Bug 204175 - print/fontforge poudriere build error
Summary: print/fontforge poudriere build error
Status: Closed FIXED
Alias: None
Product: Ports & Packages
Classification: Unclassified
Component: Individual Port(s) (show other bugs)
Version: Latest
Hardware: i386 Any
: --- Affects Some People
Assignee: Matthias Andree
URL:
Keywords: needs-qa
Depends on:
Blocks:
 
Reported: 2015-10-31 08:20 UTC by Ivan Klymenko
Modified: 2016-05-26 21:59 UTC (History)
4 users (show)

See Also:
cyberbotx: maintainer-feedback+
cyberbotx: maintainer-feedback+

Attachments
fontforge-i386.log (16.94 KB, text/x-log)
2015-10-31 08:20 UTC, Ivan Klymenko 		no flags Details
print/fontforge_i386.log (16.40 KB, text/plain)
2015-11-05 07:57 UTC, Ivan Klymenko 		no flags Details
splineutil2-xxxxxx.c.bz2 (374.33 KB, application/x-bzip)
2015-11-05 21:00 UTC, Ivan Klymenko 		no flags Details
splineutil2-xxxxxx.sh (4.69 KB, application/x-shellscript)
2015-11-05 21:01 UTC, Ivan Klymenko 		no flags Details
Diff to Makefile of print/fontforge (1.06 KB, patch)
2016-04-15 16:45 UTC, Hiroo Ono 		no flags Details | Diff
poudriere log for i386 head (as of r297216) (35.49 KB, application/x-xz)
2016-04-15 16:46 UTC, Hiroo Ono 		no flags Details
Show Obsolete (1) View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Ivan Klymenko 2015-10-31 08:20:34 UTC 
Created attachment 162634 [details]
fontforge-i386.log
Comment 1 Naram Qashat 2015-11-03 14:04:18 UTC 
I have been getting fallout messages about this already, but I really don't know what I could do about it. For one, this is on HEAD and I don't have a HEAD system at all so I can't get a full log, plus I don't know how stable HEAD is for this sort of thing. For two, this seems like something that should go to the developers of fontforge and not me. I can try and see if this happens with a 10-i386 system as well, but if it doesn't, then I still won't know how to handle this.
Comment 2 Kubilay Kocak freebsd_committer freebsd_triage 2015-11-03 15:16:34 UTC 
CC last committer in case they can help
Comment 3 Naram Qashat 2015-11-04 02:01:32 UTC 
Well, I can tell you guys this, it doesn't fail to build in a 10-i386 poudriere jail, so the problem appears to be specific to HEAD-i386.
Comment 4 Ivan Klymenko 2015-11-04 18:40:23 UTC 
After:
1. upgrade poudriere jails to snapshots from date 02.11.2015;
2. removed all options where they are OpenMP;
3. deleted in poudriere all the old packages;
4. rebuilt the entire repository.

problem is not reproduce.

conclusion: port option OpenMP - evil.

Thanks.
Comment 5 Ivan Klymenko 2015-11-05 07:57:50 UTC 
Created attachment 162800 [details]
print/fontforge_i386.log
Comment 6 Ivan Klymenko 2015-11-05 07:58:54 UTC 
Unfortunately the problem reproduced.
Comment 7 Dmitry Marakasov freebsd_committer freebsd_triage 2015-11-05 20:06:12 UTC 
CC dim@.

Usually you should collect the files failure log mentions (preprocessed source and build script) and submit them to clang bugzilla.
Comment 8 Ivan Klymenko 2015-11-05 20:32:26 UTC 
(In reply to Dmitry Marakasov from comment #7)
Alas, I can not intercept these files in the build in poudriere
/tmp/splineutil2-2024cd.c
/tmp/splineutil2-2024cd.sh
Comment 9 Dmitry Marakasov freebsd_committer freebsd_triage 2015-11-05 20:38:50 UTC 
poudriere testport -i
Comment 10 Dimitry Andric freebsd_committer freebsd_triage 2015-11-05 20:50:13 UTC 
Yes, I reported this problem upstream a long time ago:

https://llvm.org/bugs/show_bug.cgi?id=21903

For some reason, this bug did not get much traction, and it is still open after almost a year.  Currently, I have no ETA on when this gets fixed.
Comment 11 Ivan Klymenko 2015-11-05 21:00:53 UTC 
Created attachment 162823 [details]
splineutil2-xxxxxx.c.bz2
Comment 12 Ivan Klymenko 2015-11-05 21:01:17 UTC 
Created attachment 162824 [details]
splineutil2-xxxxxx.sh
Comment 13 Ivan Klymenko 2015-11-05 21:01:31 UTC 
(In reply to Dmitry Marakasov from comment #9)
Thanks!
Comment 14 Hiroo Ono 2016-04-15 16:43:37 UTC 
clang36 (and llvm36) builds fontforge fine on head/i386.
So, the workaround may be to build with it on head/i386.
Comment 15 Hiroo Ono 2016-04-15 16:45:01 UTC 
Created attachment 169350 [details]
Diff to Makefile of print/fontforge
Comment 16 Hiroo Ono 2016-04-15 16:46:30 UTC 
Created attachment 169352 [details]
poudriere log for i386 head (as of r297216)
Comment 17 Hiroo Ono 2016-05-02 14:32:56 UTC 
I proposed a workaround to use clang36 instead of clang37 or clang38 on i386.
How about this workaround?
Comment 18 Naram Qashat 2016-05-02 14:39:21 UTC 
I suppose if the patch works, I'm fine with it.
Comment 19 commit-hook freebsd_committer freebsd_triage 2016-05-26 21:57:53 UTC 
A commit references this bug:

Author: mandree
Date: Thu May 26 21:57:29 UTC 2016
New revision: 415915
URL: https://svnweb.freebsd.org/changeset/ports/415915

Log:
  Fix build on HEAD-i386, using clang 3.6, not newer.

  PR:		204175
  Submitted by:	Hiroo Ono
  Reported by:	Ivan Klymenko
  Approved by:	Naram Qashat (maintainer)

Changes:
  head/print/fontforge/Makefile