AIX prefix bootstrap issues
Hello Michael,
Yesterday I swtiched back and forth. And my e-mail (and head) got screwed up. Sorry for this. I need to keep cool and clam in front of a computer. I don't have LIBPATH set. I can access a AIX5.2 box and a AIX5.3 one. I tried prefix-launcher on both. Let me summarize what I did yesterday in detail. 1. In the AIX5.2 box, z. CONFIG_SHELL=<some path>/bash a. The gcc-4.2.4 step in prefix-launcher fails (some hpp file cannot be found. It is suggested to be a tar error, installing tar 1.15 does not solve the problem inside prefix-launcher) , while moving the gcc tree outside prefix-launcher and compiling independently works. I don't know why. The log of failing gcc-4.2.4@prefix-launcher is missing, cuz I deteled the buildroot tree, sorry. Is it worth the time to reproduce? b. With gcc-4.2.4, I set in prefix-laucher "make config gcc_SUPPLY=EXTERNAL" and went on. From this point on I use the new gcc-4.2.4 just compiled. The coreutils-6.9 step fails. I moved the coreutils source tree outside and tried to build independently, but failed again. Jee, I have also deleted the log file and the source tree. Very bad habit, I realize. c. Ignoring failure of coreutils-6.9, I went on with "make config coreutils_SUPPLY=EXTERNAL". The prefix-laucher built till end. d. I used eprefix-bootstrap generated by prefix-laucher. The bootstrap stops in gcc-4.2.4 stage. The error message indicates "install -m0700" on AIX is not compatible with that of GNU. I remembered I don't have coreutils compiled. e. I emerged the most recent coreutils and findutils in portage with success. f. eprefix-bootstrap again, gcc-4.2.4 failed in a later stage. I did not inspect carefully yet. I am regenerating the log file. g. I thought I could circumvent the gcc-4.2.4 by emerging other versions of gcc. First I tried gcc-4.3. It have a large dependency tree. The build goes to help2man and got stuck, complaining it need perl5.8 to start with. I couldn't understand this: I have perl5.10 in Prefix and perl5.8 is masked, why a most recent version of help2man in portage cannot meet its dependency within portage? I then made a symlink to perl5.8 on the host (/opt/freeware/bin/perl5.8). help2man compiles but the shebang test failed. Now I can't go beyond help2man. h. Then I tried gcc-4.1 in portage, it fails with "Arg list too long". As pointed out by Michael, it is not caused by /bin/sh, but a global parameter controling the argument length limit in AIX. I contact with the administrator and wait. Log file: http://www.awa.tohoku.ac.jp/~benda/GentooAIX52/gcc-4.1.build.log Therefore, now I cannot go beyond emerging gcc on the AIX5.2 box. 2. In the AIX5.3 box, z. CONFIG_SHELL=<some path>/bash a. Similar as above gcc-4.2.4@prefix-launcher fails, with some ".hpp" file not found. I moved out, "make distclean", built, succeeded. Failed log file in prefix-launcher: http://www.awa.tohoku.ac.jp/~benda/GentooAIX53/gcc-4.2.4.build.log b. "make config gcc_SUPPLY=EXTERNAL" then continue with prefix-launcher using newly built gcc-4.2.4. Fails at ncurses. I seems that my gcc-4.2.4 is not functioning well? cstdio can be corrupted? Log file: http://www.awa.tohoku.ac.jp/~benda/GentooAIX53/ncurses.build.log Thank you very much for your help. And thank you for your comments and criticize: It made me rethink of my situation and make things clear instead of trying things out blindly. Yours, Benda -- XU Benda Research Center for Neutrino Science Tohoku University JAPAN http://www.awa.tohoku.ac.jp/~benda |
AIX prefix bootstrap issues
Dear guys,
heroxbd@gmail.com writes: > f. eprefix-bootstrap again, gcc-4.2.4 failed in a later stage. I did not > inspect carefully yet. I am regenerating the log file. The log file is at: http://www.awa.tohoku.ac.jp/~benda/GentooAIX52/gcc-4.2.4.emerge.log It fails very early. Should I generate some more verbose log file? How? Yours, -- XU Benda Research Center for Neutrino Science Tohoku University JAPAN http://www.awa.tohoku.ac.jp/~benda |
AIX prefix bootstrap issues
On 04/09/10 09:54, heroxbd@gmail.com wrote:
> Dear guys, > > heroxbd@gmail.com writes: > >> f. eprefix-bootstrap again, gcc-4.2.4 failed in a later stage. I did not >> inspect carefully yet. I am regenerating the log file. > > The log file is at: > > http://www.awa.tohoku.ac.jp/~benda/GentooAIX52/gcc-4.2.4.emerge.log from that log i see (somewhere in the middle): make[2]: execvp: /home/k/benda/local/bin/bash: Arg list too long make[2]: *** [all-stage1-libiberty] Error 127 make[2]: *** Waiting for unfinished jobs.... so it all boils down to the argument thing...? regards, markus > > It fails very early. Should I generate some more verbose log file? How? > > Yours, |
| All times are GMT. The time now is 07:13 PM. |
VBulletin, Copyright ©2000 - 2013, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.