : :
2233 [ + - ]: 152 : rc = archive_read_open_FILE (a, fp);
2234 [ - + ]: 152 : if (rc != ARCHIVE_OK)
2235 : : {
2236 [ # # # # : 0 : obatched(clog) << "cannot open archive from pipe " << b_source0 << endl;
# # ]
2237 [ # # # # ]: 0 : throw archive_exception(a, "cannot open archive from pipe");
2238 : : }
2239 : :
2240 : : // archive traversal is in three stages, no, four stages:
2241 : : // 1) skip entries whose names do not match the requested one
2242 : : // 2) extract the matching entry name (set r = result)
2243 : : // 3) extract some number of prefetched entries (just into fdcache)
2244 : : // 4) abort any further processing
2245 : 152 : struct MHD_Response* r = 0; // will set in stage 2
2246 [ + + ]: 152 : unsigned prefetch_count =
2247 : : internal_req_p ? 0 : fdcache_prefetch; // will decrement in stage 3
2248 : :
2249 [ + + ]: 1536 : while(r == 0 || prefetch_count > 0) // stage 1, 2, or 3
2250 : : {
2251 [ + - ]: 1516 : if (interrupted)
2252 : : break;
2253 : :
2254 : 1516 : struct archive_entry *e;
2255 [ + - ]: 1516 : rc = archive_read_next_header (a, &e);
2256 [ + + ]: 1516 : if (rc != ARCHIVE_OK)
2257 : : break;
2258 : :
2259 [ + - + + ]: 1384 : if (! S_ISREG(archive_entry_mode (e))) // skip non-files completely
2260 : 1084 : continue;
2261 : :
2262 [ + - ]: 300 : string fn = canonicalized_archive_entry_pathname (e);
2263 [ + + + + ]: 300 : if ((r == 0) && (fn != b_source1)) // stage 1
2264 : 76 : continue;
2265 : :
2266 [ + - + + ]: 224 : if (fdcache.probe (b_source0, fn) && // skip if already interned
2267 [ - + ]: 24 : fn != b_source1) // but only if we'd just be prefetching, PR29474
2268 : 24 : continue;
2269 : :
2270 : : // extract this file to a temporary file
2271 : 200 : char* tmppath = NULL;
2272 : 200 : rc = asprintf (&tmppath, "%s/debuginfod-fdcache.XXXXXX", tmpdir.c_str());
2273 [ - + ]: 200 : if (rc < 0)
2274 [ # # # # ]: 0 : throw libc_exception (ENOMEM, "cannot allocate tmppath");
2275 : 200 : defer_dtor<void*,void> tmmpath_freer (tmppath, free);
2276 [ + - ]: 200 : fd = mkstemp (tmppath);
2277 [ - + ]: 200 : if (fd < 0)
2278 [ # # # # ]: 0 : throw libc_exception (errno, "cannot create temporary file");
2279 : : // NB: don't unlink (tmppath), as fdcache will take charge of it.
2280 : :
2281 : : // NB: this can take many uninterruptible seconds for a huge file
2282 [ + - ]: 200 : rc = archive_read_data_into_fd (a, fd);
2283 [ - + ]: 200 : if (rc != ARCHIVE_OK) // e.g. ENOSPC!
2284 : : {
2285 [ # # ]: 0 : close (fd);
2286 : 0 : unlink (tmppath);
2287 [ # # # # ]: 0 : throw archive_exception(a, "cannot extract file");
2288 : : }
2289 : :
2290 : : // Set the mtime so the fdcache file mtimes, even prefetched ones,
2291 : : // propagate to future webapi clients.
2292 : 200 : struct timespec tvs[2];
2293 : 200 : tvs[0].tv_sec = 0;
2294 : 200 : tvs[0].tv_nsec = UTIME_OMIT;
2295 [ + - ]: 200 : tvs[1].tv_sec = archive_entry_mtime(e);
2296 [ + - ]: 200 : tvs[1].tv_nsec = archive_entry_mtime_nsec(e);
2297 : 200 : (void) futimens (fd, tvs); /* best effort */
2298 : :
2299 : 200 : struct timespec extract_end;
2300 : 200 : clock_gettime (CLOCK_MONOTONIC, &extract_end);
2301 : 200 : double extract_time = (extract_end.tv_sec - extract_begin.tv_sec)
2302 : 200 : + (extract_end.tv_nsec - extract_begin.tv_nsec)/1.e9;
2303 : :
2304 [ + + ]: 200 : if (r != 0) // stage 3
2305 : : {
2306 : : // NB: now we know we have a complete reusable file; make fdcache
2307 : : // responsible for unlinking it later.
2308 [ + - + - : 48 : fdcache.intern(b_source0, fn,
+ - ]
2309 : : tmppath, archive_entry_size(e),
2310 : : false, extract_time); // prefetched ones go to the prefetch cache
2311 : 48 : prefetch_count --;
2312 [ + - ]: 48 : close (fd); // we're not saving this fd to make a mhd-response from!
2313 : 48 : continue;
2314 : : }
2315 : :
2316 : : // NB: now we know we have a complete reusable file; make fdcache
2317 : : // responsible for unlinking it later.
2318 [ + - + - : 152 : fdcache.intern(b_source0, b_source1,
+ - ]
2319 : : tmppath, archive_entry_size(e),
2320 : : true, extract_time); // requested ones go to the front of the line
2321 : :
2322 [ + + ]: 152 : if (!section.empty ())
2323 : : {
2324 [ + - + - ]: 4 : int scn_fd = extract_section (fd, b_mtime,
2325 [ + - + - : 8 : b_source0 + ":" + b_source1,
- + ]
2326 : : section, extract_begin);
2327 [ + - ]: 4 : close (fd);
2328 [ + - ]: 4 : if (scn_fd >= 0)
2329 : 4 : fd = scn_fd;
2330 : : else
2331 : : {
2332 [ # # ]: 0 : if (verbose)
2333 [ # # # # ]: 0 : obatched (clog) << "cannot find section " << section
2334 : : << " for archive " << b_source0
2335 [ # # # # : 0 : << " file " << b_source1 << endl;
# # # # #
# ]
2336 : 0 : return 0;
2337 : : }
2338 : :
2339 : 4 : rc = fstat(fd, &fs);
2340 [ - + ]: 4 : if (rc < 0)
2341 : : {
2342 [ # # ]: 0 : close (fd);
2343 [ # # # # ]: 0 : throw libc_exception (errno,
2344 [ # # # # : 0 : string ("fstat ") + b_source0 + string (" ") + section);
# # # # #
# # # # #
# # # # #
# ]
2345 : : }
2346 [ + - ]: 4 : r = MHD_create_response_from_fd (fs.st_size, fd);
2347 : : }
2348 : : else
2349 [ + - + - ]: 148 : r = MHD_create_response_from_fd (archive_entry_size(e), fd);
2350 : :
2351 [ + - + - : 304 : inc_metric ("http_responses_total","result",archive_extension + " archive");
+ - + - -
+ + + - -
- - ]
2352 [ - + ]: 152 : if (r == 0)
2353 : : {
2354 [ # # ]: 0 : if (verbose)
2355 [ # # # # : 0 : obatched(clog) << "cannot create fd-response for " << b_source0 << endl;
# # ]
2356 [ # # ]: 0 : close(fd);
2357 : 0 : break; // assume no chance of better luck around another iteration; no other copies of same file
2358 : : }
2359 : : else
2360 : : {
2361 [ + - ]: 152 : add_mhd_response_header (r, "Content-Type",
2362 : : "application/octet-stream");
2363 [ + - ]: 152 : add_mhd_response_header (r, "X-DEBUGINFOD-SIZE",
2364 [ + - + - ]: 152 : to_string(archive_entry_size(e)).c_str());
2365 [ + - ]: 152 : add_mhd_response_header (r, "X-DEBUGINFOD-ARCHIVE", b_source0.c_str());
2366 [ + - ]: 152 : add_mhd_response_header (r, "X-DEBUGINFOD-FILE", b_source1.c_str());
2367 [ - + - - ]: 152 : if(!ima_sig.empty()) add_mhd_response_header(r, "X-DEBUGINFOD-IMASIGNATURE", ima_sig.c_str());
2368 [ + - + - ]: 152 : add_mhd_last_modified (r, archive_entry_mtime(e));
2369 [ + - ]: 152 : if (verbose > 1)
2370 [ + - + - ]: 456 : obatched(clog) << "serving archive " << b_source0
2371 : : << " file " << b_source1
2372 : : << " section=" << section
2373 [ + - + - : 152 : << " IMA signature=" << ima_sig << endl;
+ - + - +
- + - +
- ]
2374 : : /* libmicrohttpd will close it. */
2375 [ + - ]: 152 : if (result_fd)
2376 : 152 : *result_fd = fd;
2377 : 152 : continue;
2378 : : }
2379 [ - - - - : 1816 : }
+ + ]
2380 : :
2381 : : // XXX: rpm/file not found: delete this R entry?
2382 : : return r;
2383 [ - + + + : 1608 : }
- + ]
2384 : :
2385 : :
2386 : : static struct MHD_Response*
2387 : 1381 : handle_buildid_match (bool internal_req_p,
2388 : : int64_t b_mtime,
2389 : : const string& b_stype,
2390 : : const string& b_source0,
2391 : : const string& b_source1,
2392 : : const string& section,
2393 : : int *result_fd)
2394 : : {
2395 : 1381 : try
2396 : : {
2397 [ + + ]: 1381 : if (b_stype == "F")
2398 [ + - ]: 595 : return handle_buildid_f_match(internal_req_p, b_mtime, b_source0,
2399 : : section, result_fd);
2400 [ + - ]: 786 : else if (b_stype == "R")
2401 [ + + ]: 786 : return handle_buildid_r_match(internal_req_p, b_mtime, b_source0,
2402 : : b_source1, section, result_fd);
2403 : : }
2404 [ - + ]: 58 : catch (const reportable_exception &e)
2405 : : {
2406 [ + - ]: 58 : e.report(clog);
2407 : : // Report but swallow libc etc. errors here; let the caller
2408 : : // iterate to other matches of the content.
2409 : 58 : }
2410 : :
2411 : : return 0;
2412 : : }
2413 : :
2414 : :
2415 : : static int
2416 : 3236 : debuginfod_find_progress (debuginfod_client *, long a, long b)
2417 : : {
2418 [ - + ]: 3236 : if (verbose > 4)
2419 [ # # # # : 0 : obatched(clog) << "federated debuginfod progress=" << a << "/" << b << endl;
# # # # ]
2420 : :
2421 : 3236 : return interrupted;
2422 : : }
2423 : :
2424 : :
2425 : : // a little lru pool of debuginfod_client*s for reuse between query threads
2426 : :
2427 : : mutex dc_pool_lock;
2428 : : deque<debuginfod_client*> dc_pool;
2429 : :
2430 : 620 : debuginfod_client* debuginfod_pool_begin()
2431 : : {
2432 : 620 : unique_lock<mutex> lock(dc_pool_lock);
2433 [ + + ]: 620 : if (dc_pool.size() > 0)
2434 : : {
2435 [ + - + - : 1188 : inc_metric("dc_pool_op_count","op","begin-reuse");
+ - + - -
+ - + - -
- - ]
2436 : 594 : debuginfod_client *c = dc_pool.front();
2437 : 594 : dc_pool.pop_front();
2438 : 594 : return c;
2439 : : }
2440 [ + - + - : 52 : inc_metric("dc_pool_op_count","op","begin-new");
+ - + - -
+ - + - -
- - ]
2441 [ + - ]: 26 : return debuginfod_begin();
2442 : 620 : }
2443 : :
2444 : :
2445 : 146 : void debuginfod_pool_groom()
2446 : : {
2447 : 146 : unique_lock<mutex> lock(dc_pool_lock);
2448 [ + + ]: 172 : while (dc_pool.size() > 0)
2449 : : {
2450 [ + - + - : 52 : inc_metric("dc_pool_op_count","op","end");
+ - + - -
+ - + - -
- - ]
2451 [ + - ]: 26 : debuginfod_end(dc_pool.front());
2452 : 26 : dc_pool.pop_front();
2453 : : }
2454 : 146 : }
2455 : :
2456 : :
2457 : 620 : void debuginfod_pool_end(debuginfod_client* c)
2458 : : {
2459 : 620 : unique_lock<mutex> lock(dc_pool_lock);
2460 [ + - + - : 1240 : inc_metric("dc_pool_op_count","op","end-save");
+ - + - -
+ - + - -
- - ]
2461 [ + - ]: 620 : dc_pool.push_front(c); // accelerate reuse, vs. push_back
2462 : 620 : }
2463 : :
2464 : :
2465 : : static struct MHD_Response*
2466 : 1949 : handle_buildid (MHD_Connection* conn,
2467 : : const string& buildid /* unsafe */,
2468 : : string& artifacttype /* unsafe, cleanse on exception/return */,
2469 : : const string& suffix /* unsafe */,
2470 : : int *result_fd)
2471 : : {
2472 : : // validate artifacttype
2473 : 1949 : string atype_code;
2474 [ + + + - ]: 1949 : if (artifacttype == "debuginfo") atype_code = "D";
2475 [ + + + - ]: 1211 : else if (artifacttype == "executable") atype_code = "E";
2476 [ + + + - ]: 589 : else if (artifacttype == "source") atype_code = "S";
2477 [ + + + - ]: 12 : else if (artifacttype == "section") atype_code = "I";
2478 : : else {
2479 [ + - ]: 4 : artifacttype = "invalid"; // PR28242 ensure http_resposes metrics don't propagate unclean user data
2480 [ + - + - ]: 12 : throw reportable_exception("invalid artifacttype");
2481 : : }
2482 : :
2483 [ + + ]: 1945 : if (conn != 0)
2484 [ + - + - : 4436 : inc_metric("http_requests_total", "type", artifacttype);
+ - - + -
- - + ]
2485 : :
2486 : 1945 : string section;
2487 [ + + ]: 1945 : if (atype_code == "I")
2488 : : {
2489 [ - + ]: 8 : if (suffix.size () < 2)
2490 [ # # # # ]: 0 : throw reportable_exception ("invalid section suffix");
2491 : :
2492 : : // Remove leading '/'
2493 [ + - - + ]: 8 : section = suffix.substr(1);
2494 : : }
2495 : :
2496 [ + + - + ]: 2522 : if (atype_code == "S" && suffix == "")
2497 [ # # # # ]: 0 : throw reportable_exception("invalid source suffix");
2498 : :
2499 : : // validate buildid
2500 [ + + ]: 1945 : if ((buildid.size() < 2) || // not empty
2501 [ + + + - : 3884 : (buildid.size() % 2) || // even number
+ - ]
2502 : 1939 : (buildid.find_first_not_of("0123456789abcdef") != string::npos)) // pure tasty lowercase hex
2503 [ + - - + ]: 12 : throw reportable_exception("invalid buildid");
2504 : :
2505 [ + - ]: 1939 : if (verbose > 1)
2506 [ + - + - ]: 5817 : obatched(clog) << "searching for buildid=" << buildid << " artifacttype=" << artifacttype
2507 [ + - + - : 1939 : << " suffix=" << suffix << endl;
+ - + - +
- ]
2508 : :
2509 : : // If invoked from the scanner threads, use the scanners' read-write
2510 : : // connection. Otherwise use the web query threads' read-only connection.
2511 [ + + ]: 1939 : sqlite3 *thisdb = (conn == 0) ? db : dbq;
2512 : :
2513 : 1939 : sqlite_ps *pp = 0;
2514 : :
2515 [ + + ]: 1939 : if (atype_code == "D")
2516 : : {
2517 [ - + ]: 738 : pp = new sqlite_ps (thisdb, "mhd-query-d",
2518 : : "select mtime, sourcetype, source0, source1 from " BUILDIDS "_query_d where buildid = ? "
2519 [ + - + - : 1476 : "order by mtime desc");
+ - + - +
- - - ]
2520 [ + - ]: 738 : pp->reset();
2521 [ + - ]: 738 : pp->bind(1, buildid);
2522 : : }
2523 [ + + ]: 1201 : else if (atype_code == "E")
2524 : : {
2525 [ - + ]: 616 : pp = new sqlite_ps (thisdb, "mhd-query-e",
2526 : : "select mtime, sourcetype, source0, source1 from " BUILDIDS "_query_e where buildid = ? "
2527 [ + - + - : 1232 : "order by mtime desc");
+ - + - +
- - - ]
2528 [ + - ]: 616 : pp->reset();
2529 [ + - ]: 616 : pp->bind(1, buildid);
2530 : : }
2531 [ + + ]: 585 : else if (atype_code == "S")
2532 : : {
2533 : : // PR25548
2534 : : // Incoming source queries may come in with either dwarf-level OR canonicalized paths.
2535 : : // We let the query pass with either one.
2536 : :
2537 [ - + ]: 577 : pp = new sqlite_ps (thisdb, "mhd-query-s",
2538 : : "select mtime, sourcetype, source0, source1 from " BUILDIDS "_query_s where buildid = ? and artifactsrc in (?,?) "
2539 [ + - + - : 1154 : "order by sharedprefix(source0,source0ref) desc, mtime desc");
+ - + - +
- - - ]
2540 [ + - ]: 577 : pp->reset();
2541 [ + - ]: 577 : pp->bind(1, buildid);
2542 : : // NB: we don't store the non-canonicalized path names any more, but old databases
2543 : : // might have them (and no canon ones), so we keep searching for both.
2544 [ + - ]: 577 : pp->bind(2, suffix);
2545 [ + - + - : 1776 : pp->bind(3, canon_pathname(suffix));
- + ]
2546 : : }
2547 [ + - ]: 8 : else if (atype_code == "I")
2548 : : {
2549 [ - + ]: 8 : pp = new sqlite_ps (thisdb, "mhd-query-i",
2550 : : "select mtime, sourcetype, source0, source1, 1 as debug_p from " BUILDIDS "_query_d where buildid = ? "
2551 : : "union all "
2552 : : "select mtime, sourcetype, source0, source1, 0 as debug_p from " BUILDIDS "_query_e where buildid = ? "
2553 [ + - + - : 16 : "order by debug_p desc, mtime desc");
+ - + - +
- - - ]
2554 [ + - ]: 8 : pp->reset();
2555 [ + - ]: 8 : pp->bind(1, buildid);
2556 [ + - ]: 8 : pp->bind(2, buildid);
2557 : : }
2558 : 1939 : unique_ptr<sqlite_ps> ps_closer(pp); // release pp if exception or return
2559 : :
2560 : 1939 : bool do_upstream_section_query = true;
2561 : :
2562 : : // consume all the rows
2563 : 2001 : while (1)
2564 : : {
2565 [ + - ]: 2001 : int rc = pp->step();
2566 [ + + ]: 2001 : if (rc == SQLITE_DONE) break;
2567 [ - + ]: 1381 : if (rc != SQLITE_ROW)
2568 [ # # # # ]: 0 : throw sqlite_exception(rc, "step");
2569 : :
2570 [ + - ]: 1381 : int64_t b_mtime = sqlite3_column_int64 (*pp, 0);
2571 [ + - - + : 1381 : string b_stype = string((const char*) sqlite3_column_text (*pp, 1) ?: ""); /* by DDL may not be NULL */
+ - ]
2572 [ + - - + : 1381 : string b_source0 = string((const char*) sqlite3_column_text (*pp, 2) ?: ""); /* may be NULL */
+ - - - ]
2573 [ + - + + : 1976 : string b_source1 = string((const char*) sqlite3_column_text (*pp, 3) ?: ""); /* may be NULL */
+ - - - ]
2574 : :
2575 [ + - ]: 1381 : if (verbose > 1)
2576 [ + - + - : 4143 : obatched(clog) << "found mtime=" << b_mtime << " stype=" << b_stype
- - ]
2577 [ + - + - : 1381 : << " source0=" << b_source0 << " source1=" << b_source1 << endl;
+ - + - +
- + - +
- ]
2578 : :
2579 : : // Try accessing the located match.
