Created attachment 206722 [details] nspr-4.22.log generated by poudriere I'm experiencing build errors in devel/nspr since the update to version 4.22 and have attached the poudriere build log. cc -o attach.o -c -fvisibility=hidden -O2 -pipe -fstack-protector-strong -fno-strict-aliasing -ansi -Wall -fPIC -UDEBUG -DPACKAGE_NAME=\"\" -DPACKAGE_TARNAME=\"\" -DPACKAGE_VERSION=\"\" -DPACKAGE_STRING=\"\" -DPACKAGE_BUGREPORT=\"\" -DPACKAGE_URL=\"\" -DNDEBUG=1 -DHAVE_VISIBILITY_HIDDEN_ATTRIBUTE=1 -DHAVE_VISIBILITY_PRAGMA=1 -DXP_UNIX=1 -DFREEBSD=1 -DHAVE_BSD_FLOCK=1 -DHAVE_SOCKLEN_T=1 -DHAVE_POINTER_LOCALTIME_R=1 -DHAVE_DLADDR=1 -DHAVE_LCHOWN=1 -DHAVE_SETPRIORITY=1 -DHAVE_STRERROR=1 -DHAVE_SYSCALL=1 -D_REENTRANT=1 -D_THREAD_SAFE=1 -DFORCE_PR_LOG -D_PR_PTHREADS -UHAVE_CVAR_BUILT_ON_SEM -I../../dist/include/nspr -I../../pr/include -I../../pr/include/private attach.c attach.c:170:2: error: unterminated conditional directive #ifdef _PR_PTHREADS ^ attach.c:275:2: error: expected '}' } ^ attach.c:109:1: note: to match this '{' { ^ attach.c:176:8: error: use of undeclared label 'exit_now' goto exit_now; ^ 3 errors generated. gmake[1]: *** [../../config/rules.mk:450: attach.o] Error 1 gmake[1]: Leaving directory '/wrkdirs/usr/ports/devel/nspr/work/nspr-4.22/nspr/pr/tests' *** Error code 2 Stop. make: stopped in /usr/ports/devel/nspr
Created attachment 206723 [details] svn-diff-nspr Patch for tests.
Created attachment 206724 [details] svn-diff-nspr_v2 Removed unwanted doubled lines in patch-nonblock.c.
Thanks! For me your patch solved the build. But by the way, layer and nblayer tests (run via make test) hang forever (and it's for a long time for the port). Not to mention just failed tests.
Does not hang on 11.3amd64: Passed nearly all tests except three (PR_loadlibrary failed libfilename and libmyso. libmy.so does not exist) and (fdcach: fd stack malfunctioned) but ends with: vercheck Passed version Passed writev Passed xnotify Passed zerolen Passed END Wed Aug 21 15:46:22 CEST 2019 *** Error code 1
(In reply to Walter Schwarzenfeld from comment #4) Really? I have on amd64 11-STABLE the hang of layer and nblayer for a very long time. I use only ports (with portsnap support). I have to kill "layer" and "nblayer" processes otherwise the test did not end forever. My failed are: dlltest FAILED dlltest: PR_LoadLibrary failed (-5977, 0, Cannot open "dll/libmy.so") fdcach FAILED fdcach: fd stack malfunctioned initclk FAILED initclk: Elapsed time should be 3000 ms but is 3108 ms initclk: FAIL layer FAILED layer: Beginning non-layered test layer: Ending non-layered test layer: Beginning layered test libfilename FAILED libfilename: PR_LoadLibrary failed nblayer FAILED nblayer: Beginning non-layered test nblayer: Ending non-layered test nblayer: Beginning layered test timetest FAILED timetest: PR_ParseTimeString() incorrect timetest: t is -1664496000000000, t2 is -1664495960000000, time string is Wed Apr 4 05:31:40 +0531 1917 timetest: END Thu Aug 22 00:22:43 +07 2019 *** Error code 1 And if "layer" waits something (part of ps -ax after hours of waiting shows no CPU usage): 20896 2 I+ 0:00.08 ./layer the "nblayer" uses some CPU cycles (after waiting dozens of minutes): 21724 2 I+ 39:46.26 ./nblayer
You can try chang in work/pr/tests/runtests.sh to 207 TEST_TIMEOUT=3200 208 printf " ${TEST_TIMEOUT}" 209 if [ $OS_PLATFORM = "Windows_95" ] || [ $OS_PLATFORM = "Windows_98" ] || [ $OS_PLATFORM = "Windows_NT" ] || [ $OS_PLATFORM = "OS/2"] || [ $OS_PLATFORM = "FreeBSD" ] ; then Change the value of TEST_TIMEOUT as you need. I cannot test it, cause I cannot reproduce.
Created attachment 206784 [details] svn-diff-nspr_v3 I remove the tests for libfilename and dlltest (both needs libmy.so which not exists and fdcache (which seems and old bug I found in mozilla bug PR's).
Comment on attachment 206784 [details] svn-diff-nspr_v3 - Build fails even without patches under files/ directory, so this is an upstream regression and should be reported upstream in order to get the fix right - Patches for tests should be in files/patch-tests unless they're patch of an upstream fix; splitting patches per-file doesn't scale with many unrelated changes - Tests are not installed, so bumping PORTREVISION is not necessary
I've filed upstream bug. The link is in See Also field.
A commit references this bug: Author: jbeich Date: Fri Aug 23 21:55:41 UTC 2019 New revision: 509663 URL: https://svnweb.freebsd.org/changeset/ports/509663 Log: devel/nspr: unbreak TEST=on build after r509330 attach.c:170:2: error: unterminated conditional directive #ifdef _PR_PTHREADS ^ attach.c:275:2: error: expected '}' } ^ attach.c:109:1: note: to match this '{' { ^ attach.c:176:8: error: use of undeclared label 'exit_now' goto exit_now; ^ PR: 239986 Reported by: Ross McKelvie Changes: head/devel/nspr/files/patch-bug1575821 head/devel/nspr/files/patch-tests