Subject: Re: Help compiling i386 kernel pls.: Elf32_Ehdr
To: \"Gary \" D. \" Timuss\" <G.D.Timuss@uea.ac.uk>
From: Jason Thorpe <email@example.com>
Date: 11/04/1996 08:15:18
On Fri, 1 Nov 1996 12:18:20 +0000 (GMT)
Gary "D." Timuss <G.D.Timuss@uea.ac.uk> wrote:
> (a) How did anyone _know_ this? I didn't see anything on this list,
> /doc/CHANGES doesn't say anything, and options(4) says nothing either.
> I now see these are contained within the sample options files, but there
> is no indication of their essentialness.
At least one message on this subject was posted here (current-users) by
Chris Demetriou. The pertinent entry in doc/CHANGES is:
Made EXEC_AOUT and EXEC_SCRIPT no longer mandatory. Changed
COMPAT_LINUX, COMPAT_OSF1, COMPAT_SVR4 and COMPAT_ULTRIX
so they no longer implicitly pull in executable
format support code. [cgd 961002]
> (b) If COMPAT_LINUX _requires_ (and logically implies) EXEC_ELF, why does
> one need to specify both? And if "machine i386" _requires_ EXEC_AOUT and
> EXEC_SCRIPT, then these aren't "options" either.
The point of the changes was that the old scheme was wasteful, and
somewhat messy in dealing with exec formats for architectures which
don't support all formats. If COMPAT_LINUX _requires_ EXEC_ELF, that's
a bug; if you only want to exec a.out Linux programs, you shouldn't have
to include ELF support. I think Chris has mentioned that the compat code's
exec stuff needs to be cleaned up a bit yet...
Also, EXEC_SCRIPT is a real option. I found this out the "hard way"
by forgetting to put it in my sparc's kernel. The system happily booted
into multi-user mode (/etc/rc and friends aren't invoked directly, but
are passed to a shell :-)
Jason R. Thorpe firstname.lastname@example.org
NASA Ames Research Center Home: 408.866.1912
NAS: M/S 258-6 Work: 415.604.0935
Moffett Field, CA 94035 Pager: 415.428.6939