* we often see QA warnings like: glibc-locale-2.26: glibc-locale: /glibc-binary-localedata-en-gb/usr/lib/locale/en_GB/LC_MEASUREMENT is owned by uid 3004, which is the same as the user running bitbake. This may be due to host contamination [host-user-contaminated] glibc-locale-2.26: glibc-locale: /glibc-binary-localedata-nn-no.iso-8859-1/usr/lib/locale/nn_NO.ISO-8859-1/LC_MEASUREMENT is owned by uid 3004, which is the same as the user running bitbake. This may be due to host contamination [host-user-contaminated] but we don't know the root cause of it. * don't trigger error from '^inode mismatch' * show parts of pseudo.log in console many builds are executed in tmpfs, so I cannot check the pseudo.log later * lower from error to warn as I hate all my builds reporting failure just because of this pseudo debug * the only theory we currently have is that it's a bug in pseudo when inode is being reused for different files, which is supported by pseudo.log entries:
Good build: pseudo$ grep -v "^path mismatch" pseudo.log debug_logfile: fd 2 pid 7975 [parent 7974], doing new pid setup and server start Setup complete, sending SIGUSR1 to pid 7974. db cleanup for server shutdown, 17:33:58.787 memory-to-file backup complete, 17:33:58.787. db cleanup finished, 17:33:58.787 Build with QA host-user-contaminated issue: ERROR: foo-1.0.0-r0 do_package_qa: QA Issue: foo: file-with-wrong-UID is owned by uid 2001, which is the same as the user running bitbake. This may be due to host contamination [host-user-contaminated] pseudo$ grep "file-with-wrong-UID" pseudo.log inode mismatch: 'file-with-wrong-UID' ino 13242270 in db, 13242271 in request. creat ignored for existing file 'file-with-wrong-UID'. inode mismatch: 'file-with-wrong-UID' ino 13242270 in db, 13242271 in request. inode mismatch: 'file-with-wrong-UID' ino 13242270 in db, 13242271 in request. inode mismatch: 'file-with-wrong-UID' ino 13242270 in db, 13242271 in request. inode mismatch: 'file-with-wrong-UID' ino 13242270 in db, 13242271 in request. inode mismatch: 'file-with-wrong-UID' ino 13242270 in db, 13242271 in request. inode mismatch: 'file-with-wrong-UID' ino 13242270 in db, 13242271 in request. path mismatch [1 link]: ino 13242270 db 'file-with-wrong-UID' req 'some-other-unrelated-file'. creat for 'some-other-unrelated-file' replaces existing 13242270 ['file-with-wrong-UID']. db cleanup for server shutdown, 02:16:23.685 memory-to-file backup complete, 02:16:23.685. db cleanup finished, 02:16:23.685 And some-other-unrelated-file is really some different file, not just hardlink to the same file from some different directory (like between WORKDIR and sysroot other "path mismatch" entries show). Signed-off-by: Martin Jansa <martin.ja...@gmail.com> --- meta/classes/insane.bbclass | 43 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 43 insertions(+) diff --git a/meta/classes/insane.bbclass b/meta/classes/insane.bbclass index 0a3b528ddb..76cd92709b 100644 --- a/meta/classes/insane.bbclass +++ b/meta/classes/insane.bbclass @@ -1238,6 +1238,44 @@ python do_qa_unpack() { bb.warn('%s: the directory %s (%s) pointed to by the S variable doesn\'t exist - please set S within the recipe to point to where the source has been unpacked to' % (d.getVar('PN'), d.getVar('S', False), s_dir)) } +python do_qa_pseudo() { + ########################################################################### + # Check pseudo.log for unexpected errors + # + # Typical pseudo.log contains many "^path mismatch" lines for all the hardlinked files + # e.g. in some smaller component I see 231/237 lines to be "^path mismatch" other 6 + # lines are setup and cleanup lines like this: + # debug_logfile: fd 2 + # pid 7975 [parent 7974], doing new pid setup and server start + # Setup complete, sending SIGUSR1 to pid 7974. + # db cleanup for server shutdown, 17:33:58.787 + # memory-to-file backup complete, 17:33:58.787. + # db cleanup finished, 17:33:58.787 + # + # but if there is one of: + # "^inode mismatch" + # "^creat ignored for existing file" + # "^creat for.*replaces existing" + # then there might be some bigger issue which sometimes results in host-user-contaminated QA warnings + ########################################################################### + + import subprocess + + pseudodir = d.getVar('PSEUDO_LOCALSTATEDIR') + bb.note("Checking pseudo.log for common errors") + pseudolog = os.path.join(pseudodir, "pseudo.log") + statement = "grep" \ + " -e '^creat ignored for existing file'" \ + " -e '^creat for.*replaces existing'" \ + " %s" % pseudolog + if subprocess.call("%s -q" % statement, shell=True) == 0: + statement2 = "grep -v '^path mismatch' %s" % pseudolog + statement2_count = "grep -v '^path mismatch' -c %s" % pseudolog + output = subprocess.check_output(statement2, shell=True).decode("utf-8") + output_count = subprocess.check_output(statement2_count, shell=True).decode("utf-8").replace('\n', '') + bb.warn("This %s indicates %s errors, see %s or %s [qa_pseudo]: %s" % (pseudolog, output_count, statement, statement2, output)) +} + # The Staging Func, to check all staging #addtask qa_staging after do_populate_sysroot before do_build do_populate_sysroot[postfuncs] += "do_qa_staging " @@ -1250,6 +1288,11 @@ do_configure[postfuncs] += "do_qa_configure " # Check does S exist. do_unpack[postfuncs] += "do_qa_unpack" +# Check pseudo.log for unexpected errors +# For some reason do_build postfunction isn't executed +# do_build[postfuncs] += "do_qa_pseudo" +addtask do_qa_pseudo after do_populate_sysroot do_packagedata do_package before do_build + python () { import re -- 2.14.1 -- _______________________________________________ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core