2580 : : // XXX: in case of multiple matches, attempt them in parallel?
2581 [ + - ]: 1381 : auto r = handle_buildid_match (conn ? false : true,
2582 : : b_mtime, b_stype, b_source0, b_source1,
2583 : : section, result_fd);
2584 [ + + ]: 1381 : if (r)
2585 [ + + ]: 1319 : return r;
2586 : :
2587 : : // If a debuginfo file matching BUILDID was found but didn't contain
2588 : : // the desired section, then the section should not exist. Don't
2589 : : // bother querying upstream servers.
2590 [ + + + - : 62 : if (!section.empty () && (sqlite3_column_int (*pp, 4) == 1))
- + ]
2591 : : {
2592 : 4 : struct stat st;
2593 : :
2594 : : // For "F" sourcetype, check if the debuginfo exists. For "R"
2595 : : // sourcetype, check if the debuginfo was interned into the fdcache.
2596 [ - + ]: 6 : if ((b_stype == "F" && (stat (b_source0.c_str (), &st) == 0))
2597 [ + + + - : 6 : || (b_stype == "R" && fdcache.probe (b_source0, b_source1)))
+ - - + ]
2598 : : do_upstream_section_query = false;
2599 : : }
2600 [ + - - + : 2762 : }
+ - - + ]
2601 [ + - ]: 620 : pp->reset();
2602 : :
2603 [ - + ]: 620 : if (!do_upstream_section_query)
2604 [ # # # # ]: 0 : throw reportable_exception(MHD_HTTP_NOT_FOUND, "not found");
2605 : :
2606 : : // We couldn't find it in the database. Last ditch effort
2607 : : // is to defer to other debuginfo servers.
2608 : :
2609 : 620 : int fd = -1;
2610 [ + - ]: 620 : debuginfod_client *client = debuginfod_pool_begin ();
2611 [ - + ]: 620 : if (client == NULL)
2612 [ # # # # ]: 0 : throw libc_exception(errno, "debuginfod client pool alloc");
2613 : 620 : defer_dtor<debuginfod_client*,void> client_closer (client, debuginfod_pool_end);
2614 : :
2615 [ + - ]: 620 : debuginfod_set_progressfn (client, & debuginfod_find_progress);
2616 : :
2617 [ + - ]: 620 : if (conn)
2618 : : {
2619 : : // Transcribe incoming User-Agent:
2620 [ + - - + : 620 : string ua = MHD_lookup_connection_value (conn, MHD_HEADER_KIND, "User-Agent") ?: "";
+ - ]
2621 [ + - + - : 624 : string ua_complete = string("User-Agent: ") + ua;
+ - ]
2622 [ + - ]: 620 : debuginfod_add_http_header (client, ua_complete.c_str());
2623 : :
2624 : : // Compute larger XFF:, for avoiding info loss during
2625 : : // federation, and for future cyclicity detection.
2626 [ + - + + : 1226 : string xff = MHD_lookup_connection_value (conn, MHD_HEADER_KIND, "X-Forwarded-For") ?: "";
+ - + - ]
2627 [ + + ]: 620 : if (xff != "")
2628 [ + - - + ]: 36 : xff += string(", "); // comma separated list
2629 : :
2630 : 620 : unsigned int xff_count = 0;
2631 [ + + ]: 860 : for (auto&& i : xff){
2632 [ + + ]: 240 : if (i == ',') xff_count++;
2633 : : }
2634 : :
2635 : : // if X-Forwarded-For: exceeds N hops,
2636 : : // do not delegate a local lookup miss to upstream debuginfods.
2637 [ + + ]: 620 : if (xff_count >= forwarded_ttl_limit)
2638 [ + - + - ]: 8 : throw reportable_exception(MHD_HTTP_NOT_FOUND, "not found, --forwared-ttl-limit reached \
2639 : 8 : and will not query the upstream servers");
2640 : :
2641 : : // Compute the client's numeric IP address only - so can't merge with conninfo()
2642 [ + - ]: 616 : const union MHD_ConnectionInfo *u = MHD_get_connection_info (conn,
2643 : : MHD_CONNECTION_INFO_CLIENT_ADDRESS);
2644 [ + - ]: 616 : struct sockaddr *so = u ? u->client_addr : 0;
2645 : 616 : char hostname[256] = ""; // RFC1035
2646 [ + - - + ]: 616 : if (so && so->sa_family == AF_INET) {
2647 [ # # ]: 0 : (void) getnameinfo (so, sizeof (struct sockaddr_in), hostname, sizeof (hostname), NULL, 0,
2648 : : NI_NUMERICHOST);
2649 [ + - ]: 616 : } else if (so && so->sa_family == AF_INET6) {
2650 : 616 : struct sockaddr_in6* addr6 = (struct sockaddr_in6*) so;
2651 [ + - + - : 616 : if (IN6_IS_ADDR_V4MAPPED(&addr6->sin6_addr)) {
- + ]
2652 : 616 : struct sockaddr_in addr4;
2653 [ + - ]: 616 : memset (&addr4, 0, sizeof(addr4));
2654 : 616 : addr4.sin_family = AF_INET;
2655 : 616 : addr4.sin_port = addr6->sin6_port;
2656 [ + - ]: 616 : memcpy (&addr4.sin_addr.s_addr, addr6->sin6_addr.s6_addr+12, sizeof(addr4.sin_addr.s_addr));
2657 [ + - ]: 616 : (void) getnameinfo ((struct sockaddr*) &addr4, sizeof (addr4),
2658 : : hostname, sizeof (hostname), NULL, 0,
2659 : : NI_NUMERICHOST);
2660 : : } else {
2661 [ # # ]: 0 : (void) getnameinfo (so, sizeof (struct sockaddr_in6), hostname, sizeof (hostname), NULL, 0,
2662 : : NI_NUMERICHOST);
2663 : : }
2664 : : }
2665 : :
2666 [ + - + - : 1236 : string xff_complete = string("X-Forwarded-For: ")+xff+string(hostname);
+ - + - -
+ - + - -
- + ]
2667 [ + - ]: 616 : debuginfod_add_http_header (client, xff_complete.c_str());
2668 [ + + + - : 1300 : }
+ + ]
2669 : :
2670 [ + + ]: 616 : if (artifacttype == "debuginfo")
2671 [ + - ]: 68 : fd = debuginfod_find_debuginfo (client,
2672 [ + - ]: 68 : (const unsigned char*) buildid.c_str(),
2673 : : 0, NULL);
2674 [ + + ]: 548 : else if (artifacttype == "executable")
2675 [ + - ]: 546 : fd = debuginfod_find_executable (client,
2676 [ + - ]: 546 : (const unsigned char*) buildid.c_str(),
2677 : : 0, NULL);
2678 [ + - ]: 2 : else if (artifacttype == "source")
2679 [ + - ]: 2 : fd = debuginfod_find_source (client,
2680 [ + - ]: 2 : (const unsigned char*) buildid.c_str(),
2681 : : 0, suffix.c_str(), NULL);
2682 [ # # ]: 0 : else if (artifacttype == "section")
2683 [ # # ]: 0 : fd = debuginfod_find_section (client,
2684 [ # # ]: 0 : (const unsigned char*) buildid.c_str(),
2685 : : 0, section.c_str(), NULL);
2686 : :
2687 [ + + ]: 616 : if (fd >= 0)
2688 : : {
2689 [ + - ]: 4 : if (conn != 0)
2690 [ + - + - : 624 : inc_metric ("http_responses_total","result","upstream");
+ - + - -
+ - + - -
- - ]
2691 : 4 : struct stat s;
2692 : 4 : int rc = fstat (fd, &s);
2693 [ + - ]: 4 : if (rc == 0)
2694 : : {
2695 [ + - ]: 4 : auto r = MHD_create_response_from_fd ((uint64_t) s.st_size, fd);
2696 [ + - ]: 4 : if (r)
2697 : : {
2698 [ + - ]: 4 : add_mhd_response_header (r, "Content-Type",
2699 : : "application/octet-stream");
2700 : : // Copy the incoming headers
2701 [ + - ]: 4 : const char * hdrs = debuginfod_get_headers(client);
2702 [ + - ]: 4 : string header_dup;
2703 [ + - ]: 4 : if (hdrs)
2704 [ + - - + ]: 4 : header_dup = string(hdrs);
2705 : : // Parse the "header: value\n" lines into (h,v) tuples and pass on
2706 : 20 : while(1)
2707 : : {
2708 : 12 : size_t newline = header_dup.find('\n');
2709 [ + + ]: 12 : if (newline == string::npos) break;
2710 : 8 : size_t colon = header_dup.find(':');
2711 [ + - ]: 8 : if (colon == string::npos) break;
2712 [ + - ]: 8 : string header = header_dup.substr(0,colon);
2713 [ + - ]: 8 : string value = header_dup.substr(colon+1,newline-colon-1);
2714 : : // strip leading spaces from value
2715 : 8 : size_t nonspace = value.find_first_not_of(" ");
2716 [ + - ]: 8 : if (nonspace != string::npos)
2717 [ + - + + ]: 12 : value = value.substr(nonspace);
2718 [ + - ]: 8 : add_mhd_response_header(r, header.c_str(), value.c_str());
2719 [ + - + + : 12 : header_dup = header_dup.substr(newline+1);
+ + - - ]
2720 [ + - ]: 16 : }
2721 : :
2722 [ + - ]: 4 : add_mhd_last_modified (r, s.st_mtime);
2723 [ + - ]: 4 : if (verbose > 1)
2724 [ + - + - : 8 : obatched(clog) << "serving file from upstream debuginfod/cache" << endl;
- - ]
2725 [ + - ]: 4 : if (result_fd)
2726 : 4 : *result_fd = fd;
2727 [ + - ]: 4 : return r; // NB: don't close fd; libmicrohttpd will
2728 : 4 : }
2729 : : }
2730 [ # # ]: 0 : close (fd);
2731 : : }
2732 : : else
2733 [ + + ]: 612 : switch(fd)
2734 : : {
2735 : : case -ENOSYS:
2736 : : break;
2737 : : case -ENOENT:
2738 : : break;
2739 : 532 : default: // some more tricky error
2740 [ + - + - ]: 1064 : throw libc_exception(-fd, "upstream debuginfod query failed");
2741 : : }
2742 : :
2743 [ + - - + ]: 160 : throw reportable_exception(MHD_HTTP_NOT_FOUND, "not found");
2744 [ - + - + ]: 2561 : }
2745 : :
2746 : :
2747 : : ////////////////////////////////////////////////////////////////////////
2748 : :
2749 : : static map<string,double> metrics; // arbitrary data for /metrics query
2750 : : // NB: store int64_t since all our metrics are integers; prometheus accepts double
2751 : : static mutex metrics_lock;
2752 : : // NB: these objects get released during the process exit via global dtors
2753 : : // do not call them from within other global dtors
2754 : :
2755 : : // utility function for assembling prometheus-compatible
2756 : : // name="escaped-value" strings
2757 : : // https://prometheus.io/docs/instrumenting/exposition_formats/
2758 : : static string
2759 : 739781 : metric_label(const string& name, const string& value)
2760 : : {
2761 : 739781 : string x = name + "=\"";
2762 [ + + ]: 12881830 : for (auto&& c : value)
2763 [ - - - + ]: 12142672 : switch(c)
2764 : : {
2765 [ # # ]: 0 : case '\\': x += "\\\\"; break;
2766 [ # # ]: 0 : case '\"': x += "\\\""; break;
2767 [ # # ]: 0 : case '\n': x += "\\n"; break;
2768 [ + - ]: 24285064 : default: x += c; break;
2769 : : }
2770 [ + - ]: 739158 : x += "\"";
2771 : 739281 : return x;
2772 : 0 : }
2773 : :
2774 : :
2775 : : // add prometheus-format metric name + label tuple (if any) + value
2776 : :
2777 : : static void
2778 : 1276 : set_metric(const string& metric, double value)
2779 : : {
2780 : 1276 : unique_lock<mutex> lock(metrics_lock);
2781 [ + - ]: 1276 : metrics[metric] = value;
2782 : 1276 : }
2783 : : static void
2784 : 592 : inc_metric(const string& metric)
2785 : : {
2786 : 592 : unique_lock<mutex> lock(metrics_lock);
2787 [ + - ]: 592 : metrics[metric] ++;
2788 : 592 : }
2789 : : static void
2790 : 6407 : set_metric(const string& metric,
2791 : : const string& lname, const string& lvalue,
2792 : : double value)
2793 : : {
2794 [ + - + - : 12815 : string key = (metric + "{" + metric_label(lname, lvalue) + "}");
- + - + +
+ - - ]
2795 [ + - ]: 6408 : unique_lock<mutex> lock(metrics_lock);
2796 [ + - ]: 6408 : metrics[key] = value;
2797 [ + - ]: 12816 : }
2798 : :
2799 : : static void
2800 : 357215 : inc_metric(const string& metric,
2801 : : const string& lname, const string& lvalue)
2802 : : {
2803 [ + - + - : 777754 : string key = (metric + "{" + metric_label(lname, lvalue) + "}");
- + + + +
+ - - ]
2804 [ + - ]: 357207 : unique_lock<mutex> lock(metrics_lock);
2805 [ + - ]: 357220 : metrics[key] ++;
2806 [ + - ]: 714422 : }
2807 : : static void
2808 : 344705 : add_metric(const string& metric,
2809 : : const string& lname, const string& lvalue,
2810 : : double value)
2811 : : {
2812 [ + - + - : 752745 : string key = (metric + "{" + metric_label(lname, lvalue) + "}");
- + + + +
+ - - ]
2813 [ + - ]: 344714 : unique_lock<mutex> lock(metrics_lock);
2814 [ + - ]: 344754 : metrics[key] += value;
2815 [ + - ]: 689501 : }
2816 : : static void
2817 : 592 : add_metric(const string& metric,
2818 : : double value)
2819 : : {
2820 : 592 : unique_lock<mutex> lock(metrics_lock);
2821 [ + - ]: 592 : metrics[metric] += value;
2822 : 592 : }
2823 : :
2824 : :
2825 : : // and more for higher arity labels if needed
2826 : :
2827 : : static void
2828 : 7881 : inc_metric(const string& metric,
2829 : : const string& lname, const string& lvalue,
2830 : : const string& rname, const string& rvalue)
2831 : : {
2832 [ + - - + : 15762 : string key = (metric + "{"
- - ]
2833 [ + - + - : 31524 : + metric_label(lname, lvalue) + ","
- + - + +
+ - - ]
2834 [ + - - + : 23643 : + metric_label(rname, rvalue) + "}");
- + ]
2835 [ + - ]: 7881 : unique_lock<mutex> lock(metrics_lock);
2836 [ + - ]: 7881 : metrics[key] ++;
2837 [ + - ]: 15762 : }
2838 : : static void
2839 : 7881 : add_metric(const string& metric,
2840 : : const string& lname, const string& lvalue,
2841 : : const string& rname, const string& rvalue,
2842 : : double value)
2843 : : {
2844 [ + - - + : 15762 : string key = (metric + "{"
- - ]
2845 [ + - + - : 31524 : + metric_label(lname, lvalue) + ","
- + - + +
+ - - ]
2846 [ + - - + : 23643 : + metric_label(rname, rvalue) + "}");
- + ]
2847 [ + - ]: 7881 : unique_lock<mutex> lock(metrics_lock);
2848 [ + - ]: 7881 : metrics[key] += value;
2849 [ + - ]: 15762 : }
2850 : :
2851 : : static struct MHD_Response*
2852 : 712 : handle_metrics (off_t* size)
2853 : : {
2854 : 712 : stringstream o;
2855 : 712 : {
2856 [ + - ]: 712 : unique_lock<mutex> lock(metrics_lock);
2857 [ + + ]: 68552 : for (auto&& i : metrics)
2858 [ + - ]: 67840 : o << i.first
2859 : : << " "
2860 [ + - + - ]: 67840 : << std::setprecision(std::numeric_limits<double>::digits10 + 1)
2861 [ + - + - ]: 67840 : << i.second
2862 : 67840 : << endl;
2863 : 0 : }
2864 [ + - ]: 712 : const string& os = o.str();
2865 [ + - ]: 712 : MHD_Response* r = MHD_create_response_from_buffer (os.size(),
2866 [ + - ]: 712 : (void*) os.c_str(),
2867 : : MHD_RESPMEM_MUST_COPY);
2868 [ + - ]: 712 : if (r != NULL)
2869 : : {
2870 [ + - ]: 712 : *size = os.size();
2871 [ + - ]: 712 : add_mhd_response_header (r, "Content-Type", "text/plain");
2872 : : }
2873 [ + - ]: 1424 : return r;
2874 : 712 : }
2875 : :
2876 : : static struct MHD_Response*
2877 : 0 : handle_root (off_t* size)
2878 : : {
2879 [ # # # # : 0 : static string version = "debuginfod (" + string (PACKAGE_NAME) + ") "
# # # # #
# # # ]
2880 [ # # # # : 0 : + string (PACKAGE_VERSION);
# # # # #
# ]
2881 : 0 : MHD_Response* r = MHD_create_response_from_buffer (version.size (),
2882 : 0 : (void *) version.c_str (),
2883 : : MHD_RESPMEM_PERSISTENT);
2884 [ # # ]: 0 : if (r != NULL)
2885 : : {
2886 : 0 : *size = version.size ();
2887 : 0 : add_mhd_response_header (r, "Content-Type", "text/plain");
2888 : : }
2889 : 0 : return r;
2890 : : }
2891 : :
2892 : :
2893 : : ////////////////////////////////////////////////////////////////////////
2894 : :
2895 : :
2896 : : /* libmicrohttpd callback */
2897 : : static MHD_RESULT
2898 : 5254 : handler_cb (void * /*cls*/,
2899 : : struct MHD_Connection *connection,
2900 : : const char *url,
2901 : : const char *method,
2902 : : const char * /*version*/,
2903 : : const char * /*upload_data*/,
2904 : : size_t * /*upload_data_size*/,
2905 : : void ** ptr)
2906 : : {
2907 : 5254 : struct MHD_Response *r = NULL;
2908 : 5254 : string url_copy = url;
2909 : :
2910 : : /* libmicrohttpd always makes (at least) two callbacks: once just
2911 : : past the headers, and one after the request body is finished
2912 : : being received. If we process things early (first callback) and
2913 : : queue a response, libmicrohttpd would suppress http keep-alive
2914 : : (via connection->read_closed = true). */
2915 : 5253 : static int aptr; /* just some random object to use as a flag */
2916 [ + + ]: 5253 : if (&aptr != *ptr)
2917 : : {
2918 : : /* do never respond on first call */
2919 : 2626 : *ptr = &aptr;
2920 : 2626 : return MHD_YES;
2921 : : }
2922 : 2627 : *ptr = NULL; /* reset when done */
2923 : :
2924 [ + - ]: 2627 : const char *maxsize_string = MHD_lookup_connection_value(connection, MHD_HEADER_KIND, "X-DEBUGINFOD-MAXSIZE");
2925 : 2627 : long maxsize = 0;
2926 [ + + + - ]: 2627 : if (maxsize_string != NULL && maxsize_string[0] != '\0')
2927 : 2 : maxsize = atol(maxsize_string);
2928 : : else
2929 : : maxsize = 0;
2930 : :
2931 : : #if MHD_VERSION >= 0x00097002
2932 : 2627 : enum MHD_Result rc;
2933 : : #else
2934 : : int rc = MHD_NO; // mhd
2935 : : #endif
2936 : 2627 : int http_code = 500;
2937 : 2627 : off_t http_size = -1;
2938 : 2627 : struct timespec ts_start, ts_end;
2939 : 2627 : clock_gettime (CLOCK_MONOTONIC, &ts_start);
2940 : 2627 : double afteryou = 0.0;
2941 [ + - ]: 2627 : string artifacttype, suffix;
2942 : :
2943 : 2627 : try
2944 : : {
2945 [ + - - + : 5887 : if (string(method) != "GET")
- + ]
2946 [ # # # # ]: 0 : throw reportable_exception(400, "we support GET only");
2947 : :
2948 : : /* Start decoding the URL. */
2949 : 2626 : size_t slash1 = url_copy.find('/', 1);
2950 [ + - ]: 2627 : string url1 = url_copy.substr(0, slash1); // ok even if slash1 not found
2951 : :
2952 [ + + + # ]: 4533 : if (slash1 != string::npos && url1 == "/buildid")
2953 : : {
2954 : : // PR27863: block this thread awhile if another thread is already busy
2955 : : // fetching the exact same thing. This is better for Everyone.
