Hurd on Mach on GNU/Linux (verion 0.0.0)

[John Tobey]

Hi all,

I've been toying with this for a few weeks and am no longer too
ashamed to tell y'all about it.  I want to develop Hurd programs but
am too lazy to reboot or use two computers and too brainwashed by RMS
ideology to spring for vmware.  Plus I am kind of interested to know
about Mach.

Put this all together and stir in some free weekends, and the result
is at:
http://tobeyhutchison.com/jtobey/Hurd/gnumach-otop.tar-0.0.0.bz2

What this hopes to be is enough of GNUMach ported to POSIX/Linux to be
able to run the Hurd binaries where they have been sitting on my disk
since the last time I booted them up.

What it is currently, on my machine at least:

   * compiles

   * initializes kernel data structures

   * implements task_create()/pmap_create() using fork and shared
     memory

   * implements thread_create()/pcb_init() by telling the subprocess
     to execute clone() (a pthreads implementation is half-there, but
     may never really work)

   * jumps to the ext2fs.static entry point

   * handles page faults by telling the subprocess to execute mmap()
     on a shared file descriptor

   * dispatches system calls

   * ext2fs.static loops indefinitely, slowly eating stack (which gets
     extended through page faults) and executing mach_reply_port() and
     mach_msg_trap()

   * mach_msg_trap() returns MACH_RCV_INVALID_DATA (perhaps because
     the specified root device does not exist?)

I do not guarantee that it will do even this much on your system; I
guarantee the opposite.  This code is not clean.  You do not want to
look too closely at this code.  However, I thought people might be
interested.

Regards,
-John

ps, OTOP means "On Top Of POSIX" but there are a few Linux and i386
dependencies still in there.

-- 
John Tobey <jtobey@john-edwin-tobey.org>