[MonetDB-users] Problem with BSS initialisation.
Hello.
I've spent quite some time lately debugging my issue with OCaml binding of MonetDB libraries:
http://www.mail-archive.com/caml-list@yquem.inria.fr/msg05449.html
It seems that the BSS is not properly initialised to 0 when dynamically loaded. It seems that the code in mal_box.c relies on the fact that BSS is initialised to 0 when loaded.
I'm still trying to figure out how exactly is the libmonetdb5.so library loaded. This is still an obscure point. ltrace did not provide useful information.
Advice welcome.
Guillaume Yziquel a écrit :
Hello.
I've spent quite some time lately debugging my issue with OCaml binding of MonetDB libraries:
http://www.mail-archive.com/caml-list@yquem.inria.fr/msg05449.html
It seems that the BSS is not properly initialised to 0 when dynamically loaded. It seems that the code in mal_box.c relies on the fact that BSS is initialised to 0 when loaded.
In fact, this is not the case. It is a symbol collision with libncurses. There's also a box function there, living in the .text data segment.
Thanks to Csaba Halasz (Jester01 on ##asm) for helping out with debugging.
Guillaume Yziquel wrote:
Guillaume Yziquel a écrit :
Hello.
I've spent quite some time lately debugging my issue with OCaml binding of MonetDB libraries:
http://www.mail-archive.com/caml-list@yquem.inria.fr/msg05449.html
It seems that the BSS is not properly initialised to 0 when dynamically loaded. It seems that the code in mal_box.c relies on the fact that BSS is initialised to 0 when loaded.
In fact, this is not the case. It is a symbol collision with libncurses. There's also a box function there, living in the .text data segment.
Thanks to Csaba Halasz (Jester01 on ##asm) for helping out with debugging.
Great find! A patch will be checked into the Feb2010 release branch Thanks. Martin
participants (2)
-
Guillaume Yziquel -
Martin Kersten