2956 : : // The latecomer says "... after you!" and waits.
2957 [ + - + - : 4444 : add_metric ("thread_busy", "role", "http-buildid-after-you", 1);
+ - + - -
+ + - - -
- - ]
2958 : : #ifdef HAVE_PTHREAD_SETNAME_NP
2959 : 1909 : (void) pthread_setname_np (pthread_self(), "mhd-buildid-after-you");
2960 : : #endif
2961 : 1909 : struct timespec tsay_start, tsay_end;
2962 : 1909 : clock_gettime (CLOCK_MONOTONIC, &tsay_start);
2963 [ + + + - ]: 1909 : static unique_set<string> busy_urls;
2964 [ + - ]: 1909 : unique_set_reserver<string> after_you(busy_urls, url_copy);
2965 : 1909 : clock_gettime (CLOCK_MONOTONIC, &tsay_end);
2966 : 1909 : afteryou = (tsay_end.tv_sec - tsay_start.tv_sec) + (tsay_end.tv_nsec - tsay_start.tv_nsec)/1.e9;
2967 [ + - + - : 3818 : add_metric ("thread_busy", "role", "http-buildid-after-you", -1);
+ - + - -
+ + - - -
- - ]
2968 : :
2969 [ + - + - : 3818 : tmp_inc_metric m ("thread_busy", "role", "http-buildid");
+ - + - -
+ - + - -
- - ]
2970 : : #ifdef HAVE_PTHREAD_SETNAME_NP
2971 : 1909 : (void) pthread_setname_np (pthread_self(), "mhd-buildid");
2972 : : #endif
2973 : 1909 : size_t slash2 = url_copy.find('/', slash1+1);
2974 [ - + ]: 1909 : if (slash2 == string::npos)
2975 [ # # # # ]: 0 : throw reportable_exception("/buildid/ webapi error, need buildid");
2976 : :
2977 [ + - ]: 1909 : string buildid = url_copy.substr(slash1+1, slash2-slash1-1);
2978 : :
2979 : 1909 : size_t slash3 = url_copy.find('/', slash2+1);
2980 : :
2981 [ + + ]: 1909 : if (slash3 == string::npos)
2982 : : {
2983 [ + - - + ]: 1324 : artifacttype = url_copy.substr(slash2+1);
2984 [ + - ]: 1324 : suffix = "";
2985 : : }
2986 : : else
2987 : : {
2988 [ + - - + ]: 585 : artifacttype = url_copy.substr(slash2+1, slash3-slash2-1);
2989 [ + - - + : 1211 : suffix = url_copy.substr(slash3); // include the slash in the suffix
+ + ]
2990 : : }
2991 : :
2992 : : // get the resulting fd so we can report its size
2993 : 1909 : int fd;
2994 [ + + ]: 1909 : r = handle_buildid(connection, buildid, artifacttype, suffix, &fd);
2995 [ + - ]: 1283 : if (r)
2996 : : {
2997 : 1283 : struct stat fs;
2998 [ + - ]: 1283 : if (fstat(fd, &fs) == 0)
2999 : 1283 : http_size = fs.st_size;
3000 : : // libmicrohttpd will close (fd);
3001 : : }
3002 : 2535 : }
3003 [ + + ]: 717 : else if (url1 == "/metrics")
3004 : : {
3005 [ + - + - : 1424 : tmp_inc_metric m ("thread_busy", "role", "http-metrics");
+ - + - -
+ - + - -
- - ]
3006 [ + - ]: 712 : artifacttype = "metrics";
3007 [ + - + - : 1424 : inc_metric("http_requests_total", "type", artifacttype);
+ - - + -
- ]
3008 [ + - ]: 712 : r = handle_metrics(& http_size);
3009 : 712 : }
3010 [ - + ]: 6 : else if (url1 == "/")
3011 : : {
3012 [ # # ]: 0 : artifacttype = "/";
3013 [ - - - - : 634 : inc_metric("http_requests_total", "type", artifacttype);
- - - - -
- - + ]
3014 [ # # ]: 0 : r = handle_root(& http_size);
3015 : : }
3016 : : else
3017 [ + - + - : 18 : throw reportable_exception("webapi error, unrecognized '" + url1 + "'");
+ - - + ]
3018 : :
3019 [ - + ]: 1995 : if (r == 0)
3020 [ # # # # ]: 0 : throw reportable_exception("internal error, missing response");
3021 : :
3022 [ + + + - ]: 1995 : if (maxsize > 0 && http_size > maxsize)
3023 : : {
3024 [ + - ]: 2 : MHD_destroy_response(r);
3025 [ + - + - : 6 : throw reportable_exception(406, "File too large, max size=" + std::to_string(maxsize));
+ - - + ]
3026 : : }
3027 : :
3028 [ + - ]: 1993 : rc = MHD_queue_response (connection, MHD_HTTP_OK, r);
3029 : 1993 : http_code = MHD_HTTP_OK;
3030 [ + - ]: 1993 : MHD_destroy_response (r);
3031 : 2627 : }
3032 [ - + ]: 634 : catch (const reportable_exception& e)
3033 : : {
3034 [ + - + - : 1268 : inc_metric("http_responses_total","result","error");
+ - + - -
+ - + - -
- - ]
3035 [ + - ]: 634 : e.report(clog);
3036 : 634 : http_code = e.code;
3037 [ + - ]: 634 : http_size = e.message.size();
3038 [ + - ]: 634 : rc = e.mhd_send_response (connection);
3039 : 634 : }
3040 : :
3041 : 2627 : clock_gettime (CLOCK_MONOTONIC, &ts_end);
3042 : 2627 : double deltas = (ts_end.tv_sec - ts_start.tv_sec) + (ts_end.tv_nsec - ts_start.tv_nsec)/1.e9;
3043 : : // afteryou: delay waiting for other client's identical query to complete
3044 : : // deltas: total latency, including afteryou waiting
3045 [ + - + - : 5254 : obatched(clog) << conninfo(connection)
- - ]
3046 : : << ' ' << method << ' ' << url
3047 [ + - + - : 2627 : << ' ' << http_code << ' ' << http_size
+ - + - +
- + - +
- ]
3048 [ + - + - : 2627 : << ' ' << (int)(afteryou*1000) << '+' << (int)((deltas-afteryou)*1000) << "ms"
+ - + - +
- + - +
- ]
3049 [ + - ]: 2627 : << endl;
3050 : :
3051 : : // related prometheus metrics
3052 : 2627 : string http_code_str = to_string(http_code);
3053 [ + - + - : 5254 : add_metric("http_responses_transfer_bytes_sum",
+ - + - -
+ - + - -
- - ]
3054 : : "code", http_code_str, "type", artifacttype, http_size);
3055 [ + - + - : 5254 : inc_metric("http_responses_transfer_bytes_count",
+ - + - -
+ - + - -
- - ]
3056 : : "code", http_code_str, "type", artifacttype);
3057 : :
3058 [ + - + - : 5254 : add_metric("http_responses_duration_milliseconds_sum",
+ - + - -
+ - + - -
- - ]
3059 : : "code", http_code_str, "type", artifacttype, deltas*1000); // prometheus prefers _seconds and floating point
3060 [ + - + - : 5254 : inc_metric("http_responses_duration_milliseconds_count",
+ - + - -
+ - + - -
- - ]
3061 : : "code", http_code_str, "type", artifacttype);
3062 : :
3063 [ + - + - : 5254 : add_metric("http_responses_after_you_milliseconds_sum",
+ - + - -
+ - + - -
- - ]
3064 : : "code", http_code_str, "type", artifacttype, afteryou*1000);
3065 [ + - + - : 5254 : inc_metric("http_responses_after_you_milliseconds_count",
+ - + - -
+ - + - -
- - - - ]
3066 : : "code", http_code_str, "type", artifacttype);
3067 : :
3068 [ - + ]: 2627 : return rc;
3069 [ + + - + : 12275 : }
+ + ]
3070 : :
3071 : :
3072 : : ////////////////////////////////////////////////////////////////////////
3073 : : // borrowed originally from src/nm.c get_local_names()
3074 : :
3075 : : static void
3076 : 244 : dwarf_extract_source_paths (Elf *elf, set<string>& debug_sourcefiles)
3077 : : noexcept // no exceptions - so we can simplify the altdbg resource release at end
3078 : : {
3079 : 244 : Dwarf* dbg = dwarf_begin_elf (elf, DWARF_C_READ, NULL);
3080 [ - + ]: 244 : if (dbg == NULL)
3081 : 0 : return;
3082 : :
3083 : 244 : Dwarf* altdbg = NULL;
3084 : 244 : int altdbg_fd = -1;
3085 : :
3086 : : // DWZ handling: if we have an unsatisfied debug-alt-link, add an
3087 : : // empty string into the outgoing sourcefiles set, so the caller
3088 : : // should know that our data is incomplete.
3089 : 244 : const char *alt_name_p;
3090 : 244 : const void *alt_build_id; // elfutils-owned memory
3091 : 244 : ssize_t sz = dwelf_dwarf_gnu_debugaltlink (dbg, &alt_name_p, &alt_build_id);
3092 [ + + ]: 244 : if (sz > 0) // got one!
3093 : : {
3094 : 40 : string buildid;
3095 : 40 : unsigned char* build_id_bytes = (unsigned char*) alt_build_id;
3096 [ + + ]: 840 : for (ssize_t idx=0; idx<sz; idx++)
3097 : : {
3098 : 800 : buildid += "0123456789abcdef"[build_id_bytes[idx] >> 4];
3099 : 800 : buildid += "0123456789abcdef"[build_id_bytes[idx] & 0xf];
3100 : : }
3101 : :
3102 [ + - ]: 40 : if (verbose > 3)
3103 : 40 : obatched(clog) << "Need altdebug buildid=" << buildid << endl;
3104 : :
3105 : : // but is it unsatisfied the normal elfutils ways?
3106 : 40 : Dwarf* alt = dwarf_getalt (dbg);
3107 [ + - ]: 40 : if (alt == NULL)
3108 : : {
3109 : : // Yup, unsatisfied the normal way. Maybe we can satisfy it
3110 : : // from our own debuginfod database.
3111 : 40 : int alt_fd;
3112 : 40 : struct MHD_Response *r = 0;
3113 : 40 : try
3114 : : {
3115 [ + - ]: 40 : string artifacttype = "debuginfo";
3116 [ + - + - : 40 : r = handle_buildid (0, buildid, artifacttype, "", &alt_fd);
- + - + -
- ]
3117 : 0 : }
3118 [ - - ]: 0 : catch (const reportable_exception& e)
3119 : : {
3120 : : // swallow exceptions
3121 : 0 : }
3122 : :
3123 : : // NB: this is not actually recursive! This invokes the web-query
3124 : : // path, which cannot get back into the scan code paths.
3125 [ + - ]: 40 : if (r)
3126 : : {
3127 : : // Found it!
3128 : 40 : altdbg_fd = dup(alt_fd); // ok if this fails, downstream failures ok
3129 : 40 : alt = altdbg = dwarf_begin (altdbg_fd, DWARF_C_READ);
3130 : : // NB: must close this dwarf and this fd at the bottom of the function!
3131 : 40 : MHD_destroy_response (r); // will close alt_fd
3132 [ + - ]: 40 : if (alt)
3133 : 40 : dwarf_setalt (dbg, alt);
3134 : : }
3135 : : }
3136 : : else
3137 : : {
3138 : : // NB: dwarf_setalt(alt) inappropriate - already done!
3139 : : // NB: altdbg will stay 0 so nothing tries to redundantly dealloc.
3140 : : }
3141 : :
3142 [ + - ]: 40 : if (alt)
3143 : : {
3144 [ + - ]: 40 : if (verbose > 3)
3145 : 40 : obatched(clog) << "Resolved altdebug buildid=" << buildid << endl;
3146 : : }
3147 : : else // (alt == NULL) - signal possible presence of poor debuginfo
3148 : : {
3149 [ # # ]: 0 : debug_sourcefiles.insert("");
3150 [ # # ]: 0 : if (verbose > 3)
3151 : 0 : obatched(clog) << "Unresolved altdebug buildid=" << buildid << endl;
3152 : : }
3153 : 40 : }
3154 : :
3155 : 244 : Dwarf_Off offset = 0;
3156 : 244 : Dwarf_Off old_offset;
3157 : 244 : size_t hsize;
3158 : :
3159 [ + + ]: 8880 : while (dwarf_nextcu (dbg, old_offset = offset, &offset, &hsize, NULL, NULL, NULL) == 0)
3160 : : {
3161 : 8636 : Dwarf_Die cudie_mem;
3162 : 8636 : Dwarf_Die *cudie = dwarf_offdie (dbg, old_offset + hsize, &cudie_mem);
3163 : :
3164 [ - + ]: 8636 : if (cudie == NULL)
3165 : 20 : continue;
3166 [ + + ]: 8636 : if (dwarf_tag (cudie) != DW_TAG_compile_unit)
3167 : 20 : continue;
3168 : :
3169 [ - + ]: 8616 : const char *cuname = dwarf_diename(cudie) ?: "unknown";
3170 : :
3171 : 8616 : Dwarf_Files *files;
3172 : 8616 : size_t nfiles;
3173 [ - + ]: 8616 : if (dwarf_getsrcfiles (cudie, &files, &nfiles) != 0)
3174 : 0 : continue;
3175 : :
3176 : : // extract DW_AT_comp_dir to resolve relative file names
3177 : 8616 : const char *comp_dir = "";
3178 : 8616 : const char *const *dirs;
3179 : 8616 : size_t ndirs;
3180 [ - + ]: 8616 : if (dwarf_getsrcdirs (files, &dirs, &ndirs) == 0 &&
3181 [ - + ]: 8616 : dirs[0] != NULL)
3182 : : comp_dir = dirs[0];
3183 : : if (comp_dir == NULL)
3184 : : comp_dir = "";
3185 : :
3186 [ + + ]: 8616 : if (verbose > 3)
3187 : 13268 : obatched(clog) << "searching for sources for cu=" << cuname << " comp_dir=" << comp_dir
3188 : 6634 : << " #files=" << nfiles << " #dirs=" << ndirs << endl;
3189 : :
3190 [ + - - - ]: 8616 : if (comp_dir[0] == '\0' && cuname[0] != '/')
3191 : : {
3192 [ # # ]: 0 : if (verbose > 3)
3193 : 0 : obatched(clog) << "skipping cu=" << cuname << " due to empty comp_dir" << endl;
3194 : 0 : continue;
3195 : : }
3196 : :
3197 [ + + ]: 139137 : for (size_t f = 1; f < nfiles; f++)
3198 : : {
3199 : 130521 : const char *hat = dwarf_filesrc (files, f, NULL, NULL);
3200 [ - + ]: 130572 : if (hat == NULL)
3201 : 0 : continue;
3202 : :
3203 [ + + - + ]: 248112 : if (string(hat) == "<built-in>") // gcc intrinsics, don't bother record
3204 : 0 : continue;
3205 : :
3206 [ + + ]: 130560 : string waldo;
3207 [ + + ]: 130560 : if (hat[0] == '/') // absolute
3208 [ - + ]: 83226 : waldo = (string (hat));
3209 [ + - ]: 47334 : else if (comp_dir[0] != '\0') // comp_dir relative
3210 [ - + - + : 81658 : waldo = (string (comp_dir) + string("/") + string (hat));
- + - + +
+ ]
3211 : : else
3212 : : {
3213 [ # # ]: 0 : if (verbose > 3)
3214 : 0 : obatched(clog) << "skipping hat=" << hat << " due to empty comp_dir" << endl;
3215 [ # # ]: 0 : continue;
3216 : : }
3217 : :
3218 : : // NB: this is the 'waldo' that a dbginfo client will have
3219 : : // to supply for us to give them the file The comp_dir
3220 : : // prefixing is a definite complication. Otherwise we'd
3221 : : // have to return a setof comp_dirs (one per CU!) with
3222 : : // corresponding filesrc[] names, instead of one absolute
3223 : : // resoved set. Maybe we'll have to do that anyway. XXX
3224 : :
3225 [ + + ]: 130556 : if (verbose > 4)
3226 [ - + ]: 100 : obatched(clog) << waldo
3227 [ - + ]: 50 : << (debug_sourcefiles.find(waldo)==debug_sourcefiles.end() ? " new" : " dup") << endl;
3228 : :
3229 [ + - ]: 130556 : debug_sourcefiles.insert (waldo);
3230 : 130521 : }
3231 : : }
3232 : :
3233 : 244 : dwarf_end(dbg);
3234 [ + + ]: 244 : if (altdbg)
3235 : 40 : dwarf_end(altdbg);
3236 [ + + ]: 244 : if (altdbg_fd >= 0)
3237 : 40 : close(altdbg_fd);
3238 : : }
3239 : :
3240 : :
3241 : :
3242 : : static void
3243 : 1280 : elf_classify (int fd, bool &executable_p, bool &debuginfo_p, string &buildid, set<string>& debug_sourcefiles)
3244 : : {
3245 : 1280 : Elf *elf = elf_begin (fd, ELF_C_READ_MMAP_PRIVATE, NULL);
3246 [ + - ]: 1279 : if (elf == NULL)
3247 : : return;
3248 : :
3249 : 1279 : try // catch our types of errors and clean up the Elf* object
3250 : : {
3251 [ + - + + ]: 1279 : if (elf_kind (elf) != ELF_K_ELF)
3252 : : {
3253 [ + - ]: 791 : elf_end (elf);
3254 : 832 : return;
3255 : : }
3256 : :
3257 : 488 : GElf_Ehdr ehdr_storage;
3258 [ + - ]: 488 : GElf_Ehdr *ehdr = gelf_getehdr (elf, &ehdr_storage);
3259 [ - + ]: 488 : if (ehdr == NULL)
3260 : : {
3261 [ # # ]: 0 : elf_end (elf);
3262 : : return;
3263 : : }
3264 : 488 : auto elf_type = ehdr->e_type;
3265 : :
3266 : 488 : const void *build_id; // elfutils-owned memory
3267 [ + - ]: 488 : ssize_t sz = dwelf_elf_gnu_build_id (elf, & build_id);
3268 [ + + ]: 488 : if (sz <= 0)
3269 : : {
3270 : : // It's not a diagnostic-worthy error for an elf file to lack build-id.
