On Sun, Jan 06, 2019 at 09:59:51AM +0800, Yanhao Mo wrote: > Hmm, the reproducible building is still failed. And I still cannot > reproduce the error. Accordding to the error message: > > > XPASS : TestKallsyms::testProc() 'kallsyms.parseMapping(path)' returned > > TRUE unexpectedly. () > > Loc: > > [/<<BUILDDIR>>/hotspot-1.1.0+git20180816/3rdparty/perfparser/tests/auto/kallsyms/tst_kallsyms.cpp(138)] > > QDEBUG : TestKallsyms::testParseErrors() "Mapping is empty." > > QDEBUG : TestKallsyms::testParseErrors() "Invalid address: this" > > QDEBUG : TestKallsyms::testParseErrors() "No such file or directory" > > I guess that because the build environments used by > tests.reproducible-builds.org dosen't have a normal /proc/kallsyms file. > Or maybe it doesn't have such file at all. Could you please help verify > that? Very appreciated.
It is usual and customary for build tools to mount the /proc directory from host machine inside the chroot in which the build happens. I know that sbuild does that, and it would be a real surprise if pbuilder (used by reproducible-builds) did not do that as well. In either case, since I'm using sbuild and the package fails to build for me, not having /proc properly mounted is very unlikely to be the reason for the failure. > And if what I'm guessing is true for the /proc/kallsyms file. Then what > operations am I excepted to take? Should I just disable the specific > testcase? Only as a last resort. Before that, I can offer ssh access to a machine where this seems to happen all the time. Please contact me privately for details. Thanks.
