meinproc4 segfaults at end of build

[nerdling (Jeremy Lavergne)]

Same old worthless crash from meinproc4. This drives the KDE folks crazy. This one is from kdebase4:

/bin/sh: line 1: 95059 Segmentation fault      /opt/local/bin/meinproc4 --check --cache /opt/local/var/macports/build/_Users_aeetes_dports_kde_kdebase4/kdebase4/work/build/konsole/doc/manual/index.cache.bz2 /opt/local/var/macports/build/_Users_aeetes_dports_kde_kdebase4/kdebase4/work/kdebase-4.6.5/konsole/doc/manual/index.docbook

[mkae (Marko Käning)]

Cc Me!

[mkae (Marko Käning)]

I see the same suddenly with skrooge (r81433). :-(

[nerdling (Jeremy Lavergne)]

The work around I recommend is running the meinproc4 command manually. You can do this by becoming root and then copy/pasting the line of code from the MacPorts log.

[mkae (Marko Käning)]

I guess due to r81472 this ticket can be considered closed.

[nerdling (Jeremy Lavergne)]

Potential fix applied in r81472.

[nerdling (Jeremy Lavergne)]

This continues to happen in lots of KDE ports even with docbook-xsl-ns added.

[nerdling (Jeremy Lavergne)]

If anyone has a chance, please try rebuilding an impacted KDE port after adding this:

build.asroot yes

[nerdling (Jeremy Lavergne)]

By using build.asroot I was able to build marble without meinproc setfaulting. r81664.

[nerdling (Jeremy Lavergne)]

The issue is now relegated to only when the user requests variant +docs. However, it's still a problem at that point.

[ryandesign (Ryan Carsten Schmidt)]

The ​buildbot failed to build kdenlive because of the issue reported in this ticket.

[mkae (Marko Käning)]

Same thing happened again for me with kmymoney4-devel port. I had to use build.asroot set to yes, otherwise it wouldn't work, although it worked on my own machine without trouble. It looks like that buildbot's environment seems to be different for some reason, which triggers this bug.

It looks like we'll have to try to hunt down this bug in meinproc4's sources ourselves, since the KDE folks aren't really focused on Mac OS X...

[jmroot (Joshua Root)]

Is there a crash log anywhere for this?

[mkae (Marko Käning)]

Except the above error message there is no other crash log...

The question is how could one generate more info to debug this issue?

I tried debugging this with gdb, but got NO error:

$ sudo gdb /opt/macports-test/bin/meinproc4
GNU gdb (GDB) 7.5
Copyright (C) 2012 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-apple-darwin10.8.0".
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>.
(gdb) set args --check --cache /opt/macports-test/var/macports/build/_Users_marko_WC_MacPorts_ports_kde_kmymoney4-devel/kmymoney4-devel/work/build/doc/index.cache.bz2 /opt/macports-test/var/macports/build/_Users_marko_WC_MacPorts_ports_kde_kmymoney4-devel/kmymoney4-devel/work/kmymoney/doc/index.docbook
.
.
.
/opt/macports-test/share/xsl/docbook-xsl/html/math.xsl:5: element stylesheet: error : Reference to namespace 'mml' not in scope
[Inferior 1 (process 44416) exited normally]
(gdb)

[mkae (Marko Käning)]

At least for kmymoney4-devel r103781 should solve this issue by correctly omitting the API handbook generation eventually.

[michaelld (Michael Dickens)]

Cc Me!

[nerdling (Jeremy Lavergne)]

Replying to ryandesign@…:

The ​buildbot failed to build kdenlive because of the issue reported in this ticket.

Consider attaching the build logs: when the slaves roll over the build numbers or otherwise get cleaned, we lose all the old build logs.

[nerdling (Jeremy Lavergne)]

Replying to mk@…:

I tried debugging this with gdb, but got NO error:

$ sudo gdb /opt/macports-test/bin/meinproc4

Can you run this again as the build user rather than root? Perhaps using sudo -u macports gdb… or simply su'ing to the macports user.

[mkae (Marko Käning)]

Just a few moments ago I updated ​https://bugs.kde.org/show_bug.cgi?id=261509 ...

Well, it's impossible to know when and where meinproc4 is going to crash...

My current approach is to simply try it a couple of times until it succeeds. :-(

[jmroot (Joshua Root)]

Replying to mk@…:

Except the above error message there is no other crash log...

Logs for crashes during builds should end up in /Library/Logs/DiagnosticReports (I just confirmed this with a test port that builds and runs a test program with a deliberate null pointer dereference).

[ryandesign (Ryan Carsten Schmidt)]

Similar meinproc4 crash reported in #41575.

[mkae (Marko Käning)]

Replying to jmr@…:

Logs for crashes during builds should end up in /Library/Logs/DiagnosticReports

Good to know! Will check that out next time it happens. Thanks for the hint.

[mkae (Marko Käning)]

Marked #41575 and #41326 as duplicates of this issue.

[mkae (Marko Käning)]

Fixed in r114083 (see #41326)!