3271 : : // It might just be very old.
3272 [ + - ]: 40 : elf_end (elf);
3273 : : return;
3274 : : }
3275 : :
3276 : : // build_id is a raw byte array; convert to hexadecimal *lowercase*
3277 : 448 : unsigned char* build_id_bytes = (unsigned char*) build_id;
3278 [ + + ]: 9408 : for (ssize_t idx=0; idx<sz; idx++)
3279 : : {
3280 [ + - ]: 8960 : buildid += "0123456789abcdef"[build_id_bytes[idx] >> 4];
3281 [ + - ]: 17920 : buildid += "0123456789abcdef"[build_id_bytes[idx] & 0xf];
3282 : : }
3283 : :
3284 : : // now decide whether it's an executable - namely, any allocatable section has
3285 : : // PROGBITS;
3286 [ + + ]: 448 : if (elf_type == ET_EXEC || elf_type == ET_DYN)
3287 : : {
3288 : 398 : size_t shnum;
3289 [ + - ]: 398 : int rc = elf_getshdrnum (elf, &shnum);
3290 [ - + ]: 398 : if (rc < 0)
3291 [ # # # # ]: 0 : throw elfutils_exception(rc, "getshdrnum");
3292 : :
3293 : 398 : executable_p = false;
3294 [ + + ]: 7054 : for (size_t sc = 0; sc < shnum; sc++)
3295 : : {
3296 [ + - ]: 6878 : Elf_Scn *scn = elf_getscn (elf, sc);
3297 [ - + ]: 6878 : if (scn == NULL)
3298 : 0 : continue;
3299 : :
3300 : 6878 : GElf_Shdr shdr_mem;
3301 [ + - ]: 6878 : GElf_Shdr *shdr = gelf_getshdr (scn, &shdr_mem);
3302 [ - + ]: 6878 : if (shdr == NULL)
3303 : 0 : continue;
3304 : :
3305 : : // allocated (loadable / vm-addr-assigned) section with available content?
3306 [ + + + + ]: 6878 : if ((shdr->sh_type == SHT_PROGBITS) && (shdr->sh_flags & SHF_ALLOC))
3307 : : {
3308 [ + + ]: 222 : if (verbose > 4)
3309 [ + - + - : 32 : obatched(clog) << "executable due to SHF_ALLOC SHT_PROGBITS sc=" << sc << endl;
+ - ]
3310 : 222 : executable_p = true;
3311 : 222 : break; // no need to keep looking for others
3312 : : }
3313 : : } // iterate over sections
3314 : : } // executable_p classification
3315 : :
3316 : : // now decide whether it's a debuginfo - namely, if it has any .debug* or .zdebug* sections
3317 : : // logic mostly stolen from fweimer@redhat.com's elfclassify drafts
3318 : 448 : size_t shstrndx;
3319 [ + - ]: 448 : int rc = elf_getshdrstrndx (elf, &shstrndx);
3320 [ - + ]: 448 : if (rc < 0)
3321 [ # # # # ]: 0 : throw elfutils_exception(rc, "getshdrstrndx");
3322 : :
3323 : : Elf_Scn *scn = NULL;
3324 : : bool symtab_p = false;
3325 : : bool bits_alloc_p = false;
3326 : 23578 : while (true)
3327 : : {
3328 [ + - ]: 12013 : scn = elf_nextscn (elf, scn);
3329 [ + + ]: 11984 : if (scn == NULL)
3330 : : break;
3331 : 11781 : GElf_Shdr shdr_storage;
3332 [ + - ]: 11781 : GElf_Shdr *shdr = gelf_getshdr (scn, &shdr_storage);
3333 [ - + ]: 11782 : if (shdr == NULL)
3334 : : break;
3335 [ + - ]: 11782 : const char *section_name = elf_strptr (elf, shstrndx, shdr->sh_name);
3336 [ - + ]: 11809 : if (section_name == NULL)
3337 : : break;
3338 [ + + ]: 11809 : if (startswith (section_name, ".debug_line") ||
3339 [ - + ]: 11565 : startswith (section_name, ".zdebug_line"))
3340 : : {
3341 : 244 : debuginfo_p = true;
3342 [ + - ]: 244 : if (scan_source_info)
3343 : 244 : dwarf_extract_source_paths (elf, debug_sourcefiles);
3344 : : break; // expecting only one .*debug_line, so no need to look for others
3345 : : }
3346 [ + + ]: 11565 : else if (startswith (section_name, ".debug_") ||
3347 [ + # ]: 10779 : startswith (section_name, ".zdebug_"))
3348 : : {
3349 : 770 : debuginfo_p = true;
3350 : : // NB: don't break; need to parse .debug_line for sources
3351 : : }
3352 [ + + ]: 10795 : else if (shdr->sh_type == SHT_SYMTAB)
3353 : : {
3354 : : symtab_p = true;
3355 : : }
3356 : 10771 : else if (shdr->sh_type != SHT_NOBITS
3357 [ + + ]: 10771 : && shdr->sh_type != SHT_NOTE
3358 [ + + ]: 5535 : && (shdr->sh_flags & SHF_ALLOC) != 0)
3359 : : {
3360 : 11565 : bits_alloc_p = true;
3361 : : }
3362 : 11565 : }
3363 : :
3364 : : // For more expansive elf/split-debuginfo classification, we
3365 : : // want to identify as debuginfo "strip -s"-produced files
3366 : : // without .debug_info* (like libicudata), but we don't want to
3367 : : // identify "strip -g" executables (with .symtab left there).
3368 [ - + ]: 447 : if (symtab_p && !bits_alloc_p)
3369 : 0 : debuginfo_p = true;
3370 : : }
3371 [ # # ]: 0 : catch (const reportable_exception& e)
3372 : : {
3373 [ # # ]: 0 : e.report(clog);
3374 : 0 : }
3375 : 447 : elf_end (elf);
3376 : : }
3377 : :
3378 : :
3379 : : // Intern the given file name in two parts (dirname & basename) and
3380 : : // return the resulting file's id.
3381 : : static int64_t
3382 : 31582 : register_file_name(sqlite_ps& ps_upsert_fileparts,
3383 : : sqlite_ps& ps_upsert_file,
3384 : : sqlite_ps& ps_lookup_file,
3385 : : const string& name)
3386 : : {
3387 : 31582 : std::size_t slash = name.rfind('/');
3388 [ + + ]: 31582 : string dirname, filename;
3389 [ + + ]: 31582 : if (slash == std::string::npos)
3390 : : {
3391 [ + - ]: 90 : dirname = "";
3392 [ + - ]: 90 : filename = name;
3393 : : }
3394 : : else
3395 : : {
3396 [ + - - + ]: 31492 : dirname = name.substr(0, slash);
3397 [ + - - + : 31492 : filename = name.substr(slash+1);
- - ]
3398 : : }
3399 : :
3400 : : // intern the two substrings
3401 : 31582 : ps_upsert_fileparts
3402 [ + - ]: 31582 : .reset()
3403 [ + - ]: 31580 : .bind(1, dirname)
3404 [ + - ]: 31579 : .step_ok_done();
3405 : 31582 : ps_upsert_fileparts
3406 [ + - ]: 31582 : .reset()
3407 [ + - ]: 31581 : .bind(1, filename)
3408 [ + - ]: 31581 : .step_ok_done();
3409 : :
3410 : : // intern the tuple
3411 : 31582 : ps_upsert_file
3412 [ + - ]: 31582 : .reset()
3413 [ + - ]: 31582 : .bind(1, dirname)
3414 [ + - ]: 31582 : .bind(2, filename)
3415 [ + - ]: 31581 : .step_ok_done();
3416 : :
3417 : : // look up the tuple's id
3418 : 31582 : ps_lookup_file
3419 [ + - ]: 31582 : .reset()
3420 [ + - ]: 31582 : .bind(1, dirname)
3421 [ + - ]: 31580 : .bind(2, filename);
3422 [ + - ]: 31581 : int rc = ps_lookup_file.step();
3423 [ - + - - : 31582 : if (rc != SQLITE_ROW) throw sqlite_exception(rc, "step");
- - ]
3424 : :
3425 [ + - ]: 31582 : int64_t id = sqlite3_column_int64 (ps_lookup_file, 0);
3426 [ + - ]: 31582 : ps_lookup_file.reset();
3427 [ + + ]: 31582 : return id;
3428 [ + + ]: 60992 : }
3429 : :
3430 : :
3431 : :
3432 : : static void
3433 : 974 : scan_source_file (const string& rps, const stat_t& st,
3434 : : sqlite_ps& ps_upsert_buildids,
3435 : : sqlite_ps& ps_upsert_fileparts,
3436 : : sqlite_ps& ps_upsert_file,
3437 : : sqlite_ps& ps_lookup_file,
3438 : : sqlite_ps& ps_upsert_de,
3439 : : sqlite_ps& ps_upsert_s,
3440 : : sqlite_ps& ps_query,
3441 : : sqlite_ps& ps_scan_done,
3442 : : unsigned& fts_cached,
3443 : : unsigned& fts_executable,
3444 : : unsigned& fts_debuginfo,
3445 : : unsigned& fts_sourcefiles)
3446 : : {
3447 : 974 : int64_t fileid = register_file_name(ps_upsert_fileparts, ps_upsert_file, ps_lookup_file, rps);
3448 : :
3449 : : /* See if we know of it already. */
3450 : 974 : int rc = ps_query
3451 : 974 : .reset()
3452 : 974 : .bind(1, fileid)
3453 : 974 : .bind(2, st.st_mtime)
3454 : 974 : .step();
3455 : 974 : ps_query.reset();
3456 [ + + ]: 974 : if (rc == SQLITE_ROW) // i.e., a result, as opposed to DONE (no results)
3457 : : // no need to recheck a file/version we already know
3458 : : // specifically, no need to elf-begin a file we already determined is non-elf
3459 : : // (so is stored with buildid=NULL)
3460 : : {
3461 : 368 : fts_cached++;
3462 : 368 : return;
3463 : : }
3464 : :
3465 : 606 : bool executable_p = false, debuginfo_p = false; // E and/or D
3466 [ + - ]: 606 : string buildid;
3467 [ + - ]: 606 : set<string> sourcefiles;
3468 : :
3469 [ + - ]: 606 : int fd = open (rps.c_str(), O_RDONLY);
3470 : 606 : try
3471 : : {
3472 [ + - ]: 606 : if (fd >= 0)
3473 [ + - ]: 606 : elf_classify (fd, executable_p, debuginfo_p, buildid, sourcefiles);
3474 : : else
3475 [ # # # # : 0 : throw libc_exception(errno, string("open ") + rps);
# # # # ]
3476 [ + - + - : 1212 : add_metric ("scanned_bytes_total","source","file",
+ - - + -
+ - - -
- ]
3477 [ + - ]: 606 : st.st_size);
3478 [ + - + - : 1212 : inc_metric ("scanned_files_total","source","file");
+ - + - -
+ - + - -
- - ]
3479 : : }
3480 : : // NB: we catch exceptions here too, so that we can
3481 : : // cache the corrupt-elf case (!executable_p &&
3482 : : // !debuginfo_p) just below, just as if we had an
3483 : : // EPERM error from open(2).
3484 [ - - ]: 0 : catch (const reportable_exception& e)
3485 : : {
3486 [ - - ]: 0 : e.report(clog);
3487 : 0 : }
3488 : :
3489 [ + - ]: 606 : if (fd >= 0)
3490 [ + - ]: 606 : close (fd);
3491 : :
3492 [ + + ]: 606 : if (buildid == "")
3493 : : {
3494 : : // no point storing an elf file without buildid
3495 : 510 : executable_p = false;
3496 : 510 : debuginfo_p = false;
3497 : : }
3498 : : else
3499 : : {
3500 : : // register this build-id in the interning table
3501 : 96 : ps_upsert_buildids
3502 [ + - ]: 96 : .reset()
3503 [ + - ]: 96 : .bind(1, buildid)
3504 [ + - ]: 96 : .step_ok_done();
3505 : : }
3506 : :
3507 [ + + ]: 606 : if (executable_p)
3508 : 72 : fts_executable ++;
3509 [ + + ]: 606 : if (debuginfo_p)
3510 : 72 : fts_debuginfo ++;
3511 [ + + + + ]: 606 : if (executable_p || debuginfo_p)
3512 : : {
3513 : 96 : ps_upsert_de
3514 [ + - ]: 96 : .reset()
3515 [ + - ]: 96 : .bind(1, buildid)
3516 [ + + + - ]: 120 : .bind(2, debuginfo_p ? 1 : 0)
3517 [ + + + - ]: 120 : .bind(3, executable_p ? 1 : 0)
3518 [ + - ]: 96 : .bind(4, fileid)
3519 [ + - ]: 96 : .bind(5, st.st_mtime)
3520 [ + - ]: 96 : .step_ok_done();
3521 : : }
3522 [ + + ]: 606 : if (executable_p)
3523 [ + - + - : 144 : inc_metric("found_executable_total","source","files");
+ - + - -
+ - + - -
- - ]
3524 [ + + ]: 606 : if (debuginfo_p)
3525 [ + - + - : 144 : inc_metric("found_debuginfo_total","source","files");
+ - + - -
+ - + - -
- - ]
3526 : :
3527 [ + + + - ]: 678 : if (sourcefiles.size() && buildid != "")
3528 : : {
3529 : 72 : fts_sourcefiles += sourcefiles.size();
3530 : :
3531 [ + + ]: 14676 : for (auto&& dwarfsrc : sourcefiles)
3532 : : {
3533 [ + - ]: 14604 : char *srp = realpath(dwarfsrc.c_str(), NULL);
3534 [ + + ]: 14604 : if (srp == NULL) // also if DWZ unresolved dwarfsrc=""
3535 : 220 : continue; // unresolvable files are not a serious problem
3536 : : // throw libc_exception(errno, "fts/file realpath " + srcpath);
3537 [ + - ]: 14384 : string srps = string(srp);
3538 : 14384 : free (srp);
3539 : :
3540 : 14384 : struct stat sfs;
3541 : 14384 : rc = stat(srps.c_str(), &sfs);
3542 [ - + ]: 14384 : if (rc != 0)
3543 [ # # ]: 0 : continue;
3544 : :
3545 [ + - ]: 14384 : if (verbose > 2)
3546 [ + - + - : 43152 : obatched(clog) << "recorded buildid=" << buildid << " file=" << srps
- - ]
3547 [ + - + - : 14384 : << " mtime=" << sfs.st_mtime
+ - + - ]
3548 [ + - + - : 14384 : << " as source " << dwarfsrc << endl;
+ - ]
3549 : :
3550 : : // PR25548: store canonicalized dwarfsrc path
3551 [ + - ]: 14384 : string dwarfsrc_canon = canon_pathname (dwarfsrc);
3552 [ + + ]: 14384 : if (dwarfsrc_canon != dwarfsrc)
3553 : : {
3554 [ + + ]: 2434 : if (verbose > 3)
3555 [ + - + - : 3848 : obatched(clog) << "canonicalized src=" << dwarfsrc << " alias=" << dwarfsrc_canon << endl;
+ - + - +
- ]
3556 : : }
3557 : :
3558 [ + - ]: 14384 : int64_t fileid1 = register_file_name (ps_upsert_fileparts, ps_upsert_file, ps_lookup_file, dwarfsrc_canon);
3559 [ + - ]: 14384 : int64_t fileid2 = register_file_name (ps_upsert_fileparts, ps_upsert_file, ps_lookup_file, srps);
3560 : :
3561 : 14384 : ps_upsert_s
3562 [ + - ]: 14384 : .reset()
3563 [ + - ]: 14384 : .bind(1, buildid)
3564 [ + - ]: 14384 : .bind(2, fileid1)
3565 [ + - ]: 14383 : .bind(3, fileid2)
3566 [ + - ]: 14384 : .bind(4, sfs.st_mtime)
3567 [ + - ]: 14384 : .step_ok_done();
3568 : :
3569 [ + - + - : 28768 : inc_metric("found_sourcerefs_total","source","files");
+ - + - -
+ - + + -
- - - - -
- ]
3570 [ + - ]: 28988 : }
3571 : : }
3572 : :
3573 : 606 : ps_scan_done
3574 [ + - ]: 606 : .reset()
3575 [ + - ]: 606 : .bind(1, fileid)
3576 [ + - ]: 606 : .bind(2, st.st_mtime)
3577 [ + - ]: 606 : .bind(3, st.st_size)
3578 [ + - ]: 606 : .step_ok_done();
3579 : :
3580 [ + - ]: 606 : if (verbose > 2)
3581 [ + - + - ]: 1818 : obatched(clog) << "recorded buildid=" << buildid << " file=" << rps
3582 [ + - + - : 606 : << " mtime=" << st.st_mtime << " atype="
+ - + - ]
3583 : : << (executable_p ? "E" : "")
3584 [ + - + + : 1674 : << (debuginfo_p ? "D" : "") << endl;
+ - + + +
- + - ]
3585 [ + + ]: 702 : }
3586 : :
3587 : :
3588 : :
3589 : :
3590 : :
3591 : : // Analyze given archive file of given age; record buildids / exec/debuginfo-ness of its
3592 : : // constituent files with given upsert statements.
