[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: __libc_stack_end is broken?
From: |
Samuel Thibault |
Subject: |
Re: __libc_stack_end is broken? |
Date: |
Wed, 29 Aug 2007 09:13:00 +0200 |
User-agent: |
Mutt/1.5.11 |
Hi,
Michael Casadevall, le Tue 28 Aug 2007 20:43:02 -0400, a écrit :
> While fixing Ruby on Hurd, I discovered __libc_stack_end is (seemingly)
> broken. It seems on Linux, this points to the end the stack for the
> process, but on Hurd this is broken (it points at an invalid location
> outside of a processes address space).
That's odd: the code that sets it is shared between Linux & the Hurd:
sysdeps/i386/elf/start.S that pushes %esp.
> A quick gogle says though that this symbol shouldn't even be exported,
> so I'm not sure what the proper behavior in this regard is.
The Hurd still exports it because sysdeps/mach/hurd/dl-sysdep.c define
it externally, while the generic elf/dl-sysdep.c correctly sets it
hidden.
> Any comments would be appericated because it will also influence how
> the ruby patch is done.
Both issues should get fixed :)
Samuel