1 |
I upgraded to gcc 4.1.1 along the way, while building 180 something of |
2 |
the 450 or so packages that emerge -e world wanted to build, |
3 |
dev-lang/glib-2.10.3 failed to build with the following: |
4 |
|
5 |
>>> checking ebuild checksums ;-) |
6 |
>>> checking auxfile checksums ;-) |
7 |
>>> checking miscfile checksums ;-) |
8 |
>>> checking glib-2.10.3.tar.bz2 ;-) |
9 |
>>> Unpacking source... |
10 |
>>> Unpacking glib-2.10.3.tar.bz2 to /var/tmp/portage/glib-2.10.3/work |
11 |
Applying glib-2.8.3-macos.patch ... |
12 |
>>> Source unpacked. |
13 |
>>> Compiling source in /var/tmp/portage/glib-2.10.3/work/glib-2.10.3 |
14 |
... |
15 |
Removing useless C++ checks ... |
16 |
Running elibtoolize in: glib-2.10.3 |
17 |
Applying portage-1.5.10.patch ... |
18 |
Applying max_cmd_len-1.5.20.patch ... |
19 |
Applying sed-1.5.6.patch ... |
20 |
* econf: updating glib-2.10.3/config.guess with |
21 |
/usr/share/gnuconfig/config.guess |
22 |
* econf: updating glib-2.10.3/config.sub with |
23 |
/usr/share/gnuconfig/config.sub |
24 |
/usr/bin/confcache --confcache-ignore /var/tmp/ccache:/var/tmp/.distcc |
25 |
--confcache-dir /var/tmp/confcache ./configure --prefix=/usr |
26 |
--host=x86_64-pc-linux-gnu --mandir=/usr/share/man |
27 |
--infodir=/usr/share/info --datadir=/usr/share --sysconfdir=/etc |
28 |
--localstatedir=/var/lib --enable-gtk-doc --with-threads=posix |
29 |
--libdir=/usr/lib64 --build=x86_64-pc-linux-gnu |
30 |
configure: loading cache config.cache |
31 |
checking for a BSD-compatible install... (cached) /bin/install -c |
32 |
checking whether build environment is sane... yes |
33 |
checking for gawk... (cached) gawk |
34 |
. |
35 |
. |
36 |
. |
37 |
.if x86_64-pc-linux-gnu-gcc -DHAVE_CONFIG_H -I. -I. -I.. -I.. -I../glib |
38 |
-I../gmodule -DG_DISABLE_CAST_CHECKS -pthread -march=opteron -O3 -pipe |
39 |
-fomit-frame-pointer -ftracer -Wall -MT timeloop-closure.o -MD -MP -MF |
40 |
".deps/timeloop-closure.Tpo" \ |
41 |
-c -o timeloop-closure.o `test -f 'timeloop-closure.c' || echo |
42 |
'./'`timeloop-closure.c; \ |
43 |
then mv -f ".deps/timeloop-closure.Tpo" ".deps/timeloop-closure.Po"; \ |
44 |
else rm -f ".deps/timeloop-closure.Tpo"; exit 1; \ |
45 |
fi |
46 |
/bin/sh ../libtool --mode=link x86_64-pc-linux-gnu-gcc -march=opteron |
47 |
-O3 -pipe -fomit-frame-pointer -ftracer -Wall -o |
48 |
libmoduletestplugin_a.la -rpath /usr/lib64 -Wl,--export-dynamic |
49 |
-avoid-version -module -rpath /usr/lib64 libmoduletestplugin_a.lo |
50 |
/bin/sh ../libtool --mode=link x86_64-pc-linux-gnu-gcc -march=opteron |
51 |
-O3 -pipe -fomit-frame-pointer -ftracer -Wall -o |
52 |
libmoduletestplugin_b.la -rpath /usr/lib64 -Wl,--export-dynamic |
53 |
-avoid-version -module -rpath /usr/lib64 libmoduletestplugin_b.lo |
54 |
/bin/sh ../libtool --mode=link x86_64-pc-linux-gnu-gcc -march=opteron |
55 |
-O3 -pipe -fomit-frame-pointer -ftracer -Wall -o testglib testglib.o |
56 |
../glib/libglib-2.0.la |
57 |
x86_64-pc-linux-gnu-gcc -shared .libs/libmoduletestplugin_b.o |
58 |
-march=opteron -Wl,--export-dynamic -Wl,-soname |
59 |
-Wl,libmoduletestplugin_b.so -o .libs/libmoduletestplugin_b.so |
60 |
creating libmoduletestplugin_b.la |
61 |
(cd .libs && rm -f libmoduletestplugin_b.la && ln -s |
62 |
../libmoduletestplugin_b.la libmoduletestplugin_b.la) |
63 |
x86_64-pc-linux-gnu-gcc -shared .libs/libmoduletestplugin_a.o |
64 |
-march=opteron -Wl,--export-dynamic -Wl,-soname |
65 |
-Wl,libmoduletestplugin_a.so -o .libs/libmoduletestplugin_a.so |
66 |
/bin/sh ../libtool --mode=link x86_64-pc-linux-gnu-gcc -march=opteron |
67 |
-O3 -pipe -fomit-frame-pointer -ftracer -Wall -o testgdate |