3593 : : static void
3594 : 362 : archive_classify (const string& rps, string& archive_extension, int64_t archiveid,
3595 : : sqlite_ps& ps_upsert_buildids, sqlite_ps& ps_upsert_fileparts, sqlite_ps& ps_upsert_file,
3596 : : sqlite_ps& ps_lookup_file,
3597 : : sqlite_ps& ps_upsert_de, sqlite_ps& ps_upsert_sref, sqlite_ps& ps_upsert_sdef,
3598 : : time_t mtime,
3599 : : unsigned& fts_executable, unsigned& fts_debuginfo, unsigned& fts_sref, unsigned& fts_sdef,
3600 : : bool& fts_sref_complete_p)
3601 : : {
3602 : 362 : string archive_decoder = "/dev/null";
3603 [ + + ]: 896 : for (auto&& arch : scan_archives)
3604 [ + + ]: 534 : if (string_endswith(rps, arch.first))
3605 : : {
3606 [ + - ]: 362 : archive_extension = arch.first;
3607 [ + - ]: 896 : archive_decoder = arch.second;
3608 : : }
3609 : :
3610 : 362 : FILE* fp;
3611 : 362 : defer_dtor<FILE*,int>::dtor_fn dfn;
3612 [ + + ]: 362 : if (archive_decoder != "cat")
3613 : : {
3614 [ + - + - : 64 : string popen_cmd = archive_decoder + " " + shell_escape(rps);
+ - - + -
- - - ]
3615 [ + - ]: 32 : fp = popen (popen_cmd.c_str(), "r"); // "e" O_CLOEXEC?
3616 : 32 : dfn = pclose;
3617 [ - + ]: 32 : if (fp == NULL)
3618 [ # # # # : 0 : throw libc_exception (errno, string("popen ") + popen_cmd);
# # # # ]
3619 : 32 : }
3620 : : else
3621 : : {
3622 [ + - ]: 330 : fp = fopen (rps.c_str(), "r");
3623 : 330 : dfn = fclose;
3624 [ - + ]: 330 : if (fp == NULL)
3625 [ # # # # : 0 : throw libc_exception (errno, string("fopen ") + rps);
# # # # ]
3626 : : }
3627 : 362 : defer_dtor<FILE*,int> fp_closer (fp, dfn);
3628 : :
3629 : 362 : struct archive *a;
3630 [ + - ]: 362 : a = archive_read_new();
3631 [ - + ]: 362 : if (a == NULL)
3632 [ # # # # ]: 0 : throw archive_exception("cannot create archive reader");
3633 : 362 : defer_dtor<struct archive*,int> archive_closer (a, archive_read_free);
3634 : :
3635 [ + - ]: 362 : int rc = archive_read_support_format_all(a);
3636 [ - + ]: 362 : if (rc != ARCHIVE_OK)
3637 [ # # # # ]: 0 : throw archive_exception(a, "cannot select all formats");
3638 [ + - ]: 362 : rc = archive_read_support_filter_all(a);
3639 [ - + ]: 362 : if (rc != ARCHIVE_OK)
3640 [ # # # # ]: 0 : throw archive_exception(a, "cannot select all filters");
3641 : :
3642 [ + - ]: 362 : rc = archive_read_open_FILE (a, fp);
3643 [ - + ]: 362 : if (rc != ARCHIVE_OK)
3644 : : {
3645 [ # # # # : 0 : obatched(clog) << "cannot open archive from pipe " << rps << endl;
# # ]
3646 [ # # # # ]: 0 : throw archive_exception(a, "cannot open archive from pipe");
3647 : : }
3648 : :
3649 [ + + ]: 362 : if (verbose > 3)
3650 [ + - + - : 692 : obatched(clog) << "libarchive scanning " << rps << " id " << archiveid << endl;
+ - + - +
- ]
3651 : :
3652 : : bool any_exceptions = false;
3653 : 2564 : while(1) // parse archive entries
3654 : : {
3655 [ + - ]: 2564 : if (interrupted)
3656 : : break;
3657 : :
3658 : 2564 : try
3659 : : {
3660 : 2564 : struct archive_entry *e;
3661 [ + - ]: 2564 : rc = archive_read_next_header (a, &e);
3662 [ + + ]: 2564 : if (rc != ARCHIVE_OK)
3663 : : break;
3664 : :
3665 [ + - + + ]: 2202 : if (! S_ISREG(archive_entry_mode (e))) // skip non-files completely
3666 : 1528 : continue;
3667 : :
3668 [ + - ]: 674 : string fn = canonicalized_archive_entry_pathname (e);
3669 : :
3670 [ + + ]: 674 : if (verbose > 3)
3671 [ + - + - : 1268 : obatched(clog) << "libarchive checking " << fn << endl;
+ - - - ]
3672 : :
3673 : : // extract this file to a temporary file
3674 : 674 : char* tmppath = NULL;
3675 : 674 : rc = asprintf (&tmppath, "%s/debuginfod-classify.XXXXXX", tmpdir.c_str());
3676 [ - + ]: 674 : if (rc < 0)
3677 [ # # # # ]: 0 : throw libc_exception (ENOMEM, "cannot allocate tmppath");
3678 : 674 : defer_dtor<void*,void> tmmpath_freer (tmppath, free);
3679 [ + - ]: 674 : int fd = mkstemp (tmppath);
3680 [ - + ]: 674 : if (fd < 0)
3681 [ # # # # ]: 0 : throw libc_exception (errno, "cannot create temporary file");
3682 : 674 : unlink (tmppath); // unlink now so OS will release the file as soon as we close the fd
3683 : 674 : defer_dtor<int,int> minifd_closer (fd, close);
3684 : :
3685 [ + - ]: 674 : rc = archive_read_data_into_fd (a, fd);
3686 [ - + ]: 674 : if (rc != ARCHIVE_OK) {
3687 [ # # ]: 0 : close (fd);
3688 [ # # # # ]: 0 : throw archive_exception(a, "cannot extract file");
3689 : : }
3690 : :
3691 : : // finally ... time to run elf_classify on this bad boy and update the database
3692 : 674 : bool executable_p = false, debuginfo_p = false;
3693 [ + - ]: 674 : string buildid;
3694 [ + - ]: 674 : set<string> sourcefiles;
3695 [ + - ]: 674 : elf_classify (fd, executable_p, debuginfo_p, buildid, sourcefiles);
3696 : : // NB: might throw
3697 : :
3698 [ + + ]: 674 : if (buildid != "") // intern buildid
3699 : : {
3700 : 352 : ps_upsert_buildids
3701 [ + - ]: 352 : .reset()
3702 [ + - ]: 352 : .bind(1, buildid)
3703 [ + - ]: 352 : .step_ok_done();
3704 : : }
3705 : :
3706 [ + - ]: 674 : int64_t fileid = register_file_name (ps_upsert_fileparts, ps_upsert_file, ps_lookup_file, fn);
3707 : :
3708 [ + + ]: 674 : if (sourcefiles.size() > 0) // sref records needed
3709 : : {
3710 : : // NB: we intern each source file once. Once raw, as it
3711 : : // appears in the DWARF file list coming back from
3712 : : // elf_classify() - because it'll end up in the
3713 : : // _norm.artifactsrc column. We don't also put another
3714 : : // version with a '.' at the front, even though that's
3715 : : // how rpm/cpio packs names, because we hide that from
3716 : : // the database for storage efficiency.
3717 : :
3718 [ + + ]: 616 : for (auto&& s : sourcefiles)
3719 : : {
3720 [ - + ]: 464 : if (s == "")
3721 : : {
3722 : 0 : fts_sref_complete_p = false;
3723 : 0 : continue;
3724 : : }
3725 : :
3726 : : // PR25548: store canonicalized source path
3727 : 464 : const string& dwarfsrc = s;
3728 [ + - ]: 464 : string dwarfsrc_canon = canon_pathname (dwarfsrc);
3729 [ + + ]: 464 : if (dwarfsrc_canon != dwarfsrc)
3730 : : {
3731 [ + - ]: 32 : if (verbose > 3)
3732 [ + - + - : 64 : obatched(clog) << "canonicalized src=" << dwarfsrc << " alias=" << dwarfsrc_canon << endl;
+ - + - +
- - - ]
3733 : : }
3734 : :
3735 [ + - ]: 464 : int64_t srcfileid = register_file_name(ps_upsert_fileparts, ps_upsert_file, ps_lookup_file,
3736 : : dwarfsrc_canon);
3737 : :
3738 : 464 : ps_upsert_sref
3739 [ + - ]: 464 : .reset()
3740 [ + - ]: 464 : .bind(1, buildid)
3741 [ + - ]: 464 : .bind(2, srcfileid)
3742 [ + - ]: 464 : .step_ok_done();
3743 : :
3744 [ + - ]: 464 : fts_sref ++;
3745 : 464 : }
3746 : : }
3747 : :
3748 [ + + ]: 674 : if (executable_p)
3749 : 150 : fts_executable ++;
3750 [ + + ]: 674 : if (debuginfo_p)
3751 : 202 : fts_debuginfo ++;
3752 : :
3753 [ + + + + ]: 674 : if (executable_p || debuginfo_p)
3754 : : {
3755 : 352 : ps_upsert_de
3756 [ + - ]: 352 : .reset()
3757 [ + - ]: 352 : .bind(1, buildid)
3758 [ + + + - ]: 502 : .bind(2, debuginfo_p ? 1 : 0)
3759 [ + + + - ]: 554 : .bind(3, executable_p ? 1 : 0)
3760 [ + - ]: 352 : .bind(4, archiveid)
3761 [ + - ]: 352 : .bind(5, mtime)
3762 [ + - ]: 352 : .bind(6, fileid)
3763 [ + - ]: 352 : .step_ok_done();
3764 : : }
3765 : : else // potential source - sdef record
3766 : : {
3767 : 322 : fts_sdef ++;
3768 : 322 : ps_upsert_sdef
3769 [ + - ]: 322 : .reset()
3770 [ + - ]: 322 : .bind(1, archiveid)
3771 [ + - ]: 322 : .bind(2, mtime)
3772 [ + - ]: 322 : .bind(3, fileid)
3773 [ + - ]: 322 : .step_ok_done();
3774 : : }
3775 : :
3776 [ + - + + : 674 : if ((verbose > 2) && (executable_p || debuginfo_p))
+ + ]
3777 [ + - + - ]: 1056 : obatched(clog) << "recorded buildid=" << buildid << " rpm=" << rps << " file=" << fn
3778 [ + - + - : 352 : << " mtime=" << mtime << " atype="
+ - + - +
- + - ]
3779 : : << (executable_p ? "E" : "")
3780 : : << (debuginfo_p ? "D" : "")
3781 [ + - + + : 704 : << " sourcefiles=" << sourcefiles.size() << endl;
+ - + + +
- + - + -
+ - ]
3782 : :
3783 [ + + + + ]: 1532 : }
3784 [ - - ]: 0 : catch (const reportable_exception& e)
3785 : : {
3786 [ - - ]: 0 : e.report(clog);
3787 : 0 : any_exceptions = true;
3788 : : // NB: but we allow the libarchive iteration to continue, in
3789 : : // case we can still gather some useful information. That
3790 : : // would allow some webapi queries to work, until later when
3791 : : // this archive is rescanned. (Its vitals won't go into the
3792 : : // _file_mtime_scanned table until after a successful scan.)
3793 : 0 : }
3794 : : }
3795 : :
3796 [ - + ]: 362 : if (any_exceptions)
3797 [ # # # # ]: 0 : throw reportable_exception("exceptions encountered during archive scan");
3798 [ + + ]: 378 : }
3799 : :
3800 : :
3801 : :
3802 : : // scan for archive files such as .rpm
3803 : : static void
3804 : 702 : scan_archive_file (const string& rps, const stat_t& st,
3805 : : sqlite_ps& ps_upsert_buildids,
3806 : : sqlite_ps& ps_upsert_fileparts,
3807 : : sqlite_ps& ps_upsert_file,
3808 : : sqlite_ps& ps_lookup_file,
3809 : : sqlite_ps& ps_upsert_de,
3810 : : sqlite_ps& ps_upsert_sref,
3811 : : sqlite_ps& ps_upsert_sdef,
3812 : : sqlite_ps& ps_query,
3813 : : sqlite_ps& ps_scan_done,
3814 : : unsigned& fts_cached,
3815 : : unsigned& fts_executable,
3816 : : unsigned& fts_debuginfo,
3817 : : unsigned& fts_sref,
3818 : : unsigned& fts_sdef)
3819 : : {
3820 : : // intern the archive file name
3821 : 702 : int64_t archiveid = register_file_name (ps_upsert_fileparts, ps_upsert_file, ps_lookup_file, rps);
3822 : :
3823 : : /* See if we know of it already. */
3824 : 702 : int rc = ps_query
3825 : 702 : .reset()
3826 : 702 : .bind(1, archiveid)
3827 : 702 : .bind(2, st.st_mtime)
3828 : 702 : .step();
3829 : 702 : ps_query.reset();
3830 [ + + ]: 702 : if (rc == SQLITE_ROW) // i.e., a result, as opposed to DONE (no results)
3831 : : // no need to recheck a file/version we already know
3832 : : // specifically, no need to parse this archive again, since we already have
3833 : : // it as a D or E or S record,
3834 : : // (so is stored with buildid=NULL)
3835 : : {
3836 : 340 : fts_cached ++;
3837 : 340 : return;
3838 : : }
3839 : :
3840 : : // extract the archive contents
3841 : 362 : unsigned my_fts_executable = 0, my_fts_debuginfo = 0, my_fts_sref = 0, my_fts_sdef = 0;
3842 : 362 : bool my_fts_sref_complete_p = true;
3843 : 362 : bool any_exceptions = false;
3844 : 362 : try
3845 : : {
3846 [ + - ]: 362 : string archive_extension;
3847 : 362 : archive_classify (rps, archive_extension, archiveid,
3848 : : ps_upsert_buildids, ps_upsert_fileparts, ps_upsert_file, ps_lookup_file,
3849 : : ps_upsert_de, ps_upsert_sref, ps_upsert_sdef, // dalt
3850 [ + - ]: 362 : st.st_mtime,
3851 : : my_fts_executable, my_fts_debuginfo, my_fts_sref, my_fts_sdef,
3852 : : my_fts_sref_complete_p);
3853 [ + - + - : 724 : add_metric ("scanned_bytes_total","source",archive_extension + " archive",
+ - - + +
+ - - -
- ]
3854 [ + - ]: 362 : st.st_size);
3855 [ + - + - : 724 : inc_metric ("scanned_files_total","source",archive_extension + " archive");
+ - + - -
+ + + - -
- - ]
3856 [ + - + - : 724 : add_metric("found_debuginfo_total","source",archive_extension + " archive",
+ - + - -
+ + + - -
- - ]
3857 : : my_fts_debuginfo);
3858 [ + - + - : 724 : add_metric("found_executable_total","source",archive_extension + " archive",
+ - + - -
+ + + - -
- - ]
3859 : : my_fts_executable);
3860 [ + - + - : 740 : add_metric("found_sourcerefs_total","source",archive_extension + " archive",
+ - + - -
+ + + - +
- - - - -
- ]
3861 : : my_fts_sref);
3862 : 362 : }
3863 [ - - ]: 0 : catch (const reportable_exception& e)
3864 : : {
3865 [ - - ]: 0 : e.report(clog);
3866 : 0 : any_exceptions = true;
3867 : 0 : }
3868 : :
3869 [ + - ]: 362 : if (verbose > 2)
3870 [ + - ]: 1086 : obatched(clog) << "scanned archive=" << rps
3871 [ + - + - ]: 362 : << " mtime=" << st.st_mtime
3872 [ + - ]: 362 : << " executables=" << my_fts_executable
3873 [ + - + - ]: 362 : << " debuginfos=" << my_fts_debuginfo
3874 [ + - + - ]: 362 : << " srefs=" << my_fts_sref
3875 [ + - + - ]: 362 : << " sdefs=" << my_fts_sdef
3876 [ + - + - : 362 : << " exceptions=" << any_exceptions
+ - + - ]
3877 : 362 : << endl;
3878 : :
3879 : 362 : fts_executable += my_fts_executable;
3880 : 362 : fts_debuginfo += my_fts_debuginfo;
3881 : 362 : fts_sref += my_fts_sref;
3882 : 362 : fts_sdef += my_fts_sdef;
3883 : :
3884 [ - + ]: 362 : if (any_exceptions)
3885 [ # # # # ]: 0 : throw reportable_exception("exceptions encountered during archive scan");
3886 : :
3887 [ + - ]: 362 : if (my_fts_sref_complete_p) // leave incomplete?
3888 : 362 : ps_scan_done
3889 : 362 : .reset()
3890 : 362 : .bind(1, archiveid)
3891 : 362 : .bind(2, st.st_mtime)
3892 : 362 : .bind(3, st.st_size)
3893 : 362 : .step_ok_done();
3894 : : }
3895 : :
3896 : :
3897 : :
3898 : : ////////////////////////////////////////////////////////////////////////
3899 : :
3900 : :
3901 : :
3902 : : // The thread that consumes file names off of the scanq. We hold
3903 : : // the persistent sqlite_ps's at this level and delegate file/archive
3904 : : // scanning to other functions.
