Changeset 403aae58 for general/prog/llvm.xml
- Timestamp:
- 08/03/2023 08:22:53 AM (10 months ago)
- Branches:
- 12.0, 12.1, ken/TL2024, ken/tuningfonts, lazarus, plabs/newcss, python3.11, rahul/power-profiles-daemon, renodr/vulkan-addition, trunk, xry111/llvm18
- Children:
- 61f3e811
- Parents:
- e4c821e2
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
general/prog/llvm.xml
re4c821e2 r403aae58 337 337 <screen remap="doc"><userinput>ninja docs-clang-html docs-clang-man</userinput></screen> 338 338 339 <!-- The excessive core dump made my system irresponsive. --> 340 <para> 341 LLVM test suite can produce many core dump files. They will occupy 342 a large amount of disk space, and the core dump process can 343 significantly slow down the testing. To test the results with core 344 dump disabled, <phrase revision='systemd'>ensure 345 <xref linkend='systemd'/> is rebuilt with 346 <xref linkend='linux-pam'/> support and the current login session is 347 started after updating the 348 <filename>/etc/pam.d/system-session</filename> file to include 349 <filename class='libraryfile'>pam_systemd.so</filename>, 350 then </phrase>issue: 351 </para> 352 353 <screen remap="test" revision="sysv"><userinput>sh -c 'ulimit -c 0 && ninja check-all'</userinput></screen> 354 355 <!-- From core(5): "RLIMIT_CORE will be ignored if the system is 356 configured to pipe core dumps to a program". Here "the program" 357 is systemd-coredumpd. So we need some fancy systemd setting. 358 -–user: Run it under per-user systemd manager, avoiding 359 the necessity of root priviledge. 360 -&nash;pty: Run it in a pseudo-terminal, so we can monitor the 361 process or interrupt it with Ctrl-C if we want. 362 -d: Run it with $PWD as the working directory. 363 -G: Delete the transient service unit even if the command fails, 364 to prevent "systemctl -–user status" from reporting 365 "degraded". 366 -p LimitCORE=0: Disable core dump. 367 --> 368 <screen remap="test" revision="systemd"><userinput> 369 systemctl --user start dbus && 370 systemd-run --user --pty -d -G -p LimitCORE=0 ninja check-all</userinput></screen> 371 339 372 <para> 340 373 <!-- EDITORS - if you have more than 4 cores, take some offline to measure 341 374 the elapsed time for the tests. Also, libstdc++.a and perhaps libstdc++fs.a 342 375 are used by a couple of the link-static tests. --> 343 T o test the results, issue: <command>ninja check-all</command>. Tests are376 Tests are 344 377 built with all <emphasis>available</emphasis> cores, but run using the number 345 378 of <emphasis>online</emphasis> processors. (The difference is that available … … 352 385 <option>CONFIG_USER_NS</option> is not enabled in kernel 353 386 configuration. 387 <!-- TODO: make or find a report to upstream --> 388 The tests named <filename>AtoiAndFriendsOOBTest</filename>, 389 <filename>StrtolOOBTest</filename>, 390 <filename>StrtollOOBTest</filename>, 391 <filename>interception_malloc_test.cpp</filename>, and 392 <filename>scanf-allocate.cpp</filename> are known to fail with 393 Glibc-2.38. 354 394 <!-- 16.0.3: 67327 Passed, 23510 Unsupported, 85 Skipped, 355 395 124 Expectedly Failed … … 370 410 124 Expectedly Failed, 1 Failed 371 411 FAIL: ThreadSanitizer-x86_64 :: Linux/clone_setns.cpp - xry111 412 413 16.0.5 with Glibc-2.38: 414 63716 Passed, 23512 Unsupported, 85 Skipped, 415 124 Expectedly Failed, 15 Failed 416 417 - AtoiAndFriendsOOBTest, StrtolOOBTest, StrtollOOBTest 418 (repeated 4 times with different configurations) 419 - interception_malloc_test.cpp (repeated 2 times) 420 - scanf-allocate.cpp 372 421 --> 373 422
Note:
See TracChangeset
for help on using the changeset viewer.