68 |
testgdate.o ../glib/libglib-2.0.la |
69 |
creating libmoduletestplugin_a.la |
70 |
(cd .libs && rm -f libmoduletestplugin_a.la && ln -s |
71 |
../libmoduletestplugin_a.la libmoduletestplugin_a.la) |
72 |
/bin/sh ../libtool --mode=link x86_64-pc-linux-gnu-gcc -march=opteron |
73 |
-O3 -pipe -fomit-frame-pointer -ftracer -Wall -o testgdateparser |
74 |
testgdateparser.o ../glib/libglib-2.0.la |
75 |
x86_64-pc-linux-gnu-gcc -march=opteron -O3 -pipe -fomit-frame-pointer |
76 |
-ftracer -Wall -o .libs/testglib testglib.o |
77 |
../glib/.libs/libglib-2.0.so |
78 |
testglib.o: In function `main': |
79 |
testglib.c:(.text+0x4d0): undefined reference to `glib_binary_age' |
80 |
testglib.c:(.text+0x4dc): undefined reference to `glib_interface_age' |
81 |
testglib.c:(.text+0x4e2): undefined reference to `glib_micro_version' |
82 |
testglib.c:(.text+0x4e8): undefined reference to `glib_minor_version' |
83 |
testglib.c:(.text+0x4ee): undefined reference to `glib_major_version' |
84 |
collect2: ld returned 1 exit status |
85 |
make[3]: *** [testglib] Error 1 |
86 |
make[3]: *** Waiting for unfinished jobs.... |
87 |
|
88 |
************************************************************************ |
89 |
********************** |
90 |
|
91 |
I tried masking glib-2.10.3 and building a lower version, same result. |
92 |
Since gcc 3.4.6 was my previous compiler, I tried runniing |
93 |
"fix_libtool_files.sh 3.4.6," no joy. Then, I vaguely remembered a |
94 |
message here recently, that mentioned something about the confcache |
95 |
feature of portage causing a problem for someone during a gcc upgrade. I |
96 |
commented out the FEATURES="confcache" line in make.conf and deleted |
97 |
everything from /var/tmp/confcache, afterward, lo and behold glib-2.10.3 |
98 |
built and installed without error. |
99 |
|
100 |
The description for confcache says: "GNU autoconf is a bottleneck for |
101 |
compiling packages - especially on multi-processor boxes, confcache had |
102 |
been considered as a way to lower the cost of repeatedly running |
103 |
autoconf-generated configure scripts." Since this is a dual Opteron box |
104 |
I'd like to use confcache, seems that it might save some time while |
105 |
building the 450 or so packages that I now need to start over on |
106 |
rebuilding. Of course if confcache had worked correctly I wouldn't need |
107 |
to start over. |
108 |
|
109 |
I've looked in the man pages and searched the web for confcache |
110 |
information, but haven't found much. |
111 |
|
112 |
1. Should I have known that I needed to do something to confcache after |
113 |
upgrading gcc, is that "common knowledge" that I missed somehow? |
114 |
|
115 |
2. Can I safely re-enable confcache before restarting the emerge -e |
116 |
world? |
117 |
|
118 |
I did find this http://bugs.gentoo.org/show_bug.cgi?id=133902, and |
119 |
the comments in the "master" confcache bug seem to indicate that the |
120 |
"error" is in an ebuild previous to the one that failed. If I had |
121 |
not deleted /var/tmp/confcache/config.cache I'd have sent in the |
122 |
requested information to help resolve this bug. I can't tell from the |
123 |
comments whether or not it's safe to re-enable confcache with a new |
124 |
fresh database, or if there are just some ebuilds that break when it's |
125 |
enabled. |
126 |
|
127 |
3. Is there any info regarding refreshing, updating, maintaining, |
128 |
confcache's cache, specifically how, when, and what to do? |
129 |
|
130 |
|
131 |
TIA |
132 |
Bob Young |
133 |
San Jose, CA. |
134 |
|
135 |
-- |
136 |
gentoo-user@g.o mailing list |