3905 : : static void
3906 : 264 : scan ()
3907 : : {
3908 : : // all the prepared statements fit to use, the _f_ set:
3909 [ + - + - : 528 : sqlite_ps ps_f_upsert_buildids (db, "file-buildids-intern", "insert or ignore into " BUILDIDS "_buildids VALUES (NULL, ?);");
+ - - - ]
3910 [ + - + - : 528 : sqlite_ps ps_f_upsert_fileparts (db, "file-fileparts-intern", "insert or ignore into " BUILDIDS "_fileparts VALUES (NULL, ?);");
+ - + - -
- ]
3911 [ + - ]: 264 : sqlite_ps ps_f_upsert_file (db, "file-file-intern", "insert or ignore into " BUILDIDS "_files VALUES (NULL, \n"
3912 : : "(select id from " BUILDIDS "_fileparts where name = ?),\n"
3913 [ + - + - : 528 : "(select id from " BUILDIDS "_fileparts where name = ?));");
+ - + - -
- ]
3914 [ + - ]: 264 : sqlite_ps ps_f_lookup_file (db, "file-file-lookup",
3915 : : "select f.id\n"
3916 : : " from " BUILDIDS "_files f, " BUILDIDS "_fileparts p1, " BUILDIDS "_fileparts p2 \n"
3917 [ + - + - : 528 : " where f.dirname = p1.id and f.basename = p2.id and p1.name = ? and p2.name = ?;\n");
+ - + - -
- ]
3918 [ + - ]: 264 : sqlite_ps ps_f_upsert_de (db, "file-de-upsert",
3919 : : "insert or ignore into " BUILDIDS "_f_de "
3920 : : "(buildid, debuginfo_p, executable_p, file, mtime) "
3921 : : "values ((select id from " BUILDIDS "_buildids where hex = ?),"
3922 [ + - + - : 528 : " ?,?,?,?);");
+ - + - -
- ]
3923 [ + - ]: 264 : sqlite_ps ps_f_upsert_s (db, "file-s-upsert",
3924 : : "insert or ignore into " BUILDIDS "_f_s "
3925 : : "(buildid, artifactsrc, file, mtime) "
3926 : : "values ((select id from " BUILDIDS "_buildids where hex = ?),"
3927 [ + - + - : 528 : " ?,?,?);");
+ - + - -
- ]
3928 [ + - ]: 264 : sqlite_ps ps_f_query (db, "file-negativehit-find",
3929 : : "select 1 from " BUILDIDS "_file_mtime_scanned where sourcetype = 'F' "
3930 [ + - + - : 528 : "and file = ? and mtime = ?;");
+ - + - -
- ]
3931 [ + - ]: 264 : sqlite_ps ps_f_scan_done (db, "file-scanned",
3932 : : "insert or ignore into " BUILDIDS "_file_mtime_scanned (sourcetype, file, mtime, size)"
3933 [ + - + - : 528 : "values ('F', ?,?,?);");
+ - + - -
- ]
3934 : :
3935 : : // and now for the _r_ set
3936 [ + - + - : 528 : sqlite_ps ps_r_upsert_buildids (db, "rpm-buildid-intern", "insert or ignore into " BUILDIDS "_buildids VALUES (NULL, ?);");
+ - + - -
- ]
3937 [ + - + - : 528 : sqlite_ps ps_r_upsert_fileparts (db, "rpm-fileparts-intern", "insert or ignore into " BUILDIDS "_fileparts VALUES (NULL, ?);");
+ - + - -
- ]
3938 [ + - ]: 264 : sqlite_ps ps_r_upsert_file (db, "rpm-file-intern", "insert or ignore into " BUILDIDS "_files VALUES (NULL, \n"
3939 : : "(select id from " BUILDIDS "_fileparts where name = ?),\n"
3940 [ + - + - : 528 : "(select id from " BUILDIDS "_fileparts where name = ?));");
+ - + - -
- ]
3941 [ + - ]: 264 : sqlite_ps ps_r_lookup_file (db, "rpm-file-lookup",
3942 : : "select f.id\n"
3943 : : " from " BUILDIDS "_files f, " BUILDIDS "_fileparts p1, " BUILDIDS "_fileparts p2 \n"
3944 [ + - + - : 528 : " where f.dirname = p1.id and f.basename = p2.id and p1.name = ? and p2.name = ?;\n");
+ - + - -
- ]
3945 [ + - ]: 264 : sqlite_ps ps_r_upsert_de (db, "rpm-de-insert",
3946 : : "insert or ignore into " BUILDIDS "_r_de (buildid, debuginfo_p, executable_p, file, mtime, content) values ("
3947 [ + - + - : 528 : "(select id from " BUILDIDS "_buildids where hex = ?), ?, ?, ?, ?, ?);");
+ - + - -
- ]
3948 [ + - ]: 264 : sqlite_ps ps_r_upsert_sref (db, "rpm-sref-insert",
3949 : : "insert or ignore into " BUILDIDS "_r_sref (buildid, artifactsrc) values ("
3950 : : "(select id from " BUILDIDS "_buildids where hex = ?), "
3951 [ + - + - : 528 : "?);");
+ - + - -
- ]
3952 [ + - ]: 264 : sqlite_ps ps_r_upsert_sdef (db, "rpm-sdef-insert",
3953 : : "insert or ignore into " BUILDIDS "_r_sdef (file, mtime, content) values ("
3954 [ + - + - : 528 : "?, ?, ?);");
+ - + - -
- ]
3955 [ + - ]: 264 : sqlite_ps ps_r_query (db, "rpm-negativehit-query",
3956 : : "select 1 from " BUILDIDS "_file_mtime_scanned where "
3957 [ + - + - : 528 : "sourcetype = 'R' and file = ? and mtime = ?;");
+ - + - -
- ]
3958 [ + - ]: 264 : sqlite_ps ps_r_scan_done (db, "rpm-scanned",
3959 : : "insert or ignore into " BUILDIDS "_file_mtime_scanned (sourcetype, file, mtime, size)"
3960 [ + - + - : 528 : "values ('R', ?, ?, ?);");
+ - + - -
- ]
3961 : :
3962 : :
3963 : 264 : unsigned fts_cached = 0, fts_executable = 0, fts_debuginfo = 0, fts_sourcefiles = 0;
3964 : 264 : unsigned fts_sref = 0, fts_sdef = 0;
3965 : :
3966 [ + - + - : 528 : add_metric("thread_count", "role", "scan", 1);
+ - + - -
+ - + - -
- - ]
3967 [ + - + - : 528 : add_metric("thread_busy", "role", "scan", 1);
+ - + - -
+ - + - -
- - ]
3968 [ + + ]: 1714 : while (! interrupted)
3969 : : {
3970 [ + - ]: 1451 : scan_payload p;
3971 : :
3972 [ + - + - : 2901 : add_metric("thread_busy", "role", "scan", -1);
+ - + - -
+ - + - -
- - ]
3973 : : // NB: threads may be blocked within either of these two waiting
3974 : : // states, if the work queue happens to run dry. That's OK.
3975 [ + - + - ]: 1450 : if (scan_barrier) scan_barrier->count();
3976 [ + - ]: 1450 : bool gotone = scanq.wait_front(p);
3977 [ + - + - : 2900 : add_metric("thread_busy", "role", "scan", 1);
+ - + - -
+ - + - -
- - ]
3978 : :
3979 [ + + - + ]: 1450 : if (! gotone) continue; // go back to waiting
3980 : :
3981 : 1186 : try
3982 : : {
3983 : 1186 : bool scan_archive = false;
3984 [ + + ]: 2486 : for (auto&& arch : scan_archives)
3985 [ + + ]: 1300 : if (string_endswith(p.first, arch.first))
3986 : 702 : scan_archive = true;
3987 : :
3988 [ + + ]: 1186 : if (scan_archive)
3989 [ + - ]: 702 : scan_archive_file (p.first, p.second,
3990 : : ps_r_upsert_buildids,
3991 : : ps_r_upsert_fileparts,
3992 : : ps_r_upsert_file,
3993 : : ps_r_lookup_file,
3994 : : ps_r_upsert_de,
3995 : : ps_r_upsert_sref,
3996 : : ps_r_upsert_sdef,
3997 : : ps_r_query,
3998 : : ps_r_scan_done,
3999 : : fts_cached,
4000 : : fts_executable,
4001 : : fts_debuginfo,
4002 : : fts_sref,
4003 : : fts_sdef);
4004 : :
4005 [ + + ]: 1186 : if (scan_files) // NB: maybe "else if" ?
4006 [ + - ]: 974 : scan_source_file (p.first, p.second,
4007 : : ps_f_upsert_buildids,
4008 : : ps_f_upsert_fileparts,
4009 : : ps_f_upsert_file,
4010 : : ps_f_lookup_file,
4011 : : ps_f_upsert_de,
4012 : : ps_f_upsert_s,
4013 : : ps_f_query,
4014 : : ps_f_scan_done,
4015 : : fts_cached, fts_executable, fts_debuginfo, fts_sourcefiles);
4016 : : }
4017 [ - - ]: 0 : catch (const reportable_exception& e)
4018 : : {
4019 [ - - ]: 0 : e.report(cerr);
4020 : 0 : }
4021 : :
4022 [ + - ]: 1186 : scanq.done_front(); // let idlers run
4023 : :
4024 : 1186 : if (fts_cached || fts_executable || fts_debuginfo || fts_sourcefiles || fts_sref || fts_sdef)
4025 : : {} // NB: not just if a successful scan - we might have encountered -ENOSPC & failed
4026 [ + - + - ]: 1186 : (void) statfs_free_enough_p(db_path, "database"); // report sqlite filesystem size
4027 [ + - + - ]: 1186 : (void) statfs_free_enough_p(tmpdir, "tmpdir"); // this too, in case of fdcache/tmpfile usage
4028 : :
4029 : : // finished a scanning step -- not a "loop", because we just
4030 : : // consume the traversal loop's work, whenever
4031 [ + - + - : 2372 : inc_metric("thread_work_total","role","scan");
+ - + - -
+ - + + -
- - - - -
- ]
4032 : 1450 : }
4033 : :
4034 [ + - + - : 527 : add_metric("thread_busy", "role", "scan", -1);
+ - + - -
+ - + - -
- - ]
4035 : 264 : }
4036 : :
4037 : :
4038 : : // Use this function as the thread entry point, so it can catch our
4039 : : // fleet of exceptions (incl. the sqlite_ps ctors) and report.
4040 : : static void*
4041 : 264 : thread_main_scanner (void* arg)
4042 : : {
4043 : 264 : (void) arg;
4044 [ + + ]: 792 : while (! interrupted)
4045 : 264 : try
4046 : : {
4047 [ + - ]: 264 : scan();
4048 : : }
4049 [ - - ]: 0 : catch (const reportable_exception& e)
4050 : : {
4051 [ - - ]: 0 : e.report(cerr);
4052 : 0 : }
4053 : 264 : return 0;
4054 : : }
4055 : :
4056 : :
4057 : :
4058 : : // The thread that traverses all the source_paths and enqueues all the
4059 : : // matching files into the file/archive scan queue.
4060 : : static void
4061 : 120 : scan_source_paths()
4062 : : {
4063 : : // NB: fedora 31 glibc/fts(3) crashes inside fts_read() on empty
4064 : : // path list.
4065 [ + + ]: 120 : if (source_paths.empty())
4066 : 2 : return;
4067 : :
4068 : : // Turn the source_paths into an fts(3)-compatible char**. Since
4069 : : // source_paths[] does not change after argv processing, the
4070 : : // c_str()'s are safe to keep around awile.
4071 : 118 : vector<const char *> sps;
4072 [ + + ]: 318 : for (auto&& sp: source_paths)
4073 [ + - ]: 200 : sps.push_back(sp.c_str());
4074 [ + - - - ]: 118 : sps.push_back(NULL);
4075 : :
4076 [ + + + - ]: 222 : FTS *fts = fts_open ((char * const *)sps.data(),
4077 : : (traverse_logical ? FTS_LOGICAL : FTS_PHYSICAL|FTS_XDEV)
4078 : : | FTS_NOCHDIR /* multithreaded */,
4079 : : NULL);
4080 [ - + ]: 118 : if (fts == NULL)
4081 [ # # # # ]: 0 : throw libc_exception(errno, "cannot fts_open");
4082 : 118 : defer_dtor<FTS*,int> fts_cleanup (fts, fts_close);
4083 : :
4084 : 118 : struct timespec ts_start, ts_end;
4085 : 118 : clock_gettime (CLOCK_MONOTONIC, &ts_start);
4086 : 118 : unsigned fts_scanned = 0, fts_regex = 0;
4087 : :
4088 : 118 : FTSENT *f;
4089 [ + - + + ]: 2374 : while ((f = fts_read (fts)) != NULL)
4090 : : {
4091 [ + - ]: 2138 : if (interrupted) break;
4092 : :
4093 [ - + ]: 2138 : if (sigusr2 != forced_groom_count) // stop early if groom triggered
4094 : : {
4095 [ # # ]: 0 : scanq.clear(); // clear previously issued work for scanner threads
4096 : : break;
4097 : : }
4098 : :
4099 : 2138 : fts_scanned ++;
4100 : :
4101 [ + - ]: 2138 : if (verbose > 2)
4102 [ + - + - : 4276 : obatched(clog) << "fts traversing " << f->fts_path << endl;
+ - ]
4103 : :
4104 [ + + + + : 2138 : switch (f->fts_info)
+ ]
4105 : : {
4106 : 1274 : case FTS_F:
4107 : 1274 : {
4108 : : /* Found a file. Convert it to an absolute path, so
4109 : : the buildid database does not have relative path
4110 : : names that are unresolvable from a subsequent run
4111 : : in a different cwd. */
4112 [ + - ]: 1274 : char *rp = realpath(f->fts_path, NULL);
4113 [ - + ]: 1274 : if (rp == NULL)
4114 : 0 : continue; // ignore dangling symlink or such
4115 [ + - ]: 1274 : string rps = string(rp);
4116 : 1274 : free (rp);
4117 : :
4118 [ + - ]: 1274 : bool ri = !regexec (&file_include_regex, rps.c_str(), 0, 0, 0);
4119 [ + - ]: 1274 : bool rx = !regexec (&file_exclude_regex, rps.c_str(), 0, 0, 0);
4120 [ + + ]: 1274 : if (!ri || rx)
4121 : : {
4122 [ + - ]: 88 : if (verbose > 3)
4123 [ + - ]: 176 : obatched(clog) << "fts skipped by regex "
4124 [ + + + - : 96 : << (!ri ? "I" : "") << (rx ? "X" : "") << endl;
+ + + - +
- ]
4125 : 88 : fts_regex ++;
4126 [ + + ]: 88 : if (!ri)
4127 [ + - + - : 8 : inc_metric("traversed_total","type","file-skipped-I");
+ - + - -
+ - + - -
- - ]
4128 [ + + ]: 88 : if (rx)
4129 [ + - + - : 168 : inc_metric("traversed_total","type","file-skipped-X");
+ - + - -
+ - + - -
- - ]
4130 : : }
4131 : : else
4132 : : {
4133 [ + - + - ]: 1186 : scanq.push_back (make_pair(rps, *f->fts_statp));
4134 [ + - + - : 2372 : inc_metric("traversed_total","type","file");
+ - + - -
+ - + - -
- - - - ]
4135 : : }
4136 : 0 : }
4137 : 1274 : break;
4138 : :
4139 : 4 : case FTS_ERR:
4140 : 4 : case FTS_NS:
4141 : : // report on some types of errors because they may reflect fixable misconfiguration
4142 : 4 : {
4143 [ + - + - : 8 : auto x = libc_exception(f->fts_errno, string("fts traversal ") + string(f->fts_path));
+ - + - -
+ - + -
- ]
4144 [ + - ]: 4 : x.report(cerr);
4145 : 0 : }
4146 [ + - + - : 8 : inc_metric("traversed_total","type","error");
+ - + - -
+ - + - -
- - ]
4147 : 4 : break;
4148 : :
4149 : 12 : case FTS_SL: // ignore, but count because debuginfod -L would traverse these
4150 [ + - + - : 24 : inc_metric("traversed_total","type","symlink");
+ - + - -
+ - + - -
- - ]
4151 : 12 : break;
4152 : :
4153 : 424 : case FTS_D: // ignore
4154 [ + - + - : 848 : inc_metric("traversed_total","type","directory");
+ - + - -
+ - + - -
- - ]
4155 : 424 : break;
4156 : :
4157 : 424 : default: // ignore
4158 [ + - + - : 848 : inc_metric("traversed_total","type","other");
+ - + - -
+ - + - -
- - ]
4159 : 424 : break;
4160 : : }
4161 : : }
4162 : 118 : clock_gettime (CLOCK_MONOTONIC, &ts_end);
4163 : 118 : double deltas = (ts_end.tv_sec - ts_start.tv_sec) + (ts_end.tv_nsec - ts_start.tv_nsec)/1.e9;
4164 : :
4165 [ + - + - : 354 : obatched(clog) << "fts traversed source paths in " << deltas << "s, scanned=" << fts_scanned
+ - + - ]
4166 [ + - + - : 118 : << ", regex-skipped=" << fts_regex << endl;
+ - ]
4167 [ + - ]: 236 : }
4168 : :
4169 : :
4170 : : static void*
4171 : 66 : thread_main_fts_source_paths (void* arg)
4172 : : {
4173 : 66 : (void) arg; // ignore; we operate on global data
4174 : :
4175 [ + - + - : 132 : set_metric("thread_tid", "role","traverse", tid());
+ - - + -
+ - - -
- ]
4176 [ + - + - : 132 : add_metric("thread_count", "role", "traverse", 1);
+ - - + -
+ - - -
- ]
4177 : :
4178 : 66 : time_t last_rescan = 0;
4179 : :
4180 [ + - ]: 283 : while (! interrupted)
4181 : : {
4182 : 283 : sleep (1);
4183 : 283 : scanq.wait_idle(); // don't start a new traversal while scanners haven't finished the job
4184 : 283 : scanq.done_idle(); // release the hounds
4185 [ + + ]: 283 : if (interrupted) break;
4186 : :
4187 : 217 : time_t now = time(NULL);
4188 : 217 : bool rescan_now = false;
4189 [ + + ]: 217 : if (last_rescan == 0) // at least one initial rescan is documented even for -t0
4190 : 64 : rescan_now = true;
4191 [ + + + + ]: 217 : if (rescan_s > 0 && (long)now > (long)(last_rescan + rescan_s))
4192 : 217 : rescan_now = true;
4193 [ + + ]: 217 : if (sigusr1 != forced_rescan_count)
4194 : : {
4195 : 60 : forced_rescan_count = sigusr1;
4196 : 60 : rescan_now = true;
4197 : : }
4198 [ + + ]: 217 : if (rescan_now)
4199 : : {
4200 [ + - + - : 240 : set_metric("thread_busy", "role","traverse", 1);
+ - - + -
+ - - -
- ]
4201 : 120 : try
4202 : : {
4203 [ + - ]: 120 : scan_source_paths();
4204 : : }
4205 [ - - ]: 0 : catch (const reportable_exception& e)
4206 : : {
4207 [ - - ]: 0 : e.report(cerr);
4208 : 0 : }
4209 : 120 : last_rescan = time(NULL); // NB: now was before scanning
4210 : : // finished a traversal loop
4211 [ + - + - : 240 : inc_metric("thread_work_total", "role","traverse");
+ - - + -
+ - - -
- ]
4212 [ + - + - : 240 : set_metric("thread_busy", "role","traverse", 0);
+ - - + -
+ - - -
- ]
4213 : : }
4214 : : }
4215 : :
4216 : 66 : return 0;
4217 : : }
4218 : :
4219 : :
4220 : :
4221 : : ////////////////////////////////////////////////////////////////////////
4222 : :
4223 : : static void
4224 : 72 : database_stats_report()
4225 : : {
4226 : 72 : sqlite_ps ps_query (db, "database-overview",
4227 [ + - + - : 144 : "select label,quantity from " BUILDIDS "_stats");
+ - - - ]
4228 : :
4229 [ + - + - ]: 144 : obatched(clog) << "database record counts:" << endl;
4230 : 1656 : while (1)
4231 : : {
4232 [ + - ]: 864 : if (interrupted) break;
4233 [ + - ]: 864 : if (sigusr1 != forced_rescan_count) // stop early if scan triggered
4234 : : break;
4235 : :
4236 [ + - ]: 864 : int rc = ps_query.step();
4237 [ + + ]: 864 : if (rc == SQLITE_DONE) break;
4238 [ - + ]: 792 : if (rc != SQLITE_ROW)
4239 [ # # # # ]: 0 : throw sqlite_exception(rc, "step");
4240 : :
4241 [ + - ]: 792 : obatched(clog)
4242 [ + - - + : 792 : << ((const char*) sqlite3_column_text(ps_query, 0) ?: (const char*) "NULL")
+ - ]
4243 : : << " "
4244 [ + - + - : 1584 : << (sqlite3_column_text(ps_query, 1) ?: (const unsigned char*) "NULL")
- + + - ]
4245 : 792 : << endl;
4246 : :
4247 [ + - + - : 1584 : set_metric("groom", "statistic",
- + + - +
- + - + -
- + + + -
- - - ]
4248 [ + - ]: 792 : ((const char*) sqlite3_column_text(ps_query, 0) ?: (const char*) "NULL"),
4249 : : (sqlite3_column_double(ps_query, 1)));
4250 : 792 : }
4251 : 72 : }
4252 : :
4253 : :
4254 : : // Do a round of database grooming that might take many minutes to run.
