Hi Ian!
This is a bug with strip -g which makes the stripped object wrong. HJ has checked in a patch, however it just makes the linker abort, rather than segfault, so at least we know what is wrong.
Thanks for taking the time to look into this!
It seems #342777 is in the same basket, so who knows how many libraries have been built with this incorrect strip version. Is there any way to find out?
I have slapped together a script that extracts this information from the buildd logs for binary packages in 'libs', you can take a look here:
http://people.debian.org/~sharky/toolchain/ia64.htmlThis is a quick & dirty script, it probably needs some cleaning up, but it's a start. Moreover, it should be noted that it only works if the package was built using one of the buildds, so i386 output would be pretty meaningless..
If you find the data useful I can set up a daily cron job to keep the information up to date.
Cheers, Jeremy -- http://www.jerryweb.org/ : JerryWeb.org http://sailcut.sourceforge.net/ : Sailcut CAD http://opensource.polytechnique.org/ : Polytechnique.org Free Software