Date: Fri, 6 Feb 1998 11:09:33 -0100 (GMT) From: Remy NONNENMACHER <remy@synx.com> To: Richard Foulk <richard@pegasus.com> Cc: freebsd-emulation@FreeBSD.ORG Subject: Re: NetBSD emulation Message-ID: <Pine.A32.3.91.980206104943.10787G-100000@rs1> In-Reply-To: <199802051807.IAA09715@pegasus.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 5 Feb 1998, Richard Foulk wrote: > } > NetBSD *IS* emulated by FreeBSD. The problem is that someone (Net ? Free > } > ?) reversed the exec magic number. This means that FreeBSD recognizes the > } > NetBSD Magic, load and starts the binary. The binary try to load the > } > ld.so dynamic linker *BUT*, before, it checks the ld.so magic number and > } > find it reversed !! (then it dies, spiting something about 'bad ld.so > } > magic'). > } > > > So what about fudging a copy with the magic number modified? > AFAIK (as it was time ago when i switched from Net to Free, when i got the problem) the NetBSD emulation in Free have no 'file lookup' on a specific directory (but it have for Linux and IBCS2). I think it's due to the fact that Net and Free are quiet the same at syscalls level and emul dvl decided not to make a specific lookup environment. I remember i made an horrible hack somewhere to trap the open, by a NetBSD program, of the specific file 'ld.so' to use an /emul lookup. It was so ugly and unreliable that i abandonned the whole thing. (better recompile ...). I can't precisely remember but i believe the a.out Magic Number reversing was done beetween two versions of NetBSD (1.0 to 1.1, i guess). Maybe someone could confirm this historic point ? (I also remember the respect i felt for the ones dealing with shared libs and exec loading... It's far above a simple 'load text, load data, create stack, return to start point'). > > Richard >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.A32.3.91.980206104943.10787G-100000>