4255 : 72 : void groom()
4256 : : {
4257 [ + - ]: 144 : obatched(clog) << "grooming database" << endl;
4258 : :
4259 : 72 : struct timespec ts_start, ts_end;
4260 : 72 : clock_gettime (CLOCK_MONOTONIC, &ts_start);
4261 : :
4262 : : // scan for files that have disappeared
4263 : 72 : sqlite_ps files (db, "check old files",
4264 : : "select distinct s.mtime, s.file, f.name from "
4265 : : BUILDIDS "_file_mtime_scanned s, " BUILDIDS "_files_v f "
4266 [ + - + - : 144 : "where f.id = s.file");
+ - - - ]
4267 : : // NB: Because _ftime_mtime_scanned can contain both F and
4268 : : // R records for the same file, this query would return duplicates if the
4269 : : // DISTINCT qualifier were not there.
4270 [ + - ]: 72 : files.reset();
4271 : :
4272 : : // DECISION TIME - we enumerate stale fileids/mtimes
4273 [ + - ]: 72 : deque<pair<int64_t,int64_t> > stale_fileid_mtime;
4274 : :
4275 : 72 : time_t time_start = time(NULL);
4276 : 304 : while(1)
4277 : : {
4278 : : // PR28514: limit grooming iteration to O(rescan time), to avoid
4279 : : // slow filesystem tests over many files locking out rescans for
4280 : : // too long.
4281 [ + + - + ]: 188 : if (rescan_s > 0 && (long)time(NULL) > (long)(time_start + rescan_s))
4282 : : {
4283 [ # # # # : 0 : inc_metric("groomed_total", "decision", "aborted");
# # # # #
# # # # #
# # ]
4284 : 0 : break;
4285 : : }
4286 : :
4287 [ + - ]: 188 : if (interrupted) break;
4288 : :
4289 [ + - ]: 188 : int rc = files.step();
4290 [ + + ]: 188 : if (rc != SQLITE_ROW)
4291 : : break;
4292 : :
4293 [ + - ]: 116 : int64_t mtime = sqlite3_column_int64 (files, 0);
4294 [ + - ]: 116 : int64_t fileid = sqlite3_column_int64 (files, 1);
4295 [ + - - + ]: 116 : const char* filename = ((const char*) sqlite3_column_text (files, 2) ?: "");
4296 : 116 : struct stat s;
4297 : 116 : bool regex_file_drop = 0;
4298 : :
4299 [ + + ]: 116 : if (regex_groom)
4300 : : {
4301 [ + - ]: 16 : bool reg_include = !regexec (&file_include_regex, filename, 0, 0, 0);
4302 [ + - ]: 16 : bool reg_exclude = !regexec (&file_exclude_regex, filename, 0, 0, 0);
4303 : 16 : regex_file_drop = !reg_include || reg_exclude; // match logic of scan_source_paths
4304 : : }
4305 : :
4306 : 116 : rc = stat(filename, &s);
4307 [ + + - + ]: 116 : if ( regex_file_drop || rc < 0 || (mtime != (int64_t) s.st_mtime) )
4308 : : {
4309 [ + - ]: 24 : if (verbose > 2)
4310 [ + - + - : 48 : obatched(clog) << "groom: stale file=" << filename << " mtime=" << mtime << endl;
+ - + - +
- ]
4311 [ + - ]: 24 : stale_fileid_mtime.push_back(make_pair(fileid,mtime));
4312 [ + - + - : 48 : inc_metric("groomed_total", "decision", "stale");
+ - + - -
+ - + - -
- - ]
4313 [ + - + - : 48 : set_metric("thread_work_pending","role","groom", stale_fileid_mtime.size());
+ - + - -
+ - + - -
- - ]
4314 : : }
4315 : : else
4316 [ + - + - : 184 : inc_metric("groomed_total", "decision", "fresh");
+ - + - -
+ - + - -
- - ]
4317 : :
4318 [ + - ]: 116 : if (sigusr1 != forced_rescan_count) // stop early if scan triggered
4319 : : break;
4320 : 116 : }
4321 [ + - ]: 72 : files.reset();
4322 : :
4323 : : // ACTION TIME
4324 : :
4325 : : // Now that we know which file/mtime tuples are stale, actually do
4326 : : // the deletion from the database. Doing this during the SELECT
4327 : : // iteration above results in undefined behaviour in sqlite, as per
4328 : : // https://www.sqlite.org/isolation.html
4329 : :
4330 : : // We could shuffle stale_fileid_mtime[] here. It'd let aborted
4331 : : // sequences of nuke operations resume at random locations, instead
4332 : : // of just starting over. But it doesn't matter much either way,
4333 : : // as long as we make progress.
4334 : :
4335 [ + - + - : 144 : sqlite_ps files_del_f_de (db, "nuke f_de", "delete from " BUILDIDS "_f_de where file = ? and mtime = ?");
+ - + - -
- ]
4336 [ + - + - : 144 : sqlite_ps files_del_r_de (db, "nuke r_de", "delete from " BUILDIDS "_r_de where file = ? and mtime = ?");
+ - + - -
- ]
4337 [ + - ]: 72 : sqlite_ps files_del_scan (db, "nuke f_m_s", "delete from " BUILDIDS "_file_mtime_scanned "
4338 [ + - + - : 144 : "where file = ? and mtime = ?");
+ - + - -
- ]
4339 : :
4340 [ + + ]: 96 : while (! stale_fileid_mtime.empty())
4341 : : {
4342 : 24 : auto stale = stale_fileid_mtime.front();
4343 : 24 : stale_fileid_mtime.pop_front();
4344 [ + - + - : 48 : set_metric("thread_work_pending","role","groom", stale_fileid_mtime.size());
+ - + - -
+ - + - -
- - ]
4345 : :
4346 : : // PR28514: limit grooming iteration to O(rescan time), to avoid
4347 : : // slow nuke_* queries over many files locking out rescans for too
4348 : : // long. We iterate over the files in random() sequence to avoid
4349 : : // partial checks going over the same set.
4350 [ - + - - ]: 24 : if (rescan_s > 0 && (long)time(NULL) > (long)(time_start + rescan_s))
4351 : : {
4352 [ # # # # : 0 : inc_metric("groomed_total", "action", "aborted");
# # # # #
# # # # #
# # ]
4353 : 0 : break;
4354 : : }
4355 : :
4356 [ + - ]: 24 : if (interrupted) break;
4357 : :
4358 : 24 : int64_t fileid = stale.first;
4359 : 24 : int64_t mtime = stale.second;
4360 [ + - + - : 24 : files_del_f_de.reset().bind(1,fileid).bind(2,mtime).step_ok_done();
+ - + - ]
4361 [ + - + - : 24 : files_del_r_de.reset().bind(1,fileid).bind(2,mtime).step_ok_done();
+ - + - ]
4362 [ + - + - : 24 : files_del_scan.reset().bind(1,fileid).bind(2,mtime).step_ok_done();
+ - + - ]
4363 [ + - + - : 48 : inc_metric("groomed_total", "action", "cleaned");
+ - + - -
+ - + - -
- - ]
4364 : :
4365 [ + - ]: 24 : if (sigusr1 != forced_rescan_count) // stop early if scan triggered
4366 : : break;
4367 : : }
4368 : 72 : stale_fileid_mtime.clear(); // no need for this any longer
4369 [ + - + - : 144 : set_metric("thread_work_pending","role","groom", stale_fileid_mtime.size());
+ - + - -
+ - + - -
- - ]
4370 : :
4371 : : // delete buildids with no references in _r_de or _f_de tables;
4372 : : // cascades to _r_sref & _f_s records
4373 [ + - ]: 72 : sqlite_ps buildids_del (db, "nuke orphan buildids",
4374 : : "delete from " BUILDIDS "_buildids "
4375 : : "where not exists (select 1 from " BUILDIDS "_f_de d where " BUILDIDS "_buildids.id = d.buildid) "
4376 [ + - + - : 144 : "and not exists (select 1 from " BUILDIDS "_r_de d where " BUILDIDS "_buildids.id = d.buildid)");
+ - + - -
- ]
4377 [ + - + - ]: 72 : buildids_del.reset().step_ok_done();
4378 : :
4379 [ - + ]: 72 : if (interrupted) return;
4380 : :
4381 : : // NB: "vacuum" is too heavy for even daily runs: it rewrites the entire db, so is done as maxigroom -G
4382 [ + - + - : 144 : sqlite_ps g1 (db, "incremental vacuum", "pragma incremental_vacuum");
+ - + - -
- ]
4383 [ + - + - ]: 72 : g1.reset().step_ok_done();
4384 [ + - + - : 144 : sqlite_ps g2 (db, "optimize", "pragma optimize");
+ - - + -
- ]
4385 [ + - + - ]: 72 : g2.reset().step_ok_done();
4386 [ + - + - : 144 : sqlite_ps g3 (db, "wal checkpoint", "pragma wal_checkpoint=truncate");
+ - + - -
- ]
4387 [ + - + - ]: 72 : g3.reset().step_ok_done();
4388 : :
4389 [ + - ]: 72 : database_stats_report();
4390 : :
4391 [ + - + - ]: 72 : (void) statfs_free_enough_p(db_path, "database"); // report sqlite filesystem size
4392 : :
4393 [ + - ]: 72 : sqlite3_db_release_memory(db); // shrink the process if possible
4394 [ + - ]: 72 : sqlite3_db_release_memory(dbq); // ... for both connections
4395 [ + - ]: 72 : debuginfod_pool_groom(); // and release any debuginfod_client objects we've been holding onto
4396 : : #if HAVE_MALLOC_TRIM
4397 : 72 : malloc_trim(0); // PR31103: release memory allocated for temporary purposes
4398 : : #endif
4399 : :
4400 : : #if 0 /* PR31265: don't jettison cache unnecessarily */
4401 : : fdcache.limit(0); // release the fdcache contents
4402 : : fdcache.limit(fdcache_mbs); // restore status quo parameters
4403 : : #endif
4404 : :
4405 : 72 : clock_gettime (CLOCK_MONOTONIC, &ts_end);
4406 : 72 : double deltas = (ts_end.tv_sec - ts_start.tv_sec) + (ts_end.tv_nsec - ts_start.tv_nsec)/1.e9;
4407 : :
4408 [ + - + - : 144 : obatched(clog) << "groomed database in " << deltas << "s" << endl;
+ - + - ]
4409 : 72 : }
4410 : :
4411 : :
4412 : : static void*
4413 : 72 : thread_main_groom (void* /*arg*/)
4414 : : {
4415 [ + - + - : 144 : set_metric("thread_tid", "role", "groom", tid());
+ - - + -
+ - - -
- ]
4416 [ + - + - : 144 : add_metric("thread_count", "role", "groom", 1);
+ - - + -
+ - - -
- ]
4417 : :
4418 : 72 : time_t last_groom = 0;
4419 : :
4420 : 510 : while (1)
4421 : : {
4422 : 291 : sleep (1);
4423 : 291 : scanq.wait_idle(); // PR25394: block scanners during grooming!
4424 [ + + ]: 291 : if (interrupted) break;
4425 : :
4426 : 219 : time_t now = time(NULL);
4427 : 219 : bool groom_now = false;
4428 [ + + ]: 219 : if (last_groom == 0) // at least one initial groom is documented even for -g0
4429 : 66 : groom_now = true;
4430 [ + + + + ]: 219 : if (groom_s > 0 && (long)now > (long)(last_groom + groom_s))
4431 : 219 : groom_now = true;
4432 [ + + ]: 219 : if (sigusr2 != forced_groom_count)
4433 : : {
4434 : 6 : forced_groom_count = sigusr2;
4435 : 6 : groom_now = true;
4436 : : }
4437 [ + + ]: 219 : if (groom_now)
4438 : : {
4439 [ + - + - : 144 : set_metric("thread_busy", "role", "groom", 1);
+ - - + -
+ - - -
- ]
4440 : 72 : try
4441 : : {
4442 [ + - ]: 72 : groom ();
4443 : : }
4444 [ - - ]: 0 : catch (const sqlite_exception& e)
4445 : : {
4446 [ - - - - : 0 : obatched(cerr) << e.message << endl;
- - ]
4447 : 0 : }
4448 : 72 : last_groom = time(NULL); // NB: now was before grooming
4449 : : // finished a grooming loop
4450 [ + - + - : 144 : inc_metric("thread_work_total", "role", "groom");
+ - - + -
+ - - -
- ]
4451 [ + - + - : 144 : set_metric("thread_busy", "role", "groom", 0);
+ - - + -
+ - - -
- ]
4452 : : }
4453 : :
4454 : 219 : scanq.done_idle();
4455 : 219 : }
4456 : :
4457 : 72 : return 0;
4458 : : }
4459 : :
4460 : :
4461 : : ////////////////////////////////////////////////////////////////////////
4462 : :
4463 : :
4464 : : static void
4465 : 74 : signal_handler (int /* sig */)
4466 : : {
4467 : 74 : interrupted ++;
4468 : :
4469 [ + + ]: 74 : if (db)
4470 : 72 : sqlite3_interrupt (db);
4471 [ + - ]: 74 : if (dbq)
4472 : 74 : sqlite3_interrupt (dbq);
4473 : :
4474 : : // NB: don't do anything else in here
4475 : 74 : }
4476 : :
4477 : : static void
4478 : 60 : sigusr1_handler (int /* sig */)
4479 : : {
4480 : 60 : sigusr1 ++;
4481 : : // NB: don't do anything else in here
4482 : 60 : }
4483 : :
4484 : : static void
4485 : 6 : sigusr2_handler (int /* sig */)
4486 : : {
4487 : 6 : sigusr2 ++;
4488 : : // NB: don't do anything else in here
4489 : 6 : }
4490 : :
4491 : :
4492 : : static void // error logging callback from libmicrohttpd internals
4493 : 0 : error_cb (void *arg, const char *fmt, va_list ap)
4494 : : {
4495 : 0 : (void) arg;
4496 [ # # # # : 0 : inc_metric("error_count","libmicrohttpd",fmt);
# # # # #
# # # #
# ]
4497 : 0 : char errmsg[512];
4498 : 0 : (void) vsnprintf (errmsg, sizeof(errmsg), fmt, ap); // ok if slightly truncated
4499 [ # # ]: 0 : obatched(cerr) << "libmicrohttpd error: " << errmsg; // MHD_DLOG calls already include \n
4500 : 0 : }
4501 : :
4502 : :
4503 : : // A user-defined sqlite function, to score the sharedness of the
4504 : : // prefix of two strings. This is used to compare candidate debuginfo
4505 : : // / source-rpm names, so that the closest match
4506 : : // (directory-topology-wise closest) is found. This is important in
4507 : : // case the same sref (source file name) is in many -debuginfo or
4508 : : // -debugsource RPMs, such as when multiple versions/releases of the
4509 : : // same package are in the database.
4510 : :
4511 : 817 : static void sqlite3_sharedprefix_fn (sqlite3_context* c, int argc, sqlite3_value** argv)
4512 : : {
4513 [ - + ]: 817 : if (argc != 2)
4514 : 0 : sqlite3_result_error(c, "expect 2 string arguments", -1);
4515 [ + - + + ]: 1634 : else if ((sqlite3_value_type(argv[0]) != SQLITE_TEXT) ||
4516 : 817 : (sqlite3_value_type(argv[1]) != SQLITE_TEXT))
4517 : 527 : sqlite3_result_null(c);
4518 : : else
4519 : : {
4520 : 290 : const unsigned char* a = sqlite3_value_text (argv[0]);
4521 : 290 : const unsigned char* b = sqlite3_value_text (argv[1]);
4522 : 290 : int i = 0;
4523 [ + + + - : 35878 : while (*a != '\0' && *b != '\0' && *a++ == *b++)
+ + + + ]
4524 : 35048 : i++;
4525 : 290 : sqlite3_result_int (c, i);
4526 : : }
4527 : 817 : }
4528 : :
4529 : :
4530 : : static unsigned
4531 : 144 : default_concurrency() // guaranteed >= 1
4532 : : {
4533 : : // Prior to PR29975 & PR29976, we'd just use this:
4534 : 144 : unsigned sth = std::thread::hardware_concurrency();
4535 : : // ... but on many-CPU boxes, admins or distros may throttle
4536 : : // resources in such a way that debuginfod would mysteriously fail.
4537 : : // So we reduce the defaults:
4538 : :
4539 : 144 : unsigned aff = 0;
4540 : : #ifdef HAVE_SCHED_GETAFFINITY
4541 : 144 : {
4542 : 144 : int ret;
4543 : 144 : cpu_set_t mask;
4544 : 144 : CPU_ZERO(&mask);
4545 : 144 : ret = sched_getaffinity(0, sizeof(mask), &mask);
4546 [ + - ]: 144 : if (ret == 0)
4547 : 144 : aff = CPU_COUNT(&mask);
4548 : : }
4549 : : #endif
4550 : :
4551 : 144 : unsigned fn = 0;
4552 : : #ifdef HAVE_GETRLIMIT
4553 : 144 : {
4554 : 144 : struct rlimit rlim;
4555 : 144 : int rc = getrlimit(RLIMIT_NOFILE, &rlim);
4556 [ + - ]: 144 : if (rc == 0)
4557 [ + - ]: 288 : fn = max((rlim_t)1, (rlim.rlim_cur - 100) / 4);
4558 : : // at least 2 fds are used by each listener thread etc.
4559 : : // plus a bunch to account for shared libraries and such
4560 : : }
4561 : : #endif
4562 : :
4563 [ - + - + : 144 : unsigned d = min(max(sth, 1U),
- + ]
4564 [ - + ]: 144 : min(max(aff, 1U),
4565 [ - + ]: 144 : max(fn, 1U)));
4566 : 144 : return d;
4567 : : }
4568 : :
4569 : :
4570 : : // 30879: Something to help out in case of an uncaught exception.
