Bulk build stuck on NetBSD/macppc 6.0

[Havard Eidnes <he%NetBSD.org@localhost>]

Hi,

looking at a bulk pkgsrc build which appears to get stuck while
building abcl for NetBSD/macppc 6.0:


malus# proctree
...
 |             |       |-+- 01752 root make USE_BULK_CACHE bulk-package 
 |             |       | \-+- 04727 root /bin/sh -c set -e; set +e;  
uptodate=`/usr/bin/env MAKECONF=/etc/mk.conf 
PATH=/usr/pkgsrc/lang/abcl/work/.wrapper/bin:/usr/pkgsrc/lang/abcl/work/.buildlink/bin:/usr/pkgsrc/lang/abcl/work/.g
 |             |       |   |--- 18917 root /usr/bin/tee -a 
/usr/pkgsrc/lang/abcl/.make 
 |             |       |   \-+- 24652 root /bin/sh -c set -e; set +e;  
uptodate=`/usr/bin/env MAKECONF=/etc/mk.conf 
PATH=/usr/pkgsrc/lang/abcl/work/.wrapper/bin:/usr/pkgsrc/lang/abcl/work/.buildlink/bin:/usr/pkgsrc/lang/abcl/work/
 |             |       |     \-+- 12395 root /bin/sh -c set -e; set +e;  
uptodate=`/usr/bin/env MAKECONF=/etc/mk.conf 
PATH=/usr/pkgsrc/lang/abcl/work/.wrapper/bin:/usr/pkgsrc/lang/abcl/work/.buildlink/bin:/usr/pkgsrc/lang/abcl/wor
 |             |       |       \-+- 29558 root /usr/bin/make _MAKE OPSYS 
OS_VERSION LOWER_OPSYS _PKGSRCDIR PKGTOOLS_VERSION PKG_JVM _CC _PATH_ORIG 
package package-install 
 |             |       |         \-+- 19151 root /bin/sh -c set -e;\t\t\t\t\t 
if test -n "" &&  /usr/sbin/pkg_info -K /var/db/pkg -qe abcl-1.0.1; then  echo 
===\\> "Skipping installation of already handled package";  else  cd /usr
 |             |       |           \-+- 13353 root /usr/bin/make HOST_OSTYPE 
USE_BULK_CACHE _SRC_TOP_ LOWER_OPSYS PKG_JVM _PKGSRC_BARRIER 
ALLOW_VULNERABLE_PACKAGES package package-install 
 |             |       |             \-+- 06050 root /bin/sh -c set -e; cd 
/usr/pkgsrc/lang/abcl/work/abcl-src-1.0.1 &&  /usr/bin/env USETOOLS=no 
CLASSPATH=/usr/pkg/java/kaffe/lib/tools.jar:. JAVA_HOME=/usr/pkg/java/kaffe 
CC=gcc C
 |             |       |               \--- 28053 root 
/usr/pkg/java/kaffe/jre/bin/kaffe-bin -classpath 
/usr/pkg/lib/java/ant/ant-launcher.jar -Dant.home=/usr/pkg 
-Dant.library.dir=/usr/pkg/lib/java/ant org.apache.tools.ant.launch
 |             |       |                 \--- 19263 root jikes -sourcepath 
/usr/pkgsrc/lang/abcl/work/abcl-src-1.0.1/src -d 
/usr/pkgsrc/lang/abcl/work/abcl-src-1.0.1/build/classes -classpath 
/usr/pkgsrc/lang/abcl/work/abcl-src-1.0

which is the end of this branch of processes.  Looking at the
jikes process, I see the following open FDs:

malus# fstat -p 19263
USER     CMD          PID   FD MOUNT       INUM MODE         SZ|DV R/W
root     jikes      19263 root /usr     5468288 drwxr-xr-x     512 r 
root     jikes      19263   wd /usr     9536116 drwxr-xr-x     512 r 
root     jikes      19263    0* pipe 0x1f6f4438 <- 0x0 r
root     jikes      19263    1* pipe 0x99e9600 -> 0x1f6f4510 w
root     jikes      19263    2* pipe 0x1f6f3a20 -> 0x19fe5518 w
root     jikes      19263    3 /usr     9535893 -rw-r--r--  8668982 r 
root     jikes      19263    4 /usr     9535950 -rw-r--r--   33203 r 
root     jikes      19263    5 /usr     9535893 -rw-r--r--  8668982 r 
root     jikes      19263   10* pipe 0x99e9cc0 -> 0x0 w
malus# ps axl | egrep PID\|jikes
 UID   PID  PPID   CPU PRI NI    VSZ   RSS WCHAN   STAT TTY       TIME COMMAND
   0 19263 28053 20843  70 20  24520 21644 pipe_wr IN   ttyp5  0:00.88 jikes 
-sourcepath /usr/pkgsrc/lang/abcl/work/abcl-src-1.0.1/src -d 
/usr/pkgsrc/lang/abcl/work/abcl-src-1.0.1/build/classes -classpath 
/usr/pkgsrc/lang/abcl/wor
malus# 
malus# gdb /usr/sandbox/usr/pkg/bin/jikes 
GNU gdb (GDB) 7.3.1
Copyright (C) 2011 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
...
(gdb) attach 19263
...
(gdb) where
#0  0xfdd30bf0 in write () from /usr/lib/libc.so.12
#1  0xfde13c80 in __swrite () from /usr/lib/libc.so.12
#2  0xfde125c4 in __sfvwrite () from /usr/lib/libc.so.12
#3  0xfde123d0 in fwrite () from /usr/lib/libc.so.12
#4  0xfdf35dc8 in __gnu_cxx::stdio_sync_filebuf<char, std::char_traits<char> 
>::xsputn(char const*, int) () from /usr/lib/libstdc++.so.7
#5  0xfdf30e64 in std::basic_ostream<char, std::char_traits<char> >& 
std::__ostream_insert<char, std::char_traits<char> >(std::basic_ostream<char, 
std::char_traits<char> >&, char const*, int) () from /usr/lib/libstdc++.so.7
#6  0x0189cf1c in ?? ()
#7  0x0189d60c in ?? ()
#8  0x01857808 in ?? ()
#9  0x0185a248 in ?? ()
#10 0x0185b600 in ?? ()
#11 0x018d11d8 in ?? ()
#12 0x018d0dec in ?? ()
#13 0x01801dbc in ?? ()
#14 0x01801c54 in ?? ()
(gdb) 

malus# fstat | less

searching for 0x19fe5518 gives this process:

root     kaffe-bin  28053 root /usr     5468288 drwxr-xr-x     512 r 
root     kaffe-bin  28053   wd /usr     9536116 drwxr-xr-x     512 r 
root     kaffe-bin  28053    0 /usr     5468995 crw-rw-rw-    null r 
root     kaffe-bin  28053    1* pipe 0x1f6f3438 -> 0x19fe5368 w
root     kaffe-bin  28053    2* pipe 0x1f6f3438 -> 0x19fe5368 w
root     kaffe-bin  28053    5* pipe 0x1f6f4510 <- 0x99e9600 r
root     kaffe-bin  28053    7* pipe 0x19fe5518 <- 0x1f6f3a20 r
root     kaffe-bin  28053   10* pipe 0x99e9cc0 -> 0x0 w
root     kaffe-bin  28053   11* pipe 0x1f6f35e8 -> 0x0 w

on fd 7.  This pair of processes appears to be stuck.

Any good suggestions on how to prevent this?  Is kaffe (I beleive
it is Old) insufficient for the build?  If so, should it be
excluded from the usable java VMs for this package?

For now I've just killed jikes, to let the build progress.


Regards,

- Håvard