openldap client GSSAPI authentication segfaults in fbsd8stable i386

[George Mamalakis]

On 12/2/2010 8:48 πμ, jhell wrote:
>
>
> This is a lot of information to consume.
>
> Lets start with this:
>
> All of the machines in question are of some form of FreeBSD 8.
>
> You enter gdb and very clearly it starts whining about a segfault & 
> libc.so.7
>
> Did you happen to upgrade these clients & server from FreeBSD 7 ?
>
> AFAIK the libc version on FreeBSD 8 was bumped to libc.so.8
>
> If you upgraded and from source did you run make delete-old and make 
> delete-old-libs after your make install and after your mergemaster ?
>
> Will wait here for results.
>
> Best wishes.
>

jhel,

None of my machines is upgraded from fbsd7. As I stated in my previous 
email, two of them are built from January's snapshot (no 
updates/upgrades); My gdb example is running on one of these machines. 
Nevertheless, on all machines that segfault, the outcome of gdb is the 
same. The rest machines are built either from fbsd8-release and upgraded 
to the latest stable sources, except from my laptop which was installed 
from fbsd8-beta1 and has been upgraded frequently since then. Its 
current binaries are built from January's 25 sources.

Either way, since you asked, I always use make delete-old and make 
delete-old-libs when upgrading.

Lastly, to make a long story short, what I want everybody to keep from 
my latest email is that 32bit and 64bit behave differently. 32bit 
segfaults, but not always..and it definately behaves differently when 
built on vmware esxi4 then when built on virtual box. Then again, amd64, 
on the same esxi host works correctly in only one of the two machines. 
Lastly, I want someone who knows more about the sources in question than 
I do, to see what gdb shows, and that after the little hack we made, 
instead of a segfault we got the same strange behavior as that of the 
"brokent" amd64 box.

Cheers once more.

mamalos