4571 : 0 : void my_terminate_handler()
4572 : : {
4573 : : #if defined(__GLIBC__)
4574 : 0 : void *array[40];
4575 : 0 : int size = backtrace (array, 40);
4576 : 0 : backtrace_symbols_fd (array, size, STDERR_FILENO);
4577 : : #endif
4578 : : #if defined(__GLIBCXX__) || defined(__GLIBCPP__)
4579 : 0 : __gnu_cxx::__verbose_terminate_handler();
4580 : : #endif
4581 : 0 : abort();
4582 : : }
4583 : :
4584 : :
4585 : : int
4586 : 74 : main (int argc, char *argv[])
4587 : : {
4588 : 74 : (void) setlocale (LC_ALL, "");
4589 : 74 : (void) bindtextdomain (PACKAGE_TARNAME, LOCALEDIR);
4590 : 74 : (void) textdomain (PACKAGE_TARNAME);
4591 : :
4592 : 74 : std::set_terminate(& my_terminate_handler);
4593 : :
4594 : : /* Tell the library which version we are expecting. */
4595 : 74 : elf_version (EV_CURRENT);
4596 : :
4597 [ + - - + ]: 148 : tmpdir = string(getenv("TMPDIR") ?: "/tmp");
4598 : :
4599 : : /* Set computed default values. */
4600 [ - + + - : 74 : db_path = string(getenv("HOME") ?: "/") + string("/.debuginfod.sqlite"); /* XDG? */
+ - - + -
+ + - -
- ]
4601 : 74 : int rc = regcomp (& file_include_regex, ".*", REG_EXTENDED|REG_NOSUB); // match everything
4602 [ - + ]: 74 : if (rc != 0)
4603 : 0 : error (EXIT_FAILURE, 0, "regcomp failure: %d", rc);
4604 : 74 : rc = regcomp (& file_exclude_regex, "^$", REG_EXTENDED|REG_NOSUB); // match nothing
4605 [ - + ]: 74 : if (rc != 0)
4606 : 0 : error (EXIT_FAILURE, 0, "regcomp failure: %d", rc);
4607 : :
4608 : : // default parameters for fdcache are computed from system stats
4609 : 74 : struct statfs sfs;
4610 : 74 : rc = statfs(tmpdir.c_str(), &sfs);
4611 [ - + ]: 74 : if (rc < 0)
4612 : 0 : fdcache_mbs = 1024; // 1 gigabyte
4613 : : else
4614 : 74 : fdcache_mbs = sfs.f_bavail * sfs.f_bsize / 1024 / 1024 / 4; // 25% of free space
4615 : 74 : fdcache_mintmp = 25; // emergency flush at 25% remaining (75% full)
4616 : 74 : fdcache_prefetch = 64; // guesstimate storage is this much less costly than re-decompression
4617 : :
4618 : : /* Parse and process arguments. */
4619 : 74 : int remaining;
4620 : 74 : (void) argp_parse (&argp, argc, argv, ARGP_IN_ORDER, &remaining, NULL);
4621 [ - + ]: 74 : if (remaining != argc)
4622 : 0 : error (EXIT_FAILURE, 0,
4623 : 0 : "unexpected argument: %s", argv[remaining]);
4624 : :
4625 [ + + + + : 74 : if (scan_archives.size()==0 && !scan_files && source_paths.size()>0)
- + ]
4626 [ # # ]: 0 : obatched(clog) << "warning: without -F -R -U -Z, ignoring PATHs" << endl;
4627 : :
4628 : 74 : fdcache.limit(fdcache_mbs);
4629 : :
4630 : 74 : (void) signal (SIGPIPE, SIG_IGN); // microhttpd can generate it incidentally, ignore
4631 : 74 : (void) signal (SIGINT, signal_handler); // ^C
4632 : 74 : (void) signal (SIGHUP, signal_handler); // EOF
4633 : 74 : (void) signal (SIGTERM, signal_handler); // systemd
4634 : 74 : (void) signal (SIGUSR1, sigusr1_handler); // end-user
4635 : 74 : (void) signal (SIGUSR2, sigusr2_handler); // end-user
4636 : :
4637 : : /* Get database ready. */
4638 [ + + ]: 74 : if (! passive_p)
4639 : : {
4640 : 72 : rc = sqlite3_open_v2 (db_path.c_str(), &db, (SQLITE_OPEN_READWRITE
4641 : : |SQLITE_OPEN_URI
4642 : : |SQLITE_OPEN_PRIVATECACHE
4643 : : |SQLITE_OPEN_CREATE
4644 : : |SQLITE_OPEN_FULLMUTEX), /* thread-safe */
4645 : : NULL);
4646 [ - + ]: 72 : if (rc == SQLITE_CORRUPT)
4647 : : {
4648 : 0 : (void) unlink (db_path.c_str());
4649 : 0 : error (EXIT_FAILURE, 0,
4650 : : "cannot open %s, deleted database: %s", db_path.c_str(), sqlite3_errmsg(db));
4651 : : }
4652 [ - + ]: 72 : else if (rc)
4653 : : {
4654 : 0 : error (EXIT_FAILURE, 0,
4655 : : "cannot open %s, consider deleting database: %s", db_path.c_str(), sqlite3_errmsg(db));
4656 : : }
4657 : : }
4658 : :
4659 : : // open the readonly query variant
4660 : : // NB: PRIVATECACHE allows web queries to operate in parallel with
4661 : : // much other grooming/scanning operation.
4662 : 74 : rc = sqlite3_open_v2 (db_path.c_str(), &dbq, (SQLITE_OPEN_READONLY
4663 : : |SQLITE_OPEN_URI
4664 : : |SQLITE_OPEN_PRIVATECACHE
4665 : : |SQLITE_OPEN_FULLMUTEX), /* thread-safe */
4666 : : NULL);
4667 [ - + ]: 74 : if (rc)
4668 : : {
4669 : 0 : error (EXIT_FAILURE, 0,
4670 : : "cannot open %s, consider deleting database: %s", db_path.c_str(), sqlite3_errmsg(dbq));
4671 : : }
4672 : :
4673 : :
4674 [ + - ]: 148 : obatched(clog) << "opened database " << db_path
4675 [ + + + - : 76 : << (db?" rw":"") << (dbq?" ro":"") << endl;
- + + - +
- ]
4676 [ + - + - ]: 148 : obatched(clog) << "sqlite version " << sqlite3_version << endl;
4677 [ + + + - : 220 : obatched(clog) << "service mode " << (passive_p ? "passive":"active") << endl;
+ - ]
4678 : :
4679 : : // add special string-prefix-similarity function used in rpm sref/sdef resolution
4680 : 74 : rc = sqlite3_create_function(dbq, "sharedprefix", 2, SQLITE_UTF8, NULL,
4681 : : & sqlite3_sharedprefix_fn, NULL, NULL);
4682 [ - + ]: 74 : if (rc != SQLITE_OK)
4683 : 0 : error (EXIT_FAILURE, 0,
4684 : : "cannot create sharedprefix function: %s", sqlite3_errmsg(dbq));
4685 : :
4686 [ + + ]: 74 : if (! passive_p)
4687 : : {
4688 [ + + ]: 72 : if (verbose > 3)
4689 [ + - + - ]: 92 : obatched(clog) << "ddl: " << DEBUGINFOD_SQLITE_DDL << endl;
4690 : 72 : rc = sqlite3_exec (db, DEBUGINFOD_SQLITE_DDL, NULL, NULL, NULL);
4691 [ - + ]: 72 : if (rc != SQLITE_OK)
4692 : : {
4693 : 0 : error (EXIT_FAILURE, 0,
4694 : : "cannot run database schema ddl: %s", sqlite3_errmsg(db));
4695 : : }
4696 : : }
4697 : :
4698 [ + - + - : 148 : obatched(clog) << "libmicrohttpd version " << MHD_get_version() << endl;
+ - ]
4699 : :
4700 : : /* If '-C' wasn't given or was given with no arg, pick a reasonable default
4701 : : for the number of worker threads. */
4702 [ + + ]: 74 : if (connection_pool == 0)
4703 : 70 : connection_pool = default_concurrency();
4704 : :
4705 : : /* Note that MHD_USE_EPOLL and MHD_USE_THREAD_PER_CONNECTION don't
4706 : : work together. */
4707 : 74 : unsigned int use_epoll = 0;
4708 : : #if MHD_VERSION >= 0x00095100
4709 : 74 : use_epoll = MHD_USE_EPOLL;
4710 : : #endif
4711 : :
4712 : 74 : unsigned int mhd_flags = (
4713 : : #if MHD_VERSION >= 0x00095300
4714 : : MHD_USE_INTERNAL_POLLING_THREAD
4715 : : #else
4716 : : MHD_USE_SELECT_INTERNALLY
4717 : : #endif
4718 : : | MHD_USE_DUAL_STACK
4719 : : | use_epoll
4720 : : #if MHD_VERSION >= 0x00095200
4721 : : | MHD_USE_ITC
4722 : : #endif
4723 : : | MHD_USE_DEBUG); /* report errors to stderr */
4724 : :
4725 : : // Start httpd server threads. Use a single dual-homed pool.
4726 : 74 : MHD_Daemon *d46 = MHD_start_daemon (mhd_flags, http_port,
4727 : : NULL, NULL, /* default accept policy */
4728 : : handler_cb, NULL, /* handler callback */
4729 : : MHD_OPTION_EXTERNAL_LOGGER,
4730 : : error_cb, NULL,
4731 : : MHD_OPTION_THREAD_POOL_SIZE,
4732 : : (int)connection_pool,
4733 : : MHD_OPTION_END);
4734 : :
4735 : 74 : MHD_Daemon *d4 = NULL;
4736 [ - + ]: 74 : if (d46 == NULL)
4737 : : {
4738 : : // Cannot use dual_stack, use ipv4 only
4739 : 0 : mhd_flags &= ~(MHD_USE_DUAL_STACK);
4740 [ # # ]: 0 : d4 = MHD_start_daemon (mhd_flags, http_port,
4741 : : NULL, NULL, /* default accept policy */
4742 : : handler_cb, NULL, /* handler callback */
4743 : : MHD_OPTION_EXTERNAL_LOGGER,
4744 : : error_cb, NULL,
4745 : : (connection_pool
4746 : : ? MHD_OPTION_THREAD_POOL_SIZE
4747 : : : MHD_OPTION_END),
4748 : : (connection_pool
4749 : : ? (int)connection_pool
4750 : : : MHD_OPTION_END),
4751 : : MHD_OPTION_END);
4752 [ # # ]: 0 : if (d4 == NULL)
4753 : : {
4754 : 0 : sqlite3 *database = db;
4755 : 0 : sqlite3 *databaseq = dbq;
4756 : 0 : db = dbq = 0; // for signal_handler not to freak
4757 : 0 : sqlite3_close (databaseq);
4758 : 0 : sqlite3_close (database);
4759 : 0 : error (EXIT_FAILURE, 0, "cannot start http server at port %d",
4760 : : http_port);
4761 : : }
4762 : :
4763 : : }
4764 : 74 : obatched(clog) << "started http server on"
4765 : : << (d4 != NULL ? " IPv4 " : " IPv4 IPv6 ")
4766 [ + - + - : 148 : << "port=" << http_port << endl;
+ - + - +
- ]
4767 : :
4768 : : // add maxigroom sql if -G given
4769 [ - + ]: 74 : if (maxigroom)
4770 : : {
4771 [ # # ]: 0 : obatched(clog) << "maxigrooming database, please wait." << endl;
4772 [ # # ]: 0 : extra_ddl.push_back("create index if not exists " BUILDIDS "_r_sref_arc on " BUILDIDS "_r_sref(artifactsrc);");
4773 [ # # ]: 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);");
4774 [ # # ]: 0 : extra_ddl.push_back("drop index if exists " BUILDIDS "_r_sref_arc;");
4775 : :
4776 : : // NB: we don't maxigroom the _files interning table. It'd require a temp index on all the
4777 : : // tables that have file foreign-keys, which is a lot.
4778 : :
4779 : : // NB: with =delete, may take up 3x disk space total during vacuum process
4780 : : // vs. =off (only 2x but may corrupt database if program dies mid-vacuum)
4781 : : // vs. =wal (>3x observed, but safe)
4782 [ # # ]: 0 : extra_ddl.push_back("pragma journal_mode=delete;");
4783 [ # # ]: 0 : extra_ddl.push_back("vacuum;");
4784 [ # # ]: 0 : extra_ddl.push_back("pragma journal_mode=wal;");
4785 : : }
4786 : :
4787 : : // run extra -D sql if given
4788 [ + + ]: 74 : if (! passive_p)
4789 [ - + ]: 72 : for (auto&& i: extra_ddl)
4790 : : {
4791 [ # # ]: 0 : if (verbose > 1)
4792 [ # # # # ]: 0 : obatched(clog) << "extra ddl:\n" << i << endl;
4793 : 0 : rc = sqlite3_exec (db, i.c_str(), NULL, NULL, NULL);
4794 [ # # # # ]: 0 : if (rc != SQLITE_OK && rc != SQLITE_DONE && rc != SQLITE_ROW)
4795 : 0 : error (0, 0,
4796 : : "warning: cannot run database extra ddl %s: %s", i.c_str(), sqlite3_errmsg(db));
4797 : :
4798 [ # # ]: 0 : if (maxigroom)
4799 [ # # ]: 0 : obatched(clog) << "maxigroomed database" << endl;
4800 : : }
4801 : :
4802 [ + + ]: 74 : if (! passive_p)
4803 [ + - + - ]: 144 : obatched(clog) << "search concurrency " << concurrency << endl;
4804 : 74 : obatched(clog) << "webapi connection pool " << connection_pool
4805 [ + - - + : 74 : << (connection_pool ? "" : " (unlimited)") << endl;
+ - + - ]
4806 [ + + ]: 74 : if (! passive_p) {
4807 [ + - + - ]: 144 : obatched(clog) << "rescan time " << rescan_s << endl;
4808 [ + - + - ]: 144 : obatched(clog) << "scan checkpoint " << scan_checkpoint << endl;
4809 : : }
4810 [ + - + - ]: 148 : obatched(clog) << "fdcache mbs " << fdcache_mbs << endl;
4811 [ + - + - ]: 148 : obatched(clog) << "fdcache prefetch " << fdcache_prefetch << endl;
4812 [ + - + - ]: 148 : obatched(clog) << "fdcache tmpdir " << tmpdir << endl;
4813 [ + - + - ]: 148 : obatched(clog) << "fdcache tmpdir min% " << fdcache_mintmp << endl;
4814 [ + + ]: 74 : if (! passive_p)
4815 [ + - + - ]: 144 : obatched(clog) << "groom time " << groom_s << endl;
4816 [ + - + - ]: 148 : obatched(clog) << "forwarded ttl limit " << forwarded_ttl_limit << endl;
4817 : :
4818 [ + + ]: 74 : if (scan_archives.size()>0)
4819 : : {
4820 : 50 : obatched ob(clog);
4821 [ + - ]: 50 : auto& o = ob << "accepting archive types ";
4822 [ + + ]: 152 : for (auto&& arch : scan_archives)
4823 [ + - + - : 102 : o << arch.first << "(" << arch.second << ") ";
+ - + - ]
4824 [ + - ]: 50 : o << endl;
4825 : 50 : }
4826 : 74 : const char* du = getenv(DEBUGINFOD_URLS_ENV_VAR);
4827 [ + + + + ]: 74 : if (du && du[0] != '\0') // set to non-empty string?
4828 [ + - + - ]: 28 : obatched(clog) << "upstream debuginfod servers: " << du << endl;
4829 : :
4830 [ + + ]: 74 : vector<pthread_t> all_threads;
4831 : :
4832 [ + + ]: 74 : if (! passive_p)
4833 : : {
4834 : 72 : pthread_t pt;
4835 : 72 : rc = pthread_create (& pt, NULL, thread_main_groom, NULL);
4836 [ - + ]: 72 : if (rc)
4837 : 0 : error (EXIT_FAILURE, rc, "cannot spawn thread to groom database\n");
4838 : : else
4839 : : {
4840 : : #ifdef HAVE_PTHREAD_SETNAME_NP
4841 : 72 : (void) pthread_setname_np (pt, "groom");
4842 : : #endif
4843 [ + - ]: 72 : all_threads.push_back(pt);
4844 : : }
4845 : :
4846 [ + + + + ]: 72 : if (scan_files || scan_archives.size() > 0)
4847 : : {
4848 [ + - ]: 66 : if (scan_checkpoint > 0)
4849 [ + - ]: 66 : scan_barrier = new sqlite_checkpoint_pb(concurrency, (unsigned) scan_checkpoint);
4850 : :
4851 : 66 : rc = pthread_create (& pt, NULL, thread_main_fts_source_paths, NULL);
4852 [ - + ]: 66 : if (rc)
4853 : 0 : error (EXIT_FAILURE, rc, "cannot spawn thread to traverse source paths\n");
4854 : : #ifdef HAVE_PTHREAD_SETNAME_NP
4855 : 66 : (void) pthread_setname_np (pt, "traverse");
4856 : : #endif
4857 [ + - ]: 66 : all_threads.push_back(pt);
4858 : :
4859 [ + + ]: 330 : for (unsigned i=0; i<concurrency; i++)
4860 : : {
4861 : 264 : rc = pthread_create (& pt, NULL, thread_main_scanner, NULL);
4862 [ - + ]: 264 : if (rc)
4863 : 0 : error (EXIT_FAILURE, rc, "cannot spawn thread to scan source files / archives\n");
4864 : : #ifdef HAVE_PTHREAD_SETNAME_NP
4865 : 264 : (void) pthread_setname_np (pt, "scan");
4866 : : #endif
4867 [ + - ]: 264 : all_threads.push_back(pt);
4868 : : }
4869 : : }
4870 : : }
4871 : :
4872 : : /* Trivial main loop! */
4873 [ + - + - ]: 74 : set_metric("ready", 1);
4874 [ + + ]: 214 : while (! interrupted)
4875 [ + - ]: 140 : pause ();
4876 [ + - ]: 74 : scanq.nuke(); // wake up any remaining scanq-related threads, let them die
4877 [ + + + - ]: 74 : if (scan_barrier) scan_barrier->nuke(); // ... in case they're stuck in a barrier
4878 [ + - + - ]: 74 : set_metric("ready", 0);
4879 : :
4880 [ + - ]: 74 : if (verbose)
4881 [ + - + - : 148 : obatched(clog) << "stopping" << endl;
- - ]
4882 : :
4883 : : /* Join all our threads. */
4884 [ + + ]: 476 : for (auto&& it : all_threads)
4885 [ + - ]: 402 : pthread_join (it, NULL);
4886 : :
4887 : : /* Stop all the web service threads. */
4888 [ + - + - ]: 74 : if (d46) MHD_stop_daemon (d46);
4889 [ - + - - ]: 74 : if (d4) MHD_stop_daemon (d4);
4890 : :
4891 [ + + ]: 74 : if (! passive_p)
4892 : : {
4893 : : /* With all threads known dead, we can clean up the global resources. */
4894 [ + - ]: 72 : rc = sqlite3_exec (db, DEBUGINFOD_SQLITE_CLEANUP_DDL, NULL, NULL, NULL);
4895 [ - + ]: 72 : if (rc != SQLITE_OK)
4896 : : {
4897 [ # # # # ]: 0 : error (0, 0,
4898 : : "warning: cannot run database cleanup ddl: %s", sqlite3_errmsg(db));
4899 : : }
4900 : : }
4901 : :
4902 [ + - ]: 74 : debuginfod_pool_groom ();
4903 [ + + ]: 74 : delete scan_barrier;
4904 : :
4905 : : // NB: no problem with unconditional free here - an earlier failed regcomp would exit program
4906 [ + - ]: 74 : (void) regfree (& file_include_regex);
4907 [ + - ]: 74 : (void) regfree (& file_exclude_regex);
4908 : :
4909 : 74 : sqlite3 *database = db;
4910 : 74 : sqlite3 *databaseq = dbq;
4911 : 74 : db = dbq = 0; // for signal_handler not to freak
4912 [ + - ]: 74 : (void) sqlite3_close (databaseq);
4913 [ + + ]: 74 : if (! passive_p)
4914 [ + - ]: 72 : (void) sqlite3_close (database);
4915 : :
4916 [ + + ]: 74 : return 0;
4917 : 74 : }
|
