On Jul 6, Aurelien Jarno ([EMAIL PROTECTED]) wrote: > On Sun, Jul 06, 2008 at 03:48:01PM +1000, Aníbal Monsalve Salazar wrote: > > Package: openjade > > Severity: important > > Version: 1.4devel1-18 > > > > There was an error while trying to autobuild libusb on arm: > > > > Automatic build of libusb_2:0.1.12-12 on cats by sbuild/arm 98 > > Build started at 20080611-2333 > > > > [...] > > ** Using build dependencies supplied by package: > > Build-Depends: debhelper (>= 5.0.22), autotools-dev, pkg-config, docbook, > > docbook-dsssl > > > > [...] > > Making all in doc > > make[3]: Entering directory > > `/build/buildd/libusb-0.1.12/obj-arm-linux-gnu/doc' > > rm -rf html > > mkdir html > > openjade -t sgml -d ../../doc/website.dsl\#html ../../doc/manual.sgml > > make[3]: *** [html/index.html] Segmentation fault > > make[3]: Leaving directory > > `/build/buildd/libusb-0.1.12/obj-arm-linux-gnu/doc' > > make[2]: *** [all-recursive] Error 1 > > > > I have already started to investigate this problem some time ago. The > segfaults occur in lib/Location.cxx in: > > InputSourceOrigin *InputSourceOrigin::make() > > while instancianting InputSourceOriginImpl. > > The problem does occur with gcc-4.3 and gcc-4.2, but does not occur with > gcc-4.1. Lowering the optimization level to -O0 does not change > anything. I don't know yet if it is a gcc or an opensp bug, but the > later is possible given the high number of warnings issued with -Wall.
I went through the painstaking effort of using -Wall -Werror to compile it and eventually made all the errors go away, and it still segfaulted on arm. It does this on no architecture other than arm. I'm at a loss for what to do, that's why I tagged bug 479273 "help". -- Neil Roeth -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
